The central role of the VERNALIZATION1 gene in the vernalization response of cereals

Additional keywords: barley, flowering, wheat.

Plants growing in temperate regions time flowering to coincide with favourable seasonal conditions. Winter frost can damage cold-sensitive floral organs, whereas heat and water stress during summer can reduce fertility, so flowering often occurs in spring when conditions are optimal (see King and Heide 2009). One cue that promotes spring flowering is prolonged exposure to cold during winter, or vernalization. Vernalization occurs in many plants (see Chouard 1960; Amasino 2004; King and Heide 2009) but the focus of this review is the molecular mechanisms that control vernalization-induced flowering in economically important cereal crops such as wheat (Triticum aestivum L.) and barley (Hordeum vulgare L.).

In temperate regions, wheat and barley are sown in autumn then over-winter before flowering in spring. When sown in spring, these same cereal varieties typically show delayed flowering or fail to flower altogether. Several researchers recognised that the cold of winter is a critical factor required to trigger flowering of these plants and that this is lacking when plants are sown in spring (see McKinney 1940; Chouard 1960). For example, Gassner showed that germinating wheat or rye (Secale cereale M. Bieb.) seeds at normal growth temperatures can cause a strong delay of flowering, whereas germination at low temperatures can stimulate flowering (Gassner 1918). He concluded that many cereals have a requirement for cold, or “Kaltbedürfnis”, which must be satisfied to allow flowering (Gassner 1918). This phenomenon later came to be referred to as vernalization (vernalis meaning “pertaining to spring”; see Chouard 1960).

Vernalization promotes the transition to reproductive development at the shoot apex (Flood and Halloran 1984). Thus, after vernalization, the production of leaf primordia ceases and floral primordial appear at the shoot apex (Bonnet 1935, 1936; Zadoks et al. 1974). These develop into inflorescence branches that bear the florets (flowers) (Bonnet 1935, 1936; Zadoks et al. 1974). Vernalization is sufficient to trigger the transition to reproductive development, but long days are then required for rapid inflorescence development and stem elongation (Purvis 1934; Gott et al. 1955) (Fig. 1). Critically, vernalization is a prerequisite for the acceleration of flowering by long days, so long days cause rapid flowering only after plants have been vernalized (Purvis 1934). This combination of vernalization requirement and daylength sensitivity ensures that flowering is delayed until after winter to avoid frost damage but then occurs rapidly as daylength increases during spring, thereby avoiding heat and water stress during summer (Fig. 2). A similar combination of vernalization requirement and daylength sensitivity is found in many plants from temperate zones (see Thomas and Vince-Prue 1997; King and Heide 2009).

The effect of vernalization is cumulative, so increasing durations of cold accelerate flowering to greater extents until a point when the vernalization response is saturated (Gott et al. 1955). Acceleration of flowering by vernalization is also temperature dependent and, typically, there is an optimal temperature for vernalization between 0 and 10°C that will saturate the vernalization response more rapidly than warmer or colder temperatures (Gassner 1918; Chouard 1960). Thus, vernalization triggers a quantitative flowering response in cereals, and the effectiveness of vernalization is both time and temperature dependent.

When cells from vernalized wheat are used to regenerate plants through tissue culture, the resulting plants do not require vernalization to flower (Marcińska et al. 1995). Similarly, wheat cells can be vernalized during tissue culture to give rise to plants that flower rapidly with no further requirement for vernalization (Whelan and Schaalje 1992). Furthermore, maturing seeds on the spike (inflorescence) can be exposed to prolonged cold and then allowed to undergo the normal process of seed drying and germination to give rise to seedlings that flower without vernalization (Gregory and Purvis 1936). These observations suggest that cereals retain a cellular memory of vernalization. The idea that plants retain a memory of vernalization is consistent with the way vernalization promotes flowering: when germinating seeds are exposed to prolonged cold, there are no visible signs of floral development at the end of vernalization, but plants undergo rapid floral development when placed at normal growth temperatures after vernalization (Purvis 1934; Flood and Halloran 1984; Sasani et al. 2009). So exposure to cold is remembered and this exerts an after-effect during subsequent development (Chouard 1960). After plants flower, the memory of vernalization is presumably reset in seeds to allow the vernalization response to reoccur in the next generation.

The VERNALIZATION1 gene (VRN1) controls vernalization-induced flowering in cereals (see Trevaskis et al. 2007a; Distelfeld et al. 2009). VRN1 encodes a MADS box transcription factor (a class of transcription factor named after the archetypal genes MCM1, AGAMOUS, DEFICIENS, SRF1) related to genes that promote flowering in other plant species (Danyluk et al. 2003; Trevaskis et al. 2003; Yan et al. 2003). VRN1 transcripts are present at low basal levels but increase during prolonged cold treatment (Danyluk et al. 2003; Trevaskis et al. 2003; Yan et al. 2003). This response is quantitative, with longer cold treatments inducing higher transcript levels (Danyluk et al. 2003; Yan et al. 2003; von Zitzewitz et al. 2005; Sasani et al. 2009). This parallels the degree to which flowering is accelerated by increasing lengths of cold treatment (Sasani et al. 2009). Thus, it seems likely that VRN1 is a cold-activated promoter of flowering that mediates vernalization-induced flowering in cereals. The activity of VRN1 might be essential for flowering in cereals, since a mutant of Einkorn wheat (Triticum monococcum) that lacks VRN1 is unable to flower (Shitsukawa et al. 2007). However, a recent study shows that the non-flowering phenotype of this mutant might also be partly due to deletion of genes flanking VRN1 (Distelfeld and Dubcovsky 2010).

Vernalization activates expression of VRN1 in the leaf and shoot apex, and elevated expression of VRN1 is maintained in these tissues following vernalization (Yan et al. 2003; Sasani et al. 2009). VRN1 is likely to have distinct functions in each of these organs. At the shoot apex, expression of VRN1 is likely to promote the transition to reproductive development, so the production of leaves ceases and inflorescence development begins (Gocal et al. 2001; Loukoianov et al. 2005; Hemming et al. 2008; Preston and Kellogg 2008; Sasani et al. 2009) (Fig. 3). VRN1-like genes play similar roles in promoting phase transition in other plants, including Arabidopsis thaliana (L.) Heynh (Ferrándiz et al. 2000). In the leaves, expression of VRN1 unlocks the long-day flowering response, allowing increasing daylength to accelerate flowering in spring (Trevaskis et al. 2007a; Hemming et al. 2008; Preston and Kellogg 2008) (Fig. 3).

Studies in A. thaliana have shown that FLOWERING LOCUS T (FT) mediates the long-day flowering response (Kardailsky et al. 1999; Kobayashi et al. 1999). Expression of FT is controlled by interactions between the circadian clock and light receptors, such that the FT protein is produced in leaves when daylight extends into the late afternoon and exceeds a critical length (Suárez-López et al. 2001; An et al. 2004; Valverde et al. 2004). The FT protein is then transported to the shoot apex where it promotes floral development (Corbesier et al. 2007; Tamaki et al. 2007). Thus, FT is proposed to be the main component of the ‘florigen’, the leaf-derived signal that triggers flowering at the shoot apex (Zeevaart 2008).

In temperate cereals, the FT-like 1 (FT1) gene (also known as VRN3) seems to play a role similar to that of FT in A. thaliana, and is induced by long days to accelerate flowering (Turner et al. 2005; Yan et al. 2006; Faure et al. 2007). In vernalization-responsive cereals, FT1 is expressed only after plants are vernalized (Yan et al. 2006; Hemming et al. 2008). It seems that expression of VRN1 is required for long-day induction of FT1 (Hemming et al. 2008). VRN1 might promote expression of FT1 via regulatory interactions with VRN2 (Trevaskis et al. 2007a; Hemming et al. 2008). VRN2 is a repressor of flowering that is expressed in leaves in long days (Yan et al. 2004a; Dubcovsky et al. 2006; Trevaskis et al. 2006) and it seems likely that VRN2 delays flowering by suppressing long-day induction of FT1 (Hemming et al. 2008). VRN1 downregulates VRN2 (Loukoianov et al. 2005; Trevaskis et al. 2006), so downregulation of VRN2 by VRN1 provides a mechanism to allow long-day induction of FT1 in the leaves of vernalized plants (Hemming et al. 2008; Sasani et al. 2009) (Fig. 3). It is not known whether the VRN1 protein interacts directly with the VRN2 gene or whether other genes mediate this regulatory interaction.

Wheat and barley crops are often grown in regions with mild winter conditions where vernalization does not occur. In other regions, crops are sown in spring to avoid harsh winter conditions and so do not experience vernalization. To adapt crops to these regions, breeders have produced varieties that flower without vernalization (see Chouard 1960). Varieties of wheat and barley that flower without vernalization are referred to as spring types, whereas those that require vernalization to flower are termed winter types.

Genetic analyses have shown that VRN1 is a major determinant of the vernalization requirement in wheat and barley (Pugsley 1971; Takahashi and Yasuda 1971). Some alleles of VRN1 are expressed without prior cold treatment and these allow flowering without vernalization (Danyluk et al. 2003; Trevaskis et al. 2003; Yan et al. 2003). Increased expression of VRN1 from these ‘active alleles’ is associated either with insertions and deletions in the promoter (wheat) or with deletions in the first intron (wheat and barley) of the VRN1 gene (Yan et al. 2004b; Fu et al. 2005) (Fig. 4). Presumably, these regions contain sequences required to maintain low levels of VRN1 expression until plants are vernalized.

Comparison of the promoter regions from wild-type versus active alleles of VRN1 has identified a critical region near the transcriptional start site that is altered in some active alleles of VRN1 (Pidal et al. 2009). This region might contain a binding site for a repressor protein that is required to maintain VRN1 in an inactive state until plants are vernalized. Alternatively, the conformation of DNA in this region might limit the rate of transcriptional initiation, and mutations might remove this limit.

Studies in barley have shown that the first intron of VRN1 contains a broad region that is required for repression of VRN1 before winter. Around a dozen different alleles of VRN1 have been identified in barley (Fu et al. 2005; von Zitzewitz et al. 2005; Cockram et al. 2007; Szucs et al. 2007; Hemming et al. 2009). Some lack large sections of the ~11 kb first intron, while others lack smaller sections (Fu et al. 2005; von Zitzewitz et al. 2005; Cockram et al. 2007; Szucs et al. 2007; Hemming et al. 2009). Typically, alleles lacking larger sections are more active and are associated with earlier flowering without vernalization, whereas alleles lacking smaller segments are associated with only a moderate increase in VRN1 activity and weaker promotion of flowering (Szucs et al. 2007; Hemming et al. 2009). Comparison of the regions deleted in these different active alleles has identified a core region within the first intron that seems to be important for repression of VRN1 (Hemming et al. 2009). In addition to this core region, sections at the 5′ end of the intron (towards the promoter) also seem to play an important role in maintaining repression of VRN1, and alleles with large deletions that remove both the core region and the 5′ end of the intron typically have the highest activity (Szucs et al. 2007; Hemming et al. 2009). An insertion of a mobile genetic element in the 5′ end of the first intron is also associated with active expression of VRN1 without vernalization (Stockinger et al. 2007; Hemming et al. 2009).

The state of chromatin (the packaging of DNA and associated histones into higher order structures) appears to be an important determinant of VRN1 activity (Oliver et al. 2009). Analysis of histone modifications at the VRN1 locus in barley plants that have not been vernalized has shown that the chromatin at VRN1 contains a specific histone modification typically associated with an inactive chromatin state and long-term transcriptional repression: histone-3-lysine-27-trimethylation (H3K27Me3) (Oliver et al. 2009). This modification is found within the first intron of VRN1 and also at the promoter (Oliver et al. 2009). Deposition of H3K27Me3 at these sites might contribute to repression of VRN1 before winter. In other organisms, H3K27Me3 is deposited by the polycomb repressor complex 2 (PRC2) (Cao et al. 2002). The intron of VRN1 might contain sequences targeted by such a complex, explaining why this region is required to maintain repression of VRN1; however, the sequence motifs targeted by plant PRC2 complexes are not known.

When plants are vernalized, the level of H3K27Me3 at the VRN1 locus decreases and histone modifications typically associated with an active chromatin state appear (H3K4Me3), suggesting that vernalization induces a change in the state of chromatin at the VRN1 locus (Oliver et al. 2009). Chromatin modifications can be inherited through cell divisions, so these epigenetic modifications, and the associated changes in chromatin state, might allow VRN1 to remain active after vernalization. This could contribute to a cellular memory of vernalization in cereals (Fig. 5).

Prolonged exposure to cold activates expression of VRN1, but how this occurs is not known. The regulatory regions in the first intron are not required for cold induction of VRN1, since exposure to low temperatures can further activate expression of VRN1 alleles that lack most of the first intron (Trevaskis et al. 2007b; Hemming et al. 2009). So while the first intron is important to maintain low levels of VRN1 transcripts before winter, cold activation of VRN1 is probably mediated by other parts of the gene. Transient transformation experiments using tobacco leaves (Nicotiana benthamiana L.) suggest that fusion of a 1 kb region of the VRN1 promoter to a reporter gene (Green Fluorescent Protein) can mediate cold induction of reporter gene activity (Kane et al. 2007). This suggests that cold activation of VRN1 is controlled by regulatory elements in the promoter. Presumably these elements are targeted by a temperature response mechanism that is conserved in cereals (monocots: Poaceae) and tobacco (dicot: Solanaceae). Unfortunately the study of Kane et al. (2007) did not examine whether the increase in reporter gene activity seen after cold treatment resulted from increased transcription or simply from increased stability of the reporter protein, so the results of this experiment are inconclusive.

Some wheats and barleys carry alleles of FT1 (VRN3) that are expressed at high levels and cause rapid flowering without vernalization (Takahashi and Yasuda 1971; Yan et al. 2006). Other wheats and barleys lack a functional VRN2 gene and so flower rapidly in long day conditions, without any requirement for vernalization (Takahashi and Yasuda 1971; Dubcovsky et al. 1998; Yan et al. 2004a). In these genotypes, activation of the long-day flowering response without prior cold treatment overcomes the normal requirement for vernalization.

VRN2 and FT1 (VRN3) have different effects on flowering behaviour. Active alleles of FT1 cause rapid inflorescence initiation and development, irrespective of daylength or vernalization (Yan et al. 2006). Loss of VRN2 also causes rapid inflorescence initiation and development without vernalization, but only in long-day conditions (Karsai et al. 2005; Hemming et al. 2008). By comparison, active alleles of VRN1 mimic the effects of vernalization by promoting inflorescence initiation and allowing long days to accelerate subsequent stages of inflorescence development. The way each gene influences overall flowering behaviour determines how these genes are utilised to produce wheats and barleys that flower without vernalization. For example, active alleles of FT1 seem to be useful in regions with extreme winters and short summer growing seasons, where rapid flowering is beneficial regardless of growing conditions (Takahashi and Yasuda 1971).

VRN1 is expressed without prior vernalization in varieties that carry active alleles of FT1 (Yan et al. 2006). In A. thaliana, FT interacts with a ‘basic leucine zipper’ transcription factor, FD, to activate target genes that include MADS box transcription factors that are related to VRN1; FRUITFUL and APETALLA1 (Abe et al. 2005; Teper-Bamnolker and Samach 2005; Wigge et al. 2005; Corbesier et al. 2007). The wheat FT1 protein can interact with the FD-like protein 2 (TaFDL2), which can, in turn, bind to the promoter of VRN1 (Li and Dubcovsky 2008). Thus, in cereals, FT1 might activate expression of VRN1 in leaves and the shoot apex through interactions with FD-like proteins (Li and Dubcovsky 2008). Activation of VRN1 by FT1 might also occur in varieties that lack VRN2, where rapid induction of both FT1 and VRN1 occurs when plants are exposed to long days (Yan et al. 2006; Trevaskis et al. 2007b).

Although activation of FT1 can induce expression of VRN1 in some genotypes, the relevance of this to vernalization-induced flowering is unclear. In autumn-sown (vernalization-responsive) varieties, VRN1 is induced by cold during winter when days are short (Danyluk et al. 2003; von Zitzewitz et al. 2005; Trevaskis et al. 2006) and FT1 is not expressed (Hemming et al. 2008). Similarly, vernalization induces expression of VRN1 in seeds germinated in darkness where FT1 is not expressed (Sasani et al. 2009). So FT1 is unlikely to mediate induction of VRN1 by cold. Instead, long-day induction of VRN1 might occur in spring if plants are partially vernalized – after mild winters, for example – and this could compensate for incomplete vernalization. Similarly, long-day activation of VRN1 might allow some varieties to flower eventually even if plants do not experience vernalization. Activation of VRN1 by long days might also be important after the transition to reproductive development, since long days can enhance expression of VRN1 in leaves after the initiation inflorescence (Sasani et al. 2009). This might occur through a positive feedback mechanism whereby FT1 enhances expression of VRN1, which further activates FT1 (Fig. 3). This might strengthen the long-day flowering response once inflorescence development begins.

Different alleles of VRN1 can generate a wide spectrum of vernalization requirements and have been used to adapt varieties to a wide range of growing regions and sowing times (Pugsley 1971; Takahashi and Yasuda 1971). Preliminary surveys of genetic diversity at the VRN1 locus have shown how breeders have utilised particular alleles of VRN1 to breed cultivars adapted to specific regions. For example, an active allele of VRN1 on the D genome is common in wheats from some growing regions and might offer an adaptive benefit relative to other alleles (Zhang et al. 2008). Similarly, some alleles of VRN1 gene are prevalent in European barleys (Cockram et al. 2007; Hemming et al. 2009).

The prevalence of an allele within a breeding program might indicate an adaptive benefit of this allele in a particular region, but might also reflect breeding history, since a founding variety or recurrent parent can bias breeding material towards a particular VRN1 genotype. Therefore, genetic surveys alone might not identify the optimal VRN1 genotype for a particular region. To determine which VRN1 alleles are most suitable for use in Australian wheat and barley varieties, we used DNA diagnostic techniques to identify different alleles of VRN1 amongst diverse cultivars, breeding lines and landraces of wheat (3000 accessions) and barley (4400 accessions) (Hemming et al. 2009; S. Fieg and B. Trevaskis, unpubl. data). The different alleles thus identified have been introgressed into elite wheat and barley varieties, and are undergoing recurrent backcrossing to develop near-isogenic lines that vary for VRN1 genotype but otherwise have similar genetic backgrounds. These can be tested for performance in different regions with different sowing times, and will provide valuable information to Australian cereal breeders.

An alternative approach being used to identify alleles of VRN1 that are suited to different Australian growing regions is to identify VRN1 alleles that are common among Australian cereal breeding programs and then analyse the association of these alleles with crop performance among diverse materials grown in different regions over multiple years. This approach makes good use of existing pedigree and phenotype data, and is a powerful way to identify favourable alleles of VRN1 within existing breeding material. This strategy is currently being used to identify useful alleles of VRN1 among Australian wheat breeding material (see Eagles et al. 2009).

Autumn-sown wheats and barleys, which require vernalization to flower, can acclimatise to low temperatures and are subsequently able to survive freezing conditions during winter. This process is known as cold acclimation (reviewed in Galiba et al. 2009). The longer plants are exposed to low temperatures, the more frost tolerance increases. This continues until the point when further cold treatment has no additional impact on flowering time (the vernalization saturation point) then frost tolerance begins to decrease (Vasiljev 1934 cited by McKinney 1940; Roberts 1979; Mahfoozi et al. 2001; Prasil et al. 2004). Thus, it seems that cold acclimation and the vernalization response are interconnected.

Activation of VRN1 might cause the decrease in frost tolerance that begins when plants are fully vernalized. This hypothesis is supported by genetic studies that show active alleles of VRN1 reduce frost tolerance (Roberts 1990; Hayes et al. 1993; Fowler et al. 1996; Koemel et al. 2004; Limin and Fowler 2006). For example, a comparison of near-isogenic lines that differ in VRN1 genotype has shown that varieties with active alleles of VRN1, which flower without vernalization, have a greatly reduced capacity to acclimate to cold compared with lines with wild-type alleles of VRN1, which require vernalization to flower (Koemel et al. 2004; Limin and Fowler 2006). At present, it is not clear whether this is a direct consequence of induction of VRN1 or an indirect consequence, caused by the effect that VRN1 has on other genes or on plant development per se. Regardless of the mechanism, this relationship between VRN1 activity and frost tolerance is important and has implications for cereal breeders, since altering vernalization requirement can also affect winter survival.

Vernalization occurs in a wide range of plants. Genes controlling vernalization-induced flowering have now been isolated from temperate cereals (monocots: Poaceae) and the model plant A. thaliana (dicot: Brassicaceae). In A. thaliana, the vernalization response is mediated by FLOWERING LOCUS C (FLC) (Michaels and Amasino 1999; Sheldon et al. 1999). FLC encodes a MADS box floral repressor that is downregulated by vernalization (Michaels and Amasino 1999; Sheldon et al. 1999). This contrasts with the situation in temperate cereals, where no FLC-like genes occur and instead VRN1 is activated by vernalization to promote flowering (see Trevaskis et al. 2007a; Greenup et al. 2009). It seems that the vernalization response has evolved independently in these distantly related angiosperms.

Although different genes control the vernalization response in cereals and A. thaliana, there are parallels between the pathways controlling vernalization-induced flowering in these plants. For example, vernalization is a prerequisite for long-day induction of both FT and FT1 (Michaels and Amasino 1999; Sheldon et al. 1999; Michaels et al. 2005; Yan et al. 2006; Hemming et al. 2008). In cereals, VRN2 downregulates FT1 before winter, directly or indirectly (Yan et al. 2006; Hemming et al. 2008), and in A. thaliana, FLC interacts directly with the FT gene sequence to repress transcription (Helliwell et al. 2006). Furthermore, in both cereals and in A. thaliana, the state of chromatin at key MADS box transcription factors plays an important role in regulating the vernalization response. The repressive chromatin mark H3K27Me3 is required for repression of VRN1 before winter in cereals and for repression of FLC after winter in A. thaliana (Schubert et al. 2006; Sung et al. 2006; Wood et al. 2006; Finnegan and Dennis 2007; De Lucia et al. 2008; Oliver et al. 2009). These parallels are probably the result of convergent evolution.

One key area for further research is in understanding how plants sense cold and how increasing durations of cold cause increased expression of VRN1. Another key question is how VRN1 promotes flowering. More specifically, what are the direct targets of VRN1? By using molecular tools such as reporter gene fusions, microarrays and targeted mutagenesis, it should be possible to address these questions. In this regard, barley is a useful genetic model, since it is transformable and large collections of genetically diverse barley accessions exist in cereal seed banks. There are also many mapping populations, including recombinant inbred lines and doubled haploids, which can be used to identify potential regulators of VRN1.

The activity of VRN1 underlies many of the physiological features of vernalization-induced flowering in cereals, so understanding how VRN1 is regulated will provide further insights into the biology of the vernalization response in these plants. A better understanding of the regulation and molecular functions of VRN1 will also have important implications for cereal breeding programs. It should eventually be possible to predict with precision how different alleles of VRN1 will influence flowering behaviour in different genetic backgrounds or different environments, and to tailor breeding strategies accordingly. Knowledge developed in temperate cereals will also be relevant to related grasses, including a range of economically important pasture grasses. For these reasons, the vernalization response of cereals, the plants where vernalization-induced flowering was first recognised, will be an important area for ongoing research.