Low-temperature and daylength cues are integrated to regulate FLOWERING LOCUS T in barley.

Interactions between flowering time genes were examined in a doubled haploid barley (Hordeum vulgare) population segregating for H. vulgare VERNALIZATION1 (HvVRN1), HvVRN2, and PHOTOPERIOD1 (PPD-H1). A deletion allele of HvVRN2 was associated with rapid inflorescence initiation and early flowering, but only in lines with an active allele of PPD-H1. In these lines, the floral promoter FLOWERING LOCUS T (HvFT1) was expressed at high levels without vernalization, and this preceded induction of HvVRN1. Lines with the deletion allele of HvVRN2 and the inactive ppd-H1 allele did not undergo rapid inflorescence initiation and were late flowering. These data suggest that HvVRN2 counteracts PPD-H1 to prevent flowering prior to vernalization. An allele of HvVRN1 that is expressed at high basal levels (HvVRN1-1) was associated with rapid inflorescence initiation regardless of HvVRN2 or PPD-H1 genotype. HvFT1 was expressed without vernalization in lines with the HvVRN1-1 allele and HvFT1 transcript levels were highest in lines with the active PPD-H1 allele; this correlated with rapid apex development postinflorescence initiation. Thus, expression of HvVRN1 promotes inflorescence initiation and up-regulates HvFT1. Analysis of HvVRN1 expression in different genetic backgrounds postvernalization showed that HvVRN2, HvFT1, and PPD-H1 are unlikely to play a role in low-temperature induction of HvVRN1. In a vernalization responsive barley, HvFT1 is not induced by low temperatures alone, but can be induced by long days following prolonged low-temperature treatment. We conclude that low-temperature and daylength flowering-response pathways are integrated to control expression of HvFT1 in barley, and that this might occur through regulation of HvVRN2 activity.

Flowering of temperate cereals, such as bread wheat (Triticum aestivum) and barley (Hordeum vulgare), is accelerated by prolonged exposure to cold, or vernalization. Vernalization occurs during winter when plants are exposed to temperatures between zero and 10°C (Takahashi and Yasuda, 1971;Flood and Halloran, 1984). Vernalization accelerates flowering by promoting inflorescence initiation, the first step in the transition of the shoot apex to reproductive development (Flood and Halloran, 1984). Temperate cereal varieties differ in their requirement for vernalization. Some varieties have an obligate requirement for vernalization, whereas others flower early without vernalization. Three genes have been reported to determine the requirement for vernalization in the temperate cereals; VERNALIZATION1 (VRN1), VRN2, and VRN3 (Pugsley, 1970;Takahashi and Yasuda, 1971;Dubcovsky et al., 1998). Gene se-quences for these three loci have been identified (Danyluk et al., 2003;Trevaskis et al., 2003;Yan et al., 2003Yan et al., , 2004Yan et al., , 2006. VRN1 encodes a MADS-box transcription factor that is essential for flowering (Danyluk et al., 2003;Trevaskis et al., 2003;Yan et al., 2003;Shitzukawa et al., 2007). In varieties that require vernalization to flower, VRN1 is expressed at low basal levels and is induced by exposure to low temperatures (Danyluk et al., 2003;Murai et al., 2003;Trevaskis et al., 2003;Yan et al., 2003). Some varieties of wheat and barley carry alleles of VRN1 that are expressed at high basal levels without vernalization. These alleles of VRN1 reduce or remove the requirement for vernalization and are dominant to alleles that are expressed at low basal levels without vernalization (Danyluk et al., 2003;Trevaskis et al., 2003;Yan et al., 2003). In barley, dominant alleles of H. vulgare VRN1 (HvVRN1) have deletions of regions in the first intron, which are presumably essential for maintaining low levels of HvVRN1 expression in plants that have not been vernalized (Fu et al., 2005).
VRN2 encodes a protein with a zinc finger motif and a CCT (CONSTANS, CONSTANS-LIKE, and TIMING OF CAB1-1) domain (Yan et al., 2004). VRN2 is a repressor of flowering and plants that lack a functional copy of VRN2 flower early (Yan et al., 2004). When plants are vernalized in long days, expression of VRN2 decreases, whereas expression of VRN1 increases (Yan et al., 2004). On the basis of this expression pattern, it was suggested that VRN2 represses VRN1, and that vernalization represses VRN2, permitting induction of VRN1 (Yan et al., 2004). Yan et al. (2004) proposed that VRN2 is the central regulator of the vernalization response, a role analogous to that of FLOWERING LOCUS C (FLC) in Arabidopsis (Arabidopsis thaliana; Michaels and Amasino, 1999;Sheldon et al., 1999Sheldon et al., , 2000. However, daylength is the major determinant of VRN2 expression levels, with high expression occurring in long days Trevaskis et al., 2006). When barley plants are vernalized in short days, expression of HvVRN2 stays at a similar low level throughout vernalization but expression of HvVRN1 is induced (Trevaskis et al., 2006). As vernalization typically occurs during winter, when days are short, it is unlikely that VRN2 plays a major role in regulating the response to low temperatures during winter (Trevaskis et al., 2006).
VRN3 has been mapped to a known flowering time gene, H. vulgare FLOWERING LOCUS T (HvFT1), a homolog of the FLOWERING LOCUS T gene (FT) of Arabidopsis (Kobayashi et al., 1999;Kardailsky et al., 1999Faure et al., 2007). FT is a mobile promoter of flowering; FT protein is produced in leaves when Arabidopsis plants are exposed to long days and is transported to the shoot apex where it promotes floral development (Corbesier et al., 2007). Similarly, in barley HvFT1 expression is induced by long days and may mediate the long-day flowering response (Turner et al., 2005). This requires an active version of another gene, PHOTOPERIOD1 (PPD-H1; Turner et al., 2005). In varieties that require vernalization, HvFT1 is expressed at low levels but can be induced by low temperatures (in long days) and it has been suggested that this plays a role in the vernalization response in cereals . In varieties that carry dominant alleles of VRN3, which cause early flowering and remove the requirement for vernalization, HvFT1 is expressed at elevated levels without cold treatment . The association between dominant alleles of VRN3 and elevated HvFT1 expression levels, combined with tight genetic linkage, suggest that activation of HvFT1 might be the molecular basis for dominant alleles of VRN3 . Mutations in the first intron of HvFT1 might be the basis for these high expression levels . It is not clear how daylength and low-temperature pathways are integrated to regulate HvFT1 expression.
In this article we show that loss of HvVRN2 allows rapid inflorescence initiation only when an active allele of PPD-H1 is present, and this is correlated with elevated HvFT1 expression levels. In contrast, HvVRN1 accelerates inflorescence initiation independently of PPD-H1 activity, and low-temperature induction of HvVRN1 does not require HvVRN2 or PPD-H1.

Genetic Interactions between HvVRN2 and PPD-H1 Influence Flowering Time in Barley
We examined flowering time (head emergence) in a Sloop 3 Halcyon doubled haploid population (Read et al., 2003). This population segregates for alleles of HvVRN1 and HvVRN2 (Read et al., 2003, Trevaskis et al., 2006 Table I). Lines that inherit both HvVRN1 and HvVRN2 alleles from the parent Halcyon are predicted to require vernalization and be late flowering in the absence of prolonged exposure to cold. The two loci are not linked, so the doubled haploid population (homozygous at all loci; Jain et al., 1996) is predicted to have a 3:1 ratio of early:late flowering in the absence of vernalization. When grown under long days in field experiments (Boyd et al., 2003) 56 of the 131 lines examined were late flowering (Supplemental Fig. S1). This result deviates significantly from the expected ratio of 3:1 (early:late; x 2 , 21.6; see Supplemental Data S1), due to the larger-than-expected number of late-flowering lines. All the lines with the HvVRN1-1 allele, which is expressed at high basal levels, flowered early, suggesting that expression of HvVRN1 can activate flowering regardless of genotypes at other loci segregating in this population (Supplemental Fig. S1). In contrast, 22 of the 50 lines deleted for HvVRN2 (HvVRN1;DHvVRN2), which were expected to be early flowering, flowered late (Supplemental Fig.  S1). This suggests that alleles of another unlinked locus are segregating in the Sloop 3 Halcyon population and affect flowering time in lines that lack HvVRN2.
A third major quantitative trait locus affecting flowering time in long days mapped to a region containing PPD-H1 (chromosome 2H, unlinked to HvVRN1 on chromosome 5H or HvVRN2 on chromosome 4H) in the Full-length HvVRN1 gene, which is expressed at low basal levels.
Morex allele (AY785815), with a deletion in the first intron, which is expressed at high basal levels. HvVRN2 genotype HvVRN2 DHvVRN2:

Description
HvZCCTa and HvZCCTb genes present.
HvVRN2 locus is deleted. PPD-H1 genotype ppd-H1 PPD-H1 Description An allele with loss-of-function mutation in CCT domain, which reduces daylength sensitivity.
Wild-type gene sequence.

HvFT1 genotype
HvFT1-H HvFT1-S Description Intron sequence previously associated with recessive alleles of vrn3 and low levels of HvFT1 expression.
Intron sequence previously associated with dominant alleles of VRN3 and high levels of HvFT1 expression.
Sloop 3 Halycon population (Read et al., 2003). We used a molecular marker (see ''Materials and Methods'') to genotype the Sloop 3 Halcyon population for PPD-H1. This showed that Halcyon carries an inactive allele of PPD-H1 (ppd-H1) previously identified in daylength insensitive varieties (Turner et al., 2005) whereas Sloop carries an active allele (Table I). In lines deleted for HvVRN2 there was a complete association between PPD-H1 genotype and flowering time (Supplemental Fig. S1); lines with an active allele of PPD-H1 flowered early, whereas lines with the inactive ppd-H1 allele were late flowering. This explains the deviation from the expected ratio of 3:1 (early:late) in the Sloop 3 Halcyon population.
Loss of HvVRN2 Causes Rapid Inflorescence Initiation and Increased HvFT1 Expression in Plants with the Active PPD-H1 Genotype The interaction between HvVRN2 and PPD-H1 was examined further in controlled conditions. Ten lines from the Sloop 3 Halcyon population, in which the HvVRN2 locus is deleted (HvVRN1;DHvVRN2), were grown in long days until flowering occurred (head emergence). Two flowering-time classes were observed, and these correlated with the PPD-H1 genotype (Fig.  1A). Plants with the active PPD-H1 allele flowered early, whereas those with the inactive ppd-H1 allele flowered 4 weeks later. The early flowering phenotype of plants with the active PPD-H1 allele was caused by rapid shoot apex development, as indicated by apex length (Fig. 1B) and inflorescence initiation occurred within 13 d in all plants examined (Fig. 1C). In contrast, plants carrying the inactive ppd-H1 allele remained vegetative at the same time points (Fig. 1C). These results show that deletion of HvVRN2 promotes rapid inflorescence initiation only when combined with an active allele of PPD-H1.
Expression of HvFT1 was not detected in plants that carry HvVRN2 (HvVRN1;HvVRN2; see Fig. 4, D and E), consistent with previous reports that HvFT1 is expressed at low levels in vernalization requiring varieties . HvFT1 expression was also very low in plants deleted for HvVRN2 and carrying the inactive ppd-H1 allele ( Fig. 1, D and E). In contrast, HvFT1 expression was detected in 7-d-old plants in lines carrying the active PPD-H1 allele (Fig. 1,D and E). This suggests that HvVRN2 normally downregulates HvFT1 and counteracts activation of HvFT1 expression by PPD-H1. PPD-H1 expression levels were not correlated with HvVRN2 genotype (Supplemental Fig. S2). This suggests that HvVRN2 does not counteract activation of HvFT1 by PPD-H1 simply by repressing PPD-H1 expression. Expression of HvFT1 preceded any visible change in the morphology of the shoot apex ( Fig. 1, C-E). HvVRN1 was not expressed at high levels in these plants at this early stage of development (7 d), but was expressed at high levels once the shoot apex progressed to inflorescence initiation (14 d; Fig. 1, C-E). These data suggest that increased expression of HvFT1 accelerates inflorescence initiation in lines with the HvVRN2 deletion allele and the active PPD-H1 allele, whereas HvVRN1 might only be induced as a consequence of increased HvFT1 expression or rapid floral development.

Overexpression of HvVRN2 Down-Regulates HvFT1 and Delays Flowering
The hypothesis that HvVRN2 down-regulates HvFT1 was tested by overexpressing HvVRN2 in transgenic Figure 2. Overexpression of HvVRN2 down-regulates HvFT1 and delays flowering. A, Heading date (average days to head emergence) in long-day glasshouse conditions for transgenic plants homozygous for the HvVRN2 overexpression construct (Ubi:VRN2) and sibling null controls (Null) in the T2 generation. Asterisks indicate P values of Student's t tests: ***, P , 0.001. B, Development of the shoot apex in transgenic plants homozygous for the HvVRN2 overexpression construct (Ubi:VRN2) and sibling null controls (Null), in the T2 generation grown in long-day glasshouse conditions. Apices were sampled at various time points (days). DR, Double ridge, the first visible marker of inflorescence initiation. C, Relative expression levels of HvVRN2, HvVRN1, and HvFT1 assayed by qRT-PCR and normalized to ACTIN in RNA from a transgenic plant homozygous for the HvVRN2 overexpression construct (Ubi:VRN2) and a sibling null control (Null), in the T2 generation grown in long-day glasshouse conditions. Error bars represent SE of triplicate reactions. ND, No expression was detected; NS, no significant difference between lines according to Student's t tests.
[See online article for color version of this figure.] barley plants. A genomic copy of the HvVRN2 gene was expressed under the control of the constitutive maize (Zea mays) UBIQUITIN promoter in the barley cultivar Golden Promise. Golden Promise is deleted for HvVRN2 and carries the HvVRN1-1 allele. This allele has a deletion in the first intron of the HvVRN1 sequence and is expressed at high basal levels (see Table I), but the putative MADS-box binding site in the promoter is intact and expression of the HvVRN1-1 allele is induced by low temperatures (Trevaskis et al., 2007a). A single transgenic line was obtained, and in the descendents of this line a late-flowering phenotype segregated with the transgene. In the T2 generation, plants homozygous for the overexpression construct flowered 4 weeks later than sibling null controls ( Fig.  2A). Inflorescence initiation was delayed by less than a week (Fig. 2B), so the delay in flowering observed was due mainly to slower development following inflorescence initiation. In 14-d-old seedlings, where the stage of apex development is similar in overexpression lines and in sibling null controls (Fig. 2B), HvFT1 expression levels in both the T1 and T2 generations were significantly lower in plants overexpressing HvVRN2 than in control lines (Fig. 2C). HvVRN1 expression levels did not differ between plants overexpressing HvVRN2 and sibling null control lines. Figure 3. Expression of HvVRN1 promotes rapid inflorescence initiation independently of PPD-H1 genotype. A, Heading date (average days to head emergence) of doubled haploid lines with an allele of HvVRN1 that is expressed to high levels without vernalization (HvVRN1-1;HvVRN2) grown in long-day glasshouse conditions. A, Active PPD-H1 allele. I, Inactive ppd-H1 allele. Error bars represent SE. B, Average length of the shoot apex in 14-d-old plants from the same lines. C, The shoot apices of representative lines carrying PPD-H1 or ppd-H1 alleles at different time points (days). DR, Double ridge, the first visible marker of inflorescence initiation. D, Expression levels of HvFT1, HvVRN1, and HvVRN2 assayed by RT-PCR in 14-d-old plants (403 cycles). Expression of ACTIN is shown as a loading comparison (253 cycles). E, Relative expression levels of HvFT1 and HvVRN1 in plants with the HvVRN1-1 allele but with different PPD-H1 genotypes and relative expression levels of HvVRN2 in nonvernalized plants with the HvVRN1 allele and the HvVRN1-1 allele. RNA from was pooled from five PPD-H1 lines or five ppd-H1 lines (same lines as in D), then expression levels were assayed by qRT-PCR and normalized to ACTIN.   Ten lines from the Sloop 3 Halcyon population carrying an allele of HvVRN1 that is expressed at high basal levels (HvVRN1-1;HvVRN2) were grown in long days. These lines showed inflorescence initiation within 16 d and took an average of 48 d to flower (Fig. 3, A and C). Flowering occurred approximately 14 d earlier in lines carrying the active PPD-H1 allele than in lines with the inactive ppd-H1 allele (Fig. 3A). This was due mainly to accelerated development after inflorescence initiation, as the timing of inflorescence initiation differed by less than 3 d between lines with the different PPD-H1 genotypes (Fig. 3, B and C). In 14-d-old plants, HvFT1 was expressed at higher levels in lines carrying the active PPD-H1 allele than in lines with the inactive ppd-H1 allele (Fig. 3, D and E), and this may account for the differences in apex development and flowering time. Expression of HvVRN2 was repressed in lines expressing HvVRN1 (Fig. 3, D and E). These data show that expression of HvVRN1 is sufficient to repress HvVRN2 and promote inflorescence initiation, and that this occurs regardless of PPD-H1 genotype.
Ten late-flowering lines from the Sloop 3 Halcyon population (HvVRN1;HvVRN2) were vernalized and transferred to long days. Flowering occurred approximately 8 weeks earlier than in nonvernalized plants (Fig. 4A). Two flowering times were observed among the vernalized plants and these correlated with PPD-H1 genotype (Fig. 4A). Apex development was more rapid and flowering occurred on average 16 d earlier in plants carrying the active PPD-H1 allele than in plants carrying the inactive ppd-H1 allele (Fig. 4, A-C). In all 10 lines, HvVRN1 and HvFT1 were induced by vernalization and HvVRN2 was repressed, regardless of PPD-H1 genotype (Fig. 4, D and E). HvFT1 expression levels in vernalized plants did correlate with the PPD-H1 genotype (Fig. 4, D and E) and this may account for the differences in apex development and flowering time. Lines that lack HvVRN2 and are daylength insensitive (HvVRN1;DHvVRN2;ppd-H1), which are late flowering without vernalization were also vernalized. Vernalization accelerated apex development in lines with this genotype, and head emergence occurred approximately 2 weeks earlier than in nonvernalized plants (Fig. 5, A-C). The flowering time of vernalized plants deleted for HvVRN2 (average of 60 d) was similar to that of vernalized plants carrying HvVRN2 (average of 57 d), so HvVRN2 genotype has little influence on flowering time in vernalized plants. HvVRN1 expression was induced by vernalization in all these lines (Fig. 5D). This shows that HvVRN1 is induced by prolonged exposure to low temperatures in the absence of HvVRN2, and this may account for the acceleration of flowering in these lines. There was variation in HvFT1 expression levels in vernalized plants deleted for HvVRN2 (Fig. 5D), suggesting that segregation of another locus might influence HvFT1 expression when plants with this genotype are vernalized.

HvFT1 Does Not Influence the Vernalization Requirement in the Sloop 3 Halcyon Population and Expression of HvFT1 Does Not Increase during Low-Temperature Treatment in Short Days
Single nucleotide changes in the first intron of HvFT1 may provide the molecular basis for dominant alleles of VRN3 that reduce the vernalization requirement . The sequence of the HvFT1 first intron from Sloop is identical to that isolated from barley varieties that have dominant alleles of VRN3 that reduce the vernalization requirement, whereas the HvFT1 first intron sequence from Halcyon is identical to that found in barley varieties that require vernalization (Supplemental Fig. S3). In the Sloop 3 Halcyon population there was no association between these different HvFT1 sequences and vernalization requirement (Fig. 6), and HvFT1 was not expressed in nonvernalized plants, regardless of which HvFT1 sequence was present (Fig. 4, D and E). Consistent with these observations, no quantitative trait loci for flowering time have been associated with the VRN3 region in the Sloop 3 Halcyon population (Read et al., 2003).
It has been suggested that expression of HvFT1 is induced by low temperatures, and that this plays a role in the vernalization response in cereals . Expression of HvFT1 was examined in a vernalization-requiring, daylength-sensitive barley variety (Sonja HvVRN1;HvVRN2;PPD-H1). In nonvernalized plants grown in short days, expression levels of HvFT1 were below the limit of detection by quantitative real-time PCR (qRT-PCR). There was no detectable increase in HvFT1 expression when these plants were exposed to long days (Fig. 7, A and B). . HvFT1 is not induced by low temperatures alone. A, Relative expression levels of HvFT1 assayed by qRT-PCR and normalized to ACTIN in RNA from nonvernalized plants (Sonja, minus roots) that were grown for 2 weeks in short days then the following: maintained in short days for 2 weeks (SD), shifted to long days for 2 weeks (LD), vernalized for 9 weeks in short days and harvested directly from vernalization chamber (SDV), vernalized in short days for 9 weeks then returned to glasshouse temperatures for 2 weeks in short days (V1SD), and vernalized in short days for 9 weeks then returned to glasshouse temperatures in long days for 2 weeks (V1LD). The shoot apex of all plants was vegetative. Error bars represent SE of triplicate reactions. B, Expression levels of HvVRN1 and HvFT1 assayed by RT-PCR in the same samples (403 cycles). Expression of HvVRN1 is shown as a positive control for the vernalization treatment. Expression of ACTIN is shown as a loading comparison (253 cycles). Expression of HvFT1 also remained low in plants that were maintained in long days without vernalization (Supplemental Fig. S4). To test whether HvFT1 is induced by low-temperature, HvFT1 expression levels were monitored at weekly intervals in plants grown at vernalizing temperatures in short days. HvFT1 expression remained below the limit of detection throughout the 9-week vernalization treatment. Following vernalization, expression of HvFT1 remained low in plants that were maintained in short days, but was induced when vernalized plants were exposed to long days (Fig. 7, A and B). These data show that HvFT1 is not induced by cold per se, but by long days when the vernalization requirement has been satisfied.

DISCUSSION
By analyzing flowering time in a doubled haploid barley population that segregates for genes controlling vernalization requirement and daylength sensitivity, we found that a deletion of HvVRN2 causes early flowering only when an active allele of PPD-H1 is present. This genetic interaction between HvVRN2 and PPD-H1 has not been described previously and has implications for our understanding of how different genetic pathways regulate flowering time in cereals.

Deletion of HvVRN2 Allows Expression of HvFT1 to Induce Flowering without Vernalization
Vernalization accelerates flowering by promoting inflorescence initiation, the transition to reproductive development at the shoot apex (Flood and Halloran, 1984). Our data show that deletion of HvVRN2 can bypass the vernalization requirement and promote rapid inflorescence initiation, but only when combined with the active PPD-H1 genotype. PPD-H1 mediates long-day induction of HvFT1 (Turner et al., 2005), and lines that combine deletion of HvVRN2 with an active allele of PPD-H1 have elevated HvFT1 expression levels. This suggests that HvVRN2 down-regulates expression of HvFT1 and counteracts PPD-H1. Thus, HvVRN2 regulates flowering by repressing the activity of the long-day response pathway. This proposed role is consistent with HvVRN2 being expressed only in long days Trevaskis et al., 2006), and explains why lines that lack HvVRN2 flower early only in long days (Karsai et al., 2005). HvVRN2 may interact directly with components of the longday response pathway to regulate HvFT1 expression. HvVRN2 has a CCT domain, a type of domain known to interact with CCAAT binding factors (Ben-Naim et al., 2006). In Arabidopsis a CCAAT binding factor, HAP3b, has been shown to up-regulate FT and promote flowering in long days (Cai et al., 2007). HvVRN2 might interact negatively with a HAP3b-like factor in cereals to repress HvFT1 expression. If PPD-H1, which also contains a CCT domain, interacts with the same HAP3b-like factor to enhance HvFT1 expression this could explain the antagonistic relationship between HvVRN2 and PPD-H1.
Both FT1 and VRN1 are expressed at elevated levels in early flowering wheats that lack VRN2 (Yan et al., 2003. The strong influence of PPD-H1 (a known regulator of HvFT1) on flowering time in barley lines that lack HvVRN2 suggests that elevated expression of HvFT1 is the primary cause of early flowering in lines that lack HvVRN2. Expression data support this conclusion. HvFT1 is expressed at high levels during early stages of vegetative development in barley plants that lack HvVRN2 and have the active PPD-H1 allele, whereas expression of HvVRN1 increases around the time of inflorescence initiation. Deletion of HvVRN2 may indirectly cause increased HvVRN1 expression by promoting floral development because HvVRN1 expression is induced during inflorescence development (Trevaskis et al., 2007a) and is required for inflorescence initiation (Shitzukawa et al., 2007). Consistent with this idea, HvVRN1 expression levels are not elevated in lines that combine the inactive ppd-H1 allele with deletion of HvVRN2, where reproductive development of the shoot apex is delayed. HvVRN1 might also be induced in leaves as a consequence of HvFT1 induction. In Arabidopsis, activation of FT leads to induction of FRUITFUL (the Arabidopsis ortholog of VRN1) in leaves (Corbesier et al., 2007). In summary, HvVRN2 is likely to delay flowering by down-regulating expression of HvFT1. Figure 8. Interactions between genetic pathways controlling seasonal induction of flowering in temperate cereals. VRN2 represses FT1 to counteract PPD1 dependent long-day induction of FT1 prior to winter. Prolonged exposure to low temperatures up-regulates VRN1, which promotes inflorescence meristem identity at the shoot apex, accelerating inflorescence initiation independently of the long-day response pathway. VRN1 also represses VRN2 in the leaves to allow the long-day induction of FT1. FT1 can further accelerate inflorescence initiation and also subsequent stages of inflorescence development. According to this model, VRN1 acts in a low-temperature response pathway, FT1 acts in a daylength response pathway, and VRN2 integrates the lowtemperature and daylength responses. In varieties where VRN2 is deleted, or where dominant VRN3 alleles are present, increased FT1 expression accelerates inflorescence initiation. This induces expression of VRN1 (dotted line).

HvVRN1 Promotes Inflorescence Initiation and Represses
HvVRN2 to Allow Expression of HvFT1 The HvVRN1-1 allele, which is expressed at high basal levels, accelerates inflorescence initiation and promotes flowering. Increased HvVRN1 expression probably promotes inflorescence initiation by promoting inflorescence meristem identity at the shoot apex. Expression of HvVRN2 was repressed in lines that carry HvVRN1-1 (as reported previously; Trevaskis et al., 2006). Consistent with the function proposed for HvVRN2, expression of HvFT1 was elevated in these lines and was highest in lines carrying the active allele of PPD-H1. Taken together, the molecular phenotypes of plants carrying the HvVRN1-1 allele are very similar to those of vernalized plants. This shows that, unlike deletion of HvVRN2, which seems to activate flowering in long days through a daylength response pathway, the HvVRN1-1 allele can substitute for prolonged exposure to cold.
Previous interpretations of genetic interactions between VRN1 and VRN2 have led to the suggestion that VRN1 genotype does not influence flowering time in lines that lack VRN2 . Our data show this is not the case; the HvVRN1-1 allele is associated with elevated HvVRN1 expression, accelerated inflorescence initiation, and early flowering in the absence of HvVRN2 (see DHvVRN2/ppd-H1/HvVRN1 versus DHvVRN2/ppd-H1/HvVRN1-1; Supplemental Figs. S1 and S5).

The Molecular Basis of the Vernalization Response
When studying the vernalization response it is important to differentiate between the response to low temperatures, which occurs during winter when days are short, and the response to long days, which occurs after winter. When plants are vernalized in short days (Fig. 7), HvVRN1 is induced by cold without changes in expression levels of HvFT1 or HvVRN2. Furthermore, low-temperature induction of HvVRN1 can occur in the absence of active versions of HvVRN2 or PPD-H1 (Figs. 4 and 5). This shows that HvVRN1 is regulated by a low-temperature response pathway that operates independently of the daylength response. We suggest that vernalization and daylength response pathways act sequentially to promote flowering. Before vernalization, both the daylength and vernalization pathways are inactive; expression of HvVRN1 is low, and HvVRN2 blocks long-day induction of HvFT1. During vernalization HvVRN1 expression is induced. Following vernalization, HvVRN1 is expressed and HvVRN2 is repressed, allowing long-day induction of HvFT1. According to this model, HvVRN1 acts in the vernalization response pathway, HvFT1 acts in the daylength response pathway, and HvVRN2 is an integrator of the vernalization and daylength responses (Trevaskis et al., 2007b;Fig. 8).
Our data show that expression of HvVRN1 can promote inflorescence initiation, whereas expression of HvFT1 can accelerate both inflorescence initiation and subsequent stages of inflorescence development. These observations are consistent with studies showing that low temperatures can accelerate flowering until plants reach the double ridge stage, whereas long days can accelerate flowering until the terminal spikelet differentiates (Flood and Halloran, 1984;Roberts et al., 1988). Low temperatures and long days are likely to promote flowering by accelerating different stages of floral development. Induction of HvVRN1 during winter will promote inflorescence initiation. Subsequently, as daylength increases, expression of HvFT1 will accelerate later stages of inflorescence development.
Another model for the vernalization response in cereals suggests that VRN2 is a repressor of VRN1 and FT1 (VRN3), and that VRN2 is down-regulated by cold or daylength to elicit the response to low temperature during winter (Yan et al., 2004. Our model differs in that it proposes that VRN1 is induced by low temperature independently of VRN2 and FT1, which regulate the daylength response. Both models offer similar mechanisms to account for natural variation in the vernalization requirement; alleles of VRN1 and VRN3 with high basal expression levels, or deletion of VRN2, promote flowering and remove the requirement for vernalization. The molecular basis of dominant alleles of VRN3 in barley is unclear. The Sloop 3 Halcyon population segregates for two HvFT1 sequences suggested to correspond to the different alleles of VRN3 , but HvFT1 genotype did not affect vernalization requirement in this population. Sequence changes outside the proximal promoter and transcribed region that influence HvFT1 expression may provide the molecular basis for the dominant VRN3 allele in the mapping population used by Yan et al. (2006). A third model for the vernalization response of cereals is that low temperatures repress VRT2, a MADS-box gene suggested to repress VRN1, and this allows expression of VRN1 to increase (Kane et al., 2005(Kane et al., , 2007. This model seems unlikely because in barley HvVRT2 is induced by low temperatures (Trevaskis et al., 2007a).
Understanding how genes that regulate the vernalization and daylength responses interact to control flowering is important for cereal breeding. The strong genetic interaction between HvVRN2 and PPD-H1 has implications for temperate cereal breeding strategies; PPD-H1 genotype is likely to be important when HvVRN2 is used to breed for a reduced vernalization requirement and early flowering.

Plant Growth Conditions
The field flowering time experiments have been described previously (Boyd et al., 2003). Glasshouse experiments were planted in January, 2007. For glasshouse experiments, plants were grown in pots of soil in sunlit glass-houses under long days (16-h light/8-h dark) with supplementary lighting when natural light levels dropped below 200 mE. Glasshouses had an average temperature of 19°C. Plants were harvested at the middle of the light period for short day treatments, or 12 h after the beginning of the light period (midafternoon) for long day treatments. For vernalization treatment, seeds were imbibed in foil-covered pots at 4°C for 10 weeks. Pots were then moved to standard glasshouse conditions for 1 week. At this stage, leaf number was similar to that of 14-d-old seedlings grown in standard glasshouse conditions. Experiments were terminated after 120 d of growth in glasshouse conditions.

Apex Length and Flowering Time Measurements
Apices were isolated under a binocular dissecting microscope and then digitally photographed on a Leica M8 digital camera. Apex lengths were measured from digital images using analySIS LS Professional (Olympus Soft Imaging Solutions). Apex lengths were averaged from three to six plants.
Heading date was measured as the day when the head first emerged from the sheath, and averaged for five plants.

DNA and RNA Extraction
DNA for PCR genotyping was extracted from individual dry seeds in a 96well format. Seeds were ground in an extraction buffer, (2% cetyl trimethylammonium bromide, 2% polyvinylpyrrolidone, 0.02 M EDTA, 1.4 M NaCl, 100 mM Tris-Cl, pH 8.0) using a ball bearing in a mix mill. Extracts were incubated at 65°C for 30 min then cleared by centrifugation. DNA was precipitated with 1 vol of isopropanol and 0.1 vol of 6 M ammonium acetate, then centrifuged. Pellets were washed in 70% ethanol and resuspended in 150 mL of Tris-EDTA buffer (10 mM Tris-Cl, pH 8, 1 mM EDTA). Total RNA was extracted from whole seedlings, excluding root tissue, using the method of Chang et al. (1993).

Reverse Transcription-PCR and qRT-PCR
To remove genomic DNA contamination, RNA was treated with RQ1 RNase-free DNase (Promega) according to the manufacturer's instructions. An oligo(T) primer (T 18 [G/C/A]) was used to prime first-strand complementray DNA (cDNA) synthesis from 5 mg of total RNA using the SuperScript III reverse transcriptase enzyme (Invitrogen) according to the manufacturer's instructions. A single reverse transcription (RT) reaction was performed for each RNA sample. RT-PCR was performed using Taq F1 DNA polymerase (Fischer Biotech). The primers used for HvFT1, HvVRN1, HvVRN2 (HvZCCTb), and ACTIN have been described previously (Trevaskis et al., 2006). The primers used for PPD-H1 were 5#CAAATCAAAGAGCGG-CGATG3# and 5#TCTGACTTGGGATGGTTCACA3#. Each primer pair amplifies cDNA-specific DNA products. Cycling conditions were 40 cycles (except for ACTIN, 25 cycles) 30 s at 94°C, 30 s at 57°C, and 30 s at 72°C. Fragments were visualized by agarose gel electrophoresis.
qRT-PCR was performed on a Rotor-Gene 3000 real-time cycler (Corbett Research) with the same primer pair sets. qRT-PCR was performed using Platinum Taq DNA polymerase (Invitrogen). Cycling conditions were 2 min at 95°C, 50 cycles of 10 s at 95°C, 15 s at 60°C, and 20 s at 72°C. This was followed by a melting-curve program (72°C-95°C with a 5-s hold at each temperature). Fluorescence data were acquired at the 72°C step and during the meltingcurve program. Expression levels of genes of interest relative to ACTIN were calculated using the comparative quantification analysis method (Rotogene-5; Corbett Research), which takes into account the amplification efficiency of each primer set. Quantification for each primer set and cDNA template combination was performed in triplicate, and included a no-template control, to ensure results were not influenced by primer-dimer formation or DNA contamination. Data presented are the average and SE from triplicate reactions on the same PCR run.

Plant Transformation
An HvVRN2 overexpression construct was made using the GATEWAY cloning system. An HvZCCTb gene fragment was inserted into an entry vector and then recombined into the pUBI-GATEWAY vector, a vector constructed with a GATEWAY destination site downstream of the maize (Zea mays) UBIQUITIN promoter (Christensen et al., 1992) in the pWBVEC8 binary vector backbone (Wang et al., 1998). Barley (Hordeum vulgare) plants were transformed using Agrobacterium transformation of excised embryos (Tingay et al., 1997;Mathews et al., 2001). T1 and T2 plants were screened for segregation of the transgene using primers that amplify from the hygromycin selectable marker gene. Expression analysis was carried out on plants hemizygous or homozygous for the transgene and sibling null control lines that did not inherit the transgene.

Supplemental Data
The following materials are available in the online version of this article.
Supplemental Figure S1. HvVRN1, HvVRN2, and PPD-H1 genotype versus flowering time, from a field planting of the Sloop 3 Halcyon doubled haploid population.
Supplemental Figure S2. Relative expression levels of PPD-H1 in lines carrying HvVRN2 versus lines deleted for HvVRN2.
Supplemental Figure S3. Nucleotide sequence alignment showing allelic variation in HvFT1 in Halcyon and Sloop.
Supplemental Figure S4. A time course of HvFT1 expression in nonvernalized plants growing in long days.
Supplemental Figure S5. The shoot apices of representative lines deleted for HvVRN2 and carrying the inactive ppd-H1 allele, but with different HvVRN1 genotypes.
Supplemental Data S1. Genetics of flowering time in the Sloop 3 Halcyon doubled haploid population.