A Pair of Related Genes with Antagonistic Roles in Mediating Flowering Signals

See allHide authors and affiliations

Science  03 Dec 1999:
Vol. 286, Issue 5446, pp. 1960-1962
DOI: 10.1126/science.286.5446.1960


Flowering in Arabidopsis is promoted via several interacting pathways. A photoperiod-dependent pathway relays signals from photoreceptors to a transcription factor gene, CONSTANS(CO), which activates downstream meristem identity genes such as LEAFY (LFY).FT, together with LFY, promotes flowering and is positively regulated by CO. Loss ofFT causes delay in flowering, whereas overexpression of FT results in precocious flowering independent ofCO or photoperiod. FT acts in part downstream of CO and mediates signals for flowering in an antagonistic manner with its homologous gene, TERMINAL FLOWER1 (TFL1).

In higher plants, flowering—the transition from vegetative to reproductive growth phase—is controlled via several interacting pathways influenced by both endogenous factors and environmental conditions. In Arabidopsis, a photoperiod-dependent pathway promotes flowering in response to an inductive long-day (LD) photoperiod, whereas an autonomous pathway functions independently of the photoperiod and other environmental conditions (1). Recent studies suggest that theFT gene may be regulated via both photoperiod-dependent and autonomous pathways and may act redundantly with LFY in promoting flowering (2).

We identified the FT gene by transferred DNA (T-DNA) tagging (3). The predicted FT open reading frame encodes a protein with similarity to the TFL1gene product (Fig. 1, A and B) (4). FT and TFL1 are members of a gene family in Arabidopsis (Fig. 1, B and C) (5). Putative FT orthologs in other species were found in databases (Fig. 1, B and C) (6).FT and TFL1 represent two clades that may have branched before the diversification of angiosperms (Fig. 1C).FT was expressed in all tissues in seedlings and mature plants (Fig. 2A). The FT mRNA level gradually increased with time under both LD and short-day (SD) photoperiods (Fig. 2, B and C). Under LD conditions, expression was first detected on day 4 and plateaued around day 6, preceding floral commitment around days 9 and 10 (7). Up-regulation of FT expression was delayed and reduced under SD conditions (Fig. 2C).

Figure 1

Structure of the FT gene and sequence comparison. (A) Schematic diagram of theFT locus. Boxes and lines are exons and introns of indicated length (open reading frames are hatched box segments). Mutations in six alleles are shown below. (B) Comparison of amino acid sequences (24) for FT (3), TSF (5), CiFT (6), OsFT (6), TFL1 (4), and human PEBP (hPEBP) (23). Below the sequences is a consensus. Substitutions or terminations (asterisks) in the six alleles are shown. Vertical arrows, introns conserved betweenFT and TFL1; vertical arrowhead, cleavage site in hPEBP to generate the HCNP (underlined) (23). (C) A phylogenetic tree, constructed by the neighbor-joining method, for 13 representative proteins from plants, hPEBP, andSaccharomyces TFS1 (GenBank accession number X62105), RCN1 and RCN2 from rice (25), SP (GenBank accession numberU84140), CEN (GenBank accession number S81193), and BnTFL1-1 (GenBank accession number AB017525). Bootstrap values are shown on each branch.

Figure 2

Analysis of FT expression. (A to E) Reverse transcription polymerase chain reaction analysis (14). Duplicate lanes for each sample represent duplicate reactions. A fragment of a β-tubulin gene (TUB2) or APETALA2 (AP2) was amplified as a control. Numbers in (B) and (C) indicate days of incubation in a growth chamber (germination on day 1). (A) Expression in LD-grown seedlings on day 6 (left) and mature plants (right). WS, whole seedling; SA, shoot apex; Hy, hypocotyls; Co, cotyledons; Ro, roots; FB, floral buds; Fl, flowers; iS, immature siliques; mS, mature siliques; St, stems; RL, rosette leaves, Br, bracts. (B) Temporal profile of expression of FT, TSF, andTFL1 genes in aerial parts of LD-grown plants. ΔFT indicates FT deletion line (3) on day 14. (C) Effect of photoperiods and co-1 onFT expression (LD, 16 hours light/8 hours dark; SD, 8 hours light/16 hours dark). (D) Effect of late-flowering mutations on the level of FT expression on day 7, when the level in the wild type plateaued [see (B)]; co-1 is in the Col background, and fha-1, fca-1, and fwa-2 are in the Ler background. (E) Up-regulation of FTexpression by activation of the CO-GR fusion protein (26). CO-GR fusion protein was activated by application of dex. Numbers indicate hours after application of dex (+Dex) or the solvent (–Dex); rRNA stained with ethidium bromide (EtBr) is shown as a loading control. (F) FTexpression in transgenic plants overexpressing various genes. RNA blot analysis of 10 μg of total RNA. FT/TFL1 andTFL1/FT are chimeric genes coding for FT(1–63)/TFL1(67–177) and TFL1(1–66)/FT(64–175) fusion proteins, respectively. 35S::CO:GR (+) indicates35S::CO:GR; co-2 plants 4 days after application of dex (26).

CO is up-regulated under LD conditions, and it activates meristem identity genes (8, 9). We investigated whether CO regulates FT. In theco-1 mutant, up-regulation of FT expression was delayed (Fig. 2C), suggesting that it may require CO during the early vegetative phase. In contrast, the FT mRNA level was not affected in fha-1, which lacks cryptochrome 2 (10); fca-1, which is defective in the autonomous pathway (11); or fwa-2, which is similar toft in terms of genetic interactions with LFY andAP1 (2) (Fig. 2D). When CO activity was induced from the CO–glucocorticoid receptor (CO:GR) fusion protein by treating35S::CO:GR; co-2 plants (9) with dexamethasone (dex), up-regulation of FT expression was observed within 12 hours (Fig. 2E). The FT mRNA level was elevated in dex-treated 35S::CO:GR; co-2 plants, but not in plants with similar precocious flowering phenotype due to overexpression of LFY (12), AP1(13), or TSF (see below) (Fig. 2F). Therefore, up-regulation of FT expression upon induction ofCO activity was not a secondary effect of induced acceleration of flowering. Because up-regulation of FTexpression eventually occurs in the co mutant and in the SD photoperiod, there is probably a photoperiod-independent pathway forFT up-regulation. Although the physiological relevance is not clear, early transient expression of FT was observed in the co mutant and in the SD photoperiod (Fig. 2C).

Constitutive overexpression of CO causes photoperiod-independent precocious flowering (9). IfFT promotes flowering under the control of CO, then constitutive overexpression of FT should result in flowering, independent of the photoperiod and CO function. All transgenic lines that overexpress FT(35S::FT) flowered early with determinate inflorescence similar to tfl1, as did35S::CO plants (9) (Fig. 3, A and B). Overexpression ofTSF or CiFT results in the same phenotype (14). Early flowering was correlated withFT mRNA accumulation: Plants with the highest level flowered with only two rosette leaves (14). Neither the SD photoperiod nor co-1 affected the early-flowering phenotype of transgenic plants (Table 1), which indicates that FT regulates flowering by acting in part downstream of CO. Consistent with this observation,ft-1 partially suppressed the precocious flowering phenotype of 35S::CO (15). In contrast, the semidominant fwa-2 mutation, which did not affectFT expression (Fig. 1D), partially suppressed the precocious flowering phenotype of 35S::FT (Table 1). Thus, FWA may interfere with pathways downstream ofFT (14).

Figure 3

Phenotype of transgenic plants. (A) A 35S::FT plant. (B) A “terminal flower” on the primary inflorescence of a 35S::FT plant. (C) A35S::FT; tfl1-17 plant with single terminal flowers replacing the primary and secondary inflorescences;tfl1-17 is an RNA-null allele as the result of a 1-kb deletion in TFL1 (20). (D andE) 35S::FT/–; 35S::LFY/–plants. In (A) through (E), the arrowhead indicates a rosette leaf; arrow, bract; TF, terminal flower; AF, axillary flower. (Fand G) Shoot apex of a 3-day-old seedling of35S::FT/– (F) and 35S::FT/–; 35S::LFY/– (G). The arrow indicates one of the first two leaves or bracts left intact; the arrowhead indicates the shoot apex in between. One cotyledon (large asterisk) and one leaf or bract (small asterisk) were removed. In 35S::FT/–; 35S::LFY/–, the shoot apical meristem itself was transformed into a single terminal flower (se, sepal; p, petal; s, stamen; g, gynoecium), whereas it had two bract primordia and the shoot apical meristem between them in 35S::FT/–. Scale bars, 5 mm (A, C, D, and E), 2.5 mm (B), 100 μm (F and G).

Table 1

Flowering times of transgenic and mutant plants.

View this table:

The SD photoperiod and co-1 actually enhanced the precocious flowering phenotype in 35S::FT plants (Table 1). The LD photoperiod or CO may also enhance the expression of genes antagonistic to FT. One candidate for such a gene is TFL1, which is up-regulated by the LD photoperiod and CO (4, 9).FT and TFL1 play opposite roles in flowering, as loss of function and ectopic overexpression of these genes results in nearly opposite phenotypes [this study and (16–19)]. Haploinsufficiency of FT in the35S::TFL1 background (Table 1) and ofTFL1 in the wild-type background (17) suggests the importance of the balance between the two. This is further supported by the enhancement of the phenotype of35S::FT by tfl1 and of the phenotype of35S::TFL1 by ft (Table 1 andFig. 3C).

Because a loss-of-function mutation in one gene had phenotypic effect even in an excess of the other's activity, their mutual antagonism may not simply be due to each blocking the other's activity.35S::FT; 35S::TFL1 plants and ft; tfl1 plants showed a phenotype similar to that of35S::FT and ft plants, respectively (2) (Table 1), which suggests that the level ofFT activity is likely more important in the timing of flowering.

Hyperactivity of the FT-mediated pathway alone was not sufficient to induce flowering immediately upon germination, even in the absence of the antagonistic activity of TFL1 (Fig. 3C). Because FT may function in parallel with LFY(2), we investigated whether the combined activity of these genes is a limiting factor for induction of flowering.35S::FT/–; 35S::LFY/– plants germinated normally and had cotyledons and hypocotyls indistinguishable from those of the wild type (Fig. 3, D and E). However, a terminal flower with one or two bracts developed within 3 days after germination, replacing the complex shoot system observed in the wild type (Fig. 3, D to G, and Table 1). Thus, a simultaneous excess ofFT and LFY activity from the very beginning of post-embryonic development induces flowering with almost no intervening vegetative phase. In contrast, neither 35S::FT; 35S::AP1 nor 35S::LFY; 35S::AP1 plants showed such an extreme phenotype (20, 21).

Our results and those of others (2, 9, 19) suggest that FT and TFL1 mediate signals for floral transition in part downstream of CO in an antagonistic manner (14). What, then, are the biochemical functions of FT and TFL1? They belong to a family of possibly membrane-associated proteins that includes phosphatidylethanolamine-binding protein (PEBP) (22). Interestingly, PEBP in humans and rats is a precursor of hippocampal cholinergic neurostimulating peptide (HCNP) (23) (Fig. 1B). Whether the function of FT and TFL1 involves the generation of peptide molecules as the transmissible signal is an interesting question.

  • * Present address: Research Institute for Biological Sciences, Kayo-cho, Jobo-gun, Okayama 716-1241, Japan.

  • To whom correspondence should be addressed. E-mail: taraqui{at}


Stay Connected to Science

Navigate This Article