Winter Memory throughout the Plant Kingdom: Different Paths to Flowering^,

Abstract

Molecular mechanisms contribute to the memory of winter in different plant groups.

Plants have evolved a variety of mechanisms to synchronize flowering with their environment to optimize reproductive success. Many species flower in spring when the photoperiod increases and the ambient temperatures become warmer. Winter annuals and biennials have evolved repression mechanisms that prevent the transition to reproductive development in the fall. These repressive processes can be overcome by the prolonged cold of winter through a process known as vernalization. The memory of the past winter is sometimes stored by epigenetic chromatin remodeling processes that provide competence to flower, and plants usually require additional inductive signals to flower in spring. The requirement for vernalization is widespread within groups of plants adapted to temperate climates; however, the genetic and biochemical frameworks controlling the response are distinct in different groups of plants, suggesting independent evolutionary origins. Here, we compare and contrast the vernalization pathways in different families of plants.

The timing of flowering is an important adaptive trait that often involves integrating multiple environmental cues to ensure reproductive success. In many species, the perception of daylength (photoperiod) is an essential environment cue as it provides reliable information about seasonal shifts (e.g. Song et al., 2015; Shim et al., 2017). In Arabidopsis (Arabidopsis thaliana), the so-called photoperiodic pathway is coupled to the sensing of ambient temperatures as warmer growth conditions accelerate flowering (Balasubramanian et al., 2006; Verhage et al., 2014). The perception of these environmental signals is superimposed on an internal developmental program that prevents flowering in young seedlings and promotes the transition to reproductive development in older plants (e.g. Yu et al., 2015). In many species adapted to temperate climates, the perception of seasonal changes also involves the acquisition of the competence to flower in response to an extended cold period, a process referred to as vernalization (e.g. Chouard, 1960; Preston and Sandve, 2013; Fig. 1A). In addition, some species acquire floral competence when exposed to the shorter photoperiod of winter (Purvis and Gregory, 1937; Wellensiek, 1985), but the molecular mechanisms controlling the so-called “short-day vernalization” are still unknown. Vernalization is adaptive in that it ensures that flowering does not occur before the freezing temperatures of winter, which would reduce reproductive success. After vernalization, however, many plants still require subsequent exposure to additional inductive signals to initiate reproductive development (e.g. Amasino, 2010).

Figure 1.

Vernalization pathways in different plant groups. A, Phylogenetic tree of angiosperms (data from Jansen et al. [2007]) showing, on the right, the vernalization-responsive species mentioned in the main text. B to E, Schematic representation of the mechanisms governing vernalization in Arabidopsis (B), A. alpina (C), sugar beet (D), and core Pooideae (E).

Whether vernalization is required for flowering as well as the duration of cold exposure required to fulfill the vernalization requirement varies considerably among species and even within a species (Amasino, 2010; Duncan et al., 2015). Genotypes with a vernalization requirement are typically referred to as either winter annuals or biennials. There is not a sharp distinction between winter annuals and biennials, but the difference often relates to the extent to which the plant develops before winter and/or whether there is an obligate requirement for cold exposure to flower (e.g. Salisbury and Ross, 1992). When there is variation within a species for a vernalization requirement, the vernalization-requiring genotypes are often classified as winter annuals (or winter varieties), whereas the genotypes without vernalization requirement are often referred to as spring annuals or spring varieties because they will readily flower when planted after winter in the spring. These behavioral differences have a considerable influence on agricultural practices and are key to the adaptation of plant varieties to distinct climates. Here, we review the recent progress made in our understanding of the molecular mechanisms controlling vernalization with a focus on three different plant groups, Brassicales, Caryophyllales, and Poales.

THE MEMORY OF WINTER IN ARABIDOPSIS

The first insights into the molecular mechanisms controlling flowering were obtained in the model Brassicaceae, Arabidopsis (e.g. Pajoro et al., 2014). We know from studies in this model that the timing of flowering is a complex process that involves many genes in networks coordinating the initiation of flowering with environmental cues and developmental programs (e.g. Bouché et al., 2016). An essential downstream step of floral induction involves the up-regulation of FLOWERING LOCUS T (FT), a gene that encodes a small protein with similarity to phosphatidylethanolamine-binding proteins (Kobayashi et al., 1999). The FT protein, traditionally called “florigen,” is produced in leaf vascular tissues and moves through the phloem to the shoot apical meristem (SAM), where it interacts with the bZIP transcription factor FD and 14-3-3 proteins (Fig. 1B; Abe et al., 2005; Wigge et al., 2005; Corbesier et al., 2007; Taoka et al., 2011; Ho and Weigel, 2014). Together, these proteins form a floral activator complex that triggers the expression of several downstream targets, including SUPPRESSOR OF OVEREXPRESSION OF CO1 (SOC1), resulting in switching the fate of the SAM from initiating leaves to the production of flowers (Moon et al., 2005; Yoo et al., 2005). Pathways controlling flowering, including the vernalization pathway in Arabidopsis, act primarily through the modulation of the activity of the floral integrators FT and SOC1.

In Arabidopsis, natural diversity of the vernalization requirement is largely due to allelic variation at FRIGIDA (FRI) and its downstream target FLOWERING LOCUS C (FLC); winter accessions bear dominant (i.e. active) alleles of both genes (Michaels and Amasino, 1999; Sheldon et al., 1999; Johanson et al., 2000; Gazzani et al., 2003). FRI is part of a complex involved in activating FLC, and FLC encodes a MADS-box protein that represses flowering by preventing the transcription of FT in leaves and SOC1 in the SAM (Michaels and Amasino, 1999; Sheldon et al., 2000; Hepworth et al., 2002; Helliwell et al., 2006; Searle et al., 2006). Thus, FLC repression of both leaf and meristem flowering pathways ensures a tight repression of flowering prior to cold in winter accessions. Upon cold exposure, the expression of FLC is stably repressed, thus conferring a molecular “memory” of the past winter (e.g. Amasino, 2010). Interestingly, the exposure to cold temperatures triggers a rapid decrease in FLC expression levels (i.e. within a few days), but only extended periods of cold ensure stable repression upon return to warmer growth temperatures (e.g. Finnegan, 2015). Although FLC repression is maintained throughout the plant’s life cycle, the repressed state of FLC is reset to an active state in the following generation, resulting in the re-establishment of the vernalization requirement (e.g. Schmitz and Amasino, 2007).

How cold represses FLC has been a long-standing question on which many studies have been focused. These studies have revealed multiple components of cold-mediated repression, including epigenetic modifications and antisense transcription (e.g. Kim and Sung, 2014), but the mechanisms controlling the initial decrease in FLC levels are not yet fully understood (Helliwell et al., 2015). The first vernalization-related, cold-induced change identified to date is the peak of expression of antisense FLC transcripts, collectively called COOLAIR, which are conserved in Arabidopsis relatives (Swiezewski et al., 2009; Castaings et al., 2014; Marquardt et al., 2014). The experimental reduction of COOLAIR expression prevents the vernalization-induced decrease in some activating chromatin marks at the FLC locus (Csorba et al., 2014), whereas the disruption of its promoter by T-DNA does not prevent the overall repression of FLC by vernalization (Helliwell et al., 2011). The peak of COOLAIR is followed by the increase in the expression of a sense noncoding RNA originating from the first intron of FLC, called COLDAIR (Heo and Sung, 2011). Although COLDAIR appears to be less evolutionary conserved than COOLAIR in Arabidopsis relatives (Castaings et al., 2014), the knock-down of its expression results in an increase of FLC expression associated with late flowering and reduced vernalization response (Heo and Sung, 2011). Following the expression of these noncoding RNAs, a key event is the cold-mediated induction of the gene encoding VERNALIZATION-INSENSITIVE3 (Sung and Amasino, 2004). This protein, which is necessary for the deposition of H3K27me3 repressive marks at the FLC locus, participates in the stable repression of FLC by the polycomb remodeling complex PRC2, as extensively reviewed elsewhere (e.g. Kim and Sung, 2014; Berry and Dean, 2015; Hepworth and Dean, 2015). The mitotic stability of vernalization-mediated FLC repression, as well as the subsequent resetting in the next generation, has provided a system to explore multiple aspects of the epigenetic control of gene expression. Other mechanisms have been postulated to regulate FLC at a molecular level, such as alternative splicing (Mahrez et al., 2016), and possibly posttranslational protein stabilization (Kwak et al., 2016). However, the extent to which these mechanisms participate in the control of the vernalization response in natural conditions is not well understood.

The repression of FLC has received much attention, but the regulation of additional genes appears to also contribute to the vernalization response in Arabidopsis as the flowering time of an flc null mutant is still accelerated by exposure to prolonged cold temperatures (Michaels and Amasino, 2001). Some obvious candidate genes to fulfill such a role are the paralogs of FLC, called FLOWERING LOCUS M (FLM) and MADS AFFECTING FLOWERING2-5 (MAF2-5), which also control flowering by repressing FT expression (Gu et al., 2013). Although initial studies reported somewhat contradictory results (Ratcliffe et al., 2001, 2003; Sung et al., 2006; Sheldon et al., 2009), the thorough characterization of the expression of the FLC family genes showed that they all respond to vernalizing treatments, albeit with different kinetics (Kim and Sung, 2013): FLC expression decreases rapidly upon cold exposure, whereas FLM and MAF2-3 expression only decreases after the cold period has ended, and MAF4-5 expression peaks during cold. The role of FLM and MAF2 in the vernalization response seems to be marginal (Kim and Sung, 2013); instead, these genes appear to be key to the repression of flowering at low ambient temperatures (Posé et al., 2013; Lee et al., 2013a; Rosloski et al., 2013; Airoldi et al., 2015; Sureshkumar et al., 2016). The maf3 single mutant does not show any phenotype, but the maf4 and maf5 single mutants are induced to flower by shorter cold periods, suggesting that MAF4 and MAF5 normally ensure that vernalization is not achieved by suboptimal durations of cold exposure (Kim and Sung, 2013). AGL19, another MADS-box protein closely related to SOC1, might also play a role in the vernalization pathway, as the agl19 and flc mutations show additive impairment of the vernalization response (Schönrock et al., 2006). AGL19 appears to be a floral activator up-regulated upon cold exposure through FLC-independent processes (Schönrock et al., 2006; Kang et al., 2015). Although further work is needed to assess the role of these additional components in the vernalization response, a recent study suggests that they might participate in environmental adaptation as their differential regulation is correlated with the flowering time of different accessions originating from an altitudinal gradient (Suter et al., 2014). In conclusion, there is still much to learn about FLC-independent vernalization events in Arabidopsis.

VERNALIZATION IN PERENNIAL BRASSICACEAE

In contrast to the annual habit of Arabidopsis, perennials live for many years and flower repeatedly throughout their lives. Critical to this life history strategy is that not all meristems are converted to inflorescences because some meristems must be reserved for next season’s growth. Indeed, the perennial life history of Arabis alpina, a close relative of Arabidopsis in the Brassicaceae, relies on the transient floral competence to ensure that not all meristems flower in a growing season. Some meristems undergo the floral transition in spring, while others remain vegetative to resume growth the following year. In A. alpina, the repression of flowering prior to vernalization is mediated by an FLC ortholog called PERPETUAL FLOWERING1 (PEP1), and repression of PEP1 during vernalization leads to increased expression of SOC1 and LEAFY, two essential promoters of flowering (Fig. 1C; Wang et al., 2009b, 2011a). As the gene name suggests, pep1 mutants flower rapidly without vernalization, and allelic variation at PEP1 contributes to the natural variation in flowering responses that exists among different accessions of A. alpina (Wang et al., 2009b; Albani et al., 2012). Unlike FLC in Arabidopsis, the expression of PEP1 is only transiently repressed by cold, and the meristems that transition to flowering during cold become inflorescences, whereas meristems at an immature stage remain vegetative (Wang et al., 2009b). Although FLC and PEP1 share several regulatory mechanisms, including chromatin remodeling and antisense transcription (Wang et al., 2009b; Castaings et al., 2014), the repressive H3K27me3 marks at PEP1 return to their original levels a few weeks after the end of the cold period, correlating with its transient repression (Wang et al., 2009b). Although it is not surprising that FLC and orthologs such as PEP1 are the basis of vernalization requirement in Brassicaceae, this difference in memory at the AtFLC locus versus lack of memory at the PEP1 locus is likely to be crucial for the perennial nature of A. alpina versus the annual habit of Arabidopsis.

The ability of a meristem to transition to flowering is also controlled by an age-dependent pathway. In the early stages of Arabidopsis development, the age-regulated miR156 promotes juvenility and represses flowering by posttranscriptionally down-regulating the expression of genes from the SQUAMOSA PROMOTER BINDING PROTEIN-LIKE (SPLs) family, which are positive regulators of flowering (Cardon et al., 1997). As plants age, miR156 levels decrease, leading to an up-regulation of SPLs, which in turn induce miR172 expression; the levels of miR172 are thus negatively correlated with those of miR156 (Wu and Poethig, 2006; Wang et al., 2009a). Although this balance is conserved in many species, including rice, maize, and poplar (Chuck et al., 2007; Wang et al., 2011b; Xie et al., 2012), the levels of miR156 and miR172 are uncoupled in A. alpina, and exposure to cold temperatures triggers the expression of miR172 independently of miR156 (Bergonzi et al., 2013; Fig. 1C). PEP2, also identified through the screening for perpetual-flowering mutants, encodes a miR172-regulated APETALA2-like transcription factor that positively regulates PEP1 expression, thus contributing to the maintenance of the vegetative stage (Bergonzi et al., 2013). The cold-mediated induction of miR172 thus leads to the repression of PEP2 and, concomitantly, a decrease of PEP1 levels. In young meristems, however, high levels of miR156 block flowering, even in the absence of a PEP1-repressing effect. As observed in Arabidopsis, miR156 levels decrease as meristems age (Bergonzi et al., 2013), and only meristems with low miR156 levels can be induced to flower by prolonged exposure to cold temperatures. Interestingly, the decline in miR156 is blocked by cold, ensuring that flowering occurs only in plants that had previously grown rapidly under warmer ambient temperatures. In addition, TERMINAL FLOWER1 (TFL1), a paralog of FT, represses flowering in immature meristems (Kobayashi et al., 1999; Wang et al., 2011a). The experimental down-regulation of AaTFL1 allows the vernalization-mediated floral induction of younger meristems and a response to shorter periods of cold exposure (Wang et al., 2011a). Such a role for TFL1 in the repression of flowering in young meristems has been observed in other perennial species, such as apple trees (Kotoda et al., 2006). TFL1 appears to set a minimal threshold of inductive signal that is required to trigger flowering, and participates in the selective induction of flowering in mature meristems only.

In another perennial Brassicaceae, Cardamine flexuosa, the mechanisms involved in the age-dependent ability to become vernalized appear to be less complex. As in Arabidopsis, the balance between miR156 and miR172 is maintained, and the age-driven decrease of miR156 is associated with a concomitant increase of miR172. In C. flexuosa, the prevention of flowering in young nonvernalized plants is ensured by two potent floral repressors, FLC and TARGET OF EAT1 (TOE1), an APETALA2-like transcription factor that is posttranscriptionally regulated by miR172 (Zhou et al., 2013). As the plants age, miR172 levels increase to repress TOE1, conferring to the meristem the competence to flower when exposed to a vernalizing treatment that transiently represses FLC expression (Zhou et al., 2013). Hence, distinct mechanisms using a similar framework evolved to confer perennial behavior in Brassicaceae.

VERNALIZATION SYSTEMS IN OTHER EUDICOTS

The key regulators of the vernalization pathway appear to be conserved within Brassicaceae; however, there is no strong evidence that similar components are involved in the vernalization systems of other eudicot families. Although the heterologous expression of FLC-like genes from different families is able to repress flowering in Arabidopsis flc mutants, their functional relevance in the control of vernalization response in their respective species is not clear (e.g. Reeves et al., 2007; Périlleux et al., 2013). Moreover, in Medicago truncatula, a legume with a vernalization response that diverged 90 to 100 million years ago from Arabidopsis (Zeng et al., 2014), neither FRI nor FLC orthologs have been identified (Hecht et al., 2005). However, a mutation of MtVERNALIZATION2 (MtVRN2), a member of the Polycomb Group Repressive Complex (VRN2-PRC2) responsible for the deposition of H3K27me3 repressive marks at the FLC locus in Arabidopsis (De Lucia et al., 2008), bypasses the vernalization requirement and leads to early flowering (Jaudal et al., 2016). Under long-day conditions, this phenotype requires a functional allele of the florigen FTa1 (Jaudal et al., 2016). Interestingly, FTa1 is up-regulated in the Mtvrn2 mutant, but does not display H3K27me3 alterations, suggesting that MtVRN2 represses genes upstream of FTa1 (Jaudal et al., 2016). VRN2 thus plays distinct roles in the control of flowering in different plant groups, acting as an essential component of the cold-mediated acquisition of the competence to flower in Arabidopsis and participating in the establishment of the vernalization requirement in M. truncatula.

To date, the best characterized vernalization system in a eudicot family other than Brassicaceae comes from sugar beet (Beta vulgaris), a Caryophyllales species that diverged from Arabidopsis soon after the eudicot-monocot split, about 110 million years ago (Zeng et al., 2014). Cultivated sugar beet is a biennial root crop that requires vernalization to flower, but many wild relatives behave as annuals. Unlike in the Brassicaceae, the ortholog of FLC from sugar beet plays at best a minor role in vernalization (Vogt et al., 2014). The biennial behavior of cultivated beet is associated with a recessive allele at the bolting locus B. This locus encodes the pseudo-response regulator (PRR) BOLTING TIME CONTROL1 (BvBTC1), a protein related to AtPRR7 (Pin et al., 2012), and the reduction of BvBTC1 activity by RNAi is sufficient to convert annual varieties into biennials. Recessive alleles of another bolting locus, called B2, confer a biennial behavior to plants homozygous for the annual allele BTC1 (Büttner et al., 2010). This locus encodes the DOUBLE B-BOX TYPE ZINC FINGER protein BvBBX19, which is orthologous to an Arabidopsis protein that negatively influences the induction of FT (Dally et al., 2014; Wang et al., 2014). Interestingly, the expression of both BvBBX19 and BvBTC1 is diurnally regulated (Pin et al., 2012; Dally et al., 2014), and although the functional relationship between these two proteins is still unclear, they both participate in the regulation of an antagonistic pair of FT-like proteins, BvFT1 and BvFT2 (Fig. 1D; Pin et al., 2010, 2012; Dally et al., 2014). The short-day-expressed BvFT1 represses floral transition and negatively influences BvFT2 levels, whose expression is promoted by long days (Pin et al., 2010). In biennial beets that require vernalization, cold stably represses BvFT1 expression possibly through BvBTC1, relieving its repressive effect on BvFT2; BvFT2, in turn, is able to trigger flowering if plants are exposed to long days (Pin et al., 2010, 2012). The pathway through which BvBBX19 and BvBTC1 regulate the expression of BvFT genes is not known. Transcriptomic studies have provided some new candidate genes possibly involved in the vernalization pathway (Mutasa-Göttgens et al., 2012), and the publication of the sugar beet genome will undoubtedly open new perspectives for the dissection of the molecular mechanisms controlling its floral induction (Dohm et al., 2014).

VERNALIZATION SYSTEMS IN MONOCOTS

Vernalization responsiveness is also common in different species of monocots (Chouard, 1960; Brewster, 1987; Preston and Sandve, 2013). A few years ago, some key components of the system governing bulb formation and floral induction in onion (Allium cepa), a biennial species that belongs to the Asparagales order, were identified (Lee et al., 2013b). In the model proposed by Lee et al. (2013b), the initiation of bulbing and flowering are both controlled by FT-like genes. After planting in spring, high AcFT4 activity inhibits bulb formation by repressing AcFT1. Later in the season, inductive photoperiods down-regulate AcFT4, allowing the induction of the bulb-promoting AcFT1. During winter, vernalization leads to the up-regulation of another FT-like gene, AcFT2, which is necessary to promote flowering the next summer (Lee et al., 2013b). Further experiments are required to confirm and expand this model, but these preliminary results suggest a mechanism distinct from the vernalization system in Brassicales, Caryophyllales, and, as discussed below, Pooideae.

In Pooideae (temperate grasses), a grass subfamily that includes crown pooid crops such as wheat, oats, rye, and barley, allelic variation and functional studies have advanced the understanding of the molecular basis of vernalization. As in Arabidopsis, the timing of flowering is an important trait that is tightly controlled by genetic networks that integrate environmental cues, such as photoperiod and vernalization. After the identification of FLC in Arabidopsis, many efforts were directed toward the identification of its ortholog in vernalization-sensitive monocots, and a recent study identified ODDSOC2 as an ortholog in cereals (Ruelens et al., 2013). Although ODDSOC2 expression is suppressed by cold (Greenup et al., 2010), its specific role in the vernalization response is not clear. Instead, like in sugar beet, an FLC-independent pathway appears to be the major contributor to the vernalization requirement and response to prolonged cold exposure in winter cereal varieties.

Our current understanding is that the core regulatory mechanisms of vernalization in cereals includes three genes called VRN1, VRN2, and VRN3 (Yan et al., 2003, 2004, 2006). VRN3 is an ortholog of FT that interacts with an FD-like protein to trigger the expression of downstream targets, including the AP1/FUL ortholog VRN1 (Li and Dubcovsky, 2008; Li et al., 2015). VRN4, which resulted from a duplication of VRN1, displays a distinct expression pattern and appears to be associated with a weaker vernalization requirement (Kippes et al., 2015). As in Arabidopsis, the control of FT expression in leaves is key to the acquisition of the competence to flower in response to vernalization. Prior to cold, in winter cereal varieties, VRN1 is expressed at low levels, and the CONSTANS-like VRN2 gene plays a repressive role on FT (Fig. 1E; Danyluk et al., 2003; Yan et al., 2004; Chen and Dubcovsky, 2012). The role of VRN2 in conferring the vernalization requirement is supported by the fact that nonfunctional VRN2 alleles can confer a spring habit in diploid wheat (Triticum aestivum) and barley (Hordeum vulgare), and the development of a triple VRN2 mutant in hexaploid wheat also results in rapid flowering and a reduced vernalization response (Kippes et al., 2016). In cereals, prolonged cold temperatures cause the repression of VRN2 and a quantitative induction of VRN1, which directly binds to the FT promoter to trigger its expression and also to VRN2 to presumably repress its expression (Yan et al., 2004; Shimada et al., 2009; Deng et al., 2015). The induction of VRN1 by cold is associated with removal of H3K27me3 repressive marks and deposition of H3K4me3 activating marks (Oliver et al., 2009), and also possibly involves posttranscriptional mechanisms (Xiao et al., 2014). The initial down-regulation of VRN2 during the cold occurs independently of VRN1, but VRN1 is critical for the stable repression of VRN2 after cold exposure ends (Chen and Dubcovsky, 2012). When induced, FT reinforces the expression of VRN1, thus creating a positive feedback loop that ensures the transition to the reproductive stage (Li and Dubcovsky, 2008). Hence, vernalization results in decreased VRN2 expression and increased VRN1 expression, facilitating the induction of FT expression upon extension of the photoperiod.

The photoperiodic activation of FT involves the PRR PHOTOPERIOD1 (PPD1), which is responsible for a large part of the natural variation in flowering observed in wheat and barley; varieties with active PPD1 contain high levels of FT and flower early, whereas ppd1 varieties have low FT and flower late (Turner et al., 2005; Beales et al., 2007; Kitagawa et al., 2012; Shaw et al., 2012). How PPD1 controls FT expression in grasses is not known, but it possibly acts through the transcriptional regulation of CONSTANS (Turner et al., 2005; Shaw et al., 2012). More recently, the light receptor for the photoperiodic pathway was identified as PHYTOCHROME C (PHYC); phyC mutants flower very late under inductive photoperiods (Chen et al., 2014; Woods et al., 2014b).

Although there has been great progress in understanding vernalization at a molecular level from studying wheat and barley, much remains to be learned, including identifying additional components within the vernalization pathway and the extent to which vernalization pathways are conserved throughout the grasses. Pooideae is a diverse grass subfamily comprising ∼3,800 species with a large geographical range, predominantly in higher latitudes (Mannion, 1997; Grass Phylogeny Working Group, 2001). Recently it has been shown that vernalization responsiveness is widespread throughout Pooideae, although there have been several independent losses of a vernalization requirement (McKeown et al., 2016; Woods et al., 2016). Furthermore, studies in the small temperate grass model Brachypodium distachyon, which is an early diverging pooid sister to the crown pooid clade, have contributed to understanding the evolution of vernalization systems in pooids (Woods and Amasino, 2015). Like wheat and barley, B. distachyon is a long-day plant that exhibits an extensive natural variation of flowering behavior across accessions with respect to photoperiod and vernalization responses (Ream et al., 2014; Tyler et al., 2016) and has a “memory” of winter (Woods et al., 2014a). Additionally, B. distachyon contains orthologs of all of the VRN genes discovered in wheat and barley, and these genes also likely contribute to natural variation in flowering responses among different accessions of B. distachyon (Higgins et al., 2010; Woods et al., 2016, 2017; Bettgenhaeuser et al., 2017). Genetic and physiological characterizations carried out in different Pooideae species, including B. distachyon, confirms the conservation of VRN1 and VRN3/FT as promoters and VRN2 as a repressor of flowering (Lv et al., 2014; Ream et al., 2014; Woods et al., 2016). However, the cold- and VRN1-mediated repression of VRN2 is restricted to core Pooideae, such as wheat and barley, suggesting that the establishment of the VRN1-VRN2 loop occurred late in the diversification of temperate grasses (Woods et al., 2016). In contrast, the induction of VRN1 by cold evolved early in the diversification of Pooideae, and VRN1 is induced following cold exposure even under noninductive conditions (McKeown et al., 2016). Hence, whereas the memory of winter in Arabidopsis is controlled by the stable repression of the floral inhibitor FLC, this memory appears to rely, in large part at least, on the stable activation of the floral promoter VRN1 in Pooideae (Woods et al., 2014a).

CONCLUDING REMARKS

The exploration of the vernalization systems in different plant groups reveals distinct pathways that share a common principle: in vernalization-responsive species, there is a block to flowering (a vernalization requirement), and cold provides competence to flower (overcomes the block to flowering), whether it is via the suppression of a floral repressor (e.g. Arabidopsis) or the activation of a floral promoter (e.g. Pooideae). Moreover, the induction of flowering typically requires additional inductive signals. In most vernalization-requiring species, the transition to floral development upon perception of these signals may occur long after return to warmer growth temperatures, revealing a memory of the past winter. Other species, such as the Brassicaceae A. alpina, exhibit a transient response to cold temperatures, and this absence of winter memory is key to its perennial growth habit because only meristems that are mature at the start of the cold exposure will undergo floral transition, while the other meristems remain vegetative to maintain growth from year to year. Evolution of unique vernalization systems is key to the establishment of a species life history strategy. Further dissection of the vernalization pathways throughout angiosperm lineages is likely to uncover new mechanisms establishing the memory of winter (see “Outstanding Questions”) and shed light on the role of vernalization in the diversification of plant groups.