The VERNALIZATION INDEPENDENCE 4 gene encodes a novel regulator of FLOWERING LOCUS C


*For correspondence (fax 517-353-0890; e-mail


The late-flowering, vernalization-responsive habit of many Arabidopsis ecotypes is mediated predominantly through repression of the floral programme by the FLOWERING LOCUS C (FLC) gene. To better understand this repressive mechanism, we have taken a genetic approach to identify novel genes that positively regulate FLC expression. We identified recessive mutations in a gene designated VERNALIZATION INDEPENDENCE 4 (VIP4), that confer early flowering and loss of FLC expression in the absence of cold. We cloned the VIP4 gene and found that it encodes a highly hydrophilic protein with similarity to proteins from yeasts, Drosophila, and Caenorhabditis elegans. Consistent with a proposed role as a direct activator of FLC, VIP4 is expressed throughout the plant in a pattern similar to that of FLC. However, unlike FLC, VIP4 RNA expression is not down-regulated in vernalized plants, suggesting that VIP4 is probably not sufficient to activate FLC, and that VIP4 is probably not directly involved in a vernalization mechanism. Epistasis analysis suggests that VIP4 could act in a separate pathway from previously identified FLC regulators, including FRIGIDA and the autonomous flowering promotion pathway gene LUMINIDEPENDENS. Mutants lacking detectable VIP4 expression flower earlier than FLC null mutants, suggesting that VIP4 regulates flowering-time genes in addition to FLC. Floral morphology is also disrupted in vip4 mutants; thus, VIP4 has multiple roles in development.


In many plants, flowering is not initiated until after an extended period of growth in the cold. In the natural environment, this mechanism allows flowering and seed production to occur only after winter. In the many ecotypes of Arabidopsis thaliana that exhibit this type of flowering habit, repression of flowering is mediated predominantly through the activity of the MADS-box gene FLOWERING LOCUS C (FLC) (Koornneef et al., 1994; Lee et al., 1994b; Michaels and Amasino, 1999; Sheldon et al., 1999). In current models of flowering, the repressive mechanism involving FLC acts antagonistically with promotive pathways associated with GA biosynthesis/sensitivity and perception of inductive photoperiods (Simpson et al., 1999). The accelerated flowering of plants treated with GAs, or grown in inductive photoperiods, is not accompanied by greatly decreased FLC RNA expression (Michaels and Amasino, 1999; Sheldon et al., 1999), and FLC does not appear to be developmentally regulated (Rouse et al., 2002; Sheldon et al., 1999), indicating that flowering pathways are integrated predominantly ‘downstream’ of FLC. However, both genetic and molecular experiments have suggested that some ‘crosstalk’ occurs among pathways (Koornneef et al., 1998a; Rouse et al., 2002), and thus these flowering mechanisms cannot be proposed to act completely independently.

FLC is subject to both positive and negative regulation, and several flowering-time genes are known that act as strong, ‘upstream’ regulators of FLC. For example, genes in the so-called autonomous pathway, including LUMINIDEPENDENS (LD), FLOWERING LOCUS D (FLD), FPA, FVE, FY, and FCA, act to repress FLC (Koornneef et al., 1991, 1998a). This regulation occurs at least partly at the RNA level, as FLC RNA is expressed to high levels in autonomous-pathway mutants relative to wild-type plants (Michaels and Amasino, 1999; Sheldon et al., 1999). Several of the autonomous-pathway genes have now been characterized at the molecular level. LD encodes a nuclear protein containing a diverged homeodomain and an acidic carboxyl-terminal region enriched in glutamine residues (Aukerman and Amasino, 1996; Lee et al., 1994a; van Nocker et al., 2000) suggesting that it could act as a transcriptional regulator. Two other autonomous-pathway genes, FCA and FPA, encode proteins containing potential RNA-binding domains (Macknight et al., 1997; Schomburg et al., 2001), suggesting that they function in post-transcriptional control of expression.

The FRIGIDA (FRI) gene also regulates FLC RNA expression, but, in contrast to the autonomous-pathway genes, acts in a promotive manner (Koornneef et al., 1998b; Michaels and Amasino, 1999). The predicted FRI protein does not exhibit strong homology with any other protein of known function, but exhibits coiled-coil domains, suggesting that it interacts with protein partner(s) (Johanson et al., 2000). Although it is now clear that other positive regulators of FLC exist (below), only FRI has been characterized, because allelic variation at FRI is a major determinant of the flowering habit (i.e. annual versus winter-annual) among natural Arabidopsis ecotypes (Johanson et al., 2000; Lee et al., 1993).

The mechanistic relationships among the autonomous-pathway genes, and between these genes and FRI, have not been well characterized. Koornneef et al. (1998a) explored these relationships through genetic epistasis experiments. Mutations in FY did not further enhance the late flowering conferred by loss of FCA function, suggesting that the two genes have a close functional relationship. In contrast, mutations in two other autonomous-pathway genes, FPA and FVE, greatly enhanced the lateness of fca and fy mutants. These genetic interactions were subsequently found to be reflected at the level of FLC RNA and protein expression, which were enhanced in fpa/fca, fve/fca, and fve/fy double mutants relative to the respective single mutants (Rouse et al., 2002). These findings could indicate that FPA/FVE and FCA/FY comprise two, partially redundant, mechanisms of FLC repression. However, this type of analysis is contingent on mutations creating a complete loss of function, and this has not yet been demonstrated for all of these genes. Loss of FLC function completely suppresses the late-flowering phenotype conferred either by FRI, or by loss of function at least of fve, ld, fpa, or fca (Michaels and Amasino, 2001). Thus, FRI and these autonomous-pathway genes likely act in flowering solely through mediation of FLC activity. The activation of FLC by FRI is epistatic to repression of FLC by autonomous-pathway genes (Michaels and Amasino, 2000). This is consistent with a mechanism whereby FRI limits the activity of the autonomous pathway, possibly through the negative regulation of one or more components.

The flowering-promotive effect of cold, termed vernalization, is mediated largely through repression of inhibitory FLC activity. This also occurs at the RNA level (Michaels and Amasino, 1999; Sheldon et al., 1999), and probably at the transcriptional level, as cold is not sufficient to overcome the repression of flowering associated with constitutive expression of FLC in transgenic plants (Michaels and Amasino, 1999; Sheldon et al., 1999). The molecular process(es) involved in vernalization-associated down-regulation of FLC is completely unknown. However, it is not likely to directly involve FRI or the autonomous-pathway genes; the evidence for this is that a long period of cold is fully effective to abrogate the late flowering phenotype of mutants lacking activities of both FRI and any of the known autonomous-pathway genes (Michaels and Amasino, 2000). Although vernalization by nature should involve a temperature-sensitive mechanism, no molecular components of such a mechanism have been definitively identified. Moreover, although pathways of cold signalling in Arabidopsis are becoming increasingly well characterized, the involvement of known cold-signalling components in vernalization has generally not been explored. This is, at least in part, because most studies of cold signalling have been carried out in ‘lab strains’ of Arabidopsis[e.g. Columbia (Col), Landsberg erecta (Ler)] that, because they lack effective FRI and/or FLC alleles, do not exhibit strong FLC activity and typically flower soon after germination irrespective of cold (Johanson et al., 2000; Koornneef et al., 1994; Lee et al., 1994b). The well-known CBF family of transcription factors, which act as molecular ‘switches’ to induce many elements of the cold acclimation response (Gilmour et al., 2000), do not seem to be involved in vernalization. Constitutive expression of CBF1 or CBF3 in a late-flowering genetic background containing active FRI and FLC alleles, although sufficient to activate cold-responsive genes, did not greatly affect flowering time or FLC expression (Liu et al., 2002; our unpublished results).

Once FLC is down-regulated in vernalized plants, repression is maintained through an epigenetic mechanism involving the VRN2 gene (Gendall et al., 2001). The cold-associated down-regulation of FLC is not greatly affected by loss of VRN2 function, indicating that this gene probably is not important for initial suppression of FLC. VRN2 encodes a protein with sequence similarity to a member of the Polycomb-group protein class, which has been best characterized in Drosophila. These proteins are components of large complexes that reinforce the transcriptionally suppressed state of homeotic genes, potentially by packaging and/or maintaining chromatin in states less accessible to transcriptional machinery (Pirrotta, 1997). Similarly, it is likely that VRN2 functions in some way to reduce accessibility of the FLC gene, as FLC chromatin in vrn2 mutants exhibits increased DNase sensitivity relative to that of wild-type plants, following cold treatment (Gendall et al., 2001). That chromatin structure is intimately involved in flowering and vernalization, was previously shown by the strong effect on flowering conferred by disruption of processes tied to chromatin dynamics, including DNA methylation (Finnegan et al., 1996; Finnegan, 1998; Ronemus et al., 1996) and histone deacetylation (Tian and Chen, 2001), especially in genotypes with a winter-annual flowering habit (Burn et al., 1993). Transgenic plants in which endogenous DNA methylation was disrupted exhibited decreased FLC expression in the absence of a vernalizing cold treatment (Sheldon et al., 1999), indicating that appropriate chromatin structure is crucial for the maintenance of FLC expression in non-vernalized plants, as well as its suppression in vernalized plants.

As a first step to characterize the mechanism of flowering repression involving FLC, and how this mechanism is negatively regulated by cold, we carried out a genetic screen designed to identify positive regulators of FLC. Here we present the genetic and molecular analyses of one of these regulators, designated VERNALIZATION INDEPENDENCE 4.


A genetic screen for activators of FLC

To identify potential activators of FLC, we mutagenized the winter-annual, Col:FRISF2 (hereafter referred to as ‘wild-type’) genetic background and screened for recessive mutations that conferred cold-independent, early flowering. Early flowering lines were re-screened by assaying for reduced FLC RNA expression in seedlings, where FLC RNA is typically easily detectable (below). To eliminate further consideration of lines with mutations in the FLC and FRI genes, mutants were also used in genetic complementation analysis with lines FN231 and FN235, carrying loss-of-function mutations in the FLC and FRI genes, respectively. Early flowering lines that exhibited reduced FLC RNA expression, and that were not likely to represent new alleles of FLC or FRI, were sorted into allelic groups through complementation analysis. This strategy resulted in the identification of several complementation groups representing mutants that we designated vernalization independence (vip) mutants. The vip4 group, represented by two T-DNA alleles and one fast-neutron allele, was selected for further study (Figure 1). FLC expression was not detectable in plants carrying the T-DNA allele vip4-1, as determined by gel-blot analysis of seedling RNAs, indicating that VIP4 is a strong activator of FLC (Figure 1b).

Figure 1.

(a) Phenotype of the wild-type, Col:FRISF2 introgression line and a vip4 mutant.

(i) Wild-type plant grown in the absence of cold. Plant is 8-weeks-old and has produced approximately 60 leaves.

(ii) Wild-type plant grown after a vernalizing cold treatment and flowering after approximately 3 weeks.

(iii) A vip4-1 mutant plant grown in the absence of cold, and flowering after approximately 3 weeks.

(iv) Inflorescence from a vip4-1 mutant plant. Morphologically abnormal flowers where sepals fail to enclose the developing floral buds are indicated (arrows).

(b)Analysis of FLC RNA expression in wild-type (WT) and vip4 mutant plants. RNA was extracted from aerial portions of 14-day-old, wild-type and vip4-1 seedlings and analysed by gel blotting using an FLC probe as described in Experimental procedures. The membrane was subsequently stripped and reprobed with an 18S rDNA probe to indicate the integrity and relative quantity of total RNA in each lane.

To address the relationship between VIP4, FLC, and FRI, we evaluated the effects of a vernalizing cold treatment on the flowering response of vip4-1 relative to that of wild-type, flc, and fri plants (Figure 2). Flowering time was measured from a developmental perspective, as the total number of leaves produced on the primary stem. When grown under inductive (long-day) photoperiods in the absence of cold, vip4 mutants flowered at approximately the same time as the flc and fri mutants, and vernalized wild-type plants. However, significant differences were apparent when plants were grown under non-inductive (short-day) photoperiods, where the promotive activity of genes acting through perception of inductive photoperiods is expected to be minimized. We found that flc mutants flowered earlier (23.9 ± 4.4 leaves) than fri mutants (44.1 ± 1.2 leaves), suggesting that FLC retains a small degree of activity even in the absence of FRI function, and this is in accordance with previous observations (Michaels and Amasino, 2001). However, under these conditions, vip4-1 plants flowered even earlier (17.1 ± 1.3 leaves) than flc plants. This indicates that VIP4 may also repress flowering outside of its positive regulation of FLC. Also, similar to previous observations, cold reduced the flowering time of flc mutants suggesting that vernalization targets FLC-independent as well as FLC-dependent mechanisms (Michaels and Amasino, 2001). However, even vip4-1 plants showed a slight acceleration of flowering in response to cold, and vernalized wild-type plants flowered significantly earlier (10.7 ± 0.9 leaves) than did vip4-1 plants grown in the absence of cold (Figure 2). This suggests that, if vip4-1 is a null mutation, vernalization also involves a vip4-independent mechanism.

Figure 2.

Flowering time of wild-type plants, vip4 mutants, and mutants lacking activity of FRI or FLC.

Wild-type (WT) plants, vip4-1 mutants, and the FN231 and FN235 mutants lacking activity of FLC or FRI, respectively, were grown without a cold treatment (– cold), or after a 70-day cold treatment (+ cold) under long-day (LD) or short-day (SD) photoperiods, as described in Experimental procedures. ‘Leaf number’ indicates the total number of rosette and cauline leaves produced. Data is the mean and standard deviation for at least 12 plants. Wild-type plants grown in SD without cold did not flower during the course of this experiment.

In addition to the flowering-time phenotype, vip4 plants exhibit defects in floral morphology. Among these is a widening of medial sepals, such that sepals typically fail to enclose the remainder of the floral bud in the latest stages of floral development (Figure 1a). Petals are narrower than in wild-type flowers, and occasionally are greatly reduced in size. Stamens are often reduced in number to four or five. No defect in carpel morphology was apparent, and flowers are typically fully fertile. No additional phenotypic defects were obvious in vip4 mutants.

Cloning and identification of the VIP4 gene

Because segregation data indicated that the vip4-1 mutation might be due to T-DNA integration, we recovered genomic DNA flanking the T-DNA by inverse-PCR and found that the T-DNA was inserted into the transcribed region of a predicted gene near the bottom of chromosome V, designated At5g61150 (Figure 3a). Subsequent characterization of the At5g61150 region of vip4-2 and vip4-3 plants indicated the presence in the transcribed region of a T-DNA for vip4-2, and a large genomic insertion, originating from the top of chromosome V, for vip4-3 (Figure 3a). In vip4-1 plants, RNAs hybridizing with a At5g61150 probe failed to accumulate to levels detectable by gel blotting of total RNAs, whereas these RNAs were readily detectable in wild-type plants (Figure 3b). The observed size of the transcript was approximately 2.4 kb, consistent with the size derived from the annotation of the At5g61150 intron/exon structure provided by the Arabidopsis Genome Initiative. This predicted structure was confirmed by isolation and sequencing of several cDNAs from libraries prepared from Col:FRISF2 shoot apices (data not shown).

Figure 3.

(a) Depiction of the region of chromosome V encompassing the VIP4 gene.

The direction of transcription for VIP4 and two adjacent genes is indicated by arrows. Introns are shown as white boxes, whereas exons are shown as black (translated region) or grey (untranslated region) boxes. Only the proximal portion of adjacent genes are depicted; for these, intron/exon structure is shown as annotated by the Arabidopsis Genome Initiative, and the extent of untranslated regions is based on EST sequences in the GenBank/EMBL databases. The position of the inserted DNA for the vip4-1, vip4-2, and vip4-3 alleles is indicated. A zig-zagged line in the 3′ region of the VIP4 gene indicates the probe used for RNA gel blot analyses (below). The NsiI sites delineating the region used to complement the vip4-1 mutation are indicated.

(b) Analysis of VIP4 RNA expression in wild-type (WT) and vip4 mutant plants.RNA was extracted from aerial portions of 14-day-old, wild-type and vip4-1 seedlings and analysed by gel blotting using a VIP4 probe as described in Experimental procedures. The membrane was subsequently stripped and reprobed with an 18S rDNA probe to indicate the integrity and relative quantity of total RNA in each lane.

To confirm that we had identified the VIP4 gene, we introduced the entire At5g61150 transcriptional unit, plus immediately adjacent genomic regions, into the vip4-1 background, through Agrobacterium-mediated transformation. Primary transformants (T1 plants) were grown either in the absence of cold, or after a vernalizing cold treatment. All of the 20 T1 plants recovered in both cases were phenotypically indistinguishable from wild-type plants, producing at least 60 leaves before flowering in the absence of cold, flowering very early when given a cold treatment, and exhibiting normal floral morphology (Figure 4a, and data not shown). In non-vernalized progeny of a representative T1 plant, both VIP4 and FLC RNAs were expressed to levels similar to that seen in the wild-type plant (data not shown). As additional evidence that At5g61150 is VIP4, we disrupted expression of the At5g61150 gene in wild-type plants through antisense RNA expression. For this experiment, we engineered a transgene in which the part of At5g61150 corresponding to the translated and 3′ regions, including introns, was expressed in 3′ to 5′ orientation from the 35S CaMV promoter. Approximately one-third of the > 100 T1 plants recovered flowered very early in the absence of cold, and produced flowers with a vip4-like phenotype (Figure 4a, and data not shown). Finally, in transgenic plants engineered to express the At5g61150 transcribed region in the 5′ to 3′ orientation from the 35S promoter (35S:VIP4; see below), early flowering, vip4-like plants appeared with high frequency (approximately one-third of > 100 T1 plants) (Figure 4a). The vernalization-independent early flowering of the VIP4-antisense and 35S:VIP4 plants was presumably due to suppression of the endogenous VIP4 gene, as VIP4 RNA did not accumulate to detectable levels in any of the several plants assayed (Figure 4b). In addition, in contrast to non-transgenic, wild-type plants, FLC RNA was not detectable in leaf tissues of these early flowering, VIP4-antisense and 35S:VIP4 plants (Figure 4b), indicating that early flowering was mediated at least partly through loss of FLC expression.

Figure 4.

Molecular complementation of the vip4-1 mutation and manipulation of VIP4 expression in transgenic plants.

(a) (i, ii) Transgenic, vip4-1 mutant plants carrying an introduced copy of the VIP4 gene, grown without a cold treatment (i), or after a vernalizing cold treatment (ii). Plant shown in (i) is 6-weeks-old and has produced approximately 45 leaves. (iii, iv) Transgenic, wild-type plants expressing VIP4 antisense RNA (iii) or expressing the VIP4 gene from the 35S promoter (iv) grown without a cold treatment.

(b) Analysis of VIP4 and FLC RNA expression in wild-type and transgenic plants. VIP4 and FLC RNA expression was evaluated in wild-type (WT) plants, a representative, late-flowering transgenic plant expressing the VIP4 gene from the 35S promoter (35S:VIP4#5), and representative, early flowering transgenic plants expressing VIP4 antisense RNA (VIP4-AS#4) or expressing the VIP4 gene from the 35S promoter (35S:VIP4#9). RNA was extracted from rosette leaves of non-vernalized plants grown in LD photoperiods and analysed by gel blotting, using VIP4 and FLC probes. Blots were subsequently stripped and reprobed with an 18S rDNA probe to indicate the integrity and relative quantity of total RNA in each lane.

The VIP4 gene encodes a 633-residue, 72-kDa protein with a predicted pI of 4.4 (data not shown). Almost one-half of the residues are charged (Glu, Asp, His, Lys, Arg) and thus the VIP4 protein is highly hydrophilic; this hydrophilicity is most apparent in extensive amino-terminal and carboxyl-terminal regions (data not shown). The VIP4 protein does not exhibit any motif currently defined in the PROSITE Dictionary of Protein Sites and Patterns. However, predominantly within its less hydrophilic central domain, VIP4 exhibits sequence homology with the Leo1 protein from Saccharomyces cerevisiae, and other hydrophilic proteins of unknown function from Saccharomyces pombe, C. elegans, and Drosophila (23–29% identity over 239–311-amino acid segments; data not shown). VIP4 does not exhibit strong homology with any other protein predicted to be encoded by the Arabidopsis genome, and proteins homologous to VIP4 have not been reported from other plant species.

We used RNA gel blotting to analyse the general spatial expression pattern of VIP4 in non-vernalized plants. We found that VIP4 was expressed throughout the plant, with the potential exception of rosette leaves (Figure 5a). We subsequently used RT-PCR to confirm that VIP4 was expressed in these tissues as well (data not shown). This expression pattern generally paralleled that of FLC, which was also expressed ubiquitously, but at very low levels in the leaves (Figure 5a). A search of current databases of expressed sequence tags (ESTs) resulted in the identification of a single EST (BE527160) originating from developing seeds, indicating that VIP4 is expressed in seed tissues as well.

Figure 5.

Analysis of VIP4 and FLC RNA expression in various organs and tissues of wild-type plants, in non-vernalized and vernalized wild-type plants, and in various genetic backgrounds.

(a) Expression in seedlings (SL), shoot apices (SA), rosette leaves (RL), cauline leaves (CL), inflorescence apices (IA), flowers (F), stems (S), and roots (R) of non-vernalized plants. In order to obtain reproductive tissues for analysis, plants were grown under long-day (LD) photoperiods.

(b) Expression in aerial portions of 14-day-old seedlings grown without a cold treatment (non-vernalized; NV), or after a 40-day cold treatment (vernalized; V) under short-day (SD) photoperiods.

(c) Expression in aerial portions of 14-day-old Columbia (Col), wild-type Col:FRISF2 (WT) and ld-1 (Col:ld-1) seedlings grown without a cold treatment under short-day (SD) photoperiods.RNAs were analysed by gel blotting using VIP4 and FLC probes. Blots were subsequently stripped and reprobed with an 18S rDNA probe to indicate the integrity and relative quantity of total RNA in each lane.

To determine if the suppression of FLC RNA expression associated with vernalization might be mediated through suppression of VIP4, we evaluated VIP4 RNA expression in vernalized and non-vernalized seedlings. As shown in Figure 5b, VIP4 RNA was expressed to similar levels irrespective of the vernalization status. The effectiveness of the cold treatment given to these plants was evident by the decrease of FLC RNA to non-detectable levels (Figure 5b). This suggests that VIP4 is insufficient to activate FLC in vernalized plants, and that modulation of VIP4 RNA expression is unlikely to be involved in the vernalization response.

Molecular analysis of VIP4 function

We further characterized the relationship between VIP4, FRI, and an autonomous-pathway gene, LD, through analysis of molecular epistasis. As previously reported (Michaels and Amasino, 1999), we did not detect FLC RNA expression in the Col ecotype lacking activity of FRI, but found that it was expressed to readily detectable levels in the Col:FRISF2 line, and in an ld mutant in the Col background, which lacks activity of both FRI and LD (Figure 5c). In contrast, VIP4 RNA was expressed to similar levels in all three genotypes (Figure 5c). That VIP4 RNA expression was similar between Col and Col:FRISF2 indicates that VIP4 is not likely to mediate the activation of FLC expression by FRI. Likewise, that VIP4 expression was similar between an ld mutant and its wild-type genetic background Col indicates that the de-repression of FLC conferred by loss of LD function is also unlikely to be mediated through VIP4.

To help define the role of VIP4 and especially its relationship to FLC, we evaluated the effects of enhanced expression of VIP4 in transgenic Col:FRISF2 plants. Several plants expressing high levels of VIP4 RNA were identified from a 35S:VIP4 T1 population grown in the absence of cold (Figure 4b). This RNA was apparently processed to the same extent as the endogenous VIP4 RNA, as evidenced by its co-migration with the VIP4 transcript from wild-type plants (data not shown). These 35S:VIP4 T1 plants were phenotypically similar to wild-type plants with respect to flowering time and floral morphology (data not shown). Although VIP4 RNA accumulated to high levels in leaf tissues of these plants, FLC RNA expression was not enhanced in leaves, relative to its levels in wild-type plants (Figure 4b), suggesting that ectopic VIP4 activity was not sufficient to activate FLC, even in the absence of vernalization.


As a first step towards understanding the mechanism of FLC-mediated flowering repression and its negative regulation by cold, we are taking a genetic approach to identify components of floral-repressive mechanisms that act, at least partly, through promotion of FLC expression. Although several genetic efforts have already been carried out to identify regulators of flowering, these have mainly focused on recessive mutations that delay flowering, and thus most of the genes identified are assumed to act in a flowering-promotive capacity (Koornneef et al., 1991; Lee et al., 1994a; Redei, 1962). Several genes that act as floral repressors have also been identified (see below), largely through associated developmental pleiotropy, but these genes have been characterized only in the common ‘lab strains’ or ecotypes of Arabidopsis that do not normally exhibit strong FLC activity. Our use of a synthetic genetic background, containing an active FRI locus from a natural, winter-annual ecotype, introgressed into the Col genotype (Lee et al., 1994b), permits rigorous genetic analysis of FLC-associated repressive mechanism(s), while simultaneously permitting full utilization of currently available Arabidopsis genomics tools.

The genetic and molecular analysis of VIP4 demonstrates that it acts as a repressor of flowering at least partly through its ability to strongly activate FLC. Our current knowledge of flowering is consistent with FLC being regulated predominantly through at least two mechanisms or pathways (Michaels and Amasino, 1999). One mechanism involves the autonomous-pathway genes, which repress FLC expression, and FRI, which acts antagonistically to the autonomous pathway (Simpson et al., 1999), possibly by limiting the activity of one or more components. At least a second mechanism must be proposed to promote FLC expression, based on the observation that, in plants lacking activity of the autonomous pathway, FLC is strongly expressed even in the absence of FRI. Because FLC expression is repressed by cold even in the absence of FRI and/or autonomous pathway function, vernalization likely acts to limit the activity of this second mechanism.

VIP4 could be hypothesized to occupy any of a number of positions with respect to these pathways. VIP4 RNA levels were not affected by loss of function of FRI or LD, indicating that, if VIP4 mediates activation of FLC by FRI and/or de-repression of FLC by loss of LD activity, such a mechanism would have to involve changes in VIP4 protein activity, or changes in RNA levels within restricted tissues. Our observation that vip4 mutants flower much earlier than fri null mutants also suggests that VIP4 does not act in flowering exclusively with FRI as a co-activator of FLC. Thus, it is possible that VIP4 acts independently of these genes in a distinct mechanism required for FLC expression in the absence of cold. We found that increasing VIP4 RNA expression was not sufficient to further activate FLC, even in non-vernalized plants where other elements necessary for FLC expression are active. Also, vip4 mutations appear to be completely recessive (data not shown). The lack of gene dosage effect is consistent with VIP4 acting as one non-limiting component of a more extensive mechanism. Obvious candidates for other potential components are represented by the several allelic groups of vip mutations that we have identified through our genetic approach.

A flowering-repressive mechanism involving VIP4 could function in several possible capacities. For example, because the ‘vernalized state’ is not maintained through meiosis (i.e. the requirement for cold is re-set in each generation; Lang, 1965), this mechanism could act to re-establish FLC expression in the developing embryo, possibly by disrupting the epigenetic repressive mechanism involving VRN2. If so, then this might be reflected by decreased accessibility of FLC chromatin in the vip4 mutant, relative to that in wild-type plants. Another possibility is that VIP4 acts in a hypothetical pathway of vernalization cold signalling, maintaining it in an ‘off’ state. However, if this is the case, then VIP4 is unlikely to act as a general suppressor of cold-signalling pathways, a role hypothesized for the HOS1 gene (Lee et al., 2001), because unlike hos1 mutants, vip4 plants exhibited neither ectopic expression of a representative cold-responsive gene, COR78, nor enhanced freezing tolerance as measured by electrolyte leakage assays (our unpublished results).

Irrespective of its nature, the flowering-repressive mechanism involving VIP4 could be deactivated by cold through the negative regulation of one or more components. The observation that VIP4 RNA is expressed to equivalent levels in both non-vernalized and vernalized plants suggests that if VIP4 itself were a cold-regulated component, regulation would either be mediated at the level of VIP4 protein activity, or at the RNA level within a restricted subset of tissues. However, in this respect, it is noteworthy that the subtle floral defects seen in plants lacking VIP4 activity are not observed in vernalized, wild-type plants, suggesting that VIP4 maintains activity in vernalized plants, at least in floral tissues.

The VIP4 protein exhibits sequence homology with yeast Leo1 and proteins from Drosophila and C. elegans; in addition, the highly hydrophilic nature of these proteins is conserved. These observations suggest that these proteins could function in analogous molecular mechanisms. Of these proteins, only Leo1 has been characterized. High-throughput, proteomic analyses suggest that Leo1 physically interacts with multiple protein partners in several cellular contexts (Gavin et al., 2002; Ito et al., 2001). This protein has been shown to exhibit an ATP-sensitive interaction with the 19S ‘cap’ of the proteasome (Verma et al., 2000), and is a component of the Paf1 transcriptional complex, which is required for full expression of a subset of yeast genes (Mueller and Jaehning, 2002). It is noteworthy that the defects in floral morphology seen in vip4 mutants are not observed in mutants or natural ecotypes lacking FLC activity (Michaels and Amasino, 1999), suggesting that the role of VIP4 in floral development is mediated outside of its relationship with FLC. Thus, VIP4 likely acts as a common component of distinct developmental mechanisms, possibly through interactions with multiple protein partners.

The observation that vip4 mutants flower earlier than flc null mutants indicates that VIP4 regulates flowering-time genes in addition to FLC. These hypothetical target(s) could have a role in GA biosynthesis or sensitivity, or in the perception of photoperiod, as current models of flowering predict that such mechanisms would influence flowering outside of pathway(s) involving FLC (Simpson et al., 1999). An especially attractive candidate is FLM[also known as AGL27 (Alvarez-Buylla et al., 2000) or MAF1 (Ratcliffe et al., 2001)], which encodes a MADS-box protein highly related to FLC, and acts as a floral repressor through a mechanism that is likely independent of FLC (Ratcliffe et al., 2001; Scortecci et al., 2001). Other possibilities include AGL31, a tandemly repeated cluster of four genes which also encode proteins highly related to FLC (Alvarez-Buylla et al., 2000; Scortecci et al., 2001).

A more speculative candidate for an additional target of VIP4 regulation is the MADS-box gene AGAMOUS (AG), which has functions in floral organ and meristem identity (Mizukami and Ma, 1997). Ectopic expression of AG is associated with early flowering, and in some cases, floral defects similar to that observed for vip4 (Gómez-Mena et al., 2001; Mizukami and Ma, 1997; Serrano-Cartagena et al., 2000; our unpublished results). If this were the case, then VIP4 would be included in an expanding class of gene acting both in floral repression and negative regulation of AG. These genes include CURLY LEAF (CLF), WAVY LEAVES and COTYLEDONS (WLC), INCURVATA 2 (ICU2), EMBRYONIC FLOWER (EMF)1 and 2, and EARLY BOLTING IN SHORT DAYS (EBS) (Chen et al., 1997; Gómez-Mena et al., 2001; Goodrich et al., 1997; Serrano-Cartagena et al., 2000; Simpson et al., 1999). Interestingly, Goodrich et al. (1997) found that the ectopic activity of AGAMOUS could not explain the full degree of early flowering of the clf-2 mutant plants, and suggested that CLF also regulates other flowering gene(s). It is tempting to speculate that CLF, and perhaps these other genes as well, might play a role as regulators of FLC. As these genes have been characterized only in laboratory strains of Arabidopsis that lack FRI function, and therefore express FLC only weakly, their potential to regulate FLC remains unclear. The possibility that these genes are involved in vernalization is intriguing because at least CLF and another member of this class, EMF2, encode Polycomb-group-like proteins (Goodrich et al., 1997; Yoshida et al., 2001), suggesting that they participate in epigenetic regulation of gene activity. To test this possibility will require the introduction of the respective mutations into genetic backgrounds that normally express FLC. Alternatively, new alleles of these genes may be identified through further characterization of other vip allelic groups.

Experimental procedures

Plant material and growth conditions

Introgression line Col:FRISF2 consists of the FRI locus from ecotype San Feliu-2 (FRISF2) introgressed into the Columbia (Col) ecotype through six successive backcrosses and made homozygous by self-pollination (Lee et al., 1994b). Line FN231 contains a fast-neutron flc allele isolated in the Col:FRISF2 background, and is identical with flc-1 described by Michaels and Amasino (1999). Line FN235, containing a fast-neutron fri allele isolated in the Col:FRISF2 background, is as described by Michaels and Amasino (1999). The ld-1 mutant in the Col background was obtained from the Arabidopsis Biological Resource Center (ABRC) at The Ohio State University. Standard growth conditions were 22°C under 100–180 µmol·m−2·sec−1 of cool white fluorescent lighting and 16-h light/8-h dark (long-day) or 8-h light/16-h dark (short-day) photoperiods. For vernalizing cold treatments, seeds were surface-sterilized, placed on agar-solidified germination medium as described by van Nocker et al. (2000), and grown at 4°C under SD photoperiods. To evaluate flowering time, plants were grown individually in 5.7 × 5.7 × 7.5 cm pots. Plant transformations used the floral dip method of Clough and Bent (1998) and Agrobacterium strain ABI.

Mutagenesis and screening

For T-DNA mutagenesis, a binary vector designated pPZP201:BAR, containing a 5′ mannopine synthase/glufosinate resistance/3′ octopine synthase cassette cloned into the SmaI site of pPZP201 (Hajdukiewicz et al., 1994), was introduced into Col:FRISF2 plants. Seeds from infiltrated plants (T1 seeds) were subjected to a vernalizing cold treatment, transferred to soil for further growth, and herbicide-resistant T1 plants were allowed to self-pollinate and set seed. Seeds from approximately 500 T1 plants were pooled. Approximately 5000 T2 plants from each pool were screened for early flowering in the absence of cold. Fast-neutron mutagenesis and screening were described by Michaels and Amasino (1999). FLC RNA expression was evaluated in approximately 14-day-old progeny (T3 or M3) plants grown without a cold treatment in SD conditions. To test for genetic complementation of the flc or fri mutations, T3 or M3 individuals were crossed with lines FN231 and FN235, and flowering time was evaluated in F1 progeny.

Molecular techniques

DNA was isolated essentially as described by Murray and Thompson (1980); RNA was isolated as described by Liu et al. (2002). DNA and RNA gel-blot analyses were carried out as described by Liu et al. (2002). The probe for gel-blot analyses of VIP4 was a 432-bp fragment amplified from genomic DNA using primers CVN4-F1 (5′..ATGGACGAAAGGAGAGTGAAAG..3′) and CVN4-R1 (5′..GGAATCAGAATATGAGACGGAAG..3′); the probe for gel-blot analysis of FLC was a 510-bp RT-PCR product corresponding to FLC coding region but excluding the conserved MADS-domain. This segment of the FLC gene does not exhibit significant sequence homology with any other Arabidopsis gene. For inverse-PCR, 200 ng of restriction endonuclease-digested genomic DNA was purified using the QIAquick PCR Purification Kit (QIAGEN, Valencia, CA, USA), and subsequently incubated with 10 u T4 DNA ligase (Roche, Indianapolis, IN, USA) in a final reaction volume of 30 µl at 16°C overnight. DNA was amplified directly from 2 µl of the ligation mixture using Advantage cDNA Polymerase Mix (Clontech, Palo Alto, CA, USA).

For identification of VIP4 cDNAs, shoot apex cDNA libraries were constructed. Vernalized and non-vernalized Col:FRISF2 plants were grown under SD conditions, and when plants had formed 20–25 rosette leaves, 1–2 mm-thick sections containing the shoot apex were excised. Library construction utilized the ZAP Express XR system (Stratagene, La Jolla, CA, USA). Library A contained approximately 1.5–8.0 kbp cDNAs and had a primary titre of 6.25 × 106 recombinants; library B contained approximately 0.5–3.0 kbp cDNAs and had a primary titre of 4.25 × 106 recombinants.

Construction of transgenic Arabidopsis lines

For molecular complementation analysis, the bacteriophage P1 clone MAF19 was obtained from the Kazusa DNA Research Institute (Yana, Kisarazu, Chiba, Japan), amplified in E. coli, purified using the Qiagen Plasmid Midi Kit, and subjected to restriction with NsiI. An approximate 7.1 kb fragment containing the VIP4 transcriptional unit and adjacent intergenic regions was cloned into the PstI site of binary vector PZP212 (Hajdukiewicz et al., 1994), and introduced into the vip4-1 mutant background. For antisense expression and constitutive expression, a DNA segment containing the VIP4 transcribed region was amplified by PCR using primers CVN4-F1 and CVN4-R2 (5′..AGGCAAACACAAGCTCACTATC..3′), and cloned into the BamHI site of binary vector pPZP201:BAR:35S in reverse (for antisense expression) or forward (for constitutive expression) orientations. The pPZP201:BAR:35S plasmid was engineered by inserting the cauliflower mosaic virus (CaMV) 35S promoter from plasmid pBI121 (Clontech) into the XbaI site of pPZP201:BAR (above).


We thank Scott Michaels and Richard Amasino for contributing mutant lines vip4-2 and vip4-3, and for many helpful comments on this manuscript, and the Kazusa DNA Research Institute (Japan) for providing the Arabidopsis P1 clone MAF19. This work was supported by the Michigan Agricultural Experiment Station, and a US Department of Agriculture National Research Initiative Competitive Grant (79-35304-5108) to S.V.N.