418 The control of flowering by vernalization Candice C Sheldon, E Jean Finnegan, Dean T Rouse, Million Tadege, David J Bagnall, Chris A Helliwell, W James Peacock and Elizabeth S Dennis* The process by which vernalization, the exposure of a germinating seed or a juvenile plant to a prolonged period of low temperature, promotes flowering in the adult plant has remained a mystery for many years. The recent isolation of one of the key genes involved in vernalization, FLOWERING LOCUS C, has now provided an insight into the molecular mechanism involved, including the role of DNA methylation. Addresses Commonwealth Scientific and Industrial Research Organisation, Plant Industry, GPO Box 1600, Canberra, Australian Capital Territory 2601, Australia *e-mail: e.dennis@pi.csiro.au Current Opinion in Plant Biology 2000, 3:418-422 1369-5266/00/$ - see front matter © 2000 Elsevier Science Ltd. All rights reserved. Col Columbia efs early flowering in short days FLC FLOWERING LOCUS C FRI FRIGIDA GA gibberellin Ler Landsberg erecta LFY LEAFY METI METHYLTRANSFERASEI QTL quantitative trait locus VRN1 VERNALIZATION1 Introduction The transition from vegetative growth to flowering is controlled by both environmental and developmental signals. Ambidopsis thaliana has been widely used as a model plant in the study of the molecular mechanisms that govern this process. Ecotypes of Ambidopsis are found over wide geographic and climatic ranges. Ecotypes from high latitudes or from alpine regions often take more than three months to flower when grown under controlled conditions, but flower much more rapidly (after approximately three weeks) when their germinating seeds have been exposed to a prolonged cold-treatment, a process known as vernalization. Ecotypes originating closer to the equator, and the commonly used laboratory ecotypes (e.g. Landsberg erecta [her] and Columbia [Col]), flower rapidly without exposure to cold. The promotion of flowering by low temperature, combined with the onset of long days, ensures that flowering occurs in the spring, providing the maximal opportunity for seed set. Genetic analyses of late- and early-flowering Ambidopsis ecotypes identified two major loci determining flowering time: FRIGIDA (FRI) on chromosome 4 and FLOWERING LOCUS C {FLC) on chromosome 5 [1-5]. Dominant alleles of these genes act synergistically to cause late-flowering; the late-flowering phenotype can be fully suppressed by vernalization. In Brassica species, vernalization-responsive flowering time loci segregate as two major quantitative trait loci (QTLs) that are collinear with the regions of the Ambidopsis genome in which FRI and FLC are located [6], suggesting that the same genes are important in both genera. Multiple genes that confer insensitivity to vernalization have been mapped in wheat and barley but it is not known whether they are homologues of FRI or FLC [7], In this review we discuss progress made during the past two years in understanding the molecular basis of the promotion of flowering by vernalization. Vernalization may be mediated through changes in DNA methylation Vernalization has a number of unique features that can be accounted for by the hypothesis that it causes the activation, by demethylation, of gene(s) that are essential in the promotion of flowering [8]. The observation that prolonged growth at low temperatures results in reduced genomic DNA methylation is consistent with this hypothesis [8,9], Genome-wide demethylation, induced either by treatment with 5-azacytidine or by a METHYLTRANSFERASEI (METI) antisense construct, promotes flowering in vernalization-responsive Ambidopsis ecotypes and mutants, but not in non-vernalization-responsive lines [8,9]. 5-azacytidine also largely replaces vernalization in winter wheat [10], Unlike the day-length trigger to flowering, in which the leaf perceives the stimulus and produces a signal that is transmitted to the apex, exposure to low temperature is perceived by actively dividing cells in the apex itself [11,12]. Growth at low temperatures may disrupt maintenance methylation, the process by which patterns of DNA methylation are transmitted to newly synthesized DNA in dividing cells [13], perhaps through decreasing the activity of a cold-sensitive DNA methyltransferase. Methylation at sites in 'vernalization genes' would be diluted by successive cycles of DNA replication, accounting for the requirement for cell-division for the vernalization response, and the observed correlation between the duration of the cold treatment and the extent to which flowering is promoted [14,15,16**]. As vernalization often occurs at the germinating seed stage with flowering occurring weeks later, the vernalization signal must be transmitted through a number of cycles of cell division; it is not, however, transmitted to progeny [12,17]. It could be that changes in the methylation patterns of specific genes, which are established during growth in the cold, are maintained through mitotic cell divisions, but reset in progeny. The control of flowering by vernalization Sheldon eř a/. 419 FLC is a key regulator of flowering time and the response to vernalization The recent isolation of the FLC gene has provided a major advance in uncovering the molecular mechanisms involved in the control of flowering time and the response to vernalization [18**, 19**]. The data show that FLC acts as a repressor of flowering and that the level of FLC expression correlates with the time to flowering. Late-flowering ecotypes and the over-expression mutant flc-11 (previously called flf-1) have a high level of expression, whereas early-flowering ecotypes and the loss-of-function mutant flc-13 iflf-3) have little or no activity [16**,18**]. The early-flowering ecotypes Lít and C24 have been characterized genetically as possessing recessive (i.e. inactive) FLC alleles, whereas all other ecotypes have dominant (i.e. active) alleles [4,5]. The predicted translation products of the G24, Lít and Col FLC alleles are, however, identical, suggesting that the observed differences in allelic activity are caused by differences in FLC gene regulation [16**], and not to changes in the gene product. Dominant FRI alleles function to upregulate FLC expression [18**,19**], accounting for the observed genetic interaction between FRI and FLC. The early-flowering Lít and Col ecotypes have been hypothesized to contain inactive FRI alleles [20], which would not upregulate FLC; these ecotypes do have a low level of FLC transcript. In addition to FRI and FLC, about 80 loci that influence flowering time have been identified by mutational approaches [21]. These mutants have been mapped to several flowering pathways [22,23] (Figure 1). FRI and FLC are both active in the autonomous pathway, along with FCA, FVE, FPA, LUMINIDEPENDENS (LD), FLOWERING LOCUS D (FLD) and FY; the mutation of any of these genes causes a late-flowering phenotype. Mutants of these genes can be further divided into three subgroups (as shown in Figure 1) on the basis of their epistasis relationships [24]. Many of the genes in the autonomous pathway have been shown to regulate FLC expression [16**,18**,19**] (Figure 1), placing FLC as one of the later genes in the pathway. Levels of both FLC mRNA and protein are downregulated by exposure of the germinating seed to prolonged periods at low temperatures [16**,18**,19**], showing that the FLC gene is integrally involved in mediating the response to vernalization. The extent of downregulation is proportional to the duration of the cold-treatment, as is the promotion of flowering [16**]. The downregulation of FLC activity persists throughout the development of the vernalized plant, the change in gene expression being transmitted through many mitotic divisions. The high level of FLC expression is reset in the progeny of a vernalized plant [16**]. Consistent with the idea that vernalization acts to alter expression of critical genes through changes in methylation status, plants containing a MET1 antisense construct with only 15% of the wild-type level of genomic methylation are early-flowering [9] and have a reduced level of FLC Figure 1 Vernalization VRN1 Autonomous pathway FPA FVE FY FCA LD FLD Vegetative VRN2 -----------------► Flowering Current Opinion in Plant Biology In Arabidopsis, an ability to respond to vernalization is established by an elevated level of FLC transcript [1 6"]. In the late-flowering ecotypes, this is caused by the upregulation of FLC by dominant alleles of FRI [1 8",1 9"]. In the late-flowering mutants of the autonomous pathway, elevated FLC levels are caused by the loss-of-function of one of the genes that normally act to downregulate FLC expression. Vernalization causes a decrease in genomic methylation [9], and plants with a reduced level of DNA methylation have reduced expression of the FLC gene [18"]. It is not known whether FLC expression is directly regulated by demethylation or whether it is mediated through a regulator of FLC. The downregulation of FLC expression by vernalization is mediated through the effects of the VRN1 and VRN2 genes [1 6"]. VRN2 also acts as a repressor of FLC in the unvernalized plant [18"]. FLC may act to regulate flowering time by blocking GA action at the apex, affecting either GA biosynthesis or signal transduction [18"]. It is likely that LFY acts downstream of the autonomous pathway and therefore downstream of FLC. Sucrose and GA both promote flowering and both synergistically upregulate expression from the LFY promoter [38]. Sucrose may also act in the autonomous pathway to downregulate FLC expression. transcript [18**]. Demethylation is generally associated with activation of gene expression rather than downregulation. Thus, the downregulation of FLC suggests that methylation may block expression of a repressor of FLC, or perhaps the binding of a repressor to the FLC promoter. Far-red light promotes flowering in ecotypes and mutants that have a strong vernalization response [25]. However, plants grown under far-red-rich light show little alteration in FLC expression level compared with those grown in far-red-poor light (CC Sheldon, ES Dennis, unpublished data), suggesting that this effect is not mediated through FLC. FLC expression is also not affected by growth in different photoperiods [16**, 19**], and the flc-13 mutant maintains a response to photoperiod [16**]. Thus, FLC appears not to be essential for the photoperiod flowering pathway. However, flc loss-of-function mutants have a shortened circadian period, and a QTL for circadian period is closely linked to the FLC locus [26], 420 Cell signalling and gene regulation Genes that mediate the vernalization response All mutants and ecotypes that have an elevated level of FLC are late-flowering and are able to respond to vernalization, both by a decrease in their time to flowering and by a decrease in the level of FLC transcript [16**]. This finding has two implications. First, it suggests that an elevated level of FLC is essential for the ability of a plant to respond to vernalization. This is supported by the lack of vernalization response of the flc-13 null mutant [16**]. Second, it indicates that the late-flowering phenotype of these mutants is caused by an upregulated level of FLC transcript. This was confirmed by the shortening of the time to flowering of the late-flowering^-/ mutant by an FLC antisense construct [16**]. Other Arabidopsis genes that modify the vernalization response have been identified. The pleiotropic early-flowering mutant efs {early flowering in short days) encodes a repressor of flowering that acts in the autonomous pathway to flowering [27*]. Double mutants between efs and the vernalization-responsive late-flowering mutants^« and fve flower at about the same time as the efs mutant and show no promotion of flowering following a vernalization treatment. The loss of a vernalization response in these double mutants may simply indicate that efs mutants flower as early as is possible. Alternatively, like FLC, EFS may link the autonomous and vernalization-dependent pathways to flowering [27*]; perhaps EFS regulates FLC activity either transcriptionally or posttranscriptionally by interacting directly with FLC. Mutants showing a reduced response to vernalization have been isolated in an fca mutant background; these have been placed in three complementation groups: VERNALIZATION 1 (VRNI), VRN2 and VRN3 [28]. Both fca vrnl and fca vrn2 show a reduced response to vernalization, with a corresponding decrease in the downregulation of FLC in vernalized fca vrnl and fca vrn2 plants [16**,28]. The double mutant fca vrn2 flowered later than fca and had an increased level of FLC transcript compared with the fca mutant, suggesting that VRN2 is a repressor of FLC [16**, 18**]. In contrast, whereasyhz vrnl flowered no later than fca, the vrnl mutation alone delayed flowering. This late-flowering phenotype was not associated with elevated FLC levels, indicating that VRNI may also promote flowering via another pathway [16**]. Genes that act downstream of FLC in the autonomous and vernalization-dependent pathways All of the genes discussed so far act to regulate FLC, and so far no genes have been identified that are directly regulated by FLC (see Update). FLC belongs to the MADS-box family of transcription factors and is likely to regulate the transcription of other genes, acting either as a homodimer or as a heterodimer with other MADS-box proteins. One clue as to the type of gene that FLC may regulate comes from the observation that the late-flowering flc-11 mutant has a reduced response to applied gibberellin (GA) compared to the wild-type or to other late-flowering mutants. This suggests that FLC may block GA action in the apex [18**] either by regulating the expression of GA-biosynthetic genes or by affecting signal transduction genes. Exogenous GA does not change FLC transcript levels [18**], suggesting that GA acts downstream of FLC. Consistent with this, the vrnl mutant did not affect the promotion of flowering by application of GA [28], Strengthening the evidence for a link between FLC and GA action is the observation that plants that constitutively over-express FLC display phenotypes that are consistent with reduced GA activity [18**]. Thefve mutant (which has an elevated level of FLC) has reduced internode length compared with wild-type plants [29], suggesting a deficiency in GA biosynthesis or response. The loss-of-function flc-13 mutant displays phenotypes that are associated with an enhancement of GA response or plants treated with exogenous GAs, such as early germination, elongated hypocotyls and petioles (DJ Bagnall, GG Sheldon, unpublished data), and early flowering [18**]. GA is an important flowering promoter in Arabidopsis. The GA-insensitive Arabidopsis mutant gal-3, which has low levels of endogenous GA [30], does not flower when grown in short days and its flowering is delayed under long-day conditions [31,32]. Despite these correlations, the involvement of GA in the vernalization response remains a controversial topic. An increase in the activity of the GA biosynthesis enzyme