- © 2001 American Society of Plant Physiologists
Abstract
Proper development of petals and stamens in Arabidopsis flowers requires the activities of APETALA3AP3) and PISTILLATAPI), whose transcripts can be detected in the petal and stamen primordia. Localized expression of AP3andPI requires the activities of at least three genes: APETALA1AP1), LEAFYLFY), and UNUSUALFLORALORGANSUFO). It has been proposed that UFOprovides spatial cues and thatLFYspecifies competence forAP3andPIexpression in the developing flower. To understand the epistatic relationship amongAP1,LFY, andUFOin regulatingAP3andPIexpression, we generated two versions of AP1 that have strong transcriptional activation potential. Genetic and molecular analyses of transgenic plants expressing these activated AP1 proteins show that the endogenous AP1 protein acts largely as a transcriptional activator in vivo and thatAP1specifies petals by regulating the spatial domains ofAP3and PIexpression throughUFO.
INTRODUCTION
The Arabidopsis flower is one of the best understood plant developmental systems. The flower consists of four organ types arranged in concentric whorls. Identities of the different organ types are established by three classes of floral organ identity genes, the actions of which are best summarized by the ABC model of flower development. This model stipulates that these genes function in a combinatorial fashion to specify organ types, such that whorl 1 sepals are specified by A activity, whorl 2 petals by A and B activities, stamens in the third whorl by B and C activities, and the central carpels by C activity (Bowman et al., 1991; Coen and Meyerowitz, 1991). An example of the A class gene is APETALA1 (AP1), which also is a meristem identity gene. The B class genes consist of APETALA3 (AP3) and PISTILLATA (PI). These three organ identity genes encode MADS domain transcription factors and are expressed in spatially restricted patterns within the developing flower (Jack et al., 1992; Mandel et al., 1992; Goto and Meyerowitz, 1994).
AP3 and PI are essential for the development of petals and stamens, as indicated by the fact that ap3 and pi mutants produce flowers with petals and stamens converted to sepals and carpels, respectively. Upstream regulators of AP3 and PI include the meristem identity genes AP1 and LEAFY (LFY) as well as UNUSUAL FLORAL ORGANS (UFO; Irish and Sussex, 1990; Schultz and Haughn, 1991; Huala and Sussex, 1992; Weigel et al., 1992; Bowman et al., 1993; Levin and Meyerowitz, 1995; Wilkinson and Haughn, 1995). In lfy or ufo mutants, AP3 and PI expression is reduced (Weigel and Meyerowitz, 1993; Levin and Meyerowitz, 1995; Wilkinson and Haughn, 1995; Lee et al., 1997). Moreover, coexpression of LFY and UFO leads to ectopic expression of AP3 (Parcy et al., 1998). LFY encodes a novel transcription factor, and its protein is present throughout the floral primordia during floral stages 1 to 3 (Weigel et al., 1992; Levin and Meyerowitz, 1995; Parcy et al., 1998; Busch et al., 1999). UFO encodes an F box–containing protein and is expressed in a dynamic pattern (Ingram et al., 1995; Lee et al., 1997). During floral stage 3, UFO is expressed transiently in the presumptive whorls 2 and 3, largely overlapping with the domains of AP3 and PI expression. These and other results led to the proposal that cooperation between LFY and UFO is required for the proper expression of AP3 (Lee et al., 1997; Parcy et al., 1998).
In contrast, it is not clear how AP1 might be integrated into the current framework of B class gene regulation. Although ap1 lfy double mutants show a dramatic reduction of AP3 and PI expression not observed in ap1 or lfy single mutants, the initial domain of AP3 and PI expression is not altered in ap1 mutants (Weigel and Meyerowitz, 1993). It has been shown that AP1 binds to the AP3 promoter (Hill et al., 1998) and that mutating these AP1 binding sites abolishes the activity of an AP3 minimal promoter (Tilly et al., 1998). On the basis of these results, it has been suggested that AP3 may be a direct target of AP1 (Hill et al., 1998). Despite these rapid advances, it has been difficult to define the in vivo function of AP1 in the activation of AP3 and PI expression because petal primordia usually fail to develop in strong ap1 mutants (Bowman et al., 1993).
Here, we report the phenotype of transgenic plants expressing translational fusions of AP1 with the strong transcriptional activation domain of VP16. These plants develop petals instead of sepals in whorl 1. Because AP1 has no role in controlling the initiation of these organs, AP1:VP16 transgenic plants allow us to determine the function of AP1 in specifying petal identity. Here, we show that UFO acts downstream from AP1 in activating the expression of AP3 and PI in spatially restricted domains.
RESULTS
An Activated Form of AP1
AP1 plays a key role in specifying the identity of the floral meristem and acts redundantly with LFY to activate the expression of the downstream B class floral homeotic genes AP3 and PI (Irish and Sussex, 1990; Bowman et al., 1993; Weigel and Meyerowitz, 1993; Krizek and Meyerowitz, 1996). AP1 is expressed throughout the developing flower from floral stage 1 to early stage 3 (Mandel et al., 1992). Later in development, its expression is downregulated in the central dome and is restricted to the outer two whorls. In contrast, AP3 and PI start to be expressed during floral stage 3 in a subset of AP1-expressing cells (Jack et al., 1992; Weigel and Meyerowitz, 1993; Goto and Meyerowitz, 1994). Although it has been shown that the expression of AP1 at high levels in the epidermal layer causes cell autonomous activation of AP3 (Sessions et al., 2000), ubiquitous expression of AP1 under the control of the 35S promoter does not alter floral morphology (Mandel and Yanofsky, 1995; Liljegren et al., 1999) and therefore is unlikely to alter AP3 and PI expression. Together, these results suggest that the activity of the AP1 protein must be regulated during early flower development so that the expression domains of AP3 and PI are narrower than those of AP1.
To elucidate the mechanism by which AP1 activates AP3 and PI in spatially restricted patterns, we generated translational fusions of AP1 and the strong transcriptional activation domain of the herpes virus protein VP16 (Triezenberg et al., 1988; Cousens et al., 1989). The VP16 domain should render AP1 a constitutive transcriptional activator independent of any post-translational modification. We reasoned that if AP1 normally activates transcription of its immediate targets, the fusion protein might cause hyperactivation or ectopic expression of downstream genes. Conversely, if AP1 acts to repress transcription, the fusion protein might lead to derepression of genes not normally expressed in the flower.
The AP1 protein can be subdivided into four regions: the M, I, K, and C domains (Mandel et al., 1992; reviewed in Riechmann and Meyerowitz, 1997). The N-terminal M, I, and K domains of AP1 are important for DNA binding and protein–protein interactions, whereas the C-terminal C domain of the protein may determine the transactivation potential of AP1. In view of these observations, the VP16-transacting domain was either fused to the end of AP1, generating AP1:VP16 (Figure 1C), or used to replace the C-terminal domain, generating AP1(ΔC):VP16 (Figure 1E). To establish that fusion of a foreign polypeptide does not compromise the activity of the AP1 or the AP1(ΔC) protein, we used a mutant derivative of VP16 (mVP16) that is inactive in transcription activation to generate AP1:mVP16 and AP1(ΔC):mVP16 (Figures 1D and 1F; Cress and Triezenberg, 1991). These constructs were introduced into plants homozygous for a null allele of ap1 (ap1-15; see Methods). Multiple transformants carrying each gene fusion were analyzed.
To simplify the design of the constructs used to express different fusion proteins, we generated an AP1 minigene by fusing the 5′ end of the AP1 genomic region with the 3′ end of its cDNA (Figures 1A and 1B). This minigene allows the AP1 coding region to be expressed under the control of regulatory elements in the AP1 promoter and in the first five exons and introns. It has been shown that this minigene contains all of the regulatory elements necessary for proper AP1 expression (A. Pinyopich and M.F. Yanofsky, unpublished data).
ap1 Mutants Carrying the Activated Forms of AP1 Show Bract-to-Petal Transformations
The most striking phenotype of ap1 mutants harboring the AP1(ΔC):VP16 or the AP1:VP16 transgene was the formation of petals in the medial whorl 1 position of flowers (Figures 2B to 2D, 2F, and 2G) rather than the bracts normally found in ap1 mutant flowers (Figure 2B). In addition, both transgenes largely rescued all aspects of ap1 mutant phenotypes (cf. Figure 2B with Figures 2C, 2F, and 2G). Axillary flowers were partially suppressed in these plants, and petals developed in the second whorl.
Several control experiments suggest that the bract-to-petal transformation observed in ap1 AP1:VP16 plants is a gain-of-function phenotype associated with the AP1:VP16 transgene. First, the AP1 minigene completely rescued all ap1 mutant phenotypes (Figure 2E and data not shown). Second, the AP1:mVP16 transgene also rescued ap1 mutant phenotypes but did not cause bract-to-sepal transformations. These results strongly suggest that the whorl 1 petals are specified by the gain-of-function activity of the AP1:VP16 transgene. Third, the AP1:VP16 transgene was recessive to the endogenous wild-type AP1 gene. The observation that the presence of AP1 significantly attenuated the gain-of-function phenotype of AP1:VP16 plants suggests that the AP1:VP16 protein regulates the expression of the same set of targets as does the wild-type AP1 protein. Fourth, the AP1(ΔC):mVP16 transgene failed to rescue ap1 mutant phenotypes (Figure 2H), consistent with the idea that the C domain of AP1 functions mainly as a transcriptional activation domain in vivo (Cho et al., 1999). Because plants carrying either transgene exhibit the same phenotypes and genetic interactions with different floral mutants (data not shown), we focus our discussion on plants carrying the AP1:VP16 transgene.
Scheme of Different AP1 Constructs Used in This Study.
(A) AP1 genomic region.
(B) AP1 minigene.
(C) AP1:VP16.
(D) AP1:mVP16.
(E) AP1(ΔC):VP16.
(F) AP1(ΔC):mVP16.
Exons are indicated by rectangles. The four domains (M [coarse stippled], I [checkered], K [fine stippled], and C [striped]) of the AP1 protein are indicated in (A) and (B). The VP16 transactivation domains are shown in black ([C] and [E]), and the mutant VP16 domains are shown in dotted rectangles ([D] and [F]). Arrows indicate the endogenous AP1 promoter.
Scanning electron microscopy was used to examine the whorl 1 organs of ap1 AP1:VP16 plants (Figure 3). The sepals and petals of wild-type flowers have different epidermal morphologies on the adaxial and abaxial surfaces (Figures 3A to 3D). Unlike sepals, the bracts of ap1 flowers have leaf-like epidermal characteristics (Figures 3aa to 3dd; Bowman et al., 1993). Cells on the surface of the lateral whorl 1 organs from ap1 AP1:VP16 flowers resembled cells on the surface of sepals (Figures 3A, 3B, 3aa″, and 3bb″). However, cells with epicuticular ridges, like those present on petals, were observed on the medial whorl 1 organs (Figures 3C, 3D, 3cc″, and 3dd″).
Several additional features of the transformed petals found in ap1-15 AP1:VP16 plants are noteworthy. Under our growth conditions, complete bract-to-petal transformations were observed from flowers arising at node 4 and after. Flowers arising later (after node 4) showed minimal variation with respect to whorl 1 petals. Therefore, the images shown in Figure 2 are representative of the floral morphologies of ap1-15 AP1:VP16 plants. However, in early-arising flowers, the lateral margins of medial first whorl organs had petal characteristics, whereas the central portion of the organs had characteristics of sepals or bracts. In these mosaic organs, a gradient of cell types could be observed (data not shown). The lateral whorl 1 organs of ap1 AP1:VP16 plants tended to curve toward the center of the floral buds and developed less stellate trichomes than did the bracts of ap1 mutants, consistent with the idea that they were more sepal like than bract like. In these early-arising flowers, transformation of the medial whorl 1 organs to petals was more dramatic for those in the adaxial position (closer to the stem) than for those in the abaxial position. Furthermore, the adaxial (upper) surfaces of these organs showed a more complete transformation than did the abaxial (lower) surfaces. Finally, a more complete transformation of the medial whorl 1 organs to petals was observed in later-arising flowers.
ap1 Mutants Expressing AP1:VP16 Show Bract-to-Petal Transformations.
(A) A wild-type flower (Columbia ecotype) with a first whorl sepal (se) and a second whorl petal (p) indicated. Simple trichomes can be found on the abaxial sepal surface.
(B) An ap1-15 flower from the fourth node. Note the absence of petals and the presence of secondary flowers. Bracts bearing stellate trichomes, instead of sepals, are present in the first whorl. Bracts tend to curve outward, away from the center of the flower.
(C) An ap1 AP1:VP16 flower from the fourth node. The medial whorl 1 organs are completely converted to petals (arrowheads). The lateral whorl 1 organs have features of bracts and sepals. For example, these organs curve toward the center of the flower and have both simple and stellate trichomes (not shown). The ap1-15 mutant phenotype is largely rescued by the AP1:VP16 transgene. Secondary flowers are rare, and whorl 2 petals are restored.
(D) Inflorescence apex of an ap1-15 AP1:VP16 plant. The medial positions of the whorl 1 petals are indicated (arrowheads). Arrows indicate secondary flowers.
(E) An ap1-15 AP1:mVP16 flower is essentially wild type except for the occasional presence of stellate trichomes in the outer whorl 1 organs.
(F) and (G) An ap1-15 AP1(ΔC):VP16 flower from the fourth node. The medial whorl 1 organs are either completely (arrowheads) or partially (arrows) converted to petals.
(H) An ap1-15 AP1(ΔC):mVP16 flower has the same phenotype as an ap1-15 flower.
All images are of the same magnification (×14) except for (B), which is shown at ×0.8 that of the others. Medial first whorl petals are indicated with arrowheads in (C), (D), (F), and (G).
AP3 and PI Are Required for Bract-to-Petal Transformations
It has been difficult to study the role of AP1 in specifying petal identity because petal primordia fail to initiate in ap1 mutants. The observation that petals form in the first whorl of ap1-15 AP1:VP16 plants allowed us to study the genetic pathway downstream from AP1 required to specify petals, because AP1 is not required for the initial outgrowth of whorl 1 organs. Therefore, we crossed ap1 plants expressing the activated forms of AP1 into ap3 and pi mutants.
Wild-type activities of the two B class homeotic genes AP3 and PI are required for normal petal development (Figures 4A and 4D; Bowman et al., 1989, 1991; Hill and Lord, 1989). Conversely, constitutive expression of AP3 and PI is sufficient to specify whorl 1 petals in the wild type or to partly restore whorl 2 petals in ap1 mutants (Krizek and Meyerowitz, 1996; Samach et al., 1999). ap1 ap3 and ap1 pi double mutants carrying the AP1:VP16 transgene developed sepals in both whorls 1 and 2 (Figures 4C and 4F). The suppression of whorl 1 petals by both ap3 and pi was complete, and no petaloid organs were observed. These results demonstrate that AP3 and PI are active in the ectopic petals found in ap1 AP1:VP16 flowers and suggest that the AP1:VP16 transgene may be sufficient to cause their ectopic expression (see below).
Precocious and Ectopic Expression of AP3 and PI in ap1 AP1:VP16 Flowers
Our genetic data demonstrate that AP3 and PI activities are required for whorl 1 petals to develop in ap1 AP1:VP16 plants. To determine whether AP3 and PI RNAs accumulate ectopically, we examined the patterns of AP3 and PI expression by in situ hybridization (Figure 5). In wild-type flowers, AP3 is expressed from stage 3 in the presumptive second and third whorls (Figure 5A; Jack et al., 1992). After floral stages 5 and 6, AP3 is expressed throughout the developing petals and stamens and at the adaxial base of sepals (Jack et al., 1992; Weigel and Meyerowitz, 1993). PI initially is expressed from stage 3 in the second, third, and fourth whorls (Figure 5E; Goto and Meyerowitz, 1994). By floral stage 5, PI expression is restricted to petals and stamens. These patterns of AP3 and PI expression appeared normal in ap1 mutants, although the expression domains appeared smaller, probably because the whorl 2 organs failed to develop (Figures 5B and 5F). Using a different allele of ap1, we observed similar expression patterns of B class genes (Weigel and Meyerowitz, 1993).
In ap1 AP1:VP16 plants, AP3 was activated earlier and expressed initially during floral stage 2 (Figure 5C). The domain of AP3 expression in these transgenic flower meristems was broader than in the wild type and included the emerging whorl 1 organs. AP3 continued to be expressed at high levels in the first three whorls later in flower development. By examining cross-sections, it could be seen that AP3 was expressed in the medial whorl 1 organs but not in the lateral organs (Figure 5D). Usually, higher levels of AP3 expression were detected on the adaxial than on the abaxial sides of these organs.
Scanning Electron Micrographs of Whorl 1 Organs in Wild-Type, ap1-15 Mutant, and ap1 AP1:VP16 Plants.
(A) to(D)Wild type. Adaxial views are shown in (A) and (C), and abaxial views are shown in(B) and (D). Stomata can be found on both the adaxial (A) and abaxial (B) surfaces of lateral sepals. Cells with unique epicuticular ridges are present on both sides of petals. Unlike sepals, petals lack stomata.
(aa) to (dd) ap1-15. Adaxial views are shown in(aa) and(cc), and abaxial views are shown in(bb) and (dd). Cells on the adaxial ([aa] and [cc]) and abaxial ([bb] and [dd]) surfaces of whorl one organs (bracts) are different from cells on sepals ([A] and [B]). Elongated cells, similar to those found on leaves, are found on the abaxial sides of bracts ([bb] and [dd]).
(aa″) to(dd″) ap1-15 AP1:VP16. Adaxial views are shown in (aa″) and (cc″), and abaxial views are shown in (bb″) and (dd″). The adaxial (aa″) and abaxial (bb″) surfaces of the lateral whorl one organs resemble the corresponding surfaces of sepals (compare [aa″] and [bb″]with [A] and [B]). Similarly, the adaxial (cc″) and abaxial (dd″) sides of a medial first whorl organ resemble the corresponding surfaces of a petal, except for the occasional presence of stomata (compare [cc″] and [dd″] with [C] and [D]). All images are of the same magnification. s, stomata.
PI was expressed in a pattern analogous to that of AP3 in ap1 AP1:VP16 plants (Figures 5G and 5H). PI was first detected during floral stage 2 in a broad domain of the floral meristem that gives rise to all four whorls. By stage 5, PI expression abated from the center of the developing flower, as in the wild type, but persisted in the outer three whorls at least until floral stage 9 (data not shown). Ectopic PI expression could be detected only in the medial, but not in the lateral, whorl 1 organs (data not shown). Like AP3, higher levels of PI transcripts usually were found on the adaxial than on the abaxial surfaces of the medial organs (Figure 5H). These results indicate that ectopic expression of AP3 and PI in ap1 mutants carrying the AP1:VP16 transgene correlates with the appearance and the variable morphology of medial whorl 1 petals.
UFO Also Is Required for Whorl 1 Petal Development
Early-arising flowers of ufo mutants lack petals and resemble ap3 or pi mutants (Figure 6A; Levin and Meyerowitz, 1995; Wilkinson and Haughn, 1995). This phenotype attenuates as the plant develops, such that later-arising flowers often develop some whorl 2 petals (Figure 6D; Levin and Meyerowitz, 1995; Wilkinson and Haughn, 1995). Two lines of evidence suggest that AP3 and PI act downstream from UFO to specify petal identity. First, UFO is necessary for AP3 and PI expression (Levin and Meyerowitz, 1995; Wilkinson and Haughn, 1995). Second, ectopic expression of AP3 and PI restores petals and stamens in ufo mutants, whereas ectopic expression of UFO in ap3 and pi mutants does not (Krizek and Meyerowitz, 1996; Lee et al., 1997; Samach et al., 1999).
To determine whether UFO activity is required for AP1 to activate AP3 and PI expression, and hence petal development, we constructed ap1 ufo double mutants carrying the AP1:VP16 transgene. If UFO acts downstream from AP1, we would anticipate that early-arising flowers in these double mutants carrying the transgene would lack petals. Alternatively, if UFO acts upstream of or in parallel with AP1, petals should be restored in these plants. We found that early-arising flowers of ap1 ufo AP1:VP16 plants had no petals, indicating that UFO activity is required for AP1 to induce petal formation (Figure 6C). As anticipated, later-arising flowers produced petals in whorls 1 and 2, consistent with the fact that even in ufo single mutants, later-arising flowers have petals (Figures 6D to 6F).
To further analyze the epistatic relationship between AP1 and UFO, we examined the phenotype of ufo mutants bearing the 35S::AP1 transgene (Figures 6G to 6I). Constitutive expression of AP1 under the control of the 35S promoter results in early flowering, conversion of axillary shoots to flowers, and the development of terminal flowers at the apices of inflorescence shoots (Figure 6G; Mandel and Yanofsky, 1995; Liljegren et al., 1999). If UFO acts downstream from AP1, we would expect flowers from ufo mutants carrying the 35S::AP1 transgene to lack petals. However, if AP1 acts downstream from or in parallel with UFO, we would expect the constitutive expression of AP1 to restore petals. ufo 35S::AP1 plants showed all of the gain-of-function phenotypes conferred by the 35S::AP1 transgene (Figures 6H and 6I). In addition, early-arising flowers of ufo 35S::AP1 plants lacked whorl 2 petals (Figures 6H and 6I), arguing against AP1 acting downstream from UFO. Very rarely, one or two petals could be found in later-arising flowers of ufo 35S::AP1 plants (data not shown), which is consistent with the fact that the phenotypes of ufo mutants are attenuated in later-arising flowers.
Expression Patterns of AP1 and LFY Remain Unaltered in ap1AP1:VP16 Flowers
Because AP1 and LFY positively regulate each other's expression (Bowman et al., 1993; Liljegren et al., 1999; Ratcliffe et al., 1999), we wanted to determine if the AP1:VP16 transgene alters the expression patterns of endogenous AP1 and LFY. We found the patterns of AP1 and LFY expression to be normal in ap1 AP1:VP16 plants (Figure 7), further supporting the notion that the phenotypes observed in ap1 AP1:VP16 plants were due to gain-of-function activity of AP1:VP16 rather than misexpression of AP1 or LFY.
Development of Whorl 1 Petals in ap1-15 AP1:VP16 Plants Requires AP3 and PI Activities.
(A) An ap3-3 flower. Petals and stamens are reduced. Note the similarities between an ap3-3 flower and a pi-1 flower (D).
(B) An ap3-3 ap1-15 flower. Petals and stamens are absent, but numerous secondary and tertiary flowers can be seen. Note the similarities between an ap3-3 ap1-15 flower and an pi-1 ap1-15 flower (E).
(C) Petals fail to develop in an ap3-3 ap1-15 AP1:VP16 flower. The whorl 1 and whorl 2 organs resemble each other and have characteristics of both sepals and leaves. Note that flowers of the following genotypes resembled each other: ap3-3 ap1-15 AP1:VP16 (C) and pi-1 ap1-15 AP1:VP16 (F).
(D) A pi-1 flower.
(E) A pi-1 ap1-15 flower.
(F) A pi-1 ap1-15 AP1:VP16 flower lacks first and second whorl petals. Some secondary flowers can be seen.
Magnification of (E) is ×0.8 that of the other images (×14). w1, whorl 1 organ.
DISCUSSION
AP1 Specifies Petals by ActivatingAP3 andPI throughUFO
AP1 and LFY have redundant roles in the activation of B class gene expression (Weigel and Meyerowitz, 1993). We have found that ap1 mutants carrying an activated form of AP1, AP1:VP16, show bract-to-petal transformations of medial whorl 1 organs. Furthermore, this phenotype correlates with precocious and ectopic expression of AP3 and PI and is suppressed by mutations in either AP3 or PI. Last, AP1 expression is normal in ap3 mutants (C. Gustafson-Brown and M.F. Yanofsky, unpublished data). Together, our results substantiate the proposal that AP3 and PI act downstream from AP1 (Weigel and Meyerowitz, 1993; Krizek and Meyerowitz, 1996).
It has been shown convincingly that UFO acts upstream of AP3 and PI (Krizek and Meyerowitz, 1996; Lee et al., 1997). To examine the epistatic relationship of AP1 and UFO in the activation of B class genes, we generated ap1 ufo AP1:VP16 plants. If AP1 acts downstream from or in parallel with UFO, it would be expected that expression of the constitutively active transcription factor AP1:VP16 would lead to ectopic expression of AP3 and PI in ap1 ufo mutants. However, bract-to-petal transformations were suppressed in early-arising flowers of these plants, strongly suggesting that AP1 acts upstream of UFO for B class gene activation. Four additional lines of evidence support this proposal. First, constitutive expression of AP1 fails to restore petals in ufo mutants. Second, AP1 expression in ufo mutants resembles that in the wild type (Wilkinson and Haughn, 1995). Third, UFO is not expressed beyond floral stage 3 in ap1 mutants (Lee et al., 1997). Fourth, constitutive expression of UFO restores petals in ap1 mutants (T. Wada, I. Lee, and D. Weigel, personal communication).
When AP3 and PI begin to be expressed, AP1 RNA and LFY protein can be detected throughout the developing flower (Mandel et al., 1992; Levin and Meyerowitz, 1995; Parcy et al., 1998). During this time, UFO is expressed transiently in whorls 2 and 3 (Ingram et al., 1995; Lee et al., 1997). It has been proposed that the combination of LFY and regional factors such as UFO leads to the localized expression of AP3 (Parcy et al., 1998). However, this proposal does not explain how UFO expression is established. Our results show that a chimeric AP1 protein with activated transcriptional activity causes an expansion of the expression domains of AP3 and PI. Perhaps UFO activity is expanded in ap1 mutants carrying the AP1:VP16 transgene.
A model for B class gene activation is shown in Figure 8. We propose that in early-arising flowers there is an obligatory requirement for AP1 to activate B class gene expression through UFO. Cooperation of AP1 with another factor (“X”), or modulation of the transcriptional activity of AP1 (e.g., phosphorylation) in a subset of cells in the floral meristem, may be sufficient to establish the domain of UFO activity. In later-arising flowers, the absolute requirement of UFO for B class gene expression is bypassed. Our results do not rule out the possibility that AP1 directly activates AP3 or PI in later-arising flowers.
Precocious and Ectopic Expression of AP3 and PI in ap1-15 AP1:VP16 Plants.
(A) Wild type (Columbia ecotype). AP3 RNA is not detected in the wild type during floral stages 1 or 2 (arrows). AP3 is expressed initially during stage 3 in the presumptive petal and stamen primordia (not seen in this example). The sepal and gynoecium are largely devoid of AP3 expression throughout development. Some expression can be detected at the adaxial base of sepals (arrowheads).
(B) In ap1-15 flowers, AP3 is expressed in a pattern similar to that in the wild type (A). The domain of expression is narrower.
(C) and (D) AP3 expression in ap1-15 AP1:VP16 plants. AP3 first is detected during floral stage 2 (arrowhead 2 in [C]). By stage 3, AP3 is detected in the presumptive outer three whorls of the developing flower (arrowhead 3 in [C]). This expression pattern persists until late in development ([C] and [D]). (D) shows a cross-section of an inflorescence apex showing ectopic expression of AP3 in the medial first whorl organs. Expression of AP3 is more pronounced in the adaxial organs (arrowheads) than in the abaxial organs (arrows). AP3 expression often is stronger on the lateral margins and on the adaxial sides of these organs. These observations correlate with the floral phenotypes observed in ap1-15 AP1:VP16 plants.
(E) Wild type (Columbia ecotype). PI starts to be expressed during floral stage 3, centripetal to the early sepal primordia. The expression in the fourth whorl decreases during stages 3 and 4. From stage 5 onward, PI can be detected only in the developing petals and stamens.
(F) In ap1-15, the expression pattern of PI is similar to that in the wild type (E).
(G) and (H) PI expression in ap1-15 AP1:VP16 plants. PI is first detected during floral stage 2, including the whorl 1 organ primordia (arrowhead 2 in [G]). From stage 3 onward, PI is clearly detected in whorl 1 ([G] and [H] and data not shown). Like AP3, PI transcripts can be detected only in the medial whorl 1 organs but not the lateral organs (data not shown).
Magnification of (E) is ×2 that of the other images (×58). Longitudinal sections and cross-sections of flowering apices were hybridized to antisense AP3 ([A] to [D]) or PI ([E] to [H]) probes. Numbers indicate floral stages.
Development of Whorl 1 Petals in ap1-15 AP1:VP16 Plants Requires UFO Activities.
(A) A ufo-2 flower from the first node. Note the similarities with ap3-3 and pi-1 mutant flowers (see Figures 4A and 4D).
(B) A ufo-2 ap1-15 flower from the first node. Note the similarities with ap3-3 ap1-15 and pi-1 ap1-15 double mutant flowers (see Figures 4B and 4E).
(C) A ufo-2 ap1-15 AP1:VP16 flower from the first node. Sepals develop in whorl 1 and whorl 2, as in ap3-3 ap1-15 AP1:VP16 and pi-1 ap1-15 AP1:VP16 flowers (see Figures 4C and 4F).
(D) A late-arising ufo-2 flower (after node 10). The ufo mutant phenotype attenuates; therefore, petals and stamens can be seen in whorl 2 and whorl 3, respectively.
(E) A late-arising ufo-2 ap1-15 flower (after node 10). Petals and secondary flowers are absent. A stamen can be seen.
(F) A late-arising ufo-2 ap1-15 AP1:VP16 flower (after node 10). Arrowheads indicate medial whorl 1 petals.
(G) to (I) Constitutive expression of AP1 fails to rescue petals in ufo-2 mutants.
(G) Inflorescence apex of a 35S::AP1 transgenic plant. Axillary meristems subtended by cauline leaves develop into flowers.
(H) and (I) A ufo-2 35S::AP1 plant lacks petals and stamens. Instead, sepals are found in the outer two whorls (arrows). Magnification of (G) is ×0.8 that of the other images (×14). p, petal; st, stamen; w1, whorl 1 organ; w2, whorl 2 organ.
It is interesting that AP1 binds to sequence elements in the promoter region of AP3 (Hill et al., 1998; Tilly et al., 1998). Mutations in these elements that abolish AP1 binding also eliminate AP3-specific expression (Hill et al., 1998; Tilly et al., 1998), raising the possibility that AP3 may be regulated directly by AP1 at least under some circumstances (Hill et al., 1998). In contrast, sequence elements (known as CArG boxes) to which AP1 protein can bind in vitro have not been identified in the minimal PI promoter (Honma and Goto, 2000). This observation suggests that AP1 may regulate AP3 and PI expression by using different mechanisms. Consistent with this idea, it has been shown that the autoregulation of AP3 expression is direct, whereas the autoregulation of PI expression is indirect (Honma and Goto, 2000).
Genetic Behavior of AP1 : VP16
AP1:VP16 can be regarded as an allele of ap1 with both gain-of-function and loss-of-function components. Because AP1:VP16 only partially rescues the ap1 mutant phenotype, it must be a hypomorphic allele. The conversion of medial bracts to petals in ap1 AP1:VP16 plants suggests that the transgene has a gain-of-function effect not observed with other ap1 alleles (see below). This idea is consistent with the observation that the bract-to-petal conversion depends on the transcriptional activation potential of VP16. The copy number of the AP1:VP16 transgene had been increased by crossing different lines together; however, no dominant behavior could be observed for the transgene in the wild type (data not shown). Because most gain-of-function alleles are dominant, it is surprising that AP1:VP16 is recessive to the wild-type allele.
We suggest two possible explanations for the recessive nature of AP1:VP16. AP1 might function as a component of a higher order multimer with other MADS box proteins such as CAULIFLOWER (S. Pelaz, C. Gustafson-Brown, and M.F. Yanofsky, unpublished data) and SEPALLATA3 (Honma and Goto, 2001). Perhaps two or more subunits of AP1:VP16 must be present in this large complex for efficient misexpression of downstream targets. This possibility can be confirmed by biochemical analysis of the complex. Alternatively, it is possible that the VP16 domain used in this study is not a very strong transcriptional activation domain in planta. Consequently, a strong and dominant phenotype is observed only when the VP16 domain is fused to a transcriptional repressor such as LFY (Parcy et al., 1998). In contrast, when the VP16 domain is fused to a transcriptional activator such as AP1, a weak and recessive phenotype is observed.
Expression of AP1 and LFY in ap1-15 AP1:VP16 Plants Is the Same as in Wild-Type Plants.
(A) and (D) Wild type (Columbia ecotype).
(B) and (E) ap1-15.
(C) and (F) ap1-15 AP1:VP16.
Longitudinal sections of flowering apices that were hybridized to antisense AP1 ([A] to [C]) or LFY probe ([D] to [F]) are shown. The expression of AP1 or LFY in different genetic backgrounds is very similar. Numbers indicate floral stages (Smyth et al., 1990). Magnification of (D) to (F) is ×2 that of (A) to (C) (×58).
A Model for the Activation of B Class Genes.
Expression of AP3 and PI in spatially correct domains requires input from AP1, LFY, and UFO. In early-arising flowers, AP1 acts through UFO to activate AP3 and PI. AP1 may cooperate with a hypothetical factor (“X”) to localize UFO activity. In later-arising flowers, the genetic circuit changes so that AP1 can activate AP3 and PI independent of UFO.
The medial whorl 1 organs of ap1-3 and ap1-5, two weak alleles of AP1, often are mosaics of central leaf-like tissue and marginal petaloid tissue (Bowman et al., 1993). To a first approximation, this observation may suggest that AP1 represses B class genes in whorl 1. However, we find this proposal unlikely for two reasons. First, the whorl 1 organs of other intermediate or strong alleles of AP1 do not have petaloid characteristics (Bowman et al., 1993). If AP1 acts as a repressor for B class genes, it would be expected that a more complete conversion of bracts to petals would be observed in stronger alleles of AP1. Second, lfy ap1 double mutants show a dramatic reduction of AP3 and PI expression not observed in lfy or ap1 single mutants (Weigel and Meyerowitz, 1993). If AP1 negatively regulates B class genes, it would be expected that mutations in AP1 would rescue AP3 and PI expression in lfy mutants.
Both ap1-3 and ap1-5 carry an AG-to-AA mutation at the invariant nucleotide of the splice acceptor site of the fifth intron (Mandel et al., 1992; A. Mandel and M.F. Yanofsky, unpublished data). Conceptual translation of the mutant transcript produces a polypeptide containing the M and I domains and most of the K domain. However, mutations at the splice junctions often are associated with the production of aberrantly spliced transcripts by using cryptic splice sites (Lorkovic et al., 2000). In the absence of a detailed molecular analysis of the transcripts produced by the ap1-3 and ap1-5 alleles, it is difficult to ascertain the nature of the translated protein products. As we have shown, an AP1 protein lacking any transcriptional activation domain (AP1[ΔC]:mVP16) is completely inactive and fails to rescue the ap1 mutants. Thus, the molecular basis of the ap1-3 and ap1-5 mutant phenotypes requires further analyses.
Note that AP1:VP16 causes only medial bract-to-petal transformations and that ectopic expression of AP3 and PI is not observed in the lateral whorl 1 organs or the whorl 4 organs. These results strongly suggest that other factors important for petal development have not been identified. In this context, it is interesting that the SEPALLATA1, -2, and -3 MADS box genes have been shown to play a crucial role in petal development (Pelaz et al., 2000). In view of these results, it seems likely that the genetic circuits regulating organ identity are more complex than initially suggested by the ABC model.
METHODS
Isolation and Molecular Analysis of ap1-15
A strong allele of ap1 (ap1-15) in the Columbia ecotype of Arabidopsis thaliana was isolated in the laboratory of Z. Renee Sung (University of California, Berkeley). Molecular analysis of the genomic region of this allele revealed a single base deletion at thymidine 123 (GenBank accession number Z16421; Mandel et al., 1992; A. Mandel and M.F. Yanofsky, unpublished data). Conceptual translation of the ap1-15 transcript produced a polypeptide containing a portion of the MADS domain (40 amino acids) followed by 79 amino acids encoded by the frameshifted message. ap1-15 mutants exhibit all phenotypes described for strong ap1 alleles (Bowman et al., 1993). Therefore, ap1-15 is likely to be a null allele. To genotype this allele, we amplified genomic DNA by polymerase chain reaction (Saiki et al., 1988) using two primers, 5′-TGAGCTCTTCTTTATATCTCTC-3′ (OAM7.1) and 5′-TCGAAGAGTTTTCCCTTAT-3′ (OAM7.2; corresponding to nucleotides 149 to 167). The restriction enzyme MwoI recognizes the target sequence GC(N)7GC and will cleave only the wild-type polymerase chain reaction product, yielding 39- and 225-bp fragments.
Plasmid Constructs
An AP1 minigene was generated by fusing the 5′ end of an AP1 genomic fragment, which contains 1.7 kb of the AP1 promoter, to the 3′ end of an AP1 cDNA (amino acids 167 to 256; Mandel et al., 1992). This minigene was cloned into the transformation vector pCGN1547 (McBride and Summerfelt, 1990), producing pMN112.
AP1:VP16 (pMN151) was generated by fusing the transcriptional activation domain of VP16 (amino acids 413 to 490) from herpes simplex virus to the last amino acid of the AP1 minigene (Cousens et al., 1989). AP1:mVP16 (pMN153) was generated in the same way except that a mutant truncated version of the VP16 transcription activation domain (amino acids 413 to 456) was used (Cress and Triezenberg, 1991).
To assess the function of the C domain of AP1 (amino acids 175 to 256), it was replaced by the VP16 domain. AP1(ΔC):VP16 (pMN150) was generated by fusing the transcriptional activation domain of VP16 (amino acids 413 to 490) to amino acid 174 of the AP1 minigene. AP1(ΔC):mVP16 (pMN152) was generated in the same way except that the mutant truncated VP16 domain was used. Detailed information regarding all plasmid constructs is available upon request.
Plasmid constructs were introduced into ap1-15 mutants by the floral dip method (Clough and Bent, 1998). The phenotypes of >40 primary (T1) transformants were examined for each construct. In all cases, consistent phenotypes were observed among transformants carrying the same construct. One representative T1 transformant for each construct was chosen for further analysis.
Note that plants containing any of these AP1 constructs were prone to cosuppression in subsequent generations or after genetic crosses. Furthermore, the kanamycin resistance phenotype conferred by the transgene did not necessarily cosegregate with an active transgene. Two strategies were used to identify double mutants bearing a VP16 transgene (see below for details). Also, ap1-15 AP1(ΔC):VP16 and ap1-15 AP1:VP16 plants grown under continuous light showed stronger phenotypes than corresponding plants grown under long-day conditions.
Plant Materials
The following additional mutants were used in this study: ap1-1 (Irish and Sussex, 1990; Bowman et al., 1993), ap3-3 (Jack et al., 1992), cal-1 (Bowman et al., 1993), pi-1 (Bowman et al., 1989; Hill and Lord, 1989), and ufo-2 (Levin and Meyerowitz, 1995). These are all strong alleles, and they are very likely to represent complete loss-of-function mutations of the corresponding genes. Transgenic plants carrying the 35S::AP1 construct in the Landsberg erecta background have been described (Liljegren et al., 1999).
ap1-15 ap3-3, ap1-15 pi-1, and ap1-15 ufo-2
These double mutants were generated by manual cross-pollination using homozygous mutant strains. Double mutants were identified in the F2 generation as plants segregating novel phenotypes. Using different alleles, phenotypes of the relevant double mutants were reported previously (Irish and Sussex, 1990; Bowman et al., 1993; Levin and Meyerowitz, 1995; Wilkinson and Haughn, 1995), which greatly facilitated identification of the double mutants used in this study.
ap1-15 ap3-3, ap1-15 pi-1, and ap1-15 ufo-2 Double Mutants Carrying a VP16 Transgene
To construct these double mutants carrying either the AP1(ΔC):VP16 or the AP1:VP16 transgene, we crossed homozygous ap1-15 mutants carrying the respective transgene to ap3-3, pi-1, or ufo-2. The relevant double mutants carrying the transgene were identified among the F2 progeny. The F2 seed were germinated on plant tissue culture medium containing kanamycin to select for the transgene. Because the AP1(ΔC):VP16 and AP1:VP16 transgenes partly rescue the ap1-15 mutant phenotype, the relevant double mutants could be identified easily.
ap1-1 cal-1 Carrying a VP16 Transgene
ap1-15 mutants carrying either the AP1(ΔC):VP16 or the AP1:VP16 transgene were crossed to ap1-1 cal-1 double mutants. Kanamycin-resistant F2 progeny segregating for novel phenotypes were analyzed for the absence of the ap1-15 allele by the genotyping strategy described above. The presence of the ap1-1 and cal-1 alleles was confirmed by backcrossing plants segregating for the novel phenotypes to ap1-1 cal-1 mutants.
ufo-2 Carrying a 35S:: AP1 Transgene
Wild-type plants carrying the 35S::AP1 transgene in the Landsberg erecta background were crossed to ufo-2 mutants. Kanamycin-resistant F2 progeny segregating for novel phenotypes should correspond to 35S::AP1 ufo-2 plants.
Scanning Electron Microscopy
Preparation and examination of plant tissues by scanning electron microscopy were performed as described (Gu et al., 1998).
In Situ Hybridization
Fixation of tissue and hybridization conditions were as described previously (Drews et al., 1991) with modifications (Ferrándiz et al., 1999). Antisense AP1, AP3, and PI RNA probes were labeled with digoxigenin as described (Jack et al., 1992; Mandel et al., 1992; Goto and Meyerowitz, 1994; Ferrándiz et al., 1999).
Acknowledgments
We thank S. Liljegren, S. Pelaz, A. Sessions, and D. Weigel for comments on the manuscript; S. Liljegren for ufo-2 35S::AP1 seed; Z.R. Sung for the ap1-15 allele; A. Mandel for molecular analyses of ap1-5 and ap1-15; S. Pelaz, C. Gustafson-Brown, A. Mandel, A. Pinyopich, T. Honma and K. Goto, and I. Lee and D. Weigel for unpublished results; and P. O'Hare, S. Triezenberg, K. Goto, T. Jack, I. Lee, D. Weigel, and the Arabidopsis Stock Center for plasmids and seed. M.N. received a long-term postdoctorate fellowship from the Human Frontier Science Program Organization (No. LT-367/97). This work was supported by Grant No. IBN9728402 from the National Science Foundation and Grant No. GM55328 from the National Institutes of Health to M.F.Y.
Footnotes
-
↵1 Current address: Exelixis Plant Science, Inc., 16160 S.W. Upper Boones Ferry Rd., Portland, OR 97224-7744.
- Received November 8, 2000.
- Accepted January 26, 2001.
- Published April 1, 2001.
REFERENCES
