This Article Full Text (PDF) All Versions of this Article: 16/suppl_1/S1    most recent tpc.017038v1 Alert me when this article is cited Alert me if a correction is posted Services Email this article to a friend Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Jack, T. Search for Related Content PubMed PubMed Citation Articles by Jack, T. Agricola Articles by Jack, T.

Molecular and Genetic Mechanisms of Floral Control

Department of Biological Sciences, Dartmouth College, Hanover, New Hampshire 03755

¹ To whom correspondence should be addressed. E-mail thomas.jack{at}dartmouth.edu; fax 603-646-1347.

	INTRODUCTION

TOP INTRODUCTION UNIFYING PRINCIPLES OF FLOWER... MULTIPLE FLORAL INDUCTIVE... INTEGRATION OF FLOWERING SIGNALS... INTERACTION BETWEEN AP1 AND... TFL1 FLORAL ORGAN IDENTITY GENES A NEW ADDITION TO... D CLASS AND OVULE... THE MAJORITY OF ABCE... COMBINATIONS OF ABCE GENES... HIGHER ORDER MADS COMPLEXES? ACTIVATION OF FLORAL ORGAN... INITIAL ACTIVATION OF THE... INITIAL ACTIVATION OF THE... INTERACTIONS AMONG ABC GENES NEW INSIGHTS INTO AP2... HUA1, HUA2, HEN2, AND... HEN1 FUNCTIONS TO PROCESS... miRNA172 ALSO FUNCTIONS TO... FUTURE PROSPECTS REFERENCES

 
In the last 15 years, knowledge of the molecular and genetic mechanisms that underlie floral induction, floral patterning, and floral organ identity has exploded. Elucidation of basic mechanisms has derived primarily from work in three dicot species: Antirrhinum majus, Arabidopsis thaliana, and Petunia hybrida. Although Antirrhinum and petunia have contributed fundamental breakthroughs to our understanding of flower development, it is from Arabidopsis that the most detailed and comprehensive picture of the molecular mechanisms underlying flower development has been obtained. In this review, I will outline the present state of knowledge, focusing on molecular and genetic mechanisms revealed in work on Arabidopsis, specifically in three areas: the integration of floral induction signals by a small group of floral integrators, the activation of the floral organ identity genes by the floral meristem identity genes, and interactions among the floral organ identity genes, particularly the A and C class genes.

By choosing to focus on progress in Arabidopsis, I do not mean to suggest that work in other species is unimportant or uninformative. To the contrary, without studies in Antirrhinum and petunia, our knowledge and the broad impact of what has been learned would clearly be less. One of the satisfying things about the field of flower development is the applicability of the floral patterning mechanisms to a wide range of plant species; such a conclusion only comes from careful analysis in multiple distantly related species. The general pattern in the field has been that molecular and genetic mechanisms, based on work in model species, serve as the basis for work in other species, many of which are of economic importance. Ultimately, the goal is to use information discovered in the model plants to engineer economically important plants for human and ecological benefit.

	UNIFYING PRINCIPLES OF FLOWER DEVELOPMENT

TOP INTRODUCTION UNIFYING PRINCIPLES OF FLOWER... MULTIPLE FLORAL INDUCTIVE... INTEGRATION OF FLOWERING SIGNALS... INTERACTION BETWEEN AP1 AND... TFL1 FLORAL ORGAN IDENTITY GENES A NEW ADDITION TO... D CLASS AND OVULE... THE MAJORITY OF ABCE... COMBINATIONS OF ABCE GENES... HIGHER ORDER MADS COMPLEXES? ACTIVATION OF FLORAL ORGAN... INITIAL ACTIVATION OF THE... INITIAL ACTIVATION OF THE... INTERACTIONS AMONG ABC GENES NEW INSIGHTS INTO AP2... HUA1, HUA2, HEN2, AND... HEN1 FUNCTIONS TO PROCESS... miRNA172 ALSO FUNCTIONS TO... FUTURE PROSPECTS REFERENCES

 
The first unifying principle in the flower development field is the ABC model. This model, initially proposed in the early 1990s based on genetic experiments in Antirrhinum and Arabidopsis, is striking in its simplicity and is applicable to a wide range of angiosperm species, both dicots and monocots, including economically important grass species such as rice and maize (Bowman et al., 1991; Coen and Meyerowitz, 1991; Ambrose et al., 2000; Fornara et al., 2003). The Arabidopsis flower, like most angiosperm flowers, consists of four organ types that are arranged in a series of concentric rings or whorls. From outside to inside, the flower consists of sepals in whorl 1, petals in whorl 2, stamens in whorl 3, and carpels in whorl 4. The ABC model postulates that three activities, A, B, and C, specify floral organ identity in a combinatorial manner. Specifically, A alone specifies sepals, A+B specifies petals, B+C specifies stamens, and C alone specifies carpels. A second major aspect is that A and C classes are mutually repressive. In the absence of A, C activity is present throughout the flower. Likewise, in the absence of C, A activity is present throughout the flower. Throughout the 1990s, the ABC genes were cloned from a wide range of species, and numerous molecular studies were performed. These molecular experiments largely support the major tenets of the ABC model (reviewed by Weigel and Meyerowitz, 1994; Yanofsky, 1995; Ng and Yanofsky, 2001b; Lohmann and Weigel, 2002).

The second major unifying principle involves the central role of the LEAFY (LFY) gene (Coen et al., 1990; Weigel et al., 1992). LFY orthologs are present in a wide range of flowering and nonflowering plant species (Frohlich and Parker, 2000; Gocal et al., 2001). In many developmental contexts, LFY is necessary and sufficient to specify a meristem as floral (Weigel and Nilsson, 1995). In addition, LFY serves two key roles in specifying flowers. First, LFY is a key integrator of the outputs of floral inductive pathways (Nilsson et al., 1998; Blazquez and Weigel, 2000). Second, LFY is a key activator of the floral organ identity ABC genes (Weigel and Meyerowitz, 1993; Parcy et al., 1998; Lenhard et al., 2001; Lohmann et al., 2001). Both aspects of LFY function are described in more detail below.

Broadly speaking, flower development can be divided into four steps that occur in a temporal sequence. First, in response to both environmental and endogenous signals, the plant switches from vegetative growth to reproductive growth; this process is controlled by a large group of flowering time genes. Second, signals from the various flowering time pathways are integrated and lead to the activation of a small group of meristem identity genes that specify floral identity. Third, the meristem identity genes activate the floral organ identity genes in discrete regions of the flower. Fourth, the floral organ identity genes activate downstream "organ building" genes that specify the various cell types and tissues that constitute the four floral organs.

	MULTIPLE FLORAL INDUCTIVE PATHWAYS CONTROL THE TIMING OF FLOWERING

TOP INTRODUCTION UNIFYING PRINCIPLES OF FLOWER... MULTIPLE FLORAL INDUCTIVE... INTEGRATION OF FLOWERING SIGNALS... INTERACTION BETWEEN AP1 AND... TFL1 FLORAL ORGAN IDENTITY GENES A NEW ADDITION TO... D CLASS AND OVULE... THE MAJORITY OF ABCE... COMBINATIONS OF ABCE GENES... HIGHER ORDER MADS COMPLEXES? ACTIVATION OF FLORAL ORGAN... INITIAL ACTIVATION OF THE... INITIAL ACTIVATION OF THE... INTERACTIONS AMONG ABC GENES NEW INSIGHTS INTO AP2... HUA1, HUA2, HEN2, AND... HEN1 FUNCTIONS TO PROCESS... miRNA172 ALSO FUNCTIONS TO... FUTURE PROSPECTS REFERENCES

 
The flowering time genes function on four major promotion pathways: long-day photoperiod, gibberellin (GA), autonomous, and vernalization. Mutants in the long-day photoperiod promotion pathway are late flowering when grown in long-day photoperiods. Many long-day pathway genes encode proteins involved in light perception (e.g., PHYTOCHROME A and CRYPTOCHROME2) or components of the circadian clock (e.g., GIGANTEA and ELF3) (reviewed by Reeves and Coupland, 2000; Mouradov et al., 2002; Hayama and Coupland, 2003). The light and clock components ultimately lead to the activation of CONSTANS (CO). co mutants are late flowering, particularly in long-day photoperiods (Koornneef et al., 1991). Overexpression of CO results in very early flowering (Simon et al., 1996; Onouchi et al., 2000). CO encodes a nuclear protein that contains two B-box zinc finger domains (Putterill et al., 1995). Despite the presence of zinc finger domains, there is no evidence that CO is a sequence-specific DNA binding protein. Instead, it seems likely that CO is a component of a transcriptional activation complex that is directed to specific target genes by another component of the complex that possesses sequence-specific DNA binding activity.

A second flowering time pathway involves the promotion of flowering by GA. Mutants defective in the biosynthesis of GA, such as ga1, exhibit dramatic delays in the timing of flowering when grown in short days but not long days, suggesting that GA is an important stimulator of flowering in the absence of long-day promotion (Wilson et al., 1992; Moon et al., 2003). To date, this pathway consists of only GA biosynthetic and GA response genes. In other words, no genes have been isolated that are clearly on a GA output pathway specific for flowering time control.

Genes on the third pathway, the autonomous pathway, function to control flowering in a photoperiod-independent manner. As a facultative long-day plant, Arabidopsis flowers more rapidly when grown in long days, but it does eventually flower when grown in noninductive short-day photoperiods. Autonomous pathway components play a role in this promotion. The fourth major pathway is the vernalization pathway. An extended cold treatment (vernalization) that mimics overwintering stimulates flowering in many Arabidopsis accessions.

The details of the functions and interactions among the flowering time genes have been the focus of several recent reviews (Koornneef et al., 1998; Mouradov et al., 2002; Simpson and Dean, 2002) and will not be described in detail here. Instead, I will focus on how flowering time signals are integrated and how these signals function to activate downstream meristem identity genes (Figure 1).

View larger version (19K): [in this window] [in a new window]   Figure 1. Major Floral Inductive Pathways. Signals from the four major floral inductive pathways are integrated by FLC, SOC1, FT, and LFY. Interactions demonstrated to be direct are indicated in gray.

The second subclass, the floral meristem identity genes, specify lateral meristems in Arabidopsis to develop into flowers rather than leaves or shoots. After bolting, Arabidopsis plants produce between two and five cauline leaves on the primary inflorescence before developing flowers. In the axil of each of the cauline leaves is a secondary inflorescence meristem that gives rise to a secondary shoot. In Arabidopsis, LFY and APETALA1 (AP1) specify the lateral primordia to develop as flowers rather than shoots. Both lfy and ap1 single mutants exhibit a partial conversion of flowers to shoots (Irish and Sussex, 1990; Schultz and Haughn, 1991; Weigel et al., 1992; Bowman et al., 1993). In lfy ap1 double mutants, lateral meristems in the plant are not specified as floral and instead strongly resemble shoots. lfy ap1 plants have a phenotype very similar to that of 35S:TFL1 (Weigel et al., 1992). Ectopic expression of LFY or AP1 converts the inflorescence shoot apical meristem to a flower; 35S:AP1 and 35S:LFY flowers exhibit a terminal flower phenotype similar to that of tfl1 mutants (Mandel and Yanofsky, 1995; Weigel and Nilsson, 1995). Although AP1 and LFY are the major floral meristem identity genes, other genes such as CAULIFLOWER (Bowman et al., 1993; Kempin et al., 1995), FRUITFULL (Gu et al., 1998; Ferrandiz et al., 2000), and AP2 (Jofuku et al., 1994; Okamuro et al., 1996, 1997b) play secondary roles in specifying floral meristem identity.

Both LFY and AP1 encode sequence-specific DNA binding transcription factors. AP1 is a member of the MADS family (Huijser et al., 1992; Mandel et al., 1992), whereas LFY encodes a plant-specific protein that exhibits no strong similarity to other genes in Arabidopsis (Coen et al., 1990; Weigel et al., 1992). Transcription of AP1 and LFY in lateral meristems in many developmental contexts is sufficient to specify them as floral (Mandel and Yanofsky, 1995; Weigel and Nilsson, 1995). However, the fact that 35S:LFY 35S:AP1 double transgenic plants still undergo a vegetative growth phase, as indicated by the development of a small number of vegetative rosette leaves, suggests that there are other factors, independent of LFY and AP1, that determine the competence of the plant to flower (Blazquez et al., 1997).

	INTEGRATION OF FLOWERING SIGNALS BY FLC, SOC1, FT, AND LFY

TOP INTRODUCTION UNIFYING PRINCIPLES OF FLOWER... MULTIPLE FLORAL INDUCTIVE... INTEGRATION OF FLOWERING SIGNALS... INTERACTION BETWEEN AP1 AND... TFL1 FLORAL ORGAN IDENTITY GENES A NEW ADDITION TO... D CLASS AND OVULE... THE MAJORITY OF ABCE... COMBINATIONS OF ABCE GENES... HIGHER ORDER MADS COMPLEXES? ACTIVATION OF FLORAL ORGAN... INITIAL ACTIVATION OF THE... INITIAL ACTIVATION OF THE... INTERACTIONS AMONG ABC GENES NEW INSIGHTS INTO AP2... HUA1, HUA2, HEN2, AND... HEN1 FUNCTIONS TO PROCESS... miRNA172 ALSO FUNCTIONS TO... FUTURE PROSPECTS REFERENCES

 
One of the key events in the development of flowers is the activation of LFY and AP1. LFY and AP1 respond, either directly or indirectly, to outputs of flowering time pathways. Some of the outputs of the flowering time pathways are integrated by LFY, whereas others are integrated upstream or in parallel to LFY by FLOWERING LOCUS C (FLC), SUPPRESSOR OF OVEREXPRESSION OF CONSTANS (SOC1), and FLOWERING LOCUS T (FT).

Repressive signals from the autonomous and vernalization pathways are integrated by the floral repressor FLC (Figure 1) (Michaels and Amasino, 1999; Sheldon et al., 2000b). FLC also integrates positive regulatory signals from the genes FRIGIDA (FRI) (Johanson et al., 2000) and PHOTOPERIOD INDEPENDENT EARLY FLOWERING1 (PIE1) (Noh and Amasino, 2003). FLC encodes a MADS transcription factor. Mutations in FLC result in early flowering, whereas overexpression of FLC causes late flowering. There is a strong correlation between the levels of FLC RNA/protein and the timing of flowering: high levels of FLC correlate with late flowering, and low levels of FLC correlate with early flowering. The autonomous pathway genes function to downregulate the levels of FLC RNA/protein. The best described molecular mechanism involves the autonomous pathway gene FLOWERING LOCUS D (FLD), which encodes a protein with similarity to a human protein that is a component of the histone deacetylase complex (He et al., 2003). Histone deacetylases function as transcriptional repressors by deacetylating histones, resulting in a transcriptionally inactive chromatin state. In fld mutants, histone H4 is hyperacetylated in the vicinity of the FLC transcription start site. The region of FLC that mediates FLD function is a 295-bp region of the first intron that, when deleted from FLC, results in both hyperacetylation of the FLC locus (resulting in high levels of FLC RNA) and a late-flowering phenotype (similar to fld mutants) (He et al., 2003). In a second autonomous pathway mutant, fve, histone H4 also is hyperacetylated at the FLC locus. At present, the molecular mechanisms for how other autonomous pathway genes control the levels of FLC are not well understood but are the focus of active investigation (Rouse et al., 2002). One intriguing possibility is that the autonomous pathway genes FY and FCA function to regulate the processing of FLC RNA (Quesada et al., 2003; Simpson et al., 2003).

Vernalization also results in a reduction in FLC RNA/protein levels. Several lines of evidence suggest that vernalization controls FLC epigenetically, either by altering the methylation state of FLC or by controlling chromatin structure (Sheldon et al., 2000a; Gendall et al., 2001). The establishment and maintenance of the downregulation of FLC by vernalization requires both 5' promoter sequences and intragenic sequences (Sheldon et al., 2002). Deletion of the large 2.8-kb first intron of FLC, but retention of exons 1 and 2 and 5' promoter sequences, results in a failure to maintain vernalization, suggesting that intron 1 of FLC mediates the maintenance of vernalization. This fits nicely with the putative role of VERNALIZATION2 (VRN2), a gene necessary for the maintenance of vernalization (i.e., stable downregulation of FLC levels after vernalization) (Gendall et al., 2001). VRN2 encodes a Polycomb group protein. In Drosophila and mammals, Polycomb proteins are important for stable transcriptional repression and are postulated to function by altering chromatin structure (Orlando, 2003). Interestingly, the FLC activator PIE1 encodes a protein with similarity to ATP-dependent chromatin remodeling proteins; in other systems, PIE1-like proteins function to put chromatin in a transcriptionally active state (Noh and Amasino, 2003).

In turn, FLC functions to repress the floral activator SOC1 (Figure 1) (Lee et al., 2000; Hepworth et al., 2002). SOC1, like FLC, encodes a MADS transcription factor (Lee et al., 2000). SOC1 is activated by the long-day promotion pathway via CO (Samach et al., 2000) as well as by the GA pathway (Borner et al., 2000; Moon et al., 2003). Integration of the FLC and CO signals is mediated by discrete elements in the SOC1 promoter (Hepworth et al., 2002). A consensus MADS binding sequence in the SOC1 promoter can be bound by FLC in vitro. Mutation of this binding sequence abolishes repression of SOC1 by FLC. Although a CO-responsive region of the SOC1 promoter also was defined, binding of CO to this sequence could not be demonstrated, either because the activation is indirect or because CO requires a cofactor for sequence-specific DNA binding. Future experiments will distinguish between these possibilities.

Although the GA-responsive element in the SOC1 promoter has not been defined, it is clear that removal of the FLC repression of SOC1 is not sufficient to result in high SOC1 transcript levels; upregulation of SOC1 also requires positive activation by either the GA or the long-day promotion pathway. The best evidence that the release of FLC repression is not sufficient for SOC1 upregulation comes from an analysis of ga1 mutant plants that express high levels of FLC RNA/protein because they contain functional alleles of both FRI and FLC. When short-day-grown ga1 FRI FLC plants are vernalized, levels of FLC RNA decrease in response to vernalization treatment but levels of SOC1 do not increase. Thus, the upregulation of SOC1 requires activation by the long-day pathway either via CO or via the GA pathway. In short days, the GA pathway is the only pathway that can activate SOC1 (Moon et al., 2003).

Like SOC1, LFY is a key integrator of output signals from the long-day promotion and GA pathways (Blazquez et al., 1998; Nilsson et al., 1998; Blazquez and Weigel, 2000). Separate LFY promoter elements have been shown to mediate the response to long days (photoperiod promotion) and short days (GA promotion) (Blazquez and Weigel, 2000). The GA effects on the LFY promoter require an 8-bp binding site that is a perfect match for the sequence recognized by a MYB transcription factor (Blazquez and Weigel, 2000). A MYB protein, AtMYB33, binds in vitro to a DNA probe containing the 8-bp LFY element but not to a mutant form of this element. Although AtMYB33 is upregulated by GA, it is not known if AtMYB33 is necessary for the GA activation of LFY. Analysis of atmyb33 mutants and overexpression lines should resolve this issue.

The photoperiod promotion effects on LFY may be mediated by SOC1 or by a second MADS gene, AGAMOUS-LIKE24 (AGL24). Like SOC1 loss- and gain-of-function alleles, agl24 loss-of-function mutants are late flowering, and overexpression of AGL24 results in early flowering (Yu et al., 2002; Michaels et al., 2003). At present, it is somewhat controversial whether AGL24 functions downstream of SOC1 (Yu et al., 2002) or in parallel to SOC1 (Michaels et al., 2003). However, it is clear that both SOC1 and AGL24 function upstream of LFY. What is still unclear is whether either SOC1 or AGL24 acts directly on LFY, because binding of SOC1 or AGL24 to the LFY promoter has not been demonstrated.

The third major integrator of flowering time pathways is FT (Kardailsky et al., 1999; Kobayashi et al., 1999). ft mutants are late flowering in long days (Koornneef et al., 1991). The primary input to FT activation is long-day photoperiod promotion mediated via CO. The best evidence for this is the rapid induction of FT RNA in response to an inducible form of CO (CO fused to the rat glucocorticoid receptor) (Kobayashi et al., 1999). 35S:CO plants express increased levels of FT RNA and are very early flowering, but in 35S:CO ft, flowering time is delayed, demonstrating that FT functions downstream of CO (Onouchi et al., 2000; Samach et al., 2000). FT also receives inputs from FLC. This is best illustrated by the downregulation of FT RNA that occurs when 35S:CO is expressed together with 35S:FLC (Hepworth et al., 2002). FT is negatively regulated by EARLY BOLTING IN SHORT DAYS (EBS) (Pineiro et al., 2003). EBS encodes a protein that contains conserved motifs that suggest that EBS functions via chromatin remodeling. The molecular details of how FT integrates signals from CO, FLC, and EBS are unknown. In other words, it is not known if FT itself integrates these signals or whether the signals are integrated by a gene that functions upstream of FT.

The long-day promotion pathway functions by activating LFY and AP1 via separate branches of the photoperiod pathway. Downstream of CO, the pathway splits; one branch functions via SOC1 and LFY, the other via FT. CO is the last identified component of the long-day promotion pathway that is upstream of both LFY and FT. The branch of the pathway that acts via FT appears to promote flowering by ultimately activating AP1 rather than LFY (Ruiz-Garcia et al., 1997; Nilsson et al., 1998).

	INTERACTION BETWEEN AP1 AND LFY

TOP INTRODUCTION UNIFYING PRINCIPLES OF FLOWER... MULTIPLE FLORAL INDUCTIVE... INTEGRATION OF FLOWERING SIGNALS... INTERACTION BETWEEN AP1 AND... TFL1 FLORAL ORGAN IDENTITY GENES A NEW ADDITION TO... D CLASS AND OVULE... THE MAJORITY OF ABCE... COMBINATIONS OF ABCE GENES... HIGHER ORDER MADS COMPLEXES? ACTIVATION OF FLORAL ORGAN... INITIAL ACTIVATION OF THE... INITIAL ACTIVATION OF THE... INTERACTIONS AMONG ABC GENES NEW INSIGHTS INTO AP2... HUA1, HUA2, HEN2, AND... HEN1 FUNCTIONS TO PROCESS... miRNA172 ALSO FUNCTIONS TO... FUTURE PROSPECTS REFERENCES

 
Although AP1 and LFY are necessary to specify floral meristem identity, they do not function independently of one another. Instead, AP1 functions largely downstream of LFY. The best evidence for this comes from analysis of combinations of gain-of-function and loss-of-function alleles of LFY and AP1. The floral promotion effects of 35S:LFY are blocked in an ap1 mutant (Weigel and Nilsson, 1995), but the floral promotion effects of 35S:AP1 are not blocked in a lfy mutant (Mandel and Yanofsky, 1995). However, in 35S:AP1 lfy, floral organ identity is not properly specified, demonstrating that LFY is necessary for the proper expression of floral organ identity genes, and this activity of LFY is independent of AP1.

The activation of AP1 by LFY is postulated to be direct. The best evidence for this comes from experiments that use an inducible form of LFY (fusion of LFY to the rat glucocorticoid receptor [LFY-GR]). The induction of LFY-GR in the presence of a protein synthesis inhibitor results in the rapid upregulation of AP1 RNA, suggesting that LFY directly activates AP1 (Wagner et al., 1999). The AP1 promoter contains a sequence that can be bound in vitro by the LFY protein (Parcy et al., 1998), but this sequence has not yet been demonstrated to be necessary in planta for LFY activation of AP1.

	TFL1

TOP INTRODUCTION UNIFYING PRINCIPLES OF FLOWER... MULTIPLE FLORAL INDUCTIVE... INTEGRATION OF FLOWERING SIGNALS... INTERACTION BETWEEN AP1 AND... TFL1 FLORAL ORGAN IDENTITY GENES A NEW ADDITION TO... D CLASS AND OVULE... THE MAJORITY OF ABCE... COMBINATIONS OF ABCE GENES... HIGHER ORDER MADS COMPLEXES? ACTIVATION OF FLORAL ORGAN... INITIAL ACTIVATION OF THE... INITIAL ACTIVATION OF THE... INTERACTIONS AMONG ABC GENES NEW INSIGHTS INTO AP2... HUA1, HUA2, HEN2, AND... HEN1 FUNCTIONS TO PROCESS... miRNA172 ALSO FUNCTIONS TO... FUTURE PROSPECTS REFERENCES

 
The regulation of the shoot identity gene TFL1 is poorly understood. TFL1 RNA accumulates in subapical cells in the shoot apex before the vegetative-to-reproductive phase transition, at 2 to 3 days of seedling development when plants are grown in long days (Bradley et al., 1997). TFL1 is expressed at low levels in the vegetative shoot meristem and appears to play a role in preventing premature flowering. At later stages, TFL1 is upregulated and plays a role in repressing the expression of floral meristem identity genes such as LFY and AP1 in the shoot meristem. The upstream regulators of TFL1 are unknown. TFL1 encodes a protein that likely plays a role in signaling, perhaps as an inhibitor of mitogen-activated protein kinase pathways (Corbit et al., 2003). TFL1 is closely related to FT; the two proteins are 50% identical (Kardailsky et al., 1999). This high degree of similarity is surprising because FT and TFL1 have opposite effects on flowering timing: ft mutants are late flowering, and tfl1 mutants are early flowering. TFL1 and FT are members of a six-member gene family in Arabidopsis (Mimida et al., 2001). Future work will be aimed at determining the molecular function of this enigmatic group of proteins.

There is a mutually repressive relationship between the shoot identity gene TFL1 and the meristem identity genes LFY and AP1, and the repression is mediated at the transcriptional level. In tfl1 mutants, LFY and AP1 RNAs are expressed ectopically in the shoot apex (Weigel et al., 1992; Bowman et al., 1993; Gustafson-Brown et al., 1994; Bradley et al., 1997). Similarly, in lfy mutants, TFL1 RNA is expressed in the ectopic shoots (Ratcliffe et al., 1999). In 35S:LFY and 35S:AP1, TFL1 RNA levels are reduced dramatically (Liljegren et al., 1999). Likewise, in 35S:TFL1, LFY and AP1 RNA levels are reduced dramatically (Ratcliffe et al., 1998). It is possible that LFY and/or AP1, because they are transcription factors, bind directly to the TFL1 promoter, but the TFL1 promoter has not been characterized, so the details of this regulation have not been elucidated. Because TFL1 does not encode a transcription factor, the negative regulatory effects of TFL1 on LFY and AP1 are likely to be indirect. At present, the downstream components of the TFL1 pathway have not been identified. It also is not clear where TFL1 fits with regard to the major floral inductive pathways.

	FLORAL ORGAN IDENTITY GENES

TOP INTRODUCTION UNIFYING PRINCIPLES OF FLOWER... MULTIPLE FLORAL INDUCTIVE... INTEGRATION OF FLOWERING SIGNALS... INTERACTION BETWEEN AP1 AND... TFL1 FLORAL ORGAN IDENTITY GENES A NEW ADDITION TO... D CLASS AND OVULE... THE MAJORITY OF ABCE... COMBINATIONS OF ABCE GENES... HIGHER ORDER MADS COMPLEXES? ACTIVATION OF FLORAL ORGAN... INITIAL ACTIVATION OF THE... INITIAL ACTIVATION OF THE... INTERACTIONS AMONG ABC GENES NEW INSIGHTS INTO AP2... HUA1, HUA2, HEN2, AND... HEN1 FUNCTIONS TO PROCESS... miRNA172 ALSO FUNCTIONS TO... FUTURE PROSPECTS REFERENCES

 
One of the important functions of the floral meristem identity genes is to activate the ABC floral organ identity genes. The A class genes specify the identity of sepals and petals that develop in whorls 1 and 2, respectively. A second function of A class genes is to repress C class activity in whorls 1 and 2. In Arabidopsis, there are two A class genes: AP1 and AP2. The B class genes AP3 and PISTILLATA (PI) are required to specify the identity of petals in whorl 2 and stamens in whorl 3. The C class gene AGAMOUS (AG) is necessary to specify the identity of whorl 3 stamens and whorl 4 carpels. The second major function of C class is to repress A class activity in whorls 3 and 4.

The general rule for the floral organ identity genes is that the gene products are expressed in the region of the flower that exhibits defects in mutants. For example, AG RNA is expressed in stamen and carpel primordia and throughout these organs once they have formed (Yanofsky et al., 1990). Similarly, the B class genes AP3 and PI are expressed persistently in petals and stamens (Goto and Meyerowitz, 1994; Jack et al., 1994). AP1 functions as both a floral meristem identity gene and a floral organ identity gene, and the different aspects of AP1 function are reflected in the AP1 expression pattern. During very early floral stages, when AP1 activity is required to specify floral meristem identity, AP1 RNA is present throughout the floral primordium. At later floral stages, when AP1 activity is required to specify the identity of sepals and petals, AP1 RNA is expressed exclusively in whorls 1 and 2 (Mandel et al., 1992; Gustafson-Brown et al., 1994).

The second A class gene, AP2, is the exception to the general rule stated above. Although AP2 functions only in whorls 1 and 2, AP2 RNA is present in all four floral whorls throughout flower development. This puzzling fact was explained recently by the discovery that AP2 is translationally repressed by a microRNA (miRNA) present in whorls 3 and 4 (Chen, 2004). The experiments that led to this exciting discovery are described in more detail below.

	A NEW ADDITION TO THE ABC MODEL

TOP INTRODUCTION UNIFYING PRINCIPLES OF FLOWER... MULTIPLE FLORAL INDUCTIVE... INTEGRATION OF FLOWERING SIGNALS... INTERACTION BETWEEN AP1 AND... TFL1 FLORAL ORGAN IDENTITY GENES A NEW ADDITION TO... D CLASS AND OVULE... THE MAJORITY OF ABCE... COMBINATIONS OF ABCE GENES... HIGHER ORDER MADS COMPLEXES? ACTIVATION OF FLORAL ORGAN... INITIAL ACTIVATION OF THE... INITIAL ACTIVATION OF THE... INTERACTIONS AMONG ABC GENES NEW INSIGHTS INTO AP2... HUA1, HUA2, HEN2, AND... HEN1 FUNCTIONS TO PROCESS... miRNA172 ALSO FUNCTIONS TO... FUTURE PROSPECTS REFERENCES

 
In the last several years, it has become clear that a fourth set of genes, the SEPALLATA (SEP) genes, are necessary for proper floral organ identity (Pelaz et al., 2000, 2001a). The first indication that SEP genes played an important role in petal, stamen, and carpel identity came from cosuppression experiments in petunia and tomato (Angenent et al., 1994; Pnueli et al., 1994). In petunia, a transgenic line designed to cosuppress the SEP3 ortholog FLORAL BINDING PROTEIN2 (FBP2) resulted in floral organ identity transformations in whorls 2, 3, and 4 as well as a loss of floral determinacy. In these FBP2 cosuppressed plants, the SEP/FBP2 subfamily member FBP5 also is downregulated, suggesting that both FBP2 and FBP5 are necessary for organ identity specification of petals, stamens, and carpels as well as for proper floral determinacy (Ferrario et al., 2003). These observations were extended with genetic analysis of the three SEP family members in Arabidopsis: SEP1, SEP2, and SEP3. Single and double sep mutant combinations fail to exhibit a dramatic phenotype in floral development. By contrast, sep1 sep2 sep3 triple mutants consist entirely of sepal-like organs, and their flowers are indeterminate (Pelaz et al., 2000). The phenotype of sep1 sep2 sep3 triple mutants is similar to that of double mutants that lack both B and C class activity, such as pi ag and ap3 ag (Bowman et al., 1989). Based on this fact, the three SEP genes are postulated to function redundantly to specify petals, stamens, and carpels as well as floral determinacy.

The discovery of the importance of the SEP genes has led to a revision of the ABC model (Goto et al., 2001; Theissen, 2001; Theissen and Saedler, 2001). The SEP genes also are referred to as E class genes. The revised "ABCE" model postulates that sepals are specified by A activity alone, petals by A+B+E, stamens by B+C+E, and carpels by C+E (Figure 2).

View larger version (32K): [in this window] [in a new window]   Figure 2. The Revised ABC Model of Flower Development. A, B, C, and E are four activities that are present in adjacent whorls of the flower. These four activities are postulated to function combinatorially to specify the identity of the four organs in the flower: sepals, petals, stamens, and carpels. A second major tenet of the ABC model is that A and C activities are mutually repressive. The specific molecular interactions between A and C class genes as well as their regulators are shown at right. The majority of ABC genes encode MADS domain transcription factors. Recent evidence suggests that MADS proteins function together in complexes larger than a dimer. The "quartet" model postulates that tetramers of MADS proteins specify floral organ identity (shown as colored ovals). Interactions demonstrated to be direct are indicated in red.

	D CLASS AND OVULE DEVELOPMENT

TOP INTRODUCTION UNIFYING PRINCIPLES OF FLOWER... MULTIPLE FLORAL INDUCTIVE... INTEGRATION OF FLOWERING SIGNALS... INTERACTION BETWEEN AP1 AND... TFL1 FLORAL ORGAN IDENTITY GENES A NEW ADDITION TO... D CLASS AND OVULE... THE MAJORITY OF ABCE... COMBINATIONS OF ABCE GENES... HIGHER ORDER MADS COMPLEXES? ACTIVATION OF FLORAL ORGAN... INITIAL ACTIVATION OF THE... INITIAL ACTIVATION OF THE... INTERACTIONS AMONG ABC GENES NEW INSIGHTS INTO AP2... HUA1, HUA2, HEN2, AND... HEN1 FUNCTIONS TO PROCESS... miRNA172 ALSO FUNCTIONS TO... FUTURE PROSPECTS REFERENCES

The Arabidopsis ortholog of FBP11 is AGL11, recently renamed SEEDSTICK (STK) (Pinyopich et al., 2003). STK functions redundantly with the closely related genes AG, SHATTERPROOF1 (SHP1), and SHP2 to specify ovule identity. Although carpels and ovules fail to develop in ag single mutants, carpelloid organs and ovules do develop in ap2 ag double mutants, demonstrating that genes independent of AG can specify carpel and ovule identity. This residual carpelloid identity in ap2 ag double mutants can be removed by eliminating either SPATULA or SHP1/2 activity (Alvarez and Smyth, 1999; Pinyopich et al., 2003). shp1 shp2 double mutants exhibit defects in valve margin development and seed dehiscence, but ovule development is normal (Liljegren et al., 2000). stk single mutants have defects in the development of the funiculus, the stalk that attaches the ovule to the placenta, as well as defects in release of the mature seed from the seed pod, but ovule identity is normal. By contrast, in stk shp1 shp2 triple mutants, most ovules arrest, suggesting that these three genes function redundantly to specify ovule identity. Consistent with this, ectopic expression of STK, SHP1, or SHP2 results in ectopic ovule development (Favaro et al., 2003), a phenotype similar to that observed when FBP11 is overexpressed in petunia. In the end, the Arabidopsis genes STK, SHP1, and SHP2 could be considered as D class genes because they function similarly to the petunia genes FBP7 and FBP11 in specifying ovule identity.

	THE MAJORITY OF ABCE GENES ENCODE MADS TRANSCRIPTION FACTORS

TOP INTRODUCTION UNIFYING PRINCIPLES OF FLOWER... MULTIPLE FLORAL INDUCTIVE... INTEGRATION OF FLOWERING SIGNALS... INTERACTION BETWEEN AP1 AND... TFL1 FLORAL ORGAN IDENTITY GENES A NEW ADDITION TO... D CLASS AND OVULE... THE MAJORITY OF ABCE... COMBINATIONS OF ABCE GENES... HIGHER ORDER MADS COMPLEXES? ACTIVATION OF FLORAL ORGAN... INITIAL ACTIVATION OF THE... INITIAL ACTIVATION OF THE... INTERACTIONS AMONG ABC GENES NEW INSIGHTS INTO AP2... HUA1, HUA2, HEN2, AND... HEN1 FUNCTIONS TO PROCESS... miRNA172 ALSO FUNCTIONS TO... FUTURE PROSPECTS REFERENCES

 
Most ABCDE genes are members of the MADS transcription factor family, including the A class gene AP1, the B class genes AP3 and PI, the C class gene AG, the D class genes STK, SHP1, and SHP2, and the E class genes SEP1, SEP2, and SEP3. The A class gene AP2 is the exception; AP2 encodes a putative transcription factor that is a member of a small plant-specific gene family (Okamuro et al., 1997a; Riechmann and Meyerowitz, 1998). In addition, several flowering time proteins (FLC, SOC1, AGL24, and SHORT VEGETATIVE PHASE [SVP] [Hartmann et al., 2000]) and meristem identity proteins CAULIFLOWER (CAL and FUL) also are MADS proteins. In Arabidopsis, there are >100 MADS genes (Alvarez-Buylla et al., 2000; de Bodt et al., 2003; Parenicova et al., 2003). The MADS family can be divided into two groups based on molecular evolutionary criteria. The vast majority of plant MADS genes of known function are type II MADS genes. The majority of type II MADS proteins have a characteristic domain structure. The MADS domain is located at the N-terminal end and encodes DNA binding, nuclear localization, and dimerization functions (McGonigle et al., 1996; Riechmann and Meyerowitz, 1997; Immink et al., 2002). A second conserved domain, the K domain, mediates protein–protein interaction and dimerization functions (Fan et al., 1997; Yang et al., 2003a). In a subset of plant MADS proteins, the C domain encodes a transcriptional activation domain (Moon et al., 1999; Honma and Goto, 2001). The C domain also has been reported to play a role in the formation of higher order MADS complexes (Egea-Cortines et al., 1999; Honma and Goto, 2001). Recently, the extreme C-terminal end of AP3 was demonstrated to play a role in functional specificity (Lamb and Irish, 2003).

Type I MADS genes do not encode a K domain. Even though there are 60 type I MADS genes in Arabidopsis, it was not until recently that a function was determined for the first member of this class. The type I MADS gene PHERES functions in early seed development (Kohler et al., 2003).

All MADS proteins studied to date bind to DNA as dimers, either homodimers or heterodimers. In Arabidopsis, AG has been demonstrated to bind to DNA either as a homodimer or as a heterodimer with SEP1 (Huang et al., 1996). By contrast, AP3 and PI do not form DNA binding homodimers but instead bind to DNA only as a heterodimer (Riechmann et al., 1996a, 1996b). The fact that neither A (AP1) nor C (AG) class proteins form DNA binding heterodimers in vitro with AP3 or PI makes heterodimerization an unlikely molecular explanation for why petal development requires A+B activities and stamen development requires B+C activities. At present, the in vivo significance of different MADS dimer combinations is not well understood.

	COMBINATIONS OF ABCE GENES ARE SUFFICIENT TO DIRECT FLORAL ORGAN IDENTITY

TOP INTRODUCTION UNIFYING PRINCIPLES OF FLOWER... MULTIPLE FLORAL INDUCTIVE... INTEGRATION OF FLOWERING SIGNALS... INTERACTION BETWEEN AP1 AND... TFL1 FLORAL ORGAN IDENTITY GENES A NEW ADDITION TO... D CLASS AND OVULE... THE MAJORITY OF ABCE... COMBINATIONS OF ABCE GENES... HIGHER ORDER MADS COMPLEXES? ACTIVATION OF FLORAL ORGAN... INITIAL ACTIVATION OF THE... INITIAL ACTIVATION OF THE... INTERACTIONS AMONG ABC GENES NEW INSIGHTS INTO AP2... HUA1, HUA2, HEN2, AND... HEN1 FUNCTIONS TO PROCESS... miRNA172 ALSO FUNCTIONS TO... FUTURE PROSPECTS REFERENCES

 
The failure of floral organs to develop with the correct identity in A, B, C, and E class mutants demonstrates that the ABCE genes are necessary to specify floral organ identity. When expressed ectopically, the ABCE genes are sufficient to direct organ identity in flowers but not in vegetative leaves. For example, ectopic expression of both B class genes (35S:AP3 and 35S:PI) results in flowers that consist of two outer whorls of petals and two inner whorls of stamens, but leaves remain largely vegetative (Krizek and Meyerowitz, 1996). Based on this finding, it was concluded that AP3 and PI are sufficient, within the flower, to direct petal and stamen identity. Similarly, 35S:AG, 35S:SEP1, 35S:SEP2, and 35S:SEP3 do not alter the identity of the vegetative leaves. However, when the E class gene SEP3 is expressed ectopically together with AP3 and PI, both rosette and cauline leaves are converted to organs that resemble petals (Honma and Goto, 2001; Pelaz et al., 2001b). Furthermore, when AG is expressed ectopically together with AP3, PI, and SEP3, the cauline leaves are converted to organs that resemble stamens (Honma and Goto, 2001). These studies demonstrate that the E class genes, together with the ABC genes, are sufficient to direct floral organ identity in vegetative organs.

	HIGHER ORDER MADS COMPLEXES?

TOP INTRODUCTION UNIFYING PRINCIPLES OF FLOWER... MULTIPLE FLORAL INDUCTIVE... INTEGRATION OF FLOWERING SIGNALS... INTERACTION BETWEEN AP1 AND... TFL1 FLORAL ORGAN IDENTITY GENES A NEW ADDITION TO... D CLASS AND OVULE... THE MAJORITY OF ABCE... COMBINATIONS OF ABCE GENES... HIGHER ORDER MADS COMPLEXES? ACTIVATION OF FLORAL ORGAN... INITIAL ACTIVATION OF THE... INITIAL ACTIVATION OF THE... INTERACTIONS AMONG ABC GENES NEW INSIGHTS INTO AP2... HUA1, HUA2, HEN2, AND... HEN1 FUNCTIONS TO PROCESS... miRNA172 ALSO FUNCTIONS TO... FUTURE PROSPECTS REFERENCES

 
Plant MADS proteins have been demonstrated to associate in complexes larger than dimers (Egea-Cortines et al., 1999; Honma and Goto, 2001; Ferrario et al., 2003; Yang et al., 2003b). This has led to a molecular model, the "quartet" model, which has received broad publicity but in fact is supported by only a limited amount of experimental evidence (Jack, 2001; Theissen, 2001; Theissen and Saedler, 2001; Eckardt, 2003).

The quartet model (Theissen, 2001; Theissen and Saedler, 2001) postulates that tetramers of MADS proteins determine floral organ identity (Figure 2). Each tetramer is proposed to consist of two MADS dimers, each of which binds to a single MADS binding site. The tetramers are formed by protein–protein interaction between MADS dimers, resulting in a tetramer that is simultaneously bound to two MADS binding sites. There are at least two molecular mechanisms that explain how these MADS tetramers result in an active transcription complex. One mechanism is that binding to two MADS binding sites is required, but the binding of MADS dimers is cooperative; specifically, binding of one dimer in the tetramer results in increased affinity for local binding of the second dimer in the tetramer. Some target genes have multiple consensus MADS binding sites in their promoters (e.g., GLOBOSA [GLO] in Antirrhinum [Tröbner et al., 1992] and AP3 in Arabidopsis [Hill et al., 1998; Tilly et al., 1998]). A second mechanism is that one or more subunits provides an activation domain to the tetramer to allow for efficient transcriptional activation. For example, both AP3 and PI lack activation domains, but SEP3 and AP1 possess activation domains (Honma and Goto, 2001). Thus, the inclusion of SEP3 or AP1 together with AP3/PI might result in a transcriptionally active complex.

The quartet model makes predictions about the composition of the tetramers in the four whorls of the flower (Figure 2). Specifically, in whorl 2, a combination of AP3/PI-SEP/AP1 is postulated to specify petals; in whorl 3, AP3/PI-SEP/AG is postulated to specify stamens; and in whorl 4, AG/AG-SEP/SEP is postulated to specify carpels.

One line of evidence that MADS proteins form higher order complexes comes from yeast two-hybrid and three-hybrid experiments. A two-hybrid screen using AG as bait identified SEP1, SEP2, and SEP3 as interacting proteins (Fan et al., 1997). A yeast three-hybrid screen, designed to identify proteins that interact with the AP3/PI heterodimer, but not with AP3 or PI alone, led to the isolation of SEP3 and AP1 (Honma and Goto, 2001). Interactions among AP3/PI/AP1 and AP3/PI/SEP3 were confirmed by coimmunoprecipitation experiments (Honma and Goto, 2001), lending support to the hypothesis that MADS proteins form complexes that consist of more than two monomers. Similarly, in petunia, yeast three- and four-hybrid experiments demonstrated the existence of complexes that consist of B+E and B+E+C MADS proteins (Ferrario et al., 2003; Immink et al., 2003).

Evidence suggesting that these higher order MADS complexes are functional comes from DNA binding assays performed with Antirrhinum MADS proteins. In one key experiment, a probe containing two MADS binding sites exhibited enhanced DNA binding in the presence of both SQUAMOSA (SQUA) (the AP1 ortholog) and DEFICIENS (DEF)/GLO (the AP3/PI orthologs) compared with DEF/GLO or SQUA alone (Egea-Cortines et al., 1999). Based on this finding, the authors concluded that the B class proteins DEF/GLO and the A class protein SQUA formed a multimeric DNA binding complex.

At present, the nature of MADS protein complexes in planta is completely uncharacterized. For example, even though there is abundant evidence that AP3 and PI form a heterodimer in vitro and in yeast, an AP3/PI heterodimer has not been isolated from plant cells. Even less is known about other proteins that might be components of plant MADS protein complexes. Future work will focus on the biochemical characterization of MADS protein complexes from plant cells.

	ACTIVATION OF FLORAL ORGAN IDENTITY GENES BY FLORAL MERISTEM IDENTITY GENES

TOP INTRODUCTION