Seeds of B. distachyon Bd21 (PI 254867) and Bd29-1(PI 639818) were sown in 6-inch pots containing Sun Gro Hort soil mix (Bellevue, WA, United States) in the greenhouse at 25°C, 16/8h (day/night) with an irradiance of 450 ± 50 μmol m^-2 s^-1.

A fragment of BdVRN1 (414 bp) gene was amplified by PCR (primers listed in Supplementary Table S2) with the addition of four bases of CACC at its 3′ end. PCR was performed as follows: 94°C for 5 min; 35 cycles of 94°C for 20 s, 59°C for 30 s, and 72°C for 50 s; final extension at 72°C for 1 min. The PCR product was separated on 1% agarose gel and extracted from the gel. Cleaned gene fragment was cloned into a Gateway^® entry vector pENTR^TM/D-TOPO^® (Life Technologies, Grand Island, NY, United States) containing attL and attR recombination sites according to the manufacturer’s protocol. The Gateway^® compatible pANDA vector (Miki and Shimamoto, 2004) containing two cassettes in inverse orientation linked by a small fragment of the GUS gene was used as the destination vector for BdVRN1. The attL × attR reaction is mediated by Gateway^® LR Clonase^TM II enzyme mix (Life Technologies, Grand Island, NY, United States). Kanamycin (nptII) and hygromycin (hpt) resistance genes were used for the selection in bacteria and plants, respectively. The final binary vector pANDA::BdVRN1 was verified by Sanger sequencing (Supplementary Figure S1).

The pANDA::BdVRN1 vector was introduced into Agrobacterium tumefaciens C58C1 strain for transformation of Brachypodium following the protocol developed by Vogel and Hill (2008). The selection medium contained 150 mg L^-1 Timentin (bioWORLD, Dublin, OH, United States) to suppress Agrobacterium growth and 40 mg L^-1 hygromycin B (bioWORLD, Dublin, OH, United States) to kill untransformed calli. Two cycles of selection, each lasting 2 weeks were performed under dark at 23°C. Resistant calli were transferred into the regeneration medium containing Kinetin (KT) at 1 mg L^-1 and hygromycin at 10 mg L^-1 in a tissue culture chamber at 23°C, 16/8 h (light) with an irradiance of 120 μmol m^-2 s^-1. Shoots started to appear 7–10 days after the transfer. Rooting took place on a MS medium supplemented with 0.1 mg L^-1 NAA and 10 mg L^-1 hygromycin. Well-rooted plantlets were carefully moved into 6-inch pots containing Sun Gro Hort soil mix (Bellevue, WA, United States). Plants were grown in a growth chamber at 23°C, 16/8 h (day/night) with an irradiance of 400 ± 30 μmol m^-2 s^-1.

Transgenic plants were screened by PCR using primers (HPT-F 5′ GAATTCAGCGAGAGCCTG 3′, HPT-R 5′ ACATTGTTGGAGCCGAAA 3′) designed from the sequence of the hygromycin resistant gene present in the pANDA binary vector. Eight independent T₀ lines were confirmed by PCR analyses. In two of the independent lines, RNAi-4 and RNAi-12, PCR-positive and negative plants segregated in a 3:1 ratio in the progenies, indicating the integration of the transgene occurred at a single locus in each of the two lines. Homozygous plants of T₂ RNAi-4 and RNAi-12 lines were chosen for further analyses.

Cold acclimation was done by growing plants in growth chamber at a 4°C and 8/16 h (day/night) with an irradiance of 400 ± 30 μmol m^-2 s^-1 for 10 days. To determine the vernalization requirement for flowering, 4-week-old plants of Bd 21 or Bd29-1 accession were placed into a growth chamber at 4°C and 8/16 h (day/night) with an irradiance of 400 ± 30 μmol m^-2 s^-1 for 3, 6, 9, or 12 weeks before they were moved back to a greenhouse (25°C, 16/8 h day/night) with an irradiance of 450 ± 50 μmol m^-2 s^-1.

Freezing tolerance was evaluated for Bd21, Bd29-1and BdVRN1 RNAi mutants. Five-week-old plants were treated with -5°C and 8/16 h (day/night) for 12 h. Following the freezing treatment, plants were moved into a growth chamber at 4°C for recovery for 1 day and were then transferred into a greenhouse (25°C, 16/8 h day/night) and evaluated for recovery 1 week later. Percentage of plants that survived the freezing test and regrew were recorded.

To examine the expression levels of BdVRN genes (Supplementary Table S1) in response to vernalization treatment, 4-week-old plants of each accession were placed into a growth chamber at 4°C and 8/16 h (day/night) with an irradiance of 400 ± 30 μmol m^-2 s^-1. Leaves and meristem were harvested separately 2, 4, 6, 8, 10, or 12 weeks following the conclusion of the vernalization treatment.

For semiquantitative RT-PCR analysis, total RNA was extracted from leaves or meristems with TRIzol^® Reagent (Invitrogen, Carlsbad, CA, United States). Reverse transcription was performed with the SuperScript^® III First-Strand Synthesis System for RT-PCR kit (Invitrogen, Carlsbad, CA, United States). PCR was performed with gene specific primers (See Supplementary Table S2 for primer information) with the following program: pre-denaturation at 94°C for 5 min; 25 cycles of denaturation at 94°C for 30 s, primer-annealing at 58°C for 30 s, elongation at 72°C for 50 s, and post elongation at 72°C for 5 min for all genes. The expression level of each gene was normalized to that of the BdGAPDH reference gene.

For Real-Time RT-PCR analysis, total RNA extraction and reverse transcription were done the same way as described for the semiquantitative RT-PCR analysis. Quantitative analyses were carried out on the Eco^® Real-Time PCR System (Illumina, Inc., San Diego, CA, United States) using the SYBR^® GreenER^TM qPCR SuperMix kit (Invitrogen, Carlsbad, CA, United States) according to the manufacturer’s instructions. The quantity of PCR products was determined at the end of each cycle by the Eco^® Software v4.0 (Illumina, Inc., San Diego, CA, United States). The expression level of each gene was normalized to that of the BdGAPDH gene, and the expression level for each gene in the wild type without treatment was set as 1.0. Primers used for PCR amplification are listed in Supplementary Table S3.

Time to flower was determined for accessions Bd29-1and Bd21 with or without vernalization treatment. Bd21 started flowering within 24 ± 3 days, regardless of the vernalization treatment (Figure 1), indicating vernalization treatment had no significant effect on accelerating flowering in Bd21. In contrast, without vernalization treatment Bd29-1 did not flower even after 140 days when record-taking was stopped. When subjected to a 3-week vernalization treatment, Bd29-1 started flowering after 109 ± 6.2 day. A 6-week vernalization treatment of Bd29-1 reduced the time to flower to just 27 ± 4.2 days. Longer than 6 weeks of vernalization treatment did not reduce the number of days to flowering further. Clearly, accession Bd29-1 requires vernalization treatment to flower.

Three genes of VRN1, VRN2, and VRN3 that regulate the vernalization requirement in wheat or barley have been characterized. Previous comparative genomic analysis showed that Brachypodium has a VRN1 homolog, Bradi1g08340 (BdVRN1), located in a position co-linear to the rice (OsMADS14) and wheat (TaVRN-1) homologous genes, and a VRN3 homolog, Bradi1g48830 (BdVRN3). Although there was no apparent homologous VRN2-like gene in Brachypodium, Bradi3g10010 is a member of the group IV CCT gene, an intermediate between the barley HvCO9 and cereal VRN2 genes in the phylogenetic tree (Higgins et al., 2010). For these reasons, we designated Bradi3g10010 as BdVRN2.

We next assayed the expression of the BdVRN1, 2 and 3 genes in both Bd21 and Bd29-1 with or without vernalization treatment. Without vernalization treatment, expression of BdVRN1 expression remained at a low level in leaves and the meristem during the first 3 weeks following seed germination for both Bd21 and Bd29-1 (Figure 2A and Supplementary Figure S2). Four weeks after germination, BdVRN1 expression was dramatically increased by 20-fold in the meristem in Bd21, whereas it remained unchanged in Bd29-1 (Figures 2A,B). Following various length of vernalization treatments, expression of BdVRN1 in the meristem of Bd21 remained at a high level, whereas in Bd29-1 it was gradually increased to a level similar to that of Bd21 after 6 weeks of vernalization treatment (Figures 2A,B).

BdVRN2 was expressed at a high level in leaves (Supplementary Figure S2A) but was expressed at a very low level in the meristem of both Bd21 and Bd29-1 (Figure 2C). Semiquantitative RT-PCR showed that BdVRN2 was highly expressed in leaves of Bd21 with or without vernalization and its expression in Bd29 appeared to increase slightly after a 2-week vernalization; but longer vernalization did not increase further (Supplementary Figure S2B). Interestingly, the expression of BdVRN2 in the meristem is not influenced by vernalization in both Bd21 and Bd29-1 (Figure 2C and Supplementary Figure S2B). In Bd21, expression of BdVRN3 in the meristem increased continuously before vernalization treatment took place, and did not change significantly by vernalization treatment (Figure 2D). In contrast, in Bd29-1, the expression of BdVRN3 in the meristem remained at a low level without vernalization, but increased gradually with vernalization treatment and increased by sixfolds when it reached the peak following a 8-week vernalization (Figure 2D).

To determine whether BdVRN1 indeed regulates flowering time, RNAi mutants of BdVRN1 were created for Bd21. Eight independent T₀ transgenic lines were confirmed by PCR analysis and two of them, RNAi-4 and RNAi-12 did not flower without vernalization (Figures 3A,C). Real-Time RT-PCR showed that the BdVRN1 transcripts were reduced to 0.12 ± 0.02 in RNAi-4 and 0.26 ± 0.04 in RNAi-12 (Figure 3B). After a 6-week vernalization treatment, RNAi-4 and RNAi-12 mutants started flowering after 35.5 ± 5.6 days and 28.2 ± 8.3 days, respectively, compared to 19.2 ± 5.4 days for the wild type. The phenotypic change of Bd21 from non-vernalization-requiring to vernalization-requiring in the BdVRN1 knockdown mutants strongly indicate that BdVRN1 is of paramount importance to flowering induction in Brachypodium.

To better understand how BdVRN1 affects flowering at the molecular level, we analyzed the expression patterns of BdVRN1, BdVRN2, and BdVRN3 in the RNAi-4 knockdown mutant. The expression of BdVRN1 was significantly suppressed in 4-week-old seedlings of the RNAi mutant, but it increased significantly following a 6-week vernalization treatment, although it is still only half of the value observed for the wild type without cold acclimation (Figure 4A). Moreover, the expression level of BdVRN3 also decreased in the BdVRN1 mutants, but its expression was also elevated by vernalization treatment (Figure 4C). However, the expression of BdVRN2 was not affected in the mutant (Figure 4B). These results indicated that upregulation of BdVRN1 is positively correlated with BdVRN3, but not with BdVRN2.

To determine the relationship between the vernalization requirement and freezing tolerance, we compared the freezing tolerance of Bd29-1 and Bd21. Without a vernalization treatment, Bd29-1 showed a much higher freezing tolerance than did Bd21 based on the percentage of plants that survived the freeze-thaw-regrowth test. When plants of both accessions were vernalized for 6 weeks, the recovery rate of Bd21 was nearly unchanged whereas that of Bd29-1 greatly decreased (Figure 5A). With a cold acclimation, both Bd21 and Bd29-1 had an increase in recovery rate compared with their non-cold acclimated counterparts (Figure 5B), which clearly indicated that both accessions have a well-defined cold acclimation mechanism to increase freezing tolerance upon an exposure to low temperatures. The result that vernalization treatment reduced the freezing tolerance of Bd29-1, but not Bd21suggested that vernalization requirement is negatively correlated with freezing tolerance. Furthermore the BdVRN1 RNAi-4 mutant that gained vernalization requirement is also much more freezing tolerant than that of the wild type Bd21 with or without cold acclimation (Figures 5A,B). Semiquantitative RT-PCR experiments confirmed that without vernalization treatment Bd29-1 and the Bd21 RNAi-4 mutant had lower BdVRN1 expression levels than the wild type Bd21, but a 6-week vernalization treatment increased the expression levels of BdVRN1 in both Bd29-1 and the BdVRN1 RNAi-4 mutant approaching to that in Bd21 (Figure 5C). This increase in BdVRN1 expression coincides with the reduced freezing tolerance.

The regulation of gene expression in response to cold is quite complex with more than a 1000 genes are cold induced, but a critical group of them are organized into a cascade controlled through a regulatory hub involving cold-inducible CBF genes (Chinnusamy et al., 2007; Thomashow, 2010). The CBF genes encode members of the AP2/ERF family of transcription factors, which bind to the CRT/DRE regulatory elements in target genes of the CBF regulon and increase freezing tolerance (Jaglo-Ottosen et al., 1998; Liu et al., 1998; Catala et al., 2003; Dubouzet et al., 2003; Ito et al., 2006; Xiong and Fei, 2006; Lata and Prasad, 2011). For example, CBF1 in Arabidopsis is a transcriptional activator for the COR genes, such as COR47 and COR414-TM1, which are involved in cold and dehydration responses (Xiong et al., 2002; Sakuma et al., 2006; Griffith et al., 2007). DREB2A is a transcription factor induced by dehydration in Arabidopsis, rice, soybean, and maize (Simpson et al., 2003; Sakuma et al., 2006; Qin et al., 2007; Cui et al., 2011; Mizoi et al., 2013). The RD (RESPONSIVE TO DESICCATION) genes, for example RD26 and RD29B, are induced by cold or drought stress through the regulation of the transcription factor DREBs (Yamaguchi-Shinozaki and Shinozaki, 1994; Nakashima et al., 2006; Msanne et al., 2011). To further characterize the relationship between BdVRN1 gene and freezing tolerance, we analyzed the expression of some important cold-responsive genes by real-time RT-PCR in both the wild type Bd21 and the BdVRN1 RNAi-4 mutants without vernalization treatment.

Previously, we isolated eight CBF genes (CBF1 Bradi4g35630, CBF2 Bradi4g35620, CBF3 Bradi4g35650, CBF4 Bradi4g35570, CBF5 Bradi4g35580, CBF6 Bradi4g35590, CBF7 Bradi4g35600, and CBF8 Bradi4g35650) from Brachypodium that are tandemly arranged in chromosome 4 and are all cold induced (not published). Here we compared their expression with four other important cold-responsive genes between the wild type Bd21 and the BdVRN1 RNAi-4 mutants. We showed that CBF2, CBF3, CBF5, CBF6, COR414-TM1, and COR47 were highly induced by 3- to 600-fold in both the wild type and the BdVRN1 RNAi-4 mutants after a mere 24 h of cold treatment (Figure 6). More interestingly, CBF2, CBF3, CBF5, CBF6, and DREB2A were highly expressed in the BdVRN1 RNAi-4 mutant without cold induction, and the constitutive expression of CBF3 and DREB2A in the RNAi mutant was even higher than that of the cold-treated wild type. On the other hand, RD26 and RD29B genes showed no noticeable changes in the RNAi mutant, compared to the wild type.