The Plant Cell, Vol. 5, 1183-1193, October 1993 © 1993 American Society of Plant Physiologists Regulation of Arabidopsis Flower Development Jack K. Okamuro,8'1 Bart G. W. den Boer,b and K. Diane Jofuku* a Sinsheimer Laboratories, Department of Biology, University of California, Santa Cruz, California 95064 " Laboratory of Genetics, Universiteit Gent, B-9000 Gent, Belgium INTRODUCTION Plant development is governed by intrinsic and environmental factors that regulate the identity and activity of meristems, organized tissues of pluripotent "stem" cells, that together determine plant form and architecture. However, little is known about how these factors act at the molecular level to affect meristem identity and function. Genetic studies in Arabidopsis and other plant species such as snapdragon, petunia, and maize have revealed a hierarchy of regulatory genes that function together to promote the formation of the floral meristem and to regulate floral organogenesis. Mutations in these genes result in dramatic defects in flower development that can affect both meristem identity and organ development. Table 1 lists the Arabidopsis genes that are known to control meristem and organ identity and their snapdragon equivalents. One class of regulatory genes, the homeotic organ identity genes, have provided important insights into the genetic and molecular mechanisms that govern floral organ identity and development (Bowman etal., 1991; Coen and Meyerowitz, 1991; Coen and Carpenter, 1993, this issue; van der Krol and Chua, 1993, this issue). In Arabidopsis, these genes include APETALA1 {API), APETALA2 (AP2), APETALA3 (AP3), PIST1L-LATA (PI), and AGAMOUS (AG) (Table 1). By contrast, our understanding of the genetic and molecular processes that govern the establishment and maintenance of the floral meristem is still in its infancy. Genetic studies have identified six genes in Arabidopsis that play an important role in the regulation of floral meristem identity and the pattern of meristem development (Table 1). One of these genes, TERMINAL FLOWER1 (TFL1), ensures that the inflorescence and floral meristems remain functionally distinct. The other five genes, LEAFY (LFY), CAULIFLOWER (CAL), API, AP2, and AG together control the pattern of flower development by regulating floral meristem identity. In this review, our discussion will focus on how the analysis of these six genes has provided new genetic, molecular, and physiological insights into the regulation of flower development in Arabidopsis. ' To whom correspondence should be addressed. SEPARATION BETWEEN INFLORESCENCE AND FLORAL MERISTEMS IS MAINTAINED BY TERMINAL FLOWER1 Reproductive development in Arabidopsis is controlled by the activities of the inflorescence and the floral meristems. Each of these meristems can be distinguished by its pattern of apical growth and organogenesis. For example, as illustrated in Figure 1, the inflorescence meristem is characterized by a pattern of indeterminate growth that under long-day growth conditions (16-hr light/8-hr dark) results in the production of two to four cauline leaves and lateral or secondary inflorescences, followed by the production of flowers. Each of these structures is produced by a meristem or organ primordium that was produced by and emerged from the flanks of the inflorescence meristem. By contrast, the floral meristem displays a determinate pattern of cell division and organogenesis, resulting in the production of four concentric rings or whorls of floral organs (sepals, petals, stamens, and carpels) that comprise the Arabidopsis flower (Figure 1). Although the inflorescence and floral meristems are closely related both spatially and by cell lineage, each meristem must maintain a separate identity and follow a different developmental pathway to carry out its unique functions. They do so in part through the activity of the TFL1 gene. TFL1 is responsible for the maintenance of the inflorescence meristem and the regulation of floral meristem production, till mutants are characterized by both early flowering and the conversion of the inflorescence meristem into a terminal flower (Shannon and Meeks-Wagner, 1991; Alvarez et al., 1992). This conversion limits the development of the normally indeterminate inflorescence and results in a dramatic loss of both lateral branches and flowers, as illustrated in Figure 2. In addition, till plants often produce one or two lateral branches that are terminated by a single wild-type flower. Genetic studies have shown that several genes, including LFY, API, and AP2, are involved in the establishment of the floral meristem (Irish and Sussex, 1990; Huala and Sussex, 1992; Weigel et al., 1992; Coen and Carpenter, 1993, this issue). For example, mutations in Ify cause a partial conversion of the floral meristem to an inflorescence (Table 1). It has been suggested that TFL1 and these floral meristem-promoting genes function antagonistically in the inflorescence and floral meristems to maintain the separation of meristem functions 1184 The Plant Cell Table 1. Genes Involved in the Regulation of Meristem and Floral Organ Identity in Arabidopsis Genetic Function/ Proposed Molecular Snapdragon Gene Locus Mutant Phenotype Function(s) Homolog References" Inflorescence Meristem Identity TERMINAL FLOWER {TFL1) Early flowering; conversion of Negative regulator of LFY, NDb 1, 2 inflorescence to floral AP1, and AP2 meristem Floral Meristem Identity AGAMOUS (AG) Indeterminate and repetitive Putative transcription factor PLENA (PLE) 3, 4, 5, 6 pattern of floral organogenesis resulting in a "flower-within-a-flower" [(sepal, petal, petal)n] APETALA1 (AP1) Production of axillary flowers Putative transcription factor SQUAMOSA (SQUA) 7, 8, 9 APETALA2 (AP2) Similar to ap7 under short- Negative regulator of AG ND 10, 11, 12, 13 day growth conditions CAULIFLOWER (CAL) Phenotypically wild type; ND ND 14 however, apl cal double mutants display a conversion of the floral meristem to an inflorescence LEAFY (LFY) Partial conversion of floral Putative transcription factor FLORICAULA (FLO) 15, 16, 17, 18 meristems to inflorescence Positive regulator of AP3 shoots and PI Floral Organ Identity AGAMOUS (AG) Homeotic conversion of Putative transcription factor PLENA (PLE) 3, 4, 5, 6 stamens to petals and of Negative regulator of API, carpels to sepals AP2, and AP3 APETALA1 (API) Homeotic conversion of Putative transcription factor SQUAMOSA (SQUA) 7, 8, 9 sepals to leaves; absence of petals APETALA2 (AP2) Homeotic conversion of Negative regulator of AG ND 10, 11, 12, 13 sepals to leaves or carpels and of petals to stamens APETALA3 (AP3) Homeotic conversion of Putative transcription factor DEFICIENS (DEF) 19, 20 petals to sepals and of Positive regulator of PI stamens to carpels PISTILLATA (PI) Similar to ap3 Putative transcription factor GLOBOSA (GLO) 21, 22 a (1) Shannon and Meeks-Wagner (1991); (2) Alvarez et al. (1992); (3) Bowman et al. (1989); (4) Yanofsky et al. (1990); (5) Carpenter and Coen (1990); (6) Bradley et al. (1993); (7) Irish and Sussex (1990); (8) Mandel et al. (1992); (9) Huijser et al. (1992); (10) Bowman et al. (1991); (11) Komaki et al. (1989); (12) Kunst et al. (1989); (13) Jofuku et al. (1993); (14) Bowman (1992); (15) Schultz and Haughn (1991); (16) Huala and Sussex (1992); (17) Weigel et al. (1992); (18) Weigel and Meyerowitz (1993); (19) Jack et al. (1992); (20) Sommer et al. (1990); (21) Hill and Lord (1989); (22) Trobner et al. (1992). b ND, not determined. (Shannon and Meeks-Wagner, 1993). Double mutant studies showed that TFL1 suppresses the activities of LFY, AP1, and AP2 in the inflorescence meristem, and, conversely, that LFY, API, and AP2 suppress TFL1 activity in the floral meristem (Shannon and Meeks-Wagner, 1993). In addition, in situ hybridization analysis demonstrated at the molecular level that TFL1 negatively regulates LFY gene expression in the inflorescence meristem (Weigel et al., 1992). Although it is not yet known how TFL1 and LFY, AP1, and AP2 carry out these functions, it is clear that the antagonistic relationship between these genes is essential for maintaining the separation between the inflorescence and floral meristems. The effects of till mutations on inflorescence development are determined in part by environmental conditions. Changes in both photoperiod and temperature have striking effects on mutant inflorescence development (Shannon and Meeks-Wagner, 1991; Alvarez et al., 1992). For example, till mutant plants grown under very long photoperiods (20-hr light/4-hr Arabidopsis Flower Development 1185 Primary ^ inflorescence \ Secondary inflorescence \ Cauline leaf Rosette leaf dark) are phenotypically more extreme than tfH plants grown under long-day conditions and may produce only a single flower (Shannon and Meeks-Wagner, 1991). By contrast, short day-grown tfll plants (10-hr light/14-hr dark) are phenotypically much closer to wild type, producing both lateral inflorescences and flowers. Thus, long days enhance and short days suppress the tfll mutant phenotype. tfll mutants are also conditional with respect to temperature. Plants grown at 30°C display a more extreme phenotype than plants grown at the standard temperature of 22°C, whereas plants grown at 15°C are almost wild type (Alvarez et al., 1992). Although the molecular nature of the tfll mutations are not yet known, these studies together suggest that TFL1 may mediate the activity of a physiological inhibitor of floral meristem initiation (Alvarez et al., 1992). Figure 1. Schematic Representation of a Wild-Type Arabidopsis Inflorescence and Flower. In Arabidopsis, the primary inflorescence consists of an inflorescence stem with a number of cauline leaves, lateral or secondary inflorescences, and flowers that are produced in a spiral pattern of phyllotaxis. The Arabidopsis flower consists of four concentric rings or whorls of floral organs: four sepals (S), four petals (P), six stamens (St), and two fused ovule-bearing carpels (C). The closed circle represents the position of the inflorescence meristem with respect to the flower. LEAFY PROMOTES AND MAINTAINS FLORAL MERISTEM IDENTITY LFY is a central player in the establishment of floral meristem identity and the regulation of flower homeotic gene expression. Mutations in LFY cause a partial block or delay in floral meristem production by the inflorescence meristem (Schultz WT mi Ify Ify ap 1 Figure 2. Genetic Control of Floral Meristem Production by TERMINAL FLOWER1, LEAFY, APETALA1, and APETALA2. WT, tfll, Ify, and Ify ap1 refer to schematic representations of wild-type, terminal tlcmer1, leafy, and leafy apetalal plants, respectively, tfll mutants are characterized by early flowering and the conversion of the inflorescence meristem into a terminal flower (Shannon and Meeks-Wagner, 1991; Alvarez et al., 1992). Under long-day conditions, mutant plants are characterized by the lack of lateral shoots and the production of only a few flowers. Severe Ify mutants are characterized by a partial block or delay in floral meristem production by the inflorescence meristem (Schultz and Haughn, 1991; Weigel et al., 1992). Flowers are replaced by shoots or flowerlike shoots (represented by green flowers). Ifyapl double mutants are characterized by a strong block in floral meristem production (Huala and Sussex, 1992; Weigel et al., 1992). Flowers are replaced by lateral shoots. Ify ap2-1 mutants are similar in phenotype to Ify ap1 double mutants (Huala and Sussex, 1992). 1186 The Plant Cell and Haughn, 1991; Huala and Sussex, 1992; Weigel et al., 1992). Thus, in strong Ify mutants, the first flowers produced by the primary inflorescence are replaced by shoots that are characterized by extended internodes, spiral phyllotaxis, and an indeterminate pattern of growth (Figure 2). As the inflorescence continues to grow, the lateral shoots become progressively more compact and flowerlike. However, these "flowers" are still phenotypically mutant, bearing sepals and carpel-like organs but no petals or stamens. At the molecular level, LFY belongs to an evolutionarily conserved family of plant genes that includes the snapdragon gene FLORICAULA (FLO) (Weigel et al., 1992). Like Ify mutants, flo mutants are defective in the transition from inflorescence to floral meristem (Coen et al., 1990). Phenotypically, however, flo mutants are much stronger than Ify mutants, showing a complete conversion of flowers to shoots. It has been hypothesized that the proteins encoded by both FLO and LFY may function as transcription factors because they contain a proline-rich region as well as acidic and basic regions that are often found in eukaryotic transcription factors (Coen et al., 1990; Weigel et al., 1992). However, no other sequence similarity between LFY/FLO and other eukaryotic transcription factors is detectable, suggesting that these genes encode a novel class of plant regulatory protein. One function of LFY may be to act as a genetic "trigger" for flower development by positively regulating homeotic gene expression. Consistent with this proposed function, LFY is the first known gene in the cascade of flower-specific meristem and organ identity genes to be activated during flower development (Weigel et al., 1992). As shown in Figure 3, LFY gene expression is detectable in the floral anlage even before the floral primordium becomes visible. As flower development progresses, LFY is expressed continuously and uniformly in young floral primordia. During organ development, LFY is highly expressed in the primordia of all four types of floral organs and appears to act as a positive regulator of AP3, PI, and AG gene expression (Weigel and Meyerowitz, 1993). LFY expression is transient, however, with transcript levels decreasing to undetectable levels in all four types of floral organs shortly before they reach maturity. Recent studies have shown that the LFY gene is also required for the maintenance of floral meristem identity. Changes in photoperiod can dramatically alter floral meristem identity and development in both homozygous and heterozygous Ify mutant flowers (J.K. Okamuro, B.G.W. den Boer, C. Lotys, and K.D. Jofuku, manuscript submitted). For example, under long-day conditions, heterozygous flowers are phenotypically wild type. By contrast, Figure 4 shows that under short-day Developmental Stage 12 IM, 1,2 3 v 7 I AP2 MADS BOX GENES AP1 AP3 AG ' vr^y vr^"^ w r? o Figure 3. Temporal and Spatial Regulation of Arabidopsis Meristem and Organ Identity Gene Expression during Flower Development. This diagram illustrates both the temporal and spatial patterns of gene expression during flower development for the meristem and organ identity genes APETALA2 (AP2) (K.D. Jofuku, B.G.W. den Boer, M. Van Montagu, and J.K. Okamuro, manuscript submitted), LEAFY (LFY) (Weigel et al., 1992), APETALA1 (AP1) (Mandel et al., 1992), APETALA3 (AP3) (Jack et al., 1992), and AGAMOUS (AG) (Drews et al., 1991). Successive stages of flower development are represented by longitudinal sections through the inflorescence meristem (IM), the floral meristem (FM), and stage 1, 2, 3, and 7 flowers (as described by Smyth et al., 1990). Mature stage 12 flowers are represented by transverse sections showing all four types of floral organs: sepals (S), petals (P), stamens (St), and carpels (C). Regions of darker and lighter shading within a flower represent quantitative differences in transcript levels for AP2, LFY and API. Arabidopsis Flower Development 1187 Figure 4. Physiological Control of Floral Meristem Identity in Heterozygous leafy Flowers. (A) A heterozygous leafy-6 (lfy-6) flower under long-day growth conditions. Heterozygous Ify flowers are morphologically and functionally normal. (B) Short-day-induced floral reversion in a heterozygous lfy-6 flower. The mutant flowers are initially phenotypically wild type, characterized by a normal complement of sepals, petals, stamens, and carpels. However, instead of forming a normal fruit, the ovary becomes swollen and is eventually forced open by the growth of an ectopically formed inflorescence (J.K. Okamuro, B.G.W. den Boer, C. Lotys, and K.D. Jofuku, manuscript submitted). The ectopic inflorescence is characterized by elongated internodes, spiral phyllotaxis, and the production of many lateral flowers. (C) Exogenously applied gibberellin suppresses floral meristem reversion in heterozygous Ify flowers. Gibberellin-treated heterozygous Ify flowers produce a normal complement of floral organs under short-day conditions (J.K. Okamuro, B.G.W. den Boer, C. Lotys, and K.D. Jofuku, manuscript submitted). conditions (9-hr light/15-hr dark), these flowers undergo a reversion to inflorescence development after producing a normal complement of floral organs. Thus, a heterozygous "mutant" flower consists of sepals, petals, stamens, carpels, and an ectopic inflorescence. This photoperiod-dependent transformation or reversion to inflorescence development clearly demonstrates that floral meristem identity and determinacy are not irreversibly fixed in the floral meristem. Moreover, it shows that the floral meristem is capable of sustained growth and development even after forming a normal complement of floral organs. Although the phenomenon of floral reversion is not unique to Arabidopsis and has been demonstrated for several plant species, including Triticum aestivum (wheat), Impatiens balsamina, and Anagallis (reviewed by Battey and Lyndon, 1990), LFY is the first gene to be implicated in the control of this process. Thus, the characterization of LFY function may provide clues into the molecular mechanism that regulates the maintenance of floral meristem identity under changing growth conditions. APETALA1 CONTROLS FLORAL MERISTEM AND ORGAN IDENTITY AT THE TRANSCRIPTIONAL LEVEL AP1 plays a dual role in the regulation of floral meristem identity and the control of floral organ identity and development. ap1 mutants are characterized by the partial conversion of the floral meristem into an inflorescence meristem, resulting in the production of secondary flowers within the axils of the first whorl floral organs (Irish and Sussex, 1990; Bowman, 1992). The flowers produced by ap1 mutants also show a homeotic conversion of sepals into bractlike leaves, the loss of petals, and a normal complement of stamens and carpels. In addition, recent experiments have shown that the inflorescence-like character of the apl mutant is enhanced under short-day growth conditions, suggesting that API, like the meristem identity gene LFY, is involved in the physiological regulation of floral meristem identity and development (J.K. Okamuro, B.G.W. den Boer, C. Lotys, and K.D. Jofuku, unpublished results). The role of API in the establishment of floral meristem identity is clearly revealed in double mutant studies by combining ap1 with other meristem mutations. For example, cauliflower (cal) is a genetic enhancer of the apl floral phenotype (Bowman, 1992). cal mutants have little if any discernible phenotype of their own. However, cal can transform the floral meristem into an inflorescence when in double mutant combination with apt Thus, AP1 and CAL function together to promote the transition from inflorescence to floral meristem. The molecular nature of the AP1ICAL interaction is not yet known. The establishment of the floral meristem is also due in part to a synergistic interaction between AP1 and LFY. For example, by combining the Ify and apl mutations together in a Ify apl double mutant, flower production is blocked and floral meristems are converted into shoots, as shown in Figure 2 (Huala and Sussex, 1992; Weigel et al., 1992). This synergism is also reflected at the molecular level. In situ hybridization experiments have shown that apl and Ify single mutants do not strongly suppress the flower-specific expression of the homeotic organ identity genes AP3 and PI. By contrast, AP3 and PI gene expression is completely suppressed in most "flowers" of the Ify apl double mutant (Weigel and Meyerowitz, 1188 The Plant Cell 1993). Although the molecular basis of the interaction between API and LFY and its effects on homeotic gene expression have not yet been determined, it is clear that AP1 and LFY function cooperatively to promote floral meristem identity. Molecular studies have shown that there is a strong correlation between the temporal and spatial regulation of AP7 gene expression and its genetic functions in the control of floral meristem and organ identity and development (Mandel et al., 1992). AP1 transcripts are first detectable in very young floral buds, before the emergence of the first organ primordia and after the onset of LFY gene expression (Figure 3). As flower development progresses, AP1 gene expression expands to include both sepal and petal organ primordia but is excluded from developing stamens and carpels by another homeotic gene, AG (Mandel et al., 1992). This antagonistic relationship between API and AG establishes an important boundary of homeotic gene activity between the petal and stamen primordia in the developing flower. Similarly, the temporal and spatial boundaries of AG gene expression are controlled in part by an antagonistic interaction with AP2 (Drews et al., 1991). Together, these interactions illustrate an important regulatory mechanism in pattern formation that is used by both plants and animals (Coen and Meyerowitz, 1991; Nusslein-Volhard, 1991). How does AP1 regulate flower meristem identity and organ development at the molecular level? API belongs to an evolutionary conserved family of transcription factors that includes the Arabidopsis homeotic genes AG, AP3, and PI as well as their snapdragon homologs SQUAMOSA (SQUA), PLENA (PLE), DEFICIENS (DEF), and GLOBOSA (GLO) (Table 1). Each of these regulatory factors is characterized by a highly conserved 58-amino acid DNA binding domain called the MADS box (derived from yeast MCM1 [minichromosome maintenance-1; Ammerer, 1990], Arabidopsis AG [Yanofsky et al., 1990], snapdragon DEF [Sommer et al., 1990], and human SRF [serum response factor; Norman et al., 1988]). Much of what is known about the functions of these proteins comes from in vitro studies of MCM1 and SRF (reviewed by Treisman and Ammerer, 1992). These studies showed that MCM1 and SRF function as dimers to regulate gene expression in response to extracellular signals and that their activity is regulated in part by their interactions with cell-specific accessory transcription factors. MCM1 and SRF share a high degree of sequence similarity within a 91-amino acid "core domain" that includes the MADS box. This core domain contains all sequences necessary for DNA recognition and binding, protein dimeriza-tion, and interaction with accessory transcription factors (Norman et al., 1988; Ammerer, 1990; Christ and Tye, 1991; Mueller and Nordheim, 1991; Primig et al., 1991). AP1, like other plant MADS box-containing proteins, shows no significant homology to MCM1 or SRF outside the MADS box (Mandel et al., 1992). To date, the only flower homeotic proteins studied in detail with respect to their structure and function are AG, DEF, and GLO (Mueller and Nordheim, 1991; Schwarz-Sommer et al., 1992; Trobner et al., 1992; Shiraishi et al., 1993). In vitro studies have shown that the MADS box domain from AG and DEF, and by inference AP1, is sufficient for DNA recognition and binding (Mueller and Nordheim, 1991; Shiraishi et al., 1993). However, these experiments do not address the question of what sequences in these proteins are responsible for their ability to interact with other proteins to activate or repress gene transcription. It has been shown that the conserved carboxyl region of the SRF/MCM1 core domain mediates the interactions of SRF and MCM1 with cell-specific accessory transcription factors, and it is possible that this region may serve a similar function for other MADS box-containing proteins, such as AP1. Alternatively, sequences found elsewhere in these proteins could serve this function. The plant MADS box-containing genes share a conserved 65-amino acid region that is located near the carboxyl end of the MADS box domain and is not found in either MCM1 or SRF. This second domain of homology, called the K-box because it bears structural similarity to the interacting region of keratins, is capable of forming a coiled-coil structure due to the propensity of regions within this domain to form am-phipathic a-helices (Ma et al., 1991). The formation of coiled-coils mediates protein-protein interactions in both structural and regulatory proteins such as yeast GCN4 (O'Shea et al., 1989) and the BZLF1 gene product of the Epstein-Barr virus (Flemington and Speck, 1990) and could function similarly for AP1 and other plant MADS box-containing proteins to mediate dimerization or possible interactions with accessory transcription factors. AGAMOUS IS INVOLVED IN THE TERMINATION OF FLOWER DEVELOPMENT The floral homeotic gene AG plays a critical role in regulating the second half of flower ontogeny—the development of stamens and carpels and the termination of flower development. ag flowers are characterized by the absence of carpels and the homeotic conversion of stamens to petals. In addition, the floral meristem displays an indeterminate or repetitive pattern of organogenesis that results in a (sepal, petal, petal),, structure referred to as a flower-within-a-flower (Bowman et al., 1989). Together, these mutant characteristics indicate that AG is necessary for the genetic control of both floral meristem determinacy and floral organ identity. The effects of ag mutations on floral organ identity may be due in part to the role of AG as a negative regulator of AP1 gene expression. In wild-type flowers, API activity is restricted to developing sepals and petals (Mandel et al., 1992). In ag flowers, however, the API gene is ectopically expressed in third whorl organ primordia and thus may be responsible for the homeotic conversion of stamens to petals (Irish and Sussex, 1990; Mandel et al., 1992). By contrast, the indeterminate or repetitive pattern of ag flower organogenesis does not result Arabidopsis Flower Development 1189 from the ectopic activity of either API or LFY, as shown by double mutant studies (Irish and Sussex, 1990; Weigel et al., 1993). Under short-day conditions, ag flowers may also undergo a heterochronic reversion of the floral meristem to an inflorescence meristem similar to that seen in flowers from plants heterozygous for the Ify mutation (J.K. Okamuro, B.G.W. den Boer, C. Lotys, and K.D. Jofuku, manuscript submitted). That is, after producing several whorls of sepals and petals, the floral meristem reverts to the production of axillary floral buds, extended internodes, and spiral phyllotaxis, as shown in Figure 5. These mutant flowers are usually found along the primary inflorescence at the normal point of transition from shoot to flower production. This short-day-induced reversion from flower to inflorescence meristem in ag flowers again demonstrates that floral meristem identity is not fixed but is maintained physiologically by AG as well as by LFY. AG, like API and LFY, shows a temporal and spatial pattern of gene expression that reflects its genetic functions. AG RNA transcripts appear later in flower development than either LFY ox API transcripts and are first detectable in the central region of the flower primordium after the sepal primordia have begun to emerge from the flanks of the floral meristem (Figure 3). As organogenesis progresses, AG gene expression is restricted to developing stamens and carpels and is undetectable in both sepals and petals. Both the temporal and spatial pattern of AG gene expression is determined in part by the homeotic gene AP2, which negatively regulates AG gene expression in early floral buds, in sepals, and in petals (Drews et al., 1991). The molecular nature of the AP2-AG interaction is not yet known. APETALA2 FUNCTIONS THROUGHOUT FLOWER DEVELOPMENT AP2, like AG, is involved in the genetic control of both meristem identity and floral organ identity. ap2 mutant flowers are characterized by a broad spectrum of related phenotypes, and mutations in AP2 affect the identity or development of all four floral organs (Komaki et al., 1988; Bowman et al., 1989,1991; Kunst et al., 1989; K.D. Jofuku, B.G.W. den Boer, M. Van Montagu, and J.K. Okamuro, manuscript submitted). Severe ap2 mutant flowers are characterized by the homeotic transformation of sepals into ovule-bearing carpels, the absence of petals, a reduction in stamen number, and defects in carpel fusion. By contrast, weak ap2 mutant flowers are characterized by the homeotic conversions of sepals into leaves and of petals into pollen-producing stamens, and they contain a normal complement of stamens and carpels. The effects of ap2 mutations on flower organ identity are due in part to the role of AP2 in the temporal and spatial regulation of AG gene expression. In wild-type flowers, AG gene activity is confined to developing stamens and carpels. In strong ap2 mutants, however, the AG gene is activated at an earlier stage of flower development and is ectopically expressed in both first and second whorl organ primordia, resulting in the homeotic conversion of sepals to carpels and an absence of petals (Drews et al., 1991). The loss of second and third whorl organs in strong ap2 mutants is due to ectopic AG gene activity, because these organs are restored in ap2 ag double mutants (Bowman et al., 1991). This negative effect of AG on Figure 5. Physiological Control of Floral Meristem Identity in agamous Flowers. (A) An agamous-1 (ag-1) mutant flower grown under standard long-day conditions (16-hr light/8-hr dark). The mutant flower phenotype consists of the repeating flower-within-a-flower pattern of organogenesis (sepal, petal, petal),,. (B) Short-day-induced floral meristem reversion of an ag flower. After producing several whorls of sepals and petals, the floral meristem reverts to an inflorescence meristem under short-day conditions (J.K. Okamuro, B.G.W. den Boer, C. Lotys, and K.D. Jofuku, manuscript submitted). The ectopic inflorescence displays both elongated internodes, spiral phyllotaxis, and the production of lateral flowers. (C) Exogenously applied gibberellin suppresses floral meristem reversion in short-day-induced ag flowers. 1190 The Plant Cell petal development was clearly demonstrated by Mizukami and Ma (1992), who induced "ap2-like" mutants in transgenic wild-type plants by the constitutive ectopic expression of a chimeric AG gene in the flower. In addition to its role in the control of floral organ identity, AP2 also contributes to the establishment of floral meristem identity. In response to short-day photoperiod, several weak ap2 mutants will produce secondary flowers in the axils of the first whorl organs of the flower and will fail to make petals (Komaki et al., 1988; Shannon and Meeks-Wagner, 1993; J.K. Okamuro, B.G.W. den Boer, and K.D. Jofuku, unpublished observations). This short-day phenotype resembles that of ap1 flowers grown under long-day conditions and suggests that AP2 and AP1 affect similar processes in flower development. In addition, genetic studies have shown that AP2 promotes floral meristem development through synergistic interactions with both AP1 and LFY. For example, mutations in Ify do not block flower development completely. However, weak mutations in ap2, such as ap2-1, when in combination with Ify, can completely block the formation of flowers, resulting in the conversion of whorled flowers into shoots (Huala and Sussex, 1992). Moreover, flowers are converted into indeterminate inflorescence branches in ap2-7 ap1-1 double mutants (Irish and Sussex, 1990). The mechanism by which these three genes together promote floral meristem development has not yet been determined. The molecular analysis of AP2 gene expression has shown that AP2 is expressed continuously in both the inflorescence and floral meristems and in all four types of floral organs (Figure 3; K.D. Jofuku, B.G.W. den Boer, M. Van Montagu, and J.K. Okamuro, manuscript submitted). The sustained expression of AP2 during inflorescence and early flower development provides a striking contrast to the sequential and spatially overlapping expression domains of the MADS box-containing genes API, AP3, and AG and to the transient expression of LFY in all four types of floral organs. Thus, one important conclusion from the molecular analysis of AP2 gene expression is that its expression during flower development overlaps with the expression of all three MADS box-containing genes, including AG. This indicates that AG does not suppress AP2 gene transcription in stamens and carpels as was proposed by Meyerowitz et al. (1991) and that AP2 expression in these organs does not suppress AG gene expression as it does in sepals and petals. Thus, it appears that AP2 alone is not capable of suppressing AG gene activity but may function cooperatively with an unidentified gene whose activity is restricted to sepals and petals. Finally, mRNA gel blot studies have shown that AP2 gene expression is not restricted to flowers and the inflorescence meristem. AP2 is also expressed at low levels in leaf, stem, and root (K.D. Jofuku, B.G.W. den Boer, M. Van Montagu, and J.K. Okamuro, manuscript submitted; B.G.W. den Boer, unpublished results). Although ap2 mutants have no visible defects in vegetative development under long-day growth conditions (Bowman et al., 1989), short-day studies suggest that AP2 does have a cryptic function in stem development (J.K. Okamuro, unpublished results). How does AP2 carry out its genetic functions at the molecular level? DNA sequence analysis has shown that AP2 encodes a protein that is distinct from all known plant, fungal, and animal regulatory proteins (K.D. Jofuku, B.G.W. den Boer, M. Van Montagu, and J.K. Okamuro, manuscript submitted). However, the ability of AP2 to repress AG gene expression in the flower and the presence of a putative nuclear localization signal sequence in the AP2 protein suggests that it may function as a transcription factor. In addition, the AP2 protein contains a highly acidic serine-rich domain that is structurally similar to regions found in several nucleic acid binding proteins. Another feature of the AP2 protein is a 69-amino acid repeated motif, the AP2-domain, that is evolutionary conserved in AP2-like proteins from plants as divergent as petunia, snapdragon, and rice (B.G.W. den Boer, A. Gerats, M. Van Montagu, E.S. Coen, J.K. Okamuro, and K.D. Jofuku, unpublished results). Sequences within the AP2-doma\n are theoretically capable of forming amphipathic helical structures that may mediate protein-protein interactions. Thus, one possible function of this domain may be to mediate interactions between AP2 and the products of other homeotic genes such as API or LFY. HORMONAL CONTROL OF FLORAL MERISTEM DEVELOPMENT Arabidopsis is a quantitative long-day plant. Changes in photoperiod and temperature can dramatically affect the timing and duration of vegetative and inflorescence growth (Langridge, 1957; Bernier et al., 1993, this issue) but do not normally alter the wild-type pattern of floral organogenesis or inflorescence development (Shannon and Meeks-Wagner, 1991). This developmental homeostasis or resistance to environmental perturbations occurs in both plants and animals and is referred to as canalization (Waddington, 1942). Recent experiments have shown that an important function of the network of meristem identity genes is to maintain the normal pattern of flower development under changing environmental conditions. Each of the five mutants we have just described—Ify, ag, ap1, ap2, and tfH—has a classic long-day "signature" phenotype with respect to the identity or pattern of flower or inflorescence development. Each phenotype, however, can be dramatically altered by changes in photoperiod or temperature. Thus, under short-day conditions, the floral meristems of Ify, ag, ap1, and ap2 mutants display new traits that are normally characteristics of an inflorescence meristem, and the inflorescence meristem of the tfH mutant is more wild type under short-day than long-day conditions. The effects of short-day photoperiod on flower development in these mutants suggest that these genes must normally function together to maintain the normal Arabidopsis Flower Development 1191 pattern of inflorescence and floral meristem development despite varying physiological conditions. How does wild-type Arabidopsis maintain a normal pattern of flower and inflorescence development under changing environmental conditions? A wealth of studies in numerous plant species have implicated the hormone gibberellin as an important regulator of flowering and flower development in higher plants (reviewed by Zeevaart, 1983; Kinet et al., 1985; Chailakhyan and Khrianin, 1987; Bernier et al., 1993, this issue). Moreover, gibberellin levels can be affected by changes in both photoperiod and temperature due to changes in the activities of specific enzymes in the gibberellin biosynthetic pathways (Metzger and Zeevaart, 1980; 1982; Gilmour et al., 1986; Hazebroek and Metzger, 1990; Hazebroek et al., 1993). One clue that gibberellin might be involved in the physiological control of floral induction and inflorescence development in Arabidopsis comes from a recent study by Wilson et al. (1992), who showed that flowering can be completely suppressed in the gibberellin biosynthesis mutant gal under short-day conditions. Moreover, short-day-grown gal mutants can be induced to flower if treated with exogenous gibberellin, thus demonstrating a direct correlation between flowering and gibberellin under short-day conditions in Arabidopsis. If the effects of short days on flower development in ag, apt, ap2, and Ify plants are due to reduced gibberellin levels, then the exogenous application of gibberellin should reverse the effects of short-day conditions on floral meristem identity. Recent experiments have shown that this is indeed the case (J.K. Okamuro, B.G.W. den Boer, C. Lotys, and K.D. Jofuku, manuscript submitted). The short-day-induced reversion of the floral meristem to an inflorescence in both Ify heterozygous and ag homozygous mutants is fully suppressed by the application of exogenous gibberellin (Figures 4 and 5). Moreover, the effects of short-day photoperiod on ap1 and ap2 floral meristem identity are also suppressed by gibberellin (J.K. Okamuro, B.G.W. den Boer, C. Lotys, and K.D. Jofuku, unpublished data). Together, these experiments suggest that the effects of short-day photoperiod on floral meristem identity and development in these mutants are mediated by a decrease in gibberellin levels (J.K. Okamuro, B.G.W. den Boer, C. Lotys, and K.D. Jofuku, manuscript submitted). What insight do these physiological studies provide regarding the regulation of inflorescence and flower development in Arabidopsis? One possible explanation for the effects of short-day photoperiod and gibberellin on inflorescence and floral meristem identity in these mutants is that the expression of one or more of the meristem identity genes API, AP2, AG, or LFY is positively regulated, either directly or indirectly, by gibberellin. Thus, when gibberellin levels are reduced under short-day conditions, the inflorescence character of the floral meristem in Ify, ap1, or ag mutants is conditionally enhanced, perhaps due to increased expression of TFL1, the antagonist of the floral meristem-promoting genes. Conversely, under long-day conditions, when gibberellin levels are high, till mutants are phenotypically enhanced; that is, the inflorescence meristem is transformed into a floral meristem due to the ectopic expression of the floral meristem-promoting genes. CONCLUSION Plant reproduction is governed by both intrinsic and environmental factors that coordinate the activities of the inflorescence and floral meristems. Many of the genes that regulate floral meristem identity and development have been identified and characterized in Arabidopsis. These studies have provided much of our current knowledge on the genetic and molecular regulation of flower development. In addition, recent physiological studies utilizing floral meristem mutants have implicated the plant hormone gibberellin as an important regulator of Arabidopsis flower development. Together, these studies suggest that an integrated molecular and physiological understanding of how flower development is regulated may soon be possible. ACKNOWLEDGMENTS We wish to thank Robert Goldberg and Marc Van Montagu for their support and encouragement during the course of this work. We also thank Gary Drews for critical comments regarding this manuscript. Our work (J.K.O. and K.D.J.) is supported by National Institutes of Health (NIH) Grant No. R29 GM46309, by NIH-Minority Biomedical Research Support Grant No. GM08132, and by funds from the University of California, Santa Cruz. B.G.W.d.B. was supported by a European Economic Community Predoctoral Fellowship. REFERENCES Alvarez, J., Guli, C.L., Yu, X-H., and Smyth, D.R. (1992). terminal flower. A gene affecting inflorescence development in Arabidopsis thaliana. Plant J. 2, 103-116. Ammerer, G. (1990). Identification, purification, and cloning of a polypeptide (PRTF/GRM) that binds to mating-specific promoter elements in yeast. Genes Dev. 4, 299-312. Battey, N.H., and Lyndon, R.F. (1990). Reversion of flowering. Bot. Rev. 56, 162-189. Bernier, G., Havelange, A., Houssa, C, Petitjean, A., and Lejeune, P. (1993). Physiological signals that induce flowering. Plant Cell 5, 1147-1155. Bowman, J.L. (1992). Making cauliflower out of Arabidopsis: The specification of floral meristem identity. Flowering Newslett. 14, 7-19. Bowman, J.L., Smyth, D.R., and Meyerowitz, E.M. (1989). Genes directing flower development in Arabidopsis. Plant Cell 1, 37-52. Bowman, J.L., Smyth, D.R., and Meyerowitz, E.M. (1991). Genetic interactions among floral homeotic genes of Arabidopsis. Development 112, 1-20. 1192 The Plant Cell Chailakhyan, M.K., and Khrianin, V.N. (1987). Sexuality in Plants and Its Hormonal Regulation (New York; Springer-Verlag). Coen, E.S., and Carpenter, R. (1993). The metamorphosis of flowers. Plant Cell 5, 1175-1181. Coen, E.S., and Meyerowltz, E.M. (1991). The war of the whorls: Genetic interactions controlling flower development. Nature 353, 31-37. Coen, E.S., Romero, J.M., Doyle, S., Elliott, R., Murphy, G., and Carpenter, R. (1990). floricaula: A homeotic gene required for flower development in Antirrhinum majus. Cell 63, 1311-1322. Drews, G.N., Bowman, J.L., and Meyerowitz, E.M. (1991). Negative regulation of the Arabidopsis homeotic gene AGAMOUS by the APETALA2 product. Cell 65, 991-1002. Flemington, E., and Speck, S.H. (1990). Evidence for coiled-coil dimer formation by an Epstein-Barr virus transactivator that lacks a hep-tad repeat of leucine residues. Proc. Natl. Acad. Sci. USA 87, 9459-9463. Gilmour, S.J., Zeevaart, J.A.D., Schwenen, L., and Graebe, J.E. (1986). Gibberellin metabolism in cell-free extracts from spinach leaves in relation to photoperiod. Plant Physiol. 82, 190-195. Hazebroek, J.P., and Metzger, J.D. (1990). Thermoinductive regulation of gibberellin metabolism in Thalspi arvense L. I. Metabolism of [2H]kaurenoic acid and [14C]gibberellin A,2-aldehyde. Plant Physiol. 94, 154-165. Hazebroek, J.P., Metzger, J.D., and Mansager, E.R. (1993). Thermoinductive regulation of gibberellin metabolism in Thalspi arvense L. II. Cold induction of enzymes in gibberellin biosynthesis. Plant Physiol. 101, 547-552. Huala, E., and Sussex, I.M. (1992). LEAFY interacts with floral homeotic genes to regulate Arabidopsis floral development. Plant Cell 4, 901-913. Irish, V.F., and Sussex, I.M. (1990). Function of the apetala-1 gene during Arabidopsis floral development. Plant Cell 2, 741-753. Kinet, J., Sachs, R.M., and Bernier, G. (1985). The Physiology of Flowering, Vol. 3 (Boca Raton, FL: CRC Press). Komaki, M.K., Okada, K., Nishino, E., and Shimura, Y. (1988). Isolation and characterization of novel mutants of Arabidopsis thaliana detective in flower development. Development 104, 195-203. Kunst, L., Klenz, J.E., Martinez-Zapater, J., and Haughn, G.W. (1989). AP2 gene determines the identity of perianth organs in flowers of Arabidopsis thaliana. Plant Cell 1, 1195-1208. Langridge, J. (1957). Effect of day-length and gibberellic acid on the flowering of Arabidopsis. Nature 180, 36-37. Ma, H., Yanofsky, M.F., and Meyerowitz, E.M. (1991). AGL1-AGL6, an Arabidopsis gene family with similarity to floral homeotic and transcription factor genes. Genes Dev. 5, 484-495. Mandel, M.A., Gustafson-Brown, C, Savidge, B., and Yanofsky, M.F. (1992). Molecular characterization of the Arabidopsis floral homeotic gene APETALA1. Nature 360, 273-277. Metzger, J.D., and Zeevaart, J.A.D. (1980). Effect of photoperiod on the levels of endogenous gibberellins in spinach as measured by combined gas chromatography-selected ion current monitoring. Plant Physiol. 66, 844-846. Metzger, J.D., and Zeevaart, J.A.D. (1982). Photoperiodic control of gibberellin metabolism in spinach. Plant Physiol. 69, 287-294. Meyerowitz, E.M., Bowman, J.L., Brockman, L.L., Drews, G.N., Jack, T., Sieburth, L., and Weigel, D. (1991). A genetic and molecular model for flower development in Arabidopsis thaliana. Development (suppl.) 1, 157-167. Mizukami, Y., and Ma, H. (1992). Ectopic expression of the floral homeotic gene AGAMOUS in transgenic Arabidopsis plants alters floral organ identity. Cell 71, 119-131. Mueller, C.G.F., and Nordheim, A. (1991). A protein domain conserved between yeast MCM1 and human SRF directs ternary complex formation. EMBO J. 10, 4219-4229. 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 response element. Cell 55, 989-1003. Niisslein-Volhard, C. (1991). Determination of the embryonic axes of Drosophila. Development (suppl.) 1, 1-10. O'Shea, E.K., Klemm, J.D., Kime, P.S., and Alber, T. (1991). X-ray structure of the GCN4 leucine zipper, a two-stranded, parallel coiled coil. Science 254, 539-544. Primig, M., Winkler, H., and Ammerer, G. (1991). The DNA binding and oligomerization domain of MCM1 is sufficient for its interaction with other regulatory proteins. EMBO J. 10, 4209-4218. Schultz, E.A., and Haughn, G.W. (1991). LEAFY, a homeotic gene that regulates inflorescence development in Arabidopsis. Plant Cell 3, 771-781. Schwarz-Sommer, Z., Hue, I., Huijser, P., Flor, P.J., Hansen, R., Tetens, F., Lonnig, W-E., Saedler, H., and Sommer, H. (1992). Characterization of the Antirrhinum floral homeotic MADS-box gene deficiens: Evidence for DNA binding and autoregulation of its persistent expression throughout flower development. EMBO J. 11, 251-263. Shannon, S., and Meeks-Wagner, D.R. (1991). A mutation in the Arabidopsis TFL1 gene affects inflorescence meristem development. Plant Cell 3, 877-892. Shannon, S., and Meeks-Wagner, D.R. (1993). Genetic interactions that regulate inflorescence development in Arabidopsis. Plant Cell 5, 639-655. Shiraishi, H., Okado, K., and Shimura, Y. (1993). Nucleotide sequences recognized by the AGAMOUS MADS domain of Arabidopsis thaliana in vitro. Plant J. 4, 385-398. Sommer, H., Beltran, J-P., Huijser, P., Pape, H., Lonnig, W-E., Saedler, H., and Schwarz-Sommer, Z. (1990). Deficiens, a homeotic gene involved in the control of flower morphogenesis in Antirrhinum majus: The protein shows homology to transcription factors. EMBO J. 9, 605-613. Trobner, W., Ramirez, L., Motte, P., Hue, I., Huijser, P., Lonnig, W., Saedler, H., Sommer, H., and Schwarz-Sommer, Z. (1992). GLOBOSA: A homeotic gene which interacts with DEFICIENS in the control of Antirrhinum floral organogenesis. EMBO J. 11,4693-4704. van der Krol, A.R., and Chua, N.-H. (1993). Flower development in petunia. Plant Cell 5, 1195-1203. Waddington, C.H. (1942). Canalization of development and the inheritance of acquired characters. Nature 150, 563-565. Weigel, D., and Meyerowitz, E.M. (1993). Activation of floral homeotic genes in Arabidopsis. Science 261, 1723-1726. Arabidopsis Flower Development 1193 Weigel, D., Alvarez, J., Smyth, O.R., Yanofsky, M.F., and Meyerowitz, E.M. (1992). LEAFY controls floral meristem identity in Arabidopsis. Cell 69, 843-859. Wilson, R.N., Heckman, J.W., and Somerville, C.R. (1992). Gibberellin is required for flowering in Arabidopsis thaliana under short days. Plant Physiol. 100, 403-408. Yanofsky, M.F., Ma, H., Bowman, J.L., Drews, G.N., Feldmann, K.A., and Meyerowitz, E.M. (1990). The protein encoded by the Arab/'dop-s/s homeotic gene agamous resembles transcription factors. Nature 346, 35-39. Zeevaart, J.A.D. (1983). Gibberellins and flowering. In The Biochemistry and Physiology of Gibberellins, Vol. 2, A. Crazier, ed (New York: Praeger Publishers), pp. 333-374.