Molecular Genetics and Genomics

, Volume 277, Issue 3, pp 301–313

Wheat VIN3-like PHD finger genes are up-regulated by vernalization


  • Daolin Fu
    • Department of Plant SciencesUniversity of California
  • Mignon Dunbar
    • Department of Plant SciencesUniversity of California
    • Department of Plant SciencesUniversity of California
Original Paper

DOI: 10.1007/s00438-006-0189-6

Cite this article as:
Fu, D., Dunbar, M. & Dubcovsky, J. Mol Genet Genomics (2007) 277: 301. doi:10.1007/s00438-006-0189-6


The term ‘vernalization’ describes the acceleration of the transition between the vegetative and reproductive stages after exposing plants to an extended period of low temperature. In Arabidopsis, vernalization promotes flowering by silencing the flowering repressor gene FLOWERING LOCUS C (FLC). Mitotically stable repression of FLC is the result of chromatin modifications mediated by the Vernalization-INsensitive 3 (VIN3) and VIN3-Like (VIL) proteins. In this study, we identified and characterized three VIL genes in diploid wheat (Triticum monococcum L.), named TmVIL1, TmVIL2, and TmVIL3. Similar to Arabidopsis VIN3, all three wheat VIL proteins carry three conserved domains including a plant homeodomain finger motif (PHD), a fibronectin type III domain (FNIII), and a VIN3 interacting domain (VID). Genetic mapping placed TmVIL1, TmVIL2, and TmVIL3 loci in the centromeric regions of chromosome 5, 6, and 1, respectively. The chromosome location of TmVIL1 is close to that of the vernalization gene VRN-D5, but more precise mapping information is required to validate this relationship. Transcription of the wheat VIL genes was up-regulated by vernalization, with a peak after 4–6 weeks of cold treatment. When transferred back to warm conditions, transcript levels of the wheat VIL genes returned to pre-vernalization levels. In addition, the transcript levels of wheat VIL genes are affected by photoperiod. This study indicates that wheat VIL genes have retained a similar structure and transcriptional regulation as their Arabidopsis VIN3/VIL homologues, suggesting that they might have retained some of their functions.


WheatVernalizationPHD fingerVILVIN3



Degenerate cleaved amplified polymorphic sequences


Flowering locus C


Fibronectin type III


Long day




Plant homeodomain


Short day


VIN3 interacting domain




Vernalization-INsensitive 3


The timely transition from the vegetative phase to the reproductive phase is a critical developmental process for plant reproductive success. Multiple flowering promotion pathways have evolved to respond to different environmental cues, and to integrate these signals in a coordinated flowering response (Koornneef et al. 1998; Simpson and Dean 2002). Responses to photoperiod and vernalization are two of the most important mechanisms controlling flowering in many plants (Gilbert 1926). The vernalization response, the acceleration of flowering by a long exposure to a chilling treatment (Chouard 1960), is particularly important for plants growing in cold regions.

The molecular basis of vernalization has been investigated in detail in Arabidopsis by characterizing a set of genes, defined by mutations that reduce the vernalization response (Chandler et al. 1996). In vernalization-responsive Arabidopsis ecotypes, the late-flowering habit is the result of the interaction between dominant, late-flowering genes FRIGIDA (FRI) and FLOWERING LOCUS C (FLC) (Michaels and Amasino 1999; Sheldon et al. 1999). The MADS-box gene FLC is a repressor of flowering, whose transcription is up-regulated by FRI to a level that inhibits flowering. Vernalization represses FLC, inducing flowering.

Once vernalized, the cells in Arabidopsis apices ‘remember’ their vernalized state through multiple mitotic divisions, a hallmark of an epigenetic event (Wolffe and Matzke 1999; Sheldon et al. 2000, 2002). The establishment of this vernalized state, and the mitotically stable repression of FLC in Arabidopsis require two interacting plant homeodomain (PHD) finger proteins designated VERNALIZATION-INSENSITIVE 3 (VIN3) (Sung and Amasino 2004a) and VIN3-LIKE 1 (VIL1) (Sung et al. 2006). VIN3 transcripts are detectable only after an effective cold exposure has been reached for vernalization, and are directly correlated with both the duration of the cold treatment and the degree of FLC repression (Sung and Amasino 2004a). VIL1 transcript levels show only a small increase in response to vernalization but a larger response to short days (SD). VIL1 is required for the repression of FLC and the FLOWERING LOCUS M (FLM); an FLC related flowering repressor (Sung et al. 2006).

Plant homeodomain domains, such as the ones observed in VIN3 and VIL1, are often found in proteins involved in chromatin remodeling complexes (Aasland et al 1995). Chromatin immunoprecipitation (ChIP) studies have shown a direct interaction between VIN3 and FLC chromatin regions that are remodeled during vernalization (Bastow et al. 2004; Sung and Amasino 2004a). The vernalization-mediated FLC repression and the corresponding chromatin changes in histone H3 acetylation and dimethylation are not observed in vin3 (Sung and Amasino 2004a) or vil1 mutants (Sung et al. 2006). In addition to the functional Vin3 and Vil1 alleles required to establish the FLC chromatin modification, two additional genes, designated VRN1 and VRN2 (Gendall et al. 2001; Levy et al. 2002), are necessary to maintain the repressed FLC state when plants are returned to higher temperature conditions after the vernalization treatment. Arabidopsis VRN1 and VRN2 genes are different from the wheat and barley vernalization genes with the same names (Yan et al. 2003, 2004a).

In the temperate cereals, the VRN1 gene is a MADS-box transcription factor closely related to the Arabidopsis meristem identity gene APETALA1 (Yan et al. 2003). A large proportion of the natural allelic variation in vernalization requirement in the temperate cereals is associated with mutations in the promoter and first intron of this gene (Yan et al. 2004b; Fu et al. 2005), a phenomenon that has not been reported in Arabidopsis. A more critical difference in the vernalization pathways between these species is the absence of clear homologues of wheat VRN2 in Arabidopsis, and the absence of clear homologues of FLC in the temperate cereals (Yan et al. 2004a). The VRN2 gene from the temperate cereals is a zinc-finger CCT domain gene down-regulated by both vernalization and SDs (Yan et al. 2004a; Dubcovsky et al. 2006).

In spite of these differences, it is currently not known if the temperate cereals and Arabidopsis share similar genes in the early steps of their vernalization pathways. To answer this question, we characterized the wheat homologues of the VIN3 and VIL1 genes, which are currently the most upstream genes identified in the Arabidopsis vernalization pathway. In this study, we characterized three VIL genes in diploid wheat, established the relationship between their protein products with VIN3/VIL proteins in rice and Arabidopsis, determined their chromosome locations, and examined their transcription profiles under different vernalization and photoperiod conditions.

Materials and methods

Plant materials and growth conditions

The lines of diploid wheat (Triticum monococcum L., 2n = 14, genome AmAm) analyzed in this study were T. monococcum ssp. monococcum DV92 (spring growth habit) and T. monococcum ssp. aegilopoides G3116 (winter growth habit). These two lines are the parents of the segregating population previously used to construct a linkage RFLP map in diploid wheat (Dubcovsky et al. 1996).

For the SD vernalization experiment, G3116 seeds were planted under long day conditions (LD: 16 h of light/8 h of dark) in a greenhouse at non-vernalizing temperatures (20–25°C). Then, 10-week-old plants were transferred to a cold chamber (4°C) under SD conditions (8 h of light at 30 μmol m-2 s-1/16 h of dark). After 6 weeks of cold treatment, plants were moved back to the LD greenhouse. Plant materials were evaluated immediately before transferring the plants to the cold chamber (0 week), every 2 weeks during the vernalization treatment (2, 4, and 6 weeks), and 2 weeks after returning the plants to the greenhouse under LD conditions (2 weeks out).

For the LD vernalization experiment, G3116 seeds were planted in the greenhouse under LD conditions (20–25°C, 16 h of light/8 h of dark). Two-week-old seedlings were transferred to a growth chamber under LD conditions (16°C, 16 h of light at 230 μmol m-2 s-1/8 h of dark). Three weeks later, 5-week-old plants at the six-leaf stage were transferred to a vernalization chamber under LD conditions (4°C, 16 h of light at 147 μmol m-2 s-1/8 h of dark) for 4 weeks. Meanwhile, control plants were maintained under the original growth conditions. Plants were evaluated immediately before the transfer to the cold chamber (0 week), and then 4 weeks later for both the vernalized (4°C) and control plants (16°C).

For the photoperiod response experiment, G3116 seeds were planted in the greenhouse under LD conditions (20–25°C, 16 h of light/8 h of dark). Two-week-old seedlings were transferred to two growth chambers, one under LD (16°C, 16 h of light at 230 μmol m-2s-1/8 h of dark) conditions, and the other under SD (16°C, 8 h of light at 211 μmol m-2 s-1/16 h of dark) conditions. Three weeks later, the 5-week-old plants at the six-leaf stage were split into two groups: half were maintained under the same growth conditions (control plants), and the other half was transferred to a growth chamber with the opposite photoperiod. After additional 4 weeks, leaf samples were collected for RNA extractions from the plants transferred to opposite photoperiods as well as from control plants.

DNA extraction, gene cloning, and Southern blotting

Genomic DNA was isolated from leaves of single plants by following a previously published procedure (Dvořák et al. 1988). The coding regions of wheat VIL genes were initially cloned from cDNA samples from winter plants (G3116) using the primers described in Table 1. The corresponding genomic sequences were subsequently obtained from the spring plants using clones from the T. monococcum DV92 BAC library (Lijavetzky et al. 1999). Using the FailSafe™ PCR kit (Epicenter, Madison, WI, USA), the whole fragments delimited by the cloning primers were amplified from isolated BAC clones and sequenced. Flanking sequences were recovered by the inverse PCR technique (Fu et al. 2006). Procedures for Southern blots and hybridizations were performed according to Dubcovsky et al. (1994). DNA probes used in the hybridization experiments were prepared from purified PCR products amplified with primers described in Table 1.
Table 1

PCR primers used in this study


Primer name

Primer sequence (5′ to 3′)

PCR cloning



















DNA Hybridization



















Genetic mapping



















Real-time PCR





































Genetic mapping

The wheat VIL genes were mapped on a population of 74 F2 plants from the cross DV92 × G3116 (Dubcovsky et al. 1996). The TmVIL1 locus was mapped using a Degenerate Cleaved Amplified Polymorphic Sequences (dCAPS) marker (Michaels and Amasino 1998), which used the primer pair TmVIL1/dCAPS/F1 and TaVIL1/R3 (Table 1), followed by digestion with the restriction enzyme SspI. The TmVIL2 locus was mapped using a 3-bp polymorphic indel detected with PCR primers TmVIL2/Ex2F2 and TmVIL2/I2R2 (Table 1). The TmVIL3 locus was mapped also using a dCAPS marker, with primer pair TmVIL3/F2 and TmVIL3/dCAPS/R1 (Table 1) followed by digestion with the restriction enzyme RsaI. PCR amplification conditions included an initial denaturation cycle at 94°C for 5 min, followed by 40 cycles of 30 s at 94°C, 30 s at 50°C (TmVIL1), 58°C (TmVIL2), or 54°C (TmVIL3), and 30 s at 72°C, and a final 10 min extension step at 72°C. Polymorphic fragments were separated using a 6% acrylamide gel. Genetic linkage maps were constructed using Mapmaker Version 3.0 (Lander et al. 1987) using the Kosambi function (Kosambi 1944).

Real-time quantitative PCR

Total RNA was prepared using the TRIZOL reagent (Invitrogen, Carlsbad, CA, USA) from leaves as described by Yan et al. (2003). First strand cDNA synthesis was performed using the SuperScript™ First-Strand Synthesis System (Invitrogen) according to the manufacturer’s instruction. Gene transcript levels were determined using the SYBR® GREEN PCR Master Mix (Applied Biosystems, Foster City, CA, USA). Real-time PCR assays were performed on an ABI PRISM 7000 SDS (Applied Biosystems) in a final volume of 20 μl containing 1× SYBR Green PCR Master Mix, 1 μM primers (forward and reverse), and 1 μl cDNA template. After an initial incubation at 95°C for 10 min to activate the AmpliTaq Gold® DNA polymerase, thermal conditions followed 40 cycles of 95°C for 15 s and 60°C for 1 min, and a final dissociation protocol in a temperature gradient from 60 to 95°C.

In all, five target genes, TmVIL1, TmVIL2, TmVIL3, TmVRN1, and TmVRN2, were investigated. TmVRN1 was used as reference in the vernalization experiments and TmVRN2 served as reference for photoperiod experiment. TmACT (ACTIN) was used as an endogenous control for data normalization in all experiments. PCR primers used for the expression analysis are listed in Table 1. A standard curve was constructed for each target gene using ten-fold serial dilutions of a cDNA sample. The efficiency of the primers used in this study was 96% or higher. Expression analysis was performed with 12 biological replicates for each treatment in the SD vernalization experiment, and with 14 biological replicates per treatment both in the LD vernalization experiment and the photoperiod response study. To eliminate the effect of unusually large (or low) values, data points that were more than three standard deviations apart from the mean of each treatment were eliminated from the analyses (this criteria resulted in the elimination of two data points in the SD vernalization experiment, one data point in LD vernalization experiment, and one data point in the photoperiod response experiment, respectively). The 2-ΔΔCT method (Livak and Schmittgen 2001) was used to normalize and calibrate transcript values relative to the endogenous controls.

Sequence analysis

Sequences similar to Arabidopsis VIN3 and VIL proteins were retrieved from GenBank ( using BLASTP and TBLASTN algorithms. Sequences were assembled using the Vector NTI program (Version 9.0.0, InforMax, Frederick, MD, USA). Multiple sequence alignments were obtained using the program CLUSTALW Version 1.8 (Thompson et al. 1994). Primers used for regular PCR, dCAPS marker development, and real-time PCR were designed using the PRIMER3 program (Rozen and Skaletsky 2000), the dCAPS Finder 2.0 (Neff et al. 2002), and the Primer Express software (Version 2.0, Applied Biosystems), respectively. The phylogenetic tree was based on the multiple sequence alignment of the three conserved domains and was constructed using the program MEGA 3.1 (Kumar et al. 2004).

The predicted translation product of VIN3/VIL genes was examined for the presence of protein structural features. Conserved protein motifs were identified by searching the Pfam database Version 18.0 (Bateman et al. 2004). Coiled–coil domains were predicted using the program Coils, Version 2.1 (Lupas et al. 1991). The Nuclear Localization Signals (NLS) were predicted using the PredictNLS program (Cokol et al. 2000), and the sub-cellular localization was predicted using the LOCtree program (Nair and Rost 2005).


Cloning of wheat VIL genes

Four rice proteins with regions of at least 40% similarity with Arabidopsis proteins AtVIN3 (At5g57380), AtVIL1 (At3g24440), AtVIL2 (At4g30200), and AtVIL3 (At2g18880) (Sung et al. 2006) were detected by a BLASTP search of the rice genome. All 16 comparisons, between rice and Arabidopsis VIN3/VIL proteins, showed E-values between e-44 and e-118 indicating highly significant similarities. Other rice proteins exhibited short regions of low similarity (lower than 22%) with significantly higher E-values (E > e-10) and were, therefore, excluded from further analyses. The four rice proteins showing high similarity with the four Arabidopsis genes were designated as rice VIL proteins, and named OsVIL1 (ABA98812), OsVIL2 (BAD38062), OsVIL3 (AAT94000), and OsVIL4 (BAD03519). A two-letter prefix is used to indicate the source species of the VIN3/VIL genes described in this study (At: Arabidopsis thaliana, Os: Oryza sativa, Tm: Triticum monococcum).

A recent study of the Arabidopsis VIL genes by Sung et al. (2006) reports an additional VIL gene: AtVIL4 (At2g18870), which encodes only the distal VIN3 Interacting Domain (VID). Using this additional Arabidopsis VIL protein to query the rice genome, the same four OsVIL1-4 proteins were recovered (E < e-12) from the rice genome, suggesting that these four OsVIL proteins are likely a complete set of VIL proteins present in the rice genome sequence.

The newly identified Arabidopsis and rice VIN3 and VIL proteins were then used to explore the wheat EST database using TBLASTN. Sequence assembling of the multiple wheat ESTs, recovered in the previous search, resulted in three full-length gene sequences, which closely corresponded to three of the proteins from rice. PCR primers specific for each of these three genes (Table 1) were used to recover and sequence cDNAs from T. monococcum accession G3116. The three diploid wheat genes were named TmVIL1 (DQ886919), TmVIL2 (DQ886917), and TmVIL3 (DQ886918) based on their similarity with the rice genes. Wheat ESTs corresponding to TmVIL2 were the most significant matches when OsVIL4 was used to query the wheat EST database.

To better understand the evolutionary relationship among the different VIL proteins, a phylogenetic tree was constructed using the Neighbor-Joining (NJ) method (Fig. 1). The three TmVIL proteins showed close relationship with the rice proteins designated with the same number, and those three clusters were supported by 100% bootstrap confidence values. The OsVIL4 rice protein was more closely related to the wheat and rice VIL2 proteins (71% bootstrap value). The wheat and rice VIL1 and VIL3 proteins formed a higher level cluster that was more closely related to Arabidopsis AtVIL1 (100% bootstrap value) than to the other Arabidopsis VIN3/VIL proteins. Arabidopsis AtVIL2 and AtVIL3 proteins formed a cluster supported by high-bootstrap values (100%), but the relationship between this cluster and the other VIN3/VIL proteins was not supported by high-bootstrap values and should be considered with caution. Similarly, Arabidopsis AtVIN3 showed no close relationship with the VIL proteins.
Fig. 1

Phylogenetic tree of VIN3/VIL proteins based on the simultaneous analysis of the three conserved domains PHD, FNIII, and VID presented in Tables 2, 3 and 4. Multiple sequence alignments of the conserved domains were produced by Clustal W. The neighbor-joining (NJ) tree was constructed using the complete deletion option from MEGA 3.1. Bootstrap values are indicated close to the nodes and are based on 1,000 bootstrap replicates (only bootstrap values higher than 50% are shown)

AtVIL4 (Sung et al. 2006) lacks the PHD and fibronectin type III (FNIII) domains and therefore was excluded from the previous analysis. We constructed a separate NJ tree including AtVIL4 using only the VID domain, which is conserved across all VIN3/VIL proteins (Table 3). The VID-based phylogenetic tree placed AtVIL2, AtVIL3, and AtVIL4 within one cluster supported by a high-bootstrap value (81%), and retained the same groups as Fig. 1 (data not shown).

Genomic organization of wheat VIL genes

DNA hybridization of DV92 genomic DNA Southern blots (two restriction enzymes) with probes for the three wheat VIL genes indicated that TmVIL1, TmVIL2, and TmVIL3 are all single or low-copy genes (Fig. 2). Hybridization of the probe TmVIL3 with DNA digested by DraI resulted in two restriction fragments, due to the presence of a DraI restriction site within the TmVIL3 probe. The same probes were used to screen the complete T. monococcum BAC library by DNA hybridization.
Fig. 2

DNA hybridization of wheat VIL genes. Labels for each lane indicate the probe used and the restriction enzymes used to digest the T. monococcum DV92 genomic DNA (M = 1-kb DNA ladder)

The genomic region for each VIL gene was completely sequenced by PCR-based walking using selected clones from the DV92 T. monococcum BAC library. The genomic sequences of TmVIL1 from BAC 652J2 (DQ886922), TmVIL2 from BAC 657C7 (DQ886920), and TmVIL3 from BAC 655P13 (DQ886921) were deposited in GenBank. Comparisons of these genomic regions with the corresponding cDNAs revealed that the TmVIL1 and TmVIL2 have four exons whereas TmVIL3 has only three (Fig. 3).
Fig. 3

Gene structure and domain organization of wheat VIL genes. Rectangles represent exons. Lines represent introns. ATG and TAA designate the start codon and stop codon, respectively. Protein motifs and specialized structures are indicted as CC coiled-coil domain, FNIII fibronectin type III domain, NLS nuclear localization signal, PHD plant homeodomain motif, and VID VIN3 interacting domain

Polymorphisms between parental lines DV92 and G3116 were used to map the three wheat VIL genes on the T. monococcum genetic map (Dubcovsky et al. 1996) using PCR markers (Table 1). The TmVIL1 locus was mapped completely linked to a large group of loci clustered at the centromeric region of chromosome 5Am (e.g., Xcdo749 5AmS and Xabc706 5AmL) (Fig. 4). The chromosome arm location of TmVIL1 is currently unknown. The TmVIL2 locus was mapped on the centromeric region of the short arm of chromosome 6Am, 3.2 cM proximal to RFLP locus XksuE3 and 2.1 cM distal to the Xpsr113 marker, which was completely linked to the centromere (Fig. 4). Finally, TmVIL3 was mapped on chromosome 1Am, completely linked to RFLP locus Xpsr1201, which was previously mapped on the centromeric region of the long arm of chromosome 1Am (Fig. 4). Interestingly, all three wheat VIL genes were mapped to the centromeric regions of the linkage maps.
Fig. 4

Chromosome location of TmVIL genes

Orthology between the three T. monococcum VIL genes and the corresponding rice genes (same gene number based on the phylogenetic analysis, Fig. 1) was further supported by the mapping results. OsVIL1 is located on the long arm of rice chromosome 12 (≈21.2 Mb), which is co-linear with the centromeric region of wheat homoeologous group 5. OsVIL2 is located on the short arm of rice chromosome 2 (≈2.9 Mb), which is co-linear with wheat chromosome 6, where the TmVIL2 gene was mapped. Finally, OsVIL3 is located on the short arm of rice chromosome 5 (≈2.6 Mb), which is co-linear with wheat chromosome 1, where TmVIL3 was mapped.

Analysis of wheat VIL proteins

All three wheat VIL proteins, TmVIL1 (697 amino acids), TmVIL2 (750 amino acids), and TmVIL3 (615 amino acids) include a PHD finger domain. This domain is located at the end of the second exon in TmVIL1 and TmVIL2 and at the end of the first exon in TmVIL3 (Fig. 3). All VIN3/VIL proteins (Table 2) show the conserved Cys4-His-Cys3 signature characteristic of the PHD zinc finger domain (Aasland et al. 1995).
Table 2

PHD finger domain alignment
The Arabidopsis VIN3/VIL proteins also include a conserved domain at the carboxy end of the proteins, designated VID based on its role on the physical interactions among this group of proteins (Sung et al. 2006). This region was also well conserved in all the wheat and rice VIL proteins included in this study (Fig. 3; Table 3).
Table 3

VID domain alignment
A FNIII domain was found between the PHD and VID domains in the Arabidopsis VIN3/VIL1 proteins (Sung et al. 2006). This FNIII domain is less conserved in the grasses (Table 4) than the PHD and VID domains described above. AtVIN3 and all members of VIL1 and VIL3 proteins showed higher similarity to the pfam FNIII domain than AtVIL2, OsVIL2, OsVIL4, and TmVIL2 (Table 4).
Table 4

FNIII domain alignment

The LOCtree program predicted a nuclear localization for the three wheat VIL proteins and classified them as DNA-binding proteins. The presence of a putative NLS (TVKRC) in TmVIL1 provided additional evidence of the nuclear localization of this protein (Fig. 3). A comparison with known coiled–coil proteins showed that both TmVIL1 and TmVIL3 contained a protein segment that would likely adopt a coiled–coil conformation (Fig. 3).

Transcription of wheat VIL genes under vernalization treatment

Short day vernalization

All three wheat VIL genes showed a significant up-regulation during the course of vernalization. Significant differences (P < 0.001) in transcript levels were detected between the vernalized and un-vernalized plants (Fig. 5a–c). TmVIL2 and TmVIL3 showed maximum transcription levels after 4 weeks of vernalization (4w, Fig. 5b, c), whereas the maximum for TmVIL1 was observed after 6 weeks of vernalization (6w, Fig. 5a).
Fig. 5

Time course of TmVIL transcript levels during SD vernalization. Transcript levels of TmVIL1 (a), TmVIL2 (b), TmVIL3 (c), and the vernalization reference TmVRN1 (d) were estimated by quantitative PCR. The values on the Y-axis represent normalized and calibrated values relative to the endogenous ACTIN control using the 2-ΔΔCT method. Values for each treatment are averages of 12 10-week-old G3116 plants. RNA samples were extracted from leaves immediately before vernalization (0w) and then every 2 weeks (2w, 4w, and 6w). The final samples were collected 2 weeks after transferring the plants back to warm conditions (2w out)

None of the wheat VIL genes showed significant differences (P > 0.05) in transcript levels between the samples collected before vernalization (0w, Fig. 5) and those collected 2 weeks after moving the plants out of the cold room and back to the greenhouse under LD and warm conditions (2wOut, Fig. 5). These results indicate that the increases in TmVIL transcript levels observed during the vernalization treatment are not permanent. The VRN1 gene, which was included as a reference (Fig. 5d), showed the expected up-regulation of transcript levels (P < 0.001) that is characteristic of a successful vernalization process (Yan et al. 2003).

Long day vernalization

This second experiment, using 14 5-week-old plants per treatment, showed a significant up-regulation of the three wheat VIL genes after 4 weeks of vernalization (Fig. 6a–c). In all cases, highly significant differences (P < 0.001) were observed in the statistical contrast between the vernalized plants (4w ver., Fig. 6) and the average of the un-vernalized plants before vernalization (0w, Fig. 6) and the control plants maintained under warm conditions (4w no v., Fig. 6). Of the three VIL wheat genes, the TmVIL1 transcripts had the most pronounced response to cold (Fig. 6a), a result that was also observed under SD vernalization (Fig. 5).
Fig. 6

Effect of long day vernalization on wheat VIL transcript levels. Transcript levels of TmVIL1 (a), TmVIL2 (b), TmVIL3 (c), and the vernalization control TmVRN1 (d) were estimated by quantitative PCR. The values on the Y-axis represent normalized and calibrated values relative to the endogenous ACTIN control using the 2-ΔΔCT method. Values for each treatment are averages of 14 5-week-old G3116 plants. RNA samples were extracted from leaves immediately before vernalization (0w) and 4 weeks after vernalization (4w ver.). Control plants were maintained in the chamber under non-vernalization conditions and were measured simultaneously with the plants in the cold chamber (4w no v.)

The contrast between the samples collected before vernalization (0w, Fig. 6) and the control plants maintained under warm conditions (4w no v., Fig. 6) showed no significant differences for TmVIL1 and TmVIL3 (P > 0.05). However, TmVIL2 transcript levels were higher (P < 0.01) in the un-vernalized samples collected from 9-week-old plants (4w no v., Fig. 6b) relative to the samples collected from un-vernalized 5-week-old plants (0w, Fig. 6b). The VRN1 gene, which was included as a reference (Fig. 6d), showed significantly higher transcript levels (P < 0.001) after 4 weeks at 4°C than before this treatment, confirming a successful vernalization process.

Photoperiod response of the transcription of wheat VIL genes

To simplify the description of the complex results from this experiment, we will use a two-letter abbreviation for the different treatments where the first letter indicates the photoperiod (SD or LD) during the first three initial weeks and the second letter indicates the photoperiod during the next 4 weeks. For example, the SL treatment indicates plants grown initially under SD and then transfer to LD, whereas SS indicate the control plants that were maintained under SD for the whole duration of the experiment (Fig. 7).
Fig. 7

Quantitative PCR determination of the effect of photoperiod on wheat VIL transcript levels. Transcript levels of TmVIL1 (a), TmVIL2 (b), TmVIL3 (c), and TmVRN2 (d). The photoperiod sensitive TmVRN2 gene (Dubcovsky et al. 2006) was added as reference. The values on the Y-axis represent normalized and calibrated values relative to the endogenous ACTIN control using the 2-ΔΔCT method. Values for each treatment are averages of 14 9-week-old G3116 plants. RNA samples were extracted from leaves corresponding to four different treatments: LL = 7-week LD, LS = 3-week LD followed by 4-week SD, SS = 7-week SD, and SL = 3-week SD followed by 4-week LD

Both TmVIL1 and TmVIL2 showed significantly higher transcript levels (P = 0.04 and 0.001, respectively) in plants grown initially under SD (SS and SL, grey bars, Fig. 7a, b) than in the plants grown initially under LD (LL and LS, black bars). However, the transfer from SD to LD or from LD to SD after the first 3 weeks had no significant effects on the transcript levels of these two genes (P > 0.10).

On the contrary, the transcript levels of TmVIL3 showed no significant differences (P = 0.48) between plants grown initially under SD and LD (Fig. 7c), but highly significant increases (P < 0.001) in the plants transferred from LD to SD (SL) relative to the LD controls (LL). A favorable effect of SD on TmVIL3 transcript levels was also observed on the plants grown initially under SD. Plants maintained under continuous SD (SS) showed slightly higher TmVIL3 transcript levels than those moved to LD (SL) (Fig. 7c). However, the differences were not significant (P = 0.18).

The photoperiod responsive gene TmVRN2 was measured as a reference to evaluate the effect of the conditions in each growth chamber. In agreement with a previous study (Dubcovsky et al. 2006), SD had a significant effect in reducing TmVRN2 transcript levels (P < 0.001) (LL versus LS and SS versus SL, Fig. 7d). On average, plants grown initially under SD showed 30% lower TmVRN2 transcript levels than those grown initially under LD (P < 0.001). The TmVRN2 transcription data confirmed that the plants in this experiment showed the expected SD and LD responses for other known photoperiod responsive genes.


VIL map location and relationship to loci affecting vernalization

In temperate cereals, the vernalization requirement is controlled by a number of vernalization genes designated VRN1 to VRN5 (McIntosh et al. 2003). The Triticeae vernalization genes VRN1 and VRN2 have been mapped to the middle of the long arm of homoeologous group 5 and on the distal region of the long arm of homoeologous group 4, respectively (Dubcovsky et al. 1998). Their chromosome locations suggest that VRN1 and VRN2 are not related to the VIL genes, a fact that was confirmed by the recent cloning of the two vernalization genes (Yan et al. 2003, 2004a). The VRN-B4 gene, recently re-designated VRN-B3 (Yan et al. 2006), has been mapped on the short arm of chromosome 7B (Law and Wolfe 1966), which indicates that allelic variation at this locus is not related to the VIL loci.

TmVIL1 was mapped completely linked to centromeric markers on chromosome 5Am, a location that is close but not identical to the one reported for the vernalization gene VRN-D5 from ‘Triple Dirk F’ (Kato et al. 2003). VRN-D5 was mapped 4.4 cM distal to microsatellite marker Xgdm3 (Kato et al. 2003), which is located on the 5D long arm 6–12 cM distal to the centromere (Pestsova et al. 2000). This places VRN-D5 10–16 cM distal to the centromere and not exactly in the TmVIL1 region (Fig. 4).

TmVIL2 was mapped on the centromeric region of homoeologous group 6, which does not correspond to any of the mapped vernalization loci in wheat and barley. However, in Chinese Spring, increased dosage of group 6 chromosomes delays ear-emergence whereas reduced dosage of the same chromosomes accelerates ear emergence. These effects are removed by vernalization (Islam-Faridi et al. 1996). To see if there is a relationship between the gene controlling these differences in flowering time and TmVIL2, it would be necessary to map the flowering gene on chromosome 6 more precisely.

TmVIL3 was mapped to the centromeric region of chromosome 1, which coincides with the original location reported for the VRN-H3 vernalization gene from barley. This preliminary result was based on the loose linkage of VRN-H3 with a morphological marker (45 cM) located on chromosome 1H (Yasuda 1969). However, more precise mapping results have confirmed that VRN-H3 is actually located on homoeologous group 7 (Yan et al. 2006), eliminating TmVIL3 as a candidate for VRN3.

Conservation of VIL gene structure

The Arabidopsis and the grass VIL proteins share several conserved structural elements. Proteins from both species have a conserved PHD finger motif, a zinc-finger motif characterized by a conserved Cys4-His-Cys3 (C4HC3) pattern (Aasland et al. 1995). PHD fingers are found in a large number of genes associated with chromatin-mediated transcriptional regulation (Aasland et al. 1995; Sung and Amasino 2004a). The VIN3 and VIL1 mediated chromatin modification is essential to the establishment of a mitotically stable vernalized state and the epigenetic control of vernalization in Arabidopsis (Sheldon et al. 2002; Amasino 2004; Sung et al. 2006).

The TmVIL proteins also have a putative FNIII domain as reported before for the AtVIN3/AtVIL genes (Sung et al. 2006). The FNIII domain is one of the repeated structural motifs discovered in Fibronectin (FN) (Pankov and Yamada 2002) and found to be involved in protein–protein interactions (Aspberg et al. 1997). The specific role of this domain in the VIN3/VIL genes is currently unknown.

Finally, the wheat VIL proteins carry a carboxy terminal VID domain, which is conserved among all protein members of this family. Two-hybrid assays using a series of AtVIN3 deletions and other AtVIL proteins demonstrated that the VID domain is necessary and sufficient for positive two-hybrid interactions (Sung et al. 2006).

Transcription profile of TmVIL genes

One of the most interesting results from this study was the up-regulation of the wheat VIL genes by vernalization and SD. Although a basal transcript level of the three VIL genes was observed in the leaves of wheat plants grown under warm conditions, a significant increase of the VIL transcript levels occurred after long-term exposure to cold conditions, during both SDs and LDs. Depending on the gene, an increase in transcript levels continued up to the fourth or sixth week of vernalization (Fig. 5). Another similarity between the transcription profiles of the three TmVIL genes and AtVIN3 was the transient effect of the vernalization up-regulation. When plants were returned to warm conditions after vernalization, transcript from all three TmVIL genes reverted to their pre-vernalization levels (Fig. 5).

The TmVIL genes showed relatively high-basal transcript levels (only 6–21-fold lower than the highly expressed ACTIN endogenous control) that transiently increased during vernalization. Compared to the Arabidopsis genes, this transcription profile is more similar to AtVIL1 and AtVIL2 than to AtVIN3. AtVIN3 transcript levels are almost undetectable before vernalization or within 3 days after plants were returned to a warm growth temperature after vernalization (Sung and Amasino 2004a).

In addition to responding to vernalization, Arabidopsis VIL1 is also responsive to photoperiod (Sung et al. 2006). SD promotes the transcription of Arabidopsis VIL1 and early flowering under non-vernalized SD conditions. Transcript levels of the wheat VIL genes were generally higher under SD conditions, but these differences were detected only during the early developmental stages for the TmVIL1 and TmVIL2 genes and later in development for the TmVIL3 gene.

We currently do not know if the higher transcript levels of the TmVIL genes observed under vernalization and SD are the result of transcription up-regulation or increased stability of the corresponding mRNAs.

Biological role of VIL genes

In Arabidopsis, the cellular memory of the vernalized state is mediated by histone modifications of regulatory regions located at the promoter and the first intron of FLC, a central flowering repressor down-regulated by vernalization (Michaels and Amasino 1999; Bastow et al. 2004; Sung and Amasino 2004a, b). The establishment of a silenced FLC state requires histone H3 deacetylation of Lys 9 and Lys 14 and dimethylation of Lys 9 and Lys 27 (Bastow et al. 2004; Sung and Amasino 2004a). Presumably, vernalization-induced expression of AtVIN3 is required for the H3 deacetylation in FLC chromatin, which in turn promotes the chromatin dimethylation at H3 Lys 9 and Lys 27 (Sung and Amasino 2004a).

Plant homeodomain finger containing proteins such as AtVIN3 are often members of multi-subunit chromatin-remodeling complexes. This was recently confirmed by studying other AtVIN3 like genes in Arabidopsis. The AtVIN3 protein was shown to interact with the AtVIL1 and AtVIL2 proteins in yeast-two-hybrid experiments (Sung et al. 2006). A recent study also suggests that AtVIN3 and AtVRN2 are part of a large protein complex (≈1,000 kDa) containing a number of polycomb-group (PcG) proteins, which are required for the repression of FLC by vernalization (Wood et al. 2006).

The characterization of vin3 and vil1 mutants confirmed that both genes are required for the vernalization-mediated enrichment of di- and tri-methylation at the H3 Lys 9 and Lys 27 of FLC chromatin. Similarly, both alleles are needed for chromatin changes in FLM, an FLC-related flowering repressor. The FLC-independent vernalization pathway (Michaels and Amasino 2001) also requires a functional AtVIN3 allele (Sung and Amasino 2004a). One of the genes from this pathway, the MADS-box gene AGL19, is a flowering promoter normally repressed by several PcG proteins. Vernalization relieves AGL19 from the PcG repression by a mechanism that requires a functional AtVIN3 allele (Schonrock et al. 2006). These results suggest the possibility that the vernalization and photoperiod responses of the wheat VIL genes might be related to the regulation of other MADS-box gene(s) different from FLC, which has not been detected so far in the temperate cereals.

Arabidopsis vil1 mutants exhibit a substantial delay of flowering under SD independent of vernalization, suggesting that AtVIL1 promotes flowering under SD. Sung et al. (2006) hypothesized that the AtVIL1 mediated repression of FLM under SD serves to attenuate the photoperiodic response.

The similar effects of SD and vernalization on the transcript levels of Arabidopsis and wheat VIL genes, together with their conserved structural features, suggest that this group of genes might have related functions. It would be interesting to produce wheat VIL mutants to test their effect on flowering time under SD conditions, vernalization requirement, and chromatin remodeling of genes involved in the regulation of flowering in the temperate cereals. We have initiated a wheat TILLING project to generate VIL mutants and further explore their role in the regulation of flowering in temperate cereals.


This research was supported by the United States Department of Agriculture CSREES NRI competitive grant 2006-01160, and NSF-Plant Genome grant DBI-0321462. The work has been carried out in compliance with the current laws governing genetic experimentation in the USA.

Copyright information

© Springer-Verlag 2006