An Arabidopsis F-box protein acts as a transcriptional co-factor to regulate floral development

Flowering is under the control of multiple endogenous and environmental signals, including hormonal cues, photoperiodic changes and exposure to cold temperatures. In Arabidopsis, the LEAFY (LFY) gene is a key integrator of these signals, in that its expression is upregulated in response to these inputs, and in turn LFY acts to coordinately establish the initial expression of the floral homeotic genes(Blazquez and Weigel, 2000; Nilsson et al., 1998; Weigel and Meyerowitz,1993).

LFY encodes a novel plant-specific transcription factor, and its role in activating floral homeotic genes has been shown to be separable from that of conferring floral meristem identity (Parcy et al., 1998). LFY has been shown to bind to a loosely defined CC(A/T)(A/G)(G/T)G(G/T) consensus site in the regulatory regions of the APETALA1 (AP1), APETALA3 (AP3) and AGAMOUS (AG) floral homeotic genes(Busch et al., 1999; Lamb et al., 2002; Lohmann et al., 2001; Parcy et al., 1998; Wagner et al., 1999). However, AP1, AP3 and AG are expressed in distinct spatial domains,while LFY is expressed throughout the floral meristem, implying that the mode of action of LFY in activating different classes of floral homeotic genes is distinct (Parcy et al., 1998). The spatially and temporally restricted transcriptional activation of both AP1 and AG depends on the concerted action of LFY in combination with other transcription factors. AG is expressed in a restricted region consisting of the inner two floral whorls; WUSCHEL (WUS), a homeodomain containing protein that is expressed in the center of the floral meristem, has been shown to directly activate AG in conjunction with LFY, resulting in localized expression(Lenhard et al., 2001; Lohmann et al., 2001). AP1 expression is initially activated throughout the flower in response to LFY activity (Parcy et al.,1998; Wagner et al.,1999), and becomes confined to the first two floral whorls as a consequence of AG-mediated repression in the center of the meristem(Gustafson-Brown et al.,1994).

By contrast, the activation of AP3 expression in developing petal and stamen primordia requires the activity of the F-box protein UNUSUAL FLORAL ORGANS (UFO) in conjunction with that of LFY(Ingram et al., 1995; Jack et al., 1992; Lee et al., 1997; Levin and Meyerowitz, 1995; Wilkinson and Haughn, 1995). Ectopic expression of UFO in plants containing a 35S::UFOtransgene results in a dramatic phenotype in which carpels are transformed to stamens, extra petals are formed and leaf margins are serrated(Lee et al., 1997). This gain-of-function phenotype of UFO in flowers is due to ectopic AP3 expression because the phenotype is completely abolished when functional AP3 is not present(Lee et al., 1997). Every aspect of this gain-of-function phenotype of UFO is abolished when LFY activity is absent, indicating that the function of UFOis dependent on LFY activity (Lee et al., 1997). Furthermore, ectopic expression of LFY and UFO together is sufficient to induce ectopic AP3 expression,while ectopic expression of LFY alone is insufficient(Lee et al., 1997; Parcy et al., 1998). UFO is expressed in all shoot meristems in a dynamic pattern, but in the flower the expression domain of UFO largely coincides with that of AP3, providing regional specification(Lee et al., 1997; Long and Barton, 1998; Samach et al., 1999).

Although LFY and UFO are both required for the appropriate activation of AP3 expression, the molecular processes underlying this regulation have been unclear. Most F-box proteins are components of SCF complexes that act as E3 ubiquitin ligases to mark target proteins for proteasome-dependent degradation(Deshaies, 1999). In the SCF E3 ubiquitin ligases, F-box proteins serve as adaptor components, bringing specific substrates to the complex for ubiquitylation and targeting them for degradation (Craig and Tyers,1999; Feldman et al.,1997; Skowyra et al.,1997). UFO has been shown to participate in an SCF^UFOcomplex biochemically and to genetically interact with ArabidopsisSCF complex components (Ni et al.,2004; Samach et al.,1999; Wang et al.,2003; Zhao et al.,1999), indicating that UFO has a role in targeted proteolysis.

Some F-box proteins have been shown to regulate the availability of transcription factors or their co-factors for initiating or maintaining transcription. This can occur through direct ubiquitylation and proteasome-mediated degradation of transcription factors(Kodadek et al., 2006; Muratani and Tansey, 2003), or by promoting the degradation of co-factors that can modulate associated transcription factor subcellular localization or activity(Ostendorff et al., 2002; Palombella et al., 1994; Perissi et al., 2004). In particular, controlling the activity and abundance of transcription factors by targeted proteolysis appears to be a mechanism by which cells can effectively switch on and off downstream gene expression. In addition to targeting proteins for destruction, ubiquitylation can also have a non-proteolytic role of modifying protein activity (Kaiser et al., 2000), or physically occlude protein-protein interactions(Horwitz et al., 2007).

The genetic evidence that UFO is required for activating AP3expression strongly suggests that the ubiquitin-proteasome system is involved in AP3 regulation. Here, we demonstrate that UFO acts to modulate AP3 transcription via regulating LFY transcriptional activity at the promoter. Based on our observations, we propose that UFO, as a component of an E3 ubiquitin ligase, functions as a co-factor for LFY-induced AP3floral homeotic gene transcription and regulates the activity of the LFY transcription factor in a proteasome-dependent manner.

Plants were grown in 3:1 vermiculite: soil at 22°C under 16-hour light/8-hour dark conditions. For dose studies, lfy-26 homozygotes were crossed to homozygous 35S::UFO-myc plants. Homozygotes for 35S::UFO-myc were selected in the F2 generation. Siblings from the F4 generation segregating for lfy-26 in the 35S::UFO-mychomozygous background were assessed for their phenotypes. For 35S::UFO-Myc; 35S::LFY-FLAG; ufo-2, each plant was subjected to western blot analysis to assay expression levels of epitope tagged proteins. Plants expressing both proteins at high levels were used for co-immunoprecipitations. For 35S::LFY-GR; 35S::UFO; hlr-1, kanamycin was used to identify the appropriate genotype that show a halted root mutant phenotype. Transgenic lines were generated using the floral dip method(Clough and Bent, 1998). 35S::UFO-myc plants were generated previously(Wang et al., 2003); lfy and ufo mutants were obtained from the Arabidopsis Biological Resource Center (Ohio State University, Columbus, OH) and hlr-1 seeds were a kind gift from Dr K. Okada, Kyoto University.

pUFOA and pLFYX used in EMSA assays and GST pull-down assays have been reported (Lamb et al., 2002). Yeast constructs AD-LFY and BD-UFO were generated by inserting the LFY open reading frame into pGAD424 and that of UFO into pGBT9. BD-ΔFUFO was constructed by excising the F-box of UFO (amino acids 1-87) using NcoI and XhoI sites from BD-UFO and religated. Truncated versions of LFY were constructed using Expand High Fidelity PCR (Roche Applied Science, Indianapolis, IN). 35S::LFY-FLAG was constructed by inserting three FLAG tags (DYKDDDDK) at the C terminus of LFY. To generate SRDX fused to UFO in frame, the ORF of UFO was cloned into the Gateway entry vector pDONR201 and then recombined into the destination vector pH35GEAR [a gift from Taku Demura(Kubo et al., 2005)],resulting in 35S::UFO-SRDX. The negative control(35S::UFO-mSRDX) was generated by replacing the SRDX domain with a FLAG tag in 35S::UFO-SRDX. This results in a UFO fusion protein containing an additional 19 amino acids (PAFLYKVVDNSDYKDDDKA; the FLAG tag is underlined; remaining sequences derived from vector) at the C-terminal end. 35S::UFO-SRDX contains an additional 24 amino acids(AFLYKVVVISRPLDLDLELRLGFA; SRDX domain is underlined; remaining sequences derived from vector). All binary expression constructs were transferred to Agrobacterium tumefaciens strain GV3101 by electroporation.

Proteins for EMSA were transcribed and translated in vitro using wheat germ extract (Promega, Madison, WI). Preparation of DNA probes, binding reactions and gel running conditions were as described previously(Hill et al., 1998; Lamb et al., 2002).

In frame fusion constructs were generated in pGAD424 or pGBT9 (Clontech,Palo Alto, CA) and transformed into the yeast strain Y190. lacZliquid assays were performed as described previously(Tan and Irish, 2006).

GST-LFY and GST constructs were expressed in Escherichia colistrain BL21 (DE3) codon plus (Stratagene, La Jolla, CA) and affinity purified using glutathione sepharose 4B beads (Amersham Biosciences, Piscataway, NJ). Approximately 1 μg of GST-LFY and GST-LFY-N2, based on spectrophotometric measurements, were incubated overnight with [³⁵S] Met-labeled in vitro transcribed and translated UFO at 4°C in 0.2 ml of buffer A [100 mM NaCl, 50 mM Tris-Cl (pH 7.5), 0.1 mM EDTA, 0.1 mM EGTA, 50 mM NaF, 0.2% Triton X-100, 0.1% β-mercaptoethanol, complete protease inhibitor (Roche,Applied Science, Indianapolis, IN) and 1 mM phenylmethylsulfonyl fluoride]. The beads were washed four times with ice-cold buffer B [100 mM NaCl, 50 mM HEPES (pH 7.5), 0.1 mM EDTA, 1 mM phenylmethylsulfonyl fluoride and complete protease inhibitor] at 4°C. The washed beads were boiled in SDS sample buffer, and proteins were separated by SDS-PAGE. Gels were fixed, dried and exposed to X-ray film.

Total protein from inflorescence tissue was extracted using an extraction buffer containing 20 mM Tris-HCl (pH 7.5), 150 mM NaCl, 10 mM MgCl₂, 10% glycerol, 0.1% NP-40, 1 mM phenylmethylsulfonyl fluoride, and complete protease inhibitor (Roche, Applied Science,Indianapolis, IN). Immunoprecipitation was performed using anti-FLAG M2 agarose (Sigma, St Louis, MO) at 4°C overnight. After washing four times with extraction buffer lacking NP-40, proteins were eluted using 3XFLAG(Sigma, St Louis, MO). Immune complexes were analyzed by SDS-PAGE and immunoblotted using anti-FLAG (M2, Sigma, St Louis, MO) or anti-Myc (9E10,Covance, Philadelphia, PA) antibodies and were detected using an ECL kit(Amersham Biosciences, Buckinghamshire, UK). To detect LFY ubiquitylation,immunoprecipitations were performed as above, except that the extraction buffer contained 20 μM MG132, and incubated for 3 hours at 4°C. The pellet was washed with extraction buffer four times and eluted using 3XFLAG. Samples were fractionated by SDS-PAGE and immunoblotted with either anti-FLAG(M2, Sigma, St Louis, MO) or anti-ubiquitin (Sigma, St Louis, MO) antibodies. For anti-ubiquitin immunoblot, the membrane was pre-treated with denaturing buffer [6 M guanidine HCl, 20 mM Tris-HCl (pH 7.5), 5 mMβ-mercaptoethanol and 1 mM phenylmethylsulfonyl fluoride] as described(Miller et al., 2004).

The procedure essentially followed(Wang et al., 2002) and(Ito et al., 1997). Approximately 5 g of inflorescence tissue from either UM or UM;lfy-26 was used as starting material. Immunoprecipitation (IP) was performed using 30 μl of anti-Myc affinity matrix (Covance, Philadelphia,PA). For control IP, the same amount of normal mouse IgG serum conjugated to agarose beads (Santa Cruz Biotechnology, Santa Cruz, CA) was used. Inflorescence tissue from either 35S::LFY-FLAG or 35S::LFY-FLAG;ufo-2 was used to IP the LFY-chromatin complex using 30μl of anti-FLAG resin (Sigma, St Louis, MO). LFY-FLAG was eluted from the resin using 3XFLAG peptides. Amplification of genomic fragments was performed following standard semi-quantitative PCR conditions, with 28 to 35 cycles of amplification.

Seven day-old seedlings of different genotypes were transferred to solid MS media containing 1 μM dexamethasone (DEX, Sigma, St Louis, MO) or 0.01%ethanol for mock treatment. Epoxomicin (10 μM) (Sigma, St Louis, MO) in liquid MS media was added to DEX containing MS and vacuum infiltrated for 1 minute. DMSO was used as mock control. After 4 hours of treatment, seedlings were harvested in liquid nitrogen and used for RNA preparation. For epoxomicin treatment, inflorescences were treated with 1 μM epoxomycin in 1/2 MS,0.02% silwet for 4 hours.

RNA was extracted from seedlings using Trizol (GibcoBRL, Frederick, MD). cDNA was synthesized using Superscript III Reverse Transcriptase (Invitrogen,Carlsbad, CA). Quantitative real-time PCR was performed in 25 μl using the QuantiTect SYBR Green PCR kit (Qiagen, Valencia, CA) on an ABI Prism 7900 HT Sequence Detector (Applied Biosystems, Foster City, CA). Reaction conditions were: 95°C for 15 minutes, followed by 40 cycles of 94°C for 15 seconds, 53°C for 30 seconds and 72°C for 30 seconds. Primer sequences for AP3 have been reported (Lamb et al., 2002). Triplicate assays were carried out using at least two independent RNA samples. The amount of mRNA was quantified using DART-PCR 1.0 (Peirson et al., 2003) or 2^-ΔΔCt and normalized either to the value of eukaryotic translation elongation factor-1α (EF-1α) or to that of glyceraldehyde-3-phosphate dehydrogenase (GAPDH). PCR products were subjected to agarose gel electrophoresis to test the specificity of the amplification.

Plants containing an epitope tagged version of UFO driven by the 35S promoter, 35S::UFO-Myc (Wang et al., 2003), displayed an ectopic petal and stamen phenotype(Fig. 1A,B) and serrated leaves(Fig. 1F, left). This phenotype is identical to that produced by 35S::UFO(Lee et al., 1997), indicating that the epitope tag does not interfere with UFO activity. In addition, the complete elimination of LFY function in the 35S::UFO-Myc plants also resulted in a lfy mutant phenotype, with a complete transformation of flowers to shoots (Fig. 1A,arrows) or intermediate structures between flower and shoot(Fig. 1D). We observed that when LFY is heterozygous in the 35S::UFO-Myc background, the plants showed suppression of the ectopic 35S::UFO-Myc phenotype(Fig. 1A,C,E,F). Unlike 35S::UFO-Myc plants, which are completely infertile because of a complete transformation of carpels to stamenoid organs(Fig. 1A,B,E), 35S::UFO-Myc; lfy-26/+ flowers produced short curved siliques containing viable seeds (Fig. 1A,E). Most 35S::UFO-Myc; lfy-26/+ flowers appeared similar to wild type with four petals and six stamens(Fig. 1C). Serration of leaf margins, observed in 35S::UFO-Myc plants, was also reduced in the 35S::UFO-Myc; lfy-26/+ plants(Fig. 1F, middle), consistent with a proposed role for LFY during vegetative development(Lee et al., 1997). These genetic data indicate that there are dose dependent effects of LFY on the 35S::UFO-Myc phenotype, suggesting that LFY and UFO physically interact in a complex.

To demonstrate explicitly a physical interaction between LFY and UFO, we carried out several assays. Electrophoretic mobility shift assays (EMSA) have shown that LFY binds directly to sequences in the AP3 promoter(Lamb et al., 2002)(Fig. 2A). We showed that EMSA using both LFY and UFO proteins resulted in a supershift of the DNA/protein complex, even though UFO itself did not show any DNA-binding affinity for AP3 promoter sequences (Fig. 2A). These results indicate that UFO and LFY can directly and physically interact, at least within the context of AP3 promoter sequences. To examine whether such interactions were dependent on the presence of DNA, we also performed GST-pull down assays. As shown in Fig. 2B, a GST-LFY fusion protein could successfully pull down radiolabeled UFO, whereas GST alone could not. In addition, yeast two-hybrid assays that we carried out also demonstrated that LFY and UFO can physically associate in the absence of AP3 promoter DNA. LFY directly interacted with UFO when UFO was fused to the GAL4 DNA-binding domain and LFY was fused to the GAL4 activation domain(Fig. 2C). However, the reciprocal fusion of the GAL4 domain to LFY and UFO did not result in a positive interaction (Fig. 2C). As the conformation of an unoccupied F-box has been suggested to occlude protein-protein interaction domains(Deshaies, 1999), we reasoned that the protein-protein interaction domain of UFO would be more accessible when the F-box domain of the protein was deleted. As expected, a ΔFUFO construct that lacks the F-box showed a much more robust interaction with LFY in yeast, irrespective of whether ΔFUFO was fused to the Gal4 activation domain or to the Gal4 DNA-binding domain(Fig. 2C). We also found that LFY protein can homodimerize in yeast cells(Fig. 2C), while UFO does not(data not shown).

To test whether or not this interaction can be observed in a physiologically relevant context, we examined the ability of LFY and UFO to interact in inflorescence tissue. We generated transgenic plants harboring epitope tagged versions of both LFY and UFO. Because 35S::UFO-Myc;35S::LFY-FLAG plants arrest at the seedling stage in a manner similar to that of 35S::UFO; 35S::LFY plants(Parcy et al., 1998) (and data not shown), we used 35S::LFY-FLAG; 35::UFO-Myc; ufo-2 (LF;UM) plants as a source of inflorescence tissues. These LF; UMplants produce flowers with petals or petaloid organs, indicating that both transgenes are functional, and that the dose of UFO is appropriate to bypass lethality (see Fig. S1 in the supplementary material). For co-immunoprecipitations, we used a line in which both the 35S::LFY-FLAG and 35S::UFO-Myc transgenes showed robust expression (Fig. 2D, input lanes). We performed immunoprecipitations with an anti-FLAG antibody to pull down the LFY immune complex and tested whether or not UFO was co-immunoprecipitated. As shown in Fig. 2D, UFO-Myc was detected in the anti-FLAG immunoprecipitated lane from LF; UM tissue, indicating that UFO co-immunoprecipitates with LFY. This interaction was weak but detectable, suggesting that UFO-LFY interactions may be transient. Neither control IgG immunoprecipitations nor immunoprecipitations from plants expressing only epitope-tagged UFO(UM) showed any UFO bands, indicating that this interaction is specific. Together, these observations demonstrate that UFO physically interacts with LFY both in vitro and in vivo.

Phylogenetic analyses of LFY orthologs identified conserved N-terminal and DNA-binding C-terminal domains (Maizel et al., 2005). Nonsense mutations in LFY, even those located within the C-terminus [e.g. lfy-26 that lacks the last 48 amino acids(Lee et al., 1997; Maizel et al., 2005)], cause strong lfy mutant phenotypes, implying that the C-terminal domain is crucial for LFY function. To determine which domains in LFY are required for the interaction with UFO, we generated four truncated versions of LFY: LFYN1(amino acids 1-141), LFYC1 (amino acids 142-420), LFYN2 (amino acids 1-375,equivalent to the truncated mutant form found in lfy-26) and LFYC2(amino acids 376-420) (Fig. 3A). Yeast two-hybrid assays revealed that LFYN1 failed to interact with ΔFUFO, while LFYC1 still retained the interaction albeit at somewhat reduced levels (Fig. 3B), indicating that the C-terminal domain of LFY is crucial for interaction with UFO and that the N-terminal domain may stabilize the interaction. Interestingly, LFYN2 did not show any interaction withΔFUFO. We confirmed the abrogation of the interaction of LFYN2 with full-length UFO using GST pull-down assays, which were performed with bacterially produced GST fusion proteins and protein extracts from inflorescence tissues of 35S::UFO-Myc plants(Fig. 3C). Full-length LFY protein efficiently pulled down UFO protein from plant extracts, while LFYN2 failed to do so. Considering that a LFYN2 construct has also been shown to lack DNA-binding activity (Maizel et al.,2005), it appears that an intact C terminus is important for LFY function in two ways: conferring stable DNA binding affinity and interacting with UFO.

Interestingly, LFY contains a sequence motif (DSGTH; amino acids 139-143)similar to the DSGXS `destruction box' motif found in substrates of the F-box protein β-TRCP (Winston et al.,1999; Yaron et al.,1997). We tested whether the motif present in LFY can serve as the recognition site for UFO by mutagenizing the aspartic acid or serine residue to alanine (Fig. 3A). Yeast two-hybrid assays showed that mutation of either D139 or S140 had no effect on LFY interaction with either UFO or ΔFUFO(Fig. 3B). This observation indicates that the cryptic destruction box present in LFY is not required for recognition by the F-box protein UFO.

The observation of physical interaction between LFY and UFO led us to test whether UFO could potentially act as a transcriptional co-factor and be recruited to the AP3 promoter through its physical association with LFY. Two AP3 promoter elements (termed DEE and PEE) have been shown to be essential for establishing the initial domain of AP3 expression(Hill et al., 1998). LFY has been previously shown to bind to the DEE region of the AP3 promoter in vitro (Lamb et al., 2002). We first sought to confirm the LFY-AP3 promoter association in vivo using chromatin immunoprecipitation (ChIP). ChIP carried out using inflorescence tissue from 35S::LFY-FLAG plants showed that both the DEE and PEE sequence elements are enriched in the immunoprecipitated chromatin when compared with controls (Fig. 4A,B). We confirmed that we could recapitulate the interaction of LFY protein with other known DNA targets such as the AP1 promoter(William et al., 2004), as AP1 promoter sequences were enriched in our ChIP assays(Fig. 4B). ChIP assays from inflorescence tissues of 35S::LFY-FLAG; ufo-2 plants also resulted in enrichment of the promoter region tested(Fig. 4B), indicating that functional UFO is not required for LFY binding to target sequences, although it may facilitate such binding. Control IPs or ChIP assays performed using wild-type tissue showed no significant enrichment of the promoter regions tested, nor did the control Mu genomic region show specific enrichment,indicating that the binding of LFY to the promoter fragments of AP1and AP3 was specific (Fig. 4B).

To examine if UFO can be recruited to the AP3 promoter, we performed ChIP assays using inflorescence tissues of 35S::UFO-Mycplants and examined whether AP3 promoter fragments were enriched in the immunoprecipitated chromatin. ChIP analyses(Fig. 4C) showed that UFO associates with the AP3 promoter and localizes to both AP3promoter elements. Furthermore, we observed that the association of UFO with the AP3 promoter was abolished when ChIP was performed using extracts from lfy-26 mutant plants harboring the 35S::UFO-Myctransgene (Fig. 4C). As this association is dependent on the presence of functional LFY protein, these results indicate that UFO is recruited to the AP3 promoter through its interaction with LFY.

To assay the ability of UFO to function as a transcriptional co-factor of LFY in planta, we fused the strong SRDX transcriptional repressor domain that contains an EAR motif (Hiratsu et al., 2003) to UFO and introduced it into Arabidopsis plants under the control of the constitutive 35Spromoter. If UFO acts as a LFY transcriptional co-factor, then 35S::UFO-SRDX should confer dominant repressor activity to the transcription events that LFY governs, resulting in a phenotype similar to that of lfy mutants. However, if UFO does not act via modulating transcription, the 35S::UFO-SRDX construct should generate a phenotype more similar to that of a UFO gain-of-function transgene. We also generated 35S::UFO-mSRDX transgenic plants in which the SRDX domain has been replaced with an epitope tag of similar size;these plants were used to control for the possibility that a chimeric UFO protein would non-specifically abrogate LFY activity.

As expected, we observed that ten out of 30 independent primary 35S::UFO-SRDX transformants showed a range of lfy-like mutant phenotypes (Fig. 5). In addition, one line appeared indistinguishable from the strong ufo-2 mutant (Fig. 5F), whereas none of the transgenic lines showed the gain-of-function phenotype of UFO(Fig. 5; see Fig. S2 in the supplementary material). As is prominently seen in lfy mutants, these ten lines showed defects in petal and stamen development(Fig. 5; see Fig. S2). Furthermore, we also observed a complete or partial transformation of flowers to inflorescence shoots, which is a hallmark of lfy mutants(Huala and Sussex, 1992; Schultz and Haughn, 1991; Weigel et al., 1992). For example, two severe lines had up to six secondary inflorescences (data not shown) and most lines harbored intermediate structures between inflorescence shoots and flowers (Fig. 5B-D). These intermediate structures were often subtended by bracts(Fig. 5I, arrow), with leaf-like organs arranged in a spiral pattern rather than in a whorl(Fig. 5B-D) and internode elongation was also often observed (Fig. 5C,D). The only difference we observed in 35S::UFO-SRDXplants compared with lfy mutants is that secondary flowers rarely form in the axils of leaf-like organs in 35S::UFO-SRDX plants when compared with lfy mutants, which is consistent with the proposed role of LFY in functioning as a transcriptional repressor in inhibiting secondary flower formation (Parcy et al.,2002). Later in development, the 35S::UFO-SRDX flowers tended to terminate with fused carpelloid leafy organs(Fig. 5E), which is also a typical characteristic of lfy mutants(Huala and Sussex, 1992; Schultz and Haughn, 1991; Weigel et al., 1992). In most lines, the transformation of flowers to inflorescence shoots was less severe in later arising flowers (Fig. 5D; see Fig. S2A,B). RT-PCR analyses showed that none of these lines had a reduction in endogenous UFO transcript levels when compared with wild type, indicating that the 35S::UFO-SRDX phenotypes are not caused by co-suppression of endogenous UFO expression (data not shown). Furthermore, none of plants expressing 35S::UFO-mSRDXshowed any lfy-like mutant phenotypes, indicating that the mSRDX fusion does not confer a dominant-negative effect. Rather, these control 35S::UFO-mSRDX plants showed a typical UFO overexpression phenotype. Together, these results demonstrate that the transformations seen in the 35S::UFO-SRDX plants reflected the dominant transcriptional repressor activity conferred by the fusion of the SRDX domain to UFO. The observed dominant-negative effect of the UFO-SRDX protein on LFY activity supports the idea of a direct physical interaction between LFY and UFO, and suggests that UFO acts as a LFY transcriptional co-factor.

The two most severe 35S::UFO-SRDX lines showed more drastic transformations than the strongest lfy alleles characterized to date in terms of floral organ identity defects. These two severe lines produced lateral structures consisting of only reiterating leaf-like organs with stipules arranged in a spiral pattern (Fig. 5G, see Fig. S2C in the supplementary material). This severe phenotype was maintained throughout the inflorescence, and was not ameliorated over time. This indicates that the specification of floral organ identity was almost completely repressed in those lines. This phenotype resembled that found in ap1 lfy double mutants(Bowman et al., 1993; Huala and Sussex, 1992; Weigel et al., 1992), and suggests the possibility that 35S::UFO-SRDX not only reduces the transcriptional activity of LFY, but also potentially of AP1.

To examine whether these severe phenotypic defects are correlated with transgene expression levels, we quantified transcript levels of the UFO-SRDX transgene using quantitative RT-PCR. Indeed, the two most severe lines harboring lateral structures with only leafy organs(Fig. 5G) showed the highest levels of UFO-SRDX transcripts (see Fig. S2D in the supplementary material). In addition, the two strong lines that showed lfy mutant phenotypes(Fig. 5B-E; see Fig. S2D in the supplementary material) produced a few progeny that segregated in a 1:2 ratio of plants appearing identical to the most severe lines of the primary transformants when compared with parental lines, which was consistent with homozygous and hemizygous segregants for the transgene. Therefore, expression studies of the transgene and progeny testing strongly suggest that all the observed phenotypic defects described above were due to a strong dominant repressive effect of the UFO-SRDX transgene.

Given that the F-box protein UFO associates with the AP3 promoter and functions as a transcriptional co-factor of LFY, it is probable that UFO mediates transcriptional regulation in a proteasome-dependent manner. To test if 26S proteasome activity is required for AP3 expression in vivo, we used an inducible form of LFY, 35S::LFY-GR(Wagner et al., 1999) and a proteasome mutant, halted root (hlr)(Ueda et al., 2004). This inducible form of LFY can activate AP3 transcription in seedlings upon DEX treatment only in the presence of UFO(Lamb et al., 2002). The proteasome subunit Rpt2a mutant hlr has been shown to have reduced proteasome activity during seedling development, and thus fails to degrade the substrate of the well characterized SCF complex SCF^TIR1(Ueda et al., 2004). Therefore, we introduced 35S::LFY-GR and 35S::UFO into a hlr mutant background and then tested for the requirement of proteasome activity for LFY and UFO to activate AP3 transcription in seedlings. 35S::LFY-GR; 35S::UFO; hlr-1 seedlings were grown without DEX for 7 days, treated with 1 μM DEX for 16 hours, and then harvested and the relative levels of AP3 expression assayed using qRT-PCR. We used 35S::LFY-GR; 35S::UFO seedlings as a positive control; DEX treatment resulted in the induction of AP3 expression as previously reported(Lamb et al., 2002)(Fig. 6A). However, AP3 expression was not induced in the hlr mutant background,even in the presence of both LFY and UFO, indicating that proteasome activity is required for AP3 transcription to ensue(Fig. 6A).

To independently confirm the requirement of proteasome activity for the induction of AP3 expression in seedlings, we also used epoxomicin, a potent inhibitor of the proteolytic activity of the 20S core component of the proteasome (Meng et al.,1999). We observed that epoxomicin treatment reduced the level of AP3 transcripts induced by DEX(Fig. 6B) up to 64% and 40 % in two independent biological replicates, respectively. These results indicate that proteasome activity is required for AP3 to be fully induced in seedlings by LFY and UFO. These observations support the idea that UFO acts to target proteins for 26S proteasome mediated degradation, and that such degradation is a requirement for AP3 transcription.

To test whether UFO can act to post-translationally modify LFY, we examined the ubiquitylation status of LFY protein. When LFY-FLAG protein was subjected to SDS-PAGE, we detected the fusion protein at the expected size, but also saw a faint smear of cross-reacting higher molecular weight proteins(Fig. 6C, lane 1), suggesting that LFY may be ubiquitylated in vivo. The abundance of these high molecular weight proteins is somewhat reduced in a ufo-2 mutant background,suggesting that UFO may act to modify post-translationally a subset of LFY proteins in vivo (Fig. 6C, lane 2; Fig. 6D). We further showed that the levels of high molecular weight FLAG-tagged proteins are increased in epoxomicin-treated tissue when compared with mock DMSO treatment(Fig. 6C), suggesting that, in vivo, LFY is subject to ubiquitin-dependent proteasome mediated degradation. To confirm that the high molecular weight protein smear corresponds to ubiquitylated forms of LFY, or LFY-associated proteins, the immunoprecipitated LFY-FLAG protein, isolated from the inflorescences of 35S::LFY-FLAGtransgenic plants, was examined for the presence of ubiquitylated species in both UFO and ufo-2 backgrounds(Fig. 6D). Polyubiquitylated species could be detected in immunoblots decorated with anti-ubiquitin antibody (Fig. 6D, upper panel). These observations indicate that LFY-FLAG, or tightly associated co-immunoprecipitated LFY-FLAG complex components, are polyubiquitylated in vivo. Furthermore, LFY-FLAG complex polyubiquitylation is partially abrogated in the ufo-2 mutant background(Fig. 6D). This suggests that,in vivo, LFY (or LFY complex) ubiquitylation is partly dependent on UFO activity, but probably also depends on the action of other F-box proteins.

Nearly 700 F-box proteins are predicted to be encoded in the Arabidopsis genome (Gagne et al.,2002; Kuroda et al.,2002). The striking expansion of this gene family probably reflects the recruitment of F-box proteins to regulate a variety of biological processes, implicating protein degradation as a prevalent developmental control mechanism in plants. For example, the Arabidopsis F-box protein TIR1 mediates hormonal signaling by acting as an auxin receptor; auxin binding to TIR1 promotes the activity of SCF^TIR1 in targeting Aux/IAA proteins for destruction(Dharmasiri et al., 2005; Kepinski and Leyser, 2005; Tan et al., 2007). Here, we describe a new role for a plant F-box protein, in functioning as a component of the transcriptional machinery that regulates floral homeotic gene expression.

The work presented here provides evidence that the F-box protein UFO functions as a transcriptional co-factor of LFY. We have shown that UFO physically interacts with LFY and that this interaction is necessary to recruit UFO to AP3 promoter elements. Furthermore, UFO-Myc tagged protein is localized to the nucleus (E.C., K. Geuten and V.F.I., unpublished). These results imply that the physical interaction between LFY and UFO occurs in the context of the AP3 promoter, and the abrogation of the association of UFO with the promoter in the absence of functional LFY strongly supports the idea. We also carried out transgenic studies in which a UFO-SRDX fusion construct was expressed ectopically, resulting in a lfy-like phenotype. These results indicate that converting UFO to a strong transcriptional repressor in vivo results in reduced LFY activity. Furthermore, we observed that LFY, or LFY-associated proteins, are ubiquitylated in vivo in a partially UFO-dependent manner, and that proteasome activity is required for AP3 to be fully induced. Based on these observations, it is likely that SCF^UFO stimulates LFY-induced transcription via promoter-associated proteasome-mediated degradation of LFY,or of a factor in the LFY transcriptional complex.

Recently, several yeast and mammalian F-box proteins have been shown to promote the degradation of target transcription factors, but, surprisingly,coordinately promote their transcriptional activity(Kim et al., 2003; Muratani et al., 2005; Perissi et al., 2004; von der Lehr et al., 2003). However, the mechanism by which this transcriptional control is effected is still a matter of debate (Kodadek et al.,2006; Nalley et al.,2006). These observations, though, have led to the proposition that the rapid turnover of active transcriptional activators by the ubiquitylation-proteasome pathway, allowing for the removal of `spent'activators, can explain how degradation of transcriptional activators can paradoxically stimulate target gene expression(Collins and Tansey, 2006; Kodadek et al., 2006; Lipford et al., 2005).

One possibility for how UFO functions to stimulate LFY transcriptional activity is through mediating polyubiquitylation of the LFY transcription factor (or tightly associated factors) at the AP3 promoter, thus stimulating its rapid turnover and concomitant activation of AP3expression. To test this idea, we examined ubiquitylation in vivo, by using a FLAG-tagged version of LFY overexpressed in transgenic Arabidopsisplants (Fig. 6D). The ubiquitylated species detected by this method do not appear to be exclusively dependent on UFO, suggesting that LFY complex ubiquitylation depends on multiple E3 ligases, which could potentially modulate other LFY transcriptional activities. As LFY is involved in regulating the expression of multiple floral homeotic genes (Busch et al., 1999; Lamb et al.,2002; Lohmann et al.,2001; Parcy et al.,1998) as well as several meristem identity genes(Saddic et al., 2006; Wagner et al., 1999), it is conceivable that LFY transcriptional activity is differentially regulated by distinct interactions with different F-box proteins. Furthermore, overall LFY protein levels appear to be unaffected by UFO, which may reflect the possibility that UFO specifically acts to modulate the stability of promoter bound LFY.

Additionally, SCF^UFO may act to ubiquitylate other factors that are part of the transcriptional machinery at the AP3 promoter; such factors would likely correspond to potential negative regulators of AP3 transcription (Samach et al.,1999). Several negative regulators of AP3, including SUPERMAN (SUP), EARLY BOLTING IN SHORT DAYS(EBS) and the transcriptional co-repressor LEUNIG (LUG)/SEUSS (SEU)complex have been identified (Gomez-Mena et al., 2001; Sakai et al.,1995; Liu and Meyerowitz,1995; Sridhar et al.,2004), but we have not been able to detect any interaction between UFO and those proteins (E.C. and V.F.I., unpublished; Z. Liu, personal communication), implying that they are not targets of SCF^UFOaction. Thus, an as yet unidentified factor may play this role.

Mutant phenotypes for several eudicot UFO homologs suggest they all participate in a similar process of regulating AP3 homolog gene expression, as they all disrupt petal and stamen development(Ingram, 1997; Taylor, 2001; Zhang, 2003; Ikeda et al., 2007). However,these mutations also result in ectopic flower production, a phenotype not seen in Arabidopsis ufo mutants. For example, Antirrhinum fimbriata mutants produce ectopic flowers in the axils of sepals(Ingram, 1997), similar to the phenotype produced by mutations in the Antirrhinum AP1 homolog(Huijser, 1992). In legumes, mutations in the Pisum and Lotus UFOhomologs produce indeterminate floral meristems resulting in extra floral organs (Taylor et al., 2001; Zhang et al., 2003), in a manner similar to that of mutations in the legume AP1 homologs(Benlloch et al., 2006; Taylor et al., 2002). One possibility is that UFO is not only involved in regulating LFY-dependent transcription of AP3, but may also control the activity of the AP1 transcription factor in Arabidopsis(Fig. 5)(Hepworth et al., 2006). Together, these observations suggest that UFO may have a widely conserved role in regulating both LFY and AP1 transcriptional activities, as reflected in its mutant phenotypes in different species.

In summary, our results demonstrate that the UFO F-box protein acts as a transcriptional co-factor of LFY at the AP3 promoter. These results also provide insights into how the activity of LFY, a master regulator of flowering, is tightly regulated to promote regional activation of downstream target floral homeotic genes, which in turn is necessary for specification and differentiation of individual floral organs.

We thank Dr Xiping Wang for kindly providing a 35S::UFO-Myctransgenic line and Dr Taku Demura for pH35SGEAR. We thank Dr Zhongchi Liu for sharing unpublished results and members of the Irish laboratory for helpful suggestions. This work was supported by NSF grant IOS-0516789 to V.F.I.