3ODQW(PEU\RJHQHVLV=\JRWHWR6HHG

$XWKRUV5REHUW%*ROGEHUJ*HQDURGH3DLYDDQG5DPLQ<DGHJDUL
6RXUFH6FLHQFH1HZ6HULHV9RO1R2FWSS
3XEOLVKHGE\$PHULFDQ$VVRFLDWLRQIRUWKH$GYDQFHPHQWRI6FLHQFH
6WDEOH85/http://www.jstor.org/stable/2885440 .
$FFHVVHG

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .
http://www.jstor.org/page/info/about/policies/terms.jsp

.
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of
content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms
of scholarship. For more information about JSTOR, please contact support@jstor.org.

American Association for the Advancement of Science is collaborating with JSTOR to digitize, preserve and
extend access to Science.

http://www.jstor.org

This content downloaded from 164.67.109.135 on Fri, 18 Jul 2014 20:08:48 PM

All use subject to JSTOR Terms and Conditions
 m
.......... ARTICLES

Most morphogeneticevents in flowering

Plant Embryogenesis: plants occur in the postembryonicsporo-
phyte after seed germination(Fig. 1) (2).
Zygote to Seed Vegetativeorgansystemsdifferentiatecon-
tinuouslyfromroot and shoot meristematic
regionsthat areformedinitiallyduringem-
Robert B. Goldberg,*Genaro de Paiva, RaminYadegari bryogenesis.The reproductiveorgansof the
flower are differentiated from a repro-
Mostdifferentiationevents in higherplantsoccur continuouslyinthe postembryonicadult grammedshoot meristemafterthe seedling
phase of the life cycle. Embryogenesisin plants, therefore, is concerned primarilywith has become a matureplant (Fig. 1) (25).
establishingthe basic shoot-root body patternof the plant and accumulatingfood re- Thus, a germlineanalogousto that foundin
serves that willbe used by the germinatingseedling aftera periodof embryonicdormancy animals (1) is not set aside during plant
withinthe seed. Recent genetics studies inArabidopsishave identifiedgenes that provide embryogenesis.
new insightinto how embryos formduringplantdevelopment.These studies, and others A maturefloweringplant embryocon-
usingmolecularapproaches,are beginningto revealthe underlyingprocesses thatcontrol tains two primaryorgan systems-the axis
plant embryogenesis. and cotyledon (Fig. 1) (2). These organs
have distinct developmentalfates and are
both composedof threebasic,or primordial,
tissuelayers-protoderm,procambium,and
A majorproblemin plant developmentis lecularapproachessuggestthat a plant em- groundmeristem-which will become the
to unravelthe mechanismsoperatingdur- bryohas a modularstructureand consistsof epidermal,vascular,andparenchymatissues
ing embryogenesisthat enable a plant to severalregionsthat formautonomouslydur- of the youngseedling,respectively(2). The
specifyits body plan and tissuedifferentia- ing embryogenesis. axis, or hypocotyl-radicleregionof the em-
tion patterns.Althoughprogresswith a va- bryo,containsthe shoot and root meristems
rietyof animalsystemshas been spectacular Embryos Begin the Diploid Phase and will give rise to the matureplant after
in this regard(1), a detailedunderstanding of the Higher Plant Life Cycle seed germination(Fig. 1). By contrast,the
of the events that govern plant embryo cotyledon is a terminallydifferentiatedor-
formationhas yet to be realized.One obsta- The flowering plant life cycle is divided gan that accumulatesfood reservesthat are
cle in achievingthis goal is the location of into haploid and diploid generationsthat utilizedby the seedlingfor growthand de-
embryoswithin the plant and their relative are dependent on each other (Fig. 1) (2, velopmentbeforeit becomesphotosynthet-
inaccessibilityto experimentalmanipula- 16-18). The haploid, or gametophytic, ically active (Fig. 1). The cotyledon func-
tion, particularlyat the earlystagesof em- generationbegins aftermeiosiswith spores tions primarilyduringseedgerminationand
bryogenesis.In floweringplants, reproduc- that undergomitosis and differentiateinto senescesshortlyafter the seedling emerges
tive processesoccur within floral organs eithera pollen grain(malegametophyte)or from the soil. Embryogenesisin higher
(Fig. 1) (2). The egg cell is presentin the an embryo sac (female gametophyte) (3, plants,therefore,serves(i) to specifymeri-
ovule,a multicellularstructurethat is buried 19-20). The pollen grain contains two stems and the shoot-root plant body pat-
beneathseveralcell layersof the pistil, the sperm cells, whereasthe embryosac con- tern, (ii) to differentiatethe primaryplant
female reproductiveorgan (2-4). Because tains a single egg (Fig. 1). Other accessory tissue types, (iii) to generatea specialized
egg cell formation,fertilization,and embry- cells within the haploid male and female storageorganessentialfor seed germination
ogenesisoccurwithin the pistil, it has been gametophyteshelp facilitatethe pollination and seedlingdevelopment,and (iv) to en-
difficultto dissectthe majoreventsthat take and fertilizationprocesses(3, 19, 20). The able the sporophyteto lie dormantuntil
placeduringthe earlystagesof higherplant male and female gametophytesare derived conditionsare favorablefor postembryonic
development. fromspecializedspore-formingcells within development.
Recently, it has become feasibleto iso- the reproductiveorgansof the flower(3, 4,
late plant eggsand fertilizethem in vitro in 21). By contrast, the diploid, or sporo- The Shoot-Root Body Plan Is
order to investigate the initial events of phytic, generationbegins afterfertilization Generated During Early
planteribryogenesis(5). In addition,genet- with the zygoteand formsthe matureplant Embryogenesis
ic approacheshave been used to identify with vegetative organs (leaf, stem, root)
genes requiredfor variousembryogenicpro- and flowers that contain the reproductive How the embryoacquiresits three-dimen-
cesses, includingpatternformation(6, 7). organs(antherand pistil) (Fig. 1). sional shape with specializedorgans and
Genetic manipulationof Arabidopsis thali- Two fertilizationevents occur in flower- tissues,and whatgene networksorchestrate
ana, by both chemical mutagenesis(8-12) ing plants (2, 22). One spermunites with embryonicdevelopmentremainmajorun-
and insertionalmutagenesis(13-15), has the egg cell to producea zygoteand initiate resolvedproblems.Froma descriptivepoint
identifieda large numberof mutantsthat embryogenesis. The otheruniteswith a spe- of view, plantembryogenesis can be divided
are blocked at differentstages of embryo- cializedcell within the embryosac (central into three generalphasesin which distinct
genesis.In this reviewwe outline the major cell) to initiate the differentiationof the developmentaland physiologicalevents oc-
insightsthat have been derivedfromstudies endosperm,a triploidtissue that is neither cur: (i) postfertilization-proembryo, (ii)
of Arabidopsis embryomutants,and we sum- gametophyticnorsporophyticin origin(Fig. globular-heart transition,and (iii) organex-
marizegene transcriptionexperimentsin 1) (23). The endospermis presentduring pansion and maturation(26-28) (Fig. 2
other plants that providenew information seeddevelopmentandprovidesnutrientsfor and Table 1). Although there is consider-
aboutthe processesregulatinghigherplant eitherthe developingembryo,the germinat- able variationin how embryosin different
embryogenesis.Both the genetic and mo- ing seedling,or both (23). Fertilizationalso plant taxa form(29), the overalltrendsare
causesthe ovule,containingthe embryoand remarkably similar(29). We summarizethe
Authors are in the Department of Biology, Universityof endosperm,to develop into a seed and the Capsella and Arabidopsis pattern of embryo
California,Los Angeles, CA 90024-1606, USA. ovary to differentiateinto a fruit, which development (29-33) because (i) it is one
*To whom correspondence should be addressed. facilitatesseed dispersal(Fig. 1) (24). of the most well-studied forms of plant em-

SCIENCE * VOL.266 * 28 OCTOBER1994 605

This content downloaded from 164.67.109.135 on Fri, 18 Jul 2014 20:08:48 PM

All use subject to JSTOR Terms and Conditions
Fig. 1. The life cycle of a 1 Pollen
flowering plant with em- Egg cell formation, grains
phasis on egg cell forma- pollination,and fertilization Stigma permcells
tion and seed develop-
-ube
ment. [Adapted from (16, Pistil Megaspore Surviving Embryo Female Style nucleus
26).J mothercell megaspore sac gametophyte(n) Male
(2n) (n) \p ntipodal gametoht
Nucellus CCentral cells gametophyte
/ > n~~~~Meiosis M itosis { cell XO (nr)fri
-
Ovule
Me~~~~~~~sis F~~~~~~~~J
~~~~ ~ Ovary-fruit
rynergid
1I
Flower Funiculus Integuments ~~~~
Flower Petal Egg cells ~
~~~~~~~~~~~~~~~~~~~~~cell
- Anther Micropyle //
Pistil
epal Germinating Dormant Developing Fertilizedendosperm
Seedling seed seed seed nucle s (3n)
Endosperm
aotyledo tyldo

meristemI11
LShoot
ootmeristem Root FFAA1Embryo Developing egg (2n)
CotyledonsJ embryo

Seed developmentand germination

bryogenesis,dating back to the classical the suspensor(Fig. 2). In Arabidopsis,the to be passedfrom the maternalsporophyte
studiesof Hanstein,Schaffner,and Soueges suspensorcontains only 7 to 10 cells (Fig. into the developing proembryo(Fig. 2)
with Capsella(30-32), (ii) it has an invari- 2). The suspensoranchorsthe embryoprop- (35). The suspensorsenescesafterthe heart
ant divisionpattem duringthe earlystages, er to the surroundingembryosac and ovule stage and is not a functional part of the
which allowscell lineagesto be tracedhis- tissue and servesas a conduitfor nutrients embryoin the matureseed. Derivativesof
tologically (33), and (iii) recent studies
with Arabidopsismutants have provided
Posffertilization Globular-heart transition
new insightsinto the processesthat control
embryodevelopment(6, 9).
Asymmetric Proembryo C
cleavageof dtezygoteresultsin Pd G
theformationof an embryowitha suspensor
andembryoproperthathavedistinctdevelop- MAT ~~~~JEP I:P
mentalfates. The zygote in Arabidopsis and
Capsellahas an asymmetricdistributionof
cellularcomponents-the nucleusandmost
of the cytoplasmare present in the upper Zygote2ce Transition Heart
2/4-P
E 8ce 3
portionof the cell, whereasa largevacuole EP 16-cell embryo embryo
dominatesthe middleto lowerportion(Fig. EP Globular
2). This spatialasymmetryis derivedfrom embryo
the egg cell (30). The zygotedividesasym-
Organ expansion and maturation
metricallyinto two distinct-sizeddaughter
cells-a small, upper terminal cell and a Torpedo embryo Walking-stick embryo Mature embryo
large, lower basal cell-which establish a CE s
polarizedlongitudinalaxis within the em- ~V)J~ \ SC
bryo (Fig.2) (2, 30-33). Histologicalstud-
En SC
ies over the course of the past 125 years
have indicatedthat the terminaland basal Pd
cells give rise to different regions of the
matureembryo(29-33). The small termi-
nal cell givesriseto the embryoproperthat
will form most of the matureembryo(Fig. ft RM RM
2). Cell lineagesderivedfromthe terminal
cell and embryoproperwill specifythe cot-
yledons, shoot meristem,hypocotylregion
Fig. 2. A generalized overview of plant embryogenesis. Schematic representations of embryonic stages
of the embryonicaxis (29-33), and partof
are based on light microscopy studciesof Arabidopsis (33) and Capsella (30-32) embryo development.
the radicle,or embryonicroot (Fig.2) (34). Torpedo and walking-stick refer to specific stages of embryogenesis in Arabidopsis and Capsella.
By contrast, the large basal cell derived Abbreviations:T, terminalcell; B, basal cell; EP, embryo proper; S, suspensor; Bc, suspensor basal cell;
from the lower portion of the zygote will Pd, protoderm; u, upper tier; 1,lower tier; Hs, hypophysis; Pc, procambium; Gm, ground meristem; C,
divide and forma highly specialized,termi- cotyledon; A, axis; MPE, micropylarend; CE, chalazal end; SC, seed coat; En, endosperm; SM, shoot
nally differentiatedembryonicorgancalled meristem; and RM, root meristem.

606 SCIENCE * VOL.266 * 28 OCTOBER1994

This content downloaded from 164.67.109.135 on Fri, 18 Jul 2014 20:08:48 PM

All use subject to JSTOR Terms and Conditions
 . ......
...... _U ARTICLES

Table 1. Majorevents of flowering plant embryo- layerof the embryoproper,is producedand Embryogenesis Can Occur
genesis. the hypophysisformsat the top of the sus- Without Surrounding
pensor(Fig. 2). Subsequentcell differentia- Maternal Tissue
Postfertilization-proembryo tion events within the embryoproperresult
Terminaland basal cell differentiation
Formationof suspensor and embryo proper
in the productionof an inner procambium It is unclearwhat influence, if any, mater-
tissue layer and a middle layer of ground nal tissue or accessorycells of the female
Globular-hearttransition
Differentiationof majortissue-type primordia meristemcells (Fig. 2). The spatialorgani- gametophytehave on egg cell formation
Establishment of radial(tissue-type) axis zationof protoderm,groundmeristem,and and subsequent embryonic development
Embryo proper becomes bilaterally procambiumlayersestablishesa radialaxis (Fig. 1). Forexample,do eitherthe ovule or
symmetrical of differentiatedtissueswithin the globular cells within the embryosac (for example,
Visible appearance of shoot-root embryo.By contrast, the presence of the synergids)produce morphogeneticfactors
(apical-basal) axis hypophysisat the basalend of the embryo that contributeto the establishmentof lon-
Initiationof cotyledon and axis
(hypocotyl-radicle)development proper establishesan apical-basalpolarity gitudinalasymmetrywithin the egg?Sever-
Differentiationof root meristem within this region and indicates that the al argumentssuggest that the maternal
Organ expansion and maturation globularembryois not radiallysymmetrical sporophyteprovidesonly physical support
Enlargement of cotyledons and axis by cell in the formalsense (Fig. 2). structuresandnutrientsforthe embryo(Fig.
division and expansion A dramaticchangein the morphologyof 1). First, somatic cells from a variety of
Differentiationof shoot meristem the embryo proper occurs just after the vegetativeand reproductivetissuescan un-
Formationof lipidand protein bodies globular stage. Cotyledons are specified dergoembryogenesisin cultureand lead to
Accumulation of storage proteins and lipids
Vacuolizationof cotyledon and axis cells
from two lateraldomainsat the apical end the productionof fertile plants (37-39).
Cessation of RNA and protein synthesis (top), the hypocotylregionof the axis be- Somatic embryos undergo developmental
Loss of water (dehydration) gins to elongate, and the root meristem events similar to those that occur within
Inhibitionof precocious germination becomesdifferentiatedfromthe hypophysis the embryo-proper regionof zygoticembry-
Dormancy regionat the basalend (bottom) (34). The os, exceptthat they do not becomedormant
embryoproperis now heart-shaped,has a (Fig. 2 and Table 1). In addition, spatial
bilateralsymmetry,and the body plan and and temporalgene expressionprogramsap-
the uppermost cell of the suspensor, the tissue layers of the mature embryo (and pear to be similar in somatic and zygotic
hypophysis (Fig. 2), contribute to the for- postembryonicplant)have been established embryos (39, 40). Second, embryo-like
mation of the root meristem (30-34). Thus, (Fig.2). Morphogeneticchangesduringthis structuresleadingto plantletscan formdi-
cell lineages derived from the basal cell give periodare mediatedby differentialcell di- rectly from the attached leaves of some
rise to the suspensor and part of the radicle vision and expansionratesandby asymmet- plants (41). Third,zygotesproducedby fer-
region of the embryonic axis (Fig. 2). ric cleavagesin differentcell planes(2). No tilizing egg cells in vitro undergoembryo-
Different gene sets must become active cell migrationoccurs, in contrast to the genesis in culture and give rise to flower-
in the terminal and basal cells after the migrationevents that take place in many producingplants(5). Finally,ultrastructural
division of the zygote. Whether the polar- typesof animalembryos(1). studies suggest that there is a barrierbe-
ized organization of the egg cell, the zygote, Embryogenesis terminateswitha dormancy tween the inner ovule cell layer and the
or both control differential gene expression period.A majorchangein embryonicdevel- embryo sac that prevents the transferof
events early in embryogenesis is not known. opment occursduringthe organexpansion material directly between these compart-
For example, do prelocalized regulatory fac- and maturationphase (Fig. 2). A switch ments (42). Thus, both zygoticand somatic
tors within the egg cell initiate a cascade of occurs during this period from a pattern embryogenesiscan occur in the absenceof
events leading to the lineage-dependent formationprogramto a storageproductac- surroundingovule tissue (Fig. 1).
differentiation of terminal and basal cell cumulationprogramin orderto preparethe The embryosac is necessaryfor zygotic
derivatives? Alternatively, after fertilization young sporophytefor embryonicdormancy embryogenesisbecause it contains the egg
does the zygotic genome direct the de novo and postembryonicdevelopment(Table 1). and associatedaccessorycells that are re-
synthesis of regulatory factors that are dis- The cotyledons and axis increase in size quiredfor fertilizationand endospermde-
tributed asymmetrically to the terminal and dramaticallyas a resultof cell division and velopment (Fig. 1). However, the embryo
basal cells at first cleavage? In either case, expansion events (33). Ground meristem sac is not essentialfor embryogenesisper se
these events would lead to the autonomous cells withinboth these organsbecomehigh- because(i) somaticembryosproducedfrom
specification of the terminal and basal cells ly specializedand accumulatelargeamounts sporophyticcells develop normally(5, 37-
as a consequence of intrinsic factors rather of storage proteins and oils that will be 40) and (ii) embryoscan be inducedto form
than extracellular signals (1). utilizedas a foodsourceby the seedlingafter from microsporesthat, under normal cir-
Embryonicorgansand tissue-typesdifferen- germination(Fig. 2 and Table 1) (33). One cumstances,give rise to pollen grains(43).
tiate duringthe globular-hearttransitionphase. differentiationevent does occurduringthis These resultssuggestthat normalembryo-
Two critical events must occur after the period,however-the shoot meristemforms genic processesdo not requirefactorspro-
embryo proper forms: (i) regions along the from cell layerslocalizedin the upperaxis ducedby either the femalegametophyteor
longitudinal apical-basal axis must differen- regionbetweenthe two cotyledons(Fig. 2) maternalsporophytictissue. This conclu-
tiate from each other and generate embry- (36). Thus, the differentiationof shoot and sion is supportedby the fact that the over-
onic organ systems, and (ii) the three pri- root meristemsat oppositepoles of the em- whelmingmajorityof mutationsthat alter
mordial tissue layers of the embryo need to bryonicaxisdoesnot occurat the sametime embryodevelopmentappearto be due to
be specified. Both of these events take place (34, 36). At the end of the organexpansion defects in zygoticallyacting genes (6-14,
during the globular-heart transition phase and maturation period the embryo has 44). It is possiblethat somaticcells have the
(Fig. 2 and Table 1). The embryo proper reachedits maximumsize, cells of the em- potential to produceputative maternalor
has a spherical shape during the proembryo bryo and surroundingseed layershave be- gametophyticfactorsunderthe propercon-
and globular stages (Fig. 2). The first visible come dehydrated, metabolic activities have ditions, or that somatic embryos specify
cell differentiation events occur at the 16- ceased, and a period of embryonic dormancy their longitudinal apical-basal and radial
cell stage when the protoderm, or outer cell within the seed begins (26-28, 33). tissue-type axes by different mechanisms

SCIENCE * VOL. 266 * 28 OCTOBER 1994 607

This content downloaded from 164.67.109.135 on Fri, 18 Jul 2014 20:08:48 PM

All use subject to JSTOR Terms and Conditions
than zygoticembryos.However,mostof the bryos containing chimeric ,B-glucuronidasegions (Fig.3G). This resultsuggeststhat the
availabledatasuggestthat embryomorpho- (GUS) reportergenes driven by soybean preferentiallocalizationof Kti3 mRNA at
genesis and cell specification events are embryo-specific gene promoters showed the micropylarend of a soybean globular
directed primarilyby the zygotic genome that a globularembryo is organizedinto stage embryo(Fig. 3C) is due to transcrip-
after fertilizationoccurs. If so, then this distinct, nonoverlappingtranscriptionalre- tional regulatoryprocesses. By contrast,
woulddiffersignificantlyfromthe situation gions,or territories(Fig.3, F to I) (47, 48). GUS activityis visibleas a uniformbluebelt
with many animalssuch as Drosophilaand Blue color resultingfromGUS enzymeac- that surroundsthe equatorregionof a glob-
sea urchin, in which matemally supplied tivity occurs specificallyat the micropylar ular embryo containing a soybean lectin/
factors influence the pattem of early em- end of a globularstageembryocontaininga GUS reportergene (Fig. 3H) (48). GUS
bryodevelopment(1). Moreextensivestud- KI3/GUS reportergene (Fig. 3, F and G) activityis not visibleat either the micropy-
ies with developing female gametophytes, (47). No GUS activity is observedwithin lar or the chalazal(apical)ends of the em-
egg cells, and zygoticembryosin the early the suspensoror other embryo-proper re- bryo proper (Fig. 3H) (48); nor is there
stagesof embryogenesisare needed to clar-
ify this issue.
A ~~~B
A GlobularEmbryoContains
DifferentiallyTranscribedRegions ~CE
En.
A largenumberof genes are expresseddur-
ing embryogenesisin higher plants (26).
Although it is not knownhow manygenes
are necessary to programmorphogenetic
and tissuedifferentiationprocesses,approx-
imately 15,000 diversegenes are active in
the embryosof plantsas diverseas soybean
and cotton (26). Many of these genes are
expressedin specificcell types,regions,and
organsof the embryo(26, 40) and provide
usefulentrypointsto unravelthe molecular EP G -

mechanismsthat regulatecell- and region-

specific differentiationevents duringplant -~~~~~~~~~~~~~T
embryogenesis(1).
The axis regionof the embryodoes not
becomevisiblydistinctuntil the heartstage ------
%N;:i- G ifil:X ;0 0
(Figs.2 and 3, A and B). Localizationstud-
ies with a soybeanKunitztrypsininhibitor
mRNA, designatedas Kti3 (45), indicated,
however,that cells destinedto become the
axis are alreadyspecified at the globular
stage (40). Figure 3C shows that Kti3
mRNA accumulatesspecificallyat the bas- Fig. 3. Differential activitydunngearlyembryogenesis.(Aand B)Brght-fieldphotographs
transcriptional
al, or micropylar,end of a late-globular of developingsoybeanembryossectionedlongitudinally (40).(A)Lateglobularstage embryo.(B)Heart
stage soybeanembryo.No detectableKti3 stage embryo.(C to E) Localizationof the majorKunitztrypsininhibitormRNA(Kti3)in longitudinal
mRNA is present in other regions of the data are similarto those shown in (40).
sections of soybean embryos(40, 45). In situ hybridization
embryoproperor in the suspensor(Fig.3C) Photographsweretakenby dark-field microscopy.(C)Lateglobularstage embryo.(D)Lateglobularto
transition photographof a tobacco globularstage
stage embryo.(E)Heartstage embryo.(F)Bright-field
(40). In transitionand heartstageembryos, embryosectioned longitudinally. (G and H) Localizationof GUS activitywithintransgenictobacco
Kti3mRNA remainsdistributedasymmetri- globularstage embryos.Photographsof whole-mountembryosweretakenby bright-field microscopy
cally at the embryomicropylarend (Fig. 3, (G)Embryocontainsa chimericKti3/GUSgene (47). (H)Embryocontainsa chimericlectin/GUSgene
D and E) and is localizedspecificallywithin (48).(I)Localization of GUS mRNAwithina transgenictobacco globularstage embryosimilarto that
the groundmeristemcell layer (Fig. 2) of shown in (H)containinga lectin/GUSgene (48).Abbreviations: EP, embryoproper;S, suspensor;C,
the emerging embryonic axis (Fig. 3D) cotyledon;A, axis;En,endosperm;CE,chalazalend; MPE,micropylar end; and Pd, protoderm.
(40). No detectableKti3 mRNA is present
within the newly initiatedcotyledons(Fig.
3E). This resultdiffersfrom that obtained WTseed WTseedling emb 30/gnom monopteros gurke fackel
with the carrotEP2 lipid transferprotein
mRNA, which is localizeduniformlyin the
outer protodermcell layer that surrounds H ff~~~~HH
the embryoproperat the globularand heart
stages (46). Taken together, these results
indicate that cells along the longitudinal
apical-basalaxis of the embryoproperare Complete Complete * Apical
0 Basal
8) Central
0 Basal *0Apical 8) Central
alreadydifferentiatedfrom each other at
the globularstage and that earlyin embry- Fig. 4. Schematic representations of Arabidopsis pattem mutants [adapted from (9)].The green, yellow,
ogenesisdistinct gene sets are expressedin and orange colors delineate the apical, central, and basal regions, respectively. Strong (upper)and weak
differentembryonicregionsand cell types. (lower)gnom phenotypes are depicted (9, 15). Abbreviations: WT, wild-type; RM, root menstem; SM,
Transformation studieswith tobaccoem- shoot meristem; C, cotyledon; h, hypocotyl; and R, root.

608 SCIENCE * VOL.266 * 28 OCTOBER1994

This content downloaded from 164.67.109.135 on Fri, 18 Jul 2014 20:08:48 PM

All use subject to JSTOR Terms and Conditions
 9 a V>NSC 92929g9292<99
>8<899B09 2W ND...
....
gAagj#SW&HaSaga WgNaaag5-g9*9am,a,a...........................
f............................
aG,
<,
>,9
,,g,<
?........
-XO.;,,.s,,,*a
X>*g^s.

ARTICLES|

detectableGUS activitywithin the suspen- 48). These resultssuggestthat a prepattem cotyledonsand shoot meristem,the central
sor (Fig. 3H) (48). GUS mRNA localiza- of different transcriptionalregulatorydo- region consists of the hypocotyl,or upper
tion studieswith longitudinalglobularem- mains has been establishedin the globular axis,and the basalregionincludesthe lower
bryo sections indicatedthat the lectinpro- embryo before the morphogeneticevents axis, or radicle,and the root meristem(Fig.
moteractivityoccursspecificallywithin the that leadto differentiationof cotyledonand 4). These regions are derived ultimately
groundmeristemcell layer of the equator axisregionsat the heartstage(Figs.2 and3, from the terminal and basal cell lineages
region (Fig. 31) (48). A and B). Presumably,each transcriptional and are maintainedin the young seedling
These experimentsindicate that both domainsets in motion a cascadeof events (Figs.2 and 4). Can regionsalong the lon-
the longitudinalapical-basaland radialtis- leading to the differentiationof specific gitudinal axis develop independently of
sue-type axes of a globularembryoare par- embryoregionslater in embryogenesis. each other?If territoriesestablishedwithin
titioned into discrete transcriptional terri- the embryoproperof a globularstage em-
tories(1). The longitudinalaxis of the em- Hormones Affect Embryo bryo are specifiedautonomously,then the
bryopropercontainsat leastthreenonover- Morphogenesis loss or alterationof cells within a territory
lapping transcriptionalterritories:(i) the shouldnot affectthe developmentof a con-
chalazalregion,(ii) the equatorregion,and What physiologicalevents cause the em- tiguousregion (Figs. 2 and 3). Several ex-
(iii) the micropylarregion (Fig. 3, F to H). bryo properto initiate cotyledonsand be- perimentssuggestthat this is actuallythe
The suspensor represents an additional come bilaterallysymmetricduringthe glob- case-that is, a plant embryoforms from
transcriptionaldomainalong this axis (Fig. ular-hearttransitionphase(Fig. 2)? Several modules that develop independently of
3, G and H). Eachtissuelayerof the radial experimentsimplicatea class of plant hor- each other (9-12, 54-59).
embryo-proper axis also has a distinct tran- mones, the auxins, in this morphogenetic Patternmutationsdeletespecificembryonic
scriptionalprogram(Fig. 3, E, G, and I). process(49-52). Auxins,suchas indoleace- regions. Genetic studies have uncovered
Transcriptional activitywithin these layers, tic acid (IAA), are involvedin a numberof Arabidopsis mutationsthat alterthe organi-
however, appearsto be established in a plant activities,includingphoto- and grav- zationof the embryobody plan (Fig. 4 and
territory-specificmanner-that is, ground itropism, apical dominance, and vascular Table 2) (9, 10). Embyropattem mutations
meristemcells within the equator region cell differentiation(2). Embryosin plantsas werefoundthat delete (i) the apicalregion
activatepromotersdistinctfromthosewith- diverseas bean and pine synthesizeauxins (gurke),(ii) the centralregion(fackel),(iii)
in groundmeristemcells of the micropylar (49) and transportthem in a polarized, the central and basalregions(monopteros),
region,and vice versa (Fig. 3, E to I) (47, basipetaldirectionfromthe shoot meristem and (iv) the apical and basal regions
to root tip alongthe embryonicaxis (Fig.2) (emb30/gnom)(Fig. 4) (9). Other embryo
(50). The highest auxin levels occurat the pattern-forminggenes probablyexist;how-
Table 2. Examples of Arabidopsis mutants that globular stage of embryogenesis (49). ever, they have not been identifiedin the
have defects in embryo development. Several Agents that inhibit polarizedauxin trans- genetic screens carried out thus far (6,
hundred Arabidopsis embryo-defective mutants
have been identifiedby both chemical and T-DNA
port either block the transitionfrom the 8-10, 14, 44). Each mutant gene acts zy-
mutagenesis. Most of these mutants can be ob- globularto heart stage completely (51) or gotically,indicatingthat majorspecifiersof
tained from the Arabidopsis Biological Resource prevent the bilateral initiation of cotyle- the embryobody plan act afterfertilization
Center (arabidopsis+@osu.edu). dons at the top of the globularembryo(Fig. has occurred(Fig. 2) (9, 10). In addition,
2) (52). Forexample,auxintransportinhib- the loss of a specificregion,or combination
Mutant class References itors cause carrot somatic embryosto re- of regions,does not affectthe development
main sphericallyshaped and develop into of an adjacentneighbor(Fig.4) (9, 10). For
Pattern mutants
emb3O/gnom 8-11, 15
giant globularembryos(51). By contrast, example,gurkeembryoslack an apical re-
monopteros 9, 12 zygotic embryos of the Indian mustard gion, but have normal central and basal
gurke 9 (Brassica juncea),an Arabidopsisrelative,fail regions(Fig.4) (9). monopteros embryos,by
fackel 9 to initiate two laterallypositionedcotyle- contrast,have a normal apical region but
Cell-type differentiationmutants dons when treated with auxin transport are missing the central and basal regions
keule 9 blockersin culture (52). A cotyledon-like (Fig. 4) (9).
knolle 9 organ does form, but as a collar-likering Embryopattem mutationsalter the di-
Suspensor transformationmutants aroundthe entire upper(apical) region of vision planesthat areestablishedduringthe
twvin 73 the embryo (52). Treated Indian mustard postfertilization-proembryo phaseof embry-
sus1 67
sus2 67 embryosresemblethose of the Arabidopsis ogenesis (Fig. 2) (11, 12). Thus, deletions
sus3 67 pini-1 mutant,which has a defect in polar- of mature embryo regions can be traced
raspberryl 44 izedauxin transport(52, 53). These results back to histologicaldefects at the proem-
raspberry2 44 suggest that auxin asymmetriesare estab- bryo and globular stages (Fig. 2) (1 1,
Meristem differentiationand identity mutants lished within the embryo-proper region of 12)-a resultthat providesfunctionalevi-
shoot meristemless 36 globularstageembryos(Fig.3, A andF) and dence for the embryofate mapsproposedby
embryonicflower 60 that these asymmetriescontribute to the the developmental botanists of the late
short-root 61, 62 establishmentof bilateralsymmetryat the 19th and early20th centuries(31, 32). For
hobbit 61, 62
heart stage (Figs.2 and 3, A and B). example,emb30/gnom zygotesdo not divide
Maturationprogram mutants
lec1- 1/lec1-2 44, 76-78 asymmetrically(Fig. 2) ( 11). Two similar-
Iec2 77 Plant Embryos Form from sized daughtercells are producedin the
fus3 79, 80 Regions That Develop emb30/gnomembryo-properregion instead
abi3 84, 85 Autonomously of the unequal-sized terminalandbasalcells
Seedling lethalitymutants that are found in wild-typeembryos(11).
fus1/cop1 91, 94 The longitudinal axis of a mature plant Laterdivisionevents in emb30/gnom embry-
fus2/det1 91, 95 embryo is made up of several regions that os are also highly variable and abnormal
fus6/copl11 91, 96 (11). By contrast, the monopterosdivision
fus7/cop9 91, 97 are designated as apical, central, and basal
(Fig. 4) (9). The apical region contains the pattern is normal until the 8-cell embryo-

SCIENCE * VOL. 266 * 28 OCTOBER 1994 609

This content downloaded from 164.67.109.135 on Fri, 18 Jul 2014 20:08:48 PM

All use subject to JSTOR Terms and Conditions
proper stage (Fig. 2) (12). During the there are genes which regulatethe specifi- to the cotyledonsand shoot meristemof a
monopteros globular-hearttransitionphase, cation, organization,and fate of meristems transgenictobacco embryo(Fig. 5D) (54).
lower tier cells of the embryoproperand that differentiate during embryogenesis. These data, and those of others (55-59),
derivativesof the hypophysisundergoab- Genes controlling meristem development indicatethatuniquetranscription factorsare
normal divisions (12) (Fig. 2). monopteros most likely act downstreamof embryo-re- active within each embryonicregion and
upper tier embryo-propercells, however, gion specifiers,such as monopteros andgurke that these factorsinteractwith specificpro-
develop normally-that is, defects at the (Fig. 4) (9, 10, 12), in the regulatoryhier- moter elements.The combinationof these
basalend of the embryoproperdo not affect archyneeded to forma plant embryo. elementsand factorsgives rise to the tran-
events that occurat the apicalend (Fig. 2) Meristemmutantscharacterizedto date scriptionalpattem of the whole embryo
(12). The resultis a mutantembryothat has indicatethat the shoot and root meristems (Fig. 5A). Identificationof transcription
cotyledons and a shoot meristem, but is function autonomously-that is, they do factors that interact with region-specific
missingthe hypocotyland rootregions(Fig. not affect the differentiationof contiguous DNA elementsshouldprovidereverseentry
4) (12). These observationsindicate that domains such as the cotyledons or hypo- into the independentregulatorynetworks
geneswhich areresponsible,in part,for the cotyl (36, 60-62). Meristems,therefore, requiredfor specifyingeach autonomousre-
establishmentof the embryobody plan di- representindependentsubmoduleswithin gion of a plant embryo.
rect territory-specific cell division patterns the apical and basalregionsof the embryo
duringearlyembryogenesis.Abnormaldivi- (Fig.4). A majorquestionis what effect, if Cell Differentiation and
sions in one embryonic territorydo not any, do cells adjacentto the shoot and root Morphogenesis Can Be
affect the division pattern of an adjacent meristems have on their development? Uncoupled in Plant Embryos
territory-that is, cell lineagesgivingriseto That is, arecell signalingevents involvedin
specific regions of the matureembryode- specifying the root and shoot meristems What is the relationbetweencell differen-
velop autonomously. within a specific embryonicregion, or do tiation and morphogenesisin plant embry-
AgrobacteriumT-DNA-tagged emb30/ the meristemsdifferentiateautonomously? os? Are processesrequiredfor tissue differ-
gnomalleleshave led to the isolationof the The latememutantprovidesone answerto entiation along the embryoradialaxis cou-
EMB30/GNOMgene (15). This gene en- this question (9). laterneseedlings lack pled to those that specify autonomousre-
codes a protein that is relatedto the yeast cotyledons but produce leaves indicating
Sec7 secretoryprotein,is active throughout that a shoot meristemis present(9). Thus,
the plant life cycle, and is involved in cell cotyledons are not requiredfor the differ-
division, elongation, and adhesion events entiation of the shoot meristem during
requiredat many stagesof sporophyticde- embryogenesis.
velopment, including embryogenesis(15). Promoter elermentsinterpretembryoregion-
Thus, EMB30/GNOMdoes not appearto specificregulatorynetwork1s.One consequence
establishthe embryoniccell division pat- of the modularorganizationof a plant em-
tern directly,but most likely facilitates a bryois that geneswhich areactive through-
patternset by other genes.What these pat- out the embryomust intersectwith several
tern-forminggenes are and how they inter- region-specificregulatorynetworks-that is,
act with downstreamgenes that mediate the promotersof embryo-specific genes are
events requiredfor the differentiationof requiredto senseand interpretthe transcrip-
autonomousregionsalong the longitudinal tional regulatorymachineryuniqueto each
axis remainto be determined. autonomousregion (Fig. 4). For example,
Mutationsaffectmeristernatic zonesof the Kti3 mRNA accumulateswithin the axis
embryolongitudinal axis. Arabidopsis muta- regionearly in soybeanembryogenesis, but
tions have been identified that affect the does not accumulatewithin the cotyledons
differentiationof the shoot and root meri- until muchlater(Fig.3, C to E) (40). Thus,
stemsduringembryogenesis(Table 2) (36, discretepromoterelementsshouldexist that
60-62). These mutationstarget a specific areresponsibleforinteractingwith transcrip-
meristemand have no other effect on em- tion factorsproducedby separateregulatory
bryonic development. For example, shoot circuits.
meristemless fails to differentiatea shoot A Kti3/GUSgene with 2 kb of 5' flank-
meristem during embryogenesisand pro- ing sequenceis transcribed in all regionsof a Fig. 5. DNA elements program transcrption to
ducesseedlingswithout leaves (36). embry- maturetransgenictobaccoembryo(Fig.5A) specific embryonic regions. Maturationstage em-
bryos were hand-dissected from transgenic to-
onicflower,on the other hand, generatesa (54). Deletion of 0.2 kb from the 5' end bacco seeds containing the E coi GUS gene
shoot meristemat the top of the embryonic eliminatesKti3/GUStranscriptional activity fused with differentsoybean seed protein gene 5'
axis (Figs. 2 and 4) (54). Remarkably,em- within the embryoradicleregion (Fig. 5B) regions (54). GUS activity in whole-mount embry-
bryonicflower seedlings produce flowers (54). Deletion of another 1 kb eliminates os was localized as outlined in (47), and photo-
ratherthan leaves, indicatingthat the fate Kti3/GUStranscriptionwithin the cotyle- graphs were taken with the use of dark-field mi-
of the shoot meristemis alteredduringem- dons and shoot meristem,but still permits croscopy. (A to C) Kti3 gene upstream regions
bryogenesis-a floral meristemis specified transcriptionto occurwithin the hypocotyl fused with the E coli GUS gene. (A) A 2-kb Kti3
rather than a vegetative shoot meristem region(Fig.5C) (54). These resultsindicate gene 5' fragment. (B) A 1.7-kb Kti3 gene 5' frag-
(60). By contrast,shortrootandhobbitaffect that discretecis-actingdomainsarerequired ment. (C)A 0.8-kb Kti3 gene 5' CaM\//GUS frag-
ment. (D) A minimal CaMV/GUS gene-promoter
root meristemdevelopment(61, 62). These for the transcriptional activationof the Kti3 cassette (59) fused witha 0.36-kb Gyl gene 5'
mutationslead to abnormalroot develop- gene within the radicle,hypocotyl,andcot- region(-446 to -84) (63).The whiteembryore-
ment after seed germination, indicating yledon-shoot meristem regions of the em- sults fromthe segregationof a single Gy1/GUS
that the root meristemis altered,but not bryo. Promoter analysis of the soybean Gyl gene withinthis transgenictobacco lineand rep-
eliminated,duringembryogenesis(61, 62). storage protein gene (63) also uncovered a resents a negativecontrol.Abbreviations:
C, cot-
Taken together,these resultsindicatethat regulatorydomain that directs transcription yledon;H, hypocotyl;and Rd, radicle.
610 SCIENCE * VOL.266 * 28 OCTOBER1994

This content downloaded from 164.67.109.135 on Fri, 18 Jul 2014 20:08:48 PM

All use subject to JSTOR Terms and Conditions
 ~~ ARTICLES

gions of the longitudinalapical-basalaxis genesis.Forexample,EP2lipidtransferpro- (Table 2). Surprisingly,raspberry 1 embryos

(Fig. 2)? Studiesof Arabidopsisembryopat- tein mRNA accumulatesspecificallywithin accumulateEP2and 2S2 markermRNAsin
tern mutants suggest that these processes the epidermalcell layer (Fig. 6B) (46, 64, theircorrectspatialcontext alongthe radial
are not necessarilyinterconnected.For ex- 65), and 2S2 albuminmRNA accumulates axis of both the embryo-proper and suspen-
ample, a fackel embryo does not have a within storageparenchymacells (Fig. 6C) sorregions(Fig.6, F to I) (64). EP2mRNA
hypocotyl,but epidermal,groundmeristem, (64, 66). Neither mRNA is detectable accumulatesalong the outer perimeterof
andvasculartissuesdifferentiatewithincot- within the vascularlayer(Fig. 6, B and C) raspberry 1 embryos(Fig.6, F andG), where-
yledon and radicle regions (Fig. 4) (9). (46, 64-66). Collectively,the EP2and 2S2 as 2S2 mnRNAaccumulateswithin interior
Thus, the lossof one embryonicregiondoes mRNAscan identifyembryoepidermaland cells (Fig.6, H and I) (64). By contrast,EP2
not affect the formation of tissue layers storageparenchymacell layersand, by de- and 2S2 mRNAsdo not accumulatedetect-
within the remainingregions (Fig. 4) (9- fault, the inner vasculartissueas well (Fig. ably within the central core of raspberry)
12). A more direct question, however, is 6, B and C). embryos(Fig.6, F to I) (64). Similarresults
whethermutantembryosthat arrestearlyin An Arabidopsisembryo mutant, desig- wereobtainedwith raspberry2 embryos(Ta-
embryonicdevelopmentand remainglobu- nated raspberry1, was identifiedin a screen ble 2) (44, 64).
lar-shapeddifferentiatethe specializedcell of T-DNA-mutagenizedArabidopsislines These mRNA localizationstudies indi-
and tissue layers that are found in organ (Table2) (44). This mutantfailsto undergo cate that specializedtissuescan differentiate
systemsof a mature,wild-typeembryo. the globular-heart transition(Fig.2), has an within the embryo-proper regionof mutant
A maturationstage Arabidopsis embryo embryo-proper region that remainsglobu- embryosthat remainglobularshaped,and
has specializedepidermal,storageparenchy- lar-shapedthroughoutembryogenesis,and that thesetissuesformin theircorrectspatial
ma, and vascularcell layerswithin both the does not differentiatecotyledonsand axis contexts.A similarconclusionwas inferred
cotyledonand axisregions(Fig.6A). These (Fig. 6, D and E) (64). raspberrylembryos fromhistologicalstudiesof the sus mutants
tissues are derivedfrom the three primary alsohave an enlargedsuspensorregion(Fig. (67). Tissue differentiation,therefore,can
cell layersthat arespecifiedalongthe radial 6, D and E) (64). raspberry2 (44) and sus take place independentlyof morphogenesis
axis of a globularembryo(Fig. 2), and ex- (67) embryo-defectivemutants also have in a higherplantembryo,implyingthatmor-
pressspecificmarkergenes late in embryo- phenotypes similar to that of raspberyl phogeneticcheckpointsdo not occurbefore
cell differentiationevents can proceed(68).
It does not follow, however,that morpho-
genesiscan occurwithoutpropercell differ-
entiation events. Arabidopsis embryo mu-
tants that alter tissue-specificationpatterns
have abnormalmorphologies(Table 2) (9).
Forexample,knolleembryoslack an epider-
mal cell layer and have a round,ball-like
shape without defined apical and basal re-
gions(Table2) (9). Similarly,the carrottsl 1
somaticembryomutanthas a defectivepro-
todermcell layerand fails to undergomor-
phogenesis(38, 39). Additionof either an
extracellularchitinase (69) or Rhizobium
nodulation factors (lipooligosaccharides)
(70) can rescuethe tsl 1 mutant.Lipooligo-
saccharidenodulation factors have been
4~4 shownto be signalmoleculesinvolvedin the
differentiationof Rhizobium-induced root
nodules (70). This suggeststhat in carrot
somatic embryos,the protodermcell layer
may providesignalsnecessaryfor embryo-
~~~~~~~~~~~....
-~~~~~~~-..--
genesis to occur (69, 70). Taken together,
experimentswith mutantembryosthat have
defectivecell layerssuggestthatspecification
of the radialaxis needs to occurin orderfor
Wt@=~~~~~~~~~~~k a normalshoot-rootaxis to form.An impor-
Fig. 6. Localization of cell-specific mRNAs wHthinwild-type and mutant Arabidopsis embryos. (A) tant corollaryis that cells within the radial
Brght-field photograph of a late-maturation stage Arabidopsis embryo sectioned longitudinally(64). (B axis probablyinteractwith each other (9).
and C) In situ hybndizationof labeled RNA probes with longitudinalsections of Arabidopsis maturation
stage embryos (64). Photographs taken by dark-field microscopy. (B) Localization of Arabidopsis EP2 Suspensor Cells Have the
lipid transfer protein mRNA (65). (C) Localization of 2S2 albumin mRNA (66). (D and E) Arabidopsis Potential to Generate an Embryo
raspberryl embryos (44). Embryos were harvested at a stage when wild-type embryos withinthe same
silique were in late maturation[as in (A)].(D) Longitudinalsection of a raspbenyl embryo. The photograph One intriguingaspectof the raspberry) em-
was taken by bright-fieldmicroscopy. (E)Araspberryl embryo photographed with Nomarski interference bryois its largesuspensor(Fig.6, D and E).
optics. (F to 1).In situ hybridizationof labeled RNA probes with raspberryl embryo longitudinalsections
(64). (F and G) Localization of EP2 lipid transfer protein mRNA (65). Localization experiments with
raspberry]suspensorsare indistinguishable
raspberry2 embryos, which have larger suspensors than raspberTyl embryos (44), confirmed that EP2 from wild-type during the early stages of
mRNA is present only in the suspensor outer cell layer (64). (H and 1)Localizationof 2S2 albumin mRNA embryogenesis(Fig. 2) (64). Laterin seed
(66). Insitu hybridizationwith serial sections through an entire raspberryl embryo confirmed that there is development,when neighboringwild-type
an inner core of cells with no detectable 2S2 mRNA (64). Abbreviations:A, axis; V, vascular tissue; Ed, embryosundergomaturation,cell prolifera-
epidermis; C, cotyledon; P, storage parenchyma; Ep, embryo proper; S, suspensor; and En, endosperm. tion events cause the raspbery)suspensor
SCIENCE * VOL.266 * 28 OCTOBER1994 611

This content downloaded from 164.67.109.135 on Fri, 18 Jul 2014 20:08:48 PM

All use subject to JSTOR Terms and Conditions
to enlargeat its basalend (Fig.6, D and E) Fig. 7. An Arabidopsis leafy cotyle-
(64). EP2 and 2S2 mRNAs (46, 65, 66) don seedling. An allele of the lecl
accumulatein the raspberry 1 suspensor(Fig. gene (76, 77), designated as lec1-2,
6, F to I) with a spatialpattem similarto was identifiedin a screen of T-DNA-
mutagenized Arabidopsis lines (44,
that which occursin mature,wild-typeem- __ |78). (A and B) Bright-field photo-
bryos (Fig. 6, B and C) (64). These cell- graphs of Arabidopsis seedlings
specificmRNAs do not accumulatedetect- a | Wi (A) Wild-type seedling. (B)
(78).
1
ablyin wild-typesuspensors,or in raspberry lecl-2 seedling. Abbreviations: C,
suspensorsearly in embryogenesis(64). -| | M
- cotyledon; H, hypocotyl; R, root;
These results indicate that the raspberry1 and Tr, tnchome.
suspensorhas enteredan embryogenicpath-
way and that an embryoproper-like,radial
tissueaxis has been specified.
Other Arabidopsisembryomutantshave
suspensorabnormalitiessimilar to that of
raspberryl,including raspberry2,and sus gram is inducedthat is responsiblefor (i) (76-78). Other lec mutants, such as lec2
(Table 2) (14, 35, 44, 67, 71). Although synthesizinglargeamountsof storageprod- and fus3, have phenotypic characteristics
the extent of suspensorenlargementvaries, ucts, (ii) inducingwaterloss and a desiccat- that overlapthose of lecl (77-80). Forex-
all of these mutants have morphological ed state, (iii) preventingprematuregermi- ample, fus3 and lecl embryosare almost
defects in the embryoproper(14, 35, 44, nation, and (iv) establishinga state of dor- indistinguishable fromeach other (77). lec2
67, 71). Mutant embryos that resemble mancy (Fig. 2 and Table 1) (2, 26-28). embryos,by contrast,have leaf-likecotyle-
wild-type,but arrestat specific embryonic Severalspecializedgene sets, such as those dons, but are desiccationtolerant, do not
stages,do not have aberrantsuspensors(44, encoding storageproteinsand late embryo germinateprecociously,and have normal
64). Disruptionsin embryo-proper morpho- abundant(lea) proteins,are activatedtran- levels of storagebodies in their axis region
genesis, therefore,can induce an embryo scriptionallyduringmaturationandthen re- (77). This indicates that leafy cotyledons
proper-likepathwayin terminallydifferen- pressedbeforedormancy(26, 74, 75). These can occurwithout correspondingdefectsin
tiatedsuspensorcells, a resultfirstobserved gene sets remaintranscriptionally quiescent desiccation and dormancy,and that wild-
by the embryo-proper ablationexperiments duringseedgerminationwhen germination- type LEC genes are activatedindependent-
of Haccius30 yearsago (72). The Arabidop- specific and postembryonicgene sets are ly in each embryonicorgansystem(77). A
sis twin mutantrepresentsa strikingexam- transcribed(26, 55, 75). Genetic studies corollaryis that gene networksthat func-
ple of the embryogenicpotential of the with Arabidopsis have identifiedsomeof the tion in desiccationand dormancyare inde-
suspensorregion (73). tuin causes subtle genes that regulateprocessesrequiredfor pendent of those responsiblefor cotyledon
defectsto occurin embryo-proper morphol- embryomaturationand dormancy,includ- cell specialization.
ogy, generatesa second embryowithin the ing the expressionof storageproteinand lea Molecularstudieswith lec embryoshave
seed fromproliferatingsuspensorcells, and genes (Table 2). indicatedthat the transcriptionof matura-
resultsin twin embryosthat are connected Leafycotyledonmutationsdisruptembryo tion-specificgenes, such as those encoding
by a suspensorcell bridge(73). maturation. A mutantgene class,designated storage proteins, is greatly reduced (80).
The natureof mutant genes that affect as leafycotyledon(lec), has been identified Conversely,the transcriptionof germina-
suspensordevelopment,such as raspberry 1, that causes defects in the cotyledon cell tion-specificgenes,suchas isocitratelyase,is
sus,andtwin, is not known.These genesare differentiationprocessand in maturation- activated(78). These data,coupledwith the
probablynot involved in suspensorspecifi- specificeventssuchas storageproductaccu- histologicaldescriptionsof lec embryos(Fig.
cation events, becausea normalsuspensor mulation,desiccationtolerance,and main- 7) (76-80), indicatethat LEC genes func-
formsbeforeinductionof the embryo-prop- tenanceof dormancy(Table2) (76-80). lec tion duringmaturationto activategenes in-
er pathwayin mutantembryos(44, 64, 67, mutationstransformcotyledonsof embryos volved in cotyledoncell specialization,stor-
71, 73). They reveal, however,that inter- and seedlingsinto leaf-likestructures(76- age product accumulation,induction and
actions occur between the suspensorand 80). Wild-type cotyledons do not have maintenanceof embryonicdormancy,and
embryo-properregions. One possibility is trichomes(Fig. 7A). Trichomesare present desiccationtolerance(76-80). LEC genes
that the embryopropertransmitsspecific only on leaf, stem, and sepal surfacesin also simultaneouslysuppressthe manifesta-
inhibitorysignalsto the suspensorthat sup- wild-type plants and are markers for tion of leaf-likecharacteristics
in cotyledons
pressthe embryonicpathway(35, 67, 71- postembryonicdevelopment(81). By con- during embryogenesis,including trichome
73). Altematively,a balanceof growthreg- trast, lecl cotyledons have trichomes on specification.In the absenceof LEC prod-
ulatorsmightbe establishedwithin the en- their adaxial surface which differentiate ucts, embryoniccotyledonsenter a default
tire embryo that maintains the develop- duringembryogenesisfrom the protoderm state and express leaf-like characteristics
mental states of both the embryo-proper cell layer (Fig. 7B) (76-78). lecl cotyle- (76, 77), manyof whichdevelopnormallyin
and suspensorregions.Disruptionsof such dons have other leaf-like characteristics postgermination, wild-typecotyledons(82).
signalswouldcausethe suspensorto takeon includingstomata,mesophyllcells, and an Abscisicacidmaintaunsembiryonic donman-
an embryoproper-likefate, a resultanalo- absenceof protein and lipid storagebodies cy. The planthormoneabscisicacid (ABA)
gous to embryoinductionin differentiated (77). The axis region of lecl embryosalso is involved in several plant processes,in-
sporophyticor gametophyticcells (37-39). lacks storage organelles, indicating that cluding senescence, responsesto environ-
the wild-type LEC1 gene functions in mental stresses, growth inhibition, and
The Embryo Is Reprogrammed both embryonicorgans (77). maintenanceof a dormantstate (83). Ex-
Late in Embryogenesis In addition to the leaf-like cotyledon ogenousABA preventsseed germinationas
transformation,lecl embryos(i) germinate well as the precociousgerminationof em-
How does the embryo preparefor dormancy precociouslyabout5%of the time, (ii) have bryos in culture. In addition, Arabidopsis
and postembryonic development (Fig. 1)? leaf primordiaemergingfrom their shoot mutantsthat eithercannotsynthesizeABA
Late in embryogenesis a maturation pro- apex, and (iii) fail to survive desiccation or fail to respondto ABA germinatepreco-
612 SCIENCE * VOL.266 * 28 OCTOBER1994

This content downloaded from 164.67.109.135 on Fri, 18 Jul 2014 20:08:48 PM

All use subject to JSTOR Terms and Conditions
 _ _
_ m __ A RTICLES|
gS11 sol.r.1-111.1".

ciously (84, 85). These data indicate that ering plants (10, 90-92). Several fusca Biology of Plants (Worth, New York, 1992).
3. L. Reiser and R. L. Fischer, Plant Cell 5, 1291 (1993).
ABA preventsgerminationwhile seeds are genes have been shown to be alleles of 4. C. S. Gasser and K. Robinson-Beers, ibid., p. 1231.
still dormantor presentwithin siliques. constitutive photomorphogenic(cop)/deeti- 5. E. Kranzand H. Lorz,ibid., p. 739; J.-E. Faure, H. L.
Three mutantArabidopsis loci, designat- olated (det) genes that function in light- Mogensen, C. Dumas, H. Lorz, E. Kranz, ibid., p.
747; C. Dumas and H. L. Mogensen, ibid., p. 1337;
ed as abil, abi2, and abi3,have been iden- regulateddevelopmentduringseed germi- J.-E. Faure, C. Digonnet, C. Dumas, Science 263,
tified that result in ABA insensitivityand nation (Table 2) (91-97). The productsof 1598 (1994).
allowseedgerminationto occurin the pres- COP/DET loci appear to suppresslight- 6. D. W. Meinke, Dev. Genet. 12, 382 (1991); Plant Cell
ence of ABA (84). In addition to preco- regulatedgene activities in the dark and 3, 857 (1991).
7. W. F. Sheridan, Annu. Rev. Genet. 22, 353 (1988); J.
cious germination,abi3 embryosare also activate these activities in the presenceof K. Clarkand W. F. Sheridan, Plant Cell 3, 935 (1991).
desiccationintolerantand defective in the light by way of a light-mediatedsignal 8. D. W. Meinke and 1. M. Sussex, Dev. Biol. 72, 50
synthesis of maturation-specificmRNAs, transductionpathway(91-97). Becausede- (1979); ibid., p. 62; D. W. Meinke, Theor.Appl. Gen-
et. 69, 543 (1985).
such as those encodingstorageproteinsand fective copldetgenes are detected as fusca 9. U. Mayer, R. A. Torres-Ruiz,T. Berlath,S. Mis6ra,G.
lea proteins (85, 86). This indicates that embryomutants,their wild-typeCOP/DET Jurgens, Nature 353, 402 (1991); in CellularCom-
the wild-typeABI3 gene is a positive regu- alleles are active duringmaturation.Thus, munication in Plants, R. M. Amasino, Ed. (Plenum,
New York, 1993), p. 93.
lator of gene networksleading to storage regulatorygenes expressedat the end of 10. G. JOrgens, U. Mayer, R. A. Torres-Ruiz, T. Berlath,
product accumulation, desiccation, and embryogenesispreparethe plant for life af- S. Mis6ra, Development 1 (suppl.), 27 (1991).
dormancy(85, 86). The ABI3gene encodes ter the seed. 11. U. Mayer,G. BOttner,G. Jurgens, Development 117,
149 (1993).
a transcriptionfactor (87) related to the 12. T. Berlath and G. Jurgens, ibid. 118, 575 (1993).
corn viviparous-Iprotein, which can acti- Conclusion 13. K. A. Feldmann, Plant J. 1, 71 (1991); N. R.
vate the transcriptionof chimericGUS re- Forsthoefel, Y. Wu, B. Schulz, M. J. Bennett, K. A.
Feldmann, Aust. J. Plant Physiol. 19, 353 (1992).
porter genes containing embryo matura- Plant embryogenesisprovidesa vital bridge 14. D. Errampalliet al., Plant Cell 3, 149 (1991); L. A.
tion-specific gene promoters(88). Thus, between the gametophyticgenerationand Castle et al., Mol. Gen. Genet. 241, 504 (1993).
ABI3 mediatesits effect on embryomatu- postembryonicdifferentiationevents that 15. D. E. Shevell et al., Cell 77, 1051 (1994).
rationat the transcriptionallevel. Because 16. R. B. Goldberg, Science 240, 1460 (1988).
occur continuouslyin the shoot and root 17. V. Walbot, Trends Genet. 1, 165 (1985).
aba embryos, which fail to synthesize meristems of the sporophytic plant. As 18. R. Chasan and V. Walbot, Plant Cell 5,1139 (1993).
ABA, are normal with respect to most such, plant embryosmust establishthe po- 19. S. McCormick,ibid., p. 1265.
maturation-specific processes (84-86), larizedsporophyticplant bodyplan and en- 20. J. P. Mascarenhas, ibid., p. 1303.
21. R. B. Goldberg, T. P. Beals, P. M. Sanders, ibid., p.
ABA probablydoes not regulatethe ABI3 able the youngplant to surviveharshenvi- 1217.
gene (84, 86). Rather,ABI3 probablyop- ronmentalconditions and a period of be- 22. S. D. Russell, ibid., p. 1349.
erates through an ABA-independent low-groundgrowth from seeds. Plant em- 23. M. A. Lopes and B. A. Larkins,ibid., p. 1383.
24. G. Gillaspy, H. Ben-David, W. Gruissen, ibid., p.
pathway that is involved in establishing bryos are simpler than their animal 1439.
desiccation and dormancy states late in counterparts,yet they must carryout the 25. E. S. Coen and R. Carpenter, ibid., p. 1175; J. K.
embryogenesis(85, 86). Failureto achieve same developmentaltasks-that is, form a Okamuro, B. G. W. den Boer, K. D. Jofuku, ibid., p.
1183.
these developmentalstates resultsin ABA three-dimensionalorganism with special- 26. R. B. Goldberg, S. Barker, L. Perez-Grau, Cell 56,
insensitivity-that is, ABA responsive- ized regions,compartments,and cell-types 149 (1989).
ness is a consequenceof ABI3 gene activ- from a single-celled zygote. These events 27. K. Lindsey and J. F. Topping, J. Exp. Bot. 44, 359
ity (85, 86). lec embryosare sensitive to occur early in plant embryogenesisand are (1993).
28. M. A. L. West and J. J. Harada, Plant Cell 5, 1361
ABA and fail to germinate if ABA is poorlyunderstood.Genetic studiesin Ara- (1993).
present (77, 78). Thus, LEC and ABI3 bidopsishave begunto revealgenes that are 29. V. Raghavan,Experimental
Embryogenesisin Vas-
genes are part of independent regulatory necessaryfor embryogenicevents such as cular Plants (Academic Press, New York, 1976); S.
Natesh and M. A. Rau, in Embryology of Angio-
networksthat control embryomaturation- patternformation,cell differentiation,and sperms, B. M. Johri, Ed. (Springer-Verlag, Berlin,
specific events. organdevelopment. 1984), chap. 8; B. M. Johri, K. B. Ambegaokar, P. S.
In contrastto abi3embryos,abil andabi2 The precise molecularmechanismsre- Srvastava, ComparativeEmbryologyof Angio-
sperms (Springer-Verlag,Berlin, 1992).
embryoscarryout normalmaturation-specif- sponsible for specifyingdifferentcell lin- 30. M. Schaffner, Ohio Nat. 7, 1 (1906); R. Schulz and
ic events,includingthe activationof storage eages early in plant embryogenesisare not W. A. Jensen, J. Ultrastruct. Res. 22, 376 (1968);
protein and lea genes (84-86). ABI1 and known.A majorvoid in ourknowledgecon- Am. J. Bot. 55,541 (1968); ibid., p. 807; ibid., p. 915.
ABI2 loci, therefore,are involved only in 31. J. Hanstein, Bot. Abhadl., Bonn. 1, 1 (1870).
cerns the events that occur within the egg 32. E. C. Soueges, Ann. Sci. Nat. 9, Bot. 19, 311 (1914);
maintaining embryonic dormancy. The cell and in the earlyembryoafterfertiliza- Ann.Sci. Nat. 10, Bot. 1,1 (1919).
ABI1gene encodesa Ca2+-dependent phos- tion. In this respectit is crucialto obtain 33. S. G. Mansfield,L. G. Briarty,S. Erni,Can. J. Bot. 69,
phatasewith similarityto serine-threonine molecularmarkersin order to follow the 447 (1991); S. G. Mansfieldand L. G. Briarty,ibid., p.
461; ibid. 70,151 (1992).
phosphatasesinvolved in signal transduc- specificationevents that take place during 34. L. Dolan et al., Development 119, 71 (1993).
tion processes(89). In responseto ABA, the earlyembryogenesis and gain entryinto reg- 35. E. C. Yeung and D. W. Meinke, Plant Cell 5, 1371
ABII phosphatasemight counteractphos- ulatorynetworksthat areactivatedin differ- (1993).
36. M. K. Barton and R. S. Poethig, ibid., p. 823.
phorylationevents requiredfor the initia- ent embryonicregionsafterfertilization.Al- 37. J. L. Zimmerman, ibid., p. 141 1.
tion of root meristemcell division,resulting thougha largeamountof progresshas been 38. F. A. Van Engelen and S. C. de Vries, Trends Genet.
in a dormantembryonicstate (89). madein recentyearsin understanding how a 8, 66 (1992).
Regulatory loci requiredfor postembryonic plant embryoforms,there is still a long way 39. A. J. de Jong, E. D. L. Schmidt, S. C. de Vries, Plant
Mol. Biol. 22, 367 (1993).
development are activelate in embryogenesis. to go. The next few yearsshouldbe a very 40. L. Perez-Grau and R. B. Goldberg, Plant Cell 1, 1095
A largenumberof Arabidopsis fuscamutants excitingtime to studyplant embryos. (1989).
have been identified that accumulatean- 41. R. L. Taylor, Can. J. Bot. 45,1553 (1967).
42. M. A. Chamberlin, H. T. Horner, R. G. Palmer, ibid.
thocyanins,or red pigments,on their coty- REFERENCESAND NOTES 71, 1153 (1993).
ledons late in embryogenesis(10, 90-92). 43. J. P. Nitsch, Phytomorphology 19, 389 (1969); B.
With the exception of fus3 (Table 2), em- 1. E. H. Davidson,Development105, 421 (1989);ibid. Norreel, Bull. Soc. Bot. Fr. 117, 461 (1970); R. S.
bryogenesis is normal in fuscamutants (10, 108, 365 (1990);ibid.113, 1 (1991);ibid.118, 665 Sangwanand B. S. Sangwan-Norreel, Int.Rev.Cy-
(1993). tol.107, 221 (1987).
90-92). After germination, however, fusca 2. K. Esau, Anatomy of Seed Plants (John Wiley, New 44. The Embryo21st CenturyGroup,unpublishedre-
seedlings fail to develop into mature flow- York, 1977); P. H. Raven, R. F. Evert,S. E. Eichhorn, sults.Thisgroupis a collaborative
effortbetweenthe

SCIENCE * VOL. 266 * 28 OCTOBER 1994 613

This content downloaded from 164.67.109.135 on Fri, 18 Jul 2014 20:08:48 PM

All use subject to JSTOR Terms and Conditions
 laboratories of R. B. Goldberg [Universityof Califor- Shimura, Plant Cell 3, 677 (1991). 77. D. Meinke, L. H. Franzmann, T. C. Nickle, E. C.
nia, Los Angeles (UCLA)],J. J. Harada (UC Davis), R. 54. G. de Paiva and R. B. Goldberg, unpublished results. Yeung, Plant Cell 6, 1049 (1994).
L. Fischer (UC Berkeley),J. L. Zimmerman(Universi- The chimeric Kti3/GUS genes contained the Kti3 78. The Embryo 21st Century Group (44), unpublished
ty of Maryland,Baltimore),and A. Koltunow (Univer- gene TATA-boxregion (-28) (47). The chimeric Gyll data. Research on the lecl-2 gene was carried out
sity of Adelaide, Australia).A total of 5822 T-DNA- GUS gene contained the cauliflower mosaic virus by M. West and J. J. Harada (UC Davis).
mutagenized Arabidopsis lines (13) were screened (CaMV) TATA-box region within the CaMV minimal 79. K. Keith, M. Krami,N. G. Dengler, P. McCourt,Plant
for mutants defective in embryogenesis (6, 8, 14). promoter (-42 to +0) (59). Cell 6, 589 (1994).
Sixty-six heterozygous lines were identifiedthat seg- 55. T. L. Thomas, Plant Cell 5, 1401 (1993). 80. H.,Baumlein et al., Plant J., in press.
regated seeds in a ratio of 3 wild-type to 1 mutant 56. A. N. Nunberg, et al., ibid. 6, 473 (1994). 81. M. Hulskamp, S. Mis6ra, G. Jurgens, Cell 76, 555
within their siliques, indicating that they contained 57. M. A. Bustos, D. Begum, F. A. Kalkan,M. J. Battraw, (1994).
embryo-defective alleles that were inherited as sim- T. C. Hall,EMBOJ. 10, 1468 (1990). 82. Y. Hou, A. G. van Arnim,X.-W. Deng, Plant Cell 5,
ple Mendelian recessive genes. The raspberryl, 58. A. da Silva Conceic,o and E. Krebbers, Plant J. 5, 329 (1993).
raspberry2, and lecl-2 genes were uncovered in 493 (1994). 83. L. Taiz and E. Zeiger, Plant Physiology (Benjamin,
three different heterozygous lines. 59. P. N. Benfey, L. Ren, N.-H. Chua, EMBOJ. 9,1677 New York, 1991).
45. K. D. Jofuku, R. D. Schipper, R. B. Goldberg, Plant (1990). 84. M. Koorneef, M. L. Jorna, D. L. C. Brinkhorstovan
Cell 1, 427 (1989); K. D. Jofuku and R. B. Goldberg, 60. Z. R. Sung, A. Belancheur, B. Shunong, R. Bertrand- der Swan, C. M. Karssen, Theor. Appl. Genet. 61,
ibid., p. 1079. Garcia, Science 258, 1645 (1992). 385 (1982); M. Koorneef, G. Reuling, C. M. Karssen,
46. P. Sterk, H. Booij, G. A. Schellekens, A. Van Kam- 61. P. N. Benfey et al., Development 119, 57 (1993). Physiol. Plant. 61, 377 (1984); M. Koorneef, C. J.
men, S. C. DeVries, ibid. 3, 907 (1991). 62. R. A. Aeschbacher, J. W. Schiefelbein, P. N. Benfey, Hanhart, H. M. W. Hilhorst, C. M. Karssen, Plant
47. G. de Paiva and R. B. Goldberg, unpublished results. Annu. Rev. Plant Physiol. Plant Mol. Biol. 45, 25 Physiol. 90, 463 (1989).
A 2-kb 5' region from the soybean Kti3 gene was (1994). 85. E. Nambara, S. Nito, P. McCourt, Plant J. 2, 435
fused with the Escherichia coli GUS gene [R. A. Jef- 63. N. G. Nielsen et al., Plant Cell 1, 313 (1989). (1992).
ferson, T. A. Kavanagh, M. W. Swan, EMBO J. 6, 64. The Embryo 21 st Century Group (44), unpublished 86. R. R. Finkelstein,Mol. Gen. Genet. 238, 401 (1993).
3901 (1987)], and the resulting chimeric Kti3/GUS results. Research on the raspberryl and raspberry2 87. J. Giraudatet al., Plant Cell 4, 1251 (1992).
gene was transferredto tobacco plants byAgrobac- genes was carriedout by R. Yadegari, G. de Paiva, T. 88. D. R. McCartyet al., Cell 66, 895 (1991).
terium-mediated transformation as described (45). Laux, and R. B. Goldberg (UCLA).Wild-type and 89. J. Leung etal., Science 264, 1448 (1994); K. Meyer,
GUS activity in isolated whole-mount embryos was raspberryl seeds were harvested from heterozy- M. P. Leube, E. Grill,ibid., p. 1452.
analyzed as outlined in G. N. Drews, T. P. Beals, A. gous +/raspberryl Arabidopsis plants (44), fixed, 90. A. J. Muller,Biol. Zentralbl. 82, 133 (1963).
Q. Bui, and R. B. Goldberg [Plant Cell 4, 1383 embedded in paraffin,and hybridizedwith 35S-RNA 91. L. A. Castle and D. W. Meinke, Plant Cell 6, 26
(1992)]. antisense probes according to the procedures out- (1994).
48. R. Yadegari and R. B. Goldberg, unpublished re- lined in (40, 48). raspberryl seeds were harvested 92. S. Mis6ra, A. J. Muller, U. Weiland-Heidecker, G.
sults. A 2-kb 5' region from the soybean lectin gene from siliques at a time when neighboring wild-type Jurgens, Mol. Gen. Genet. 244, 242 (1994).
[R. B. Goldberg, G. Hoschek, L. 0. Vodkin, Cell 33, seeds were in the maturationstage of development. 93. X.-W. Deng et al., Cell 71, 791 (1992).
465 (1983)] was fused with the E coli GUS gene and 65. The Arabidopsis EP2 gene probe (46) was provided 94. T. W. McNellis et al., Plant Cell 6, 487 (1994).
transferredto tobacco plants as outlined in (47). Lo- by S. DeVries. 95. A. Pepper, T. Delaney, T. Washburn, D. Poole, J.
calization of GUS activity in staged, whole-mount 66. P. Guerche et al., Plant Cell 2, 469 (1990). The Ara- Chory, Cell 78, 109 (1994).
embryos was determined by the procedure outlined 96. N. Wei et al., Plant Cell 6, 629 (1994).
bidopsis 2S2 gene probe was provided by E. Kreb-
in (47). In situ hybridizationof a GUS antisense 35S- 97. N. Wei, D. A. Chamovitz, X.-W. Deng, Cell 78, 117
bers.
RNA probe with a longitudinal globular-stage em- (1994).
67. B. W. Schwartz, E. C. Yeung, D. W. Meinke, Devel-
bryo section was carriedout according to the proce- 98. We express our gratitude to J. Harada, R. Fischer,
opment, in press.
dures outlined in (40) and in K. H. Cox and R. B. 1. Okamuro, D. Jofuku, L. Zimmerman, A. Kol-
68. R. Losick and L. Shapiro, Science 262,1227 (1993).
Goldberg [inPlant MolecularBiology:A PracticalAp- tunow, T. Laux, L. Perez-Grau, and G. Drews for
69. A. J. DeJong et al., Plant Cell 4, 425 (1992).
proach, C. H. Shaw, Ed. (IRL,Oxford, 1988), pp. many critical and stimulating discussions of plant
1-35]. 70. A. J. De Jong et al., ibid. 5, 615 (1993). embryogenesis. We thank B. Timberlake, R.
49. L. Michalczuk, T. J. Cooke, J. D. Cohen, Phyto- 71. M. P. F. Marsden and D. W. Meinke,Am. J. Bot. 72, Chasan, and E. Davidson for insightful comments
chemistry 31, 1097 (1992). 1801 (1985). on this manuscript. We gratefully acknowlodge K.
50. S. C. Fryand E. Wangermann, New Phytol. 77, 313 72. B. Haccius, Phytomorphology 13,107 (1963). Brill,B. Phan, and T. Zeyen for diligent typing of this
(1976). 73. D. M. Vernon and D. W. Meinke, Dev. Biol., in press. manuscript and M. Kowalczyk for the colorful art
51. F. M. Schiavone and T. J. Cooke, Cell Differ.21, 53 74. L. Walling,G. N. Drews, R. B. Goldberg, Proc. Nat/. work. We dedicate this article to Professor E. H.
(1987). Acad. Sci. U.S.A. 83, 2123 (1986). Davidson, who has provided intellectual leadership
52. C.-M. Liu,Z.-H. Xu, N.-H. Chua, ibid., p. 621. 75. L: Comai and J. J. Harada, ibid. 87, 2671 (1990). and direction to the field of development biology
53. K. Okada, J. Ueda, M. K. Komaki, C. J. Bell, Y. 76. D. Meinke, Science 258,1647 (1992). over the past 25 years.

614 SCIENCE * VOL. 266 * 28 OCTOBER 1994

This content downloaded from 164.67.109.135 on Fri, 18 Jul 2014 20:08:48 PM

All use subject to JSTOR Terms and Conditions