NATURE|Vol 466|1 July 2010 NEWS & VIEWS

can cycle from one state to another when main- in mind is the relevance of a study to tumour The rocks have been dated with great precision
tained under conditions that support stem growth in patients. ■ (to 2,100 ± 30 million years old), and the pos-
cells. This highlights the fact that stochastic Peter Dirks is at the Hospital for Sick Children, sibility that the structures are a younger min-
influences may be at work within stem-cell University of Toronto, Toronto, Ontario M5G 1X8, eral assemblage that grew within the rock has
populations8. Roesch and colleagues’ results Canada. been excluded by sulphur-isotope analyses.
will be crucial if the same principles apply to e-mail: peter.dirks@sickkids.ca These indicate that the fossil-replacing min-
patients’ primary tumours that have not been eral pyrite (an iron sulphide) was precipitated
1. Boiko, A. D. et al. Nature 466, 133–137 (2010).
maintained in culture. 2. Roesch, A. et al. Cell 141, 583–594 (2010).
by sulphate-reducing bacteria as the sediments
The field of cancer-stem-cell research is 3. Quintana, E. et al. Nature 456, 593–598 (2008). were deposited. Further, contrasts between the
young and emerging, so it is not surprising that 4. Takahashi, K. & Yamanaka, S. Cell 126, 663–676 (2006). carbon-isotope signature of the structures and
many questions remain. Each study must be 5. Hamburger, A. W. & Salmon, S. E. Science 197, 461–463 (1977). their surrounding sediment indicate that the
6. Lee, J. et al. Cancer Cell 9, 391–403 (2006).
carefully assessed, particularly when consider- 7. Hayashi, K., Lopes, S. M., Tang, F. & Surani, M. A. Cell Stem
pyrite grew in an organic framework.
ing the experimental methodology and models Cell 3, 391–401 (2008). However, it is the fossil morphology, revealed
used. One point that should always be borne 8. Chang, H. H. et al. Nature 453, 544–547 (2008). through X-ray microtomography, that provides
the most compelling evidence of biogenicity.
The fossils take the form of three-dimensional
sheets with scalloped margins defined by radial
EARLY LIFE slits and an internal radial fabric (as seen in

Origins of multicellularity
Fig. 4 of the paper on page 102). The authors
interpret this structure as evidence of coordi-
nated growth. The relative complexity of the
fossils, and their co-occurrence with solu-
Philip C. J. Donoghue and Jonathan B. Antcliffe ble organic compounds containing sterane
Interpreting truly ancient fossils is an especially tricky business. (derived from sterol precursors), lead El Albani
and colleagues to conclude that they are unlike
The conclusion that 2.1-billion-year-old structures from Gabon are the any living bacterium. They don’t rule out the
remains of large colonial organisms will get palaeobiologists talking. possibility that these remains might even rep-
resent the earliest multicellular eukaryotes
It is a peculiar but widely held view that modern levels. This bacterial world was under- — that is, colonies of organisms with a mem-
Charles Darwin used the palaeontological going the greatest episode of climate change in brane-bound nucleus that represent a distinct,
record as one of the principal lines of evi- the history of the planet: pumping out oxygen, more complex form of life from bacteria.
dence for biological evolution. He did not. To drawing down carbon dioxide, slowly trans- The null hypothesis, however, has to be
modern eyes, On the Origin of Species presents forming the Earth into the world we know. that these remains represent bacterial colo-
a shocking account of the fossil record as an The fossils are not much to look at — yet, nies. Future work must determine whether
archive of evolutionary history. given their antiquity, the fact that they can be the sterane signature, a hallmark of eukary-
For instance, Darwin highlights the idea seen by the naked eye is astonishing. Out of otes, is derived from soluble organics gener-
that the then earliest-known fossil-bearing their geological context, these structures are ated within the sediments, or whether they
rocks, from the Cambrian period, beginning unremarkable and would probably have been migrated into these sediments from younger
about 542 million years ago, contain records ignored. But the geological context is as remark- rock sequences.
of modern groups — implying an extensive able as are the potential evolutionary implica- In the interim, however, is there anything in
prehistory teeming with life. One-and-a-half tions. El Albani and colleagues1 have, therefore, the geometry of these presumed multicellular
centuries of subsequent research have revealed been painstaking in attempting to demonstrate sheets that sets them apart from bacterial
a vast microscopic fossil record of unicellular the age and biological nature of the structures. colonies? Multicellularity represents one
protists and bacteria extending, some would
argue, as far back as there are sedimentary Hitherto oldest record of macroscopic multicellularity
rocks from which they could be recovered. But Gabon fossils Probable eukaryotic fossils
although fossils of millimetre- to metre-scale Oldest certain bacterial fossils Ediacara biota
multicellular organisms characterize the 90 mil- Earliest stromatolites Oldest unequivocal animal fossils
lion years of the Ediacaran period that precedes 0
Atmosphere
Oxygen level (log scale)

the Cambrian, pre-Ediacaran macroscopic

fossils are exceedingly rare. Great Oxidation –2
Event
Hence the excitement that will surround
the report on page 100 of this issue1, in which –4
El Albani and colleagues describe macroscopic
fossils from 2.1-billion-year-old rocks from –6
southeastern Gabon. These centimetre-scale
fossils, which the authors interpret as repre- Oceans
sentatives of multicellular organisms, push Anoxic Sulphidic Oxic
the record of macroscopic life back more than 4 3 2 1 0
200 million years into the Palaeoproterozoic, Time (billions of years ago)
an interval of Earth’s history so alien to us that
Figure 1 | Evidence of early life, and the chemical state of the atmosphere and oceans. The earliest
we would not be able to breathe the air. Even
stromatolites date to 3.4 billion years ago. But truly macroscopic multicellular organisms, such as
though these fossils postdate the Great Oxida- the putative examples from Gabon described by El Albani et al.1, do not appear until after the Great
tion Event — the notable rise in atmospheric Oxidation Event, 2.4 billion years ago, when atmospheric oxygen reached a few per cent of present
oxygen 2.4 billion years ago — the atmosphere levels. That timing fits in with speculation that the rise in oxygen levels is linked causally to the
was still a toxic mix of greenhouse gases, with evolutionary origin of eukaryotes5 and multicellularity6. (Figure modified from ref. 7. The scale for
oxygen making up only a few per cent of atmospheric oxygen is the log of the partial pressure of O2 expressed in atmospheres.)
41
© 2010 Macmillan Publishers Limited. All rights reserved
NEWS & VIEWS NATURE|Vol 466|1 July 2010

of the principal thresholds in evolutionary than it answers. But within the confines of a change in free energy as an enzyme and inhibi-
history2. This threshold has been exceeded very patchy global record of Proterozoic and tor are brought together is difficult, owing to
tens of times3, perhaps because much of the Archaean rocks, which extend from about the flexibility of the two components and the
requisite molecular machinery to facilitate 3.8 billion years ago to the beginning of the changes in their hydration (the arrangement
cell–cell coordination is a shared primitive Cambrian, these remains contribute to a fossil of water molecules around them) that occur
feature of living organisms4, but also because record that belies the dated caricature painted on binding2. So alternative approaches are
some definitions of multicellularity encompass in the Origin. It was Darwin’s view that absence being explored. In a popular one, known as
everything from simple bacterial colonies to of organisms in these early intervals of Earth’s double decoupling, the free-energy changes
badgers. Stricter definitions of multicellularity history would prove his theory of biological associated with making the inhibitor ‘disap-
are met in far fewer instances. Although the evolution wrong. The discovery and continu- pear’ from the complex with the enzyme and
fossils are macroscopic, they do not seem to ing elucidation of the Precambrian fossil record for making the inhibitor disappear in bulk
represent anything other than the basic type of has met Darwin’s predictions on the extent and water are computed. The difference in these
multicellularity, which occurs earlier in time in structure of evolutionary history. ■ results yields ΔGb. However, such calculations
the form of stromatolites. Philip C. J. Donoghue and Jonathan B. Antcliffe are still computationally taxing and are often
Nevertheless, the size, form and thickness are in the Department of Earth Sciences, imprecise for compounds as large as typical
of these sheets may be significant. It has been University of Bristol, Wills Memorial Building, drugs. It is easier to compute the differences
argued that the paucity of ancient records of Queen’s Road, Bristol BS8 1RJ, UK. in ΔGb for structurally similar inhibitors, and
macroscopic multicellularity could reflect the e-mail: phil.donoghue@bristol.ac.uk this approach has been used with much suc-
physiological limitations enforced by atmos- cess in guiding the discovery of, for example,
pheric and oceanic chemistry during the first 1. El Albani, A. et al. Nature 466, 100–104 (2010). potent anti-HIV agents3. Nevertheless, there is
2. Maynard Smith, J. & Szathmáry, E. The Major Transitions in
2 billion years of Earth’s history5,6. The proxim- Evolution (Oxford Univ. Press, 1997).
a pressing need for new methods that address
ity in the age of these fossils to the timing of 3. Butterfield, N. J. Precambr. Res. 173, 201–211 (2009). the binding problem.
the Great Oxidation Event (Fig. 1) fits elegantly 4. Shapiro, J. A. Annu. Rev. Microbiol. 52, 81–104 (1998). It is against this backdrop that Colizzi et al.1
with speculative hypotheses on the co-evolu- 5. Anbar, A. D. & Knoll, A. H. Science 297, 1137–1142 (2002). report their new approach for gaining insight
6. Bengtson, S., Rasmussen, B. & Krapez, B. Paleobiology 33,
tion of life and the chemistry of the oceans. 351–381 (2007). into protein–ligand binding affinities. Mimick-
This latest discovery raises more questions 7. Anbar, A. D. Science 322, 1481–1483 (2008). ing single-molecule pulling experiments and
simulations4, they used ‘steered’ molecular
dynamics (SMD) to compute the force that is
required to extract inhibitors from complexes
DRUG DISCOvERY with enzymes. To model each extraction,

Pulled from a protein’s embrace

the authors carried out molecular-dynamics
simulations to predict what would happen if
a harmonic spring was attached to the inhibi-
tor and then pulled at constant velocity into
William L. Jorgensen the surrounding water (Fig. 1a). The force that
It is hard to predict how strongly a small molecule will bind to a protein, but was applied to the inhibitor was determined
from the extension of the spring and recorded
this is a crucial goal of computer-aided drug discovery. A new approach as a function of time, producing a force profile
models the forcible removal of molecules from a protein’s active site. (Fig. 1b). Strongly bound inhibitors yield pro-
files with higher peak forces, whereas the force
Most drugs are small ligand molecules that out, whereby an inhibitor is constructed to be profiles of weaker inhibitors are flatter.
bind to a protein. Some disrupt protein–protein complementary to the enzyme’s active site. The To illustrate their approach, Colizzi et al.
interactions, whereas many others — enzyme minimum requirements are the target’s struc- investigated the binding of five structurally
inhibitors — inhibit reactions carried out by ture and computer-graphics tools that enable related compounds to the protein FabZ from
proteins. Computational methods for accu- one to build and examine how molecules fit the malaria-causing parasite Plasmodium fal-
rately predicting the binding affinity of a into the active site. Additional insight provided ciparum, an enzyme that is a potential target
ligand for a protein are therefore central to by evaluating the molecular energetics of the for antimalarial drugs5. The tested compounds
rational drug design. Reporting in the Journal binding process is, however, crucial to most included two (luteolin and myricetin) that
of the American Chemical Society, Colizzi et al.1 current activities in structure-based design. inhibit FabZ and three that do not. The authors
describe a stimulating advance in this field: a So what is the goal of structure-based design? began by computationally generating putative
computational method for predicting protein– Simply, it is to assist in the discovery of safe and structures for the FabZ–luteolin complex by
ligand binding affinities that works by model- effective drugs. To achieve this, it is essential to using protein–ligand docking methods. Two
ling the force that is required to pull inhibitors design inhibitors that bind tightly to active sites alternative binding orientations for luteolin
out of their complexes with proteins. (with a large, negative ‘free energy of binding’, emerged, but the authors’ SMD results demon-
Traditionally, drugs were discovered by trial ΔGb, to use the thermodynamics jargon). This strated that one of these modes was preferred.
and error. Powders made from willow bark, helps the inhibitors to exert their biological They then used the preferred mode in SMD
for example, have been known to be useful effect at low doses. It also promotes selectiv- calculations to generate force profiles for the
for treating headaches, pains and fevers since ity for the target enzyme, which helps prevent five compounds.
the time of Hippocrates. The active ingredi- undesirable side effects that might arise if The resultant profiles for luteolin and myri-
ent was eventually isolated in the 1800s, and other enzymes are inhibited. The development cetin differed qualitatively from those for the
a close analogue became aspirin in 1899. of methods for accurately computing ΔGb is three inactive analogues, with peak forces
Drug discovery evolved to be increasingly therefore an active area of research. that were twice as large as those of the inac-
deliberate and, with the advent of structural The most sophisticated approaches for tive compounds — as would be expected if the
biology in the 1960s, the rational design of calculating ΔGb use molecular dynamics or inhibitors bind to FabZ more strongly than
enzyme inhibitors was made possible. Given ‘Monte Carlo’ statistical mechanics to model the non-inhibitors. Emboldened by this suc-
the three-dimensional structure of a target the enzyme and the inhibitor in water, with all cess, the authors went on to use their method
enzyme, ‘structure-based design’ can be carried atoms represented. A direct calculation of the to predict the activity of another structurally
42
© 2010 Macmillan Publishers Limited. All rights reserved