Characterization of FLOWERING LOCUS T1 (FT1) Gene in Brachypodium and Wheat

The phase transition from vegetative to reproductive growth is a critical event in the life cycle of flowering plants. FLOWERING LOCUS T (FT) plays a central role in the regulation of this transition by integrating signals from multiple flowering pathways in the leaves and transmitting them to the shoot apical meristem. In this study, we characterized FT homologs in the temperate grasses Brachypodium distachyon and polyploid wheat using transgenic and mutant approaches. Downregulation of FT1 by RNAi was associated with a significant downregulation of the FT-like genes FT2 and FT4 in Brachypodium and FT2 and FT5 in wheat. In a transgenic wheat line carrying a highly-expressed FT1 allele, FT2 and FT3 were upregulated under both long and short days. Overexpression of FT1 caused extremely early flowering during shoot regeneration in both Brachypodium and hexaploid wheat, and resulted in insufficient vegetative tissue to support the production of viable seeds. Downregulation of FT1 transcripts by RNA interference (RNAi) resulted in non-flowering Brachypodium plants and late flowering plants (2–4 weeks delay) in wheat. A similar delay in heading time was observed in tetraploid wheat plants carrying mutations for both FT-A1 and FT-B1. Plants homozygous only for mutations in FT-B1 flowered later than plants homozygous only for mutations in FT-A1, which corresponded with higher transcript levels of FT-B1 relative to FT-A1 in the early stages of development. Taken together, our data indicate that FT1 plays a critical role in the regulation of flowering in Brachypodium and wheat, and that this role is associated with the simultaneous regulation of other FT-like genes. The differential effects of mutations in FT-A1 and FT-B1 on wheat heading time suggest that different allelic combinations of FT1 homoeologs could be used to adjust wheat heading time to improve adaptation to changing environments.


Introduction
The optimization of flowering and seed production is critical for plant survival and, in seed crops, to maximize grain yields. This is particularly important for cereal crops, which contribute significantly to global food production. However, the regulatory gene network controlling flowering is complex and its manipulation requires a precise understanding of the roles of different components and their interactions. In the temperate grasses, which include wheat, barley and the model species Brachypodium distachyon, vernalization (long exposures to cold temperatures) and photoperiod (variation in day length) are the main seasonal signals regulating flowering time. Over the last two decades, substantial progress has been made toward our understanding of the genes involved in these two pathways [1][2][3].
Vernalization requirement is conferred by the genes VERNAL-IZATION1 (VRN1) [4,5], VRN2 [6], VRN3 [7], and VRN4 [8,9]. Among these four genes, only the first three have been cloned and characterized to date. VRN1 is a MADS-box meristem identity gene homologous to Arabidopsis AP1, which is upregulated during vernalization and promotes flowering in the spring [4]. VRN2 is a floral repressor that encodes a protein containing a zinc finger motif and a CCT domain (CONSTANS, CONSTANS-LIKE, and TIMING OF CAB1-1) [6], and is downregulated by vernalization and short days (SD) [10,11]. The VRN3 gene encodes a RAF kinase inhibitor-like protein with high similarity to Arabidopsis protein FLOWERING LOCUS T (FT) and is a flowering promoter [7]. This gene will be referred to hereafter as FT1 to differentiate it from other FT-like genes present in the temperate grasses ( [12]).
These three vernalization genes are interconnected by complex interactions [3]. VRN2 acts as a repressor of FT1 in the leaves, preventing flowering in the fall. During the winter, FT1 and VRN2 transcripts are reduced to almost undetectable levels whereas VRN1 transcripts increase proportionally to the duration of the cold period [7]. At the end of the winter, the presence of VRN1 represses VRN2 facilitating the upregulation of FT1 during the lengthening days of spring [13]. FT1 expression then further upregulates VRN1 to reinforcing this inductive pathway and result in an irreversible transition to flowering.
Under long days (LD), FT1 is upregulated by the photoperiod gene PHOTOPERIOD 1 (PPD1), which is responsible for most of the natural variation in photoperiod sensitivity in wheat and barley [14][15][16]. The convergence of the photoperiod and vernalization pathways in the regulation of FT1, place this gene at the center of the gene network regulating flowering time in the temperate cereals [3].
In addition to its role in the integration of flowering signals, the small proteins encoded by FT and FT-like genes are essential for the transmission of environmental signals from the leaves to the shoot apical meristem, as demonstrated in Arabidopsis, rice and other species [17][18][19][20]. FT1 protein transport has yet to be demonstrated in the temperate cereals, but indirect evidence suggests that a similar mechanism exists in these species. FT1 is known to interact with the bZIP transcription factor FDL2, which binds to the promoter of VRN1 [21]. Although FT1 is not expressed in the wheat apex [22], a good correlation has been observed between the transcript levels of FT1 in the leaves and of VRN1 in the apex [21]. A significant correlation between flowering and FT1 transcript levels in the leaves has also been observed in a sample of Brachypodium accessions [23].
Previous studies have shown that over-expression of FT1 accelerates flowering in Brachypodium and wheat [24][25][26] but the effects of the downregulation of this gene in these species have not been reported before. In this study, we validate the FT1 overexpression results and determine the effect of the downregulation of this gene by RNA interference (RNAi) in Brachypodium and wheat. We also use these transgenic lines to characterize the expression of the closest FT-like genes (FT2 to FT6). Finally, we test the effect of mutations in FT1 homoeologs in tetraploid wheat to determine their individual and combined contribution to heading time, and discuss their potential value for engineering wheat flowering time.
Brachypodium plants were grown in a growth chamber at 25uC and LD photoperiod (
The FT1 overexpression (FT1 OE ) constructs were developed in binary vectors pCAMBIA1300 and pGWB5 ( Figure 1A-B) [27]. The pCAMBIA1300 vector was used for the overexpression of FT1 cDNAs of Bd21-3 and Langdon, and the FT1 genomic region of Langdon. The pGWB5 vector was used for the overexpression For FT1 RNAi , we selected fragments from Brachypodium Bd21-3 cDNA (position 79 to 447) and G3116 cDNA (position 62 to 367). The selected RNAi triggers did not share any stretch of more than 18 identical nucleotides with other FT-like genes, preventing co-silencing of other FT-like genes. The Brachypodium FT1 RNAi trigger was cloned in the pCAMBIA1300-based vector ( Figure 1C), and the wheat FT1 RNAi trigger was cloned in the pANDA-based vector ( Figure 1D) [28]. In both constructs, expression of the selected RNAi triggers was driven by the maize Ubiquitin promoter (Ubi).

Generation of Transgenic Plants
Tissue culture and Agrobacterium-mediated transformation of Brachypodium was conducted as reported by Dr. Vogel and his colleagues [29]. Transgenic Brachypodium plants were established in soil in the growth chamber and in the greenhouse.
Protocols for the tissue culture and biolistic bombardment of wheat were adapted from a previous study [30]. Immature caryopses from T. aestivum cultivar Bobwhite were harvested two weeks after anthesis, sterilized with 70% (v/v) ethanol containing 0.05% (v/v) Tween 20 for 5 min, then with 20% (v/v) bleach supplemented with 0.05% (v/v) Tween 20 for 15 min, and washed 3-5 times using sterile distilled water. Immature embryos (ca. 1 mm long) were isolated from the sterilized caryopses, placed with the scutellum facing upward on the dissection media (MS  shoots with well-developed roots were established in soil in the greenhouse. The biolistic bombardment was performed using the PDS-1000/He Particle Delivery System (Bio-Rad Laboratories, USA). To prepare three bombardments, 2.1 mg of microcarriers (Gold particles of 0.6 mm in diameter; Bio-Rad, USA) were measured into a 1.5 ml microcentrifuge tube, sterilized by mixing with 35 ml pure ethanol, recovered by spinning (12,000 rpm for 5 s) and removing the supernatant, rinsed in 200 ml ice-cold sterile distilled water, and collected by spinning and removing the supernatant. The pre-treated microcarriers were resuspended in 245 ml prechilled sterile water containing 20 mg plasmid DNA, and combined with another 250 ml pre-chilled CaCl 2 (2.5 M). Where required, solutions in the previous steps were mixed thoroughly by pipetting. The microcarrier suspension was then supplied with 50 ml pre-chilled spermidine solution (1.45%, v/v) and mixed immediately by vortexing in the cold room (4uC) for 15-20 min. The plasmid-coated microcarriers were recovered by centrifuga-    tion (12,000 rpm for 10 s) and, following supernatant removal, were resuspended in 36 ml pure ethanol. For each bombardment, 10 ml gold suspension was loaded to the center of a macrocarrier disk (Bio-Rad), air-dried in the laminar flow hood, and placed in the microcarrier launch assembly under the 1100 psi rupture discs. Sixty immature embryos arranged in a 3.5-cm diameter circle were placed 6-cm below the macrocarrier assembly. The PDS-1000/He System was operated according to the manufacturer's instruction. Bombardment conditions were 1,300 psi helium pressure and 25 mm Hg vacuum.
Putative transgenic plants were confirmed by PCR analysis. The presence of the selection marker (BAR or HPT) and other vectorspecific fragments were tested in putative transgenic plants using PCR. Quantitative reverse transcription PCR (qRT-PCR) was used to quantify the downregulation of FT1 transcription in the RNAi transgenic plants and to confirm the expression of the FT1 RNAi constructs. In addition, transgenic wheat plants were confirmed by testing their resistance to 0.3% (v/v) Finale herbicide.

Mutant Screening and Genetic Analyses in Tetraploid Wheat
Genome-specific primers were designed for both FT-A1 and FT-B1 genes ( Table 1) and were used to screen a tetraploid wheat TILLING population consisting of 1,384 EMS-treated individuals of the tetraploid wheat cultivar Kronos [31]. The screening was performed using a Cel I-based method described previously [31]. For the mutations described in this study, the first letter indicates the original base or amino acid and the last letter represents the mutant base or amino acid. The number in the middle indicates the position of the mutation counted from the ATG start codon in the genomic DNA sequence, or from the initial methionine in the predicted protein. For FT-A1, we selected mutant line T4-474 that carries a G655A mutation resulting in a premature stop codon. This mutant will be referred to hereafter as ft-A1. For FT-B1 no truncation mutations were found, so two different mutations resulting in amino acid substitutions at conserved amino acids were selected from lines T4-263 (C856T) and T4-344 (G98A), designated hereafter as ft-B1 263 and ft-B1 344 , respectively. The predicted amino acid changes in each of the mutant lines are described in the results section.
We backcrossed ft-A1 three times to wild type Kronos and then crossed it with an M 3 plant of the mutant line ft-B1 263 to generate a BC 1 F 2 population segregating for both mutations. Epistatic interactions between FT-A1 and FT-B1 were analyzed in this population using a factorial ANOVA. Data from this population was also used to calculate the degree of dominance for FT-A1 and FT-B1 using the formula: where X 1 , X 2 and X 3 are the heading time values, respectively, of the plants homozygous for the mutant late flowering allele (ft), the heterozygotes, and the plants homozygous for the wild type early flowering allele (FT). The degree of dominance for FT-A1 was calculated using only ft-B1 263 homozygous plants, and the degree of dominance for FT-B1 was calculated using only ft-A1 homozygous plants.
Mutant line ft-B1 344 was backcrossed to Kronos and a separate BC 1 F 2 segregating population was developed to evaluate the effect of this mutation on heading time. Both segregating populations were evaluated simultaneously for heading time in the same greenhouse experiment.
The transcript levels of FT-A1 and FT-B1 homoeologs were compared in three-week-old wild type Kronos plants grown in growth chambers under LD (   Quantitative Reverse Transcription PCR (qRT-PCR) Analysis Gene expression in FT1 RNAi transgenic plants was determined from six biological replicates. The youngest leaves of adult plants were collected at 12:00 p.m. for Brachypodium and 4:00 p.m. for hexaploid wheat. Total RNA was extracted from leaf tissues using the Trizol method (Life Technologies, Grand Island, NY, USA) according to the manufacturer's instructions. cDNA templates were prepared using the Fermentas First Strand cDNA Synthesis Kit (Thermo Scientific, Waltham, MA, USA). In total, eight target genes were studied (Table 1). Ten-fold serial dilutions of cDNA templates were used to test the amplification efficiency of each primer pair. In Brachypodium and in Bobwhite transgenic wheat and control lines, qRT-PCR was performed using the FastStart SYBR Green Master (Roche Applied Science, Indianapolis, IN, USA) on the StepOnePlus Real-Time PCR Systems (Life Technologies). The amplification conditions were one cycle of 10 min at 95uC, 40 cycles of two consecutive steps of 15 s at 95uC and 1 min at 60uC and a standard dissociation protocol. ACTIN was used as an endogenous control using primers designed by Fu, et al. (Table 1; [33]).
In tetraploid wheat and in FT1 HOPE lines, the youngest leaf was collected at 10:00 a.m. and RNA was extracted using the Spectrum Plant Total RNA Kit (Sigma-Aldrich, St. Louis, MO, USA). cDNA was synthesized using the High Capacity cDNA Reverse Transcription kit (Life Technologies) and used in qRT-PCR reactions performed on a 7500 Fast Real-Time PCR System (Life Technologies). Transcript levels of FT-A1 and FT-B1 homoeologs in tetraploid wheat were compared using genome specific primers (Table 1). ACTIN was used as an internal control using primers originally designed by Uauy, et al. (Table 1; [34]). The qRT-PCR conditions were one cycle of 20 s at 95uC and 40 cycles of two consecutive steps of 3 s at 95uC and 30 s at 60uC.
Transcript levels of target genes were calculated using the formula 1000*2 (ACTIN CT -TARGET CT) , which indicates the relative number of target molecules per 1000 molecules of ACTIN. The expression data was analyzed using SAS version 9.0 (SAS Institute Inc, Cary, NC, USA). Data that did not meet the assumptions of the ANOVA was transformed using power transformations to restore homogeneity of variances (Levene's test) and normality of residuals (Shapiro-Wilk test). Independent transgenic events were compared with the wild type control using Dunnett's tests, and the information was summarized using a contrast comparing the wild type versus all transgenic lines. Graphs were prepared using the GraphPad Prism version 5.01 (GraphPad Software, San Diego, CA, USA).

Pollen Staining
The viability of pollen grains in FT1 transgenic plants was evaluated using a simplified staining protocol [35]. Mature but non-dehiscent anthers were collected and submerged in 100 ml staining solution for 24 h at room temperature and in darkness. After rinsing in distilled water, anthers were transferred to a fresh slide containing a drop of distilled water, and then inspected under the stereomicroscope. To measure the percentage of viable pollen, anthers were gently pressed to release pollen grains. Pollen grains stained magenta-red were viable, and those stained blue-green were non-viable. Two-hundred pollen grains within the microscopic view, performed in triplicate, were used to calculate the proportion of viable pollen. Table 3. Summary of the ft1 mutants in tetraploid wheat cultivar Kronos.

Overexpression of FT1 Promoted Floral Organogenesis During Tissue Culture
To study the effect of increased FT1 expression on floral development, we developed transgenic Brachypodium and wheat plants overexpressing FT1. In the Brachypodium transformation experiment using the FT1 OE construct for Bd21-3 FT1cDNA driven by the maize UBIQUITIN promoter ( Figure 1A), none of the 180 calli regenerated under LD (16 h light/8 h dark) and only two out of 100 calli regenerated under SD (8 h light/16 h dark), grew weakly, and eventually died. In the second Brachypodium transformation experiment using the fusion construct of Bd21-3 FT1 and GFP driven by the 35S promoter ( Figure 1B), six of 280 infected calli showed regenerated shoots that immediately developed floral organs under LD (Figure 2A-C). As a result of the limited development of vegetative tissue, none of the T 0 plants produced seeds, and they did not survive when transplanted to soil. We then adjusted the light conditions during callus differentiation and shoot regeneration to a SD photoperiod and weak light intensity (29 mmol m 22 s 21 ). Under these conditions, the floral organogenesis was delayed and there was an increase in vegetative growth, but it was still insufficient to support the production of seeds. The lack of normal vegetative tissue also precluded the use of the GFP tag to study the localization of the FT protein in Brachypodium leaves.
In common wheat, we tested two different FT1 OE constructs, both driven by the maize Ubiquitin promoter: the first one using the FT-B1 coding region (Ubi::cFT-B1) and the second one using the FT-B1 genomic region (Ubi::gFT-B1) from tetraploid wheat Langdon ( Figure 1A). Of the 1,876 calli bombarded with the Ubi::cFT-B1 construct, 12 developed floral organs during culture. Of the 523 calli bombarded with the Ubi::gFT-B1 construct, three flowered during the culture stage. PCR analysis confirmed that all early-flowering T 0 plants were positive for the presence of the transgene (data not shown). In both cases, calli were maintained in recovery culture under dark conditions for four weeks and were then transferred to regeneration media under LD. Green shoots appeared five days later, and clusters of florets developed 3 to 4 weeks later ( Figure 2D). At this stage, florets were opened wide and exhibited a normal feathery stigma but with smaller and shriveled stamens. Most florets had complete structures (including lemma, palea, pistil and stamens, Figure 2E), but some florets did not develop stamens. In comparison to wild type plants, the anthers of the transgenic plants had fewer pollen grains and a large proportion of non-viable pollen, as determined by pollen staining ( Figure 2F). Similar to the transgenic Brachypodium plants, there was limited leaf development and none of the T 0 plants produced seeds in culture or were able to survive when transplanted to soil.

Downregulation of FT1 Delayed Heading Time of Transgenic Plants
To determine the effect of reduced FT1 expression on flowering development, we developed Brachypodium and wheat FT1 RNAi constructs that target only the FT1 gene. Of the 280 Brachypodium calli transformed with the Brachypodium FT1 RNAi construct ( Figure 1C), a total of 38 calli generated putative transgenic plants. PCR analyses demonstrated that all putative transgenic plants were positive for the FT1 RNAi construct. All Brachypodium transgenic T 0 plants grew vigorously, but failed to flower ( Figure 3A) even after two weeks of vernalization at 4uC, which is sufficient to induce flowering in the non-transgenic Bd21-3 control plants.
In the common wheat cultivar Bobwhite, the FT1 RNAi construct shares a continuous stretch of at least 179 identical nucleotides with the FT-A1, FT-B1, and FT-D1 homoeologs, and therefore, is expected to silence all three copies simultaneously. Of the 1,675 wheat calli bombarded with the FT1 RNAi construct, 13 independent plants were confirmed to be transgenic from their resistance to the herbicide Finale and by PCR analysis. Transgenic lines showed heterogeneity in the effect of the transgene on heading time, floral organ morphology and fertility ( Figure 3B-D, Table 2). Ten of the 13 transgenic plants showed moderate delays in heading time ranging from 2 to 4 weeks, two showed very late heading time (lines '1965' and '2547'), and one (line '2548') remained vegetative for ten months, before dying without flowering. Spikelets in line '2547' had two complete florets, but additional florets were undeveloped. Of the two intact florets, each consisted of one pistil and three stamens, but the stigmas remained closed ( Figure 3E) and the anthers stayed green and did not shed pollen. Of the two late transgenic lines only '1965' set seeds.
Progeny tests including 5 to 10 T 1 plants were performed for six independent transgenic lines, including the late-flowering line '1965' and five moderately late-flowering lines. The T 1 progeny of the late-flowering T 0 transgenic plant '1965' were not as late as the original T 0 plant (only 10-20 days delay compared to the wild type) suggesting that late flowering of the T 0 plant was not determined solely by the downregulation of FT1. Unfortunately, no seeds were available for the T 0 transgenic lines showing the greatest delay in flowering (lines '2547' and '2548') and, therefore, we were unable to verify the linkage between these late flowering phenotypes and the transgene.
Progeny for lines '140D', '152E' and '1965' segregated for early and late heading time (14 to 20 days later than the Bobwhite control). Progeny of '152E' developed relatively normal florets and set seeds. In a few lines, e.g. '140D-19 spikes failed to emerge from the sheath of the flag leaf resulting in abnormal curling of the spike and awns ( Figure 3D). Additionally, the florets developed pistil and stamen, but some florets had abnormal stigmas which failed to open after heading and did not set seeds. All plants from the progeny of lines '157B', '1381' and '1382' headed at the same time as the wild type control plants suggesting that they were not functional transgenic lines (Table 2). In summary, FT1 RNAi transgenic wheat lines flowered 2 to 4 weeks later than the wild type control, exhibited reduced pollen viability and in some cases mature stigmas did not open as widely as those in control plants ( Table 2).

Mutations in FT-A1 and FT-B1 Delay Flowering in Tetraploid Wheat
We screened a TILLING population of the tetraploid wheat cultivar Kronos and identified 38 mutant alleles for FT-A1 and 13 for FT-B1, and selected three for functional characterization ( Table 3). The G655A mutation present in the selected ft-A1 mutant disrupts the splice site located at the beginning of the second intron. Sequencing of the resulting ft-A1 cDNA using homoeolog-specific primers confirmed the elimination of the splice site, which resulted in a 4-bp insertion and an in-frame premature stop codon at position 88 (W88*). This premature stop codon eliminates the last 90 amino acids (50.8% of the protein), and almost certainly produces a non-functional protein.
No truncations or splice site mutations were identified for FT-B1, so two substitution mutations within the PEBP motif were selected based on high evolutionary conservation of the targeted amino acid and negative BLOSUM62 scores, which are predictive of altered structural and functional properties [36]. In mutant line ft-B1 263 , a C856T mutation resulted in a change between proline and serine (P77S, BLOSUM62 score = 21). The proline at position 77 is highly conserved in FT1 orthologs in Arabidopsis, maize, rice, diploid, tetraploid and hexaploid wheat and in all six FT-like proteins (FT1 to FT6) in barley and wheat (Table 3). Mutant line ft-B1 344 carries a G to A mutation at position 98 (G98A), which results in the amino acid change G33E (BLO-SUM62 score = 22). The original amino acid (glycine) is conserved in the FT1 orthologs in Arabidopsis, maize, rice, barley, diploid, tetraploid and hexaploid wheat. In the other FTlike proteins this amino acid position is substituted by alanine, asparagine or aspartic acid, but in no case by glutamic acid as in the G33E mutation (Table 3).
To quantify the effect of the ft-A1 truncation mutation, the ft-B1 263 substitution mutation, and their interaction we genotyped 110 plants from a BC 1 F 2 population segregating for these two mutations and determined their heading time under LD without vernalization (Kronos has a spring growth habit). The wild type alleles for early flowering showed partial dominance over the mutant alleles, with an estimated degree of dominance of approximately 0.46 and 0.59 for FT-A1 and FT-B1, respectively.
Based on this partial dominance we decided to merge the homozygous wild type and heterozygous classes into a single nonmutant class. This strategy has the advantage of simplifying the analysis and description of the interactions between these two genes, and of increasing the power of the statistical analyses. The resulting 262 factorial ANOVA (using FT-A1 and FT-B1 as factors) revealed significant differences in heading time both for FT-A1 (P = 0.0075) and FT-B1 (P,0.0001), as well as a significant interaction (P = 0.0143, Figure 4A) between the two genes. This interaction indicates that the effect of each FT1 homoeolog on flowering time is dependent on the allele present in the other homoeolog.
To describe these interactions, we analyzed the simple effects of each FT1 homoeolog within the two classes of the other FT1 homoeolog. The FT-B1 alleles showed highly significant differences in heading time (P,0.0001) within both alleles of FT-A1, but the differences between the adjusted means of the FT-B1 alleles were 34% larger in the lines homozygous for the mutant ft-A1 allele (25.0 day delay in flowering) than in the lines homozygous or heterozygous for the wild type FT-A1 allele (18.6 day delay). The effects of the FT-A1 alleles on heading time were smaller than those of the FT-B1 alleles, and the differences were significant only within the homozygous ft-B1 263 mutant class (6.7 day delay; P = 0.0374). In all cases, plants carrying mutant ft1 alleles flowered later than plants homozygous or heterozygous for the wild type alleles. Plants homozygous for both ft-A1 and ft-B1 263 mutant alleles flowered on average 25.0 days later than plants homozygous or heterozygous for the two wild type alleles, confirming an important role of FT1 in the acceleration of wheat flowering under LD.
To characterize the effect of the ft-B1 344 mutation (G98A) on heading time we genotyped and phenotyped 23 BC 1 F 2 plants segregating for the ft-B1 344 mutation in a genetic background fixed for the wild type FT-A1 allele. Plants homozygous for the wild type FT-B1 allele headed on average in 43.5 days while plants homozygous for the mutant ft-B1 344 allele headed in 51.2 days (7.7 day delay). This difference was smaller than that detected between homozygous lines of wild type and ft-B1 263 mutation (20.4 day delay within the same homozygous wild type FT-A1 Figure 5. Phylogenetic analysis of FT-like genes in temperate grasses. Phylogenetic analysis was performed using the full-length proteins. A neighbor-joining tree was constructed using pairwise deletions and 1,000 bootstrap iterations with the program MEGA 5.0 [52]. The scale bar 0.05 represents 5% base substitution. Bootstrap numbers larger than 50 are shown in the respective nodes. To simplify the tree, only the wheat A-genome homoeologs of wheat were included. Accession information: background), suggesting that the ft-B1 344 mutation has a less deleterious effect on FT-B1 function than the ft-B1 263 mutation.
The predicted proteins encoded by the FT-A1 and FT-B1 genes in tetraploid wheat are identical, so we hypothesized that the larger effect of the ft-B1 263 mutant relative to the ft-A1 truncation mutant on heading time could be associated with relative differences in their respective expression levels. To test this hypothesis, we designed genome-specific primers for FT-A1 and FT-B1 and quantified their transcript levels using qRT-PCR (Table 1) in three-week-old wild type Kronos plants. Transcript levels of the FT-B1 homoeolog were three-fold higher than those of the FT-A1 homoeolog ( Figure 4B), which correlates with the stronger effect of the FT-B1 locus on heading time.

Expression of FT-like Genes in Transgenic FT1 Plants
Based on the observation that the transcript levels of FT2 were altered in transgenic wheat plants with higher transcript levels of FT1 [21] we decided to investigate the effect of FT1 down-and up-regulation on the five closest paralogs to FT1. To provide an evolutionary framework for the different wheat and Brachypodium FT-like genes included in the expression analyses, we first characterized their phylogenetic relationships ( Figure 5). There are multiple FT-like genes in both Brachypodium and wheat [12], but only the closest five are included in this study. The predicted proteins from Brachypodium, wheat, barley and rice were grouped into five clusters. Both the FT1 and FT2 clusters include homologs from all four species, which suggests an ancient duplication that predates the divergence between rice and the temperate grasses. The FT3 and FT5 clusters are more closely related to each other than to the FT4 cluster, which includes two closely related proteins, FT4 and a previously undescribed protein, which we have named FT6 (Table 1). Barley and Brachypodium homologs of this novel FT protein were also identified ( Figure 5).
Brachypodium FT1 RNAi T 0 plants '705', '706', '707', and '710' grown under LD showed a significant reduction in FT1 transcript levels (,1%, P,0.001) relative to the wild type control ( Figure 6A-B). The same transgenic plants also showed a significant downregulation of the FT-like genes FT2 (average 3% of wild type, P,0.001, Figure 6C) and FT4 (average 36% of wild type, P,0.001, Figure 6E), but their expression of FT3 or FT5 was not significantly different from the wild type ( Figures 6A,D,F). The expression of FT6 was nearly undetectable in both wild type and transgenic plants (data not shown). The FT1 RNAi transgenic plants also showed a strong downregulation of the FT1 downstream target VRN1 (,3% of wild type, P,0.001, Figure 6G), which may explain the non-flowering phenotype of these lines. As expected, no significant effects were detected in the expression of GIGANTEA (GI) (Figure 6A), an upstream regulator of FT in the photoperiodic pathway in Arabidopsis [37].
In the five selected wheat FT1 RNAi T 1 transgenic lines ('140D-19, '152E-19, '152E- 29, '1965-29, and '1965-59), FT1 transcript levels were significantly reduced relative to the levels observed in the wild type Bobwhite (,26%, P,0.001, Figure 7A-B). The downregulation of FT1 was associated with significant reductions in the transcript levels of FT2 (49%, P,0.05, Figures 7A,C) and FT5 (28%, P,0.001, Figures 7A,F). No significant differences between mutant and wild type lines were detected for FT-like genes FT3 and FT4 (Figures 7A,D-E) or for the GI upstream control ( Figure 7A). As in Brachypodium, the expression of the wheat FT6 gene was also nearly undetectable in both wild type and transgenic plants (data not shown). Transcript levels of VRN1 were slightly reduced (76% of wild type) in the transgenic plants, but these differences were not significant (Figures 7A,G). The smaller differences in VRN1 transcript levels and in flowering time between the wheat FT1 RNAi and wild type compared with those observed in Brachypodium, might be associated with the presence of dominant VRN1 alleles for spring growth habit in both the Bobwhite transgenic and control plants (Vrn-A1, Vrn-B1 and Vrn-D1; [38]).
To further test the effect of FT1 (or of the developmental changes induced by FT1) on the expression of other FT-like genes, we quantified the transcript levels of these genes in wheat transgenic plants carrying the highly-expressed Hope FT-B1 allele [7]. In the transgenic FT1 HOPE lines, the significant increase in FT1 transcript levels was associated with a significant upregulation of FT2 and FT3 both under LD ( Figure 8A) and SD ( Figure 8B) photoperiods. Expression levels of FT4, FT5 and FT6 were not significantly different between the wild type and FT1 HOPE lines. Taken together these results indicate that the developmental changes induced by FT1 are associated with changes in the expression of other FT-like genes.

FT1 is a Strong Promoter of Flowering Initiation in Both Brachypodium and Wheat
In the temperate grasses, the shoot apical meristem is located at the base of the plant covered by multiple layers of nested leaf primordia, sometimes below the soil and/or a layer of dead leaves, and is thus relatively isolated from some environmental signals. In contrast, leaves are broadly exposed and act as the main sensory organ for monitoring changes in the environment. FT1 and possibly other FT-like genes play a central role in the integration and transport of these signals [7,18]. The convergence of multiple signaling pathways in the regulation of these genes is likely related to the distinctive ability of their encoded proteins to move through the phloem transporting environmental signals from the leaves to the shoot apical meristems [17,20].
In this study we show that the ectopic expression of FT1 under constitutive promoters results in precocious flowering of differentiating wheat and Brachypodium calli, bypassing the generation of normal vegetative tissues. Similar results have been reported before for both Brachypodium and wheat. Two recent studies have shown that overexpression of FT1 (and also of FT2) in Brachypodium results in precocious flowering [25,26]. Similarly, transformation of the FT-D1 wheat homoeolog driven by the 35S promoter into the spring wheat variety 'Norin 619 [24], or of a highly expressed FT-B1 allele driven by its native promoter into the winter wheat Jagger (FT1 HOPE ) [7], accelerate wheat heading time. Taken together, these results demonstrate that FT1 is a strong activator of the flowering pathway in both wheat and Brachypodium.
FT1 has the ability to induce the transcription of VRN1 meristem identity gene even in the absence of vernalization. In varieties with a winter growth habit, chromatin in VRN1 regulatory regions is in a repressed state before vernalization and cold treatment promotes a more active state leading to the upregulation of VRN1 [39]. We show here that the overexpression of FT1 in the vernalization-responsive Brachypodium accession Bd21-3 [26,40] results in extremely early flowering, even in the absence of vernalization. Similarly, the introgression of natural FT1 alleles expressed at high levels in early developmental stages for both wheat [7] and barley [41] into varieties with a winter growth habit also results in early flowering in the absence of vernalization.

Downregulation of FT1 Showed Stronger Effects in Brachypodium than in Wheat
Results from the FT1 overexpression experiments presented here and from previous studies [7,24,25] indicate that the expression of FT1 is sufficient to induce flowering in both Brachypodium and wheat. However, these studies have not described the effect of the downregulation of FT1, which is important to determine if this gene is essential for flowering in the temperate cereals. In Brachypodium FT1 RNAi transgenic lines, the drastic downregulation of FT1 was associated with a non-flowering phenotype, even after the plants were vernalized for two weeks. This result indicates that FT1 is essential for flowering in this Brachypodium line under the conditions tested in this study.
Rice FT1 homologs also seem to be essential for flowering [42]. In this species, there are two duplicated FT1 orthologs, Hd3a and the Rice Flowering Locus T1 (RFT1) that share 90% identity at the protein level, suggestive of a relatively recent duplication in the rice lineage [42]. The simultaneous silencing of Hd3a and RFT1 by RNAi results in non-flowering rice plants, even when grown in an inductive SD photoperiod [42]. In contrast, non-functional mutations in all six FT-like genes in Arabidopsis delay, but do not abolish flowering [43]. These studies in rice and Arabidopsis indicate that the importance of FT-like genes in flowering regulation carries across species. Our Brachypodium and wheat results support this conclusion. While FT1 seems to be essential for flowering in Brachypodium, in wheat the downregulation by RNAi or disruptive mutation in the coding regions result only in delayed flowering.
In wheat, most FT1 RNAi transgenic plants (10 out of 13 T 0 lines) flowered only two to four weeks later than the wild type Bobwhite plants ( Table 2) and were able to set seeds. Among the three extremely late or non-flowering wheat FT1 RNAi transgenic lines only one set seed (line '1965'). A progeny test of this line showed that the derived T 1 progeny flowered 10 to 20 days later than the wild type, similar to the other transgenic lines ( Table 2). This result suggests that other factors (e.g. tissue culture) may have contributed to the later flowering time of the '1965' T 0 plant relative to its T 1 progeny. The other two late flowering wheat plants (lines '2547' and '2548') failed to produce seeds and, therefore, it was not possible to verify whether their unusually late flowering phenotype was linked to the FT1 RNAi transgene.
The stronger effect of the FT1 RNAi transgene on flowering time in Brachypodium relative to wheat was correlated with a stronger downregulation of FT1 in the Brachypodium transgenic lines (,1% of wild type) than in wheat (,26% of wild type). We currently do not know if this is caused by differences in the efficiency of the RNAi constructs, or by the different genetic backgrounds. The Bd21-3 line selected for Brachypodium transformation has a facultative winter growth habit [40] whereas the Bobwhite genotype used in wheat transformation has a spring growth habit. The Vrn-A1, Vrn-B1 and Vrn-D1 alleles present in Bobwhite are dominant for early flowering and epistatic to the winter alleles of other vernalization genes. These VRN1 alleles may contribute, directly or indirectly, to the relatively higher FT1 transcript levels observed in the wheat RNAi lines than in the Brachypodium RNAi lines.
The tetraploid line Kronos used to generate the ft1 TILLING mutants also carries a Vrn-A1 allele that is dominant for spring growth habit and epistatic to other vernalization genes [44]. This may contribute to the limited effect of the ft-A1/ft-B1 263 double mutant, which showed a flowering delay (25 days) similar to that observed in the FT1 RNAi transgenic hexaploid wheat plants. Unfortunately, we were unable to find a truncation mutation for the FT-B1 homoeolog and therefore, we cannot rule out the possibility that the selected ft-B1 263 mutant has some residual ability to induce flowering.  Triple ppd1-null mutants in the photoperiod sensitive hexaploid spring variety 'Paragon', which have almost undetectable FT1 transcript levels, flower approximately 30 days later than their wild type control [45]. This delay in flowering is similar to that observed in the ft-A1/ft-B1 263 double mutant and in the Bobwhite FT1 RNAi transgenic lines. Taken together, these results support the hypothesis that FT1 is not essential for flowering in spring wheat. Even if other genes can induce flowering in wheat, our results demonstrate that FT1 still plays a critical role in the timely induction of reproductive development in both Brachypodium and wheat.
Changes Induced by FT1 are Associated with Changes in Other FT-like Genes FT-like genes FT2, FT3, FT4, FT5 and FT6 are the closest paralogs of FT1 in the temperate grasses [12]. Among these, FT2 is the most similar to FT1 (.75% protein identity), and is part of a duplication that occurred before the divergence of the rice and wheat lineages ( Figure 5). The FT3 and FT5 clusters (80 to 84% protein identity) appear to have arisen from a more recent duplication. Wheat and barley FT3 and FT5 proteins form well supported clusters, but their relationship with the Brachypodium homologs is unclear. This suggests that this duplication occurred prior to the divergence of wheat and barley, but close to the time of divergence between the Brachypodium and the Triticeae lineages. A separate cluster includes two groups of wheat, Brachypodium and barley paralogs; the FT4 group that is related to rice OsFTL12 and the FT6 group that is related to rice OsFTL13 ( Figure 5). The FT4 and FT6 paralogs have a similar level of divergence (71 to 75% protein identity) to that between the FT1 and FT2 paralogs.
Downregulation of FT1 in the FT1 RNAi transgenic plants in Brachypodium and wheat results in a significant downregulation of FT2, while upregulation of FT1 in FT1 HOPE wheat plants results in a significant upregulation of FT2. These results suggest that, directly or indirectly, FT2 transcription is regulated by FT1. Overexpression of FT2 homologs in Brachypodium [25] and rice (OsFTL1) [46] results in extremely early flowering similar to the results observed for FT1. These results suggest that FT1 and FT2 paralogs have both retained the ability to induce flowering. Therefore, the changes in FT2 transcript levels observed in the FT1 RNAi and FT1 HOPE transgenic plants may also contribute to the observed differences in flowering time.
Significant changes in the expression of other FT-like genes were detected in the FT1 RNAi and FT1 HOPE transgenic plants, but these effects were less consistent across genotypes and environmental conditions. The association between reduced FT1 transcript levels in the FT1 RNAi lines and the downregulation of FT4 was significant only in Brachypodium, whereas the association with the downregulation of FT5 was significant only in wheat. While the effect of FT1 on FT3 expression was significant in both LD and SD grown FT1 HOPE transgenic plants, the transcript levels of wheat FT3 were 25-fold higher under SD than under LD (Figure 8). This has also been reported previously in barley, where FT3 is a candidate gene for a major QTL affecting barley flowering under SD (Ppd-H2) [12]. Overexpression of barley FT3 in rice also accelerates flowering [47]. However, this effect is significantly smaller than the acceleration of flowering time in rice transgenic lines transformed with the barley FT1 and FT2 genes and is modulated by photoperiod [47]. These results suggest that FT3 may play a role distinct from that of FT1 and FT2 in the regulation of flowering time.
No significant differences in FT4, FT5 or FT6 transcript levels were detected between Jagger and the Jagger FT1 HOPE transgenic line, but the transcript levels of these three genes were higher under LD than under SD in the wild type Jagger plants suggesting some role in the photoperiodic response ( Figure 8). The roles of FT4, FT5 and FT6 and their rice homologs remain poorly understood.
The observed reduction in the expression of FT-like genes in the FT1 RNAi transgenic plants is unlikely to be the result of a direct targeting of the RNAi construct, because the region of FT1 selected as target for RNAi did not share any stretches of more than 18 identical nucleotides with other FT-like genes. A continuous stretch of at least 21 identical nucleotides between the trigger and a target gene is required to produce efficient silencing [48]. Therefore, the transcriptional changes observed in other FT-like genes in the leaves of the transgenic plants are likely a result of the changes in FT1 or of the developmental changes induced by FT1. Additional studies will be required to validate the effect of FT1 on the expression of other FT-like genes.
In rice, co-silencing of the FT1 homologs Hd3a and RFT1 also resulted in the downregulation of the FT2 homolog OsFTL1, but showed no effect on the expression of the rice homologs of FT3 and FT5 (OsFTL10) or FT4 (OsFTL12) [42]. Therefore, the observed changes in FT4 and FT5 transcription could be species specific, or be affected by different developmental stages of the plants in the different experiments.
The fact that FT1 interacts with the bZIP transcription factors FDL2 and FDL6 (which can bind the VRN1 promoter), while FT2 interacts with FDL13 (which cannot bind the VRN1 promoter) suggests that FT1 and FT2 may have different gene targets [21]. This observation is consistent with the functional divergence observed for the FT1 and FT2 orthologs in rice (Hd3a and OsFTL1). Whereas overexpression of Hd3a in rice results in an early flowering phenotype [49], overexpression of OsFTL1 produces a more complex phenotype including elongation of internodes, loss of apical dominance, and formation of a terminal tissue at the apical meristem [46].
In rice, Hd3a forms a hexameric protein complex comprised of two 14-3-3 proteins, two FD-like proteins and two Hd3a proteins, designated as the florigen activation complex (FAC) [50]. This complex has been shown to bind the OsMADS15 promoter (a homologue of VRN1) and induce flowering. Different combinations of FT-like, FD-like and 14-3-3 proteins can theoretically form a large number of possible different FACs, each varying in their constituent components. This system may provide flexibility to translate signals carried by FT-like mobile protein into tissue and developmental stage specific responses by modulating the expression of FD-like or 14-3-3 proteins interacting proteins in those tissues or developmental stages [51]. Our understanding of these complex interactions is in its infancy, but the results presented here suggest that changes in FT1 are associated with changes in the expression of other FT-like genes, which may help to coordinate the complex responses required during the transition from the vegetative to the reproductive phase. It would be interesting to test if the alteration of the expression of other FT-like genes affects the expression of FT1, or if this is a unidirectional hierarchical regulatory system in which FT1 initiates a cascade of changes that affects the expression of the other members of the FT-like family.

TILLING Mutants of FT-A1 and FT-B1 Homoeologs Provide New Variability to Engineer Wheat Flowering Time
The ft-A1 mutant has a premature stop codon that eliminates more than half of the encoded protein whereas the ft-B1 263 has an amino acid substitution within the conserved PEBP domain. Therefore, we initially expected the ft-A1 mutant to exhibit a larger effect on flowering time than the ft-B1 263 mutant. However, we observed the opposite; a significantly larger delay in flowering was associated with the ft-B1 263 mutants than with the ft-A1 mutant. This result suggests that the FT-B1 homoeolog has a more important role in the regulation of flowering in tetraploid wheat than the FT-A1 homoeolog.
The FT-A1 and FT-B1 proteins are identical, so differences in their transport efficiency or in their ability to induce flowering are unlikely to contribute to the observed differences in heading time. By contrast, the higher transcript levels of FT-B1 relative to FT-A1, correlates well with the larger effect of FT-B1 relative to FT-A1 on wheat heading time, providing a simple explanation for the observed phenotypic effects. Similar differences have been described for hexaploid wheat, where FT-B1 exhibits much higher expression levels than either the A or D homoeologs [15]. Although the ft-A1 mutation showed no effect on flowering time, it was associated with a delay in heading time of 6.7 days in plants homozygous for the ft-B1 263 mutation. This suggests that FT-A1 is a hypomorphic allele, but that it still retains some ability to induce flowering under LD.
The slight differences in heading time between the two different ft-B1mutants and among the different combination of ft-A1 and ft-B1 alleles suggest that different combinations of FT1 homoeologs (both mutant and natural alleles) can be used to precisely regulate heading time in wheat. This will become more important as wheat breeders try to adjust heading time of current wheat varieties in environments affected by global climate change.