The Arabidopsis flowering-repressor gene FLOWERING LOCUS C (FLC) is a developmental switch used to trigger floral induction after extended growth in the cold, a process termed vernalization. In vernalized plants, FLC becomes transcriptionally silenced through a process that involves an epigenetic mechanism. We identified recessive mutations designated vernalization independence (vip) that confer cold-independent flowering and suppression of FLC. These mutations also lead to developmental pleiotropy, including specific defects in floral morphology, indicating that the associated genes also have functions unrelated to flowering time. We identified the VIP3 gene by positional cloning and found that it encodes a protein consisting almost exclusively of repeated Trp-Asp (WD) motifs, suggesting that VIP3 could act as a platform to assemble a protein complex. Constitutive transgenic expression of VIP3 in vernalized plants is insufficient to activate FLC, and thus VIP3 probably participates in the regulation of FLC as one component of a more extensive mechanism. Consistent with this, genetic analyses revealed that the VIP loci define a functional gene class including at least six additional members. We suggest that VIP3 and other members of this gene class could represent a previously unrecognized flowering mechanism.
GENETIC and molecular studies in Arabidopsis have shown that flowering results from the action of several interdependent regulatory mechanisms or pathways, each mediating the effect of separate endogenous or environmental influences (Koornneefet al. 1998b; Simpson and Dean 2002). The Arabidopsis FLOWERING LOCUS C (FLC) MADS-domain protein is a key flowering-time regulator that integrates the activity of the so-called autonomous pathway and the FRIGIDA (FRI) gene, which influence flowering independently of day length, and the “vernalization pathway,” which moderates the promotive effects of extended growth in the cold (Simpson and Dean 2002). FLC in turn represses flowering through negative regulation of the SOC1 (AGL20) and FT genes (Leeet al. 2000; Samachet al. 2000; Michaels and Amasino 2001). In this manner, FLC acts antagonistically with CONSTANS (CO), which moderates the promotive effects of inductive photoperiods (Hepworthet al. 2002).
At least six genes, designated FY, FCA, FPA, FLD, LD, and FVE, have been proposed to participate in the autonomous pathway. These genes collectively act to promote flowering through repression of FLC (Michaels and Amasino 1999; Rouseet al. 2002). In contrast, the activity of FRI represses flowering by positively regulating FLC expression (Michaels and Amasino 1999; Sheldonet al. 1999). The antagonistic relationship between FRI and the autonomous pathway has not been clearly defined, but the observation that FRI activates FLC and represses flowering even in the presence of autonomous pathway activity (i.e., in wild-type plants that carry a functional FRI allele) suggests that FRI could function to limit the activity of one or more autonomous pathway components.
The early flowering conferred by loss of FLC is apparently completely epistatic to the repressive effects of FRI or mutation in at least FCA, FVE, and LD, suggesting that the only function of FRI or these autonomous pathway genes, with respect to flowering, is to regulate FLC (Michaels and Amasino 2001). In noninductive photoperiods, where the promotive influence of the photoperiodic pathway involving CO is minimized, plants lacking only FLC activity flower earlier than plants lacking only FRI activity. This suggests that FLC represses flowering to some extent even in the absence of FRI and suggests that FLC expression is also promoted through mechanism(s) not involving FRI (Michaels and Amasino 2001).
Although FRI and the known autonomous pathway genes have now been identified at the molecular level, the nature of the corresponding regulatory mechanisms is undefined. Both transcriptional and post-transcriptional events are likely involved, as at least one autonomous pathway gene, LD, encodes a homeodomain-transcription factor-like protein (Aukerman and Amasino 1996), and two others, FCA and FPA, encode proteins with structural features suggestive of RNA binding (Macknightet al. 1997; Schomburget al. 2001). Repression of FLC by the autonomous pathway is presumed to occur at least partly at the level of FLC transcriptional activity, because FLC represses flowering when expressed from the constitutive CaMV 35S promoter, even in a genotype lacking strong FRI activity (i.e., where the autonomous pathway is active; Michaels and Amasino 1999).
Vernalization is an epigenetic effect (Wellensiek 1962, 1964) associated with suppression of the FLC gene (Michaels and Amasino 1999). Vernalization likely targets FLC regulatory mechanisms that do not directly involve FRI or the autonomous pathway, because cold promotes flowering even in genotypes carrying combined loss-of-function mutations in autonomous pathway genes and in FRI (Koornneefet al. 1998b). Vernalization-associated FLC repression is mediated by at least two genes, VRN1 and VRN2 (Simpson and Dean 2002). VRN2 may act to initiate or maintain a relatively silent state of FLC chromatin, as a vrn2 mutant shows increased accessibility of the FLC locus to DNAse I (Gendallet al. 2001). Also, the VRN2 protein resembles the Drosophila Polycomb-group transcriptional regulator Su(z)12, which potentially acts at the level of chromatin structure (Gendallet al. 2001). Transgenic overexpression of VRN1, which encodes a nuclear-localized DNA-binding protein, results in early flowering that is associated with increased expression of FT, but not decreased expression of FLC (Levyet al. 2002). This suggests both that VRN1 requires vernalization-specific auxiliary factors to target FLC and that VRN1 may also regulate flowering through an FLC-independent mechanism. In addition, VRN1 overexpression disrupts seemingly unrelated developmental processes, indicating that its role may be wider than that of regulating flowering. Neither VRN1 nor VRN2 seems to be regulated in a vernalization-associated manner, also revealing that specificity is derived from relationships with cold-regulated factors (Levyet al. 2002; C. Dean, personal communication). Genetic analyses suggest that VRN1 and VRN2 could represent members of a larger group of genes with similar function (Chandleret al. 1996).
Most genotypes of Arabidopsis commonly used in laboratory studies carry a dysfunctional fri allele (Johansonet al. 2000) and flower soon after germination. Thus, genetic pathways of floral repression in this reference plant have not been extensively characterized. We previously reported the identification and cloning of the VERNALIZATION INDEPENDENCE 4 (VIP4) gene, which acts as a flowering repressor by promoting expression of FLC (Zhang and van Nocker 2002). We report here the cloning and characterization of a novel positive regulator of FLC designated VIP3. On the basis of genetic and molecular evidence presented here, we propose that VIP3 and, possibly, VIP4 as well could promote FLC expression through a previously unrecognized mechanism and that these genes are members of a functionally similar gene class in Arabidopsis.
MATERIALS AND METHODS
Growth conditions: Arabidopsis seeds were either planted directly into artificial soil mix or surface sterilized and germinated under sterile conditions as described previously (van Nockeret al. 2000). Standard growth conditions were 60–100 μmol m–2 sec–1 of fluorescent lighting in a 16 hr light/8 hr dark photoperiod at 22° and ∼50% relative humidity. Shortday growth conditions were identical with standard growth conditions but utilized 8 hr light/16 hr dark photoperiods. For a vernalizing cold treatment, seeds on germination medium were first placed at 4° under 20–50 μmol m–2 sec–1 of fluorescent lighting in an 8 hr light/16 hr dark photoperiod for 30 or 70 days. For growth under far-red light-enriched conditions, lighting was supplied entirely by household incandescent bulbs.
Strains and genetic techniques: Introgression line Col:FRISF2 consists of the dominant FRI locus from ecotype San Feliu-2 (FRISF2) introgressed into the Columbia (Col) ecotype (Leeet al. 1994). Introgression line Ler:FRISF2:FLCSF2 consists of FRISF2 and the semidominant FLC locus from ecotype San Feliu-2 (FLCSF2) both introgressed into the Landsberg erecta (Ler) genetic background through seven successive backcrosses and made homozygous through self-pollination (Leeet al. 1994). The ld-1 mutant (ecotype Col-1) was obtained from the Arabidopsis Biological Resource Center (ABRC) at The Ohio State University (Columbus, Ohio). The null flc-3 mutant was a generous gift from S. Michaels and R. Amasino (University of Wisconsin). The Ler::vip3 introgression line was created by carrying out four successive outcrosses to wild-type Ler and selecting for plants that carried Ler alleles of FRI and FLC after the second outcross. The Escherichia coli strain harboring bacterial artificial chromosome (BAC) F27B13 was obtained from the ABRC.
Polymerase chain reaction (PCR)-based molecular markers were utilized to discriminate between wild-type and mutant alleles of VIP3, FRI, and LD. A marker for presence of the wild-type VIP3 allele was designed to amplify, from the wild-type allele, a region spanning the site corresponding to the vip3 mutation [primers: F27B13.7F2 (5′-TTGCAGGTGGAA GTAGTGCCTC-3′) and F27B13.7 R2 (5′-TGTCATCAGAGA CACTAGCAAGTCG-3′)]. To determine presence of the vip3 allele, a marker was designed to amplify the right junction of the genomic insertion [primers: F27B13.7F2 and T16L4F (5′-GCCACTGCCGCCAGTTTTATCAAG-3′)]. A marker for discrimination between the FRISF2 and friCol alleles was based on a 16-bp length polymorphism within the FRI promoter as described by Johanson et al. (2000) and employed primers FRI16F (5′-TGGTGTTCCTTCAAACTTTAGG-3′) and FRI16R (5′-GCTCAATCAGTCATTGCACTC-3′). A marker for discrimination between the LD and ld-1 alleles was based on the ld-1 mutation, which is localized within the LD transcribed region, and was generously provided by S. Michaels (University of Wisconsin).
A vip3/fri double mutant was created by crossing vip3 with wild-type Col (carrying the strong, loss-of-function friCol allele). A VIP3/vip3, friCol/friCol plant was identified in the respective F2 population and allowed to self-pollinate, and double mutants were analyzed in the corresponding progeny. A vip3/fri/ld triple mutant was created by crossing vip3 with a plant carrying the strong ld-1 allele in the Col background. F2 progeny from this cross that were friCol/friCol, VIP3/vip3, and LD/ld were allowed to self-pollinate, and triple mutants were analyzed in the corresponding progeny. A vip3/vip4 double mutant was created by crossing vip3 with vip4-1 and backcrossing the corresponding F1 plant with a vip3 mutant. A vip3/vip3 VIP4-1/vip4-1 plant was allowed to self-pollinate, and double mutants were analyzed in the corresponding progeny.
Mutagenesis and cloning of VIP3: For mutagenesis of introgression line Col:FRISF2, seeds were exposed to ∼165 Gy of fastneutron radiation using the fast-neutron beam at the Michigan State University Cyclotron Laboratory or incubated with 0.15% ethyl methanesulfonate (EMS) overnight and subsequently rinsed extensively with distilled water. Seeds were then subjected to a vernalizing cold treatment and planted in soil, and plants were allowed to self-pollinate. M2 seed was collected in pools each representing ∼1000 M1 individuals. Approximately 5000 plants from each of 24 fast neutron-derived M2 families and 20 EMS-derived M2 families were screened. T-DNA mutagenesis and screening were described previously (Zhang and van Nocker 2002). The vip3 mutant was backcrossed three times in succession to wild-type plants before phenotypic analysis. Phenotypic analysis of other vip mutants was performed with progeny derived from a backcross of M2 plants to wild type. For genetic complementation analysis, mutants were grown at 18°, under which conditions all mutants were fertile.
Positional cloning of the VIP3 gene utilized F2 progeny of a single F1 individual derived from a cross between vip3 and introgression line Ler:FRISF2:FLCSF2. Bulked-segregant analysis was performed with 24 F2 individuals and molecular markers described by Lukowitz et al. (2000). Fine mapping was done entirely using molecular markers based on small insertion-deletion polymorphisms as characterized and cataloged by Cereon (http://www.arabidopsis.org/cereon/index.html; courtesy of S. Rounsley) and noted in Figure 3.
Molecular techniques: For use as probes in DNA gel blotting, BAC DNA was purified from 250-ml bacterial cultures using a commercially available kit (QIAGEN, Valencia, CA). For PCR purposes, DNA was prepared from plant tissues using the CTAB-based method described by Lukowitz et al. (2000). DNA and RNA gel blotting was performed essentially as previously described (Zhang and van Nocker 2002). For detection of VIP3 RNA, the probe was a DNA corresponding to the entire VIP3 coding region, amplified from flower-derived cDNA using primers F27B13.7FBam (5′-AAAGGATCCATGA AACTCGCAGGTCTGAAATCG-3′) and F27B13.7RBam2 (5′-AAAGGATCCGAATTGTTCATGAGTAATCATAGAGC-3′). For detection of FLC RNA, the probe was as described by Zhang and van Nocker (2002). For molecular complementation of the vip3 mutation, an ∼6.4-kb BamHI fragment derived from BAC F27B13 was ligated into the BamHI site of vector pPZP:BAR (Zhang and van Nocker 2002) and introduced into wild-type plants through floral dipping. This DNA contained the entire predicted transcriptional units At4g29830 (VIP3) and At4g29820, as well as part of transcriptional units At4g29840 and At4g29810. Several independent T1 lines were crossed to the vip3 mutant, herbicide-resistant progeny from these crosses were allowed to self-pollinate, and the resulting progeny were again subjected to herbicide selection. All vip3-like progeny were found to be herbicide sensitive. PCR analysis of several wild-type-like progeny indicated homozygosity for the vip3 allele and presence of the transgene (data not shown).
For overexpression or antisense expression of VIP3 in transgenic plants, the VIP3 coding and 3′ untranslated region was amplified from genomic DNA using primers F27B13.7FBam and F27B13.7RBam (5′-AAAGGATCCAATGCCATCCCTGA CATGGCTTGC-3′). These primers incorporate a BamHI restriction endonuclease site into both termini. The PCR products were ligated into vector pGEM-T (Promega, Madison, WI), the resulting construction was subjected to digestion with BamHI, and the fragment containing the VIP3 coding and 3′ region was ligated into the BamHI site of vector pPZP:BAR:35S (Zhang and van Nocker 2002). Ligation products were obtained that contained the VIP3 fragment in both forward (sense) and reversed (antisense) orientation. These were introduced into Agrobacterium strain ABI, and the resulting strains were used to infect wild-type plants through floral dipping.
For immunological studies, recombinant, hexahistidinetagged, full-length VIP3 protein was expressed in E. coli and purified using Ni2+-affinity chromatography and a commercially available kit (Novagen; His-Bind) according to the manufacturer's instructions. This purified protein was used to generate anti-VIP3 sera in rabbits. For immunoblotting, plant protein extracts were prepared by grinding tissues under liquid nitrogen, adding the frozen tissue powder to sample buffer containing 4% SDS (Laemmli 1970), incubating at 100° in a boiling water bath, and clarifying by centrifugation for 5 min at 12,000 × g. Immunoblotting was done as described by Harlow and Lane (1988), using PVDF membranes (Bio-Rad, Richmond, CA) blocked with Tween-20 in phosphate-buffered saline and alkaline-phosphatase-labeled goat anti-rabbit IgGs (Bio-Rad).
Sequence analysis: WD motifs in the VIP3 protein were identified using the Protein Sequence Analysis server (http://bmerc-www.bu.edu/psa/index.html) at the BioMolecular Engineering Research Center, Boston University. Other sequence analyses were performed using BLAST on web servers maintained by the National Center for Biotechnology Information (NCBI; http://www.ncbi.nlm.nih.gov) or The Arabidopsis Information Resource (http://www.arabidopsis.org) and programs of the Genetics Computer Group (Madison, WI).
Identification and genetic analysis of the VIP3 locus: To identify new floral repressors important for regulation of FLC, we extensively mutagenized the late-flowering, vernalization-responsive (winter-annual) genetic background Col:FRISF2 (hereafter referred to as wild-type) using fast neutrons, T-DNA, and EMS and identified individuals that flowered very early independently of a vernalizing cold exposure. Several early flowering individuals were recovered that displayed defects in floral morphology similar to that described for the previously identified vip4 mutants and that were found to be nonallelic with vip4 (see below). One of these mutants, designated vip3, was selected for further study (Figure 1A). To determine if the defect conferred by the vip3 mutation could be in a flowering-repressive mechanism involving FLC, we evaluated FLC expression by RNA gel blotting of seedling RNAs. In contrast to wild-type plants, FLC RNA expression was not detectable in the vip3 mutant, even with phosphorimaging and extended exposures (Figure 1B).
In addition to the defect in flowering time (see below), vip3 plants exhibited several other defects in growth and development when grown under standard conditions. Specifically, rosette leaves of vip3 plants were smaller than those of wild-type plants, and overall plant size was reduced (Figure 1A, c). In addition, flowers of vip3 plants exhibited abnormalities of organs in the outer three whorls (Figure 1A, h). Sepals typically had irregular, translucent margins, and lateral sepals were always reduced in size. Petals were also reduced in size and were often variable in number; whereas wild-type flowers have four petals, vip3 flowers had up to six. Stamens were typically decreased in number from six to four or five. Organs in the outer three whorls were also often replaced by filamentous structures. The gynoecium was morphologically normal, but slightly reduced in size. Flowers were typically male sterile, and self-pollination was rare. When plants were grown at a lower temperature (18°), these floral defects were attenuated, and plants were typically fertile (data not shown). The vip3 mutation conferred essentially identical pleiotropy when introgressed into the commonly used Ler ecotype (Figure 1A, d and i).
Similar to wild-type plants, heterozygous VIP3/vip3 plants resulting from a backcross between vip3 and wild-type plants flowered very late under photoperiodically inductive (long-day) conditions in the absence of cold. In addition, none of the phenotypic defects described above for the vip3 mutant were apparent in VIP3/vip3 plants (data not shown). These observations indicate that the vip3 mutation is effectively recessive. In the progeny of a VIP3/vip3 plant, mutant individuals were found with a frequency expected for Mendelian segregation of a single recessive locus (data not shown). In addition, analysis of the progeny of reciprocal crosses between a wild-type plant and a VIP3/vip3 plant indicated that the vip3 mutant allele was transmitted through both male and female gametes with a frequency similar to that of the wild-type allele (data not shown).
Flowering response of the vip3 mutant: To better define the position of VIP3 with respect to flowering pathways involving FLC, we evaluated the effects of photoperiod, extended cold, and light quality on the flowering response of the vip3 mutant relative to that of wild-type plants and plants carrying a null mutation in FLC (Figure 2, A and B). When grown in the absence of cold, wild-type plants produced ∼67 leaves under long-day conditions but did not flower under short-day (noninductive) conditions, even after producing >100 rosette leaves. We found that when given an extended (70 days) cold treatment, the photoperiodic response of wild-type plants was effectively eliminated, and plants flowered after producing ∼10 or 11 leaves irrespective of photoperiod (Figure 2A). This is consistent with current models of flowering where vernalization, acting through suppression of FLC, can bypass the lack of floral promotion from inactivity of the photoperiodic pathway (Simpson and Dean 2002). Both the flc null mutant and the vip3 mutant exhibited strong photoperiodic responses when grown in the absence of cold and still showed slight photoperiodic responses even after being given a 70-day cold treatment (Figure 2A).
When grown under photoperiodically inductive conditions, the vernalization response of winter-annual types of Arabidopsis appears saturated after ∼40 days of cold; i.e., longer cold periods have no further flowering-promotive effects (Lee and Amasino 1995). This probably reflects the ability of the photoperiodic pathway to partially bypass repression of flowering by FLC, because much longer (e.g., ∼70 days) cold periods are necessary to saturate the vernalization response when flowering is evaluated in noninductive photoperiods (Lee and Amasino 1995). Under short-day conditions, we found that vip3 plants grown in the absence of cold flowered appreciably later (17.8 ± 0.9 leaves) than wild-type plants given a 70-day cold treatment (10.7 ± 0.9 leaves; Figure 2). A 70-day cold treatment reduced flowering time of vip3 plants to that of vernalized wild-type plants (Figure 2). The observation that cold effectively promotes flowering in vip3 plants suggests that, if vip3 is a null mutation, vernalization could promote flowering at least partially independently of VIP3.
Wild-type plants flowered much earlier (44.7 ± 10.2 leaves) when grown under far-red-enriched light supplied by incandescent bulbs than when grown under the fluorescent lights used in our standard growth conditions (75.2 ± 8.2 leaves; Figure 2B), and this is consistent with previous observations that far-red-enriched light promotes flowering in genotypes that strongly express FLC (Bagnall 1993; Lee and Amasino 1995). We found that flc mutant plants flowered earlier when grown under incandescent lighting (9.6 ± 0.9 leaves) relative to fluorescent lighting (12.2 ± 1.1 leaves; Figure 2B), suggesting that promotion of flowering associated with light quality is mediated at least partly independently of FLC. The vip3 mutant plants also flowered earlier under these conditions (9.2 ± 0.6 leaves) than under standard growth conditions (11.1 ± 0.8 leaves), and the net reduction in flowering time was similar to that observed for the flc mutant. This observation suggests that mechanisms that promote flowering in response to light quality are intact in the vip3 mutant.
Under all conditions evaluated, vip3 mutant plants flowered earlier than flc mutants (Figure 2, A and B). This was most apparent when plants were grown in short-day photoperiods, irrespective of cold. This observation indicates that VIP3 has an additional flowering-repressive role that is mediated outside of its positive regulation of FLC.
Interactions with FRI and LD: One of several potential positions of the VIP3 gene within the regulatory hierarchy of flowering is as a negative regulator of the activity of the autonomous pathway, a function that has been proposed for FRI (above). If this were the case, then loss of VIP3 function would not be expected to suppress the late flowering associated with loss of autonomous pathway activity. To test this, we evaluated the epistatic interactions between VIP3 and the autonomous pathway gene LD. We introduced the vip3 mutation into the Col::ld-1 genetic background, which carries strong loss-of-function alleles in both FRI and LD. Col::ld-1 plants behave as winter annuals, because loss of LD activity leads to derepression of FLC; these plants were otherwise phenotypically indistinguishable from wild-type plants (data not shown). We found that vip3/fri/ld triple-mutant plants were phenotypically similar to vip3 plants, exhibiting aberrant floral morphology and reduced plant size (data not shown). Under long-day conditions, there was no apparent difference in flowering time between vip3 plants and the vip3/fri/ld triple mutant (Figure 2C). Under short-day conditions, however, the vip3/fri/ld triple mutant flowered notably later (26.7 ± 2.9 leaves) than vip3 plants (20.8 ± 3.8 leaves; Figure 2C). That the vip3 phenotype was predominantly epistatic to the late-flowering ld phenotype indicates that VIP3 is unlikely to function as an upstream regulator of the autonomous pathway and that it has a function that is distinct from that of FRI. The observation that this epistasis was incomplete also suggests that VIP3 could function in a pathway that is distinct from the autonomous pathway mechanism involving LD. A caveat to this analysis is that the incomplete epistasis observed could potentially result from weak function of VIP3,ifthe vip3 mutation were not null.
To determine if FRI has any flowering-repressive effect in a vip3 genetic background, we evaluated the effect of the strong loss-of-function friCol allele on flowering time of vip3 plants. When grown under long-day photoperiods, there was no significant difference in flowering time between vip3 single mutants and vip3/fri double mutants (Figure 2C). Under short-day photoperiods, vip3/fri double mutants flowered marginally earlier (16.7 ± 2.2 leaves) than vip3 single mutants (20.8 ± 3.8 leaves; Figure 2C). Thus, with respect to flowering time, the effect of the vip3 mutation was strongly epistatic to the effect of FRI. Mutants lacking both VIP3 and FRI were otherwise phenotypically similar to vip3 single mutants, exhibiting aberrant floral morphology and reduced plant size (data not shown).
Positional cloning of the VIP3 gene: Through genetic mapping, we localized the vip3 mutation to an ∼2.4-Mb region of the lower arm of chromosome IV, represented by three overlapping BAC clones (Figure 3A). Subsequently, we analyzed genomic DNA from vip3 and wild-type plants by gel blotting using these three BACs as probes. This approach resulted in the indication of an insertion within the predicted coding region of a transcriptional unit designated At4g29830 by the Arabidopsis Genome Initiative (AGI; Figure 3A). Further analysis using inverse PCR and sequencing indicated that the insertion was associated with the translocation of up to ∼320 kb, a possibly contiguous sequence from a proximal region of chromosome IV (data not shown). We found that, in the vip3 mutant, RNAs hybridizing with an At4g29830 probe accumulated to detectable levels, but were shorter than RNAs seen in wild-type plants, suggesting that the insertion in the vip3 mutant resulted in a truncation of the At4g29830 gene (Figure 1B). DNA and RNA gel-blot analyses of the adjacent genes, At4g29820 and At4g29840, in the vip3 mutant revealed that both genes were intact and expressed to levels similar to those seen in wild-type plants (data not shown). Using immunoblotting and antisera generated against recombinant At4g29830 protein, we observed a highly reactive protein species in wild-type plant extract approximating the predicted size of the At4g29830 protein (Figure 1B). This protein was not detectable in equivalent extracts from vip3 plants (Figure 1B). Also, there were no immunoreactive protein species of larger or smaller molecular mass unique to vip3 extracts (Figure 1B and data not shown), suggesting that any aberrant protein produced from the At4g29830 gene in vip3 plants is unstable. A query of Arabidopsis expressed sequence tag databases resulted in the identification of six independent cDNAs corresponding to the At4g29830 gene. These cDNAs collectively defined a transcribed region and intron/exon structure that is consistent with that predicted by the AGI and with the size of At4g29830 RNAs as determined by gel blotting (Figure 3A). RNA gel-blot analysis and immunoblot analysis indicated that At4g29830 RNA and protein are expressed throughout the plant (data not shown).
To determine if disruption of At4g29830 was the lesion causing the vip3 phenotype, we performed molecular complementation in transgenic plants using an ∼6.4-kb DNA containing the entire At4g29830 transcriptional unit. Because the vip3 mutant was predominately male sterile when grown under standard conditions and was therefore incompatible with the standard floral-dip method of plant transformation (Clough and Bent 1998), we first introduced the transgene into wild-type plants and then introduced the transgene into the vip3 mutant through crossing. Several independent lines were generated that were homozygous for the translocation mutation and contained at least one copy of the VIP3 transgene. These plants were phenotypically indistinguishable from wild-type plants, flowering extremely late in the absence of cold and producing morphologically normal flowers (Figure 1A, f and data not shown). On the basis of these data and the observation that antisense expression of At4g29830 resulted in a phenocopy of the vip phenotype (below), we concluded that At4g29830 is the VIP3 gene.
On the basis of annotation provided by the AGI, VIP3 encodes a 321-amino-acid protein that is composed almost entirely of seven repeats of a motif designated the Trp-Asp (WD) motif (also known as the WD-40 repeat; Neeret al. 1994; Smithet al. 1999; Figure 3B). No closely related genes exist in the Arabidopsis genome, and the predicted VIP3 protein does not show extensive sequence homology with any protein cataloged in current protein databases.
VIP3 expression: Genetic epistasis analysis (above) indicated that VIP3 could function downstream from the strong FLC regulators FRI and LD. To determine if the promotive activity of FRI on FLC expression might be mediated through activation of FLC by VIP3, we compared VIP3 RNA and protein abundance in wild-type plants with that in the Col ecotype (lacking strong FRI activity). Likewise, to determine if VIP3 might mediate the derepression of the FLC gene due to loss of LD activity, we evaluated VIP3 RNA and protein levels in the ld-1 mutant. Although loss of FRI or LD activity resulted in obvious differences in FLC RNA expression, no effect on VIP3 expression was apparent (Figure 1B). To determine if the repressive effect of cold on FLC expression might be mediated through loss of VIP3 activity, we evaluated VIP3 RNA and protein levels in vernalized and nonvernalized wild-type plants. In both situations, VIP3 was expressed to similar levels (Figure 1B). These findings suggest that modulation of VIP3 RNA or protein levels is unlikely to be involved in the regulation of FLC by FRI, the autonomous pathway, or cold.
Constitutive and antisense expression of VIP3 in transgenic plants: To study the potential effects of manipulated expression of VIP3 on growth and development, we engineered transgenic plants in which the wild-type genomic copy of VIP3 was expressed in either sense or antisense orientation, under control of the constitutive CaMV 35S promoter. For both the sense (35S:VIP3) and the antisense (VIP3-AS) strategies, at least 150 transgenic plants were recovered. For the VIP3-AS strategy, self-pollinated offspring from infiltrated plants (designated T1 plants) were grown without a vernalizing cold treatment. Approximately one-half of VIP3-AS plants surviving selection flowered very early, with as few as 5 rosette leaves (Figure 1A, e). In contrast, nonvernalized wild-type plants grown under similar conditions produced at least 60 rosette leaves without flowering (Figure 1A, a). In addition, the typical early flowering VIP3-AS plants were smaller than wild-type plants and produced morphologically abnormal flowers similar to those seen on vip3 plants (Figure 1A, e and j). For the 35S:VIP3 strategy, a population of T1 plants was grown without a vernalizing cold treatment. Similar to nonvernalized wild-type plants, the great majority of these plants flowered extremely late or did not flower during the course of this experiment. Analysis of VIP3 RNA and protein levels in leaf tissues of eight of these late-flowering plants indicated that all expressed the 35S:VIP3 transgene to high levels relative to wild-type nontransgenic plants (data not shown). The few very early flowering plants observed in this population possibly resulted from transgene-associated suppression of the endogenous VIP3 gene, as VIP3 protein was not detectable in leaf tissues of these plants (data not shown).
To determine if constitutive expression of VIP3 could overcome the repressive effect of cold on FLC expression, we analyzed another population of 35S:VIP3 T1 plants grown after being subjected to a vernalizing cold treatment. In this population of ∼250 individuals, all plants flowered very early, and there was no large variation in flowering time among the plants (data not shown). VIP3 was expressed to high levels in several of these plants, as determined by RNA gel-blot and immunoblot analyses of leaf tissues (Figure 1C). Even in these VIP3-expressing plants, FLC expression was not detectable (Figure 1C). These findings indicate that VIP3 is probably insufficient to activate FLC in vernalized plants.
VIP3 is a member of a class of functionally related genes in Arabidopsis: The phenotype of the vip3 mutant was similar to that of plants with mutations in the previously identified flowering-time gene VIP4 (Zhang and van Nocker 2002). For example, vip4 mutants do not express detectable levels of FLC, flower earlier than an FLC null mutant, and exhibit defects in floral morphology in whorls 1–3 (Zhang and van Nocker 2002). The observation that several vip3/vip4-like mutations identified through our genetic screens were nonallelic to either VIP3 or VIP4 (see below) indicated that a class of gene with similar roles in flowering timing and floral development could exist in Arabidopsis. To attempt to define the extent of this potential gene class, we mapped all of the seven vip3/4-like mutations recovered in our screens to a limited region of the genome, and those plants harboring mutations on the same chromosome were subjected to genetic complementation analysis (Tables 1 and 2). In addition to VIP3 and VIP4, five new loci were identified, which we tentatively designated VIP1, VIP2, VIP5, VIP6, and VIP7 (Tables 1 and 2; Figure 4). All of the mutants exhibited early flowering, reduced plant size, and floral defects in whorls 1–3 similar to those described for vip3 above. In all cases these phenotypes cosegregated with early flowering in the small mapping populations and with a frequency expected for Mendelian segregation of a single, recessive locus (Figure 4 and data not shown). When evaluated under short-day photoperiods in the absence of cold, most of these mutants flowered as early as, or earlier than, an flc null mutant (Figure 4E). The exception was vip2, originating in an EMS-mutagenized population, which flowered slightly later (Figure 4E). Analysis of FLC RNA levels in the mutants revealed that in all cases except for vip2, this early flowering was associated with loss of detectable FLC expression (Figure 4, B and C and data not shown). For the vip2 mutant, FLC expression was detectable using phosphorimaging and extended exposures (data not shown). None of the mutations mapped to previously described flowering-time genes, with the exception of VIP2 and VIP5, which are located within a broad region described for EARLY FLOWERING IN SHORT DAYS (EFS; Soppeet al. 1999).
To help clarify the potential relationships among these genes, we analyzed VIP3 and VIP4 mRNA levels and VIP3 protein abundance in all of the vip mutant backgrounds. In the vip1, vip2, vip5, vip6-3, and vip7 backgrounds, VIP3 and VIP4 RNA levels and VIP3 protein abundance were similar to that seen in wild-type plants (Figure 4, B–D and data not shown). In addition, VIP3 RNA and protein was expressed to apparent wild-type levels in plants carrying the strong vip4-3 mutation, and VIP4 RNA was expressed to apparent wild-type levels in the vip3 mutant (Figure 4, C and D). These findings suggest that, if these genes function in a pathway involving VIP3 or VIP4, they probably do not act as direct regulators of VIP3 or VIP4.
Current models of flowering propose that FLC is regulated through several mechanisms, including the autonomous pathway, FRI, and vernalization (Simpson and Dean 2002). In addition, both genetic and molecular studies indicate that FLC is weakly regulated by genes operating outside of these pathways, as a result of poorly defined “feedback” mechanisms (Koornneefet al. 1998a; Rouseet al. 2002). VIP3 encodes a strong activator of FLC, as FLC RNA expression appears to be abolished in the vip3 mutant. Importantly, FLC expression is apparently ubiquitous in wild-type plants and does not appear to be developmentally regulated (Sheldonet al. 1999), indicating that VIP3 acts in a more direct manner. VIP3 is also expressed throughout the plant, similar to FLC (Sheldonet al. 1999), and this is consistent with a role for VIP3 as a direct activator of FLC.
As an activator of FLC, VIP3 functions in a manner distinctly different from FRI, which has been proposed to limit that activity of the autonomous pathway (Zhang and van Nocker 2002). This was evident by the epistasis analysis employing a strong mutation in the autonomous pathway gene LD. The derepression of FLC and late flowering conferred by the ld mutation is effectively epistatic to loss of FLC activation and early flowering conferred by the strong loss-of-function friCol allele. This effect was largely suppressed when the vip3 mutation was introduced into the ld/friCol background. This indicates that VIP3 probably does not act to limit the activity of LD. However, this epistasis was incomplete, and the small inhibitory effect caused by loss of LD function in a vip3 background could suggest that vip3 functions at least partly independently of LD. In addition, we found that an active FRI allele marginally delays flowering in a vip3 background, suggesting that VIP3 might have FRI-independent functions. These conclusions are dependent on the vip3 mutation creating a total loss of function of the VIP3 gene. Although full-length VIP3 RNA or protein was not detectable in the vip3 mutant in our experiments, it is possible that VIP3 is still expressed at a very low level.
The lack of effect of disruption of FRI or LD on VIP3 RNA or protein expression suggests that modulation of VIP3 expression is unlikely to be involved in the regulation of FLC by FRI or LD. Likewise, because VIP3 RNA and protein are expressed to similar levels in vernalized and nonvernalized plants, VIP3 is unlikely to be a direct regulator of the vernalization response. Possibly, regulation of VIP3 by these factors is carried out through modification of protein activity or within a small spatial domain. However, the simple observation that the developmental pleiotropy conferred by the vip3 mutation is not apparent in a fri null mutant, where the autonomous pathway is actively suppressing FLC, or in vernalized plants, also suggests that VIP3 retains activity under these circumstances. Thus, our data are most consistent with VIP3 acting outside of mechanisms involving FRI or LD.
A mechanism of FLC regulation in which VIP3 participates could be a major target of the vernalization pathway. This is suggested by the observation that both vip3 and vernalization affect flowering predominately through FLC, but also through FLC-independent mechanisms. Our findings that a long cold treatment slightly accelerates flowering of the vip3 mutant and that vip3 plants flower slightly later than vernalized wild-type plants could indicate that vernalization is mediated at least partly outside of VIP3 activity. However, a slight vernalization response could be mediated by weak VIP3 activity in vip3 plants. These possibilities can be resolved only through the identification and analysis of an unambiguous vip3 null mutation.
VIP3 encodes a protein containing WD motifs. The WD motif is found in a large variety of proteins that do not share any obvious function (Neeret al. 1994). The crystal structure of the well-known WD-repeat protein Gβ shows that each of the seven WD motif units takes the form of four antiparallel β-strands, with the seven repeated WD motifs forming a symmetrical structure termed a β-propeller (Smithet al. 1999). A distinctive feature of the VIP3 protein is a 13-residue extension within the region predicted for strand D of repeat IV. The analogous region of Gβ takes the form of a loop comprising the exterior surface of the propeller structure, and additional amino acids at this position may comprise an independently folded domain that would protrude from the structure. VIP3 appears to lack extensive amino- or carboxyl-terminal domains outside of the potential β-propeller, suggesting that it could act exclusively in the context of a molecular scaffold.
We formerly identified the Arabidopsis VIP4 gene, an FLC activator that encodes a highly hydrophilic protein with similarity to the Leo1 protein from Saccharomyces cerevisiae and similar proteins from Drosophila and Caenorhabditis elegans (Zhang and van Nocker 2002). Leo1 is involved in the expression of a small subset of yeast genes, as a component of the Paf1 transcriptional regulator, which may represent a transcriptional endpoint of protein kinase C-mitogen-activated protein kinase signaling (Mueller and Jaehning 2002). On the basis of the phenotypic similarity between the vip3 and vip4 mutants, the observation that the VIP4 gene exhibits epistatic relationships with FRI and LD that are similar to that seen for VIP3, and our observations that the vip4-1 mutation does not obviously enhance the phenotype of vip3 plants (our unpublished results), it is likely that the VIP3 and VIP4 genes act in close functional proximity. The relationship between VIP3 and VIP4 at the molecular level is not known, but our results suggest that it does not involve modulation of RNA expression of either gene or modulation of VIP3 protein abundance.
In addition to VIP4, mutations at five other loci create phenotypes that are superficially indistinguishable from that of vip3. Although two of the VIP loci, VIP2 and VIP5, map roughly to the previously identified flowering-time gene EFS, the vip2 and vip5 mutants do not exhibit specific pleiotropic phenotypes described for efs mutants. For example, efs mutants show increased seed dormancy, decreased apical dominance, and normal development of more apical flowers (Soppeet al. 1999), and these phenotypes were not observed in the vip mutants. In addition, the specific floral defects seen in the vip mutants were not reported in efs mutants (Soppeet al. 1999). Thus the VIP loci probably define a previously unreported group of flowering repressors. In spite of the large numbers of mutagenized plants screened in this study, the screen does not appear to approach saturation, as five of the seven VIP loci are defined by only a single allele. Therefore this group could be extensive. Although the relationships among these genes remain largely undefined, evidence presented here indicates that they are unlikely to act to modify VIP4 RNA, VIP3 RNA, or protein levels. One possibility is that these genes define components of a protein complex, potentially analogous to the yeast Paf1 transcriptional complex. However, at least VIP3 does not exhibit strong homology with known Paf1 components or with any other yeast protein.
In addition to its early flowering phenotype, vip3, vip4, and the other vip mutants described here display similar defects in floral development. Because plants lacking FLC do not display floral defects, the role of these genes in floral development is mediated outside of their regulation of FLC. We formerly proposed (Zhang and van Nocker 2002) that VIP4 may have a floral function similar to that of a class of gene involved in repressing the expression of AGAMOUS (AG) and/or other floral homeotic genes outside of their typical spatial or temporal domains. We analyzed AG RNA abundance in vip3 fully developed flowers and found that it was elevated substantially (∼50%) over wild-type flowers (data not shown) but it remains unclear if this resulted from a direct role in AG expression or was merely an indirect effect of altered morphology of vip3 flowers.
We propose that the VIP gene class defines a mechanism involved in multiple developmental processes, including flowering (through activation of FLC) and floral development (through interaction with yet-undefined factors). The activity of such a mechanism in specific contexts could be directed by spatial or temporal cues provided by specific auxiliary factors. The functions in plant development that we propose for VIP3 are similar, but opposite, to those described for the VRN1 gene (Simpson and Dean 2002). For example, gain-of-function studies suggest that, like VIP3, VRN1 is involved in flowering through both FLC-dependent and FLC-independent mechanisms and that VRN1 is also involved in developmental processes apparently unrelated to timing of flowering, including floral development (Levyet al. 2002). Thus, these two genes could act in an antagonistic manner. Because the silencing of FLC associated with vernalization might involve changes in chromatin environment, one possibility is that a VIP3 mechanism could maintain FLC chromatin in a configuration that is relatively accessible to transcription. In additional developmental contexts, this mechanism may act on other genes subject to chromatin-associated silencing. Further characterization of the VIP genes will require the identification of additional regulatory targets and the definition of elements that specify these as targets.
We thank Rick Amasino and Scott Michaels for contributing several mutants (vip1, vip6-3, and vip7) for analysis and Steve Rounsley (Cereon Corporation) for making the Cereon Arabidopsis Polymorphism Collection available. We are indebted to Reginald Ronnigen and the MSU Cyclotron Laboratory for assistance with fast-neutron mutagenesis. This work was supported by the Michigan Agricultural Experiment Station and a U.S. Department of Agriculture National Research Initiative Competitive grant (79-35304-5108) to S.v.N.
Communicating editor: C. S. Gasser
- Received October 19, 2002.
- Accepted January 24, 2003.
- Copyright © 2003 by the Genetics Society of America