Adaptation of barley to mild winters: A role for PPDH2

Background Understanding the adaptation of cereals to environmental conditions is one of the key areas in which plant science can contribute to tackling challenges presented by climate change. Temperature and day length are the main environmental regulators of flowering and drivers of adaptation in temperate cereals. The major genes that control flowering time in barley in response to environmental cues are VRNH1, VRNH2, VRNH3, PPDH1, and PPDH2 (candidate gene HvFT3). These genes from the vernalization and photoperiod pathways show complex interactions to promote flowering that are still not understood fully. In particular, PPDH2 function is assumed to be limited to the ability of a short photoperiod to promote flowering. Evidence from the fields of biodiversity, ecogeography, agronomy, and molecular genetics was combined to obtain a more complete overview of the potential role of PPDH2 in environmental adaptation in barley. Results The dominant PPDH2 allele is represented widely in spring barley cultivars but is found only occasionally in modern winter cultivars that have strong vernalization requirements. However, old landraces from the Iberian Peninsula, which also have a vernalization requirement, possess this allele at a much higher frequency than modern winter barley cultivars. Under field conditions in which the vernalization requirement of winter cultivars is not satisfied, the dominant PPDH2 allele promotes flowering, even under increasing photoperiods above 12 h. This hypothesis was supported by expression analysis of vernalization-responsive genotypes. When the dominant allele of PPDH2 was expressed, this was associated with enhanced levels of VRNH1 and VRNH3 expression. Expression of these two genes is needed for the induction of flowering. Therefore, both in the field and under controlled conditions, PPDH2 has an effect of promotion of flowering. Conclusions The dominant, ancestral, allele of PPDH2 is prevalent in southern European barley germplasm. The presence of the dominant allele is associated with early expression of VRNH1 and early flowering. We propose that PPDH2 promotes flowering of winter cultivars under all non-inductive conditions, i.e. under short days or long days in plants that have not satisfied their vernalization requirement. This mechanism is indicated to be a component of an adaptation syndrome of barley to Mediterranean conditions.


Background
Temperature and photoperiod are the main environmental cues that regulate flowering time in winter cereals [1,2]. Barley (Hordeum vulgare L.) is classified as a long-day plant, which means that it will flower earlier when exposed to increasing day lengths. Some genotypes of barley require a period of prolonged exposure to cold during winter (vernalization) to accelerate the transition of the shoot apex from vegetative to reproductive development [3]. This combination of a requirement for vernalization and sensitivity to photoperiod ensures that flowering is postponed until after winter to avoid frost damage, but then occurs rapidly as daylength increases during spring, thereby avoiding heat and water stress during summer [4].
Wheat and barley cultivars are classified on the basis of their flowering behavior into two types of growth habit, namely winter and spring. The former requires prolonged exposure to low temperature to flower, whereas the latter group flowers rapidly without exposure to cold. Genetic studies have revealed that the epistatic relationships between three genes, VRNH1, VRNH2, and VRNH3, control the response to vernalization [5,6]. The winter growth habit depends on the combination of recessive alleles at VRNH1 and VRNH3 with the dominant allele at VRNH2 [5]. Genotypes that possess other allelic combinations for these genes exhibit a spring growth habit to different degrees. In agronomic classifications of barley germplasm, a third category of cultivars, termed facultative [7], is recognized, in which cultivars show winter hardiness but do not require vernalization.
The activity of VRNH1 is essential for flowering [8]. VRNH1 acts as a promoter of flowering, is induced by vernalization, and regulates the transition to the reproductive stage at the shoot apex [9]. Allelic variation at VRNH1 has been described, mainly in relation to deletions within the first intron [10][11][12]. These deletions are presumed to be responsible for the different vernalization requirements that are associated with different alleles. In plants that have not been vernalized, the deletions lead to differences in the levels of the VRNH1 transcript and, consequently, the allelic variation results in diverse flowering times [13,14].
VRNH2 is a floral repressor that delays flowering until plants are vernalized [5,15]. Allelic diversity at VRNH2 arises from the presence or deletion of a cluster of three genes (ZCCT-H) [7]. The null allele of VRNH2 corresponds to the recessive spring allele and is associated with rapid flowering [7,16,17]. Day length is the major determinant of the level of VRNH2 expression, with high levels of expression occurring during periods with long days [15,18,19].
HvFT1, candidate gene for VRNH3, is a homolog of the FLOWERING LOCUS T gene (FT) of Arabidopsis thaliana [20,21]. Strong evidence indicates that VRNH3 plays a central role in promoting flowering as an integrator of the vernalization and photoperiod pathways in temperate cereals [6,9,22]. Recently, novel VRNH3 alleles that show different adaptive effects have been identified by analyzing sequence polymorphisms and their phenotypic effects [23].
Two major photoperiod response genes, PPDH1 and PPDH2, have been reported in barley [1,24]. PPDH1 confers sensitivity to a long photoperiod and accelerates flowering under long days [25]. HvFT3 has been identified as a candidate gene for PPDH2, which is described as a gene that is responsive to a short photoperiod [21,22]. As far as can presently be determined, only two known HvFT3 (PPDH2) alleles exist, one of which is a null [14,31]. However, it is not possible to rule out that additional alleles are also present in cultivated barley. The dominant allele is the functional one, comprises four exons, and produces faster development towards flowering under short days. The recessive allele is a truncated gene, retaining only the 3' portion of exon 4 [22], and produces flowering delay under short days (Additional file 1).
The complexity and strength of the interactions reported among these genes indicate that they share the same regulatory network [26][27][28]. VRNH1, VRNH2, and VRNH3 form a feedback regulatory loop [6]. VRNH1 is probably the principal target of the vernalization signal [2]. Levels of the VRNH2 transcript are downregulated by short days and by a high level of VRNH1 expression [19]. Expression of VRNH2 delays flowering by inhibiting expression of VRNH3 [9]. After vernalization, transcription of VRNH2 decreases, which facilitates the upregulation of VRNH3 by long days in spring, and triggers flowering [4,6]. It is likely that the downregulation of VRNH2 is mediated by VRNH1. Photoperiod response genes also participate in the promotion to flowering. The dominant PPDH1 allele accelerates flowering by upregulating VRNH3 under long days [9]. PPDH2 is thought to upregulate VRNH3 expression under short-day conditions [22]. In addition, expression of PPDH2 has been detected both under short days [21,22] and under long days when the levels of VRNH2 transcript decrease [14].
As a result of these interactions, phenotypic responses of barley to environmental signals are complex. Natural allelic variation at these flowering time genes has been found in several studies in relation to responses to vernalization [9,11,29], photoperiod [21,30], or both [31,32]. This natural variation might be related to adaptation to different environmental conditions.
In the study reported herein, we investigated further the patterns of expression and interactions of VRN and PPD genes in a selection of vernalization-responsive barley cultivars. These cultivars represented different allelic combinations of VRNH1, VRNH3, and PPDH2 in a dominant PPDH1 and VRNH2 genetic background. The geographic distribution of PPDH2 alleles was analyzed in a wide array of barley germplasm that represented cultivars and landraces. In addition, the possible role of PPDH2 in the acceleration of flowering under long days was examined in a collection of winter cultivars, by analyzing their response to vernalization treatments of different duration.

Distribution of PPDH2 alleles among domesticated barleys
We investigated the distribution of the PPDH2 alleles over a sample of 162 barley cultivars of different geographic origins (Additional file 2) and 159 Spanish landrace-derived inbred lines from the Spanish Barley Core Collection (SBCC) [33]. Lines were classified according to their seasonal growth habit, on the basis of the allelic constitution at VRNH1 and VRNH2 (Table 1). To enlarge the sample, we included previously published results for an additional 202 barley cultivars [21,31]. The dominant allele of PPDH2 gene was found in most of the spring cultivars (189 out of 206), whereas the majority of winter cultivars (102 out of 140) possessed the recessive (null) ppdH2 allele (Table 1). Facultative genotypes, characterized by having a winter allele at VRNH1, and a null allele (vrnH2) at VRNH2, did not show such a clear genetic distinction and approximately half (seven out of 18 cultivars) carried the dominant (functional) PPDH2 allele (Table 1). Strikingly, the allelic distribution among SBCC landraces differed from that observed in the commercial cultivars. Most of the winter Spanish landraces (127 out of 140) carried the functional PPDH2 allele ( Table 1). The 140 winter SBCC landraces all carried the dominant allele at VRNH2 and PPDH1 but possessed two different alleles at VRNH1. According to the terminology for VRNH1 alleles proposed by Hemming et al. [13], 93 of these landraces carried VRNH1-6 and 47 carried the earlier flowering VRNH1-4 allele [14]. PPDH2 was carried at the same frequency among landraces carrying the VRNH1-6 and VRNH1-4 alleles. The wild-type recessive VRNH1 allele was not detected among the Spanish landraces.

Cold-induced gene expression under a long photoperiod
Expression of the vernalization and photoperiod response genes was studied in eight barley lines, which represented four typical winter cultivars and four Spanish landraces ( Table 2). The lines had been exposed to low temperature treatments of increasing length (15,30 or 45 days) under short days, in every case ensued by growth for 15 days under long days (16 h light). All of the genotypes carried VRNH2 and the long-photoperiod-sensitive allele PPDH1, which allowed observation of possible interactions between these genes and the other vernalization and photoperiod genes. The expression profiles of the vernalization and photoperiod response genes were assessed by quantitative reverse-transcription PCR (qRT-PCR; Figures 1 and 2). Differences in expression among genotypes and treatments were found for VRNH1, VRNH2, VRNH3, and PPDH2. Although  a Alleles based on the size of intron 1 [12]. b Presence/absence of HvZCCT [15]. c Promoters identified previously [23]. d Alleles based on two SNPs in intron 1, as reported previously [20]. e Alleles based on SNP22 [25]. f Presence/absence of PPDH2 [13].
expression data for PPDH1 were also analyzed, its expression is not shown in Figures 1 and 2 because it was consistently high in all treatments and genotypes, and thus did not contribute to the variability of responses observed. In all genotypes, VRNH1 expression increased gradually with increasing duration of vernalization treatment, although differences in response between VRNH1 alleles were evident. After 15 d of cold treatment, VRNH1 expression was only detected in genotypes that carried the larger~4 kb deletion in intron 1 (allele VRNH1-4), namely SBCC058 and SBCC114 (Figure 1a). The level of VRNH1 expression was significantly higher in SBCC058 than in SBCC114 (Figure 1a). After vernalization for 30 d, VRNH1 expression was detected in five genotypes ( Figure 1b). VRNH1 expression was detected in all genotypes only after 45 d of cold treatment ( Figure 1c). The expression level was highest for the VRNH1-4 and VRNH1-6 alleles (namely SBCC106 and SBCC016), and lowest for the wild-type recessive winter allele vrnH1, which was carried by Plaisant, Rebelle, Arlois, and Hispanic. Even though these four cultivars carried the same VRNH1 allele, they showed differences in VRNH1 expression (Figure 1b-c).
Although all lines carried the active VRNH2 allele, differences in its expression were observed (Figure 1a Effect of VRNH3, PPDH1, and PPDH2 and different vernalization treatments on heading date in winter cultivars To assess a possible effect of the major vernalization and photoperiod response genes on flowering time under natural conditions, we analyzed the time from planting to heading of 70 winter cultivars that were exposed to five different periods of vernalization, which ranged from 0 to 60 d, before transplantation to the field in Martonvásár, Hungary, on March 25 th , which corresponds to a day length of 12 h 25 min. The list of barley lines and their genetic constitution for the major flowering-time genes is presented in Additional file 3. All of the lines carried the dominant allele at VRNH2. Although polymorphisms have been reported for the candidate genes (the ZCCT-H family), VRNH2 seems to be quite conserved among winter barleys, and just two alleles are usually assumed [12]. We evaluated the differences between the PPDH1, PPDH2, and VRNH3 alleles as a function of the duration of vernalization (Table 3). This was possible because there were enough individuals in each of the 8 classes resulting from the combination of the three genes to perform an analysis. Although the cultivars presented three different VRNH1 alleles (all showing a response to vernalization), they were so unevenly distributed over the sample (60 vrnH1, five VRNH1-6, five VRNH1-4) that it was not possible to include it as an additional factor in the analysis. All three genes analyzed showed significant effects on flowering time. On average, the dominant allele at PPDH1 accelerated the onset of flowering by 4 d. Lines that carried the functional allele at PPDH2 flowered 6 d earlier, and genotypes that carried the TC haplotype for VRNH3 flowered 2 d earlier. Consistent with the expectation for winter genotypes, different durations of vernalization had a significant effect on flowering time. In the present analysis, a significant interaction between PPDH2 and the different cold treatments was detected ( Figure 3). Exposure to a cold treatment before transplanting reduced the time to heading, although the reduction was not significant for vernalization treatments longer than 30 d. The presence of the dominant PPDH2 allele was associated with earlier flowering in plants that had not been vernalized fully (0 or 15 d cold treatment; Figure 3).
We also analyzed the geographic distribution of 125 winter barley cultivars, which were assigned to their predicted phenotype on the basis of the presence of a complete HvFT3 gene, and classified into three classes according to latitude. The PPDH2 dominant allele was predominant in winter cultivars from southern latitudes, whereas the proportion of cultivars with the recessive (null) allele ppdH2 was greater at higher latitudes (Figure 4).

Discussion
Heading date is a crucial trait for the adaptation of barley to different areas of cultivation and cropping seasons. Traditionally, cultivars are classified into spring, facultative, and winter types on the basis of their flowering habit. This is an agronomic classification that is based on phenotypic behavior. It is a useful simplification that summarizes a more complex and diverse array of responses at the genetic level. Study of the genes involved in the photoperiod and vernalization pathways in cereals is continuously producing new information that is shedding light on the nature of adaptation of cultivars and on the variety of phenotypic responses produced by the combination of photoperiod and vernalization genes carried by individual cultivars.

PPDH2 is not distributed randomly in barley germplasm
The spread of cultivated barley out of its area of origin was driven by the occurrence of phenotypic variation that resulted from the appearance of new multilocus flowering-time haplotypes at VRNH1, VRNH2, PPDH1, and PPDH2 [32]. Mutations in VRNH1 allowed the expansion of cultivated barley from midlatitudinal regions to lower and higher latitudes, where spring types are common [29,32,34]. The entry of barley to Europe occurred via several routes [34]; One of them, to the North and then West, via the Balkan Peninsula, and another one towards the Southwest, then through North Africa, reaching Europe through Spain. In the first case the environmental conditions (long winters, shorter days than in the Mediterranean region) favoured the recessive allele in PPDH2 so its frequency increased significantly within the winter forms. In the latter case, the ancestral form was not selected out from the winter barleys, which is exactly the case for the Spanish landraces.
In midlatitudinal regions, including North Africa, southern Europe, Nepal, China, and Japan, both spring and winter barley types are cultivated. However, in these regions, the dichotomic agronomic classification is insufficient to describe the range of vernalization responses found, in which VRNH1 plays a central role. Allelic diversity at VRNH1 has been described by several authors [11][12][13][14]. This diversity is the result of deletions or insertions within the first intron of the gene, and is associated with a gradation of vernalization responses from strict winter to spring types. In general, the larger the deletion, the shorter the vernalization period required.
Although, originally, wild barley carried the photoperiod-responsive alleles PPDH1 and PPDH2 (dominant allele), mutant, nonresponsive alleles of these genes   originated before domestication [32]. The appearance of the nonresponsive ppdh1 allele allowed the cultivation of barley to spread to more northerly regions [30]. Regarding PPDH2, some authors have already pointed out the prevalence of the dominant allele in spring cultivars, and its relative scarcity in winter cultivars [21,31].
In the present study, exclusively done with winter types, we found that, the dominant PPDH2 allele was frequent at lower latitudes (<44°N) but not at higher latitudes.
The dominant allele was also prevalent in a large set of winter landraces cultivated on the Iberian Peninsula (35-44°N). This pattern is remarkable, because latitudes below 44°N include almost the entire Mediterranean region. In this region, barley is sown predominantly during autumn and, to a large extent, using winter cultivars. The adaptive role of PPDH2 is confirmed by its influence on key agronomic traits. It was identified originally as a short-photoperiod quantitative trait locus in winter × spring barley crosses [1,35]. Its effect is especially large in Mediterranean latitudes, where it has been identified as the main QTL that affects flowering, together with Eam6 [35,36]. It also affects grain yield indirectly, through flowering date, under Mediterranean conditions [37].

PPDH2 expression is mediated by the vernalization pathway in winter cultivars
Analysis of gene expression can provide indications of the role of PPDH2 and interacting genes. To be meaningful for the Mediterranean region, we chose to carry out this study with winter genotypes, unlike previous studies [21,22] which focused on the effect of PPDH2 on spring genotypes.
Expression of photoperiod and vernalization response genes show strong interactions [6,26,28]. A long photoperiod induces VRNH2 expression [19], which then represses expression of VRNH3 [9] and PPDH2 [14]. The model currently accepted proposes that during autumn and winter (low temperature and short days), vernalization induces VRNH1 expression and the short photoperiod downregulates VRNH2 expression [2,19]. Subsequently, in spring, VRNH1 is relatively high, much more rapidly if vernalization was sufficient. Although the long photoperiod conditions in spring are favorable for VRNH2 expression, VRNH2 is repressed by the expression of VRNH1. Once the vernalization requirement has been satisfied, VRNH3 expression is induced by long days [9], after which the plants are committed irreversibly to reproductive development.
In our expression analysis, we compared three different VRNH1 alleles. At each time-point examined, the expression level was lower in the four winter cultivars that carried the full-length intron than in the four SBCC lines that carried two different deletions. As proposed previously [19], vernalization did not block the induction of VRNH2 in response to increasing day length, which was detected under long days after 15 or 30 d of cold treatment. Once VRNH1 is expressed, it can then begin to repress VRNH2 expression. However, the differences in responses observed among the four winter cultivars that carried the strict winter allele at VRNH1 were unexpected. Two of these cultivars (Plaisant and Rebelle) behaved as expected; a long period (45 d) of cold induction was needed to detect VRNH1 expression. Interestingly, for the other two cultivars (Arlois and Hispanic), we detected expression of VRNH1 after only 30 d of cold treatment, and the transcript level increased further after 45 d of treatment. These four cultivars carry identical, recessive alleles at VRNH1 and VRNH3, and dominant alleles at VRNH2 and PPDH1. Among the genes investigated, they differ only at PPDH2, which leads us to think of a possible role of this gene in the earlier induction of VRNH1 expression. However, we cannot rule out the possibility that additional genes might be responsible for this induction.
In a previous study, we did not detect VRNH3 expression in some of these genotypes when they were grown without vernalization under a long photoperiod (SBCC058 and Plaisant) or vernalized under a short photoperiod (SBCC058, SBCC106 and Plaisant) [14]. In winter genotypes, a period of cold induction is required before VRNH3 expression can be induced by long days, as reported already for the wild-type vrnH1 winter allele [9]. In the present study, we included cultivars that represented several recessive alleles at VRNH3, because we had previous evidence that they might produce differences in heading date in the field among these cultivars [23]. Different expression between VRNH3 alleles was detected only for the pair of lines with the largest deletion in VRNH1 (SBCC058 and SBCC114). The TC allele showed higher expression than the AG allele, in the same direction as reported in a previous study [23]. However, there was no difference in VRNH3 expression between SBCC016 and SBCC106, which showed the same polymorphism at VRNH3 among them than SBCC114 and SBCC058. Either the duration of the experiment was insufficient to reveal possible differences or other genes that are unaccounted for at present influence this pathway.
Although the expression of PPDH2 is higher under a short photoperiod, we and other authors [14,22] have reported PPDH2 expression under a long photoperiod. Expression of PPDH2 was detected at some time -point in all genotypes that carried the dominant allele of PPDH2, irrespective of day length. In winter genotypes, VRNH2 must be absent or clearly receding (either because lack of induction under short days, or repression by expression of VRNH1) for PPDH2 to be expressed.

PPDH2 promotes flowering irrespective of photoperiod under noninductive conditions
An additional question concerns the nature of the role of PPDH2. PPDH2 has been suggested to affect the promotion of the transition of the shoot apical meristem from vegetative to reproductive, in the end affecting flowering. The two experiments that support this hypothesis, however, propose different modes of action for PPDH2. On one hand [21], it was proposed that HvFT3 (PPDH2) substituted HvFT1 (VRNH3) as the trigger to induce flowering under short days (8 h), although its expression was not sufficient to induce the transition to the reproductive stage. They did not find HvFT1 (VRNH3) induction with 8 h of light, even after the transition of the meristem had taken place. Another study [22], concluded that HvFT3 acts as a floral promoter under short days (12 h this time), but through the induction of HvFT1 (VRNH3). Therefore, it seems proven that PPDH2 promotes flowering under short days, but the mechanism (or mechanisms) of action are not clear yet. The experiments just reported used different genotypes and, probably more important, different day lengths. Differences in induction of genes may have been caused by different critical day length thresholds needed for expression of these genes. In any case, all the genotypes tested in those studies were spring lines, and the interaction of PPDH2 with the vernalization pathway in winter genotypes, at gene expression level, remained largely unexplored.
By investigating simultaneously the expression of the flowering response genes, we observed that VRNH1 and PPDH2 were expressed before VRNH3 in all six vernalization-responsive genotypes tested. Our results agree with a comparative model proposed by Higgins et al. [28]. In that scheme, PPDH2 promotes VRNH1 expression under short-day conditions. We propose that PPDH2 has a more general role for winter cultivars, and promotes flowering under all noninductive conditions, i. e. under short days or long days in plants that have not satisfied their vernalization requirement. This hypothesis is supported by the field trial observations. Heading date in our trial occurred from May 10 th until July 13 th . The photoperiod experienced by the plants increased from 12 h 25 min at transplanting to 14 h 53 min when the first genotype reached heading, and then kept increasing until 15 h 58 min on June 21 st . Therefore, most of the growth period of the plants occurred in photoperiods well above 12 h. We observed a concurrent effect of PPDH1 and PPDH2 on flowering, which agrees with the concurrent effect for these two genes found under a 12 h photoperiod [38]. During this period of the year (May-July), and even earlier, the effect of long days on heading date in experiments carried out in temperate latitudes can be detected through its effect on PPDH1 [39].
Heading date was distinctly earlier for winter genotypes that carried the dominant PPDH2 allele than for cultivars that possessed the recessive allele. The difference was especially marked for plants that had not been vernalized or had experienced only a short cold period. The 70 genotypes used might show some intrinsic difference in earliness per se that might account for some of the differences that could be attributed to PPDH2 as the main factor. However, the differences in heading that were caused by PPDH2 decreased gradually as the duration of vernalization increased. This interaction between PPDH2 and duration of vernalization treatment was quite reliable, and is consistent with the role for PPDH2 suggested above. Other authors [40] have also reported an effect of PPDH2 on flowering time under long photoperiods, but only with the application of synchronous photo and thermo cycles, and when specific allelic configurations are present at the PPDH1 and VRNH1 loci.
Winter genotypes are cultivated normally in areas where they are exposed to sufficient vernalization during winter. As a consequence, these genotypes do not need to express other genes that promote flowering. By contrast, in spring cultivars, PPDH2 can facilitate flowering and ensure timely completion of such a short vital cycle. However, in winter cultivars with lower requirements for vernalization, such as those adapted to geographical areas with traditionally mild winters, as exemplified by Mediterranean climates, the presence of PPDH2 might help to induce flowering when the vernalization requirement has not been satisfied fully (which is a not unusual phenomenon under natural conditions in the Iberian Peninsula). This could explain why the majority of SBCC winter lines carry the dominant PPDH2 allele. SBCC winter lines are adapted to a typical mild Mediterranean winter, in which temperatures are not very low. If the cold period is insufficiently long to satisfy the vernalization requirement of these genotypes, PPDH2 could act as a compensatory mechanism to accelerate flowering and ensure it occurs at the optimal time, possibly before the effect of a sensitive PPDH1 is noticeable. In some barley and wheat cultivars the vernalization requirement can be replaced, partially or completely by exposure to short photoperiods [18,41,42]. This phenomenon, known as short-day vernalization [42] has been reported in barley genotypes with winter alleles in VRNH1 and VRNH2 and dominant PPDH2 [1,35]. In these genotypes, a dual short day-long day induction of flowering could take place [18]. This dual mechanism is present in many species, including many Festucoideae [43]. King and Heide [43], proposed that "...as an evolutionary mechanism, the versatility of the alternative short day/vernalization primary induction system offers a beautiful safety mechanism with short days acting as a fall-back alternative in case of inadequate winter chill". The involvement of VRN2 in the genetic basis of this mechanism was already put forward by Dubcovsky et al. [18], because "the convergence of photoperiod and vernalization signals at the VRN2 gene, provides a possible explanation to the interchangeability of short day and vernalization treatments." The presence of the dominant PPDH2 allele would not be necessary under conditions in which vernalization occurred inevitably year after year, as it is common in more northerly latitudes. Actually, other authors have claimed that the presence of the dominant allele at PPDH2 is not a desirable feature for winter barley [44,45], because it would induce progress towards flowering too early [21], with undesirable agronomic consequences, including loss of frost tolerance. This may well be true for strict winter cultivars (strict winter vrnH1 allele plus dominant VRNH2) in more northerly latitudes. The null, late-flowering allele would be more suitable for an autumn-sown cultivar because it would keep plants in the vegetative growth phase longer [46], perhaps through maintaining the expression of genes that confer tolerance to low temperature [47]. On the basis of these studies, negative agronomic effects of the dominant PPDH2 allele should be investigated, especially in relation to freezing tolerance. However, a dominant PPDH2 allele could be a good option for cultivars cultivated in geographic areas where the winters are not that cold. The adaptation syndrome for barley landraces in the Iberian Peninsula seems to be the combination of an appropriate VRNH1 allele with dominant PPDH1, to ensure that flowering will occur before temperature rises too high, and with a dominant PPDH2 to ensure that plant growth will be not too delayed even in the years that conditions do not produce full vernalization.

Conclusions
It is crucial to study the main genes involved in the vernalization and photoperiod pathways simultaneously, because this enables the interactions and functions of these genes to be interpreted more accurately, and their involvement in the induction of flowering to be elucidated.
There is a wide agreement over the central role of VRNH1 on the control of the progress of barley towards flowering. Nevertheless, different flowering-time responses seem to be modulated by the alleles present at the other vernalization and photoperiod genes VRNH2, VRNH3, PPDH1, and PPDH2. Of these genes, PPDH2 might have an important role in the regulation of VRNH1, especially under a long photoperiod, by upregulating VRNH1 expression and indirectly reducing the time to flower.
PPDH2 has a strong effect on heading date in a wide array of winter genotypes. The dominant allele at PPDH2 accelerates flowering under long days in plants in which the vernalization requirement has not been satisfied. The presence of PPDH2 in most winter landrace-derived lines of the SBCC indicates this allele could promote adaptation to geographic areas with milder winters, such as Mediterranean environments.
We also suggest the PPDH2-dependent mechanism proposed in this study could be complementary to the mechanism governed by PPDH1. The sensitive PPDH1 allele is typical of winter cultivars and PPDH2 is more common in spring cultivars. Both mechanisms promote flowering in different environments. Furthermore, in Mediterranean environments, these two mechanisms could be combined to facilitate flowering in optimal conditions.

Genotyping
A set of 162 barley genotypes (Additional file 1) and 159 landraces from the SBCC were genotyped for the vernalization (VRNH1, VRNH2, and VRNH3) and photoperiod (PPDH1 and PPDH2) genes as described previously [14,23]. Genotyping was conducted on single plants of each accession, partly at ARI-HAS (Hungary) and partly at EEAD-CSIC (Spain).

Gene expression analysis Plant material
Eight winter genotypes of barley were chosen to assess differences in the expression of the five major genes involved in responses to temperature and photoperiod. The genotypes consisted of the French cultivars Rebelle ((Barbarrosa × Monarca) × Pirate), Plaisant (Ager × Nymphe), Hispanic (Mosar × (Flika × Lada)), and Arlois (unknown pedigree), and four inbred lines, derived from landraces, that belong to the SBCC [33]. The genotypes studied have different VRNH1-VRNH3 allelic combinations and all can be classified as 'winter' genotypes. The genotypes could be grouped into four pairs, with each pair sharing the same VRNH1 allele, as defined by the length of the first intron. Each pair defined on the basis of VRNH1 structure was polymorphic for VRNH3 (Table 2), as defined by single nucleotide polymorphisms (SNPs) in intron 1, as reported previously [20], and by indels in the promoter region [23]. All genotypes carried an active VRNH2 and the sensitive allele at PPDH1, and all carried the PPDH2 functional allele except Rebelle and Plaisant ( Table 2).

Conditions of plant growth
For expression studies, seeds of the eight genotypes were sown in pots and germinated in a sunlit glasshouse at 19 ± 1°C with a 16-h light/8-h dark photoperiod. Ten days after sowing, when the plants had reached the twoleaf stage (Z12, Zadoks scale [48]), the seedlings were moved to a growth chamber and exposed to 7 ± 1°C for 15, 30 or 45 d under a short photoperiod (8-h light/16h dark) and low light intensity (12 μmol m -2 s -1 ). After vernalization, the plants were transferred sequentially to an additional growth chamber maintained at 22 ± 1°C under a 16-h light/8-h dark photoperiod with light intensity of 220 μmol m -2 s -1 , where they were kept for 15 d, after which whole seedlings, excluding root tissue, were harvested. Harvesting took place in the middle of the light period. Four individual plants were harvested per sampling time-point and genotype, and were treated as four biological replicates.

RT-PCR and real-time PCR analysis
Extraction of RNA and preparation of cDNA followed the methods reported previously [14]. qRT-PCR was performed for all of the samples harvested. Amplifications were carried out in 20-μl reactions that contained 10 μl of SYBR Green Quantimix Easy SYG Kit (Biotools, Madrid, Spain), 0.3 μM each primer, 4 mM MgCl 2 , and 4 μl of cDNA, which corresponded to 300 ng of total RNA. Reactions were run on an ABI7500 real-time PCR system (Applied Biosystems). Cycling conditions for VRNH1, VRNH2, VRNH3, Actin, and PPDH2 were 6 min at 95°C, followed by 40 cycles of 15 s at 95°C, 15 s at 60°C, and 50 s at 72°C, and this was followed immediately by a melting curve program (60-95°C). Fluorescence data were acquired during the elongation step at 72°C and during the melting curve program. Two identical reactions (technical repeats) were performed per sample for each cDNA-primer combination. Levels of Actin expression were also quantified in the same run as an internal control. Four biological repeats were analyzed and showed similar trends. Expression levels were calculated using the ABI 7500 software package (Applied Biosystems). Gene expression at each timepoint was normalized to the expression of Actin.

Statistical analysis of differences in gene expression
Differences in relative expression between genotypes and treatments were evaluated using the analysis of variance (ANOVA) procedure in SAS [49]. The variable used for the analysis of each treatment and genotype was ΔC T (C T actin -C T target gene). This variable was preferred over the more commonly used 2 -ΔCT because of the concerns expressed regarding its use for statistical analysis [50]. The ANOVA model included biological replication, genotype, treatments, and genotype-by-treatment interactions. Genotypes and treatments were considered as fixed factors. The variability that resulted from biological repeats and their interaction with the other factors was used as the error term to test genotype and treatment, as well as their interaction. A multiple means separation was carried out using the least significant difference (LSD) test (P < 0.05) for the main effects that were significant in the ANOVA. Each value included in the analysis was the average of two technical repeats to protect against slight fluctuations in reading and small pipetting errors.

Field trial of winter cultivars after vernalization treatment
Sensitivity to vernalization and the subsequent flowering behavior of a set of 70 winter barley genotypes (Additional file 3) were evaluated. A vernalization period was imposed using the Martonvásár Phytotron (Hungary), in accordance with procedures described previously [51]. Vernalization was applied in 15-day increments, to give a total of five treatments that ranged from no vernalization to 60 d of vernalization at a temperature of 3°C, under an 8-h light/16-h dark photoperiod and low light intensity (12-13 μmol m -2 s -1 ). After vernalization, seedlings were transplanted by hand to the field at Martonvásár, Hungary, on March 25 th , 2010, when the average photoperiod was 12 h. Two plants were evaluated per genotype and treatment. For each plant, the number of days to flowering (Z49, Zadoks scale [48]) was scored. The trial was terminated after 100 d. Plants that had not headed were given a value of 150 d to heading.