bZIP transcription factors
The bZIP transcription factors are characterized by a basic region that binds DNA and a leucine zipper dimerization motif. This family is found in all eukaryotes but has more members in plants than in humans (Homo sapiens) or worm (Caenorhabditis elegans) (Riechmann et al., 2000). Approximately 75 bZIP transcription factors involved in pathogen resistance, light and stress signaling, flower development and seed maturation have been described in Arabidopsis (Jakoby et al., 2002).
The bZIP transcription factor FD plays a central role in the regulation of flowering in Arabidopsis (Abe et al., 2005; Wigge et al., 2005). FD is expressed in the SAM before floral induction, and its transcript levels increase with time after germination under short- or long-day photoperiods (Abe et al., 2005). Under inductive photoperiods, the FT protein is expressed in the phloem and travels to the shoot apex where it interacts with FD (Corbesier et al., 2007; Jaeger and Wigge, 2007;Mathieu et al., 2007; Tamaki et al., 2007). The interaction between these two proteins in planta has been demonstrated in Arabidopsis by both bimolecular fluorescent complementation (Abe et al., 2005) and chromatin immunoprecipitation (ChIP) using FT antibodies in plants over-expressing FD (Wigge et al., 2005).
The ChIP experiment showed enrichment for the C-box region in the AP1 promoter only under conditions promoting the expression of FT, suggesting that a stable FT–FD–AP1 complex was formed in Arabidopsis (Wigge et al., 2005). The presence of a similar interaction in wheat is indirectly supported by the in vitro interactions (TaFT–TaFDL2 and TaFDL2–VRN1 promoter) described above and by the spatial correlation between the expression of TaFDL2 and VRN1. High levels of TaFDL2 and VRN1 are observed in the leaves of wheat plants expressing TaFT, paralleling the induction of AP1 in the leaves of Arabidopsis when FD is expressed ectopically in this tissue (Wigge et al., 2005).
To explore further whether a similar TaFT–TaFDL2–VRN1 complex could be detected in wheat, we included both TaFT and TaFDL2 proteins in the same binding reaction with the VRN1 promoter segments (Appendix S1). Unfortunately, we failed to observe any supershift relative to the TaFDL2 protein used alone (Appendix S1). These results suggest that either there is no stable TaFT–TaFDL2–VRN1 complex in wheat, or, more likely, that the conditions used in the binding assays were not conductive to the formation of a stable complex. It is also possible that additional proteins are necessary to stabilize this complex. ChIP studies will be necessary to confirm the existence of a TaFT–TaFDL2–VRN1 complex in wheat.
Identification of the wheat FD homologue was not a trivial task due to the limited sequence conservation among bZIP transcription factors, which is generally restricted to the bZIP motif (Figure 1). Therefore, we relied on simultaneous interaction of the TaFDL candidates with TaFT in yeast two-hybrid tests and with the VRN1 promoter in an EMSA to select the best candidate. Only TaFDL2 fulfilled both selection criteria among the five TaFDL genes identified in wheat leaves, making it the best candidate for a functional homologue of Arabidopsis FD. Expression results showed that TaFDL2 transcripts accumulate in the vegetative and reproductive apices (Figure 5), suggesting that the TaFDL2 protein is present in the correct tissue and at the correct developmental stage to be involved in flowering induction once TaFT becomes available.
Duplication and functional differentiation of FT and FT2 genes
Phylogenetic and comparative mapping analyses have shown that orthologues of wheat TaFT and TaFT2 are present in barley and rice, indicating that the duplication that originated these paralogous genes pre-dated the divergence of the grasses (Faure et al., 2007; Yan et al., 2006). The same phylogenetic analyses also showed that the TaFT/TaFT2 duplication occurred after the divergence of the grass species from the dicots. Therefore, this duplication is independent of the FT/TSF (twin sister of FT) duplication in the Arabidopsis lineage (Faure et al., 2007; Yan et al., 2006). Arabidopsis tsf mutations delay flowering and enhance the phenotype of ft mutants, whereas TSF over-expression causes precocious flowering under short days (Michaels et al., 2005; Yamaguchi et al., 2005).
Unfortunately, similarities and differences between Arabidopsis FT and TSF (Michaels et al., 2005; Sung et al., 2006; Yamaguchi et al., 2005) cannot be used to infer the functions of TaFT and TaFT2 in the grasses because independently duplicated genes may have different sub-functionalization. Therefore, the specific roles of the TaFT and TaFT2 genes need to be studied directly in the grass species. Detailed micro-co-linearity studies between the barley HvFT gene region on chromosome arm 7HS and the rice region on rice chromosome arm 6S, including Hd3a (the functional orthologue of Arabidopsis FT), have confirmed that these genes are true orthologues (Yan et al., 2006). However, this relationship is complicated by a recent tandem duplication that occurred only in the rice lineage, resulting in closely linked genes Hd3a (OsFTL2) and RFT1 (OsFTL3) (Izawa et al., 2002; Kojima et al., 2002; Yan et al., 2006). Comparative mapping studies also confirmed that HvFT2, which is located on the short arm of barley chromosome 3H, is orthologous to rice OsFTL1, which is located on the co-linear region of the short arm of rice chromosome 1 (Faure et al., 2007).
In rice, a short-day plant, Hd3a transcript levels are rapidly induced under short days (SD), but this effect can be suppressed by short exposures to light in the middle of the night (night breaks). These characteristics are not observed for other rice FT-like genes (including OsFTL1 and OsFTL3), suggesting that Hd3a is the central gene in the photoperiod pathway in rice (Ishikawa et al., 2005). Recent studies have confirmed that the Hd3a protein moves from the leaves to the apices, where it induces flowering, confirming that it is the functional homologue of Arabidopsis FT (Tamaki et al., 2007).
Whereas over-expression of Hd3a results in early flowering in rice (Kojima et al., 2002), over-expression of OsFTL1 produces more complex phenotypes including elongation of internodes, loss of apical dominance, and a terminal tissue at the apical meristem (Izawa et al., 2002). This terminal tissue is composed of multiple glumes, and occasionally a terminal floret at the tip instead of a panicle, suggesting that OsFTL1 may be involved in regulation of panicle and floral development rather than in the regulation of flowering initiation, which is mainly regulated by Hd3a (Izawa et al., 2002).
The functions described above for Hd3a and OsFTL1 may also apply to the orthologous wheat TaFT and TaFT2 genes. The finding that these wheat paralogues interact with different FD-like proteins provides a putative molecular explanation for functional differentiation. TaFT interacts with the TaFDL2 and TaFDL6 proteins, whereas TaFT2 interacts with TaFDL13, which belongs to a different clade of FD-like proteins (Figure 1). We showed here that the interaction between TaFT and TaFDL2 probably confers on TaFT the ability to regulate VRN1 transcription and therefore to affect the fate of the SAM. In contrast, TaFDL13 (the TaFT2 partner) showed no interactions with the VRN1 promoter. As TaFDL13 is mainly expressed in the apical region, it may provide TaFT2 with specificity for targets expressed in this tissue.
A role for TaFT2 in processes occurring after the SAM reproductive differentiation is also supported by the transcription profiles of HvFT2. The proteins coded by TaFT2 and HvFT2 are 98% identical (excluding the seven initial amino acids), suggesting that they may also share similar structural and functional characteristics. Faure et al. (2007) found that HvFT2 is up-regulated by LD, but this up-regulation occurs 2–3 weeks after differentiation of the SAM. Our results in wheat exhibit a similar trend. The transcript levels of TaFT in the transgenic plants over-expressing TaFT are very high from the first leaf, whereas TaFT2 transcript levels were low at the first-leaf stage and reached high levels at the fifth-leaf stage (Figure 6e,f), suggesting a delayed induction compared with TaFT. In the non-transgenic control plants, the transcript levels of TaFT were 40–70-fold higher than those of TaFT2, also suggesting a more central role for TaFT. More importantly, the strong up-regulation of TaFT2 in the two TaFT over-expressing transgenic wheat lines relative to the control suggests that TaFT2 may be regulated directly or indirectly by TaFT (Figure 6e,f).
Similarities between the Arabidopsis and wheat flowering pathways
FT transcripts in Arabidopsis and wheat are rapidly up-regulated upon transfer of plants to LD (Faure et al., 2007; Imaizumi et al., 2003; Turner et al., 2005; Yan et al., 2006). In addition, over-expression of FT in transgenic plants results in precocious flowering, confirming the conserved role of this gene as a promoter of flowering (Kobayashi et al., 1999; Yan et al., 2006). The observed protein–protein interaction between TaFT and TaFDL2, and the protein–DNA interaction between TaFDL2 and the VRN1 promoter suggest that the molecular interactions downstream of FT are also conserved between wheat and Arabidopsis. These interactions are incorporated into our current model of flowering regulation in temperate cereals (Figure 7).
Figure 7. Model for the regulation of flowering time in temperate cereals. The horizontal dotted line represents the VRN1 promoter and the vertical bars indicate the eight exons of VRN1. The white oval represents the TaFDL2 protein interacting with the G-box (or hybrid boxes, grey rectangle) in the VRN1 promoter. The grey circle represents the TaFT protein interacting with TaFDL2. Arrows indicate induction and lines ending in a bar indicate repression. The doted line from VRN1 to VRN2 indicates a negative regulatory feedback loop. LD, long days; SD, short days.
Download figure to PowerPoint
The model presented in Figure 7 shows the interaction between TaFDL2 and the G-box and hybrid boxes in the VRN1 promoter (Figure 4). A similar protein–DNA interaction has been previously identified in Arabidopsis, in which FD binds to a 130 bp region of the AP1 promoter including a C-box motif (Wigge et al., 2005).
In this study, we observed a close correlation between VRN1 and TaFT transcript levels in transgenic wheat plants over-expressing TaFT. Lower transcript levels of TaFT in the transgenic plant Tr-1 relative to Tr-2, and at the first-leaf stage compared with the fifth-leaf stage (Figure 6a,b) were paralleled by similar differences in VRN1 transcript levels (Figure 6c,d). More importantly, VRN1 transcript levels were 500–1000-fold higher in the transgenic wheat plants over-expressing TaFT than in the non-transgenic control plants. Additional experiments in Arabidopsis showed that addition of a strong transcriptional activation domain to FT increases its flowering promoting activity (Wigge et al., 2005). Taken together, these results suggest that TaFT is a limiting factor in the activation of VRN1.
As we detected no binding between the TaFT protein and the VRN1 promoter by EMSA (data not shown), it is unlikely that TaFT can regulate transcription of VRN1 through direct protein–DNA interaction. These results, combined with the observed yeast two-hybrid interaction between TaFDL2 and TaFT (Figure 2) and the TaFDL2 interaction with the VRN1 promoter, suggest a model in which TaFT and TaFDL2 act in concert to regulate the transcription of VRN1 (Figure 7).
Differences between the Arabidopsis and wheat flowering pathways
Although conservation of the last steps of the photoperiod pathway suggests that their origin pre-dates the divergence of the monocot and dicot lineages, the differences in the vernalization pathways between Arabidopsis and the temperate cereals suggest independent origins. This is consistent with the view that the temperate Pooideae grasses are a specialized monophyletic group that evolved from subtropical bambusoid-like species (Clayton and Renvoize, 1986; Preston and Kellogg, 2008).
The central repressor in the vernalization pathway of Arabidopsis, the MADS box gene FLOWERING LOCUS C (FLC) (Michaels and Amasino, 1999; Sheldon et al., 1999), has not been found in the grass species, and VRN2, the central repressor of flowering in the vernalization pathway of the temperate cereals, has not been found in Arabidopsis (Yan et al., 2004b). FLC delays flowering by interacting with regulatory regions of FT in the leaves and SOC1 in the meristems (Helliwell et al., 2006; Searle et al., 2006; Wigge et al., 2005). Vernalization permanently down-regulates FLC, releasing FT and SOC1 to promote the transcription of AP1 (Michaels and Amasino, 1999).
In the temperate cereals, VRN2 is also down-regulated by vernalization (Yan et al., 2004b), but, in contrast to FLC, can also be down-regulated by short days (Dubcovsky et al., 2006; Trevaskis et al., 2006) (Figure 7). The position of VRN2 upstream of VRN1 and TaFT in the model is supported by results from transgenic winter wheat plants with reduced VRN2 transcript levels (Yan et al., 2004b) and isogenic lines for VRN2 (Yan et al., 2006). Plants with reduced or non-functional VRN2 transcripts showed up-regulation of TaFT and VRN1 and a significant acceleration of flowering (Yan et al., 2004b, 2006). The increase in VRN1 transcript levels is then followed by down-regulation of VRN2 (Figure 7), suggesting the existence of a regulatory feedback loop that has not been described in Arabidopsis (Loukoianov et al., 2005).
Arabidopsis and the temperate cereals also differ in the genes that show natural variation in vernalization requirement. In Arabidopsis, most of the natural mutants with reduced or non-vernalization requirements are concentrated in FRI and FLC (Gazzani et al., 2003; Michaels et al., 2003), whereas, in the temperate cereals, they were detected in VRN1 (Fu et al., 2005; Yan et al., 2003, 2004a), VRN2 (Cockram et al., 2007; Yan et al., 2004b; von Zitzewitz et al., 2005) and TaFT (Yan et al., 2006). The known natural mutations for TaFT and VRN1 are dominant for spring growth habit and are located within regulatory, rather than coding, regions of these genes, suggesting the disruption of recognition sites for a flowering repressor, probably VRN2.
An additional characteristic that seems to be unique for the temperate cereals is the high level of expression of VRN1 in the leaves (Schmitz et al., 2000; Yan et al., 2003). In contrast, AP1 transcripts are abundant in the induced SAM and floral primordia in Arabidopsis, but are undetectable or present at much lower levels in some vegetative tissues (e.g. vascular tissues of cotyledons, Abe et al., 2005). Interestingly, transgenic 35S:FD Arabidopsis plants with ectopic expression of FD show high levels of AP1 transcripts in the leaves (Wigge et al., 2005), suggesting that the spatial differences between AP1 and VRN1 transcription profiles are the result of differences between the spatial transcription profiles of Arabidopsis FD and its wheat homologue (TaFDL2). The results presented here support this hypothesis. We first identified TaFDL2 as the functional homologue of Arabidopsis FD by its ability to interact with TaFT and the VRN1 promoter, and then showed that TaFDL2 exhibits high transcript levels in the leaves of both vernalized and non-vernalized winter wheat plants from very early developmental stages (Figure 5). The simultaneous presence of TaFDL2 and TaFT in the leaves of spring or vernalized winter wheat plants provides a simple explanation for the simultaneous presence of VRN1. The regulation of TaFT by various environmental cues provides temporal specificity to this interaction.
Although the results presented above explain the molecular mechanism by which VRN1 transcription is regulated in the leaves, they do not explain why such a complex regulation of a meristem identity gene has been conserved in the leaves of the temperate grasses analyzed so far. Up-regulation of VRN1 transcripts in the leaves of winter genotypes has been confirmed in wheat (Danyluk et al., 2003; Trevaskis et al., 2003; Yan et al., 2003), barley (Trevaskis et al., 2003), Lolium (Petersen et al., 2004) and oats (Preston and Kellogg, 2008), suggesting conservation across several tribes of temperate cereals (Triticeae, Poeae and Aveneae).
An interesting observation is that VRN1 transcript levels are negatively associated with the accumulation of COR (cold-responsive) genes and with the degree of frost tolerance (Danyluk et al., 2003). Frost tolerance increases during the acclimatization of wheat plants to cold but non-freezing temperatures, but decreases after the transition between the vegetative and reproductive apices. Near-isogenic lines for the VRN1 gene carrying the recessive vrn1 allele (winter growth habit) can tolerate 11°C lower freezing temperatures than lines carrying the dominant Vrn1 allele (spring growth habit) (Limin and Fowler, 2006). Similarly, spring lines grown under SD, which down-regulates the VRN1 transcript levels, can tolerate 8.5°C lower temperatures than the same lines under LD (Limin and Fowler, 2006). A similar result has been reported for barley. In a double-haploid population segregating for VRN-H1, lines carrying the recessive vrn-H1 allele showed higher transcript levels of CBF (C-repeat binding factors) than those carrying the dominant Vrn-H1 allele (Stockinger et al., 2007). In addition, lines grown under SD (reduced VRN-H1 levels) showed higher CBF transcript levels than lines grown under LD. CBF transcription factors are rapidly up-regulated by cold temperatures, inducing the expression of COR genes and playing a critical role in cold acclimatization and frost tolerance in temperate cereals (Francia et al., 2004; Knox et al., 2008; Miller et al., 2006; Skinner et al., 2005; Vágújfalvi et al., 2003). Based on these results, it is tempting to speculate that the temperate cereals have developed the ability to use the presence of VRN1 in the leaves as a signal to down-regulate the cold-tolerance regulatory network. As VRN1 is up-regulated upon the arrival of the spring, its presence would prevent up-regulation of the frost tolerance genes in the spring but not in the fall (autumn), when cold temperatures are an indication of future frost events.