Feedback regulation of COOLAIR expression controls seed dormancy and flowering time

See allHide authors and affiliations

Science  01 Jun 2018:
Vol. 360, Issue 6392, pp. 1014-1017
DOI: 10.1126/science.aar7361

Controls on seed dormancy

Herbivores and an inopportune cold snap can destroy fragile plant seedlings. Plants control the dormancy of their seeds in anticipation of more favorable growth conditions. Chen and Penfield analyzed the molecular controls on seed dormancy in the model plant Arabidopsis thaliana. Two genes and an antisense RNA, known from the process of vernalization, integrate ambient temperature to control seed dormancy via their opposing configurations.

Science, this issue p. 1014


Plants integrate seasonal signals, including temperature and day length, to optimize the timing of developmental transitions. Seasonal sensing requires the activity of two proteins, FLOWERING LOCUS C (FLC) and FLOWERING LOCUS T (FT), that control certain developmental transitions in plants. During reproductive development, the mother plant uses FLC and FT to modulate progeny seed dormancy in response to temperature. We found that for regulation of seed dormancy, FLC and FT function in opposite configuration to how those same genes control time to flowering. For seed dormancy, FT regulates seed dormancy through FLC gene expression and regulates chromatin state by activating antisense FLC transcription. Thus, in Arabidopsis the same genes controlled in opposite format regulate flowering time and seed dormancy in response to the temperature changes that characterize seasons.

In Arabidopsis, seasonal changes in temperature are sensed by the epigenetic state of the floral repressor gene FLOWERING LOCUS C (FLC) (13). Prolonged periods of cold promote the expression of antisense transcripts at FLC known as COOLAIR, which facilitate silencing of FLC mRNA transcription via recruitment of Polycomb repressive complex 2 (PRC2) and deposition of the trimethylated histone H3 Lys27 (H3K27me3) chromatin mark (46). A key target of FLC is FLOWERING LOCUS T (FT), and FT and FLC are believed to act sequentially to deliver seasonal information that promotes flowering (79).

After flowering, FLC and FT control plant architecture and progeny seed dormancy, the latter by regulation of seed coat development and seed hormone levels (1015). Temperature changes applied to plants before buds are visible affect progeny seed dormancy in a process that requires maternal FLC and FT, which pass seasonal information to progeny (12). In this way, the mother plant exploits environmental temperature variation to create diversity in progeny seed behavior, which is important for bet-hedging reproductive strategies (15). Here, we investigated how FT controls seed dormancy.

To repress FT expression, FLC forms heterodimers with the transcription factor SHORT VEGETATIVE PHASE (SVP) (16, 17). To determine the genetic architecture of maternal temperature signaling pathways controlling seed dormancy, we exposed ft-1 svp-3 and ft-1 flc-21 double mutant rosettes to high and low ambient temperatures and analyzed the effect on progeny dormancy relative to the wild type and single mutants (Fig. 1A). Restricting temperature treatments to the pre-anthesis period enables dormancy outcomes to be ascribed solely to parental responses (12). Genetic interactions showed that the effect of SVP requires FT, because mutation of FT suppressed the high germination of svp-3 mutants (Fig. 1A). However, ft-1 flc-21 double mutant plants produced seeds whose germination resembled that of flc-21, and flc-21 reversed the stronger dormancy of ft-1 mutants (Fig. 1A). Therefore, in the context of seed dormancy, the FLC locus acts genetically downstream of FT. To test whether FT requires maternal FLC for dormancy control, we made reciprocal crosses between ft-1 and ft-1 flc-21 double mutants. Maternal FLC was required for seed dormancy control by FT (fig. S1A) (10). FT is required for temperature regulation of seed coat development (12). Indeed, flc-21 prevented increases in seed coat tannin content and decreases in seed coat permeability observed in ft mutant fruit tissue, and reversed the up-regulation of transcription factors controlling seed coat tannin deposition in ft-1 (Fig. 1, B and C, and fig. S1, B and C).

Fig. 1 FT acts upstream of FLC to regulate seed dormancy.

(A) Germination of seeds from mother plants subjected to either 22° or 16°C before anthesis. Error bars represent SE (n ≥ 5 biological replicates). (B) Insoluble tannin content in wild-type (Ler), ft-1, flc-21, and ft-1 flc-21 double mutant seeds. Error bars represent SE (n = 5 biological replicates). (C) Quantitative reverse transcription PCR (qRT-PCR) shows that FLC is required for up-regulation of regulators of tannin synthesis TRANSPARENT TESTA 12 (TT12) and TT16 in ft-1. Error bars represent SE (n = 3 biological replicates). *P < 0.05 (difference from wild type).

Next, we tested whether FT controls FLC expression, focusing our analysis on female reproductive tissues. Relative to 16°C conditions, warm conditions (22°C) down-regulated FLC mRNA expression in wild-type carpels, as it does in leaves (18). Loss of FT strongly reduced the magnitude of this response (Fig. 2A). To locate the tissues in which FT regulates FLC, we crossed a translational GUS fusion containing the entire FLC locus (3) (FLC-GUS) into ft-1 and analyzed the expression in reproductive tissues. In flowers at 22°C, FLC is expressed only in the stamen of wild-type plants, but in the ft-1 mutant we observed ectopic expression of FLC-GUS in carpels (Fig. 2, B and C). We also tested the effect of ft-1 on antisense COOLAIR expression in carpels (Fig. 2D). COOLAIR is transcribed into both short (proximal) and longer (distal) transcripts (19). We found that at 22°C, levels of distal COOLAIR were reduced in ft-1 but levels of proximal COOLAIR transcript were unchanged (Fig. 2D). Therefore, FT influences the balance of sense and antisense gene expression at the FLC locus. Further analysis of FLC:GUS in seed tissues during seed maturation showed no effect of FT on FLC:GUS expression in seeds themselves (fig. S2), consistent with the maternal action of FLC in seed dormancy control; this finding suggests that FT and FLC act to control dormancy either during carpel development or shortly after fertilization.

Fig. 2 Maternal temperature experience and FT affect FLC expression in carpels.

(A) FLC mRNA transcripts in carpel tissue harvested from plants grown at either 16° or 22°C. Error bars represent SE (n = 3 biological replicates). (B) FLC-GUS expression in the stage 12 floral organs of representative Ler and ft-1 flowers. Scale bar, 1 mm. (C) Quantitative measurement of FLC-GUS activity in wild-type and ft-1 isolated carpels at 22°C. Error bars represent SE (n = 3 biological replicates). (D) COOLAIR transcripts in wild-type and ft-1 carpels at 22°C. Error bars represent SE (n = 3 biological replicates). *P < 0.05 (difference from wild type).

Because of the differences in FLC expression between wild-type and ft-1 mutants, we used FT tagged with green fluorescent protein (GFP) to test whether FLC is a direct target of FT. Plants expressing FT-GFP from the phloem-specific SUC2 promoter (9) were used to determine whether FT-GFP could be detected at the FLC locus by chromatin immunoprecipitation (ChIP). This construct is functional and complements the late-flowering ft-7 mutation, and also increases germination of ft-7 seeds (fig. S3). We could detect enrichment of FT-GFP toward the last exon and 3′ region of the FLC locus, corresponding to the promoter and first exon of COOLAIR transcripts (Fig. 3A) (4, 5). Given that COOLAIR transcripts silence FLC mRNA expression (5), we asked whether FT affects the temperature-regulation of FLC mRNA levels by promoting COOLAIR transcription. We used strand-specific quantitative reverse transcription polymerase chain reaction (qRT-PCR) to compare COOLAIR expression in wild-type and ft-1 siliques produced on plants that began reproductive development at either 22° or 16°C. We found no evidence that either temperature or FT regulated proximal COOLAIR transcripts (Fig. 3B). However, 22°C increased distal COOLAIR levels in wild-type fruits compared to 16°C, and this effect required FT (Fig. 3C). This down-regulation of distal COOLAIR in ft-1 suggested that this transcript opposed FLC mRNA expression (20), which is higher in the ft-1 mutant than in the wild type (Fig. 2).

Fig. 3 Maternal temperature experience affects FLC antisense transcription in fruit tissue.

(A) FT-GFP ChIP performed across the FLC locus in silique tissues. Data are represented relative to the negative control ACTIN2. Numbers indicate positions of primers used. Error bars represent SE (n = 3 biological replicates). *P < 0.05 (difference from wild type). (B and C) Proximal and distal COOLAIR levels in silique tissues from plants growing at either 16° or 22°C. Error bars represent SE (n = 3 biological replicates). Primers used are indicated by arrows in (A). *P < 0.05 (difference from wild type). (D) COOLAIR:LUC expression imaged in wild-type and ft-1 seedlings, carpels, and developing siliques from 3 to 14 days after pollination (DAP; left to right). Scale bars, 5 mm (seedlings and siliques), 1 mm (carpels). (E) Quantitative analysis of COOLAIR:LUC expression at 22°C in the indicated organs in long days, and the effect of 16-hour days (LD) and 8-hour days (SD) on COOLAIR:LUC expression in Ler seedlings. Data are means ± SE (n ≥ 5 biological replicates). P values were calculated by two-tailed Student t test between genotypes or treatments.

We next tested whether the FT protein can activate transcription from an isolated COOLAIR promoter fused to the LUCIFERASE (LUC) reporter. After introgressing the COOLAIR:LUC transgene from Ler to ft-1, we compared LUC activity between wild-type plants and ft-1 mutants. In the wild type, we could detect COOLAIR:LUC activity in whole plants, carpels, and fruits, with LUC activity in fruits declining during fruit and seed maturation (Fig. 3D). In ft-1 plants, COOLAIR:LUC expression was reduced to 20 to 30% of wild-type levels in all organs tested (Fig. 3, D and E). Moving wild-type COOLAIR:LUC plants from long days to short days reduced LUC expression (Fig. 3E), consistent with the known role of FT in photoperiod responses (9). Because FT bound the FLC locus close to the COOLAIR transcription start site, and because FT was required for COOLAIR promoter activity in plants grown at 22°C in long days, we conclude that FT suppresses FLC mRNA expression by activating COOLAIR transcription.

During vernalization, changes in COOLAIR expression precede stable chromatin state changes at FLC (4, 5). To test whether FT regulation of FLC is chromatin-based, we analyzed histone methylation at the FLC locus in wild-type and ft-1 plants at 22°C. Wild-type plants showed more H3K27me3 at the nucleation region (6) in siliques and leaves and less H3K36me3 than did ft-1 plants (Fig. 4A). This shows that FT can switch FLC between active and passive transcription states. The effect of ft-1 on FLC expression in leaves could be overcome by vernalization (Fig. 4, B and C). We also found that an FLC-dependent effect of FT on flowering time can be observed if plants are grown at 16°C (Fig. 4D). The ft-1 mutant is later-flowering at 16°C than at 22°C, and this effect can also be overcome by vernalization. This is reminiscent of fca and fy mutants, which are also affected in COOLAIR processing (21). Therefore, the feedback regulation of FLC by FT has a general relevance, controlling both flowering time and seed dormancy processes.

Fig. 4 FT affects FLC chromatin state in leaves and siliques.

(A) ChIP comparing Ler and levels of H3K27me3 and H3K36me3 across the FLC locus in silique and ft-1 leaf tissues. Data are means ± SE (n ≥ 2 biological replicates). (B) The effect of vernalization (V) at 4°C for the indicated number of weeks (W) on distal COOLAIR transcript levels in Ler and ft-1. Data are means ± SE (n = 3 biological replicates). *P < 0.05 (difference from Ler control levels without vernalization). (C) GUS activity from FLC:GUS-expressing Ler and ft-1 plants in seedlings without vernalization (–V) versus 7 days after vernalization for 4 weeks (4W V+7). Data are means ± SE (n = 3 biological replicates). *P < 0.05 (difference from Ler control levels without vernalization). (D) Rosette and cauline leaf numbers at flowering in Ler and ft-1 plants grown at either 22° or 16°C with or without prior vernalization. Data are means ± SE (n ≥ 5 biological replicates). Significant differences in response to vernalization are indicated.

Supplementary Materials

Materials and Methods

Figs. S1 to S3

Table S1

References (2230)

References and Notes

Acknowledgments: We thank C. Dean for FLC:GUS seeds and COOLAIR:LUC construct, G. Coupland for ft-7 pSUC2:FT-GFP, and G. Calder and E. Wegel for microscopy support. Funding: Supported by BBSRC grant BB/P013511/1 to the John Innes Centre. Author contributions: Experimental design by M.C. and S.P. Experimental work was conducted by M.C. Data analysis by M.C. and S.P. S.P. wrote the manuscript. Competing interests: None declared. Data and materials availability: All data needed to evaluate the conclusions in the paper are present in the paper or the supplementary materials.
View Abstract

Stay Connected to Science

Navigate This Article