a genetic and molecular model for flower development in...

Development Supplement I, 1991, 157-167Printed in Great Britain © The Company of Biologists Limited 1991

157

A genetic and molecular model for flower development in Arabidopsis

thaliana

ELLIOT M. MEYEROWITZ*, JOHN L. BOWMAN, LAURA L. BROCKMAN, GARY N. DREWS,

THOMAS JACK, LESLIE E. SIEBURTH and DETLEF WEIGEL

Division of Biology 156-29, California Institute of Technology, Pasadena, California 91125, USA

•Corresponding author

Summary

Cells in developing organisms do not only differentiate,they differentiate in defined patterns. A striking exampleis the differentiation of flowers, which in most plantfamilies consist of four types of organs: sepals, petals,stamens and carpels, each composed of characteristiccell types. In the families of flowering plants in whichthese organs occur, they are patterned with the sepals inthe outermost whorl or whorls of the flower, with thepetals next closest to the center, the stamens even closerto the center, and the carpels central. In each species offlowering plant the disposition and number (or range ofnumbers) of these organs is also specified, and the floral'formula' is repeated in each of the flowers on eachindividual plant of the species. We do not know how cellsin developing plants determine their position, and inresponse to this determination differentiate to the cell

types appropriate for that position. While there havebeen a number of speculative proposals for themechanism of organ specification in flowers (Goethe,1790; Goebel, 1900; Heslop-Harrison, 1964; Green,1988), recent genetic evidence is inconsistent with all ofthem, at least in the forms in which they were originallypresented (Bowman etal. 1989; Meyerowitz etal. 1989).We describe here a preliminary model, based onexperiments with Arabidopsis thaliana. The model is byand large consistent with existing evidence, and haspredicted the results of a number of genetic andmolecular experiments that have been recently per-formed.

Key words: Arabidopsis thaliana, pattern formation, floralmutants.

Processes in flower development

From many descriptions of flower development (Payer,1857; Sattler, 1973), we can divide the developmentalprocesses that lead to flowers into four successivestages. The first is floral induction. This is the process bywhich the shoot apical meristem, which is the set ofdividing cells that gives rise to most of the plant partsthat are above the roots, decides that it is time to stopmaking leaves, and to start making flowers (Bernier,1988). Induction is effected in different ways indifferent plants; in Arabidopsis a combination ofinternal and environmental signals, including age of theplant, its nutritional state, day length, and in somestrains temperature, determines that the apical meri-stem will change its fate from vegetative to floral. Oncethis has happened the second stage of flower develop-ment begins. This is the production of individualflowers on the flanks of the shoot apical meristem.These floral primordia arise, by patterned cell divisions,in a defined temporal and spatial pattern (Fig. 1A). Thethird stage is the formation of organ primordia on each

flower primordium. The wild-type Arabidopsis flowercontains 15 separate organs: four sepals, four petals, sixpollen-bearing stamens, and a pistil (Fig. 1). InArabidopsis, the third stage of floral development isthus the formation of 15 undifferentiated organprimordia (Fig. 1B,C), at stereotyped positions withinthe flower primordium. These organ primordia are atthis time not only undifferentiated, but are also notirreversibly determined, so that their fates are as yeteither partly or completely unspecified (Bowman et al.1989). The final stage of flower development is thespecification of the fates of the organ primordia, andtheir consequent differentiation into organs appropri-ate for their positions (Fig. ID). Each organ iscomposed of a small number of defined cell types, someof which (e.g. pollen sperm cells) are organ-specific,and some of which (e.g. guard cells) are found inseveral kinds of organs. Detailed descriptions of flowerdevelopment in Arabidopsis thaliana are available (Hilland Lord, 1989; Smyth et al 1990).

That each of these stages of flower development is tosome degree a separate process, effected by a different

158 E. M. Meverowitz and others

Fig. 1. Scanning electron microscope views of the development of wild-type Arabidopsis flowers. (A) A florally-inducedshoot apical meristem (inflorescence meristem) surrounded by a series of flower primordia. The developmentally mostadvanced flower primordium is at the right, and is showing the earliest stages of sepal development (stage 3). (B) Anindividual flower primordium at stage 6, when organ primordia have formed, but are not yet fully specified. Two sepalshave been removed to reveal the underlying inner whorl primordia. (C) Another developing flower, at stage 7, with all ofthe sepals removed to show the first signs of organ-specific differentiation of the six third whorl organs (which arebecoming stamens) and of the fourth-whorl ovary. (D) Developing flower at stage 11, when the organ-specificdifferentiation of the epidermal cells of each organ type is evident. Sepals have been removed to reveal the inner organs.The bars in panels A,B and C represent lOum, that in panel D 100/«n.

group of regulatory genes in Arabidopsis, is indicatedby the existence of many mutations that affect only oneor only a subset of the stages. That these processes areinterrelated is equally demonstrated by the fact thatmany mutations affect more than one of them. Themutations that interfere with the fourth stage arehomeotic: they cause normal organs to develop inabnormal places (Pruitt et al. 1987; Meyerowitz et al.1989). Some of these mutants are normal up to the timewhen the undifferentiated organ primordia are formed,and these primordia generally appear in their wild-typenumbers and places. Other mutants affect organnumber and organ position as well as organ identity.

There is much more known about the fourth stage inflower development than about the others: we havedone a series of genetic experiments that have led us toa working hypothesis for the way in which three classesof homeotic genes specify organ identity in each floralwhorl (Bowman et al. 1989; Bowman et al. 1991). Thismodel does not. however, explain the phenotypes of thehomeotic mutants that relate to organ number andposition. Nonetheless, the model has correctly pre-dicted patterns of organ identity in doubly and triplymutant strains, and predicts both the pattern ofexpression of the RNA of one of the homeotic genes,and the changes in that pattern that result frommutations in one of the other homeotic genes.

A genetic model for organ identity in flowerdevelopment

Wild-type Arabidopsis flowers have four concentric setsof organs, which consist of (from the outside of theflower) four sepals, four petals, six stamens, and anovary (Figs 2A and 3). Mutations that alter this patternwere first recognized in Arabidopsis by Braun (1873),who recognized a homeotic phenotype in a wildpopulation of plants in Berlin. To begin our genetic andmolecular characterization of Arabidopsis flower devel-

opment, we collected from others, and producedourselves, a large series of mutations (induced by ethylmethanesulfonate or X-rays) with abnormal floralphenotypes. As described above, these mutations fallinto several classes, each apparently affecting one ormore processes in flower development. We haveconcentrated on the homeotic mutations that affect thefourth stage in flower development, including theprocess by which unspecified organ primordia learntheir fate. Of the mutations initially screened, manygave clear, reproducible homeotic phenotypes. Comp-lementation analysis (by us and others) has shown thatmost of these fall into four complementation groups,and thus represent four genetic loci with major effectson organ type specification (Pruitt et al. 1987; Meyerow-itz, 1987; Komaki et al. 1988; Bowman et al. 1989; Kunstetal. 1989; Meyerowitz et al. 1989; Yanofsky etal. 1990;Bowman et al. 1991).

Four mutations in three classesThese loci are agamous {ag), at position 37.6 cM on thefourth chromosome in the morphological marker mapof Arabidopsis (Koornneef, 1989); apetalal (ap2) at4-63.5; pistillata (pi) at 5-23.6; and apetala3 (ap3) at3-73.9. All of the mutations at these loci are recessive(except for one semidominant ap2 allele which werecently obtained). Fig. 3 shows the map positions ofthese four loci, and schematically represents thephenotypes of one or more mutant alleles of each of thegenes. To describe the phenotypes we will use the termwhorl to mean a region of the flower or flowerprimordium (and not the collection of organs found inthat region). The first whorl is the outer ring of tissue,from which the sepals arise in wild-type flowers. Thesecond whorl is the next inner concentric ring, fromwhich petals normally arise. The third whorl is the ringfrom which stamens develop in wild type, and thefourth whorl is the bull's-eye. It is the normal locationof the ovary.

Each of these mutations affects the organs of two

Fig. 2. Mature Arabidopsis flowers. (A) Wild type, showing the four petals, six stamens, and ovary.(B) agamous-3 homozygote, showing the extra petals and sepals, as well as the absence of stamens andovary. (C) apetala3-l homozygote, with the organs of the outer two whorls removed to show the thirdwhorl organs, which develop as carpels rather than stamens, as in wild type. (D) apelala2-2/apeiala2-lheterozygous flower, showing first whorl carpels, absence of second whorl organs, and normal organidentity (though not number) in the third whorl. The fourth whorl is normal. (E) apetala2-l homozygousflowers, showing first whorl leaves and second whorl stamens. The third and fourth whorl organs arenormal.

Fig. 5. Flowers of doubly and triply mutant strains. (A) Flowers homozygous for both ap2-l and ap3-l. All organs are carpelloid.(B) Rowers homozygous for both ap3-l and ag-I. All floral organs are sepals. (C) Flower homozygous for ap2-I as well as ag-l.showing that the whorl 1 and 4 organs are leaves, and the whorl 2 and 3 organs are intermediate between petals and stamens.(D) Triply mutant inflorescence, homozygous for the three mutations ap2-I. ag-I. and pi-1. In each flower, all of the floral organshave developed as leaves.

Fig. 6. Flower of a plant homozygous forsupennan-I. Note the four sepals, fourpetals. 10 stamens, and reduced ovary.

Arabidopsis flower development 159

adjacent whorls. The agamous mutations (of whichthree alleles have been studied, i.e. ag-1, ag-2 and ag-3)affect only whorls three and four. The six organs ofwhorl three develop as petals rather than stamens, andthe whorl four primordium forms four sepals, inside ofwhich are additional whorls of petals and sepals. Thus,if the normal pattern of sepals, petals, stamens, ovary iscalled an ABCD pattern, agamous mutants have anABBA(BBABBA...) pattern (Fig. 2B). We have ex-tensively studied three pi alleles, pi-1, pi-2 and pi-3(Bowman el al. 1989; Bowman et al. 1991) and two ap3alleles (one is described by Bowman et al. 1989), andboth pi and ap3 affect whorls two and three only, havingessentially the same effects (except for some details ofthe pi-1 third whorl, where the organ primordia do notform normally). Whorl two contains four sepals ratherthan the wild-type four petals, and instead of the wild-type pattern of six third whorl stamens, the mutantwhorl three consists of six carpels (the subunit of theovary; the normal Arabidopsis ovary consists of twocarpels). These mutations thus have the phenotypeAADD (Fig. 2C). Apetala2 mutants, four alleles ofwhich have been studied in our laboratory (Pruitt et al.1987; Bowman et al. 1989; Bowman etal. 1991), and fivemore by others (Komaki et al. 1988; Kunst et al. 1989),affect organ identity in whorls one and two. In mostalleles (the single exception will be explained below)the organs of the first whorl are carpels, and the secondwhorl consists of four (or sometimes fewer) stamens in

the places normally occupied by the four petals. Mostalleles also have effects on organ number in the first,second and third whorls, and on the position ofinitiation of organ primordia in whorls 1, 2 and 3. Sincewe have only succeeded in producing a model for organidentity, and not for organ number, we will for presentpurposes consider the relevant pattern of most alleles tobe DCCD (Fig. 2D). Discussion of the effects of AP2on organ number and position will be reserved for later.

A working hypothesis, in the form of a modelWe have proposed a simple model to explain the wild-type functions of the homeotic genes (Bowman et al.1989; Bowman et al. 1991). It proposes that these fourgenes fall into three classes, each class affecting adifferent pair of whorls. APETA LA2 is the only knowngene in the first class, which affects whorls 1 and 2; A P3and PI constitute the presently-known second class,with effects in whorls 2 and 3; and AG is the only knownmember of the final class, which affects whorls 3 and 4.Each class of genes is expressed in the developingflower in those whorls affected in the mutants. Whorlone is thus characterized by expression of AP2\ whorltwo by the combination AP2, PI and AP3; whorl threeby the combination AG, PI and AP3; and whorl four bythe expression of AG. Each of the products of thesegenes has as its function the induction of expression of aseries of downstream genes, the expression of which

LINKAGE GROUP

sepal

apetala3-l apetala2-2 apetala2-l agamoua-1 pistillata-2

Fig. 3. Schematic representation of the phenotypes of some mutant alleles of the four best-studied homeotic genes. Thestructure of a wild-type Arabidopsis flower, and the genetic map positions of the homeotic genes, are also depicted.


controls or constitutes the differentiation of specificfloral organs. When AP2 (and any other genes in thesame class, if there are any) is expressed alone in anyorgan primordium, the primordium differentiates to asepal. The combination of two classes of genes, AP2plus PI/AP3, directs a primordium to differentiate to apetal; the third-whorl combination PI/AP3 plus AGdirects stamen differentiation. If only AG (and othergenes in the same class, if they exist) is expressed, theprimordium differentiates to a carpel. Thus, thesegenes are expected to begin to be active before anyorgan differentiation, and to direct the fate which organprimordia will adopt.

One further aspect is needed for the model to explainthe mutant phenotypes. We propose an interactionbetween the whorl 1/2 {APT) and whorl 3/4 (AG)genes. In addition to its action on downstream genes,AP2 represses AG expression in whorls 1 and 2. Inaddition to its action on downstream genes, AGrepresses expression of AP2 in whorls 3 and 4.

The model now explains the single mutant pheno-types. Fig. 4 shows, at the top, the wild-type overlap-ping patterns of expression of these genes in thedifferent whorls, as proposed by the model. Beneaththis is shown the expected pattern of expression in anap2 mutant. Since the wild-type AP2 product is absent,AG is expressed in all four whorls. The expression ofthe PI and AP3 genes is unaffected. Whorl 1 organsthus differentiate into carpels, as directed by AG, whorl2 and 3 organs into stamens, as directed by the AG andPI plus AP3 combination, and whorl four is normal.The effect of ag mutations is shown below: the patternof expression of regulatory genes is normal in whorls 1and 2. The absence of wild-type AG product allowsAP2 to express in whorls 3 and 4, and dictates that theorder of organs thus be sepals, petals, petals, sepals.The existence of the further whorls that are presentinside whorl 4 in ag mutants is not specifically predictedby the model (in Fig. 4 this is noted by the asterisk), andindicates that one of the functions of AG is to turn offdownstream genes that allow continued cell division inthe center of the developing flower. There is additionalevidence for this AG function, which will be describedbelow. The phenotype of pi or ap3 mutants and itsorigin is also shown in Fig. 4: in the absence of thefunction provided by the activity of these genes, the firsttwo whorls express only AP2 and differentiate tosepals, while the third and fourth whorl organs expressonly AG and differentiate to carpels.

Genetic tests of the modelThe first test we have applied to see if this model canactually predict anything other than the original mutantphenotypes (on which, after all, it is based) is to predictthe phenotypes of double and triple mutant combi-nations, and then to test the predictions by makingplants homozygous for all possible pairs of mutations.All of the double and triple mutant combinations agreewith the predictions of the model (Fig. 5, Bowman et al.1989; Bowman et al. 1991).

If the double mutant ap2 ap3 (or ap2 pi) is made, it is

expected that all whorls will express only AG, and thatall organ primordia will differentiate to carpels. This isindeed what is seen in the double mutant (Fig. 5A). Anap3 (or pi) ag double mutant retains only the AP2function, which is predicted to be expressed in all fourwhorls. This is expected to convert all floral organs tosepals. The doubly mutant plants do in fact give theexpected pattern of all organs developing as sepals, andin addition, because of the indeterminate growth of ag

AP3, PIAP2 AG

Se Pe St Ca

apl( AP3, PI

AG

Ca St St Ca

AP3, PIAP2

Se Pe Pe Se*

ap3, pi

apl ap3

ap3 ag

apl ag

apl ag ap3

whorl:

AP2 AG )

Se Se Ca Ca

AG

Ca Ca Ca Ca

AP2

Se Se Se Se*

AP3, PI

Leaf Pe/St Pe/St Leaf*

Leaf Leaf Leaf Leaf

Fig. 4. Schematic representation of the model. A sectionthrough one-half of a floral primordium is represented as apair of boxes, with the regions representing each whorllabeled at the bottom. The genotype under consideration islisted at the left, and in the boxes are the predicted patternof activity of the products of the three classes of homeoticgenes. The gene products are represented in upper case.Under each diagram is the predicted phenotype of theorgans in each whorl. These predictions have all beenconfirmed by experiment. +, wild type; Se, sepal; Pe,petal; St, stamen; Ca, carpel. Pe/St represents theintermediate organs, with characteristics of both stamensand petals, found in ag apl double mutants. Leaf, eitherleaf or carpelloid leaf. Either can be found in thesepositions, depending on the apl allele involved. Theasterisk is a reminder that all ag combinations haveadditional whorls inside whorl four.


flowers, the double mutants have many more than fourwhorls, see Fig. 5B.

The ap2 ag double mutant prediction is the mostimportant. The model (see Fig. 4) predicts that whorls 1and 4 will have no expression of any of the classes ofhomeotic genes, and thus if the three known classes ofhomeotic gene are necessary for floral organ develop-ment, the whorl 1 and 4 primordia will differentiate intoorgans unlike any normally seen in flowers. In fact, bymany criteria, the whorl 1 and 4 organs in the mutantare leaves (stellate trichomes, stipules, shape, patternof guard cells, appearance of spongy mesophyll, leafand not sepal senescence pattern) or carpelloid leaves.This implies that the three classes of genes are almostsufficient to explain floral organ differentiation (an as-yet unknown class may contribute the carpelloid aspectof these organs). The phenotype of the whorl 2 and 3organs in the double mutant is expected to beintermediate between that of petals and stamens, sincethe ap3/pi class of genes is expressed, giving theseorgans the potential to develop as either petals orstamens, but the gene activities that would decidebetween these two fates are absent. The double mutantphenotype in these whorls is in accord with expectation(Fig. 5C). The phenotype demonstrates the AP2-AGinteraction that the model postulates: in ag mutantsalone there are no whorl 1 or 2 effects, but adding ag toan ap2 mutant does create new whorl 1 and whorl 2phenotypes. Likewise, ap2 normally has no effect onorgan identity in whorl 3 or whorl 4, but it does havesuch an effect in the absence of wild-type AG product.

The triple mutant combinations ap2 pi ag and ap2 ap3ag should reveal the ground state in every whorl, andshould not contain any floral organs, only vegetative, orclose to vegetative, ones. This does indeed occur, and inthe triple mutants each organ in each whorl is convertedto leaves, or slightly carpelloid leaves. It is worth notingthat it is now just past the 200th anniversary of Goethe'sstatements that all floral organs are derived from thebasic ground state of leaf, by the action of differentqualities of sap on the basic organ (Goethe, 1790).

There are several observations that were quitepuzzling before the flower development model wasconceived, but which are now seen to fit the model verywell. The best example is that of the 'weak' ap2 allele,ap2-l (Fig. 2E). Unlike the other eight described allelesof ap2, which have first whorl carpels, ap2-l whengrown at 25°C has leaves in the first whorl. All secondwhorl ap2 organs are stamens, or are staminoid. Wehave called this allele 'weak' because its phenotypeseems to be that of a partial loss of AP2 productfunction. At low temperatures (16°C) it is very close towild type, with leafy first whorl organs, but petals in thesecond whorl. At intermediate temperatures (25°C) it isless normal, with first whorl leaves and second whorlstamens. At 29°C this mutant has carpelloid leaves inthe first whorl, and stamens in the second whorl. (For amore detailed description, see Bowman et al. 1989.)Thus, the intermediate leaf phenotype in the first whorlappears on a graded path of loss of function. The modelaccounts for this nicely: it predicts two functions for the

wild-type AP2 product. One is repression of AGactivity in whorls 1 and 2, the other is induction of aseries of downstream genes, the activity of whichproduces the differentiation of outer whorl organs. If,in the first whorl, the first activity is present and thesecond absent, then none of the homeotic geneactivities will be expressed there, and we expect (andget) leaves. When both activities are absent, as in thestrong ap2 mutants, carpels form in the first whorl.

A working genetic model is insufficientOur model is very effective at explaining and predictinggenetic results as far as regulation of organ identity isconcerned. The model, however, makes no explicitpredictions about the actual mechanisms involved in thepostulated activities of the homeotic genes, other thanthat they regulate each other's activities, and activate orrepress downstream genes. It does not predict thenature of the downstream genes on which the homeoticgenes act, and it does not indicate the nature of theearly pattern of gene inducers that exists prior toexpression of the homeotic genes, and to which theyrespond by turning on in pairs of adjacent whorls. Itsays nothing of the detailed patterns of different celltypes which actually constitute the floral organs; ittreats each organ as a homogeneous structure, not amosaic of different cell types, arrayed in stereotypedpatterns. The model may also not be unique. Thoughwe have not thought of any other model as effective inpredicting genetic results, there may be one. Thus, weneed to know the structure, activity, expression patternand nature of the products of the genes whose activitiesare only described in a very general fashion by ourcurrent working model. The model needs to be tested atthe molecular level, and if the tests confirm it, it needsto be refined. The tests in progress involve cloning thehomeotic genes, and analysis of their expression and oftheir products, and of the function of these products.

The molecular cloning of the Agamous locus

The molecular tests of the genetic model have startedwith the molecular cloning of AG (Yanofsky et al.1990). In the course of a large series of Arabidopsistransformation experiments, whose goal was T-DNAtagging of Arabidopsis genes, Dr K. Feldmann ob-tained what appeared to be (and on complementationtesting was) a new agamous allele (Feldmann et al.1989), and made it available to us. Since at that time wehad only carefully characterized one EMS-inducedallele, ag-1, we called the new one ag-2. Since it arose ina transformation experiment, it seemed possible thatthe new mutation might be due to the chance insertionof the introduced DNA into the AG locus. We clonedthe DNA next to the insertion point of the introducedDNA, and RFLP (restriction fragment length poly-morphism) mapped it (using the markers and methodsdescribed in Chang et al. 1988). The map positions ofthe flanking DNA and agamous were identical, and ondirect cosegregation analysis, the cloned DNA and theag-1 mutation were not separated. The cloned fragment


from the mutant was used as a probe to isolate cosmidclones from wild-type plants, and one of these cloneswas transformed into ag-2 mutant plants using Agrobac-lerium tumefaciens T-DNA insertion via leaf discs. Theregenerated plants had wild-type flowers. This showedthat the cloned DNA contains the AG gene.

To find which of the transcripts coded in the cosmidcorresponds to the AG gene, a series of wild-typecDNA clones coded by the cloned genomic DNA wasobtained. The introduced T-DNA in the ag-2 strain,which is a 35 kb DNA insertion, was found to be in themiddle of a large intron of a gene with (at least) 9 exonsand 8 introns, which codes for a 1.1kb RNA that isexpressed in inflorescences. To be certain that this wasthe AG gene, the wild-type gene and the ag-1 mutantversion of the gene were sequenced. The ag-1 mutanthad been induced in the Landsberg erecta strain, thesame strain from which the wild-type gene used for theinitial DNA sequencing was cloned. The sequence ofthe ag-1 DNA showed a single nucleotide change fromwild-type, in the same gene that was tagged in ag-2. Thechange is a G to A transition at the acceptor site for thefourth intron, a change that presumably preventsproper splicing. Thus, by criteria of complementationand mutant sequencing, the cloned gene is AGA-MOUS.

Comparison of genomic and cDNA sequences hasallowed the gene structure to be determined. The onlyunusual feature is that the second intron is 2,985 bases,the largest identified plant intron. There are also twovery small exons, each 42 bases. These are the sixth andseventh exons (of the gene as so far understood: the 5'end is not yet in hand, and may be in an as-yet-undiscovered, additional exon). The spliced RNA, asindicated by the cDNA clone sequences, contains asingle long open reading frame of, so far, 285 aminoacids. The deduced amino acid sequence of the codedprotein is revealing; the second exon of the AG geneshows a striking homology with the DNA binding (andprotein dimerization) domain of two previously ident-ified DNA-binding transcription factors. These areserum response factor (SRF) of vertebrates, which is atranscriptional regulator of the c-fos oncogene inhumans, and of a cytoskeletal actin gene in Xenopus(Norman et al. 1988; Boxer et al. 1989), and PRTF, theproduct of the Saccharomyces MCM1 gene, a keytranscriptional regulator of mating-type specific genes(Passmore et al. 1989; Jarvis et al. 1989; Herskowitz,1990). The region of similarity between the threeproteins spans 58 amino acids, of which 32 are identical(with no gaps introduced) between the Arabidopsis andyeast genes, and 25 are identical between the humanand Arabidopsis genes. The Arabidopsis gene showsidentical or conserved amino acids at more than 80 % ofthe positions in this 58 amino acid region, whencompared to either of the other genes. The sequenceoutside this DNA binding (and protein dimerization)region is not similar for any pairwise comparison ofyeast, human and plant genes. Details of the cloning ofAGAMOUS, and of the sequencing results, aredescribed by Yanofsky et al. (1990).

Molecular tests of the working model

The sequence of AG is therefore consistent with thegenetic model of its action. It is postulated that AGregulates AP2 expression, and that it acts to regulate aseries of downstream genes as well. A transcriptionfactor could certainly act this way, though of coursemuch more will have to be known about the specificinteractions of AG and other genes and their productsbefore the function and role of the AG protein areunderstood. Many tests of the working model havebeen made possible by obtaining cloned copies of theAG gene, and some tests have already been started.

The first is RNA blot hybridization to determine thespecificity of AG RNA expression. Whole plants weretaken prior to flowering, and poly(A)+ RNA made.Poly(A)+ RNA was also extracted from flowering stemswhich had their apical inflorescence removed. Inaddition, both poly(A)+ RNA and polysomal poly(A)+

RNA were made from young inflorescences. Thesewere subjected to agarose gel electrophoresis, and thegel pattern blotted to a nylon membrane. AG cDNAprobes were hybridized to this blot at high stringency,and a 1.1kb RNA identified in both inflorescencesamples. Quantitative analysis of the blot indicates thatthe RNA is present as one part in 104 of poly(A)+ orpolysomal RNA from whole inflorescences. The RNAwas also identified in the pre-flowering plant RNA andthe floral stem lanes, but at a level about 100 times lessthan in the inflorescence lanes. Since pre-floweringrosette leaves each have small, developmentally ar-rested floral buds in their axils, and since the cauline(stem) leaves of the flowering stem also have arrestedbuds in their axils, it is possible that the signal fromthese lanes is due to AG expression in early floralprimordia, and that AG is exclusively expressed indeveloping flowers. In any event, most if not all of theexpression of the gene is confined to developingflowers.

A more refined examination of the time and place ofAG expression has been started (Yanofsky et al. 1990;G. N. Drews, J. L. Bowman and E. M. Meyerowitz,submitted). An RNA transcript of a portion of one ofthe AG cDNAs (chosen so as not to cross-hybridizewith any of the AG homologues, see below) was labeledwith 35S and hybridized in situ to tissue sections of floralapices. After autoradiography, the pattern of AGexpression was examined. So far, the expressionappears to be confined to developing flowers, and noexpression was seen either in leaves, stems, or in theinflorescence (shoot apical) meristem. The expressionstarts at the time the sepal primordia are first beginningto form (stage 3, Smyth et al. 1990), and before theprimordia of other organs appear. Expression is in thecentral part of the undifferentiated floral primordium,where whorl 3 and 4 organ primordia will later arise.AG RNA continues to be present when organprimordia form in whorls 2, 3 and 4, and theautoradiographic signal is confined to the stamen andovary primordia. The RNA remains present throughoutflower development, with expression in all of the cells of


the developing stamen except for pollen mother cellsand pollen, and in almost all cells of the ovary. TheRNA signal is especially strong in stigmatic tissue and indeveloping ovules (though apparently not over theembryo sac).

The early expression pattern is exactly that predictedby the model, a very encouraging result. One additionalpreliminary set of in situ hybridizations has been done,and these are a key test of the working model. AGprobe has been hybridized to sections of ap2-l andap2-2 floral apices (G. N. Drews, J. L. Bowman and E.M. Meyerowitz, submitted). AG expression expands toinclude all whorls in an ap2-2 background, and inwhorls 2, 3 and 4, but not (or only slightly) in 1, in theap2-l mutant. This is a striking molecular confirmationof the genetic model, which predicted exactly thisinteraction of the different AP2 alleles and the activityof the AG gene (or its product). It is already clear fromthe in situ experiments that the expression pattern ofAG is more complex than that predicted by the model;it is expressed in many differentiated cell types in lateflower development, and there is at least a hint ofdifferent strengths of expression in different flowerregions. There is also apparent down-regulation in thegametophytes of both sexes. Molecular studies of AGare thus certain to lead to refinements of the model, andto begin to take the general and idealized model closerto reality.

The Agamous gene family

One final set of preliminary studies is that of AGhomologues in the Arabidopsis genome. Low strin-gency genomic blots show a series of about 15 bands inthe Arabidopsis thaliana genome that cross-hybridizewith AG cDNA probes, and low-stringency hybridiz-ations with synthetic oligonucleotides representingportions of AG (particularly the SRF-homologousregion) show additional bands. We have so far cloned atleast a dozen of these genes, and have sequenceinformation on six of them (Ma et al. 1991). Each ofthese six genomic regions is expressed as an RNA ininflorescences, which were the source of the RNA forthe cDNA library. The sequence analysis shows that thecoded proteins are related to AG most strongly in theSRF-homologous region, though homology extendsbeyond this. All six of these clones have been analyzedin RNA blots, and five of the six appear to be specific tofloral apices and stems. The sixth is expressed invegetative tissues as well as in flowers. There is thus anintimation that this gene family includes many genesinvolved in flower development, and that the regionthrough which they appear to be most related, the partwith homology to known DNA-binding regions in otherkingdoms, may be some sort of 'floreo' box, much asthe unrelated Drosophila homeobox is a DNA bindingmotif shared by a large fraction of the genes involved inpattern formation processes in fly embryos. Thispossibility is strengthened by the report that a homeoticgene from Antirrhinum (the only plant homeotic gene

other than AG which has been cloned and for which asequence is published) has AGAMOUS homology inthe putative DNA binding site region. This gene, DEFA, has a mutant phenotype rather like that of pi or ap3in Arabidopsis, and not at all like that of ag (Sommer etal. 1990). Thus, there are at least two cloned andsequenced plant homeotic genes that clearly encoderelated proteins. This raises hopes that others of thegenes will be in the same gene family (Schwarz-Sommeret al. 1990). Indeed, preliminary screening of theArabidopsis genome with the DEF A cDNA clone hasrevealed a series of relatives, which seem to be moreclosely related to DEF A than are any of the AGrelatives so far analyzed (T. Jack, L. L. Brockman andE. M. Meyerowitz, unpublished observations). In otherwords, the AG/DEF gene family has an AG and a DEFsubfamily, each with many members. We have startedthe RFLP mapping of members of both subfamilies,and there is evidence that one member of the DEF Asubfamily maps very close to the AP3 gene. Theremarkable homology of homeotic genes of differentfunction, from two widely diverged species, maytherefore serve as a tool for the isolation of additionalhomeotic genes in both species.

Unanswered questions

Other than the apparent gaps, such as the still-unclonedhomeotic genes, and the yet-to-be-performed molecu-lar tests of the genetic model that will only be possiblewhen the cloning of the other genes is complete, thereare additional unanswered questions, for which newtypes of classical and molecular genetic experiments arerequired. These include questions about the exact celltypes in which the homeotic genes act, and the degreeto which they act autonomously in one cell layer andinductively on others. These also include aspects of thephenotypes of the homeotic genes not considered in theworking model, such as the effects that some of thesegenes have on the formation and position of organprimordia, prior to their differentiation. There is alsothe key question of prepattern. Our model postulatesthat each of the homeotic genes comes to be expressedin a specific pattern, encompassing a field thatcomprises two adjacent whorls. While the patterns ofeventual expression of AG and AP2 are to some degreedependent on an interaction between these genes, wedo not know what their initial patterns of expressionmay be, since AP2 is as yet uncloned, and our in situhybridization experiments may be too insensitive toreveal the earliest expression of AG. We also do notknow the mechanism by which PI and AP3 come to beexpressed in the second and third whorls. Finally, thereis another frontier: that of new, as yet undiscoveredmutations. Though we have multiple alleles of each ofthe four homeotic genes, many other, undiscoveredhomeotic loci may exist. They may already be in ourmutant collections, unrecognized, or they may havephenotypes in vegetative as well as floral cells, such thattheir floral phenotypes do not become apparent

164 E. M. Meyerowitz and others

because of earlier lethal effects, or because flowers failto form.

Inductive interactions in organ developmentThe inflorescence meristem and floral primordia of anArabidopsis flowering stem contain two outer (tunica)layers, and internal cells. In other dicotyledonousplants where extensive mosaic studies have been done,it has been shown that each of these three sets of cellsare, for the most part, clonally distinct throughout thedevelopment of the plant, and that each gives rise todifferent parts of each of the floral organs. Epidermalstructures of flowers usually derive from the outertunica layer (LI); the internal cells of sepals and petals,as well as some of the internal cells of stamens andovaries (including the germ cells) derive from thesecond layer (LII); and the deepest cells in the sexorgans usually derive from the inner cells (LIU) (Satinaet al. 1940; Tilney-Basset, 1986). Since each floral organderives from cells of different, clonally distinct, layers,and since all of the cells of each organ are of organ-specific types (each Arabidopsis floral organ has its owndistinguishable subtypes of epidermal and deeper cells:Pruitt et al. 1987; Hill and Lord, 1989; Smyth et al.1990), each of these clonally distinct populations of cellsmust at some point learn its fate, and coordinate itsdevelopment with that of the adjacent cell populations.Cells of each layer may either learn their fatesindependently, or the cells of the different layers maycommunicate, and the phenotypes of some cells resultfrom inductive influences of adjacent cells. It has beenknown for over 70 years, from the study of geneticmosaics, that such induction does indeed occur. Forexample, an epidermis from a hairy tomato straindevelops far fewer hairs than normal if it is growingover second layer cells derived from a hairless strain ofnightshade (Kriiger, 1931). A particularly appositeexample from the old literature is Bateson's discussionof doubleness in chimeric Bouvardia (Bateson, 1916)('double' is the horticultural term for mutations likeagamous, which increase petal number, usually at theexpense of stamens). He cites earlier examples whereroot cuttings from double Bouvardia plants grow up togive single (wild-type) flowers, implying that theoriginal, double-flowered plants were genetic mosaicsin which the clonally distinct layers were not all of thesame genotype (since plants derived from root cuttingsusually derive only from the inner layers). These resultsindicate that the absence of a wild-type flower-formgene in one or more outer cell layers was sufficient tomake the entire flower mutant in phenotype. Thisdemonstrates an inductive influence, of some sort,between the clonally-derived layers of the developingplant. Another example is that of 'Briarcliff roses,which have double flowers, but from which adventitiousbuds or root cuttings give rise to single flowers(Zimmerman and Hitchcock, 1951). This again showsthat, at least in roses, mutations like agamous can havea direct effect in some cell layers, and an inductiveeffect in others. One of the most striking examples of adouble flower mutation showing nonautonomous ex-

pression is in a species chimera of a double (all stamensand carpels converted to petals) Camellia japonica anda wild-type Camellia sasanqua. In this case wild-typelayer I cells, which develop into epidermal cells, directunderlying mutant LII and LIII derivatives to formstamens and carpels, even though these cells wouldnever form these organs, or any of their specialized celltypes, were they covered by an epidermis of their owngenotype (Stewart et al. 1972; Stewart, 1978). Again,the phenotype of the flower seems to depend largely onthe genotype of the epidermal cell layer. A finalexample of the determination of flower form by a singlecell layer is the remarkable case of Antirrhinum majuswettsteinii. Antirrhinum flowers are normally bilaterallysymmetrical. It appears from a considerable series ofexperiments that a radially symmetric flower (like thatseen in cycloidea mutants) is due to a mutation onlypresent in the LI layer, which gives rise to the epidermalcells of the flower. In these flowers a number of testsshows that the LII and LIII cells, which comprise thesubstance of most of the flower, and whose mutantphenotype is directed by the genotype of the epidermis,are genotypically normal (Melchers, 1960; Pohlheim,1978).

The homeotic gene products in Arabidopsis (or anyother plant) could thus determine organ type in at leasttwo different fashions. They could act in cells derivedfrom all of the layers of the flower primordium, or theycould act autonomously in only a subset of the clonallydistinct layers, and influence the differentiation of theother layers inductively. Only genetic mosaic exper-iments will tell us in which distinct layer of cells thesegenes have their primary effect, and which layers, ifany, are affected secondarily. This sort of experimentwill then reveal which of the homeotic phenotypes aredue directly to loss of gene regulators, and which aredue to loss of communication between cells which mustcommunicate developmental information to each other.At this point it should be made explicit that AGappears, from the in situ hybridization experiments, toexpress in cells derived from all three layers. Nonethe-less, this is no guarantee that it acts in all three layers;even if it does, it could act quite differently in cells fromeach layer.

Organ number and patternThe genetic model that we propose does a good job ofpredicting results in terms of organ type, but does notexplain how the homeotic genes under study mayregulate organ number, or the spatial patterns in whichorgan primordia appear. Two examples will suffice toshow the failure of the model in this area. Our modelpredicts A P2 activity to be critical for establishing organtype in whorls 1 and 2 only. There are no A P2 effects onorgan type in whorl 3 (unless AG is also mutant).Nonetheless, ap2-2 and the other strong ap2 mutantsare missing most of their third whorl organs, and theseorgans can be abnormally positioned when they doappear. The strong ap2 mutants also lack some first andall second whorl organs (Komaki et al. 1988; Meyero-witz et al. 1989; Kunst et al. 1989; Bowman et al. 1991).


The absence of these organs is due to failure of organprimordia ever to appear, and not to failure of initiatedprimordia to develop (Bowman et al. 1991).

Another example of the failure of our organ typemodel to treat organ number is the phenotype of the agmutants. While the model explains the identity oforgans in the first four whorls of these flowers (whichmay be considered equivalent to the four whorls of wildtype, but are more reasonably considered whorls 1,2,3and then a repeat of whorl 1, since the organ numberand position in the fourth ag whorl is that of a wild-typewhorl 1, not a wild-type whorl 4), the model does notaddress the indeterminacy of the mutant flowers: theexistence of whorls 5,6,7 and so on. It is possible thatthe absence of organs in ap2 mutants is related to theantagonistic interaction of AP2 and AG. One functionof the wild-type AG product is determinacy, that is, itcauses cessation of organ formation in the center of theflower. The absence of AP2 product in ap2 mutantsallows AG to be expressed in whorls 1 and 2, and at thesame time prevents organ formation in these whorls. Ifthe organ suppression in ap2 mutants is indeed due tothe ectopic expression of AG, the organs should berestored in double mutants of the genotype ap2 ag. Thisis largely true: the double mutant has slightly less thanthe wild-type number of organs, but always has manymore organs than ap2-2 single mutants (Bowman et al.1991). This still does not explain the organ numbereffects of ap2 in the third whorl, however, nor give asuitable mechanism for the AG effect in the central partof developing flowers.

Pattern establishmentAnother question that our model emphasizes, but doesnot answer, is that of the initial expression of thehomeotic genes. For example, we postulate that PI andAP3 are active in only the second and third whorls,whether the other homeotic products are present ornot. In addition, while the model suggests (and themolecular experiments demonstrate) that AP2 and AGinteract in producing their final patterns of expression,the question of their initial patterns of expressionremains. Since the phenotypic effects of ap2 mutants(stage 2 in the system of Smyth et al. 1990) can be seenearlier in flower development than the initial expressionof ag (stage 3), it seems clear that AP2 expressionprecedes that of AG. Thus, the pattern of AGexpression may result entirely from a prepattern of AP2activity. It is not known how AP2 comes to be active inonly the outer whorls. While the homeotic genescontrol downstream genes, they are themselves underthe control of yet other genes, which establish aprepattern to which the homeotic genes respond. Thereis thus a sequence of gene activities that control spatialpatterning in flower development, and the homeoticgenes act at only one time in the overall process ofpattern formation.

We do not yet know which genes may regulate thepattern of homeotic gene expression, though we havecandidate mutations whose phenotypes indicate thepossibility that the wild-type function of the mutated

genes may be in establishment of this pattern. One isthe mutation superman-1, a third chromosome recess-ive mutation whose phenotype appears to be expansionof the third whorl at the expense of the fourth, andperhaps some expansion of the second whorl at theexpense of the third (U. Mayer, G. Jiirgens, D. Weigel,J. Bowman and E. Meyerowitz, unpublished data). Thesuperman-1 mutants have normal numbers of sepals inthe first whorl (4), and of petals in the second (4), butinside these can have additional petalloid sepals, and upto two (or more) whorls of six stamens. Some flowershave normal carpels in their center, though others haveonly rudimentary carpels, or no organs at all (Fig. 6).The phenotype is thus consistent with an alteration inthe prepattern that regulates the initial expression ofthe homeotic genes PI and AP3, and perhaps AP2 aswell. Other explanations are also possible.

Another potential prepattern locus is LEAFY(Haughn and Somerville, 1988; D. Weigel and E.Meyerowitz, unpublished observations). There areeight characterized recessive alleles at this fifth-chromosome locus, and their phenotypes cover a rangeof effects. In some, the transition from vegetative toinflorescence meristem is disrupted, such that the apicalmeristem produces new leaves (with attendant vegetat-ive meristems) in the place of flowers. The mature plantis highly branched, and contains only leaves. Otheralleles have a weaker effect: flowers do form, but all ofthe floral organs are phylloid, with the appearance, forthe most part, of leaves or carpelloid leaves. Thisphenotype resembles that of the triply mutant strainsdescribed above, with one major difference: the patternof organs in these flowers is spiral, not whorled. Onecan imagine, then, that LEAFY expression is anecessary precursor to the formation of the prepatternto which the homeotic genes respond, and in addition,that the formation of this prepattern is related to theprocess of floral induction by a common requirementfor the LEAFY product. LEAFY is thus anothermutation which may be involved in formation of theprepattern. There is little doubt that there are others tobe discovered, as well, and that when the prepattern isbetter understood, the activity of the homeotic genes(and our model for it) will be viewed in a new light.

New genes?One additional question raised by our model is that ofundiscovered genes. A process as complex as patternformation, even in organisms such as plants with theirsimple construction and limited numbers of cell types,must rely on far more regulatory genes than we have asyet discovered. This must indicate that other genes,with roles as critical as those incorporated in our model,have gone undiscovered. Although we have multiplealleles of those genes that we so far recognize ashomeotic, this does not mean that the genome has beenmutagenically saturated. Some may occur more rarelythan those studied; others may have vegetative pheno-types in addition to those in the flower, and haveconsequently gone unrecognized; and yet others mayhave lethal phenotypes, so that only a screen for

166 E. M. Meyerowiiz and others

conditional mutations will reveal them. It seems clearthat one frontier in the genetic study of flowerdevelopment is the design of screens to identify suchmutations.

Conclusions

By studying the phenotypes of a series of homeoticmutations, each of which affects flower development inthe mustard Arabidopsis thaliana, we have derived asimple working model for the process by which groupsof cells in the developing flower 'choose their fate'. Themodel is consistent with the phenotypes of themutations, and correctly predicts the phenotypes ofdoubly and triply mutant strains. In addition, themolecular cloning of one of the Arabidopsis genes hasallowed a series of molecular tests of the genetic model.The results of these tests are also consistent with themodel, and confirm the hypothesis that the normalspatial pattern of expression of the AGAMOUS gene ispartly a result of the activity of the APETALA2 gene.Despite these successes, the model does not explainhow floral organs appear in appropriate positions andappropriate numbers; it only speaks to the specificationof organ fate.

The fact that similar mutations to those described inArabidopsis have been found in Antirrhinum, a plant ina distant family of dicots, indicates the possibility thatthe processes of flower development in all floweringplants are comparable (Coen and Carpenter, 1986;Meyerowitz et al. 1989; Sommer et al. 1990; Carpenterand Coen, 1990; Schwarz-Sommer et al. 1990). Themolecular cloning of homeotic genes from bothArabidopsis and Antirrhinum, and the discovery thatthese genes are members of the same gene family,emphasizes this possibility. It also opens to experimen-tal investigation the molecular basis of the evolutionarydifferences in organ form and pattern that arecharacteristics of flowers of different taxonomic groups.Continued investigation should thus lead not only toanswers to the questions of pattern formation indevelopment, but also to insights into the mechanismsof morphological evolution.

We would like to acknowledge the contributions of theformer student and postdoctoral fellows in our lab whoworked on the flower project: Hong Ma (now at Cold SpringHarbor Labs), Robert E. Pruitt (now at the University ofMinnesota), Usha Vijayraghavan (now at the Indian Instituteof Science, Bangalore) and Martin F. Yanofsky (now at theUniversity of California, San Diego). Also David R. Smyth,of Monash University, Melbourne, who contributed to theproject while here on sabbatical leave.

Our work on flower development is funded by U.S.National Science Foundation grant DCB-8703439, and byfunds from the Lucille P. Markey Charitable Trust. J.L.B.and L.L.B. have been supported by NIH training grant 5T32-GM07616; G.N.D. and T.P.J. are supported by NIHpostdoctoral fellowships, L.E.S. is a fellow of the DamonRunyon-Walter Winchell Cancer Fund, and D.W. is anEMBO long-term fellow.

References

BATESON, W. (1916). Root cuttings, chimaeras and 'sports' J.Genet. 6, reprinted in Scientific Papers of William Bateson. vol.II, 1928. Cambridge University Press: Cambridge.

BERNIER, G (1988). The control of floral evocation andmorphogenesis. A. Rev Plant Phvsiol. Plant molec. Biol. 39.175-219.

BOWMAN, J. L., SMITH, D. R. AND MEYEROWITZ, E. M. (1989).

Genes directing flower development in Arabidopsis. Plant Cell1, 37-52.

BOWMAN, J. L., SMITH, D. R. AND MEYEROWITZ. E. M. (1991).

Genetic interactions between the floral homeotic genes ofArabidopsis. Development 112, 1-19.

BOXER, L. M., PRYWES, R.. ROEDER, R. G. AND KEDES, L. (1989).

The sarcomeric actin CArG-binding factor is indistinguishablefrom the c-fos serum response factor. Molec. cell Biol. 9.515-522.

BRAUN. A. (1873). Ueber ein gefillltes und durchwachsenesExemplar von Arabis Thaliana. Sitzungs Gesell. Naturforsch.Freunde z. Berlin, 75.

CARPENTER, R. AND COEN. E. S. (1990). Floral homeotic mutationsproduced by transposon mutagenesis in Antirrhinum majus.Genes Dev. 4, 1483-1493.

CHANG, C , BOWMAN, J. L., DEJOHN. A W , LANDER, E. S. AND

MEYEROWITZ, E M (1988). Restriction fragment lengthpolymorphism linkage map for Arabidopsis thaliana. Proc. natn.Acad. Sci. U.S.A. 85, 6856-6860.

COEN, E. S. AND CARPENTER, R. (1986). Transposable elements inAntirrhinum majus: generators of genetic diversity. TrendsGenet 2, 292-296.

FELDMANN, K. A., MARKS, M. D., CHRISTIANSON, M. L. AND

QUATRANO, R. S. (1989). A dwarf mutant of Arabidopsisgenerated by T-DNA insertion mutagenesis. Science 243,1351-1354.

GOEBEL, K. (1900). Organographv of Plants (translated by I.Balfour), Clarendon Press: Oxford

GREEN, P. B. (1988). A theory for inflorescence development andflower formation based on morphological and biophysicalanalysis in Echeveria. Planta 175, L53-169.

HAUGHN, G. W. AND SOMERVILLE, C. R. (L988). Genetic control ofmorphogenesis in Arabidopsis Dev Genet. 9, 73-89.

HERSKOWTTZ, I. (1990). A regulatory hierarchy for cellspecialization in yeast. Nature 342. 749-757

HESLOP-HARRISON, J. (1964). Sex expression in flowering plants.Brookhaven Symp. Biol 16, 109-125.

HILL, J. P. AND LORD, E. M. (1989). Floral development inArabidopsis thaliana. a comparison of the wild type and thehomeotic pistillata mutant. Can. J. Bot. 67. 2922-2936.

JARVIS, E. E., CLARK, K. L. AND SPRAGUE, JR, G. F. (1989) The

yeast transcription factor PRTF, a homolog of the mammalianserum response factor, is encoded by the MCM1 gene. GenesDev. 3, 936-945.

KOMAKI, M. K., OKADA. K., NISHINO, E. AND SHIMURA. Y. (1988).

Isolation and characterization of novel mutants of Arabidopsisthaliana defective in flower development. Development 104,195-203.

KOORNNEEF, M. (1989). Linkage map of Arabidopsis thaliana. InGenetic Maps, (5th edition). Cold Spring Harbor.

KROCER, M (1931) Vergleichend-entwicklungsgeschichteUntersuchungen an den Fruchtkoten und Frtichten zweierSolanum-chimaren und lhrer Elternarten. Planta 17, 372-436

KUNST, L., KLENZ, J. E., MARTINEZ-ZAPATER, J. AND HAUGHN, G.

W. (1989). AP2 gene determines the identity of perianth organsin flowers of Arabidopsis thaliana. Plant Cell 1, 1195-1208.

MA, H , YANOFSKY, M. F. AND MEYEROWITZ. E. M. (1991).

AGL1-AGL6, an Arabidopsis gene family with similarity tofloral homeotic and transcription factor genes. Genes Dev 5.484-495.

MELCHERS, G. (1960). Haploide BlUtenpflanzen als Material derMutationszuchtung. Der Ziichter 30. 129-134.

MEYEROWITZ, E. M. (1987). In situ hybridization to RNA in planttissue. Plant Mol Biol. Reporter S, 242-250.


MEYEROWITZ, E. M., SMYTH, D. R. AND BOWMAN, J. L. (1989).

Abnormal flowers and pattern formation in floral development.Development 106, 209-217.

NORMAN, C , RUNSWICK, M., POLLOCK, R. AND TREISMAN, R.

(1988). Isolation and properties of cDNA clones encoding SRF,a transcription factor that binds to the c-fos serum responseelement. Cell 55, 989-1003.

PASSMORE, S., ELBLE, R. AND TYE, B-K. (1989). A proteininvolved in minichromosome maintenance in yeast binds atranscriptional enhancer conserved in eukaryotes. Genes Dev 3,921-935.

PAYER, J-B. (1857) Traiti d'Organogtnie Comparie des la Fleur.Masson: Paris.

POHLHEIM, F. (1978). Untersuchungen an Antirrhinum majuswelsteinii, einer an vegetativ haploiden Pflanzen entstandenenSproBvariante. Zbl. Biol. 97, 53-67.

PRUITT, R. E., CHANG, C , PANG, P. P. Y. AND MEYEROWITZ, E.

M. (1987). Molecular genetics and development of Arabidopsis.In Genetic Regulation of Development, (ed. W. R. Loomis), pp.327-338. A. R. Liss Inc.: New York.

SATINA, S., BLAKESLEE, A. F. AND AVERY, A. G. (1940).

Demonstration of the three germ layers in the shoot apex ofDatura by means of induced polyploidy in periclinal chimeras.Am. J. Bot. 27, 895-905.

SATTLER, R. (1973). Organogenesis of flowers: a photographic text-atlas. Toronto.

SCHWARZ-SOMMER, Z . , HUUSER, P . , N A C K E N , W . , S A E D L E R , H . AND

SOMMER, H. (1990). Genetic control of flower development byhomeotic genes in Antirrhinum majus. Science 250, 931-936.

SMYTH, D. R., BOWMAN, J. L. AND MEYEROWTTZ, E. M. (1990).

Early flower development in Arabidopsis. Plant Cell 2, 755-767.SOMMER, H., BELTRAN, J. P., HUUSER, P., PAPE, H., LONNIG, W.

E., SAEDLER, H. AND SCHWARZ-SOMMER, Z. (1990). Deficiens, ahomeotic gene involved in the control of flower morphogenesisin Antirrhinum majus: the protein shows homology totranscription factors. EMBO J. 9, 605-613.

STEWART, R. N. (1978). Ontogeny of the primary body in chimeralforms of higher plants. In The Clonal Basis of Development,(ed. S. Subtelny and I. Sussex). Academic Press: New York.

STEWART, R. N., MEYER, F. G. AND DERMAN, H. (1972). Camellia

+ 'Daisy Eagleson,' a graft chimera of Camellia sasanqua andC. japonica. Am. J. Bot. 59, 515-524.

TILNEY-BASSET, R. A. E. (1986). Plant Chimeras. E. Arnold:London.

VON GOETHE, J. W. (1790). Versuch die Metamorphose derPflanzen zu erklaren. Ettinger, Gotha (translated by A. Arber(1946). Goethe's botany, Chron. Bot. 10, 63-126.

YANOFSKY, M. F., MA, H., BOWMAN, J. L., DREWS, G. N.,

FELDMANN, K. A. AND MEYEROWTTZ, E. M. (1990). Agamous: anArabidopsis gene whose product resembles transcription factors.Nature 346, 35-39.

ZIMMERMANN, P. W. AND HITCHCOCK, A. E. (1951). Rose 'sports'from adventitious buds. Contr. Boyce Thompson Inst. PI. Res.16, 221-224.

a genetic and molecular model for flower development in...

Documents