Science 2013-7-19

CONTENTS Volume 341 Issue 6143
EDITORIAL BOOKS ET AL.

215 Pathways for Conservation 238 Toms River
Susan M. Haig et al. D. Fagin, reviewed by J. Tuomisto
239 The Cost Disease
NEWS OF THE WEEK W. J. Baumol, reviewed by A. Chandra
220 A roundup of the week’s top stories
EDUCATION FORUM
NEWS & ANALYSIS 240 Crowdsourcing and Curating
223 Tampered Data Cast Shadow Online Education Resources
on Drug Trial D. Porcello and S. Hsi
224 Latest Skirmish Over Ancestral Violence PERSPECTIVES

Strikes Blow for Peace
>> Report p. 270 242 Trans-HSF1 Express
V. Gandin and I. Topisirovic
225 Spain’s Research Council Approaches >> Research Article p. 250
Bankruptcy
243 Droplets Out of Equilibrium
226 Ever-Bigger Viruses Shake Tree of Life pages 230 & 234
T. M. Hermans et al.
>> Report p. 281 >> Report p. 253
227 Germany Debates How to 244 100 Years of Atomic Theory
Strengthen Universities D. C. Clary
229 Mars Rover Plans Roll While Asteroid 245 Enhancing Pluripotency and Lineage
Acrimony Continues Specification
W. Xie and B. Ren
NEWS FOCUS 247 Multiscale Design and Integration
230 BIOGEOGRAPHY of Sustainable Building Functions
Battle for the Americas M. P. Gutierrez and L. P. Lee
Salvage Paleontology on the Seaway >> Science Podcast
>> Science Podcast
234 The Amazon in 4D REVIEW
249 Applications of Acceptorless
LETTERS Dehydrogenation and Related page 240
236 Turkey Must End Violent Response Transformations in Chemical Synthesis
C. Gunanathan and D. Milstein
to Protests
Review Summary; for full text:
E. Altindis et al.
http://dx.doi.org/10.1126/science.1229712
Optimizing Peer Review of Software Code
P. Sliz and A. Morin
RESEARCH ARTICLE
Response 250 Tight Coordination of Protein Translation
L. N. Joppa et al. ON THE WEB THIS WEEK
and HSF1 Activation Supports the Anabolic
237 CORRECTIONS AND CLARIFICATIONS Malignant State >> Science Podcast
237 TECHNICAL COMMENT ABSTRACTS S. Santagata et al. Listen to stories on interdisciplinary buildings,
Chemical and genetic screening links measuring the martian atmosphere, an
ribosome activity levels and a transcriptional isthmus miracle, and more.
regulator in malignant cells.
Research Article Summary; for full text: >> Find More Online
http://dx.doi.org/10.1126/science.1238303 Check out Science Express, our podcast,
>> Perspective p. 242 videos, daily news, our research journals, and
Science Careers at www.sciencemag.org.
CONTENTS continued >>
COVER DEPARTMENTS
Enhanced transmission electron microscopy image 214 This Week in Science
of a “Pandoravirus” particle (length: 1.2 micrometers). 216 Editors’ Choice
Despite obeying all criteria to discriminate viruses from 218 Science Staff
cells (no ribosome, no adenosine triphosphate production, 299 New Products
no division), these Acanthamoeba viruses, unrelated 300 Science Careers
to previously recognized virus families, exhibit genomes
of up to 2.5 megabases, encoding more genes than some
microsporidia. See pages 226 and 281.
Image: O. Poirot/Structural and Genomic Information Laboratory,
CNRS Aix-Marseille Université
www.sciencemag.org SCIENCE VOL 341 19 JULY 2013 211

Published by AAAS
CONTENTS
REPORTS 275 Loss of Function of the Melanocortin 2

251 Bright Hot Impacts by Erupted Fragments Receptor Accessory Protein 2 Is Associated
Falling Back on the Sun: A Template with Mammalian Obesity
for Stellar Accretion M. Asai et al.
F. Reale et al. Disruption of a protein required for effective
Analysis of plasma downflows after a solar signaling by a melanocortin receptor causes
eruption suggests that these events can be severe obesity in mice.
used to understand stellar accretion. 278 Developmental Control of the
253 Switchable Static and Dynamic Melanocortin-4 Receptor by MRAP2
Self-Assembly of Magnetic Droplets Proteins in Zebrafish
on Superhydrophobic Surfaces J. A. Sebag et al.
J. V. I. Timonen et al. A study in zebrafish sheds light on the
Magnetic droplets oscillate between signaling properties of a protein implicated
static and dynamic self-assembly patterns in severe obesity in mice.
in a magnetic field. 281 Pandoraviruses: Amoeba Viruses
>> Perspective p. 243 with Genomes Up to 2.5 Mb
page 251
257 Ultrahigh Magnetoresistance Reaching That of Parasitic Eukaryotes
at Room Temperature in Molecular Wires N. Philippe et al.
R. N. Mahato et al. Giant viruses that infect Southern
The conduction of molecular wires embedded Hemisphere Acanthamoeba and
in a zeolite host crystal is almost entirely are visible under the light microscope
blocked in small magnetic fields. have been identified.
>> News story p. 226
260 Isotope Ratios of H, C, and O in CO2
and H2O of the Martian Atmosphere 286 Sept4/ARTS Regulates Stem Cell Apoptosis
C. R. Webster et al. and Skin Regeneration
>> Science Podcast Y. Fuchs et al.
Elimination of a proapoptotic gene increases
263 Abundance and Isotopic Composition hair follicle stem cells and improves skin
of Gases in the Martian Atmosphere regeneration and wound repair in mice.
from the Curiosity Rover 290 How the Red Queen Drives Terrestrial
P. R. Mahaffy et al.
Mammals to Extinction
Data from the Curiosity rover provide a
T. B. Quental and C. R. Marshall
detailed account of the chemical and isotopic
Loss of diversity among Cenozoic land
composition of Mars’ atmosphere.
mammals suggests a failure to keep pace
266 Ice-Shelf Melting Around Antarctica with a deteriorating environment.
E. Rignot et al.
Basal melting of Antarctic ice shelves 292 Exceptional Convergence on the
accounts for as much mass loss as does Macroevolutionary Landscape in Island
iceberg calving. Lizard Radiations
D. L. Mahler et al.
270 Lethal Aggression in Mobile Forager Bands A comparative study describes morphological
and Implications for the Origins of War evolution across the entire Greater Antillean
page 292
D. P. Fry and P. Söderberg anole lizard fauna.
Nomadic foragers are less warlike than
assumed, suggesting that war may not have 295 Predicting and Manipulating Cardiac Drug
been an early component of human behavior. Inactivation by the Human Gut Bacterium
>> News story p. 224 Eggerthella lenta
H. J. Haiser et al.
273 Interactions of Multisensory Components The heart drug digoxin can be inactivated
Perceptually Rescue Túngara Frog by a strain of gut microbe bearing a cardiac
Mating Signals glycoside reductase operon.
R. C. Taylor and M. J. Ryan
Mating signals that females find unattractive
when presented singly become attractive
when combined.
CREDIT: (BOTTOM) MIGUEL A. LANDESTOY
SCIENCE (ISSN 0036-8075) is published weekly on Friday, except the last week in December, by the American Association for the Advancement of Science, 1200
New York Avenue, NW, Washington, DC 20005. Periodicals Mail postage (publication No. 484460) paid at Washington, DC, and additional mailing offices. Copyright © 2013
by the American Association for the Advancement of Science. The title SCIENCE is a registered trademark of the AAAS. Domestic individual membership and subscription (51 issues):
$149 ($74 allocated to subscription). Domestic institutional subscription (51 issues): $990; Foreign postage extra: Mexico, Caribbean (surface mail) $55; other countries (air assist
delivery) $85. First class, airmail, student, and emeritus rates on request. Canadian rates with GST available upon request, GST #1254 88122. Publications Mail Agreement Number
1069624. Printed in the U.S.A.
Change of address: Allow 4 weeks, giving old and new addresses and 8-digit account number. Postmaster: Send change of address to AAAS, P.O. Box 96178, Washington, DC
20090–6178. Single-copy sales: $10.00 current issue, $15.00 back issue prepaid includes surface postage; bulk rates on request. Authorization to photocopy material for
internal or personal use under circumstances not falling within the fair use provisions of the Copyright Act is granted by AAAS to libraries and other users registered with the Copyright
Clearance Center (CCC) Transactional Reporting Service, provided that $30.00 per article is paid directly to CCC, 222 Rosewood Drive, Danvers, MA 01923. The identification code for
Science is 0036-8075. Science is indexed in the Reader’s Guide to Periodical Literature and in several specialized indexes.

Published by AAAS
EDITED BY STELLA HURTLEY however, reexamined the standard cross-cultural
sample, the main repository for behavioral data
on forage bands, and found little evidence for
<< More large-scale conflicts or wars. Instead, the major-
Magnetoresistance ity of incidences of lethal aggression in these
When data is read off your com- societies were homicides driven by a variety of
puter’s hard drive, chances are that factors relevant at the individual or family scale.
the read head is using the phe-
nomenon of magnetoresistance
(MR)—the dependence of electri-
Accessory to Obesity?
cal resistance on applied magnetic Melanocortin receptors are a family of cell
field—to interpret the magnetic membrane receptors that control diverse
signature of the data on the disk. physiological functions. Mutations in the gene
Devices that have the large MR encoding melanocortin 4 receptor (MC4R) are
necessary for such tasks are usu- a cause of familial early-onset obesity. Asai et
ally made of layers of magnetic al. (p. 275) studied the function of an acces-
materials. Mahato et al. (p. 257, sory protein for MC4R signaling, MRAP2, and
Downloaded from www.sciencemag.org on July 18, 2013

published online 4 July) observed found that mice genetically deficient in MRAP2
a large MR effect in a nonmag- develop severe obesity. Sequencing of MRAP2
netic material—organic molecules in unrelated, severely obese humans revealed
squeezed into a zeolite crystal. Im- one individual with a clearly disruptive genetic
portantly for potential future ap- variant, suggesting that MRAP2 mutations might
plications, the effect was observed also be a rare cause of human obesity. In a
at room temperature and at low zebrafish model, Sebag et al. (p. 278) studied
magnetic fields. two paralogs of the MRAP2 accessory protein,
one of which enhanced MC4R responsiveness
to α–melanocyte-stimulating hormone, which
Accretion Analog the origin of these meteorites and implies that regulates feeding and growth.
the current atmospheric reservoirs of CO2 and
Mass flow from a circumstellar disk onto a young H2O were largely established after the period of
star’s surface plays an important role in the early atmospheric loss some 4 billion years ago.
Romancing the Frog
final stages of star formation but the details of In túngara frogs, auditory and visual compo-
this complex process are not well understood. nents of mate calling do not naturally occur
Reale et al. (p. 251, published online 20 June)
Major Meltdown together. Taylor and Ryan (p. 273, published
analyzed a solar flare that led to bright impacts of The ice shelves and floating ice tongues that online 6 June) now show that two signals that are
plasma onto the solar surface. Numerical simula- surround Antarctica cover more than 1.5 million unattractive to female frogs when presented alone
tions suggest that these events can be seen as square kilometers—approximately the size of become highly attractive when presented together.
analogs to accretion of matter onto stars and can the entire Greenland Ice Sheet. Conventional In a kind of “perceptual
thus be used to understand stellar accretion. wisdom has held that ice shelves around rescue,” the unique com-
Antarctica lose mass mostly by iceberg calving, bination of two signals
but recently it has become increasingly clear increased the receiver’s
Mars’ Atmosphere that melting by a warming ocean may also be interest in the previously
important. Rignot et al. (p. 266, published uninteresting signals.
from Curiosity 13 June) present detailed glaciological esti-
The Sample Analysis at Mars (SAM) instrument mates of ice-shelf melting around the entire
on the Curiosity rover that landed on Mars in continent of Antarctica, which show that basal Stem Cells in
August last year is designed to study the chemi- melting accounts for as much mass loss as does
cal and isotopic composition of the martian calving.
Wound Healing
atmosphere. Mahaffy et al. (p. 263) present Although excessive numbers of stem cells (SCs)
CREDITS (TOP TO BOTTOM): NYMUS3D; RYAN TAYLOR
volume-mixing ratios of Mars’ five major atmo- may increase the risk of cancer, elevated SC
spheric constituents (CO2, Ar, N2, O2, and CO) Ancient Warriors numbers may be desirable, at least transiently,
and isotope measurements of 40Ar/36Ar and C for the promotion of tissue repair and regenera-
and O in CO2, based on data from one of SAM’s
or Murderers? tion. Fuchs et al. (p. 286, published online 20
instruments, obtained between 31 August and Some have suggested that the human predilec- June) found that mice deficient for the proapop-
21 November 2012. Webster et al. (p. 260) tion for war is ancient, perhaps dating back totic Sept4/ARTS gene have elevated numbers of
used data from another of SAM’s instruments to the emergence of our species, while others apoptosis-resistant hair follicle SCs and display
obtained around the same period to determine maintain that evidence for such early warring dramatic improvement in wound healing and
isotope ratios of H, C, and O in atmospheric CO2 is scant. Past studies that looked at nomadic regeneration. Inactivation of the caspase inhibi-
and H2O. Agreement between the isotopic ratios foraging bands as models of early humans and tor XIAP, a direct target for the proapoptotic
measured by SAM with those of martian meteor- their potential for conflict concluded that war activity of ARTS, abrogated these phenotypes
ites, measured in laboratories on Earth, confirms is in our blood. Fry and Söderberg (p. 270), and impaired wound healing.
214 19 JULY 2013 VOL 341 SCIENCE www.sciencemag.org

Published by AAAS
THIS WEEK IN SCIENCE
Additional summaries
Setting Hydrogen Free edly diverse transcriptional network involved in Playing the Tape of Life
Oxidation of organic compounds has tradition- oncogenesis. Santagata et al. (p. 250; see the Should the tape of life be replayed, would it
ally been considered to involve the transfer of Perspective by Gandin and Topisirovic) found produce the same music? Many influential evolu-
hydrogen atoms in the molecular framework to that reduced translation may be used to sense tionary biologists, notably Stephen J. Gould, have
an oxidant such as O2, peroxide, or a metal oxide a cell’s metabolic status and regulate transcrip- argued that the answer is “no.” However, patterns
complex. Gunanathan and Milstein (p. 249) tion, in particular by inactivating HSF1 with con- of convergence among different species filling
review the ongoing development of an alternative sequent affects on its targets. Small-molecule similar niches all over the world have argued that
process, in which a catalyst coaxes the H atoms drugs that affected this link were able to inhibit the answer is neither so simple nor perhaps so
to depart on their own in the form of H2. These the growth of transformed cells in culture and of negative. Classic cases of convergence, such as
acceptorless dehydrogenations are appealing an animal tumor model. marsupials on the Australian continent or cichlids
because they generate so little waste. In one across the African rift lakes, have demonstrated
class of reactions, the liberated H2 gas is actively that similar ecological pressures can result in
expelled from the reaction mixture and collected
Zeus’ Revenge species with similar ecological traits. Such classic
for potential use elsewhere. In another class, the Sediment-dwelling amoebae appear to have an examples, however, do not allow for the influence
H atoms return to the source molecule after it unhappy affinity for huge viruses. Giant icosahe- of niche filling based purely on chance. Mahler
has undergone an intermediate transformation in dral Mimiviruses with genomes of the order of 1 et al. (p. 292) take advantage of the well-studied

their absence. megabase (Mb) were first identified in Acanth- species clades of Caribbean anoles to examine
amoeba. Digging into antipodean sediments patterns of adaptation and niche filling across
has once again been fruitful where Philippe species and islands. Across-islands convergence
Magnetic Self-Assembly et al. (p. 281; see the cover) discovered some on a few distinct adaptive peaks (or niches) has
During self-assembly, objects spontaneously enormous viruses in Acanthamoeba, visible by driven diversification of species. Anomalies from
assemble into larger ordered patterns as light microscopy and having genomes up to these ecotypes are only found on the largest, most
observed, for example, in the phase segrega- 2.5 Mb. The Pandoraviruses are phagocytosed diverse islands. Thus, ecological niches power-
by target cells and, after fusing fully shape species and convergence on particular
with the phagosome membrane, forms is an inherent component of adaptation.
their contents are released into the Thus, it seems that the tape of life might play the
cytoplasm where they wreak terrible same music, despite being produced by different
havoc on its nucleus. These viruses instruments.
are encased into a tegument-like
envelope and lack genes for capsid
proteins, and there are no genes
Digoxin Dangers
tion of block copolymers or the assembly of for protein translation, adenosine triphosphate A proportion of patients treated with digoxin,
micrometer-sized objects and components generation, or binary fission—confirming their a cardiac glycoside used to treat heart function
in electronics. In dynamic self-assembly, the classification as viruses. abnormalities, generate the inactive metabo-
ordered patterns require an external energy lite, dihydrodigoxin, resulting in poor efficacy.
source, but still form because of intrinsic Haiser et al. (p. 295) examined a potential
interactions within the system. Timonen et al.
Background Extinction culprit responsible for this transformation—the
(p. 253; see the Perspective by Hermans et al.) Diversity results through both the processes of actinobacterium, Eggerthella lenta—to probe
studied the organization of magnetic droplets, species origination and extinction. However, the microbiota-digoxin interaction. Microbe
in the form of a ferrofluid, placed on a low- studies of extinction have tended to focus on growth was promoted by arginine, and differen-
friction surface. A time-varying magnetic field mass extinctions, despite the fact that the back- tial expression analysis revealed a two-gene car-
transformed the statically arranged droplets ground extinction represents a greater loss in diac glycoside reductase (cgr) operon that was
into a dynamic pattern. terms of the absolute number of extinct taxa. In induced by digoxin in low arginine conditions.
order to identify what factors affect this rate of Not all strains of E. lenta could reduce digoxin
background extinction, Quental and Marshall and, when fecal samples from healthy people
Sensing Reduced (p. 290, published online 20 June) explored were tested, a spectrum of digoxin inactivation
the dynamics of 19 mammalian clades and was detected. When the digoxin-reducing strain
Translation compared the rates of expansions and declines of E. lenta was given to germ-free mice that
The interplay between metabolic pathways among taxa to expected models assuming were fed a high-protein (that is, high-arginine)
and the cellular survival programs that enable random processes. Most clades decline to extinc- diet, digoxin levels stayed high in serum, and
tumors to grow are poorly understood. Heat tion in a “driven” manner—that is, faster than drug inactivation was suppressed.
shock factor 1 (HSF1) coordinates an unexpect- expected by chance alone.
CREDIT: TIMONEN ET AL.
www.sciencemag.org SCIENCE VOL 341 19 JULY 2013 214-b

Published by AAAS
EDITORIAL
Susan M. Haig is a
supervisory wildlife
ecologist with the U.S.
Geological Survey
Forest and Rangeland
Ecosystem Science
Center, Corvallis, OR; Pathways for Conservation
a professor of wildlife
ecology at Oregon State NEXT WEEK, CONSERVATION SCIENTISTS WILL GATHER AT THE INTERNATIONAL CONGRESS FOR
University, Corvallis, OR; Conservation Biology (ICCB) in Baltimore, Maryland, to grapple with the challenges of
and president of the preserving our natural world in the face of a growing and increasingly consumptive human
American Ornithologists' population. The natural world provides countless services, such as clean water, protection
Union, Washington, DC. from flooding, and carbon sequestration, while offering opportunities for new medicines,
E-mail: susan_haig@ foods, and energy production. Yet these valuable services and opportunities are disappear-
usgs.gov. ing along with the species and natural areas that supply them. The needs of a growing human
population must be met while conserving the planet’s natural systems. Accomplishing both
Thomas E. Martin is a
will depend on making clearer connections between scientific results regarding issues such

senior scientist with the
as biodiversity loss and the critical decisions that must be made about conditions that under-
U.S. Geological Survey
lie change, such as greenhouse gas emissions and freshwater availability. The good news is
Montana Cooperative
that today’s conservation scientists are developing innovative tools
Wildlife Research Unit,
and strategies.
Missoula, MT; and a
New technical concepts include applying electronic circuit the-
professor at the University
ory to better understand how environmental features influence
of Montana, Missoula,
the genetic structure of multiple species in a particular landscape.
MT. E-mail: tmartin@
This landscape genetics approach has provided guidance for deci-
usgs.gov.
sions about timber management in the disappearing mature for-
ests of the U.S. Pacific Northwest. Advances in automated wireless
sensors, deployed by the thousands, will reduce the need for labor-
Charles van Riper III intensive manual sampling of water, soil, air, vegetation, and wild-
is a senior scientist life, providing an unprecedented opportunity to track the effects of
and supervisory climate change. Likewise, identifying animal responses to environ-
research ecologist with mental change throughout their life cycle will benefit from the use
the U.S. Geological of unmanned aircraft. This effort will be greatly enhanced when a
Survey at the University global animal tracking system operated from the International Space
of Arizona, Tucson, AZ. Station (called ICARUS) is launched and then expanded with the use of cell phone tech-
E-mail: charles_van_ nology to monitor animal migratory connectivity. Even de-extinction technologies may be
riper@usgs.gov. considered in future efforts.
Successful strategies for maximizing biodiversity while supporting human needs depend
T. Douglas Beard Jr. is on understanding how species differ in their resilience and adaptability to broad environ-
the chief of the U.S. mental change. Those with little plasticity or genetic variance are at highest risk because
Geological Survey of changing conditions; these include corals, amphibians, and island birds. Climate change
National Climate Change may lead to completely new species assemblages, and conservation decision-makers must
and Wildlife Science understand species responses so that responsible actions can be implemented. Another chal-
Center, Reston, VA; and lenge is identifying the responses of invasive species, because they can so easily adapt to
president of the World changing conditions and negatively alter biodiversity. This can be seen in the largely inef-
Council of Fisheries fective efforts to thwart loss of biodiversity in the U.S. Great Lakes as a result of a nonnative
Societies, Bethesda, zebra mussel or in the loss of native plant diversity caused by invasive spotted knapweed.
MD. E-mail: dbeard@ Most importantly, conservation scientists must redouble efforts to communicate their
usgs.gov. research to the public, agencies, and policy-makers in ways that are easily understood and
implemented. This approach is currently playing out in the California State Legislature
as conservation scientists convey the negative effects of lead ammunition on wildlife and
humans. A statewide ban on lead would also substantially boost recovery of the California
CREDIT: HANZL/ISTOCKPHOTO.COM
condor and other scavenging birds and mammals. Overall, conservation decisions must be
made by considering the fair-value impact on the ecosystem as well as the human need for
the resource. If appropriately valued, nature and society should both benefit. The ICCB con-
ference will continue this discussion to identify and address the most important of these
challenges for preserving our natural world.
— Susan M. Haig, Thomas E. Martin, Charles van Riper III, T. Douglas Beard Jr.
10.1126/science.1242710

Published by AAAS
EDITORS’CHOICE
EDITED BY GILBERT CHIN AND MARIA CRUZ
cell differentiation. Both groups hy-

pothesize that the hypermethylation
phenotype is due to the aberrant
accumulation of an oncometabolite
that inhibits DNA-demethylating
enzymes, with succinate being a
strong candidate. — PAK
Cancer Discov. 3, 648 (2013);
ASTRONOMY Cancer Cell 23, 739 (2013).
Merger Relics EPIDEMIOLOGY

Massive galaxies are thought to form through a succession of mergers between smaller galaxies.
Communicable Disease
These interactions are expected to be more frequent in regions where there are fewer galaxies; in
high-density regions, galaxies move faster and mergers are less likely. However, a recent study has In the United States, states have
CREDITS (TOP TO BOTTOM): NASA, HOLLAND FORD (JHU), THE ACS SCIENCE TEAM AND ESA; B. LIM ET AL., PROC. NATL. ACAD. SCI. U.S.A. 110, 25 (3 JUNE 2013) © 2013 NATIONAL ACADEMY OF SCIENCES
revealed that 38% of massive galaxies in heavy clusters of galaxies show features that are consistent the authority to grant exemp-

with the merging of galaxies (such as tidal tails). To try to explain this unexpected result, Yi et al. tions so that children can begin
ran a cosmological-volume simulation to derive the merger history of dark-matter halos, then to attend school without having
constructed semi-analytical models of galaxies to populate those halos, and finally performed been vaccinated against child-
hydrodynamical simulations of galaxy-galaxy mergers to estimate the lifetime of merger features. hood diseases. Medical exemptions
Merger features are expected to last very long in clusters of galaxies. Thus, when observed there, can be granted when a child has
they may not be the result of recent in situ mergers but relics from interactions that happened when a history of allergic reactions or
the galaxies were in a different environment. — MJC is immunocompromised. How-
Astron. Astrophys. 554, A122 (2013). ever, there has been a noticeable
increase in the numbers of unvac-
cinated children resulting from
nonmedical exemptions, based on
religious or philosophical grounds;
SIGNAL TRANSDUCTION on Cic, indicating that Cic degradation may be in 2011–2012, roughly 80% of all exemptions
a consequence of an earlier signal. — LBR were nonmedical. Blank et al. have gathered
Minute Regulation Proc. Natl. Acad. Sci. U.S.A. 110, 10330 (2013). information from public health officials, health
A more thorough understanding of cellular departments, the Centers for Disease Control
signaling mechanisms depends on resolv- CANCER and Prevention, the National Conference of State
ing the spatial and temporal characteristics of Legislatures, and state legislature databases.
Tumor Epigenetics
such signals in vivo. During the third hour of Policies were characterized as easy, medium,
development of the fruit fly embryo, formation That human tumors display both genetic muta- or difficult, according to the level of effort they
of the membranes that partition the syncytium tions and epigenetic alterations—for example, would pose for parents requesting exemptions.
into individual cells occurs with such regularity in DNA methylation—has been known for many The lower the barrier, the more nonmedical
that it can be used as a marker of developmental years; with the completion of cancer genome exemptions were observed, with a twofold dif-
age of the embryo with an accuracy of about sequencing projects, possible causal links ference between the easiest and most difficult
between the two have come into procedures. For 2011–2012, at least 21 bills
sharper focus. The discovery of re- were introduced at the state level to change
current tumor-associated mutations the exemption procedures, and exemptions
in genes that encode chromatin- would have been made easier if bills in 10
modifying enzymes or DNA meth- states had passed. As of February 2013, three
yltransferases represents a clear bills have been introduced in two states to
link between tumor genotype and tighten exemptions, and five bills have been
“epigenotype.” Emerging evidence introduced in four states to loosen them. The
suggests that a link can be subtle, authors advocate social and policy efforts to
as illustrated by two studies describ- promote parental education and to stem the
ing consistent epigenetic alterations spread of vaccine-preventable diseases. — BJ
1 min. Lim et al. analyzed the kinetics of the in tumors with mutations in the gene encoding Health Affairs 32, 1282 (2013).
effects of the mitogen-activated protein kinase the metabolic enzyme succinate dehydrogenase
ERK on transcription in dozens of stained and (SDH). Killian et al. find that gastrointestinal OCEAN SCIENCE
fixed embryos oriented on microfluidics chips. stromal tumors harboring SDH mutations are
Heating the Deep Ocean
Activated ERK was known to reduce the nuclear characterized by dramatic and widespread DNA
localization of the transcriptional repressor hypermethylation, whereas Letouzé et al. report As greenhouse gas concentrations in the
protein Cic, thus increasing Cic’s degradation. that SDH-mutant paragangliomas display DNA atmosphere increase, the resulting heat warms
However, these more refined measurements hypermethylation that is associated with the not only the atmosphere but also the oceans.
showed that ERK has a faster regulatory effect silencing of genes involved in neuroendocrine However, temperature increases at the sea

Published by AAAS
EDITORS’CHOICE
surface and in the upper ~700 m of the oceans and sulfur in 2.4-billion-year-old drill cores
stalled in the 2000s. Recent modeling and from South Africa. Synchrotron-based x-ray
observational studies have suggested that the absorption spectroscopy revealed that abundant
deep ocean has taken up the heat, particularly Mn oxides were hosted in carbonate deposits,
at depths between 700 and 2000 m. However, which were probably formed via oxidation of
regular, spatially homogenous ocean tempera- soluble Mn(II). Several lines of geochemical
ture data are only available since 2003 from evidence based on redox-sensitive proxies,
the Argo ocean observing system, complicating however, preclude oxygen as the primary oxi-
comparison with earlier data sets. Balmaseda dant acting on Mn. Moreover, because the rise
et al. used a new observation-based reanalysis of oxygen from oxygenic cyanobacteria would
of the ocean to investigate how the heat content not occur for another 200 million years after
has changed over the period from 1958 to the sediments were deposited, a primitive (or
2009. They provide evidence for an overall transitional) form of photosynthesis may have
warming trend, punctuated by cooling episodes been responsible for forming the Mn oxides. In Join the
that can be attributed to volcanic eruptions. this scenario, the water-oxidizing complex in
Conversation!

The massive 1997–1998 El Niño event also the early photosynthesizing enzyme machinery
had a noticeable cooling impact. In the past utilized Mn(II) as an electron donor. — NW
Twitter is a great way to
decade, the deep ocean has continued to warm, Proc. Natl. Acad. Sci. U.S.A. 110, 11238 (2013).
whereas temperatures in the upper 300 m have connect with AAAS
stabilized; the ocean as a whole has continued M AT E R I A L S S C I E N C E
to warm at an increasing rate. The qualitative members and staff about
warming patterns persisted when Argo data were
I Sense the Force
the issues that matter
removed. The authors attribute the changing Although composite materials can show enhanced
heat distribution to changes in surface winds, properties, understanding the way these materials to you most. Be a
particularly an intensification of the trade winds deform and fail is not straightforward. Besides
in subtropical gyres. — JFU the individual properties of the matrix and the part of the discussion
Geophys. Res. Lett. 40, 1754 (2013). filler, one has to worry about the distribution while staying up-to-date
of the filler particles, the strength of the interfa-
GEOCHEMISTRY cial bonding between the filler and matrix, and on the latest news
the way that the combined material will deform
Oxidation Before Oxygen and information about
and debond. Raja et al. show that luminescent
CREDIT: J. E. JOHNSON ET AL., PROC. NATL. ACAD. SCI. U.S.A. 110, 28 (24 JUNE 2013) © 2013 NATIONAL ACADEMY OF SCIENCES
The emergence of oxygen-producing photosyn- semiconductor nanocrystal tetrapods, composed your personal
thesis had a profound effect on Earth’s surface of zinc-blend CdSe cores with epitaxially grown
environment. It eventually oxidized the oceans wurtzie CdS arms, can be used as stress sensors. member benefits.
and atmosphere, paving the way for aerobic life. The tetrapods were electrospun into a matrix
Follow us
of poly L-lactic acid at
weight fractions rang- @AAASmember
ing from 3.6 to 40%.
During extension of and join the
the composite fibers,
conversation with
both the elastic and
plastic deformation re- #AAAS
gions could be tracked
through shifts in the
fluorescence of the
nanocrystals. The au-
thors note that because
the particles tend to
clump and because
there is incomplete
stress transfer from
the polymer to the
nanocrystals, they do
not deform plastically
Determining the timing of photosynthesis and when the polymer does, thus giving greater re-
the subsequent transformation of biogeochemi- versibility to the stress measurements, as seen
cal cycles relies on analyzing clues in ancient in comparing the fluorescence observations with MemberCentral.aaas.org
buried sediments. Johnson et al. analyzed the traditional tensile measurements. — MSL
mineralogy and isotopic signatures of carbon Nano Lett. 10.1021/nl401999t (2013).
www.sciencemag.org SCIENCE VOL 341 19 JULY 2013

Published by AAAS
EDITOR-IN-CHIEF Marcia McNutt EXECUTIVE PUBLISHER Alan I. Leshner
www.sciencemag.org EXECUTIVE EDITOR NEWS EDITOR PUBLISHER Beth Rosner
Monica M. Bradford Tim Appenzeller FULFILLMENT SYSTEMS AND OPERATIONS (membership@aaas.org); CUSTOMER
MANAGING EDITOR, RESEARCH JOURNALS Katrina L. Kelner SERVICE SUPERVISOR Pat Butler; SPECIALISTS LaToya Casteel, Michelle Ofordire,
DEPUTY EDITORS Barbara R. Jasny, Andrew M. Sugden, Valda J. Vinson April Marshall; MANAGER, DATA ENTRY Mickie Napoleoni; DATA ENTRY SPECIALISTS
EDITORIAL SENIOR EDITORS/COMMENTARY Lisa D. Chong, Brad Wible; SENIOR JJ Regan, Jaimee Wise, Fiona Giblin
EDITORS Gilbert J. Chin, Pamela J. Hines, Paula A. Kiberstis (Boston), Marc BUSINESS OPERATIONS AND ADMINISTRATION DIRECTOR Deborah Rivera-Wienhold;
S. Lavine (Toronto), Beverly A. Purnell, L. Bryan Ray, Guy Riddihough, BUSINESS SYSTEMS AND FINANCIAL ANALYSIS DIRECTOR Randy Yi; MANAGER OF FULFILLMENT
1200 New York Avenue, NW
H. Jesse Smith, Phillip D. Szuromi (Tennessee), Jake S. Yeston, Laura M. Zahn SYSTEMS Neal Hawkins; SYSTEMS ANALYST Nicole Mehmedovich;
Washington, DC 20005 MANAGER , BUSINESS ANALYSIS Eric Knott; MANAGER , BUSINESS OPERATIONS
(San Diego); ASSOCIATE EDITORS Melissa R. McCartney (Education Projects),
Editorial: 202-326-6550, FAX 202-289-7562 Jessica Tierney; BUSINESS ANALYSTS Cory Lipman, Cooper Tilton,
Kristen L. Mueller, Jelena Stajic, Sacha Vignieri (Oregon), Nicholas S.
News: 202-326-6591, FAX 202-371-9227 Wigginton (Michigan); BOOK REVIEW EDITOR Sherman J. Suter; ASSOCIATE LETTERS Celeste Troxler; FINANCIAL ANALYST Jeremy Clay; RIGHTS AND PERMISSIONS:
Bateman House, 82-88 Hills Road EDITOR Jennifer Sills; EDITORIAL MANAGER Cara Tate; SENIOR COPY EDITORS Jeffrey E. ASSISTANT DIRECTOR Emilie David; ASSOCIATE Elizabeth Sandler;
Cambridge, UK CB2 1LQ Cook, Cynthia Howe, Harry Jach, Lauren Kmec, Barbara P. Ordway, Trista MARKETING DIRECTOR Ian King; MARKETING MANAGERS Alison Chandler,
+44 (0) 1223 326500, FAX +44 (0) 1223 326501 Wagoner; COPY EDITOR Chris Filiatreau; SENIOR EDITORIAL COORDINATORS Carolyn Julianne Wielga, Justin Sawyers; MARKETING ASSOCIATES Mary Ellen
SUBSCRIPTION SERVICES For change of address, missing issues, new Kyle, Beverly Shields; EDITORIAL COORDINATORS Ramatoulaye Diop, Joi S. Crowley, Elizabeth Sattler, Rebecca Riffkin; SENIOR MARKETING EXECUTIVE
Granger, Lisa Johnson, Anita Wynn; PUBLICATIONS ASSISTANTS, Aneera Dobbins, Jennifer Reeves; DIRECTOR , SITE LICENSING Tom Ryan; DIRECTOR ,
orders and renewals, and payment questions: 866-434-AAAS (2227)
Jeffrey Hearn, Dona Mathieu, Le-Toya Mayne Flood, Shannon McMahon, CORPORATE RELATIONS Eileen Bernadette Moran; SENIOR PUBLISHER
or 202-326-6417, FAX 202-842-1065. Mailing addresses: AAAS, P.O. RELATIONS SPECIALIST Kiki Forsythe; PUBLISHER RELATIONS MANAGER
Scott Miller, Jerry Richardson, Teresa R. Sakon, Brian White; EDITORIAL ASSISTANT
Box 96178, Washington, DC 20090-6178 or AAAS Member Services, Catherine Holland; PUBLISHER RELATIONS, EASTERN REGION Keith Layson;
Patricia M. Moore; EXECUTIVE EDITORIAL ASSISTANT Yolanda O'Bannon (San
1200 New York Avenue, NW, Washington, DC 20005 Francisco); EXECUTIVE ASSISTANT Alison Crawford; ADMINISTRATIVE SUPPORT PUBLISHER RELATIONS, WESTERN REGION Ryan Rexroth; CUSTOMER RELATIONS
INSTITUTIONAL SITE LICENSES please call 202-326-6755 for any questions Maryrose Madrid MANAGER Iquo Edim; CUSTOMER RELATIONS ANALYSTS Simon Chong, Lana
or information EDITORIAL DIRECTOR, WEB AND NEW MEDIA Stewart Wills; SENIOR WEB EDITOR Sarah Guz; ASSOCIATE DIRECTOR, MARKETING Christina Schlecht; MARKETING ASSOCIATES
REPRINTS: Author Inquiries 800-635-7181 Crespi; WEB DEVELOPMENT MANAGER Martyn Green; WEB DEVELOPER Corinna Paulina Curto, Mitchell Edmund; ELECTRONIC MEDIA: DIRECTOR Lizabeth Harman;
Commercial Inquiries 803-359-4578 Cohn; INTERNS Kristine Hamilton, Linda Poon ASSISTANT MANAGER Lisa Stanford; PRODUCTION SPECIALISTS Antoinette Hodal,
NEWS DEPUTY NEWS EDITORS Robert Coontz, Elizabeth Culotta, David Grimm Nichele Johnston, Lori Murphy, Kimberly Oster; WEB AND NEW MEDIA:
PERMISSIONS 202-326-6765, permissions@aaas.org
(Online), Eliot Marshall, Jeffrey Mervis, Leslie Roberts, Richard Stone, SENIOR PROJECT MANAGER Trista Smith, PROJECT LEADER Luke Johnson COMPUTER
MEMBER BENEFITS AAAS Travels: Betchart Expeditions 800-252-4910; John Travis; CONTRIBUTING EDITOR Polly Shulman; NEWS WRITERS Yudhijit SPECIALISTS Walter Jones, Kai Zhang, WEB DEVELOPER Chris Coleman; PROGRAM
Apple Store www.store.apple.com/us/go/eppstore/aaas; NASA Federal, Bhattacharjee, Adrian Cho, Jennifer Couzin-Frankel, Carolyn Gramling, DIRECTOR, AAAS MEMBER CENTRAL Peggy Mihelich
1-888-NASA-FCU (1-888-627-2328) or www.nasafcu. Jocelyn Kaiser, Richard A. Kerr, David Malakoff, Elizabeth Pennisi, Robert DIRECTOR, GLOBAL COLLABORATION, CUSTOM PUBLICATIONS, ADVERTISING Bill Moran
com; Cold Spring Harbor Laboratory Press Publications F. Service (Pacific NW), Erik Stokstad, Emily Underwood; WEB DEVELOPER EDITOR, CUSTOM PUBLISHING Sean Sanders: 202-326-6430
www.cshlpress.com/affiliates/aaas.htm; GEICO Auto Insurance Daniel Berger; SOCIAL MEDIA STRATEGIST Meghna Sachdev; INTERNS Kelly ASSISTANT EDITOR, CUSTOM PUBLISHING Tianna Hicklin 202-326-6463
www.geico.com/landingpage/go51.htm?logo=17624; Hertz Servick, Nisha Giridharan, Senah Yeboah-Sampong, Rachel Walther; SPONSORED CONTENT SPECIALIST Candice Nulsen 202-256-1528
800-654-2200 CDP#343457; Office Depot https://bsd. CONTRIBUTING CORRESPONDENTS John Bohannon, Jon Cohen (San Diego,
ASSOCIATE DIRECTOR, COLLABORATION, CUSTOM PUBLICATIONS/CHINA/TAIWAN/KOREA/
officedepot.com/portalLogin.do; Seabury & Smith Life Insurance 800- CA), Ann Gibbons, Sam Kean, Eli Kintisch, Andrew Lawler, Mitch Leslie, SINGAPORE Ruolei Wu +86-1367-101-5294
424-9883; Subaru VIP Program 202-326-6417; VIP Moving Services Charles C. Mann, Virginia Morell; COPY EDITOR Kara Estelle; ADMINISTRATIVE
www.vipmayflower.com/domestic/index.html; Other Benefits: AAAS PRODUCT (science_advertising@aaas.org); MIDWEST Rick Bongiovanni:
SUPPORT Scherraine Mack; BUREAUS San Diego, CA: 760-942-3252, FAX
Member Services 202-326-6417 or www.aaasmember.org. 330-405-7080, FAX 330-405-7081; EAST COAST/E. CANADA Laurie Faraday:
760-942-4979; Pacific Northwest: 503-963-1940 508-747-9395, FAX 617-507-8189; WEST COAST/W. CANADA Lynne Stickrod:
science_editors@aaas.org (for general editorial queries) PRODUCTION DIRECTOR Wendy K. Shank; ASSISTANT MANAGER Rebecca Doshi; 415-931-9782, FAX 415-520-6940; UK EUROPE/ASIA Roger Goncalves: TEL/
science_letters@aaas.org (for queries about letters) SENIOR SPECIALISTS Steve Forrester, Anthony Rosen; SPECIALIST Jennifer Dahl;
FAX +41 43 243 1358; JAPAN, Makiko Hara: +81 (0) 3 6802 4616, FAX
science_reviews@aaas.org (for returning manuscript reviews) PREFLIGHT DIRECTOR David M. Tompkins; MANAGER Marcus Spiegler; SPECIALISTS +81 (0) 3 6802 4615; ads@sciencemag.jp; CHINA/TAIWAN Ruolei Wu: +86
science_bookrevs@aaas.org (for book review queries) Jason Hillman, Tara Kelly 1367 1015 294 rwu@aaas.org
ART DIRECTOR Yael Fitzpatrick; ASSOCIATE ART DIRECTOR Laura Creveling; WORLDWIDE ASSOCIATE DIRECTOR OF SCIENCE CAREERS Tracy Holmes: +44 (0)
Published by the American Association for the Advancement of Sci- SENIOR ILLUSTRATORS Chris Bickel, Katharine Sutliff; SENIOR ART ASSOCIATES
ence (AAAS), Science serves its readers as a forum for the presentation 1223 326525, FAX +44 (0) 1223 326532
Holly Bishop, Preston Huey; ART ASSOCIATES Kay Engman, Garvin Grullón,
and discussion of important issues related to the advancement of sci- CLASSIFIED (advertise@sciencecareers.org); U.S./CANADA/SOUTH AMERICA
Chrystal Smith; PHOTO EDITOR Leslie Blizard
ence, including the presentation of minority or conflicting points of view, Tina Burks: 202-326-6577; U.S. CORPORATE Candice Nulsen 202-256-1528;
SCIENCE INTERNATIONAL SALES ADMINISTRATOR Marci Gallun; EUROPE / ROW SALES Axel Gesatzki;
rather than by publishing only material on which a consensus has been
reached. Accordingly, all articles published in Science—including EUROPE (science@science-int.co.uk) EDITORIAL: INTERNATIONAL MANAGING SALES ASSISTANT Kelly Grace; JAPAN Yuri Kobayashi +81 (0)90-
EDITOR Andrew M. Sugden; SENIOR EDITOR/COMMENTARY Julia Fahrenkamp-
editorials, news and comment, and book reviews—are signed and reflect 9110-1719; careerads@sciencemag.jp; C H I N A / TA I WA N
the individual views of the authors and not official points of view adopted by Uppenbrink; SENIOR EDITORS Caroline Ash, Stella M. Hurtley, Ian S. Osborne, Ruolei Wu: +86 1367 1015 294 rwu@aaas.org; ADVERTISING
AAAS or the institutions with which the authors are affiliated. Peter Stern; ASSOCIATE EDITOR Maria Cruz; CONTRIBUTING EDITOR Helen SUPPORT MANAGER Karen Foote: 202-326-6740; ADVERTISING PRODUCTION
Pickersgill; EDITORIAL SUPPORT Rachel Roberts, Alice Whaley; ADMINISTRATIVE OPERATIONS MANAGER Deborah Tompkins; SENIOR PRODUCTION SPECIALIST/GRAPHIC
AAAS was founded in 1848 and incorporated in 1874. Its mission is to SUPPORT Alex Brown, Janet Clements; NEWS: DEPUTY NEWS EDITOR, U.K. Dan-
DESIGNER Amy Hardcastle; PRODUCTION SPECIALIST Yuse Lajiminmuhip; SENIOR
advance science, engineering, and innovation throughout the world iel Clery; CONTRIBUTING EDITOR, EUROPE Martin Enserink; CONTRIBUTING TRAFFIC ASSOCIATE Christine Hall; SALES COORDINATOR Shirley Young; MARKETING
for the benefit of all people. The goals of the association are to: enhance CORRESPONDENTS Michael Balter (Paris), Kai Kupferschmidt (Berlin),
MANAGER Allison Pritchard; MARKETING ASSOCIATE Aimee Aponte
communication among scientists, engineers, and the public; promote and Gretchen Vogel (Berlin)
defend the integrity of science and its use; strengthen support for the AAAS BOARD OF DIRECTORS RETIRING PRESIDENT, CHAIR William H. Press;
ASIA Japan Office: Asca Corporation, Tomoko Furusawa, Rustic Bldg. 7F, 77 PRESIDENT Phillip A. Sharp; PRESIDENT-ELECT Gerald R. Fink; TREASURER David
science and technology enterprise; provide a voice for science on societal
issues; promote the responsible use of science in public policy; strengthen Tenjin-cho, Shinjuku-ku, Tokyo 162-0808, Japan; +81 3 6802 4616, FAX Evans Shaw; CHIEF EXECUTIVE OFFICER Alan I. Leshner; BOARD Bonnie L. Bassler,
and diversify the science and technology workforce; foster education in +81 3 6802 4615, inquiry@sciencemag.jp; CONTRIBUTING EDITOR, ASIA Mara May R. Berenbaum, Claire M. Fraser, Elizabeth Loftus, Stephen L. Mayo,
science and technology for everyone; increase public engagement with Hvistendahl [China: mhvisten@aaas.org]; CONTRIBUTING CORRESPONDENTS Raymond Orbach, Sue V. Rosser, Inder M. Verma
science and technology; and advance international cooperation in science. Dennis Normile [Japan: dnormile@gol.com]; Christina Larson [China:
christina.larson@gmail.com]; Hao Xin [China: cindyhao@gmail.com];
INFORMATION FOR AUTHORS Pallava Bagla [South Asia: pallava.bagla@gmail.com]
See pages 716 and 717 of the 8 February 2013 issue or LATIN AMERICA CONTRIBUTING CORRESPONDENT Lizzie Wade [Mexico City: lizziewade@
access www.sciencemag.org/about/authors outlook.com]
SENIOR EDITORIAL BOARD Robert Desimone, MIT Daniel Kammen, Univ. of California, Berkeley Jim Roberts, Fred Hutchinson Cancer Research Ctr.
Claude Desplan, New York Univ. Joel Kingsolver, Univ. of North Carolina at Chapel Hill Barbara A. Romanowicz, Univ. of California, Berkeley
A. Paul Alivisatos, Lawrence Berkeley Nat'l. Laboratory Ap Dijksterhuis, Radboud Univ. of Nijmegen Robert Kingston, Harvard Medical School Jens Rostrup-Nielsen, Haldor Topsoe
Ernst Fehr, Univ. of Zurich Dennis Discher, Univ. of Pennsylvania Roberto Kolter, Harvard Medical School Mike Ryan, Univ. of Texas, Austin
Michael S. Turner, University of Chicago Gerald W. Dorn II, Washington Univ. School of Medicine Alberto R. Kornblihtt, Univ. of Buenos Aires Shimon Sakaguchi, Kyoto Univ.
Jennifer A. Doudna, Univ. of California, Berkeley Leonid Kruglyak, Princeton Univ. Miquel Salmeron, Lawrence Berkeley National Lab
BOARD OF REVIEWING EDITORS Julian Downward, Cancer Research UK Thomas Langer, Univ. of Cologne Jürgen Sandkühler, Medical Univ. of Vienna
Adriano Aguzzi, Univ. Hospital Zürich Bruce Dunn, Univ. of California, Los Angeles Mitchell A. Lazar, Univ. of Pennsylvania Alexander Schier, Harvard Univ.
Takuzo Aida, Univ. of Tokyo Christopher Dye, WHO David Lazer, Harvard Univ. Randy Seeley, Univ. of Cincinnati
Leslie Aiello, Wenner-Gren Foundation Todd Ehlers, University of Tuebingen Virginia Lee, Univ. of Pennsylvania Vladimir Shalaev, Purdue Univ.
Sonia Altizer, Univ. of Georgia David Ehrhardt, Carnegie Inst. of Washington Stanley Lemon, Univ. of North Carolina at Chapel Hill Joseph Silk, Institut d'Astrophysique de Paris
Virginia Armbrust, Univ. of Washington Tim Elston, Univ. of North Carolina at Chapel Hill Ottoline Leyser, Cambridge Univ. Denis Simon, Arizona State Univ.
Sebastian Amigorena, Institut Curie Gerhard Ertl, Fritz-Haber-Institut, Berlin Marcia C. Linn, Univ. of California, Berkeley Alison Smith, John Innes Centre
Angelika Amon, MIT Barry Everitt, Univ. of Cambridge Jianguo Liu, Michigan State Univ. Davor Solter, Inst. of Medical Biology, Singapore
Kathryn Anderson, Memorial Sloan-Kettering Cancer Center Paul G. Falkowski, Rutgers Univ. Luis Liz-Marzan, CIC biomaGUNE Peter Sorger, Harvard Medical School
Siv G. E. Andersson, Uppsala Univ. Ernst Fehr, Univ. of Zurich Jonathan Losos, Harvard Univ. John Speakman, Univ. of Aberdeen
Peter Andolfatto, Princeton Univ. Tom Fenchel, Univ. of Copenhagen Ke Lu, Chinese Acad. of Sciences Allan C. Spradling, Carnegie Institution of Washington
Meinrat O. Andreae, Max Planck Inst., Mainz Michael Feuer, The George Washington Univ. Christian Lüscher, Univ. of Geneva Jonathan Sprent, Garvan Inst. of Medical Research
Paola Arlotta, Harvard Univ. Alain Fischer, INSERM Laura Machesky, CRUK Beatson Inst. for Cancer Research Eric Steig, Univ. of Washington
Johan Auwerx, EPFL Susan Fiske, Princeton Univ. Anne Magurran, Univ. of St Andrews Paula Stephan, Georgia State Univ. and National Bureau
David Awschalom, Univ. of California Santa Barbara Anne C. Ferguson-Smith, Univ. of Cambridge Oscar Marin, CSIC & Univ. Miguel Hernández of Economic Research
Ben Barres, Stanford Medical School Peter Fratzl, Max Planck Inst. Charles Marshall, Univ. of California, Berkeley Elsbeth Stern, ETH Zürich
Jordi Bascompte, Estación Biológica de Doñana, CSIC Elaine Fuchs, Rockefeller Univ. Chris Marshall, Inst. of Cancer Research V. S. Subrahmanian, Univ. of Maryland
Facundo Batista, London Research Inst. Wulfram Gerstner, EPFL Lausanne Martin M. Matzuk, Baylor College of Medicine Ira Tabas, Columbia Univ.
Ray H. Baughman, Univ. of Texas, Dallas Daniel Geschwind, UCLA C. Robertson McClung, Dartmouth College Yoshiko Takahashi, Kyoto University
David Baum, Univ. of Wisconsin Andrew Gewirth, Univ. of Illinois Graham Medley, Univ. of Warwick Sarah Teichmann, Cambridge Univ.
Mark Bear, Massachusetts Inst. of Technology Karl-Heinz Glassmeier, TU Braunschweig Yasushi Miyashita, Univ. of Tokyo John Thomas, North Carolina State Univ.
Yasmine Belkaid, NIAID, NIH Elizabeth Grove, Univ. of Chicago Richard Morris, Univ. of Edinburgh Christopher Tyler-Smith, The Wellcome Trust Sanger Inst.
Philip Benfey, Duke Univ. Kip Guy, St. Jude's Children's Research Hospital Edvard Moser, Norwegian Univ. of Science and Technology Herbert Virgin, Washington Univ.
Stephen J. Benkovic, Penn State Univ. Taekjip Ha, Univ. of Illinois at Urbana-Champaign Sean Munro, MRC Lab. of Molecular Biology Bert Vogelstein, Johns Hopkins Univ.
Christophe Bernard, Aix-Marseille Univ. Christian Haass, Ludwig Maximilians Univ. Thomas Murray, The Hastings Center Cynthia Volkert, Univ. of Göttingen
Gregory C. Beroza, Stanford Univ. Steven Hahn, Fred Hutchinson Cancer Research Center Naoto Nagaosa, Univ. of Tokyo Bruce D. Walker, Harvard Medical School
Gabriele Bergers, Univ. of California, San Francisco Gregory J. Hannon, Cold Spring Harbor Lab. James Nelson, Stanford Univ. School of Med. Douglas Wallace, Dalhousie Univ.
Peer Bork, EMBL Martin Heimann, Max Planck Inst., Jena Daniel Neumark, Univ. of California, Berkeley Ian Walmsley, Univ. of Oxford
Bernard Bourdon, Ecole Normale Superieure de Lyon Yka Helariutta, Univ. of Finland Stuart Newman, New York Medical College David A. Wardle, Swedish Univ. of Agric Sciences
Chris Bowler, Ecole Normale Supérieure Isaac Held, NOAA Timothy W. Nilsen, Case Western Reserve Univ. David Waxman, Fudan Univ.
Ian Boyd, Univ. of St. Andrews James A. Hendler, Rensselaer Polytechnic Inst. Pär Nordlund, Karolinska Inst. Jonathan Weissman, Univ. of California, San Francisco
Christian Büchel, Universitätsklinikum Hamburg-Eppendorf Janet G. Hering, Swiss Fed. Inst. of Aquatic Helga Nowotny, European Research Advisory Board Kathy Willis, Oxford Univ.
Joseph A. Burns, Cornell Univ. Science & Technology Luke O'Neill, Trinity College, Dublin Ian A. Wilson, The Scripps Res. Inst.
William P. Butz, Population Reference Bureau Ray Hilborn, Univ. of Washington Stuart Newman, New York Medical College Timothy D. Wilson, Univ. of Virginia
Gyorgy Buzsaki, New York Univ., School of Medicine Michael E. Himmel, National Renewable Energy Lab. N. Phuan Ong, Princeton Univ. Rosemary Wyse, Johns Hopkins Univ.
Mats Carlsson, Univ. of Oslo Kai-Uwe Hinrichs, Univ. of Bremen Joe Orenstein, Univ. of California, Berkeley & Lawrence Jan Zaanen, Leiden Univ.
Mildred Cho, Stanford Univ. Kei Hirose, Tokyo Inst. of Technology Berkeley National Lab Kenneth Zaret, Univ. of Penn. School of Medicine
David Clapham, Children’s Hospital, Boston David Hodell, Univ. of Cambridge Harry Orr, Univ. of Minnesota Jonathan Zehr, Univ. of California, Santa Cruz
David Clary, Univ. of Oxford David Holden, Imperial College Andrew Oswald, Univ. of Warwick Maria Zuber, MIT
Jonathan D. Cohen, Princeton Univ. Lora Hooper, UT Southwestern Medical Ctr at Dallas Steve Palumbi, Stanford Univ.
Robert Cook-Deegan, Duke Univ. Jeffrey A. Hubbell, EPFL Lausanne Jane Parker, Max-Planck Inst. of Plant Breeding Research BOOK REVIEW BOARD
James Collins, Boston Univ. Thomas Hudson, Ontario Inst. for Cancer Research Donald R. Paul, Univ. of Texas at Austin John Aldrich, Duke Univ.
Alan Cowman, Walter & Eliza Hall Inst. Raymond Huey, Univ. of Washington P. David Pearson, Univ. of California, Berkeley David Bloom, Harvard Univ.
Robert H. Crabtree, Yale Univ. Steven Jacobsen, Univ. of California, Los Angeles John H. J. Petrini, Memorial Sloan-Kettering Cancer Center Angela Creager, Princeton Univ.
Wolfgang Cramer, Mediterranean Inst. of Biodiversity and Ecology Kai Johnsson, EPFL Lausanne Simon Phillpot, Univ. of Florida Richard Shweder, Univ. of Chicago
Jeff L. Dangl, Univ. of North Carolina Peter Jonas, Inst. of Science & Technolgy (IST) Australia Joshua Plotkin, Univ. of Pennsylvania Ed Wasserman, DuPont
Tom Daniel, Univ. of Washington Matt Kaeberlein, Univ. of Washington Philippe Poulin, CNRS Lewis Wolpert, Univ. College London
Frans de Waal, Emory Univ. William Kaelin Jr., Dana-Farber Cancer Inst. Colin Renfrew, Univ. of Cambridge
Stanislas Dehaene, Collège de France Daniel Kahne, Harvard Univ. Trevor Robbins, Univ. of Cambridge

Published by AAAS
NEWS OF THE WEEK
AROUND THE WORLD
3
2 1
4
5
Double the money. Geoghegan-Quinn announces
new plans for Joint Technology Initiatives.
2014 and 2020, up from €3.12 billion in

2007 to 2013, research commissioner Máire
Geoghegan-Quinn announced on 10 July.
Industry says that it will put about €10 bil-
lion on the table for the next 7 years, up from
€4.66 billion in the current period.

Rome 1 joint FAO-OIE expert committee that they JTIs bring together academia, research
can meet the new safety criteria. “We want organizations, and competing businesses
Rinderpest Research Poised to ensure [rinderpest research] is done in the in research areas considered complex or
For a Comeback safest possible environment,” Lubroth says. risky, such as aeronautics or pharmaceu-
In 2011, rinderpest, a devastating cattle ill- ticals. The initiatives have been slammed
ness, became the second infectious disease Bethesda, Maryland 2 for having complex rules and procedures,
after smallpox that humans have eradicated but the funding boost is a testament to the
from nature. In June 2012, the U.N. Food
U.S. Senate Panel Gives NIH participants’ overall satisfaction with this
and Agricultural Organization (FAO), in $31 Billion in 2014 collaborative model. Future incarnations
Rome, and the Paris-based World Organi- A U.S. Senate spending panel last week will be more ambitious, cut down on red
sation for Animal Health (OIE) banned approved a bill giving the National Institutes tape, and align disparate rules, a commis-
research using live rinderpest virus, fearing of Health (NIH) $31 billion in 2014, a 7% sion spokesman says.
increase over this year’s budget, which was The commission will release a detailed
Former scourge. Cows dead from rinderpest in depressed by the across-the-board federal legal proposal early next week. The plans
South Africa in 1896. budget cuts known as sequestration. The must then be signed off on by the European
NIH funding level approved by the Senate Parliament and member states before the
Appropriations Committee is $147 million programs are rolled out next year.
below the president’s request for 2014, but http://scim.ag/EUJointTech
$2 billion more than the $28.9 billion that
NIH has to spend this year. Rome 4
CREDITS (TOP TO BOTTOM): EUROPEAN UNION/SHIMERA/JENNIFER JACQUEMART; WIKIMEDIA COMMONS

Jennifer Zeitzer, legislative relations
director for the Federation of American
Scientists, Stem Cell Provider
Societies for Experimental Biology, said Meet to Discuss Trial
that her group is “thrilled” with the figure. A panel of top science administrators and
“It’s far better than the current situation and stem cell scientists met on 12 July with
an accident could cause a resurgence. Now, it’s a move in the right direction,” she says. Davide Vannoni, president of the Stamina
research on the virus is set to resume fol- The bill now awaits a vote by the entire Foundation, which provides controversial
lowing a 10 July decision by FAO and OIE Senate. The House Appropriations Commit- stem cell treatments, to hash out details for
to lift the moratorium. The two organiza- tee, which has a much smaller pot of money a government-sponsored clinical trial of the
tions used the hiatus to develop safety proto- to fund NIH and other agencies, has not yet therapy. Many stem cell scientists remain
cols and risk-versus-benefit criteria. taken up its version of the bill. opposed to the study, saying there’s little to
Juan Lubroth, FAO’s chief veterinar- suggest that the therapy—which is based on
ian, says that there is interest in sequencing Brussels 3 cultured mesenchymal stem cells but whose
some of the hundreds of isolates that were details remain unpublished—will work.
collected over decades to understand how
E.C. Beefs Up Research In May, the Italian Parliament ordered
the virus changed over time. They also want Partnerships With Industry an 18-month clinical trial of the therapy and
to see if vaccines developed for peste des The European Commission wants to double allocated €3 million to carry it out. But prior
petits ruminants, a closely related disease, public money available for Joint Technology to the 12 July meeting, Vannoni had not yet
can protect cattle from rinderpest. If so, Initiatives (JTIs), five research programs that provided the committee appointed to coordi-
stores of rinderpest virus used to replen- provide public support to private research nate the study with protocols describing how
ish vaccines could be destroyed, eliminat- and innovation, with industry matching his treatment works, because he says that he’s
ing another potential source of accidental E.U. funding. Together, the JTIs will receive still working on their standardization. The
release. Researchers will have to convince a €6.44 billion from the E.U. budget between start of the trial has been delayed by at least

Published by AAAS
NEWS
a month. Vannoni has his own conditions, Also last week, four Japanese electric exploded in size in the first moments after
including that there can be no changes to his utilities applied for approval to restart the big bang. The work laid the foundations
protocol and the cells need to be produced at 10 nuclear power plants that have been idled for the idea of an inflationary universe, a
one facility under the supervision of a Stam- since shortly after the Fukushima accident. bedrock of modern cosmology.
ina biologist. Barring these, Vannoni says NRA will review the applications based on Mukhanov’s contribution dates back
that he may move the research abroad. new safety standards unveiled last month. to 1981, when he was a researcher at the
http://scim.ag/Staminatrial The process is expected to take at least Moscow Physical Technical Institute. Along
6 months. The utilities will also need the with a fellow scientist named G. V. Chibisov,
Okuma, Japan 5 approval of local governments. Mukhanov showed that quantum fluctuations
in the dense and tiny newborn universe could
Contaminated Water Likely NEWSMAKERS have served as the seeds of modern day cos-
Polluting Ocean From Reactors mic structure. As the universe swelled, these
Japanese and international groups moni-
Origin-of-Everything Studies Nab fluctuations gave rise to the formation of gal-
toring Pacific Ocean waters near the Fuku- Gruber Prize axies and galaxy clusters.
shima Daiichi Nuclear Power Plant have Two physicists who made

continued to find radioactive cesium at lev- fundamental contributions FINDINGS
els higher than expected, given the effects to understanding how the
of dilution and radioactive decay. They universe evolved into its Cancer Knife Sniffs Out Tumor Cells
have claimed that this suggests contami- current structure have won When cancer surgeons can’t see the edges of
nated cooling water is leaking from the the 2013 Gruber Cosmol- Mukhanov a tumor, they often send some tissue to the
crippled reactors, though plant operator ogy Prize. Viatcheslav pathology lab, which can take up to
Tokyo Electric Power has long denied Mukhanov, a professor at 30 minutes per analysis. Now, there’s an
this possibility. At a press briefing on Ludwig Maximilians alternative: The “intelligent knife,” or
10 July, Shunichi Tanaka, chair of Japan’s University in Munich, iKnife. When tissue is seared by a standard
new Nuclear Regulation Authority (NRA), Germany, and Alexei electrosurgical scalpel, the iKnife sucks the
sided with the experts, saying that the evi- Starobinsky, a researcher resulting smoke into a modified mass spec-
dence suggests that radiological contamina- at the Landau Institute Starobinsky trometry machine. The instrument aims
tion of the ocean probably “has been con- for Theoretical Physics in to instantly detect whether cells are cancer-
tinuing for the last 2 years.” Tanaka added Moscow, will share the $500,000 award. ous or healthy by analyzing this smoke for
that it was important to take measures to Starobinsky’s calculations in the late certain lipids, or fats, and comparing
minimize the leaks. 1970s showed that the universe could have this profile to molecular signatures >>
Meet ‘Big-Nose Horned-Face’

CREDITS (TOP TO BOTTOM): LMU MUNICH/CHRISTOPH OLESINSKI; THE GRUBER FOUNDATION; RAUL MARTIN
There’s a new contender for wackiest-looking

dinosaur: Nasutoceratops, which means, in
Latin, “big-nose horned-face.” The 5-meter-
long dino had a giant schnoz and sharp,
curved horns measuring nearly a meter long.
Nasutoceratops, a distant relative of the
famed Triceratops, was recently discovered
by paleontologists digging in Utah’s Grand
Staircase-Escalante National Monument.
As the team reports online this week in the
Proceedings of the Royal Society B, the creature
lived about 76 million years ago in a swampy,
subtropical region that geologists call Lar-
amidia, formed when an inland sea filled the
center of North America and divided it in half.
The dino’s nose wasn’t used for smelling; the
scent organs of horned dinosaurs were further
back in their heads, next to the brain. So why
such a big honker? The researchers don’t know,
but suggest that the elaborate horns might
have been used to attract females and fend off
competing males.

Published by AAAS
NEWS OF THE WEEK
BY THE NUMBERS if using the iKnife helps patients develop That didn’t stop microbiologist Tanja
fewer recurring tumors and live longer. Woyke from the Department of Energy Joint
23.82 months Average time http://scim.ag/cancerknife Genome Institute in Walnut Creek, Califor-
from publication to retraction nia. She and her colleagues used an approach
for articles published after 2002, Shedding Light on Microbial that allowed them to sequence the DNA in
according to a PLOS ONE analysis. individual cells (rather than requiring many
‘Dark Matter’ copies of the cells). They characterized 200
Papers published between 1973 and When researchers began sequencing DNA new microbes from 29 largely uncharted
2002 took nearly 50 months before from environmental samples more than a phyla, then used the genomes to determine
retraction, perhaps due in part to decade ago, they discovered a vast, diverse the microbes’ phylogenetic relationships and
higher barriers to pulling papers. world of microbes. But this microbial dark to assess how each lives, naming 18 phyla
CREDITS (CLOCKWISE FROM LEFT): J. BALOG ET AL., SCIENCE TRANSLATIONAL MEDICINE 5, 194 (17 JULY); XIANGZHEN LI AND WEN-TSO LIU/U. ILLINOIS, URBANA-CHAMPAIGN; © KOCA LITTLE COMPANY/CULTURA/CORBIS
accordingly, the team reported
64% Fraction of 455 women Mysteries revealed. online this week in Nature.
who experienced infertility after Now microbes from Novelties emerged: Some
environmental samples archaea possess genes previ-
surviving childhood cancer and

can be characterized ously thought to exist only in
were eventually able to become from their genomes. bacteria; some bacteria have
pregnant, according to a report in archaeal genes. One group
of microbes has even altered
The Lancet Oncology.
the three-base coding system
8000 years ago First known for proteins.
In the future, single-cell
use of manure as fertilizer among sequencing could fill in more
Europe’s first farmers—thousands of branches of the microbial
years earlier than thought—according tree of life. “It’s an inflec-
to a study in the Proceedings of the tion point in environmental
microbiology and micro-
National Academy of Sciences. biology in general,” says
matter has defied further description—and Norman Pace, a microbiologist at the Uni-
definitive placement on the tree of life— versity of Colorado, Boulder.
>>FINDINGS because so few kinds will grow in the lab. http://scim.ag/microbedark
for various cancer types. When tested during
81 surgeries, the iKnife matched pathology
results for cancerous and normal tissues for
every single patient, the researchers report Random Sample
Cool by the Numbers

When it comes to a hot, crowded
room, hell really is other
people. But once they exit, air-
conditioning the empty room is
an energy drain. The solution: a
ventilation system that adjusts
to the number of people in a
room. Such a system could save
three times as much energy as
simply turning off the heat and
air conditioning to empty rooms, computer models show. Installing people-counting sensors
could save 18% of the energy used by a typical large office building in the United States,
report researchers at the Department of Energy’s (DOE’s) Pacific Northwest National Labora-
tory (PNNL) in Richland, Washington. By comparison, turning off heating, ventilation, and air
conditioning to empty rooms saves 6%, says PNNL engineer Guopeng Liu.
this week in Science Translational Medicine. But building owners are likely to install such a system only if it’s cheap and doesn’t require
The iKnife provides “real-time infor- rewiring the building, says Aravind Dasu, a computer scientist with the University of Southern
mation,” says its inventor Zoltán Takáts, a California in Washington, D.C.: “If it’s all wireless it’s a game-changer.” James Freihaut, chief
Hungarian chemist who collaborated with scientist at DOE’s Energy Efficient Buildings hub in Philadelphia, Pennsylvania, notes that local
surgeon Jeremy Nicholson at Imperial Col- building codes may also present a challenge, as many of them require constant ventilation
lege London to develop the tool. Their next within an occupied commercial structure.
step is to conduct clinical trials to find out

Published by AAAS
NEWS
a month. Vannoni has his own conditions, Also last week, four Japanese electric exploded in size in the first moments after
including that there can be no changes to his utilities applied for approval to restart the big bang. The work laid the foundations
protocol and the cells need to be produced at 10 nuclear power plants that have been idled for the idea of an inflationary universe, a
one facility under the supervision of a Stam- since shortly after the Fukushima accident. bedrock of modern cosmology.
ina biologist. Barring these, Vannoni says NRA will review the applications based on Mukhanov’s contribution dates back
that he may move the research abroad. new safety standards unveiled last month. to 1981, when he was a researcher at the
http://scim.ag/Staminatrial The process is expected to take at least Moscow Physical Technical Institute. Along
6 months. The utilities will also need the with a fellow scientist named G. V. Chibisov,
Okuma, Japan 5 approval of local governments. Mukhanov showed that quantum fluctuations
in the dense and tiny newborn universe could
Contaminated Water Likely NEWSMAKERS have served as the seeds of modern day cos-
Polluting Ocean From Reactors mic structure. As the universe swelled, these
Japanese and international groups moni-
Origin-of-Everything Studies Nab fluctuations gave rise to the formation of gal-
toring Pacific Ocean waters near the Fuku- Gruber Prize axies and galaxy clusters.
shima Daiichi Nuclear Power Plant have Two physicists who made

continued to find radioactive cesium at lev- fundamental contributions FINDINGS
els higher than expected, given the effects to understanding how the
of dilution and radioactive decay. They universe evolved into its Cancer Knife Sniffs Out Tumor Cells
have claimed that this suggests contami- current structure have won When cancer surgeons can’t see the edges of
nated cooling water is leaking from the the 2013 Gruber Cosmol- Mukhanov a tumor, they often send some tissue to the
crippled reactors, though plant operator ogy Prize. Viatcheslav pathology lab, which can take up to
Tokyo Electric Power has long denied Mukhanov, a professor at 30 minutes per analysis. Now, there’s an
this possibility. At a press briefing on Ludwig Maximilians alternative: The “intelligent knife,” or
10 July, Shunichi Tanaka, chair of Japan’s University in Munich, iKnife. When tissue is seared by a standard
new Nuclear Regulation Authority (NRA), Germany, and Alexei electrosurgical scalpel, the iKnife sucks the
sided with the experts, saying that the evi- Starobinsky, a researcher resulting smoke into a modified mass spec-
dence suggests that radiological contamina- at the Landau Institute Starobinsky trometry machine. The instrument aims
tion of the ocean probably “has been con- for Theoretical Physics in to instantly detect whether cells are cancer-
tinuing for the last 2 years.” Tanaka added Moscow, will share the $500,000 award. ous or healthy by analyzing this smoke for
that it was important to take measures to Starobinsky’s calculations in the late certain lipids, or fats, and comparing
minimize the leaks. 1970s showed that the universe could have this profile to molecular signatures >>
Meet ‘Big-Nose Horned-Face’

CREDITS (TOP TO BOTTOM): LMU MUNICH/CHRISTOPH OLESINSKI; THE GRUBER FOUNDATION; RAUL MARTIN
There’s a new contender for wackiest-looking

dinosaur: Nasutoceratops, which means, in
Latin, “big-nose horned-face.” The 5-meter-
long dino had a giant schnoz and sharp,
curved horns measuring nearly a meter long.
Nasutoceratops, a distant relative of the
famed Triceratops, was recently discovered
by paleontologists digging in Utah’s Grand
Staircase-Escalante National Monument.
As the team reports online this week in the
Proceedings of the Royal Society B, the creature
lived about 76 million years ago in a swampy,
subtropical region that geologists call Lar-
amidia, formed when an inland sea filled the
center of North America and divided it in half.
The dino’s nose wasn’t used for smelling; the
scent organs of horned dinosaurs were further
back in their heads, next to the brain. So why
such a big honker? The researchers don’t know,
but suggest that the elaborate horns might
have been used to attract females and fend off
competing males.

Published by AAAS
NEWS OF THE WEEK
BY THE NUMBERS if using the iKnife helps patients develop That didn’t stop microbiologist Tanja
fewer recurring tumors and live longer. Woyke from the Department of Energy Joint
23.82 months Average time http://scim.ag/cancerknife Genome Institute in Walnut Creek, Califor-
from publication to retraction nia. She and her colleagues used an approach
for articles published after 2002, Shedding Light on Microbial that allowed them to sequence the DNA in
according to a PLOS ONE analysis. individual cells (rather than requiring many
‘Dark Matter’ copies of the cells). They characterized 200
Papers published between 1973 and When researchers began sequencing DNA new microbes from 29 largely uncharted
2002 took nearly 50 months before from environmental samples more than a phyla, then used the genomes to determine
retraction, perhaps due in part to decade ago, they discovered a vast, diverse the microbes’ phylogenetic relationships and
higher barriers to pulling papers. world of microbes. But this microbial dark to assess how each lives, naming 18 phyla
CREDITS (CLOCKWISE FROM LEFT): J. BALOG ET AL., SCIENCE TRANSLATIONAL MEDICINE 5, 194 (17 JULY); XIANGZHEN LI AND WEN-TSO LIU/U. ILLINOIS, URBANA-CHAMPAIGN; © KOCA LITTLE COMPANY/CULTURA/CORBIS
accordingly, the team reported
64% Fraction of 455 women Mysteries revealed. online this week in Nature.
who experienced infertility after Now microbes from Novelties emerged: Some
environmental samples archaea possess genes previ-
surviving childhood cancer and

can be characterized ously thought to exist only in
were eventually able to become from their genomes. bacteria; some bacteria have
pregnant, according to a report in archaeal genes. One group
of microbes has even altered
The Lancet Oncology.
the three-base coding system
8000 years ago First known for proteins.
In the future, single-cell
use of manure as fertilizer among sequencing could fill in more
Europe’s first farmers—thousands of branches of the microbial
years earlier than thought—according tree of life. “It’s an inflec-
to a study in the Proceedings of the tion point in environmental
microbiology and micro-
National Academy of Sciences. biology in general,” says
matter has defied further description—and Norman Pace, a microbiologist at the Uni-
definitive placement on the tree of life— versity of Colorado, Boulder.
>>FINDINGS because so few kinds will grow in the lab. http://scim.ag/microbedark
for various cancer types. When tested during
81 surgeries, the iKnife matched pathology
results for cancerous and normal tissues for
every single patient, the researchers report Random Sample
Cool by the Numbers

When it comes to a hot, crowded
room, hell really is other
people. But once they exit, air-
conditioning the empty room is
an energy drain. The solution: a
ventilation system that adjusts
to the number of people in a
room. Such a system could save
three times as much energy as
simply turning off the heat and
air conditioning to empty rooms, computer models show. Installing people-counting sensors
could save 18% of the energy used by a typical large office building in the United States,
report researchers at the Department of Energy’s (DOE’s) Pacific Northwest National Labora-
tory (PNNL) in Richland, Washington. By comparison, turning off heating, ventilation, and air
conditioning to empty rooms saves 6%, says PNNL engineer Guopeng Liu.
this week in Science Translational Medicine. But building owners are likely to install such a system only if it’s cheap and doesn’t require
The iKnife provides “real-time infor- rewiring the building, says Aravind Dasu, a computer scientist with the University of Southern
mation,” says its inventor Zoltán Takáts, a California in Washington, D.C.: “If it’s all wireless it’s a game-changer.” James Freihaut, chief
Hungarian chemist who collaborated with scientist at DOE’s Energy Efficient Buildings hub in Philadelphia, Pennsylvania, notes that local
surgeon Jeremy Nicholson at Imperial Col- building codes may also present a challenge, as many of them require constant ventilation
lege London to develop the tool. Their next within an occupied commercial structure.
step is to conduct clinical trials to find out

Published by AAAS
NEWS & ANALYSIS NEWS & ANALYSIS
Chagrined. KPUM President Toshikazu Yoshikawa
(center) and other officials apologize after report-
ing fraud.
group. Once anomalous data were corrected,

the claim that valsartan reduces the incidence
of cardiovascular events such as angina or
stroke by about half “is not supported,” the
report says. Eight papers authored by Matsub-
ara have been retracted.
In an e-mail to Science, Matsubara
writes that he “was not involved in and gave
no instructions” for the event data analy-
sis, which he said was the responsibility of
a Novartis employee. Novartis headquarters

in Basel declined to comment on KPUM’s
J A PA N findings, stating in an e-mail to Science that
“we are unfamiliar with how the university
Tampered Data Cast conducted its review.” But the company has
confirmed that employees of its Japanese sub-
sidiary participated in the Kyoto Heart Study
Shadow on Drug Trial and several similar studies at other Japanese

institutions without noting their affiliation on
resulting papers. In a statement posted to its
TOKYO—In a scandal reverberating across The 4-year study followed 3000 patients given website on 12 July, Novartis blamed the dis-
Japan’s biomedical research landscape, a uni- valsartan or alternative medications. A main closure failures on a lack of guidelines and a
versity in Kyoto last week acknowledged data outcome, reported on 31 August 2009 in the misunderstanding of the appropriate level of
manipulation in a university-run clinical trial European Heart Journal, was that valsartan, involvement in such trials. “Preventive and
for a blockbuster hypertension drug, valsar- which reduces blood pressure by blocking the corrective measures have been implemented,”
tan. Japanese media have turned the episode receptor for the hormone angiotensin, “pre- the statement said.
into a cause célèbre; the Yomiuri Shimbun, vented more cardiovascular events” in high- The debacle raises troubling questions
one of Japan’s biggest newspapers, labeled it a risk patients than did drugs that lower blood about research oversight. KPUM officials said
“serious betrayal” for patients. Repercussions pressure through another mechanism. that they could not probe more deeply into
could extend beyond valsartan, marketed In late 2011, bloggers started raising ques- who manipulated data because they cannot
under the trade name Diovan by the Swiss tions about alleged image manipulation in compel cooperation from Novartis employ-
pharmaceutical giant Novartis. Matsubara’s papers. Then, in an April 2012 ees. Universities in Japan have no authority
“This incident is causing a loss of confi- letter to The Lancet, Yui expressed concerns to question people outside their institution,
dence in Japan’s research internationally and about Kyoto Heart Study’s statistics and con- says Tetsuya Tanimoto, a physician at the Uni-
making Japanese patients skeptical [about clusions, writing that the effectiveness in pre- versity of Tokyo who studies pharmaceutical
treatments],” says Yoshiki Yui, a medical venting angina was not seen regulatory issues. Japan needs
doctor at Kyoto University. Health minis- in other trials of valsartanlike “Data were something like the U.S. Office
ter Norihisa Tamura has vowed to appoint a drugs or in clinical practice. of Research Integrity, he says.
committee that would propose measures to Responding to requests from manipulated.” Japan’s medical research
prevent a recurrence. journals, KPUM started its —TOSHIKAZU YOSHIKAWA establishment could come in
Sold worldwide, valsartan was approved own investigation. Matsubara KPUM PRESIDENT for broad scrutiny. KPUM
in Japan in 2000 and has become the coun- resigned from KPUM in Febru- confirmed in an e-mail to
try’s best-selling drug, pulling in roughly ary, after the European Heart Journal, citing Science that since 2008, Matsubara received
$1 billion last year, according to analysts. “[c]ritical problems … with some of the data,” about $1.4 million for his research from
A clinical trial claiming that valsartan also retracted the heart study paper. Novartis. Such grants to clinical researchers
reduces angina and stroke risk helped boost “Data were manipulated,” KPUM Presi- are not unusual, but they highlight inadequate
its popularity in Japan. That claim has fallen dent Toshikazu Yoshikawa bluntly stated at public support for clinical research in Japan,
apart, in the process raising concerns about an 11 July press conference. According to Tanimoto says. The revelations could also
cozy ties between researchers and drug com- the university’s investigative report, which erode public support for emerging plans to
CREDIT: KYODO/NEWSCOM
panies and about impediments to investigat- Science has obtained, in the Kyoto Heart shake up medical research by establishing an
ing research misconduct in Japan. Study there were 34 discrepancies between outcome-focused version of the U.S. National
At the heart of the scandal are data from the clinical medical records and the data set Institutes of Health. In the wake of the scan-
the Kyoto Heart Study, launched in 2003 by used for analysis; these overstated adverse dal, the Japanese public may question whether
Hiroaki Matsubara, a cardiologist at Kyoto cardiovascular events in the nonvalsartan the money will be well-spent and the results
Prefectural University of Medicine (KPUM). group and missed such events in the valsartan trustworthy. –DENNIS NORMILE

Published by AAAS
Play, not war. !Kung men like these playing a
traditional game in Namibia may attack other
individuals, but they rarely gather to make war on
other societies.
and blurred by intermarriage. “In my view the

default for nomadic foragers is nonwarring.”
The new study is “a very valuable contri-
bution to the study of mobile foragers,” says
anthropologist Richard B. Lee of the Uni-
versity of Toronto in Canada, who has stud-
ied hunter-gatherers in the field for 40 years.
Endicott agrees that the paper offers a “valu-
able corrective” to an “erroneous” view of
mobile foragers as warlike.
However, critics find fault with Fry and
Söderberg’s selection of societies and their
restricted samples and ask how well these
A N T H R O P O LO G Y societies represent the past. The descriptive
data of the SCCS can’t provide the key metric
Latest Skirmish Over Ancestral on how many people died in war compared
with other causes of death, Hill says. He
and economist Samuel Bowles of the Santa
Violence Strikes Blow for Peace Fe Institute also argue that Fry should have
included sedentary hunter-gatherers, such as
Are hunter-gatherer societies warlike? That nographies,” says anthropologist Kim Hill warring fishing societies in British Colum-
question has sparked a war of its own among of Arizona State University, Tempe, whose bia. And limiting the data to early ethnogra-
scientists. Some anthropologists see these own detailed field studies found warfare in phies ignores a host of additional data—and
traditional societies as largely peaceful, rely- three South American groups. He and oth- deaths—says Harvard University anthropolo-
ing on trade networks with outsiders for sur- ers cite recent work suggesting that ancient gist Richard Wrangham. “There is lots of evi-
vival, while other researchers cite frequent war was frequent enough to have influenced dence of war in some of the societies that they
deadly clashes among neighboring groups. our evolution, for example by encourag- list as having no killings,” he says. For exam-
The question has implications beyond today’s ing altruistic cooperation among warriors ple, the Andamanese Islanders are reported as
dwindling foraging peoples, because our (Science, 5 June 2009, p. 1293). having few killing events, but other research-
ancestors lived as hunter-gatherers for most Fry and Söderberg drew on data from the ers have documented additional killings
of prehistory. If war is a common feature of Standard Cross-Cultural Sample (SCCS), there, often between groups.
the foraging way of life, then perhaps it was a a respected ethnographic database set up Part of the dispute comes down to the
driving force in human evolution. by other anthropologists in the 1980s. The definition of war. For Fry, war implies the
On page 270 of this issue, two research- pair pulled out all mobile societies in which killing of any outgroup member because
ers fire a salvo in support of peace. Using hunting, gathering, and fishing brought in they are in the outgroup; having a personal
existing ethnographic data, Douglas Fry and 95% of the people’s food—21 societies in motive makes a homicide. His critics argue
Patrik Söderberg of Åbo Akademi University all—then studied the oldest ethnographic that personal motives may indeed spark
in Vasa, Finland, conclude that people liv- descriptions that were highly rated by the “wars” in small-scale societies. “Feuds are
ing in mobile foraging societies, such as the SCCS scholars. “To be purists, we took only warfare, revenge raiding is warfare,” Hill
!Kung of southern Africa and the Semang of the oldest high-quality sources for each cul- says. Researchers on both sides caution
the Malay Peninsula, today rarely engage in ture,” says Fry, an anthropologist; he argues against using living people as direct models
CREDIT: © KIM WALKER/ROBERT HARDING WORLD IMAGERY/CORBIS
what most modern people call “war.” Rather, that the oldest studies best reflect a society’s of our ancestors.
two-thirds of killings in such societies occur traditional ways. Anthropologist Polly Wiessner of the Uni-
among people of the same group, and most The pair then scrutinized every instance versity of Utah in Salt Lake City, who has
lethal events stem from personal disputes. of lethal aggression recounted in the ethnog- studied both the relatively peaceful !Kung
“These findings imply that warfare was prob- raphies. They found that more than half of and the relatively violent Enga people of
ably not very common before the advent of societies did not practice what they would Papua New Guinea (Science, 28 September
agriculture, when most if not all humans call war on outside groups. Overall, 55% 2012, pp. 1593 and 1651), agrees that few
lived as nomadic foragers,” says cultural of cases had a single killer and single vic- mobile foragers often wage war today. But
anthropologist Kirk Endicott of Dartmouth tim. Also, most killings were driven by per- given that some foragers do fight fiercely,
College, who was not part of the study. sonal motives, such as fights over women and she hopes the battle lines among scientists
But those on the other side of the debate revenge, and are better classified as homi- will shift to asking what promotes and what
say that the paper lacks the numerical data cides or sometimes feuds than as war, Fry inhibits warfare. “We should be asking how
to evaluate how common war and homicide says. Such societies are too small to wage coalitionary aggression, which does appear
actually are. “This is essentially a list of anec- wars, he notes, and groups rarely fight each in our phylogeny, was harnessed among our
dotes—there’s no real method in these eth- other because group membership is flexible successful ancestors.” –ELIZABETH CULOTTA

Published by AAAS
NEWS&ANALYSIS
S PA I N
Spain’s Research Council Approaches Bankruptcy

BARCELONA, SPAIN—Spain’s flagship people rather than laying them off, Beltrán competitive grants and industry contracts,
research agency is in deep trouble. The explains. But a letter circulating among CSIC without researchers’ knowledge. Group
Spanish National Research Council (CSIC) researchers to gauge their interest in legal leaders have traditionally used these stashes
has run massive deficits for several years. action also blames “bad management” by suc- as emergency funds, to hire extra pairs of
If the government—which has made dras- cessive CSIC governing bodies. (The CSIC hands or bridge the gap between two con-
tic cuts in research funding the past few presidency declined interview requests.) tracts. Lora-Tamayo has said that CSIC
years—doesn’t pick up the tab, it may go Lora-Tamayo issued a directive on 2 July intends to return the money “once the
into bankruptcy before the year is over. At a ordering CSIC directors and managers to agency has been successfully stabilized.”
press conference last week, CSIC President slow down spending.
Emilio Lora-Tamayo warned of an impend- Institutes were asked
ing “cataclysm” and said that without a cash to prioritize salaries for
injection, CSIC “wouldn’t be able to go on.” staff members on short-
On 28 June, the Spanish government term contracts and to
agreed to give CSIC an extra €25 million, as

f inish research proj-
part of an €104 million package for research; ects if not completing
but CSIC says that it needs €75 million them by the end of the
more. Meanwhile, Lora-Tamayo has issued year would mean hav-
a range of measures to stem the agency’s ing to return funds. But
financial hemorrhage, which scientists say is for some institutes, the
bringing research projects to a halt. CSIC’s available money doesn’t
researchers are angry—particularly about come close to even pay-
the revelation that the agency has dipped ing for short-term con-
into the savings accumulated by research tracts, says Luis Sanz-
groups. Some 140 researchers are consid- Menéndez, director of
ering taking legal action to get those funds the CSIC Institute of
back. “The situation is chaotic,” says plant Public Goods and Poli-
biologist José Pío Beltrán, who heads the cies in Madrid. As a Mad in Madrid. Spanish researchers demonstrated against cuts in the science
office coordinating CSIC institutes in the result, researchers are budget on 14 June. Their banner reads: “There is no future without science.”
Valencia region. CSIC researchers and trade barred from spending
unions held meetings this week to discuss money and projects are grinding to a halt. “We A petition launched on 10 July that
the situation. can’t pay … laboratory expenses, bills for drew 69,000 signatures in just 5 days urges
With some 6000 researchers in areas from equipment that we have bought. … We can’t Spain’s state secretary for research Carmen
biology and materials science to the humani- buy plane tickets” to go to meetings, says Vela to “do whatever needs to be done so
ties, CSIC is the largest research organiza- plant ecologist Fernando Valladares of the the CSIC doesn’t die.” Vela said last
tion in Spain; its 125 institutes—more than CSIC National Museum of Natural Sciences week that she would give CSIC an extra
PHOTO CREDIT: © SUSANA VERA/REUTERS/CORBIS; GRAPH SOURCE: CSIC INTERNAL PRESENTATION
50 of them joint ventures with universities in Madrid. €50 million by September and find other
and other research bodies—account for 20% Lora-Tamayo has also revealed that for solutions, if necessary. But Spain’s fiscal sit-
of the country’s research output. The gov- years, CSIC has been using savings that uation is so dire that many researchers worry
ernment provides roughly 60% of CSIC’s research groups accumulated from past the money won’t come. The prospect of
budget; individual researchers raise the rest. losing the savings permanently is “spine-
CSIC’s revenues peaked at €879 mil- An Agency in Crisis chilling,” says Valladares, who says he’s
lion in 2008, when the agency still had a “still a little bit in a state of shock.”
1000
€26 million surplus. But since then, fund- CSIC’s woes aren’t the only headache
ing streams have declined much faster than for Spanish science. The current govern-
800
expenditures, leading to annual deficits that ment has slashed the national science bud-
have depleted reserves (see graphic). With 600
get, delayed competitive funding rounds,
a projected €101 million deficit in 2013,
Millions of €
and failed to honor past commitments, caus-

CSIC would go deeply into the red. 400 ing “absolute bewilderment,” says a 15 July
Part of the problem is that, in his first man- Spending statement by the Confederation of Spanish
Revenues
date between 2004 and 2008, Spanish presi- 200 Deficit Scientific Societies, the Conference of
dent José Luis Rodríguez Zapatero “asked *projected Rectors of Spanish Universities, trade
us to be really driving strongly the Spanish 0 unions, and other associations. The entire
[research] system, so we started very ambi- -100 scientific enterprise is affected, the letter
tious programs to incorporate people,” Bel- -200
2008 2009 2010 2011 2012 2013*
says: “This strangulation of R&D is under-
trán says. When Spain’s economy tanked mining a public research system that took
and funds declined during Zapatero’s sec- Seeing red. Since 2009, big deficits have depleted more than 30 years to build.”
ond term, CSIC continued supporting those CSIC’s reserves. –ELISABETH PAIN

Published by AAAS
M I C R O B I O LO G Y
Ever-Bigger Viruses Shake Tree of Life

“It’s like finding a sasquatch,” says Elodie a cell and coax their host to replicate them, as mimivirus, plus a few more, and seems
Ghedin, a virologist at the University of because they can’t make their own proteins. to represent an intermediate step between a
Pittsburgh in Pennsylvania. That’s one of the A decade ago, the discovery in an amoeba free-living ancestor and mimivirus
amazed reactions to the discovery, reported of a virus that rivals the size of a small Now, scans of the water and sediment
on page 281, of two new viruses with by bacterium prompted a rethinking of how samples that Claverie, Abergel, and other
far the largest genomes ever seen in a virus, viruses originated and what they could do. lab members gather whenever they travel
including one that’s bigger than the genomes Didier Raoult, a microbiologist at the Uni- have yielded the pandoraviruses. (The
of some parasitic eukaryotes. The virolo- versity of the Mediterranean in Marseille; researchers inoculate amoeba with the sam-
gists in France who unearthed the massive Claverie; and their colleagues sequenced ples to see if any viruses replicate and burst
viruses—the biggest one is 1 micron long, the genome of mimivirus, for “microbe out). The one with the smaller genome came
a hundred times the size of many viruses— mimicking virus.” Its 1.18 million bases from mud in an Australian pond, while the
suggest that their finds challenge the long- contained more than 900 putative genes, new king of the viral genomes was in coastal
standing view that viruses don’t qualify as life. some closely resembling genes in non- sediments collected off Chile. “The fact that

Genome Sizes
s
iru
s
av
ru
or
ivi
M e
ag
nd
im
Ph
Pa
Viruses
Eukaryotes Parasitic Free-living
Bacteria Parasitic Free-living
Archaea Parasitic Free-living
0 2 4 6 8 10 12 14 16
Number of bases (in millions)
“It is clear that the paradigm that viruses viruses that are involved in protein production two of them were found almost simultane-
have small genomes and are relatively sim- (Science 28 March 2003, p. 2033; ously from very distant locations either indi-
ple in comparison to cellular life has been 19 November 2004, p. 1344). cates we were incredibly lucky or that they
overturned,” says Curtis Suttle, a virologist Mimivirus could have acquired those are not rare,” Claverie says. “They are prob-
at the University of British Columbia in Van- genes from its cellular hosts, but the mimi- ably everywhere.”
couver. The genome of one of the viruses virus genes are so different from those of the Because of their size, the pandoraviruses
is 1.91 million DNA bases long, while the host amoeba and other cells that Raoult and appeared bacterialike at first. But using light
other runs 2.47 million bases. That dwarfs Claverie instead proposed that mimivirus and electron microscopy, the French group
some bacterial genomes and edges into the descended from a free-living cell that gradu- followed the newfound entities through a
eukaryotic realm (see chart). ally lost most of its other genes as it became a replication cycle, which proved viruslike.
Jean-Michel Claverie and Chantal parasite. That mimivirus precursor, they sug- Instead of dividing in two like a typical bac-
Abergel from CNRS, the French national gested, represented a previously unknown terium or cell, they generated hundreds or
research agency, at Aix-Marseille University branch of life, one predating the emergence more new viral particles, Claverie’s team
in France, and their colleagues have dubbed of the three major branches, or domains, of reports. Both pandoraviruses lack genes for
the new viruses pandoraviruses because of life—bacteria, archaea, and eukaryotes. energy production and can’t actually produce
their amphora shape and the surprises they This theory remains controversial, but it a protein on their own, fulfilling the defini-
may portend. They have strikingly different has motivated Claverie to keep hunting for tion of virus. “The authors seem to have gone
genes and physical appearances from other viral giants. “According to this scenario, the proverbial extra mile to show that these
viruses. The finding “expands our view of looking for even bigger viruses with bigger agents are actually viruses rather than some
the virus world,” Ghedin says. genomes was a way to go back in time, to sort of unusual bacteria,” says Eugene Koo-
CREDIT: K. ENGMAN/SCIENCE
After their late 19th century discovery, take a closer, earlier look at this postulated nin, a computational evolutionary biologist
viruses were quickly demoted to inert parti- ancestor,” he says. He and other researchers at the National Center for Biotechnology
cles, too simple to belong to the realm of the have since come across several other giant Information in Bethesda, Maryland.
living. Considered little more than a protein viruses, including Megavirus chilensis, at But unlike other viruses, the pandora-
package of genetic material with no meta- 1.25 million bases the previous viral genome viruses lack the gene for the capsid pro-
bolic capabilities, viruses must get inside record-holder. It has many of the same genes tein that typically forms a capsule around a

Published by AAAS
NEWS&ANALYSIS
virus’s genes and are missing some ing Uncomplete

key genes found in all other giant Classif ication—
viruses and their relatives, including and the French
ones for replication. “They seem to word for stuff.
be a new family unto themselves,” It’s too soon
Ghedin says. to redraw the tree
Indeed, most of the pandoravirus of life, several
genes don’t look like any in known researchers cau-
databases. “The lack of similar- tion. But some
ity might be an indication that they revision is already
originated from a totally different warranted, Suttle
primitive cellular lineage than bacte- argues. “What the discovery of pan-
ria, archaea, and eukarya,” Claverie A class of their own. Pandoraviruses have a much bigger genome, doraviruses and mimiviruses makes
says. Add in other giant viruses, he an atypical shape, and different genes from megaviruses (inset), the increasingly clear is that the ‘domains’
says, and “these viruses might indi- next largest viruses known to date. of life is an archaic concept that does
cate that not only a fourth domain nothing other than keep viruses from
existed but also a fifth, a sixth, etc.” Raoult plus archaea, bacteria, and microbial their rightful place at the table where the

goes so far as to suggest lumping all com- eukaryotes—into a new grouping he would story of the evolution of life is told.”
plex microbes—the various giant vir uses call TRUC, an acronym for Things Resist- –ELIZABETH PENNISI
FUNDING
Germany Debates How to Strengthen Universities

BERLIN—Germany’s recent push to boost a Change is coming because the budget- Research Center in Heidelberg, floated a con-
handful of its universities to the world’s top boosting agreements end between 2015 and troversial idea late last year for starting its own
ranks got a yellow light this week. A promi- 2019, and national elections in September granting program that would fund cooperative
nent advisory council recommended taking a will decide who gets to set the policies that projects with universities.
break from the national competition to find, will take their place. Funding bodies and This week’s report from the Wissen-
and fund, Germany’s top schools. Funding organizations of research institutes have in schaftsrat, which includes scientists and poli-
should focus on strengthening a broad base of recent months offered varying views of how ticians, was the most eagerly awaited. An ear-
CREDITS (TOP TO BOTTOM): IGS – CNRS/AMU; IGS CNRS-AMU/CHANTAL ABERGEL; HEIKE ZAPPE/HUMBOLDT-UNIVERSITÄT ZU BERLIN
research—and teaching—at the country’s uni- the system should be changed. But they all lier, rejected draft reportedly said that Ger-
versities, according to a 15 July report from share one thing: “We all agree that the fund- many should strive for up to five world-class
the German Council of Science and Humani- ing for universities has to be increased,” says universities. The final version, in contrast,
ties (Wissenschaftsrat). Peter Gruss, president of the Max Planck takes a more egalitarian approach, saying
A relatively robust economy has allowed Society, which funds institutes independent that universities are the core of the research
the government to increase funding for the of universities. system. It proposes two new funding mecha-
country’s major nonuniversity research orga- On 9 July, the Max Planck Society pro- nisms to support top research at a broad spec-
nizations by 5% per year since 2011. But posed establishing a system of Max Planck trum of schools: establishing Merian profes-
many university-based scientists see little Professors and Max Planck Centers at uni- sorships, named for 17th century naturalist
of the money because the constitution pro- versities neighboring its institutes. The Helm- and illustrator Maria Sibylla Merian, which
hibits the federal government from funding holtz Association, which runs Germany’s large would provide €1 million per year to each of
universities directly. A pair of agreements research centers such as the DESY accelera- 250 leading academics; and setting up roughly
between the federal and state governments tor lab in Hamburg and the German Cancer 40 Liebig Centers (named for chemist Justus
did pump billions of euros into universi- von Liebig), to boost key research
ties through a “Higher Education Pact” that areas. The competition for top schools
funds the growing number of students and could be revisited in 10 to 15 years, the
an “Excellence Initiative” to support univer- report says.
sity research and encourage a few schools to Council Chair Wolfgang Mar-
strive for the world’s top ranks. But not all quardt says that the recommendations
universities have thrived. are affordable if Germany takes seri-
In Germany, education is controlled by ously the goal of spending 3.5% of
the Länder (states). Many state budgets gross domestic product on research
have not been generous to research so uni- and development. But others are more
versity-based researchers have become cautious. “There’s still a huge gap
increasingly dependent on grant-based between these proposals and the reali-
funding, particularly from the German ties faced by politicians,” says Wil-
Research Foundation (DFG). “The basic helm Krull, secretary general of the
funding for the universities is eroding,” says Under pressure. German universities haven’t all benefited Volkswagen Foundation, a private
DFG President Peter Strohschneider. from government funding largess. research funder. –GRETCHEN VOGEL

Published by AAAS
NEWS&ANALYSIS
virus’s genes and are missing some ing Uncomplete

key genes found in all other giant Classif ication—
viruses and their relatives, including and the French
ones for replication. “They seem to word for stuff.
be a new family unto themselves,” It’s too soon
Ghedin says. to redraw the tree
Indeed, most of the pandoravirus of life, several
genes don’t look like any in known researchers cau-
databases. “The lack of similar- tion. But some
ity might be an indication that they revision is already
originated from a totally different warranted, Suttle
primitive cellular lineage than bacte- argues. “What the discovery of pan-
ria, archaea, and eukarya,” Claverie A class of their own. Pandoraviruses have a much bigger genome, doraviruses and mimiviruses makes
says. Add in other giant viruses, he an atypical shape, and different genes from megaviruses (inset), the increasingly clear is that the ‘domains’
says, and “these viruses might indi- next largest viruses known to date. of life is an archaic concept that does
cate that not only a fourth domain nothing other than keep viruses from
existed but also a fifth, a sixth, etc.” Raoult plus archaea, bacteria, and microbial their rightful place at the table where the

goes so far as to suggest lumping all com- eukaryotes—into a new grouping he would story of the evolution of life is told.”
plex microbes—the various giant vir uses call TRUC, an acronym for Things Resist- –ELIZABETH PENNISI
FUNDING
Germany Debates How to Strengthen Universities

BERLIN—Germany’s recent push to boost a Change is coming because the budget- Research Center in Heidelberg, floated a con-
handful of its universities to the world’s top boosting agreements end between 2015 and troversial idea late last year for starting its own
ranks got a yellow light this week. A promi- 2019, and national elections in September granting program that would fund cooperative
nent advisory council recommended taking a will decide who gets to set the policies that projects with universities.
break from the national competition to find, will take their place. Funding bodies and This week’s report from the Wissen-
and fund, Germany’s top schools. Funding organizations of research institutes have in schaftsrat, which includes scientists and poli-
should focus on strengthening a broad base of recent months offered varying views of how ticians, was the most eagerly awaited. An ear-
CREDITS (TOP TO BOTTOM): IGS – CNRS/AMU; IGS CNRS-AMU/CHANTAL ABERGEL; HEIKE ZAPPE/HUMBOLDT-UNIVERSITÄT ZU BERLIN
research—and teaching—at the country’s uni- the system should be changed. But they all lier, rejected draft reportedly said that Ger-
versities, according to a 15 July report from share one thing: “We all agree that the fund- many should strive for up to five world-class
the German Council of Science and Humani- ing for universities has to be increased,” says universities. The final version, in contrast,
ties (Wissenschaftsrat). Peter Gruss, president of the Max Planck takes a more egalitarian approach, saying
A relatively robust economy has allowed Society, which funds institutes independent that universities are the core of the research
the government to increase funding for the of universities. system. It proposes two new funding mecha-
country’s major nonuniversity research orga- On 9 July, the Max Planck Society pro- nisms to support top research at a broad spec-
nizations by 5% per year since 2011. But posed establishing a system of Max Planck trum of schools: establishing Merian profes-
many university-based scientists see little Professors and Max Planck Centers at uni- sorships, named for 17th century naturalist
of the money because the constitution pro- versities neighboring its institutes. The Helm- and illustrator Maria Sibylla Merian, which
hibits the federal government from funding holtz Association, which runs Germany’s large would provide €1 million per year to each of
universities directly. A pair of agreements research centers such as the DESY accelera- 250 leading academics; and setting up roughly
between the federal and state governments tor lab in Hamburg and the German Cancer 40 Liebig Centers (named for chemist Justus
did pump billions of euros into universi- von Liebig), to boost key research
ties through a “Higher Education Pact” that areas. The competition for top schools
funds the growing number of students and could be revisited in 10 to 15 years, the
an “Excellence Initiative” to support univer- report says.
sity research and encourage a few schools to Council Chair Wolfgang Mar-
strive for the world’s top ranks. But not all quardt says that the recommendations
universities have thrived. are affordable if Germany takes seri-
In Germany, education is controlled by ously the goal of spending 3.5% of
the Länder (states). Many state budgets gross domestic product on research
have not been generous to research so uni- and development. But others are more
versity-based researchers have become cautious. “There’s still a huge gap
increasingly dependent on grant-based between these proposals and the reali-
funding, particularly from the German ties faced by politicians,” says Wil-
Research Foundation (DFG). “The basic helm Krull, secretary general of the
funding for the universities is eroding,” says Under pressure. German universities haven’t all benefited Volkswagen Foundation, a private
DFG President Peter Strohschneider. from government funding largess. research funder. –GRETCHEN VOGEL

Published by AAAS
NEWS&ANALYSIS
Curiosity 2.0. A blueprint for a more capable Mars

rover includes instruments to enable the caching of
samples (above).
the rover a reality by issuing a call for pro-

posals to design the science instruments.
The growing consensus around the Mars
rover stands in stark contrast to the continu-
ing acrimony surrounding NASA’s asteroid

S PA C E S C I E N C E capture plan. At a workshop held at the U.S.
National Academies in Washington, D.C., on
the same day that NASA released the rover
Mars Rover Plans Roll While blueprint, engineers and scientists questioned
the fundamentals of the proposed Asteroid
Asteroid Acrimony Continues Retrieval Mission, which would involve iden-
tifying, capturing, and dragging a 500-tonne
NASA officials are getting some contrasting ing the samples in more capable laborato- asteroid into lunar orbit and then sending
messages about two of their boldest projects. ries on Earth, researchers should be able to astronauts to sample it in the early 2020s.
An agency science panel last week gave a address “an enormous number of science Organized by an independent group of
strong endorsement to sending a more capa- questions,” Mustard says, including whether planetary scientists, the meeting included pre-
ble robotic rover to Mars in 2020 that could life ever existed on an earlier, damper Mars sentations from NASA scientists and engi-
cache samples for a later mission to pick up and how it evolved. neers. Many attendees, however, were not
and return to Earth. But researchers and key Although planetary scientists ranked Mars persuaded by the agency’s explanation of the
members of Congress continued to raise sample caching and return as a top priority in project’s technical details or relatively short
serious doubts about the agency’s proposal a 2010 decadal plan, the idea isn’t devoid of timeline, which calls for launching the cap-
to capture a small asteroid and then land an controversy. Some scientists have argued that ture spacecraft by 2018. A fatal flaw, accord-
astronaut on it. the time a rover spends caching rocks could ing to some, is the difficulty of identifying an
The rover proposal, unveiled on 9 July by be better spent collecting real-time data that asteroid of suitable size. Another problem,
a NASA-appointed panel, buoyed the spir- could be beamed back to Earth. And some according to engineers, is that NASA will not
its of Mars scientists in the United States, White House budget officials have been resis- know how to design the equipment needed to
who have been worried about NASA’s com- tant to caching on the grounds that it would capture the asteroid until it has determined
mitment to exploring the Red Planet since it force NASA to commit to flying additional the mass, spin, shape, and other characteris-
pulled out of a European-led effort in 2012 missions to retrieve the samples. tics of the object—which it won’t be able to
(Science, 24 February 2012, p. 900). The Ed Weiler, the former head of NASA do without first getting close to it. The agency
panel, led by planetary scientist John Mus- science, hopes that the panel’s unambigu- needs to take a deep breath and slow down, or
tard of Brown University, says that a 2020 ous support for caching will end the debate. risk technical delays and cost overruns, said
rover should look a lot like Curiosity, the “The cache will last on Mars for many Gentry Lee, chief engineer for solar system
NASA robot now surveying Mars. But it years,” he says. “You can plan a sample exploration at NASA’s Jet Propulsion Labora-
would have a host of new capabilities. In return mission when money becomes avail- tory in Pasadena, California.
addition to Curiosity-style instruments that able.” There was “unanimous enthusiasm” Republicans in the U.S. House of Rep-
take broad pictures of sites, for instance, it for caching among the panelists, Mustard resentatives are also sending that message.
would also carry sensors that can make more adds. “We said—Go big or go home. Are we The House science committee this week is
detailed images of rocks and geological fea- going to do this or not?” expected to approve a NASA authorization
tures. Other instruments would allow fine- The caching capability will make it bill that would stop the agency from pursuing
scale mineralogy studies, especially of rocks harder to estimate the rover’s final cost, how- asteroid capture without further study. The
through which water might have flowed in ever, the panel says. NASA hopes to build Obama administration “has not been able
CREDITS: NASA/JPL-CALTECH (2)
the past, and chemically detect and analyze it for less than $1.5 billion, about $1 billion to adequately justify the rationale or budget
organic carbon (a possible sign of life). less than Curiosity. It’s not yet clear whether for such a mission,” which could cost more
The biggest add-on, however, would be the White House and Congress will back that than $2 billion, panel chair Lamar Smith
the capability to drill into rock sites chosen kind of spending, but NASA officials seem (R–TX) argued in The Hill newspaper on
during the exploration and insert the cores optimistic. Jim Green, head of NASA’s plan- 9 July. The Senate has yet to offer its views
into a container on the rover. Later missions etary sciences division, said that the agency on the mission. –YUDHIJIT BHATTACHARJEE
would haul the container home. By analyz- will soon take another step toward making With reporting by Richard Kerr.

Published by AAAS
NEWSFOCUS

Battle for the followed by others on a sturdy land bridge
that formed 4 million to 3 million years ago.
A radical new model envisions a much
Americas earlier bridge, however. After dating the

uplift of land masses born of volcanism in
Central America and analyzing the geo-
The formation of the Isthmus of Panama allowed the fauna of two continents chemistry and magnetic alignment of rocks
in the region, Camilo Montes of the Uni-
to mingle, transforming biogeography. A radical new hypothesis holds that versity of the Andes in Bogotá; Carlos
the land bridge formed millions of years earlier than scientists thought Jaramillo of the Smithsonian Tropical
Research Institute (STRI) in Balboa, Pan-
Panama, 3 million years ago. A narrow strip Biotic Interchange. The story has a dark end- ama; and colleagues argue that the land
of land bridging the Americas turns into ing: South American fauna compete poorly at bridge was largely formed as early as 15 mil-
a migratory highway. Heading north is an home and away, and many species go extinct. lion years ago, with the last deep-water con-
assemblage of creatures isolated for tens of More intriguing is how the battle for the nection between the oceans closing at most
millions of years: porcupines and armadillos; Americas got started. Decades ago the fos- 5 million years later. By then, “most of the
the elephant-sized ground sloth Megathe- sil record showed, indisputably, modern landscape was already
rium; opossums and other marsupials; and that intercontinental migrations
bizarre life forms like the terror bird, a flight- exploded between 3 million to Online uplifted,” Jaramillo says, although
shallow links between the oceans
less carnivore that reached 3 meters in height 2.5 million years ago—evidence sciencemag.org persisted until full closure 4.2 mil-
Podcast interview
and sprinted like a cheetah. Beating a south- to most scientists that the isth- with author
lion to 3.5 million years ago. The
ward path are more familiar animals, includ- mus had just formed by then. Richard Stone (http:// new model had its coming-out
ing horses, cougars, and saber-toothed cats. Other studies have bolstered scim.ag/pod_6143). party at a feisty symposium at the
CREDIT: FLORILEGIUS/ALAMY
Alfred Russel Wallace, the father of bio- the idea, although it grew more annual meeting of the Geological
geography, recognized the faunal melting nuanced as evidence accumulated of “her- Society of America last November. Their
pot created when the tip of Central America ald” species moving from one continent to scenario poses a number of conundrums,
collided with South America and forged the the other millions of years earlier. That led not least of which is why it took so long—
Isthmus of Panama, and in 1876 he postu- some to suggest that fragmentary marshes several million years—for the Great Ameri-
lated what is now called the Great American allowed some species to cross the gap early, can Biotic Interchange to gain momentum.

Published by AAAS
BIOGEOGRAPHY: PANAMA AND THE AMAZON | NEWSFOCUS
Intercontinental interloper. The elephant-sized in response to shoaling of the Panama isth- Another consequence of isthmus closure
ground sloth Megatherium made it as far north mus,” he wrote. Combined with changes in in the Caribbean was the collapse of upwell-
as Texas. carbon-13 enrichment, the oxygen isotope ing, in which deeper, colder waters rise to
data suggested that modern ocean circu- the surface, bringing nutrients to shallower
“Why did that vast diversity of land mam- lation patterns in the Caribbean and east- waters. Proxies of this collapse include local
mals stay penned up in their respective con- ern Pacific developed about 3 million years extinctions—seafloor organisms “with-
tinents if there was a bridge to cross?” asks ago. “It’s a brilliant paper,” Coates says. He ered on the vine,” Coates says—and a shift
Jeremy Jackson, a STRI ecologist and pale- credits Keigwin for “launching the standard in the Caribbean from fast-growing oysters
ontologist and former director of the Center model of isthmus closure.” to slower-growing species about 3.5 million
for Marine Biodiversity and Conservation at Over the next quarter century, a string years ago.
the Scripps Institution of Oceanography in of paleoceanographic studies have sup- The bottom line, Jackson says, is that
San Diego, California. ported the standard model. For example, until 4 million years or so ago, deep-ocean
The timing of isthmus formation indicators in the Caribbean looked
matters, Jackson says. “The join- like those in the Eastern Pacific.
ing of the continents and severing of “The ocean core data are the killer of
ocean contacts stands as one of the the new hypothesis,” he says. “How
most important and best documented could the two oceans have been so

events in earth system evolution,” well mixed geochemically if there
he says. The new model, if correct, were a land barrier in between?”
would force scientists to reinterpret
dozens of studies—a prospect that Putting the s in isthmus
delights Jaramillo. “We’re unravel- Jaramillo and Montes’s journey into
ing the whole story,” he says. tectonic heresy began with a $1 mil-
The debate is heating up. In lion grant from the Panama Canal
April, geologists Anthony Coates Authority that Jaramillo landed to
of STRI and Robert Stallard of the conduct geological studies along
U.S. Geological Survey in Boul- banks of the Panama Canal slated
der published a report online in the for obliteration to widen the water-
Bulletin of Marine Science (BMS) way. “He started a fantastic project,”
defending the standard model of late Coates says. A companion effort
closure and offering an analogy for in rescue paleontology has netted
the late stages of closure: the Indo- thousands of specimens, including
nesian archipelago, where a narrow 10 species new to science (see side-
but deep strait between the islands of bar, p. 232).
Bali and Lombok—identified in the As part of the projects, Montes
19th century by Wallace—is mostly and Jaramillo used radioisotope,
impassable to animals moving south geochemical, and paleomagnetic
from mainland Asia. And two more analyses to date rocks and discern
papers in press marshal data from relationships between rock forma-
disparate disciplines supporting late tions. They wanted more accurate
closure, while acknowledging that timelines for recovered fossils, and
the prelude to closure could have they hoped that these would lead to a
been messier—and more interesting better understanding of how the link
—than thought. between the continents was forged.
“When we started our study, we
Oceans of evidence didn’t have any problem with 3 mil-
While the first clues to a late clo- Across the divide. Jeremy Jackson (top) says paleocean data are proof lion years” as the date of isthmus for-
sure of the isthmus came from fos- positive that a deep-water barrier between North and South America mation, Montes says.
CREDITS (TOP TO BOTTOM): AARON O’DEA; STRI ARCHIVES
sils, the oceans on either side of the lasted until at least 4 million years ago. Geological data suggest the The gold standard for dating rock
isthmus yielded more compelling barrier vanished much earlier, says Carlos Jaramillo. formations is a technique that relies
evidence. In 1982, Lloyd Keigwin on the radioactive decay of ura-
of the Woods Hole Oceanographic Institu- researchers found that carbonate began to nium-235 and uranium-238. The mineral zir-
tion published a seminal paper in Science pile up in the Caribbean sea floor from about con is chock full of these isotopes and their
comparing oxygen isotopes in the shells of 4.6 million years ago, pointing to more slug- lead and thorium decay products. By mea-
bottom-dwelling foraminifera recovered gish deep-ocean circulation. Then, about suring the proportions of these isotopes in a
from deep-sea drilling in the western Carib- 3.5 million years ago, new kinds of carbon- zircon crystal, researchers can determine the
bean and the eastern Pacific. Keigwin found ate-loving foraminifera and corals appear in time since the crystal and surrounding rock
that oxygen isotopes on either side of Pan- the Caribbean’s fossil record, and molecular hardened and cooled, as well as the exhuma-
ama began to diverge about 4 million years genetics data also back rapid diversification tion, or exposure, of rock from tectonic pro-
ago, suggesting that the Caribbean’s salin- around this time. “It’s a classic Darwinian cesses. The team recovered zircon crystals,
ity increased around that time—“possibly evolutionary experiment,” Coates says. formed during the cooling of magma, from

Published by AAAS
NEWSFOCUS
Salvage Paleontology on the Seaway pump,” MacFadden says, spawning new life forms that spread through
North America.
PARAISO, PANAMA—On an embankment hewn from the jungle dur- The canal’s construction in the early 20th century revealed a fossil
ing the construction of the Panama Canal a century ago, Jason Bourque trove, which scientists secured piecemeal over the years. The expansion
sits hunched over a 19-million-year-old turtle. Under a broiling mid- is opening up whole new exposures. That’s especially exciting because
day sun, Bourque spends a half hour patiently brushing away grit from throughout the isthmus, lush forest has made it difficult for identify rich
a fractured black carapace the size of a dinner plate before covering it fossil sites, says Aaron Wood, a postdoc at the Florida Museum of Natu-
in a plaster bandage for removal. By the time he’s finished, a dozen or ral History. With the canal expansion slated for completion as early as
so container ships, oil tankers, and other vessels steam past along the next year, the window of opportunity for paleontologists is closing fast.
77-kilometer-long seaway. Racing to gather up as many
Bourque, a preparator at the Flor- fossils as they can, MacFadden’s
ida Museum of Natural History, is tak- crew over 4 years has hauled in more
ing advantage of a fossil windfall. As than 3000 specimens, among them
part of a $5.5 billion expansion of the 10 new species. A top fossil hunter on
canal that will move 152 million cubic the team is Aldo Rincon, a Ph.D. stu-
meters of earth, the canal authority dent at the University of Florida, lead

has opened up sections of the seaway author on a report published in the
to salvage paleontology: collecting Journal of Vertebrate Paleontology
fossils in newly exposed rock faces. last March describing a new species of
“It’s a once-in-century opportunity,” anthracothere. Rincon’s favorite spec-
says Bruce MacFadden, curator of ver- imen, he says, is the fragile jawbone
tebrate paleontology at the museum, and teeth of a bat that he unearthed
run by the University of Florida in from the Cucaracha Formation, in the
Gainesville, and director of the Pan- shadow of the Centennial Bridge, last
ama Canal Project. year. “I rolled down the hill, I was so
The project’s geological stud- Race against time. Ongoing expansion of the Panama Canal has excited,” he says.
ies have provided grist for a radical yielded a treasure trove of fossils. Jason Bourque (background) exca- The bat, descended from a South
hypothesis that the Isthmus of Pan- vates a 19-million-year-old turtle. American species, is a rare example of
ama formed a land bridge several a nonswimmer to have traversed the
million years earlier than thought (see main story, p. 230). Its fossil deep-water gap that separated Colombia and Panama about 19 mil-
finds are also revelatory: a menagerie of miniature horses and cam- lion years ago. (Crocodiles, turtles, and snakes crossed much earlier.)
els, gopherlike rodents, peccaries, and caimans, as well as weird crea- A view of life in this ancient cradle of biodiversity is coming into sharper
tures such as the giant bear-dog and the hippolike anthracothere. The focus. The Panama Canal, says Andrés Cárdenas, a postdoc at the
assemblage recovered so far from the prehistoric rainforest suggests Smithsonian Tropical Research Institute in Balboa, Panama, “has every-
that the volcanic arc from which the isthmus formed was a “species thing you need to tell the story.” –R. S.
the Azuero Peninsula west of the Panama form a peninsula, allowing North American are stunningly similar,” says Coates, referring
Canal and from the San Blas mountains, a fauna to colonize it, as fossils found among to a pair of studies in the Geological Society of
range to the east that extends into Colom- the remnants of the arc testify. By then, pos- America Bulletin in 2003 and 2004. However,
bia. The zircons revealed that a volcanic arc sibly the final barrier to intercontinental he argues, “There’s no way you can tell from
began forming about 70 million years ago movement was a deep-water gap between their data whether parts of the arc were still
in the gap between the two continents, with the peninsula and the coast of modern-day submerged” in the final stages of filling in.
major exhumation events lifting up blocks of Colombia as it subducted under the arc. And the spatial resolution of their data is not
the arc about 47 million, 25 million, and 11 “It’s terrific work,” Jackson says about the fine enough, he says, to rule out deep-water
million years ago. By measuring the direction scenario, which Montes and Jaramillo pub- gaps in the arc.
of the ancient magnetic fields frozen into the lished in the Geological Society of America Coates and Stallard think that Indonesia
volcanic rocks, the researchers were able to Bulletin in January 2012 and in the Journal of may offer a modern-day analog of pre-
reconstruct how tectonic forces had shifted Geophysical Research: Solid Earth that April. isthmus Panama. Over millennia, Asian spe-
and deformed the arc. But he and others reject the headline-grabbing cies have readily hopped or drifted between
Piecing together the evidence, the conclusion that Montes and Jaramillo arrived islands in the Sunda region, west of Bali.
researchers, with geologist David Farris of at: that by 10 million years ago, the geological There, the migration mostly stopped at Bali.
Florida State University in Tallahassee, pro- shape-shifting they meticulously documented Lombok, the next major island to the east, is
posed that the collision between South Amer- had squeezed away the last deep-sea connec- only 35 kilometers away, but the strait is sev-
CREDIT: R. STONE/SCIENCE
ica’s northern Andean blocks and the Central tion between the oceans. eral hundred meters deep and has an espe-
American volcanic arc began about 25 mil- cially swift current, thwarting purposeful or
lion years ago. Panama’s ‘S’ shape, Montes Wallace would be proud accidental migration and forming a species
points out, betrays the tremendous strain that Coates himself accepts Montes and barrier—Wallace’s Line. Several hundred
the arc underwent. Around that time, the arc Jaramillo’s account of the first stages of isth- kilometers to the east is a similar species bar-
melded with the rest of Central America to mus closure. “Our geological reconstructions rier: Lydekker’s Line, the edge of the world for

Published by AAAS
marsupials and other creatures from the Sahul ida and Texas. One gave rise to Megalonyx, by the Maya in Central America, arose in
region, encompassing Australia and New a 1-ton sloth named by Thomas Jefferson, a the Amazon; they reached Central Amer-
Guinea. In between the Wallace and Lydekker fossil buff, who mistook a claw of the beast ica only after the isthmus formed. In 2011,
lines is a collection of islands stretching from for that of a gigantic meat-eater. (Sloths are STRI ecologist David Roubik, a specialist
Borneo to New Guinea that biogeographers vegetarians.) The first emigrant from North on bees, discovered two new, closely related
refer to as Wallacea for its unique faunal America appears to have been the raccoon- stingless bee species: one in Colombia and
assemblage. like carnivore Cyonasua, fossils of which one on Coiba Island, off the Pacific coast of
Coates and Stallard propose that deep- show up in Argentina 7.3 million years ago. Panama. The forest-dwelling bees are not
water gaps in the present-day Panama Canal Terror birds stormed Texas 5 million years accomplished fliers. “They can’t establish a
Basin and between the volcanic arc and South ago, with giant armadillolike pampatheriids new nest across more than a short stretch of
America were equivalent to the Wallace and hot on their heels. open water,” Roubik says. It’s possible that
Lydekker lines. They argue that even as the Scientists have generally chalked up these rafts carried nests to Coiba, he says. But to
isthmus was filling in, the path of migration early migrations to island-hopping or rafting. Roubik, a simpler explanation is that a land
was interrupted for 10 million years or more. Modern sloths and raccoons are good swim- bridge well before 3 million years ago paved
“Essentially, South America is a the way for the bees. “My opinion is
few million years ahead of Austra- that there was something akin to a
lia,” in that Indonesia’s volcanic arc Central America: 6 Million Years Ago little isthmus around 6 to 12 million

and Australia should eventually form years ago,” he says, before portions
a contiguous land mass, says Coates, of the volcanic arc were submerged
whose report with Stallard appeared again, only to reemerge about
online in BMS on 4 April. “It’s a very 3 million years ago.
elegant analog,” says STRI paleobi- That concept could be gaining
ologist Aaron O’Dea. traction. Most scientists accept that
O’Dea and others who adhere to present-day Panama rose from the sea
the late-formation thesis have also over millions of years as the volcanic
taken up the gantlet. In a paper in arc and South America collided. In a
press at BMS, Jackson and O’Dea paper in press at Biological Reviews,
comprehensively review studies on O’Dea; STRI ecologist Egbert
the paleocean environment, disper- Leigh; and Geerat Vermeij of the
sal of terrestrial and marine crea- University of California, Davis, pro-
tures, and molecular phylogeny of pose that about 10 million years ago,
sea life, all pointing, they say, to clo-
Indonesia: Today the shoaling resulted in a land bridge
sure of the isthmus about 3 million that was “briefly near-complete.”
years ago. That would explain how key her-
One especially compelling find, ald species, such as the sloths up
presented at the geology meeting north and the Cyonasua down south,
last November, is shark teeth and pioneered new terrain long before
fish fossils such as otoliths recovered the Great American Biota Inter-
from 6-million-year-old sediments change. By 7 million years ago,
in the Chagres Formation in the Pan- they propose, deep-water gaps in the
ama Canal Basin. The otoliths, or ear emerging isthmus had disrupted the
bones, are from marlins and bluefish: land bridge. “Multiple stages of isth-
species that generally don’t ply shal- mus formation are not inconsistent
CREDITS: COATES ET AL. BULLETIN OF MARINE SCIENCE 5-7. VOL 89, NO 3. © 2013 (2)
low waters. STRI’s Carlos De Gracia with the biological evidence,” Far-
and colleagues argue that the fossil ris says.
assemblage points to waters between That explanation may satisfy
100 and 700 meters deep. Coates some researchers, but at STRI,
calls the Chagres fossils a “smoking Elegant analog? According to the standard model of isthmus formation where many of the debate’s antago-
gun” for the existence of a deep-sea (top), final closure occurred well after 6 million years ago. Deep-water nists are either based or spend time
gap millions of years later than the gaps between the Panama Canal Basin and South America may have been for research, the sparring continues.
closure envisioned by Jaramillo and equivalent to Wallace’s Line and Lydekker’s Line in present-day Indonesia. “I’ll see Carlos in the hall, and flash
Montes. him three fingers. He flashes back
Yet, other pieces of the puzzle don’t fit the mers, Jackson notes, while globally, the annals 15,” says Coates, who calls Jaramillo “a very
late closure thesis. Most troubling are the so- of biogeography are replete with accounts of fine scientist” and says that their interactions
called herald animals. Biogeographers have accidental émigrés borne to distant lands on “are always polite. This is the way science is
long recognized that the Great American clumps of vegetation. supposed to be.”
Biotic Interchange was never a stampede, as if But some scientists find that explanation Good manners can’t disguise the rift,
a drawbridge were lowered over a moat. About hard to swallow. And then there is the case of however. “Both sides,” Montes says, “have
9 million years ago, well before the surge of the stingless bees. entrenched positions.” For opponents in the
migrations, two sloth species descended from About 22 million years ago, ancestors debate, that means, for now, no closure.
South American ancestors appear in Flor- of these honey bees, which were cultivated –RICHARD STONE

Published by AAAS
NEWSFOCUS
Salvage Paleontology on the Seaway pump,” MacFadden says, spawning new life forms that spread through
North America.
PARAISO, PANAMA—On an embankment hewn from the jungle dur- The canal’s construction in the early 20th century revealed a fossil
ing the construction of the Panama Canal a century ago, Jason Bourque trove, which scientists secured piecemeal over the years. The expansion
sits hunched over a 19-million-year-old turtle. Under a broiling mid- is opening up whole new exposures. That’s especially exciting because
day sun, Bourque spends a half hour patiently brushing away grit from throughout the isthmus, lush forest has made it difficult for identify rich
a fractured black carapace the size of a dinner plate before covering it fossil sites, says Aaron Wood, a postdoc at the Florida Museum of Natu-
in a plaster bandage for removal. By the time he’s finished, a dozen or ral History. With the canal expansion slated for completion as early as
so container ships, oil tankers, and other vessels steam past along the next year, the window of opportunity for paleontologists is closing fast.
77-kilometer-long seaway. Racing to gather up as many
Bourque, a preparator at the Flor- fossils as they can, MacFadden’s
ida Museum of Natural History, is tak- crew over 4 years has hauled in more
ing advantage of a fossil windfall. As than 3000 specimens, among them
part of a $5.5 billion expansion of the 10 new species. A top fossil hunter on
canal that will move 152 million cubic the team is Aldo Rincon, a Ph.D. stu-
meters of earth, the canal authority dent at the University of Florida, lead

has opened up sections of the seaway author on a report published in the
to salvage paleontology: collecting Journal of Vertebrate Paleontology
fossils in newly exposed rock faces. last March describing a new species of
“It’s a once-in-century opportunity,” anthracothere. Rincon’s favorite spec-
says Bruce MacFadden, curator of ver- imen, he says, is the fragile jawbone
tebrate paleontology at the museum, and teeth of a bat that he unearthed
run by the University of Florida in from the Cucaracha Formation, in the
Gainesville, and director of the Pan- shadow of the Centennial Bridge, last
ama Canal Project. year. “I rolled down the hill, I was so
The project’s geological stud- Race against time. Ongoing expansion of the Panama Canal has excited,” he says.
ies have provided grist for a radical yielded a treasure trove of fossils. Jason Bourque (background) exca- The bat, descended from a South
hypothesis that the Isthmus of Pan- vates a 19-million-year-old turtle. American species, is a rare example of
ama formed a land bridge several a nonswimmer to have traversed the
million years earlier than thought (see main story, p. 230). Its fossil deep-water gap that separated Colombia and Panama about 19 mil-
finds are also revelatory: a menagerie of miniature horses and cam- lion years ago. (Crocodiles, turtles, and snakes crossed much earlier.)
els, gopherlike rodents, peccaries, and caimans, as well as weird crea- A view of life in this ancient cradle of biodiversity is coming into sharper
tures such as the giant bear-dog and the hippolike anthracothere. The focus. The Panama Canal, says Andrés Cárdenas, a postdoc at the
assemblage recovered so far from the prehistoric rainforest suggests Smithsonian Tropical Research Institute in Balboa, Panama, “has every-
that the volcanic arc from which the isthmus formed was a “species thing you need to tell the story.” –R. S.
the Azuero Peninsula west of the Panama form a peninsula, allowing North American are stunningly similar,” says Coates, referring
Canal and from the San Blas mountains, a fauna to colonize it, as fossils found among to a pair of studies in the Geological Society of
range to the east that extends into Colom- the remnants of the arc testify. By then, pos- America Bulletin in 2003 and 2004. However,
bia. The zircons revealed that a volcanic arc sibly the final barrier to intercontinental he argues, “There’s no way you can tell from
began forming about 70 million years ago movement was a deep-water gap between their data whether parts of the arc were still
in the gap between the two continents, with the peninsula and the coast of modern-day submerged” in the final stages of filling in.
major exhumation events lifting up blocks of Colombia as it subducted under the arc. And the spatial resolution of their data is not
the arc about 47 million, 25 million, and 11 “It’s terrific work,” Jackson says about the fine enough, he says, to rule out deep-water
million years ago. By measuring the direction scenario, which Montes and Jaramillo pub- gaps in the arc.
of the ancient magnetic fields frozen into the lished in the Geological Society of America Coates and Stallard think that Indonesia
volcanic rocks, the researchers were able to Bulletin in January 2012 and in the Journal of may offer a modern-day analog of pre-
reconstruct how tectonic forces had shifted Geophysical Research: Solid Earth that April. isthmus Panama. Over millennia, Asian spe-
and deformed the arc. But he and others reject the headline-grabbing cies have readily hopped or drifted between
Piecing together the evidence, the conclusion that Montes and Jaramillo arrived islands in the Sunda region, west of Bali.
researchers, with geologist David Farris of at: that by 10 million years ago, the geological There, the migration mostly stopped at Bali.
Florida State University in Tallahassee, pro- shape-shifting they meticulously documented Lombok, the next major island to the east, is
posed that the collision between South Amer- had squeezed away the last deep-sea connec- only 35 kilometers away, but the strait is sev-
CREDIT: R. STONE/SCIENCE
ica’s northern Andean blocks and the Central tion between the oceans. eral hundred meters deep and has an espe-
American volcanic arc began about 25 mil- cially swift current, thwarting purposeful or
lion years ago. Panama’s ‘S’ shape, Montes Wallace would be proud accidental migration and forming a species
points out, betrays the tremendous strain that Coates himself accepts Montes and barrier—Wallace’s Line. Several hundred
the arc underwent. Around that time, the arc Jaramillo’s account of the first stages of isth- kilometers to the east is a similar species bar-
melded with the rest of Central America to mus closure. “Our geological reconstructions rier: Lydekker’s Line, the edge of the world for

Published by AAAS
BIODIVERSITY
Atlantic and forming the beginnings of what
The Amazon in 4D is now the Amazon River.

As the wetlands drained and gave way to
forest, the biodiversity that we see in today’s
An interdisciplinary team is linking data from field research in the Amazon into a Amazon began to take shape, Lohmann says.
comprehensive digital collection, aiming to map evolution in space and time New evidence garnered from DNA sequenc-
ing and other evolutionary data point to an
NEW YORK CITY—In a chilly specimen stor- Amazon for more than 45 years. Now, doz- image of the region not as an environmental
age room on the sixth floor of the American ens of scientists from Brazil and the United museum but as what Lohmann calls a biolog-
Museum of Natural History (AMNH), Joel States have teamed up across disciplines to ical “cradle” where large numbers of species
Cracraft opens a cabinet and slides out two try to find an answer. Called Dimensions of have diversified fairly recently. As a result,
trays filled with delicate, palm-sized song- Amazonian Biodiversity, their project aims says Alan Graham, a paleobotanist at the Mis-
birds, inhabitants of the Amazon rainforest. to create a “four-dimensional evolutionary souri Botanical Garden in St. Louis, scientists
One tray holds a jumble of red-headed atlas” of the Amazon. By combining data now “need a whole new suite of explana-
manakins—eyeless, boneless, and stuffed from a variety of fields such as geology, tory factors” to account for the region’s bio-
with cotton. The other holds their close evolutionary biology, DNA analysis, and diversity. Dimensions, launched in Septem-
relative, golden-headed manakins. The two climate modeling, the team plans to track ber 2012 with a promise of $2 million over
species seem to differ only in the color of the region’s biodiversity and environment 5 years from the U.S. government and an
the males’ crown feathers. But they are through space and time to understand how equivalent offer from the state of São Paulo in
never found in the same place. Red-headed they interacted and evolved. Brazil, aims to find them.
manakins live south of the Amazon River; To untangle the Amazon’s complicated
golden-headed manakins live north of it. A volatile place history, scientists from many disciplines will
How that pattern arose is a mystery, says For many years, biologists took the Ama- need to work together, says Carina Hoorn, a
Cracraft, the museum’s curator of ornithol- zon’s abundance of plant and animal species geologist at the University of Amsterdam.
ogy. Scientists don’t know what drove the to mean that its rainforest was an old, stable Hoorn is involved in CLIM-AMAZON, a
birds to evolve different colored crowns. ecosystem—a kind of ecological “museum” European-Brazilian collaboration begun in
They don’t know when the two species began that has “had more time to accumulate more 2011 that is gleaning data on the region’s cli-
to diverge, or the precise age of the river that species,” says botanist Lúcia Lohmann of mate and environmental history from sedi-
separates them—much less, when it became the University of São Paulo in Brazil, who ments throughout the Amazon River Basin.
wide enough to deter birds from flying is, along with Cracraft, a principal investiga- It’s not enough to focus on a single critical
across. They don’t know just how these birds, tor for Dimensions. But geological evidence field, such as geology or evolutionary biol-
and the plants and animals living alongside points to a much more volatile history for ogy, she says. “You really have to see it in an
them, may have adapted to the Amazonian the region, according to Lohmann and her integrated, interdisciplinary way.”
climate, or exactly how that climate may collaborators. From at least 17 million to Cracraft’s manakins provide an example
have changed over time. In short, scientists 11 million years ago, in the middle of an of how such a strategy might work. If you
don’t know how the Amazon rainforest and epoch known as the Miocene, the Amazon were studying Amazonian birds, he says,
its stunning biodiversity came to be. Basin was covered in an extensive system you might plot the location where each of
CREDIT: KIKE CALVO
“It’s one of the great questions of all of wetlands that scientists refer to as Lake AMNH’s manakins was collected and see
time,” says Thomas Lovejoy, biodiversity Pebas. At some point—exactly when is still a sharp north-south separation between the
chair of the Heinz Center in Washington, disputed—the lake’s water began flowing golden-headed and red-headed species. But
D.C., and an ecologist who has worked in the east, eventually breaking through into the only if you overlaid the data on a map of the

Published by AAAS
Uncharted. The origins of Amazonia’s stunning bio- to match up places mentioned in collectors’ This can skew conclusions and bias ongo-
diversity aren’t well understood. original notes with the localities included ining research, Lohmann agrees. “We often go
geographical directories called gazetteers, back to the most collected areas as being the
region’s waterways would it be clear that the or by pulling out paper maps and retracing a centers of diversity just because our data sets
barrier separating them is the Amazon River, collector’s route. don’t allow a very fair comparison through-
the formation of which may have isolated two Some patterns line up, like the distribution
out the area,” she says. If the Dimensions
groups that evolved into separate species. of plants and their pollinators, atlas can reveal which
Knowing precisely how long ago the two while others diverge in curious places and species are
manakins split off from their common ances- ways. Amazonian birds, for exam- already well studied, it
tor might give a good clue as to when the river ple, tend not to fly across rivers, could help scientists tar-
became the major waterway that it is today. meaning that scientists often find get future fieldwork on
“This is one of the purposes of the grant, to strikingly different species on undercollected areas.
try to tease apart that layered history of Ama- either side of a waterway. Many Coordinating such
zonia and [its] biota in a way that can explain plant species, however, “seem a sprawling, inter-
it all,” Cracraft says. “Or at least most of it.” to cross rivers without any prob- disciplinary effort is not
The team plans to go through a similar lems,” Lohmann says. So while easy, and often requires
mapping process with all the region’s birds, the formation of rivers might be getting “out of your com-

primates, and butterflies, as well as two fam- a driving force behind bird evolu- fort zone,” Lohmann
ilies of plants. “To start with, we just want tion in the basin, researchers need admits. As the Dimensions
to understand general patterns of project progresses, it’s
diversity and see how those relate to “essential” that research
environmental features,” Lohmann leaders “be very closely
explains. A project like this is “long attuned to other special-
overdue,” Lovejoy says. ists, and especially those
that have alternate view-
Finding hidden patterns points,” agrees Graham,
Although museum specimens are who is not involved in the
at the heart of the Dimensions proj- project. For example, the Dimensions
ect, much of the information that and CLIM-AMAZON researchers
they contain is not easily accessible disagree on some fundamental ques-
because it’s not yet online. Improv- tions about the region, particularly the
ing digital archives is one of the proj- age of the Amazon River. Hoorn says
ect’s main objectives. The team plans that she welcomes the discussions.
to create a comprehensive, freely “The better people work together, the
available database, including details better the results will be,” she says.
on when, where, and how specimens Although Dimensions focuses
were acquired, which anyone will be on Amazonia’s history, it may help
CREDITS (TOP TO BOTTOM): MICHEL GIRAUD-AUDINE; ALMIR CÂNDIDO DE ALMEIDA; TATIANA ESCOVAR/AMNH
able to use to analyze the region’s bio- scientists and policymakers protect
diversity. “The key will be to make Mapping manakins. Dots represent related manakin species, includ- the increasingly at-risk ecosystem.
this information comparable so that ing the golden-headed (red dots) and red-headed (green dots) species. The atlas, for example, may pinpoint
we can combine stuff and analyze areas where new species are likely
bigger data sets,” Cracraft says. If this suc- to come up with other ideas about how popu- to evolve. Preserving those hotspots of bio-
ceeds, “suddenly it will be possible to look at lations of plants might have become isolated diversity might give Amazonia’s inhabit-
patterns and questions that have been hard to and evolved into separate species. ants a leg up in the race to adapt to Earth’s
actually get at before,” predicts Lovejoy, who The atlas coordinates should reveal not changing climate, says George Gilchrist, the
is not involved in the Dimensions project. just how species are distributed but also National Science Foundation (NSF) pro-
In this first year of their grant, Dimensions where collectors in the past focused their gram director in charge of the Dimensions
researchers have been focusing on adding efforts. Most fieldwork in the Amazon has grant. “If we can’t keep the full diversity of
precise geographical coordinates to records been done along major rivers or near cities. species there, can we keep the evolutionary
of museum and herbarium specimens. Loca- But collecting in just a few areas, no mat- potential that generated [it]?” he wonders.
tion is the common factor that runs through ter how extensive, does not guarantee a rep- Ultimately, scientists and NSF officials
everything, from plants to birds to primates, resentative sample of diversity—especially hope that Dimensions will serve as a model
and even “the paleo and geological data,” because Amazonian species tend to be very for interdisciplinary studies about the evo-
says Barbara Thiers, director of the New York particular about where they live, even when lutionary history of all kinds of ecosystems.
Botanical Garden’s herbarium, which is con- barriers separating one area from another Cracraft, for one, seems to have already taken
tributing to the Dimensions project. “Col- aren’t yet clear to scientists. “People have the Dimensions’ strategy to heart. Outside the
lectors didn’t routinely record [latitude and basically been looking at nice, cute things room where the AMNH manakins are stored,
longitude] information on the specimens” at eye-height … things that were easy to get he calls to a group of museum employees
until handheld GPS devices came along, at,” says Alexandre Antonelli, an evolution- about to go bird watching in Central Park,
she says. So the Dimensions team is recon- ary biologist at the University of Gothen- “Tell me what you see and where!”
structing coordinates as best they can, trying burg in Sweden. –LIZZIE WADE

Published by AAAS
COMMENTARY
A “cybrarian” for online States of self-assembly
education tools
240 243
LETTERS I BOOKS I POLICY FORUM I EDUCATION FORUM I PERSPECTIVES
LETTERS
edited by Jennifer Sills
5. H. Hu, D. Christiani, Lancet 339, 1535 (1992).

Turkey Must End Violent Response to Protests 6. B. F. Bessac, S.-E. Jordt, Proc. Am. Thorac. Soc. 7, 269
(2010).
7. B. C. Howard, “The surprising history and science
THE TURKISH MEDICAL ASSOCIATION (TTB) REPORTED THAT 8121 PEOPLE WERE OFFICIALLY of tear gas,” National Geographic Daily News (12
admitted to hospitals resulting from police violence between 31 May and 26 June (1). This June 2013); http://news.nationalgeographic.com/
number includes 5 deaths, 61 life-threatening injuries, 104 head traumas, and 11 ophthalmic news/2013/06/130612-tear-gas-history-science-turkey-
protests/.
injuries, one of which led to loss of an eye due to shots of tear gas canisters from short range
8. H. G. Atkinson, “Under the gun: Ongoing assaults on
(1). Turkish police have used excessive amounts of Bahrain’s health system,” Physicians for Human Rights
tear gas (lachrymatory agents) in public (2) and con- (May, 2012); http://physiciansforhumanrights.org/library/
fined spaces such as hospitals or infirmaries, accord- reports/under-the-gun-ongoing-assaults-on-bahrains-
health-system.html.
ing to international media reports and the TTB 9. United Nations Geneva Protocol (1925); www.icrc.org/
(1–4). Such use of asphyxiating gases in confined eng/war-and-law/treaties-customary-law/geneva-
spaces is not only extremely dangerous for public conventions/index.jsp.
10. “Turkish police to buy 100,000 gas bomb cartridges,”
health (5–8), but also strictly limited by interna- Hurriyet Daily News (19 June 2013); www.hurriyetdai
tional agreements, such as the Geneva Protocol (9), lynews.com/turkish-police-to-buy-100000-gas-bomb-
to which Turkey is a signatory. Security forces have cartridges-.aspx?PageID=238&NID=49075&NewsCa
used 130,000 tear-gas cartridges in 20 days, and tID=341.
11. Lancet 381, 2067 (2013).
Turkey is planning to buy 100,000 new cartridges 12. E. Güne, “Police search houses of Gezi protests’ starter
(10). Doctors and nurses treating patients affected group members,” Hurriyet Daily News (9 July 2013);
by tear gas and other police brutality, as well as the www.hurriyetdailynews.com/police-search-houses-of-
gezi-protests-starter-group-members.aspx?pageID=238
Show of force. Turkish riot police fire tear Istanbul Medical Chamber General Secretary, have &nID=50349&NewsCatID=339.
gas at protestors in Istanbul’s Taksim Square. been apprehended by police (11, 12), a clear viola- 13. Convention (IV) Relative to the Protection of Civilian Per-
tion of customary international and human rights sons in Time of War (Geneva, 1949); www.icrc.org/applic/
ihl/ihl.nsf/Treaty.xsp?documentId=AE2D398352C5B028C
law (13). More than 4000 academics around the world have already signed a petition to protest
12563CD002D6B5C&action=openDocument.
the police brutality (14). We call upon the Turkish government to obey international law in the 14. Academics for Gezi (http://academicsforgezi.com/
treatment of protesters and those providing medical treatment to them, and to start a good-faith our-call/).
dialogue with the protest movement.
EMRAH ALTINDIS,1* M. ALI ALPAR,2 EMRE AKSAY,3 JONATHAN BECKWITH,1 CHRISTIAN BÖKEL,4 ROBERT F.
CURL,5 ROBERT B. DARNELL,6 STEPHEN J. ELLEDGE,7 BURAK ERMAN,8 JENS FRAHM,9 STEPHEN P. GOFF,10
Optimizing Peer Review of
PAUL GREENGARD,6 ROALD HOFFMANN,11 BAYAZIT ILHAN,12 JAN KASLIN,13 STEVEN M. LIPKIN,3 CORNELIA
POULOPOULOU,14 EREZ RAZ,15 MARK A. RUBIN,3 MEHMET SALTURK,16 RICHARD R. SCHROCK,17 ALAIN
Software Code
TRAUTMANN,18 DERYA UNUTMAZ,19 HAREL WEINSTEIN,3 CAGHAN KIZIL4* IN THEIR POLICY FORUM “TROUBLING TRENDS
1
Harvard Medical School, Boston, MA 02115, USA. 2The Science Academy (Bilim Akademisi), Istanbul, Turkey. 3Weill Cornell in scientific software use” (17 May, p. 814),
Medical College of Cornell University, New York, NY 10065, USA. 4Center for Regenerative Therapies Dresden, Technische L. N. Joppa et al. make a case for increased
Universität, 01307, Dresden, Germany. 5Rice University, Houston, TX 77251, USA. 6The Rockefeller University and The New
York Genome Center, New York, NY 10065, USA. 7Harvard Medical School, Brigham and Women’s Hospital, Boston, MA education of scientists in computer program-
02115, USA. 8Koç University, Istanbul, Turkey. 9Max Planck Institute for Biophysical Chemistry, 37077, Göttingen, Germany. ming skills and requirements for peer review
10
Columbia University, NY 10027, USA. 11Cornell University, Department of Chemistry and Chemical Biology, Ithaca, NY, of scientific software code. We agree in prin-
14853, USA. 12Turkish Medical Association (TTB), Ankara, Turkey. 13Monash University, Victoria, 3800, Australia. 14University
of Athens, Athens, 10679, Greece. 15University of Münster, 48149, Münster, Germany. 16University of Cologne, Cologne, ciple but believe that some of the specific rec-
50923, Germany. 17Massachusetts Institute of Technology, Cambridge, MA 02139, USA. 18Centre National de la Recherche ommendations by Joppa et al. are unfeasible.
Scientifique (CNRS), 75794, Paris, France. 19New York University, New York, NY 10012, USA. In particular, we believe that requiring
CREDIT: OSMAN ORSAL/REUTERS
*Corresponding authors. E-mail: emrah_altindis@hms.harvard.edu (E.A.); caghan.kizil@crt-dresden.de (C.K.)

prepublication peer review of computer
source code by journal reviewers would
References place impossible strain on an already over-
1. TTB Report (www.ttb.org.tr/index.php/Haberler/veriler-3842.html).
2. Gas Burns—Taksim Gezi Parki 12.06.2013 (www.youtube.com/watch?v=66e-jRqMEZg). burdened system. Many scientific journals
3. A. O. Aktan, Br. Med. J. 346, f3801 (2013). currently have great difficulty finding suffi-
4. Police Throwing Tear Gas into German Hospital İstanbul 16.06.13 2.57am #occupygezi (www.youtube.com/ cient numbers of qualified reviewers to eval-
watch?v=AL06wWNqTY4).
uate submissions in a timely and construc-

Published by AAAS
Chemistry pioneer Waste not, by far the more challenging issue being
want not software validation. Addressing this issue,
together with the equally pressing issue of
244 247 uncertainty quantification in complex [com-

putational] models, has been the focus of
intensive research efforts in other scientific
disciplines (2). These efforts might provide
a good starting point for equivalent efforts in
the life sciences.
tive fashion. Requiring that reviewers be less of the strength of processes for dealing LUCAS N. JOPPA,1* DAVID GAVAGHAN,2 RICHARD
able to evaluate not only the scientific merit with corrections and retractions, putting “the HARPER,1 KENJI TAKEDA,1 STEPHEN EMMOTT1
of a manuscript but also parse, understand, genie back in the bottle” is always going to be 1
Microsoft Research, Cambridge, CB3 0FB, UK. 2Depart-
and evaluate what can be thousands or tens a difficult task after a result has been reported ment of Computer Science, University of Oxford, Oxford,
OX1 3QD, UK.
of thousands of lines of source code written in the literature. At a minimum, code needs to
in one or more of a variety of programming be available to reviewers should they choose *Corresponding author. E-mail: lujoppa@microsoft.com
languages is impractical. to scrutinize it. Moreover, prepublication
Reference
A more tenable solution for computer review of code need not necessarily rely on 1. Iron Mountain, “How verification services fortify your
codes is postpublication peer review, where the current review system. Just as English- software escrow solution” (Iron Mountain, 2011).
the release of source code is a requirement of language editing services have emerged to 2. National Academies Press, Assessing the Reliability of
Complex Models: Mathematical and Statistical Foundations
publication (1) and interested and appropri- ensure a minimum standard of accessibility
of Verification, Validation, and Uncertainty Quantification
ately skilled members of the broader scien- of articles in many major journals, so might (National Academies Press, Washington, DC, 2012).
tific community may download and evaluate software-reviewing services provide a stamp
the code at will. This would bring review of of approval that code actually implements the
computer code in line with existing policies algorithm reported in a paper. Indeed, in the
pertaining to data and materials availability commercial sector, software escrow provid-
Letters to the Editor
already in place at most journals. ers routinely provide full verification services Letters (~300 words) discuss material published in
Increased dependence on postpublica- to companies purchasing (or investing in) Science in the past 3 months or matters of gen-
eral interest. Letters are not acknowledged upon
tion review will require strengthening proce- business-critical software [e.g., (1)], and the
receipt. Whether published in full or in part, Let-
dures and facilities for reporting corrections approaches used by such companies might ters are subject to editing for clarity and space.
and retractions of published research articles, provide pointers for a new model for aca- Letters submitted, published, or posted elsewhere,
a course of action long advocated by those demic software verification services. in print or online, will be disqualified. To submit a
concerned over increasing rates of retraction, Of course, verification of software is just Letter, go to www.submit2science.org.
especially in the biomedical sciences (2). the first essential step in the process, with
PIOTR SLIZ* AND ANDREW MORIN
BCMP, Harvard Medical School, Boston, MA 02115, USA.
CORRECTIONS AND CLARIFICATIONS
*Corresponding author. E-mail: piotr_sliz@hms.harvard.edu
News & Analysis: “NIH to phase out most chimp research” by J. Kaiser (5 July, p. 17). The article failed to note
References that Gabon has not banned biomedical research on chimpanzees. The HTML and PDF versions online have
1. A. Morin et al., Science 336, 159 (2012). been corrected.
2. Retraction Watch (http://retractionwatch.wordpress.com/).
Reports: “A population of fast radio bursts at cosmological distances,” by D. Thornton et al. (5 July, p. 53). A mistake
in the script used to produce Table 1 caused the energy released (E) values in the last row of the table to be off by
Response 106. The correct values are ~1033, ~1031, ~1032, and ~1031. The conclusions of the paper are unaffected. The HTML
and PDF versions online have been corrected.
SLIZ AND MORAN QUESTION THE FEASIBILITY
of our recommendation to both peer-review
TECHNICAL COMMENT ABSTRACTS
computer code and release it, and they prof-
fer an alternative: postpublication commu-
nity review and stronger procedures and
Comment on “Can We Name Earth’s Species Before They Go Extinct?”
facilities for dealing with corrections and Camilo Mora, Audrey Rollo, Derek P. Tittensor
retractions of published results. These are Costello et al. (Review, 25 January 2013, p. 413) challenged the common view that many species are disappear-
not incompatible. Encouraging the broader ing before being described. We suggest that their conclusion is overly optimistic because of a limited selection and
interpretation of available evidence that tends to overestimate rates of species description and underestimate the
scientific community to inspect computer number of species on Earth and their current extinction rate.
code postpublication would help in identi- Full text at http://dx.doi.org/10.1126/science.1237254
fying scientific errors currently unnoticed in
the scientific literature. Improving the pro- Response to Comment on “Can We Name Earth’s Species Before They Go
cess of corrections and retractions would Extinct?”
have positive benefits far beyond this issue.
Mark J. Costello, Robert M. May, Nigel E. Stork
However, neither negates the need for pre-
Mora et al. disputed that most species will be discovered before they go extinct, but not our main recommendations
publication review of code. to accelerate species’ discoveries. We show that our conclusions would be unaltered by discoveries of more micro-
The scientific publishing process relies scopic species and reinforce our estimates of species description and extinction rates, that taxonomic effort has
on prepublication peer review as a filter for never been greater, and that there are 2 million to 8 million species on Earth.
robust results. This is so because, regard- Full text at http://dx.doi.org/10.1126/science.1237381

Published by AAAS
COMMENTARY
A “cybrarian” for online States of self-assembly
education tools
240 243
LETTERS I BOOKS I POLICY FORUM I EDUCATION FORUM I PERSPECTIVES
LETTERS
edited by Jennifer Sills
5. H. Hu, D. Christiani, Lancet 339, 1535 (1992).

Turkey Must End Violent Response to Protests 6. B. F. Bessac, S.-E. Jordt, Proc. Am. Thorac. Soc. 7, 269
(2010).
7. B. C. Howard, “The surprising history and science

THE TURKISH MEDICAL ASSOCIATION (TTB) REPORTED THAT 8121 PEOPLE WERE OFFICIALLY of tear gas,” National Geographic Daily News (12
admitted to hospitals resulting from police violence between 31 May and 26 June (1). This June 2013); http://news.nationalgeographic.com/
number includes 5 deaths, 61 life-threatening injuries, 104 head traumas, and 11 ophthalmic news/2013/06/130612-tear-gas-history-science-turkey-
protests/.
injuries, one of which led to loss of an eye due to shots of tear gas canisters from short range
8. H. G. Atkinson, “Under the gun: Ongoing assaults on
(1). Turkish police have used excessive amounts of Bahrain’s health system,” Physicians for Human Rights
tear gas (lachrymatory agents) in public (2) and con- (May, 2012); http://physiciansforhumanrights.org/library/
fined spaces such as hospitals or infirmaries, accord- reports/under-the-gun-ongoing-assaults-on-bahrains-
health-system.html.
ing to international media reports and the TTB 9. United Nations Geneva Protocol (1925); www.icrc.org/
(1–4). Such use of asphyxiating gases in confined eng/war-and-law/treaties-customary-law/geneva-
spaces is not only extremely dangerous for public conventions/index.jsp.
10. “Turkish police to buy 100,000 gas bomb cartridges,”
health (5–8), but also strictly limited by interna- Hurriyet Daily News (19 June 2013); www.hurriyetdai
tional agreements, such as the Geneva Protocol (9), lynews.com/turkish-police-to-buy-100000-gas-bomb-
to which Turkey is a signatory. Security forces have cartridges-.aspx?PageID=238&NID=49075&NewsCa
used 130,000 tear-gas cartridges in 20 days, and tID=341.
11. Lancet 381, 2067 (2013).
Turkey is planning to buy 100,000 new cartridges 12. E. Güne, “Police search houses of Gezi protests’ starter
(10). Doctors and nurses treating patients affected group members,” Hurriyet Daily News (9 July 2013);
by tear gas and other police brutality, as well as the www.hurriyetdailynews.com/police-search-houses-of-
gezi-protests-starter-group-members.aspx?pageID=238
Show of force. Turkish riot police fire tear Istanbul Medical Chamber General Secretary, have &nID=50349&NewsCatID=339.
gas at protestors in Istanbul’s Taksim Square. been apprehended by police (11, 12), a clear viola- 13. Convention (IV) Relative to the Protection of Civilian Per-
tion of customary international and human rights sons in Time of War (Geneva, 1949); www.icrc.org/applic/
ihl/ihl.nsf/Treaty.xsp?documentId=AE2D398352C5B028C
law (13). More than 4000 academics around the world have already signed a petition to protest
12563CD002D6B5C&action=openDocument.
the police brutality (14). We call upon the Turkish government to obey international law in the 14. Academics for Gezi (http://academicsforgezi.com/
treatment of protesters and those providing medical treatment to them, and to start a good-faith our-call/).
dialogue with the protest movement.
EMRAH ALTINDIS,1* M. ALI ALPAR,2 EMRE AKSAY,3 JONATHAN BECKWITH,1 CHRISTIAN BÖKEL,4 ROBERT F.
CURL,5 ROBERT B. DARNELL,6 STEPHEN J. ELLEDGE,7 BURAK ERMAN,8 JENS FRAHM,9 STEPHEN P. GOFF,10
Optimizing Peer Review of
PAUL GREENGARD,6 ROALD HOFFMANN,11 BAYAZIT ILHAN,12 JAN KASLIN,13 STEVEN M. LIPKIN,3 CORNELIA
POULOPOULOU,14 EREZ RAZ,15 MARK A. RUBIN,3 MEHMET SALTURK,16 RICHARD R. SCHROCK,17 ALAIN
Software Code
TRAUTMANN,18 DERYA UNUTMAZ,19 HAREL WEINSTEIN,3 CAGHAN KIZIL4* IN THEIR POLICY FORUM “TROUBLING TRENDS
1
Harvard Medical School, Boston, MA 02115, USA. 2The Science Academy (Bilim Akademisi), Istanbul, Turkey. 3Weill Cornell in scientific software use” (17 May, p. 814),
Medical College of Cornell University, New York, NY 10065, USA. 4Center for Regenerative Therapies Dresden, Technische L. N. Joppa et al. make a case for increased
Universität, 01307, Dresden, Germany. 5Rice University, Houston, TX 77251, USA. 6The Rockefeller University and The New
York Genome Center, New York, NY 10065, USA. 7Harvard Medical School, Brigham and Women’s Hospital, Boston, MA education of scientists in computer program-
02115, USA. 8Koç University, Istanbul, Turkey. 9Max Planck Institute for Biophysical Chemistry, 37077, Göttingen, Germany. ming skills and requirements for peer review
10
Columbia University, NY 10027, USA. 11Cornell University, Department of Chemistry and Chemical Biology, Ithaca, NY, of scientific software code. We agree in prin-
14853, USA. 12Turkish Medical Association (TTB), Ankara, Turkey. 13Monash University, Victoria, 3800, Australia. 14University
of Athens, Athens, 10679, Greece. 15University of Münster, 48149, Münster, Germany. 16University of Cologne, Cologne, ciple but believe that some of the specific rec-
50923, Germany. 17Massachusetts Institute of Technology, Cambridge, MA 02139, USA. 18Centre National de la Recherche ommendations by Joppa et al. are unfeasible.
Scientifique (CNRS), 75794, Paris, France. 19New York University, New York, NY 10012, USA. In particular, we believe that requiring
CREDIT: OSMAN ORSAL/REUTERS
*Corresponding authors. E-mail: emrah_altindis@hms.harvard.edu (E.A.); caghan.kizil@crt-dresden.de (C.K.)

prepublication peer review of computer
source code by journal reviewers would
References place impossible strain on an already over-
1. TTB Report (www.ttb.org.tr/index.php/Haberler/veriler-3842.html).
2. Gas Burns—Taksim Gezi Parki 12.06.2013 (www.youtube.com/watch?v=66e-jRqMEZg). burdened system. Many scientific journals
3. A. O. Aktan, Br. Med. J. 346, f3801 (2013). currently have great difficulty finding suffi-
4. Police Throwing Tear Gas into German Hospital İstanbul 16.06.13 2.57am #occupygezi (www.youtube.com/ cient numbers of qualified reviewers to eval-
watch?v=AL06wWNqTY4).
uate submissions in a timely and construc-

Published by AAAS
Chemistry pioneer Waste not, by far the more challenging issue being
want not software validation. Addressing this issue,
together with the equally pressing issue of
244 247 uncertainty quantification in complex [com-

putational] models, has been the focus of
intensive research efforts in other scientific
disciplines (2). These efforts might provide
a good starting point for equivalent efforts in
the life sciences.
tive fashion. Requiring that reviewers be less of the strength of processes for dealing LUCAS N. JOPPA,1* DAVID GAVAGHAN,2 RICHARD
able to evaluate not only the scientific merit with corrections and retractions, putting “the HARPER,1 KENJI TAKEDA,1 STEPHEN EMMOTT1
of a manuscript but also parse, understand, genie back in the bottle” is always going to be 1
Microsoft Research, Cambridge, CB3 0FB, UK. 2Depart-
and evaluate what can be thousands or tens a difficult task after a result has been reported ment of Computer Science, University of Oxford, Oxford,
OX1 3QD, UK.
of thousands of lines of source code written in the literature. At a minimum, code needs to
in one or more of a variety of programming be available to reviewers should they choose *Corresponding author. E-mail: lujoppa@microsoft.com
languages is impractical. to scrutinize it. Moreover, prepublication
Reference
A more tenable solution for computer review of code need not necessarily rely on 1. Iron Mountain, “How verification services fortify your

codes is postpublication peer review, where the current review system. Just as English- software escrow solution” (Iron Mountain, 2011).
the release of source code is a requirement of language editing services have emerged to 2. National Academies Press, Assessing the Reliability of
Complex Models: Mathematical and Statistical Foundations
publication (1) and interested and appropri- ensure a minimum standard of accessibility
of Verification, Validation, and Uncertainty Quantification
ately skilled members of the broader scien- of articles in many major journals, so might (National Academies Press, Washington, DC, 2012).
tific community may download and evaluate software-reviewing services provide a stamp
the code at will. This would bring review of of approval that code actually implements the
computer code in line with existing policies algorithm reported in a paper. Indeed, in the
pertaining to data and materials availability commercial sector, software escrow provid-
Letters to the Editor
already in place at most journals. ers routinely provide full verification services Letters (~300 words) discuss material published in
Increased dependence on postpublica- to companies purchasing (or investing in) Science in the past 3 months or matters of gen-
eral interest. Letters are not acknowledged upon
tion review will require strengthening proce- business-critical software [e.g., (1)], and the
receipt. Whether published in full or in part, Let-
dures and facilities for reporting corrections approaches used by such companies might ters are subject to editing for clarity and space.
and retractions of published research articles, provide pointers for a new model for aca- Letters submitted, published, or posted elsewhere,
a course of action long advocated by those demic software verification services. in print or online, will be disqualified. To submit a
concerned over increasing rates of retraction, Of course, verification of software is just Letter, go to www.submit2science.org.
especially in the biomedical sciences (2). the first essential step in the process, with
PIOTR SLIZ* AND ANDREW MORIN
BCMP, Harvard Medical School, Boston, MA 02115, USA.
CORRECTIONS AND CLARIFICATIONS
*Corresponding author. E-mail: piotr_sliz@hms.harvard.edu
News & Analysis: “NIH to phase out most chimp research” by J. Kaiser (5 July, p. 17). The article failed to note
References that Gabon has not banned biomedical research on chimpanzees. The HTML and PDF versions online have
1. A. Morin et al., Science 336, 159 (2012). been corrected.
2. Retraction Watch (http://retractionwatch.wordpress.com/).
Reports: “A population of fast radio bursts at cosmological distances,” by D. Thornton et al. (5 July, p. 53). A mistake
in the script used to produce Table 1 caused the energy released (E) values in the last row of the table to be off by
Response 106. The correct values are ~1033, ~1031, ~1032, and ~1031. The conclusions of the paper are unaffected. The HTML
and PDF versions online have been corrected.
SLIZ AND MORAN QUESTION THE FEASIBILITY
of our recommendation to both peer-review
TECHNICAL COMMENT ABSTRACTS
computer code and release it, and they prof-
fer an alternative: postpublication commu-
nity review and stronger procedures and
Comment on “Can We Name Earth’s Species Before They Go Extinct?”
facilities for dealing with corrections and Camilo Mora, Audrey Rollo, Derek P. Tittensor
retractions of published results. These are Costello et al. (Review, 25 January 2013, p. 413) challenged the common view that many species are disappear-
not incompatible. Encouraging the broader ing before being described. We suggest that their conclusion is overly optimistic because of a limited selection and
interpretation of available evidence that tends to overestimate rates of species description and underestimate the
scientific community to inspect computer number of species on Earth and their current extinction rate.
code postpublication would help in identi- Full text at http://dx.doi.org/10.1126/science.1237254
fying scientific errors currently unnoticed in
the scientific literature. Improving the pro- Response to Comment on “Can We Name Earth’s Species Before They Go
cess of corrections and retractions would Extinct?”
have positive benefits far beyond this issue.
Mark J. Costello, Robert M. May, Nigel E. Stork
However, neither negates the need for pre-
Mora et al. disputed that most species will be discovered before they go extinct, but not our main recommendations
publication review of code. to accelerate species’ discoveries. We show that our conclusions would be unaltered by discoveries of more micro-
The scientific publishing process relies scopic species and reinforce our estimates of species description and extinction rates, that taxonomic effort has
on prepublication peer review as a filter for never been greater, and that there are 2 million to 8 million species on Earth.
robust results. This is so because, regard- Full text at http://dx.doi.org/10.1126/science.1237381

Published by AAAS
Comment on ''Can We Name Earth's Species Before They Go
Extinct?''
Camilo Mora et al.
Science 341, 237 (2013);
DOI: 10.1126/science.1237254
This copy is for your personal, non-commercial use only.
If you wish to distribute this article to others, you can order high-quality copies for your
colleagues, clients, or customers by clicking here.

Permission to republish or repurpose articles or portions of articles can be obtained by
following the guidelines here.
The following resources related to this article are available online at

www.sciencemag.org (this information is current as of July 19, 2013 ):
Updated information and services, including high-resolution figures, can be found in the online
version of this article at:
http://www.sciencemag.org/content/341/6143/237.3.full.html
A list of selected additional articles on the Science Web sites related to this article can be
found at:
http://www.sciencemag.org/content/341/6143/237.3.full.html#related
This article cites 13 articles, 3 of which can be accessed free:
http://www.sciencemag.org/content/341/6143/237.3.full.html#ref-list-1
Science (print ISSN 0036-8075; online ISSN 1095-9203) is published weekly, except the last week in December, by the
American Association for the Advancement of Science, 1200 New York Avenue NW, Washington, DC 20005. Copyright
2013 by the American Association for the Advancement of Science; all rights reserved. The title Science is a
registered trademark of AAAS.
TECHNICAL COMMENT
ered with caution as several other sources of
duplication may exist, including basionyms (i.e.,
Comment on “Can We Name Earth’s

valid species that receive new names to control
for homonyms and changes to different taxo-
nomic ranks or positions) and typographical er-
Species Before They Go Extinct?” rors. For comparison, the combined record of
species from all kingdoms of life in the Catalog
of Life, the World Register of Marine Species,
Camilo Mora,1 Audrey Rollo,2 Derek P. Tittensor3,4,5 and the List of Prokaryotic Names with Stand-
ing in Nomenclature indicates that on average
Costello et al. (Review, 25 January 2013, p. 413) challenged the common view that many ~8000 valid species were described annually
species are disappearing before being described. We suggest that their conclusion is overly from 1990 to 2000 (this period was chosen to
optimistic because of a limited selection and interpretation of available evidence that tends to avoid the time lag it takes to store species names
overestimate rates of species description and underestimate the number of species on Earth into these databases).
and their current extinction rate. Are current extinction rates overestimated?
Costello et al. considered that “realistic” extinc-
n an attempt to draw attention to the chal- basis and considerable subjectivity (9, 10). In tion rates likely range between ~0.001% and
I lenge of describing and monitoring Earth’s turn, a broader consideration of available esti- ~0.1% per year (6), which is equivalent to ~50 to

biodiversity in the face of looming environ- mates would have given values above 8 million ~5000 species going extinct annually, assuming
mental crises, it has been frequently claimed that species even without including controversial hy- 5 million species on Earth. Although there re-
many species are likely going extinct before being perdiverse estimates (6). For example, the esti- mains great uncertainty about current extinction
named (1–5). Costello and colleagues (6) chal- mates compiled by Scheffers et al. (5) yield 7.8 rates (12, 13), an assessment of available evi-
lenged this view and suggested that such a worry to 8.7 million species of animals, 0.29 to 0.39 dence supports a higher rate. For instance, anal-
“result[s] from overestimates of how many spe- million plants, and 0.6 to 1.5 million fungi (ex- ysis of the estimates compiled by Stork (13),
cies may exist, beliefs that the expertise to describe cluding hyperdiverse estimates), for a total of which included estimates derived from species-
species is decreasing, and alarmist estimates 8.7 to 10.6 million species (without including area relationships (i.e., extinctions induced by
of extinction rates.” We suggest that their con- protists and prokaryotes). Costello et al. also deforestation and climate change), Red List cat-
clusions are overly optimistic because of a lim- suggested that the fact that the number of spe- egory changes, and co-extinctions, showed that
ited selection and interpretation of the available cies per author is decreasing is an indication the average rate of species extinction was 0.72%
evidence. We consider each strand of evidence that the pool of undescribed species on Earth is
individually. getting smaller. However, the fact that ~0.5
Are there fewer species? Previous estimates million species may be in museum jars awaiting Millions
of the number of species on Earth have fallen description (6) suggests that the downward trend 8
broadly between 3 and 100 million species [see in the number of species per author is unlikely to
table 1 in (3) and table 1 in (5)]. Costello et al.’s be due to a shortfall of species to name; another
consensus estimate of 5 T 3 million species falls possibility that should also be considered regard- 6
Number of species
at the low end of this range primarily because ing the downward trend in the number of species
of the inclusion of two very low values as part per author is that there may now be more coau-
of a small sample of estimates. We argue that thors per described species. 4
there is little support for a preferential selec- Is taxonomic effort overestimated? The num-
tion of those two estimates and that a broader ber of species described each year is a key metric
review of available estimates will give higher of taxonomic effort for estimating the time it 2 MEE
consensus values. One of these studies was de- would take to describe the unknown species on
rived from extrapolating species description rates Earth. Costello et al. suggest that some 18,000
and provides perhaps the lowest estimate yet species are described every year, citing work 0
of the total number of species on Earth: 2 mil- derived mainly from The International Plant 2000 2100 2200 2300 2400
lion (7). The other combined estimates from Names Index and The Zoological Record (11). Year
species description rates, ratios of undescribed However, these databases are repositories for
species in samples, and expert opinions and nominal species (i.e., all species that have re- Fig. 1. Contemporary rates and extrapolations
put the global number of marine species at 0.7 ceived a name regardless of their current status), of species description (blue lines) and extinc-
to 1 million (8). The limitations of these meth- and thus their use may err on the side of over- tion (red lines). Although neither description nor
odologies are well known: Modeling rates of estimation when quantifying rates of description extinction rates will remain constant over time,
species description underpredict true values of valid species. For instance, we cross-checked Costello et al.’s (6) approach is useful at indicating
when all extant species will be described by iden-
when using incomplete data [figure 3, K to O, all 229,309 species names reported since 2000
tifying the year when the trend of described species
in (3)]; ratios are highly variable depending on in the Zoological Record with authoritative data-
(on average, 8000 species a year, blue line) inter-
how well the sampled areas have been studied bases such as the Catalog of Life and the World cepts with the trend of extinct species (on average,
(9); and expert opinions have limited empirical Register of Marine Species and found that only 0.72% a year, solid red line; dotted red line in-
56,397 were recognized as valid species and dicates the average rate after removing extinction
1
Department of Geography, University of Hawaii, Honolulu, 47,395 as synonyms, invalid names, and/or dupli-
Hawaii, USA. 2Pacific Islands Fisheries Science Center, Honolulu,
rates yielded by species-area relationships: 0.22%
cates due to either variations in the names of au- a year). Individual rates analyzed here are shown in
Hawaii, USA. 3United Nations Environment Programme World
Conservation Monitoring Centre, Cambridge, UK. 4Microsoft Re- thors or by also being named as subspecies and/or light colors. As reference, the extinction level in the
search Computational Science Laboratory, Cambridge, UK. 5De- as subgenera. Unfortunately, 125,517 species names five previous mass extinction events (MEE, >75%
partment of Biology, Dalhousie University, Halifax, Canada. in the Zoological Record could not be matched of all species going extinct) is indicated with the
*Corresponding author. E-mail: cmora@hawaii.edu to authoritative databases and should be consid- horizontal gray line.
www.sciencemag.org SCIENCE VOL 341 19 JULY 2013 237-c

TECHNICAL COMMENT
per year (~36,000 species a year if there are 5 mil- proach as Costello et al. and assuming 8 million 5. B. R. Scheffers, L. N. Joppa, S. L. Pimm, W. F. Laurance,
lion species). Only two out of the 15 case studies species, 8000 species described per year, and a Trends Ecol. Evol. 27, 501–510 (2012).
6. M. J. Costello, R. M. May, N. E. Stork, Science 339,
were as low as the highest “realistic” rate proposed yearly extinction rate of 0.72%—all intermediate 413–416 (2013).
by Costello et al. Removing all rates derived from values—we estimate that by the year 2164 all 7. M. J. Costello, S. P. Wilson, B. Houlding, Syst. Biol. 61,
species-area relationships, which are currently extant species will be described, at which point 871–883 (2012).
debated [(12), but see (14)], still yields a mean ~5 million species will become extinct (solid black 8. W. Appeltans et al., Curr. Biol. 22, 2189–2202
(2012).
extinction rate of ~ 0.22% or ~11,000 species a arrow in Fig. 1). If extinction rates are set to 0.22% 9. P. Bouchet, in The Exploration of Marine Biodiversity:
year if there are 5 million species. Although it a year (i.e., the rate after excluding those derived Scientific and Technological Challenges, C. M. Duarte, Ed.
is true that our current record of extinct spe- by species-area), by 2327 all extant species would (Fundación BBVA, Madrid, 2006), pp. 31–62.
cies is very low [~800 recorded extinctions by be discovered and ~4 million would be extinct 10. T. L. Erwin, Conserv. Biol. 5, 330 (1991).
the IUCN [www.iucnredlist.org/about/summary- (dashed black arrow in Fig. 1). These values are 11. Q. D. Wheeler, S. Pennak, State of Observed Species
(International Institute for Species Exploration, 2011).
statistics#Tables_3_4 (15)], this number needs far less comforting than those of Costello et al. (6). 12. F. He, S. P. Hubbell, Nature 482, E5–E6 (2012).
to be considered within the context that only Although we agree with Costello et al. that 13. N. E. Stork, Biodivers. Conserv. 19, 357–371 (2010).
~ 65,000 species have been assessed to deter- alarmist overestimates of biodiversity loss are 14. J. Rybicki, I. Hanski, Ecol. Lett. 16 (suppl. 1), 27–38
mine their current extinction vulnerability (www. unhelpful, we also contend that underestimating (2013).
15. C. Mora, F. A. Zapata, The Balance of Nature and Human
iucnredlist.org/documents/summarystatistics/ the task ahead and overestimating our capacity Impact, K. Krohde, Ed. (Cambridge Univ. Press,
2012_2_ RL_Stats_Table_1.pdf ) and that of those to address it could lead to a false sense of con- Cambridge, 2013).
only ~55% have assessments based on complete fidence. Clearly, multiple sources of uncertainty 16. S. Pimm, P. Raven, A. Peterson, Ç. H. Şekercioğlu,
information [www.iucnredlist.org/about/red-list- remain in our knowledge of biodiversity and its P. R. Ehrlich, Proc. Natl. Acad. Sci. U.S.A. 103,

10941–10946 (2006).
overview#expanding_coverage (15)]. This lack rate of loss on Earth; however, the magnitude of
of information also accentuates the reluctance the challenge ahead is considerable and should Acknowledgments: We thank the Catalog of Life
to declare species prematurely extinct [due to the not be underestimated, because the unique di- (www.species2000.org), the World Register of Marine
skepticism generated when declared-extinct spe- versity of life on our planet and the services it Species (www.marinespecies.org), the List of Prokaryotic
Names with Standing in Nomenclature (www.bacterio.net),
cies are rediscovered (16)]; thus, for many species provides to humankind are at stake.
the Thomson Reuters Zoological Record (www.organismnames.
it can take years before being declared extinct com), The International Plant Names Index (www.ipni.org),
References and Notes
(15, 16). 1. A. Balmford, R. E. Green, M. Jenkins, Trends Ecol. Evol. and The International Union for Conservation of Nature Red
Unfortunately, Costello’s overarching question— 18, 326–330 (2003). List of Threatened Species (www.iucnredlist.org) for making
can we name Earth’s species before they go 2. R. Dirzo, P. H. Raven, Annu. Rev. Environ. Resour. 28, their data available. We thank M. Costello, N. Stork,
137–167 (2003). K. Gaston, B. Worm, and L. Joppa for comments on the
extinct?—is very sensitive to changes in the es- manuscript.
3. C. Mora, D. P. Tittensor, S. Adl, A. G. B. Simpson,
timates discussed above. For instance, although B. Worm, PLoS Biol. 9, e1001127 (2011).
neither taxonomic effort nor extinction rates will 4. S. Bacher, Trends Ecol. Evol. 27, 65–66, author reply 66 1 March 2013; accepted 14 June 2013
remain constant over time, using the same ap- (2012). 10.1126/science.1237254
237-c 19 JULY 2013 VOL 341 SCIENCE www.sciencemag.org

Response to Comments on ''Can We Name Earth's Species Before
They Go Extinct?''
Mark J. Costello et al.
Science 341, 237 (2013);
DOI: 10.1126/science.1237381


http://www.sciencemag.org/content/341/6143/237.4.full.html
found at:
http://www.sciencemag.org/content/341/6143/237.4.full.html#related
http://www.sciencemag.org/content/341/6143/237.4.full.html#ref-list-1
TECHNICAL COMMENT
from this study related only to the expert opin-
ion, a method they say has “limited empirical
Response to Comments on
basis and considerable subjectivity.” Yet, they over-
looked the more empirical and objective esti-
mates in this study.
“Can We Name Earth’s Species Laurance (2) and Mora et al. (3) suggested
that we have greatly underestimated the numbers
Before They Go Extinct?” of microscopic species. Depending on which

size categories of species are compared, there
can be more species of smaller body size [e.g.,
Mark J. Costello,1* Robert M. May,2 Nigel E. Stork3 (4, 10)]. This is because taxa with typically large
body size have relatively few species (e.g., 3%
Mora et al. disputed that most species will be discovered before they go extinct, but not our species in taxa >10 cm) (Table 1). About 85% of
main recommendations to accelerate species’ discoveries. We show that our conclusions would be species are arthropods, worms, and other macro-
unaltered by discoveries of more microscopic species and reinforce our estimates of species invertebrates. However, microscopic taxa account
description and extinction rates, that taxonomic effort has never been greater, and that there are for just 11% of known species (Table 1). The rea-
2 to 8 million species on Earth. son for this is likely that these microtaxa can be
superabundant; can disperse without any energy

ere, we respond to comments on our dicted by statistical extrapolation of the rate of cost by air, water, and attaching to drifting ma-
H Review that concluded that most species

will be discovered and named before
they go extinct due to human activities, because
species description. However, their experts’ opin-
ions predicted that two-thirds of species remained
to be discovered. Thus, the three methods esti-
terials and other species; and can survive until
they find suitable conditions to grow (8, 11–13).
Thus, they are relatively more widespread than
taxonomic effort has never been greater, conser- mated the number of marine species to be 0.3 to larger taxa. Their wide geographic dispersal mit-
vation efforts are reducing extinction rates, and 1.0 million. The figure of 0.7 to 1.0 million (two- igates against speciation and extinction. There
there are 5 T 3 million species on Earth rather thirds species not named) quoted by Mora et al. (3) is no evidence that rates of species description
than tens of millions (1). No responses questioned
our primary recommendations regarding what
is required to accelerate progress in discovering Table 1. The number of species per higher taxonomic group. Data from http://species.asu.edu/
files/SOS2010.pdf (27). Higher taxa are classified into those generally large (mega, >10 cm), small
life on Earth, but some questioned some of the
(macro, >1 mm), and microscopic (micro, <1 mm) when at reproductive age.
details of our arguments or presented more pes-
simistic views. For example, Laurance (2) and Higher taxonomic group Described living species
Mora et al. (3) concluded that “major uncertain-
Mega
ties still cloud our understanding of the amount
Pisces: fishes 31,658
of extant biodiversity and its fate in the coming
Aves: birds 9,997
centuries” and “multiple sources of uncertainty
Reptilia: snakes, lizards, turtles 8,863
remain in our knowledge of biodiversity and
Amphibia: frogs, toads, salamanders 6,644
its rate of loss on Earth…the magnitude of the
Mammalia 5,528
challenge ahead is considerable and should not
Other chordates 149
be underestimated,” respectively, which reaffirm
Subtotal 62,839
our conclusions. However, our Review had not
Macro
“challenged the…view that many species are dis-
Insecta 1,013,692
appearing before being described” as stated by
Plantae: vascular, including flowering 283,701
Mora et al. (3). Other comments require a more
Arachnida: spiders, mites 103,568
detailed response.
Mollusca: snails, slugs, bivalves, squid, octopus 85,880
Estimates of 3 to 5 million species on Earth
Crustacea: crabs, shrimp, krill 48,014
made over 20 years ago (4–6) are still reasonable
Platyhelminthes: flat worms 20,225
(7). They were ignored by Mora et al. (3) when
Annelida: segmented worms 16,897
they incorrectly stated that our estimate of global
Algae 12,304
species richness of 5 T 3 million species was
Cnidaria: corals, jellyfish 9,926
because we included some recent low estimates
Echinodermata: sea urchins, starfish, sea cucumbers 7,037
(8). Gaston (6) cited estimates of 2 to 3 million
Porifera: sponges 6,057
insects on Earth that tend to have been overlooked
Tunicata: sea squirts 2,808
in favor of the more memorable hyperestimates.
Hemichordata: acorn worms 110
In an assessment of all marine species, Appeltans
Other invertebrates 40,575
et al. (9) found about one-third of species in
Subtotal 1,650,794
samples in less studied areas (n = 100 studies)
Micro
were new to science, a similar proportion pre-
Fungi: yeasts, molds, mushrooms 100,342
1
Leigh Marine Laboratory, Institute of Marine Science, Uni- Protozoa: protists, nucleated algae 29,104
versity of Auckland, Post Office Box 349, Warkworth, New Nematoda: round worms 25,205
Zealand. 2Department of Zoology, University of Oxford, South Prokaryota: bacteria, archaea 8,216
Parks Road, Oxford OX1 3PS, UK. 3Environmental Futures Others, e.g., micro-Chromista, Bryophyta 46,210
Centre, School of Environment, Griffith University, 170 Kessels
Road, QLD 4111, Brisbane, Australia. Subtotal 209,077
Total 1,922,710
*Corresponding author. E-mail: m.costello@auckland.ac.nz
www.sciencemag.org SCIENCE VOL 341 19 JULY 2013 237-d

TECHNICAL COMMENT
are getting relatively higher for these micro- committed to extinction,” but the lag time for changed over time. Until recently, estimates of
taxa, because they are generally proportional this extinction debt may be centuries or millen- global species richness had not considered the
to the number of known species in higher taxa nia, and habitat changes during that period of size of the taxonomic workforce or assumed it
(Fig. 1). Thus, insects make up about half of time can be considerable. We favor a more care- was declining.
all species and continue to provide the most fully deliberated weighting of the evidence and In recent years, species descriptions may
new species. It is true that some of the micro- arguments for the estimates. While agreeing with take more time to prepare because of higher ex-
scopic taxa have high local species richness, us that the current record of species extinction is pectations of journals and the need to account
but this does not necessarily translate to high “very low,” Mora et al. (3) attributed this to the for more literature than before. On the other
global species richness because the species may limited number of species whose extinction risk hand, there are technological efficiencies in com-
be more widespread than sampling indicates, has been assessed. However, the species first as- munication, access to literature, laboratory meth-
although it may have narrow habitat requirements sessed tend to be groups where extinction rates ods, photography, paper preparation, and the
(14, 15). Species’ distributions are underesti- are highest (e.g., amphibians), whereas other taxa publication process (24). One metric of taxo-
mated by limited geographic sampling, which (e.g., fish) have much lower recent extinction nomic productivity is the number of publications
can overestimate habitat and host specificity rates per number of known species. Nevertheless, and new species described. The last decade saw
and endemicity. we emphasize that the global loss of any species a record number of new marine species named
A key finding in our Review (1) was that we is a tragic event and the rapid movement of many (9). The rate of marine species descriptions has
now have better estimates of how many species species up the scale of threat toward extinction, been relatively higher than for terrestrial since
may be on Earth than of extinction rates. We as shown in the International Union for Con- the 1950s, reflecting the new capabilities to ex-
concurred with recent studies that recent extinc- servation of Nature List of Threatened Species plore the ocean. It has generally taken twice as

tion rates have not been as severe as may have (19), is of great concern. many authors to name marine species compared
been expected, in large part because species A series of studies on fossils, regional and with terrestrial species over time (21). However,
survive in secondary habitats, the success of con- global fauna and flora, all marine species, and the greatest period of terrestrial and overall spe-
servation efforts, and a so-far unquantified ex- all algae have shown that several times as many cies discovery was 100 years ago (9), and there
tinction debt. Human-mediated species extinctions people have been describing species as ever be- has been a significant decline in the number of
have been mostly due to hunting and introduced fore (9, 20, 21). Critics may counter that this species described new to science overall per
predators on islands, whereas current rates may reflects a new fashion to put more people’s names author since then (21). This fact has not pre-
be more due to habitat loss and future rates due against a new species (the “et al.” effect) and that viously been considered in estimates of global
to climate change. Laurance (2) agreed with us there is a greater proportion of people describ- species richness and indicates that it has been
that present rates are in the range 0.01 to 1.0% ing new species part-time. However, the “et al.” getting relatively harder to discover new species
per decade. However, Mora et al. (3) proposed effect only appeared in the 1980s when this trend because most have been found (8, 21).
that the average of extinction estimates given in was already under way, and two independent Mora et al. (3) suggested that if 0.5 million
Stork (16) should be used. This is a simplistic analyses showed it did not significantly alter the species remained to be described in collections,
and flawed approach, first because many of trend (8, 22). Moreover, the data show no indi- this indicates that it is not hard to find new spe-
these estimates are based on a range of untested cations of relatively more “part-timers,” whether cies. However, there has always been a backlog
assumptions and also because it assumes that measured as the proportion of authors who de- of species to describe because no institution has
estimates would be normally distributed around scribed one or more species per decade for the experts in all taxa; the specimens may not be of
the true mean. Early estimates of species extinc- past century, the average duration of authors’ sufficient number or quality to confidently describe
tions of 10 to 30% loss of species per decade publication lives, or the skewness of the relative as new species; the specimens may have gone
were based on a flawed species-area model (17) number of species per author over time [reviewed unnoticed due to having been originally mislabeled
and clearly have not eventuated; otherwise, we in (21)]. The number of authors of new species based on the more limited knowledge at the time;
should already have seen the loss of 50% or has increased most in Asia and South America and it can sometimes be easier to collect new
more of all species. As Pereira et al. (18) indi- and represents a younger generation employed species than to describe them. It takes on aver-
cated, some of the models using habitat loss mea- in universities (23). Thus, the relative average age 21 to 25 years from their date of collection
sures were predicting the numbers of “species effort per taxonomist does not appear to have for species to be described (25, 26). Hence, rec-
ommendations for greater online access to in-
formation on what taxa occur in what collections
Number of species described in 2009
10,000 and for collaboration between institutions and

experts to address this backlog (21, 25).
8,000 Considering the size of the workforce and its
current productivity, we may expect that tax-
onomy at a global level is in a healthier state
6,000
than some have believed. The number of new
species described in recent years would be a
4,000 reasonable measure of current effort. This is most
commonly estimated from the Index of Orga-
2,000
nism Names, which encompasses the Zoological
Record and tracks a wide range of literature for
studies that name new species, subspecies, and
0 higher taxa (27). Since the late 1970s, the pro-
0 200,000 400,000 600,000 800,000 1,000,000 cess has been digitized, appears accurate, and has
been used in several studies [e.g., (23, 28–30)].
Number of described species per taxon However, Mora et al. (3) dispute this data, and
Fig. 1. New species named in 2009 versus already named species. The number of species for their analysis could only confirm the validity of
higher taxonomic groups described in a typical year compared with the number already described (r 2 = 45% of the species described since 2000, and
0.99). The trend is similar if the insects (r 2 = 0.85) and plants (r 2 = 0.80) are omitted. [Data from (27)] they alleged that a further 46% were synonyms
237-d 19 JULY 2013 VOL 341 SCIENCE www.sciencemag.org

TECHNICAL COMMENT
5,000,000 4. R. M. May, Science 241, 1441–1449 (1988).
5. R. M. May, R. J. H. Beverton, Philos. Trans. R. Soc.
London B Biol. Sci. 330, 293–304 (1990).
6. K. J. Gaston, Conserv. Biol. 5, 283–296 (1991).
Number of species on Earth 4,000,000 7. N. E. Stork, Biodivers. Conserv. 2, 215–232 (1993).
8. M. J. Costello, S. P. Wilson, B. Houlding, Syst. Biol. 61,
871–883 (2012).
9. W. Appeltans et al., Curr. Biol. 22, 2189–2202 (2012).
3,000,000
10. T. Fenchel, Oikos 68, 375–378 (1993).
11. B. J. Finlay, G. F. Esteban, K. J. Clarke, J. L. Olmo, Protist
152, 355–366 (2001).
2,000,000 12. B. J. Finlay, T. Fenchel, Protist 155, 237–244 (2004).
13. M. J. Costello, S. P. Wilson, Glob. Ecol. Biogeogr. 20,
319–330 (2011).
14. B. J. Finlay, J. A. Thomas, G. C. McGavin, T. Fenchel,
1,000,000 R. T. Clarke, Proc. Biol. Sci. 273, 1935–1941 (2006).
15. N. E. Stork, Nature 448, 657–658 (2007).
16. N. E. Stork, Biodivers. Conserv. 19, 357–371 (2010).
17. F. He, S. P. Hubbell, Nature 473, 368–371 (2011).
0
18. H. M. Pereira et al., Science 330, 1496–1501 (2010).
2000 2050 2100 2150 2200 19. F. D. M. Smith, R. M. May, R. Pellew, T. H. Johnson,
Years K. S. Walter, Nature 364, 494–496 (1993).
20. O. De Clerck, M. D. Guiry, F. Leliaert, Y. Samyn,
Fig. 2. Predicted species extinction and description rates in future years. The number of

H. Verbruggen, J. Phycol. 49, 215–225 (2013).
species described per year (gray line) in Costello et al. (1) and adjusted to 13,500 per decade (black 21. M. J. Costello, S. Wilson, B. Houlding, Syst. Biol. 62,
line) to exclude new fossil species and subspecies. Colored lines show species remaining based on 616–624 (2013).
whether there are 5 or 2 million species on Earth and whether extinction rates are 1% (red line), 0.1% 22. L. N. Joppa, D. L. Roberts, N. Myers, S. L. Pimm,
(blue line) or 0.01% (green line) per decade. Proc. Natl. Acad. Sci. U.S.A. 108, 13171–13176 (2011).
23. E. Tancoigne, C. Bole, A. Sigogneau, A. Dubois,
Front. Zool. 8, 5 (2011).
24. W. N. Eschmeyer, R. Fricke, J. D. Fong, D. Polack,
or invalid names. It is not clear exactly how they new species per year but did not take account of Zootaxa 2525, 19–50 (2010).
derived these figures because they did not cite the the fact that about 19% of these were fossils and 25. D. P. Bebber et al., Proc. Natl. Acad. Sci. U.S.A. 107,
22169–22171 (2010).
data sources as their licenses required (31, 32), 6% were subspecies (27). We do so here, but this 26. B. Fontaine, A. Perrard, P. Bouchet, Curr. Biol. 22,
and these databases will change in quality and has negligible effect on the general trends in R943–R944 (2012).
quantity over time. It would be extraordinary species discovery versus extinction rates (Fig. 2). 27. T. Reuters, Index to Organism Names, World Wide
that about two-thirds of all recently named spe- Of course, some of these species will be found Web electronic resource; accessed at www.
organismnames.com on 7 March 2013.
cies have already been found to be synonyms or to be synonyms, as will future named species. In
28. Q. D. Wheeler, S. Pennak, 2011 State of Observed Species,
otherwise invalid. Their analysis of three data- our Review, we emphasized the importance of International Institute for Species Exploration; retrieved
bases, including the World Register of Marine the need to revise species descriptions and re- from http://species.asu.edu/SOS on 20 February 2012.
Species (WoRMS) (31), indicated that about 8000 examine type specimens to resolve synonyms. 29. V. Lohrmann, K. Vohland, M. Ohl, C. Häuser,
new species were described from 1990 to 2000. The important point is that we proposed prac- Taxonomische Forschung in Deutschland: Eine
Übersichtsstudie (Museum für Naturkunde, Berlin);
This analysis was clearly wrong because WoRMS tical recommendations to accelerate this pace, accessed at www.biodiversity.de/images/stories/
alone contained at least 17,286 accepted species which have not been criticized. Following these Downloads/taxo-studie-2012.pdf on 17 June 2012.
that had been named in those years, and it was recommendations will improve data and under- 30. E. Tancoigne, A. Dubois, Cladistics 2013, 1–4 (2013).
incomplete (9, 33). If this represents about 15% standing of species’ biogeography and how hab- 31. W. Appeltans et al., Eds., World Register of Marine
Species; accessed at www.marinespecies.org on
of all species (the proportion that marine spe- itat change may affect species extinctions, provide 14 May 2013, 2012.
cies are of all species on Earth), then more than more accurate estimates of global species richness 32. F. A. Bisby et al., Eds., Species 2000 & ITIS Catalogue
115,000 species would have been named in that and extinction rates, and enable the discovery of of Life: Annual Checklist (CD-ROM, Species 2000,
decade, which is 14 times the figure calculated most species before they go extinct. Reading, UK, 2011).
33. M. J. Costello et al., PLoS ONE 8, e51629 (2013).
by Mora et al. (3). From 1990 to 2012, the In-
dex to Organism Names reported an average of References and Notes Acknowledgments: We thank R. Pyle, D. Remsen, N. Robinson,
16,500 new species of animals per year, and more 1. M. J. Costello, R. M. May, N. E. Stork, Science 339, É. Tancoigne, and K. Kelly for helpful information and/or
413–416 (2013). discussion.
than 17,000 from 2006 to 2010, to which should 2. W. F. Laurance, Science 339, 1275 (2013).
be added new species of plants and fungi. In our 3. C. Mora, A. Rollo, D. P. Tittensor, Science 340, 237 (2013); 22 May 2013; accepted 14 June 2013
Review, we used a conservative figure of 16,000 www.sciencemag.org/cgi/content/full/341/6143/237-c. 10.1126/science.1237381
www.sciencemag.org SCIENCE VOL 341 19 JULY 2013 237-d

BOOKS ET AL.
ENVIRONMENT people who then applied their knowledge to
new areas and challenges. Absorbing read-
ing, the book also serves up excellent edu-
A Saga of Industrial Pollution cational environmental case studies and sto-
ries. Although not a textbook (it contains no
Jouni Tuomisto equations or chemical formulae), it provides
informative discussions of the
T
oms River offers a fascinating, care- coal tar chemistry, or assem- limitations and opportunities
fully written description of chemi- bling elaborate patient studies Toms River of various methods. These
cal industry malpractices during the to understand the development A Story of Science should help readers under-
past five decades and the subsequent actions of cancer and other diseases. and Salvation stand the capabilities of envi-
of citizens, authorities, companies, employ- He presents dozens of key by Dan Fagin ronmental health.
ees, and lawyers. It helps us understand why researchers from Paracelsus Bantam, New York, 2013. Much of the regretful look-
chemical industry tends to produce health (in the 16th century) to Richard 559 pp. $28, C$34. ing back in the book should
hazards, why that is happening in new areas Doll (in the 20th) along with ISBN 9780553806533. not have been hindsight.
(such as China) even today, and how these their contributions to the under- Fagin convincingly demon-
problems could be reduced. standing and practices of what strates that in very many cases

In 1952, the Swiss company Ciba (later we now call environmental health sciences. enough was known for people to have made
Ciba-Geigy and now Novartis) built a chemi- A similar case of drinking water contami- better decisions. So why did the chemical
cal factory for producing dyes and, later, other nation in Woburn, Massachusetts (1), became companies go on dumping and the authorities
chemicals in sleepy Toms River, New Jersey. the basis of a 1998 feature film. The complex- keep ignoring future trouble? There seem to
The town’s enthusiastic welcome gradually ities Fagin presents suggest that the saga of be two main reasons. Either the existing infor-
turned into complaints and eventually out- Toms River could not be captured in a two- mation did not reach the decision-makers, or
rage. Through illegal dumping of Union Car- hour movie. However, it has the makings of the decision-makers’ outlook prevented them
bide waste, Toms River came to host another a fine television series: numerous interesting from recognizing the hazards and opportu-
hazardous site. The careless disposal of solid characters, the spectacle of a growing indus- nities. In other words, the business-as-usual
and liquid hazardous wastes at the two sites trial town with a complex social ecosystem, approach seems to be such a strong default
polluted the nearby river, soil, groundwater, and a story spanning more than five decades that even clear alternatives tend to go unno-
and ocean and caused suspicion of a child- (with flashbacks into even earlier history of ticed. As the book shows, the approach was
hood cancer cluster. Both sites were included chemistry and science). Indeed, Toms River shared not only by the chemical industry and
in the Superfund program, which identifies synthesizes Fagin’s extensive work. He inter- authorities but also by most researchers and
and attempts to clean up contaminated sites viewed 140 people and mentions more than citizens. Leaps of progress came only after a
in the United States. twice that number. determined few succeeded in shaking the sta-
Describing in depth and with credible Despite the huge amounts of detail and tus quo.
data what happened, the book also recounts sometimes-difficult scientific concepts, the This implies that we should systematically
who made the decisions and the reasons for lucid text remains fairly easy to read. Fagin promote an open flow of environmental infor-
and the outcomes of the actions. Dan Fagin impressively discloses the web of contacts mation, especially among people holding dif-
(who teaches environmental journalism at that stimulated the flow of ideas among ferent views and different interpretations. In
New York University) lays Toms River, fear that open
out the practices through information would cause
the different decades. He public outrage was repeat-
shows that many actions edly expressed. But the
that readers may think of as book shows that when at
outrageously stupid were at some point the informa-
the time widely accepted tion leaked out anyway,
standard practice and were the largest outrage usually
based on rational although stemmed from the secrecy,
short-sighted thinking. not the bad news per se.
Fagin weaves his nar- In the book’s final part,
rative from stories of indi- Fagin describes how law-
viduals passionate about yer Jan Schlichtmann (who
something—whether pro- also was deeply involved in
tecting children against the Woburn case) pushed
cancer, finding methods to strongly for such a shared
produce bright colors using understanding about child-
hood cancer cases in Toms
CREDIT: GOOGLE EARTH
The reviewer is at the Department River. Not only did he suc-

of Environmental Health, National ceed, his open approach
Institute for Health and Welfare was able to create trust and
(THL), Post Office Box 95, FI-70701
Kuopio, Finland. E-mail: jouni. a feeling of understand-
tuomisto@thl.fi Troubled environment. The site of the Ciba chemical plant at Toms River, New Jersey. ing among the opposing

Published by AAAS
BOOKS ET AL.
groups. That promising outcome supports with the fast sector and the slow sector is nec- proton-beam therapy. More things are con-
claims that openness improves outcomes essary (as is the case with health care), then stantly being discovered in health care (which
(2). Therefore, we seem to need more open wages will rise in the slow sector in order to pushes against the pokey view of health care
assessments as early as possible. Once the persuade people to enter it. The innovation), and they have
damage is done, all we can do is estimate its faster the growth of productiv- high prices because of market
The Cost Disease
scope and seek the culprits—as events played ity in the fast lane, the faster power, patents, or research and
Why Computers Get
out in New Jersey. wages will also rise in the slow development costs.
Cheaper and Health
A balanced book, Toms River does not lane. And so the rapid produc- In addition, several com-
Care Doesn’t
push a political agenda, unlike Rachel Car- tivity growth that we have seen mentators have noted that
son’s Silent Spring (3). But both books offer in computers and automobiles by William J. Baumol; the price of health care on a
many valuable lessons to those around the pulls up wages in health care, with contributions by per-unit basis and adjusted
world who wish to improve environmental education, the arts, and garbage David de Ferranti, Monte for quality is actually falling.
Malach, Ariel Pablos-
and occupational health. collection. The central claim Quality in health care is admit-
Méndez, Hilary Tabish,
in The Cost Disease is that the tedly difficult to measure,
References and Lilian Gomory Wu
“crisis” of health care spending Yale University Press, but at least for heart-attack
1. J. Harr, A Civil Action (Vintage, New York, 1996).
2. M. V. Pohjola, J. T. Tuomisto, Environ. Health 10, 58 may not be a crisis at all: inno- New Haven, CT, 2012. treatments, a careful quality-
(2011). vation in some sectors increases 271 pp. $30, £20. adjusted evaluation shows fall-

3. R. Carson, Silent Spring (Houghton Mifflin, Boston, 1962). incomes (or reduces prices), ISBN 9780300179286. ing prices (1). Connecting the
allowing us to spend more on dots to the first point about
10.1126/science.1240379 health care. In Baumol’s view, use, it’s possible for prices to
as long as there is a steady stream of innova- fall but use to increase. In other words, the
tion outside of health care—and he believes price of treating cardiovascular disease may
HEALTH ECONOMICS that there will continue to be—we’ll be able be dropping as bypass is replaced with stents,
to spend even more on health care. and brand-name statins with generics, but we
Can We Afford More Baumol is careful not to argue that the may be still spending more because we are
health care system lacks wasteful spending, diagnosing and treating more people.
Health Care? and he devotes a fair bit of space to noting the Furthermore, the distinction between the
scope of that problem. His “however” is that level of spending and the growth of spending
Amitabh Chandra the cost-disease hypothesis explains the rapid is central for policy debates. There is no deny-
growth of health care spending but doesn’t ing that growth of health care today will influ-
W
illiam Baumol is among the most defend the level of spending (which could be ence its level tomorrow. But if that’s true, the
thoughtful economists of our rife with waste). The distinction between the growth of health care yesterday influenced its
time—over his prolific career he level of spending and the growth of spend- level today. Baumol is open to waste being an
has covered everything from entrepreneur- ing is central. For Baumol, the waste lies explanation for the level of health care spend-
ship to institutions to operations research. In in the level of spending, but the growth of ing—yet as this simple example illustrates,
the 1960s, he along with Princeton colleague spending simply reflects the lack of produc- his forthrightness opens up a Pandora’s box
William Bowen put together a clever thesis to tivity growth in health care. Prices in health of concerns about waste in health care.
explain the increasing share of incomes that care rise in order for it to be produced. The Lastly, regardless of the cause of the rise,
go to relatively unproductive sectors. That same logic can be used to explain why the one thing that everyone agrees on is that
thesis is now referred to as “Baumol’s cost wages of barbers have increased over time, health care spending is increasing rapidly.
disease,” and in The Cost Disease, Baumol even though there has not been innovation in Given government’s role as the largest pur-
applies his theory to the debate on health care barbering (at least not since barbers stopped chaser of health care, this means that taxes
spending. being surgeons and dentists). have to go up. The U.S. Congress has shown
Assume (as we economists are superb at The Cost Disease offers a fresh take on an little appetite for that, and other Organiza-
doing) that there are two sectors in the econ- important phenomenon. It uses tantalizingly tion for Economic Co-operation and Devel-
omy that differ in the speed of innovation; call simple ideas to illustrate the perils of curtail- opment countries can’t raise their taxes any
them fast and slow. In the fast lane, innovation ing the growth of health care spending. While further. Confronted with this reality, it is dif-
is rapid. Wages grow because employees are Cassandras have sounded the alarm over the ficult to ignore the pressures to reduce waste,
more productive as a result of new technolo- rapid growth of health care spending, Bau- increase competition, and adopt high-value
gies that allow them to do more for less. In the mol tells us to keep calm and carry on. Who technologies. In The Cost Disease, Baumol
slow sector, where innovation is jaundiced, should one listen to? cautions us that in the zeal to reduce health
we might expect wages to stagnate and the There are four issues that one should think care spending, we should not forget the cen-
industry to wither and ultimately disappear. about in refereeing this debate. The first is to tral role of innovations outside of the sector.
This would be true if the two sectors compete note that health care spending is the product As those enrich us, we can surely afford more
with each other, as was the case of steamships of health care price and use. Baumol’s study health care.
and sailboats. But the cost-disease hypothesis focuses only on the rapid increase in prices
posits that if the slow sector doesn’t compete and is silent on the quantity of health care References
1. D. M. Cutler, M. McClellan, J. P. Newhouse, D. Remler,
that people receive. But we’re surely getting Q. J. Econ. 113, 991 (1998).
more done to us today—e.g., stents, cardiac
The reviewer is at the Harvard Kennedy School of Govern-
ment, Harvard University, Cambridge, MA 02138, USA. computed tomography, visits to the intensive
E-mail: amitabh.chandra@harvard.edu care unit, new oncology drugs, robots, and 10.1126/science.1232064

Published by AAAS
BOOKS ET AL.
groups. That promising outcome supports with the fast sector and the slow sector is nec- proton-beam therapy. More things are con-
claims that openness improves outcomes essary (as is the case with health care), then stantly being discovered in health care (which
(2). Therefore, we seem to need more open wages will rise in the slow sector in order to pushes against the pokey view of health care
assessments as early as possible. Once the persuade people to enter it. The innovation), and they have
damage is done, all we can do is estimate its faster the growth of productiv- high prices because of market
The Cost Disease
scope and seek the culprits—as events played ity in the fast lane, the faster power, patents, or research and
Why Computers Get
out in New Jersey. wages will also rise in the slow development costs.
Cheaper and Health
A balanced book, Toms River does not lane. And so the rapid produc- In addition, several com-
Care Doesn’t
push a political agenda, unlike Rachel Car- tivity growth that we have seen mentators have noted that
son’s Silent Spring (3). But both books offer in computers and automobiles by William J. Baumol; the price of health care on a
many valuable lessons to those around the pulls up wages in health care, with contributions by per-unit basis and adjusted
world who wish to improve environmental education, the arts, and garbage David de Ferranti, Monte for quality is actually falling.
Malach, Ariel Pablos-
and occupational health. collection. The central claim Quality in health care is admit-
Méndez, Hilary Tabish,
in The Cost Disease is that the tedly difficult to measure,
References and Lilian Gomory Wu
“crisis” of health care spending Yale University Press, but at least for heart-attack
1. J. Harr, A Civil Action (Vintage, New York, 1996).
2. M. V. Pohjola, J. T. Tuomisto, Environ. Health 10, 58 may not be a crisis at all: inno- New Haven, CT, 2012. treatments, a careful quality-
(2011). vation in some sectors increases 271 pp. $30, £20. adjusted evaluation shows fall-

3. R. Carson, Silent Spring (Houghton Mifflin, Boston, 1962). incomes (or reduces prices), ISBN 9780300179286. ing prices (1). Connecting the
allowing us to spend more on dots to the first point about
10.1126/science.1240379 health care. In Baumol’s view, use, it’s possible for prices to
as long as there is a steady stream of innova- fall but use to increase. In other words, the
tion outside of health care—and he believes price of treating cardiovascular disease may
HEALTH ECONOMICS that there will continue to be—we’ll be able be dropping as bypass is replaced with stents,
to spend even more on health care. and brand-name statins with generics, but we
Can We Afford More Baumol is careful not to argue that the may be still spending more because we are
health care system lacks wasteful spending, diagnosing and treating more people.
Health Care? and he devotes a fair bit of space to noting the Furthermore, the distinction between the
scope of that problem. His “however” is that level of spending and the growth of spending
Amitabh Chandra the cost-disease hypothesis explains the rapid is central for policy debates. There is no deny-
growth of health care spending but doesn’t ing that growth of health care today will influ-
W
illiam Baumol is among the most defend the level of spending (which could be ence its level tomorrow. But if that’s true, the
thoughtful economists of our rife with waste). The distinction between the growth of health care yesterday influenced its
time—over his prolific career he level of spending and the growth of spend- level today. Baumol is open to waste being an
has covered everything from entrepreneur- ing is central. For Baumol, the waste lies explanation for the level of health care spend-
ship to institutions to operations research. In in the level of spending, but the growth of ing—yet as this simple example illustrates,
the 1960s, he along with Princeton colleague spending simply reflects the lack of produc- his forthrightness opens up a Pandora’s box
William Bowen put together a clever thesis to tivity growth in health care. Prices in health of concerns about waste in health care.
explain the increasing share of incomes that care rise in order for it to be produced. The Lastly, regardless of the cause of the rise,
go to relatively unproductive sectors. That same logic can be used to explain why the one thing that everyone agrees on is that
thesis is now referred to as “Baumol’s cost wages of barbers have increased over time, health care spending is increasing rapidly.
disease,” and in The Cost Disease, Baumol even though there has not been innovation in Given government’s role as the largest pur-
applies his theory to the debate on health care barbering (at least not since barbers stopped chaser of health care, this means that taxes
spending. being surgeons and dentists). have to go up. The U.S. Congress has shown
Assume (as we economists are superb at The Cost Disease offers a fresh take on an little appetite for that, and other Organiza-
doing) that there are two sectors in the econ- important phenomenon. It uses tantalizingly tion for Economic Co-operation and Devel-
omy that differ in the speed of innovation; call simple ideas to illustrate the perils of curtail- opment countries can’t raise their taxes any
them fast and slow. In the fast lane, innovation ing the growth of health care spending. While further. Confronted with this reality, it is dif-
is rapid. Wages grow because employees are Cassandras have sounded the alarm over the ficult to ignore the pressures to reduce waste,
more productive as a result of new technolo- rapid growth of health care spending, Bau- increase competition, and adopt high-value
gies that allow them to do more for less. In the mol tells us to keep calm and carry on. Who technologies. In The Cost Disease, Baumol
slow sector, where innovation is jaundiced, should one listen to? cautions us that in the zeal to reduce health
we might expect wages to stagnate and the There are four issues that one should think care spending, we should not forget the cen-
industry to wither and ultimately disappear. about in refereeing this debate. The first is to tral role of innovations outside of the sector.
This would be true if the two sectors compete note that health care spending is the product As those enrich us, we can surely afford more
with each other, as was the case of steamships of health care price and use. Baumol’s study health care.
and sailboats. But the cost-disease hypothesis focuses only on the rapid increase in prices
posits that if the slow sector doesn’t compete and is silent on the quantity of health care References
1. D. M. Cutler, M. McClellan, J. P. Newhouse, D. Remler,
that people receive. But we’re surely getting Q. J. Econ. 113, 991 (1998).
more done to us today—e.g., stents, cardiac
The reviewer is at the Harvard Kennedy School of Govern-
ment, Harvard University, Cambridge, MA 02138, USA. computed tomography, visits to the intensive
E-mail: amitabh.chandra@harvard.edu care unit, new oncology drugs, robots, and 10.1126/science.1232064

Published by AAAS
EDUCATIONFORUM
SCIENCE EDUCATION
Crowdsourcing and Curating The diverse wealth of crowdsourced online

tools can benefit from professional
Online Education Resources coordination and quality control.
Darrell Porcello* and Sherry Hsi
T
he Internet is a growing source The National Science Digital
of open educational resources Library (NSDL), initially funded by
(OERs) focused on Science, the U.S. National Science Founda-
Technology, Engineering, and Math tion, set out to establish a common
(STEM). These STEM OERs are set of metadata fields and controlled
not only shared openly and free for vocabulary—a well-defined list of
all to use, but often provide licenses words to choose from when populat-
that permit modification and reuse. ing a metadata field, e.g., “biology” or

Educators must have access to tools “physics” for subject area—for online
that pinpoint valuable resources STEM educational resources by knit-
while avoiding substandard ones. ting together disciplinary and audi-
We discuss how multiple informa- ence-specific communities (3). This
tion sources, user communities, and provided the foundation for a long-
online platforms might be coordi- standing consortium of online plat-
nated to craft effective experiences in forms for STEM OERs [e.g., (4)]. The
digital-rich learning environments. core infrastructure team of the NSDL
There is no shortage of STEM recently updated to the Learning
OER collections. Public education and gov- Although dedicated users from the result- Application Readiness (LAR) metadata for-
ernment agencies, television stations, state- ing online communities might be more will- mat in response to the increasing demand for
level after-school organizations, science ing to detail their instructional experiences STEM OERs to be “aligned to educational
museums, and other organizations produce and personal connections to OERs, overall goals, curriculum, or professional develop-
high-quality educational content for the Web quality may be quite variable. On the other ment needs of users” (5).
and form extensive collections of collabora- hand, having professional staff with dis- Synergistic efforts of the Learning
tive, project-based, and open-ended STEM ciplinary expertise populate, catalog, and Resource Metadata Initiative (LRMI), funded
educational activities. Individual educa- maintain curated collections usually results by the Bill and Melinda Gates Foundation,
tors, scientists, and hobbyists use online in fine-tuned, coherent, and smaller collec- establish a simplified set of standards to tag all
platforms like Instructables, MakeProjects, tions. Although these collections grow more educational online content, making it easier
and Pinterest to self-publish instructions for slowly, often with less vocal user communi- to publish, discover, and deliver quality OERs
creative projects. ties, they can quickly respond to new quality on the Web (6). Careful cataloging through
But as the number of STEM OERs grows or educational standards. metadata fuels advanced search engines and
online, how do educators decide what collec- Over the past decade, online platforms for more effective discovery of STEM OERs by
tions to use when searching for digital con- STEM OERs have vacillated between these educators. Metadata can also provide infor-
tent? In a large kindergarten-to–high school two poles. New efforts are combining the best mation about how a resource aligns to cur-
(K–12) annual U.S. national survey, 41% of aspects of both strategies to create sustainable ricular standards for STEM OERs.
principals responded that it was “difficult to online platforms to make it easier for educa- Balancing Expert and Community Defi-
evaluate quality of digital content” (1), but tors to discover and use high-quality STEM nitions of Quality. Learning resource and
>50% of teachers responded that the most OERs, confident in their scientific accuracy metadata quality are required for STEM
important factors in evaluating content were and pedagogical approach. OER online platforms to be accepted by par-
“being referred by a colleague,” “free,” and ents, educators, and administrators, but who CREDIT: HAYES THRONTON/LAWRENCE HALL OF SCIENCE
“created by educators” (2), none of which is Four Components Essential to OER Success decides what is quality information? In the
necessarily a hallmark of quality. Convergence Toward Common Metadata. A Wikipedia model, the written articles are
This highlights a difference within the standard set of terms, or metadata, used by largely crowdsourced, although core com-
online world of OERs. On one hand, col- a dedicated community of users to tag digi- munity members monitor and control the
lections using crowdsourcing allow a wide tal resources is a necessary characteristic of quality of information. As one of the most-
range of online users to contribute, choos- STEM OER collections. Grade level, subject visited sites on the Internet, with more than
ing their own descriptions and keywords area, cost per group, and resource type are just 23 million articles, the open digital encyclo-
to catalog, review, and manage OERs. This a few of the metadata fields relevant to edu- pedia has almost 80,000 volunteer editors
can produce large and loose collections. cators. For example, metadata could allow an who regularly contribute and edit content (7).
educator using Howtosmile.org, a free digital Online STEM OER communities do not have
Lawrence Hall of Science, University of California, Berkeley, library of almost 3500 STEM OERs for out- this critical mass to use crowdsourcing as the
Berkeley, CA 94720USA.
of-school educators, to identify 82 hands-on sole source of quality control. It can play a
*Author for correspondence: porcello@berkeley.edu activities on evolution for 8- to 11-year-olds. part, but a professional or trained volunteer

Published by AAAS
EDUCATIONFORUM
staff is required to curate a high-quality col- orative, is introducing an easy to use LRMI- metadata harvesting, resource vitality, and
lection of STEM OERs. based tagging tool for educators to catalog analytics protocols. Following good Linked
In addition, as standards alignment and online resources, whereas a forthcoming Open Data (19) practices will better intercon-
student achievement metrics become more student-centered platform called Sparticl nect STEM OERs and enable greater flexibil-
critical parameters for teachers and adminis- from Twin Cities Public Television (14) uses ity for application development based on data
trators choosing STEM OERs, professional “folksonomies”—tags offered by commu- from online platforms.
services are required for increased reliabil- nity folks—and user ranking to better deliver STEM OERs are becoming important
ity and accuracy in metadata creation. These STEM OERs to teens. No matter how it is parts of teaching experiences inside and out-
points are demonstrated in NASAWave- eventually used, genuine community input side the classroom. As search and discovery
length.org, a new STEM OER collection for an online platform can only begin after for STEM OERs become more seamless and
from NASA where resources are reviewed by there is first a critical mass of cataloged, high- natural to the online workflow of the educa-
educators using a defined set of criteria (8), quality STEM OERs established. tor and student, the four critical components
aligned to the AAAS Project 2061 Bench- Interoperability. When an educator is described above will drive advanced architec-
marks for Science Literacy (9) when appro- searching for a resource, like an interactive ture to streamline delivery within browsers,
priate, and offered through a powerful search animation illustrating the steps of mitosis, mobile devices, smart boards, and future edu-
engine and a visual browse organized by rel- it would be more efficient if a single query cational technologies. A distributed, yet fully
evant educational concepts. The peer educa- could search across multiple online collec- interconnected, online landscape of STEM

tor and subject-area expert review process tions, rather than having the educator visit OERs will improve quality and the user expe-
for the CPALMS online platform (10), aimed multiple Web sites. Interoperability enables rience and will help avoid inadvertently rec-
at Florida K–12 teachers, can only accept information flow across multiple online plat- reating resources that already exist during
STEM OERs based on rigorous and stan- forms to ensure broad-based access to STEM times of dwindling budgets for new mate-
dards-driven instructional materials. OERs. Imagine searching for STEM OERs rials. When creating or disseminating new
As digital libraries evolve to identify similar to using a single site to search for tick- STEM OERs, a sensible strategy would be to
effective teaching strategies and just-in-time ets from multiple airlines. leverage existing projects to deliver OERs in
information for educators, a new specialized One necessary step toward interoperabil- a responsible and efficient manner online.
profession is emerging for a dedicated cyber- ity is the adoption of an openly licensed, stan-
librarian. This person would review incom- dardized metadata format for STEM OERs. References and Notes
ing resources against a set of quality criteria Pushing beyond a basic set of library record 1. Project Tomorrow, 2011 Speak-Up Report (Project Tomor-
row, Irvine, CA, 2012); www.tomorrow.org/speakup/
to ensure only the best materials with strong fields from the long-standing Dublin Core zSU11_PersonalizedClassroom_EducatorsReport.html.
STEM content, pedagogy, and standards metadata standard, this new, expanded meta- 2. 2010 Speak-Up Report (2011); www.tomorrow.org/
alignments are included in STEM OER col- data standard, with an updated controlled speakup/pdfs/SU10_3EofEducation_Educators.pdf.
3. L. L. Zia, D-Lib 6, (2000). www.dlib.org/dlib/october00/
lections. The “cybrarian” also can serve as vocabulary, can address specific character- zia/10zia.html10.1045/october2000-zia.
the key decision-maker in weighing adminis- istics of STEM OERs and their users. Addi- 4. Including comPADRE (Physics), ChemEd DL (Chemistry),
trator goals, adding regional-specific content, tions might include reading level, learning CLEAN (Climate and Energy), AMSER (Technical and
and determining use with traditionally under- style preferences, assessment data, material Community Colleges), Engineering Pathway, Teachers’
Domain (Media), MSP2 (Middle School Teachers), CPALMS
represented populations in STEM or for stu- costs, standards alignment, or user acces- (Florida K-12 Teachers), DLESE (Geosciences), and
dents with disabilities. sibility. The effort toward common meta- Howtosmile.org.
Community Input. User voices can shape data has led to several aggregator Web sites 5. NSDL LAR, https://wiki.ucar.edu/display/nsdldocs/
LAR+Concepts.
the public face and enhance the usability of that search over large sets of STEM OERs 6. LRMI, www.lrmi.net/about.
online platforms for STEM OERs. Although including NSDL.org, Informalscience.org 7. Wikimedia Foundation, 2011–2012 Annual Report;
users can author and submit new content to from the Center for Advancement of Infor- http://wikimediafoundation.org/wiki/FAQ/en.
8. NASA Wavelength.org Review Criteria; http://nasareviews.
grow a collection, the most practical com- mal Science Education (CAISE) (15), the strategies.org/reviewcriteria.html.
munity input lies within “digital footprints,” cross-U.S. federal agency Learning Registry 9. Project 2016, www.project2061.org/publications/bsl/.
e.g., page views, comments, ratings, or newly (16), and the EU Open Science Resources 10. CPALMS review process: www.cpalms.org/cpalms/review.
formed subcollections, left by users’ online Project (17). aspx.
11. Howtosmile.org lists: http://howtosmile.org/userlists/
interactions and behaviors across an online Although this work represents prog- recent.
platform. Users often leave footprints dur- ress, an agreement on a STEM OER meta- 12. Gooru, http://about.goorulearning.org/.
ing activities they do regardless of their ben- data standard has not been reached within 13. InBloom, https://www.inbloom.org/.
14. Sparticl from Twin Cities Public Television,
efit to the community. For example, users on the education community. But interopera- www.sparticl.org/about/.
Howtosmile.org are >10 times more likely bility does not stop at metadata. NSDL has 15. CAISE’s Informalscience.org project,
to create and save a subcollection (11) of begun to promote a standard set of commu- http://informalscience.org/perspectives/blog/
welcome-to-informalscienceorg.
STEM hands-on activities than leave a com- nity input called “paradata” (18) as another 16. The Learning Registry, http://learningregistry.org.
ment or rating for an individual resource. A way to share contextualized usage data about 17. Open Science Resources (OSR) Project,
user’s subcollection might inform other users a resource and other relevant information www.openscienceresources.eu.
exploring a set of STEM OERs. Web sites across multiple online platforms. The hope 18. NSDL, Paradata; https://wiki.ucar.edu/display/nsdldocs/
Paradata.
like Gooru (12) are combining STEM OERs is that STEM OERs listed on multiple plat- 19. Linked Open Data—What is it? A video from Europeana;
into lesson plans to attract educators. forms will retain usage information to estab- http://vimeo.com/36752317.
Another strategy for community input is lish a common set of digital footprints to be
Acknowledgments: Authors are affiliated with Howtosmile.org,
to have users participate in cataloging duties viewable at all points of access. Interoperabil- NASAWavelength.org, and Informalscience.org.
or earn rewards for ranking content. InBloom ity must stretch to the technical back-end sys-
(13), formally the Shared Learning Collab- tems of online platforms to include standard 10.1126/science.1234722

Published by AAAS
Trans-HSF1 Express
Valentina Gandin and Ivan Topisirovic
Science 341, 242 (2013);
DOI: 10.1126/science.1242359


http://www.sciencemag.org/content/341/6143/242.full.html
found at:
http://www.sciencemag.org/content/341/6143/242.full.html#related
http://www.sciencemag.org/content/341/6143/242.full.html#ref-list-1
PERSPECTIVES
CELL BIOLOGY
Ribosome activity is linked to the activity of

Trans-HSF1 Express a transcription factor in cancer cells.
Valentina Gandin and Ivan Topisirovic
P
rotein synthesis is ele- genes, only rohinitib selec-
vated in many can- tively attenuated proliferation
cers (1). To cope with of malignant cells. Rocaglates
increased protein load and pro- prevent the initiation of transla-
tein folding defects caused by tion by interfering with eIF4A,
genetic abnormalities, malig- an RNA helicase component
nant cells bolster their chap- of the eIF4F translation initia-
erone system . Heat-shock tion complex (6). The eIF4A
transcription factor 1 (HSF1) helicase unwinds second-
is a major activator of chaper- ary structure present in the 5′

one-encoding genes. In can- untranslated region (5′UTR) of
cer cells, HSF1 also affects mRNA so that a ribosome can
the transcription of genes that scan toward the codon that ini-
are implicated in tumorigen- tiates translation. It is thought
esis (2), thereby shaping the that eIF4A inhibitors [silves-
“malignant transcriptome.” On trol (a rocaglate), pateamine
page 250 of this issue, Santag- A, and hippuristanol] exhibit
ata et al. (3) suggest that HSF1 robust antineoplastic effects
acts as a central node of a reg- in vitro and in vivo in mul-
ulatory network that “senses” tiple cancer models by selec-
messenger RNA (mRNA) tively blocking the translation
translation rates and enables Coupled programs. In cancer cells, mRNA translation is increased, boosting the overall of oncogenic mRNAs that bear
malignant cells to respond protein load (stress) and generating a repertoire of proteins that promote tumorigenesis. long, highly structured 5′UTRs
to increased protein loads by In the model shown, the transcription factor HSF1 “senses” this increase in mRNA transla- that depend on the unwinding
reprogramming transcription. tion and modifies transcription to increase chaperone capacity and ribosome biogenesis, activity of eIF4A (6, 7). Santa-
Santagata et al. deployed thereby bolstering translation. The hyperactivation of this translation-transcription net- gata et al. show that in addition
a series of unbiased genome- work, centered on HSF1, leads to metabolic reprogramming and drives neoplastic growth. to restraining production of
wide approaches to demon- pro-oncogenic proteins at the
strate that suppression of global mRNA trans- of UPR genes (such as activating transcription level of translation, rohinitib may also sup-
lation abrogates HSF1 binding to target genes. factor 4) (4). Like the HSF1-ribosome path- press neoplastic growth by inactivating the
The authors further determined that a class of way, UPR plays a central role in tumorigen- HSF1 transcription network. It remains to be
compounds (rocaglates) that blocks transla- esis, wherein it is thought to promote cancer determined whether rohinitib induces selec-
tion also inhibits the transcriptional activ- cell survival under various types of stress and tive changes in the profile of mRNAs that are
ity of HSF1. Rohinitib, a rocaglate with the therefore to contribute to tumor progression translated, particularly those encoding pro-
highest potency of HSF1 inhibition, abolished and the development of chemoresistance (5). teins that favor tumorigenesis.
glucose uptake and glycolysis in cancer cells The findings of Santagata et al. point to a part- Cancer cells increase aerobic glycolysis
and blocked proliferation of premalignant and nership between translational and transcrip- (the “Warburg effect”) to meet high bioener-
malignant cells, while only marginally affect- tional programs that plays an important role getic and biosynthetic demands of neoplas-
ing proliferation of normal cells. not only in stress response but also in cancer tic growth. Translation of mRNA is thought
Cross talk between translational and tran- initiation and progression (see the figure). The to be the most energy-consuming process in
scriptional machineries in this HSF1-ribosome precise mechanism that underlies modulation the cell (8). Accordingly, cancer cells adapt
pathway bears resemblance to the unfolded of HSF1 transcription activity in response to to nutrient deprivation by decreasing energy
protein response (UPR) that is triggered by changes in mRNA translation remains to be consumption. This is accomplished in part by
the accumulation of unfolded and misfolded deciphered. Analogous to UPR, it is plausi- reducing the rate of translation elongation, a
proteins in the endoplasmic reticulum (4). The ble that changes in translation of a subset of process that is controlled by the adenosine
UPR shuts down global protein synthesis by mRNAs affect HSF1 activity. Alternatively, monophosphate–activated protein kinase–
inactivating eukaryotic translation initiation inhibition of translation and subsequent disso- eukaryotic elongation factor 2 kinase pathway
factor 2 (eIF2) while stimulating translation ciation of chaperones from ribosomes should (9). Santagata et al. demonstrate that rohinitib
CREDIT: C. BICKEL/SCIENCE
of a specific subset of stress-related mRNAs be expected to increase the binding of chaper- reduces aerobic glycolysis in a panel of can-
that encode major transcriptional regulators ones to HSF1 and its inactivation. cer cell lines, which correlates with changes
Intriguingly, whereas inhibitors of the in the amounts of mRNAs that encode fac-
Lady Davis Institute for Medical Research, Sir Mortimer B. elongation step (cycloheximide) and the ini- tors that function in glucose uptake (thiore-
Davis–Jewish General Hospital, and Department of Oncol-
ogy, McGill University, Montréal, Quebec, H3A 1A3, Canada. tiation step (rohinitib) of mRNA translation doxin interacting protein) or glucose metabo-
E-mail: ivan.topisirovic@mcgill.ca both decreased expression of HSF1 target lism (phosphoglycerate kinase 1). Although it

Published by AAAS
PERSPECTIVES
remains to be resolved whether HSF1 directly decrease in mRNA translation profoundly References
1. D. Silvera et al., Nat. Rev. Cancer 10, 254 (2010).
controls expression of genes implicated in alters transcriptional programs by altering the 2. M. L. Mendillo et al., Cell 150, 549 (2012).
energy metabolism in response to rohinitib, activity of HSF1. In turn, HSF1 activates the 3. S. Santagata et al., Science 341, 1238303 (2013).
these findings imply that the HSF1 regulatory transcription of ribosomal genes (2), whereas DOI: 10.1126/science.1238303
network that couples translational and tran- ribosome-associated chaperones stimulate 4. G. D. Pavitt, D. Ron, Cold Spring Harb. Perspect. Biol. 4,
a012278 (2012).
scriptional programs contributes to metabolic translation elongation by promoting the tran- 5. Y. Ma, L. M. Hendershot, Nat. Rev. Cancer 4, 966 (2004).
reprogramming of malignant cells. sition of nascent mRNA chains through the 6. A. Malina et al., Cold Spring Harb. Perspect. Biol. 4,
The steady-state compositions of the ribosome exit tunnel (11, 12). a012377 (2012).
eukaryotic transcriptomes only loosely corre- The plasticity of malignant proteomes 7. Y. V. Svitkin et al., RNA 7, 382 (2001).
8. D. F. Rolfe, G. C. Brown, Physiol. Rev. 77, 731 (1997).
spond to the composition of their proteomes, that enables cancer cells to adapt to various
9. G. Leprivier et al., Cell 153, 1064 (2013).
indicating that posttranscriptional regulatory types of stress appears to be achieved by feed- 10. J. D. Keene, Nat. Rev. Genet. 8, 533 (2007).
mechanisms collaborate with the transcrip- back loops that coordinate transcription and 11. R. Shalgi et al., Mol. Cell 49, 439 (2013).
tional networks to modulate cellular activities, translation. Networks that orchestrate these 12. B. Liu, Y. Han, S. B. Qian, Mol. Cell 49, 453 (2013).
including cell proliferation and survival (10). responses could potentially be exploited to
Santagata et al. show that in cancer cells, a devise new and better cancer treatments. 10.1126/science.1242359

MATERIALS SCIENCE
Droplets Out of Equilibrium Magnetic droplets can be switched between

static and dynamic structures.
Thomas M. Hermans1, Holger Frauenrath2, Francesco Stellacci2
L
iving systems create structures and their thermodynamic equilibrium but are kept into entropy and vice versa (6, 7). The prime
functions of remarkable complexity by in a local state by deep potential or kinetic example is the cytoskeleton, where nucleoside
mastering self-assembly in different barriers. Examples include martensitic steel, triphosphates are hydrolyzed to their diphos-
equilibrium and nonequilibrium states. Three most folded proteins (5), and cell membranes. phate analogs, providing chemical energy
states can be distinguished: equilibrium, non- Biological systems control the formation of for forming long fibrous assemblies that give
dissipative nonequilibrium (or kinetically these structures with sophisticated processing structural support to the cell membrane. At
trapped), and dissipative (or dynamic) non- techniques, such as directed crystallization or first glance, the continuous consumption of
equilibrium. On page 253 of this issue, Tim- the use of chaperones in protein folding (2). energy seems wasteful, but it enables the nat-
onen et al. (1) report on a model system in However, life is enabled by dissipative (or ural systems to repair or rapidly adjust their
which all three states are accessible (see the dynamic) assemblies that do not fall in either shape and functions.
figure) and show how this leads to a range of of these two categories. Prigogine’s seminal In synthetic systems, a lot of progress has
well-ordered structures. works on complexity showed that there are been made in developing equilibrium struc-
Most inorganic structures such as gems steady thermodynamic states that, in order to tures (8–10), but dissipative structures are still
and minerals are equilibrium structures, exist, must either consume energy or continu- very scarce (11). Timonen et al. now report
which generally form spontaneously in such ously and quasi-reversibly exchange energy that magnetic droplets spontaneously split
a way that the free energy is and assemble into well-defined
minimized. Some organic geometric patterns under equi-
structures also form in this B librium conditions. These pat-
way, such as some folded pro- terns can be altered by supply-
teins (2) or structure materials ing them with energy, leading
like collagen. Most supramo- to dissipative nonequilibrium
C
lecular materials, which aim to structures, and all equilibrium
mimic the hierarchical struc- and dissipative structures can
A
tures of biomaterials, are equi- be kinetically trapped.
∆G
librium structures (3, 4). Dissipative state The authors show that a
The next level of complex- Trapped state single droplet of ferrofluid on
ity is found in structures that a superhydrophobic surface
reside in kinetically trapped or divides into multiple drop-
metastable states. These struc- lets in a way that depends
tures are typically far from Equilibrium state on the strength and gradient
of the magnetic field of the
1
Institut de Science et d’Ingénierie
magnet placed below the sur-
Supramoléculaires, Université de Three self-assembly states. Timonen et al. started with a single droplet that can be face. Each little droplet can
Strasbourg, 67083 Strasbourg Cedex,
described in terms of equilibrium thermodynamics (A). When a magnetic field is applied, be considered as an individ-
France. 2Institute of Materials, Ecole
Polytechnique Fédérale de Lausanne, the system forms an ordered arrangement of multiple droplets, an example of a kinetically ual dipole being attracted to
1015 Lausanne, Switzerland. E-mail: trapped state (B). When an oscillating field is applied to the kinetically trapped state, put- the center of the magnet. In
francesco.stellacci@epfl.ch ting energy on the system, a dissipative assembly state develops (C). G, Gibbs free energy. addition, droplets that come

Published by AAAS
Droplets Out of Equilibrium
Thomas M. Hermans et al.
Science 341, 243 (2013);
DOI: 10.1126/science.1241793


found at:
PERSPECTIVES
remains to be resolved whether HSF1 directly decrease in mRNA translation profoundly References
1. D. Silvera et al., Nat. Rev. Cancer 10, 254 (2010).
controls expression of genes implicated in alters transcriptional programs by altering the 2. M. L. Mendillo et al., Cell 150, 549 (2012).
energy metabolism in response to rohinitib, activity of HSF1. In turn, HSF1 activates the 3. S. Santagata et al., Science 341, 1238303 (2013).
these findings imply that the HSF1 regulatory transcription of ribosomal genes (2), whereas DOI: 10.1126/science.1238303
network that couples translational and tran- ribosome-associated chaperones stimulate 4. G. D. Pavitt, D. Ron, Cold Spring Harb. Perspect. Biol. 4,
a012278 (2012).
scriptional programs contributes to metabolic translation elongation by promoting the tran- 5. Y. Ma, L. M. Hendershot, Nat. Rev. Cancer 4, 966 (2004).
reprogramming of malignant cells. sition of nascent mRNA chains through the 6. A. Malina et al., Cold Spring Harb. Perspect. Biol. 4,
The steady-state compositions of the ribosome exit tunnel (11, 12). a012377 (2012).
eukaryotic transcriptomes only loosely corre- The plasticity of malignant proteomes 7. Y. V. Svitkin et al., RNA 7, 382 (2001).
8. D. F. Rolfe, G. C. Brown, Physiol. Rev. 77, 731 (1997).
spond to the composition of their proteomes, that enables cancer cells to adapt to various
9. G. Leprivier et al., Cell 153, 1064 (2013).
indicating that posttranscriptional regulatory types of stress appears to be achieved by feed- 10. J. D. Keene, Nat. Rev. Genet. 8, 533 (2007).
mechanisms collaborate with the transcrip- back loops that coordinate transcription and 11. R. Shalgi et al., Mol. Cell 49, 439 (2013).
tional networks to modulate cellular activities, translation. Networks that orchestrate these 12. B. Liu, Y. Han, S. B. Qian, Mol. Cell 49, 453 (2013).
including cell proliferation and survival (10). responses could potentially be exploited to
Santagata et al. show that in cancer cells, a devise new and better cancer treatments. 10.1126/science.1242359

MATERIALS SCIENCE
Droplets Out of Equilibrium Magnetic droplets can be switched between

static and dynamic structures.
Thomas M. Hermans1, Holger Frauenrath2, Francesco Stellacci2
L
iving systems create structures and their thermodynamic equilibrium but are kept into entropy and vice versa (6, 7). The prime
functions of remarkable complexity by in a local state by deep potential or kinetic example is the cytoskeleton, where nucleoside
mastering self-assembly in different barriers. Examples include martensitic steel, triphosphates are hydrolyzed to their diphos-
equilibrium and nonequilibrium states. Three most folded proteins (5), and cell membranes. phate analogs, providing chemical energy
states can be distinguished: equilibrium, non- Biological systems control the formation of for forming long fibrous assemblies that give
dissipative nonequilibrium (or kinetically these structures with sophisticated processing structural support to the cell membrane. At
trapped), and dissipative (or dynamic) non- techniques, such as directed crystallization or first glance, the continuous consumption of
equilibrium. On page 253 of this issue, Tim- the use of chaperones in protein folding (2). energy seems wasteful, but it enables the nat-
onen et al. (1) report on a model system in However, life is enabled by dissipative (or ural systems to repair or rapidly adjust their
which all three states are accessible (see the dynamic) assemblies that do not fall in either shape and functions.
figure) and show how this leads to a range of of these two categories. Prigogine’s seminal In synthetic systems, a lot of progress has
well-ordered structures. works on complexity showed that there are been made in developing equilibrium struc-
Most inorganic structures such as gems steady thermodynamic states that, in order to tures (8–10), but dissipative structures are still
and minerals are equilibrium structures, exist, must either consume energy or continu- very scarce (11). Timonen et al. now report
which generally form spontaneously in such ously and quasi-reversibly exchange energy that magnetic droplets spontaneously split
a way that the free energy is and assemble into well-defined
minimized. Some organic geometric patterns under equi-
structures also form in this B librium conditions. These pat-
way, such as some folded pro- terns can be altered by supply-
teins (2) or structure materials ing them with energy, leading
like collagen. Most supramo- to dissipative nonequilibrium
C
lecular materials, which aim to structures, and all equilibrium
mimic the hierarchical struc- and dissipative structures can
A
tures of biomaterials, are equi- be kinetically trapped.
∆G
librium structures (3, 4). Dissipative state The authors show that a
The next level of complex- Trapped state single droplet of ferrofluid on
ity is found in structures that a superhydrophobic surface
reside in kinetically trapped or divides into multiple drop-
metastable states. These struc- lets in a way that depends
tures are typically far from Equilibrium state on the strength and gradient
of the magnetic field of the
1
Institut de Science et d’Ingénierie
magnet placed below the sur-
Supramoléculaires, Université de Three self-assembly states. Timonen et al. started with a single droplet that can be face. Each little droplet can
Strasbourg, 67083 Strasbourg Cedex,
described in terms of equilibrium thermodynamics (A). When a magnetic field is applied, be considered as an individ-
France. 2Institute of Materials, Ecole
Polytechnique Fédérale de Lausanne, the system forms an ordered arrangement of multiple droplets, an example of a kinetically ual dipole being attracted to
1015 Lausanne, Switzerland. E-mail: trapped state (B). When an oscillating field is applied to the kinetically trapped state, put- the center of the magnet. In
francesco.stellacci@epfl.ch ting energy on the system, a dissipative assembly state develops (C). G, Gibbs free energy. addition, droplets that come

Published by AAAS
PERSPECTIVES
sufficiently close will repel each other. The (1) start out with an ordered equilibrium struc- Timonen et al.’s report shows that liquid
balance between global attraction and local ture, which gradually decreases its order with droplets on nonwetting surfaces provide a rich
repulsion leads to the observed, highly sym- increasing distance from equilibrium. Here, platform for studying dissipative structures.
metric equilibrium droplet patterns. When energy dissipation decreases the symmetry of The gradual transition of equilibrium to dis-
the magnetic field is removed, the patterns the system. sipative structures reported by Timonen et
remain constant, showing that the system is The main challenge in studying dissipa- al. could potentially be applied to molecular
in a kinetically trapped state. tive structures is to identify the exact form self-assembling structures, which are usually
The equilibrium droplet patterns can be and magnitude of dissipation. In Timonen well-ordered in their equilibrium state. Push-
altered reproducibly by feeding energy into et al.’s system, this could be magnetoelastic ing supramolecular structures out of equilib-
the system. A periodically moving magnetic losses, viscous dissipation, or losses due to rium might just be the way to create emergent
field gives rise to energy dissipation that leads eddy-currents. Grzybowski and co-workers structures with functions as complex as those
to much more complex assembly geometries. (12) have shown that viscous dissipation can of living systems.
For slow oscillations, the equilibrium drop- be computed numerically for nonequilib- References
let pattern simply translocates as a whole, but rium structures of magnetic spinners assem- 1. J. V. I. Timonen et al., Science 341, 253 (2013).
less ordered dissipative patterns emerge for bling via vortex-vortex interactions, but their 2. D. Baker, D. A. Agard, Biochem. 33, 7505 (1994).
3. M. Muthukumar et al., Science 277, 1225 (1997).
fast oscillations, probably once the dissipation system does not allow for a simple experi- 4. L. Brunsveld et al., Chem. Rev. 101, 4071 (2001).
energy flux has passed a certain threshold. mental verification of the theory. For mag- 5. C. M. Dobson, Nature 426, 884 (2003).

In classical dissipative structures, such as netic nonequilibrium structures like those 6. G. Nicolis, I. Prigogine, Self-Organization in Nonequilib-
rium Systems (Wiley, New York, 1977).
reaction-diffusion patterns, Rayleigh-Bénard presented by Timonen et al., it could be ben- 7. G. Nicolis, I. Prigogine, G. Nocolis, Exploring Complexity
convection cells, or schools of fish, there is eficial to use an electromagnet to drive the (Freeman, New York, 1989).
no order in or very close to the thermody- system. In analogy to a calorimeter, where 8. J.-M. Lehn, Supramolecular Chemistry: Concepts and
Perspectives (VCH, Weinheim, 1995).
namic equilibrium. Only at a certain distance current is used to keep the temperature con- 9. G. M. Whitesides, B. Grzybowski, Science 295, 2418
from the equilibrium state, beyond a criti- stant, the current in the electromagnet could (2002).
cal point, do highly ordered configurations be measured to keep the magnetic field con- 10. S. I. Stupp et al., Science 276, 384 (1997).
11. B. A. Grzybowski et al., Soft Matter 5, 1110 (2009).
emerge (6). Moreover, order in such systems stant. The current needed to create dissipa- 12. K. V. Tretiakov et al., Soft Matter 5, 1279 (2009).
arises abruptly in a manner resembling a sharp tive structures would provide further insights
phase transition. By contrast, Timonen et al. into the rates of dissipation. 10.1126/science.1241793
CHEMISTRY
One hundred years ago, Niels Bohr’s pioneering

100 Years of Atomic Theory paper on the electronic structure of the
hydrogen atom revolutionized atomic theory.
David C. Clary
I
n 1913, Niels Bohr published a ground- absorption of radiation. In addi-
breaking paper that introduced a new tion, Bohr used Planck’s constant h
way of understanding atomic phenomena to identify the frequency ν of radia-
(1). Entitled “On the Constitution of Atoms tion as
and Molecules,” the article brought together
for the first time the model of the atom devel- ν = (W2 – W1)/h (1)
oped by Rutherford, which consisted of a
positively charged nucleus surrounded by for a transition from an initial
negatively charged electrons, with the theory state of energy W1 to a final state
for quantization of radiation developed by with energy W2. These principles
Planck. The paper became one of the most are second nature to scientists in
influential of the 20th century. 2013, but were radical a hundred
Although the model for the hydrogen years ago.
atom described in the paper was superseded At the time of Bohr’s paper, it was known stants as R = 2π2me4/h3, where m and e are
within 13 years by developments in quantum through the work of Rydberg, following on the mass and charge of the electron, respec-
CREDIT: NIELS BOHR ARCHIVE/COPENHAGEN
theory and wave mechanics, the work intro- from Balmer, that the frequency of a line in tively. It was this finding in particular that
duced several new concepts that have stood the visible spectrum of the hydrogen atom caught the attention of the notables of the
the test of time. These include the existence was given by day, with Einstein describing it as “very
of stationary states, in which an atomic or remarkable” and Bertrand Russell stating
molecular system can have a precise energy ν = R (1/n12 – 1/n22) (2) this was “perhaps the most sensational evi-
and the transition from one state to another dence in favor of Bohr’s theory” (2).
can be accompanied by the emission or where R is the empirically determined Ryd- To justify the introduction of quantum
berg constant and n1 and n2 are integers. The numbers in his theory, Bohr assumed that
Department of Physical and Theoretical Chemistry, Univer-
sity of Oxford, Oxford OX1 3QZ, UK. E-mail: david.clary@ special feature of Bohr’s theory was that it the angular momentum of the electron in
chem.ox.ac.uk gave R purely in terms of fundamental con- the hydrogen atom was quantized and that

Published by AAAS
100 Years of Atomic Theory
David C. Clary
Science 341, 244 (2013);
DOI: 10.1126/science.1240200


Supporting Online Material can be found at:
http://www.sciencemag.org/content/suppl/2013/06/20/science.1240200.DC1.html
found at:
PERSPECTIVES
sufficiently close will repel each other. The (1) start out with an ordered equilibrium struc- Timonen et al.’s report shows that liquid
balance between global attraction and local ture, which gradually decreases its order with droplets on nonwetting surfaces provide a rich
repulsion leads to the observed, highly sym- increasing distance from equilibrium. Here, platform for studying dissipative structures.
metric equilibrium droplet patterns. When energy dissipation decreases the symmetry of The gradual transition of equilibrium to dis-
the magnetic field is removed, the patterns the system. sipative structures reported by Timonen et
remain constant, showing that the system is The main challenge in studying dissipa- al. could potentially be applied to molecular
in a kinetically trapped state. tive structures is to identify the exact form self-assembling structures, which are usually
The equilibrium droplet patterns can be and magnitude of dissipation. In Timonen well-ordered in their equilibrium state. Push-
altered reproducibly by feeding energy into et al.’s system, this could be magnetoelastic ing supramolecular structures out of equilib-
the system. A periodically moving magnetic losses, viscous dissipation, or losses due to rium might just be the way to create emergent
field gives rise to energy dissipation that leads eddy-currents. Grzybowski and co-workers structures with functions as complex as those
to much more complex assembly geometries. (12) have shown that viscous dissipation can of living systems.
For slow oscillations, the equilibrium drop- be computed numerically for nonequilib- References
let pattern simply translocates as a whole, but rium structures of magnetic spinners assem- 1. J. V. I. Timonen et al., Science 341, 253 (2013).
less ordered dissipative patterns emerge for bling via vortex-vortex interactions, but their 2. D. Baker, D. A. Agard, Biochem. 33, 7505 (1994).
3. M. Muthukumar et al., Science 277, 1225 (1997).
fast oscillations, probably once the dissipation system does not allow for a simple experi- 4. L. Brunsveld et al., Chem. Rev. 101, 4071 (2001).
energy flux has passed a certain threshold. mental verification of the theory. For mag- 5. C. M. Dobson, Nature 426, 884 (2003).

In classical dissipative structures, such as netic nonequilibrium structures like those 6. G. Nicolis, I. Prigogine, Self-Organization in Nonequilib-
rium Systems (Wiley, New York, 1977).
reaction-diffusion patterns, Rayleigh-Bénard presented by Timonen et al., it could be ben- 7. G. Nicolis, I. Prigogine, G. Nocolis, Exploring Complexity
convection cells, or schools of fish, there is eficial to use an electromagnet to drive the (Freeman, New York, 1989).
no order in or very close to the thermody- system. In analogy to a calorimeter, where 8. J.-M. Lehn, Supramolecular Chemistry: Concepts and
Perspectives (VCH, Weinheim, 1995).
namic equilibrium. Only at a certain distance current is used to keep the temperature con- 9. G. M. Whitesides, B. Grzybowski, Science 295, 2418
from the equilibrium state, beyond a criti- stant, the current in the electromagnet could (2002).
cal point, do highly ordered configurations be measured to keep the magnetic field con- 10. S. I. Stupp et al., Science 276, 384 (1997).
11. B. A. Grzybowski et al., Soft Matter 5, 1110 (2009).
emerge (6). Moreover, order in such systems stant. The current needed to create dissipa- 12. K. V. Tretiakov et al., Soft Matter 5, 1279 (2009).
arises abruptly in a manner resembling a sharp tive structures would provide further insights
phase transition. By contrast, Timonen et al. into the rates of dissipation. 10.1126/science.1241793
CHEMISTRY
One hundred years ago, Niels Bohr’s pioneering

100 Years of Atomic Theory paper on the electronic structure of the
hydrogen atom revolutionized atomic theory.
David C. Clary
I
n 1913, Niels Bohr published a ground- absorption of radiation. In addi-
breaking paper that introduced a new tion, Bohr used Planck’s constant h
way of understanding atomic phenomena to identify the frequency ν of radia-
(1). Entitled “On the Constitution of Atoms tion as
and Molecules,” the article brought together
for the first time the model of the atom devel- ν = (W2 – W1)/h (1)
oped by Rutherford, which consisted of a
positively charged nucleus surrounded by for a transition from an initial
negatively charged electrons, with the theory state of energy W1 to a final state
for quantization of radiation developed by with energy W2. These principles
Planck. The paper became one of the most are second nature to scientists in
influential of the 20th century. 2013, but were radical a hundred
Although the model for the hydrogen years ago.
atom described in the paper was superseded At the time of Bohr’s paper, it was known stants as R = 2π2me4/h3, where m and e are
within 13 years by developments in quantum through the work of Rydberg, following on the mass and charge of the electron, respec-
CREDIT: NIELS BOHR ARCHIVE/COPENHAGEN
theory and wave mechanics, the work intro- from Balmer, that the frequency of a line in tively. It was this finding in particular that
duced several new concepts that have stood the visible spectrum of the hydrogen atom caught the attention of the notables of the
the test of time. These include the existence was given by day, with Einstein describing it as “very
of stationary states, in which an atomic or remarkable” and Bertrand Russell stating
molecular system can have a precise energy ν = R (1/n12 – 1/n22) (2) this was “perhaps the most sensational evi-
and the transition from one state to another dence in favor of Bohr’s theory” (2).
can be accompanied by the emission or where R is the empirically determined Ryd- To justify the introduction of quantum
berg constant and n1 and n2 are integers. The numbers in his theory, Bohr assumed that
Department of Physical and Theoretical Chemistry, Univer-
sity of Oxford, Oxford OX1 3QZ, UK. E-mail: david.clary@ special feature of Bohr’s theory was that it the angular momentum of the electron in
chem.ox.ac.uk gave R purely in terms of fundamental con- the hydrogen atom was quantized and that

Published by AAAS
PERSPECTIVES
electrons have orbits of fixed radius depend- papers did discuss several phenomena that explain all properties of atoms and mole-
ing on their quantum number. This concept proved to be significant later, including the cules was tantalizing. In 1926, Schrödinger
gave the famous planetary picture of the understanding of x-rays, magnetic proper- published his equation that essentially
atomic age, with electrons moving in orbits ties of atoms and molecules, and radioactiv- achieves this aim (8). Today, 100 years after
with well-defined radii around a nucleus. ity. Bohr also deduced correctly that β par- Bohr’s breakthrough, it is not just spectra
This was a period when several new theo- ticles were electrons emitted by the nucleus. that can be calculated accurately to explain
ries of the atom were being developed, and Some other theoretical concepts intro- and predict experiments but also the detailed
Bohr’s ideas were not readily accepted by duced in Bohr’s trilogy of papers did not energetics and dynamics of atoms and mol-
all. However, a subsequent note (3) in which hold up. More sophisticated theories gave no ecules (9). Any modern publication con-
Bohr showed that his theory also accurately angular momentum in the electronic ground cerned with understanding the properties of
explained the spectrum of the He+ ion, with a state of the hydrogen atom. Furthermore, his molecules, not just in physics and chemistry
correction introduced to account for the mass theory could not explain the spectra of atoms but also in fields from materials science to
of the nucleus, did much to convert doubters. with more than one electron, the intensities molecular biology, will use concepts derived
In addition, his theory explained precisely of spectral lines, effects of magnetic fields from Bohr’s pioneering paper.
the lines observed for the hydrogen atom on atoms, fine structure, and hyperfine struc-
from the infrared to the ultraviolet, including ture. However, subsequent extensions of References
1. N. Bohr, Philos. Mag. 26, 1 (1913).
new observations made after 1913. Bohr’s theory by himself, Sommerfeld, and
2. H. Kragh, Niels Bohr and the Quantum Atom 1913-25

Bohr’s revolutionary paper was part of others did make some improvements such (Oxford Univ. Press, Oxford, 2012).
a trilogy, with the second and third papers as the introduction of elliptical orbits for the 3. N. Bohr, Nature 92, 231 (1913).
dealing with many-electron atoms and mol- electrons and the quantization of the radial 4. N. Bohr, Philos. Mag. 26, 476 (1913).
5. N. Bohr, Philos. Mag. 26, 857 (1913).
ecules, respectively (4, 5). The two subse- motion of the electron together with its
6. H. Kragh, Phys. Today 66, 36 (2013).
quent papers extended some ideas of the angular momentum. This modified approach 7. E. Bianchi, H. M. Haggard, Phys. Rev. Lett. 107, 011301
first paper, but the results did not agree as has found useful applications in understand- (2011).
well with experiments. Furthermore, Bohr’s ing the dynamics of atoms and molecules to 8. E. Schrödinger, Ann. Phys. 79, 361 (1926).
assumptions that H2O was a linear mole- the present day (7). 9. D. C. Clary, Science 314, 265 (2006).
cule and that H2+ was unstable did not gain Bohr’s bold idea that a unifying theory
much favor from chemists (6). However, the of electronic structure must exist that can 10.1126/science.1240200
DEVELOPMENTAL BIOLOGY
Enhancing Pluripotency The identification of large numbers of human

embryonic stem cell enhancers provides a
and Lineage Specification useful tool to study cell fate and disease.
Wei Xie1,3 and Bing Ren1,2
E
mbryonic stem cells (ESCs) and with the challenges to understand the func- (2–5). Enhancers frequently harbor binding
induced pluripotent stem cells (iPSCs) tions and regulation of these elements. sites or DNA motifs of transcription factors
hold great promise in regenerative that are important for lineage identity (4,
medicine. Realizing this potential requires a Characteristics of Enhancers 6–8). In particular, genes coding for key reg-
thorough understanding of the genetic pro- Enhancers have been proposed to act as a ulators of cell identities are surrounded and
grams regulating their pluripotency and pivot of lineage identity and developmen- regulated by “super enhancers” (9), which
lineage commitment. Recent studies have tal potential (1). Hundreds of thousands of harbor unusually high levels of transcrip-
generated a wealth of information regard- putative enhancers have been annotated in tion factor binding. Binding of transcription
ing the transcriptional circuitry underlying the human genome (see the table). In par- factors to enhancers appears to precede their
self-renewal and lineage commitment of ES ticular, putative enhancers involved in pluri- binding to promoters during reprogramming
cells. Here we review these studies, focusing potency and lineage specification have been of somatic cells to iPSCs, which suggests
on a type of cis-regulatory sequences called found by analysis of human embryonic stem that enhancer programming is among one of
enhancers and their potential roles in pluripo- cells (hESCs) and several hESC-derived cell the first cellular events necessary for cell fate
tency and cell fate determination. We also dis- types and constitute a sizable portion of reg- transition (10).
cuss how such information can help ES cell ulatory elements in their genomes. During
research and regenerative medicine, together differentiation of ESCs, enhancer activity is Exploiting Stem Cell Enhancers
highly dynamic, transitioning through sev- Although the full catalog of enhancers in the
1
Ludwig Institute for Cancer Research and 2Department of
eral states characterized by distinct patterns human genome is still unknown, the identi-
Cellular and Molecular Medicine, University of California of deoxyribonuclease I (DNase I) hypersen- fication of enhancers in hESCs and iPSCs
San Diego School of Medicine, 9500 Gilman Drive, La Jolla, sitivity, chromatin modifications, and tran- will likely have a large impact on several
CA 92093–0653, USA. 3Peking-Tsinghua Center for Life scription factor binding (1). This transition fronts. For example, it will facilitate studies
Sciences, MOE Key Laboratory of Bioinformatics, School of
Life Sciences, Tsinghua University, Beijing 100084, China. between different enhancer states is corre- of molecular mechanisms underlying disease-
E-mail: biren@ucsd.edu; xiewei121@gmail.com lated with lineage-specific gene expression linked sequence variants. Of noncoding dis-

Published by AAAS
Enhancing Pluripotency and Lineage Specification
Wei Xie and Bing Ren
Science 341, 245 (2013);
DOI: 10.1126/science.1236254


PERSPECTIVES
electrons have orbits of fixed radius depend- papers did discuss several phenomena that explain all properties of atoms and mole-
ing on their quantum number. This concept proved to be significant later, including the cules was tantalizing. In 1926, Schrödinger
gave the famous planetary picture of the understanding of x-rays, magnetic proper- published his equation that essentially
atomic age, with electrons moving in orbits ties of atoms and molecules, and radioactiv- achieves this aim (8). Today, 100 years after
with well-defined radii around a nucleus. ity. Bohr also deduced correctly that β par- Bohr’s breakthrough, it is not just spectra
This was a period when several new theo- ticles were electrons emitted by the nucleus. that can be calculated accurately to explain
ries of the atom were being developed, and Some other theoretical concepts intro- and predict experiments but also the detailed
Bohr’s ideas were not readily accepted by duced in Bohr’s trilogy of papers did not energetics and dynamics of atoms and mol-
all. However, a subsequent note (3) in which hold up. More sophisticated theories gave no ecules (9). Any modern publication con-
Bohr showed that his theory also accurately angular momentum in the electronic ground cerned with understanding the properties of
explained the spectrum of the He+ ion, with a state of the hydrogen atom. Furthermore, his molecules, not just in physics and chemistry
correction introduced to account for the mass theory could not explain the spectra of atoms but also in fields from materials science to
of the nucleus, did much to convert doubters. with more than one electron, the intensities molecular biology, will use concepts derived
In addition, his theory explained precisely of spectral lines, effects of magnetic fields from Bohr’s pioneering paper.
the lines observed for the hydrogen atom on atoms, fine structure, and hyperfine struc-
from the infrared to the ultraviolet, including ture. However, subsequent extensions of References
1. N. Bohr, Philos. Mag. 26, 1 (1913).
new observations made after 1913. Bohr’s theory by himself, Sommerfeld, and
2. H. Kragh, Niels Bohr and the Quantum Atom 1913-25

Bohr’s revolutionary paper was part of others did make some improvements such (Oxford Univ. Press, Oxford, 2012).
a trilogy, with the second and third papers as the introduction of elliptical orbits for the 3. N. Bohr, Nature 92, 231 (1913).
dealing with many-electron atoms and mol- electrons and the quantization of the radial 4. N. Bohr, Philos. Mag. 26, 476 (1913).
5. N. Bohr, Philos. Mag. 26, 857 (1913).
ecules, respectively (4, 5). The two subse- motion of the electron together with its
6. H. Kragh, Phys. Today 66, 36 (2013).
quent papers extended some ideas of the angular momentum. This modified approach 7. E. Bianchi, H. M. Haggard, Phys. Rev. Lett. 107, 011301
first paper, but the results did not agree as has found useful applications in understand- (2011).
well with experiments. Furthermore, Bohr’s ing the dynamics of atoms and molecules to 8. E. Schrödinger, Ann. Phys. 79, 361 (1926).
assumptions that H2O was a linear mole- the present day (7). 9. D. C. Clary, Science 314, 265 (2006).
cule and that H2+ was unstable did not gain Bohr’s bold idea that a unifying theory
much favor from chemists (6). However, the of electronic structure must exist that can 10.1126/science.1240200
DEVELOPMENTAL BIOLOGY
Enhancing Pluripotency The identification of large numbers of human

embryonic stem cell enhancers provides a
and Lineage Specification useful tool to study cell fate and disease.
Wei Xie1,3 and Bing Ren1,2
E
mbryonic stem cells (ESCs) and with the challenges to understand the func- (2–5). Enhancers frequently harbor binding
induced pluripotent stem cells (iPSCs) tions and regulation of these elements. sites or DNA motifs of transcription factors
hold great promise in regenerative that are important for lineage identity (4,
medicine. Realizing this potential requires a Characteristics of Enhancers 6–8). In particular, genes coding for key reg-
thorough understanding of the genetic pro- Enhancers have been proposed to act as a ulators of cell identities are surrounded and
grams regulating their pluripotency and pivot of lineage identity and developmen- regulated by “super enhancers” (9), which
lineage commitment. Recent studies have tal potential (1). Hundreds of thousands of harbor unusually high levels of transcrip-
generated a wealth of information regard- putative enhancers have been annotated in tion factor binding. Binding of transcription
ing the transcriptional circuitry underlying the human genome (see the table). In par- factors to enhancers appears to precede their
self-renewal and lineage commitment of ES ticular, putative enhancers involved in pluri- binding to promoters during reprogramming
cells. Here we review these studies, focusing potency and lineage specification have been of somatic cells to iPSCs, which suggests
on a type of cis-regulatory sequences called found by analysis of human embryonic stem that enhancer programming is among one of
enhancers and their potential roles in pluripo- cells (hESCs) and several hESC-derived cell the first cellular events necessary for cell fate
tency and cell fate determination. We also dis- types and constitute a sizable portion of reg- transition (10).
cuss how such information can help ES cell ulatory elements in their genomes. During
research and regenerative medicine, together differentiation of ESCs, enhancer activity is Exploiting Stem Cell Enhancers
highly dynamic, transitioning through sev- Although the full catalog of enhancers in the
1
Ludwig Institute for Cancer Research and 2Department of
eral states characterized by distinct patterns human genome is still unknown, the identi-
Cellular and Molecular Medicine, University of California of deoxyribonuclease I (DNase I) hypersen- fication of enhancers in hESCs and iPSCs
San Diego School of Medicine, 9500 Gilman Drive, La Jolla, sitivity, chromatin modifications, and tran- will likely have a large impact on several
CA 92093–0653, USA. 3Peking-Tsinghua Center for Life scription factor binding (1). This transition fronts. For example, it will facilitate studies
Sciences, MOE Key Laboratory of Bioinformatics, School of
Life Sciences, Tsinghua University, Beijing 100084, China. between different enhancer states is corre- of molecular mechanisms underlying disease-
E-mail: biren@ucsd.edu; xiewei121@gmail.com lated with lineage-specific gene expression linked sequence variants. Of noncoding dis-

Published by AAAS
PERSPECTIVES
Recent genomic studies for annotating putative enhancers critical to determine how well
in hESCs and their derivatives these hESC- and hiPSC-based
# Of putative enhancers differentiation processes reca-
[or DHSs or transcription factor (TF) pitulate in vivo development.
Cell type Identification method
binding sites if indicated] For instance, aberrant gene
expression and promoter regula-
Studies focused on hESCs and/or their derivatives
tion were found when compar-
hESC H1 (16) ~58,000 Histone modifications ing hESC-derived pancreatic
hESC H9 and derived neuroectoderm cells (3) ~7,000 (hESCs) p300, BRG1 and histone
endoderm cells differentiated
modifications in vivo and in vitro (13). Cell
culture can also induce expan-
hESC H1 and BMP4 differentiated hESC (2) ~29,000 (hESCs) Histone modifications
~33,000 (hESC-BMP4)
sion of large histone H3 lysine
9 trimethylated (H3K9me3)
hESC H9 and derived neural crest cells (NCCs) (11) ~5,000 (NCC) p300 and histone domains (6), which are associ-
modifications
ated with constitutive hetero-
hESC H1 and derived mesendoderm, trophoblast-like cells, ~13,000 to ~43,000 for each Histone modifications chromatin. Given that enhancer
neural progenitor cells, and mesenchymal stem cells (4) cell type. ~89,000 in all activity is highly dynamic, it is
hESCs and hESC derivatives

conceivable that they may be
hESC HUES64 and derived mesoderm, neuroectoderm, ~54,000 to ~78,000 for each Histone modifications more sensitive to environmental
and definitive endoderm (5) cell type changes and culture conditions.
Other large-scale studies that include hESCs and/or their derivatives Remaining Challenges
Despite the identification of
9 cell types, including hESC H1 (8) 3.8% of the genome Histone modifications
sequences (hESC)
large numbers of hESC enhanc-
ers, it remains challenging to
125 cell types, including several hESC and iPSC lines (15) ~1,176,000 (total distal DHSs DHS mapping elucidate the gene-regulatory
in all cell types)
programs underlying pluri-
5 cell lines, including hESC H1 (17) ~4,476,000 (total distal TF TF binding potency and cell differentia-
binding sites)
tion. One reason is that the tar-
349 cell types, including several hESC and iPSC lines, and ~90,000 to ~370,000 (distal DHS mapping gets of these enhancers are not
hESC derivatives (7) and proximal DHSs for each easy to define, as an enhancer
cell type) often controls more than one
29 cell types, including several hESC and iPSC lines, ~94,000 per cell type Histone modifications gene, and the target gene is
hESC-derived neurons, and neural progenitor cells (6) not necessarily the nearest one
(1, 14). Current approaches to
ease-linked single-nucleotide polymorphisms lators of specific cell lineages. For example, address this problem are still imperfect. In
(SNPs) identified from genome-wide associa- such analysis has led to the identification of one approach, chromatin state at enhancers
tion studies (GWAS), 76.6% either fall into a novel regulators (nuclear receptors NR2F1/ is used to identify promoters sharing similar
DNase I–hypersensitive site (DHS), a marker F2) of human neural crest cells (11). In addi- chromatin profiles in nearby regions across
for putative enhancers, or are in complete tion, cell type–specific enhancers can facili- multiple cell types (8, 15). Although corre-
linkage with SNPs in a nearby DHS (7). One tate the development of lineage-specific bio- lated activity may reflect coordinated regula-
hypothesis is that these SNPs may disrupt markers to specify appropriate cell types. tion between enhancers and promoters, it can
transcription factor binding sites at enhanc- Such markers may help further optimize also be indirect or even irrelevant. In another
ers, which leads to misregulation of their tar- hESC and iPSC differentiation protocols, or approach, chromosome conformation cap-
geted genes. This can now be tested in hESCs identify subsets of cells within a differentiated ture techniques (3C) and derivatives (such as
or patient-derived iPSCs, in which disease- population that more closely resemble their in HiC) were used to directly assess the interac-
linked SNPs can be engineered in the genome vivo counterparts. Another potential benefit tion between enhancers and promoters (14).
to create risk variant–carrying and isogenic of using enhancer-based biomarkers is that However, such interaction-based analyses are
“wild-type” pluripotent cell lines. These lines enhancers can be primed or poised before acti- not yet satisfactory owing to the low resolu-
can be further differentiated into cell types of vation, which allows preselection of hESCs tion of identified long-range interactions. It
interest as disease models to study the causal and iPSCs before induction of differentia- is also unclear if interaction alone can accu-
effects of these SNPs on potential disease- tion. For example, different iPSC lines dem- rately predict functional regulation.
related gene expression patterns. onstrate distinct lineage differentiation effi- The extent to which each enhancer con-
Locations of enhancers could also help ciencies (12). It would be interesting to exam- tributes to target gene expression is gen-
pinpoint binding sites of lineage-specific ine whether enhancer activity can be used erally undefined except for a handful of
transcription factors, which often occupy to distinguish iPSCs for their differentiation enhancers such as those at the β-globin gene
hundreds of thousands of loci in the genome. potentials. Further, global enhancer activity (14). To examine the effect of enhancers on
Transcription factors falling within active can be monitored for benchmarking hESC- transcription, one would need to mutate the
enhancers are more likely to be conserved and and hiPSC (human iPSC)-derived cell types element in its endogenous locus and charac-
functional (11). Furthermore, motif analysis against tissues derived in vivo. As ES cells terize the target gene expression. Given the
of enhancer sequences can help identify regu- are raised and maintained in cell culture, it is rapid progress in genome editing technol-

Published by AAAS
PERSPECTIVES
ogy, it is tempting to predict that large num- Recent “omics” studies have uncovered a 7. M. T. Maurano et al., Science 337, 1190 (2012).
bers of putative enhancers will eventually be large number of putative enhancers that can 8. J. Ernst et al., Nature 473, 43 (2011).
9. W. A. Whyte et al., Cell 153, 307 (2013).
genetically engineered in a high-throughput program the transcriptional networks in plu- 10. A. Soufi, G. Donahue, K. S. Zaret, Cell 151, 994 (2012).
manner. Further, how enhancers are regu- ripotent cells. However, these rich data sets 11. A. Rada-Iglesias et al., Cell Stem Cell 11, 633 (2012).
lated needs to be more precisely elucidated. have revealed more questions than answers. 12. C. Bock et al., Cell 144, 439 (2011).
Molecular mechanisms governing enhanc- This is only the first step toward understand- 13. R. Xie et al., Cell Stem Cell 12, 224 (2013).
14. J. H. Gibcus, J. Dekker, Cell 49, 773 (2013).
ers are much less clear compared with those ing the molecular mechanisms of pluripo- 15. R. E. Thurman et al., Nature 489, 75 (2012).
for promoters. For example, more needs to tency, and eventually realizing the potentials 16. R. D. Hawkins et al., Cell Stem Cell 6, 479 (2010).
be learned about the factors that initiate, of hESCs and iPSCs. 17. M. B. Gerstein et al., Nature 489, 91 (2012).
establish, and maintain the active, poised,
and silenced states of enhancers, as well as References and Notes Acknowledgments: We apologize to those authors whose
works are not covered here owing to space limitations. This
the extent to which DNA looping partici- 1. C. Buecker, J. Wysocka, Trends Genet. 28, 276 (2012).
work is supported by funds from the Ludwig Institute for Can-
pates in gene activation. This will enable a 2. R. D. Hawkins et al., Cell Res. 21, 1393 (2011).
cer Research, NIH (grants U01ES017166 and 5U01HL107442),
3. A. Rada-Iglesias et al., Nature 470, 279 (2011).
fuller understanding of the role of enhancers 4. W. Xie et al., Cell 153, 1134 (2013).
and the California Institute for Regenerative Medicine (RN2-
00905) to B.R.
in the genetic programs of pluripotency and 5. C. A. Gifford et al., Cell 153, 1149 (2013).
lineage specification (16, 17). 6. J. Zhu et al., Cell 152, 642 (2013). 10.1126/science.1236254

ENGINEERING
Multiscale Design and Integration Designing systems from nano- to macroscale

can optimize the synergistic integration of
sustainable energy and waste treatment in
of Sustainable Building Functions buildings.
Maria Paz Gutierrez1 and Luke P. Lee2
B
uildings and cities must move toward nia of solar-oriented architecture, an MIT office of the National Science Founda-
balancing resources and advanc- team in 1947 demonstrated the Solar House tion, 10 teams are participating in the Sci-
ing risk preparedness in support of II, which used sunlight to heat water tanks ence in Energy and Environmental Design
robust sustainability and climate change on a facade to heat interior space (7). More (SEED) group. Among these interdisci-
action (1–5). To date, the design of sustain- recently built examples of net-zero design at plinary teams, a few are researching new
able buildings has optimized the genera- the residential scale demonstrated integra- building technologies that integrate nano-
tion of energy separately from the regeneration of water-based radiation through ceil- and microengineering principles for build-
tion of water and the processing of waste. ing panels that absorb sunlight. These panels ing-scale applications (10). The focus is on
However, the integration of macro-, micro-, transfer the heat into thermal storage facil- developing self-regulated and multifunc-
and nanoscale engineering principles has ities that heat the residence in winter by a tional building enclosures that act as “living
enabled examples of synergistic optimiza- heat exchange process (8). walls, materials and systems.” A Univer-
tion of energy generation with water and the In skyscrapers, architects and engineers sity of Colorado team is researching a liv-
processing of waste. Innovative multiscale have piloted multiple built and research ing wall for automatic thermal regulation,
design can allow buildings to wholly contain models to advance an integrated optimi- with a goal of net-zero energy and water
utilities, rather than merely providing routes zation of energy, water, and waste (9). The conservation. The living wall comprises
for their delivery. Pearl River Tower in China demonstrates the two optimized microvascular networks and
As early as the fourth century BCE, integration of reduction, reclamation, pas- a distributed phase-change medium (PCM)
CREDIT: MARIA-PAZ GUTIERREZ, PETER SUEN, YOUNG G. PARK, AND LUKE P. LEE
Greek philosophers and builders harnessed sive absorption, and generation of energy. that absorbs or releases heat. The research
the Sun for energy in houses, communities, This project features multiple integrated develops a new polymer and microvascu-
and even cities, such as Priene and Olyn- technologies, such as a reclamation scheme lar fluid channels to establish a wall sys-
thus. To prevent further destruction of the accomplished through water harvested from tem with autonomous movement of air and
forests around their city, Romans adopted chilled surfaces to control interior humidity. water to charge or discharge the PCM for
solar-oriented architecture from classic Once filtered, this water is reused for interior optimized thermal regulation (10).
Greece, advancing it to a much wider range plant irrigation and toilet flushing. Here, a A University of Pennsylvania team is
of climate applications. Roman builders single integrated technology serves both developing eSkin, which is inspired by
invented solar-oriented openings with glass functions for energy reduction (dehumidifi- human smooth muscle cells that alter their
(i.e., south-facing windows) to heat rooms cation) and water recovery. The implemen- extracellular matrices and their surround-
(6). With the spirit of more than two millen- tation of the triple net-zero concept (zero ing environments. The facade is composed
energy, zero emissions, zero waste) has been of engineered microstructure patterns in
1
Department of Architecture, University of California, demonstrated at the residential level, and passively responsive materials to gener-
Berkeley, CA 94270, USA. 2Departments of Bioengineer- many other buildings have shown the feasi- ate three-dimensional surface and opti-
ing, Electrical Engineering, and Computer Science, Berke-
ley Sensor and Actuator Center, University of California,
bility of nearly net-zero energy (7, 8). cal effects for light transmission calibra-
Berkeley, CA 94270, USA. E-mail: mpazgut@berkeley.edu; Currently, through the Emerging Fron- tion. With embedded sensors and imagers,
lplee@berkeley.edu tiers in Research and Innovation (EFRI) eSkin modulates passive solar energy, light,

Published by AAAS
Multiscale Design and Integration of Sustainable Building Functions
Maria Paz Gutierrez and Luke P. Lee
Science 341, 247 (2013);
DOI: 10.1126/science.1237278


http://www.sciencemag.org/content/suppl/2013/07/17/341.6143.247.DC1.html
found at:
PERSPECTIVES
ogy, it is tempting to predict that large num- Recent “omics” studies have uncovered a 7. M. T. Maurano et al., Science 337, 1190 (2012).
bers of putative enhancers will eventually be large number of putative enhancers that can 8. J. Ernst et al., Nature 473, 43 (2011).
9. W. A. Whyte et al., Cell 153, 307 (2013).
genetically engineered in a high-throughput program the transcriptional networks in plu- 10. A. Soufi, G. Donahue, K. S. Zaret, Cell 151, 994 (2012).
manner. Further, how enhancers are regu- ripotent cells. However, these rich data sets 11. A. Rada-Iglesias et al., Cell Stem Cell 11, 633 (2012).
lated needs to be more precisely elucidated. have revealed more questions than answers. 12. C. Bock et al., Cell 144, 439 (2011).
Molecular mechanisms governing enhanc- This is only the first step toward understand- 13. R. Xie et al., Cell Stem Cell 12, 224 (2013).
14. J. H. Gibcus, J. Dekker, Cell 49, 773 (2013).
ers are much less clear compared with those ing the molecular mechanisms of pluripo- 15. R. E. Thurman et al., Nature 489, 75 (2012).
for promoters. For example, more needs to tency, and eventually realizing the potentials 16. R. D. Hawkins et al., Cell Stem Cell 6, 479 (2010).
be learned about the factors that initiate, of hESCs and iPSCs. 17. M. B. Gerstein et al., Nature 489, 91 (2012).
establish, and maintain the active, poised,
and silenced states of enhancers, as well as References and Notes Acknowledgments: We apologize to those authors whose
works are not covered here owing to space limitations. This
the extent to which DNA looping partici- 1. C. Buecker, J. Wysocka, Trends Genet. 28, 276 (2012).
work is supported by funds from the Ludwig Institute for Can-
pates in gene activation. This will enable a 2. R. D. Hawkins et al., Cell Res. 21, 1393 (2011).
cer Research, NIH (grants U01ES017166 and 5U01HL107442),
3. A. Rada-Iglesias et al., Nature 470, 279 (2011).
fuller understanding of the role of enhancers 4. W. Xie et al., Cell 153, 1134 (2013).
and the California Institute for Regenerative Medicine (RN2-
00905) to B.R.
in the genetic programs of pluripotency and 5. C. A. Gifford et al., Cell 153, 1149 (2013).
lineage specification (16, 17). 6. J. Zhu et al., Cell 152, 642 (2013). 10.1126/science.1236254

ENGINEERING
Multiscale Design and Integration Designing systems from nano- to macroscale

can optimize the synergistic integration of
sustainable energy and waste treatment in
of Sustainable Building Functions buildings.
Maria Paz Gutierrez1 and Luke P. Lee2
B
uildings and cities must move toward nia of solar-oriented architecture, an MIT office of the National Science Founda-
balancing resources and advanc- team in 1947 demonstrated the Solar House tion, 10 teams are participating in the Sci-
ing risk preparedness in support of II, which used sunlight to heat water tanks ence in Energy and Environmental Design
robust sustainability and climate change on a facade to heat interior space (7). More (SEED) group. Among these interdisci-
action (1–5). To date, the design of sustain- recently built examples of net-zero design at plinary teams, a few are researching new
able buildings has optimized the genera- the residential scale demonstrated integra- building technologies that integrate nano-
tion of energy separately from the regeneration of water-based radiation through ceil- and microengineering principles for build-
tion of water and the processing of waste. ing panels that absorb sunlight. These panels ing-scale applications (10). The focus is on
However, the integration of macro-, micro-, transfer the heat into thermal storage facil- developing self-regulated and multifunc-
and nanoscale engineering principles has ities that heat the residence in winter by a tional building enclosures that act as “living
enabled examples of synergistic optimiza- heat exchange process (8). walls, materials and systems.” A Univer-
tion of energy generation with water and the In skyscrapers, architects and engineers sity of Colorado team is researching a liv-
processing of waste. Innovative multiscale have piloted multiple built and research ing wall for automatic thermal regulation,
design can allow buildings to wholly contain models to advance an integrated optimi- with a goal of net-zero energy and water
utilities, rather than merely providing routes zation of energy, water, and waste (9). The conservation. The living wall comprises
for their delivery. Pearl River Tower in China demonstrates the two optimized microvascular networks and
As early as the fourth century BCE, integration of reduction, reclamation, pas- a distributed phase-change medium (PCM)
CREDIT: MARIA-PAZ GUTIERREZ, PETER SUEN, YOUNG G. PARK, AND LUKE P. LEE
Greek philosophers and builders harnessed sive absorption, and generation of energy. that absorbs or releases heat. The research
the Sun for energy in houses, communities, This project features multiple integrated develops a new polymer and microvascu-
and even cities, such as Priene and Olyn- technologies, such as a reclamation scheme lar fluid channels to establish a wall sys-
thus. To prevent further destruction of the accomplished through water harvested from tem with autonomous movement of air and
forests around their city, Romans adopted chilled surfaces to control interior humidity. water to charge or discharge the PCM for
solar-oriented architecture from classic Once filtered, this water is reused for interior optimized thermal regulation (10).
Greece, advancing it to a much wider range plant irrigation and toilet flushing. Here, a A University of Pennsylvania team is
of climate applications. Roman builders single integrated technology serves both developing eSkin, which is inspired by
invented solar-oriented openings with glass functions for energy reduction (dehumidifi- human smooth muscle cells that alter their
(i.e., south-facing windows) to heat rooms cation) and water recovery. The implemen- extracellular matrices and their surround-
(6). With the spirit of more than two millen- tation of the triple net-zero concept (zero ing environments. The facade is composed
energy, zero emissions, zero waste) has been of engineered microstructure patterns in
1
Department of Architecture, University of California, demonstrated at the residential level, and passively responsive materials to gener-
Berkeley, CA 94270, USA. 2Departments of Bioengineer- many other buildings have shown the feasi- ate three-dimensional surface and opti-
ing, Electrical Engineering, and Computer Science, Berke-
ley Sensor and Actuator Center, University of California,
bility of nearly net-zero energy (7, 8). cal effects for light transmission calibra-
Berkeley, CA 94270, USA. E-mail: mpazgut@berkeley.edu; Currently, through the Emerging Fron- tion. With embedded sensors and imagers,
lplee@berkeley.edu tiers in Research and Innovation (EFRI) eSkin modulates passive solar energy, light,

Published by AAAS
PERSPECTIVES
Thermal insulation ity. The disinfected water is recirculated at

Aluminum panel
Glass night for radiant floor heating as a by-prod-
uct of the process, providing synergistic
water and waste regeneration and energy
Light generation. Once the water is used as a ther-
mal source, it is brought into the building
for reuse in toilet flushing and/or laundry.
Integrated optimization of resources in
buildings can decentralize the delivery of
utilities in cities. Potential benefits include
Graywater substantial cost savings in energy inputs
collection
and fewer disruptions from power outages
Disinfected or infrastructure failures. Further advances
water in sustainable building technology will
demand research collaborations among
architects, engineers, and scientists. Build-
ing control functions must monitor and

forecast interior and exterior conditions and
occupancy in order to optimize energy use
A B
(12). New active materials may self-gener-
Light Pathogenic Nano-TiO2 ate, regenerate, or undergo phases from the
Microcavity Microlens microorganism photocatalysts nanoscale to the architectural scale (13). To
streamline this innovation, architects, sci-
entists, and engineers will have to make
substantial advances in building materi-
als, as well as multiscale building analysis
OH* (14, 15). Multiscale designs in which water,
energy, and waste are synergistically inte-
grated can mark a new era for “resourcing”
resources in buildings.

1. J. P. Holdren, Science 319, 424 (2008).
2. I. Sartori et al., Energy Build. 48, 220 (2012).
3. C. Rosenzweig et al. Nature 467, 909 (2010).
4. P. Hernandez, P. Kenny, Energy Build. 42, 815 (2010).
5. U.S. Department of Energy, Zero Energy Buildings; http://
zeb.buildinggreen.com.
6. K. Butti, J. Perlin, A Golden Thread: 2500 Years of Solar
C 1 cm D 2 µm Architecture and Technology (Van Nostrand, New York,
1980).
Integrative sustainable building technologies. Multiscale design is illustrated from macro- to nanoscales. 7. M. Davies, Int. J. Energy Res. 10, 305 (1986).
(A) A schematic axonometric drawing of the SOAP decentralization system: graywater collection, solar-acti- 8. W. Sobek, K. Sedlbauer, H. Schuster, in Technology
vated disinfection (facade), and recirculation (radiant floor heating). (B) A single SOAP panel containing Guide: Principles, Applications and Trends, H.-J. Bull-
(C) a photocatalytic optofluidic network is created by multiple layers of integrated photocatalytic reactors. inger, Ed. (Springer, New York, 2009), pp. 432–435.
9. R. R. Frechette, R. Gilchrist, in CTBUH: Proceedings of the
(D) Pathogenic microorganisms are disinfected via photocatalytic TiO2 nanoparticles. The larger support Council on Tall Buildings and Urban Habitat’s 8th World
particles act as microlenses to concentrate solar energy and to generate an efficient photocatalytic effect. Congress, Dubai, United Arab Emirates, 3 to 5 March
2008, pp. 7–16.
10. Science in Energy and Environmental Design (SEED)
and moisture by using microstructured facade, where an efficient photocatalytic Awards; www.nsf.gov/eng/efri/fy10awards_SEED.jsp.
patterns to respond to local performance disinfection process is activated by sun- 11. Solar Optics-based Active Pasteurization (SOAP) for
criteria (10). light throughout the day. Effective disin- Greywater Reuse and Integrated Thermal (GRIT) Building
Control; www.nsf.gov/awardsearch/showAward?AWD_
Another example is solar optics–based fection of pathogenic microorganisms can ID=1038279.
active panels (SOAP) technology for gray- be achieved via integrated photocatalytic 12. N. Gershenfeld et al., Science 327, 1086 (2010).
water reuse and integrated solar energy har- titanium dioxide (TiO2) nanoparticles on 13. M. Ashby et al., Nanomaterials, Nanotechnologies and
vesting (11). In this project, graywater is microlens arrays. Design: An Introduction for Engineers and Architects
(Butterworth-Heinemann, Amsterdam, 2009).
recovered from washing (sinks, bathtubs, By integrating nanophotocatalytic ele- 14. A. Malkawi, G. Augenbroe, Advanced Building Simulation
and laundry) by disinfection, in buildings ments on microlenses, the panel system (Spon, New York, ed. 1, 2004), pp. 1–5.
that require extreme water conservation, improves light capture, transmission, and 15. J. E. Fernández, Science 315, 1807 (2007).
cooling through the day and heating dur- the efficiency of water disinfection in a
ing the night because of major diurnal tem- broad range of geometries and facade ori- Acknowledgments: Supported by the NSF EFRI SEED award.
perature swings (as found in arid climates). entations. Expanded nanophotonic archi-
Once collected, graywater is exposed to tectures can increase the density of catalytic
a solar-based optofluidic platform in the sites if the panels must deliver more capac- 10.1126/science.1237278

Published by AAAS
Applications of Acceptorless Dehydrogenation and Related
Transformations in Chemical Synthesis
Chidambaram Gunanathan and David Milstein
Science 341, (2013);
DOI: 10.1126/science.1229712


REVIEW SUMMARY
Applications of Acceptorless READ THE FULL ARTICLE ONLINE

Dehydrogenation and Related Cite this article as C. Gunanathan, D. Milstein,

Science 341, 1229712 (2013).
DOI: 10.1126/science.1229712
Transformations in Chemical Synthesis

Chidambaram Gunanathan and David Milstein*
ARTICLE OUTLINE
Precursors to Modern AD
Background: Acceptorless dehydrogenation (AD) reactions can result not only in simple removal
of hydrogen gas from various substrates but also, importantly, in surprisingly efficient and environ- Alkane Dehydrogenation
mentally benign (“green”) synthetic methodologies when intermediates resulting from the initial
Alcohol Dehydrogenation
dehydrogenation process undergo further reactions.
Dehydrogenative Coupling of Alcohols
Advances: Traditionally, dehydrogenation/oxidation reactions of organic compounds have been
to Form Esters
performed using stoichiometric amounts of inorganic oxidants, in addition to employing various

additives, cocatalysts, and catalytic systems that result in generation of copious stoichiometric, Dehydrogenative Coupling of Alcohols
often toxic, waste. Catalytic transfer hydrogenation methods, in which stoichiometric amounts of with Amines to Form Amides
sacrificial organic acceptor compounds are used, also generate stoichiometric amounts of organic Dehydrogenative Coupling with Concomitant
waste. Recent developments in catalysis by metal complexes have resulted in AD reactions that Condensation Reactions
release hydrogen gas and in related reactions in which dehydrogenation is followed by in situ
consumption of the generated hydrogen equivalents and no net hydrogen gas is liberated. These Alkane Metathesis
reactions circumvent the need for conventional oxidants or sacrificial acceptors and provide an Alkylation of Amines by Borrowing Hydrogen
assortment of applications in organic synthesis, including several methods based on further reactiv- Methodology
ity of the dehydrogenated intermediate compounds. Moreover, the evolved hydrogen gas is valuable
in itself. Alcohols as a Source of Electrophiles
and Nucleophiles
Outlook: Further development of new ADs for green, efficient chemical synthesis is expected to be
greatly influenced by fundamental organometallic chemistry as a basis for catalyst design. Such Outlook
processes are highly desirable and are expected to gradually displace elaborate conventional labo-
ratory and industrial synthetic methods. They may also provide opportunities for hydrogen storage ADDITIONAL RESOURCES
cycles, because the dehydrogenation reactions can be reversed under hydrogen pressure using the J. Choi et al., Dehydrogenation and related reactions
same catalyst. In general, AD and related dehydrogenative coupling reactions have the potential catalyzed by iridium pincer complexes. Chem. Rev.
for redirecting synthetic strategies to the use of sustainable resources, devoid of toxic reagents and 111, 1761–1779 (2011). doi: 10.1021/cr1003503
deleterious side reactions, with no waste generation.
C. Gunanathan, D. Milstein, Metal-ligand coopera-
tion by aromatization-dearomatization. Acc. Chem.
Dehydrogenation strategies in Res. 44, 588–602 (2011). doi: 10.1021/ar2000265
organic synthesis. (A) Successive
J. F. Bower, M. J. Krische, Formation of C-C Bonds
AD with release of hydrogen gas.
The catalyst liberates H2 from both via iridium-catalyzed hydrogenation and transfer
starting compound and intermedi- hydrogenation. Top. Organomet. Chem. 34, 107–
ate, exemplified by dehydrogena- 138 (2011). doi: 10.1007/978-3-642-15334-1_5
tive coupling of primary alcohols
with amines to form amides. (B) AD RELATED ITEMS IN SCIENCE
with H2 and water release. A dehy-
drogenated intermediate couples J. R. Zbieg et al., Enantioselective C-H crotylation
with nucleophiles, exemplified by of primary alcohols via hydrohydroxyalkylation
dehydrogenative coupling of alco- of butadiene. Science 336, 324–327 (2012).
hols with amines (liberating water) doi: 10.1126/science.1219274
to form imines that can be isolated
or carried on to products such as A. J. A. Watson, J. M. J. Williams, The give and
pyrazines. (C) Borrowing hydrogen. take of alcohol activation. Science 329, 635–636
The catalyst dehydrogenates the (2010). doi: 10.1126/science.1191843
substrate and formally transfers the
H atoms to an unsaturated interme- C. Gunanathan et al., Direct synthesis of amides
diate, exemplified by coupling of from alcohols and amines with liberation of H2.
ammonia or amines with alcohols Science 317, 790–792 (2007).
to form new amines, liberating doi: 10.1126/science.1145295
water, but not H2. (D) Coupling of
redox pairs. Dehydrogenation gen- A. S. Goldman et al., Catalytic alkane metathesis
erates an electrophile and a nucleo- by tandem alkane dehydrogenation-olefin
phile that react to form C−C bonds. metathesis. Science 312, 257–261 (2006).
Neither H2 nor water is evolved. doi: 10.1126/science.1123787
The list of author affiliations is available in the full article online.

*Corresponding author. E-mail: david.milstein@weizmann.ac.il

Published by AAAS
REVIEW
intermediate using the hydrogen removed from
the starting compound. This method is also
Applications of Acceptorless
called “hydrogen autotransfer” (Fig. 2C); it does
not involve net hydrogen evolution, and the over-
all process is redox neutral. Dehydrogenation re-
Dehydrogenation and Related actions can also couple a redox pair such as an
alcohol and an alkene, to provide products of
Transformations in Chemical Synthesis formal alcohol C−H functionalization; upon al-

cohol dehydrogenation (in the presence of catalytic
base), the generated metal hydride interme-
Chidambaram Gunanathan1 and David Milstein2* diate adds to the alkene to give a nucleophilic
metal alkyl, followed by reaction of the latter
Conventional oxidations of organic compounds formally transfer hydrogen atoms from the with the intermediate keto compound to form a
substrate to an acceptor molecule such as oxygen, a metal oxide, or a sacrificial olefin. In C–C bond (Fig. 2D). AD reactions provide en-
acceptorless dehydrogenation (AD) reactions, catalytic scission of C−H, N−H, and/or O−H bonds vironmentally benign synthetic methodologies
liberates hydrogen gas with no need for a stoichiometric oxidant, thereby providing efficient, for the preparation of an assortment of useful
nonpolluting activation of substrates. In addition, the hydrogen gas is valuable in itself as a products, in addition to the generation of valuable
high-energy, clean fuel. Here, we review AD reactions selectively catalyzed by transition metal hydrogen gas. For example, amines and amides,

complexes, as well as related transformations that rely on intermediates derived from reversible which are traditionally prepared by multistep pro-
dehydrogenation. We delineate the methodologies evolving from this recent concept cesses using stoichiometric amounts of activating
and highlight the effect of these reactions on chemical synthesis. reagents, can be obtained in a single synthetic step
from alcohols with minimal waste
nticipation of fossil fuel de- by means of AD.
A pletion and growing envi-

ronmental concerns urge
chemists and chemical industries to
A
R XH
+ Oxidants
Metal salts R X
+
Copious
toxic
waste
Because X−H bonds (X = C, O,
or N) are ubiquitous among organic
molecules, selective AD followed by
search for alternative raw materials Additives further tandem functionalization can
and for atom-economical, environmen- Substrate Product provide a diverse stream of products.
tally benign synthetic methods. In this This strategy has been achieved most-
context, acceptorless dehydrogenation X = CH2, CH, NH, O ly as a result of advancement in the
(AD) reactions, in which hydrogen field of catalysis by transition metal
is liberated and new bonds prospec- B complexes. Dehydrogenation re-
tively generated by further reactions actions classified by the substrates
of the dehydrogenated products, are R XH
+
Catalyst R X
+
and catalysts have been reviewed
emerging as a powerful approach, (1–3). Here, we highlight AD reac-
circumventing the need for stoichi- Substrate Sacrificial Product Sacrificial tions catalyzed by soluble transition
ometric oxidants or prefunctional- hydrogen waste metal complexes from the perspec-
ization of substrates. acceptor tive of the strategies outlined in
Removal of hydrogen atoms from X = CH2, CH, NH, O Fig. 2, with particular emphasis on
adjacent atomic centers of a hydrogen- selective coupling reactions lead-
rich organic molecule is in most cases C ing to useful products efficiently
a thermodynamically unfavorable and atom-economically.
process. Thus, dehydrogenation of R XH Catalyst R X
organic compounds often requires stoi- + H2 Precursors to Modern AD
chiometric or excess molar amounts In organic synthesis, the oxidation/
Substrate Product Only byproduct
of oxidants such as oxygen, perox- (non-poluting,
dehydrogenation is carried out using
ides, iodates, and metal oxides (Fig. X = CH2, CH, NH, O valuable) conventional methods, which use stoi-
1A) or sacrificial hydrogen accep- chiometric amounts or excess of inor-
tors (Fig. 1B), leading to wasteful Fig. 1. Classes of dehydrogenation reaction. (A) Dehydrogenation/oxidation ganic oxidants such as chromium(IV)
by-product generation. In the more by conventional oxidants. (B) Hydrogen-transfer reactions. Liberated hydrogen reagents (4), pressurized oxygen (5),
atom-economical AD reaction, mo- binds to a sacrificial acceptor molecule. (C) AD. Dehydrogenation leads to lib- or peroxides, in addition to employ-
lecular hydrogen must be effectively eration of hydrogen gas, which is removed from the reaction mixture under ing various additives, cocatalysts,
removed from the reaction mixture reflux conditions or by vacuum. and catalytic systems combined with
to drive the equilibrium toward the metal complexes and TEMPO (2,2,6,6-
products (Fig. 1C). Alternatively, the liberated hy- Our group has developed a class of AD re- tetramethylpiperidinyl-1-oxy) that result in stoichi-
drogen can also be used in situ to hydrogenate actions in which the catalyst dehydrogenates ometric waste generation, which is undesirable
unsaturated intermediates generated from a con- both the starting compound and an intermediate environmentally and economically (6). In addi-
densation reaction. compound, leading to the net-oxidized product tion, pressurized oxygen and peroxides pose ex-
1
with liberation of two equivalents of hydrogen plosion hazards. To circumvent these problems,
School of Chemical Sciences, National Institute of Science Edu- (Fig. 2A). Reactions have also been developed dehydrogenation methods without use of conven-
2
cation and Research (NISER), Bhubaneswar 751005, India. De-
partment of Organic Chemistry, The Weizmann Institute of in which both liberation of hydrogen and elim- tional oxidants were developed. Early investigations
Science, Rehovot 76100, Israel. ination of water take place (Fig. 2B). In a related of AD emanated from heterogeneous catalysis.
*Corresponding author. E-mail: david.milstein@weizmann. class of reactions, termed the “borrowing hy- Dehydrogenation of linear primary alcohols re-
ac.il drogen” approach, the catalyst hydrogenates an sulted in b-branched primary alcohols as a result
www.sciencemag.org SCIENCE VOL 341 19 JULY 2013 1229712-1

REVIEW
of condensation of the intermediate aldehydes of pincer complexes in the dehydrogenation of AD of primary alcohols to yield the corre-
followed by dehydration and hydrogenation, as alcohols (21). However, complex 2 required activa- sponding aldehydes (Fig. 3, D and E) is less
reported in the late 1800s by Guerbet, much before tion by a base. The modified catalysts 3a and 3b common, as often ruthenium complexes are de-
hydrogen-transfer reactions were reported (7). The can catalyze the reaction under neutral conditions activated by decarbonylation of the aldehydes.
simple hydrogen-transfer reaction has its origin in using low loading of 0.1 mol % (22). The iridium Synthesis of aldehydes from alcohols was reported
the Oppenauer oxidation of secondary alcohols to complexes 4 (23) and 5 (24) also catalyze this recently by Fujita and Yamaguchi (26) using the
ketones in the presence of acetone, mediated by reaction. The catalytic activity of 5 is comparable iridium catalyst 6a. The modified water-soluble
aluminum tert-butoxide (8) and later catalyzed to that of ruthenium complex 3b, whereas com- catalyst 6b catalyzes the dehydrogenation of both
by transition metal complexes. Hydrogen transfer plex 4 shows higher efficiency. In addition to the secondary and primary alcohols in water. This
using alkanes as the hydrogen source is much more synthetic potential, the dehydrogenation of second- catalyst family operates by a mechanism involving
difficult due to the generally unreactive C−H bonds. ary alcohols to ketones is also of interest from the metal-ligand cooperation (27, 28).
In 1979, Crabtree achieved stoichiometric dehydro- point of view of hydrogen production from simple
genation of alkanes using a cationic iridium(III) and biorenewable alcohols. Combining the ruthe- Dehydrogenative Coupling of
metal complex (9) in the presence of a hydrogen nium precursor [RuH2(CO)(PPh3)3] [used earlier Alcohols to Form Esters
acceptor. Pioneering examples of catalytic alkane for this reaction (20)] with PNP-type pincer ligands Esterification is one of the most important fun-
hydrogen-transfer reactions by soluble complexes was shown to be highly effective in hydrogen pro- damental reactions in synthetic organic chem-
were independently reported by Felkin and col- duction from iso-propanol (25). istry, with applications in the production of an
leagues (10) and Crabtree and colleagues (11).

Alkane Dehydrogenation
In pioneering work, Aoki and Crabtree reported A B
O
AD of cyclooctane using [IrH2(O2CCF2CF3)(PCy3)] 1 Aromatic
R OH R R OH
as catalyst, which gave 36 turnovers to cyclooc- R N product
H
tene under reflux conditions (12). Stability of the
applied homogeneous catalysts at higher temper- Catalyst Catalyst
atures, which is essential for these reactions, se-
verely limited scope and efficiency. However, an Cyclization
2H2 H2 reactions
efficient AD reaction was achieved by Xu et al.
(13) using the Ir(PCP) pincer complex 1a (Fig.
3A). Later, the sterically less-crowded complex 1b Catalyst-H2 Catalyst-H2
developed by Liu and Goldman provided close
to 1000 turnovers in AD of cyclooctane; linear OH
R1
R O R O
alkanes were also dehydrogenated (14). The high R1 R N
R N
thermal stability of pincer complexes, coupled R1NH2 H
R1NH2 H2O Product or
with the effectiveness of iridium complexes in intermediate
C−H activation, resulted in catalysts 1a and 1b
being the most effective complexes for AD of C D
alkanes. OH
R1 R1
Alcohol Dehydrogenation R OH R N R OH R
H
Traditionally, alcohol oxidations are primarily
performed using toxic strong oxidants such as Catalyst Catalyst
periodates or chromium oxides, which generate + +
toxic stoichiometric waste (4, 6). Greener alterna- Base Base Carbonyl
addition
tives have also been developed (15). For example,
a method of TEMPO-catalyzed dehydrogenation
of alcohols with sodium hypochlorite has been Catalyst-H2 Catalyst-H Base-H
developed and commonly used in both small-
and large-scale applications (Fig. 3B). However, condensation Cat.
the method suffers from the need for a stoichio- R R1 + R
O R N R O R1 O
metric amount of sodium hypochlorite, the need R1
for a cocatalyst [e.g., 10 mole percent (mol %) NaBr] R1NH2 H2O
in addition to the use of chlorinated solvents, and
the equivalent amount of sodium chloride produced
Fig. 2. Dehydrogenation strategies in organic synthesis, exemplified by reactions of alcohols.
for every molecule of alcohol dehydrogenated
(A) Successive AD with release of hydrogen gas. The catalyst liberates H2 (the sole by-product) from both
(16). In contrast, several examples of oxidant- starting compound and intermediate generated by reaction with a nucleophilic substrate. “Catalyst-H2”
free catalytic AD of secondary alcohols to the indicates formal abstraction of two hydrogen atoms by the catalyst. (B) AD with hydrogen and water release.
corresponding ketones were reported (Fig. 3C) in An intermediate formed by dehydrogenation of the starting compound can couple with nucleophiles; the
which hydrogen gas is the only by-product. Early resulting products can be isolated or can undergo further addition or cyclization reactions with or without
examples required the presence of an acid as a further H2 liberation. (C) Borrowing hydrogen. The catalyst dehydrogenates the substrate at the outset and
hydride ion acceptor (17–20). The ruthenium PNP formally transfers the H atoms to an unsaturated intermediate. Hydrogen gas is not evolved, and the reaction
[2,6-bis-(di-tert-butylphosphinomethyl)pyridine] often involves elimination of water as a by-product. (D) Coupling of redox pairs. The catalyst dehydrogenates
pincer complex 2, reported by our group, catalyzes the substrate to generate an electrophile and metal-hydride; addition of the latter to an unsaturated
the dehydrogenation of secondary alcohols using substrate forms a nucleophilic metal alkyl that reacts with the electrophile to form a C–C bond. Neither
low catalyst loading, demonstrating the potential hydrogen gas nor water are produced.
1229712-2 19 JULY 2013 VOL 341 SCIENCE www.sciencemag.org

REVIEW
assortment of fine chemicals ranging from fra- efficiency (Fig. 4A) (29). The aromatic, coor- affords unsaturated, dearomatized complex 9, an
grances to pharmaceuticals. Elaborating on the dinatively saturated complexes 7 and 8 require excellent catalyst for dehydrogenative coupling
dehydrogenation of secondary alcohols catalyzed the presence of a catalytic amount of a base, for of alcohols to form esters with liberation of H2
by complex 2, rational design of the pincer com- in situ generation of the corresponding dearo- under neutral conditions. Only traces of alde-
plexes 7-9, based on metal-ligand cooperation matized complexes by deprotonation, which are hydes are formed.
involving aromatization-dearomatization se- the actual catalysts. Installing a hemilabile amine This catalytic reaction provides an efficient,
quences of the pyridine-based ligand (see Fig. 5G), arm, which can play an important role in the atom-economical, environmentally friendly path-
resulted in the direct catalytic dehydrogenative catalytic cycle by opening a coordination site, way for the synthesis of esters. It can be carried out
coupling of primary alcohols to esters with high resulted in precatalyst 8. Treatment with a base in a solvent or with neat liquid reagents. Previous
examples of dehydrogenative coupling of alcohols
to esters were considerably less efficient (30–32).
Shvo’s catalyst provided benzyl benzoate and pentyl
pentanoate from the corresponding neat alcohols at
137° to 145°C, although yields and reaction times
were not reported (31). Murahashi demonstrated
dehydrogenative coupling of various alcohols to
esters using RuH2(PPh3)3 (2 mol %) in refluxing
mesitylene (180°C) for 24 hours (32). The acridine
catalyst 10 prepared by our group also catalyzes

this transformation in the presence of a catalytic
amount of base in refluxing solvent or under neat
conditions (33). Complexes 3a and 3b (22), as
well as the PNS [2-((tert-butylthio)methyl)-6-((di-tert-
butylphosphino)methyl)pyridine] pincer ruthenium
complex [(PNS)RuHCl(CO)] (34), analogous to
complex 8, and the iridium complex 5 (24) also
catalyze the dehydrogenative esterification of al-
cohols. Catalytic conversion of ethanol, a biore-
newable alcohol, to ethyl acetate and molecular
hydrogen is of particular interest because ethyl
acetate is a widely used industrial bulk chemical.
In this context, further fine-tuning of steric and
electronic factors of the pincer backbone (35, 36)
and screening of known pincer complexes (37)
provided efficient catalysts.
Dehydrogenative cross-coupling of pri-
mary with secondary alcohols to form mixed
esters was achieved for an assortment of pri-
mary and secondary alcohols, in a molar ratio
of 1/2.5, using the bipyridine-based, dearom-
atized catalyst 12 (Fig. 4B) (38). In general,
transesterification (ester to ester transforma-
tion) is not an atom-economical process, because
it produces, in addition to the desired ester, an
equivalent amount of alcohol. Our group has
developed a distinct mode of transesterifica-
tion, in which hydrogen gas is formed as a by-
product, rather than alcohols, upon reaction
of esters with secondary alcohols catalyzed by
complex 9 (Fig. 4C). When symmetrical esters
(i.e., having the same R groups) are used, both
the acyl and alkoxy fragments of the substrate
ester are incorporated into the product ester
with liberation of hydrogen (39). This cross-
selectivity is a result of slower dehydrogenation
of the secondary alcohol to the corresponding
ketone as compared with the dehydrogenative
coupling of the primary alcohol to ester. Diols
can undergo an intramolecular reaction to pro-
vide the corresponding lactones with hydro-
Fig. 3. Examples of dehydrogenation. (A) Dehydrogenation of an alkane catalyzed by iridium pincer gen liberation (22) with complex 3b as catalyst
complexes. (B) Dehydrogenation of primary and secondary alcohols by conventional oxidants. (C) De- (Fig. 4D).
hydrogenation of secondary alcohols by well-defined ruthenium and iridium complexes. (D and E) Dehy- Currently practiced methods for the synthesis
drogenation of primary alcohols. (D) Synthesis of aldehydes. (E) Synthesis of aldehydes and ketones in of polyesters are generally not atom-economical and
water solution. not environmentally benign. They are normally

REVIEW
based on carboxylic acid derivatives, prepared
using toxic reagents and generating toxic by-
products and salts; subsequent polycondensa-
tion with diols also generates salt waste, which A
can be challenging to remove from viscous poly- H
Catalyst O N PiPr2
mer solutions, often resulting in low conversion 2 R OH + 2H2
and poor polymer properties. By using the in Neat or solvent R O R Ru
reflux i
situ–generated complex 9 (from commercially Pr2P
Cl
CO
available complex 8) as a catalyst, Robertson and 7
colleagues have demonstrated a remarkable pro-
cess (40) for the synthesis of polyesters from O
7-10 (0.1 mol%)
diols (Fig. 4E). Efficient removal of the generated OH + 2H2
O H
hydrogen was achieved by performing the polym- N PtBu2
erization reaction under reduced pressure, re- Ru
sulting in formation of high-molecular-weight 7/KOH : neat, 24 h, 67% Et2N CO
8/KOH : toluene, 24 h, 95% Cl
polyesters. 9 : toluene, 6 h, 99% 8
Very recently, primary alcohols were oxidized 10/KOH : neat, 26 h, 92%
by our group to the corresponding carboxylic
acid salts using water as the terminal oxidant,

with liberation of hydrogen (41). The precata- B N H
PtBu2
lyst 11 was used for the in situ generation of
Ru
catalyst 12 (Fig. 4F), which catalyzes this trans- OH 12 (1 mol%) O R2 Et2N CO
formation under acceptorless conditions. Inter- R1 OH + + 2H2
estingly, water plays the role of both oxygen R2 R3 Toluene, reflux R O R 1
9
donor and reaction medium. Complex 9 is also 46-99%
known to stoichiometrically split water, by con-
secutive thermal H2 and light-induced O2 gen-
eration (42). C N H PiPr2
Dehydrogenative Coupling of Alcohols OH O 9 (1 mol%) O R2 Ru

2 + 2 + 2H2 i
Pr2P
with Amines to Form Amides R1 2 Toluene, reflux CO
R R O R R O R1
Intermolecular dehydrogenative coupling of al- Cl
39-95% 10
cohols with amines is the most atom-economical
method for amide synthesis (43). However, this
catalytic reaction was difficult to envision, since D
hemiaminal formation was expected to follow N H
R
O PtBu2
alcohol dehydrogenation. Subsequent spontane- 3b (0.3 mol%)
O + 2H2 Ru
ous water elimination would then form an imine HO n OH Toluene, reflux Et2N
n H
that could undergo hydrogenation with the lib- n = 2-3
R H
BH2
erated H2 to yield a secondary amine (from pri- n = 1-2 3b
mary amine reagents). We discovered that this 33-90%
transformation (44) could indeed be catalyzed
by the dearomatized pincer complex 9 (Fig. 5A). E
A range of alcohols reacted with alkyl and aryl N H
O
O PtBu2
amines and diamines to produce amides with 9 (0.2 mol%)
HO n OH ± + mH2 Ru
liberation of hydrogen under low catalyst load- O O N
n CO
n = 1-8 n Cl
ing (0.1 mol %). Moreover, the reactions are se-
n= ≥4 n=2
lective toward the primary amine functionality 11
70-97%
under these conditions. Following this discov-
ery, a number of catalytic systems were reported
for this transformation with various ruthenium F
precursors in combination with carbene and/or H
11 (0.2 mol%) O N
phosphine ligands, although these systems re- PtBu2
R OH + OH + 2H2
quired higher loading of metal complex and lig- H2O, reflux R O Ru
N CO
ands in addition to the need for substoichiometric 61-100%
amounts of base; selected examples (45–47) are
given in Fig. 5B. 12
Synthesis of amides from esters and amines
is also a potentially attractive method. Complex Fig. 4. Direct synthesis of esters from alcohols. (A) Synthesis of esters by dehydrogenative cou-
9 efficiently catalyzes this transformation (Fig. 5C) pling of primary alcohols. These reactions are catalyzed by the rationally designed dearomatized,
with liberation of H2 under neutral conditions unsaturated PNN pincer complex 9, and its precursor 8, as well as the less reactive PNP complex 7. (B
(48). Similar to the alcohol acylation process to F) Application of alcohol dehydrogenative coupling reactions under acceptorless conditions. (B)
(Fig. 4C), both the acyl and alkoxo fragments of Synthesis of esters from cross-coupled primary and secondary alcohols. (C) Synthesis of mixed esters by
the symmetrical esters are incorporated in the transesterification of esters using secondary alcohols. (D) Synthesis of lactones from diols. (E) Synthesis
amide product. One outcome of this reaction is of polyesters from diols. (F) Synthesis of carboxylic acid salts from alcohols using water.

REVIEW
that ethyl acetate, a cheap and abundant ester, can
be used as a convenient, atom-economical acetyla-
A tion agent of amines, producing hydrogen as the
O only by-product; this is clearly advantageous
9 (0.1 mol%)
+ R2NH2 R2 + 2H2 N H over the commonly employed acetylation agents.
R1 OH 1 PtBu2
Toluene, reflux R N
H Ru Amino alcohols also participate in the acylation
7-12 h
R1, R2 = alkyl, aryl 58-99% Et2N of amines. Gratifyingly, chiral amino-alcohols
CO
react with retention of configuration (Fig. 5D), a
9 likely attribute of the neutral reaction conditions
(49). As complex 9 catalyzes the amidation of
B amines using amino-alcohols, we reasoned that use
[Ru(COD)Cl2] (5 mol%) Ph2 H2N of amino-alcohols alone might result in formation
P
Cl
of linear or cyclic peptides. Indeed, complex 9
Cl (5 mol%) Ru i Pr
Fe (2-4 mol%) Ru Cl (5 mol%) catalyzes the conversion of various amino-alcohols
i Pr N N i Pr N
P
N
Cl (bearing substituents larger than methyl a to the
Ph Cl N
PCyp3 HBF4 (5 mol%)
2
i Pr amine group) to the corresponding cyclic dipep-
13 tides (diketopiperazines) as the only products in
KOH (10-15 mol%)
KOtBu (20 mol%) KOtBu (15-35 mol%)
very good yields with liberation of H2 (Fig. 5E).
Toluene, 110 °C, 24 h Neat, 125 C, 3.5 h Toluene, reflux, 24 h In the case of alaninol, oligopeptides were formed.

Using catalyst 9, we have also developed the cat-
Yields of amides: 21-100% 42-89% 40-92%
alytic synthesis of polyamides (50) from diols and
diamines. Before we published this work, Zeng
C and Guan reported (51) the direct polyamidation
R1 O reaction using the now commercially available
O 9 (0.1 mol%) 1
2 NH + 2 R N R3 +
2H2 catalyst 9. Optimization studies by them revealed
R2 R3 O R3 Toluene or benzene
reflux R2 the need for polar solvents for successful polym-
52-99% erization of diols and diamines, anisole being a
suitable polar solvent, resulting in high number-
average molecular weights (Mn) of the polyamides
D
O (3.2 kD in toluene versus 13.8 kD in anisole). Mn
H2N
OH H2N of the polyamides were further improved (22.6 kD)
NH2 9 (1 mol%) N
+ H + 2H2 by the addition of small amounts of dimethyl sulf-
Toluene, reflux, 6h oxide. Polymers bearing secondary amine groups
58% in the backbone, potentially useful for gene de-
(no recemization) livery, were obtained by Zeng and Guan (51), with
no need for a wasteful protection-deprotection se-
E R quence, due to the selectivity of 9 toward the
R 9 (1 mol%) O amidation of primary amine groups (Fig. 5F).
HN
2 HO
NH2 1,4-dioxane NH
+ 4H2 Ru-pincer complexes (7 to 12) developed by our
reflux, 19 h O group are believed to operate by a mode of metal-
R ligand cooperation (Fig. 5G) involving aromatization-
64-92% dearomatization of the pincer ligand (52, 53).
Dearomatized pincer complexes can activate pri-
F mary alcohols, yielding the corresponding saturated
hydrido alkoxo complexes, with aromatization. The
n HO R OH O O
9 (1 mol%) mechanism of further dehydrogenation of the
+
120 oC, N2 flow R N R' N n
+ 2(2n+1)H2 alkoxy ligand, which could follow b-hydride elim-
n H2N R' NH2 Anisole or H H ination, remains unclear at this stage (54). How-
anisole/DMSO (4:1) 65-89% ever, esterification and amidation reactions likely
proceed through hemiacetal and hemiaminal in-
H termediates, respectively, formed by nucleophilic
G L1 attack by the alcohol or amine on an intermediate
L1 H Y
aldehyde that is either coordinated to the metal or
N MLn
Y = H, OH, OR
N MLm Y free in solution (55, 56). Catalyst 9, and its bipyri-
NH2, NR2, C L2
dine analog 12, also effectively catalyze under mild
L2
conditions the hydrogenolysis reactions of esters
Fig. 5. Direct synthesis of amides from alcohols and amines. (A) Discovery of dehydrogenative to alcohols (57), amides to alcohols and amines
coupling of alcohols with amines to form amides with liberation of hydrogen, catalyzed by the pincer (58), and the hydrogenation of the CO2-derived
complex 9. (B) Selected examples of other catalytic systems developed later for the same transfor- organic carbonates and formates as mild, green,
mation. (C to F) Applications of catalyst 9 in acceptorless dehydrogenative coupling processes involving two-step routes to methanol (when dimethyl car-
amines. (C) Synthesis of amides from esters and amines. (D) Synthesis of a chiral amide from (S)-2-amino-3- bonate or methyl formate are used) (59). Methyl
phenylpropan-1-ol and benzylamine. (E) Synthesis of cyclic dipeptides from b-amino-alcohols. (F) Synthesis carbamates and urea derivatives were also hydro-
of polyamides from diols and diamines. (G) Metal-ligand cooperation by facile aromatization and de- genated to alcohols and amines under mild con-
aromatization sequences (highlighted in blue). The dearomatized ligand participates in various bond ditions (60). Density functional theory (DFT)
activation and reversible bond-formation reactions and plays a key role in catalysis. calculations carried out by other groups on our

REVIEW
amidation reaction (55, 56) and on the microscopic alyzed by the bipyridine-based RuPNN complex catalyst (18) in the second step. The hydrogen ob-
reverse, hydrogenation of amides (61), further sup- 11, following a similar mechanism, as reported by tained in the first step hydrogenates the alkenes,
port the involvement of suggested intermediates. our group very recently (71). thus providing a near thermodynamically neutral
process (Fig. 7). Although the terminal alkene is
Dehydrogenative Coupling with Concomitant Alkane Metathesis the kinetic dehydrogenation product, isomeriza-
Condensation Reactions Combining alkane dehydrogenation and the tion of the double bond along the chain and then
The acridine pincer ruthenium catalyst 10 cat- borrowing approach, Goldman, Brookhart and cross metathesis of the isomerized products lead
alyzes the conversion of alcohols to acetals in co-workers achieved alkane metathesis in a three- to a broad distribution of hydrocarbons.
very good yields, liberating hydrogen and water step tandem sequence using two different catalysts
(Fig. 6A) (33, 62). The reaction proceeds via enol- (72). The iridium complex (17) dehydrogenates Alkylation of Amines by Borrowing
ether intermediates, which upon further alcohol alkanes in the first step to produce the respective Hydrogen Methodology
addition yield acetals. The recently developed ru- alkenes, which are converted into either longer or The limited reactivity of alcohols toward nu-
thenium complex 14 also catalyzes this transfor- shorter alkenes by the Mo-based Schrock metathesis cleophiles can be readily overcome upon de-
mation (63).
The RuPNP pincer complex 15 catalyzes
the dehydrogenative coupling of alcohols with
amines, unexpectedly leading to imine products A
(Fig. 6B) rather than to amides, as catalyzed by the
N
analogous RuPNN complex 9 [PNN is (2-(di-tert- Ph3P O O

O Ru S
butylphosphinomethyl)-6-diethylaminomethyl)pyridine] 10 (0.1 mol%), 137 ° C R R
3R OH + H2 + H2O Ph3P O O
(64). It is possible that in the case of complex 15, or 14 (0.1 mol%), 72 h O
R
N
intermediate aldehyde dissociates from the metal Toluene, reflux
14
complex, forming free hemiaminal in solution,
whereas in the case of 9, a coordinated aldehyde is
attacked by the amine and no free hemiaminal is B
involved. The reason for the discrepancy might be 15 (0.2 mol%)
R1 OH + R2 NH2 R1 N R2 + H2 + H2O
a lack of the hemilabile amine “arm” and higher Toluene, reflux
steric hindrance in 15. The free hemiaminal elimi- 22-56 h
R1, R2 = alkyl, aryl 57-92%
nates water, providing a method for the synthesis N H
PtBu2
of imines, with no subsequent hydrogenation to
Ru
the corresponding amines. DFT calculations on R tBu
C 2P CO
imine formation catalyzed by 15 are in line with R 15 (1 mol%) N
the observed selectivity (65). An analogous RuPNP 2 HO + 3H2 + 2H2O 15
NH2 Toluene N
catalyst with amine-based pincer “arms” (66), a
165 ° C, 24 h
Ru-carbene complex (67), and an OsPOP [POP R
is 4,6-bis(diisopropylphosphino)dibenzofuran] 35-53%
complex (68) were also later reported to catalyze
this transformation. Using amino alcohols, com-
D
plex 15 catalyzes formation of pyrazines (Fig. 6C).
OH
Apparently, the reaction proceeds through a cyclic- 13 (1.5 mol%)
+ RNH2 N + 2H2 + 2H2O
diimine intermediate, which undergoes further NaOOCH, neat
OH R
dehydrogenation to provide the aromatic pyrazines 125° C, 16 h 45-48%
(49). Dehydrogenative coupling of diols and amines
can lead to pyrroles. Thus, reaction of 2,5-hexandiol
and alkylamines catalyzed by complex 13 in the Ru3(CO)12 (1 mol%)
E O Xantphos (3 mol%) R2
presence of sodium formate resulted in forma-
R2
+ R3NH2 + HO + 3H2O
tion of N-alkyl-2,5-dimethylpyrroles in a dehy- R1
OH
K2CO3 (20 mol%)
N
R3
R1
drogenative Paal-Knorr pyrrole synthesis (Fig. t amyl alcohol
6D) (46). Very recently, synthetic approaches to 130 °C, 18 h 30-90%
pyrroles were reported, based on alcohol dehy-
drogenation, imine formation, and base-promoted
condensation. Thus, Ru3(CO)12 with an added di- F Ph
phosphine and a base catalyzes the reaction of 16 (0.1-0.03 mol%)
N N
1,2 diols with amines and ketones, resulting in a KOtBu (1.1 equiv)
OH R2 NH2 THF, 90° C, 24 h R2 HN N N
variety of functionalized pyrroles (Fig. 6E) (69). + + 2H2 + 2H2O i
Pr2P PiPr2
The IrPNP complex 16, developed by Michlik R1 or 11 (0.5 mol%)
NH Ir
OH R1
and Kempe, in the presence of base catalyzes KOtBu (0.5 equiv)
the dehydrogenative coupling of b-amino-alcohols Toluene, reflux, 24 h
with secondary alcohols to form pyrroles; the pro- 16
posed mechanism involves ketimine formation
from the ketone and amino-alcohol, followed by
Ir-catalyzed dehydrogenation and base-promoted Fig. 6. Acceptorless dehydrogenative coupling reactions that proceed with loss of water. (A)
condensation to result in elegant synthesis of pyr- Direct synthesis of acetals from alcohols. (B) Synthesis of imines from alcohols and amines. (C) Synthesis of
roles in very good yields with diverse substituents pyrazines from amino-alcohols. (D) Synthesis of pyrroles from diols and amines. (E) Three-component
(Fig. 6F) (70). This reaction is also efficiently cat- synthesis of pyrroles. (F) Synthesis of pyrroles from alcohols and amino-alcohols.

REVIEW
hydrogenation to the corresponding carbonyl
A B compounds, which are amenable to nucleophil-
17a or 17b, and 18 Mixture of C2-C21
ic addition reactions (Fig. 2) (73). Nitrogen-
linear alkanes containing compounds, ranging from primary
R Neat, 125 °C
R R amines to heterocycles, were obtained using the
Catalyst
metal-catalyzed, alcohol-borrowing hydrogen
pathway (Fig. 8) (74). In pioneering work,
X PtBu2
Grigg et al. reported that [RhH(PPh3)4] catalyzed
Ir CF3 N-alkylation of amines by alcohols (Fig. 8C)
Catalyst-H2 O
N
Mo
(75). Murahashi et al. and Tsuji et al. reported
X PtBu2 CF3 catalysis by ruthenium complexes for the prepa-
Metathesis O
R R
R Ph ration of a range of secondary and tertiary amines
F 3C CF3
R
17a: X = CH2 (76), including indoles (Fig. 8E) (77). Selective
17b: X = O 18, Schrock catalyst synthesis of primary amines from ammonia and
electrophiles is a challenging task, because the
Fig. 7. Alkane metathesis via borrowing methodology. (A) General strategy. Catalytic C–H activation is coupled primary amine intermediate is more nucleophilic
with catalytic olefin metathesis for net alkane metathesis. (B) Metathesis of n-hexane by 17a or 17b and 18. than ammonia, and it undergoes competing al-
kylation reactions when conventional alkylating
reagents such as alkyl halides are used, result-

ing in a mixture of products. Alkylation of amines
R1 [M] R1 R3NH2 R1 [MH2] R1 by the borrowing hydrogen pathway can cir-
OH O NR3 NHR3 + H2O
R2 R2 R2 R2 cumvent this problem, and selectivity could be
R1,R2,R3 = H, alkyl, aryl Imine reached by a suitable choice of ligands and cat-
alyst design. For example, primary amines were
N H P selectively synthesized by our group from pri-
A mary alcohols and ammonia, using the acridine-
Ru derived ruthenium pincer complex 10 under mild
OH 10 (0.1 mol%) NH2 P conditions and low catalyst loading (Fig. 8A);
+ NH3 CO
Water, 135 °C, 18 h Cl selectivity in this case is enhanced by the steric
(7.5 atm) 95% 10 bulk around the metal center, lending preference
to ammonia coordination. Reactions can also be
B performed using water as reaction medium, re-
sulting in enhanced selectivity (78). Recently, 10
[Cp*IrCl2]2 (1-5%) was also used as a catalyst for the preparation
OH + NH4BF4 N
NaHCO3 (30 mol%) H of diamines from the diols derived from vegeta-
140 °C, 17 h 98% ble oils (79). Secondary alcohols could also be
employed using different ligands and metal pre-
cursors (80, 81). Secondary and tertiary amines
were selectively obtained from ammonium salts
R1 [M] R1 R3NHR4 R1 R3 [MH2] R1 R3 (Fig. 8B) and primary alcohols using an iridium
OH O N N + H2O
R2 R2 R2 R4 R2 R4
complex by Fujita and colleagues (82, 83) and
Eary and Clausen (84). Beller and co-workers
R1 = H, alkyl, aryl 2 3 4
R ,R ,R = alkyl Iminium ion
developed phosphine ligands that, in combina-
tion with Ru(0), were effective catalysts for the
C synthesis of tertiary amines (Fig. 8D) (85). In
[RhH(PPh3)4 (5 mol%)
general, ruthenium and iridium complexes were
OH + NH found to be good catalysts for alkylation of amines
N
Reflux, 6 h
as well as amides (86). Whereas reaction of am-
74%
monia and primary amines with carbonyl com-
pounds proceeds by imine intermediates, use of
D secondary amines leads to the formation of imi-
Ru3(CO)12 (2 mol%)
nium ion intermediate. Both imine and iminium
OH + N PCy2 ion intermediates are hydrogenated by the cat-
N L (6 mol%), tamyl alcohol L= N
H alyst using the hydrogen obtained from starting
140° C, 24 h Ph
97%
alcohols to deliver the primary or secondary and
tertiary amines, respectively.
Williams and co-workers have demonstrated
E that borrowing hydrogen tactics can be applied
OH NH2 [RuH2(PPh3)3 (2.5 mol%) N N in alkylation processes often used in the syn-
+ +
OH thesis of drugs (Fig. 9A); on the laboratory scale,
Neat, 20 h, 155 °C
they obtained various pharmaceuticals employing
40% 20%
alcohols in place of conventional alkyl halides (87).
Berliner reported the synthesis of PF 03463275,
Fig. 8. Alkylation of amines using alcohols. Products targeted include (A) primary amines, (B) a GlyT1 inhibitor developed for the treatment
secondary amines, (C and D) tertiary amines, and (E) heterocycles, all with very good selectivities. of schizophrenia, by applying the borrowing-

REVIEW
hydrogen methodology on a multikilogram scale b-alkylation of alcohols was achieved using irid- reactions. Very recently, Krische and co-workers
(88), with (Cp*IrCl2)2 as catalyst (Fig. 9B). The ium and ruthenium catalysts (Fig. 10, E and F) also uncovered an alternative mechanism for such
strategic advantage of the borrowing hydrogen (97, 98). In addition, upon dehydrogenation, amines C−C bond formation (100). Excellent stereose-
methodology was also applied in the synthesis of can also give rise to electrophilic reactivity and lectivities were achieved in several transforma-
natural products such as noranabasamine (89), can undergo self-coupling or coupling reactions tions; for example, a catalyst generated in situ
isolated from the dart frog (Fig. 9C). A combina- with other amines (74). from RuH2(CO)(PPh3)3, a diphosphine and a chiral
tion of RuHCl(CO)(PPh3)3 and xantphos ligand Unlike the borrowing-hydrogen strategies de- acid catalyzes anti-diastereoselective and enantio-
was used by Beller and co-workers for the se- scribed in Fig. 10, A and B, C−C coupling can selective C−H crotylation of primary alcohols
lective diamination of isosorbide, which is ob- also be achieved without using preformed nucleo- (Fig. 10G) (101).
tained from D-glucose (81). The versatility of the philes. Bower and Krische developed reactions
alkylation of amines by borrowing-hydrogen that involve alcohols and partially unsaturated Outlook
methodology allowed the preparation of pri- substrates such as alkenes, dienes, alkynes, and AD is a rapidly growing area, propelled by the
mary, secondary, and tertiary amines from al- allenes (99), which result in products of formal profound influence of fundamental organome-
cohols, including biomass-derived alcohols alcohol a-C−H functionalization (Fig. 2D). This tallic chemistry, in part based on metal-ligand
(3, 85). Because the methodology is already strategy involves alcohol dehydrogenation to cooperation. This has led to reactions such as
being adopted in large-scale synthesis and tol- generate a metal-hydride intermediate that adds the dehydrogenative coupling of amines with al-
erates various functional groups, it is well on its to the unsaturated substrate, generating a nucleo- cohols to form amides, peptides, and polyamides
way to displacing the conventional alkylation philic intermediate capable of aldehyde addition under neutral conditions with liberation of hy-
reactions in organic synthesis that rely on alkyl

halides.
Alcohols as a Source of Electrophiles A

and Nucleophiles [Ru(p-cymene)Cl2]2 (1.25 mol%)
N N O
Construction of C−C bonds through the borrowing- OH
dppf (2.5 mol%)
hydrogen concept has been achieved using two + N N O
N N O Toluene, reflux, 24 h
different approaches, in which the alcohols are NH O N
87%
modified to manifest either electrophilic or nu- Piribedil
cleophilic reactivity. Upon alcohol dehydrogena-
tion, the generated carbonyl compounds can act
as electrophiles and undergo coupling reactions [Ru(p-cymene)Cl2]2 (2.5 mol%)
with nucleophiles to generate unsaturated inter- Ar OH DPEphos (5 mol%) Ar N
N N
mediates; further hydrogenation by using hydro- + HNMe2
Toluene, reflux, 24 h
gen borrowed from the alcohols in the first step Ph Ph
provides the product of the net redox-neutral Ar = Ph, Antergan, 75%
[Ru(p-cymene)Cl2]2 (2.5 mol%)
tandem process (Fig. 10A). In contrast, dehydro- 70%
Ar = 2-Py, Tripelennamine, 75%
DPEphos (5 mol%)
genation of secondary alcohols in the presence of Toluene, reflux, 24 h
a base can turn the resulting electrophilic carbonyl Ar
compounds into nucleophilic enolates, which can NH OH
+ HO
react with electrophiles through the b-carbon center Ph
(Fig. 10B). Like amine alkylation by borrowing-
hydrogen methods (C−N bond formation), several
different types of carbon nucleophiles can be used B F
F
for C−C bond formation by this strategy. Grigg HO H2N
reported early examples of C−C bond formation [Cp*IrCl2]2 Cl N
Cl
through borrowing hydrogen in which aryl aceto- (0.033 mol%) NH
H H + N
2HCl N
nitriles were alkylated by primary alcohols using 0.2 l/kg toluene
Cl H H O
ruthenium and rhodium catalysts and a stoichio- N 110 °C, 40 h H H
metric base (90); other metal catalysts and nu- F then HCl/IPA
2.4 kg filter, 76% N
cleophiles were employed later (91). Williams N
TON > 1500
and co-workers devised efficient alkylation meth-
4.8 kg PF 03463275
ods of carbon nucleophiles catalyzed by ruthe-
nium and iridium complexes using low catalyst
loading. For example, an active methylene com-
pound was alkylated (92) by benzyl alcohol C
N
using RuH2(CO)(PPh3)3 and xantphos ligand in O
very good yield (Fig. 10C). Recently, oxindole [Cp*IrCl2]2 (1 mol%)
N
was also alkylated by the borrowing-hydrogen Ph KOAc (1 mol%) N
+ N
method (93). Use of [Ir(cod)Cl]2 + PPh3 al- NH2 Toluene, 110 °C, 17 h
lowed Obora and Ishii to perform such reactions HO O N Ph
72%, 92% de
without an additional base at elevated tempera- HN
HO
ture (94). The borrowing-hydrogen approach
can also be applied to the Wittig reaction in which Noranabasamine
the alcohol functions as a surrogate to alde-
hydes (Fig. 10D) (95). Amine products can also Fig. 9. Application of the borrowing hydrogen methodology. (A) Laboratory synthesis of various
be obtained from the aza-Wittig reaction (96). pharmaceuticals. (B) Demonstration on multi-Kg scale. (C) Application in total synthesis of a natural product.

REVIEW
drogen gas and no waste generation. The related useful products. The strides made thus far exem- renewable molecules, we feel that many addi-
dehydrogenation reactions that do not evolve plify the power of dehydrogenation as an activation tional useful applications are bound to unfold.
hydrogen gas—namely the borrowing-hydrogen tactic by generation of more reactive unsaturated
methodology and the coupling of redox pairs intermediate compounds, which can then couple References and Notes
1. G. E. Dobereiner, R. H. Crabtree, Dehydrogenation as
via intermediate generation of nucleophiles and in situ with other substrates and provide effec- a substrate-activating strategy in homogeneous
electrophiles—allow construction of both C−N tive catalytic processes under mild, environmen- transition-metal catalysis. Chem. Rev. 110, 681–703
and C−C bonds from alcohols and provide effi- tally benign conditions. As AD is increasingly www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db=
cient, atom-economical access to an assortment of applied in transformations of complex and bio- PubMed&list_uids=19938813&dopt=Abstract (2010).
doi: 10.1021/cr900202j
2. J. Choi, A. H. MacArthur, M. Brookhart, A. S. Goldman,
Dehydrogenation and related reactions catalyzed by
iridium pincer complexes. Chem. Rev. 111, 1761–1779
A B www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=
OH R3 OH OH Retrieve&db=PubMed&list_uids=21391566&dopt=
E Abstract (2011). doi: 10.1021/cr1003503
R1 R2 R R 3. A. C. Marr, Organometallic hydrogen transfer and
R1 R2
dehydrogenation catalysts for the conversion of
Catalyst bio-renewable alcohols. Catal. Sci. Technol. 2, 279
Catalyst
(2012). doi: 10.1039/c1cy00338k
4. J. Muzart, Chromium-catalyzed oxidations in organic
Base synthesis. Chem. Rev. 92, 113–140 (1992).
Catalyst-H2
Catalyst-H2 doi: 10.1021/cr00009a005

5. T. Punniyamurthy, S. Velusamy, J. Iqbal, Recent
R3 advances in transition metal catalyzed oxidation of
O Carbon O O
organic substrates with molecular oxygen. Chem. Rev.
nucleophile E+, Electrophile E
R1 R2 R1 R2 R R 105, 2329–2364 www.ncbi.nlm.nih.gov/entrez/query.
Electrophile Nucleophile fcgi?cmd=Retrieve&db=PubMed&list_uids=
15941216&dopt=Abstract (2005). doi: 10.1021/cr050523v
R3NuH H2O
6. G. Tojo, M. Fernández, Oxidation of Alcohols to
Aldehydes and Ketones: A Guide to Current Common
Practice (Springer Science, New York, 2007).
C [RuH2(CO)(PPh 3)3] (0.5 mol%) t
Bu 7. M. C. R. Guerbet, Acad. Sci. Paris 128, 1002 (1899).
N
xantphos (0.5 mol%) 8. R. V. Oppenauer, Eine methode der dehydrierung von
OH O sekundären alkoholen zu ketonen. I. Zur herstellung von
+ O Piperidium acetate (5 mol%) sterinketonen und sexualhormonen. Recl. Trav. Chim. Pays
t
Bu Toluene, 110 ° C, 3 h N Bas 56, 137–144 (1937). doi: 10.1002/recl.19370560206
9. R. H. Crabtree, J. M. Mihelcic, J. M. Quirk, Iridium
94% complexes in alkane dehydrogenation. J. Am. Chem. Soc.
101, 7738–7740 (1979). doi: 10.1021/ja00520a030
D 10. D. Baudry, M. Ephritikhine, H. Felkin, R. Holmes-Smith,
[Ir(cod)Cl]2 (5 mol%) The selective catalytic conversion of cycloalkanes into
OH
PPh3 dppp (5 mol%) CO2Bn cycloalkenes using a soluble rhenium polyhydride system.
+ J. Chem. Soc. Chem. Commun. 1983, 788 (1983).
CO2Bn Cs2CO3 (5 mol%) doi: 10.1039/c39830000788
Toluene, 150 ° C, 72 h 79% 11. M. J. Burk, R. H. Crabtree, C. P. Parnell, R. J. Uriarte,
Selective stoichiometric and catalytic carbon-hydrogen
bond cleavage reactions in hydrocarbons by iridium
E complexes. Organometallics 3, 816–817 (1984).
[Cp*IrCl 2 ] 2 (2 mol%) doi: 10.1021/om00083a034
OH NaOtBu (1 equiv) 12. T. Aoki, R. H. Crabtree, Homogeneous tungsten, rhenium,
+ and iridium catalysts in alkane dehydrogenation driven
Toluene, 110 ° C, 17 h by reflux of substrate or of cosolvent or by inert-gas
OH OH 78% flow. Organometallics 12, 294–298 (1993).
doi: 10.1021/om00026a013
13. W. Xu et al., Thermochemical alkane dehydrogenation
F catalyzed in solution without the use of a hydrogen
OH Ru(dmso)4Cl2 (2 mol%) acceptor. Chem. Commun. (Camb.) 1997, 2273–2274
OH (1997). doi: 10.1039/a705105k
KOH (2 equiv) O
+ 14. F. Liu, A. S. Goldman, Efficient thermochemical alkane
O Ph
Dioxane, 100 ° C, 7 days dehydrogenation and isomerization catalyzed by an
OH 98% iridium pincer complex. Chem. Commun. (Camb.) 1999,
655–656 (1999). doi: 10.1039/a900631a
15. L. Que Jr., W. B. Tolman, Biologically inspired oxidation
catalysis. Nature 455, 333–340 www.ncbi.nlm.
G [RuH2(CO)(PPh3)3] (5 mol%) nih.gov/entrez/query.fcgi?cmd=Retrieve&db=
OH
OH dppf (5 mol%) PubMed&list_uids=18800132&dopt=Abstract (2008).
+ doi: 10.1038/nature07371
Chiral acid (10 mol%) 16. R. A. Sheldon, I. W. C. E. Arends, G. J. Ten Brink,
THF, 95° C A. Dijksman, Green, catalytic oxidations of alcohols.
86 % Acc. Chem. Res. 35, 774–781 www.ncbi.nlm.nih.gov/entrez/
8:1 dr; 95:5 er
query.fcgi?cmd=Retrieve&db=PubMed&list_uids=
12234207&dopt=Abstract (2002). doi: 10.1021/ar010075n
Fig. 10. C−C bond formation using alcohols as sources of electrophiles and nucleophiles. (A) 17. H. B. Charman, Hydride transfer reactions catalysed by
Borrowing hydrogen strategy for C−C bond formation, in which the alcohol transformed into an electrophilic rhodium-tin complexes. J. Chem. Soc. B 1970, 584
intermediate. (B) Borrowing hydrogen strategy for b-alkylation, in which the alcohol is transformed into a (1970). doi: 10.1039/j29700000584
18. A. Dobson, S. D. Robinson, Catalytic dehydrogenation of
nucleophilic intermediate. (C and D) Typical examples of the strategy shown in (A). (E and F) Typical primary and secondary alcohols by Ru(OCOCF3)2(CO)(PPh3)2.
examples of the strategy shown in (B). (G) A nucleophilic intermediate is generated from the alkene: J. Organomet. Chem. 87, C52 (1975). doi: 10.1016/S0022-
enantioselective C−H functionalization of primary alcohols. 328X(00)88159-0

REVIEW
19. D. Morton, D. J. Cole-Hamilton, Rapid thermal hydrogen 33. C. Gunanathan, L. J. W. Shimon, D. Milstein, Direct from alcohols and secondary amines: Involvement of
production from alcohols catalysed by [Rh(2,2′-bipyridyl)2]Cl. conversion of alcohols to acetals and H2 catalyzed by esters. J. Org. Chem. 76, 10005–10010 www.ncbi.
J. Chem. Soc. Chem. Commun. 1987, 248 (1987). an acridine-based ruthenium pincer complex. J. Am. nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db=
doi: 10.1039/c39870000248 Chem. Soc. 131, 3146–3147 www.ncbi.nlm.nih.gov/ PubMed&list_uids=22084902&dopt=Abstract (2011).
20. G. B. W. L. Ligthart et al., Highly sustainable catalytic entrez/query.fcgi?cmd=Retrieve&db=PubMed&list_ doi: 10.1021/jo201756z
dehydrogenation of alcohols with evolution of hydrogen uids=19216551&dopt=Abstract (2009). 48. B. Gnanaprakasam, D. Milstein, Synthesis of amides from
gas. Tetrahedron Lett. 44, 1507–1509 (2003). doi: 10.1021/ja808893g esters and amines with liberation of H2 under neutral
doi: 10.1016/S0040-4039(02)02842-3 34. M. Gargir et al., PNS-type ruthenium pincer conditions. J. Am. Chem. Soc. 133, 1682–1685
21. J. Zhang, M. Gandelman, L. J. W. Shimon, H. Rozenberg, complexes. Organometallics 31, 6207–6214 (2012). www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=
D. Milstein, Electron-rich, bulky ruthenium PNP-type doi: 10.1021/om300516j Retrieve&db=PubMed&list_uids=21247162&dopt=Abstract
complexes: Acceptorless catalytic alcohol 35. D. Spasyuk, S. Smith, D. G. Gusev, From esters to (2011). doi: 10.1021/ja109944n
dehydrogenation. Organometallics 23, 4026–4033 alcohols and back with ruthenium and osmium catalysts. 49. B. Gnanaprakasam, E. Balaraman, Y. Ben-David,
(2004). doi: 10.1021/om049716j Angew. Chem. Int. Ed. 51, 2772–2775 www.ncbi.nlm. D. Milstein, Synthesis of peptides and pyrazines from
22. J. Zhang, E. Balaraman, G. Leitus, D. Milstein, nih.gov/entrez/query.fcgi?cmd=Retrieve&db=PubMed&list_ b-amino alcohols through extrusion of H2 catalyzed
Electron-rich PNP- and PNN-type ruthenium(II) hydrido uids=22319022&dopt=Abstract (2012). by ruthenium pincer complexes: Ligand-controlled
borohydride pincer complexes: Synthesis, structure, and doi: 10.1002/anie.201108956 selectivity. Angew. Chem. Int. Ed. 50, 12240–12244
catalytic dehydrogenation of alcohols and hydrogenation 36. D. Spasyuk, D. G. Gusev, Acceptorless dehydrogenative www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=
of esters. Organometallics 30, 5716–5724 (2011). coupling of ethanol and hydrogenation of esters and Retrieve&db=PubMed&list_uids=22031234&dopt=Abstract
doi: 10.1021/om200595m imines. Organometallics 31, 5239–5242 (2012). (2011). doi: 10.1002/anie.201105876
23. K. Fujita, N. Tanino, R. Yamaguchi, Ligand-promoted doi: 10.1021/om300670r 50. B. Gnanaprakasam, E. Balaraman, C. Gunanathan,
dehydrogenation of alcohols catalyzed by Cp*Ir 37. M. Nielsen, H. Junge, A. Kammer, M. Beller, Towards a D. Milstein, Synthesis of polyamides from diols and
complexes: A new catalytic system for oxidant-free green process for bulk-scale synthesis of ethyl acetate: diamines with liberation of H2. J. Polym. Sci. A Polym.
oxidation of alcohols. Org. Lett. 9, 109–111 www.ncbi. Efficient acceptorless dehydrogenation of ethanol. Chem. 50, 1755–1765 (2012). doi: 10.1002/pola.25943

nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db= Angew. Chem. Int. Ed. 51, 5711–5713 www.ncbi.nlm. 51. H. Zeng, Z. Guan, Direct synthesis of polyamides
PubMed&list_uids=17192097&dopt=Abstract (2007). nih.gov/entrez/query.fcgi?cmd=Retrieve&db= via catalytic dehydrogenation of diols and diamines.
doi: 10.1021/ol062806v PubMed&list_uids=22517628&dopt=Abstract (2012). J. Am. Chem. Soc. 133, 1159–1161 www.ncbi.nlm.
24. S. Musa, I. Shaposhnikov, S. Cohen, D. Gelman, doi: 10.1002/anie.201200625 nih.gov/entrez/query.fcgi?cmd=Retrieve&db=
Ligand-metal cooperation in PCP pincer complexes: rational 38. D. Srimani, E. Balaraman, B. Gnanaprakasam, Y. Ben-David, PubMed&list_uids=21204554&dopt=Abstract (2011).
design and catalytic activity in acceptorless dehydrogenation D. Milstein, Ruthenium pincer-catalyzed cross-dehydrogenative doi: 10.1021/ja106958s
of alcohols. Angew. Chem. Int. Ed. 50, 3533–3537 coupling of primary alcohols with secondary alcohols 52. C. Gunanathan, D. Milstein, Metal-ligand cooperation
www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db= under neutral conditions. Adv. Synth. Catal. 354, by aromatization-dearomatization: A new paradigm in
PubMed&list_uids=21412960&dopt=Abstract (2011). 2403–2406 (2012). doi: 10.1002/adsc.201200438 bond activation and “green” catalysis. Acc. Chem. Res. 44,
doi: 10.1002/anie.201007367 39. B. Gnanaprakasam, Y. Ben-David, D. Milstein, Ruthenium 588–602 www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=
25. M. Nielsen et al., Efficient hydrogen production from pincer-catalyzed acylation of alcohols using esters Retrieve&db=PubMed&list_uids=21739968&dopt=Abstract
alcohols under mild reaction conditions. Angew. Chem. with liberation of hydrogen under neutral conditions. (2011). doi: 10.1021/ar2000265
Int. Ed. 50, 9593–9597 www.ncbi.nlm.nih.gov/entrez/ Adv. Synth. Catal. 352, 3169–3173 (2010). 53. C. Gunanathan, D. Milstein, Bond activation by metal-
query.fcgi?cmd=Retrieve&db=PubMed&list_ doi: 10.1002/adsc.201000663 ligand cooperation: Design of “green” catalytic reactions
uids=21948701&dopt=Abstract (2011). 40. D. M. Hunsicker, B. C. Dauphinais, S. P. Mc Ilrath, based on aromatization-dearomatization of pincer
doi: 10.1002/anie.201104722 N. J. Robertson, Synthesis of high molecular weight complexes. Top. Organomet. Chem. 37, 55–84 (2011).
26. K. Fujita, T. Yoshida, Y. Imori, R. Yamaguchi, polyesters via in vacuo dehydrogenation polymerization doi: 10.1007/3418_2011_6
Dehydrogenative oxidation of primary and secondary of diols. Macromol. Rapid Commun. 33, 232–236 54. M. Montag, J. Zhang, D. Milstein, Aldehyde binding
alcohols catalyzed by a Cp*Ir complex having a functional www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd= through reversible C-C coupling with the pincer ligand
C,N-chelate ligand. Org. Lett. 13, 2278–2281 Retrieve&db=PubMed&list_uids=22173989&dopt=Abstract upon alcohol dehydrogenation by a PNP-ruthenium
www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd= (2012). doi: 10.1002/marc.201100653 catalyst. J. Am. Chem. Soc. 134, 10325–10328
Retrieve&db=PubMed&list_uids=21473566&dopt= 41. E. Balaraman, E. Khaskin, G. Leitus, D. Milstein, Catalytic www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=
Abstract (2011). doi: 10.1021/ol2005424 transformation of alcohols to carboxylic acid salts and H2 Retrieve&db=PubMed&list_uids=22703115&dopt=Abstract
27. R. Kawahara, K. Fujita, R. Yamaguchi, Dehydrogenative using water as the oxygen atom source. Nat. Chem. 5, (2012). doi: 10.1021/ja303121v
oxidation of alcohols in aqueous media using water-soluble 122–125 www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd= 55. G. Zeng, S. Li, Insights into dehydrogenative coupling
and reusable Cp*Ir catalysts bearing a functional Retrieve&db=PubMed&list_uids=23344447&dopt=Abstract of alcohols and amines catalyzed by a (PNN)-Ru(II) hydride
bipyridine ligand. J. Am. Chem. Soc. 134, 3643–3646 (2013). doi: 10.1038/nchem.1536 complex: Unusual metal-ligand cooperation. Inorg. Chem.
www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db= 42. S. W. Kohl et al., Consecutive thermal H2 and 50, 10572–10580 www.ncbi.nlm.nih.gov/entrez/query.fcgi?
PubMed&list_uids=22339738&dopt=Abstract (2012). light-induced O2 evolution from water promoted by a cmd=Retrieve&db=PubMed&list_uids=21942421&dopt=
doi: 10.1021/ja210857z metal complex. Science 324, 74–77 www.ncbi.nlm. Abstract (2011). doi: 10.1021/ic200205e
28. Y. Maenaka, T. Suenobu, S. Fukuzumi, Hydrogen nih.gov/entrez/query.fcgi?cmd=Retrieve&db= 56. H. Li et al., Computational study on the catalytic role of
evolution from aliphatic alcohols and 1,4-selective PubMed&list_uids=19342584&dopt=Abstract (2009). pincer ruthenium(II)-PNN Complex in directly synthesizing
hydrogenation of NAD+ catalyzed by a [C,N] and a doi: 10.1126/science.1168600 amide from alcohol and amine: The origin of selectivity
[C,C] cyclometalated organoiridium complex at room 43. V. R. Pattabiraman, J. W. Bode, Rethinking amide of amide over ester and imine. Organometallics 30,
temperature in water. J. Am. Chem. Soc. 134, 9417–9427 bond synthesis. Nature 480, 471–479 www.ncbi.nlm. 5233–5247 (2011). doi: 10.1021/om200620n
www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db= nih.gov/entrez/query.fcgi?cmd=Retrieve&db= 57. J. Zhang, G. Leitus, Y. Ben-David, D. Milstein, Efficient
PubMed&list_uids=22577897&dopt=Abstract (2012). PubMed&list_uids=22193101&dopt=Abstract (2011). homogeneous catalytic hydrogenation of esters to
doi: 10.1021/ja302788c doi: 10.1038/nature10702 alcohols. Angew. Chem. Int. Ed. 45, 1113–1115
29. J. Zhang, G. Leitus, Y. Ben-David, D. Milstein, Facile 44. C. Gunanathan, Y. Ben-David, D. Milstein, Direct www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=
conversion of alcohols into esters and dihydrogen synthesis of amides from alcohols and amines with Retrieve&db=PubMed&list_uids=16389607&dopt=Abstract
catalyzed by new ruthenium complexes. J. Am. Chem. liberation of H2. Science 317, 790–792 www.ncbi.nlm. (2006). doi: 10.1002/anie.200503771
Soc. 127, 10840–10841 www.ncbi.nlm.nih.gov/entrez/ nih.gov/entrez/query.fcgi?cmd=Retrieve&db= 58. E. Balaraman, B. Gnanaprakasam, L. J. W. Shimon,
query.fcgi?cmd=Retrieve&db=PubMed&list_uids= PubMed&list_uids=17690291&dopt=Abstract (2007). D. Milstein, Direct hydrogenation of amides to alcohols
16076184&dopt=Abstract (2005). doi: 10.1126/science.1145295 and amines under mild conditions. J. Am. Chem. Soc. 132,
doi: 10.1021/ja052862b 45. L. U. Nordstrøm, H. Vogt, R. Madsen, Amide synthesis 16756–16758 www.ncbi.nlm.nih.gov/entrez/
30. S.-I. Murahashi, K. Ito, T. Naota, Y. Maeda, Ruthenium from alcohols and amines by the extrusion of dihydrogen. query.fcgi?cmd=Retrieve&db=PubMed&list_uids=
catalyzed transformation of alcohols to esters and J. Am. Chem. Soc. 130, 17672–17673 www.ncbi.nlm. 21049928&dopt=Abstract (2010). doi: 10.1021/ja1080019
lactones. Tetrahedron Lett. 22, 5327–5330 (1981). nih.gov/entrez/query.fcgi?cmd=Retrieve&db= 59. E. Balaraman, C. Gunanathan, J. Zhang, L. J. W. Shimon,
doi: 10.1016/S0040-4039(01)92493-1 PubMed&list_uids=19061316&dopt=Abstract (2008). D. Milstein, Efficient hydrogenation of organic carbonates,
31. Y. Blum, Y. Shvo, Catalytically reactive (h4-tetracyclone) doi: 10.1021/ja808129p carbamates and formates indicates alternative routes
(CO)2(H)2Ru and related complexes in dehydrogenation 46. N. D. Schley, G. E. Dobereiner, R. H. Crabtree, Oxidative to methanol based on CO2 and CO. Nat. Chem. 3,
of alcohols to esters. J. Organomet. Chem. 282, C7–C10 synthesis of amides and pyrroles via dehydrogenative 609–614 www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=
(1985). doi: 10.1016/0022-328X(85)87154-0 alcohol oxidation by ruthenium diphosphine diamine Retrieve&db=PubMed&list_uids=21778980&dopt=Abstract
32. S.-I. Murahashi, T. Naota, K. Ito, Y. Maeda, H. Taki, complexes. Organometallics 30, 4174–4179 (2011). (2011). doi: 10.1038/nchem.1089
Ruthenium-catalyzed oxidative transformation of alcohols doi: 10.1021/om2004755 60. E. Balaraman, Y. Ben-David, D. Milstein, Unprecedented
and aldehydes to esters and lactones. J. Org. Chem. 52, 47. C. Chen, Y. Zhang, S. H. Hong, N-heterocyclic carbene catalytic hydrogenation of urea derivatives to amines and
4319–4327 (1987). doi: 10.1021/jo00228a032 based ruthenium-catalyzed direct amide synthesis methanol. Angew. Chem. Int. Ed. 50, 11702–11705

REVIEW
www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd= Retrieve&db=PubMed&list_uids=19928825&dopt=Abstract 90. R. Grigg, T. R. B. Mitchell, S. Sutthivaiyakit, N. Tongpenyai,
Retrieve&db=PubMed&list_uids=22052711&dopt=Abstract (2010). doi: 10.1021/cr9002159 Oxidation of alcohols by transition metal complexes part V.
(2011). doi: 10.1002/anie.201106612 75. R. Grigg, T. R. B. Mitchell, S. Sutthivaiyakit, N. Tongpenyai, Selective catalytic monoalkylation of arylacetonitriles
61. D. Cantillo, Mechanistic insights on the ruthenium-catalyzed Transition metal-catalysed N-alkylation of amines by by alcohols. Tetrahedron Lett. 22, 4107–4110 (1981).
hydrogenation of amides: C-N vs. C-O cleavage. alcohols. J. Chem. Soc. Chem. Commun. 1981, 611 (1981). doi: 10.1016/S0040-4039(01)82078-5
Eur. J. Inorg. Chem. 2011, 3008–3013 (2011). doi: 10.1039/c39810000611 91. S. Whitney, R. Grigg, A. Derrick, A. Keep, [Cp*IrCl2]2-
doi: 10.1002/ejic.201100443 76. S.-I. Murahashi, K. Kondo, T. Hakata, Ruthenium catalyzed indirect functionalization of alcohols:
62. C. Gunanathan, B. Gnanaprakasam, M. A. Iron, catalyzed synthesis of secondary or tertiary amines from Novel strategies for the synthesis of substituted
L. J. W. Shimon, D. Milstein, “Long-range” metal-ligand amines and alcohols. Tetrahedron Lett. 23, 229–232 indoles. Org. Lett. 9, 3299–3302 www.ncbi.nlm.
cooperation in H2 activation and ammonia-promoted (1982). doi: 10.1016/S0040-4039(00)86792-1 nih.gov/entrez/query.fcgi?cmd=Retrieve&db=
hydride transfer with a ruthenium-acridine pincer 77. T. Tsuji, K.-T. Huh, Y. Ohsugi, Y. Watanabe, Ruthenium PubMed&list_uids=17658835&dopt=Abstract (2007).
complex. J. Am. Chem. Soc. 132, 14763–14765 complex catalyzed N-heterocyclization: Syntheses of doi: 10.1021/ol071274v
www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd= N-substituted piperidines, morpholines, and piperazines 92. P. A. Slatford, M. K. Whittlesey, J. M. J. Williams,
Retrieve&db=PubMed&list_uids=20925343&dopt= from amines and 1,5-diols. J. Org. Chem. 50, C–C bond formation from alcohols using a Xantphos
Abstract (2010). doi: 10.1021/ja107770y 1365–1370 (1985). doi: 10.1021/jo00209a004 ruthenium complex. Tetrahedron Lett. 47, 6787–6789
63. E. Kossoy, Y. Diskin-Posner, G. Leitus, D. Milstein, 78. C. Gunanathan, D. Milstein, Selective synthesis of (2006). doi: 10.1016/j.tetlet.2006.07.069
Selective acceptorless conversion of primary alcohols to primary amines directly from alcohols and ammonia. 93. T. Jensen, R. Madsen, Ruthenium-catalyzed alkylation of
acetals and dihydrogen catalyzed by the ruthenium(II) Angew. Chem. Int. Ed. 47, 8661–8664 www.ncbi.nlm. oxindole with alcohols. J. Org. Chem. 74, 3990–3992
complex Ru(PPh3)2(NCCH3)2(SO4). Adv. Synth. Catal. nih.gov/entrez/query.fcgi?cmd=Retrieve&db= www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=
354, 497–504 (2012). doi: 10.1002/adsc.201100672 PubMed&list_uids=18846519&dopt=Abstract (2008). Retrieve&db=PubMed&list_uids=19371090&dopt=
64. B. Gnanaprakasam, J. Zhang, D. Milstein, Direct synthesis doi: 10.1002/anie.200803229 Abstract (2009). doi: 10.1021/jo900341w
of imines from alcohols and amines with liberation of H2. 79. G. Walther et al., a,w-Functionalized C19 monomers. 94. Y. Obora, Y. Ishii, Iridium-catalyzed reactions involving
Angew. Chem. Int. Ed. 49, 1468–1471 www.ncbi.nlm. ChemSusChem 4, 1052–1054 www.ncbi.nlm.nih. transfer hydrogenation, addition, N-Heterocyclization,

nih.gov/entrez/query.fcgi?cmd=Retrieve&db= gov/entrez/query.fcgi?cmd=Retrieve&db= and alkylation using alcohols and diols as key
PubMed&list_uids=20101664&dopt=Abstract (2010). PubMed&list_uids=21656697&dopt=Abstract (2011). substrates. Synlett 2011, 30–51 (2011).
doi: 10.1002/anie.200907018 doi: 10.1002/cssc.201100187 doi: 10.1055/s-0030-1259094
65. H. Li, X. Wang, M. Wen, Z. Wang, Computational insight 80. D. Pingen, C. Müller, D. Vogt, Direct amination of 95. M. G. Edwards, J. M. J. Williams, Catalytic electronic
into the mechanism of selective imine formation from secondary alcohols using ammonia. Angew. Chem. activation: indirect “Wittig” reaction of alcohols.
alcohol and amine catalyzed by the ruthenium(II)-PNP Int. Ed. 49, 8130–8133 www.ncbi.nlm.nih.gov/ Angew. Chem. Int. Ed. 41, 4740–4743 www.ncbi.nlm.
pincer complex. Eur. J. Inorg. Chem. 2012, 5011–5020 entrez/query.fcgi?cmd=Retrieve&db=PubMed&list_ nih.gov/entrez/query.fcgi?cmd=Retrieve&db=
(2012). doi: 10.1002/ejic.201200473 uids=20672264&dopt=Abstract (2010). PubMed&list_uids=12481344&dopt=Abstract (2002).
66. L. He, T. Chen, D. Gong, Z. Lai, K. Huang, Enhanced doi: 10.1002/anie.201002583 doi: 10.1002/anie.200290034
reactivities toward amines by introducing an imine arm 81. S. Imm et al., Improved ruthenium-catalyzed amination 96. G. Cami-Kobeci, J. M. J. Williams, Conversion of
to the pincer ligand: Direct coupling of two amines to of alcohols with ammonia: Synthesis of diamines and alcohols into N-alkyl anilines via an indirect aza-Wittig
form an imine without oxidant. Organometallics 31, amino esters. Angew. Chem. Int. Ed. 50, 7599–7603 reaction. Chem. Commun. (Camb.) 2004, 1072
5208–5211 (2012). doi: 10.1021/om300422v www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd= www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=
67. A. Maggi, R. Madsen, Dehydrogenative synthesis of Retrieve&db=PubMed&list_uids=21732514&dopt= Retrieve&db=PubMed&list_uids=15116190&dopt=
imines from alcohols and amines catalyzed by a Abstract (2011). doi: 10.1002/anie.201103199 Abstract (2004). doi: 10.1039/b402020k
ruthenium N-heterocyclic carbene complex. 82. R. Yamaguchi, S. Kawagoe, C. Asai, K. Fujita, Selective 97. K. Fujita, C. Asai, T. Yamaguchi, F. Hanasaka,
Organometallics 31, 451–455 (2012). synthesis of secondary and tertiary amines by R. Yamaguchi, Direct beta-alkylation of secondary
doi: 10.1021/om201095m Cp*iridium-catalyzed multialkylation of ammonium salts alcohols with primary alcohols catalyzed by a CpIr
68. M. A. Esteruelas, N. Honczek, M. Oliván, E. Oñate, with alcohols. Org. Lett. 10, 181–184 www.ncbi.nlm.nih. complex. Org. Lett. 7, 4017–4019 www.ncbi.nlm.
M. Valencia, Direct access to POP-type osmium(II) and gov/entrez/query.fcgi?cmd=Retrieve&db=PubMed&list_ nih.gov/entrez/query.fcgi?cmd=Retrieve&db=
osmium(IV) complexes: Osmium a promising alternative uids=18076182&dopt=Abstract (2008). PubMed&list_uids=16119956&dopt=Abstract (2005).
to ruthenium for the synthesis of imines from alcohols doi: 10.1021/ol702522k doi: 10.1021/ol051517o
and amines. Organometallics 30, 2468–2471 (2011). 83. K. Fujita, Z. Li, N. Ozeki, R. Yamaguchi, N-alkylation 98. R. Martínez, D. J. Ramón, M. Yus, RuCl2(DMSO)4
doi: 10.1021/om200290u of amines with alcohols catalyzed by a Cp*Ir catalyzes the b-alkylation of secondary alcohols with
69. M. Zhang, H. Neumann, M. Beller, Selective complex. Tetrahedron Lett. 44, 2687–2690 (2003). primary alcohols through a hydrogen autotransfer
ruthenium-catalyzed three-component synthesis of doi: 10.1016/S0040-4039(03)00371-X process. Tetrahedron 62, 8982–8987 (2006).
pyrroles. Angew. Chem. Int. Ed. 52, 597–601 www.ncbi. 84. C. T. Eary, D. Clausen, Preparation of substituted doi: 10.1016/j.tet.2006.07.012
nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db= 1,2,3,4-tetrahydroquinoxalines and 2,3,4,5-tetrahydro- 99. J. F. Bower, M. J. Krische, Formation of C-C bonds via
PubMed&list_uids=23184875&dopt=Abstract (2013). 1H-benzo[b][1,4]diazepines from catalytic Cp*Ir iridium-catalyzed hydrogenation and transfer
doi: 10.1002/anie.201206082 hydrogen transfer N-heterocyclization of anilino hydrogenation. Top. Organomet. Chem. 34, 107 (2011).
70. S. Michlik, R. Kempe, A sustainable catalytic pyrrole alcohols. Tetrahedron Lett. 47, 6899–6902 (2006). www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=
synthesis. Nat. Chem. 5, 140–144 (2013). www.ncbi.nlm. doi: 10.1016/j.tetlet.2006.07.033 Retrieve&db=PubMed&list_uids=21822399&dopt=
nih.gov/entrez/query.fcgi?cmd=Retrieve&db= 85. S. Bähn et al., The catalytic amination of alcohols. Abstract.
PubMed&list_uids=23344449&dopt=Abstract Chem. Cat. Chem. 3, 1853–1864 (2011). doi: 10.1002/ 100. J. C. Leung, L. M. Geary, T. Y. Chen, J. R. Zbieg,
doi: 10.1038/nchem.1547; pmid: 23344449 cctc.201100255 M. J. Krische, Direct, redox-neutral prenylation and
71. D. Srimani, Y. Ben-David, D. Milstein, Direct synthesis of 86. M. H. S. A. Hamid, P. A. Slatford, J. M. J. Williams, geranylation of secondary carbinol C-H bonds:
pyrroles by dehydrogenative coupling of b-aminoalcohols Borrowing hydrogen in the activation of alcohols. C4-regioselectivity in ruthenium-catalyzed C-C couplings
with secondary alcohols catalyzed by ruthenium pincer Adv. Synth. Catal. 349, 1555–1575 (2007). of dienes to a-hydroxy esters. J. Am. Chem. Soc. 134,
complexes. Angew. Chem. Int. Ed. 52, 4012–4015 doi: 10.1002/adsc.200600638 15700–15703 www.ncbi.nlm.nih.gov/entrez/query.
www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd= 87. M. H. S. A. Hamid et al., Ruthenium-catalyzed fcgi?cmd=Retrieve&db=PubMed&list_uids=22985393&dopt=
Retrieve&db=PubMed&list_uids=23468418&dopt= N-alkylation of amines and sulfonamides using Abstract (2012). doi: 10.1021/ja3075049
Abstract (2013). doi: 10.1002/anie.201300574 borrowing hydrogen methodology. J. Am. Chem. Soc. 101. J. R. Zbieg, E. Yamaguchi, E. L. McInturff, M. J. Krische,
72. A. S. Goldman et al., Catalytic alkane metathesis by tandem 131, 1766–1774 www.ncbi.nlm.nih.gov/entrez/ Enantioselective C-H crotylation of primary alcohols via
alkane dehydrogenation-olefin metathesis. Science 312, query.fcgi?cmd=Retrieve&db=PubMed&list_uids= hydrohydroxyalkylation of butadiene. Science 336,
257–261 www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd= 19191700&dopt=Abstract (2009). 324–327 www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=
Retrieve&db=PubMed&list_uids=16614220&dopt=Abstract doi: 10.1021/ja807323a Retrieve&db=PubMed&list_uids=22442385&dopt=
(2006). doi: 10.1126/science.1123787 88. M. A. Berliner et al., Use of an iridium-catalyzed Abstract (2012). doi: 10.1126/science.1219274
73. A. J. A. Watson, J. M. J. Williams, The give and take of redox-neutral alcohol-amine coupling on kilogram scale
alcohol activation. Science 329, 635–636 www.ncbi.nlm. for the synthesis of a GlyT1 inhibitor. Org. Process Acknowledgments: Supported by the European Research
nih.gov/entrez/query.fcgi?cmd=Retrieve&db= Res. Dev. 15, 1052–1062 (2011). doi: 10.1021/op200174k Council (ERC) under the FP7 framework (no. 246837) and
PubMed&list_uids=20689005&dopt=Abstract (2010). 89. L. Miao, S. C. Dimaggio, H. Shu, M. L. Trudell, by the Kimmel Center for Molecular Design. C.G. thanks the
doi: 10.1126/science.1191843 Enantioselective syntheses of both enantiomers of Department of Science and Technology and NISER, and he is a
74. G. Guillena, D. J. Ramón, M. Yus, Hydrogen autotransfer in noranabasamine. Org. Lett. 11, 1579–1582 www.ncbi. Ramanujan Fellow. D.M. is the Israel Matz Professorial Chair of
the N-alkylation of amines and related compounds using nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db= Organic Chemistry.
alcohols and amines as electrophiles. Chem. Rev. 110, PubMed&list_uids=19320505&dopt=Abstract (2009).
1611 www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd= doi: 10.1021/ol9002288 10.1126/science.1229712

Tight Coordination of Protein Translation and HSF1 Activation
Supports the Anabolic Malignant State
Sandro Santagata et al.
Science 341, (2013);
DOI: 10.1126/science.1238303


found at:
This article has been cited by 1 articles hosted by HighWire Press; see:
http://www.sciencemag.org/content/341/6143/1238303.full.html#related-urls
RESEARCH ARTICLE SUMMARY
Tight Coordination of Protein READ THE FULL ARTICLE ONLINE

Translation and HSF1 Activation Cite this article as S. Santagata et al.,

Science 341, 1238303 (2013).
DOI: 10.1126/science.1238303
Supports the Anabolic Malignant State

Sandro Santagata, Marc L. Mendillo, Yun-chi Tang, Aravind Subramanian, Casey C. Perley,
Stéphane P. Roche, Bang Wong, Rajiv Narayan, Hyoungtae Kwon, Martina Koeva, FIGURES IN THE FULL ARTICLE
Angelika Amon, Todd R. Golub, John A. Porco Jr., Luke Whitesell,* Susan Lindquist*
Fig. 1. Inhibiting protein translation
inactivates HSF1.
Introduction: Ribosome biogenesis is commonly up-regulated to satisfy the increased anabolic Fig. 2. LINCS analysis reveals that targeting
demands associated with malignant transformation and tumor growth. Many different oncogenic sig- protein translation inactivates HSF1.
naling pathways converge on the ribosome to increase translational flux. Despite the detailed under- Fig. 3. Chemical screens reveal that targeting
standing of ribosome regulation in cancer, it is not clear whether the net translational activity of the translation control inactivates HSF1.
ribosome can itself regulate transcriptional programs that support and promote the malignant state.

Fig. 4. Inhibiting translation initiation with
Methods: To investigate the transcriptional effects of modulating translational activity in malignant rocaglates ablates HSF1 DNA binding.
cells, we used integrated chemical and genetic approaches, including a gene signature–based genetic
and chemical screen of more than 600,000 gene expression profiles (LINCS database) and an indepen- Fig. 5. Rocaglates modulate tumor energy
dent, reporter-based chemical screen of more than 300,000 compounds. A lead compound was tested metabolism.
in several cell lines unified by their increased dependence on HSF1 activation for growth and survival, Fig. 6. Rocaglates selectively target aneuploid
and in an in vivo cancer model. cancer cells and nontransformed cells with
Results: Inhibiting translation led to large changes in the transcriptome. The single most enriched cancer-associated genetic aberrations.
category consisted of genes regulated by the heat-shock transcription factor, HSF1. The most down- Fig. 7. Rocaglates suppress tumor growth,
regulated mRNA was HSPA8, which encodes the constitutive HSP70 chaperone that helps to fold HSPA8 mRNA levels, and glucose uptake
nascent polypeptides. The expression of many other genes that HSF1 coordinates to support cancer in vivo.
was also strongly affected. HSF1 protein levels were unchanged, but HSF1 DNA occupancy was nearly
eliminated. Inhibition of the HSF1-regulated gene expression program is thus a dominant transcrip- SUPPLEMENTARY MATERIALS
tional effect elicited by inhibiting protein translation.
Using a gene signature of HSF1 inactivation to query the LINCS database revealed a strong con- Materials and Methods
nection between HSF1 inactivation and perturbations that inhibit protein translation, including a Figs. S1 to S9
broad spectrum of chemical and genetic interventions that target the ribosome, eukaryotic initiation Tables S1 to S5
factors (eIFs), aminoacyl tRNA synthetases, and upstream signaling/regulatory pathways that control References and Notes
translation.
Our high-throughput small-molecule screen identified rocaglamide A, an inhibitor of translation RELATED ITEMS IN SCIENCE
initiation, as the strongest inhibitor of HSF1 activation. An analog of this compound, RHT, increased V. Gandin, I. Topisirovic, Trans-HFS1 express.
thioredoxin-interacting protein (TXNIP) mRNA and protein levels and decreased glucose uptake and Science 341, 242-243 (2013).
lactate production. Cell-based cancer models characterized by high dependence on HSF1 activation DOI: 10.1126/science.1242359
for growth and survival were highly sensitive to RHT, as were cells derived from diverse hematopoietic
malignancies. RHT had a strong antitumor effect—with marked inhibition of HSF1 activity and glucose
uptake—against xenografted acute myeloid leukemia cells.
Discussion: The ribosome functions as a central information hub in malig-
nant cells: Translational flux conveys information about the cell’s metabolic
status to regulate the transcriptional programs that support it. Multiple unbi-
ased chemical and genetic approaches establish HSF1 as a prime transducer
of this information, centrally poised to regulate the transcription of genes that
support protein folding, biomass expansion, anabolic metabolism, cellular pro-
liferation, and survival. Targeting translation initiation may offer a strategy for
reversing HSF1 activation, disabling metabolic and cytoprotective pathways in
malignant cells.
HSF1 at the crossroads of protein translation and metabolism. (Left) Cancers

activate an HSF1-regulated transcriptional program to adapt to the anabolic demands of
relentless biomass expansion. Glucose uptake increases, and expression of TXNIP, an
inhibitor of glucose uptake, drops. (Right) Down-regulating translation with rocaglate
scaffold initiation inhibitors reverses cancer-associated HSF1 activation. Glucose uptake
drops as TXNIP levels rise.
The list of author affiliations is available in the full article online.

*Corresponding author. E-mail: whitesell@wi.mit.edu (L.W.); lindquist_admin@wi.mit.edu (S.L.)

Published by AAAS
RESEARCH ARTICLE
by HSF1 were down-regulated when translation-
al flux through the ribosome was reduced. These
Tight Coordination of Protein

genes included drivers of cell proliferation and
mitogenic signaling (such as CENPA, CKS1B, and
PRKCA), transcription and mRNA processing
Translation and HSF1 Activation (such as LSM2 and LSM4), protein synthesis
(such as FXR1 and MRPL18), energy metabolism
Supports the Anabolic Malignant State (such as MAT2A, SLC5A3, PGK1, MBOAT7, and
SPR), and invasion/metastasis (such as EMP2 and
LTBP1). In a complementary fashion, genes that
Sandro Santagata,1,2,3* Marc L. Mendillo,3,4* Yun-chi Tang,4,5† Aravind Subramanian,6 are negatively regulated by HSF1 were up-regulated
Casey C. Perley,3,4 Stéphane P. Roche,7 Bang Wong,6 Rajiv Narayan,6 Hyoungtae Kwon,3,4 when translational flux through the ribosome was
Martina Koeva,3,4 Angelika Amon,4,5 Todd R. Golub,6 John A. Porco Jr.,7 reduced. These included genes that promote dif-
Luke Whitesell,3‡ Susan Lindquist3,4‡ ferentiation (such as NOTCH2NL), cellular adhe-
sion (such as EFEMP1 and LAMA5), and apoptosis
The ribosome is centrally situated to sense metabolic states, but whether its activity, in turn, (such as BCL10, CFLAR, and SPTAN1).
coherently rewires transcriptional responses is unknown. Here, through integrated chemical-genetic This effect of translation inhibition on HSF1-
analyses, we found that a dominant transcriptional effect of blocking protein translation in regulated transcription led us to examine the

cancer cells was inactivation of heat shock factor 1 (HSF1), a multifaceted transcriptional regulator genome-wide pattern of DNA occupancy by HSF1
of the heat-shock response and many other cellular processes essential for anabolic metabolism, in breast cancer cells. After a 6-hour exposure to
cellular proliferation, and tumorigenesis. These analyses linked translational flux to the regulation cycloheximide, we performed chromatin immu-
of HSF1 transcriptional activity and to the modulation of energy metabolism. Targeting this link noprecipitation coupled with massively parallel
with translation initiation inhibitors such as rocaglates deprived cancer cells of their energy and DNA sequencing (ChIP-Seq) using a previously
chaperone armamentarium and selectively impaired the proliferation of both malignant and validated antibody against HSF1 (8). Despite cy-
premalignant cells with early-stage oncogenic lesions. cloheximide treatment, HSF1 protein levels them-
selves remained unchanged (Fig. 1D). In contrast
egulation of ribosome activity is crit- Results to DNA occupancy by RNA-polymerase II (which
R ical for supporting cellular prolifera-

tion. In cancer, ribosome biogenesis is
commonly increased to satisfy the increased
Inhibiting Translational Flux Inactivates HSF1
To investigate the transcriptional effects of re-
was not globally reduced), HSF1 occupancy was
nearly eliminated (Fig. 1, E to G, fig. S2, and table
S3). This held true for genes that were either
anabolic demands associated with malignant ducing translational flux through the ribosome in positively or negatively regulated by HSF1, as well
transformation and tumor growth (1–4). In ad- malignant cells, we analyzed the mRNA expres- as for genes shared with the classic heat-shock
dition, many different oncogenic signaling path- sion profiles of breast cancer cells after treatment response and genes specific to the HSF1 cancer
ways converge on the ribosome to modulate with various inhibitors of translation elongation program (table S3). Together, these data pointed
its function (5, 6). These inputs are integrated, (anisomycin, emetine, cephaeline, and cyclohex- to a link between the activity of the ribosome and
and the net ribosomal translational activity is imide). Large changes in the transcriptome were the activity of HSF1.
tuned to reflect the metabolic and proliferative highly correlated across all four inhibitors [Pearson
state of the cell. Despite the detailed under- correlation coefficient (r) between 0.85 to 0.97 LINCS Establishes Translation as a Potent
standing of ribosome regulation in cancer, it is for all pairwise correlations]. The most strongly Regulator of HSF1 in Cancer Cells
not well understood whether the net trans- enriched category consisted of genes regulated To further investigate the link between the HSF1
lational activity of the ribosome can itself be by promoters that contain DNA binding motifs activity and translational program, we turned to an
conveyed to regulate transcriptional programs for the heat-shock transcription factor known as expression-profiling resource created by the Li-
that support and promote the malignant state. heat shock factor 1 (HSF1) (P = 9.87 × 10–7) brary of Integrated Network-based Cellular Sig-
Is a coherent and coordinated transcriptional (Fig. 1A and table S1). Of the 13,258 genes mea- natures (LINCS) program (Fig. 2 and supplementary
response triggered by modulating translation sured, the single most down-regulated mRNA materials, materials and methods). The LINCS
activity? was HSPA8, which encodes a constitutive HSP70 database is a large catalog of gene-expression
chaperone that folds nascent polypeptides as they profiles collected from human cells treated with
1
Department of Pathology, Brigham and Women’s Hospital emerge from the ribosome (Fig. 1B and table S2) chemical (small-molecule) and genetic [short hair-
(BWH), and Harvard Medical School, Boston, MA 02215, (7). HSPA1A, a cancer-induced HSP70 gene, was pin RNA (shRNA)] perturbations.
USA. 2Dana Farber Cancer Institute, Boston, MA 02215, USA. also among the 10 most down-regulated mRNAs. We generated a query signature for HSF1 in-
3
Whitehead Institute for Biomedical Research (WIBR), Cam- This transcriptional response suggested that re- activation from expression profiles of breast can-
bridge, MA 02142, USA. 4Howard Hughes Medical Institute,
Department of Biology, Massachusetts Institute of Technology
duced translational flux causes a profound shift in cer cells that had been treated with HSF1 shRNAs
(MIT), Cambridge, MA 02142, USA. 5David H. Koch Institute the activity of HSF1. (supplementary materials, materials and methods)
for Integrative Cancer Research and Howard Hughes Medical In a wide range of cancers, HSF1 regulates a (8). This signature included both genes that were
Institute, MIT, Cambridge, MA 02142, USA. 6Broad Institute of transcriptional network that is distinct from the up-regulated by HSF1 inactivation and down-
MIT and Harvard, Cambridge, MA 02142, USA. 7Department
conventional network activated by thermal stress regulated by HSF1 inactivation. We compared our
of Chemistry, Center for Chemical Methodology and Library
Development (CMLD-BU), Boston University, Boston, MA (8). This cancer network includes many classic HSF1 query signature with LINCS expression
02215, USA. “heat-shock” genes. But, it also includes a broad profiles from nine cell lines that are currently the
*These authors contributed equally to this work. cadre of other genes that play critical roles in ma- most extensively characterized in this database
†Present address: Institute of Health Sciences, Shanghai Insti- lignancy, some of which are positively regulated (Fig. 2A). Eight of these are cancer lines of di-
tutes for Biological Sciences, Chinese Academy of Sciences & by HSF1 and some negatively regulated. All four verse histopathologic origin. These lines have been
Shanghai Jiao Tong University School of Medicine, 20025
Shanghai, China.
inhibitors of translation elongation affected genes treated individually with 3866 small-molecule com-
‡Corresponding author. E-mail: whitesell@wi.mit.edu (L.W.); in the HSF1 cancer network (P = 0.016) (Fig. 1C pounds or 16,665 shRNAs targeting 4219 genes.
lindquist_admin@wi.mit.edu (S.L.) and fig. S1). Genes that are positively regulated The compounds used for these gene expression

RESEARCH ARTICLE
profiles encompassed U.S. Food and Drug Ad- signaling pathways that regulate protein transla- the high-throughput 384-well format (supplemen-
ministration (FDA)–approved drugs and known tion: phosphatidylinositol 3-kinase (PI3Kinase)/ tary materials, materials and methods), we used
bioactives. The shRNAs used were directed against mammalian target of rapamycin (mTOR) inhib- a reporter cell line stably transduced with a
the known targets of these compounds, against itors (Fig. 2B and table S4). Of the nearly 200 luminescence-based reporter and induced HSF1
genes in related pathways, or against other genes gene ontology classes analyzed, the ribosome sub- activation with a simple proteotoxic stressor (the
that have been implicated in a variety of human unit family was the single most enriched (Fig. 2, proteasome inhibitor MG132).
diseases. In all, we compared our HSF1 signa- B and C, and table S4). In addition, eukaryotic Approximately 2500 hit compounds from the
ture with 161,636 LINCS signatures, each gen- initiation factors (eIFs) and aminoacyl tRNA primary screen, which blocked induction of the
erated from at least three replicates (for a total of synthetases were also highly enriched. This un- reporter, were then counter-screened with an in-
614,216 profiles.) biased analysis using the LINCS database dem- dependent dual reporter cell line (Fig. 3B) so as
As expected, the LINCS perturbations that onstrates the connection between translational flux to eliminate nonselective inhibitors. This second
negatively correlated with our HSF1 inactivation and the function of HSF1 in cancer. line had been stably transduced with two constructs,
signature were enriched for known activators of one encoding a green fluorescent protein (GFP)
HSF1. They included shRNAs that target com- An Unbiased High-Throughput Chemical driven by HSEs and the other encoding a red flu-
ponents of the proteasome. They also included Screen for HSF1 Inhibitors orescent protein (RFP) driven by a doxycycline-
compounds that inhibit the proteasome and that To find alternate ways to inhibit HSF1, we per- regulated control promoter. Compounds that
inhibit HSP90 (Fig. 2, B and C and table S4). formed a large high-throughput chemical screen. selectively inhibit HSF1 activity should suppress
The LINCS perturbations that positively cor- We screened 301,024 compounds through the GFP expression in this cell line but should not
related with our HSF1 inactivation signature were U.S. National Institutes of Health (NIH) Molec- suppress doxycycline-mediated induction of RFP.

most highly enriched for translation inhibitors ular Libraries Probe Center Network (MLPCN; Notably, compounds that are thought to selec-
(cephaeline, cycloheximide, and emetine) (Fig. 2, Pubchem AID, 2118) (Fig. 3A) using an HSF1- tively inhibit HSF1—such as triptolide, quercetin,
B and C, and table S4). These perturbations were regulated reporter driven by consensus heat-shock KNK423, and KNK437 (9)—all suppressed
also highly enriched for compounds that target elements (HSEs). To accommodate constraints of both reporters (fig. S3). Thus, these compounds
Fig. 1. Inhibiting protein transla-

tion inactivates HSF1. (A) Gene set
enrichment analysis was performed
by using the MSigDB web service re-
lease version 3.84 (www.broadinstitute.
org/gsea/index.jsp) on genes nega-
tively regulated in breast cancer cells
following a 6-hour incubation with in-
hibitors of protein translation elongation. Selected results are displayed; control tubulin after a 6-hour exposure to the indicated concentrations of
complete GSEA results are provided in table S1. (B) Scatter plot of levels of cycloheximide (CHX). (E) Heat map of RNA polymerase II ChIP-Seq read density
mRNA transcripts (log2) after a 6-hour incubation with the indicated inhibitors in MCF7 cells that were treated with DMSO or 10 mM CHX for 6 hours. Genomic
of protein translation elongation versus control dimethyl sulfoxide (DMSO). The regions from –2 kb to +2 kb relative to the transcription start site for all RefSeq
levels of HSPA1A and HSPA8 are indicated for each elongation inhibitor. (C) genes are shown. (F) Heat map of HSF1 ChIP-Seq read density in MCF7 cells that
Translation elongation inhibitors alter the basal transcriptional program of were treated with DMSO or 10 mM CHX for 6 hours. Genomic regions from
HSF1 in breast cancer cells. Genes bound by HSF1 in MCF7 were ranked by –1 kb to +1 kb relative to the peak of HSF1 binding for all HSF1-enriched
their differential expression between cells treated with translation elongation regions (union of all HSF1-enriched regions in the four data sets depicted
inhibitors (TI) and control DMSO (D). Each column represents a gene and is here) are shown. (G) Representative genes bound by HSF1 in MCF7 cells (HSPA8,
normalized across the column, with high expression in red and low expression HSPA1A, CKS2, and RBM23). The x axis depicts from −2kb from the transcription
in blue. (D) An immunoblot shows the levels of HSF1 protein and the loading start site (TSS) to 5 kb from the TSS for each gene.

RESEARCH ARTICLE
are far less specific for HSF1 than commonly additional rocaglates (fig. S4). These included itself (Fig. 3D and fig. S6A). However, mRNA
assumed. both natural products and totally synthetic ana- levels of Hsp40 (DNAJA1) and Hsp70 genes
More to the point, the chemical screen results logs prepared with photocycloaddition methods (HSPA1B and HSPA8) dropped substantially. The
also suggested a link between HSF1 activation and (12, 13). Five hydroxamate analogs were more most dramatically affected was the constitutively
the translation machinery. By far the most potent potent than rocaglamide A at inhibiting the HSE expressed HSPA8 gene (>90% reduction) (Fig.
and selective hit to emerge from the 301,024 reporter while retaining similar selectivity (table 3D). This was also the gene that we had found
compounds we tested was the rocaglate known S5). The most potent inhibitor had an IC50 of to be the most strongly repressed by translation
as rocaglamide A [median inhibitory concentra- ~20 nM (table S5). We named this compound elongation inhibitors (Fig. 1B).
tion (IC50) of ~50 nM for the heat shock reporter [previously reported as “8e” (13)] Rohinitib (or The effects of RHT were not due to reductions
versus IC50 >1000 nM for the control reporter] RHT), for Rocaglate Heat Shock, Initiation of in HSF1 protein levels, which remained constant
(Fig. 3C). This natural product inhibits the func- Translation Inhibitor. (Fig. 3E and fig. S6B). The sharp decrease in
tion of the translation initiation factor eIF4A, a HSP70 mRNA levels in response to RHT held
DEAD box RNA helicase (10, 11). Presumably, Characterizing the Effects of RHT on true across a histologically diverse panel of hu-
it passed counterscreening in our secondary assay Cancer Cells man cancer cell lines (MCF7, breast adenocarci-
with the dual reporter system because translation To validate findings from our engineered reporter noma; MO91, myeloid leukemia; CHP100, sarcoma;
of the doxycycline-regulated RFP control does system, we measured the effects of RHT on the and HeLa, cervical carcinoma) as well as in ar-
not require the classical cap-dependent initiation basal expression of several endogenous HSF1- tificially transformed 293T kidney cells (Fig. 3D
complex. regulated transcripts (Fig. 3D and figs. S5 and and fig. S6, A and C). RHT had a much smaller
To define structure-activity relationships for S6). RHT did not reduce the transcript levels of effect on HSP70 mRNA levels in proliferating

inhibition of the HSE reporter by rocaglamide A, the control housekeeping genes B2M and GAPDH. but nontumorigenic diploid cells (WI38 and IMR90)
we used our dual reporter system to test 38 Nor did it reduce the transcript levels of HSF1 (fig. S6C).
A 161K unique gene HSF1 query signature of Results ordered by B NES Score
expression signatures up/down regulated genes connectivity to HSF1 2.28 Translation inhibitors *
2.00 Ribosome subunits *
1 161k 1 161k 1.71 PI3K/mTOR inhibitors *
+ 1.35 Aminoacyl tRNA synthetases *
+ 1.26 EIF subunits *
0
0 * p < 10-2
- -1.20 Proteasome subunits*

-
Genes ranked by Analytics calculate Negative Positive
change in expression connectivity score connection connection -2.80 Proteasome inhibitors *
vs. control -3.05 HSP90 inhibitors *
Total
C Negative connections Positive connections signatures
All
161k
perturbagens
Null
Proteasome
56
inhibitors
HSP90
390
inhibitors
Translation
142
inhibitors
Ribosome
442
subunits
-0.70 -0.50 -0.30 -0.10 0.10 0.30 0.50 0.70
Connectivity score
Fig. 2. LINCS analysis reveals that targeting protein translation in- Normalized enrichment score (NES) of selected results are plotted (complete
activates HSF1. (A) Schematic representation of the LINCS analysis used GSEA results are provided in table S4). (C) Barcode plot of the connectivity
to identify chemical and genetic modulators that are correlated with HSF1 score of all of the individual perturbations comprising the indicated enriched
inactivation (supplementary materials, materials and methods). Pink repre- chemical or gene sets. The bagel plot in the center of the barcode plot
sents genes whose levels increase, and green represents genes whose levels summarizes the positive, negative, and null (not connected) fractions for the
decrease, after shRNA-mediated knockdown of HSF1. (B) GSEA results of our indicated enriched class. All perturbations that are positively or negatively
HSF1 inactivation signature LINCS analysis. Perturbation signatures were rank- connected for the indicated enriched classes are shown. Total perturbations in
ordered by connectivity with the HSF1 inactivation signature, and enrichment each class are indicated on the right of the plot. Blue represents negatively
was determined for KEGG pathway gene sets and ATC chemical classes (de- connected, and red represents positively connected, classes of enriched
tails are available in the supplementary materials, materials and methods). perturbations.

RESEARCH ARTICLE
To obtain a more direct and global view of (TXNIP) were up-regulated. TXNIP is a power- increased dependence on HSF1 activation for
RHT’s effects on HSF1 activity, we examined ful negative regulator of glucose uptake and is a growth and survival. Although it occurs very
genome-wide promoter occupancy by means of well-established regulator of cellular energy status early during oncogenesis, simple loss of the tu-
ChIP-Seq analysis. RHT virtually abolished HSF1 (16, 17). Its expression is dramatically reduced mor suppressor Nf1 leads to an increase in HSF1
binding throughout the genome (Fig. 4, A and B; in malignant cells, leading to increased glucose protein levels, nuclear localization, and transcrip-
fig. S6D; and table S3). As had occurred with uptake (18). Conversely, increasing TXNIP lev- tional activation (19). We treated mouse embryonic
cycloheximide, RHT affected both genes that are els leads to reduced glucose uptake (16). The in- fibroblasts (MEFs) in which Nf1 was knocked
positively regulated by HSF1 and genes that are duction of TXNIP mRNA by RHT was observed out and wild-type littermate control MEFs in
negatively regulated by HSF1 (Fig. 4A). Further- across a diverse panel of tumor cell lines (Fig. 5A). which HSF1 was not activated, with either RHT
more, it affected both classic heat-shock genes TXNIP protein levels also increased sharply de- or with cycloheximide. The two cell types were
and genes specific to the HSF1 cancer program spite a marked reduction in the levels of other similarly sensitive to cycloheximide. However,
(table S3). The effects on HSF1 DNA occupancy short-lived proteins, such as p53 (Fig. 5B). Al- Nf1-null MEFs were more sensitive than were
occurred at concentrations of cycloheximide and though we did not detect HSF1 bound to the wild-type MEFs to RHT (Fig. 6A). In this model
RHT that inhibit the ribosome activity to a sim- TXNIP locus, HSF1 null cells showed higher for an early event in tumorigenesis, targeting trans-
ilar extent (Fig. 4C). levels of TXNIP (fig. S7). In addition, HSF1 did lation initiation rather than translation elongation
directly regulate a group of other genes involved seems to provide a more selective, better tolerated
Rocaglates Modulate Tumor Energy Metabolism in energy metabolism (including MAT2A, SLC5A3, approach for disrupting the link between trans-
While characterizing the effects of RHT on the and PGK1). At a functional level, the effects of lation and HSF1 activation.
transcriptome, we noted that treated cells failed to RHTwere associated with concentration-dependent A second engineered system allowed us to ask

acidify the culture medium (detected incidentally reductions in both glucose uptake and lactate whether rocaglates would selectively inhibit the
by the color of the pH indicator phenol red in- production (Fig. 5C). Thus, the effects of RHT on growth of cells carrying a simple chromosomal
cluded in standard media). This suggested a re- protein translation, HSF1 activation, and energy aberration that models another common early event
versal of the “Warburg effect,” a metabolic shift metabolism—processes lying at the core of the in the development of cancer: aneuploidy. Chro-
responsible for increased lactic acid production anabolic state of cancer—appear to be coordinated. mosomal imbalances lead to both increased ener-
by many cancers. Genetic compromise of HSF1 gy and proteotoxic stress. This is reflected by the
drives a shift in metabolism in both cell culture Rocaglates Selectively Target Pre-Malignant elevation of the HSF1-regulated chaperone pro-
and animal models (14, 15). Hence, this effect of Cells with Early-Stage Oncogenic Lesions tein HSP72, encoded by HSPA1A (20). We isolated
RHT is consistent with inactivation of HSF1. Does this tight coordination create vulnerabilities MEFs from mice carrying Robertsonian fusions
Our mRNA expression profiling of rocaglate- for the malignant phenotype that could be ex- for chromosome 13 (21). These MEFs (TS13) car-
treated breast cancer cells also revealed that ploited as a therapeutic strategy? We looked at a ry a single extra copy of 120 Mb of chromosome
mRNA levels for thioredoxin-interacting protein range of cell-based cancer models unified by their 13, introducing an additional copy of 843 genes.
Fig. 3. Chemical screens reveal that targeting

translation control inactivates HSF1. (A) Flow-
chart outlining the steps in the high-throughput
MLPCN screen for inhibitors of HSF1 activation. (B)
Schematic of dual-reporter cell line used to counter-screen primary screen hits. GFP expression is regulated by a heat shock–inducible promoter. RFP expression
is regulated by a doxycycline response element (TetR). (C) Effect of rocaglamide A on the HSE-driven GFP and doxycycline-driven RFP signals after incubation
with 2.5 mM MG132 and 2 mg/ml doxycycline. Chemical structure of rocaglamide A is displayed in the inset. Error bars indicate mean T SEM of quadruplicates.
(D) Effect of RHT on HSF1-regulated and control endogenous mRNA transcript levels in M0-91 leukemia cells measured by means of nanostring nCounter after a
6-hour incubation with indicated concentrations of RHT. Levels of endogenous transcript are shown as percent of DMSO-treated control. Mean of duplicates are
reported. (E) HSF1 protein levels are not affected in M0-91 leukemia cells treated with RHT. Immunoblot shows the levels of HSF1 protein and the loading control
(Tubulin) after a 6-hour exposure to the indicated concentrations of RHT.

RESEARCH ARTICLE
Cycloheximide, as well as conventional cy- transformed colon epithelial cell lines with eu- mice (fig. S9). We assessed aqueous solubility,
totoxic chemotherapeutics (taxol and hydroxyurea), ploid chromosome content were the least sensitive plasma stability, plasma protein binding, hepatic
inhibited the growth of both trisomic and litter- of all the lines we tested (Fig. 6F). microsome stability, and cellular permeability
mate control MEFs to an equal extent (Fig. 6B (fig. S9A). No severe liabilities were found. We
and fig. S8). But, trisomic MEFs (P < 0.0001) were Rocaglates Suppress the Growth of Cancer next established minimally toxic parameters for
more sensitive than wild-type MEFs to RHT Cells and Tumors dosing mice with RHT and performed a plasma
(Fig. 6B). Thus, again in this model for an early Some rocaglates have previously been shown to pharmacokinetic study after administration of
neoplastic change that activates HSF1, targeting exert profound anticancer activity (10, 24–26). 1 mg/kg subcutaneously (fig. S9, B and C). Peak
translation initiation seems to provide a better We tested RHT against a collection of cell lines plasma levels were far in excess of those required
tolerated, more selective approach for targeting including nontransformed diploid lines and for the key biological activities we had demon-
the malignant state. cancer cell lines with diverse histopathological strated in cell culture. Moreover, levels required
HSF1 activation is even more prominent in origins and oncogenic lesions (Fig. 7A). The non- for anticancer activity in vitro were maintained in
advanced malignancies (8, 22, 23). For example, transformed cell lines were relatively resistant to excess of 2 hours in vivo.
colon cancers frequently show immunohistochem- RHT (IC50 from 100 to 300 nM). All cancer cell We next established subcutaneous tumor xeno-
ical evidence of strong HSF1 activation (Fig. 6C) lines were sensitive to RHT (IC50 < 30 nM); the grafts of the human myeloid leukemia cell line
and this correlates with poor clinical outcome (8). hematopoietic tumor cell lines were especially M091 in nonobese diabetic (NOD)–severe com-
We mined publicly available expression profiling sensitive (IC50 ≤ 6 nM). We used one of these bined immunodeficient (SCID) immunocompro-
from colon cancer lines with highly aneuploid hematopoietic tumor lines, the M0-91 cell line mised mice. When the mean tumor volume reached
karyotypes [chromosomal instability (CIN)] and originally derived from a patient with acute mye- 100 mm3, we administered RHT at 1 mg/kg for 4

from colon cancer lines with near-euploid kar- loid leukemia (27), to further characterize the consecutive days each week for 3 weeks (Fig. 7D).
yotypes but microsatellite instability (MIN). The effects of RHT. RHT strongly suppressed HSPA8 Over the treatment period, there was no evidence
CIN lines expressed markedly higher levels of mRNA levels in M0-91 cells and induced TXNIP of gross systemic toxicity (fig. S9D). RHT mediated
HSPA1A, which is consistent with greater levels mRNA (Fig. 7B). In addition, RHT sharply de- marked, sustained inhibition of the growth of this
of proteotoxic stress and greater activation of the creased glucose uptake by these cells (Fig. 7C). very aggressive myeloid malignancy (Fig. 7D).
HSF1-regulated cancer program (Fig. 6, D and Are the effects of RHT in cell culture achiev- We then pursued pharmacodynamics studies.
E). Next, we tested several patient-derived colon able at drug exposures that are systemically tol- Mice bearing xenografts were given a single dose
cancer lines with CIN and several patient-derived erable in animals? To directly address this critical of RHT. Tumors were explanted 4 hours later,
colon cancer lines with MIN for sensitivity to issue of therapeutic index, we first used standard and HSPA8 and TXNIP mRNA levels were deter-
inhibition by RHT. The CIN lines were much in vitro assays to test whether RHT had suffi- mined by means of reverse transcription polymer-
more sensitive than were the MIN lines. Non- ciently drug-like properties to justify testing in ase chain reaction (RT-PCR) (Fig. 7E). Similar to
Fig. 4. Inhibiting translation

initiation with rocaglates
ablates HSF1 DNA binding.
(A) Heat map of HSF1 ChIP-
Seq read density in M0-91 cells
that were treated with DMSO,
20 nM RHT, 100 nM RHT, or
10 mM CHX for 6 hours. Ge-
nomic regions from –1 kb to
+1 kb relative to the peak of
HSF1 binding for all HSF1 en-
riched regions (union of all
HSF1-enriched regions in the
seven data sets depicted here)
are shown. (B) Representative
HSF1-bound genes in M0-91
cells (HSPA8, HSPA1B, CKS2,
and RBM23). The x axis depicts from –2 kb from the transcription start site (TSS) to 5 kb from the TSS for each gene. (C) Autoradiograph of S35-labeled protein
lysates from MCF7 cells treated for 6 hours with the indicated concentrations of RHT or CHX. Graphs show the counts per minute from acetone precipitation of
proteins in each sample, quantitated by using a scintillation counter.

RESEARCH ARTICLE
the effects we observed in cell culture, RHT caused Although cancer cells often co-opt powerful, were used to ensure that additional karyotypic
a strong decrease in HSPA8 transcript levels and a adaptive nononcogene systems for their benefit changes had not yet occurred. Two primary hu-
strong increase in TXNIP transcript levels. In a (40), it now appears that by co-opting the link be- man cell lines (CCD112 CoN and CCD841 CoN),
separate experiment, we monitored the uptake of tween the ribosome and HSF1, cancers become five MIN lines (HCT-116, HCT-15, DLD-1,
fluorescently labeled 2-deoxyglucose 48 hours especially vulnerable to agents that target transla- SW48, and LoVo), and five CIN lines (Caco2,
after RHT dosing. RHTstrongly suppressed uptake tion and its upstream regulatory pathways. In this HT-29, SW403, SW480, and SW620) were ob-
of this glucose analog by these tumors (Fig. regard, our animal experiments suggest that tar- tained from ATCC. Chromosome number and
7F). Clearly, the effects of RHT achieved in cell geting translation initiation may offer a strategy karyotype information were obtained from the Na-
culture could also be achieved in whole animals. for reversing HSF1 activation and disabling the tional Cancer Institute database and the COSMIC
metabolic and cytoprotective addictions of ma- data set at the Sanger Institute. M0-91 cells were
Discussion lignant cells. previously described (27). The M0-91 cell line
HSF1 provides essential support for the malig- used in this study were established from explanted
nant state by blocking apoptotic responses and Materials and Methods M0-91 tumors that had been xenografted once in
promoting protein synthesis, anabolic energy mice. All cell cultures were maintained under 5%
metabolism, mitogenic signaling pathways, and Cell Lines CO2 in media according to their specifications.
pathways that facilitate invasion and metastasis WI38, CHP100, HeLa, 293T, PC3, MCF7, and
(8, 14, 15, 19, 23, 28–30). Here, we show that the NIH3T3 cells were purchased from American mRNA Expression Profiling and Analysis
ability of HSF1 to maintain this cancer program Type Culture Collection (ATCC). Immortalized Expression profiles for MCF7 cells treated for
is exquisitely sensitive to translational activity. Nf1 knockout MEFs and littermate wild-type con- 6 hours with anisomycin (15 mM), emetine (7 mM),

Our work establishes that the ribosome could trol MEFs were kind gifts from K. Cichowski. cephaeline (6 mM), and cycloheximide (14 mM)
function as a central information hub: Transla- Littermate-derived euploid and trisomic primary were previously deposited in the Connectivity
tional flux conveys information about the cell’s MEFs were described previously (20). RHT treat- Map (41). MCF7 cells were treated with 200 nM
metabolic status to regulate the transcriptional ments experiments were performed by using chro- rocaglamide A or 50 nM RHT for 6 hours, and
programs that support it. The specific molecular mosome 13 trisomic cell lines and littermate RNA was then purified after extraction with
mechanisms are likely to be multifaceted, but HSF1 control euploid cell lines that carried a single TRIzol reagent [Invitrogen (Carlsbad, CA), cata-
is clearly a linchpin in this process. One plausible Robertsonian translocation. Early-passage MEFs log no. 15596-026]. Gene expression analysis was
mechanism for the effects of translation inhibitors
on HSF1 activity could involve the translation of
mRNA for a modifier of HSF1 transcriptional
activity that is sensitive to subtle changes in trans-
lation activity, involving eIF4A and/or other initia-
tion factors. Because HSF1 regulates the expression
of genes encoding for ribosomal subunits and
other regulators of translation (8, 15, 31), the ex-
istence of a feedforward regulatory circuit involv-
ing the protein translation machinery and HSF1
is also possible.
HSF1 is centrally poised to support protein
folding and biomass expansion as well as many
other functions to which malignant cells are
addicted (8, 14, 15, 32). We postulate that the
translation/HSF1 link we have uncovered in
cancer may derive from ancient systems geared
to align and synchronize essential cellular func-
tions for growth and survival. In comparison, in the
nematode HSF1 is a longevity factor, and in yeast
it is an essential gene that participates in cotransla-
tional quality control (33–35).
In man, the translation/HSF1 link is particu-
larly important in supporting the malignant phe-
notype because it can respond to varied metabolic
inputs that are commonly dysregulated in cancer Fig. 5. Rocaglates modulate
(5, 6, 36–38). This translation/HSF1 link allows tumor energy metabolism.
(A) TXNIP mRNA transcript lev-
these metabolic inputs to bolster the cytoprotec-
els in a panel of cancer cell
tive milieu, helping tumor cells to accommodate
lines measured by nanostring
the drastic internal imbalances arising during on- nCounter after a 6-hour incu-
cogenesis as well as the severe external stresses bation with 50 nM RHT. Mean
arising from therapeutic interventions (39). The of duplicates are reported. (B)
tight coordination of protein translation and HSF1 Immunoblot showing TXNIP
activation, together with the many ways that cells levels in the indicated cancer
integrate the derangements of malignancy with cell lines after a 6-hour incu-
translational activity, suggests that unifying prin- bation with the indicated con-
ciples drive HSF1 activation across the extraor- centration of RHT. b-actin is the loading control. The effect on p53, a short half-life protein, is shown. (C)
dinarily wide range of human cancers in which Effects of the indicated amount of RHT on [H3]-2-deoxyglucose uptake (left) and lactate production (right)
that activation occurs (8, 22). in a panel of cancer cell lines after 8 hours of incubation. Error bars indicate mean T SEM of triplicates.

RESEARCH ARTICLE
performed by using Affymetrix (Santa Clara, CA) posited in a public database [National Center for after treatment of MCF7 cells with translation
GeneChip HT Human Genome U133A 96-Array Biotechnology Information Gene Expression elongation inhibitors was performed by using
Plates, and data was analyzed as previously de- Omnibus (GEO) pending]. Gene set enrichment the gene set enrichment analysis (GSEA) Web
scribed (8). All microarray raw data were de- analysis of the differentially expressed genes site release version 3.84 (42). Enrichment for

Fig. 6. Rocaglates selectively target aneuploid cancer cells and nontransformed
cells with cancer-associated genetic aberrations. (A) Photomicrographs of Nf1 wild type
and Nf1 null MEFs that were treated for 14 days with 25 nM RHT. The relative viable cell
number of RHT-treated (middle) and CHX-treated (right) are shown. Error bars indicate
mean T SEM of n = 6 replicates. (B) Effect of 72 hours treatment with either RHT (left) or
cycloheximide (right) on the proliferation of MEFs carrying a single extra copy of 120 Mb of chromosome 13 (TS13) compared with MEFs derived from littermate
controls (WT), [mean T SD, n = 3 replicates, ***P < 0.001, two-way analysis of variance (ANOVA)]. (C) Photomicrographs of normal colon epithelial cells and
invasive colon adenocarcinoma (hematoxylin and eosin stains and HSF1 immunohistochemistry) from the same section of a human tumor resection (im-
munostained simultaneously). HSF1-expressing cells stain brown, and HSF1-negative cells stain blue from the toluidine blue counterstain. Scale bar, 50 mm. (D)
HSPA1A mRNA transcript levels are elevated in colorectal adenocarcinomas with high-grade aneuploid karyotypes. Data from three MIN and nine CIN colon
cancer cell lines from the GSK Cancer Cell Line Genomic Profiling Data as described in the methods. Box plot, bar is median, and whiskers are min and max (three
cell lines in MIN category in triplicate 9 cell lines in CIN category in triplicate). (E) RT-PCR analysis of HSPA1A mRNA levels in the indicated euploid nontransformed
cell lines, and MIN and CIN cancer cell lines. Error bars indicate mean T SD of triplicates. (F) Effect of RHT on the proliferation of a panel of cell lines with high-grade
aneuploid karyotypes (CIN lines: Caco2, HT29, SW403, SW480, and SW620); near-euploid karyotypes with microsatellite instability (MIN lines: HCT-116, HCT-15,
DLD-1, SW48, and LoVo); or nontransformed colon epithelial cell lines with a euploid chromosomal number (CCD112CoN and CCD841CoN), (mean T SD, n = 3
replicates, ***P < 0.001, two-way ANOVA) treated for 72 hours.

RESEARCH ARTICLE
HSF1-bound genes among the genes differen- Evaluation of HSPA1A mRNA levels was per- ChiP-Seq and ChIP-PCR
tially expressed after treatment of MCF7 cells with formed by using data from the GlaxoSmithKline Described in supplementary materials, materials
translation elongation inhibitors was conducted (GSK) Cancer Cell Line Genomic Profiling Data and methods.
by using GSEA v2.08 software (42). HSF1-bound https://cabig.nci.nih.gov/community/caArray_GSK-
genes in MCF7 cells were defined as those genes data. MIN lines used were HCT15, LS174T, SW48. Immunoblot
bound in at least two of the four data sets [two CIN lines used were NCIH508, NCIH747, SW1116, Described in supplementary materials, materials
data sets from this study and two from (8)]. SW1417, SW403, SW480, SW620, T84, and SW948. and methods.

Fig. 7. Rocaglates suppress tumor growth, HSPA8 mRNA levels and RHT (1 mg/kg), on days marked by downward pointing arrows]. Eight mice were in
glucose uptake in vivo. (A) Scatter plot of IC50 values of the growth of a diverse each treatment group (error bars indicate mean T SEM; P < 0.0001). (E) RT-PCR
panel of cell lines treated with RHT. Cells were treated for 5 days. Red indicates analysis of HSPA8 and TXNIP mRNA levels from tumor xenografts after a single
hematopoietic cancer lines, dark blue indicates solid tumor cell lines, and light treatment of either vehicle or RHT (1 mg/kg, subcutaneous; five mice in each group).
blue indicates euploid nontransformed cells. (B) mRNA levels of HSPA8, TXNIP, Tumors were harvested 4 hours after treatment (error bars indicate mean T SEM;
and control housekeeping genes in M0-91 cells treated with RHT. Mean of du- HSPA8, P < 0.005; TXNIP, P < 0.0005, two-tailed t test, n = 5 replicates). (F)
plicates are reported. (C) Glucose uptake of of IR Dye 800CW 2-deoyglucose (2-DG) Representative image of epifluorescence of IRDye 800CW 2-deoxyglucose (2-DG)
in M0-91 cell lines treated with RHT. Imaging was performed by using LICOR. uptake in M0-91 xenografts. Mice bearing tumors were treated with vehicle or
(Right) Quantitation of measured intensity (error bars indicate mean T SEM; P < RHT (1 mg/kg) as described in the supplementary materials, materials and
0.005, two-tailed t test, n = 4 replicates). (D) Plot of the tumor volume of M0-91 methods; four mice in each group. Images were acquired 48 hours after the last
acute myeloid leukemia xenografts treated with vehicle or RHT. The mean tumor treatment. (Right) Quantitation of measured radiant efficiency from epifluo-
volume (in cubic millimeters) is plotted over time. Mice were treated with sub- rescence of IRDye 800CW 2-DG from images of M0-91 xenografts (error bars
cutaneous injections starting on day 18 after implantation [either vehicle alone or indicate mean T SEM; P = 0.031, two-tailed t test, n = 4 replicates).

RESEARCH ARTICLE
LINCS Analysis pTRIPZ-nonsilencing construct (RHS4743). sodium salicylate and 10% glycerol. The dried
To identify chemical and genetic modulators that These cells were heat shocked and incubated with gel was used to expose film. Counts per minute
are correlated with HSF1 inactivation, we queried doxycycline and then sorted by use of flow cy- were assessed by using a scintillation counter.
LINCS supported by the NIH Common Fund. tometry to isolate strong enhanced GFP (eGFP)
This resource at the Broad Institute is a massive and Turbo RFP (tRFP) expressors. Sorting was Glucose Uptake
expression-profiling initiative to catalog the cel- repeated twice for enrichment. Noninduced cells 500,000 cells plated in 24-well tissue culture
lular consequences of both small-molecule and were sorted to remove cells expressing eGFP and plates were treated for 8 hours with RHT. The
genetic perturbations. The expression data were tRFP at baseline to make R4.1.B4 cells. cells were then washed in 1× PBS (×2) and
generated by using a high-throughput Luminex- To make the high-throughput screening cell line placed in glucose-free and serum-free DMEM
based assay (Luminex, Austin, TX) as described NIH3T3HGL, the parent vector LV-eGFPfLUC for 20 min. Glucose uptake was measured by
previously (43). Whole-genome expression pro- was modified by removing the cytomegalovirus using 3H-2-Deoxyglucose (3H-2DG), incubation
files are inferred from changes in the expression promoter and introducing a 470-bp fragment of for 15 min (final 1 mCi/ml, 50 mM 2DG). The
of 1000 landmark genes. The changes in gene the human HSP70B′ construct upstream of the cells were washed with 1 ml of cold 1× PBS (×2),
expression resulting from each of the genetic eGFP promoter. NIH3T3 cells were infected with lysed with 0.2 M sodium hydroxide, and then
and chemical perturbations are rank-ordered from lentivirus generated from this construct, and the counted by use of scintillation. Experiments were
highest to lowest according to their differential high eGFP expressors were isolated by means of performed in triplicate. Parallel treated cultures
expression relative to control treatments. Changes flow cytometry after heat shock. cells were stained with Sytox-green (Invitrogen,
in gene expression caused by a novel perturbation catalog no. S7020) for normalization. Each an-
(the “query” signature) can then be compared with High-Throughput Small-Molecule Screen alysis was performed three times. The SEM is

the cataloged expression profiles. Profiles that Described in supplementary materials, materials displayed.
are positively correlated with the query signature and methods. Glucose uptake was also measured by using
are given a positive score, whereas profiles that are IRD800 2-deoxyglucose (Fig. 7C). M0-91 cells
negatively correlated to the query signature are Dual Reporter Cell Assay were washed and resuspended in glucose-free
given a negative score. Described in supplementary materials, materials medium (106 cells/ml). Cells were dispensed
For the analysis, we generated an HSF1 in- and methods. (100 mL/well) in triplicate wells and incubated
activation signature (table S4) of the 50 genes for 20 min with 5 mM of the IRDye 800CW 2-DG
most positively regulated (reduced expression upon Rocaglamide/Rocaglate Derivatives Optical Probe [LI-COR Biosciences (Lincoln, NB)
HSF1 depletion with shRNA) and 10 genes most Rocaglamide/rocaglate derivatives were prepared catalog no. 926-08946]. The cells were then washed
negatively regulated (increased expression upon with total synthesis methods as previously de- ×4 with cold PBS, and the signal was acquired
HSF1 depletion with shRNA) in the breast cancer scribed (12, 13). by using an infrared Odyssey imaging system
cell lines, MCF7 and BPLER (44) [average of the (LI-COR Biosciences).
difference between the ha6 shRNA and scram- Nanostring/nCounter Analysis
bled shRNA control values between the two cell The cells were lysed at a concentration of 10,000 Lactate Production
lines (8)], that were also bound by HSF1 in our cells/mL with RTL buffer [Qiagen (Valencia, CA) 500,000 cells were plated in 24-well tissue cul-
ChIP-seq experiments. This signature was used catalog no. 79216] and dissociated using a cell ture plates and were treated for 8 hours. with
to query all 161,636 shRNA and compound sig- shredder (Qiagen catalog no. 79656). The total RHT. After two washes with 1× PBS, the cells
natures (collapsed from a total of 614,216 indi- RNA in 5 ml of lysate was hybridized with the were incubated for 30 min at 37°C in 500 ml of
vidual profiles from at least three biological capture and reporter probes overnight at 65°C filter sterilized 1× Krebs buffer (126 mM NaCl,
replicates) in the LINCS data set produced in and processed according to the nCounter recom- 2.5 mM KCl, 1.2 mM NaH2PO4, 1.2 mM MgCl2,
nine cell lines (MCF7, breast cancer; HT29, co- mended protocol. Target/probe complexes were 2.5 mM CaCl2, 10 mM Glucose, 25 mM NaHCO3,
lon cancer; HEPG2, hepatoblastoma; A549, lung immobilized in nCounter Cartridges for data col- 10 mM HEPES-KOH pH 7.4). The supernatant
cancer; HCC515, lung cancer; A375, melanoma; lection using an nCounter Digital. The data was was collected, and the lactate was measured with
PC3, prostate cancer; VCAP, prostate cancer; HA1E, analyzed according to the manufacture’s guide- a Lactate Assay Kit [BioVision (Milpitas, CA) cat-
immortalized but nontransformed kidney epithe- lines. All data were normalized to the expres- alog no. K-607) according to the manufacturer’s
lium). A connectivity score was assigned to each sion levels of house keeping genes (ACTB, B2M, guidelines. Parallel treated cultures cells were stained
of the expression profiles from the 161,636 per- GAPDH, GUSB, HPRT1, and RPL10). with Sytox Green (Invitrogen, catalog no. S7020)
turbations on the basis of a weighted Kolmogorov- for normalization. Each analysis was performed
Smirnov statistic as previously described (42, 43). S35 Labeling three times. The SEM is displayed.
GSEA (42) was performed on this rank-ordered MCF7 cells were grown to confluence in six-well
list in order to determine gene or chemical classes dishes in standard Dulbecco’s minimum essential Cell Viability Assay
that were most enriched among the positively and medium (DMEM) (+10% fetal bovine serum). Relative cell growth and survival were measured
negatively connected signatures. The sets analyzed The cells were rinsed twice in 1× phosphate- in 96-well microplate format by using the fluo-
by means of GSEA encompassed the shRNAs buffered saline (PBS) and then placed in DMEM rescent detection of resazurin dye reduction as an
corresponding to the genes comprising all 186 without methionine or cysteine [Life Technolo- endpoint (544 nm excitation and 590 nm emis-
KEGG pathway gene sets. The sets also included gies (Guilford, CT) no. 21013024] for 30 min. sion). Two thousand adherent cells and 10,000
110 chemical classes grouped according to the EasyTag S35 protein labeling mix (NEG772002MC) suspension cells were plated 24 hours before com-
Anatomical Therapeutic Chemical (ATC) Classi- was added for 15 min. The cells were rinsed pound exposure (for 72 hours). Each analysis
fication System. In addition, we added a set com- twice in 1X PBS and then lysed in TNEK buffer. was performed three times. For all bar graphs, the
posed of elongation initiation factors. Statistical Cell lysates were prepared in TNEK buffer {50 mM SEM is displayed, unless indicated otherwise.
significance was tested by using 100 random sets Tris, pH 7.4; NP-40 1%; EDTA 2 mM; KCl 200 mM
size-matched to the set being tested. and protease inhibitor cocktail [Roche Diagnos- Immunohistochemistry
tics (Indianapolis, IN) catalog no. 11836153001]}. Paraffin blocks of human colon adenocarcinoma
Reporter cell lines Samples (15 mg total protein/lane) were analyzed tissue were obtained from the archives of BWH
Y9 reporter NIH3T3 cells (45) were infected by means of SDS–polyacrylamide gel electro- in accordance with the regulations for excess tis-
with lentivirus for the doxycycline regulatable phoresis. The gel was incubated for 10 min in 0.7 M sue use stipulated by the BWH institutional review

RESEARCH ARTICLE
board. Immunohistochemistry for HSF1 was per- 5. D. Silvera, S. C. Formenti, R. J. Schneider, Nat. Rev. Cancer 34. J. S. Hahn, Z. Hu, D. J. Thiele, V. R. Iyer, Mol. Cell. Biol.
formed as previously described (8). 10, 254–266 (2010). doi: 10.1038/nrc2824; 24, 5249–5256 (2004). doi: 10.1128/MCB.24.12.5249-
pmid: 20332778 5256.2004; pmid: 15169889
6. P. P. Roux, I. Topisirovic, Cold Spring Harb. Perspect. Biol. 4, 35. A. L. Hsu, C. T. Murphy, C. Kenyon, Science 300, 1142–1145
Drug Metabolism and Pharmacokinetic Studies a012252 (2012). doi: 10.1101/cshperspect.a012252; (2003). doi: 10.1126/science.1083701; pmid: 12750521
Described in supplementary materials, materials pmid: 22888049 36. S. D. Chou, T. Prince, J. Gong, S. K. Calderwood,
7. F. U. Hartl, M. Hayer-Hartl, Science 295, 1852–1858 PLoS ONE 7, e39679 (2012). doi: 10.1371/journal.
and methods.
(2002). doi: 10.1126/science.1068408; pmid: 11884745 pone.0039679; pmid: 22768106
8. M. L. Mendillo et al., Cell 150, 549–562 (2012). 37. S. M. Hensen et al., Cell Stress Chaperones 17, 743–755
Xenograft Experiment doi: 10.1016/j.cell.2012.06.031; pmid: 22863008 (2012). doi: 10.1007/s12192-012-0347-1; pmid: 22797943
5e7 M0-91 cells were implanted with Matrigel 9. L. Whitesell, S. Lindquist, Expert Opin. Ther. Targets 13, 38. T. Peng, T. R. Golub, D. M. Sabatini, Mol. Cell. Biol. 22,
469–478 (2009). doi: 10.1517/14728220902832697; 5575–5584 (2002). doi: 10.1128/MCB.22.15.5575-
(BD Biosciences, San Jose, CA) subcutaneously in
pmid: 19335068 5584.2002; pmid: 12101249
the right inguinal region of NOD-SCID mice. 10. M. E. Bordeleau et al., J. Clin. Invest. 118, 2651–2660 39. L. Whitesell, S. Santagata, N. U. Lin, Curr. Mol. Med. 12,
When the mean tumor volume reached 100 mm3, (2008). pmid: 18551192 1108–1124 (2012). doi: 10.2174/156652412803306657;
RHT formulated in hydroxypropyl beta-cyclodextrin 11. J. M. Chambers et al., Org. Lett. 15, 1406–1409 pmid: 22804235
(2013). doi: 10.1021/ol400401d; pmid: 23461621 40. N. L. Solimini, J. Luo, S. J. Elledge, Cell 130, 986–988
was administered by means of subcutaneous par- 12. C. M. Rodrigo, R. Cencic, S. P. Roche, J. Pelletier, (2007). doi: 10.1016/j.cell.2007.09.007; pmid: 17889643
enteral administration (1 mg/kg) according to J. A. Porco Jr., J. Med. Chem. 55, 558–562 (2012). 41. J. Lamb et al., Science 313, 1929–1935 (2006).
the treatment schedule shown in Fig. 7D. Tumor doi: 10.1021/jm201263k; pmid: 22128783 doi: 10.1126/science.1132939; pmid: 17008526
size was measured twice each week by a lab 13. S. P. Roche, R. Cencic, J. Pelletier, J. A. Porco Jr., 42. A. Subramanian et al., Proc. Natl. Acad. Sci. U.S.A. 102,
Angew. Chem. Int. Ed. Engl. 49, 6533–6538 (2010). 15545–15550 (2005). doi: 10.1073/pnas.0506580102;
member (M.D.) who was blinded to the treatment doi: 10.1002/anie.201003212; pmid: 20687060 pmid: 16199517
groups. There were eight mice in each treatment 14. X. Jin, D. Moskophidis, N. F. Mivechi, Cell Metab. 14, 43. D. Peck et al., Genome Biol. 7, R61 (2006).

group (RHT-treated or vehicle-treated). Body weight 91–103 (2011). doi: 10.1016/j.cmet.2011.03.025; doi: 10.1186/gb-2006-7-7-r61; pmid: 16859521
was measured twice each week. Body condition pmid: 21723507 44. T. A. Ince et al., Cancer Cell 12, 160–170 (2007).
15. C. Dai, L. Whitesell, A. B. Rogers, S. Lindquist, Cell 130, doi: 10.1016/j.ccr.2007.06.013; pmid: 17692807
score (BCS) as monitored by facility vets did not
1005–1018 (2007). doi: 10.1016/j.cell.2007.07.020; 45. T. J. Turbyville et al., J. Nat. Prod. 69, 178–184
go below 2+ in either treatment group throughout pmid: 17889646 (2006). doi: 10.1021/np058095b; pmid: 16499313
the experiment. 16. H. Parikh et al., PLoS Med. 4, e158 (2007). doi: 10.1371/
journal.pmed.0040158; pmid: 17472435 Acknowledgements: We thank T. Volkert, J. Love, S. Gupta,
In Vivo Glucose Uptake Experiment 17. C. A. Stoltzman et al., Proc. Natl. Acad. Sci. U.S.A. 105, and the WIBR–Genome Technology Core for sequencing
6912–6917 (2008). doi: 10.1073/pnas.0712199105; support; S. Malstrom (Koch Institute for Integrative Cancer
M0-91 cells were inoculated into the inguinal re- pmid: 18458340 Research) for assistance with in vivo imaging; G. Bell, P. Thiru,
gion of NOD-SCID mice. Seventeen days later, 18. S. Y. Kim, H. W. Suh, J. W. Chung, S. R. Yoon, I. Choi, and A. Lancaster for assistance with informatics analysis; the
the mice were treated with a dose of RHT (1 mg/kg; Cell. Mol. Immunol. 4, 345–351 (2007). pmid: 17976314 Connectivity Map team at the Broad Institute for generation
4 mice) or vehicle control (4 mice). Four hours 19. C. Dai et al., J. Clin. Invest. 122, 3742–3754 (2012). of the LINCS data set and query tools; Joe Negri and the
doi: 10.1172/JCI62727; pmid: 22945628 MLPCN team at the Broad Institute for chemical screening;
later, the mice were given retro-orbital injections 20. Y. C. Tang, B. R. Williams, J. J. Siegel, A. Amon, Cell 144, and M. Duquette for assistance with animal experiments.
of 100 ml IRDye 800CW 2-DG Optical Probe 499–512 (2011). doi: 10.1016/j.cell.2011.01.017; We also thank C. Rodrigo (Boston University) for compound
(10 nmol; LI-COR Biosciences, no. 926-08946), pmid: 21315436 synthesis. We thank the Lindquist laboratory for helpful
and then an additional 4 hours later, these mice 21. B. R. Williams et al., Science 322, 703–709 (2008). discussions and suggestions. The work was supported by
doi: 10.1126/science.1160058; pmid: 18974345 the Johnson & Johnson’s Corporate Office of Science and
were again treated with RHT (1mg/kg) or vehicle
22. S. Santagata et al., Proc. Natl. Acad. Sci. U.S.A. 108, Technology focused funding program (L.W.), the Marble
control. Thirty-six hours after the last RHT dose, 18378–18383 (2011). doi: 10.1073/pnas.1115031108; Fund (S.L.), and NIH R01 CA175744-01 (L.W.). The MLPCN
mice were imaged (IVIS; excitation 745 nm, emis- pmid: 22042860 screen was supported by R03 MH086465-01 and R03
sion 800 nm). Data were analyzed by using Living 23. F. Fang, R. Chang, L. Yang, Cancer 118, 1782–1794 DA027713-01 to L.W. This work was supported by the NIH
Image software (PerkinElmer, Waltham, MA). (2012). doi: 10.1002/cncr.26482; pmid: 22009757 Common Fund’s LINCS program (5U54HG006093, “Large
24. L. Alinari et al., Clin. Cancer Res. 18, 4600–4611 (2012). scale gene expression analysis of cellular states”) to T.R.G.
doi: 10.1158/1078-0432.CCR-12-0839; pmid: 22791882 J.A.P. Jr. is supported by R01 GM073855. S.L. is an
Real-Time PCR 25. R. Cencic et al., PLoS ONE 4, e5223 (2009). Investigator of the Howard Hughes Medical Institute. M.L.M.
RNAwas purified with RNEasy columns (Qiagen, doi: 10.1371/journal.pone.0005223; pmid: 19401772 was supported by American Cancer Society New England
catalog no. 74104). Quantitative PCR to evaluate 26. D. M. Lucas et al., Blood 113, 4656–4666 (2009). Division–SpinOdyssey (PF-09-253-01-DMC). S.S. is supported
doi: 10.1182/blood-2008-09-175430; pmid: 19190247 by NIH (K08NS064168), the Brain Science Foundation, the
mRNA levels was performed by using RT2 SYBR 27. M. Okabe et al., Leuk. Res. 19, 933–943 (1995). American Brain Tumor Association, the Beez Foundation,
Green qPCR Mastermix (SABiosciences) and primer doi: 10.1016/0145-2126(95)00039-9; pmid: 8632663 the V Foundation, and the Jared Branfman Sunflowers for
assay pairs (SABiosciences, Valencia, CA) on a 28. L. Meng, V. L. Gabai, M. Y. Sherman, Oncogene 29, Life Fund. The expression profiling and ChIP-Seq data are
7900HT ABI Detection System. 5204–5213 (2010). doi: 10.1038/onc.2010.277; deposited in GEO (GSE45853). The MLPCN chemical screening
pmid: 20622894 data are deposited in Pubchem (AID: 2118).
29. S. Santagata et al., ACS Chem. Biol. 7, 340–349 (2012).
References and Notes doi: 10.1021/cb200353m; pmid: 22050377
1. D. Ruggero, P. P. Pandolfi, Nat. Rev. Cancer 3, 179–192 30. K. L. Scott et al., Cancer Cell 20, 92–103 (2011).
Supplementary Materials
(2003). doi: 10.1038/nrc1015; pmid: 12612653 doi: 10.1016/j.ccr.2011.05.025; pmid: 21741599 www.sciencemag.org/content/full/341/6143/1238303/suppl/
2. J. C. Chan et al., Sci. Signal. 4, ra56 (2011). 31. V. L. Gabai et al., Mol. Cell. Biol. 32, 929–940 (2012). DC1
Materials and Methods
doi: 10.1126/scisignal.2001754; pmid: 21878679 doi: 10.1128/MCB.05921-11; pmid: 22215620
3. R. J. White, Nat. Rev. Mol. Cell Biol. 6, 69–78 (2005). 32. S. D. Westerheide, J. Anckar, S. M. Stevens Jr., Figs. S1 to S9
Reference (46)
doi: 10.1038/nrm1551; pmid: 15688068 L. Sistonen, R. I. Morimoto, Science 323, 1063–1066
Tables S1 to S5
4. K. M. Hannan et al., Mol. Cell. Biol. 23, 8862–8877 (2009). doi: 10.1126/science.1165946; pmid: 19229036
(2003). doi: 10.1128/MCB.23.23.8862-8877.2003; 33. O. Brandman et al., Cell 151, 1042–1054 (2012). 25 March 2013; accepted 24 May 2013
pmid: 14612424 doi: 10.1016/j.cell.2012.10.044; pmid: 23178123 10.1126/science.1238303

Bright Hot Impacts by Erupted Fragments Falling Back on the Sun: A
Template for Stellar Accretion
Fabio Reale et al.
Science 341, 251 (2013);
DOI: 10.1126/science.1235692


This article appears in the following subject collections:
Astronomy
http://www.sciencemag.org/cgi/collection/astronomy
REPORTS
hit the solar surface, they drive intense bright-
enings in the impact region, visible in all AIA
Bright Hot Impacts by Erupted

XUV and UV channels. Here, we focus our at-
tention on these impacts (movies S2 and S3).
Before hitting the surface, many downfalling
Fragments Falling Back on the Sun: fragments follow similar parabolic trajectories
(movie S2); one trajectory was traced to be close to
A Template for Stellar Accretion a free fall from combining AIA and STEREO-
SECCHI EUVI (22) images. We have measured
several impact speeds to be in the range of 300
Fabio Reale,1,2* Salvatore Orlando,2 Paola Testa,3 Giovanni Peres,1,2 to 450 km/s (see supplementary materials, sec-
Enrico Landi,4 Carolus J. Schrijver5 tion S1.2), similar to typical stellar accretion
flow speeds (23). We know that the fragments
Impacts of falling fragments observed after the eruption of a filament in a solar flare on are dense and cool, because we see them in ab-
7 June 2011 are similar to those inferred for accretion flows on young stellar objects. As imaged sorption in the XUV channels. From the amount
in the ultraviolet (UV)–extreme UV range by the Atmospheric Imaging Assembly onboard the of absorption, we constrain their density in the
Solar Dynamics Observatory, many impacts of dark, dense matter display uncommonly intense, range of 2 to 10 × 1010 cm−3 with a temperature
compact brightenings. High-resolution hydrodynamic simulations show that such bright spots, of ~3 × 104 K (see supplementary materials,

with plasma temperatures increasing from ~104 to ~106 kelvin, occur when high-density plasma section S1.1). The impacts are dispersed over a
(>>1010 particles per cubic centimeter) hits the solar surface at several hundred kilometers per large fraction of the solar surface (Fig. 1). The
second, producing high-energy emission as in stellar accretion. The high-energy emission comes typical cross section of impacting fragments is
from the original fragment material and is heavily absorbed by optically thick plasma, possibly around 2000 to 4000 km, whereas the long side
explaining the lower mass accretion rates inferred from x-rays relative to UV–optical–near of elongated fragments can extend to >>104 km.
infrared observations of young stars. The duration of the brightenings ranges between
1 and ≥6 min in the XUVand UV spectra. During
ass accretion from the circumstellar disk structures (14), as recently suggested for the clas- each brightening, the XUV emission typically in-
M onto the stellar surface plays an im-

portant role in the late phases of star
formation (1, 2). Young low-mass stars are pre-
sical T Tauri star TW Hydrae (accretion-fed co-
rona) (15, 16). The density and velocity of the
accreting material determine the temperature of
creases by factors of 2 to 5 over the unperturbed
conditions.
According to magnetograms measured with
sumably connected to and interact with circum- the hot plasma and the sinking depth (13). This the SDO/Helioseismic and Magnetic Imager
stellar disks through magnetic funnels (3). The complexity hampers a deep understanding of the (HMI) (fig. S4), the trajectories of the falling frag-
plasma accretes along the funnels at nearly free- dynamical and radiative properties of the plasma ments and most impacts occur in regions where
fall velocity (hundreds of kilometers/second) onto in the impact region and, ultimately, of the ac- the magnetic field is rather weak. Most fragments
the central young star. Most evidence for accre- cretion processes in young stars. do not decelerate while falling, and the ram and
tion originates from the region of impact of the Here, we study plasma impacts observed after thermal pressures of the fragments at the impacts
disk material onto the stellar chromosphere and the solar flare on 7 June 2011 at 6 UT that show are estimated to be substantially larger than the
includes infrared and optical excess emission in interesting analogies with inferred stellar accre- local magnetic pressure (supplementary materials,
lines and continuum (4–6) and a soft (energy < tion impacts. The flare was categorized as “M section S1.3). Thus, we can assume that the
0.7 keV) x-ray excess above the coronal emission class” on the Global Online Enrollment System magnetic field does not play an important role in
from dense (1011 to 1013 cm−3) and hot (2 to 4 MK) scale of the National Oceanic and Atmospheric the impacts, and we can describe the plasma evo-
plasma (7–9). Hot impact spots are observed in Administration. It was observed by the ultraviolet lution with a purely hydrodynamic model of
photospheric emission (10). (UV) and extreme UV (XUV) narrowband chan- plasma blobs, falling in a low-density corona with
Current models suggest that the impact region nels of the Atmospheric Imaging Assembly (AIA) an impact speed of 400 km/s and a density of
is rather complex because of the interplay be- (17) on board the Solar Dynamics Observatory 5 × 1010 cm−3 (both in the measured ranges). We
tween radiation and hydrodynamics (11–14): The (SDO) (18), with high spatial resolution (~ 0.6 arc ran several two-dimensional simulations (cylin-
streams might be highly structured in both den- sec per pixel) and high cadence (12 s) (19) (see drical geometry) to match the observed variety
sity and velocity, leading to inhomogeneous im- supplementary materials, section S1). During the of cases: a spherical droplet (with a radius of
pact spots; the impacting material is heated to flare, we clearly see, in all SDO/AIA XUV chan- 2000 km), a train of droplets, and an elongated
millions of degrees by kinetic energy dissipation nels, that a dense, dark filament is broken and fragment (stream). An accurate description of the
and might partially sink into the chromosphere, violently ejected (movie S1). The cloud propagates evolution required a very high spatial resolution
its emission being substantially absorbed by the outward at a speed of several hundred kilometers (down to ~5 km; see supplementary materials,
thick chromosphere and by the dense falling ma- per second and fragments in all directions. The section S2.1).
terial itself. The impact can also drive strong mo- fastest fragments escape in the form of a typical In our simulations, the very dense model frag-
tions and feed material into surrounding coronal coronal mass ejection (20), observed by the Solar ments fell freely in the tenuous ambient atmosphere
and Heliospheric Observatory/Large Angle and with a faint bow shock preceding them. As they hit
1 Spectrometric Coronagraph (21) white light tele- the chromosphere, they penetrated and smashed
Dipartimento di Fisica e Chimica, Universitàdi Palermo, Piazza
del Parlamento 1, 90134 Palermo, Italy. 2Istituto Nazionale di scope. Slower fragments fall back onto the solar against the layers whose pressure was equal to the
Astrofisica (INAF)/Osservatorio Astronomico di Palermo, Piazza surface. They are visible in all AIA XUV chan- ram pressure of the infalling fragment (movie S4).
del Parlamento 1, 90134 Palermo, Italy. 3Harvard-Smithsonian nels as dark, irregular, and moving strips on the The single spherical droplet was soon squashed
Center for Astrophysics, 60 Garden Street, MS 58, Cambridge, brighter background corona. During their fall, to a thin, dense, and hot layer and then bounced
MA 02138, USA. 4Department of Atmospheric, Oceanic and
Space Sciences, University of Michigan, Ann Arbor, MI 48109, the fragments change their morphology, stretch- back, forming an outward surge that expanded at a
USA. 5Lockheed Martin Advanced Technology Center, Palo Alto, ing and dividing further (19), but they generally speed of ~200 km/s. The surge reached tempera-
CA 94304, USA. maintain a coherent structure and remain dark tures above 105 K, but it cooled down in a few sec-
*Corresponding author. E-mail: reale@astropa.unipa.it throughout their trajectory. When the fragments onds, due to the rapid expansion and radiative losses.

REPORTS
For comparison with observations, we com- was almost featureless and hemispherical. The phases when coupling to the surrounding mag-
puted the emission that would be detected with plasma did not rise substantially above the sur- netic field becomes important. The size of the
the AIA XUV channels where we expected the face. We traced that most of the emission came emitting region and the temporal evolution of the
plasma to be mostly optically thin; that is, 94, from the original impacting droplet, with very emission in the simulations are in quantitative
131, 171, 193, 211, and 335 Å. In agreement little contribution from the ambient material. agreement with the observations. The observed
with the observation, the emission evolution was The simulations with the train of droplets and light curves (fig. S5) have shapes, flux, and time
similar in all the channels (Fig. 1 and movie S5). the stream showed longer-living (to more than scales similar to those derived from the simu-
The plasma became visible in this channel when 100 s) and larger-scale (up to ~10,000 km) bounc- lations (fig. S9). The observed brightenings gen-
it rose above the transition region while it bounced ing structures (movies S6 to S9). Although the erally last longer than the simulated ones, as can
back. Although the surge expanded over several emitting plasma was in a shell, integration along be expected because the impacting fragments may
tens of seconds, the plasma remained bright for a the line of sight kept the emission concentrated be much more extended than the modeled ones or
much shorter time (~20 s) (Fig. 1) and then faded along the central axis. The longer-lasting stream can be made by showers of smaller-scale blobs,
out rapidly, due to the cooling and rarefaction. emission was structured in sequences of evolving slightly displaced from one another.
The brightening region also remained small, not blobs and fringes (Fig. 1). Simulations either with a much smaller den-
much larger than the initial size of the droplet; it The morphology of the emission in the sim- sity of the falling droplet (5 × 109 cm−3) (movie
was the internal part of the outward surge and ulations matches the observations, up to the later S10) or a much smaller impact speed (250 km/s)
120 700

600
500
100
400
300
80 200
100
[arcsec]
0
60
40
20
0
0 20 40 60 80 100 120 20 40 60 80 100 120 20 40 60 80 100 120
Time = 136 sec Time = 116 sec Time = 140 sec
7 15 3 0.020 3 0.020 3 0.020
4 4 4
0.015 0.015 0.015
5 5 5
7 10 6 0.010 6 0.010 6 0.010
7 7 7
0.005 0.005 0.005
8 8 8
7 05 0.000 0.000 0.000
z [1010 cm]
7 00
6 5
6 0
7 15 16 6.5 16 6.5 16 6.5

14 6.0 14 6.0 14 6.0
5.5 5.5 5.5
7 10 12 12 12
5.0 5.0 5.0
10 4.5 10 4.5 10 4.5
7 05 8 4.0 8 4.0 8 4.0
z [1010 cm]
7 00
6 5
6 0
10 0 10 10 0 10 10 0 10
x [108 cm] x [108 cm] x [108 cm]
Fig. 1. Observed and modeled impact brightenings. (Top) Observed and stream) with a hydrodynamic model (at the labeled times since the
impacts of dense fragments erupted after an M-class solar flare on 7 June 2011 beginning of the simulation). The left side of each map is a cross section of the
in the SDO/AIA 171 Å channel. Three cases are shown, representative of small- emission in the 171 Å channel (logarithmic scale, DN cm−3 s−1), and the right
to large-scale impact brightenings (from left to right, taken at 7:26:25, side is the emission integrated along the horizontal line of sight; i.e., tan-
7:29:02, and 08:09:49 UT, with the plot origin at [x, y] = [357′′, –132′′], [468′′, gential to the solar surface (4223 DN per pixel per second). When integrating,
–137′′], [335′′, –130′′] from the disk center). The color palette is data number we exclude the heavily absorbed emission; i.e., by plasma at high density
(DN) per second per pixel. The white lines indicate how the width of the lower (>1010 cm−3) or below the transition region (see supplementary materials,
rows scales to the first row. (Middle) Simulated impacts of fragments that may section S2.2). (Bottom) Cross sections of the logarithm of the density (left, per
drive the respective brightenings (from left to right: droplet, train of droplets, cubic centimeter) and temperature (right, kelvin).

REPORTS
(movie S11) were unable to produce substantial neled by the magnetic field), the structure of the 19. D. E. Innes, R. H. Cameron, L. Fletcher, B. Inhester,
emission (fig. S9A), indicating that the sce- impact region (presenting hot plasma partially S. K. Solanki, Astron. Astrophys. 540, L10 (2012).
20. X. Cheng et al., Astrophys. J. 745, L5 (2012).
nario is coherent. Thus, our hydrodynamic sim- rooted in the chromosphere) is similar in the two 21. G. E. Brueckner et al., Sol. Phys. 162, 357–402 (1995).
ulations show that the hot and bright impacts cases. Moreover, as suggested recently (19), the 22. J.-P. Wuelser et al., Society of Photo-Optical
after the eruption are caused by the high den- accretion flows are likely to be frayed, even when Instrumentation Engineers (SPIE) Conference Series,
sity (>>1010 cm−3) and speed (free-fall speed, they are magnetically confined and stream along S. Fineschi, M. A. Gummin, Eds. (SPIE, San Diego, 2004),
vol. 5171, pp. 111–122.
300 to 450 km/s) of the downfalling debris. These straight tubes. Therefore, the dynamics and ener- 23. N. Calvet, E. Gullbring, Astrophys. J. 509, 802–818
values are close to those of the plasma involved getics spatially resolved in the solar observations (1998).
in stellar accretion flows. are a template and laboratory to study accretion 24. E. Gullbring, L. Hartmann, C. Briceno, N. Calvet,
For most stellar accretion flows [see (15) for processes in astrophysics. Astrophys. J. 492, 323–341 (1998).
25. J. Muzerolle, N. Calvet, C. Briceño, L. Hartmann,
an exception], the mass accretion rates derived L. Hillenbrand, Astrophys. J. 535, L47–L50 (2000).
from x-rays are consistently lower (by one or more References and Notes
1. D. Lynden-Bell, J. E. Pringle, Mon. Not. R. Astron. Soc. 26. R. L. Curran et al., Astron. Astrophys. 526, A104
orders of magnitude) than the corresponding rates 168, 603–637 (1974). (2011).
derived from UV–optical–near infrared observa- 2. L. Hartmann, Accretion Processes in Star Formation,
Acknowledgments: We thank B. De Pontieu for help and
tions (24–26). According to our analysis of a so- Cambridge Astrophysics Series (Cambridge Univ. Press,
New York, 1998). suggestions. F.R., G.P., and S.O. acknowledge support from
lar event, the impact of the dense fragments leads Italian Ministero dell’Università e Ricerca and Agenzia Spaziale
3. A. Koenigl, Astrophys. J. 370, L39–L43 (1991).
to detectable high-energy emission. From the mod- 4. C. Bertout, G. Basri, J. Bouvier, Astrophys. J. 330, Italiana, contract I/023/09/0. P.T. was supported by contract
el, we find that the mass of plasma responsible 350–373 (1988). SP02H1701R from Lockheed-Martin to the Smithsonian
for the brightenings in the 171 Å channel (see 5. A. Natta, L. Testi, S. Randich, Astron. Astrophys. 452, Astrophysical Observatory. P.T. and C.J.S. are supported by

245–252 (2006). NASA contract NNG04EA00C for the SDO AIA. E.L. is supported
supplementary materials, section S3) ranges by NASA grants NNX11AC20G and NNX10AQ58G and NSF
6. G. J. Herczeg, L. A. Hillenbrand, Astrophys. J. 681,
between 5 and 30% of the original mass of the 594–625 (2008). grant AGS-1154443. S.O. acknowledges partial support from
fragments. This is mainly because of the absorp- 7. J. H. Kastner, D. P. Huenemoerder, N. S. Schulz, the INAF. The software used in this work was, in part,
tion by the optically thick chromosphere and/or C. R. Canizares, D. A. Weintraub, Astrophys. J. 567, developed by the U.S. Department of Energy–supported
434–440 (2002). Advanced Simulation and Computing/Alliance Center for
by the dense part of the material outflowing after Astrophysical Thermonuclear Flashes at the University of
8. A. Telleschi, M. Güdel, K. R. Briggs, M. Audard, L. Scelsi,
the impact (depending on the orientation of the Chicago. We acknowledge the CINECA Award HP10CEG9MW
Astron. Astrophys. 468, 443–462 (2007).
impact region with respect to the line of sight). In 9. C. Argiroffi, A. Maggio, G. Peres, Astron. Astrophys. 465, and INAF/Osservatorio Astronomico di Palermo for
addition, the simulations tell us that droplets with L5–L8 (2007). high-performance computing resources and support. SDO
10. J.-F. Donati et al., Mon. Not. R. Astron. Soc. 386, data were supplied courtesy of the SDO/HMI and SDO/AIA
low density or low velocity are unable to produce consortia. SDO is the first mission to be launched for NASA’s
substantial emission; the observations tell us that 1234–1251 (2008).
11. J. J. Drake, 13th Cambridge Workshop on Cool Stars, Living With a Star Program. See the supplementary materials
some of the falling fragments produce no de- Stellar Systems and the Sun, vol. 560 of ESA Special for details on the data, data analysis, and modeling.
tectable brightening. Therefore, both the absorp- Publication, F. Favata, G. A. J. Hussain, B. Battrick, Eds.
tion of x-ray emission from dense plasma and the (European Space Agency, Noordwijk, 2005), p. 519.
www.sciencemag.org/cgi/content/full/science.1235692/DC1
wide range of velocity and density values of the 12. H. M. Günther, J. H. M. M. Schmitt, J. Robrade, C. Liefke,
infalling fragments contribute to underestimating 13. G. G. Sacco et al., Astron. Astrophys. 522, A55 (2010).
Supplementary Text
the mass accretion rate from the XUV band. Figs. S1 to S9
14. S. Orlando et al., Astron. Astrophys. 510, A71 (2010).
Table S1
Our simulations also show that the XUVemis- 15. N. S. Brickhouse, S. R. Cranmer, A. K. Dupree, G. J. M. Luna,
References (27–40)
sion arises from the original impacting material. S. Wolk, Astrophys. J. 710, 1835–1847 (2010).
Movies S1 to S18
16. A. K. Dupree et al., Astrophys. J. 750, 73 (2012).
Although our fragmented solar downflows differ 17. J. R. Lemen et al., Sol. Phys. 275, 17–40 (2012). 28 January 2013; accepted 5 June 2013
from the conceptual stellar accretion flows (which 18. W. D. Pesnell, B. J. Thompson, P. C. Chamberlin, Published online 20 June 2013;
are hypothesized to be continuous streams, chan- Sol. Phys. 275, 3–15 (2012). 10.1126/science.1235692
Switchable Static and Dynamic reaches an energy minimum (equilibrium) wherein

the ordered structure appears. Archetypical exam-
ples are structured block copolymers (3, 4), nano-
Self-Assembly of Magnetic Droplets particles (5, 6), nanorods (7), liquid crystals (8),
and hierarchical supramolecular systems (9). They
on Superhydrophobic Surfaces find applications in data storage (10) and struc-
tural colors (11), for instance. On the other hand,
dynamic self-assembly denotes a process in which
Jaakko V. I. Timonen,1*† Mika Latikka,1 Ludwik Leibler,2 Robin H. A. Ras,1* Olli Ikkala1* the structure forms when the system is forcefully
kept away from an energy minimum (out of equi-
Self-assembly is a process in which interacting bodies are autonomously driven into ordered librium) by continuous energy supply and dis-
structures. Static structures such as crystals often form through simple energy minimization, sipation (12, 13). Dynamic self-assembly is most
whereas dynamic ones require continuous energy input to grow and sustain. Dynamic systems are notably encountered in biological systems (14–16),
ubiquitous in nature and biology but have proven challenging to understand and engineer. Here,
we bridge the gap from static to dynamic self-assembly by introducing a model system based on 1
Department of Applied Physics, Aalto University (formerly
ferrofluid droplets on superhydrophobic surfaces. The droplets self-assemble under a static external Helsinki University of Technology), P.O. Box 15100, FI-02150
magnetic field into simple patterns that can be switched to complicated dynamic dissipative Espoo, Finland. 2Matière Molle et Chimie, UMR 7167 CNRS-
structures by applying a time-varying magnetic field. The transition between the static and dynamic ESPCI, Ecole Supérieure de Physique et Chimie Industrielles,
patterns involves kinetic trapping and shows complexity that can be directly visualized. 10 rue Vauquelin, 75005 Paris, France.
*Corresponding author. E-mail: jaakko.timonen@aalto.fi
(J.V.I.T.); robin.ras@aalto.fi (R.H.A.R.); olli.ikkala@aalto.fi (O.I.)
unctional patterns and structures are essen- subunits by autonomous self-assembly, which is
F tial in a wide variety of natural and engi-

neered systems (1). They often form of small
driven by free-energy gradients (2). Static self-
assembly denotes a process in which the system
†Present address: Non-Equilibrium Energy Research Center,
Northwestern University, 2145 Sheridan Road, Evanston, IL
60208, USA.

Switchable Static and Dynamic Self-Assembly of Magnetic Droplets
on Superhydrophobic Surfaces
Jaakko V. I. Timonen et al.
Science 341, 253 (2013);
DOI: 10.1126/science.1233775


found at:
REPORTS
(movie S11) were unable to produce substantial neled by the magnetic field), the structure of the 19. D. E. Innes, R. H. Cameron, L. Fletcher, B. Inhester,
emission (fig. S9A), indicating that the sce- impact region (presenting hot plasma partially S. K. Solanki, Astron. Astrophys. 540, L10 (2012).
20. X. Cheng et al., Astrophys. J. 745, L5 (2012).
nario is coherent. Thus, our hydrodynamic sim- rooted in the chromosphere) is similar in the two 21. G. E. Brueckner et al., Sol. Phys. 162, 357–402 (1995).
ulations show that the hot and bright impacts cases. Moreover, as suggested recently (19), the 22. J.-P. Wuelser et al., Society of Photo-Optical
after the eruption are caused by the high den- accretion flows are likely to be frayed, even when Instrumentation Engineers (SPIE) Conference Series,
sity (>>1010 cm−3) and speed (free-fall speed, they are magnetically confined and stream along S. Fineschi, M. A. Gummin, Eds. (SPIE, San Diego, 2004),
vol. 5171, pp. 111–122.
300 to 450 km/s) of the downfalling debris. These straight tubes. Therefore, the dynamics and ener- 23. N. Calvet, E. Gullbring, Astrophys. J. 509, 802–818
values are close to those of the plasma involved getics spatially resolved in the solar observations (1998).
in stellar accretion flows. are a template and laboratory to study accretion 24. E. Gullbring, L. Hartmann, C. Briceno, N. Calvet,
For most stellar accretion flows [see (15) for processes in astrophysics. Astrophys. J. 492, 323–341 (1998).
25. J. Muzerolle, N. Calvet, C. Briceño, L. Hartmann,
an exception], the mass accretion rates derived L. Hillenbrand, Astrophys. J. 535, L47–L50 (2000).
from x-rays are consistently lower (by one or more References and Notes
1. D. Lynden-Bell, J. E. Pringle, Mon. Not. R. Astron. Soc. 26. R. L. Curran et al., Astron. Astrophys. 526, A104
orders of magnitude) than the corresponding rates 168, 603–637 (1974). (2011).
derived from UV–optical–near infrared observa- 2. L. Hartmann, Accretion Processes in Star Formation,
Acknowledgments: We thank B. De Pontieu for help and
tions (24–26). According to our analysis of a so- Cambridge Astrophysics Series (Cambridge Univ. Press,
New York, 1998). suggestions. F.R., G.P., and S.O. acknowledge support from
lar event, the impact of the dense fragments leads Italian Ministero dell’Università e Ricerca and Agenzia Spaziale
3. A. Koenigl, Astrophys. J. 370, L39–L43 (1991).
to detectable high-energy emission. From the mod- 4. C. Bertout, G. Basri, J. Bouvier, Astrophys. J. 330, Italiana, contract I/023/09/0. P.T. was supported by contract
el, we find that the mass of plasma responsible 350–373 (1988). SP02H1701R from Lockheed-Martin to the Smithsonian
for the brightenings in the 171 Å channel (see 5. A. Natta, L. Testi, S. Randich, Astron. Astrophys. 452, Astrophysical Observatory. P.T. and C.J.S. are supported by

245–252 (2006). NASA contract NNG04EA00C for the SDO AIA. E.L. is supported
supplementary materials, section S3) ranges by NASA grants NNX11AC20G and NNX10AQ58G and NSF
6. G. J. Herczeg, L. A. Hillenbrand, Astrophys. J. 681,
between 5 and 30% of the original mass of the 594–625 (2008). grant AGS-1154443. S.O. acknowledges partial support from
fragments. This is mainly because of the absorp- 7. J. H. Kastner, D. P. Huenemoerder, N. S. Schulz, the INAF. The software used in this work was, in part,
tion by the optically thick chromosphere and/or C. R. Canizares, D. A. Weintraub, Astrophys. J. 567, developed by the U.S. Department of Energy–supported
434–440 (2002). Advanced Simulation and Computing/Alliance Center for
by the dense part of the material outflowing after Astrophysical Thermonuclear Flashes at the University of
8. A. Telleschi, M. Güdel, K. R. Briggs, M. Audard, L. Scelsi,
the impact (depending on the orientation of the Chicago. We acknowledge the CINECA Award HP10CEG9MW
impact region with respect to the line of sight). In 9. C. Argiroffi, A. Maggio, G. Peres, Astron. Astrophys. 465, and INAF/Osservatorio Astronomico di Palermo for
addition, the simulations tell us that droplets with L5–L8 (2007). high-performance computing resources and support. SDO
10. J.-F. Donati et al., Mon. Not. R. Astron. Soc. 386, data were supplied courtesy of the SDO/HMI and SDO/AIA
low density or low velocity are unable to produce consortia. SDO is the first mission to be launched for NASA’s
substantial emission; the observations tell us that 1234–1251 (2008).
11. J. J. Drake, 13th Cambridge Workshop on Cool Stars, Living With a Star Program. See the supplementary materials
some of the falling fragments produce no de- Stellar Systems and the Sun, vol. 560 of ESA Special for details on the data, data analysis, and modeling.
tectable brightening. Therefore, both the absorp- Publication, F. Favata, G. A. J. Hussain, B. Battrick, Eds.
tion of x-ray emission from dense plasma and the (European Space Agency, Noordwijk, 2005), p. 519.
wide range of velocity and density values of the 12. H. M. Günther, J. H. M. M. Schmitt, J. Robrade, C. Liefke,
infalling fragments contribute to underestimating 13. G. G. Sacco et al., Astron. Astrophys. 522, A55 (2010).
Supplementary Text
the mass accretion rate from the XUV band. Figs. S1 to S9
14. S. Orlando et al., Astron. Astrophys. 510, A71 (2010).
Table S1
Our simulations also show that the XUVemis- 15. N. S. Brickhouse, S. R. Cranmer, A. K. Dupree, G. J. M. Luna,
sion arises from the original impacting material. S. Wolk, Astrophys. J. 710, 1835–1847 (2010).
Movies S1 to S18
16. A. K. Dupree et al., Astrophys. J. 750, 73 (2012).
Although our fragmented solar downflows differ 17. J. R. Lemen et al., Sol. Phys. 275, 17–40 (2012). 28 January 2013; accepted 5 June 2013
from the conceptual stellar accretion flows (which 18. W. D. Pesnell, B. J. Thompson, P. C. Chamberlin, Published online 20 June 2013;
are hypothesized to be continuous streams, chan- Sol. Phys. 275, 3–15 (2012). 10.1126/science.1235692
Switchable Static and Dynamic reaches an energy minimum (equilibrium) wherein

the ordered structure appears. Archetypical exam-
ples are structured block copolymers (3, 4), nano-
Self-Assembly of Magnetic Droplets particles (5, 6), nanorods (7), liquid crystals (8),
and hierarchical supramolecular systems (9). They
on Superhydrophobic Surfaces find applications in data storage (10) and struc-
tural colors (11), for instance. On the other hand,
dynamic self-assembly denotes a process in which
Jaakko V. I. Timonen,1*† Mika Latikka,1 Ludwik Leibler,2 Robin H. A. Ras,1* Olli Ikkala1* the structure forms when the system is forcefully
kept away from an energy minimum (out of equi-
Self-assembly is a process in which interacting bodies are autonomously driven into ordered librium) by continuous energy supply and dis-
structures. Static structures such as crystals often form through simple energy minimization, sipation (12, 13). Dynamic self-assembly is most
whereas dynamic ones require continuous energy input to grow and sustain. Dynamic systems are notably encountered in biological systems (14–16),
ubiquitous in nature and biology but have proven challenging to understand and engineer. Here,
we bridge the gap from static to dynamic self-assembly by introducing a model system based on 1
Department of Applied Physics, Aalto University (formerly
ferrofluid droplets on superhydrophobic surfaces. The droplets self-assemble under a static external Helsinki University of Technology), P.O. Box 15100, FI-02150
magnetic field into simple patterns that can be switched to complicated dynamic dissipative Espoo, Finland. 2Matière Molle et Chimie, UMR 7167 CNRS-
structures by applying a time-varying magnetic field. The transition between the static and dynamic ESPCI, Ecole Supérieure de Physique et Chimie Industrielles,
patterns involves kinetic trapping and shows complexity that can be directly visualized. 10 rue Vauquelin, 75005 Paris, France.
*Corresponding author. E-mail: jaakko.timonen@aalto.fi
(J.V.I.T.); robin.ras@aalto.fi (R.H.A.R.); olli.ikkala@aalto.fi (O.I.)
unctional patterns and structures are essen- subunits by autonomous self-assembly, which is
F tial in a wide variety of natural and engi-

neered systems (1). They often form of small
driven by free-energy gradients (2). Static self-
assembly denotes a process in which the system
†Present address: Non-Equilibrium Energy Research Center,
Northwestern University, 2145 Sheridan Road, Evanston, IL
60208, USA.

REPORTS
but seminal works have shown that it can also local changes in the interactions between the superhydrophobic surface (23–27) with an ex-
be created artificially (12, 13, 17). elementary units and the external energy supply ternal magnetic field. In the following text, we first
Dynamic self-assembly has been suggested as (13). We approach this problem by introducing show how an external magnetic trigger can be
a route to adaptive systems beyond what static a model system that functions in the interface used for creating self-assembled ferrofluid drop-
self-assembly can offer (2, 13, 18). However, of static and dynamic self-assembly and can let populations from a single parent droplet. Then,
dynamic assembly is challenging to realize and be switched between them. The system relies on we demonstrate that the static equilibrium pat-
understand because it cannot be predicted through manipulation of mobile magnetic ferrofluid drop- terns can be transformed reversibly to dynamic
energy minimization. It is also sensitive to small, lets (19–21) on a low-friction lotus-leaf–like (22) dissipative patterns by feeding them energy through
Fig. 1. Magnetically trig-

gered ferrofluid droplet
division on a superhy-
drophobic surface. (A)
Schematic side-view of the
magnetic field geometry
of a cylindrical permanent
magnet (white lines, mag-
netic field; black lines, con-
stant field contours). State

of the droplet is indicated
as follows: 1, near-zero field
(nearly spherical droplet); 2,
weak field (slightly deformed
droplet); 3, strong field (con-
ical spiked droplet); and
4, above critical field (two
daughter droplets). (B) Photographs of a 20-ml ferrofluid droplet (movie S1) upon increasing the field
from 80 Oe (dH/dz 3.5 Oe/mm) to 680 Oe (dH/dz 66 Oe/mm). Notice the small asymmetry in daughter-
droplet sizes and the small satellite droplet between the two (fig. S3). (C) Frames of a high-speed video of the division (movie S2) and (D) the corresponding
distance between the daughter droplets as a function of time, approaching the distance determined by static self-assembly.
Fig. 2. Self-assembled droplet patterns in axisymmetric magnetic field. kinetic trapping. P1; high magnetic field; P2, low magnetic field. Error bars
(A) Scheme of the division instability. lc, critical wavelength; d, diameter. (B) indicate SD of three data sets. (D) Scheme of controlling the lattice constant
Photographs of the stepwise division and self-assembly of a 10-ml droplet. in the kinetically trapped patterns by adjusting the magnetic field curvature (c)
The largest droplet marked with a red circle (the diameter corresponds to lc,; and the magnetic moment of the droplets (m). a, nearest-neighbor distance. (E)
see Eq. 1) divides in each step (movie S3). Magnetic field strength was in- Snapshots of a 19-droplet pattern with two extreme periodicities at high and low
creased from 1.0 kOe (dH/dz 82 Oe/mm) for one droplet to 1.9 kOe (dH/dz magnetic
pffiffiffiffiffiffiffi fields. (F) The corresponding nearest-neighbor distance as a function
193 Oe/mm) for eight droplets. (C) Number of droplets as a function of of 5 m/c. Red dots denote experimental measurements; the black line indicates
increasing and decreasing magnetic field, showing hysteresis due to the the best linear fit.

REPORTS
an oscillating magnetic field. We highlight the face tension and density of the fluid, respective- droplet when the static magnetic field is switched
important role of kinetic trapping as a stabilizer ly, and g is the gravitational acceleration (30). In off (Fig. 2C and movie S3). This irreversibility
on both static and dynamic patterns and directly contrast, the gravitational force in our system is is due to the daughter droplets being kinetically
visualize the complexity of the transition. negligible compared with the magnetic force due trapped to remain separate; that is, there is a po-
In an elementary experiment, one droplet to the vertical field gradient; that is, dzd ðm0 HMV Þ, tential energy barrier between the global one-droplet
of aqueous ferrofluid (fig. S1) (28) is placed where m0 is permeability of vacuum, M is mag- energy minimum and any multidroplet pattern.
on a superhydrophobic surface and subjected to netization, and V is volume of the droplet. This The kinetic trapping is caused by the magnetic
a confining field of a cylindrical permanent mag- magnetic force can be up to two orders of mag- repulsion of the droplets and the nonwetting
net below the substrate (Fig. 1A and fig. S2). nitude larger than the gravitational force (fig. S5). nature of the substrate (which ensures that the
Gradually increasing the field strength H and the Thus, we can approximate the critical wavelength droplets are not physically connected). Later, we
vertical gradient dH/dz acting on the droplet (by in our system as show a route back to the original one-droplet state
decreasing the gap between the magnet and the through dynamic self-assembly.
surface) leads to a deformation of the droplet sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi The droplet pattern quickly rearranges to ac-
s
into a spiked cone and cleavage into two smaller lc ≈ 2p d
ð1Þ commodate the newly formed droplet after each
droplets at the critical field strength (Fig. 1B and dz ðm 0 HM Þ division (Fig. 2B and movie S3). The rearrange-
movie S1). The division takes a few tens of milli- ment is driven by the minimization of energy,
seconds (Fig. 1C and movie S2), after which the Importantly, Eq. 1 has a different interpretation which is a sum of the dipolar repulsion between
daughter droplets briefly oscillate before settling compared with the Rosensweig pattern. In this the magnetized droplets and their attraction toward
at their equilibrium separation (Fig. 1D). case, Eq. 1 does not determine the periodicity of the increasing gradient of the external magnet-

The division of the droplet is due to a com- the pattern but instead gives the criterion for the ic field
bination of high magnetic field and high vertical splitting: a droplet divides when the critical wave-
N N
magnetic field gradient and, therefore, does not length becomes smaller than the droplet diam- m0 mi mj
take place in a homogeneous magnetic field (fig. eter. This implies that the largest droplet is always U ¼ ∑ ∑ −
4p i¼1 j¼iþ1 jri − rj j3
S4). The division is related to the normal-field the most susceptible to dividing (Fig. 2A), which
instability of ferrofluids (Rosensweig instability) was experimentally confirmed when the magnetic N
m0 ∑ mi H − cjri j2
1
(19), but differs from it in several ways. The classic field was increased to generate larger droplet ð2Þ
Rosensweig pattern [see (29) for a demonstra- populations (Fig. 2B and movie S3), up to 75 drop- i¼1 2
tion] appears in a homogeneous perpendic- lets (fig. S6). In further contrast, the Rosensweig
ular magnetic field on a horizontal flat surface pattern is reversible (the pattern decays when where mi and mj are magnetic moments of the
of ferrofluid. Itp has the critical periodicity of
ffiffiffiffiffiffiffiffiffiffi the field is removed), but our patterns are not: droplets; ri and rj are positions of the droplets;
and c ¼ − ddrH2 is the confining curvature of the
2
lRosensweig
c ¼ 2p s=rg , where s and r are sur- the droplets do not coalesce back to a single
Fig.3.Morecomplicated
patterns. (A) Scheme
of changing the geom-
etry of the magnetic field
(top view). (B) Axisym-
metric magnetic field
(cylindrical magnet) and
the corresponding pat-
terns with five- and six-
fold symmetries and (C)
nonaxisymmetric magnetic
field (rectangular cuboid
magnet) and the corre-
sponding close-packed
ribbon patterns with an
overall twofold symme-
try (movie S5). All four
patterns in (B) and (C)
were created from a sin-
gle 20-ml parent droplet.
Hmax, maximum magnet-
ic field strength. (D)
Scheme for changing the
number of parent drop-
lets. (E) An example of
switching a hierarchical
pattern of 9 droplets with
fourfold symmetry with
three droplet sizes to
another hierarchical pat-
tern of 14 droplets with
twofold symmetry (movie
S6). The dividing droplet is marked with a solid color. The magnetic field strength was increased from 1060 Oe (dH/dz 61 Oe/mm) to 1240 Oe (dH/dz
72 Oe/mm).

REPORTS
magnetic field (fig. S2). The rearrangement close-packed ribbons with an overall twofold which cause shortening in the distances between
after each division is nearly perfect due to the symmetry (Fig. 3C and movie S5). On the other the droplets (fig. S8). Thus, the dynamic mag-
low friction that originates from the surface’s hand, starting with multiple differently sized par- netic force is an external trigger that can free the
high contact angles and low contact angle hys- ent droplets (Fig. 3D) makes switching between static pattern from its kinetic trap (Fig. 2C). It
teresis. In contrast, other surfaces with higher hierarchical patterns possible (Fig. 3E, fig. S7, and allows the number and sizes of the droplets to
contact angle hysteresis did not allow the mini- movie S6). be changed, leading to the rearrangement of the
mum energy patterns to be reached (movie S4). The static patterns transform to dynamic pattern.
The lattice constant of the patterns could be ones when they are provided with a sufficient Simulations showed that the coalescence and
adjusted by changing the ratio between the re- continuous energy feed to keep them away from rearrangement starts from droplet pairs whose
pulsion and attraction by changing the distance the energy minimum (Fig. 4A). We realized the distance is affected the most by the energy feed
between the surface and the magnet (Fig. 2, D energy feed by oscillating the permanent magnet (Fig. 4F and fig. S8). However, immediately
to F), in good agreement with the scaling relation
pffiffiffiffiffiffiffiffi horizontally below the substrate with amplitude after the first coalescence, the behavior of the
for the nearest-neighbor distance aº 5 m=c pre- A and frequency f (Fig. 4B). At low energy feed pattern becomes difficult to predict, as seen from
dicted from Eq. 2. Importantly, the periodicity rates (small amplitude and frequency), a typical the complex and seemingly chaotic coalescence
changes are completely reversible (provided that static seven-droplet pattern simply moves with and division of droplets (movie S8, mode IX).
the field is not increased above the next divi- the oscillating magnet as a whole and, thus, re- This complexity originates mostly from small
sion threshold). Thus, the droplets are free to mains close to the energy minimum (Fig. 4, C differences in initial positions and sizes of the
move, despite the number of droplets being ki- and D, mode I, and movie S7). However, the droplets (fig. S6). However, complexity can arise,
netically trapped (Fig. 2C). droplets start to coalesce above a threshold am- even in individually pipetted droplet patterns

More complicated patterns are readily achieved plitude and frequency, leading to the emergence (movie S10), and is actually seen also in switch-
by using other magnetic field geometries, such as of numerous dynamic patterns consisting of elon- ing of some static patterns (fig. S9 and movie
that of a nonaxisymmetric rectangular cuboid gated droplets and/or regular circular droplets S11). Yet, even the most chaotic transient drop-
magnet (Fig. 3A). The lacking axial symmetry (Fig. 4, D and E, modes II to X, and movies S7 let motions eventually stabilize into kinetical-
changes the typical five- and sixfold patterns and S8). The transition threshold is determined ly trapped patterns that are permanent until the
observed with cylindrical magnets (Fig. 3B) to by the time-dependent dissipative magnetic forces, oscillating field is switched off, after which the
Fig. 4. Reversible switching

between static and dynamic
self-assembly. (A) Scheme
of the switching. (B) Practical
realization of the energy feed
by horizontally oscillating the
permanent magnet below the
superhydrophobic surface. (C)
Phase diagram of the seven-
droplet pattern in the field of
1130 Oe (dH/dz 132 Oe/mm),
showing the transition bound-
ary (dashed lines). (D) Scheme
of mode I and photographs
of dynamic patterns that do
not change their appearance
during the cycle of the driving
field (movie S7). (E) Scheme of
mode VI and photographs of
dynamic patterns that change
their shape (but not the num-
ber of droplets) periodically
with the external magnetic
field (movie S8). (F) Photo-
graphs of the intermediate steps
in the formation of mode VIII
during 2.5 oscillations of the
driving field ( f = 3 Hz, A =
12.5 mm) (movie S9).

REPORTS
patterns decay back to the static seven-droplet to demonstrate switching between static and dy- 18. K. V. Tretiakov, K. J. M. Bishop, B. A. Grzybowski,
pattern within a fraction of a second. namic self-assembly and to show the usefulness Soft Matter 5, 1279–1284 (2009).
19. R. E. Rosensweig, Ferrohydrodynamics (Dover
Dynamic patterns rely critically on low friction of kinetic trapping that most often is viewed only Publications, New York, 1997).
and energy dissipation (movie S12). The geometry as a hindrance for assembly. The most diverse 20. J.-C. Bacri, F. Elias, in Morphogenesis: Origins of Patterns
of the dynamic patterns cannot be described in patterns were found to occur near the boundary and Shapes, P. Bourgine, A. Lesne, Eds. (Springer,
terms of energy minimization (Eq. 2), because where static patterns switch to dynamic ones. The Heidelberg, 2011), pp. 15–19.
21. R. Massart, IEEE Trans. Magn. 17, 1247–1248 (1981).
they are far from the energy minimum. Instead, transition is complex, and its detailed investigations 22. W. Barthlott, C. Neinhuis, Planta 202, 1–8 (1997).
the patterns are better described and rationalized will pave the way toward a better understanding 23. I. A. Larmour, S. E. J. Bell, G. C. Saunders, Angew. Chem.
as dynamic states wherein the droplet number is of the onset of dynamic dissipative self-assembly. Int. Ed. 46, 1710–1712 (2007).
kinetically trapped [as in the static patterns (Fig. 24. X. F. Gao et al., Adv. Mater. 19, 2213–2217 (2007).
References and Notes 25. H. Mertaniemi et al., Adv. Mater. 23, 2911–2914
2C)]. Patterns can be classified into those that al- (2011).
ternate (Fig. 4E) or do not alternate (Fig. 4D) during 1. P. Ball, Nature's Patterns: A Tapestry In Three Parts
(Oxford Univ. Press, Oxford, 2011). 26. H. Mertaniemi, R. Forchheimer, O. Ikkala, R. H. A. Ras,
the cycled energy feed. Alternation is caused by 2. G. M. Whitesides, B. Grzybowski, Science 295, Adv. Mater. 24, 5738–5743 (2012).
different mobilities of the droplets: Elongated ones 2418–2421 (2002). 27. T. Verho et al., Proc. Natl. Acad. Sci. U.S.A. 109,
10210–10213 (2012).
are more mobile because of the higher ratio be- 3. F. S. Bates et al., Science 336, 434–440 (2012).
4. A.-V. Ruzette, L. Leibler, Nat. Mater. 4, 19–31 (2005). 28. R. Massart, E. Dubois, V. Cabuil, E. Hasmonay, J. Magn.
tween the driving and dissipative forces. Magn. Mater. 149, 1–5 (1995).
5. Y. Xia et al., Nat. Nanotechnol. 6, 580–587 (2011).
Dynamic self-assembly can be used as a 6. E. V. Shevchenko, D. V. Talapin, N. A. Kotov, S. O’Brien, 29. D. Castelvecchi, Phys. Rev. Focus 15, 18 (2005).
switch between droplet patterns with different C. B. Murray, Nature 439, 55–59 (2006). 30. B. Berkovsky, V. Bashtovoi, IEEE Trans. Magn. 16,
numbers of droplets. In contrast to the division 7. K. Liu, N. Zhao, E. Kumacheva, Chem. Soc. Rev. 40, 288–297 (1980).

instability, dynamic self-assembly decreases the 656–671 (2011).
8. T. Kato, N. Mizoshita, K. Kishimoto, Angew. Chem. Int. Ed. Acknowledgments: We acknowledge financial support from
number of droplets. This allows the irrevers- the National Doctoral Programme in Materials Physics, Nokia
45, 38–68 (2006).
ibility of the droplet formation in static self- 9. O. Ikkala, G. ten Brinke, Science 295, 2407–2409 Research Center, the Academy of Finland, a European Research
assembly (Fig. 2C) to be overcome. For example, (2002). Council Advanced Grant, and the Finnish Agency of Technology
10. S. Park et al., Science 323, 1030–1033 (2009). and Innovation (TEKES). Electron microscopy was performed
driving any static pattern to the mode V dynamic using the devices of the Nanomicroscopy Center of Aalto
pattern (single elongated droplet) and decreas- 11. J. Yoon, W. Lee, E. L. Thomas, MRS Bull. 30, 721–726
(2005). University. We thank A. Walther (RWTH Aachen University),
ing the magnetic field while oscillating the mag- 12. B. A. Grzybowski, C. J. Campbell, Chem. Eng. Sci. 59, A. Kuzyk, and M. Kostiainen for comments on the manuscript
net allows recovery of the original one-droplet 1667–1676 (2004). and T. Huhtamäki and J. Korhonen for preparing the silicone
nanofilament surfaces.
state (movie S13). This is the final unit opera- 13. B. A. Grzybowski, C. E. Wilmer, J. Kim, K. P. Browne,
tion required to realize a complete cycle from a K. J. M. Bishop, Soft Matter 5, 1110–1128 (2009).
14. S. Kauffman, At Home in the Universe: The Search for
single liquid droplet to complicated static and Laws of Self-Organization and Complexity (Oxford Univ.
www.sciencemag.org/cgi/content/full/341/6143/253/DC1
dynamic patterns and, finally, back to the start- Materials and Methods
Press, Oxford, 1995).
Supplementary Text
ing state (fig. S10). 15. E. Karsenti, Nat. Rev. Mol. Cell Biol. 9, 255–262
(2008). Figs. S1 to S10
Externally driven magnetic droplets on super- References (31–35)
16. A. M. Mateus, N. Gorfinkiel, A. M. Arias, Semin. Cell
hydrophobic surfaces form a versatile model Dev. Biol. 20, 877–884 (2009).
Movies S1 to S14
system for studying and visualizing complicated 17. B. A. Grzybowski, H. A. Stone, G. M. Whitesides, Nature 7 December 2012; accepted 10 June 2013
phenomena in self-assembly. We used this model 405, 1033–1036 (2000). 10.1126/science.1233775
Ultrahigh Magnetoresistance at Room triplet configuration was first observed in quan-

tum dots (QDs) at cryogenic temperatures (5).
Temperature in Molecular Wires

Spin blockade can be lifted by spin relaxation,
mixing in singlet character. It has been shown
that hyperfine interaction can lift spin blockade
in QDs (6). The importance of hyperfine inter-
R. N. Mahato,1 H. Lülf,2 M. H. Siekman,1,3 S. P. Kersten,4 P. A. Bobbert,4 M. P. de Jong,1
action on spin dynamics has also been recognized
L. De Cola,2,5 W. G. van der Wiel1*
in the context of an intrinsic, room-temperature
MR effect in organic semiconductors (7, 8). Car-
Systems featuring large magnetoresistance (MR) at room temperature and in small magnetic fields are rier transport is influenced by spin-dependent
attractive owing to their potential for applications in magnetic field sensing and data storage. reactions, which are subject to the competition
Usually, the magnetic properties of materials are exploited to achieve large MR effects. Here, we report between an external magnetic field B and the
on an exceptionally large (>2000%), room-temperature, small-field (a few millitesla) MR effect in random hyperfine fields Bhf (~1 mT) of the nu-
one-dimensional, nonmagnetic systems formed by molecular wires embedded in a zeolite host crystal. clei. At small B (Fig. 1A), hyperfine interactions
This ultrahigh MR effect is ascribed to spin blockade in one-dimensional electron transport. Its generic
nature offers very good perspectives to exploit the effect in a wide range of low-dimensional systems.
1
NanoElectronics Group, MESA+ Institute for Nanotechnology,
n spintronic devices, the electron’s spin is netic materials separated by a nonmagnetic metal University of Twente, P.O. Box 217, 7500 AE, Enschede, Neth-
I exploited for information processing. Typ-

ically, these devices contain layered structures
with an electrical resistance that is dependent on
spacer layer, and tunnel magnetoresistance (TMR)
devices (3, 4), which have a tunnel barrier as the
spacer. Here, we explore entirely different physics
erlands. 2Institut de Science et d’Ingénierie Supramoléculaires
(ISIS), Université de Strasbourg, 8 Allée Gaspard Monge, 67000
Strasbourg, France. 3Transducers Science and Technology Group,
MESA+ Institute for Nanotechnology, University of Twente, P.O.
the relative orientation of the magnetization of in a nonmagnetic system, relying on a mecha- Box 217, 7500AE Enschede, Netherlands. 4Theory of Polymers
their magnetic layers; thus, the resistance can be nism akin to spin blockade in quantum devices. and Soft Matter, Department of Applied Physics, Eindhoven
University of Technology, P.O. Box 513, 5600 MB, Eindhoven,
altered by an external magnetic field, a phenom- The Pauli principle precludes that an electron Netherlands. 5Karlsruher Institut für Technologie (KIT), Institut für
enon called magnetoresistance (MR). Examples can tunnel into a state already occupied by an- Nanotechnologie Hermann-von-Helmholtz-Platz 1, D-76344
include giant magnetoresistance (GMR) devices other electron with the same spin. This spin Eggenstein-Leopoldshafen, Germany.
(1, 2), which are multilayer stacks of ferromag- blockade for two electrons starting from a spin- *Corresponding author. E-mail: w.g.vanderwiel@utwente.nl

Isotope Ratios of H, C, and O in CO2 and H2O of the Martian
Atmosphere
Chris R. Webster et al.
Science 341, 260 (2013);
DOI: 10.1126/science.1237961


found at:
REPORTS
RðBÞ − Rð0Þ Ið0Þ − IðBÞ tions show that with these ingredients, MC values 21. A. M. Nardes et al., Org. Electron. 9, 727–734 (2008).
MRðBÞ ≡ ¼ ð3Þ close to –100% indeed can be obtained with MC 22. H. Bentmann, A. A. Demkov, R. Gregory, S. Zollner,
Rð0Þ IðBÞ Phys. Rev. B 78, 205302 (2008).
(B) curves having the line shape of Eq. 2 (10). 23. S. K. Lee et al., J. Am. Chem. Soc. 121, 3513–3520
Further analysis should provide more insight (1999).
where R(B) is the resistance at magnetic field into the details of the mechanism behind the ef- 24. S. Ikeda et al., Appl. Phys. Lett. 93, 082508 (2008).
B and R(0) the resistance at zero field. We have fect, but the present experimental results clearly 25. G. Q. Gong et al., Appl. Phys. Lett. 67, 1783 (1995).
26. P. K. Siwach, H. K. Singh, O. N. Srivastava, J. Phys.
derived MRmax(V ) from the I-Vs for different point at spin blockade tuned by a competition Condens. Matter 20, 273201 (2008).
channel lengths at zero magnetic field and max- between the external magnetic field and the ran- 27. Y.-T. Zhang, Z.-Y. Chen, C.-C. Wang, Q. Jie, H.-B. Lü,
imum magnetic field, Bmax, according to MRmax(V) = dom internal hyperfine field. J. Magn. Magn. Mater. 321, 1199–1201 (2009).
[I(V,0) – I(V,Bmax)]/I(V,Bmax), (Fig. 4C). The MR 28. F. J. Yue et al., J. Phys. D Appl. Phys. 44, 025001 (2011).
References and Notes 29. J. Bai et al., Nat. Nanotechnol. 5, 655–659 (2010).
increases rapidly with decreasing voltage, reach- 30. V. N. Prigodin, J. D. Bergeson, D. M. Lincoln,
ing a maximum value of more than 2000% for 1. M. N. Baibich et al., Phys. Rev. Lett. 61, 2472–2475 (1988).
2. G. Binasch, P. Grünberg, F. Saurenbach, W. Zinn, A. J. Epstein, Synth. Met. 156, 757–761 (2006).
60-nm wire length when approaching 0 V. Be- Phys. Rev. B 39, 4828–4830 (1989). 31. P. Desai, P. Shakya, T. Kreouzis, W. P. Gillin, Phys. Rev. B
cause the current levels are below the noise limit 3. M. Jullière, Phys. Lett. 54, 225–226 (1975). 76, 235202 (2007).
close to 0 V, we have not been able to determine 4. J. S. Moodera, L. R. Kinder, T. M. Wong, R. Meservey,
Phys. Rev. Lett. 74, 3273–3276 (1995). Acknowledgments: We thank B. Koopmans, L. Abelmann,
MRmax here. L. P. Kouwenhoven, Y. Tokura, and S. Tarucha for fruitful
5. K. Ono, D. G. Austing, Y. Tokura, S. Tarucha, Science
The MR values in our molecular wires com- 297, 1313–1317 (2002). discussions. We acknowledge financial support from the
pare favorably with values reported for other 6. R. Hanson, L. P. Kouwenhoven, J. R. Petta, S. Tarucha, Netherlands Technology Foundation STW, the European
systems under similar conditions. The highest L. M. K. Vandersypen, Rev. Mod. Phys. 79, 1455–1455 Research Council, ERC Starting Grant nos. 240433 and

(2007). 280020, ERC Advanced Grant no. 247365, NanoScience
room-temperature TMR value reported to date Europe program (NanoSci-ERA), grant no. 11003, and the
7. T. Francis, Ö. Mermer, G. Veeraraghavan, M. Wohlgenannt,
is 600% for an epitaxial CoFeB/MgO/CoFeB New J. Phys. 6, 185 (2004). Foundation for Fundamental Research on Matter (FOM), part
magnetic tunnel junction (24). Collosal mag- 8. W. Wagemans et al., Spin 1, 93–108 (2011). of the Netherlands Organisation for Scientific Research (NWO).
netoresistance (CMR) manganites exhibit very 9. S. P. Kersten, S. Meskers, P. A. Bobbert, Phys. Rev. B 86, A link to the data reported here is included in the
045210 (2012). supplementary materials. R.N.M. and M.H.S. carried out the
large low-temperature MR at several tesla (25); magnetoresistance experiments and performed the data
10. Materials and methods are available as supplementary
however, room-temperature, low-field MR val- analysis. H.L. performed the zeolite L crystal synthesis and
materials on Science Online.
ues are a few tens of percents (26, 27). For nano- 11. G. Calzaferri, S. Huber, H. Maas, C. Minkowski, molecular loading. W.G.v.d.W. conceived the experiments and
composites containing magnetic nanoparticles, Angew. Chem. Int. Ed. 42, 3732–3758 (2003). planned and supervised the project. L.D.C. conceived the project
together with W.G.v.d.W. and supervised the zeolite synthesis
MR values similar to those in CMR experiments 12. P. Bornhauser, G. Calzaferri, J. Phys. Chem. 100,
and loading. S.P.K. and P.A.B. performed the theoretical analysis
have been demonstrated under comparable exper- 2035–2044 (1996).
13. P. A. Anderson et al., Dalton Trans. 19, 3122–3128 and numerical simulations. M.P.d.J. contributed to planning and
imental conditions (28). A large room-temperature (2004) and references therein. supervision. All authors discussed the results, provided important
MR effect of a few tens of percents was also re- 14. M. J. Kelly, J. Phys. Condens. Matter 7, 5507–5519 (1995). insights, and helped write the manuscript.
ported for a graphene nanoribbon field-effect 15. T. Bein, in Recent Advances and New Horizons in Zeolite
transistor (29). Science and Technology, H. Chon, S. I. Woo, S. E. Park, Supplementary Materials
Eds. (Elsevier, Amsterdam, 1996), pp. 295–322. www.sciencemag.org/cgi/content/full/science.1237242/DC1
We ascribe our very large MR values to the 16. D. J. Cardin, Adv. Mater. 14, 553 (2002). Materials and Methods
unique 1D character of our system. Indeed, non- 17. U. Simon, M. E. Franke, in Host-Guest-Systems Based on Figs. S1 to S4
structured DXP thin films show much lower Nanoporous Crystals, F. Laeri, F. Schüth, U. Simon, Data Files
MR values (Fig. 4D). When a ~40-nm DXP film M. Wark, Eds. (Wiley, Weinheim, 2003), pp. 364–378. References (32–37)
18. M. Álvaro et al., Chem. Mater. 18, 26–33 (2006).
is contacted with a PtSi CP-AFM tip in the 19. F. J. Jansen, R. A. Schoonheydt, J. Chem. Soc., Faraday 1 March 2013; accepted 14 June 2013
same setup, we measure a maximum MC of Trans. 69, 1338 (1973). Published online 4 July 2013;
around –20%. With a Pt wire of 250-mm diameter 20. Z. K. Tang et al., Science 292, 2462 (2001). 10.1126/science.1237242
instead of the PtSi tip, the maximum MC is
reduced to about –5%. These results strongly
Isotope Ratios of H, C, and O in CO2

suggest that confinement of the current path is
crucial for explaining our results, in line with
recent numerical studies (9). Explanations based
on spin-dependent interactions involving excited
states (30, 31) can be ruled out, because the MR
and H2O of the Martian Atmosphere
is also present—and even more pronounced—
Chris R. Webster,1* Paul R. Mahaffy,2 Gregory J. Flesch,1 Paul B. Niles,6 John H. Jones,7
below the DXP HOMO-LUMO gap (~2 eV).
Laurie A. Leshin,3 Sushil K. Atreya,4 Jennifer C. Stern,2 Lance E. Christensen,1 Tobias Owen,5
The fact that for DXP the energy required to form
Heather Franz,2 Robert O. Pepin,8 Andrew Steele,9 the MSL Science Team†
doubly negatively charged states is remarkably
small [~0.2 eV from cyclic voltammetry mea-
surements (23)] suggests that such states are Stable isotope ratios of H, C, and O are powerful indicators of a wide variety of planetary geophysical
involved in transport. We therefore propose that processes, and for Mars they reveal the record of loss of its atmosphere and subsequent interactions
the current is carried by electrons and that spin with its surface such as carbonate formation. We report in situ measurements of the isotopic ratios
blockade is caused by two electrons residing on of D/H and 18O/16O in water and 13C/12C, 18O/16O, 17O/16O, and 13C18O/12C16O in carbon dioxide, made
neighboring molecules attempting to form a doubly in the martian atmosphere at Gale Crater from the Curiosity rover using the Sample Analysis at
negatively charged molecule in a spin-singlet con- Mars (SAM)’s tunable laser spectrometer (TLS). Comparison between our measurements in the modern
figuration by hopping of one of the electrons. The atmosphere and those of martian meteorites such as ALH 84001 implies that the martian reservoirs
energetic disorder generally present in organic of CO2 and H2O were largely established ~4 billion years ago, but that atmospheric loss or surface
systems facilitates formation of doubly charged interaction may be still ongoing.
molecules at favorable locations. Moreover, the
presence of a positively charged potassium ion close he Sample Analysis at Mars (SAM) suite organic compounds and volatiles in rocks and
to a DXP molecule strongly reduces its LUMO
level, also facilitating double charging. Simula- T (1) on the Curiosity rover that landed in
August 2012 is conducting a search for
soils and characterizing the chemical and isotopic
composition of the modern atmosphere. Atmo-

REPORTS
spheric characterization is one of the exploration and 4600, respectively. These results have been It has been suggested that they are buffered by
goals of the Mars Science Laboratory (MSL) interpreted (17) as evidence for a two-stage evo- interaction with a larger O reservoir such as the
mission (2), and it is accomplished using SAM’s lution for Martian water—a significant early loss silicates in the crust, or crustal ice deposits (20),
tunable laser spectrometer (TLS) and its quadru- of water to space [before 3.9 billion years ago although this is complicated by evidence for dis-
pole mass spectrometer (QMS). Here we focus (Ga)], followed by only modest loss to space equilibrium between the crust and the atmosphere
on TLS measurements; a companion paper (3) during the past 4 billion years. Until Curiosity (21). Oxygen isotopes in CO2 and H2O are there-
focuses on those from the QMS. Results for non- landed, there had been no in situ measurements of fore likely indicators of more complex interactions
detection by TLS of atmospheric methane are the water isotopic species HDO and H218O. between the large reservoir of O in the hydro-
reported elsewhere (4). Oxygen isotopes in carbonates and sulfates sphere, lithosphere, and atmosphere of Mars.
Previous measurements of isotopes of H, N, from martian meteorites do not show any en- TLS is a two-channel tunable laser spectrom-
and noble gases in the martian atmosphere to date richment in d18O and therefore have not been eter that uses direct and second harmonic de-
(5) have indicated enrichment in the heavier used as indicators of atmospheric escape (18, 19). tection of infrared (IR) laser light absorbed after
isotopes, consistent with the idea of atmospheric
loss to space of the lighter isotopes (6, 7). Al-
though meteoritic analyses of d13C and d18O (8)
Table 1. Carbon dioxide isotope ratios ‰ T 2 SEM (standard error of the mean). *, not
in shergottite, nakhlite, and chassigny (SNC)– measured.
class meteorites are made at higher precision than
the atmospheric measurements to date, they are Measurement d13C d18O d17O d13C18O
challenged to correctly account for possible ter-
46 T 4 48 T 5 24 T 5 109 T 31

SAM-TLS
restrial contamination (9). Measurements of CO2
SAM-QMS (3) 45 T 12 * * *
isotopes at Mars and in particular d13C values
Phoenix lander (12) –2.5 T 4.3 31.0 T 5.7 * *
have not been consistent with atmospheric loss
Viking Neutral Mass Spectrometer (11) 23 T 43 7 T 44 * *
(10). Viking (11) measured d13C and d18O values
SNC meteorites (8, 12, 32) 36 T 10 3.9–5.4 T 0.1 ~0.53* d18O ~ d13C+ d18O
of 23 T 43 per mil (‰) and 7 T 44‰. Earth-based
ALH84001 meteoritic carbonate range (30, 31) 27 to 64 –9 to 26 ~0.53* d18O ~ d13C+ d18O
spectroscopy has suggested depleted values for
ALH84001 meteoritic carbonate mean value (31) 46 T 8 4.6 T 1.2 * *
d13C of –22 T 20‰ and d18O of 18 T 18‰ (9).
Earth telescopes (9) –22 T 21 18 T 18 * *
The recent Phoenix lander measured d13C and
d18O values for CO2 in the martian atmosphere of
–2.5 T 4.3‰ and 31 T 5.7‰, respectively (12).
Although uncertainties in these earlier atmo- Fig. 1. Spectral scan re-
spheric measurements of d13C and d18O overlap gions used by the TLS
(Table 1), their d13C values are in marked con- instrument. Calculated
trast to measurements of trapped CO2 in martian spectra from the HITRAN
meteorite EETA 79001, generally considered to database (36) for mea-
be closest to the true martian atmosphere and suring CO2 (A and B) and
which yielded a d13C of 36 T 10‰ (8). H2O isotope ratios (B).
For D/H in water, the difference in ground- The HDO line intensity has
state energies of HDO and its parent HHO are been increased by a factor
large enough to cause large changes in dD in of 6 to better represent
equilibrium and nonequilibrium (kinetic) pro- the martian environment.
cesses (13, 14), especially where condensation or
freezing occurs. For this reason, D/H has become
a universally important ratio to identify planetary
origin and history (7, 15). The 1988 telescopic
observation of D/H values in the martian at-
mosphere that were ~6 times that of Earth (7)
were pivotal in the idea of atmospheric loss to
space from a dense, warm, ancient atmosphere.
Initial measurements in meteorites (16) gave a
wide range of D/H values that may have included
terrestrial contributions. A more recent analy-
sis (17) of the ancient meteorite ALH84001
(~4 billion years old) and young meteorite Shergotty
(0.17 billion years old) produced dD values of 3000
1
Jet Propulsion Laboratory, California Institute of Technology,
Pasadena, CA 91109, USA. 2NASA Goddard Space Flight
Center, Greenbelt, MD 20771, USA. 3Rensselaer Polytechnic
Institute, Troy, NY 12180, USA. 4University of Michigan, Ann
Arbor, MI 48105, USA. 5University of Hawaii, Honolulu, HI 96822,
USA. 6NASA Johnson Space Center, Houston, TX 77058, USA.
7
University of Arizona, Tucson, AZ 85721, USA. 8University of
Minnesota, Minneapolis, MN 55455, USA. 9Carnegie Institu-
tion of Washington, Washington, DC 20015, USA.
*Corresponding author. E-mail: chris.r.webster@jpl.nasa.gov
†Mars Science Laboratory (MSL) Science Team authors and
affiliations are listed in the supplementary materials.

REPORTS
multipassing a sample cell (1). One laser source Fig. 2. Observed versus calcu-
is a near-IR tunable diode laser at 2.78 mm that lated spectra. A single spectrum
can scan two spectral regions containing CO2 and (middle section) downloaded from
H2O isotopic lines; the second laser source is an Curiosity (black), showing observed
interband cascade laser at 3.27 mm used for meth- enrichment in 13CO2 and 18OCO
ane detection alone (4). The near-IR laser makes compared to the calculated HITRAN
43 passes of a 20-cm-long sample (Herriott) cell spectrum (red) based on terrestrial
that is evacuated with a turbomolecular pump for (VPDB and VSMOW) isotope ratios
background scans, then filled to 0.7 mbar using (36). Both spectra are normalized
volume expansion of Mars air originally at ~7 mbar. in depth to the 16O12C16O line near
3590.1 cm−1 (Fig. 1). Ringing to
TLS scans over individual rovibrational lines in
the left side of the lines is ex-
two spectral regions near 2.78 mm; one centered
plained in (22).
at 3590 cm−1 for CO2 isotopes and a second
centered at 3594 cm−1 for both CO2 and H2O
isotopes (Figs. 1 and 2). The lines used in both
regions have no significant interferences. In the
3594 cm−1 region, the CO2 and H2O lines we
used interleave across the spectrum without in-
terference, allowing the determination of accu-

rate isotope ratios across widely varying CO2
and H2O abundances in both atmospheric and
evolved gas experiments. The laser scans every
second through the target spectral regions. Each 1-s
spectrum is then co-added on board in 2-min
periods, and the averaged spectra are then down-
linked as raw data during a given run, typically
of ~30 min duration. Data reported here were
collected from 6 days (martian sols 28, 53, 73,
79, 81, and 106). During data collection, the
Herriott cell and other optics are kept at 47° T 3°C
using a ramped heater that also serves to in-
crease the signal-to-noise ratio in spectra by
reducing the effect of interference fringes oc-
curring during the 2-min sample period. The
measured background amounts (empty cell) of
both CO2 and H2O are negligible and also re-
flect an insignificant contribution to the signal
from the instrument foreoptics. TLS is calibrated
using certified isotopic standards (22) that im-
prove the accuracy of isotope ratios over using
the more uncertain HITRAN (high-resolution
transmission molecular absorption) database spec-
tral parameters.
Our CO2 isotope ratios (Table 1, table S1, and
Fig. 3. Sol-by-sol mean values for CO2 isotope ratios. The mean values for all sols combined
Fig. 3) are given relative to Vienna Pee Dee
(dashed lines) are given in Table 1. See (22) for values and uncertainties of the individual sol data
belmnite (VPDB) for d13C and relative to Vienna plotted.
standard mean ocean water (VSMOW) for all
oxygen isotopes (13). The measured value of
d13C18O agrees within uncertainty to the sum of milion by volume) in martian air, and allowed us seasonal cycling (25) is unknown. The enriched
the individual d13C and d18O measurements, to retrieve a value for atmospheric dD, although atmospheric values contrast with the low primor-
providing a valuable check-sum on our results. with high uncertainty. Because our measured dial D/H values postulated for the martian mantle
Also, our measured value for d17O is half that of highly enriched dD values (Table 2 and table S2) (26) and are higher than those from our Rocknest
d18O, as predicted from mass-dependent fractiona- are clearly martian and not terrestrial, we attribute higher-temperature studies (23).
tion (d17O = 0.528 × d18O) and consistent with the high water mixing ratios to either high near- Modeling estimates of escape processes and
previous SNC meteorite analysis. The independent surface humidity (natural or from enhanced tem- atmospheric stability during Mars’ initial history
SAM QMS result for d13C of 45 T 12‰ (3) agrees peratures in the vicinity of the rover) or to water point to catastrophic loss of atmospheric mass,
well with that from TLS at 46 T 4‰, both values entrained from frozen or liquid sources on or near and suggest that many atmospheric species car-
notably disagreeing with the much lower Phoenix the heated inlet valve. Also, in evolved gas ex- rying records of early isotopic evolution did not
lander result (12) of –2.5 T 4.3‰. The sol-by-sol periments from pyrolysis of Rocknest fines (23), survive beyond approximately 3.7 to 4 Ga (27, 28).
data plotted in Fig. 3 is not over a sufficiently long water was seen coming off at relatively low tem- Carbonates in the ALH 84001 meteorite derived
period to assess possible seasonal variation in d13C peratures that we here identify as representative from an alteration event that occurred at ~3.9 Ga
or d18O. of the dD and d18O values of the martian at- (29) preserve our best record of these events. Mea-
Our measured water abundances of up to 1% mosphere. The TLS measurement of dD agrees surements of ALH 84001 carbonates show en-
by volume in our Herriott cell after atmospheric well with observations from ground-based tele- riched isotopic values of d13C = +27 to +64‰
intake exceed those expected (~150 parts per scopes (24), but the contribution from expected (30, 31), dD values of ~3000‰ (16, 17), and low

REPORTS
Table 2. Water isotope ratios ‰ T 2 SEM. *, not measured. 24. D. A. Fisher, Icarus 187, 430–441 (2007) and references
therein.
Measurement dD d18O 25. R. E. Novak et al., Bull. Am. Astron. Soc. 35, 660
(2005).
SAM-TLS atmosphere 4950 T 1,080 * 26. T. Usui, C. Alexander, J. Wang, J. Simon, J. Jones, Earth
SAM-TLS evolved water: Rocknest fines 230° to 430°C (23) 5880 T 60 84 T 10 Planet. Sci. Lett. 357-358, 119–129 (2012).
27. H. Lammer et al., Space Sci. Rev. 174, 113–154
Meteoritic crustal reservoirs (26) ~5000 * (2013).
Earth telescopes (24) 1700–8900 * 28. R. O. Pepin, Icarus 111, 289–304 (1994).
ALH 84001 (17) 3000 * 29. L. Borg, M. J. Drake, J. Geophys. Res. 110, E12S03
Shergotty USNM 321-1 (17) 4600 * (2005).
30. P. B. Niles, L. A. Leshin, Y. Guan, Geochim. Cosmochim.
Acta 69, 2931–2944 (2005).
31. J. W. Valley et al., Science 275, 1633–1638 (1997).
32. J. Farquhar, D. T. Johnston, Rev. Mineral. Geochem. 68,
d18O values (32). These values are similar to the 10. P. B. Niles et al., Space Sci. Rev. 174, 301–328 (2013). 463–492 (2008).
composition of the modern martian atmosphere, 11. A. O. Nier, M. B. McElroy, Science 194, 1298–1300 33. B. M. Jakosky, J. H. Jones, Nature 370, 328–329
(1976).
suggesting that the d13C, dD, and d18O of the 12. P. B. Niles, W. V. Boynton, J. H. Hoffman, D. W. Ming,
(1994).
34. M. Grott, A. Morschhauser, D. Breuer, E. Hauber, EPSL
martian atmosphere were enriched early and have D. Hamara, Science 329, 1334–1337 (2010). 308, 391–400 (2011).
not changed much over ~4 billion years. Our 13. R. E. Criss, Principles of Stable Isotope Composition 35. J. P. Bibring et al., Science 312, 400–404 (2006).
higher values of dD and d18O measured in the (Oxford Univ. Press, Oxford, 1999). 36. L. S. Rothman et al., J. Quant. Spectrosc. Radiat. Transf.
14. C. R. Webster, A. J. Heymsfield, Science 302, 1742–1745 110, 533–572 (2009).
atmosphere suggest that escape processes may (2003).

have also continued since 4.0 Ga, in accordance 15. P. Hartogh et al., Nature 478, 218–220 (2011). Acknowledgments: The research described here was carried
with a two-stage evolutionary process (17) de- 16. L. A. Leshin, S. Epstein, E. M. Stolper, Geochim. out at the Jet Propulsion Laboratory, California Institute of
scribed above. Cosmochim. Acta 60, 2635–2650 (1996). Technology, under a contract with NASA.
17. J. P. Greenwood, S. Itoh, N. Sakamoto, E. P. Vicenzi,
We observe large enrichments of d18O in at- H. Yurimoto, Geophys. Res. Lett. 35, L05203 (2008).
mospheric water vapor and CO2. The d18O val- 18. J. Farquar, M. H. Thiemens, J. Geophys. Res. 105,
ues of the water vapor are much larger than the www.sciencemag.org/cgi/content/full/341/6143/260/DC1
(2000).
d18O observed in carbonates and sulfates in mar- 19. C. S. Romanek et al., Nature 372, 655–657 (1994).
Supplementary Text
20. B. Jakosky, A. Zent, R. Zurek, Icarus 130, 87–95
tian meteorites and suggest that the oxygen in Figs. S1 to S3
(1997).
water vapor in the martian atmosphere is not 21. H. R. Karlsson, R. N. Clayton, E. K. Gibson Jr.,
Tables S1 to S4
in equilibrium with the crust (33, 34) and could T. K. Mayeda, Science 255, 1409–1411 (1992).
Reference (37)
MSL Science Team Authors and Affiliations
have been enriched in heavy isotopes through 22. See the supplementary materials on Science Online.
atmospheric loss. Another possibility is that the 23. L. A. Leshin et al., Lunar Planet. Sci. Conf., abstract 2220 18 March 2013; accepted 17 June 2013
(2013). 10.1126/science.1237961
elevated oxygen isotope values in the more abun-
dant martian CO2 are being transferred to the
water vapor through photochemical reactions in
the atmosphere. However, d18O values of CO2 in
Earth’s atmosphere are similarly elevated because Abundance and Isotopic Composition
of Gases in the Martian Atmosphere
of low-temperature equilibration between CO2
and H2O, and this process could also be operative
on Mars (12).
In addition to atmospheric loss, other pro-
cesses such as volcanic degassing and weathering from the Curiosity Rover
might act to change the isotopic composition
of the atmosphere through time. Estimates for Paul R. Mahaffy,1* Christopher R. Webster,2 Sushil K. Atreya,3 Heather Franz,1 Michael Wong,3
the magnitude of these two contributions over Pamela G. Conrad,1 Dan Harpold,1 John J. Jones,4 Laurie A. Leshin,5 Heidi Manning,6
the ~4-billion-year history of Mars vary widely Tobias Owen,7 Robert O. Pepin,8 Steven Squyres,9 Melissa Trainer,1 MSL Science Team†
(30, 34, 35), yet could have a strong impact on
the isotopic composition of the atmosphere and Volume mixing and isotope ratios secured with repeated atmospheric measurements taken
challenge the status quo model described above. with the Sample Analysis at Mars instrument suite on the Curiosity rover are: carbon dioxide
(CO2), 0.960(T0.007); argon-40 (40Ar), 0.0193(T0.0001); nitrogen (N2), 0.0189(T0.0003);
oxygen, 1.45(T0.09) × 10−3; carbon monoxide, < 1.0 × 10−3; and 40Ar/36Ar, 1.9(T0.3) × 103.
1. P. R. Mahaffy et al., Space Sci. Rev. 170, 401–478 The 40Ar/N2 ratio is 1.7 times greater and the 40Ar/36Ar ratio 1.6 times lower than values
(2012). reported by the Viking Lander mass spectrometer in 1976, whereas other values are generally
2. J. P. Grotzinger et al., Space Sci. Rev. 170, 5–56 consistent with Viking and remote sensing observations. The 40Ar/36Ar ratio is consistent with
(2012).
martian meteoritic values, which provides additional strong support for a martian origin of these
3. P. R. Mahaffy et al., Science 341, 263–266 (2013).
4. C. R. Webster, P. R. Mahaffy, S. K. Atreya, G. J. Flesch, rocks. The isotopic signature d13C from CO2 of ~45 per mil is independently measured with
K. A. Farley, Lunar Planet. Sci. Conf., abstract 1366 two instruments. This heavy isotope enrichment in carbon supports the hypothesis of substantial
(2013). atmospheric loss.
5. B. M. Jakosky, R. J. Phillips, Nature 412, 237–244
(2001).
he science and exploration goal of the tion requires comprehensive chemical charac-
6. M. B. McElroy, Y. L. Yung, A. O. Nier, Science 194, 70–72
(1976).
7. T. Owen, J.-P. Maillard, C. de Bergh, B. L. Lutz, Science
240, 1767–1770 (1988).
8. R. H. Carr, M. M. Grady, I. P. Wright, C. T. Pillinger,
T Mars Science Laboratory (MSL) (1) is to
advance our understanding of the po-
tential of the present or past martian environ-
terization. The first set of experiments of the
Sample Analysis at Mars (SAM) investigation
(2) (Fig. 1) of the Curiosity rover included mea-
Nature 314, 248–250 (1985).
ments to support life. An understanding of how surements of the chemical and isotopic com-
9. V. A. Krasnopolsky, J. P. Maillard, T. C. Owen, R. A. Toth, the present environment in Gale crater differs position of the atmosphere with sequences that
M. D. Smith, Icarus 192, 396–403 (2007). from the environment at the time of its forma- employed two of SAM’s three instruments. When

Abundance and Isotopic Composition of Gases in the Martian
Atmosphere from the Curiosity Rover
Paul R. Mahaffy et al.
Science 341, 263 (2013);
DOI: 10.1126/science.1237966


found at:
REPORTS
Table 2. Water isotope ratios ‰ T 2 SEM. *, not measured. 24. D. A. Fisher, Icarus 187, 430–441 (2007) and references
therein.
Measurement dD d18O 25. R. E. Novak et al., Bull. Am. Astron. Soc. 35, 660
(2005).
SAM-TLS atmosphere 4950 T 1,080 * 26. T. Usui, C. Alexander, J. Wang, J. Simon, J. Jones, Earth
SAM-TLS evolved water: Rocknest fines 230° to 430°C (23) 5880 T 60 84 T 10 Planet. Sci. Lett. 357-358, 119–129 (2012).
27. H. Lammer et al., Space Sci. Rev. 174, 113–154
Meteoritic crustal reservoirs (26) ~5000 * (2013).
Earth telescopes (24) 1700–8900 * 28. R. O. Pepin, Icarus 111, 289–304 (1994).
ALH 84001 (17) 3000 * 29. L. Borg, M. J. Drake, J. Geophys. Res. 110, E12S03
Shergotty USNM 321-1 (17) 4600 * (2005).
30. P. B. Niles, L. A. Leshin, Y. Guan, Geochim. Cosmochim.
Acta 69, 2931–2944 (2005).
31. J. W. Valley et al., Science 275, 1633–1638 (1997).
32. J. Farquhar, D. T. Johnston, Rev. Mineral. Geochem. 68,
d18O values (32). These values are similar to the 10. P. B. Niles et al., Space Sci. Rev. 174, 301–328 (2013). 463–492 (2008).
composition of the modern martian atmosphere, 11. A. O. Nier, M. B. McElroy, Science 194, 1298–1300 33. B. M. Jakosky, J. H. Jones, Nature 370, 328–329
(1976).
suggesting that the d13C, dD, and d18O of the 12. P. B. Niles, W. V. Boynton, J. H. Hoffman, D. W. Ming,
(1994).
34. M. Grott, A. Morschhauser, D. Breuer, E. Hauber, EPSL
martian atmosphere were enriched early and have D. Hamara, Science 329, 1334–1337 (2010). 308, 391–400 (2011).
not changed much over ~4 billion years. Our 13. R. E. Criss, Principles of Stable Isotope Composition 35. J. P. Bibring et al., Science 312, 400–404 (2006).
higher values of dD and d18O measured in the (Oxford Univ. Press, Oxford, 1999). 36. L. S. Rothman et al., J. Quant. Spectrosc. Radiat. Transf.
14. C. R. Webster, A. J. Heymsfield, Science 302, 1742–1745 110, 533–572 (2009).
atmosphere suggest that escape processes may (2003).

have also continued since 4.0 Ga, in accordance 15. P. Hartogh et al., Nature 478, 218–220 (2011). Acknowledgments: The research described here was carried
with a two-stage evolutionary process (17) de- 16. L. A. Leshin, S. Epstein, E. M. Stolper, Geochim. out at the Jet Propulsion Laboratory, California Institute of
scribed above. Cosmochim. Acta 60, 2635–2650 (1996). Technology, under a contract with NASA.
17. J. P. Greenwood, S. Itoh, N. Sakamoto, E. P. Vicenzi,
We observe large enrichments of d18O in at- H. Yurimoto, Geophys. Res. Lett. 35, L05203 (2008).
mospheric water vapor and CO2. The d18O val- 18. J. Farquar, M. H. Thiemens, J. Geophys. Res. 105,
ues of the water vapor are much larger than the www.sciencemag.org/cgi/content/full/341/6143/260/DC1
(2000).
d18O observed in carbonates and sulfates in mar- 19. C. S. Romanek et al., Nature 372, 655–657 (1994).
Supplementary Text
20. B. Jakosky, A. Zent, R. Zurek, Icarus 130, 87–95
tian meteorites and suggest that the oxygen in Figs. S1 to S3
(1997).
water vapor in the martian atmosphere is not 21. H. R. Karlsson, R. N. Clayton, E. K. Gibson Jr.,
Tables S1 to S4
in equilibrium with the crust (33, 34) and could T. K. Mayeda, Science 255, 1409–1411 (1992).
Reference (37)
MSL Science Team Authors and Affiliations
have been enriched in heavy isotopes through 22. See the supplementary materials on Science Online.
atmospheric loss. Another possibility is that the 23. L. A. Leshin et al., Lunar Planet. Sci. Conf., abstract 2220 18 March 2013; accepted 17 June 2013
(2013). 10.1126/science.1237961
elevated oxygen isotope values in the more abun-
dant martian CO2 are being transferred to the
water vapor through photochemical reactions in
the atmosphere. However, d18O values of CO2 in
Earth’s atmosphere are similarly elevated because Abundance and Isotopic Composition
of Gases in the Martian Atmosphere
of low-temperature equilibration between CO2
and H2O, and this process could also be operative
on Mars (12).
In addition to atmospheric loss, other pro-
cesses such as volcanic degassing and weathering from the Curiosity Rover
might act to change the isotopic composition
of the atmosphere through time. Estimates for Paul R. Mahaffy,1* Christopher R. Webster,2 Sushil K. Atreya,3 Heather Franz,1 Michael Wong,3
the magnitude of these two contributions over Pamela G. Conrad,1 Dan Harpold,1 John J. Jones,4 Laurie A. Leshin,5 Heidi Manning,6
the ~4-billion-year history of Mars vary widely Tobias Owen,7 Robert O. Pepin,8 Steven Squyres,9 Melissa Trainer,1 MSL Science Team†
(30, 34, 35), yet could have a strong impact on
the isotopic composition of the atmosphere and Volume mixing and isotope ratios secured with repeated atmospheric measurements taken
challenge the status quo model described above. with the Sample Analysis at Mars instrument suite on the Curiosity rover are: carbon dioxide
(CO2), 0.960(T0.007); argon-40 (40Ar), 0.0193(T0.0001); nitrogen (N2), 0.0189(T0.0003);
oxygen, 1.45(T0.09) × 10−3; carbon monoxide, < 1.0 × 10−3; and 40Ar/36Ar, 1.9(T0.3) × 103.
1. P. R. Mahaffy et al., Space Sci. Rev. 170, 401–478 The 40Ar/N2 ratio is 1.7 times greater and the 40Ar/36Ar ratio 1.6 times lower than values
(2012). reported by the Viking Lander mass spectrometer in 1976, whereas other values are generally
2. J. P. Grotzinger et al., Space Sci. Rev. 170, 5–56 consistent with Viking and remote sensing observations. The 40Ar/36Ar ratio is consistent with
(2012).
martian meteoritic values, which provides additional strong support for a martian origin of these
3. P. R. Mahaffy et al., Science 341, 263–266 (2013).
4. C. R. Webster, P. R. Mahaffy, S. K. Atreya, G. J. Flesch, rocks. The isotopic signature d13C from CO2 of ~45 per mil is independently measured with
K. A. Farley, Lunar Planet. Sci. Conf., abstract 1366 two instruments. This heavy isotope enrichment in carbon supports the hypothesis of substantial
(2013). atmospheric loss.
5. B. M. Jakosky, R. J. Phillips, Nature 412, 237–244
(2001).
he science and exploration goal of the tion requires comprehensive chemical charac-
6. M. B. McElroy, Y. L. Yung, A. O. Nier, Science 194, 70–72
(1976).
7. T. Owen, J.-P. Maillard, C. de Bergh, B. L. Lutz, Science
240, 1767–1770 (1988).
8. R. H. Carr, M. M. Grady, I. P. Wright, C. T. Pillinger,
T Mars Science Laboratory (MSL) (1) is to
advance our understanding of the po-
tential of the present or past martian environ-
terization. The first set of experiments of the
Sample Analysis at Mars (SAM) investigation
(2) (Fig. 1) of the Curiosity rover included mea-
Nature 314, 248–250 (1985).
ments to support life. An understanding of how surements of the chemical and isotopic com-
9. V. A. Krasnopolsky, J. P. Maillard, T. C. Owen, R. A. Toth, the present environment in Gale crater differs position of the atmosphere with sequences that
M. D. Smith, Icarus 192, 396–403 (2007). from the environment at the time of its forma- employed two of SAM’s three instruments. When

REPORTS
combined with composition and isotope data from rovers at latitudes of –2° and –15°. The GRS- large error bars with reported carbon isotopic
atmospheric gases trapped in martian meteor- derived Ar mixing ratio exhibits a large seasonal composition equivalent to d13CVPDB of 23 T
ites, measurements of the rate of atmospheric change by as much as a factor of 6 over the 50‰ (5).
escape from orbiting spacecraft, and studies of southern pole in winter (12) as the atmospheric Detailed characterization of the SNC (Shergotty,
atmosphere-surface exchange, SAM atmosphere CO2 undergoes an annual cycle of condensation Nakhla, and Chassigny) meteorites (14, 16, 21–24)
measurements are intended to constrain models and sublimation, producing a 25% change in the has revealed a combination of volatile abun-
of atmospheric loss and climate evolution over surface pressure. Although the GRS data exhibit dances and isotope systematics (14, 15, 21, 25–27)
geological time. no seasonal change in Ar in the equatorial region for noble gases, N2, and CO2 that is possible only
We report here on results from samples of the (12), APXS finds that Ar nearly tracks the sea- with origin on Mars or a very Mars-like parent
martian atmosphere analyzed by SAM’s quadru- sonal changes in surface pressure with a 2- to body (28). Although the Viking abundance and
pole mass spectrometer (QMS) and tunable laser 3-month phase lag (13). isotope measurements provided evidence sup-
spectrometer (TLS) during the first 105 sols (1 sol The mixing ratio of nitrogen can best be de- porting the hypothesis that the SNCs are from
is a martian day) of the landed mission. These termined by in situ measurements because Mars, the meteorites contain volatiles from other
experiments were among the first carried out meteorite measurements do not give definitive sources [for example, magmatic or possible com-
by SAM (Fig. 1) after several health checks of answers for this atmospheric gas. Variations in etary delivery (29)], in addition to trapped atmo-
the instrument. The experiments took place over isotopic composition of nitrogen in impact glasses spheric gases that cause some variations among
a period of several weeks from Mars solar of the martian shergotite meteorites EET79001 the meteorite values and differences between
longitude (3) of 163.7 to 211.2 (31 August to (14, 15), Zagami (16), and Tissint [e.g., (16, 17)] meteorite and Viking measurements. In addi-
21 November 2012) in Gale crater south of the suggest that, in these samples, atmospheric nitro- tion to the uncertainties introduced by multiple

equator (4.5°S, 137°E). All measurements were gen is mixed with an interior component with a sources of volatiles in the SNC meteorites, the
taken at night (table S1), and weighted means lower 15N/14N ratio. solubility of the volatiles and their partitioning
(table S2) are reported. The atmospheric CO2 isotope d13CVPDB (VPDB, in glass and in the constituent mineral phases
The mixing ratios of CO2, N2, Ar, O2, CO, Vienna Pee Dee belemnite) (18) has been re- affects both the abundance value and the iso-
Ne, Kr, and Xe at the martian surface were de- ported as –2.5 T 4.3 per mil (‰) from the Ther- topic signature, including those of the noble
termined by the mass spectrometers on the 1976 mal and Evolved Gas Analyzer (TEGA) mass gases (30, 31). The SAM data are therefore key
Viking Landers (4) more than 3 decades ago. spectrometer on the Phoenix lander (19) and as to constraining the atmospheric component
Mass spectrometers on the Viking aeroshells –22 T 20 ‰ from Fourier transform Earth-based of data obtainable from meteorites with in situ
also detected CO2, N2, Ar, CO, O2, O, and NO spectroscopy (20). The higher-uncertainty mea- observations.
(5) over an altitude range from 200 to 120 km, surements of the Viking lander found CO2 iso- Many previous composition measurements
approaching or reaching the homopause or the topes to be within 50‰ of terrestrial isotopes analyzed only a single or a small number of spe-
altitude below which the atmosphere is well (4). The aeroshell measurements had similarly cies. The SAM instrument suite, with the use of
mixed. Spectroscopic measurements of CO have
also been obtained from the Mars Reconnaissance
Orbiter [e.g., (6)], the Mars Express Spacecraft Fig. 1. The SAM suite SAM Components
located in the interior
(7, 8), and a number of ground-based observa- 1 – Solid sample inlet
1
of the Curiosity rover 2 – Quadrupole mass
tions [e.g., (9)], revealing long-term variations
uses three instruments spectrometer
correlated with solar activity (9). Recent Herschel to test either atmospher- 3 – Turbomolecular pump,
submillimeter observations (10) have provided heatpipe
ic gas or solid samples. 4 – Tunable laser
an additional measurement of the mixing ratios (Top) An image of SAM spectrometer
for CO (10) of 9.8(T1.5) × 10−4 and for O2 (11) 5 – Gas chromatograph
2
with the side panels re-
of 1.40(T0.12) × 10−3. The CO mixing ratio is moved. (Bottom) Mass
columns
found to vary by more than a factor of 4 (from spectrum of the martian

~3 × 10−4 to 1.2 × 10−3) seasonally at polar lat- atmosphere from sol 45, 3
itudes, with smaller changes in the equatorial with mass peaks labeled
region (6). The relative change in CO reflects for the main atmospheric
enrichment and depletion of noncondensable species. Isotopes of argon
5
4
volatiles during the condensation and sublima- appear above the back-
tion of CO2, the principal component of the mar- ground level (blue traces)
tian atmosphere. at mass/charge ratio (m/z)
Argon (40Ar) has also been monitored glob- 36, 38, and 40 (green ticks Sample (7.7 mbar)
ally from orbit by the gamma-ray spectrometer at top of plot). Primary Background
CO
(GRS) on the Mars Odyssey spacecraft and from ions from isotopologues O2
108 Ar
the martian surface by the alpha particle x-ray of CO2, containing 13C, 17C, N2
spectrometers (APXS) on the Mars Exploration and/or 18O, appear at m/z CO2
Detector counts sec–1
45, 46, and 47 (black ticks 107

1
NASA Goddard Space Flight Center, Greenbelt, MD 20771, at top of plot).
106
USA. 2Jet Propulsion Laboratory, California Institute of Technol-
ogy, 4800 Oak Grove Drive, Pasadena, CA 91109, USA. 3Uni-
versity of Michigan, Ann Arbor, MI 48109, USA. 4NASA Johnson 105
Space Center, Houston, TX 77058, USA. 5Rensselaer Polytechnic
Institute, Troy, NY 12180, USA. 6Concordia College, Moorhead, 104
MN 56562, USA. 7University of Hawaii, Honolulu, HI 96822,
USA. 8University of Minnesota, Minneapolis, MN 55455, USA.
9 103
Cornell University, Ithaca, NY 14853, USA.
*Corresponding author. E-mail: paul.r.mahaffy@nasa.gov
102
†MSL Science Team authors and affiliations are listed in the 12 16 20 24 28 32 36 40 44 48
supplementary materials. m/z

REPORTS
Table 1. Volume mixing ratio measurements Table 2. Isotopic composition measurements from Curiosity during the first 105 sols of the
from Curiosity during the first 105 sols of the landed mission. N/A, not applicable.
landed mission.
Isotopes Isotopic composition (QMS) Isotopic composition (TLS)
Gas Volume mixing ratio (QMS) 40 36 3
Ar/ Ar 1.9(T0.3) × 10 N/A
CO2 0.960(T0.007) d13CVPDB 45(T12) ‰* 46(T4) ‰†
Ar 0.0193(T0.0003) d18OVPDB N/A 48(T5) ‰
N2 0.0189(T0.0003) *d13CVPDB is derived from m/z 12 and 13. †d13CVPDB, as derived from m/z 45 and 46, is described in the supplementary
O2 1.45(T0.09) × 10−3 materials.
CO <1.0 × 10−3
measured by SAM is ~1.7 times greater than the radiogenic 40Ar over nonradiogenic 36Ar has
value reported from Viking measurement (4). Both been interpreted as evidence for significant loss
both the TLS and QMS, is able to make multiple, Ar and N2 are noncondensable and practically of the primordial martian atmosphere early in
high-precision composition measurements over inert gases on Mars, so their relative abundances the planet’s history, followed by partial degass-
the course of the mission. In addition, SAM’s are not expected to change considerably with time. ing of Ar. Subsequent loss to space is expected
QMS and TLS provide fully independent analy- We suspect that the difference from Viking re- to lead to enrichment of the 40Ar over 36Ar (42, 43)
ses of carbon isotopes. Repeat runs reported sults is due to different instrumental character- by the same processes that have reduced the
36
here were carried out at nearly the same time in istics rather than some unknown atmospheric Ar/38Ar ratio in the martian atmosphere. The

the early evening on Mars to validate results. process, although seasonal variation in N2 is yet latter ratio as inferred from EETA79001 glasses
Each measurement set of the type implemented to be tracked. The use on Mars of a turbomo- (15, 38) was found to be ~4, much different from
to date (32) represents a comprehensive anal- lecular pumping system (33), as well as repeated the terrestrial, chondritic, solar, and jovian (44)
ysis of the main constituents of the martian SAM analyses are expected to produce a more values which range in order from 5.3 to ~5.5. It is
atmosphere. accurate determination of the ratio of these gases notable that the 40Ar/36Ar ratio has not changed
SAM confirms the identity of the four most than the previous Viking in situ measurements appreciably since the ejection of EETA79001 from
abundant gases in the martian atmosphere, with whose mass spectrometers employed small ion the planet ~700,000 years ago. This provides a
CO2 being by far the major constituent. The pumps. constraint on the extent of very recent inputs of
SAM results for O2 (Tables 1 and 2) are con- The SAM QMS offers independent validation gas to the atmosphere from volcanic or com-
sistent with the recent Herschel (11) observations. of the d13CVPDB value in CO2 measured by the etary sources. The carbon dioxide isotope data
SAM secures an upper limit for the CO mixing TLS (34). The average of three SAM QMS at- support the hypothesis that a significant amount
ratio (Tables 1 and 2) that is consistent with the mosphere measurements gives d13CVPDB value of carbon has escaped from the martian atmo-
Herschel data and the mean of all remote sensing of 45 T 12 ‰, which is fully consistent with the sphere over time, resulting in preferential loss
spectroscopic measurements (~9 × 10−4). Differ- independently measured TLS value of 46 T4 ‰ of the lighter isotope of carbon and the observed
ences in CO mixing ratios are expected and are (34). This observed ~5% enrichment in the heavier enrichment in 13C (45). This implies that atmo-
related to the abovementioned seasonal effects, carbon isotope in the martian atmosphere com- spheric escape has dominated over exchange with
as dynamics and mixing rather than chemistry are pares well with previous measurements of 13C- unfractionated surface reservoirs that exist in the
expected to dominate the behavior of CO in the enriched carbon of atmospheric origin in martian crust or mantle.
homosphere due to the 3-year photochemical meteorite EETA79001 (22, 35). The data support
lifetime of CO. In addition to seasonal effects, the hypothesis that significant carbon has been References and Notes
localized, heterogeneous surface effects may also lost from the martian atmosphere over time by 1. J. P. Grotzinger et al., Space Sci. Rev. 170, 5–56
affect SAM measurements of CO because of sputtering (36). (2012).
2. P. R. Mahaffy et al., Space Sci. Rev. 170, 401–478
possible adsorption of CO onto the surface dur- The 40Ar/36Ar ratio of 1900 T 300 measured (2012).
ing cold martian nights—when SAM data were by SAM is within error of the trapped atmosphere 3. A Mars solar longitude of 180° represents the southern
collected—and reevaporation during warmer measured (15) to be 2050 T 170 in quenched spring equinox, where the southern polar region would
daytime. The Herschel observations, on the other shock-produced melts in martian meteorite be covered with carbon dioxide ice.
4. T. Owen et al., J. Geophys. Res. 82, 4635–4639
hand, are weighted to higher in the atmosphere. EETA79001 (27, 37, 38) but is considerably (1977).
Unlike CO, seasonal variation in O2 has not yet smaller than the value of 3000 T 500 reported 5. A. O. Nier, M. B. McElroy, Science 194, 1298–1300
been observed. by Viking (4). Laboratory studies of shock im- (1976).
The most notable differences between the plantation into silicate liquid have demonstrated 6. M. D. Smith, M. J. Wolff, R. T. Clancy, S. L. Murchie,
J. Geophys. Res. 114, E00D03 (2009).
SAM measurements and previous data are in that this process is a nearly quantitative recorder
7. F. Billebaud et al., Planet. Space Sci. 57, 1446–1457
the relative abundances of Ar and N2 and in the of atmospheric composition (39, 40), and the (2009).
isotopic compositions of Ar and CO2. The Ar/N2 implanted gases in meteorite shock-produced 8. G. Sindoni, V. Formisano, A. Geminale, Planet. Space Sci.
ratio and the N isotopes provide important con- melts compared with the Viking in situ measure- 59, 149–162 (2011).
straints to models for assessing the relative con- ments of the atmosphere have been used as the 9. V. A. Krasnopolsky, Icarus 190, 93–102 (2007).
10. P. Hartogh et al., Astron. Astrophys. 521, L48 (2010).
tributions of internal and atmospheric sources to best evidence to tie these meteorites to Mars 11. P. Hartogh et al., Astron. Astrophys. 521, L49 (2010).
gas inclusions in shock-produced glassy martian (14, 15, 21, 41). However, noble gases released 12. A. L. Sprague et al., Science 306, 1364–1367 (2004).
meteorites. The isotope data are important for from shock-produced glasses in EETA79001 13. T. E. Economou, R. T. Pierrehumbert, paper presented
constraining models of atmospheric evolution. contained at least three components (27): (i) at the 41st Lunar and Planetary Institute Science
Conference, abstract 2179, The Woodlands, TX, 1 March
Whereas Viking found nitrogen and argon to be martian air, (ii) terrestrial contamination, and 2010.
the second and third most abundant atmospheric (iii) a martian interior component with low 14. R. H. Becker, R. O. Pepin, Earth Planet. Sci. Lett. 69,
40
gases at 2.7 and 1.6% by volume, respectively, Ar/36Ar. 225–242 (1984).
SAM determines nearly equal volume mixing Even with the somewhat lower value mea- 15. R. C. Wiens, R. H. Becker, R. O. Pepin, Earth Planet.
Sci. Lett. 77, 149–158 (1986).
ratios for these constituents. Ar is found to be sured by SAM, the 40Ar/36Ar of the martian 16. K. Marti, J. S. Kim, A. N. Thakur, T. J. McCoy, K. Keil,
21% greater, whereas N2 is 30% lower than the atmosphere is highly elevated relative to the Science 267, 1981–1984 (1995).
Viking values. The resulting Ar/N2 ratio of 1.02 terrestrial ratio of 296. The enrichment in the 17. H. C. Aoudjehane et al., Science 338, 785–788 (2012).

REPORTS
18. VPDB is a terrestrial isotopes standard. 30. T. D. Swindle, AIP Conf. Proc. 341, 175 (1994). 39. R. C. Wiens, R. O. Pepin, Geochim. Cosmochim. Acta 52,
19. P. B. Niles, W. V. Boynton, J. H. Hoffman, D. W. Ming, 31. T. D. Swindle, J. H. Jones, J. Geophys. Res. 102, 1671 295–307 (1988).
D. Hamara, Science 329, 1334–1337 (2010). (1997). 40. D. Bogard, F. Horz, Meteoritics 21, 337 (1986).
20. V. A. Krasnopolsky, J. P. Maillard, T. C. Owen, R. A. Toth, 32. Details of measurement procedures and treatment of 41. R. O. Pepin, Nature 317, 473–475 (1985).
M. D. Smith, Icarus 192, 396–403 (2007). uncertainties are provided in the supplementary 42. R. O. Pepin, Icarus 111, 289–304 (1994).
21. D. D. Bogard, P. Johnson, Science 221, 651–654 (1983). materials on Science Online. 43. T. Owen, A. Bar-Nun, Icarus 116, 215–226 (1995).
22. R. H. Carr, M. M. Grady, I. P. Wright, C. T. Pillinger, 33. The turbomolecular pumps on SAM are expected to 44. P. R. Mahaffy et al., J. Geophys. Res. Planets 105,
Nature 314, 248–250 (1985). provide a more stable pressure of noble gas in the mass 15061–15071 (2000).
23. H. Y. McSween Jr., Meteoritics 29, 757–779 (1994). spectrometer ion source compared with the small ion 45. B. M. Jakosky, J. H. Jones, Rev. Geophys. 35, 1–16 (1997).
24. U. Ott, Geochim. Cosmochim. Acta 52, 1937–1948 (1988). pumps used on Viking.
25. U. Ott, F. Begemann, Meteoritics 20, 721 (1985). 34. C. R. Webster et al., Science 341, 260–263 (2013).
26. S. V. S. Murty, R. K. Mohapatra, Geochim. Cosmochim. 35. A. J. T. Jull, C. J. Eastoe, S. Cloudt, J. Geophys. Res. 102,
Acta 61, 5417–5428 (1997). 1663 (1997). www.sciencemag.org/cgi/content/full/341/6143/263/DC1
27. D. D. Bogard, D. H. Garrison, Meteorit. Planet. Sci. 33 36. B. M. Jakosky, R. O. Pepin, R. E. Johnson, J. L. Fox, Icarus Materials and Methods
Tables S1 and S2
(suppl.), 19 (1998). 111, 271–288 (1994).
28. A. H. Treiman, J. D. Gleason, D. D. Bogard, Planet. Space 37. D. H. Garrison, D. D. Bogard, Meteorit. Planet. Sci. 35 References (46, 47)
MSL Science Team Author List
Sci. 48, 1213–1230 (2000). (suppl.), A58 (2000).
29. T. Owen, A. Bar-Nun, AIP Conf. Proc. 341, 123–138 38. D. D. Bogard, R. N. Clayton, K. Marti, T. Owen, G. Turner, 18 March 2013; accepted 4 June 2013
(1994). Space Sci. Rev. 96, 425–458 (2001). 10.1126/science.1237966
Ice-Shelf Melting Around Antarctica sublimation. The products have been validated

with field data and an error propagation anal-
ysis (17) to a precision of 7 to 25%, average 10%,
E. Rignot,1,2* S. Jacobs,3 J. Mouginot,1 B. Scheuchl1 depending on location. We use the average SMB
for the years 1979 to 2010 to represent a longer-
We compare the volume flux divergence of Antarctic ice shelves in 2007 and 2008 with 1979 term state.
to 2010 surface accumulation and 2003 to 2008 thinning to determine their rates of melting and Ice-shelf thickness is from Operation IceBridge
mass balance. Basal melt of 1325 T 235 gigatons per year (Gt/year) exceeds a calving flux of (OIB) (18, 19) and BEDMAP-2 (1) (fig. S1, sup-
1089 T 139 Gt/year, making ice-shelf melting the largest ablation process in Antarctica. The giant plementary materials). It combines direct measure-
cold-cavity Ross, Filchner, and Ronne ice shelves covering two-thirds of the total ice-shelf area ments from radio-echo sounding, with indirect
account for only 15% of net melting. Half of the meltwater comes from 10 small, warm-cavity estimates from altimetry-derived ice-shelf sur-
Southeast Pacific ice shelves occupying 8% of the area. A similar high melt/area ratio is found face elevation assuming hydrostatic equilibri-
for six East Antarctic ice shelves, implying undocumented strong ocean thermal forcing on um with a nominal precision of 15 to 50 m (20).
their deep grounding lines. Flux gates are selected at the location of Inter-
ferometric Synthetic Aperture Radar (InSAR)–
he Antarctic Ice Sheet and its 58-m sea interactions extend from individual ice shelves derived grounding lines, which are more precise
T level equivalent (1) is buttressed along

most of its periphery by floating exten-
sions of land ice called ice shelves and floating
to circumpolar models at various resolutions,
but comparisons with observations are lim-
ited, and estimates of total ice-shelf meltwater
than those derived from photogrammetric tech-
niques or visible imagery (21), with accompanying
impacts on estimates of volume fluxes. Ice-front
ice tongues (Fig. 1). Ice shelves cover an area production have varied from 357 to 1600 giga- flux gates are at the seaward limit of the volume
>1.561 million km2, comparable in size to the tons per year (1 Gt = 1012 kg) (3, 7, 11). Gla- flux data, within 1 to 3 km of ice-front positions
Greenland Ice Sheet, and fringe 75% of Antarc- ciological estimates have focused on few ice digitized from a 150-m spacing mosaic of Ad-
tica’s coastline while collecting 20% of its snow- shelves (6, 12, 13) or near a fraction of glacier vanced Land Observing System (ALOS) Po-
fall over 11% of its area (2, 3). These features are grounding lines (14) due to incomplete velocity larimetric SAR (PALSAR) data for the years
nourished by the inflow of continental ice from and thickness mapping. 2007 and 2008.
grounded glaciers, surface accumulation, and Here, we present more accurate, higher-resolution Ice-shelf flow vector velocities are from InSAR
freezing of marine ice on their undersides. They glaciological estimates of ice-shelf melting around data collected in 2007 and 2008 and processed
lose mass to iceberg calving and basal melting the entire continent. At any point on an ice shelf at 450-m spacing (22). The average precision
along with topside sublimation and wind drift. Ice of thickness H and velocity vector v, the rate of in speed is 4 m/year and 1.7° in direction (fig.
shelves exert considerable control on glacier sta- ice-shelf thickening ∂H/∂t equals the sum of net S2). In the absence of vertical shear on floating
bility and Antarctic Ice Sheet mass balance (4–6) surface mass balance SMB minus net basal melt- ice, the surface-derived velocity is equivalent to a
and play important roles in ocean stratification ing B minus the lateral divergence in volume flux depth-averaged velocity. We surveyed 99.5%
and bottom water formation (7). Hv (15). A negative value of B indicates the of Antarctic ice-shelf area in 2007 and 2008
The traditional view of ablation from Ant- freeze-on of marine ice. The calculation of vol- (Table 1), or 1.554 million km2, excluding a few
arctic ice shelves has been that it occurs most- ume flux divergence on a point per point basis smaller ice shelves where ice thickness is not
ly by iceberg calving, with basal melting only yields the distribution of freeze/melt (Fig. 1). well known (table S1). Drainage boundaries
contributing 10 to 28% of the total mass loss The integration of the total inflow and outflow between ice shelves, including the eastern and
(3–6). Estimates of ice-shelf meltwater produc- within the ice-shelf perimeters yields the area- western Ross, are defined by flow vector direc-
tion derived from oceanographic data (8–10, average melt rate and total meltwater production tion. Ice rises and islands are excluded from the
e.g.) are impractical for synoptic circumpolar (Table 1). ice-shelf area estimates but included in the SMB
coverage. Numerical simulations of ice-ocean For SMB, we use output products from the Re- calculation.
gional Atmospheric and Climate Model RACMO2 Ice-shelf thickening ∂H/∂t for the period 2003
1 (16), which is forced at the lateral boundary and to 2008 is calculated using the procedure in
Department of Earth System Science, University of California,
Irvine, CA 92697, USA. 2Jet Propulsion Laboratory, Pasadena, sea surface by global reanalyses of the European (23), with an error dependent on firn depth cor-
CA 91109, USA. 3Lamont-Doherty Earth Observatory, Columbia Centre for Medium-Range Weather Forecasts. rections (fig. S3). The results are combined
University, Palisades, NY 10964, USA. RACMO2 includes surface meltwater retention with SMB and the flux divergence to calculate
*Corresponding author. E-mail: erignot@uci.edu due to refreezing, evaporation, wind drift, and B, with a precision dominated by uncertainties in

Ice-Shelf Melting Around Antarctica
E. Rignot et al.
Science 341, 266 (2013);
DOI: 10.1126/science.1235798


Atmospheric Science
http://www.sciencemag.org/cgi/collection/atmos
Oceanography
http://www.sciencemag.org/cgi/collection/oceans
REPORTS
18. VPDB is a terrestrial isotopes standard. 30. T. D. Swindle, AIP Conf. Proc. 341, 175 (1994). 39. R. C. Wiens, R. O. Pepin, Geochim. Cosmochim. Acta 52,
19. P. B. Niles, W. V. Boynton, J. H. Hoffman, D. W. Ming, 31. T. D. Swindle, J. H. Jones, J. Geophys. Res. 102, 1671 295–307 (1988).
D. Hamara, Science 329, 1334–1337 (2010). (1997). 40. D. Bogard, F. Horz, Meteoritics 21, 337 (1986).
20. V. A. Krasnopolsky, J. P. Maillard, T. C. Owen, R. A. Toth, 32. Details of measurement procedures and treatment of 41. R. O. Pepin, Nature 317, 473–475 (1985).
M. D. Smith, Icarus 192, 396–403 (2007). uncertainties are provided in the supplementary 42. R. O. Pepin, Icarus 111, 289–304 (1994).
21. D. D. Bogard, P. Johnson, Science 221, 651–654 (1983). materials on Science Online. 43. T. Owen, A. Bar-Nun, Icarus 116, 215–226 (1995).
22. R. H. Carr, M. M. Grady, I. P. Wright, C. T. Pillinger, 33. The turbomolecular pumps on SAM are expected to 44. P. R. Mahaffy et al., J. Geophys. Res. Planets 105,
Nature 314, 248–250 (1985). provide a more stable pressure of noble gas in the mass 15061–15071 (2000).
23. H. Y. McSween Jr., Meteoritics 29, 757–779 (1994). spectrometer ion source compared with the small ion 45. B. M. Jakosky, J. H. Jones, Rev. Geophys. 35, 1–16 (1997).
24. U. Ott, Geochim. Cosmochim. Acta 52, 1937–1948 (1988). pumps used on Viking.
25. U. Ott, F. Begemann, Meteoritics 20, 721 (1985). 34. C. R. Webster et al., Science 341, 260–263 (2013).
26. S. V. S. Murty, R. K. Mohapatra, Geochim. Cosmochim. 35. A. J. T. Jull, C. J. Eastoe, S. Cloudt, J. Geophys. Res. 102,
Acta 61, 5417–5428 (1997). 1663 (1997). www.sciencemag.org/cgi/content/full/341/6143/263/DC1
27. D. D. Bogard, D. H. Garrison, Meteorit. Planet. Sci. 33 36. B. M. Jakosky, R. O. Pepin, R. E. Johnson, J. L. Fox, Icarus Materials and Methods
Tables S1 and S2
(suppl.), 19 (1998). 111, 271–288 (1994).
28. A. H. Treiman, J. D. Gleason, D. D. Bogard, Planet. Space 37. D. H. Garrison, D. D. Bogard, Meteorit. Planet. Sci. 35 References (46, 47)
MSL Science Team Author List
Sci. 48, 1213–1230 (2000). (suppl.), A58 (2000).
29. T. Owen, A. Bar-Nun, AIP Conf. Proc. 341, 123–138 38. D. D. Bogard, R. N. Clayton, K. Marti, T. Owen, G. Turner, 18 March 2013; accepted 4 June 2013
(1994). Space Sci. Rev. 96, 425–458 (2001). 10.1126/science.1237966
Ice-Shelf Melting Around Antarctica sublimation. The products have been validated

with field data and an error propagation anal-
ysis (17) to a precision of 7 to 25%, average 10%,
E. Rignot,1,2* S. Jacobs,3 J. Mouginot,1 B. Scheuchl1 depending on location. We use the average SMB
for the years 1979 to 2010 to represent a longer-
We compare the volume flux divergence of Antarctic ice shelves in 2007 and 2008 with 1979 term state.
to 2010 surface accumulation and 2003 to 2008 thinning to determine their rates of melting and Ice-shelf thickness is from Operation IceBridge
mass balance. Basal melt of 1325 T 235 gigatons per year (Gt/year) exceeds a calving flux of (OIB) (18, 19) and BEDMAP-2 (1) (fig. S1, sup-
1089 T 139 Gt/year, making ice-shelf melting the largest ablation process in Antarctica. The giant plementary materials). It combines direct measure-
cold-cavity Ross, Filchner, and Ronne ice shelves covering two-thirds of the total ice-shelf area ments from radio-echo sounding, with indirect
account for only 15% of net melting. Half of the meltwater comes from 10 small, warm-cavity estimates from altimetry-derived ice-shelf sur-
Southeast Pacific ice shelves occupying 8% of the area. A similar high melt/area ratio is found face elevation assuming hydrostatic equilibri-
for six East Antarctic ice shelves, implying undocumented strong ocean thermal forcing on um with a nominal precision of 15 to 50 m (20).
their deep grounding lines. Flux gates are selected at the location of Inter-
ferometric Synthetic Aperture Radar (InSAR)–
he Antarctic Ice Sheet and its 58-m sea interactions extend from individual ice shelves derived grounding lines, which are more precise
T level equivalent (1) is buttressed along

most of its periphery by floating exten-
sions of land ice called ice shelves and floating
to circumpolar models at various resolutions,
but comparisons with observations are lim-
ited, and estimates of total ice-shelf meltwater
than those derived from photogrammetric tech-
niques or visible imagery (21), with accompanying
impacts on estimates of volume fluxes. Ice-front
ice tongues (Fig. 1). Ice shelves cover an area production have varied from 357 to 1600 giga- flux gates are at the seaward limit of the volume
>1.561 million km2, comparable in size to the tons per year (1 Gt = 1012 kg) (3, 7, 11). Gla- flux data, within 1 to 3 km of ice-front positions
Greenland Ice Sheet, and fringe 75% of Antarc- ciological estimates have focused on few ice digitized from a 150-m spacing mosaic of Ad-
tica’s coastline while collecting 20% of its snow- shelves (6, 12, 13) or near a fraction of glacier vanced Land Observing System (ALOS) Po-
fall over 11% of its area (2, 3). These features are grounding lines (14) due to incomplete velocity larimetric SAR (PALSAR) data for the years
nourished by the inflow of continental ice from and thickness mapping. 2007 and 2008.
grounded glaciers, surface accumulation, and Here, we present more accurate, higher-resolution Ice-shelf flow vector velocities are from InSAR
freezing of marine ice on their undersides. They glaciological estimates of ice-shelf melting around data collected in 2007 and 2008 and processed
lose mass to iceberg calving and basal melting the entire continent. At any point on an ice shelf at 450-m spacing (22). The average precision
along with topside sublimation and wind drift. Ice of thickness H and velocity vector v, the rate of in speed is 4 m/year and 1.7° in direction (fig.
shelves exert considerable control on glacier sta- ice-shelf thickening ∂H/∂t equals the sum of net S2). In the absence of vertical shear on floating
bility and Antarctic Ice Sheet mass balance (4–6) surface mass balance SMB minus net basal melt- ice, the surface-derived velocity is equivalent to a
and play important roles in ocean stratification ing B minus the lateral divergence in volume flux depth-averaged velocity. We surveyed 99.5%
and bottom water formation (7). Hv (15). A negative value of B indicates the of Antarctic ice-shelf area in 2007 and 2008
The traditional view of ablation from Ant- freeze-on of marine ice. The calculation of vol- (Table 1), or 1.554 million km2, excluding a few
arctic ice shelves has been that it occurs most- ume flux divergence on a point per point basis smaller ice shelves where ice thickness is not
ly by iceberg calving, with basal melting only yields the distribution of freeze/melt (Fig. 1). well known (table S1). Drainage boundaries
contributing 10 to 28% of the total mass loss The integration of the total inflow and outflow between ice shelves, including the eastern and
(3–6). Estimates of ice-shelf meltwater produc- within the ice-shelf perimeters yields the area- western Ross, are defined by flow vector direc-
tion derived from oceanographic data (8–10, average melt rate and total meltwater production tion. Ice rises and islands are excluded from the
e.g.) are impractical for synoptic circumpolar (Table 1). ice-shelf area estimates but included in the SMB
coverage. Numerical simulations of ice-ocean For SMB, we use output products from the Re- calculation.
gional Atmospheric and Climate Model RACMO2 Ice-shelf thickening ∂H/∂t for the period 2003
1 (16), which is forced at the lateral boundary and to 2008 is calculated using the procedure in
Department of Earth System Science, University of California,
Irvine, CA 92697, USA. 2Jet Propulsion Laboratory, Pasadena, sea surface by global reanalyses of the European (23), with an error dependent on firn depth cor-
CA 91109, USA. 3Lamont-Doherty Earth Observatory, Columbia Centre for Medium-Range Weather Forecasts. rections (fig. S3). The results are combined
University, Palisades, NY 10964, USA. RACMO2 includes surface meltwater retention with SMB and the flux divergence to calculate
*Corresponding author. E-mail: erignot@uci.edu due to refreezing, evaporation, wind drift, and B, with a precision dominated by uncertainties in

REPORTS
ice-front thickness and firn depth corrections (ta- western sides, consistent with an oceanic circu- 0.076°C to the thermal driving of seawater that
ble S1). We also calculate the results for ∂H/∂t = lation during which seawater is first cooled, fresh- may have started out near the sea-surface freezing
0, i.e., no ice-shelf thickness change, to obtain a ened, and made more buoyant by melting. point. Differences in observed melt rate may also
reference rate Bss corresponding to the amount The highest melt rates are detected in the be accentuated by variations in flushing time and
of freezing or melting that would be required to Southeast Pacific sector of the Antarctic Penin- tidal activity (24).
maintain an ice shelf in “steady state” for 2007 sula and West Antarctica, from the northern end Total ice inflow and outflow for each ice
and 2008 (fig. S4). of George VI to the western end of Getz Ice Shelf. shelf is summarized in Fig. 1 and Table 1. Ice-
The freeze/melt distribution confirms that On slow-moving to nearly stationary ice shelves front flux is a proxy for, but not identical to,
basal melting is strongest near the grounding like the Wilkins, George VI, Abbot, and Sulzberger, iceberg calving, which occurs at irregular time
zones of major glaciers and along the ice fronts basal melting entirely consumes the inflow of intervals ranging from days to decades. The
of some of the largest ice shelves, especially individual glaciers within a few km of their higher basal melting near some ice-shelf fronts
Ronne (Fig. 1). Ice-shelf melting decreases away grounding zones. High melt rates are also re- (12, 25) results from stronger tidal currents and
from grounding lines and becomes negative vealed in the grounding zones of the Amery, mixing, especially in combination with a shallow
(accretion of marine ice) on all large ice shelves Moscow University, Shackleton, and Totten in water column (24), as along the eastern front
and some smaller ice shelves. This general pat- East Antarctica. of Ronne [150 T 50 m in (1) versus 350 T 100 m
tern of melting and freezing beneath ice shelves In contrast, low melt rates are found under for Ross or 500 T 250 m for Filchner]. Ice-
is well understood (4–6, 15) and is governed the largest ice shelves—for example, the Ross front fluxes may overestimate iceberg calving
by the Coriolis-influenced transport and verti- West—except near deep grounding lines. Max- where near-ice-front melting is substantial and
cal mixing of ocean heat, the pressure depen- imum grounding-line depth is only 0.9 km under calving is infrequent; conversely, large icebergs

dence of the freezing point of seawater, and the Ross West but 2.1 km under the Filchner and may on average be thicker than the ice front,
the sea floor and cavity morphology. On some Ronne, 1.8 km under the Ross East, and 2.4 km in which case ice-front fluxes underestimate
large ice shelves, freezing is concentrated on the under the Amery (1). Each additional 100 m adds calving.
Fimbul Vigrid
Melt rate (m/yr) Jelbart
Nivl
Atka Lazarev
<-5 0 >5 Ekström Borchgrevink
Quar Baudouin
Riiser-Larsen Prince Harald
Weddell Sea Shirase
Rayner
Stancomb Thyer
Brunt
Larsen B Ronne
Edward VIII
Larsen C
Filchner Wilma
Robert
Larsen D Downer
Lars. E
Wordie Lars. F Amery
Lars. G
ANTARCT
George VI I
CP
Wilkins
ENINSULA
Publications
Bach
Stange
Bellingshausen
Sea Ferrigno EAST West
Venable
Cosgrove ANTARCTICA
WEST
Abbot
Shackleton
ANTARCTICA
Tracy
Pine Tremenchus
Island Conger
Crosson
Thwaites Vincennes
Dotson
Amundsen Totten
Sea
nd
Getz Withrow
Land
La
Nickerson Swinburne Drygalski

Sulzberger Nansen Moscow
es University
Aviator
Mariner
ilk
Ross Ross W Holmes
West East Dibble
Melt
Ross Sea Lillie Mertz
Calving Rennick Ninnis
100 Gt/yr 10 Gt/yr 1 Gt/yr 0 km 500 km 1000 km Cook East
Fig. 1. Basal melt rates of Antarctic ice shelves. Color coded from black lines. Each circle graph is proportional in area to the mass loss from
<–5 m/year (freezing) to >+5 m/year (melting) and overlaid on a 2009 Mod- each shelf, in Gt/year, partitioned between iceberg calving (hatch fill) and
erate Resolution Imaging Spectroradiometer mosaic of Antarctica. Ice-shelf basal melting (black fill). See Table 1 and table S1 for additional details on
perimeters in 2007 and 2008, excluding ice rises and ice islands, are thin ice-shelf locations, areas, and mass balance components.

REPORTS
Table 1. Meltwater production of Antarctic ice shelves, with ice shelves last row includes nonsurveyed coastal sectors. Ice-shelf names are from United
named in Fig. 1. Areas in square kilometers exclude ice rises and islands. States Geological Survey and (3). Surveyed ice-shelf mass loss of 287 T 89 Gt/year
Grounding-line flux (GL), surface mass balance (SMB), ice-front (proxy for in 2003 to 2008 (∂H/∂t) is 28 T 9% higher than that required to maintain
calving) flux (Ice Front), ice-shelf mass gain (∂H/∂t in water mass equivalent), the ice shelves in steady state for 2003 to 2008. *, Larsen B data (velocity,
and basal meltwater production in Gt/year, with area-average basal melt rate thickness) before the 2002 collapse; thinning rate from the remnant part of
in meters of water per year indicated in parenthesis. Total Antarctica in the the ice shelf only. Additional details in table S1.
Area GL SMB Ice front ∂H/∂t Basal melt

Name
km2 Gt/year Gt/year Gt/year Gt/year Gt/year (m/year)
Larsen G 412 0.9 T 0.2 0.1 T 0 0.7 T 1 0.0 T 0 0.3 T 0.2 (0.71 T 0.6)
Larsen F 828 1.5 T 0.3 0.3 T 0.1 0.6 T 1 −0.7 T 0.5 1.2 T 0.4 (1.4 T 0.5)
Larsen E 1,184 3.6 T 0.7 0.4 T 0.1 1.5 T 1 1.1 T 0.7 1.4 T 1 (1.2 T 0.9)
Larsen D 22,548 18.5 T 4 9.8 T 2 6.3 T 1 20.5 T 14 1.4 T 14 (0.1 T 0.6)
Larsen C 46,465 29.6 T 3 23.8 T 4 31.3 T 3 1.4 T 67 20.7 T 67 (0.4 T 1)
Larsen B* 6,755 13.6 T 3 3.0 T 0.6 8.9 T 1 −4.5 T 13 12.2 T 14 (1.8 T 2)
Wordie 277 13.8 T 1 0.3 T 0 7.6 T 3 −0.1 T 0 6.5 T 3 (23.6 T 10)
Wilkins 12,866 7.8 T 2 8.3 T 2 0.7 T 0.4 −3.4 T 16 18.4 T 17 (1.5 T 1)
Bach 4,579 5.4 T 1 1.8 T 0.3 0.8 T 0.2 −4.0 T 0.3 10.4 T 1 (2.3 T 0.3)

George VI 23,434 68.2 T 5 12.7 T 2 5.7 T 1.2 −13.8 T 16 89.0 T 17 (3.8 T 0.7)
Stange 8,027 21.0 T 3 6.0 T 1 4.6 T 0.8 −5.6 T 5 28.0 T 6 (3.5 T 0.7)
Antarctic Peninsula 127,375 184 T 26 66 T13 69 T 13 −9 T 74 191 T 80 (1.5 T 0.6)
Ronne 338,887 156.1 T 10 59.3 T 11 149.2 T 22 −47.4 T 22 113.5 T 35 (0.3 T 0.1)
Ferrigno 117 11.2 T 1 0.16 T 0 6.6 T 2 −0.3 T 0 5.1 T 2 (43.4 T 17)
Venable 3,194 14.6 T 2 3.5 T 1 6.5 T 1 −7.7 T 1 19.4 T 2 (6.1 T 0.7)
Abbot 29,688 34.0 T 4 25.0 T 5 2.4 T 0.5 4.7 T 18 51.8 T 19 (1.7 T 0.6)
Cosgrove 3,033 5.2 T 1 1.5 T 0.3 1.3 T 1.2 −3.1 T 2 8.5 T 2 (2.8 T 0.7)
Pine Island 6,249 126.4 T 6 4.6 T 0.9 62.3 T 5 −33.2 T 2 101.2 T 8 (16.2 T 1)
Thwaites 5,499 113.5 T 4 4.8 T 0.9 54.5 T 5 −33.7 T 3 97.5 T 7 (17.7 T 1)
Crosson 3,229 27.4 T 2 3.7 T 0.7 11.7 T 2 −19.2 T 1 38.5 T 4 (11.9 T 1)
Dotson 5,803 28.4 T 3 5.7 T 1 5.5 T 0.7 −16.6 T 2 45.2 T 4 (7.8 T 0.6)
Getz 34,018 96.7 T 5 34.2 T 7 53.5 T 2 −67.6 T 12 144.9 T 14 (4.3 T 0.4)
Land 640 14.5 T 1 0.8 T 0.1 12.2 T 1 −0.7 T 0.3 3.8 T 1 (5.9 T 2)
Nickerson 6,495 7.8 T 1 4.6 T 0.9 4.3 T 0.6 3.9 T 1 4.2 T 2 (0.6 T 0.3)
Sulzberger 12,333 15.1 T 2 8.2 T 2 1.0 T 0.2 4.1 T 2 18.2 T 3 (1.5 T 0.3)
Swinburne 900 4.9 T 0.4 0.9 T 0.2 1.5 T 0.3 0.6 T 0.2 3.8 T 0.5 (4.2 T 0.6)
Withrow 632 1.3 T0.2 0.3 T 0.0 1.2 T 0.3 0.1 T 0.1 0.3 T 0.4 (0.5 T 0.6)
Ross West 306,105 73.0 T 4 33.5 T 6 100.4 T 8 7.6 T 17 −1.4 T 20 (0.0 T 0.1)
West Antarctica 756,822 730 T 47 191 T 36 494 T 57 −208 T 36 654 T 89 (0.9 T 0.1)
Ross East 194,704 56.1 T 4 31.0 T 6 45.9 T 4 −7.8 T 11 49.1 T 14 (0.3 T 0.1)
Drygalski 2,338 9.6 T 0.6 0.3 T 0.1 3.0 T 1 −0.8 T 0.4 7.6 T 1 (3.3 T 0.5)
Nansen 1,985 1.3 T 0.5 0.3 T 0.1 0.2 T 0.1 0.4 T 0.1 1.1 T 0.6 (0.6 T 0.3)
Aviator 785 1.1 T 0.2 0.2 T 0. 0.2 T 0.1 −0.3 T 0.1 1.4 T 0.2 (1.7 T 0.3)
Mariner 2,705 2.5 T 0.4 1.1 T 0.2 0.6 T 0.2 0.6 T 0.3 2.4 T 0.6 (0.9 T 0.2)
Lillie 770 3.6 T 0.3 0.2 T 0. 0.5 T 0.1 0.0 T 0. 3.4 T 0.3 (4.4 T 0.4)
Rennick 3,273 4.8 T 1 0.7 T 0.1 0.8 T 0.2 −2.3 T 0.9 7.0 T 1 (2.2 T 0.3)
Cook 3,462 36.0 T 3 1.7 T 0.3 27.6 T 3 5.5 T 1 4.6 T 5 (1.3 T 1)
Ninnis 1,899 27.6 T2 1.3 T 0.2 24.6 T 3 2.0 T 0.9 2.2 T 3 (1.2 T 2)
Mertz 5,522 20.0 T 1 3.6 T 0.7 12.0 T 2 3.6 T 1 7.9 T 3 (1.4 T 0.6)
Dibble 1,482 12.5 T 1 1.5 T 0.3 8.2 T 0.9 −2.3 T 0.7 8.1 T 1 (5.5 T 0.9)
Holmes 1,921 26.0 T 2 2.8 T 0.5 24.7 T 4 −2.5 T 1 6.7 T 4 (3.5 T 2)
Moscow Univ. 5,798 52.3 T 1 4.7 T 0.9 29.6 T 3 −0.1 T 3 27.4 T 4 (4.7 T 0.8)
Totten 6,032 71.0 T 3 6.2 T 1 28.0 T 2 −14.0 T 2 63.2 T 4 (10.5 T 0.7)
Vincennes 935 12.7 T 1 0.5 T 0.1 6.8 T 1 1.3 T 0.6 5.0 T 2 (5.3 T 2)
Conger/Glenzer 1,547 1.7 T 0.4 0.9 T 0.2 1.1 T 0.8 −2.1 T 1 3.6 T 1 (2.3 T 0.9)
Tracy/Tremenchus 2,845 0.6 T0.4 1.0 T 0.2 0.2 T 0.1 −1.7 T 2 3.0 T 2 (1.5 T 0.7)
Shackleton 26,080 55.0 T 4 16.2 T 3 30.3 T 3 −31.7 T 14 72.6 T 15 (2.8 T 0.6)
West 15,666 41.9 T 4 6.9 T 1 32.6 T 7 −11.1 T 7 27.2 T 10 (1.7 T 0.7)
Publications 1,551 5.8 T 0.8 0.4 T 0.1 5.2 T 1 −0.5 T 0.8 1.5 T 2 (1.0 T 1)
Amery 60,654 56.0 T 0.5 8.5 T 2 50.4 T 8 −21.4 T 21 35.5 T 23 (0.6 T 0.4)
Wilma/Robert/Downer 858 10.3 T0.5 0.6 T 0.1 0.8 T 0.4 0.0 T 0 10.0 T 0.6 (11.7 T 0.7)
Edward VIII 411 4.1 T 0.8 0.4 T 0.1 0.3 T 0.1 0.0 T 0 4.2 T 0.8 (10.2 T 2)
continued on next page

REPORTS
Table 1 continued
Area GL SMB Ice front ∂H/∂t Basal melt
Name km2 Gt/year Gt/year Gt/year Gt/year Gt/year (m/year)
Edward VIII 411 4.1 T 0.8 0.4 T 0.1 0.3 T 0.1 0.0 T 0 4.2 T 0.8 (10.2 T 2)
Rayner/Thyer 641 14.2 T 1 0.3 T 0.1 7.8 T 0.6 0.0 T 0 6.7 T 1 (10.5 T 2)
Shirase 821 15.0 T 1 0.4 T 0.1 9.6 T 1 0.0 T 0 5.7 T 1 (7.0 T 2)
Prince Harald 5,392 8.3 T 1 4.1 T 0.8 10.3 T 2 4.0 T 2 −2.0 T 3 (-0.4 T 0.6)
Baudouin 32,952 22.0 T 3 8.4 T 2 6.5 T 1 9.8 T 11 14.1 T 12 (0.4 T 0.4)
Borchgrevink 21,580 19.6 T 3 6.1 T 1 17.5 T 3 0.7 T 4 7.5 T 6 (0.3 T 0.3)
Lazarev 8,519 3.7 T 0.6 2.0 T 0.4 3.1 T 1 −3.6 T 2 6.3 T 2 (0.7 T 0.2)
Nivl 7,285 3.9 T 0.8 1.8 T 0.3 1.3 T 0.4 0.6 T 1 3.9 T 2 (0.5 T 0.2)
Vigrid 2.089 2.7 T 0.4 0.4 T 0.1 2.0 T 0.4 −2.0 T 0.4 3.2 T 0.7 (1.5 T 0.3)
Fimbul 40,843 24.9 T 4 12.7 T 2 18.2 T 2 −4.0 T 7 23.5 T 9 (0.6 T 0.2)
Jelbart 10,844 9.9 T 1 4.9 T 0.9 8.8 T 2 6.9 T 2 −1.0 T 3 (-0.1 T 0.3)
Atka 1,969 0.9 T 0.2 0.8 T 0.1 1.0 T 0.2 1.1 T 0.2 −0.5 T 0.4 (-0.2 T 0.2)
Ekstrom 6,872 4.1 T 0.8 2.6 T 0.5 2.3 T 0.6 0.0 T 0 4.3 T 2 (0.6 T 0.2)
Quar 2,156 1.0 T0.2 0.5 T 0.1 0.6 T 0.1 −0.5 T 0.4 1.4 T 0.5 (0.7 T 0.2)
Riiser-Larsen 43,450 21.5 T 3 12.7 T 2 12.1 T 2 13.4 T 8 8.7 T 9 (0.2 T 0.2)
Brunt/Stancomb 36,894 20.3 T 3 11.4 T 2 28.1 T 4 2.6 T 4 1.0 T 7 (0.03 T 0.2)

Filchner 104,253 97.7 T 6 13.4 T 2 82.8 T 4 −13.6 T 7 41.9 T 10 (0.4 T 0.1)
East Antarctica 669,781 782 T 80 174 T 33 546 T 70 −70 T 34 480 T 116 (0.7 T 0.2)
Total surveyed 1,553,978 1,696 T 146 430 T 81 1,089 T 139 −287 T 89 1,325 T 235 (0.85 T 0.1)
Total Antarctica 1,561,402 2,048 T 149 1,265 T 141 1,500 T 237
The total ice-shelf grounding-line inflow of that of their glacier source regions over the same for a warm-cavity Southeast Pacific ice shelf,
1696 T 146 Gt/year combined with an SMB input time period (26). Similarly, the total imbalance of but the moderate-sized, shallow-draft Abbot
of 430 T 81 Gt/year is partitioned into an ice-front all Antarctic ice shelves combined is more than (29) ranks eighth overall in meltwater produc-
flux of 1,089 T 139 Gt/year and a basal meltwater twice that of the grounded ice (26). tion, while maintaining a positive mass balance
production of 1,325 T 235 Gt/year. Basal melting The ratio of calving to melting averages (B < Bss).
thus accounts for 55 T 10% of ice-shelf mass ab- 0.45 T 0.3, but exhibits considerable regional Meltwater production from several small
lation. The corresponding area-average melt rate variability (Table 1), with area-average melt East Antarctic ice shelves in the Wilkes Land
of 85 T 15 cm/year is three times as large as the rates varying from negative to > 40 m/year. This sector is larger than expected. Area-average
average SMB on ice shelves (28 T 5 cm) and five wide range reflects diverse ocean environments, melt rates from Dibble through Vincennes (4
times the average SMB on grounded ice sheet (16 T which include seawater temperature, the depths to 11 m/year) are comparable to Amundsen Sea
1 cm) (16), illustrating the considerable importance of troughs and sills that influence the access of ice-shelf rates from Crosson through Land (4
of ocean interactions in freshwater transfers be- oceanic heat to ice-shelf cavities, and the sea-ice to 11 m/year), whereas meltwater produced by
tween the ice and ocean. formation and drifts resulting from atmospheric Shackleton and West (73 and 27 Gt/year, respec-
The grounding-line flux of all surveyed ice forcing. tively) rivals that from Thwaites and Sulzberger
shelves accounts for 83 T 7% of the total ice dis- Large ice shelves generate a disproportion- (98 and 18 Gt/year, respectively). Except for
charge into the Southern Ocean (Table 1). Total ally small portion of the total ice-shelf melt- the region from 140° to 150°W where Mertz and
Antarctic grounded ice discharge (26) is 352 T water despite high production rates in their deep Ninnis melting is dominated by shelf waters,
30 Gt/year higher than our grounding-line flux grounding zones and along lengthy ice fronts. oceanographic data are sparse along the Wilkes
because of additional discharge from smaller ice The four giants with areas >100,000 km2 (Ross Land coastline. “Modified” warm deep water at
shelves and ice walls that terminate in the ocean East, Ross West, Filchner, and Ronne) cover a temperature near 0°C has been reported 40 km
(27). An equal partitioning of these missing areas 61% of the total ice-shelf area but contribute south of the continental shelf break northeast
between calving and basal melting (see supplemen- only 15% of the meltwater at an average rate of of Totten (30). By analogy with observations in
tary materials) would increase in situ meltwater 13 cm/year. The low melt rates result from the the Amundsen Sea, our results suggest the pres-
production to 1500 T 237 Gt/year and ice-front relatively weak ocean heat source provided by ence of seawater at similar temperatures under
flux to 1265 T 139 Gt/year. cold shelf waters, in turn leading to substantial several East Antarctic ice shelves. Even 0° sea-
The comparison of basal melting B (Fig. 1) marine ice accretion (28). Despite areas 3 to water at outer continental shelf depths could
with steady state melting Bss (fig. S4, Table 1, and 10 times as large as the Getz, none of the big four expose ice shelves with deep grounding lines
table S1) shows that many ice shelves are near ice shelves produce as much meltwater, with the like the Totten (2.2 km), Moscow (2.0 km), and
equilibrium (B ~ Bss), whereas some are thicken- Ross West contributing no net melt. Meltwater Shackleton (1.8 km) to temperatures more than
ing (B < Bss) and others are thinning (B > Bss). from the Southeast Pacific-Antarctic sector (George 3°C above their melting points. To evaluate the
High basal melting is therefore not synonymous VI through Getz) accounts for 48% of the total impact of these warm deep waters on ice-shelf
with thinning. Ice shelves with high melt rates can meltwater over only 8% of the area, with the melting, more information is needed about their
be in a state of mass balance, but meltwater pro- Getz being the largest meltwater source in Ant- spatial and temporal variability on the outer shelf
duction is 28 T 9% higher than required to main- arctica during the study period. B averages and links through glacially scoured troughs to the
tain the ice shelves in overall steady state (1037 T 5.1 m/year in this region, from a maximum vulnerable glacier grounding lines.
218 Gt/year). Ice shelves in the Amundsen Sea of 43 m/year under the short Ferrigno Glacier Our glaciological estimates are generally con-
sector (Pine Island to Getz) contribute 59% of the tongue to a minimum of 1.8 m/year beneath the sistent with recent results from high-resolution
287 T 89 Gt/year imbalance, an attrition rate twice Abbot. That area-average rate may seem low ocean models in the Amundsen, Bellingshausen,

REPORTS
and Weddell Seas (29, 31–33) (see supplemen- shelf thinning, but if major shifts in sea ice cover 21. E. Rignot, J. Mouginot, B. Scheuchl, Geophys. Res. Lett.
tary materials), but total melting of 10 of the and ocean circulation tip even large ice-shelf 38, L10504 (2011).
22. E. Rignot, J. Mouginot, B. Scheuchl, Science 333,
larger ice shelves is notably less here than in cavities from cold to warm (35), there could be 1427–1430 (2011).
circumpolar models (7, 11). Discrepancies be- major changes in ice shelf and thus ice-sheet 23. H. D. Pritchard et al., Nature 484, 502–505 (2012).
tween model results and observations have mass balance. 24. K. Makinson et al., Geophys. Res. Lett. 38, L06601
been attributed to deficiencies in atmospheric (2011).
References and Notes 25. H. J. Horgan, R. T. Walker, S. Anandakrishnan,
forcing, the representation of sea-ice cover, the R. B. Alley, J. Geophys. Res. 116, C02005 (2011).
1. P. Fretwell et al., The Cryosphere 7, 375–393 (2013).
smoothing of bottom topography, and assump- 2. C. W. Swithinbank, Satellite Image Atlas of Glaciers of the 26. A. Shepherd et al., Science 338, 1183–1189 (2012).
tions regarding cavity shape. Some models yield World: Antarctica, R. S. Williams, J. G. Ferrigno, Eds. 27. S. Neshyba, E. G. Josberger, J. Phys. Oceanogr. 10,
annual cycles and decadal variability (29) that (USGS Prof. Paper 1386-B, 1988). 1681–1685 (1980).
3. N. I. Barkov, Ice Shelves of Antarctica (New Delhi, NY, 28. K. Grosfeld et al., Antarct. Res. Ser. 75, 319–339 (1998).
can now be compared for specific periods with
Amerind Publishing Company, 1985). 29. M. P. Schodlok, D. Menemenlis, E. Rignot, M. Studinger,
glaciological measurements, which need to be 4. R. LeB. Hooke, Principles of Glacier Mechanics Ann. Glaciol. 53, 156–162 (2012).
extended in time. (Cambridge University Press, Cambridge, 2005). 30. G. D. Williams et al., Deep Sea Res. Part II Top. Stud.
Our results indicate that basal melting accounts 5. K. M. Cuffey, W. S. B. Paterson, The Physics of Glaciers Oceanogr. 58, 1194–1210 (2011).
for a larger fraction of Antarctic ice-shelf attrition (Elsevier, Burlington, MA, ed. 4, 2010). 31. P. R. Holland, A. Jenkins, D. Holland, Geophys. Res. Lett.
6. S. S. Jacobs et al., J. Glaciol. 38, 375 (1992). 115, (C5), C05020 (2010).
than previously estimated. These improved glaci- 7. H. H. Hellmer, Geophys. Res. Lett. 31, L10307 (2004). 32. L. Padman et al., J. Geophys. Res. 117, C01010
ological estimates provide not only more accurate 8. A. Jenkins, S. S. Jacobs, J. Geophys. Res. 113, (C4), (2012).
and detailed reference values for modeling but C04013 (2008). 33. P. R. Holland et al., Geophys. Res. Lett. 36, L11604
also a baseline for similar future studies. Ice-shelf 9. S. S. Jacobs, A. Jenkins, C. F. Giulivi, P. Dutrieux, Nature (2009).

Geosci. 4, 519–523 (2011). 34. P. R. Holland, A. Jenkins, D. M. Holland, J. Clim. 21,
meltwater production exhibits a complex spatial 2558–2572 (2008).
10. A. Foldvik, T. Gammelsrod, E. Nygaard, S. Osterhus,
pattern around the continent, with an outsized J. Geophys. Res. Oceans 106, 4463 (2001). 35. H. H. Hellmer et al., Nature 485, 225–228 (2012).
contribution of smaller, fast-melting ice shelves 11. R. Timmermann, Q. Wang, H. H. Hellmer, Ann. Glaciol.
in both West and East Antarctica. Warm-cavity ice 53, 303–314 (2012). Acknowledgments: We thank three anonymous reviewers for
12. I. Joughin, L. Padman, Geophys. Res. Lett. 30, 1477 their constructive criticism of the manuscript. This work was
shelves along the Southeast Pacific coastline, pre- performed at the University of California, Irvine, and at the Jet
(2003).
dicted and observed to be sensitive to ocean warm- 13. J. Wen et al., J. Glaciol. 56, 81–90 (2010). Propulsion Laboratory, California Institute of Technology, under
ing and circulation strength (9, 34), were thinning 14. E. Rignot, S. S. Jacobs, Science 296, 2020–2023 grants from NASA’s Cryospheric Science Program and
rapidly in 2003 to 2008 (23). Nearly half of the East (2002). Operation IceBridge (OIB) and at the Lamont-Doherty Earth
15. A. Jenkins, C. S. M. Doake, J. Geophys. Res. 96, 791 Observatory of Columbia University under grants from the
Antarctic ice shelves were also thinning, some due (1991). National Science Foundation and the National Oceanic and
to probable exposure to “warm” seawater, with 16. J. T. M. Lenaerts et al., J. Geophys. Res. 117, D05108 Atmospheric Administration.
connections to ice drainage basins grounded below (2012).
sea level. 17. E. Rignot et al., Nat. Geosci. 1, 106–110 (2008).
18. C. Allen, IceBridge MCoRDS L2 Ice Thickness. Boulder, Supplementary Materials
Continued observations of ice-shelf velocity Colorado, USA: NASA DAAC at the National Snow and Ice www.sciencemag.org/cgi/content/full/science.1235798/DC1
and thickness change, along with more detailed Data Center (2010). Supplementary Text
Figs. S1 to S4
information on cavity shape, seafloor topography, 19. D. D. Blankenship, S. Kempf, D. Young, IceBridge
HiCARS 2 L2 Geolocated Ice Thickness. Boulder, Colorado, Tables S1 and S2
and atmospheric and oceanic forcing variability References (36–58)
USA: NASA DAAC at the National Snow and Ice Data
are critical to understand the temporal variability Center (2012). 29 January 2013; accepted 31 May 2013
and evolution of Antarctic ice shelves. Continued 20. J. A. Griggs, J. L. Bamber, J. Glaciol. 57, 485–498 Published online 13 June 2013;
warming of the ocean will slowly increase ice- (2011). 10.1126/science.1235798
Lethal Aggression in Mobile an foragers and found that most MFBS (62%)
were nonwarring, whereas all of the complex
Forager Bands and Implications

and equestrian societies made war. On the other
hand, Wrangham and Glowacki [(8), p. 7] de-
veloped a chimpanzee-based lethal raiding mod-
for the Origins of War el, asserting that “humans evolved a tendency
to kill members of other groups,” and they pro-
vided ethnographic quotations on MFBS to il-
Douglas P. Fry1,2* and Patrik Söderberg1,3 lustrate the model. They [(8), p. 8] define war as
when “coalitions of members of a group seek to
It has been argued that warfare evolved as a component of early human behavior within inflict bodily harm on one or more members of
foraging band societies. We investigated lethal aggression in a sample of 21 mobile forager band another group; ‘groups’ are independent politi-
societies (MFBS) derived systematically from the standard cross-cultural sample. We hypothesized, cal units.” Bowles (9) examined war deaths in
on the basis of mobile forager ethnography, that most lethal events would stem from personal eight societies, six of which were MFBS, and re-
disputes rather than coalitionary aggression against other groups (war). More than half of the ported the occurrence of war in all eight societies,
lethal aggression events were perpetrated by lone individuals, and almost two-thirds resulted which he takes as confirmation that war has been
from accidents, interfamilial disputes, within-group executions, or interpersonal motives such as pervasive during human evolution.
competition over a particular woman. Overall, the findings suggest that most incidents of
1
lethal aggression among MFBS may be classified as homicides, a few others as feuds, and a Peace, Mediation and Conflict Research, Åbo Akademi Uni-
minority as war. versity in Vasa, Post Office Box 311, FIN-65101, Vasa, Finland.
2
Bureau of Applied Research in Anthropology, School of An-
thropology, Post Office Box 210030, Tucson, AZ 85721–0030,
controversy exists regarding mobile for- with MFBS generally report that warfare is ab- USA. 3Developmental Psychology, Åbo Akademi University in
A ager band societies (MFBS) and war-

fare. Field researchers who have worked
sent or rudimentarily developed (1–6). Fry (7)
compared MFBSs with complex and equestri-
Vasa, Post Office Box 311, FIN-65101, Vasa, Finland.
*Corresponding author. E-mail: dfry@abo.fi

Lethal Aggression in Mobile Forager Bands and Implications for the
Origins of War
Douglas P. Fry and Patrik Söderberg
Science 341, 270 (2013);
DOI: 10.1126/science.1235675


found at:
REPORTS
and Weddell Seas (29, 31–33) (see supplemen- shelf thinning, but if major shifts in sea ice cover 21. E. Rignot, J. Mouginot, B. Scheuchl, Geophys. Res. Lett.
tary materials), but total melting of 10 of the and ocean circulation tip even large ice-shelf 38, L10504 (2011).
22. E. Rignot, J. Mouginot, B. Scheuchl, Science 333,
larger ice shelves is notably less here than in cavities from cold to warm (35), there could be 1427–1430 (2011).
circumpolar models (7, 11). Discrepancies be- major changes in ice shelf and thus ice-sheet 23. H. D. Pritchard et al., Nature 484, 502–505 (2012).
tween model results and observations have mass balance. 24. K. Makinson et al., Geophys. Res. Lett. 38, L06601
been attributed to deficiencies in atmospheric (2011).
References and Notes 25. H. J. Horgan, R. T. Walker, S. Anandakrishnan,
forcing, the representation of sea-ice cover, the R. B. Alley, J. Geophys. Res. 116, C02005 (2011).
1. P. Fretwell et al., The Cryosphere 7, 375–393 (2013).
smoothing of bottom topography, and assump- 2. C. W. Swithinbank, Satellite Image Atlas of Glaciers of the 26. A. Shepherd et al., Science 338, 1183–1189 (2012).
tions regarding cavity shape. Some models yield World: Antarctica, R. S. Williams, J. G. Ferrigno, Eds. 27. S. Neshyba, E. G. Josberger, J. Phys. Oceanogr. 10,
annual cycles and decadal variability (29) that (USGS Prof. Paper 1386-B, 1988). 1681–1685 (1980).
3. N. I. Barkov, Ice Shelves of Antarctica (New Delhi, NY, 28. K. Grosfeld et al., Antarct. Res. Ser. 75, 319–339 (1998).
can now be compared for specific periods with
Amerind Publishing Company, 1985). 29. M. P. Schodlok, D. Menemenlis, E. Rignot, M. Studinger,
glaciological measurements, which need to be 4. R. LeB. Hooke, Principles of Glacier Mechanics Ann. Glaciol. 53, 156–162 (2012).
extended in time. (Cambridge University Press, Cambridge, 2005). 30. G. D. Williams et al., Deep Sea Res. Part II Top. Stud.
Our results indicate that basal melting accounts 5. K. M. Cuffey, W. S. B. Paterson, The Physics of Glaciers Oceanogr. 58, 1194–1210 (2011).
for a larger fraction of Antarctic ice-shelf attrition (Elsevier, Burlington, MA, ed. 4, 2010). 31. P. R. Holland, A. Jenkins, D. Holland, Geophys. Res. Lett.
6. S. S. Jacobs et al., J. Glaciol. 38, 375 (1992). 115, (C5), C05020 (2010).
than previously estimated. These improved glaci- 7. H. H. Hellmer, Geophys. Res. Lett. 31, L10307 (2004). 32. L. Padman et al., J. Geophys. Res. 117, C01010
ological estimates provide not only more accurate 8. A. Jenkins, S. S. Jacobs, J. Geophys. Res. 113, (C4), (2012).
and detailed reference values for modeling but C04013 (2008). 33. P. R. Holland et al., Geophys. Res. Lett. 36, L11604
also a baseline for similar future studies. Ice-shelf 9. S. S. Jacobs, A. Jenkins, C. F. Giulivi, P. Dutrieux, Nature (2009).

Geosci. 4, 519–523 (2011). 34. P. R. Holland, A. Jenkins, D. M. Holland, J. Clim. 21,
meltwater production exhibits a complex spatial 2558–2572 (2008).
10. A. Foldvik, T. Gammelsrod, E. Nygaard, S. Osterhus,
pattern around the continent, with an outsized J. Geophys. Res. Oceans 106, 4463 (2001). 35. H. H. Hellmer et al., Nature 485, 225–228 (2012).
contribution of smaller, fast-melting ice shelves 11. R. Timmermann, Q. Wang, H. H. Hellmer, Ann. Glaciol.
in both West and East Antarctica. Warm-cavity ice 53, 303–314 (2012). Acknowledgments: We thank three anonymous reviewers for
12. I. Joughin, L. Padman, Geophys. Res. Lett. 30, 1477 their constructive criticism of the manuscript. This work was
shelves along the Southeast Pacific coastline, pre- performed at the University of California, Irvine, and at the Jet
(2003).
dicted and observed to be sensitive to ocean warm- 13. J. Wen et al., J. Glaciol. 56, 81–90 (2010). Propulsion Laboratory, California Institute of Technology, under
ing and circulation strength (9, 34), were thinning 14. E. Rignot, S. S. Jacobs, Science 296, 2020–2023 grants from NASA’s Cryospheric Science Program and
rapidly in 2003 to 2008 (23). Nearly half of the East (2002). Operation IceBridge (OIB) and at the Lamont-Doherty Earth
15. A. Jenkins, C. S. M. Doake, J. Geophys. Res. 96, 791 Observatory of Columbia University under grants from the
Antarctic ice shelves were also thinning, some due (1991). National Science Foundation and the National Oceanic and
to probable exposure to “warm” seawater, with 16. J. T. M. Lenaerts et al., J. Geophys. Res. 117, D05108 Atmospheric Administration.
connections to ice drainage basins grounded below (2012).
sea level. 17. E. Rignot et al., Nat. Geosci. 1, 106–110 (2008).
18. C. Allen, IceBridge MCoRDS L2 Ice Thickness. Boulder, Supplementary Materials
Continued observations of ice-shelf velocity Colorado, USA: NASA DAAC at the National Snow and Ice www.sciencemag.org/cgi/content/full/science.1235798/DC1
and thickness change, along with more detailed Data Center (2010). Supplementary Text
Figs. S1 to S4
information on cavity shape, seafloor topography, 19. D. D. Blankenship, S. Kempf, D. Young, IceBridge
HiCARS 2 L2 Geolocated Ice Thickness. Boulder, Colorado, Tables S1 and S2
and atmospheric and oceanic forcing variability References (36–58)
USA: NASA DAAC at the National Snow and Ice Data
are critical to understand the temporal variability Center (2012). 29 January 2013; accepted 31 May 2013
and evolution of Antarctic ice shelves. Continued 20. J. A. Griggs, J. L. Bamber, J. Glaciol. 57, 485–498 Published online 13 June 2013;
warming of the ocean will slowly increase ice- (2011). 10.1126/science.1235798
Lethal Aggression in Mobile an foragers and found that most MFBS (62%)
were nonwarring, whereas all of the complex
Forager Bands and Implications

and equestrian societies made war. On the other
hand, Wrangham and Glowacki [(8), p. 7] de-
veloped a chimpanzee-based lethal raiding mod-
for the Origins of War el, asserting that “humans evolved a tendency
to kill members of other groups,” and they pro-
vided ethnographic quotations on MFBS to il-
Douglas P. Fry1,2* and Patrik Söderberg1,3 lustrate the model. They [(8), p. 8] define war as
when “coalitions of members of a group seek to
It has been argued that warfare evolved as a component of early human behavior within inflict bodily harm on one or more members of
foraging band societies. We investigated lethal aggression in a sample of 21 mobile forager band another group; ‘groups’ are independent politi-
societies (MFBS) derived systematically from the standard cross-cultural sample. We hypothesized, cal units.” Bowles (9) examined war deaths in
on the basis of mobile forager ethnography, that most lethal events would stem from personal eight societies, six of which were MFBS, and re-
disputes rather than coalitionary aggression against other groups (war). More than half of the ported the occurrence of war in all eight societies,
lethal aggression events were perpetrated by lone individuals, and almost two-thirds resulted which he takes as confirmation that war has been
from accidents, interfamilial disputes, within-group executions, or interpersonal motives such as pervasive during human evolution.
competition over a particular woman. Overall, the findings suggest that most incidents of
1
lethal aggression among MFBS may be classified as homicides, a few others as feuds, and a Peace, Mediation and Conflict Research, Åbo Akademi Uni-
minority as war. versity in Vasa, Post Office Box 311, FIN-65101, Vasa, Finland.
2
Bureau of Applied Research in Anthropology, School of An-
thropology, Post Office Box 210030, Tucson, AZ 85721–0030,
controversy exists regarding mobile for- with MFBS generally report that warfare is ab- USA. 3Developmental Psychology, Åbo Akademi University in
A ager band societies (MFBS) and war-

fare. Field researchers who have worked
sent or rudimentarily developed (1–6). Fry (7)
compared MFBSs with complex and equestri-
Vasa, Post Office Box 311, FIN-65101, Vasa, Finland.
*Corresponding author. E-mail: dfry@abo.fi

REPORTS
There are two likely explanations for such di- a factor that dampens intergroup hostility. (iv) would expect lethal aggression in MFBS to be
vergent interpretations about warfare and MFBS: MFBS tend not to be segmented into subgroups. mostly interpersonal, not intergroup. Additional-
differences in how warfare is defined and dif- In terms of residence, MFBS tend to be multi- ly, in mammals the killing of conspecifics is an
ferences in sampling procedure and composition local, not patrilocal, thus lacking a residential atypical and infrequent form of aggression com-
(7, 8). Recognizing that definitional opinions pattern known to facilitate coalitions among pared to displays, noncontact threats, and restrained
differ, rather than making an a priori determi- male kin useful for war. (v) The social order is aggression, so perhaps also for humans the de-
nation regarding which events are classified as egalitarian and leadership is lacking; no one velopment of an evolutionary model based on
manslaughter, homicide, feud, or war, we instead has the authority to order others to fight. (vi) restraint as a widely documented phenomenon
consider the salient characteristics of each and Foraging areas are large, population densities across species, rather than on rare killing behav-
every actual event involving lethal aggression in low, and resources spread-out, making defense ior, merits consideration (10).
a systematically derived, representative sample of territory difficult or impossible. (vii) Bound- We extracted a subsample of purely MFBS
of MFBS. aries often are controlled socially through use- (n = 21) from the standard cross-cultural sample
There are numerous reasons to predict a requests and permission-granting. (viii) Typical (SCCS). To circumvent sampling bias, rather than
paucity of warfare among MFBS (see supple- spoils of war—material goods or stored food— self-selecting cases, we derived the sample of
mentary materials). (i) In MFBS, group size is are largely lacking, and the necessity of mobility MFBS based on the published rating criteria of
too small to support warfare. (ii) MFBS have makes the capture and containment of individ- others researchers (11, 12). During data collection,
egocentric social networks with descent gen- uals against their will (e.g., slaves or brides) im- we used only the principal authority sources
erally figured bilaterally through both parental practical (and runs counter to the MFBS ethos (PAS) as the earliest, high-quality ethnographic
lines. This does not facilitate coalitional alliance of egalitarianism). (ix) Conflicts within and be- descriptions available (12). We considered every

formation useful in war. (iii) Group composition tween groups are easily handled by separation instance of lethal aggression reported for these
fluctuates over time, resulting in kinship and so- and other conflict-resolution mechanisms. On 21 MFBS (13).
cial networks that cut across different groups, the basis of these foregoing characteristics, we The 21 MFBS produced a total of 148 lethal
aggression events. The median number was 4
(mean = 7.05; SD = 14.64), with a range from
0 to 69. One society, the Tiwi of Australia, had
an exceptionally large number of lethal events
!Kung (4) (n = 69). If the Tiwi case is removed, the me-
dian number of lethal events for the remaining
Hadza (5)
20 societies drops to 3.5, the mean is almost
Mbuti (0) cut in half (mean = 3.95; SD = 3.69), and the
Semang (0) range is reduced to 0 to 15.
Of 135 lethal events with unambiguous per-
Andamanese (2) petrator and victim information, 55% consisted
Vedda (4) of one killer and one victim. In 23% of the le-
thal events, more than one person participated
Tiwi (69)
in killing a single individual, and in 22% of
Aranda (5) the events, more than one person participated
Gilyak (2) in killing more than one person (Fig. 1). In only
one lethal event (0.7%), did a single killer dis-
Yukaghir (1) patch more than one person (table S4, case
Ingalik (6) 18), and the two victims were children. Tiwi so-
ciety reflects a different pattern wherein 44%
Copper Inuit (15) of the lethal events involved one killer and one
Montagnais (4) Lethal events with victim, whereas the corresponding figure for the
more than one other 20 societies combined was 64% (supple-
Micmac (0) perpetrator but only mentary text).
Northern Salteaux (2) one victim Thirty-six percent of the lethal events took
Slave (1) place within the local band; for example, be-
Lethal events with tween brothers, father and son, mother and child,
Kaska (3) more than one in-laws, husbands and wives, companions, friends,
Paiutes (8) perpetrator and more clan “brothers,” neighbors, and so on (table S2).
than one victim Six percent of all incidents involved husbands
Botocudos (3)
killing wives. In most lethal events (85%), the
Aweikoma (10) killers and victims were members of the same
Yahgan (4) society. The remaining lethal aggression events
involved persons from outside the society, such
0 5 10 15 20 25 as shipwreck victims, colonists, missionaries, or
Observations neighboring indigenous cultures. Almost all of
the killers were male, whether they acted alone
Fig. 1. Lethal aggression events with multiple perpetrators in 21 MBFS. The distribution of all or with others. Females were the killers or co-
lethal events that involved multiple perpetrators is shown, based on whether there was a single victim perpetrators in only 4% of the events.
(31 events) or multiple victims (29 events), out of a grand total of 148 lethal events. The figure reflects The reasons for the lethal events varied.
how the Tiwi are more violent than the other MFBS. Nearly half of the sample societies (10 of 21) had no Given that most lethal aggression involved one
lethal events perpetrated by two or more persons. Three societies had no lethal events at all. For each killer and one victim, the large number of personal
society, the total number of lethal events is in parentheses. motives for killing is not surprising (Table 1 and

REPORTS
table S3). Specifically, many lethal disputes in- one-person reflects homicide or manslaughter, theory on the evolutionary logic of fighting,
volved two men competing over a particular wom- not coalitional killings or war. Additionally, 36% and the observation that killing is an exceptional
an (sometimes the wife of one of them), revenge of all lethal events occurred within the same event in human societies leads to the counter-
homicide exacted by family members of a victim local group (62% if the atypical Tiwi are re- hypothesis that lethal behavior has been strongly
(often aimed at the specific person responsible for moved), and violence within a local group is not selected against, not favored, in comparison to
the previous killing), and interpersonal quarrels coalitional war. Only 15% of the lethal events more restrained conflict behavior (7, 10, 15).
of various kinds; for instance, stealing of honey, occurred across societal lines. Some such events Taken together, the current findings contra-
insults or taunting, incest, self-defense or the pro- might fall within a definition of war, whereas dict recent assertions that MFBS regularly en-
tection of a loved-one, and so on. Additionally, in others might not (such as when shipwreck sur- gage in coalitionary war against other groups
some killing events, another person or persons vivors were killed). Finally, very few lethal dis- (8), that “chronic raiding and feuding charac-
supported a companion who acted out of per- putes were over resources. Overall, a consideration terize life in a state of nature” [(14), p. xxiv], or
sonal, not group, motives, such as when a friend of reasons for lethal aggression reveals that most that MFBS war deaths are substantial in recent
assisted a husband in killing his wife’s lover (see cases stemmed from personal motives consistent millennia and in the Pleistocene (9) (supplemen-
table S4, case 109). with homicide and, in some cases, family feuds, tary text). Perhaps discrepancies between the fore-
About one third of the lethal events involved but much less often with lethal aggression be- going propositions and the current findings can
disputes between people of different groups (Table tween political communities, or warfare (supple- be accounted for by proposing that self-selection
1). However, three-quarters (38 of 50) of in- mentary text). of ethnographic material may have exaggerated
tergroup disputes took place among the Tiwi Approximately half of the societies had no war (10). Additionally, methodological factors
alone. The percentage of intergroup disputes for lethal events that involved more than one per- such as relying heavily on only a few secondary

the Tiwi exceeded 50% of their events, whereas petrator. This observation is incongruent with sources (8), or estimating war mortality on the
the corresponding percentage for the other twenty assertions by Bowles (9) and Pinker (14) that basis of genocidal massacres and murders of
societies was only about 15%. Another feature war is prevalent in MFBS or by Wrangham indigenous peoples by armed ranchers and
of the Tiwi data is the regular occurrence of and Glowacki (8) that humans have an evolved settlers (9), can lead to misimpressions in com-
strings of killings. Thirty-nine percent (27 of 69) tendency to form coalitions to kill members of parison to the use of systematic sampling pro-
of Tiwi lethal events occurred in seven separate neighboring groups. Additionally, two or more cedures, reliance on primary ethnographic data,
strings (table S4, cases 129 to 131, 133 to 139, persons killing a third person might or might not and a focus on the specific circumstances of le-
140 to 141 and 145, 142 to 144, 146 to 149, 150 occur in the context of a coalition against an- thal aggression cases in MFBS (7, 10).
to 153, and 156 to 158), whereas only two strings other group or war. In some instances, motives In conclusion, when all cases are examined
of killings occurred in the other 20 societies such as sexual jealousy (table S4, e.g., cases 29 for a systematically drawn sample of MFBS,
(table S4, cases 10 to 13 and 82 to 89). and 109) or avenging the murder of a family most incidents of lethal aggression can aptly be
The findings suggest that MFBS are not member (table S4, e.g., case 157) are clearly per- called homicides, a few others feud, and only a
particularly warlike if the actual circumstances sonal rather than stemming from hypothesized minority warfare. The findings do not lend sup-
of lethal aggression are examined. Fifty-five per- general hostility toward other groups. Most port to the coalitionary model. The predictions
cent of the lethal events involved a sole perpe- mammalian aggression also is between individ- are substantiated that MFBS, as a social type,
trator killing only one individual (64% if the uals. A different evolutionary perspective sup- possess many features that make warfare unlike-
atypical Tiwi are removed). One-person-killing- ported by comparative mammalian data, game ly. The actual reasons for lethal aggression are
most often interpersonal, and consequently, the
particulars of most of the lethal events in these
Table 1. Reasons for lethal aggression, from the personal to the communal. The atypical societies do not conform to the usual concep-
Tiwi findings are shown separately, followed by the other societies minus the Tiwi (n = 20), and tualization of war.
the entire sample (n = 21), all in number of cases (with percentages in parentheses). A more
detailed version of the table with case numbers for lethal aggression events is presented in
table S3. 1. E. Leacock, Curr. Anthropol. 19, 247–275 (1978).
2. E. R. Service, The Hunters (Prentice-Hall, Englewood
Reason Tiwi only All others Total sample Cliffs, NJ, 1966).
Interpersonal events 24 (34.8%) 50 (63.3%) 74 (50.0%) 3. J. Steward, in Man the Hunter, R. Lee, I. DeVore, Eds.
(Aldine, Chicago, 1968), pp. 321–334.
Revenge against a particular killer(s) 9 (13.0%) 8 (10.1%) 17 (11.5%) 4. R. Lee, R. Daly, in The Cambridge Encyclopedia of
Over a particular woman 8 (11.6%) 6 (7.6%) 14 (9.5%) Hunters and Gatherers, R. B. Lee, R. Daly, Eds.
Over a particular man 0 (0.0%) 1 (1.3%) 1 (0.7%) (Cambridge Univ. Press, Cambridge, 1999), 1–19.
Husband kills wife 2 (2.9%) 7 (8.9%) 9 (6.1%) 5. R. Tonkinson, in Keeping the Peace, G. Kemp, D. P. Fry,
Eds. (Routledge, New York, 2004), pp. 89–104.
Wife kills husband 0 (0.0%) 0 (0.0%) 0 (0.0%) 6. R. Tonkinson, in War, Peace, and Human Nature:
Miscellaneous interpersonal disputes* 5 (7.2%) 28 (35.4%) 33 (22.3%) Convergence of Evolutionary and Cultural Views,
Interfamilial feud 0 (0.0%) 6 (7.6%) 6 (4.1%) D. P. Fry, Ed. (Oxford Univ. Press, New York, 2013),
Within-group execution 0 (0.0%) 3 (3.8%) 3 (2.0%) pp. 262–277.
7. D. P. Fry, The Human Potential for Peace (Oxford Univ.
Execution of outsiders 4 (5.8%) 3 (3.8%) 7 (4.7%)
Press, New York, 2006).
Intergroup events 38 (55.1%) 12 (15.2%) 50 (33.8%) 8. R. W. Wrangham, L. Glowacki, Hum. Nat. 23, 5–29
Over borders/resources (e.g., fruit tree) 0 (0.0%) 2 (2.5%) 2 (1.4%) (2012).
Theft of women from another group 0 (0.0%) 2 (2.5%) 2 (1.4%) 9. S. Bowles, Science 324, 1293–1298 (2009).
Interclan revenge-seeking 17 (24.6%) 0 (0.0%) 17 (11.5%) 10. D. P. Fry, in War, Peace, and Human Nature: Convergence
of Evolutionary and Cultural Views, D. P. Fry, Ed. (Oxford
During a general fight 4 (5.8%) 0 (0.0%) 4 (2.7%) Univ. Press, New York, 2013), pp. 1–21.
Miscellaneous intergroup disputes* 17 (24.6%) 8 (10.1%) 25 (16.9) 11. G. Murdock, Ethnology 6, 109–236 (1967).
Accident 3 (4.3%) 3 (3.8%) 6 (4.1%) 12. D. White, Behav. Sci. Res. 23, 1–145 (1989).
Starvation cannibalism 0 (0.0%) 2 (2.5%) 2 (1.4%) 13. Materials and Methods are available as supplementary
Grand total 69 (100%) 79 (100%) 148 (100%) 14. S. Pinker, The Better Angels of Our Nature (Viking,
*For a listing of miscellaneous events, see table S3. New York, 2011).

REPORTS
15. J. Maynard Smith, G. Price, Nature 246, 15–18 with the costs of publication. The data reported in this study Supplementary Text
(1973). are located in the supplementary materials. Tables S1 to S4
Acknowledgments: Some of the data reported here were Supplementary Materials
collected during research funded by the NSF (grant 03-13670). www.sciencemag.org/cgi/content/full/341/6143/270/DC1 25 January 2013; accepted 10 June 2013
We are grateful to the Svenska Kulturfonden for assisting Material and Methods 10.1126/science.1235675
Interactions of Multisensory cues are also associated with the sexual display.
Male frogs have inflatable vocal sacs that shut-
Components Perceptually Rescue

tle air to and from the lungs while calling. Similar
to the movement of lips during human speech
(2), they are a biomechanical consequence of
Túngara Frog Mating Signals the sound production system (7), but, as with
lips and speech, they can also influence the per-
ception of the call (8, 9). We have shown previ-
R. C. Taylor1 and M. J. Ryan2,3* ously that female túngara frogs prefer a multimodal
signal (a call associated with a robotic frog) to a
Sexual signals are often complex and perceived by multiple senses. How animals integrate signal call by itself (10), a result reconfirmed here

components across sensory modalities can influence signal evolution. Here we show that two (Fig. 1D).
relatively unattractive signals that are perceived acoustically and visually can be combined in In túngara frogs, the temporal relationship
a pattern to form a signal that is attractive to female túngara frogs. Such unanticipated between acoustic components influences the
perceptual effects suggest that the evolution of complex signals can occur by alteration of the signal’s attractiveness (11). When the chuck in a
relationships among already-existing traits. whine-chuck call is displaced by 500 ms, the call
becomes merely as attractive as a whine only
uman perception of stimuli in multiple results (Fig. 1, A, B, and D). Females were tested (Fig. 1B) and less attractive than a normal whine-
H sensory modalities can positively influ-

ence signal detection, selective attention,
learning, and memory (1). One example is “hear-
only once.
The acoustic component of a frog’s mating
call is its most distinguishing feature, but visual
chuck (Fig. 1C). The temporal relationship be-
tween the acoustic and visual components of the
signal also influences the signal’s attractiveness
ing lips and seeing voices” in the McGurk effect
(2), which provided the foundation for speech
auditory-visual research (3). Studies of multi-
modal communication in animals have often
asked whether individual signal components in
different sensory modalities are redundant or car-
ry different information (4), but few studies have
investigated how specific interactions influence
signal perception (5).
Female túngara frogs base their mate choices
on male mating calls. Specifically, males produce
calls consisting of a whine alone or they may add
up to seven chucks; they do not produce only
chucks (6). Females exhibit phonotaxis (move-
ment toward a call, a bioassay of call recognition
and preference) to a whine only, but exhibit a
fivefold preference for calls with a whine-chuck
over a whine only [N = 3662 (11); see also Fig.
1A]. We tested female mate preferences in a
series of two-choice tests. Synthetic male vocal-
izations were broadcast from two speakers, one
of which was paired with a robotic frog that pro-
vided the visual stimulus of a calling male. Fe-
males were released equidistant from the two Fig. 1. Preference responses. Each portion of the figure illustrates the acoustic components of the
speakers (with a 60° separation relative to the túngara frog mating call: a whine only [(A, B, and J), right gray], a chuck only [(J), left black], or a whine-
female release point) and allowed to choose a chuck (all other calls). The natural whine-chuck is depicted in (A), left black; (C), right gray; (D to G), all
stimulus. Because our experimental configura- acoustic signals; and (I), right gray. The rectangle represents the inflation-deflation cycle of the robofrog’s
vocal sac and its temporal relationship to the call [(D) to (J), left black]. The x axis represents 1000 ms,
tion differed from those of previous experiments,
green indicates the significantly preferred stimulus, and red indicates the unpreferred stimulus. In each of
we replicated some studies and obtained similar
the 10 experiments [(A) to (J)], 20 females were given a choice between the signal in black versus the
signal in gray. The vertical black and gray bars represent the number of females that chose the respective
1
Department of Biology, Salisbury University, Salisbury, MD
signal, and the blue dashed horizontal lines represent the null hypothesis of equal preference. Experiments
21801, USA. 2Section of Integrative Biology, University of Texas, highlighted in the solid blue box are tests of the perceptual rescue versus template-matching hypotheses, and
Austin, TX 78712, USA. 3Smithsonian Tropical Research Insti- those in the dashed blue box are the test of the component substitution hypothesis. The results of binomial tests
tute, Post Office Box 0843-03092, Balboa, Ancón, Republic of are noted as *** = P < 0.001, ** = P < 0.01, * = P < 0.05, ns (not significant) = P > 0.05. The exact P values for
Panama. each experiment are as follows: (A) P = 0.0003, (B) P = 0.744, (C) P = 0.019, (D) P = 0.034, (E) P = 0.323,
*Corresponding author. E-mail: mryan@utexas.edu (F) P = 0.0049, (G) P = 0.019, (H) P = 0.039, (I) P = 0.583, (J) P = 0.0001.

Interactions of Multisensory Components Perceptually Rescue
Túngara Frog Mating Signals
R. C. Taylor and M. J. Ryan
Science 341, 273 (2013);
DOI: 10.1126/science.1237113


Ecology
http://www.sciencemag.org/cgi/collection/ecology
REPORTS
15. J. Maynard Smith, G. Price, Nature 246, 15–18 with the costs of publication. The data reported in this study Supplementary Text
(1973). are located in the supplementary materials. Tables S1 to S4
Acknowledgments: Some of the data reported here were Supplementary Materials
collected during research funded by the NSF (grant 03-13670). www.sciencemag.org/cgi/content/full/341/6143/270/DC1 25 January 2013; accepted 10 June 2013
We are grateful to the Svenska Kulturfonden for assisting Material and Methods 10.1126/science.1235675
Interactions of Multisensory cues are also associated with the sexual display.
Male frogs have inflatable vocal sacs that shut-
Components Perceptually Rescue

tle air to and from the lungs while calling. Similar
to the movement of lips during human speech
(2), they are a biomechanical consequence of
Túngara Frog Mating Signals the sound production system (7), but, as with
lips and speech, they can also influence the per-
ception of the call (8, 9). We have shown previ-
R. C. Taylor1 and M. J. Ryan2,3* ously that female túngara frogs prefer a multimodal
signal (a call associated with a robotic frog) to a
Sexual signals are often complex and perceived by multiple senses. How animals integrate signal call by itself (10), a result reconfirmed here

components across sensory modalities can influence signal evolution. Here we show that two (Fig. 1D).
relatively unattractive signals that are perceived acoustically and visually can be combined in In túngara frogs, the temporal relationship
a pattern to form a signal that is attractive to female túngara frogs. Such unanticipated between acoustic components influences the
perceptual effects suggest that the evolution of complex signals can occur by alteration of the signal’s attractiveness (11). When the chuck in a
relationships among already-existing traits. whine-chuck call is displaced by 500 ms, the call
becomes merely as attractive as a whine only
uman perception of stimuli in multiple results (Fig. 1, A, B, and D). Females were tested (Fig. 1B) and less attractive than a normal whine-
H sensory modalities can positively influ-

ence signal detection, selective attention,
learning, and memory (1). One example is “hear-
only once.
The acoustic component of a frog’s mating
call is its most distinguishing feature, but visual
chuck (Fig. 1C). The temporal relationship be-
tween the acoustic and visual components of the
signal also influences the signal’s attractiveness
ing lips and seeing voices” in the McGurk effect
(2), which provided the foundation for speech
auditory-visual research (3). Studies of multi-
modal communication in animals have often
asked whether individual signal components in
different sensory modalities are redundant or car-
ry different information (4), but few studies have
investigated how specific interactions influence
signal perception (5).
Female túngara frogs base their mate choices
on male mating calls. Specifically, males produce
calls consisting of a whine alone or they may add
up to seven chucks; they do not produce only
chucks (6). Females exhibit phonotaxis (move-
ment toward a call, a bioassay of call recognition
and preference) to a whine only, but exhibit a
fivefold preference for calls with a whine-chuck
over a whine only [N = 3662 (11); see also Fig.
1A]. We tested female mate preferences in a
series of two-choice tests. Synthetic male vocal-
izations were broadcast from two speakers, one
of which was paired with a robotic frog that pro-
vided the visual stimulus of a calling male. Fe-
males were released equidistant from the two Fig. 1. Preference responses. Each portion of the figure illustrates the acoustic components of the
speakers (with a 60° separation relative to the túngara frog mating call: a whine only [(A, B, and J), right gray], a chuck only [(J), left black], or a whine-
female release point) and allowed to choose a chuck (all other calls). The natural whine-chuck is depicted in (A), left black; (C), right gray; (D to G), all
stimulus. Because our experimental configura- acoustic signals; and (I), right gray. The rectangle represents the inflation-deflation cycle of the robofrog’s
vocal sac and its temporal relationship to the call [(D) to (J), left black]. The x axis represents 1000 ms,
tion differed from those of previous experiments,
green indicates the significantly preferred stimulus, and red indicates the unpreferred stimulus. In each of
we replicated some studies and obtained similar
the 10 experiments [(A) to (J)], 20 females were given a choice between the signal in black versus the
signal in gray. The vertical black and gray bars represent the number of females that chose the respective
1
Department of Biology, Salisbury University, Salisbury, MD
signal, and the blue dashed horizontal lines represent the null hypothesis of equal preference. Experiments
21801, USA. 2Section of Integrative Biology, University of Texas, highlighted in the solid blue box are tests of the perceptual rescue versus template-matching hypotheses, and
Austin, TX 78712, USA. 3Smithsonian Tropical Research Insti- those in the dashed blue box are the test of the component substitution hypothesis. The results of binomial tests
tute, Post Office Box 0843-03092, Balboa, Ancón, Republic of are noted as *** = P < 0.001, ** = P < 0.01, * = P < 0.05, ns (not significant) = P > 0.05. The exact P values for
Panama. each experiment are as follows: (A) P = 0.0003, (B) P = 0.744, (C) P = 0.019, (D) P = 0.034, (E) P = 0.323,
*Corresponding author. E-mail: mryan@utexas.edu (F) P = 0.0049, (G) P = 0.019, (H) P = 0.039, (I) P = 0.583, (J) P = 0.0001.

REPORTS
[fig. 1, E to G, from (12)]. When the acoustic and data support the hypothesis that temporally dis- lution with only a few key changes. Different
visual cues are offset by 100 ms, the visual cue no junct stimuli can be linked into a common percept components of a complex phenotype need not
longer adds to the attractiveness of the acoustic of a mating call by their strategic placement in time, arise simultaneously and de novo but could result
cue (Fig. 1E). When the visual cue is displaced even when the resulting stimulus complex has from temporal or spatial shifts in previously in-
further in time, 200 ms after the call’s onset (Fig. never been experienced by females in nature. coherent traits. Put another way, perceptual in-
1F) or immediately after the call’s offset (Fig. One explanation for our results is that the tegration in a multisensory universe may yield
1G), the manipulated multimodal signals are both displaced vocal sac causes the whine and dis- emergent psychological percepts that provide the
significantly less attractive than the call alone (12). placed chuck to be perceptually bound. An al- basis for positive selection of complex signals.
Thus, displacement of the visual cue can reverse the ternative, the component substitution hypothesis, This type of unanticipated perceptual bias could
valence of the multimodal signal (Fig. 1, F and G). posits that the vocal sac inflation substitutes for be responsible for the evolution of some of the
Two hypotheses may explain why the acous- the whine, creating the context for perceiving the extreme and elaborate signals that evolve under
tic (chuck) and visual (vocal sac) cues lose sa- chuck as part of the mating signal (i.e., “whine- sexual selection by mate choice (23, 24).
lience when temporally displaced from the whine. chuck”) and making the whine itself irrelevant.
The template-matching hypothesis predicts that This is not the case. Females significantly pre- References and Notes
females have an internal neural template of the ferred a whine to a vocal sac inflation followed 1. L. E. Bahrick, R. Lickliter, R. Flom, Curr. Dir. Psychol. Sci.
13, 99 (2004).
species’ call that facilitates recognition (13); dis- by a chuck (sac-chuck) (P < 0.0002, Fig. 1J). 2. H. McGurk, J. MacDonald, Nature 264, 746 (1976).
rupting components of mating signals will dis- These results further support our interpretation of 3. J. Driver, Nature 381, 66 (1996).
rupt their recognition by females. An alternative perceptual binding. 4. S. R. Partan, P. Marler, Am. Nat. 166, 231 (2005).
hypothesis, which we term perceptual rescue, pos- What is the mechanistic basis of perceptual 5. E. Hebets, D. Papaj, Behav. Ecol. Sociobiol. 57, 197

(2005).
its that stimulus saliency is influenced by the rescue? Humans can group streams of speech 6. M. Griddi-Papp, A. S. Rand, M. J. Ryan, Nature 441, 38
relative and not the absolute relationships of into perceptual units, and when sound is inter- (2006).
signal components to one another. If so, a tem- rupted, a continuity illusion can be generated by 7. G. B. Pauly, X. E. Bernal, A. S. Rand, M. J. Ryan,
porally disrupted and less attractive stimulus introducing broadband noise into silent gaps (17). Physiol. Biochem. Zool. 79, 708 (2006).
8. P. M. Narins, W. Hödl, D. S. Grabul, Proc. Natl. Acad.
could be rescued by strategic association with Although one study in frogs failed to find a
Sci. U.S.A. 100, 577 (2003).
another less attractive stimulus, causing females continuity illusion in the auditory channel (18), 9. R. C. Taylor, B. Buchanan, J. Doherty, Anim. Behav. 74,
to bind these components into the percept of a our results are consistent with a continuity-type 1753 (2007).
more attractive mating signal. Perceptual rescue illusion that combines sensory modalities. Thus, 10. R. C. Taylor, B. A. Klein, J. Stein, M. J. Ryan, Anim. Behav.
76, 1089 (2008).
predicts more flexibility in signal recognition than one possible mechanism for perceptual rescue is
11. W. Wilczynski, S. A. Rand, M. J. Ryan, Anim. Behav. 58,
template matching. that the presence of the visual cue generates a 841 (1999).
We tested the perceptual rescue hypothesis multisensory continuity illusion. 12. R. C. Taylor, B. A. Klein, J. Stein, M. J. Ryan, J. Exp. Biol.
against the mutually exclusive template-matching Although túngara frogs do not produce dis- 214, 815 (2011).
hypothesis [in the sense of “strong inference” sociated acoustic and visual signal components, 13. P. Marler, J. Neurobiol. 33, 501 (1997).
14. J. R. Platt, Science 146, 347 (1964).
(14)]. Specifically, we asked whether placing a these manipulations are not as ecologically irrel- 15. A. Agresti, Stat. Med. 20, 2709 (2001).
visual cue between two separated acoustic cues evant as they might seem. Females are challenged 16. A. Agresti, An Introduction to Categorical Data Analysis
would cause the components to be bound into by an auditory world similar to the cocktail party (Wiley-Blackwell, New York, 2007).
one coherent signal. To do this, we placed the problem in humans (19). At their breeding cho- 17. A. Bregman, Auditory Scene Analysis: The Perceptual
Organization of Sound (MIT Press, Cambridge, MA,
vocal sac inflation from the relatively unattractive ruses, they need to perceptually bind the whine 1994).
whine-chuck-sac (Fig. 1G, left black) into the and chuck, and assignment of the two acoustic 18. F. Seeba, J. J. Schwartz, M. A. Bee, Anim. Behav. 79,
gap between the whine and the chuck (Fig. 1B, components to their sources is not always accu- 1317 (2010).
left black). This generated a whine followed by rate (20). The cross-modal interactions we reveal 19. M. A. Bee, C. Micheyl, J. Comp. Psychol. 122, 235
(2008).
the vocal sac inflation-deflation, which in turn was here suggest that the problem might be even more
20. H. E. Farris, M. J. Ryan, Nat. Commun. 2, 410
followed by a chuck 500 ms after the whine’s challenging when auditory and visual scene analy- (2011).
offset (Fig. 1H, left black). This multimodal sig- ses are combined. 21. C. T. Rowe, T. Guilford, Nature 383, 520 (1996).
nal was competed against the unimodal whine-gap- These dynamic and context-dependent inter- 22. T. Guilford, M. Dawkins, Anim. Behav. 42, 1 (1991).
chuck, the same acoustic stimulus but lacking a actions among multimodal signal components 23. M. J. Ryan, Integr. Comp. Biol. 51, 756 (2011).
24. M. J. Ryan, M. E. Cummings, Annu. Rev. Ecol. Evol. Syst.,
visual stimulus between the whine and the chuck. could form a basis for signal recognition, but it in press.
Fourteen of 20 females preferred the multi- would be radically different from the standard
modal signal to the unimodal signal [mid-value P template-matching model (13). Our study sug- Acknowledgments: We are grateful to the NSF
reported throughout (15, 16), P = 0.039, Fig. 1H]. gests a need to reconsider the neural basis by (grants IBN 0517328, IBN 0078150, and IOS 1120031)
and the Clark Hubbs Regents Professorship (M.J.R.) for funding
Thus, adding the visual cue to the gap between which animals recognize signals, to account for and the Smithsonian Tropical Research Institute for logistical
the whine and the chuck, which is equivalent to these cognitive and perceptual biases that lead to support. We thank E. Balaban for helpful discussion and
adding the chuck to the end of the temporally the emergence of hidden preferences. Emergent H. Farris, S. Partan, S. Pika, G. Rosenthal, and three
displaced visual cue, rescued the perceptual ef- properties arising from interactions among sen- anonymous reviewers for their comments on the manuscript.
We thank B. Klein and Moey, Incorporated for design and
fects of these stimuli; it made this stimulus com- sory modalities are not restricted to the recog- fabrication of the robotic frogs. We are especially grateful to
plex more attractive. Further, the addition of the nition of communication signals. In chicks, odor the interns who assisted in data collection. All research
visual cue restored the signal’s attractiveness to and color can interact to generate an aversive re- reported here complied with IACUC protocols from Salisbury
that of a normal whine-chuck, also predicted by sponse that does not occur with either component University, the University of Texas, and the Smithsonian
Tropical Research Institute. We obtained all required permits
the perceptual rescue hypothesis (9 responses to in isolation (21). In this domain as well, there is a
from the Government of Panama. Data will be archived in the
the multimodal signal, 11 to the unimodal, P = psychological response that is hidden when only Dryad Data Repository at dx.doi.org/10.5061/dryad.pk85.
0.41, Fig. 1I). isolated stimuli are encountered.
These results reject the hypothesis that the Our findings also have implications for un- www.sciencemag.org/cgi/content/full/science.1237113/DC1
negative influence of temporally displaced com- derstanding receiver psychology (22) and specif- Materials and Methods
ponents on the signal’s attractiveness (Fig. 1, B ically how perceptual processes can drive the 26 February 2013; accepted 8 May 2013
and C, and E to G) was due to these stimuli not evolution of complex signal phenotypes. The in- Published online 6 June 2013;
matching a conspecific call template. Rather, these teractions we report could facilitate signal evo- 10.1126/science.1237113

Loss of Function of the Melanocortin 2 Receptor Accessory Protein 2
Is Associated with Mammalian Obesity
Masato Asai et al.
Science 341, 275 (2013);
DOI: 10.1126/science.1233000


found at:
REPORTS
Loss of Function of the Melanocortin 2 to ACTH (3, 4). Loss of either MC2R or MRAP
in humans causes severe resistance to ACTH, with
resulting glucocorticoid deficiency (5, 6).
Receptor Accessory Protein 2 Is All mammals have a paralogous gene, MRAP2,
which, like MC3R and MC4R, is predominantly
Associated with Mammalian Obesity expressed in the brain (7), most prominently in
the pons and cerebellum but also in regions in-
volved in energy homeostasis, such as the hypo-
Masato Asai,1,2 Shwetha Ramachandrappa,3 Maria Joachim,1 Yuan Shen,1 Rong Zhang,1 thalamus and brainstem (fig. S1, A to C). Within
Nikhil Nuthalapati,1 Visali Ramanathan,1 David E. Strochlic,1 Peter Ferket,4 Kirsten Linhart,1* the paraventricular nucleus of the hypothalamus
Caroline Ho,1 Tatiana V. Novoselova,5 Sumedha Garg,3 Martin Ridderstråle,6 Claude Marcus,7 (PVN), Mrap2 and Mc4r mRNAs are coexpressed
Joel N. Hirschhorn,1,8 Julia M. Keogh,3 Stephen O’Rahilly,3 Li F. Chan,5 Adrian J. Clark,5 in many cells (fig. S1D). We hypothesized that
I. Sadaf Farooqi,3† Joseph A. Majzoub1† Mrap2 might modulate signaling through a MCR
and potentially affect energy homeostasis. We there-
Melanocortin receptor accessory proteins (MRAPs) modulate signaling of melanocortin receptors fore performed targeted deletion of Mrap2 in mice
in vitro. To investigate the physiological role of brain-expressed melanocortin 2 receptor accessory using Cre-lox–mediated excision of the 100-bp
protein 2 (MRAP2), we characterized mice with whole-body and brain-specific targeted deletion exon 3 [which encodes the highly conserved trans-
of Mrap2, both of which develop severe obesity at a young age. Mrap2 interacts directly with membrane domain (7)] to create mice with nor-
melanocortin 4 receptor (Mc4r), a protein previously implicated in mammalian obesity, and it mal levels of an mRNA predicted to encode a

enhances Mc4r-mediated generation of the second messenger cyclic adenosine monophosphate, truncated protein that includes the first 55 amino
suggesting that alterations in Mc4r signaling may be one mechanism underlying the association acids of Mrap2, with the transmembrane domain
between Mrap2 disruption and obesity. In a study of humans with severe, early-onset obesity, replaced by 11 aberrant amino acids specified by
we found four rare, potentially pathogenic genetic variants in MRAP2, suggesting that the gene the out-of-frame exon 4, followed by a stop co-
may also contribute to body weight regulation in humans. don (fig. S1, E to H). Normal levels of the mutant
mRNA indicate preservation of Mrap2-containing
embrane-expressed G protein–coupled protein for MC2R, MC2R accessory protein (MRAP), neurons in null mice, although these neurons
M receptors (GPCRs) modulate cellular

responses to numerous physiological
stimuli. The melanocortin receptors (MCRs) are
is required for the trafficking of MC2R to the sur-
face of adrenal cells and for signaling in response
probably do not express the predicted mutant
protein because mutant Mrap2 mRNA, but not
a subfamily of GPCRs that mediate signaling

in response to the pro-opiomelanocortin–derived
peptides, adrenocorticotropic hormone (ACTH),
and a-melanocyte–stimulating hormone (aMSH)
and their competitive antagonists, agouti and agouti-
related protein. The MCRs mediate a diverse
range of physiological functions: MC1R is in-
volved in skin pigmentation, MC2R plays a crit-
ical role in the hypothalamic-pituitary-adrenal
axis, MC3R and MC4R are involved in energy
homeostasis, and MC5R is implicated in exo-
crine function (1).
There is increasing recognition that accessory
proteins can modulate GPCR trafficking, as well
as ligand binding and signaling (2). An accessory
1
Division of Endocrinology, Department of Medicine, Boston
Children’s Hospital, Harvard Medical School, 300 Longwood
Avenue, Boston, MA 02115, USA. 2Departments of Pathology,
Endocrinology and Diabetes, Nagoya University Graduate School
of Medicine, 65 Tsurumai-cho, Showa-ku, Nagoya 466-8550,
Japan. 3University of Cambridge Metabolic Research Labora-
tories and National Institute for Health Research (NIHR) Cam-
bridge Biomedical Research Centre, Institute of Metabolic Science,
Addenbrooke’s Hospital, Cambridge CB2 0QQ, UK. 4Prestage
Department of Poultry Science, North Carolina State University,
Raleigh, NC 27695, USA. 5William Harvey Research Institute,
Centre for Endocrinology Queen Mary, University of London
Barts and The London School of Medicine and Dentistry, Fig. 1. Phenotype of Mrap2−/− mice. (A) Weight curves for Mrap+/+ versus Mrap+/− versus Mrap2−/− mice on
London EC1M 6BQ, UK. 6Department of Clinical Sciences, Lund standard-chow (Chow, top: male n = 9 versus 28 versus 15 mice, female n = 12 versus 18 versus 10 mice) or
University, Malmö, Sweden, and Steno Diabetes Center, DK- high-fat diets (HFD; ages 56 to 95 days, bottom: superimposed on standard-chow curves: male n = 10 versus
2820 Gentofte, Denmark. 7Department for Clinical Science, 8 versus 10 mice; female n = 7 versus 12 versus 7 mice). For both genders, the weight curves of Mrap+/+ and
Intervention and Technology, Karolinska Institute, Division of Mrap+/− mice on standard chow differ significantly at older ages (161 to 175 days) and at younger ages
Pediatrics, National Childhood Obesity Centre, S-141 86
(56 to 95 days) on a high-fat diet. *P = 0.02, **P = 0.001, ***P = 0.0003. (B) Fat depots on standard-
Stockholm, Sweden. 8Department of Genetics, Harvard Medical
School and Broad Institute, Cambridge, MA 02142, USA. chow diet. (Top) White adipose tissue (WAT) weights in Mrap+/+ versus Mrap2−/− (males and females, ages
117 to 122 days, n = 5 versus 4 mice, respectively). (Bottom left) Brown adipose tissue (BAT) weight in
*Present address: Department of Internal Medicine, KH Salem,
University of Heidelberg, 69120 Heidelberg, Germany. Mrap+/+ versus Mrap2−/− mice (males and females, ages 117 to 122 days, n = 5 versus 4 mice). (Bottom
†Corresponding author. E-mail: joseph.majzoub@childrens. right) WAT cell size in in Mrap+/+ versus Mrap2−/− mice (females, 50 cells counted from each mouse). *P =
harvard.edu (J.A.M.); isf20@cam.ac.uk (I.S.F.) 0.009, **P = 0.003, ***P = 0.0003, ****P < 0.00001.

REPORTS

Fig. 2. Energy balance in Mrap2−/− mice. (A) Cumulative food intake (top) 3 mice), ages 30 to 34 days. (Bottom) Body weight versus energy ex-
and weight (bottom) in ad libitum–fed Mrap+/+ versus Mrap2−/− males (n = penditure, integrated over 24 hours, males (n = 18 versus 14 mice, ages 30
10 versus 11 mice) and females (n = 11 versus 8 mice). (B) Energy ex- to 45 days), females (n = 16 versus 11 mice, ages 30 to 42 days). Analysis
penditure in ad libitum–fed Mrap+/+ versus Mrap2−/− mice. (Top) Continuous with ANCOVA showed no differences between genotypes (males, P = 0.38;
measurement over 3 days, males (n = 3 versus 4 mice), females (n = 4 versus females, P = 0.67).
Fig. 3. Interaction between Mrap2 and Mc4r. (A) Conditional deletion of Mrap2 in
Sim1 neurons. (Top right) Cre DNA analysis by means of polymerase chain reaction
(PCR). HT DNA from Sim1CreBAC::Mrap2 f/f mice contains Cre (374 bp), but from Mrap2 f/f
mice does not. Molecular weight marker (M) is shown on right (base pairs). (Top left)
Mrap2 DNA analysis in Sim1CreBAC::Mrap2 f/f and Mrap2 f/f mice by means of PCR. Both
genotypes contain floxed, intact Mrap2 DNA in CX, HT, and BS (314 bp in top
electropherogram, and 1013 bp in bottom electropherogram, and molecular weight
markers on left). Only Sim1CreBAC::Mrap2 f/f mice contain Mrap2Del (400 bp, bottom
electropherogram), and only in HT and BS, but not in CX, which is consistent with
fluorescent reporter data (fig. S3A). No PCR products are present without added DNA
(H2O). (Bottom) Mrap2 mRNA expression in Sim1CreBAC::Mrap2 f/f and Mrap2 f/f mice
by means of reverse transcriptase (RT)–PCR. Both genotypes express floxed, intact
Mrap2 mRNA in CX, HT, and BS (247 bp). Only Sim1CreBAC::Mrap2 f/f mice express
Mrap2Del mRNA (147 bp), and only in HT. Global Mrap2Del/Del mice express Mrap2Del
mRNA in all three sites. (B) Body weights of Mrap2+/+ (male n = 6 mice, female n =
11 mice), Mrap2−/− (male n = 11 mice, female n = 7 mice), Mrap2 f/f (male n =
8 mice, female n = 12 mice), and conditional Sim1CreBAC::Mrap2 f/f (male n = 8 mice,
female n = 7 mice) mice, all age 133 days. *P = 0.04, **P = 0.007, ***P = 0.0002,
****P < 0.0001. (C) Effect of Mrap2 on Mc4r signaling. (Left) Level of cAMP reporter
activity (CRE Luc) in CHO cells alone or cotransfected with Mc4r, with or without Mrap2 or the Mrap2 knockout construct, Mrap2delE3, 5 hours after exposure to
0 to 10 nM aMSH (n = 3 mice per group). (Right) cAMP activity of these same constructs, expressed as percent induction after 0 to 10 nM aMSH, relative to 0 nM
aMSH. *P < .0001, Mc4r+Mrap2 versus Mc4r at same [aMSH], by means of analysis of variance. For most data points, error bars are obscured by symbols.

REPORTS
protein, is present in cells transfected with the take under a variety of conditions. At 42 (fig. S2J) body temperature, in young (30 to 45 days of
same Mrap2 mutant construct used to create the and 84 (fig. S2K) days of age, when Mrap2−/− age) wild-type and Mrap2-null mice, just as their
null mice (fig. S1I). mice were clearly overweight, no difference in weights began to diverge (Fig. 2A). Surprisingly,
Mrap2-null mice appeared normal at birth, food intake was detected between the two geno- the wild-type and mutant mice had indistinguish-
with normal weight gain and post-weaning food types when analyzed over a 4-day interval. Obe- able 24-hour total energy expenditure, as analyzed
intake during early life (0 to 32 days and 23 to sity was not caused by more efficient absorption by means of analysis of covariance (ANCOVA)
32 days, respectively), although young Mrap2−/− of calories in null mice (fig. S2L). Only when (Fig. 2B) (9). There were also no differences be-
male mice trended toward greater weight and food monitored daily over 50 days (ages 34 to 84 days) tween Mrap2+/+ and Mrap2−/− mice in RER (fig.
intake with advancing age (fig. S1J). However, was a subtle increase in cumulative food intake S2R), locomotor activity (fig. S2S), or core body
null mice of both genders gradually became ex- discernable in the null animals (Fig. 2A), with the temperature at 22°C (fig. S2T), with both geno-
tremely obese on a diet of regular chow ad lib- onset of obesity preceding hyperphagia (Fig. 2A types exhibiting the expected increase in all three
itum (figs. 1A and S2A). Heterozygous mice were and fig. S2M). To further understand the contri- parameters during the active night period. After
significantly heavier than were wild-type animals bution of hyperphagia to obesity in Mrap2−/− exposure to 4°C for 18 hours, null and wild-type
on standard chow (160 to 175 days; males, Mrap2+/+ mice, we limited their food intake to that amount mice became significantly hypothermic to the same
26.0 T 0.4 g, Mrap2+/− 29.9 T 0.9 g; females, consumed by their normal siblings (pair feeding). extent (fig. S2T).
Mrap2+/+ 24.5 T 0.9 g, Mrap2+/− 28.1 T 0.7 g), and Even when fed the same amount of chow, null Because (i) MRAP is essential for signaling
at younger ages (56 to 95 days) on a high-fat diet mice gained more weight than did wild-type mice through MC2R (3, 4), (ii) MRAP’s paralog, Mrap2,
(Fig. 1A). In addition, Mrap2−/− mice had increased (fig. S2, N and O). Only when the amount of food is expressed principally in the brain, and (iii)
length (fig. S1K) and percent of weight due to fat intake in null mice was further restricted to 10% Mc2r’s paralog, Mc4r, has a key role in energy

and decreased percent of weight due to lean mass (females) and 13% (males) less than that of wild- balance in Sim1-containing neurons (10), we asked
(fig. S1L). Both genders of Mrap2−/− mice had type mice was there equivalent weight gain (fig. whether deletion of Mrap2 causes obesity in part
increased visceral adiposity, greater than twice S2P) in the two genotypes. To determine whether by altering signaling through centrally expressed
the normal white adipose tissue cell size, enlarged the late-onset hyperphagia in Mrap2−/− mice (Fig. Mc4r. We created a Sim1Cre::Mrap2 flox/flox mouse
brown adipose tissue depots, normal liver histol- 2A) could simply be the consequence of an in- with conditional deletion of Mrap2 exclusively in
ogy on a regular chow diet, but much greater hepatic creased body mass at this older age caused by a these neurons and expression of Mrap2Del mRNA
steatosis as compared with those of wild-type separate metabolic defect, we switched null mice only in hypothalamus and not cerebral cortex or
mice on a high-fat diet (Fig. 1B and fig. S2, A to ad libitum access to chow after 40 days of re- brainstem (Fig. 3A and fig. S3A). Like global
and B). Adult Mrap2-null mice had, as expected, stricted feeding (fig. S2P, upward arrow). During null mice, conditional mutants were similarly obese
elevated leptin concentrations corresponding to the first 24 hours of ad libitum feeding, food in- (Fig. 3B), and pair-feeding to a normal dietary
their increased fat mass, which normalized with take almost doubled in null mice (from 2.9 T 0.1 intake only partially reversed their obesity (fig. S3B).
diet-induced weight normalization (fig. S2C). to 5.6 T 0.5 g/day in males, and from 2.8 T 0.1 to If Mrap2 facilitates the action of Mc4r, then
Obese adult mice had normal fasting insulin (fig. 5.3 T 0.2 g/day in females), with a corresponding Mc4r deficiency should create an equivalent or
S2D) and normal tolerance to intraperitoneal glu- marked increase in body weight. Thus, hyper- more severe obesity phenotype than does Mrap2
cose injection (fig. S2E). Mrap2 has been postu- phagia develops in an age-dependent manner in deficiency, depending on the degree to which Mrap2
lated to play a role in the adrenal response to older mice, independent of body weight. Consist- interferes with Mc4r function. Supporting this,
ACTH (8). We therefore measured diurnal rhyth- ent with this, young (age 38 to 45 days) Mrap2−/− Mrap2+/− mice of both genders were less obese
micity and stress responsiveness of the adrenal mice had an intact anorectic response to the MCR than either Mc4r+/− or doubly heterozygous mice
axis in Mrap2-null mice, which were normal (fig. (Mc4r and Mc3r) agonist, MTII (fig. S2Q), cor- (fig. S3C). The differences between Mc4r+/− and
S2F). Thyroid hormone levels were also normal responding to their normal ad libitum food intake doubly heterozygous mice were not statistically
(table S1). Epinephrine and norepinephrine ex- at this age. significant, although the latter trended toward
cretion were reduced in male Mrap2−/− mice only We hypothesized that young Mrap2−/− mice being heavier. Among homozygous knockouts,
(fig. S2G), but Ucp1 mRNA concentrations in- might display abnormal energy expenditure be- those with Mc4r deficiency alone were more
creased appropriately in both genders of null mice cause obesity develops early during ad libitum obese than those with Mrap2 deficiency alone
after exposure to 4°C for 18 hours (fig. S2H). Hy- feeding before the onset of hyperphagia, persists (fig. S3C). The mice in which Mc4r was knocked
pothalamic Agrp mRNA concentration was re- in mutant mice pair-fed to a normal dietary in- out were more obese than were mice with dele-
duced in Mrap2-null mice, whereas Pomc mRNA take, and is abolished only by underfeeding. To tion of both Mc4r and Mrap2 (in males, with a
was normal (fig. S2I). explore this, we measured energy expenditure and trend in females), suggesting that Mrap2 may
To characterize the mechanisms underlying respiratory exchange ratio (RER) with indirect promote weight gain through both Mc4r-dependent
the obesity in these mice, we measured food in- calorimetry, as well as locomotor activity and core and -independent actions.
To determine whether mouse Mrap2 and Mc4r
can interact directly, we coimmunoprecipitated
Table 1. MRAP2 variants detected in obese subjects and controls. transiently expressed, N-terminally Myc-tagged
Mrap2 and N-terminally green fluorescent protein–
Subject sex/ MAF†: MAF†: tagged Mc4r in Chinese hamster ovary (CHO)
MRAP2 Subjects Controls ***PolyPhen-2
age/BMI/ European African cells (devoid of endogenous Mrap, Mrap2, and
variant with variant with variant prediction‡
BMI SDS* American American MCRs). We found that mouse Mrap2 and Mc4r
E24X 1/488 M/19/63/4.7 0/488 0.000% (0/8600) 0.000% (0/4406) Damaging interact (fig. S3D), which is consistent with pre-
N88Y 1/376 M/11/29.6/3.3 0/376 0.000% (0/8600) 0.000% (0/4406) Possibly vious data (7). We next investigated the impact of
damaging Mrap2 on Mc4r (Fig. 3C) and Mc3r (fig. S3E)
L115V 1/488 M/5/24/4.2 0/488 0.012% (1/8600) 0.000% (0/4406) Benign signaling. The combined expression of Mc4r and
R125C 1/488 F/8/29/3.5 0/488 0.047% (4/8600) 0.045% (2/4406) Possibly Mrap2 in CHO cells suppressed basal cyclic aden-
damaging osine monophosphate (cAMP)–dependent protein
kinase (PKA) signaling compared with Mc4r
*Subject sex (male, M; female, F)/age (years)/body mass index (BMI) (kilograms per square meter)/standard deviation score
(SDS). †MAF, minor allele frequency; available at the National Heart, Lung, and Blood Institute exome variant server: alone (Fig. 3C, left), as previously reported with
http://evs.gs.washington.edu/EVS. ‡PolyPhen-2; available at http://genetics.bwh.harvard.edu/pph2. the human orthologs (7). But in contrast to that

REPORTS
report (which used NDP-MSH), we found that that rare heterozygous variants in MRAP2 are as- 12. L. E. Johansson et al., PLoS ONE 4, e5327 (2009).
aMSH caused a fivefold increase above basal sociated with early-onset, severe obesity in hu- 13. J. A. Sebag et al., Science 341, 278 (2013).
PKA activity (Fig. 3C, right) compared with less mans. The mechanism (or mechanisms) by which Acknowledgments: We thank T. Nguyen for DNA analysis;
than a twofold increase with Mc4r alone or Mc4r Mrap2 exerts its effects on body weight regulation H. Feldman and A. Fleisch for statistical advice; M. Mulcahey
plus the Mrap2-null construct, Mrap2delE3 (our remain to be firmly established but likely involve for thyroid assays; H. Turkova for catecholamine assays;
in vitro model for in vivo disruption of Mrap2). altered signaling through Mc4r and perhaps other S. Cabi for creating the software program used to analyze
calorimetry data; M. Geibel for bioinformatics analyses; and
The presence of Mrap2 increased signaling through MCRs. Under conditions comparable with those D. Margulies, B. Lowell, J. Flier, and Boston Children’s Hospital
Mc3r at the two highest aMSH doses (fig. S3E). we describe, in which Mrap2 greatly enhances Endocrinology Division scientists for helpful discussions. We
These findings suggest Mrap2 may alter signal- cAMP signaling through Mc4r, Sebag et al. (13) are indebted to the patients and their families for their
ing through Mc4r and perhaps other receptors. have found that the zebrafish ortholog of Mrap2 participation and to the physicians involved in GOOS and
the Swedish obese children’s cohort study. This work was
To investigate whether alterations in MRAP2 (zMRAP2b) similarly affects zMC4R signaling. supported by grants from the National Institutes of Health,
are associated with human obesity, we sequenced This evolutionary conservation, plus the extreme including NIHP30-HD18655 ( J.A.M.), the Timothy Murphy
the coding region and intron/exon boundaries of disease phenotype caused by loss of Mrap2 func- Fund ( J.A.M.), the National Alliance for Research on
MRAP2 in obese and control individuals from the tion, supports the importance of Mrap2 in verte- Schizophrenia and Depression (M.A.), the Wellcome Trust
(I.S.F. and S.O’R.), the Medical Research Council [I.S.F.
Genetics of Obesity Study (GOOS) cohort (11) brate biology.
and L.F.C. (grant no. G0802796)], the NIHR Cambridge
and the Swedish obese children’s cohort (12). Biomedical Research Centre (I.S.F. and S.O’R.), and
Four rare heterozygous variants that were absent References and Notes R01DK075787 ( J.N.H.). S.O’R. is a paid Scientific Adviser
from cohort-specific controls and 1000 genomes 1. R. D. Cone, Endocr. Rev. 27, 736–749 (2006). for Pfizer in the area of cardiometabolic disease. Until
(Table 1) were found in unrelated, nonsyndromic, 2. D. L. Hay, D. R. Poyner, P. M. Sexton, Pharmacol. Ther. 2010, J.A.M. was on the Board of, and was a paid Scientific

109, 173–197 (2006). Advisor for, Correlagen Diagnostics, a company whose
severely obese individuals, with all but one var- projects included molecular diagnostic tests related to
3. P. M. Hinkle, J. A. Sebag, Mol. Cell. Endocrinol. 300,
iant in the C-terminal region of the protein (fig. 25–31 (2009). obesity. The authors (M.A., J.A.M., Boston Children’s Hospital)
S4). In three of these subjects, no pathogenic 4. S. N. Cooray, A. J. Clark, Mol. Cell. Endocrinol. 331, have filed a patent application related to modulating
variants were found in the coding region or 215–221 (2011). Mrap2 to alter growth.
intron/exon boundaries of all known nonsyndromic 5. L. A. Metherell et al., Nat. Genet. 37, 166–170 (2005).
6. L. F. Chan, L. A. Metherell, A. J. Clark, Eur. J. Pharmacol. Supplementary Materials
human obesity genes (table S2). Only one of the 660, 171–180 (2011). www.sciencemag.org/cgi/content/full/341/6143/275/DC1
variants (E24X) is clearly disruptive, and overall, 7. L. F. Chan et al., Proc. Natl. Acad. Sci. U.S.A. 106, Materials and Methods
few rare variants were found in the obese cohorts, 6146–6151 (2009). Figs. S1 to S4
indicating that if MRAP2 mutations contribute to 8. J. A. Sebag, P. M. Hinkle, Sci. Signal. 3, ra28 (2010). Tables S1 to S3
9. M. H. Tschöp et al., Nat. Methods 9, 57–63 (2012). References (14–19)
severe human obesity, they do so rarely. 10. N. Balthasar et al., Cell 123, 493–505 (2005).
We have found that global or brain-specific 11. S. Farooqi, S. O’Rahilly, Endocr. Rev. 27, 710–718 20 November 2012; accepted 13 June 2013
inactivation of Mrap2 causes obesity in mice and (2006). 10.1126/science.1233000
Developmental Control of the decrease in growth, a decrease in growth hor-

mone gene expression, and a compensatory in-
Melanocortin-4 Receptor by MRAP2

crease in growth hormone–releasing hormone
(ghrh) gene expression (6), thus providing quan-
titative assays for MC4R activity in vivo. The
Proteins in Zebrafish melanocortin receptors have been shown to inter-
act with the melanocortin receptor accessory pro-
teins MRAP1 and MRAP2 (7–13), which are
Julien A. Sebag,1* Chao Zhang,1* Patricia M. Hinkle,2 Amanda M. Bradshaw,1 Roger D. Cone1† single-transmembrane proteins that form unusual
antiparallel homo- and heterodimers (7–9). Whereas
The melanocortin-4 receptor (MC4R) is essential for control of energy homeostasis in MRAP1 is essential for adrenocorticotropic hor-
vertebrates. MC4R interacts with melanocortin receptor accessory protein 2 (MRAP2) in vitro, mone receptor (MC2R) trafficking to the plasma
but its functions in vivo are unknown. We found that MRAP2a, a larval form, stimulates growth membrane, ligand binding, and downstream sig-
of zebrafish by specifically blocking the action of MC4R. In cell culture, this protein binds naling (7, 8, 11), the functions of MRAP2 remain
MC4R and reduces the ability of the receptor to bind its ligand, a–melanocyte-stimulating unclear. In the zebrafish, MRAP2 exists in two
hormone (a-MSH). A paralog, MRAP2b, expressed later in development, also binds MC4R but isoforms, a and b (14). Here, we investigated
increases ligand sensitivity. Thus, MRAP2 proteins allow for developmental control of MC4R the role of MRAP2a and MRAP2b in the reg-
activity, with MRAP2a blocking its function and stimulating growth during larval development, ulation of MC4R activity in vivo in the zebrafish
whereas MRAP2b enhances responsiveness to a-MSH once the zebrafish begins feeding, thus and in vitro in human embryonic kidney (HEK)
increasing the capacity for regulated feeding and growth. 293T cells.
We first characterized the distribution and
he melanocortin-4 receptor (MC4R), a G and somatic growth (1, 2, 5). Mutations in the developmental expression kinetics of mc4r, mrap2a,
T protein–coupled receptor (GPCR), plays a

central role in energy homeostasis (1–4)
gene encoding MC4R are the most common
monogenic cause of severe early-onset obesity in
humans (1). In the zebrafish, as in mammals,
and mrap2b gene expression in the zebrafish em-
bryo at 1, 2, 3, or 4 days post-fertilization (dpf) by
reverse transcription polymerase chain reaction
MC4R is prominently involved in the regulation (RT-PCR) (Fig. 1A). mc4r and mrap2a mRNA
1
Department of Molecular Physiology and Biophysics, Vanderbilt of energy homeostasis and somatic growth (6). were detectable from 1 dpf and their expression
University School of Medicine, Nashville, TN 37232, USA. 2De- Dominant negative mutations in MC4R are a increased every day until 4 dpf, whereas mrap2b
partment of Pharmacology and Physiology, School of Medicine
and Dentistry, University of Rochester Medical Center, Rochester, natural cause of increased growth rate and final was hardly detectable. To identify the larval tis-
NY 14642, USA. size in some teleost species (5). An artificially sue distribution of mrap2 mRNAs, we performed
*These authors contributed equally to this work. induced increase in MC4R activity early in the whole-mount in situ hybridization on zebrafish
†Corresponding author. E-mail: roger.cone@vanderbilt.edu development of the zebrafish embryo causes a embryos at 5 dpf. mrap2a was ubiquitously ex-

Developmental Control of the Melanocortin-4 Receptor by MRAP2
Proteins in Zebrafish
Julien A. Sebag et al.
Science 341, 278 (2013);
DOI: 10.1126/science.1232995


found at:
REPORTS
report (which used NDP-MSH), we found that that rare heterozygous variants in MRAP2 are as- 12. L. E. Johansson et al., PLoS ONE 4, e5327 (2009).
aMSH caused a fivefold increase above basal sociated with early-onset, severe obesity in hu- 13. J. A. Sebag et al., Science 341, 278 (2013).
PKA activity (Fig. 3C, right) compared with less mans. The mechanism (or mechanisms) by which Acknowledgments: We thank T. Nguyen for DNA analysis;
than a twofold increase with Mc4r alone or Mc4r Mrap2 exerts its effects on body weight regulation H. Feldman and A. Fleisch for statistical advice; M. Mulcahey
plus the Mrap2-null construct, Mrap2delE3 (our remain to be firmly established but likely involve for thyroid assays; H. Turkova for catecholamine assays;
in vitro model for in vivo disruption of Mrap2). altered signaling through Mc4r and perhaps other S. Cabi for creating the software program used to analyze
calorimetry data; M. Geibel for bioinformatics analyses; and
The presence of Mrap2 increased signaling through MCRs. Under conditions comparable with those D. Margulies, B. Lowell, J. Flier, and Boston Children’s Hospital
Mc3r at the two highest aMSH doses (fig. S3E). we describe, in which Mrap2 greatly enhances Endocrinology Division scientists for helpful discussions. We
These findings suggest Mrap2 may alter signal- cAMP signaling through Mc4r, Sebag et al. (13) are indebted to the patients and their families for their
ing through Mc4r and perhaps other receptors. have found that the zebrafish ortholog of Mrap2 participation and to the physicians involved in GOOS and
the Swedish obese children’s cohort study. This work was
To investigate whether alterations in MRAP2 (zMRAP2b) similarly affects zMC4R signaling. supported by grants from the National Institutes of Health,
are associated with human obesity, we sequenced This evolutionary conservation, plus the extreme including NIHP30-HD18655 ( J.A.M.), the Timothy Murphy
the coding region and intron/exon boundaries of disease phenotype caused by loss of Mrap2 func- Fund ( J.A.M.), the National Alliance for Research on
MRAP2 in obese and control individuals from the tion, supports the importance of Mrap2 in verte- Schizophrenia and Depression (M.A.), the Wellcome Trust
(I.S.F. and S.O’R.), the Medical Research Council [I.S.F.
Genetics of Obesity Study (GOOS) cohort (11) brate biology.
and L.F.C. (grant no. G0802796)], the NIHR Cambridge
and the Swedish obese children’s cohort (12). Biomedical Research Centre (I.S.F. and S.O’R.), and
Four rare heterozygous variants that were absent References and Notes R01DK075787 ( J.N.H.). S.O’R. is a paid Scientific Adviser
from cohort-specific controls and 1000 genomes 1. R. D. Cone, Endocr. Rev. 27, 736–749 (2006). for Pfizer in the area of cardiometabolic disease. Until
(Table 1) were found in unrelated, nonsyndromic, 2. D. L. Hay, D. R. Poyner, P. M. Sexton, Pharmacol. Ther. 2010, J.A.M. was on the Board of, and was a paid Scientific

109, 173–197 (2006). Advisor for, Correlagen Diagnostics, a company whose
severely obese individuals, with all but one var- projects included molecular diagnostic tests related to
3. P. M. Hinkle, J. A. Sebag, Mol. Cell. Endocrinol. 300,
iant in the C-terminal region of the protein (fig. 25–31 (2009). obesity. The authors (M.A., J.A.M., Boston Children’s Hospital)
S4). In three of these subjects, no pathogenic 4. S. N. Cooray, A. J. Clark, Mol. Cell. Endocrinol. 331, have filed a patent application related to modulating
variants were found in the coding region or 215–221 (2011). Mrap2 to alter growth.
intron/exon boundaries of all known nonsyndromic 5. L. A. Metherell et al., Nat. Genet. 37, 166–170 (2005).
6. L. F. Chan, L. A. Metherell, A. J. Clark, Eur. J. Pharmacol. Supplementary Materials
human obesity genes (table S2). Only one of the 660, 171–180 (2011). www.sciencemag.org/cgi/content/full/341/6143/275/DC1
variants (E24X) is clearly disruptive, and overall, 7. L. F. Chan et al., Proc. Natl. Acad. Sci. U.S.A. 106, Materials and Methods
few rare variants were found in the obese cohorts, 6146–6151 (2009). Figs. S1 to S4
indicating that if MRAP2 mutations contribute to 8. J. A. Sebag, P. M. Hinkle, Sci. Signal. 3, ra28 (2010). Tables S1 to S3
9. M. H. Tschöp et al., Nat. Methods 9, 57–63 (2012). References (14–19)
severe human obesity, they do so rarely. 10. N. Balthasar et al., Cell 123, 493–505 (2005).
We have found that global or brain-specific 11. S. Farooqi, S. O’Rahilly, Endocr. Rev. 27, 710–718 20 November 2012; accepted 13 June 2013
inactivation of Mrap2 causes obesity in mice and (2006). 10.1126/science.1233000
Developmental Control of the decrease in growth, a decrease in growth hor-

mone gene expression, and a compensatory in-
Melanocortin-4 Receptor by MRAP2

crease in growth hormone–releasing hormone
(ghrh) gene expression (6), thus providing quan-
titative assays for MC4R activity in vivo. The
Proteins in Zebrafish melanocortin receptors have been shown to inter-
act with the melanocortin receptor accessory pro-
teins MRAP1 and MRAP2 (7–13), which are
Julien A. Sebag,1* Chao Zhang,1* Patricia M. Hinkle,2 Amanda M. Bradshaw,1 Roger D. Cone1† single-transmembrane proteins that form unusual
antiparallel homo- and heterodimers (7–9). Whereas
The melanocortin-4 receptor (MC4R) is essential for control of energy homeostasis in MRAP1 is essential for adrenocorticotropic hor-
vertebrates. MC4R interacts with melanocortin receptor accessory protein 2 (MRAP2) in vitro, mone receptor (MC2R) trafficking to the plasma
but its functions in vivo are unknown. We found that MRAP2a, a larval form, stimulates growth membrane, ligand binding, and downstream sig-
of zebrafish by specifically blocking the action of MC4R. In cell culture, this protein binds naling (7, 8, 11), the functions of MRAP2 remain
MC4R and reduces the ability of the receptor to bind its ligand, a–melanocyte-stimulating unclear. In the zebrafish, MRAP2 exists in two
hormone (a-MSH). A paralog, MRAP2b, expressed later in development, also binds MC4R but isoforms, a and b (14). Here, we investigated
increases ligand sensitivity. Thus, MRAP2 proteins allow for developmental control of MC4R the role of MRAP2a and MRAP2b in the reg-
activity, with MRAP2a blocking its function and stimulating growth during larval development, ulation of MC4R activity in vivo in the zebrafish
whereas MRAP2b enhances responsiveness to a-MSH once the zebrafish begins feeding, thus and in vitro in human embryonic kidney (HEK)
increasing the capacity for regulated feeding and growth. 293T cells.
We first characterized the distribution and
he melanocortin-4 receptor (MC4R), a G and somatic growth (1, 2, 5). Mutations in the developmental expression kinetics of mc4r, mrap2a,
T protein–coupled receptor (GPCR), plays a

central role in energy homeostasis (1–4)
gene encoding MC4R are the most common
monogenic cause of severe early-onset obesity in
humans (1). In the zebrafish, as in mammals,
and mrap2b gene expression in the zebrafish em-
bryo at 1, 2, 3, or 4 days post-fertilization (dpf) by
reverse transcription polymerase chain reaction
MC4R is prominently involved in the regulation (RT-PCR) (Fig. 1A). mc4r and mrap2a mRNA
1
Department of Molecular Physiology and Biophysics, Vanderbilt of energy homeostasis and somatic growth (6). were detectable from 1 dpf and their expression
University School of Medicine, Nashville, TN 37232, USA. 2De- Dominant negative mutations in MC4R are a increased every day until 4 dpf, whereas mrap2b
partment of Pharmacology and Physiology, School of Medicine
and Dentistry, University of Rochester Medical Center, Rochester, natural cause of increased growth rate and final was hardly detectable. To identify the larval tis-
NY 14642, USA. size in some teleost species (5). An artificially sue distribution of mrap2 mRNAs, we performed
*These authors contributed equally to this work. induced increase in MC4R activity early in the whole-mount in situ hybridization on zebrafish
†Corresponding author. E-mail: roger.cone@vanderbilt.edu development of the zebrafish embryo causes a embryos at 5 dpf. mrap2a was ubiquitously ex-

REPORTS
pressed, whereas no mrap2b-specific staining was contrast, when the same experiment was con- expected because mrap2b is not expressed in
detected (fig. S1). ducted in mc4r-null fish, only a small decrease the embryo.
To determine whether MRAP2a was modulat- in linear growth and small increase in ghrh We previously reported that MC4R is consti-
ing MC4R in vivo, we knocked down mrap2a ex- were observed (Fig. 1, E to G). The specificity tutively inhibited by high levels of agouti-related
pression by using two distinct antisense morpholino of the morpholinos was demonstrated by in- protein (AgRP), the endogenous MC4R inverse
oligonucleotides targeting two different sites of hibition of a green fluorescent protein reporter agonist, in the zebrafish embryo (6). To determine
the mrap2a transcript. Relative to fish injected plasmid containing the target sequence, and by whether both AgRP and MRAP2a contribute to
with a nontargeting morpholino, embryos in- the rescue of both length and ghrh expression the silencing of MC4R, we injected wild-type
jected with either of the mrap2a morpholinos phenotype by co-injection of mrap2a mRNA zebrafish zygotes with control morpholinos or
showed a significant decrease in linear growth lacking the morpholino target sequences (fig. morpholinos targeting agrp alone, agrp and mrap2a,
[11% with the morpholino targeting the start S2, A to C). Overall, these results indicate that or agrp and mrap2b. As previously reported,
of the coding sequence, ATG, and 13% with the MRAP2a suppresses MC4R signaling in the lar- agrp morpholino caused a measurable decrease
morpholino targeting the 5′ untranslated region val zebrafish. The use of morpholinos to down- (7%) in fish growth by increasing MC4R activity,
(5′UTR)] (Fig. 1, B and D) and a factor of 3 to regulate mrap2b did not have any impact on as measured at 5 dpf (6). When co-injected, agrp
4 increase in ghrh expression (Fig. 1C), con- the size of the zebrafish or ghrh expression, and mrap2a morpholinos caused a more profound
sistent with an increase in MC4R activity. In regardless of genotype (Fig. 1); this result was impairment of growth (13%) than agrp morpholino

Fig. 1. Regulation of MC4R
by MRAP2a in the zebra-
fish embryo. (A) RT-PCR show-
ing the level of mRNA expression
for mc4r, mrap2a, and mrap2b
in the first 4 days of zebrafish
development. (B to G) Length
[(B) and (E)], ghrh expression
[(C) and (F)], and representa-
tive pictures [(D) and (G)] of
5 dpf wild-type or mc4r-null
zebrafish injected with the in-
dicated morpholinos. (H and
I) Length (H) and represent-
ative picture (I) of zebrafish embryos
injected with (top to bottom) control
morpholino or morpholino target-
ing agrp, mrap2a, or both. *P < 0.05,
***P < 0.001; ns, not significant.
Fig. 2. Regulation of MC4R by MRAP2a. (A) Surface expression of MC4R petition binding assay in HEK293T cells transfected with mc4r without or with
measured by whole-cell enzyme-linked immunosorbent assay (ELISA) in mrap2a at mc4r/mrap2a ratios of 1:1 or 1:6. (D) Concentration-response
nonpermeabilized HEK293T cells transfected with mc4r and increasing curves of a-MSH–induced cAMP production in HEK293T cells expressing
amounts of mrap2a. (B) Europium-labeled NDP–a-MSH binding in HEK293T the CRE-luciferase reporter, mc4r, and different amounts of mrap2a.
cells transfected with mc4r and increasing amounts of mrap2a. (C) Com- ***P < 0.001.

REPORTS
alone (Fig. 1, H and I), which suggests that AGRP dose-response curve for agonist, increasing a-MSH mutant fish, together suggest that the MRAP2
and MRAP2a both participate in keeping MC4R potency by a factor of 17 (MC4R alone, median proteins regulate multiple GPCRs. Nonetheless,
inactive in the zebrafish embryo. Co-injection of effective concentration EC50 = 140 T 60 nM; the MRAP2-GPCR interaction appears highly se-
agrp and mrap2b morpholinos had the same ef- MC4R + MRAP2b, EC50 = 8.3 T 1.7 nM). MRAP2b lective. MRAP2a and MRAP2b did not modify
fect as agrp alone. also amplified the maximal cAMP response to zebrafish MC3R signaling (fig. S7A), and mouse
We next investigated the mechanism by which a-MSH stimulation, from a factor of 1.5 in its MRAP2 had no effect on the function of the
MRAP2a regulates MC4R. Surface expression absence to a factor of 17.5 in its presence. The mouse corticotropin-releasing hormone 1 and 2
of MC4R at the plasma membrane of transfected mouse MRAP2 replicated all of the signaling receptors or the mouse neuropeptide Y2 and Y5
HEK293T cells was not changed by MRAP2a ex- effects caused by MRAP2b on the zebrafish receptors (fig. S7, B to E). A small change in ef-
pression (Fig. 2A). However, because the num- MC4R (Fig. 3F), thereby confirming that MRAP2b ficacy was observed in the presence of MRAP2 at
ber of receptors at the plasma membrane may not is the homologous isoform to the mammalian the mouse glucagon-like peptide–1 receptor and
reflect the number of receptors that are competent MRAP2. After expression in HEK293T cells, a b2-adrenergic receptor (fig. S7, F and G).
to bind ligand and signal, we measured the effect coimmunoprecipitation assay showed that MC4R Our findings reveal a new level of complexity
of MRAP2a on the density of high-affinity bind- copurified with MRAP2b (fig. S4A) and that in the regulation of MC4R signaling by uncover-
ing sites for the MC4R agonist NDP–a-MSH. MRAP2b copurified with MC4R (fig. S4B), which ing the ability of MRAP2 proteins to modify the
MRAP2a caused up to an 80% decrease in suggests that those proteins are in the same pharmacology and physiology of MC4R in zebra-
europium-labeled NDP–a-MSH binding to MC4R complex. fish. The zebrafish MC4R displays an atypical
(Fig. 2B). MRAP2a did not significantly change For MRAP2a or MRAP2b to have direct ef- signaling profile with very high constitutive ac-
the affinity of MC4R for a-MSH, as measured fects on the pharmacology and physiology of MC4R tivity and a modest cyclic adenosine monophos-

in a competition binding assay (median inhibito- in vivo, they must be expressed in the same cells. phate (cAMP) response to its agonist a-MSH. In
ry concentration IC50 = 45 T 15 nM) (Fig. 2C). We used in situ hybridization to localize mc4r and the embryo, the MC4R inverse agonist AgRP is
These results suggest that MRAP2a reduces lig- mrap2 mRNAs in adult zebrafish brain slices. expressed at a high level (6). We show here that
and binding by decreasing the number of binding mc4r and mrap2a were largely colocalized, as mrap2a is also expressed in the embryo and in-
sites but not by altering affinity. Both constitutive were mc4r and mrap2b (fig. S5). This result sup- hibits MC4R signaling. AgRP and MRAP2a col-
activity and a-MSH–inducible activity were supports the concept that the diverse effects of laborate to stabilize MC4R in an inactive state,
pressed in parallel by increasing expression of MRAP2s on MC4R signaling observed in vitro inhibiting both constitutive activity and ligand-
MRAP2a (Fig. 2D), which suggests that MRAP2a are relevant in vivo and may play a major role in induced signaling, and thus maximizing growth
stabilizes an inactive conformation of MC4R. energy homeostasis. during the larval period.
We performed a coimmunoprecipitation assay The broad tissue expression of MRAP2 pro- mrap2b is expressed at high levels only in
and found that MC4R and MRAP2a are part of teins in vertebrates, and our observation that mrap2a adult fish, where it could act to reduce the con-
the same complex; a large fraction of the receptor blockade continues to have a small effect in mc4r stitutive activity of MC4R and simultaneously
copurified with MRAP2a (fig. S3A), and con-
versely, a large fraction of MRAP2a coprecipi-
tated with MC4R (fig. S3B).
As mentioned previously, the zebrafish has
two mrap2 genes, mrap2a and mrap2b. As shown
above, mrap2b expression was barely detectable
in the embryo (Fig. 1A). To determine whether
its expression increased later in the life of the
zebrafish, we harvested RNA from the brain of
adult fish and measured the expression of mc4r,
mrap2a, and mrap2b, using RT-PCR and quan-
titative PCR (qPCR). The expression of all three
genes increased in the adult brain relative to em-
bryos (Fig. 3, A and B). Most notably, the ex-
pression of mrap2b increased in the adult by a
factor of 10, raising the possibility that MRAP2b
regulates MC4R function in the adult fish.
MRAP2 expression in the mouse is also not
seen during embryogenesis and appears only af-
ter birth, reaching maximal levels close to the time
of weaning, at day 18 (Fig. 3C). This suggests a
conservation of the expression kinetics for the
mouse MRAP2 and teleost MRAP2b proteins. In
contrast to MRAP2a, which had no effect on the
trafficking of MC4R, MRAP2b caused a modest
increase in MC4R surface expression in trans- Fig. 3. Regulation of MC4R by MRAP2b. (A) RT-PCR depicting the expression of mc4r, mrap2a,
fected HEK293T cells (Fig. 3D). MRAP2b also mrap2b, and the housekeeping gene ef1a in zebrafish adult brain. (B) qPCR measuring the change in
caused a dose-dependent increase in maximal expression of mc4r, mrap2a, and mrap2b in the adult brain relative to the 4 dpf embryo. (C) qPCR (top)
binding, probably due to the increased number and RT-PCR (bottom) depicting the expression of mouse MRAP2 at different embryonic and postnatal
of MC4 receptors at the cell surface; however, stages; E, embryonic day. (D) Surface expression of MC4R measured by whole-cell ELISA in nonper-
MRAP2b did not change the affinity of MC4R meabilized HEK293T cells transfected with mc4r and different amounts of mrap2b. (E) Competition
for a-MSH (IC50 = 50 T 15 nM) (Fig. 3E). binding assay in HEK293T cells transfected with mc4r alone or with mrap2b at the indicated ratio. (F)
Remarkably, MRAP2b suppressed the consti- Concentration-response curves of a-MSH–induced cAMP production in HEK293T cells transfected with
tutive activity of MC4R by 93% and shifted the mc4r and the indicated amount of mrap2b or mouse MRAP2. ***P < 0.001.

REPORTS
sensitizes the receptor to agonist. In this manner, switch is aided by MRAP2b, which forms a com- 7. J. A. Sebag, P. M. Hinkle, Proc. Natl. Acad. Sci. U.S.A.
MRAP2b would convert the adult zebrafish MC4R plex with MC4R and renders it highly sensitive to 104, 20244–20249 (2007).
8. J. A. Sebag, P. M. Hinkle, J. Biol. Chem. 284,
from a constitutively active to a ligand-dependent a-MSH. 22641–22648 (2009).
receptor. Additionally, mrap2b expression kinet- MRAP2 introduces a previously unappreciated 9. J. A. Sebag, P. M. Hinkle, Sci. Signal. 3, ra28 (2010).
ics matches that of the mouse MRAP2 and is level of complexity in the control of MC4R, with 10. P. M. Hinkle et al., Eur. J. Pharmacol. 660, 94–102
functionally homologous. MRAP2a and MRAP2b developmentally regulated paralogs in the fish (2011).
11. L. A. Metherell et al., Nat. Genet. 37, 166–170
proteins share a strong homology with each oth- that can either inhibit (MRAP2a) or stimulate (2005).
er and with mammalian MRAP2 in most of the (MRAP2b) ligand-mediated receptor activation 12. L. F. Chan et al., Proc. Natl. Acad. Sci. U.S.A. 106,
N-terminal region and the transmembrane do- (fig. S8). A component of this complexity is re- 6146–6151 (2009).
main, whereas the N-terminal 15 amino acids and tained in mammals: Asai et al. (16) show that, 13. S. Roy, M. Rached, N. Gallo-Payet, Mol. Endocrinol. 21,
1656–1669 (2007).
C terminus of these proteins are vastly divergent like MRAP2b, mouse MRAP2 expression is acti- 14. M. J. Agulleiro et al., Mol. Cell. Endocrinol. 320,
(fig. S6). Interestingly, the N-terminal and trans- vated proximal to weaning and increases the re- 145–152 (2010).
membrane domain of MRAP1 are sufficient for sponsiveness of MC4R to a-MSH. Their observation 15. J. A. Sebag, P. M. Hinkle, J. Biol. Chem. 284, 610–618
full activity of the mammalian MC2R (15). The that MRAP2 deletion causes an obesity syndrome (2009).
16. M. Asai et al., Science 341, 275–278 (2013).
first 15 amino acids of MRAP2a and MRAP2b in the mouse can likely be attributed, in part, to
could thus represent an important regulatory re- reduced function of MC4R (16). However, given Acknowledgments: Supported by NIH grants DK075721
gion of MRAP2s, possibly responsible for the the ubiquitous expression of MRAP2 proteins, and DK070332 (R.D.C.), DK19974 (P.M.H.), and
differential regulation of MC4R. we hypothesize that these proteins also modulate F23DK091055 ( J.A.S.); United States–Israel Binational
During zebrafish embryonic development, all the activity of GPCRs and perhaps other mem- Agricultural Research and Development Fund (BARD)

research grant IS-4489-12 (R.D.C. and C.Z.); and Vanderbilt
the energy consumed is obtained from the yolk brane proteins as well. Diabetes Research and Training Center grant DK020593
sac. Our findings suggest that the embryo bene- (R.D.C.). All data and methods are publicly available in
fits from having MC4R locked in an inactive state the supplementary materials.
by the joint actions of AgRP and MRAP2a. Ac- References and Notes
1. I. S. Farooqi et al., N. Engl. J. Med. 348, 1085–1095
tivation of MC4R at this stage would slow the rapid (2003). Supplementary Materials
maturation to the mobile free-feeding juvenile stage 2. D. Huszar et al., Cell 88, 131–141 (1997). www.sciencemag.org/cgi/content/full/341/6143/278/DC1
reached at 5 dpf. Upon maturation and depletion of 3. C. Vaisse, K. Clement, B. Guy-Grand, P. Froguel, Materials and Methods
Nat. Genet. 20, 113–114 (1998). Figs. S1 to S8
the yolk sac, the zebrafish must regulate nutrient
4. G. S. Yeo et al., Nat. Genet. 20, 111–112 (1998). References (17–19)
intake. Appropriate behavioral response to diurnal, 5. K. P. Lampert et al., Curr. Biol. 20, 1729–1734 (2010).
seasonal, and other inputs requires a function- 6. C. Zhang, P. M. Forlano, R. D. Cone, Cell Metab. 15, 20 November 2012; accepted 20 June 2013
al adipostat and energy balance sensor. This 256–264 (2012). 10.1126/science.1232995
Pandoraviruses: Amoeba Viruses lation (14, 15). In particular, seven virus-encoded

amino acid–transfer RNA (tRNA) ligases (8)
with Genomes Up to 2.5 Mb Reaching

and other enzymes thought to be the hallmark
of cellular organisms were found in these viruses
(16, 17). Their study also led to the discovery of
That of Parasitic Eukaryotes “virophages” that replicate within the virion fac-
tory of the Megaviridae (18–20).
After our discovery of M. chilensis with
Nadège Philippe,1,2* Matthieu Legendre,1* Gabriel Doutre,1 Yohann Couté,3 Olivier Poirot,1 laboratory-grown Acanthamoeba for amplifica-
Magali Lescot,1 Defne Arslan,1 Virginie Seltzer,1 Lionel Bertaux,1 Christophe Bruley,3 tion, we searched for new giant viruses in sedi-
Jérome Garin,3 Jean-Michel Claverie,1† Chantal Abergel1† ments where Acanthamoeba are more prevalent
than in the water column (21, 22). We identified
Ten years ago, the discovery of Mimivirus, a virus infecting Acanthamoeba, initiated a reappraisal samples demonstrating strong cellular lytic ac-
of the upper limits of the viral world, both in terms of particle size (>0.7 micrometers) and tivity. Some of these cocultures revealed the in-
genome complexity (>1000 genes), dimensions typical of parasitic bacteria. The diversity of these tracellular multiplication of particles larger than
giant viruses (the Megaviridae) was assessed by sampling a variety of aquatic environments that of the previously isolated Megaviridae, al-
and their associated sediments worldwide. We report the isolation of two giant viruses, one off beit without their icosahedral appearance. As
the coast of central Chile, the other from a freshwater pond near Melbourne (Australia), the multiplication of these particles was found to
without morphological or genomic resemblance to any previously defined virus families. Their be insensitive to antibiotics, they were retained for
micrometer-sized ovoid particles contain DNA genomes of at least 2.5 and 1.9 megabases, further investigation.
respectively. These viruses are the first members of the proposed “Pandoravirus” genus, a term Parasite 1 originated from the superficial ma-
reflecting their lack of similarity with previously described microorganisms and the surprises rine sediment layer (~10 m deep) taken at the
expected from their future study. mouth of the Tunquen river (coast of central
he serendipitous discovery of the first giant (>1000) than those of bacteria. In the past dec-
T
1
Structural and Genomic Information Laboratory, UMR 7256
DNA virus Mimivirus (1, 2), initially mis- ade, several Mimivirus relatives have been fully CNRS Aix-Marseille Université, 163 Avenue de Luminy, Case
interpreted as a Gram-positive parasitic characterized, including the largest known viral 934, 13288 Marseille cedex 9, France. 2Laboratory of Molecular
bacterium, challenged criteria and protocols his- genome of Megavirus chilensis (1.259 Mb en- Biophysics, Department of Cell and Molecular Biology, Uppsala
torically established to separate viruses from coding 1120 proteins) (6–8). The study of this new University, Husargatan 3 (Box 596), SE-751 24 Uppsala, Sweden.
3
CEA, IRTSV, Biologie à Grande Echelle, INSERM, U1038, Université
cellular organisms (3–5). It was then realized family of viruses (referred to as “Megaviridae”) Joseph Fourier Grenoble 1, F-38054 Grenoble, France.
that virus particles could be large enough to be revealed distinctive features concerning the virion *These authors contributed equally to this work.
visible under light microscope and contain DNA structure and core delivery mechanism (9, 10), †Corresponding author. E-mail: chantal.abergel@igs.cnrs-mrs.fr
genomes larger in size (>1 Mb) and gene contents transcription signaling (11–13), and protein trans- (C.A.); jean-michel.claverie@univ-amu.fr (J.-M.C.)

REPORTS
cannot be traced back to any known cellular 13. M. Legendre et al., Genome Res. 20, 664–674 39. D. Moreira, P. López-García, Nat. Rev. Microbiol. 7,
lineage. However, their DNA polymerase does (2010). 306–311 (2009).
14. C. Abergel, J. Rudinger-Thirion, R. Giegé, J.-M. Claverie, 40. J.-M. Claverie, H. Ogata, Nat. Rev. Microbiol. 7, 615,
cluster with those of other giant DNA viruses, J. Virol. 81, 12406–12417 (2007). author reply 615 (2009).
suggesting the controversial existence of a fourth 15. S. Jeudy, C. Abergel, J.-M. Claverie, M. Legendre, 41. P. Scheid, B. Hauröder, R. Michel, Parasitol. Res. 106,
domain of life (fig. S6) (1, 5, 39, 40). The absence PLoS Genet. 8, e1003122 (2012). 1371–1377 (2010).
of Pandoravirus-like sequences from the rapidly 16. J.-M. Claverie, C. Abergel, H. Ogata, Curr. Top. 42. R. Hoffmann, R. Michel, K.-D. Müller, E. N. Schmid,
Microbiol. Immunol. 328, 89–121 (2009). Endocytobiosis Cell Res. 12, 185 (1998).
growing environmental metagenomic databases 17. F. Piacente et al., J. Biol. Chem. 287, 3009–3018 (2012). 43. M. Krzywinski et al., Genome Res. 19, 1639–1645
suggests either that they are rare or that their eco- 18. B. La Scola et al., Nature 455, 100–104 (2008). (2009).
logical niche has never been prospected. However, 19. J.-M. Claverie, C. Abergel, Annu. Rev. Genet. 43, 49–66 44. K. Katoh, H. Toh, Brief. Bioinform. 9, 286–298 (2008).
the screening of the literature on Acanthamoeba (2009). 45. K. Tamura et al., Mol. Biol. Evol. 28, 2731–2739
20. C. Desnues et al., Proc. Natl. Acad. Sci. U.S.A. 109, (2011).
parasites does reveal that Pandoravirus-like par- 18078–18083 (2012).
ticles had been observed 13 years ago (41, 42), 21. D. A. Munson, T. A. Paget, J. Eukaryot. Microbiol. 53, Acknowledgments: We thank S. Faugeron and R. Finke
although not interpreted as viruses. This work S12–S14 (2006). from the Estación de Investigaciones Marinas in Chile for help
is a reminder that our census of the microbial 22. H. Liu et al., Korean J. Parasitol. 44, 117–125 (2006). during the sampling expedition. We also thank J. Hajdu for
23. A. Dolan, et al., J. Virol. 72, 2010–2021 (1998). invaluable support and J.-P. Chauvin and A. Aouane for expert
diversity is far from comprehensive and that some
24. S. Kurtz et al., Genome Biol. 5, R12 (2004). assistance on the Institut de Biologie du Développement de
important clues about the fundamental nature 25. E. W. Sayers et al., Nucleic Acids Res. 40, D13–D25 Marseille Luminy imagery facility, as well as A. Bernadac and
of the relationship between the viral and the (2012). A. Kosta from the Institut de Microbiologie de la Méditerranée.
cellular world might still lie within unexplored 26. M. Punta et al., Nucleic Acids Res. 40, D290–D301 We thank E. Fabre and V. Schmidt for technical assistance, and
environments. (2012). P. Bonin and R. Claverie for helpful discussions. This work was

27. M. Clarke et al., Genome Biol. 14, R11 (2013). supported by Centre National de la Recherche Scientifique,
28. J. Shi, T. L. Blundell, K. Mizuguchi, J. Mol. Biol. 310, Institut National de la Santé et de la Recherche Médicale,
References and Notes 243–257 (2001). Centre de l’Energie Atomique, and Agence National pour la
1. B. La Scola et al., Science 299, 2033 (2003). 29. D. W. Adams, J. Errington, Nat. Rev. Microbiol. 7, Recherche (ANR-BLAN08-0089, ANR-09-GENM-032-001, and
2. D. Raoult et al., Science 306, 1344–1350 (2004). 642–653 (2009). ANR-10-INBS-08-01). The sampling expedition was sponsored
3. A. Lwoff, J. Gen. Microbiol. 17, 239–253 (1957). 30. M. T. Cabeen, C. Jacobs-Wagner, Annu. Rev. Genet. 44, by the ASSEMBLE grant 227799. The genome sequences of
4. J.-M. Claverie, C. Abergel, Trends Genet. 26, 431–437 365–392 (2010). P. salinus and P. dulcis have been deposited in GenBank (accession
(2010). 31. R. Y. Samson, S. D. Bell, Trends Microbiol. 17, 507–513 numbers KC977471 and KC977470, respectively). The mass
5. J.-M. Claverie, C. Abergel, Adv. Virus Res. 85, 25–56 (2009). spectrometry proteomics data have been deposited to the
(2013). 32. L. M. Iyer, L. Aravind, E. V. Koonin, J. Virol. 75, ProteomeXchange Consortium (http://proteomecentral.
6. P. Colson et al., Genome Biol. Evol. 3, 737–742 11720–11734 (2001). proteomexchange.org) via the PRIDE partner repository with the
(2011). 33. J. K. Hyun et al., J. Virol. 81, 11075–11083 (2007). data set identifier PXD000213 and DOI 10.6019/PXD000213.
7. N. Yoosuf et al., Genome Biol. Evol. 4, 1324–1330 34. M. G. Fischer, M. J. Allen, W. H. Wilson, C. A. Suttle,
(2012). Proc. Natl. Acad. Sci. U.S.A. 107, 19508–19513
8. D. Arslan, M. Legendre, V. Seltzer, C. Abergel, (2010).
J.-M. Claverie, Proc. Natl. Acad. Sci. U.S.A. 108, 35. L. M. Iyer, S. Balaji, E. V. Koonin, L. Aravind, Virus Res.
17486–17491 (2011). 117, 156–184 (2006).
Figs. S1 to S6
9. C. Xiao et al., PLoS Biol. 7, e92 (2009). 36. H. Ogata et al., ISME J. 5, 1143–1151 (2011).
Tables S1 to S5
10. N. Zauberman et al., PLoS Biol. 6, e114 (2008). 37. N. Corradi, J. F. Pombert, L. Farinelli, E. S. Didier,
11. K. Suhre, S. Audic, J.-M. Claverie, Proc. Natl. Acad. P. J. Keeling, Nat. Commun 1, 77 (2010).
Sci. U.S.A. 102, 14689–14693 (2005). 38. J. F. Pombert et al., Proc. Natl. Acad. Sci. U.S.A. 109, 15 April 2013; accepted 13 June 2013
12. D. Byrne et al., Genome Res. 19, 1233–1242 (2009). 12638–12643 (2012). 10.1126/science.1239181
Sept4/ARTS Regulates Stem Cell malian gene Septin4 (Sept4) (10, 11). Deletion
of the Sept4/ARTS gene results in increased
Apoptosis and Skin Regeneration

numbers of hematopoietic stem and progenitor
cells and elevated XIAP levels. This causes in-
creased apoptotic resistance and accelerated tu-
Yaron Fuchs,1 Samara Brown,1 Travis Gorenc,1 Joe Rodriguez,1 Elaine Fuchs,2* Hermann Steller1* mor development (12). Here, we report crucial
roles of XIAP and Sept4/ARTS in regulating hair
Adult stem cells are essential for tissue homeostasis and wound repair. Their proliferative follicle stem cell (HFSC) apoptosis and show that
capacity must be tightly regulated to prevent the emergence of unwanted and potentially apoptotic alterations have profound consequences
dangerous cells, such as cancer cells. We found that mice deficient for the proapoptotic for wound healing and regeneration.
Sept4/ARTS gene have elevated numbers of hair follicle stem cells (HFSCs) that are protected Hair follicles cycle between phases of growth
against apoptosis. Sept4/ARTS−/− mice display marked improvement in wound healing and (anagen), destruction (catagen), and rest (telogen).
regeneration of hair follicles. These phenotypes depend on HFSCs, as indicated by lineage tracing. This process requires distinct populations of HFSCs
Inactivation of XIAP, a direct target of ARTS, abrogated these phenotypes and impaired wound that reside within the bulge (13–16). ARTS was
healing. Our results indicate that apoptosis plays an important role in regulating stem cell–dependent the only Sept4 isoform detected in HFSCs (fig.
regeneration and suggest that this pathway may be a target for regenerative medicine. S1, A to C). To investigate the consequences of
ARTS deficiency, we examined bulge HFSCs
he ability of stem cells (SCs) to self-renew tively little is known about the role of apoptosis with specific bulge markers (CD34 and K15)
T and differentiate enables them to replace

cells that die during tissue homeostasis or
upon injury. Elevated SC numbers might be de-
in controlling SC numbers and its possible effect
on SC-dependent regeneration. Apoptosis is exe-
cuted by caspases that are negatively regulated
1
Strang Laboratory of Apoptosis and Cancer Biology, Howard
Hughes Medical Institute, The Rockefeller University, New York,
sirable, at least transiently, to enhance tissue repair by IAPs (inhibitor of apoptosis proteins) (5, 6).
NY 10065, USA. 2Laboratory of Mammalian Cell Biology and
(1, 2). However, a large SC pool may potentially The best-studied mammalian IAP is XIAP (7). In Development, Howard Hughes Medical Institute, The Rockefeller
increase the risk of cancer (3). cells destined to die, IAPs are inactivated by University, New York, NY 10065, USA.
One major mechanism that eliminates unde- specific antagonists (8, 9). One mammalian IAP *Corresponding author. E-mail: steller@rockefeller.edu (H.S.);
sired and dangerous cells is apoptosis (4). Rela- antagonist is ARTS, a splice variant of the mam- fuchslb@rockefeller.edu (E.F.)

Sept4/ARTS Regulates Stem Cell Apoptosis and Skin Regeneration
Yaron Fuchs et al.
Science 341, 286 (2013);
DOI: 10.1126/science.1233029


Development
http://www.sciencemag.org/cgi/collection/development
REPORTS
cannot be traced back to any known cellular 13. M. Legendre et al., Genome Res. 20, 664–674 39. D. Moreira, P. López-García, Nat. Rev. Microbiol. 7,
lineage. However, their DNA polymerase does (2010). 306–311 (2009).
14. C. Abergel, J. Rudinger-Thirion, R. Giegé, J.-M. Claverie, 40. J.-M. Claverie, H. Ogata, Nat. Rev. Microbiol. 7, 615,
cluster with those of other giant DNA viruses, J. Virol. 81, 12406–12417 (2007). author reply 615 (2009).
suggesting the controversial existence of a fourth 15. S. Jeudy, C. Abergel, J.-M. Claverie, M. Legendre, 41. P. Scheid, B. Hauröder, R. Michel, Parasitol. Res. 106,
domain of life (fig. S6) (1, 5, 39, 40). The absence PLoS Genet. 8, e1003122 (2012). 1371–1377 (2010).
of Pandoravirus-like sequences from the rapidly 16. J.-M. Claverie, C. Abergel, H. Ogata, Curr. Top. 42. R. Hoffmann, R. Michel, K.-D. Müller, E. N. Schmid,
Microbiol. Immunol. 328, 89–121 (2009). Endocytobiosis Cell Res. 12, 185 (1998).
growing environmental metagenomic databases 17. F. Piacente et al., J. Biol. Chem. 287, 3009–3018 (2012). 43. M. Krzywinski et al., Genome Res. 19, 1639–1645
suggests either that they are rare or that their eco- 18. B. La Scola et al., Nature 455, 100–104 (2008). (2009).
logical niche has never been prospected. However, 19. J.-M. Claverie, C. Abergel, Annu. Rev. Genet. 43, 49–66 44. K. Katoh, H. Toh, Brief. Bioinform. 9, 286–298 (2008).
the screening of the literature on Acanthamoeba (2009). 45. K. Tamura et al., Mol. Biol. Evol. 28, 2731–2739
20. C. Desnues et al., Proc. Natl. Acad. Sci. U.S.A. 109, (2011).
parasites does reveal that Pandoravirus-like par- 18078–18083 (2012).
ticles had been observed 13 years ago (41, 42), 21. D. A. Munson, T. A. Paget, J. Eukaryot. Microbiol. 53, Acknowledgments: We thank S. Faugeron and R. Finke
although not interpreted as viruses. This work S12–S14 (2006). from the Estación de Investigaciones Marinas in Chile for help
is a reminder that our census of the microbial 22. H. Liu et al., Korean J. Parasitol. 44, 117–125 (2006). during the sampling expedition. We also thank J. Hajdu for
23. A. Dolan, et al., J. Virol. 72, 2010–2021 (1998). invaluable support and J.-P. Chauvin and A. Aouane for expert
diversity is far from comprehensive and that some
24. S. Kurtz et al., Genome Biol. 5, R12 (2004). assistance on the Institut de Biologie du Développement de
important clues about the fundamental nature 25. E. W. Sayers et al., Nucleic Acids Res. 40, D13–D25 Marseille Luminy imagery facility, as well as A. Bernadac and
of the relationship between the viral and the (2012). A. Kosta from the Institut de Microbiologie de la Méditerranée.
cellular world might still lie within unexplored 26. M. Punta et al., Nucleic Acids Res. 40, D290–D301 We thank E. Fabre and V. Schmidt for technical assistance, and
environments. (2012). P. Bonin and R. Claverie for helpful discussions. This work was

27. M. Clarke et al., Genome Biol. 14, R11 (2013). supported by Centre National de la Recherche Scientifique,
28. J. Shi, T. L. Blundell, K. Mizuguchi, J. Mol. Biol. 310, Institut National de la Santé et de la Recherche Médicale,
References and Notes 243–257 (2001). Centre de l’Energie Atomique, and Agence National pour la
1. B. La Scola et al., Science 299, 2033 (2003). 29. D. W. Adams, J. Errington, Nat. Rev. Microbiol. 7, Recherche (ANR-BLAN08-0089, ANR-09-GENM-032-001, and
2. D. Raoult et al., Science 306, 1344–1350 (2004). 642–653 (2009). ANR-10-INBS-08-01). The sampling expedition was sponsored
3. A. Lwoff, J. Gen. Microbiol. 17, 239–253 (1957). 30. M. T. Cabeen, C. Jacobs-Wagner, Annu. Rev. Genet. 44, by the ASSEMBLE grant 227799. The genome sequences of
4. J.-M. Claverie, C. Abergel, Trends Genet. 26, 431–437 365–392 (2010). P. salinus and P. dulcis have been deposited in GenBank (accession
(2010). 31. R. Y. Samson, S. D. Bell, Trends Microbiol. 17, 507–513 numbers KC977471 and KC977470, respectively). The mass
5. J.-M. Claverie, C. Abergel, Adv. Virus Res. 85, 25–56 (2009). spectrometry proteomics data have been deposited to the
(2013). 32. L. M. Iyer, L. Aravind, E. V. Koonin, J. Virol. 75, ProteomeXchange Consortium (http://proteomecentral.
6. P. Colson et al., Genome Biol. Evol. 3, 737–742 11720–11734 (2001). proteomexchange.org) via the PRIDE partner repository with the
(2011). 33. J. K. Hyun et al., J. Virol. 81, 11075–11083 (2007). data set identifier PXD000213 and DOI 10.6019/PXD000213.
7. N. Yoosuf et al., Genome Biol. Evol. 4, 1324–1330 34. M. G. Fischer, M. J. Allen, W. H. Wilson, C. A. Suttle,
(2012). Proc. Natl. Acad. Sci. U.S.A. 107, 19508–19513
8. D. Arslan, M. Legendre, V. Seltzer, C. Abergel, (2010).
J.-M. Claverie, Proc. Natl. Acad. Sci. U.S.A. 108, 35. L. M. Iyer, S. Balaji, E. V. Koonin, L. Aravind, Virus Res.
17486–17491 (2011). 117, 156–184 (2006).
Figs. S1 to S6
9. C. Xiao et al., PLoS Biol. 7, e92 (2009). 36. H. Ogata et al., ISME J. 5, 1143–1151 (2011).
Tables S1 to S5
10. N. Zauberman et al., PLoS Biol. 6, e114 (2008). 37. N. Corradi, J. F. Pombert, L. Farinelli, E. S. Didier,
11. K. Suhre, S. Audic, J.-M. Claverie, Proc. Natl. Acad. P. J. Keeling, Nat. Commun 1, 77 (2010).
Sci. U.S.A. 102, 14689–14693 (2005). 38. J. F. Pombert et al., Proc. Natl. Acad. Sci. U.S.A. 109, 15 April 2013; accepted 13 June 2013
12. D. Byrne et al., Genome Res. 19, 1233–1242 (2009). 12638–12643 (2012). 10.1126/science.1239181
Sept4/ARTS Regulates Stem Cell malian gene Septin4 (Sept4) (10, 11). Deletion
of the Sept4/ARTS gene results in increased
Apoptosis and Skin Regeneration

numbers of hematopoietic stem and progenitor
cells and elevated XIAP levels. This causes in-
creased apoptotic resistance and accelerated tu-
Yaron Fuchs,1 Samara Brown,1 Travis Gorenc,1 Joe Rodriguez,1 Elaine Fuchs,2* Hermann Steller1* mor development (12). Here, we report crucial
roles of XIAP and Sept4/ARTS in regulating hair
Adult stem cells are essential for tissue homeostasis and wound repair. Their proliferative follicle stem cell (HFSC) apoptosis and show that
capacity must be tightly regulated to prevent the emergence of unwanted and potentially apoptotic alterations have profound consequences
dangerous cells, such as cancer cells. We found that mice deficient for the proapoptotic for wound healing and regeneration.
Sept4/ARTS gene have elevated numbers of hair follicle stem cells (HFSCs) that are protected Hair follicles cycle between phases of growth
against apoptosis. Sept4/ARTS−/− mice display marked improvement in wound healing and (anagen), destruction (catagen), and rest (telogen).
regeneration of hair follicles. These phenotypes depend on HFSCs, as indicated by lineage tracing. This process requires distinct populations of HFSCs
Inactivation of XIAP, a direct target of ARTS, abrogated these phenotypes and impaired wound that reside within the bulge (13–16). ARTS was
healing. Our results indicate that apoptosis plays an important role in regulating stem cell–dependent the only Sept4 isoform detected in HFSCs (fig.
regeneration and suggest that this pathway may be a target for regenerative medicine. S1, A to C). To investigate the consequences of
ARTS deficiency, we examined bulge HFSCs
he ability of stem cells (SCs) to self-renew tively little is known about the role of apoptosis with specific bulge markers (CD34 and K15)
T and differentiate enables them to replace

cells that die during tissue homeostasis or
upon injury. Elevated SC numbers might be de-
in controlling SC numbers and its possible effect
on SC-dependent regeneration. Apoptosis is exe-
cuted by caspases that are negatively regulated
1
Strang Laboratory of Apoptosis and Cancer Biology, Howard
Hughes Medical Institute, The Rockefeller University, New York,
sirable, at least transiently, to enhance tissue repair by IAPs (inhibitor of apoptosis proteins) (5, 6).
NY 10065, USA. 2Laboratory of Mammalian Cell Biology and
(1, 2). However, a large SC pool may potentially The best-studied mammalian IAP is XIAP (7). In Development, Howard Hughes Medical Institute, The Rockefeller
increase the risk of cancer (3). cells destined to die, IAPs are inactivated by University, New York, NY 10065, USA.
One major mechanism that eliminates unde- specific antagonists (8, 9). One mammalian IAP *Corresponding author. E-mail: steller@rockefeller.edu (H.S.);
sired and dangerous cells is apoptosis (4). Rela- antagonist is ARTS, a splice variant of the mam- fuchslb@rockefeller.edu (E.F.)

REPORTS
as well as Sox9, which, depending on the hair follicles, this strand forms during mid- to late found no effect of Sept4/ARTS (fig. S3, E to K).
follicle phase, labels HFSCs and/or progeny. Al- catagen phase, when most matrix cells have died These findings suggest that loss of ARTS pro-
though Sept4/ARTS−/− hair follicles had bulges and only a small number of outer root sheath tects HFSCs against apoptosis.
with overall normal morphology, fluorescence- (ORS) cells remain (18). Because ARTS is a pro- In wild-type catagen follicles, a new bulge is
activated cell sorting (FACS) revealed that within apoptotic protein, the elongated epithelial strand formed from CD34+ HFSCs along the upper ORS,
the a6+b1+ population of skin epithelial progen- suggested that more ORS cells survive catagen whereas surviving CD34– ORS cells form the new
itors there were more than twice as many CD34+ when ARTS is absent. hair germ (18). Both wild-type and Sept4/ARTS−/−
cells in Sept4/ARTS−/− as in the wild type (Fig. 1, A To test whether ARTS affects the survival of hair follicles formed new bulges (fig. S4, A and
and B, and fig. S1D). Similarly, for the Tg(Krt1-15- skin progenitors, we cultured CD34+Sca1– and B). Apoptotic cells are rarely seen in telogen or
EGFP)2Cot/J reporter mouse, which specifically CD34–Sca1+ keratinocytes from telogen-phase anagen but are prevalent in catagen (19). There-
marks K15+ bulge and hair germ cells (17), the backskins and evaluated their colony-forming fore, we surmised that the pronounced epithelial
numbers of K15-GFP+a6hi cells were elevated efficiency (≥4 cells per colony) and cell number. strand seen in Sept4/ARTS−/− hair follicles might
when backskin hair follicles lacked Sept4/ARTS Sept4/ARTS−/− HFSCs generated ~2.5 times as reflect enhanced apoptotic resistance of HFSCs.
(Fig. 1, C and D). many colonies as wild-type HFSCs (fig. S3A). To test this hypothesis, we examined apoptosis in
ARTS-dependent differences were even more Moreover, the number of cells within these colo- the first catagen phase [postnatal day 16 (P16)]
striking in tailskin (fig. S2). At 8 weeks of age, nies was significantly higher (fig. S3C). By con- and found a striking decrease of cell death in
Sept4/ARTS−/− tailskin hair follicles displayed an trast, epidermal keratinocytes were unaffected Sept4/ARTS−/− mice (Fig. 1E and fig. S4, C and
expanded epithelial strand populated by K15+ (fig. S3, B and D). We also examined the prolifera- D). These results suggest that in Sept4/ARTS−/−
and Sox9+ cells (fig. S2). In normal backskin tion rates of HFSCs and of other skin cells, and mice, more ORS HFSCs are spared during catagen,

thereby yielding a larger HFSC pool. Because
there is only a single bulge at this stage, it also
explains why the bulge region appeared larger in
P21 Sept4/ARTS−/− hair follicles (Fig. 1A).
In contrast to backskin, tailskin had CP3+
(cleaved caspase3+) cells in the bulge of what ap-
peared to be late catagen-phase wild-type hair
follicles (Fig. 1F and fig. S4E). Furthermore,
these CD34+K15+ bulge cells displayed mem-
brane blebbing, nuclear condensation, and TUNEL
(terminal deoxynucleotidyl transferase–mediated
deoxyuridine triphosphate nick end labeling), in-
dicating that they were bona fide apoptotic cells
(fig. S4E). This suggests that HFSC apoptosis
can occur within the bulge of tailskin hair follicles.
Next, we examined whether deletion of
Sept4/ARTS protects HFSCs from staurosporine-
and etoposide-induced apoptosis in vitro. Rela-
tive to wild-type HFSCs, Sept4/ARTS−/− HFSCs
displayed a marked decrease in apoptotic markers
(Fig. 1G). Because Sept4/ARTS−/− HFSCs are re-
sistant to apoptosis, we examined whether they
have greater capacity to cope with stress by cul-
turing HFSCs without sustaining feeder cells.
In contrast to wild-type HFSCs, Sept4/ARTS−/−
HFSCs exhibited an increase in cell number, reach-
ing confluence under conditions that severely
hindered the growth of control cells (fig. S5). Collec-
tively, these results reveal an important physiolog-
ical function of Sept4/ARTS for HFSC apoptosis
in response to stress.
HFSCs do not contribute to normal epidermal
homeostasis. However, in response to wounding,
Fig. 1. Sept4/ARTS−/− mice display HFSCs that are resistant to apoptosis. (A) Dorsal whole mounts they participate in repopulating the epidermis
(DWMs) stained for CD34, K15, Sox9, and 4´,6-diamidino-2-phenylindole (DAPI) at P21. WT, wild type.
(20, 21). Because Sept4/ARTS−/− HFSCs are pro-
(B) FACS analysis of dorsal skins assessing the percentage of bulge HFSCs within control (WT) and
tected against cell death, we investigated whether
Sept4/ARTS−/− skins. Equivalent numbers of FACS-purified a6+b1+ progenitors were resorted for
CD34+Sca1– (pink, a6+b1+CD34+SCa1–; blue, a6+b1+CD34–SCa1+; orange, a6+b1+CD34–SCa1–). they have enhanced capacity for tissue repair.
Data shown are from four mice combined and sorted together. Experiments were repeated four times Full-thickness excision wounds (1 cm2) were
independently (n ≥ 11, P < 0.001). (C) DWM images of control and Sept4/ARTS−/−:Tg(Krt1-15-EGFP)2Cot/J generated on the dorsal skins of 8- and 15-week-
reporter skins. (D) FACS analyses of a6+K15+ cells from control and Sept4/ARTS−/−:Tg(Krt1-15-EGFP)2Cot/J old mice and monitored for wound coverage. In
reporter skins. (E) Quantifications of K15+TUNEL+ and Sox9+TUNEL+ cells during first catagen (P16). S/A−/− 8-week-old Sept4/ARTS−/− mice, the wound size
denotes Sept4/ARTS−/−. (F) Quantifications of CD34+CP3+ HFSCs in control and Sept4/ARTS−/− tailskin hair was reduced by 80% after only 5 days, whereas
follicles. Horizontal line defines average. (G) Top: Z-stack of control cultured HFSCs undergoing apoptosis in wild-type mice the wound size was reduced by
in vitro. Bottom: Percentage of sorted HFSCs (CD34+Sca1–) placed into culture that are CP3+ with and only 10% (Fig. 2A and fig. S6A). Accelerated
without staurosporine (STS) and etoposide (ETO) treatment. Scale bars, 100 mm (A), 20 mm (C), 10 mm (G). healing was seen at all time points. In 15-week-
***P < 0.001. old mice, healing was generally slower, but again

REPORTS
Fig. 2. Loss of Sept4/ARTS func-

tion accelerates wound healing and
improves skin regeneration. (A and
B) Reepithelialization dynamics of skins
at different times PWI. Excision wounds
(1 cm2) were inflicted on dorsal skin of
8-week-old [(A), n = 12] or 15-week-
old mice [(B), n = 8]. Percentage of
wound coverage was calculated versus
original wound size. D denotes day PWI.
(C) Hematoxylin and eosin staining of
full-thickness excision wounds of 8-week-
old mice, 18 days PWI. Hair follicle for-
mation within the wound bed is observed
in Sept4/ARTS−/− mice. (D to F) HFSC
niches within regenerated hair follicles
are positive for CD34, K15, and Sox9.
(D) Regenerated hair follicle displays a
HFSC niche positive for K15 and Sox9.
(E) Zoom-in of the HFSC niche in (D).

(F) Regenerated hair follicle niche is
positive for CD34. (G and H) Immuno-
fluorescence for PCNA (G) and Ki67 (H),
indicating proliferative activity with-
in the regenerated hair follicle niche of
Sept4/ARTS−/− mice. Scale bars, 500 mm
(C), 50 mm (D), 20 mm [(E), (F), and (G)], 10 mm (H).
Fig. 3. Sept4/ARTS−/− sometimes appeared hyperplastic but otherwise

HFSCs are responsible appeared relatively normal (Fig. 2, C to H). Long
for enhanced tissue re- after wild-type HFSCs had returned to quiescence,
generation. (A to H) some Sept4/ARTS−/− HFSCs were still prolifera-
Reporter expression was tive, as indicated by PCNA (proliferating cell
induced from P20 to P25 nuclear antigen) and Ki67 staining (Fig. 2, G and
[(A) to (F)] or from P45 H, and fig. S6B). Together, these results sug-
to P50 [(G) and (H)], and gest that the enhanced survival of HFSCs in
wounding was executed Sept4/ARTS−/− mice is responsible for acceler-
at P26 or P56, respec-
ated repair and hair follicle regeneration in re-
tively (see fig. S7). Skins
sponse to injury.
were analyzed at t = 0,
3, 7, 18, and 60 days To examine the contribution of HFSCs to ac-
PWI. WT [(A), (D), (G)] and celerated wound healing in Sept4/ARTS−/− mice,
Sept4/ARTS−/− [(B), (E), (H)] we used Tg(Krt1-15-cre/PGR)22Cot; Rosa26-YFP
lineage tracings are shown reporter mice. When induced by RU486, this re-
at t = 0, 3, and 60 days. porter specifically activates Cre in HFSCs, and there-
(H), right panel: Close-up of hair fol- after they and their progeny are marked by yellow
licle at 60 days PWI. Dashed line indi- fluorescent protein (YFP) (22). Sept4/ARTS−/−
cates dermis-epidermis border; B denotes follicles displayed about 3.5 times as many YFP-
bulge. Scale bars, 50 mm [(A), (B), (G)], marked HFSCs at t = 0 (Fig. 3, A to C, and fig.
100 mm [(D), (E), (H)]. (C) Quantifica- S7, A and B). By t = 3 days and t = 7 days after
tions of YFP+ cells in hair follicles of WT wounding, 4 to 5 times as many YFP+ cells were
and Sept4/ARTS−/− mice (t = 0). (F) Quan- present in the upper hair follicles and epider-
tification of YFP+ cells in hair follicles mis of Sept4/ARTS−/− skin (Fig. 3, D to F, and
and epidermis of WT and Sept4/ARTS−/− fig. S7, C and D). Enhanced repair was also ob-
mice (t = 3 and 7 days). (I and J) Scar size of WT served when Sept4/ARTS−/− mice were lineage-
and Sept4/ARTS−/− mice at 60 days PWI (2 cm2) marked and wounded during the protracted second
(n = 3); photograph (I) and quantification (J) telogen (fig. S7, E to G). Moreover, in contrast to
are shown. Dashed line indicates scar border. wild-type regeneration, Sept4/ARTS−/− HFSC prog-
**P < 0.002, ***P < 0.001.
eny remained in the skin epidermis at 2 months
PWI, and the scars of Sept/ARTS−/− were significant-
ly smaller (Fig. 3, G to J). Taken together, these
results suggest that apoptosis-impaired Sept4/
Sept4/ARTS−/− mice were more efficient in ARTS−/− mice had healed. Histological analy- ARTS−/− HFSCs are functional and robustly con-
wound repair (Fig. 2B). ses revealed that in contrast to wild-type mice, tribute to healing and hair follicle regeneration.
By 18 days post–wound infliction (PWI), high numbers of hair follicles were seen in the XIAP is a direct downstream target for the
1-cm2 wounds of 8-week-old wild-type and Sept4/ wound beds of Sept4/ARTS−/− skins, which proapoptotic activity of ARTS (11, 12, 23, 24).

REPORTS
XIAP−/− mice are viable and do not display overt from SX−/− mutants. In contrast to Sept4/ARTS−/− that targeting apoptotic pathways in HFSCs may
phenotypes (7, 25). However, microarray and chro- (CD34+) HFSCs, SX−/− HFSCs exhibited signif- have therapeutic benefits to promote wound healing
matin immunoprecipitation sequencing analy- icantly increased apoptosis (Fig. 4D and fig. S10). and regeneration.
ses indicate that XIAP is expressed throughout XIAP directly inhibits caspases and contains a
skin epithelium, including the bulge (26). Because RING domain required for E3 ligase activity References and Notes
ARTS can promote ubiquitination and degrada- (29). To address its importance for SC apoptosis, 1. I. L. Weissman, Science 287, 1442–1446 (2000).
2. S. Rafii, D. Lyden, Nat. Med. 9, 702–712 (2003).
tion of XIAP (27, 28), we compared XIAP pro- we analyzed XIAPDRING mice, in which the XIAP 3. Y. C. Hsu, E. Fuchs, Nat. Rev. Mol. Cell Biol. 13, 103–114
tein levels in wild-type and Sept4/ARTS−/− HFSCs. RING domain was deleted (7). CD34+ SCs from (2012).
In wild-type mice, anti-XIAP immunolabeling Sept4/ARTS−/−;XIAPDRING (SXDRING) and SX−/− 4. Y. Fuchs, H. Steller, Cell 147, 742–758 (2011).
was detected in bulge HFSCs, sebaceous glands, mice had virtually identical phenotypes (Fig. 5. M. Gyrd-Hansen, P. Meier, Nat. Rev. Cancer 10, 561–574
(2010).
and dermal papillae (Fig. 4, A and B, and fig. 4D). Finally, inactivation of XIAP also impaired 6. E. Kuranaga, M. Miura, Trends Cell Biol. 17, 135–144
S8). Most notable, however, was the striking in- the growth of CD34+ HFSCs in vitro (fig. S10). (2007).
crease of XIAP in tailskin Sept4/ARTS−/− HFSCs Together, these results show that XIAP is a phys- 7. A. J. Schile, M. García-Fernández, H. Steller, Genes Dev.
(Fig. 4B and fig. S9A). iological downstream target for the proapoptotic 22, 2256–2266 (2008).
8. S. Kornbluth, K. White, J. Cell Sci. 118, 1779–1787
To determine the biological role of elevated and regenerative activity of Sept4/ARTS in skin, (2005).
XIAP in Sept4/ARTS−/− hair follicles, we gen- and that XIAP RING function is required for 9. A. Bergmann, A. Y. Yang, M. Srivastava, Curr. Opin.
erated double knockout Sept4/ARTS−/−;XIAP−/− the antiapoptotic activity of XIAP in HFSCs. Cell Biol. 15, 717–724 (2003).
(SX−/−) animals and examined them for wound Our findings reveal the importance of 10. S. Larisch et al., Nat. Cell Biol. 2, 915–921 (2000).
repair and hair follicle regeneration. Intriguingly, apoptosis in regulating hair follicle homeostasis 11. Y. Gottfried, A. Rotem, R. Lotan, H. Steller, S. Larisch,

EMBO J. 23, 1627–1635 (2004).
loss of XIAP function abolished the enhanced and wound repair, and the nonredundant func- 12. M. García-Fernández et al., Genes Dev. 24, 2282–2293
reepithelialization of Sept4/ARTS−/− mice. In tions of ARTS and XIAP in this process. At the (2010).
addition, wound repair in SX−/− mutants was surface, our results may appear at odds with a 13. E. Fuchs, Nature 445, 834–842 (2007).
markedly delayed relative to the wild type (Fig. previously reported role of apoptosis in promot- 14. S. J. Morrison, A. C. Spradling, Cell 132, 598–611
(2008).
4C). XIAP−/− single mutants were severely coming skin wound healing, which was ascribed to 15. C. A. Jahoda, A. M. Christiano, Cell 146, 678–681
promised in wound repair (Fig. 4C and fig. S9B). mitogenic signaling by apoptotic cells (30). How- (2011).
Taken together, these results demonstrate a crit- ever, depending on the point of interference in the 16. W. M. Woo, A. E. Oro, Cell 146, 334 (2011).
ical physiological role of XIAP in wound heal- apoptotic pathway, the outcome for mitogenic sig- 17. R. J. Morris et al., Nat. Biotechnol. 22, 411–417
(2004).
ing. Moreover, because ARTS was unable to naling can vary considerably (31). Furthermore, 18. Y. C. Hsu, H. A. Pasolli, E. Fuchs, Cell 144, 92–105
stimulate healing in the absence of XIAP, this our studies suggest that ARTS and XIAP function (2011).
suggests that XIAP is a major target for the pro- specifically in regulating apoptosis of progenitors 19. S. Müller-Röver et al., J. Invest. Dermatol. 117, 3–15
apoptotic activity of ARTS in this system (11, 12). in the hair follicle, but not in more differentiated (2001).
20. M. Ito et al., Nat. Med. 11, 1351–1354 (2005).
To investigate whether XIAP participates in skin cells. Therefore, targeting these genes is very
21. I. Brownell, E. Guevara, C. B. Bai, C. A. Loomis
wound healing by suppressing apoptosis, we an- different from a general inhibition of effector cas- A. L. Joyner, Cell Stem Cell 8, 552–565 (2011).
alyzed the cell death sensitivity of CD34+ HFSCs pases in differentiated skin cells. Our results suggest 22. M. Ito et al., Nature 447, 316–320 (2007).
23. J. C. Reed, Cancer Cell 3, 17–22 (2003).
24. E. C. LaCasse et al., Oncogene 27, 6252–6275
Fig. 4. XIAP is required (2008).
25. H. Harlin, S. B. Reffey, C. S. Duckett, T. Lindsten
for proper wound heal- C. B. Thompson, Mol. Cell. Biol. 21, 3604–3608
ing. (A) Immunofluores- (2001).
cence indicates that XIAP 26. W. H. Lien et al., Cell Stem Cell 9, 219–232 (2011).
protein is present in dor- 27. J. B. Garrison et al., Mol. Cell 41, 107–116 (2011).
sal telogen bulge (P20) 28. N. Edison et al., Cell Death Differ. 19, 356–368
hair follicles. (B) Immu- (2012).
29. S. Fulda, D. Vucic, Nat. Rev. Drug Discov. 11, 109–124
nofluorescence indicates (2012).
increased XIAP in Sept4/ 30. F. Li et al., Sci. Signal. 3, ra13 (2010).
ARTS−/− tailskin hair fol- 31. A. Bergmann, H. Steller, Sci. Signal. 3, re8 (2010).
licles. B denotes bulge;
SG, sebaceous gland. (C) Acknowledgments: We apologize to colleagues whose
contributions we could not adequately cite because of space
Wound coverage dynam-
constraints. We thank B. Keyes, Y. C. Hsu, N. Stokes, and
ics in WT, S/A−/−, XIAP−/−, I. Shachrai for discussion, advice, and technical assistance;
and SX−/− mice at various S. Mazel, X. Li, S. Semova, and S. Tadesse for FACS
times PWI, as described sorting; the Comparative Biology Center (AAALAC-accredited)
in Fig. 3A (n = 8). (D) for health care to our mice; and members of the Steller lab.
Quantifications of WT, H.S. and E.F. are Investigators of the Howard Hughes Medical
S/A−/−, SX−/−, and SXDRing Institute. Supported by NIH grants RO1GM60124 (H.S.) and
R01-AR050452 (E.F.). The Sept4/ARTS and XIAP mouse mutant
CD34+CP3+ HFSCs. Scale strains used in this study are available from the Rockefeller
bars, 20 mm (A), 100 mm University to academic groups through a materials transfer
(B). P < 0.001. agreement and to for-profit groups through a license.
Figs. S1 to S10
21 November 2012; accepted 13 June 2013
Published online 20 June 2013;
10.1126/science.1233029

How the Red Queen Drives Terrestrial Mammals to Extinction
Tiago B. Quental and Charles R. Marshall
Science 341, 290 (2013);
DOI: 10.1126/science.1239431


Evolution
http://www.sciencemag.org/cgi/collection/evolution
REPORTS
How the Red Queen Drives Terrestrial pothesis that the waxing and waning in the clades’
diversity was random, the evolutionary equivalent
of “gamblers ruin” (7–9). If the rise and fall in
Mammals to Extinction diversity was deterministic, then we would expect
the longevities of the clades to be shorter than if
Tiago B. Quental1* and Charles R. Marshall2* their diversity trajectories were due to stochastic
fluctuations in intrinsically constant rates of orig-
Most species disappear by the processes of background extinction, yet those processes are poorly ination and extinction, where diversity would have
understood. We analyzed the evolutionary dynamics of 19 Cenozoic terrestrial mammalian clades simply drifted up and then down. Indeed, this is
with rich fossil records that are now fully extinct or in diversity decline. We find their diversity what we find: On average, the longevities of the
loss was not just a consequence of “gamblers ruin” but resulted from the evolutionary loss to the clades are too short to simply be the result of
Red Queen, a failure to keep pace with a deteriorating environment. Diversity loss is driven stochastic processes [figs. S3 and S4; see also
equally by both depressed origination rates and elevated extinction rates. Although we find (10)], suggesting a deterministic component to
diversity-dependent origination and extinction rates, the diversity of each clade only transiently the diversity dynamics.
equaled the implied equilibrium diversity. Thus, the processes that drove diversity loss in terrestrial Although the exact causes of the decline are
mammal clades were fundamentally nonequilibrial and overwhelmed diversity-dependent processes. hard to determine, we were able to characterize
the dynamics responsible for the diversity trajec-
he majority of all species that have ever sil records. To qualify for analysis, each family tories. Quantitative analysis revealed four gen-
T lived are now extinct (1), yet we know little had to be monophyletic and have at least 100 eralities. First, on average the duration of the rise

about the dynamics of extinction. Most prior genus occurrences (average = 419), a total di- phase is statistically indistinguishable from the
work has examined the mechanisms and selec- versity of at least five genera, a longevity of at duration of the decline phase—the diversity tra-
tivity of mass extinctions (2), although mass least eight stratigraphic stages, and an average jectories are temporally symmetrical [see (10) for
extinctions only account for a minority of ex- preservation potential of at least 0.6 per genus how we dealt with clades that are not extinct]. A
tinctions in the history of life. To examine the per stage (average = 0.89) (see supplementary ma- similar pattern has also been observed in Pa-
dynamics of background extinction, we analyzed terials). We only analyzed Cenozoic clades to avoid leozoic marine invertebrates (11).
the generic diversity trajectories of 19 Cenozoic any complicating factors that might have been Second, both the per-genus origination rates
terrestrial mammalian families with excellent fos- introduced by the end-Cretaceous mass extinc- and the extinction rates exhibit diversity depen-
1
tion event. dence [fig. S8 and table S2; see also (10)]: When
Universidade de São Paulo (USP), Departamento de Ecologia, We first tested the hypothesis that the ob- diversity increased, origination rates dropped and
São Paulo, SP, Brazil. 2University of California Museum of Pale-
ontology, 1101 Valley Life Sciences Building, University of Cali- served diversity trajectories were a consequence extinction rates increased. Thus, these mamma-
fornia Berkeley, Berkeley, CA 94720–4780, USA. of the loss to the Red Queen, that is, an evolu- lian clades exhibit the macroevolutionary equiv-
*Corresponding author. E-mail: tbquental@usp.br (T.B.Q.); tionary loss driven by the deterioration of the alent of MacArthur and Wilson’s (12) model for
crmarshall@berkeley.edu (C.R.M.) environment (3–6), against the alternative hy- diversity change during island colonization. In
Fig. 1. Changes in per-genus origination and

extinction rates for events of increasing and
decreasing diversity within the rise and de-
cline phases of mammalian diversity trajec-
tories. Increases in diversity during the rise phases
(A) are mostly controlled by increases in the per-
genus origination rate (the change in the orig-
ination rate is significantly larger than the change
in the extinction rate, N = 32, V (test statistic) =
438, P = 0.00071, Wilcoxon rank paired test). In
contrast, increases in the per-genus origination
rates and decreases in the per-genus extinction
rates contribute equally to increases in diversity
during the decline phases (B) (the magnitude
of these changes is not significantly different,
N = 11, V = 44, P = 0.3652, Wilcoxon rank paired
test). Decreases in the per-genus origination rates
and increases in the per-genus extinction rates also
contribute equally to decreases in diversity during
both rise (C) and decline (D) phases (the magni-
tude of these changes is not significantly differ-
ent, N = 9, V = 18, P = 0.6523, Wilcoxon rank
paired test, for the rise phases, and N = 42, V =
512, P = 0.4571, Wilcoxon rank paired test, for
the decline phases). Changes in rates and diver-
sity were measured between adjacent stages (fig.
S5). The changes in the per-genus origination and
extinction rates for each stage are connected by a
line, and each pair constituted a replicate in the
Wilcoxon rank paired test. Solid lines indicate that change in origination plots show the median and first and third quartiles of the data. The whiskers
rate is more important than change in extinction rate in driving diversity indicate the datum still within 1.5 interquartile range defined by the first and
change; dashed lines, that change in extinction is more important. The box- third quartiles. n.s., not significant.

REPORTS

Fig. 2. Origination rates decrease and extinction rates increase as the mammalian clades age.
There is a significant and roughly equal change in the average per-genus origination (A) and the
average per-genus extinction (B) rates between the diversity-rise and the diversity-decline phases across
the 19 families analyzed. Each pair of values corresponds to one of the analyzed families. For
origination, N = 19, V = 184, P = 0.000053, Wilcoxon rank paired test; for extinction, N = 19, V = 25,
P = 0.0033, Wilcoxon rank paired test. The crosses in the legend identify extinct clades. LMY, lineage
million years.
both scenarios, the existence of diversity-dependent (fig. S6). This disparity is due to the fact that
rates implies that each island (in MacArthur and changes in origination rate dominated the diver-
Wilson’s model) or clade (in the macroevolutionary sification phases of the diversity trajectories (Fig. 1),
equivalent of their model) has an equilibrium whereas changes in origination and extinction
diversity, the diversity at which the origination rates contributed equally during the decline phases.
rate equals the extinction rate. Diversity depen- Similarly, Gilinsky and Bambach (17) found that
dence in origination rates, but not in extinction family diversity within marine orders and sub-
rates, has also been reported in Cenozoic North orders was largely driven by changes (decreases)
American mammals [(13), but see (14)]. in family origination rates.
Third, we unexpectedly find that, during the The simplest way of modeling these observed
decline phase, decreases in the per-genus origi- diversity dynamics is with the macroevolution-
nation rate are just as important as increases in the ary equivalent of MacArthur and Wilson’s model
per-genus extinction rate in driving the observed (12). However, this model leads to logistic diver-
diversity losses (Fig. 1). In fact, on average the sification with a stable equilibrium diversity and
initial origination rate is of a similar magnitude to thus requires modification to accommodate diver-
the final extinction rate, and the final origination sity loss. Whereas Whittaker et al. (18) provides
rate is as low as the initial extinction rate (Fig. 2). a qualitative modification of the model that in-
Most discussions of clade extinction focus only corporates the formation and ultimate demise of
on the processes and rates of extinction and sel- oceanic islands with the extinction of their terres-
dom consider the possibility that diversity can trial biotas, we quantitatively extended MacArthur
also be lost because of a failure to replace extinct and Wilson’s framework to incorporate loss to
taxa. However, Bambach et al. (15) showed that the Red Queen by adding in a temporal decay Fig. 3. How a clade loses to the Red Queen via a
decay in its intrinsic per-genus rate of diver-
the loss in generic diversity in the end-Devonian in the intrinsic diversification rate, the diversi-
sification. (A to C) The change of the intrinsic orig-
and end-Triassic mass extinctions was primarily fication rate at the inception of the clade. We
ination and extinction rates (shown by the arrows); the
driven by a lack of origination. Similarly, Van Valen achieved this by decreasing the intrinsic origina- decay of the equilibrium diversity (shown by the mov-
(3) noted that the decline in generic diversity of tion rate and increasing the intrinsic extinction ing position of the dashed orange line); and the
perissodactyl mammals was largely due to a drop rate at constant rates with time (10), which trans- realized per-genus origination (blue points) and per-
in origination rate. The causes of a failure to lates into a constant rate of decay in the expected genus extinction (red points) rates at different times in
originate, the evolutionary sterility that we call equilibrium diversity (Fig. 3 and eq. S13). Thus, its history. (D) The diversity trajectory generated by
the Entwives effect (16), are not understood and under this model the expected equilibrium diver- the diversity dynamics shown in (A) to (C). Light blue
require more attention. sity steadily decays to zero and then becomes in- points show the diversity for the time points shown in
Last, on average the overall diversity trajec- creasingly negative, driving the clade to extinction. (A) to (C). The graphs depict solution to eqs. S10, S11,
tories were more influenced by changes in orig- We began with a slow rate of decay in the and S17 (10). The running Red Queen symbolizes the
ination rate than by changes in extinction rate intrinsic diversification rate (from 0.03% to 0.3% deterioration of the environment. Myr, million years.

REPORTS
per million years) but could not generate di- mine the clades’ fates: the deterioration of their 14. J. Alroy, in Speciation and Patterns of Diversity, R. Butlin,
versity trajectories with temporally symmetrical environment. Our results suggest that diversity J. Bridle, D. Schluter, Eds. (Cambridge Univ. Press,
Cambridge, 2009), pp. 301–323.
waxing and waning phases nor the switch in the dependence plays a role in diversity dynamics 15. R. K. Bambach, A. H. Knoll, S. Wang, Paleobiology 30,
magnitudes of the initial origination and extinc- similar to the role that friction plays in the dy- 522–524 (2004).
tion rates with their final rates (10). Instead, clades namics of motion—although it must be accounted 16. In J. R. R. Tolkein’s Middle Earth [ J. R. R. Tolkien,
quickly reached equilibrium diversity and then for in the dynamics of diversity change, the dom- The Lord of the Rings (Mariner Books, Boston, 2012)],
the Ents lost their wives and thus had no means of
slowly rode the decaying equilibrium diversity inant forces of diversity change lie beyond the regenerating their race, hence the term the Entwives effect.
down to extinction—the decline phase was longer existence of diversity dependence. 17. L. Gilinsky, R. K. Bambach, Paleobiology 13, 427–445
than the diversification phase, and the final orig- (1987).
ination and extinction rates remained at interme- 18. R. J. Whittaker, K. A. Triantis, R. J. Ladle, J. Biogeogr. 35,
References and Notes 977–994 (2008).
diate values between the high initial origination 1. D. M. Raup, Science 231, 1528–1533 (1986). 19. In Whittaker et al.’s model (18), their time axis is on a
rate and the low initial extinction rate, rather than 2. R. K. Bambach, Annu. Rev. Earth Planet. Sci. 34, log scale.
switching in value. 127–155 (2006). 20. J. J. Sepkoski Jr., Paleobiology 7, 36–53 (1984).
The only way to accommodate the observed 3. L. M. Van Valen, Evol. Theory 1, 1–30 (1973). 21. M. J. Benton, B. C. Emerson, Paleontology 50, 23–40
4. We used Van Valen’s original definition of the Red Queen (2007).
diversity dynamics is if the intrinsic diversifica- as a measure of environmental deterioration regardless 22. S. M. Stanley, Paleobiology 33 (suppl.), 1–55 (2007).
tion rate (and thus the equilibrium diversity) de- of the role that biotic and abiotic factors might have 23. S. M. Stanley, Paleobiology 34, 1–21 (2008).
teriorated sufficiently fast. For example, when we played in that deterioration (3). More recently, some
modeled the decay in the intrinsic diversification have restricted the meaning of the Red Queen to biotic Acknowledgments: We thank all those who generated the
factors (5, 6), using the term Court Jester for abiotic mammal data as well as those who entered the data into the
at a rate of ~3% per million years, the clade was factors (5, 6).

Paleobiology Database, especially J. Alroy, K. Behrensmeyer,
left with a standing diversity that increasingly 5. A. D. Barnosky, J. Vertebr. Paleontol. 21, 172–185 (2001). A. Turner, M. Uhen, and M. Carrano. This is the Paleobiology
lagged behind the equilibrium diversity as the 6. M. J. Benton, Science 323, 728–732 (2009). Database publication number 178. We thank S. Finnegan,
clade went extinct (Fig. 3C). This resulted in the 7. D. M. Raup, S. J. Gould, T. J. M. Schopf, D. S. Simberloff, H. Morlon, and S. P. Quek for discussion. T.B.Q. thanks
J. Geol. 81, 525–542 (1973). Fundação de Amparo à Pesquisa do Estado de São Paulo
switching in the values of the initial and final per- 8. The idea that the major patterns in Phanerozoic diversity (2012/04072-3) and USP for funding. All of the data are
genus origination and extinction rates and led to a change could be attributed to purely stochastic process available from the Paleobiology Database (http://paleodb.org).
sufficiently negative diversification rate during the was later rejected (9).
9. S. M. Stanley, P. W. Signor III, S. Lidgard, A. F. Karr,
clade’s decline to produce the temporally symmet- www.sciencemag.org/cgi/content/full/science.1239431/DC1
Paleobiology 7, 115–127 (1981).
ric waxing and waning phases of diversity change. Materials and Methods
10. Material and methods and supplementary materials can
An unexpected consequence of the rapid de- be found on Science Online.
Figs. S1 to S8
cline in the per-genus rate of diversification is Tables S1 and S2
11. M. Foote, Paleobiology 33, 517–529 (2007).
that a clade’s diversity only transiently equals 12. R. H. MacArthur, E. O. Wilson, The Theory of Island
the equilibrium diversity. In contrast, in typical Biogeography (Princeton Univ. Press, Princeton, NJ, 1967). 22 April 2013; accepted 5 June 2013
13. J. Alroy, Palaeogeogr. Palaeoclimatol. Palaeoecol. 127, Published online 20 June 2013;
diversity-dependent models, species diversity 285–311 (1996). 10.1126/science.1239431
remains at or close to the equilibrium diversity
after the initial radiation, even when the equilib-
Exceptional Convergence on the

rium diversity decays with time, for example, as
in Whittaker et al.’s modeling of the disappear-
ance of islands through erosion (18, 19). Under
our model, the diversification phase involves a
gain toward an equilibrium diversity, as in stan-
Macroevolutionary Landscape in
dard logistic growth. However, as diversity in-
creases, the equilibrium diversity is decaying in Island Lizard Radiations
response to an already deteriorating environment,
and the clade reaches its peak diversity at an equi- D. Luke Mahler,1* Travis Ingram,2 Liam J. Revell,3 Jonathan B. Losos2
librium value less than the initial equilibrium di-
versity. Then, as the clade moves into the decline G. G. Simpson, one of the chief architects of evolutionary biology’s modern synthesis, proposed
phase, the decay in its intrinsic rate of diversifi- that diversification occurs on a macroevolutionary adaptive landscape, but landscape models
cation leads to a sufficiently rapid decrease in its are seldom used to study adaptive divergence in large radiations. We show that for Caribbean
equilibrium diversity that the clade’s realized di- Anolis lizards, diversification on similar Simpsonian landscapes leads to striking convergence of
versity increasingly lags behind the decaying equi- entire faunas on four islands. Parallel radiations unfolding at large temporal scales shed light
librium diversity (Fig. 3). Thus, although diversity on the process of adaptive diversification, indicating that the adaptive landscape may give rise
dependence in the per-genus origination and ex- to predictable evolutionary patterns in nature, that adaptive peaks may be stable over
tinction rates plays a role in determining the dura- macroevolutionary time, and that available geographic area influences the ability of lineages
tion of the clade’s history, the diversity dynamics to discover new adaptive peaks.
is dominated by the decay in the intrinsic diver-
sification rates, not by the diversity-dependent he concept of a macroevolutionary adapt- 1944 (1–4). Although influential as a metaphor,
equilibrium processes.
The secondary role that diversity-dependent
rates of origination and extinction play in the di-
T ive landscape—a multivariate phenotype
surface on which species evolve up local
adaptive peaks—has guided thinking about adapt-
Simpson’s landscape has only rarely been ap-
plied to study large adaptive radiations in nature.
Moreover, when applied the macroevolutionary
versity dynamics of the mammalian clades in ive radiation since G. G. Simpson proposed it in landscape generally has been invoked to describe
decline offers a resolution to a debate in the 1
evolutionary dynamics within a single lineage
paleontological literature, where diversity depen- Center for Population Biology, University of California in a particular ecological setting. In recent years,
Davis, Davis, CA 95616, USA. 2Department of Organismic
dence has been proposed (13, 20) but where the and Evolutionary Biology, Harvard University, Cambridge, however, a number of studies have suggested that
evidence of equilibrium is scarce (21–23). In our MA 02138, USA. 3Department of Biology, University of entire evolutionary radiations can exhibit pheno-
model, the mechanism of diversity dependence is Massachusetts Boston, Boston, MA 02125, USA. typic convergence when they diversify in similar
decoupled from the ultimate factors that deter- *Correspondence to: lmahler@ucdavis.edu environments (5–10). To the extent that such

Exceptional Convergence on the Macroevolutionary Landscape in
Island Lizard Radiations
D. Luke Mahler et al.
Science 341, 292 (2013);
DOI: 10.1126/science.1232392


REPORTS
per million years) but could not generate di- mine the clades’ fates: the deterioration of their 14. J. Alroy, in Speciation and Patterns of Diversity, R. Butlin,
versity trajectories with temporally symmetrical environment. Our results suggest that diversity J. Bridle, D. Schluter, Eds. (Cambridge Univ. Press,
Cambridge, 2009), pp. 301–323.
waxing and waning phases nor the switch in the dependence plays a role in diversity dynamics 15. R. K. Bambach, A. H. Knoll, S. Wang, Paleobiology 30,
magnitudes of the initial origination and extinc- similar to the role that friction plays in the dy- 522–524 (2004).
tion rates with their final rates (10). Instead, clades namics of motion—although it must be accounted 16. In J. R. R. Tolkein’s Middle Earth [ J. R. R. Tolkien,
quickly reached equilibrium diversity and then for in the dynamics of diversity change, the dom- The Lord of the Rings (Mariner Books, Boston, 2012)],
the Ents lost their wives and thus had no means of
slowly rode the decaying equilibrium diversity inant forces of diversity change lie beyond the regenerating their race, hence the term the Entwives effect.
down to extinction—the decline phase was longer existence of diversity dependence. 17. L. Gilinsky, R. K. Bambach, Paleobiology 13, 427–445
than the diversification phase, and the final orig- (1987).
ination and extinction rates remained at interme- 18. R. J. Whittaker, K. A. Triantis, R. J. Ladle, J. Biogeogr. 35,
References and Notes 977–994 (2008).
diate values between the high initial origination 1. D. M. Raup, Science 231, 1528–1533 (1986). 19. In Whittaker et al.’s model (18), their time axis is on a
rate and the low initial extinction rate, rather than 2. R. K. Bambach, Annu. Rev. Earth Planet. Sci. 34, log scale.
switching in value. 127–155 (2006). 20. J. J. Sepkoski Jr., Paleobiology 7, 36–53 (1984).
The only way to accommodate the observed 3. L. M. Van Valen, Evol. Theory 1, 1–30 (1973). 21. M. J. Benton, B. C. Emerson, Paleontology 50, 23–40
4. We used Van Valen’s original definition of the Red Queen (2007).
diversity dynamics is if the intrinsic diversifica- as a measure of environmental deterioration regardless 22. S. M. Stanley, Paleobiology 33 (suppl.), 1–55 (2007).
tion rate (and thus the equilibrium diversity) de- of the role that biotic and abiotic factors might have 23. S. M. Stanley, Paleobiology 34, 1–21 (2008).
teriorated sufficiently fast. For example, when we played in that deterioration (3). More recently, some
modeled the decay in the intrinsic diversification have restricted the meaning of the Red Queen to biotic Acknowledgments: We thank all those who generated the
factors (5, 6), using the term Court Jester for abiotic mammal data as well as those who entered the data into the
at a rate of ~3% per million years, the clade was factors (5, 6).

Paleobiology Database, especially J. Alroy, K. Behrensmeyer,
left with a standing diversity that increasingly 5. A. D. Barnosky, J. Vertebr. Paleontol. 21, 172–185 (2001). A. Turner, M. Uhen, and M. Carrano. This is the Paleobiology
lagged behind the equilibrium diversity as the 6. M. J. Benton, Science 323, 728–732 (2009). Database publication number 178. We thank S. Finnegan,
clade went extinct (Fig. 3C). This resulted in the 7. D. M. Raup, S. J. Gould, T. J. M. Schopf, D. S. Simberloff, H. Morlon, and S. P. Quek for discussion. T.B.Q. thanks
J. Geol. 81, 525–542 (1973). Fundação de Amparo à Pesquisa do Estado de São Paulo
switching in the values of the initial and final per- 8. The idea that the major patterns in Phanerozoic diversity (2012/04072-3) and USP for funding. All of the data are
genus origination and extinction rates and led to a change could be attributed to purely stochastic process available from the Paleobiology Database (http://paleodb.org).
sufficiently negative diversification rate during the was later rejected (9).
9. S. M. Stanley, P. W. Signor III, S. Lidgard, A. F. Karr,
clade’s decline to produce the temporally symmet- www.sciencemag.org/cgi/content/full/science.1239431/DC1
Paleobiology 7, 115–127 (1981).
ric waxing and waning phases of diversity change. Materials and Methods
10. Material and methods and supplementary materials can
An unexpected consequence of the rapid de- be found on Science Online.
Figs. S1 to S8
cline in the per-genus rate of diversification is Tables S1 and S2
11. M. Foote, Paleobiology 33, 517–529 (2007).
that a clade’s diversity only transiently equals 12. R. H. MacArthur, E. O. Wilson, The Theory of Island
the equilibrium diversity. In contrast, in typical Biogeography (Princeton Univ. Press, Princeton, NJ, 1967). 22 April 2013; accepted 5 June 2013
13. J. Alroy, Palaeogeogr. Palaeoclimatol. Palaeoecol. 127, Published online 20 June 2013;
diversity-dependent models, species diversity 285–311 (1996). 10.1126/science.1239431
remains at or close to the equilibrium diversity
after the initial radiation, even when the equilib-
Exceptional Convergence on the

rium diversity decays with time, for example, as
in Whittaker et al.’s modeling of the disappear-
ance of islands through erosion (18, 19). Under
our model, the diversification phase involves a
gain toward an equilibrium diversity, as in stan-
Macroevolutionary Landscape in
dard logistic growth. However, as diversity in-
creases, the equilibrium diversity is decaying in Island Lizard Radiations
response to an already deteriorating environment,
and the clade reaches its peak diversity at an equi- D. Luke Mahler,1* Travis Ingram,2 Liam J. Revell,3 Jonathan B. Losos2
librium value less than the initial equilibrium di-
versity. Then, as the clade moves into the decline G. G. Simpson, one of the chief architects of evolutionary biology’s modern synthesis, proposed
phase, the decay in its intrinsic rate of diversifi- that diversification occurs on a macroevolutionary adaptive landscape, but landscape models
cation leads to a sufficiently rapid decrease in its are seldom used to study adaptive divergence in large radiations. We show that for Caribbean
equilibrium diversity that the clade’s realized di- Anolis lizards, diversification on similar Simpsonian landscapes leads to striking convergence of
versity increasingly lags behind the decaying equi- entire faunas on four islands. Parallel radiations unfolding at large temporal scales shed light
librium diversity (Fig. 3). Thus, although diversity on the process of adaptive diversification, indicating that the adaptive landscape may give rise
dependence in the per-genus origination and ex- to predictable evolutionary patterns in nature, that adaptive peaks may be stable over
tinction rates plays a role in determining the dura- macroevolutionary time, and that available geographic area influences the ability of lineages
tion of the clade’s history, the diversity dynamics to discover new adaptive peaks.
is dominated by the decay in the intrinsic diver-
sification rates, not by the diversity-dependent he concept of a macroevolutionary adapt- 1944 (1–4). Although influential as a metaphor,
equilibrium processes.
The secondary role that diversity-dependent
rates of origination and extinction play in the di-
T ive landscape—a multivariate phenotype
surface on which species evolve up local
adaptive peaks—has guided thinking about adapt-
Simpson’s landscape has only rarely been ap-
plied to study large adaptive radiations in nature.
Moreover, when applied the macroevolutionary
versity dynamics of the mammalian clades in ive radiation since G. G. Simpson proposed it in landscape generally has been invoked to describe
decline offers a resolution to a debate in the 1
evolutionary dynamics within a single lineage
paleontological literature, where diversity depen- Center for Population Biology, University of California in a particular ecological setting. In recent years,
Davis, Davis, CA 95616, USA. 2Department of Organismic
dence has been proposed (13, 20) but where the and Evolutionary Biology, Harvard University, Cambridge, however, a number of studies have suggested that
evidence of equilibrium is scarce (21–23). In our MA 02138, USA. 3Department of Biology, University of entire evolutionary radiations can exhibit pheno-
model, the mechanism of diversity dependence is Massachusetts Boston, Boston, MA 02125, USA. typic convergence when they diversify in similar
decoupled from the ultimate factors that deter- *Correspondence to: lmahler@ucdavis.edu environments (5–10). To the extent that such

REPORTS
“replicated adaptive radiations” exist (11), they sug- instances of convergent evolution (16). However, island in morphospace, and assessed whether the
gest not only that diversification over macro- these analyses omitted an important dimension of average among-island distance for anoles was
evolutionary time scales may be surprisingly the anole radiations: so-called “unique” species lower than expected by chance via comparison to
deterministic but also that ecological factors can that evolved a morphology and ecology not a phylogenetic null distribution (18). We simu-
give rise to highly similar adaptive landscapes in found on other islands (14, 15). The number of lated null morphospaces using the empirical max-
geographically distinct regions (1, 2, 12). unique species varies among islands, from none imum clade credibility (MCC) phylogeny from
Laboratory investigations of microbial diver- on Puerto Rico to 14 on Cuba, and they collec- a Bayesian analysis of mitochondrial DNA and
gence (13) and computer simulation studies (1) tively constitute approximately 20% of Greater evolutionary models that account for temporal and
both suggest that stable adaptive landscapes can Antillean Anolis diversity. Thus, although repeated among-trait evolutionary rate variation (17, 19). We
generate predictable evolutionary patterns, but test- evolution of the ecomorphs has been demonstrated, repeated the analysis across a Bayesian posterior
ing this hypothesis in naturally evolving radiations whether the anole radiations themselves are con- tree sample, obtaining qualitatively identical results
requires the comparison of adaptive landscapes vergent remains an open question. (18). These analyses provide strong evidence for
over macroevolutionary time scales. Investigations Thus, we began by asking whether the four exceptional species-for-species matching among
of this type require a two-step process in which we Greater Antillean anole faunas exhibit exception- island anole faunas: Most species on each island
first test the hypothesis that radiations are more al species-for-species matching (2); i.e., greater are more morphologically similar to species from
convergent than expected by chance (11), and then pairwise phenotypic similarity between species other islands than expected by chance (P = 0.003;
we ask if this convergence has resulted from diver- on different islands than expected by chance, when Fig. 1A).
sification on similar macroevolutionary landscapes. the full ecomorphological diversity of lineages is Next, we investigated whether a macroevolu-
We used this approach to study the putative considered. We compiled phylogenetic and phe- tionary model involving convergent shifts to com-

replicated adaptive radiations (2, 14) of Anolis notypic data for 100 of 119 Greater Antillean mon peaks on a Simpsonian landscape can explain
lizards (anoles) on Caribbean islands. On each Anolis species, representing the diversity of both the evolution of remarkably similar faunas on
island in the Greater Antilles (Cuba, Hispaniola, the ecomorph and unique species. We tested for these four Caribbean islands. We applied a new
Puerto Rico, and Jamaica), anoles have indepen- species-for-species matching among islands using method (SURFACE) to infer the history of adapt-
dently evolved a similar set of habitat specialists a phylogenetic comparative analysis of species ive diversification in anoles using a phylogeny
termed “ecomorphs” (such as “twig” or “grass- similarity in a four-dimensional principal com- and phenotypic data. This method fits a model of
bush”) (14, 15). Each ecomorph is composed of ponents morphospace generated from 11 traits adaptive radiation in which lineages may undergo
morphologically and behaviorally similar species important for niche partitioning in Anolis, includ- shifts to adaptive peaks on a macroevolutionary
that occupy similar microhabitats. Prior studies ing body size, limb and tail lengths, and adhesive landscape without reference to a priori hypotheses
have documented that members of the same toepad lamella number (14, 17). We measured the specifying which lineages correspond to particular
ecomorph category from different islands cluster among-island Euclidean distance between each peaks (18, 20). In this model, lineages may
morphologically, providing evidence for repeated species and its nearest neighbor from each other undergo shifts to otherwise unoccupied peaks or
to those shared with other lineages, which gives us
the ability to explicitly model the macroevo-
A lutionary convergence of independent lineages in
a common phenotype space. Starting with an
150
P = 0.003
Ornstein-Uhlenbeck model (21, 22) in which all
Frequency
species are attracted to a single adaptive peak in

100
trait space, SURFACE uses a stepwise model se-

lection procedure based on the finite-samples
50
Akaike information criterion (AICc) (23, 24) to

fit increasingly complex multipeak models. At
0
0.2 0.3 0.4 0.5 0.6 0.7 0.8

each step, a new peak shift is added to the branch
Among-island Euclidean distance of the phylogeny that most improves model fit
across all traits, and shifts are added until none
result in further improvement. To identify con-
B vergence, the method then evaluates whether the
8 10
AICc score is further improved by permitting

P = 0.01
Frequency
these independent lineages to shift toward shared

adaptive peaks rather than requiring each to oc-
6
cupy a unique peak. We compared the conver-

4
gent landscape model to several alternatives,

2
including three variants of the “early burst” mod-

0
el of adaptive radiation, which features a diversity-

0 5 10 15 20 25 30 dependent decline in evolutionary rate but does
Number of convergent shifts not explicitly model adaptation or convergence
Fig. 1. Morphological similarity among Greater Antillean Anolis faunas is due to exceptional on a macroevolutionary landscape (17, 19).
convergence. (A) Anole species are more similar to their closest matches from other islands than A Simpsonian model of peak shifts on a shared
expected by chance. The vertical line indicates the average distance separating each Greater Antillean macroevolutionary adaptive landscape best explains
species from the most morphologically similar species on other islands (among-island Euclidean distance); the evolution of ecomorphological traits in Greater
the histogram depicts a null distribution of the same scores calculated from 999 data sets generated by Antillean Anolis. Convergence of lineages to shared
evolutionary simulation. (B) The great inter-island similarity (low distance) results from an exceptionally adaptive peaks was the dominant mode of macro-
large number of convergent shifts to shared peaks on a macroevolutionary adaptive landscape during evolutionary trait change for these replicated island
anole diversification. Here, the vertical line depicts the number of convergent peak shifts detected for radiations: A landscape model with explicit con-
Greater Antillean anoles, and the histogram depicts a null distribution of convergent shifts detected from vergence was strongly favored over the best non-
99 simulated data sets. convergent landscape model (DAICc = 162.6;

REPORTS
fig. S2), and even more strongly favored over remains to be seen whether this stability is a general tably, we found evidence for a number of non-
models without peaks (DAICc ≥ 279.3; fig. S2). feature of adaptive radiation, but our observations convergent shifts to island-specific peaks. However,
Peak shifts on the adaptive landscape were com- support the hypothesis that stable macroevolu- such peaks are occupied only on the larger islands
mon as anoles diversified (29 lineage-specific tionary landscapes underlie the common pattern of Hispaniola and Cuba, a predicted “area effect”
shifts), with 76% of shifts involving convergence of long-term phenotypic stasis in both fossil and from adaptive radiation theory (1). Larger areas
in morphospace (Figs. 1B and 2 and Table 1). comparative data sets (28, 29). provide greater opportunities for diversification
The estimated landscape contained 15 adaptive Although convergence is the most conspicu- (1, 30), which may permit lineages to more fully
peaks, with 8 occupied by more than one lineage. ous feature of Greater Antillean anole radiations, explore the macroevolutionary landscape and dis-
These convergent peaks attracted 2.8 lineages on the island faunas are far from identical. Most no- cover peaks not reached on smaller islands (1).
average, and all but one hosted lineages from
multiple islands. Overall, the number of conver-
gent adaptive peak shifts was significantly greater
than expected by chance (P = 0.01; Fig. 1B), and
these shifts account for the exceptional similarity
among island faunas (18). The number and po-
sition of peak shifts varied across 100 phylogenies,
but the number of convergent shifts was similar
for all trees (Table 1). Species traditionally grouped
in the same ecomorph class (14–16) tended to be

attracted toward the same adaptive peak (fig. S4).
Our comparison of macroevolutionary models
suggests that the adaptive landscape plays an im-
portant role in shaping parallel diversification. The
only model to account for the observed convergence
of entire island anole faunas was a Simpsonian
model (3, 5, 20–22), in which lineages experience
selection toward common peaks on the adaptive
landscape (fig. S1). Fitted peaks on the anole
landscape correspond to trait combinations that
have been shown experimentally to be adaptive
for microhabitat partitioning (14) (fig. S4). Al-
though it is possible that evolutionary constraints
may play a role in shaping whole-fauna conver-
gence, in the case of anoles the evidence points to
a dominant role for selection. The Anolis radiation
unfolded over tens of millions of years (14), a time
scale over which constraints on the production of
variation are unlikely to be maintained, especially
for quantitative traits (25). Constraint seems an even
less likely culprit considering that diverse radia-
tions of Central and South American Anolis, which
occur in ecologically different communities, ex-
hibit many morphologies not seen in Caribbean
forms (26), strongly suggesting that repeated Greater
Antillean convergence is not due to intrinsic limits
on morphological variation.
Replication of adaptive radiations is readily
attainable in simple systems over short time scales
(2, 13), but convincing examples at a grander mac-
roevolutionary scale have so far been lacking. Fig. 2. Phenotypic convergence on the macroevolutionary adaptive landscape in island ra-
Why this is the case is not yet clear, but our results diations of Greater Antillean Anolis. MCC phylogeny (left panel), painted to depict the estimated
argue against the possibility that the temporal phylogenetic history of adaptive peak shifts. Branches and island silhouettes representing geographic
lability of adaptive landscapes precludes faunal location are colored according to adaptive peak. Convergent peaks are colored and are connected by solid
convergence over long time scales. The Greater lines; with one exception, these peaks attract lineages from multiple islands. Nonconvergent peaks are
light gray (single-island radiations), or black or dark gray (radiations that spread to multiple islands; these
Antillean anole radiation is old (with a minimum
islands are connected by dashed lines). Right panels show that anole species (small circles) cluster near
age of 30 to 40 million years) (14), and most eco-
their inferred adaptive peak (large circles) in morphospace, especially for pPCs 1 to 3 (primary trait
morphological diversity among anoles arose long correlations and variance explained are reported in the axis labels). Inferred convergent peaks broadly
ago (17) (Fig. 2), suggesting not only that the correspond to Anolis ecomorph classes, a consistent finding across phylogenies, although specific peak
adaptive landscape for anoles is similar across is- assignments vary from tree to tree (18) (fig. S4). In the MCC estimate plotted here, grass-bush specialists
lands but also that it is relatively static in its prin- are attracted to dark or light yellow peaks. Larger and smaller twig specialists occur on purple and red
cipal features (14, 27). Stable landscapes are further peaks, respectively. Green peaks contain trunk-crown specialists, and brown peaks contain trunk-ground
indicated by the similarity of adaptive peaks dis- specialists. Dark blue peaks contain large crown-giant anoles; light blue peaks contain smaller giant
covered by anole lineages from different islands, anoles. A single dark gray peak contains additional trunk-ground and trunk-crown species; although
which would be highly unlikely if adaptive land- similar because of inherited ancestral condition rather than convergence in this reconstruction phe-
scapes were themselves highly labile (3, 27). It notypically similar species on this peak occur on three different islands.

REPORTS
Table 1. SURFACE convergence parameters estimated using the MCC tree, as well as 100 trees 19. L. J. Harmon et al., Evolution 64, 2385–2396 (2010).
from the Bayesian posterior probability distribution of Anolis phylogeny. 20. T. Ingram, D. L. Mahler, Methods Ecol. Evol. 4, 416–425
(2013).
21. M. A. Butler, A. A. King, Am. Nat. 164, 683–695 (2004).
Mean (SD) 22. T. F. Hansen, Evolution 51, 1341 (1997).
MCC
for 100 sampled 23. M. E. Alfaro et al., Proc. Natl. Acad. Sci. U.S.A. 106,
phylogeny
phylogenies 13410–13414 (2009).
24. G. H. Thomas, R. P. Freckleton, Methods Ecol. Evol 3,
Adaptive peak shifts 29 25.7 (2.1) 145–151 (2012).
Convergent adaptive peak shifts 22 20.2 (1.9) 25. G. L. Conte, M. E. Arnegard, C. L. Peichel, D. Schluter,
Adaptive peaks 15 12.7 (1.4) Proc. Biol. Sci. 279, 5039–5047 (2012).
Convergent adaptive peaks 8 7.2 (0.97) 26. G. Pinto, D. L. Mahler, L. J. Harmon, J. B. Losos, Proc.
Biol. Sci. 275, 2749–2757 (2008).
Convergence fraction (convergent peak shifts/total peak shifts) 0.76 0.79 (0.045) 27. T. F. Hansen, in The Adaptive Landscape in Evolutionary
Average number of lineages converging to each shared adaptive peak 2.8 2.8 (0.30) Biology, E. Svensson, R. Calsbeek, Eds. (Oxford Univ.
Fraction of convergent peaks with lineages from multiple islands 0.88 0.94 (0.074) Press, Oxford, 2012), pp. 205–226.
28. S. Estes, S. J. Arnold, Am. Nat. 169, 227–244 (2007).
29. J. C. Uyeda, T. F. Hansen, S. J. Arnold, J. Pienaar,
Proc. Natl. Acad. Sci. U.S.A. 108, 15908–15913
Anoles have diversified into many more species 10. K. A. Young, J. Snoeks, O. Seehausen, PLoS ONE 4, (2011).
on the larger islands (14, 31). Given the number of e4740 (2009). 30. Y. Kisel, T. G. Barraclough, Am. Nat. 175, 316–334 (2010).
11. Past studies of convergence between species-rich radiations 31. D. L. Rabosky, R. E. Glor, Proc. Natl. Acad. Sci. U.S.A.
convergent and unique adaptive peak shifts that have been selective in scope and have not compared 107, 22178–22183 (2010).
have occurred across the Greater Antilles in entire radiations. Even lineages diversifying randomly under 32. S. J. Gould, Wonderful Life: The Burgess Shale and

anoles, the area effect hypothesis predicts that all genetic drift will generate many convergent pairs (34), as the Nature of History (Norton, New York, 1989).
endemic unique peaks should occur on Cuba and will radiations adapting on a labile macroevolutionary 33. C. E. Wagner, L. J. Harmon, O. Seehausen, Nature 487,
landscape (27). Much of the diversity unsampled by past 366–369 (2012).
Hispaniola, as observed (18) (fig. S3). Thus, in studies may be nonconvergent, so that different radiations 34. C. T. Stayton, J. Theor. Biol. 252, 1–14 (2008).
this system, even the apparently contingent evolu- may share many convergent species pairs while being
tion of unique ecomorphologies may be, to some unexceptionally similar at the whole-radiation scale. Acknowledgments: We thank G. Bradburd, C. Davis, L. Harmon,
extent, predictable—in this case a result of the 12. M. L. Cody, H. A. Mooney, Annu. Rev. Ecol. Syst. 9, and F. Jenkins for discussion and advice; the National Evolutionary
265–321 (1978). Synthesis Center and NSF for financial support; and three
speciation-area relationship (30). 13. R. Kassen, Ann. N. Y. Acad. Sci. 1168, 3–22 (2009). anonymous reviewers for insightful feedback. Data are archived in
Gould famously argued that evolution over long 14. J. B. Losos, Lizards in an Evolutionary Tree: Ecology and Dryad (http://datadryad.org, doi: 10.5061/dryad.9g182).
time scales is “utterly unpredictable and quite un- Adaptive Radiation of Anoles (Univ. of California Press,
Berkeley, CA, 2009).
repeatable” (32, p. 14) due to historical contingency. 15. E. E. Williams, in Lizard Ecology: Studies of a Model Organism,
Widespread convergence among entire faunas of www.sciencemag.org/cgi/content/full/341/6143/292/DC1
R. B. Huey, E. R. Pianka, T. W. Schoener, Eds. (Harvard Univ.
Greater Antillean Anolis refutes Gould’s claim and Press, Cambridge, MA, 1983), pp. 326–370.
Figs. S1 to S9
shows that adaptation can overcome the influence 16. J. B. Losos, T. R. Jackman, A. Larson, K. de Queiroz,
Tables S1 to S5
L. Rodríguez-Schettino, Science 279, 2115–2118 (1998).
of chance events on the course of evolution. Our References (35–42)
17. D. L. Mahler, L. J. Revell, R. E. Glor, J. B. Losos, Evolution
demonstration of deterministic convergence on a ma- 64, 2731–2745 (2010).
Author Contributions
croevolutionary adaptive landscape complements 18. Information on materials and methods is available as 5 November 2012; accepted 5 June 2013
studies of diversification in species numbers in supplementary material on Science Online. 10.1126/science.1232392
showing that many features of large-scale radiations
may be surprisingly predictable. A recent analysis
discovered that both island diversification rate and
standing species richness in Greater Antillean anoles Predicting and Manipulating Cardiac
Drug Inactivation by the Human Gut
could be predicted from island size and time since
colonization (31). In cichlids, whether colonizing
lineages will radiate in African lakes can be predicted
from the intrinsic traits of the colonist and the eco-
logical opportunities provided by the new habitat Bacterium Eggerthella lenta
(33). Together, these studies suggest that the pri-
mary aspects of evolutionary radiation—adaptation Henry J. Haiser,1 David B. Gootenberg,1 Kelly Chatman,1 Gopal Sirasani,2
and the proliferation of species—may in some Emily P. Balskus,2 Peter J. Turnbaugh1*
cases be largely deterministic.
Despite numerous examples of the effects of the human gastrointestinal microbiome on drug
efficacy and toxicity, there is often an incomplete understanding of the underlying mechanisms.
1. S. Gavrilets, A. Vose, Proc. Natl. Acad. Sci. U.S.A. 102, Here, we dissect the inactivation of the cardiac drug digoxin by the gut Actinobacterium
18040–18045 (2005). Eggerthella lenta. Transcriptional profiling, comparative genomics, and culture-based assays
2. D. Schluter, The Ecology of Adaptive Radiation (Oxford revealed a cytochrome-encoding operon up-regulated by digoxin, inhibited by arginine, absent
Univ. Press, Oxford, 2000). in nonmetabolizing E. lenta strains, and predictive of digoxin inactivation by the human gut
3. G. G. Simpson, Tempo and Mode in Evolution (Columbia
Univ. Press, New York, 1944). microbiome. Pharmacokinetic studies using gnotobiotic mice revealed that dietary protein reduces
4. E. I. Svensson, R. Calsbeek, Eds., The Adaptive Landscape the in vivo microbial metabolism of digoxin, with significant changes to drug concentration in
in Evolutionary Biology (Oxford Univ. Press, Oxford, the serum and urine. These results emphasize the importance of viewing pharmacology from the
2012). perspective of both our human and microbial genomes.
5. B. Frédérich, L. Sorenson, F. Santini, G. J. Slater,
M. E. Alfaro, Am. Nat. 181, 94–113 (2013).
umans are home to large and diverse mi- the human gut microbiome, including alterations
H
6. R. Gillespie, Science 303, 356–359 (2004).
7. T. J. Givnish, in The Biology of Biodiversity, M. Kato, Ed. crobial communities, the most abundant to the bioavailability, activity, and toxicity of ther-
(Springer-Verlag, Tokyo, 1999), pp. 67–90.
8. M. Muschick, A. Indermaur, W. Salzburger, Curr. Biol. 22,
of which resides in the gastrointestinal apeutic drugs (1, 2). Although >40 drugs are
2362–2368 (2012). tract. Recent studies have highlighted the clinical metabolized by the gut microbiome, little is
9. M. L. J. Stiassny, A. Meyer, Sci. Am. 280, 64–69 (1999). relevance of the biotransformations catalyzed by known about the underlying mechanisms. This

Predicting and Manipulating Cardiac Drug Inactivation by the Human
Gut Bacterium Eggerthella lenta
Henry J. Haiser et al.
Science 341, 295 (2013);
DOI: 10.1126/science.1235872


REPORTS
Table 1. SURFACE convergence parameters estimated using the MCC tree, as well as 100 trees 19. L. J. Harmon et al., Evolution 64, 2385–2396 (2010).
from the Bayesian posterior probability distribution of Anolis phylogeny. 20. T. Ingram, D. L. Mahler, Methods Ecol. Evol. 4, 416–425
(2013).
21. M. A. Butler, A. A. King, Am. Nat. 164, 683–695 (2004).
Mean (SD) 22. T. F. Hansen, Evolution 51, 1341 (1997).
MCC
for 100 sampled 23. M. E. Alfaro et al., Proc. Natl. Acad. Sci. U.S.A. 106,
phylogeny
phylogenies 13410–13414 (2009).
24. G. H. Thomas, R. P. Freckleton, Methods Ecol. Evol 3,
Adaptive peak shifts 29 25.7 (2.1) 145–151 (2012).
Convergent adaptive peak shifts 22 20.2 (1.9) 25. G. L. Conte, M. E. Arnegard, C. L. Peichel, D. Schluter,
Adaptive peaks 15 12.7 (1.4) Proc. Biol. Sci. 279, 5039–5047 (2012).
Convergent adaptive peaks 8 7.2 (0.97) 26. G. Pinto, D. L. Mahler, L. J. Harmon, J. B. Losos, Proc.
Biol. Sci. 275, 2749–2757 (2008).
Convergence fraction (convergent peak shifts/total peak shifts) 0.76 0.79 (0.045) 27. T. F. Hansen, in The Adaptive Landscape in Evolutionary
Average number of lineages converging to each shared adaptive peak 2.8 2.8 (0.30) Biology, E. Svensson, R. Calsbeek, Eds. (Oxford Univ.
Fraction of convergent peaks with lineages from multiple islands 0.88 0.94 (0.074) Press, Oxford, 2012), pp. 205–226.
28. S. Estes, S. J. Arnold, Am. Nat. 169, 227–244 (2007).
29. J. C. Uyeda, T. F. Hansen, S. J. Arnold, J. Pienaar,
Proc. Natl. Acad. Sci. U.S.A. 108, 15908–15913
Anoles have diversified into many more species 10. K. A. Young, J. Snoeks, O. Seehausen, PLoS ONE 4, (2011).
on the larger islands (14, 31). Given the number of e4740 (2009). 30. Y. Kisel, T. G. Barraclough, Am. Nat. 175, 316–334 (2010).
11. Past studies of convergence between species-rich radiations 31. D. L. Rabosky, R. E. Glor, Proc. Natl. Acad. Sci. U.S.A.
convergent and unique adaptive peak shifts that have been selective in scope and have not compared 107, 22178–22183 (2010).
have occurred across the Greater Antilles in entire radiations. Even lineages diversifying randomly under 32. S. J. Gould, Wonderful Life: The Burgess Shale and

anoles, the area effect hypothesis predicts that all genetic drift will generate many convergent pairs (34), as the Nature of History (Norton, New York, 1989).
endemic unique peaks should occur on Cuba and will radiations adapting on a labile macroevolutionary 33. C. E. Wagner, L. J. Harmon, O. Seehausen, Nature 487,
landscape (27). Much of the diversity unsampled by past 366–369 (2012).
Hispaniola, as observed (18) (fig. S3). Thus, in studies may be nonconvergent, so that different radiations 34. C. T. Stayton, J. Theor. Biol. 252, 1–14 (2008).
this system, even the apparently contingent evolu- may share many convergent species pairs while being
tion of unique ecomorphologies may be, to some unexceptionally similar at the whole-radiation scale. Acknowledgments: We thank G. Bradburd, C. Davis, L. Harmon,
extent, predictable—in this case a result of the 12. M. L. Cody, H. A. Mooney, Annu. Rev. Ecol. Syst. 9, and F. Jenkins for discussion and advice; the National Evolutionary
265–321 (1978). Synthesis Center and NSF for financial support; and three
speciation-area relationship (30). 13. R. Kassen, Ann. N. Y. Acad. Sci. 1168, 3–22 (2009). anonymous reviewers for insightful feedback. Data are archived in
Gould famously argued that evolution over long 14. J. B. Losos, Lizards in an Evolutionary Tree: Ecology and Dryad (http://datadryad.org, doi: 10.5061/dryad.9g182).
time scales is “utterly unpredictable and quite un- Adaptive Radiation of Anoles (Univ. of California Press,
Berkeley, CA, 2009).
repeatable” (32, p. 14) due to historical contingency. 15. E. E. Williams, in Lizard Ecology: Studies of a Model Organism,
Widespread convergence among entire faunas of www.sciencemag.org/cgi/content/full/341/6143/292/DC1
R. B. Huey, E. R. Pianka, T. W. Schoener, Eds. (Harvard Univ.
Greater Antillean Anolis refutes Gould’s claim and Press, Cambridge, MA, 1983), pp. 326–370.
Figs. S1 to S9
shows that adaptation can overcome the influence 16. J. B. Losos, T. R. Jackman, A. Larson, K. de Queiroz,
Tables S1 to S5
L. Rodríguez-Schettino, Science 279, 2115–2118 (1998).
of chance events on the course of evolution. Our References (35–42)
17. D. L. Mahler, L. J. Revell, R. E. Glor, J. B. Losos, Evolution
demonstration of deterministic convergence on a ma- 64, 2731–2745 (2010).
Author Contributions
croevolutionary adaptive landscape complements 18. Information on materials and methods is available as 5 November 2012; accepted 5 June 2013
studies of diversification in species numbers in supplementary material on Science Online. 10.1126/science.1232392
showing that many features of large-scale radiations
may be surprisingly predictable. A recent analysis
discovered that both island diversification rate and
standing species richness in Greater Antillean anoles Predicting and Manipulating Cardiac
Drug Inactivation by the Human Gut
could be predicted from island size and time since
colonization (31). In cichlids, whether colonizing
lineages will radiate in African lakes can be predicted
from the intrinsic traits of the colonist and the eco-
logical opportunities provided by the new habitat Bacterium Eggerthella lenta
(33). Together, these studies suggest that the pri-
mary aspects of evolutionary radiation—adaptation Henry J. Haiser,1 David B. Gootenberg,1 Kelly Chatman,1 Gopal Sirasani,2
and the proliferation of species—may in some Emily P. Balskus,2 Peter J. Turnbaugh1*
cases be largely deterministic.
Despite numerous examples of the effects of the human gastrointestinal microbiome on drug
efficacy and toxicity, there is often an incomplete understanding of the underlying mechanisms.
1. S. Gavrilets, A. Vose, Proc. Natl. Acad. Sci. U.S.A. 102, Here, we dissect the inactivation of the cardiac drug digoxin by the gut Actinobacterium
18040–18045 (2005). Eggerthella lenta. Transcriptional profiling, comparative genomics, and culture-based assays
2. D. Schluter, The Ecology of Adaptive Radiation (Oxford revealed a cytochrome-encoding operon up-regulated by digoxin, inhibited by arginine, absent
Univ. Press, Oxford, 2000). in nonmetabolizing E. lenta strains, and predictive of digoxin inactivation by the human gut
3. G. G. Simpson, Tempo and Mode in Evolution (Columbia
Univ. Press, New York, 1944). microbiome. Pharmacokinetic studies using gnotobiotic mice revealed that dietary protein reduces
4. E. I. Svensson, R. Calsbeek, Eds., The Adaptive Landscape the in vivo microbial metabolism of digoxin, with significant changes to drug concentration in
in Evolutionary Biology (Oxford Univ. Press, Oxford, the serum and urine. These results emphasize the importance of viewing pharmacology from the
2012). perspective of both our human and microbial genomes.
5. B. Frédérich, L. Sorenson, F. Santini, G. J. Slater,
M. E. Alfaro, Am. Nat. 181, 94–113 (2013).
umans are home to large and diverse mi- the human gut microbiome, including alterations
H
6. R. Gillespie, Science 303, 356–359 (2004).
7. T. J. Givnish, in The Biology of Biodiversity, M. Kato, Ed. crobial communities, the most abundant to the bioavailability, activity, and toxicity of ther-
(Springer-Verlag, Tokyo, 1999), pp. 67–90.
8. M. Muschick, A. Indermaur, W. Salzburger, Curr. Biol. 22,
of which resides in the gastrointestinal apeutic drugs (1, 2). Although >40 drugs are
2362–2368 (2012). tract. Recent studies have highlighted the clinical metabolized by the gut microbiome, little is
9. M. L. J. Stiassny, A. Meyer, Sci. Am. 280, 64–69 (1999). relevance of the biotransformations catalyzed by known about the underlying mechanisms. This

REPORTS
knowledge is critical to enable the rational design arginine inhibits this reaction (Fig. 1A). The dependent effects exerted by arginine were con-
of pharmaceutical or dietary interventions. growth of E. lenta DSM2243 was stimulated by firmed on independent samples using quantitative
The inactivation of the cardiac drug digoxin arginine supplementation (Fig. 1A and fig. S2), reverse transcription polymerase chain reaction
provides a promising starting point for under- indicative of using the arginine dihydrolase path- (QRT-PCR) (Fig. 1C, fig. S10C, and table S4).
standing microbial drug metabolism. Digoxin way for adenosine 5′-triphosphate (ATP) (8). Unlike arginine, ornithine did not repress cgr2 ex-
and other cardiac glycosides have been widely Citrulline (an intermediate upstream of ATP pro- pression (fig. S11). These results are consistent
used for hundreds of years to treat heart failure duction) stimulated growth, whereas ornithine with the hypothesis that arginine represses cgr
and arrhythmias. Therapeutic effects are accom- (an end product) did not (figs. S2 and S3). operon expression and thereby inhibits digoxin
plished indirectly when inhibition of the Na+- and E. lenta cultures were grown anaerobically reduction.
K+-dependent adenosine triphosphatase (Na+/K+ in rich medium supplemented with low and Next, we tested three strains of E. lenta
ATPase) in cardiac myocytes raises the intra- high levels of arginine (0.25% and 1.25%, respec- (DSM2243, FAA 1-3-56, and FAA 1-1-60) (9, 10)
cellular Ca2+ concentration (3). Digoxin has a nar- tively) in the presence or absence of digoxin for digoxin reduction; the type strain was the
row therapeutic range (0.5 to 2.0 ng/ml) (3), and (10 mg/ml), and we performed RNA sequencing sole strain capable of digoxin reduction in vitro
some patients excrete the inactive digoxin metab- (RNA-Seq) on the resultant cellular biomass (Fig. 1D). Comparative genomics revealed that
olite, dihydrodigoxin, in which the lactone ring is (figs. S4 to S6 and table S1). A two-gene operon the type strain was nearly indistinguishable from
reduced (fig. S1A) (4). This modification disrupts was highly up-regulated after exposure to di- the other two strains when common marker genes
ring planarity, which is thought to shift posi- goxin during exponential growth (>100-fold) were used (fig. S12). Reciprocal BLASTP com-
tioning within the binding pocket of the Na+/K+ (Fig. 1B and tables S2 and S3). These two genes, parisons of all protein-coding sequences of the
ATPase, resulting in decreased target affinity (5). referred to here as the cardiac glycoside reductase three fully sequenced E. lenta strains revealed

Coadministration of broad-spectrum antibiotics (cgr) operon (gene labels: cgr1 and cgr2), encode that the type strain shared 79.4% and 90.5% of
increases serum digoxin (4), and Eggerthella lenta proteins that are homologous to bacterial cyto- its proteome with strains FAA 1-3-56 and FAA
reduces digoxin in vitro (6). Before this work, the chromes and are therefore potentially capable of 1-1-60, respectively (fig. S12). The cgr operon
molecular mechanism of digoxin reduction and using digoxin as an alternative electron acceptor. was unique to the type strain (table S6); further-
the factors that alter microbial drug inactivation Incubation of E. lenta with multiple cardiac gly- more, the two nonreducing E. lenta strains were
in vivo were unknown. cosides and their reduced forms revealed that missing three genomic loci, which were also
We confirmed that E. lenta DSM2243, the the cgr operon is broadly responsive to com- up-regulated by digoxin, and are predicted to en-
type strain, reduces digoxin in vitro (7) and that pounds with an a,b-unsaturated butyrolactone code membrane transporters for the uptake of
1
ring (figs. S7 to S9 and table S5). small molecules and glycosides (fig. S13). Argi-
Faculty of Arts and Sciences (FAS) Center for Systems Biology, Digoxin induction was increased in low- nine did not significantly decrease the expression
Harvard University, Cambridge, MA 02138, USA. 2Department
of Chemistry and Chemical Biology, Harvard University, Cam- arginine conditions during both its exponential level of these transporters (fig. S14).
bridge, MA 02138, USA. and stationary phases, relative to cultures exposed Strain-level variation provides an explanation
*Corresponding author. E-mail: pturnbaugh@fas.harvard.edu to high levels of arginine (fig. S10, A and B). for the difficulties in predicting dihydrodigoxin
(P.J.T.) cgr induction by digoxin and the growth phase– levels in cardiac patients by the presence or ab-
Fig. 1. Discovery of a bacterialoperon

induced by digoxin. (A) Arginine stim-
ulates the growth of E. lenta DSM2243
in vitro while blocking the reduction of
digoxin. Maximum optical density (ab-
sorbance) at 600 nm (OD600) (solid line;
values are means T SEM; n = 3) and
digoxin percentage reduction efficiency
(dashed line; values are means; n = 2)
after 48 hours of growth. (B) RNA-Seq
profiles of the cgr operon are shown with
and without digoxin during exponential
growth in medium containing low or high
arginine. The height is proportional to
the natural log of the number of unam-
biguous sequencing reads mapped to
each base. (C) cgr2 transcription as de-
termined by QRT-PCR. Asterisks indicate
statistical significance by Student’s t test
(P < 0.05). Horizontal lines are means;
n = 2 to 3. (D) Identification of two
strains of E. lenta incapable of reducing
digoxin. Values are means T SEM; n = 3.
ND, no reduction detected.

REPORTS
sence of E. lenta (6, 11). We used quantitative linked to reduction efficiency (R2 = 0.74, P < 0.0001). (fig. S18A) and exhibited high levels of expres-
PCR (QPCR) to measure the abundance of the cgr An explanation for the observed microbial syn- sion of the cgr operon (fig. S18B). Quantification
operon relative to the E. lenta 16S ribosomal RNA ergy is that the fastidious growth of E. lenta is of serum and urine digoxin (7) revealed signif-
(rRNA) gene (the “cgr ratio”) in microbial com- promoted by growth factors supplied by the gut icant increases in mice fed the high-protein diet,
munity DNA from 20 unrelated healthy people, microbiota, a phenomena that is known to affect indicative of suppressed digoxin reduction by
along with ex vivo digoxin reduction assays. The the metabolism of environmental pollutants by E. lenta (Fig. 3, A and B). These trends were
results stratified our cohort into low reducers soil microbial communities (12), along with also consistent with fecal analysis of samples
(12.82 T 10.68% reduction; n = 6) and high competition for arginine that boosts digoxin from each group of mice 4 to 16 hours after di-
reducers (96.25 T 7.69% reduction; n = 14) (Fig. reduction by E. lenta. Consistent with these hy- goxin administration (Fig. 3). We also confirmed
2A). The cgr ratio was significantly increased for potheses, the abundance of the E. lenta type that the high-protein diet significantly elevated
the high reducers (1.058 T 0.562) when com- strain was significantly increased in the pres- the amino acids in the distal small intestine (7),
pared with low reducers (0.425 T 0.582; P < 0.05, ence of a complex microbial community (1.6e6 T which resulted in a fold increase of 1.71 T 0.06
Student’s t test) (Fig. 2B and fig. S15). Linear 4.8e5 versus 1.8e5 T 8.4e3 in isolation; P < 0.05, (P < 0.001, Wilcoxon test) (tables S9 and 10).
regression of reduction efficiency with the cgr Mann-Whitney test), and arginine supplementa- We controlled for the indirect effects of host
ratio revealed a significant correlation (R2 = 0.22, tion suppressed the reduction of digoxin during diet and colonization that might alter digoxin
P < 0.05), whereas the abundance of E. lenta coculture (fig. S16). pharmacokinetics irrespective of reduction by
failed to predict the extent of reduction (R2 = 0.06, Diet could also explain interindividual vari- E. lenta. Germ-free mice were colonized with
P = 0.30). The optimal cgr ratio cutoff (0.6) pre- ations in digoxin reduction. In vitro growth of either the digoxin-reducing type strain or the
dicted digoxin reduction efficiency with a sen- E. lenta showed that, although arginine stimu- nonreducing FAA 1-3-56 strain and subsequent-

sitivity of 86%, specificity of 83%, and precision lated cell growth, it decreased cgr operon expres- ly fed the same two diets (fig. S17B). As seen
of 92%. sion and prevented the conversion of digoxin to before, we detected colonization with both strains,
Coculture of E. lenta with the fecal micro- dihydrodigoxin (Fig. 1, A and C, and fig. S10). high–cgr operon expression, and elevated serum
biome enhanced the efficiency of digoxin re- These observations led us to hypothesize that and urine digoxin on the high-protein diet for mice
duction. Each low-reducing fecal sample was increased consumption of dietary protein, and colonized with the type strain (Fig. 3, C and D, and
incubated with the type (reducing) and FAA 1-3-56 the corresponding increase in arginine, would in- fig. S18, C and D). Diet did not significantly af-
(nonreducing) strains of E. lenta. The commu- hibit the in vivo reduction of digoxin by E. lenta. fect the serum or urine digoxin levels of mice
nities incubated with the type strain reduced more Germ-free adult male Swiss-Webster mice were colonized with the nonreducing strain (Fig. 3, C
digoxin (95.39 T 2.41%) than the type strain colonized with the type strain before being fed and D). Serum digoxin was significantly lower
alone (68.91 T 7.70%; P < 0.05, Mann-Whitney diets differing only in the amount of total pro- in mice colonized with the type strain fed the
U test) (Fig. 2C). The cgr ratio was significantly tein (n = 5 mice/group) (tables S7 and S8 and 0% protein diet, relative to those colonized with
elevated after coculture (Fig. 2D) and was tightly fig. S17A). E. lenta colonized mice fed both diets the nonreducing strain (4.91 T 1.56 ng/ml vs.
Fig. 2. A microbial biomarker

predicts the inactivation of di-
goxin. (A) Liquid chromatography–
mass spectrometry (LC-MS) was
used to quantify digoxin reduc-
tion in the fecal microbiomes of
20 unrelated individuals. (B) The
cgr ratio was significantly differ-
ent between low and high reduc-
ers. Data represent QPCR with the
cgr2 gene, and E. lenta–specific
16S rDNA primers (table S4). (C)
Five low-reducing fecal microbial
communities were incubated for
5 days in the presence or absence
of E. lenta DSM2243 or FAA
1-3-56. LC-MS was used to quan-
tify the completion of digoxin re-
duction. Supplementation with
the nonreducing strain of E. lenta
did not significantly affect digoxin
reduction efficiency. (D) The cgr
ratio was obtained for each of
the low-reducing microbial com-
munities after incubation. Out-
liers were identified using Grubbs’
test (P < 0.01) and removed.
Values are means T SEM. Points
in (A) and (B) represent biolog-
ical replicates. Asterisks indicate
statistical significance by Student’s
t test (*P < 0.05; ***P < 0.001;
****P < 0.0001).

REPORTS
1. H. J. Haiser, P. J. Turnbaugh, Science 336, 1253–1255
(2012).
2. B. D. Wallace et al., Science 330, 831–835 (2010).
3. L. S. Goodman et al., Goodman & Gilman's the
Pharmacological Basis of Therapeutics (McGraw-Hill,
New York, ed. 12, 2011).
4. J. Lindenbaum, D. G. Rund, V. P. Butler Jr., D. Tse-Eng
J. R. Saha, N. Engl. J. Med. 305, 789–794 (1981).
5. C. D. Farr et al., Biochemistry 41, 1137–1148
(2002).
6. J. F. Dobkin, J. R. Saha, V. P. Butler Jr., H. C. Neu,
J. Lindenbaum, Science 220, 325–327 (1983).
7. Materials and methods are available as supplementary
8. J. F. Sperry, T. D. Wilkins, J. Bacteriol. 127, 780–784
(1976).
9. E. Saunders et al., Stand. Genomic Sci. 1, 174–182
(2009).
10. K. E. Nelson et al., Science 328, 994–999 (2010).
11. V. I. Mathan, J. Wiederman, J. F. Dobkin, J. Lindenbaum,
Gut 30, 971–977 (1989).
12. X. Maymó-Gatell, Y. Chien, J. M. Gossett, S. H. Zinder,

Science 276, 1568–1571 (1997).
Acknowledgments: B. Budnik and S. Trauger for liquid

Fig. 3. Dietary protein blocks the inactivation of digoxin. Serum (A) and urinary (B) digoxin levels chromatography–mass spectrometry (LC-MS) analyses;
from the type strain experiment. Fecal digoxin levels showed a consistent trend: the mean area under V. Yeliseyev, A. Liou, and R. Carmody for mouse studies;
the curve was 6.226 ng digoxin per hour per ml in germ-free mice, 3.576 for mice fed the 0% protein C. Reardon and C. Daly for sequencing support; C. Maurice,
diet, and 6.364 for mice fed the 20% protein diet. Serum (C) and urinary (D) digoxin levels from each L. David, R. Dutton, B. Wolfe, J. Button, M. Elliot, Y. Falanga,
group. Digoxin levels were quantified by enzyme-linked immunosorbent assay (ELISA) (7). Values are R. Losick, A. Murray, and B. Stern for helpful discussions.
Mouse experiments were done with the generous support of
means T SEM. Asterisks indicate statistical significance by Student’s t test (*P < 0.05; **P < 0.01). n = 4 the Harvard Digestive Diseases Center and the University
to 5 mice per group. NS, not significant. of North Carolina gnotobiotic cores. This work was supported
by grants from the NIH (P50 GM068763) and the
Harvard Digestive Diseases Center (2P30DK034854-26).
H.J.H. is supported by the Canadian Institutes of Health
13.8 T 1.25; P < 0.01, Student’s t test) (Fig. 3C). based biomarker assessments of the gut micro- Research (MFE-112991). RNA-Seq data are deposited
Together, these results suggest that the enhanced biome can guide dosage regimes. It may also be in the Gene Expression Omnibus (GEO) database
free amino acids available to E. lenta inhibited possible to provide dietary guidelines or supple- (accession GSE43919).
the activity of the cgr operon and increased the ments that prevent microbial drug metabolism.
bioavailability of digoxin. More broadly, our results emphasize that a com- Supplementary Materials
An expanded model of digoxin pharmaco- prehensive view of pharmacology includes the www.sciencemag.org/cgi/content/full/341/6143/295/DC1
kinetics is now emerging: Colonization by dis- structure and activity of our resident microbial Figs. S1 to S19
tinct strains of E. lenta, microbial interactions, communities and a deeper understanding of their Tables S1 to S10
and host diet act together to influence drug levels interactions with each other, with their host hab- References (13–25)
(fig. S19). Follow-up studies in cardiac patients itat, and with the nutritional milieu of the gastro- 30 January 2013; accepted 14 June 2013
are necessary to determine whether rapid QPCR- intestinal tract. 10.1126/science.1235872

Life Science Technologies
Produced by the Science/AAAS Custom Publishing Office New Products
DIGITAL HD MICROSCOPE CAMERAS

Easy to use, fast live images, brilliant results—microscope users achieve all of this and more with
the new Leica MC120 HD and MC170 HD cameras. The new high-performance cameras provide
high-speed, precise images in real time at a rate of up to 30 frames per second. They are directly
connectable to an HD monitor as a stand-alone solution or to a computer via USB interface.
While the 2.5 megapixel Leica MC120 HD is ideal for almost all microscopic applications, the
Leica MC170 HD with a resolution of 5 megapixels is especially well suited for acquiring the fin-
est details at low magnifications. For fast documentation or hands-on training the user can record
brilliant full HD movie clips as MP4. Images and movies are easily, cost-effectively stored with
high quality onto a portable SD-card. The infrared remote control allows direct and convenient
control of all camera parameters such as brightness, gain, and contrast of the image as well as
white balance operations, changing to other camera modes, and image capture.
Leica Microsystems
For info: 800-248-0123 www.leica-microsystems.com
EVAPORATORS MULTISAMPLE HOMOGENIZER

The third generation Acid-Resistant EZ-2 Plus Evaporator uses inert- and The Multi-Prep Laboratory Homogenizing System makes multisample ho-
corrosion-proof materials to enable it to withstand up to 6N Hydrochloric mogenizing simple.This compact, benchtop unit was designed with decades
acid, 70% Nitric acid, and most acid chlorides including thionyl chloride. of homogenizing experience and was created to address the challenges and
These strongly acidic materials can be routinely removed from samples rigorous nature of multisample homogenization. The Multi-Prep’s unique
without any loss of performance or long-term deterioration in the system. automated system improves lab productivity by homogenizing up to six
Acid resistant components in the Acid-Resistant EZ-2 Plus include a PTFE samples at a time quickly and with ease. All samples are homogenized si-
coated evaporation chamber, a glass condenser, and all metallic parts that multaneously in mere seconds, ensuring accurate homogenizing speed and
come into contact with removed solvent, which are manufactured from results. Precision crafted 316 Stainless Steel Multi-Prep Sample Probes de-
acid resistant Hasteloy C steel. The innovative design of the Acid-Resistant liver accurate and quick homogenizing results that are superior to plastic
EZ-2 Plus presents real advantages for all scientists and engineers tasked disposable generators while still addressing the concerns of avoiding cross
with removing both regular solvents and strongly acidic chemicals. Intuitive contamination. Their multipack design means a clean, durable multisample
controls and large LCD display provide ease of use comparable to a typical probe can be used each and every time. Typical run time is mere seconds.
rotary evaporator yet the Acid-Resistant EZ-2 Plus can process many more The short sample runtime coupled with the ability to process six samples
samples per unit time. per cycle can result in 300 homogenized samples per hour.
Genevac ProScientific
For info: +44-(0)-1473-240000 www.genevac.com For info: 203-267-4600 www.proscientific.com
DIGITAL SLIDE SCANNER PLATE HEATER

Axio Scan.Z1 is an automated microscope system that allows researchers The new Plate Heater is designed specifically for cell biologists who re-
to digitalize fixed tissue sections and cytologic specimens in brightfield quire accurate and precise temperature control for plates in laminar flow
and fluorescence. Thanks to the tray concept, Axio Scan.Z1 captures the hoods. The new Plate Heater has been developed to solve issues associated
entire specimen area of the microscope slide—including the edge. Just with short-term assay incubations in a laminar flow hood environment. The
a few minutes later, the self-calibrating automized slide scanner presents Plate Heater has a defined temperature range suitable for cell biology use
specimens on high-quality virtual slides. Up to 100 microscope slides can (ambient + 5˚C to 50˚C) and maintains a consistent temperature across its
be digitalized at a time. For fluorescence applications, filter wheels switch surface ensuring all cells, no matter where their location on the plate, are
wavelengths in just 50 ms. Sensitive cameras and maximally corrected op- incubated at the same temperature. Furthermore, the unit also incorporates
tics achieve optimal image quality. The Colibri.2 ultraviolet-free LED light a lid to help maintain a stable temperature, even when being used inside
source and a focus finder with oblique illumination, the Ring Aperture a laminar flow hood. These features make the Plate Heater ideal for cell-
Contrast, ensure maximum protection for the sample. Axio Scan.Z1 is op- based and biochemical applications where consistent temperature control is
erated by ZEN imaging software from Carl Zeiss. ZEN allows users either important, such as incubating cell-based assays in microplates, trypsinising
to work with predefined recording parameters automatically or to select all cells in cell culture T-flasks, or even generating reproducible 3-D cell cul-
settings individually. tures with the RAFT (Real Architecture for 3D Tissue) System.
Carl Zeiss TAP Biosystems
For info: 800-233-2343 www.zeiss.com/micro For info: +44-(0)-1763-227333 www.tapbiosystems.com
Electronically submit your new product description or product literature information! Go to www.sciencemag.org/products/newproducts.dtl for more information.
Newly offered instrumentation, apparatus, and laboratory materials of interest to researchers in all disciplines in academic, industrial, and governmental organizations are
featured in this space. Emphasis is given to purpose, chief characteristics, and availability of products and materials. Endorsement by Science or AAAS of any products or
materials mentioned is not implied. Additional information may be obtained from the manufacturer or supplier.
www.sciencemag.org/products SCIENCE VOL 341 19 JULY 2013 299

International Journal of
AIDS Research Journal of
and Treatment
Advances in
Condensed Matter Physics Endocrinology SPECTROSCOPY
Hindawi Publishing Corporation Volume 2013
http://www.hindawi.com
International Journal of BioMed Research

Genomics International
Submit your manuscripts at

http://www.hindawi.com
Photoenergy
International Journal of
Nanoma
PPAR
Advances in
Nanomaterials Meteorology
Journal of
Research
The Scientifc
World Journal
Neural Bioinorganic Chemistry Advances in

High Energy Physics
Journal of
Plasticity
and Applications
Clinical & Chemistry
Developmental
Immunology
Hindawi Publishing Corporation

http://www.hindawi.com Volume 2013
BD Horizon™ Brilliant Violet™ Polymer Conjugates
Bright dyes help resolve rare and dim cell populations
For more discerning research.

Now you can resolve rare or dim cell populations The complete portfolio of BD conjugated
with BD Horizon™ Brilliant Violet™ polymer antibodies can be used to explore cellular
conjugates from BD Biosciences. features and characterize cells through surface,
Developed from pioneering polymer dye intracellular, or secreted markers. To ensure you
technology acquired from Sirigen Ltd., can use BD reagents across your entire multicolor
BD Horizon Brilliant Violet dyes are brighter panel, our portfolio contains a broad selection of
than conventional dyes. Improved brightness fluorochrome-conjugated antibodies.
enables you to identify cell populations Request a free sample, or find out how you
with lower receptor density than was can use BD Horizon Brilliant Violet polymer
previously possible, resolving cell populations conjugates with our expanded set of tools and
previously obscured. information at bdbiosciences.com/go/brilliant.
BD Biosciences
For Research Use Only. Not for use in diagnostic or therapeutic procedures. 2350 Qume Drive
BD, BD Logo and all other trademarks are property of Becton, Dickinson and Company. © 2013 BD San Jose, CA 95131
23-14921-00 bdbiosciences.com
Want to win a rather
special prize in Stockholm,
Sweden this December?
Henrik Trygg / Image bank, Sweden

Winner’s essay published in the journal Science
$25,000 dollars grand prize
Awards held in Stockholm in December
Tis December a rather special prize will be awarded in Stockholm, Sweden. Te journal Science and SciLifeLab For over 130 years the
have come together to recognize and celebrate excellence in PhD research. Te Science and SciLifeLab Prize has journal Science has been the
been established to support young scientists at the start of their career. world’s leading journal of
original scientifc research,
“Scienceandhasencourage
never been more exciting and, as leaders in science, we need to support global news and commentary.
young researchers today and tomorrow. Tis prize is a way of doing just that.
Professor Mathias Uhlén, Director SciLifeLab
” SciLifeLab is a collaboration
between four universities in
Te grand prize winner of this major global award will have their essay published in the journal Science and receive Stockholm and Uppsala, Sweden,
$25,000. Tree runners up will receive a combined total of an additional $10,000 in prize money. and is a pioneering center for
Te prizes will be presented in Stockholm, Sweden in the middle of December 2013. large-scale biosciences with a
focus on health and
To enter environmental research.
You must be a recent Ph.D. graduate (awarded between January 1, 2011 and December 31, 2012).
Submissions must be in the form of a 1000 word essay, in English, on your thesis, highlighting the signif- With the kind support
cance of its contribution and overall implications in the feld. Te four submission areas for this prize are: of the Knut and Alice Wallenberg
(1) Genomics / Proteomics / Systems Biology (2) Developmental Biology (3) Molecular and Cellular Biology Foundation.
(4) Environmental Life Science.
Te deadline for submissions is August 15, 2013. Te overall winning essay will be published in Science.
For further details and to enter, please go to: www.sciencemag.org/scilifelabprize
Eppendorf Consumables—
it’s Your sample
Future included
50 years of experience in highest quality
Eppendorf consumables are the result Purest assay results are no coincidence:
of 50 years constant improvement and > Unique features to make every day
development. With the new Eppendorf routines faster and easier
Tubes® 5.0 mL a new tube format in the > Minimized risk of chemical leaching
medium volume range will complete the from consumables
legendary Eppendorf Tubes family. > Purity grades tailored to even the
Laboratory processes will become highest requirements
easier and more convenient.
www.eppendorf.com/consumables
Eppendorf®, the Eppendorf logo and Eppendorf Tubes® are registered Trademarks of Eppendorf AG, Germany.
All rights reserved, including graphics and images. Copyright © 2013 by Eppendorf AG.

6FLHQFH

141 1 1 "1 13 13 . 34 41 3 1 "1141 1 1431 1
1 1 1 $(%1,1 1 $)%1,13 11 1 343 13 134" 1
13 1 1 131 1113141141#%13 131 4 1131 31
4 1 41 1 ,1 4 431 1 1 4 1 4 4 1 431 41 1 431 " 1 -1 31
0 11(+1 1 1 $(%1,1 1 14314 11 34 4 1 4 11
1 $)%1,1 113 4 4 141+1 1 1 11 1431! 3 11
1 1141 4 1&431 1 4 4 1431 4 13 11 31 1343 1
3 11,14" 1 1 12*1 1 14" 131 14 "1 43 1 1
1! 14 411431,3 11 33 13414 3414 1 31 14 " 1
4 3414113133 1 1 13 1 1 14 3 1411 1 11 1
1 143 4 1 141431314 1 1 131

&431 4'1/%%(*/%$(# 34 4

/1 / 12 / /'/0 / 2121/ / 1/ / % &'%('('%( ( % % % %(
2112 2 122/ 1 /2/ //2/ //2/#/ 1221/ ( % % %( % (% % % % % %
/($./1/ / / 2 / /21 / /2 /21 / (%( ( %' % % %' %(% ' %% % %
/ 12 / / 1 / / / 12 / 1 2 / 12 / / ' ('( % '%(% %( ( ( %% &' % %
2/ /2 /2/121 /21/2 1 / 1212 / // / ( % % '( % % '( % % ( ( % % (% %
/1 /2 2 / // /'/0 / //0%/ %%% % % % % % % %'%( ( %
2 / 212 / 1// /2 1/ // /1 // ( % % '% ( % ' %'%( ( % % %
2 / 2/ 2 / / 1 2 / 2 / / 1/ 1 / 12 / ' %&' ( %' %!"#% %% &'% %&'( %
/1 / 2/ / // 2 / /2// / '%'% % %( ( %' %%'% ' ('%(% %
'/0 /1 /1/ /21// / / 1 / / ( % '(' % % % ' %%( ' %(% ( %'( %
/1 2 /2/11/ 2 / / 12 / / / / ( ( % '% % % %% % '% %
2 12 / /1/ -/ /12 / /2/ /2 1/2/// '(% %% %% % '% % %' % % % '% ( %
121/ 2121/// /'/0 / /12 / / 21/ % ('% %' %( % %% %(%'( % % %
/1/ / / '%% %' % %!$$%( ( % %'%('%

%21/ 2&/+))*$,")(!')$$$$///--- #/ ---/0 0

! /- $- - -/ - ./ /- --/ -. .. - - - .- /- - - - 0/- - 00 - 0- /
/- - ! - - / - - // - - - . - /-/- -/ -//-0 /0-0/- - - /-0-
- /. - - /- - .- / - ! /- $- . - - 00 -- -.-/- - - 0 -0- 0- - 0 -
.- - .- /- - ./ /- * - - - + - - 0//- - 0 - -- /-0-00 - /0 --
- - - . - - . - / - - .- . - .-/- -- -//-/ - -0/-- 00- -
-/ - " -# .- --/-$%%- ./ /- - - ' -(-)* -0-)!* +- - --0 -//-/0 - -
- --- -&/.-/. - / - .- - - /- / -- - 0-/-/- /-0 0 -0 ---/-
# - -, -(%- - #-. - -! -/.. -/ --- -//-"//0/-- - 0- 0/0/ -
- #-/ - -" -- / . '-.# /.-)- - - -0-- -- -//- - - - - -
/.- - - / - .- - / "- / - - - ..- - /-0-00 - -/ ---.-/- -0/-
/ . - .-! -./ / -/.-- - ! /- $- -/ - - 0 - 0 -/-0 -//-0 /0- -
. --- - /.-./- .- --/ - . - .- 0/ - - - - - - - /0 -/ -
/-/.- -. -./. -.. -/ -/.-/- -- - --/- -0/- - / -//0 -,--
- # -- / - -- "-'- / /-0/-,- +- --

111 4 34 "0/- 0#-(&&'!+% $$%------ 0 0
Electronically submit your new product description or product literature information! Go to www.sciencemag.org/products/newproducts.dtl for more information.
Newly offered instrumentation, apparatus, and laboratory materials of interest to researchers in all disciplines in academic, industrial, and governmental organizations are
featured in this space. Emphasis is given to purpose, chief characteristics, and availability of products and materials. Endorsement by Science or AAAS of any products or
materials mentioned is not implied. Additional information may be obtained from the manufacturer or supplier.
www.sciencemag.org/products SCIENCE VOL 341 19 JULY 2013 299

STAY INFORMED!
STAY CONNECTED!
Get more from your AAAS membership
memb
bership
Are you currently registered to receive e-mails from AAAS and Science?
E-mail is the primary way that AAAS communicates with our members
about AAAS programs, new member benefits, invitations to special
events, and, of course, the latest news and research being published
in Science.
Sign up today to receive e-mails from AAAS and ensure that you are
getting the most out of your membership and Science subscription.*
To get started visit: promo.aaas.org/stayconnected

You’ll need your AAAS Member number. Find it above your name on
your Science mailing label.
Don’t miss a thing. Sign up for e-mail communications from AAAS today!
*AAAS follows CAN-SPAM and European Safe Harbor guidelines for protecting your privacy. We will never sell your e-mail address and you can opt-out of receiving e-mails at any time.
By your side since 1962
One company, one family of trusted brands. Here to support your research.
As your research needs evolve, we evolve to meet them. We continuously innovate high-quality
solutions for every lab and every application, building from technology chosen by scientists for up to
50 years. And we’re staffed to support your success, with 3,000 technical sales, service, and support
professionals ready to assist you.
Find the most cited life science brands at

lifetechnologies.com/trustedbrands
©2013 Life Technologies Corporation. All rights reserved. The trademarks mentioned herein are the property of
Life Technologies Corporation and/or its affiliate(s) or their respective owners. CO05619 0613
online @sciencecareers.org
There’s only one
Department of Molecular Virology and Microbiology

The Alkek Center for Metagenomics and Microbiome Research (CMMR) and the Department of Molecular
Virology and Microbiology at Baylor College of Medicine seek to fill (3) tenure track faculty positions at
the ASSISTANT PROFESSOR level. We are seeking investigators with promising track records who are
emphasizing Microbiome Research in humans and/or model systems. While specific research interests may
vary, preferred candidates will be pursuing studies in host-microbe interactions, mucosal immunology, cell
Science Careers signaling, metabolomics and/or other ‘multi–omic’ related strategies and analyses.
At Baylor College of Medicine (BCM), you will find a unique collaborative spirit, a dedication to excellence
Advertising in patient care and an environment that surpasses its reputation as a research powerhouse and educational
giant. BCM is located in the heart of the Texas Medical Center (TMC), the largest incorporated medical
For full advertising details, go to center in the world with more than 42 member institutions in 100 buildings. The CMMR is building a highly
ScienceCareers.org and click collaborative environment where exciting advances in our understanding of how microbes impact human
For Employers, or call one of health and disease will result from synergistic interactions between collaborating faculty, clinicians, centers
our representatives. and institutions. This position features a competitive start-up package and laboratory/office space and will
Tracy Holmes leverage the facilities and resources located in the CMMR and at BCM/TMC, including the Baylor College
Worldwide Associate Director of Medicine Human Genome Sequencing Center, one of the three NHGRI-funded Large Scale Sequencing
Science Careers Centers with production scale platforms for the latest generation sequencing technologies.
Phone: +44 (0) 1223 326525 Successful candidates will synergize with our strengths in metagenomics research to delve into the mecha-
nisms by which microbes impact human health. They will maintain an outstanding research program that
THE AMERICAS focuses on solving core problems in this area, and have a commitment to excellence in teaching. Required
E-mail: advertise@sciencecareers.org qualifications include a Ph.D. and a recognized reputation for innovative scholarship, a distinguished track
Fax: 202-289-6742 record for research support and publications, leadership experience of students and trainees, and continued
faculty development.
Tina Burks
East Coast/West Coast/South America Applicants should submit a cover letter, a curriculum vitae including a publication list, a statement of
Phone: 202-326-6577 research accomplishments and future research plans, a description of teaching experience and philosophy,
Marci Gallun and the names and addresses of three potential referees to the address or email below by September 13,
Midwest/Canada 2013. Search Committee, Alkek Center for Metagenomics and Microbiome Research, Baylor College
Phone: 202-326-6582 of Medicine, One Baylor Plaza BCM MS385, Houston, Texas 77030; Telephone: 713-798-5867; email:
bcm-mvm-facultypos1@bcm.edu.
Candice Nulsen
Corporate Baylor College of Medicine is an Equal Opportunity, Affirmative Action and Equal Access Employer.
Phone: 202-256-1528
Online Job Posting Questions
Phone: 202-312-6375
EUROPE / I NDIA / A USTRALIA /

NEW ZEALAND / R EST OF WORLD
E-mail: ads@science-int.co.uk
Fax: +44 (0) 1223 326532
Neuroscience Faculty Positions
Axel Gesatzki
Phone: +44 (0)1223 326529
PRESIDENT
TUCSON, ARIZONA The California Institute of Technology invites
Kelly Grace applications for tenure-track professorial posi-
Phone: +44 (0) 1223 326528 The Research Corporation for Science
Advancement (RCSA) invites applications for tions in the Division of Biology and Biologi-
the position of President. The President reports cal Engineering. We are searching for talented
JAPAN to the Board of Directors and will carry out investigators in Molecular, Cellular, and
Yuri Kobayashi responsibilities of chief executive officer. RCSA Systems Neuroscience. Successful applicants
is the oldest foundation in the United States
Phone: +81-(0)90-9110-1719 devoted wholly to science and its mission is to
are expected to develop an innovative research
E-mail: ykobayas@aaas.org support the pursuit of high-potential early-stage program and should have a commitment to
scientific research. It focuses on providing teaching. Preference will be given to candi-
CHINA / KOREA / S INGAPORE / catalytic funding for highly innovative research dates at the Assistant Professor level. Initial
that will have a lasting impact on science and appointments at the assistant professor level
TAIWAN / T HAILAND society. Its historic emphasis is on the physical
sciences and closely related fields.
are for four years, and are contingent upon
Ruolei Wu completion of the Ph.D. degree.
Phone: +86-1367-1015-294 The President will join RCSA at an exciting time.
E-mail: rwu@aaas.org Having recently celebrated its centennial, the
organization is engaged in strategic planning to Please submit an on-line application at http:
identify ways of amplifying its impact. The //biology.caltech.edu/Positions and include
All ads submitted for publication must comply President will lead the completion of the plan a brief cover letter, curriculum vitae, relevant
with applicable U.S. and non-U.S. laws. Science and drive its implementation. The President will
reserves the right to refuse any advertisement publications, and a description of proposed
oversee the Foundation’s science programs and
at its sole discretion for any reason, including
peer review process, enhance relationships with research. Instructions will be given for sub-
without limitation for offensive language or mission of letters of reference when you apply
inappropriate content, and all advertising is external communities, and steward the
subject to publisher approval. Science encour- organization’s resources. RCSA seeks an on-line. The application deadline is October
ages our readers to alert us to any ads that innovative and visionary leader with a deep 31, 2013. Applicants must be prepared to
they feel may be discriminatory or offensive. history of scientific excellence and demonstrated attend a recruiting symposium at Caltech
management skills and financial acumen.
on January 9-10, 2014.
Inquiries, nominations, and applications should be
sent in confidence to: John Muckle, Principal,
Isaacson, Miller at RCSA@imsearch.com. The California Institute of Technology is
an Equal Opportunity/Affirmative Action
Employer. Women, minorities, veterans, and
disabled persons are encouraged to apply.
ScienceCareers.org
With the recent opening of the
Stanley M. Truhlsen Eye Institute,
the Department of Ophthalmology
and Visual Sciences at the University Chugai Pharmabody Research Pte. Ltd. (“CPR”) was established
of Nebraska Medical Center in in Singapore as a wholly-owned subsidiary of Chugai Pharma-
Omaha, NE is seeking to expand its research faculty. Candidates ceutical Co., Ltd (Chugai) in January 2012 and started opera-
for Assistant Professor should have a strong record of scholarly tions on July 2, 2012. Located at Biopolis in Singapore, it aims
activity with evidence of initial research funding success. Candidates to generate new antibody drugs based on Chugai’s proprietary
for Associate or Full Professor should have a strong, well-funded antibody engineering technology.
eye research program with qualifications appropriate to appointment
level. Research interests should complement strengths of existing We are inviting talented and highly motivated individuals
faculty in retinal diseases, neuroscience, stem cells, corneal diseases, with postdoctorate qualifications to join us as postdoc
glaucoma, cataract, and lens. Approaches of particular interest include
Researchers.
epigenetics, pharmacogenomics, imaging, expertise with zebra fish or
mouse as model animal systems, and ocular drug delivery.
Job Requirements
Interested candidates should submit their curriculum vitae and state-
ment of research interests by email to: Wallace Thoreson, Ph.D., You must have completed and be awarded with a PhD Degree in
Vice Chair for Research (wbthores@unmc.edu). The successful Science or Technology (not in the process of achieving it). You
candidate will have a primary appointment in the Department of must be able to demonstrate technical expertise in areas which
Ophthalmology and Visual Sciences and may also have secondary may include molecular biology, cell biology, protein chemistry,
appointments with other departments. protein engineering and pharmacology as evidenced by projects,
presentations, or peer-reviewed scientific publications. You will
UNMC is an Equal Opportunity Employer. have a strong advantage if you possess prior work experience
in antibody drug discovery work, coupled with strong interper-
sonal, team working and communication skills.
If you are interested to join us as a postdoc Researcher, please

visit our website (Careers Page) at http://www.chugai-
pharmabody.com/ for more information and application
procedure.
We regret that only shortlisted candidates will be notified.
The Alberta Terrestrial Imaging Centre (ATIC) at the

University of Lethbridge undertakes a concentrated effort
towards the advancement of scientific knowledge in
remote sensing. The Centre places special emphasis on
imaging spectroscopy and LiDAR, applying research and
development to the monitoring of natural resources and
the environment.
The ATIC, the Faculty of Arts & Science, and the
School of Graduate Studies at the University of
Lethbridge invite applications for:
2 MSc Studentships
LiDAR based environmental modeling Dr. C. Hopkinson
Forest ecosystem modeling Dr. D. Peddle
2 PhD Studentships
Vegetation health monitoring Dr. C. Coburn
2 Post Doctoral Fellowships
Terrestrial or atmospheric remote sensing Dr. Karl Staenz
Applicants to the above positions/studentships should clearly specify which project they
are applying and submit a curriculum vitae, a letter of application outlining research
experiences and aspirations, and the names and contact information for at least three
referees to trevor.armstrong@uleth.ca by September 10, 2013
The University of Lethbridge offers Masters and Doctoral studentship/
assistantship opportunities in a range of multi-disciplinary areas in the
Sciences.
For more information, contact the School of Graduate Studies:
www.uleth.ca/graduatestudies
Phone: 403 329-5194
Email: sgsinquires@uleth.ca
Women in Science Booklet
Science and the L’Oréal Foundation present
In partnership with
ence Business Ofce

to you by the AAAS/Sci
This booklet is brought
Read inspiring profiles of women

making a difference in biology.
Free download at
ScienceCareers.org/LOrealWIS
Cardiac Electrophysiology/Molecular Physiology
Oregon Health & Science University (OHSU) Department of Physiology and
Pharmacology and the Knight Cardiovascular Institute invite applications Abraham A. Mitchell Chair
for a tenure-track faculty position from individuals with training in Bioen-
gineering, Biophysics or Physiology who are interested in investigating the Urologic Oncology
molecular basis of cardiac arrhythmias. The successful candidate will join
a multidisciplinary, translational research team comprising basic science and The highly qualified candidate for this position at
clinical faculty focused on the underlying mechanisms of arrhythmia develop- the University of South Alabama Mitchell Cancer
ment and generation, with the aim of developing novel therapeutic approaches.
Preference will be given to candidates for the position of Assistant Professor, Institute will be a clinically active physician-
but exceptional candidates for the position of Associate and Full Professor will scientist with appropriate training and experience,
also be considered. We seek an individual who will develop an independent exemplary credentials in clinical urologic oncology,
research program, contribute to the teaching of medical and graduate students
and interact with investigators studying cardiovascular and reproductive biol- and an established, well-funded basic/translational
ogy, drug metabolism, signal transduction, ion channel biology and G-protein research program focused on urologic cancers. The
coupled receptors. OHSU offers a highly interactive research environment and ideal candidate additionally will have demonstrated
superb opportunities for career development in a spectacular Pacific Northwest
setting. The recent founding of the Knight Cardiovascular Institute offers an
leadership potential and business acumen, as
opportunity to be a part of the development of this new, multidisciplinary responsibilities of the position will include leading the
translational research effort. development of an entrepreneurial, financially sound,
OHSU values a diverse and culturally competent workforce. Candidates should translational research-focused department of urologic
model a positive attitude regarding diversity and inclusion while ensuring oncology that brings specialized clinical services and
service delivery is provided in a culturally competent manner. Individuals who
promote diversity and a culture of inclusion are encouraged to apply. OHSU clinical trials to the patients of the Gulf coastal region
provides reasonable accommodations for applicants with disabilities. We are and beyond.
proud to be an equal opportunity, affirmative action organization.
A completed application should consist of the following: curriculum vitae, Applicationsaretobee-mailedto:sallen@usouthal.edu
a brief summary of research accomplishments, an outline of future research by August 30, 2013.
plans, and three letters of recommendation. The successful candidate will have
a Ph.D. in a relevant discipline and a track record of independent publications
and funding. Applications and letters of reference may be sent electronically
USA is an Affirmative Action and Equal Opportunity
to: Beth Habecker, Ph.D., Department of Physiology & Pharmacology, Employer.
Oregon Health & Science University, habecker@ohsu.edu.
Download your free copy today at

ScienceCareers.org/booklets
AAAS is here – helping scientists achieve career success.
Every month, over 400,000 students and scientists visit ScienceCareers.org in search of the information, advice, and
opportunities they need to take the next step in their careers.
A complete career resource, free to the public, Science Careers offers a suite of tools and services developed specifically
for scientists. With hundreds of career development articles, webinars and downloadable booklets filled with practical
advice, a community forum providing answers to career questions, and thousands of job listings in academia, govern-
ment, and industry, Science Careers has helped countless individuals prepare themselves for successful careers.
As a AAAS member, your dues help AAAS make this service freely available to the scientific community. If you’re not
a member, join us. Together we can make a difference.
To learn more, visit aaas.org/plusyou/sciencecareers

AAAS is here – promoting universal science literacy.
In 1985, AAAS founded Project 2061 with the goal of helping all Americans become literate in science, mathematics, and
technology. With its landmark publications Science for All Americans and Benchmarks for Science Literacy, Project 2061 set out
recommendations for what all students should know and be able to do in science, mathematics, and technology by the time they
graduate from high school. Today, many of the state standards in the United States have drawn their content from Project 2061.
Every day Project 2061 staff use their expertise as teachers, researchers, and scientists to evaluate textbooks and assessments,
create conceptual strand maps for educators, produce groundbreaking research and innovative books, CD-ROMs, and profes-
sional development workshops for educators, all in the service of achieving our goal of universal science literacy.
As a AAAS member, your dues help support Project 2061 as it works to improve science education. If you are not yet a AAAS
member, join us. Together we can make a difference.
To learn more, visit aaas.org/plusyou/project2061

Your
Download career Get help
your free copy is our from the
today. cause. experts.
ScienceCareers.org/booklets www.
sciencecareers.org
• Job Postings
• Job Alerts
re ers Away
CAREER Cfraom the Bench • Resume/CV
Database
TRENDS tions for Scientists
Advice and Op
• Career Advice
• Career Forum
Stop searching
for a job;
start your career.
you by
is brought to
This booklet ce Business Office
ien
the AAAS/Sc
www.ScienceCareers.org
From technology specialists to patent
attorneys to policy advisers, learn more
MARKETPLACE
about the types of careers that scientists
can pursue and the skills needed in order
to succeed in nonresearch careers.
306 19 JULY 2013 VOL 341 SCIENCE www.sciencecareers.org

Science 2013-7-19

Uploaded by

Copyright:

Available Formats

Science 2013-7-19

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Science 2013-7-19

Uploaded by

Copyright:

Available Formats

CONTENTS Volume 341 Issue 6143

EDITORIAL BOOKS ET AL.

224 Latest Skirmish Over Ancestral Violence PERSPECTIVES

www.sciencemag.org SCIENCE VOL 341 19 JULY 2013 211

REPORTS 275 Loss of Function of the Melanocortin 2

www.sciencemag.org SCIENCE VOL 341 19 JULY 2013 213

Downloaded from www.sciencemag.org on July 18, 2013

214 19 JULY 2013 VOL 341 SCIENCE www.sciencemag.org

Downloaded from www.sciencemag.org on July 18, 2013

www.sciencemag.org SCIENCE VOL 341 19 JULY 2013 214-b

Downloaded from www.sciencemag.org on July 18, 2013

www.sciencemag.org SCIENCE VOL 341 19 JULY 2013 215

cell differentiation. Both groups hy-

Merger Relics EPIDEMIOLOGY

Downloaded from www.sciencemag.org on July 18, 2013

216 19 JULY 2013 VOL 341 SCIENCE www.sciencemag.org

Downloaded from www.sciencemag.org on July 18, 2013

www.sciencemag.org SCIENCE VOL 341 19 JULY 2013

218 19 JULY 2013 VOL 341 SCIENCE www.sciencemag.org

AROUND THE WORLD

2014 and 2020, up from €3.12 billion in

Downloaded from www.sciencemag.org on July 18, 2013

CREDITS (TOP TO BOTTOM): EUROPEAN UNION/SHIMERA/JENNIFER JACQUEMART; WIKIMEDIA COMMONS

220 19 JULY 2013 VOL 341 SCIENCE www.sciencemag.org

Downloaded from www.sciencemag.org on July 18, 2013

Meet ‘Big-Nose Horned-Face’

There’s a new contender for wackiest-looking

www.sciencemag.org SCIENCE VOL 341 19 JULY 2013 221

Downloaded from www.sciencemag.org on July 18, 2013

Cool by the Numbers

222 19 JULY 2013 VOL 341 SCIENCE www.sciencemag.org

Downloaded from www.sciencemag.org on July 18, 2013

Meet ‘Big-Nose Horned-Face’

There’s a new contender for wackiest-looking

www.sciencemag.org SCIENCE VOL 341 19 JULY 2013 221

Downloaded from www.sciencemag.org on July 18, 2013

Cool by the Numbers

222 19 JULY 2013 VOL 341 SCIENCE www.sciencemag.org

group. Once anomalous data were corrected,

Downloaded from www.sciencemag.org on July 18, 2013

Shadow on Drug Trial and several similar studies at other Japanese

www.sciencemag.org SCIENCE VOL 341 19 JULY 2013 223

and blurred by intermarriage. “In my view the

224 19 JULY 2013 VOL 341 SCIENCE www.sciencemag.org

Spain’s Research Council Approaches Bankruptcy

Downloaded from www.sciencemag.org on July 18, 2013

and failed to honor past commitments, caus-

www.sciencemag.org SCIENCE VOL 341 19 JULY 2013 225

Ever-Bigger Viruses Shake Tree of Life

Downloaded from www.sciencemag.org on July 18, 2013

Eukaryotes Parasitic Free-living

Bacteria Parasitic Free-living

Archaea Parasitic Free-living

226 19 JULY 2013 VOL 341 SCIENCE www.sciencemag.org

virus’s genes and are missing some ing Uncomplete

Downloaded from www.sciencemag.org on July 18, 2013

Germany Debates How to Strengthen Universities

www.sciencemag.org SCIENCE VOL 341 19 JULY 2013 227

virus’s genes and are missing some ing Uncomplete

Downloaded from www.sciencemag.org on July 18, 2013