Flower development Enrico S. Coen John lnnes Institute,
Norwich,
UK
Several homeotic genes controlling flower development have been characterized in Antirrhinum and Arabidopsis. Comparisons of their mutant phenotypes, expression patterns and genetic interactions have revealed that many of the basic mechanisms controlling flower development have been conserved in evolution, although important differences in the balance and interactions of genes also exist.
Current
Opinion
in Cell
Biology
The growth of a flower depends on the activity of a group of dividing cells, the floral me&tern. Primordia arise as bulges on the flanks of the dome-shaped meristem and these differentiate to the dilferent organs of the flower: sepals, petals, stamens (male organs) and carpels (female organs). In 1990, the first homeotic genes controlling flower development were isolated from two species, Antirrhinum and Arabidopsti. Molecular and genetic studies in these species have provided a general framework for how these genes interact to control flower development (reviewed in [ 1,2] >. In the past year there has been much progress in extending and refining this framework and in determining its degree of conservation in evolution. This review will concentrate on two inter-related areas: the control of me&tern identity and the control of organ identity.
of meristem
4:92%933
and structural comparison of these genes between Antirrhinum and Arabidopsis, species that are thought to have diverged about 70 million years ago, has revealed that homologous genes control similar but not identical processes in each species. In floricauka (flo) mutants of Antirrhinum, inllorescence shoots grow in place of flowers, although very occasionally structures bearing carpel-like organs are produced. The feaJ5) (lb) mutant in A~abidopsis confers a similar phenotype to fro but produces carpel-bearing structures much more readily [3*,4*-l. Using a clone of thejo gene as a probe, the 1& gene was isolated and shown to encode a protein 70% identical to that of the Jo gene product [4*-l. The expression pattern of @ is very similar to that of fro, the RNA being detected by in situ hybridization in very young floral meristems and transiently in flower organ primordia (Fig.la). One difference is that f& RNA is detected in stamen primordia whereas fro RNA is not. Another gene involved in floral meristem identity is squamosa (squu) in Antirrbinum and its counterpart apetalal (apZ> in Arabidopsis [ 5**,6**]. In squa mutants, instead of flowers, inflorescence-like shoots are produced that may eventually give rise to malformed flowers. In apl mutants, flowers are formed in their normal positions but they have inllorescence-like features because further axillary flowers are produced within them. Both squu and apl gene products belong to the MCMl MADS family of transcription factors, which includes mammalian serum response factor and yeast MCMI. Members of this family share a region of homology covering approximately 50 amino acids (the MADS box), but squa and ap1 are more similar to each other over their entire length (68 % identity) than to any other member of this family. Their expression patterns are also very similar; in both cases RNA is detected in very young floral meristerns and in sepal and petal primordia, although squu is also expressed in carpel primordia whereas up1 is not.
Introduction
Control
1992,
identity
Several transitions in meristem identity normally occur before flowers form. During early plant growth, leaves are produced from a vegetative meristem at the shoot apex. When plants enter the reproductive phase of growth, the apical meristem switches to become an inAorescence meristem. In many species, floral meristems arise as small bulges on the flanks of the inflorescence meristem (Fig.la). A pair of key genes involved in the switch of meristems to the floral state has been isolated both from Antirrhinum and Arabidopsis. Mutations in these genes result in floral meristems being replaced by meristems that have some or alI of the characteristics of inflorescence meristems, so that there is a failure or delay in the production of flowers and a proliferation of inflorescence-like structures in their place. Functional Abbreviations ag-agamous g/o-globosa;
MM-mini
; ap-ape&; /Q-leafy;
chromosome
maintenance; @
; def-deliciens; deficiens serum
ten--cenfroradia/is MADGMCMl agamous
Current
p/e--plena; Biology
Ltd
squa-squamosa;
ISSN
0955+X74
f/o-f/oricaula; response factor; tfk-terminal
flower.
930
Cell differentiation (a)
of meristem identity genes between Antirrbintm and Arabiclops& but with important differences in some as-
Growth lnflorescence
(b)
cl
Regions
expressing
b function
genes
q
Regions
expressing
c function
genes
Fig. 1. Typical RNA expression patterns for meristem or organ identity genes. (a) lnflorescence apex with floral meristems initiating on its flanks. Lower meristems represent progressively older stages of development. The regions expressing meristem identity genes are shaded. In the case of Antirrhinum, the floral meristerns are subtended by bracts (not shown). (b) Floral meristem at the floritypic stage showing whorls 14: whorl 1 contains sepals; whorl 2 contains petals; whorl 3 contains stamens; whorl 4 contains carpels. In Antirrhinum, it is not clear whether whorl 4 is defined during or just after the floritypic stage. b and c are homeotic functions specifying organ identity. Mutants lacking b have sepals and carpels instead of petals and stamens. Mutants lacking c have petals and sepals (or petaloid organs) in the place of stamens and carpels, respectively.
The interaction between each pair of floral meristem identity genes, jlo/squu in Antirrbinum or l&jl/apl in Arabidopsti, has also been studied. In j’o mutants, the squu gene is expressed and vice versa, showing that Jo and squa are independently activated [5**] Similarly, lb is expressed in apl mutants [4**]. This independence is continned by the analysis of double mutants. Plants that are mutant for both apl and I& are more extreme in phenotype than either single mutant: they do not produce flowers or carpel-like organs but continue to proliferate inflorescence-like shoots [4*-,7-l. This indicates that these genes are independently activated and that they act synergistically to promote floral development. Similarly, a genetic interaction between Jo and squa has been observed (E Coen, S Doyle, R Carpenter, unpublished data). Thus, in each species, at least a pair of meristem identity genes (flolsqua or I&j/apl) act together to promote the floral programme. The general picture to emerge from these studies is one of an overall concordance in the structure and function
pects of expression or function. A clue to understanding these differences is that their extent often depends on the developmental age or environment of the plant. For example, the production of carpel-like structures on 3~ mutants occurs much more readily at later stages of inflorescence growth and in plants grown in long rather than short day-length conditions [7-l. Similarly, the rare carpels seen on fro mutants are only observed on very old plants (R Carpenter, E Coen, unpublished data). This suggests that the balance between the action of meristem identity genes and their interactions with other gene products could vary, depending on the species, age or environmental circumstances, accounting for the phenotypic variation observed. One area of future research will be to identify and isolate further genes involved in the control of floral meristern identity. Good candidates are centroradialis ( ten) in Atltirrbinum and terminal flower (tji) in Arabidop sis [8-lo]. The inflorescences of both Antirrbinum and Arabidopsis are normally indeterminate: they continue to produce flowers in lateral positions but never at the apex, which would terminate their growth. This correlates with the observed expression ofjo/@ and sqdapl in flanking meristems but not in apical dome of the inflorescence (Fig.la). However, flowers are produced at the apex of ce?zand @mutants and, as expected, this correlates with a gain in expression of genes such as IJ$ [4-l. Furthermore, environmental conditions, such as short day length, which exacerbate the fi phenotype, have the complementary effect of attenuating the @ phenotype. Thus ten and t$ appear to be antagonistic to genes promoting floral development and may prevent their expression in the inflorescence apex.
Control
of organ
identity
Flowers typically consist of concentric whorls of four different types of organs: whorl 1 (outermost) contains small green leaf-like organs, called sepals; whorl 2 consists of petals; whorl 3 contains stamens; and whorl 4 (innermost) contains carpels. Several homeotic mutations affect the identity of organs, and the phenotypes conferred by most of these can be divided into three classes: class a mutants have carpels growing in place of sepals and stamens in place of petals; class b have sepals and carpels instead of petals and stamens, respectively; and class c have petals instead of stamens and sepals or petaloid organs instead of carpels. Each class of mutation has been proposed to correspond to the loss in one of three homeotic functions a, b and c. These homeotic functions act in combination to specify organ type, so that in wild type the combinations in whorls l-4 are a, ab, bc and c, corresponding to sepals, petals, stamens, carpels, respectively [l] To account for single and double mutant phenotypes it is proposed that a and c are antagonistic to each other and act in mutually exclusive domains, but b is established independently
Flower
of a or c Sign&ant progress has been made in understanding the action and regulation of b and the basis of the antagonism between a and c. Two genes, dejcierzs(deJ) and globosa (glo), required for the 6 function in Antirrbinum, have been studied in detail [ 11**,12**]. Both of these genes belong to the MADS family of transcription factors and this has allowed the isolation of glo using a probe that spans the MADS box of dej Homology between these genes is very low outside the MADS box but they show some homology in a region termed the K-box, located downstream of the MADS box. The K-box plays an important role because aminoacid substitutions in this region confer temperature-sensitive mutations in both Antirrhinum and Arabidopsh [11==,13**]. In vitro translation of defand glo cDNAs indicates that the encoded proteins can bind to DNA as heterodimers but not homodimers. The core consensus DNA sequence recognized by the DEF/GLO heterodimer is closely related to that recognized by yeast MCMl and mammalian serum response factor. These in vitro studies can readily explain why mutations in either de/or gfo confer the same class b phenotype: both gene products are partners necessary for the same transcriptional activity. Two genes in Arabidopsis, ap3 and pistilhta, also confer the class b phenotype. Using defas a probe, ap3 has been isolated and shown to encode a product 58% identical to that of defand it seems likely that pistilkzta will turn out to be the counterpart of glo [ 13**]. The expression patterns of deJ glo and ap3 have been determined by RNA in situ hybridizations [ 1loo-13**]. In each case, expression is first detected at a developmental stage when sepal primordia (whorl 1) start to appear as small bulges on the perimeter of the floral meristem. Expression becomes strongest in the regions destined to form petal and stamen pnmordia (whorls 2 and 3; Fig. lb). Thus, even though Arabidopsis meristems are much smaller than those of Antirrhinum, activation of b function genes occurs at a very similar morphological stage of development. This stage, when the sepal primordia are visible but the petal, stamen and carpel primordia are not, has been called the floritypic stage because it may represent an evolutionarily conserved stage at which many of the key homeotic genes have been activated (D Bradley, R Carpenter, H Sommer, N Hartely, E Coen, unpublished data). After the floritypic stage, expression of the b genes continues to be high in whorls 2 and 3 as their primordia appear and develop into mature organs. This expression pattern correlates very well with the proposed action of the b function in whorls 2 and 3, indicating that transcriptional regulation of these genes plays an important role in determining the domain of b activity. Several experiments have addressed the question of how expression of these genes is established and maintained. For early developmental stages, in situ hybridizations using a defprobe on glo mutants, or vice versa, gives similar results to hybridizations with wild type, showing that the establishment of defand glo expression patterns are mutually independent [ 12**]. Similarly, early up3 expression is untiected by ptitilfutu mutations [ 130.1. However, at later developmental stages, similar hybridizations reveal
development
Coen
a marked reduction of expression in mutants relative to wild type, indicating that each gene is involved in the upregulation of its partner. To account for this, it has been proposed that the DEF/GLO heterodimer is directly involved in upregulation of defand glo during late stages of development and possible binding sites for DEF/GLO in the promoters of these genes have been identified [ 120.1. One gene that appears to be involved in the establishment of the 6 expression pattern, called Jo10 or superman, has been identified in Arabidopsis [ 14*,15e]. Plants carrying mutations in this gene have flowers with extra whorls of stamens in their centre instead of carpels. Genetic analysis indicates that the b function is ectopically expressed in whorl 4 of these flowers so that the combination of homeotic functions is a, ab, bc and bc, instead of a, ab, bc and c. This is confirmed by in situ hybridization, which shows that expression of ap3 extends into whorl 4 in young floral meristems of supermun mutants [15-l. Thus one role of superman in wild-type plants is to prevent up3 expression in whorl 4. A second area of intensive research has been analysis of the a and c functions. It has been proposed that a and c are antagonistic so that they act in mutually exclusive domains; a in whorls 1 and 2, which contain sterile organs (sepals and petals); and c in whorls 3 and 4, which contain sexual organs (stamens and carpels). Each of these functions can become active in all four whorls when its antagonist is inactivated by mutation. This is reflected in their complementary mutant phenotypes: flowers with mutations affecting a have sex organs in the outer whorls in place of sterile organs, whereas flowers lacking c have sterile organs in the central whorls instead of sex organs. The agamous (ug) gene is required for the c function in Arabidopstk and encodes a protein belonging to the MADS family [ 11. Using ag as a probe, its counterpart required for the c function in Antirrhinum, plena (pie), was isolated and shown to encode a protein with 64% homology to the aggene product (D Bradley, R Carpenter, H Sommer, N Hartely, E Coen, unpublished data). In situ hybridization shows that ug and pfe RNAs are first detected during the floritypic stage, in a region destined to form stamen and carpel primordia (whorls 3 and 4; Fig lb); expression continues in these whorls as their primordia emerge and develop into mature organs (D Bradley, R Carpenter, H Sommer, N Hartely, E Coen, unpublished data) [16**,17]. This correlates with the proposed region of c activity in whorls 3 and 4, indicating that transcriptional regulation of pfe plays a role in determining the domain of c action. According to this hypothesis, mutants showing a gain of c activity in whorls 1 and 2 should show a corresponding gain in ag or pfe RNA This has been tested in Arabidopsis by in situ hybridization using an ag probe against recessive ap2 mutants that lack the a function. As predicted, high levels of ug RNA were detected in whorls 1 and 2 [16**]. Thus, one role of the a function in Arabidopsis is to prevent ag RNA accumulation in the outer two whorls of the flower. A similar conclusion has been reached from studies in Antirrhinum but through the analysis of a very different type of mutation
931
(D Bradley, R Carpenter, H Sommer, N HaAely, E Coen, unpublished data). Unlike the other mutations so far described, all of which are recessive, mutations affecting the a function in Antirrbinum are semi-dominant, suggesting that they may prevent the action of a rather than causing its loss. Molecular analysis revealed that these mutations were caused by transposon insertions in an intron of the c function gene, pfe, resulting in ectopic expression of ple in whorls 1 and 2. One explanation is that the intron plays a role in’preventing ple expression in whorls 1 and 2 of wild-type flowers and this process can be prevented by insertion of a transposon. These observations also suggest that expression of pfe in whorls 1 and 2 is sufficient to promote the c function and hence sex organ development. A similar conclusion has been derived from studying the effects of ectopically expressing ag in transgenic plants: sex organs are produced in the outer whorls of the flower [ 180•,19.-1.
2.
SCHWNU.SOMMER 2, HUIJSEH P, NACKEN W, SAEDLER H. SOWER H: Genetic Control of Flower Development by Homeotic Genes in Antiwblnum majus. Science 1990. 250:931-936.
SCHVLT~. EA. HAUCHN GW: LEAFY a Homeotic Gene that Regulates Inflorescence Development in Arubidopsis. Plant Cell 1991, 3:771-781. First detailed description of the Ieub phenotype in Arabid@& which shows that flowers are replaced by inflorescence-like structures.
3. .
4. . .
WEICEL D, ALVAREZ J, Sm DR, YANOFSKY MF, MEYEROWIIZ EM: LEAFYControls Floral Me&tern Identity in Arubidop sis Cell 1992, 69843-859. Isolation and chamctetization of I@ from Arubidc@ss showing its homology with the J7o gene of Antiwhinum. Description of & expression pattern in wild-type and mutant background and its synergistic interaction with apI. HUIJSER P. KLEIN J, L~NNIC WE, MEIJER H, SAEDLER H, SOMMER H: Bructeomania an lntlorescence Anomaly is Caused by the Loss of Function of the MAD?+box Gene squamosu in Antiwhinum majus. EMBO J 1992. 11:1239-1250. Isolation and characterization of qua from Antiwbinum; describes its expression pattern and shows that it belongs to the MADS family of transcription factors.
5. . .
6. ..
Concluding
remarks
Rapid strides have been made in our understanding of the control of both meristem and organ identity. Although it has been convenient to distinguish between these two processes, they are unlikely to be independent of each other: for example, mutations affecting organ identity may influence meristem identity and vice versa [4**,7*,10]. One challenge will be to understand the molecular basis of these interactions. A great advantage of studying plants is the large number of genetically well characterized species. Studies in Anfirrhinum and Arabidopsis have revealed that many of the basic mechanisms controlling flower development have been conserved in evolution but important differences in the balance and interactions of genes also exist. These studies are now being extended to other species such as tomato and petunia [ 20,211. The flower therefore provides us with a very attractive system for understanding basic developmental processes and how these can be modulated to provide a diversity of shapes and forms.
References
Desmond Bradley, Bob their helpful comments
and recommended
Papers of particular interest, published view, have been highlighted as: . of special interest .. of outstanding interest 1.
7. .
H~IAIA E, SLISSEX IM: LEAFY Interacts with Floral Homeotic Genes to Regulate Arabidopsis Floral Development. P&ml Cell 1992, 4901-913. Genetic analysis of the interactions between I& and up1 and up2 of Arabidopsis 8.
SHANNON S, MEEKS.WAGNER DR: A Mutation in the sis 7FLl Gene Affects lnllorescence Meristem ment. Pianl Cell 38774392.
9.
ALVAIUZ J, Guu CL, Ylr X.H. WYIH DR: terminal jlower Gene Affecting Inflorescence Development in Arubidopsis thaliuna Plant J 1992, 2:103-116.
10.
COEN ment 1991.
ES: The Role and Evolution. 42:241-279.
Arabidop Developa
of Homeotic Genes in Flower DevelopAnnu Reel Pkznt PLysiol Plan1 Mot Biol
11. . .
SCHWARZ.SOMMER 2, HCIE 1, HUIJSEH P, FLOR PJ, HANSEN R, TETENS F, IDNNIC W-E, SGDIIR H, SOMMER H: Characterization of the AnNwbinum Floral Homeotic MADS-box Gene dejiciens Evidence for DNA Binding and Autoregulation of its Persistent Expression throughout Flower Development. EMBO J 1992, 11:251-263. Demonstration that the proteins encoded by de/and glo can bind to DNA in t&o as heterodimers. Analysis of a temperature..sensitive mutation of dejprovides evidence for autoregulation. 12. ..
Acknowledgements I would like to thank and Sandra Doyle for
MANDEL MA, GUSTAF~ON-BROWN C, SAV~DGE B. YANOFSKY MF: Molecular Characterization of the Arubidopsis Floral Homeotic Gene APETALAI. Nature 1992, in press. Isolation and characterization of up1 from Arubid@sis describing irs expression pattern and its homology with squu of Anfiwhinum.
COEN ES, MEYEROWTIZ Interactions Controlling 353:31-37.
EM: The Flower
Elliott, Paula McSteen on the manuscript.
the annual
JACK T, BROCKMAN LL, MEYEROWTIZ EM: The Homoetic Gene APETU.3 of Arabidopsis thaliana Encodes a MAJX Box and is Expressed in Petals and Stamens. Cell 1992, 68:68ti97. Isolation and chatacterization of up3 from Arubidopsis showing its homology to dejiciens and its expression pattern in wild-type and mutant backgrounds. Demonstration that up3 expression is independent of pisrilhfu expression in young floral meristems.
13. ..
reading within
TROBNER W. RAMIREZ L, Morr~ P, HUE I, HUIJSER P, LONNIG W-E, SAEDUZR H, SUMMER H. SG-IWARZ-SOMMER z: GLOBOSA: a Homeotic Gene which Interacts with deficiens in the Control of Antiwhinum Floral Organogenesis. EMBO J 1992, 11:4693+704. Isolation and characterization of glo from Antiwbinum, showing its expression pattern in various mutant backgrounds. Demonstrates the in. dependence of dejand gloexpression in young meristems but a mutual dependence in later developmental stages.
period
of re-
War of the Whorls: Genetic Development. Nufure 1331,
14. .
SCHULIZ Product
EA, PICKY FB, HAUGHN Regulates the Expression
GW: The /lo10 Gene Domain of Homeotic
Flower Genes AP3 and PI in Arabidopsis 3:1221-1237. Description and genetic analysis ofjlolOof jlol0 negatively regulates ap.3 and pklillala.
Flowers. Plant Cell Arabidr@s&
suggesting
1991, that
11. JACK T, WEICEL D. MAYER U, MEYEROWTIZ a Regulator of Floral Homeotic Genes in Arabidopsis Declelopmenr 1992, 114599. Description and genetic analysis of superman in ArahidopsL~suggesting that superman negatively regulates ap3 and pWil/afa. In sifu hybrid& tions show that this operates at the RNA level in the case of ap.?. 15. .
BOWMAN JL Sw EM: SUPERMAN,
DR~ZWS GN, BOWMAN JL, MIXROWIT~. EM: Negative Regulation of the Arabidopsis Homeotic Gene AGAMOUS by the APElXL42 Product. Ceil 1991, 65991-1002. In situ localization of ag RNA showing that it is expressed ectopically in the outer whorls of up-3 mutants. BOWMAN
JL
Arabidopsis to Specific Cell 1991, 18. ..
Diuws GN, MEYEROWI’IZ EM: Expression of the Floral Homeotic Gene AGAMOUS is Restricted Cell Types Late in Flower Development. Plant 3:749-758.
MANDEL MA, BOWMAN JL, KEMPIN Sq MA H, MEYEROWIIZ EM, YANOFSKY MF: Manipulation of Flower Structure in Transgenie Tobacco. Ce// 1992. 71:133-143.
Coen
Production of transgenic tobacco plants that ectopically express the ag gene from Bras&a napus The plants have sex organs in the outer whorls of their flowers, showing that expression of ag is sufficient to promote the sexual pathway within the context of the flower. MV.UKAMI Y. MA H: Ectopic Expression of the Floral Homeotic Gene agamous in Transgenic Arabidopsis Plants Alters Floral Organ Identity. Cefl 1992, 71:133-143. Ecopic expression of ag in tnnsgenic Arabidqfks plants, showing that expression of ag is suficient to promote sex organ development within the context of the flower. 19. ..
20.
AGENENT GC, BLISCHIZ M, FRANKEN J, MOL JNM, VAN TLINEN AJ: Differential Expression of Two MADS Box Genes in Wild-Type and Mutant Petunia Flowers. Plan1 Cell 1992, 4:983-993.
21.
PNUEU L, AI~AH~III M, ZAMIR D. NAC~N W. SCHWA&SOMMER Z. LIFXHI’II E: The MADS Box Gene Family in Tomato: Temporal Expression during Floral Development, Conserved Secondary Structures and Homology with Homeotic Genes from Antirrbinum and Arabidopsis Plum / 1991. I :255-266.
16. ..
17.
develoDment
ES Coen, Department of Genetics, Norwich NR4 7UH. Norfolk, UK.
John
lnnes
Institute,
Colney
lane,
933
Norwich,
UK
Several homeotic genes controlling flower development have been characterized in Antirrhinum and Arabidopsis. Comparisons of their mutant phenotypes, expression patterns and genetic interactions have revealed that many of the basic mechanisms controlling flower development have been conserved in evolution, although important differences in the balance and interactions of genes also exist.
Current
Opinion
in Cell
Biology
The growth of a flower depends on the activity of a group of dividing cells, the floral me&tern. Primordia arise as bulges on the flanks of the dome-shaped meristem and these differentiate to the dilferent organs of the flower: sepals, petals, stamens (male organs) and carpels (female organs). In 1990, the first homeotic genes controlling flower development were isolated from two species, Antirrhinum and Arabidopsti. Molecular and genetic studies in these species have provided a general framework for how these genes interact to control flower development (reviewed in [ 1,2] >. In the past year there has been much progress in extending and refining this framework and in determining its degree of conservation in evolution. This review will concentrate on two inter-related areas: the control of me&tern identity and the control of organ identity.
of meristem
4:92%933
and structural comparison of these genes between Antirrhinum and Arabidopsis, species that are thought to have diverged about 70 million years ago, has revealed that homologous genes control similar but not identical processes in each species. In floricauka (flo) mutants of Antirrhinum, inllorescence shoots grow in place of flowers, although very occasionally structures bearing carpel-like organs are produced. The feaJ5) (lb) mutant in A~abidopsis confers a similar phenotype to fro but produces carpel-bearing structures much more readily [3*,4*-l. Using a clone of thejo gene as a probe, the 1& gene was isolated and shown to encode a protein 70% identical to that of the Jo gene product [4*-l. The expression pattern of @ is very similar to that of fro, the RNA being detected by in situ hybridization in very young floral meristems and transiently in flower organ primordia (Fig.la). One difference is that f& RNA is detected in stamen primordia whereas fro RNA is not. Another gene involved in floral meristem identity is squamosa (squu) in Antirrbinum and its counterpart apetalal (apZ> in Arabidopsis [ 5**,6**]. In squa mutants, instead of flowers, inflorescence-like shoots are produced that may eventually give rise to malformed flowers. In apl mutants, flowers are formed in their normal positions but they have inllorescence-like features because further axillary flowers are produced within them. Both squu and apl gene products belong to the MCMl MADS family of transcription factors, which includes mammalian serum response factor and yeast MCMI. Members of this family share a region of homology covering approximately 50 amino acids (the MADS box), but squa and ap1 are more similar to each other over their entire length (68 % identity) than to any other member of this family. Their expression patterns are also very similar; in both cases RNA is detected in very young floral meristerns and in sepal and petal primordia, although squu is also expressed in carpel primordia whereas up1 is not.
Introduction
Control
1992,
identity
Several transitions in meristem identity normally occur before flowers form. During early plant growth, leaves are produced from a vegetative meristem at the shoot apex. When plants enter the reproductive phase of growth, the apical meristem switches to become an inAorescence meristem. In many species, floral meristems arise as small bulges on the flanks of the inflorescence meristem (Fig.la). A pair of key genes involved in the switch of meristems to the floral state has been isolated both from Antirrhinum and Arabidopsis. Mutations in these genes result in floral meristems being replaced by meristems that have some or alI of the characteristics of inflorescence meristems, so that there is a failure or delay in the production of flowers and a proliferation of inflorescence-like structures in their place. Functional Abbreviations ag-agamous g/o-globosa;
MM-mini
; ap-ape&; /Q-leafy;
chromosome
maintenance; @
; def-deliciens; deficiens serum
ten--cenfroradia/is MADGMCMl agamous
Current
p/e--plena; Biology
Ltd
squa-squamosa;
ISSN
0955+X74
f/o-f/oricaula; response factor; tfk-terminal
flower.
930
Cell differentiation (a)
of meristem identity genes between Antirrbintm and Arabiclops& but with important differences in some as-
Growth lnflorescence
(b)
cl
Regions
expressing
b function
genes
q
Regions
expressing
c function
genes
Fig. 1. Typical RNA expression patterns for meristem or organ identity genes. (a) lnflorescence apex with floral meristems initiating on its flanks. Lower meristems represent progressively older stages of development. The regions expressing meristem identity genes are shaded. In the case of Antirrhinum, the floral meristerns are subtended by bracts (not shown). (b) Floral meristem at the floritypic stage showing whorls 14: whorl 1 contains sepals; whorl 2 contains petals; whorl 3 contains stamens; whorl 4 contains carpels. In Antirrhinum, it is not clear whether whorl 4 is defined during or just after the floritypic stage. b and c are homeotic functions specifying organ identity. Mutants lacking b have sepals and carpels instead of petals and stamens. Mutants lacking c have petals and sepals (or petaloid organs) in the place of stamens and carpels, respectively.
The interaction between each pair of floral meristem identity genes, jlo/squu in Antirrbinum or l&jl/apl in Arabidopsti, has also been studied. In j’o mutants, the squu gene is expressed and vice versa, showing that Jo and squa are independently activated [5**] Similarly, lb is expressed in apl mutants [4**]. This independence is continned by the analysis of double mutants. Plants that are mutant for both apl and I& are more extreme in phenotype than either single mutant: they do not produce flowers or carpel-like organs but continue to proliferate inflorescence-like shoots [4*-,7-l. This indicates that these genes are independently activated and that they act synergistically to promote floral development. Similarly, a genetic interaction between Jo and squa has been observed (E Coen, S Doyle, R Carpenter, unpublished data). Thus, in each species, at least a pair of meristem identity genes (flolsqua or I&j/apl) act together to promote the floral programme. The general picture to emerge from these studies is one of an overall concordance in the structure and function
pects of expression or function. A clue to understanding these differences is that their extent often depends on the developmental age or environment of the plant. For example, the production of carpel-like structures on 3~ mutants occurs much more readily at later stages of inflorescence growth and in plants grown in long rather than short day-length conditions [7-l. Similarly, the rare carpels seen on fro mutants are only observed on very old plants (R Carpenter, E Coen, unpublished data). This suggests that the balance between the action of meristem identity genes and their interactions with other gene products could vary, depending on the species, age or environmental circumstances, accounting for the phenotypic variation observed. One area of future research will be to identify and isolate further genes involved in the control of floral meristern identity. Good candidates are centroradialis ( ten) in Atltirrbinum and terminal flower (tji) in Arabidop sis [8-lo]. The inflorescences of both Antirrbinum and Arabidopsis are normally indeterminate: they continue to produce flowers in lateral positions but never at the apex, which would terminate their growth. This correlates with the observed expression ofjo/@ and sqdapl in flanking meristems but not in apical dome of the inflorescence (Fig.la). However, flowers are produced at the apex of ce?zand @mutants and, as expected, this correlates with a gain in expression of genes such as IJ$ [4-l. Furthermore, environmental conditions, such as short day length, which exacerbate the fi phenotype, have the complementary effect of attenuating the @ phenotype. Thus ten and t$ appear to be antagonistic to genes promoting floral development and may prevent their expression in the inflorescence apex.
Control
of organ
identity
Flowers typically consist of concentric whorls of four different types of organs: whorl 1 (outermost) contains small green leaf-like organs, called sepals; whorl 2 consists of petals; whorl 3 contains stamens; and whorl 4 (innermost) contains carpels. Several homeotic mutations affect the identity of organs, and the phenotypes conferred by most of these can be divided into three classes: class a mutants have carpels growing in place of sepals and stamens in place of petals; class b have sepals and carpels instead of petals and stamens, respectively; and class c have petals instead of stamens and sepals or petaloid organs instead of carpels. Each class of mutation has been proposed to correspond to the loss in one of three homeotic functions a, b and c. These homeotic functions act in combination to specify organ type, so that in wild type the combinations in whorls l-4 are a, ab, bc and c, corresponding to sepals, petals, stamens, carpels, respectively [l] To account for single and double mutant phenotypes it is proposed that a and c are antagonistic to each other and act in mutually exclusive domains, but b is established independently
Flower
of a or c Sign&ant progress has been made in understanding the action and regulation of b and the basis of the antagonism between a and c. Two genes, dejcierzs(deJ) and globosa (glo), required for the 6 function in Antirrbinum, have been studied in detail [ 11**,12**]. Both of these genes belong to the MADS family of transcription factors and this has allowed the isolation of glo using a probe that spans the MADS box of dej Homology between these genes is very low outside the MADS box but they show some homology in a region termed the K-box, located downstream of the MADS box. The K-box plays an important role because aminoacid substitutions in this region confer temperature-sensitive mutations in both Antirrhinum and Arabidopsh [11==,13**]. In vitro translation of defand glo cDNAs indicates that the encoded proteins can bind to DNA as heterodimers but not homodimers. The core consensus DNA sequence recognized by the DEF/GLO heterodimer is closely related to that recognized by yeast MCMl and mammalian serum response factor. These in vitro studies can readily explain why mutations in either de/or gfo confer the same class b phenotype: both gene products are partners necessary for the same transcriptional activity. Two genes in Arabidopsis, ap3 and pistilhta, also confer the class b phenotype. Using defas a probe, ap3 has been isolated and shown to encode a product 58% identical to that of defand it seems likely that pistilkzta will turn out to be the counterpart of glo [ 13**]. The expression patterns of deJ glo and ap3 have been determined by RNA in situ hybridizations [ 1loo-13**]. In each case, expression is first detected at a developmental stage when sepal primordia (whorl 1) start to appear as small bulges on the perimeter of the floral meristem. Expression becomes strongest in the regions destined to form petal and stamen pnmordia (whorls 2 and 3; Fig. lb). Thus, even though Arabidopsis meristems are much smaller than those of Antirrhinum, activation of b function genes occurs at a very similar morphological stage of development. This stage, when the sepal primordia are visible but the petal, stamen and carpel primordia are not, has been called the floritypic stage because it may represent an evolutionarily conserved stage at which many of the key homeotic genes have been activated (D Bradley, R Carpenter, H Sommer, N Hartely, E Coen, unpublished data). After the floritypic stage, expression of the b genes continues to be high in whorls 2 and 3 as their primordia appear and develop into mature organs. This expression pattern correlates very well with the proposed action of the b function in whorls 2 and 3, indicating that transcriptional regulation of these genes plays an important role in determining the domain of b activity. Several experiments have addressed the question of how expression of these genes is established and maintained. For early developmental stages, in situ hybridizations using a defprobe on glo mutants, or vice versa, gives similar results to hybridizations with wild type, showing that the establishment of defand glo expression patterns are mutually independent [ 12**]. Similarly, early up3 expression is untiected by ptitilfutu mutations [ 130.1. However, at later developmental stages, similar hybridizations reveal
development
Coen
a marked reduction of expression in mutants relative to wild type, indicating that each gene is involved in the upregulation of its partner. To account for this, it has been proposed that the DEF/GLO heterodimer is directly involved in upregulation of defand glo during late stages of development and possible binding sites for DEF/GLO in the promoters of these genes have been identified [ 120.1. One gene that appears to be involved in the establishment of the 6 expression pattern, called Jo10 or superman, has been identified in Arabidopsis [ 14*,15e]. Plants carrying mutations in this gene have flowers with extra whorls of stamens in their centre instead of carpels. Genetic analysis indicates that the b function is ectopically expressed in whorl 4 of these flowers so that the combination of homeotic functions is a, ab, bc and bc, instead of a, ab, bc and c. This is confirmed by in situ hybridization, which shows that expression of ap3 extends into whorl 4 in young floral meristems of supermun mutants [15-l. Thus one role of superman in wild-type plants is to prevent up3 expression in whorl 4. A second area of intensive research has been analysis of the a and c functions. It has been proposed that a and c are antagonistic so that they act in mutually exclusive domains; a in whorls 1 and 2, which contain sterile organs (sepals and petals); and c in whorls 3 and 4, which contain sexual organs (stamens and carpels). Each of these functions can become active in all four whorls when its antagonist is inactivated by mutation. This is reflected in their complementary mutant phenotypes: flowers with mutations affecting a have sex organs in the outer whorls in place of sterile organs, whereas flowers lacking c have sterile organs in the central whorls instead of sex organs. The agamous (ug) gene is required for the c function in Arabidopstk and encodes a protein belonging to the MADS family [ 11. Using ag as a probe, its counterpart required for the c function in Antirrhinum, plena (pie), was isolated and shown to encode a protein with 64% homology to the aggene product (D Bradley, R Carpenter, H Sommer, N Hartely, E Coen, unpublished data). In situ hybridization shows that ug and pfe RNAs are first detected during the floritypic stage, in a region destined to form stamen and carpel primordia (whorls 3 and 4; Fig lb); expression continues in these whorls as their primordia emerge and develop into mature organs (D Bradley, R Carpenter, H Sommer, N Hartely, E Coen, unpublished data) [16**,17]. This correlates with the proposed region of c activity in whorls 3 and 4, indicating that transcriptional regulation of pfe plays a role in determining the domain of c action. According to this hypothesis, mutants showing a gain of c activity in whorls 1 and 2 should show a corresponding gain in ag or pfe RNA This has been tested in Arabidopsis by in situ hybridization using an ag probe against recessive ap2 mutants that lack the a function. As predicted, high levels of ug RNA were detected in whorls 1 and 2 [16**]. Thus, one role of the a function in Arabidopsis is to prevent ag RNA accumulation in the outer two whorls of the flower. A similar conclusion has been reached from studies in Antirrhinum but through the analysis of a very different type of mutation
931
(D Bradley, R Carpenter, H Sommer, N HaAely, E Coen, unpublished data). Unlike the other mutations so far described, all of which are recessive, mutations affecting the a function in Antirrbinum are semi-dominant, suggesting that they may prevent the action of a rather than causing its loss. Molecular analysis revealed that these mutations were caused by transposon insertions in an intron of the c function gene, pfe, resulting in ectopic expression of ple in whorls 1 and 2. One explanation is that the intron plays a role in’preventing ple expression in whorls 1 and 2 of wild-type flowers and this process can be prevented by insertion of a transposon. These observations also suggest that expression of pfe in whorls 1 and 2 is sufficient to promote the c function and hence sex organ development. A similar conclusion has been derived from studying the effects of ectopically expressing ag in transgenic plants: sex organs are produced in the outer whorls of the flower [ 180•,19.-1.
2.
SCHWNU.SOMMER 2, HUIJSEH P, NACKEN W, SAEDLER H. SOWER H: Genetic Control of Flower Development by Homeotic Genes in Antiwblnum majus. Science 1990. 250:931-936.
SCHVLT~. EA. HAUCHN GW: LEAFY a Homeotic Gene that Regulates Inflorescence Development in Arubidopsis. Plant Cell 1991, 3:771-781. First detailed description of the Ieub phenotype in Arabid@& which shows that flowers are replaced by inflorescence-like structures.
3. .
4. . .
WEICEL D, ALVAREZ J, Sm DR, YANOFSKY MF, MEYEROWIIZ EM: LEAFYControls Floral Me&tern Identity in Arubidop sis Cell 1992, 69843-859. Isolation and chamctetization of I@ from Arubidc@ss showing its homology with the J7o gene of Antiwhinum. Description of & expression pattern in wild-type and mutant background and its synergistic interaction with apI. HUIJSER P. KLEIN J, L~NNIC WE, MEIJER H, SAEDLER H, SOMMER H: Bructeomania an lntlorescence Anomaly is Caused by the Loss of Function of the MAD?+box Gene squamosu in Antiwhinum majus. EMBO J 1992. 11:1239-1250. Isolation and characterization of qua from Antiwbinum; describes its expression pattern and shows that it belongs to the MADS family of transcription factors.
5. . .
6. ..
Concluding
remarks
Rapid strides have been made in our understanding of the control of both meristem and organ identity. Although it has been convenient to distinguish between these two processes, they are unlikely to be independent of each other: for example, mutations affecting organ identity may influence meristem identity and vice versa [4**,7*,10]. One challenge will be to understand the molecular basis of these interactions. A great advantage of studying plants is the large number of genetically well characterized species. Studies in Anfirrhinum and Arabidopsis have revealed that many of the basic mechanisms controlling flower development have been conserved in evolution but important differences in the balance and interactions of genes also exist. These studies are now being extended to other species such as tomato and petunia [ 20,211. The flower therefore provides us with a very attractive system for understanding basic developmental processes and how these can be modulated to provide a diversity of shapes and forms.
References
Desmond Bradley, Bob their helpful comments
and recommended
Papers of particular interest, published view, have been highlighted as: . of special interest .. of outstanding interest 1.
7. .
H~IAIA E, SLISSEX IM: LEAFY Interacts with Floral Homeotic Genes to Regulate Arabidopsis Floral Development. P&ml Cell 1992, 4901-913. Genetic analysis of the interactions between I& and up1 and up2 of Arabidopsis 8.
SHANNON S, MEEKS.WAGNER DR: A Mutation in the sis 7FLl Gene Affects lnllorescence Meristem ment. Pianl Cell 38774392.
9.
ALVAIUZ J, Guu CL, Ylr X.H. WYIH DR: terminal jlower Gene Affecting Inflorescence Development in Arubidopsis thaliuna Plant J 1992, 2:103-116.
10.
COEN ment 1991.
ES: The Role and Evolution. 42:241-279.
Arabidop Developa
of Homeotic Genes in Flower DevelopAnnu Reel Pkznt PLysiol Plan1 Mot Biol
11. . .
SCHWARZ.SOMMER 2, HCIE 1, HUIJSEH P, FLOR PJ, HANSEN R, TETENS F, IDNNIC W-E, SGDIIR H, SOMMER H: Characterization of the AnNwbinum Floral Homeotic MADS-box Gene dejiciens Evidence for DNA Binding and Autoregulation of its Persistent Expression throughout Flower Development. EMBO J 1992, 11:251-263. Demonstration that the proteins encoded by de/and glo can bind to DNA in t&o as heterodimers. Analysis of a temperature..sensitive mutation of dejprovides evidence for autoregulation. 12. ..
Acknowledgements I would like to thank and Sandra Doyle for
MANDEL MA, GUSTAF~ON-BROWN C, SAV~DGE B. YANOFSKY MF: Molecular Characterization of the Arubidopsis Floral Homeotic Gene APETALAI. Nature 1992, in press. Isolation and characterization of up1 from Arubid@sis describing irs expression pattern and its homology with squu of Anfiwhinum.
COEN ES, MEYEROWTIZ Interactions Controlling 353:31-37.
EM: The Flower
Elliott, Paula McSteen on the manuscript.
the annual
JACK T, BROCKMAN LL, MEYEROWTIZ EM: The Homoetic Gene APETU.3 of Arabidopsis thaliana Encodes a MAJX Box and is Expressed in Petals and Stamens. Cell 1992, 68:68ti97. Isolation and chatacterization of up3 from Arubidopsis showing its homology to dejiciens and its expression pattern in wild-type and mutant backgrounds. Demonstration that up3 expression is independent of pisrilhfu expression in young floral meristems.
13. ..
reading within
TROBNER W. RAMIREZ L, Morr~ P, HUE I, HUIJSER P, LONNIG W-E, SAEDUZR H, SUMMER H. SG-IWARZ-SOMMER z: GLOBOSA: a Homeotic Gene which Interacts with deficiens in the Control of Antiwhinum Floral Organogenesis. EMBO J 1992, 11:4693+704. Isolation and characterization of glo from Antiwbinum, showing its expression pattern in various mutant backgrounds. Demonstrates the in. dependence of dejand gloexpression in young meristems but a mutual dependence in later developmental stages.
period
of re-
War of the Whorls: Genetic Development. Nufure 1331,
14. .
SCHULIZ Product
EA, PICKY FB, HAUGHN Regulates the Expression
GW: The /lo10 Gene Domain of Homeotic
Flower Genes AP3 and PI in Arabidopsis 3:1221-1237. Description and genetic analysis ofjlolOof jlol0 negatively regulates ap.3 and pklillala.
Flowers. Plant Cell Arabidr@s&
suggesting
1991, that
11. JACK T, WEICEL D. MAYER U, MEYEROWTIZ a Regulator of Floral Homeotic Genes in Arabidopsis Declelopmenr 1992, 114599. Description and genetic analysis of superman in ArahidopsL~suggesting that superman negatively regulates ap3 and pWil/afa. In sifu hybrid& tions show that this operates at the RNA level in the case of ap.?. 15. .
BOWMAN JL Sw EM: SUPERMAN,
DR~ZWS GN, BOWMAN JL, MIXROWIT~. EM: Negative Regulation of the Arabidopsis Homeotic Gene AGAMOUS by the APElXL42 Product. Ceil 1991, 65991-1002. In situ localization of ag RNA showing that it is expressed ectopically in the outer whorls of up-3 mutants. BOWMAN
JL
Arabidopsis to Specific Cell 1991, 18. ..
Diuws GN, MEYEROWI’IZ EM: Expression of the Floral Homeotic Gene AGAMOUS is Restricted Cell Types Late in Flower Development. Plant 3:749-758.
MANDEL MA, BOWMAN JL, KEMPIN Sq MA H, MEYEROWIIZ EM, YANOFSKY MF: Manipulation of Flower Structure in Transgenie Tobacco. Ce// 1992. 71:133-143.
Coen
Production of transgenic tobacco plants that ectopically express the ag gene from Bras&a napus The plants have sex organs in the outer whorls of their flowers, showing that expression of ag is sufficient to promote the sexual pathway within the context of the flower. MV.UKAMI Y. MA H: Ectopic Expression of the Floral Homeotic Gene agamous in Transgenic Arabidopsis Plants Alters Floral Organ Identity. Cefl 1992, 71:133-143. Ecopic expression of ag in tnnsgenic Arabidqfks plants, showing that expression of ag is suficient to promote sex organ development within the context of the flower. 19. ..
20.
AGENENT GC, BLISCHIZ M, FRANKEN J, MOL JNM, VAN TLINEN AJ: Differential Expression of Two MADS Box Genes in Wild-Type and Mutant Petunia Flowers. Plan1 Cell 1992, 4:983-993.
21.
PNUEU L, AI~AH~III M, ZAMIR D. NAC~N W. SCHWA&SOMMER Z. LIFXHI’II E: The MADS Box Gene Family in Tomato: Temporal Expression during Floral Development, Conserved Secondary Structures and Homology with Homeotic Genes from Antirrbinum and Arabidopsis Plum / 1991. I :255-266.
16. ..
17.
develoDment
ES Coen, Department of Genetics, Norwich NR4 7UH. Norfolk, UK.
John
lnnes
Institute,
Colney
lane,
933
