An Antagonistic Pair of FT Homologs Mediates the Control of Flowering Time in Sugar Beet

See allHide authors and affiliations

Science  03 Dec 2010:
Vol. 330, Issue 6009, pp. 1397-1400
DOI: 10.1126/science.1197004

Just Beet It

Flowering time regulation is important for plants to maximize their reproductive output. By investigating copies of genes that are strong and central activators of flowering in many different species (homologs of the FT gene in Arabidopsis), Pin et al. (p. 1397) found that during evolution, the regulation of flowering time in sugar beet (Beta vulgaris) has come under the control of two FT-like genes. Functional differences in these genes owing to small mutations in a critical domain have caused a duplicated copy of the flowering promoter FT to turn into a flowering repressor in sugar beet. These changes may explain why cultivated beets are unable to flower until their second year after passing through the winter, a behavior important for increasing crop yield.


Cultivated beets (Beta vulgaris ssp. vulgaris) are unable to form reproductive shoots during the first year of their life cycle. Flowering only occurs if plants get vernalized, that is, pass through the winter, and are subsequently exposed to an increasing day length (photoperiod) in spring. Here, we show that the regulation of flowering time in beets is controlled by the interplay of two paralogs of the FLOWERING LOCUS T (FT) gene in Arabidopsis that have evolved antagonistic functions. BvFT2 is functionally conserved with FT and essential for flowering. In contrast, BvFT1 represses flowering and its down-regulation is crucial for the vernalization response in beets. These data suggest that the beet has evolved a different strategy relative to Arabidopsis and cereals to regulate vernalization.

In flowering plants, the timing of the transition from the vegetative to the reproductive stage is determined by interactions between the developmental state of the plant and exogenous stimuli, such as changes in day length and temperature. Vernalization is a checkpoint through which many plants must pass, where flowering is conditional on prolonged exposure to cold temperature and is subsequently induced by other inductive pathways (1). Vernalization of the winter-annual Arabidopsis and other members of the Brassicaceae family involve several steps that culminate in the repression of the major floral inhibitor gene FLOWERING LOCUS C (FLC) (25). In cereals, cold temperature stimuli are integrated by the activation of the FRUITFULL (FUL)/APETALA1 (AP1) homolog VRN1 (6), which represses the CONSTANS, CONSTANS-like, and TIMING OF CAB EXPRESSION 1 (CCT)-domain transcription factor VRN2 (7, 8).

The phosphatidylethanolamine-binding protein (PEBP) gene family, found in both angiosperms and gymnosperms (9), has evolved both activators and repressors of flowering. The Arabidopsis PEBP gene FLOWERING LOCUS T (FT) (10, 11) is a major output of the photoperiod pathway and controls the floral transition in response to changes in day length, whereas its close homolog TERMINAL FLOWER 1 (TFL1) acts as a strong floral repressor (1012).

Cultivated sugar beet (Beta vulgaris ssp. vulgaris) is a biennial crop that requires vernalization to enter the generative phase. The vernalization period typically needs to be followed by exposure to long days (LDs). Without LDs, floral induction does not occur in beets, despite the fact that they become competent to flower after vernalization (13). One of the main factors controlling flowering time in beets is the bolting gene B, a dominant promoter of flowering that overrides the need for vernalization (14). Plants carrying the dominant B allele behave as annuals and proceed to bolting and flowering as a direct response to LDs.

To determine whether FT function is conserved between beets and Arabidopsis, we initiated a first characterization of the PEBP genes in Beta (15), isolating two genes grouped within the FT-like clade [BvFT1 and BvFT2 (fig. S1A)]. Genetic mapping showed that none of the PEBP homologs maps in vicinity to B (fig. S2) on chromosome II (16). Complementation experiments and gene expression analysis showed that BvFT2 is the functional FT ortholog in beets. Transgenic expression of BvFT2 in both Arabidopsis and sugar beet (fig. S3, B and G, and fig. S4) strongly promoted flowering. Furthermore, BvFT2 expression levels were correlated with the initiation of flowering in both annual and biennial beets (Fig. 1B). When BvFT2 expression was suppressed by RNA interference (RNAi) in annual plants, a continuous vegetative growth was observed in LDs (Fig. 2A and fig. S5A), suggesting that BvFT2 is essential for flowering. Unexpectedly, BvFT1 repressed flowering when ectopically expressed in transgenic sugar beet and Arabidopsis plants (Fig. 2, B and C, fig. S5, B and C, and fig. S3, C and G). These data suggest that an FT-like gene, evolutionarily more related to FT than to TFL1/CEN (figs. S1A and S6), has evolved into a true flowering repressor. BvFT1 expression was not affected in the BvFT2 RNAi plants (Fig. 2D), whereas BvFT2 expression was dramatically reduced compared with wild type in BvFT1-ox plants, both before (Fig. 2E) and after (Fig. 2F) vernalization.

Fig. 1

Expression pattern of BvFT1 and BvFT2 in beets (15). (A) Expression of BvFT1 and BvFT2 in various tissues, including se., seed; hy., hypocotyl; co., cotyledon; le., leaf; ap., apex; ro., root; st., stem; bu., floral bud, and fl., flower. Samples were harvested in LDs at zeitgeber time (ZT) 8 (i.e., 8 hours after lights on). Error bars, mean ± SE (n = 3). (B) Expression of BvFT1 and BvFT2 in leaf samples in LDs across different developmental stages in annuals and biennials. Samples were harvested at ZT6. Error bars, mean ± SE (n = 3). (C and D) Diurnal rhythms of BvFT1 and BvFT2 in annual, biennial, and vernalized-biennial beets in SDs (C) and LDs (D). Error bars, mean ± SE (n = 5). (E) Leaf number at time of bolting in biennial, vernalized-biennial, and annual plants grown under the different photoperiodic conditions described in (C) and (D). Error bars, mean ± SD (n = 6). (F) BvFT1 transcript accumulation in biennials in response to vernalization. Leaf samples were harvested at ZT6. Error bars, mean ± SE (n = 3).

Fig. 2

Misexpression of BvFT1 or BvFT2 control bolting in sugar beet. (A to C) Nonbolting phenotype observed in BvFT2 RNAi annual (A), BvFT1-ox annual (B), and vernalized BvFT1-ox biennial (C) plants when grown in LDs for 6 weeks. (D to F) BvFT1 and BvFT2 expression in BvFT2 RNAi (D), BvFT1-ox (E), and vernalized BvFT-ox (F) plants (15). Expression levels were set to unity in wild type (WT). Error bar, mean ± SE (n = 3).

In sunflower, neofunctionalization (the evolution of new function and/or expression of a duplicated gene) (17) of FT function was postulated to have occurred as a result of a frameshift mutation in an FT-like gene, ensuing in the HaFT1 gene acquiring a role in delaying flowering through a dominant-negative interference with an activating paralog, HaFT4 (18) (fig. S6). This does not appear to be the case for BvFT1, which acts as a true repressor on the basis of the Arabidopsis overexpression phenotypes (fig. S3, J and K), the differences between the BvFT2 RNAi and BvFT1-ox phenotypes in beets (Fig. 2 and fig. S5), and the fact that expression of BvFT1 and BvFT2 is mutually exclusive rather than overlapping (Fig. 1, A, C, and D).

As for all previously characterized FT-like genes, transcripts of both BvFT1 and BvFT2 were mainly detected in leaves (Fig. 1A), but their respective temporal expression patterns were very different. Whereas BvFT1 was expressed in the leaves (Fig. 1A) of plants that were not competent to flower, that is, plants grown in short days (SDs) or nonvernalized biennials (Fig. 1, C and D), BvFT2 expression always correlated with flower induction and reproductively mature plants (Fig. 1, A to D). These observations suggest that neofunctionalization of the pair of Beta FT paralogs occurred in terms of both expression profiles and protein function. Expression of BvFT1 and BvFT2 displays diurnal oscillations (Fig. 1, C and D), suggesting that both genes might be under the control of the circadian clock. However, BvFT1 is preferentially expressed in the morning in SDs (Fig. 1C), whereas BvFT2 is expressed in the evening and in LDs (Fig. 1D). In annuals, expression of both genes responded quickly to changes in photoperiod, but in opposing ways (fig. S7). Transcription of BvFT1 was not repressed in LDs in biennials before vernalization, and BvFT2 expression was not induced (Fig. 1D). This suggests that biennials might be insensitive to the photoperiodic signal before vernalization and that a vernalization-dependent factor acts to restore the photoperiodic regulation of BvFT1 and BvFT2. Consequently, these observations suggest that BvFT1 and BvFT2 regulation is dependent on the presence of the dominant allele B.

Vernalization and photoperiod pathways are connected by the coordinated repression of BvFT1. BvFT1 transcription in biennials was both reduced during vernalization (Fig. 1F) and maintained after winter at low levels if plants were grown in LDs. These data indicate that cold treatment induces BvFT1 repression and also affects the vernalization-dependent factors that maintain BvFT1 repression in LDs. Given that a transient exposure to SDs directly after vernalization [so called “devernalization,” a condition known to abolish flowering in Beta (19)] resulted in the restoration of BvFT1 expression (fig. S8), unfavorable conditions for flowering appear to be coupled to expression of BvFT1, which suggests that BvFT1 is important for the devernalization phenomenon.

On the basis of these results, we propose a model whereby BvFT1 prevents flowering during SDs and before vernalization by repressing the expression of BvFT2 (Fig. 3). Biennials, lacking B, would be unable to sense the long-day signal before vernalization because of high levels of BvFT1 expression. During vernalization, BvFT1 expression decreases and the plant becomes competent to respond to LDs, which is required to maintain low levels of BvFT1 expression. This appears to be necessary to induce flowering. However, formally, we cannot exclude that other factors might also contribute to the vernalization response and the regulation of BvFT2 (Fig. 3, dotted arrow).

Fig. 3

Simplified flowering models in beets, Arabidopsis, and barley, showing the key integrators of the vernalization pathway, which mediate transcriptional regulation of the FT orthologs.

Unexpectedly, BvFT2 RNAi plants responded to vernalization and bolted after cold treatment (fig. S9A), indicating that additional mechanisms are acting to promote bolting in beets in parallel to BvFT2. Although BvFT2 RNAi plants proceeded to bolting, flowering was completely abolished or initiated at very late stages (fig. S9, B and C), suggesting that BvFT2 is needed for a normal flower initiation. In contrast, because BvFT1-ox plants do not respond to vernalization followed by LDs (Fig. 2C and fig. S5C), BvFT1 itself is likely to regulate downstream bolting and flowering promoters. Such a scenario suggests that in beets an FT-like gene represses bolting and flowering as part of the vernalization response (Fig. 3). In Arabidopsis, FLC inhibits flowering, whereas in cereals, VRN2 (a member of the CCT gene family) performs this role (Fig. 3). This suggests that the vernalization response has developed differently not only in monocots and eudicots but also in the two eudicot families Brassicaceae and Amaranthaceae (formerly Chenopodiaceae) within the evolutionarily diverged eudicot groups of the Rosids and Caryophyllales, respectively. Sequence comparison between the two proteins showed that both BvFT1 and BvFT2 carry the functionally important FT signatures, Tyr85(Y) and Gln140(Q) (fig. S1C) (20). However, the proteins differ within the segment B of the fourth exon [encoding an external loop of PEBP (21)] (Fig. 4, A and B, and fig. S1C), which is important for FT versus TFL1 function in Arabidopsis (21). BvFT2 is identical in this region to most other FTs (consistent with its role in promoting flowering), whereas BvFT1 has three substitutions out of 14 amino acids (Fig. 4A and fig. S1C). By swapping regions between BvFT1 and BvFT2 (Fig. 4C and table S1), we demonstrated that divergence within these three amino acids of the external loop is the major cause of their antagonistic functions. Remarkably, expression of BvFT1B2 (chimeric BvFT1 carrying the external loop of BvFT2 with the three altered amino acids) can completely revert its repressing function to promoting function (Fig. 4C and table S1). This suggests that BvFT1 was initially a promoter of flowering but that mutations within the external loop of its protein resulted in a shift in function to flowering repression. We are currently investigating which of the three amino acids are critical for the activator versus repressor function.

Fig. 4

Characterization of important regions of BvFT1 and BvFT2 involved in their antagonistic functions. (A and B) Folding prediction of BvFT1 and BvFT2. The structures are shown as solid three-dimensional traces with an alignment diversity color setting. (C) Chimeric FT proteins containing domains of BvFT1 and BvFT2 show antagonistic function when expressed in Arabidopsis.

To get further insights into the evolution of the BvFT1 repressor function, we isolated BvFT1- and BvFT2-like sequences from other Beta species, as well as other members of the Amaranthaceae family. BvFT2-like genes showed high homology across the Amaranthaceae (fig. S10, A and B). In contrast, we could only isolate BvFT1-like sequences from Beta species, suggesting that FT1 arose through gene duplication during the diversification of the genus Beta. We cannot exclude that FT1 may be present in other Amaranthaceae, but given the high level of conservation of FT-like genes, our results suggest that FT1 is likely to be restricted to Beta. Gene expression analysis showed that in LDs FT1 was expressed in all cultivated varieties of B. vulgaris, but not in wild Beta species, except in some B. vulgaris ssp. maritima accessions from northern latitudes (fig. S10C). In contrast, FT2 was expressed in these species, consistent with their flowering behavior. Indeed, B. macrocarpa, B. precumbens, and B. vulgaris ssp. maritima accessions from southern latitudes bolt and flower quickly without exposure to cold temperature, while maritima accessions from northern latitudes and all cultivated beets require vernalization to enter the generative phase (22). These results are consistent with our findings in sugar beet, clearly linking the BvFT1 expression in LDs to the biennial growth behavior. Interestingly, domesticated beets (ssp. vulgaris) most likely originated from B. vulgaris ssp. maritima plants (23).

Because BvFT1 overexpression leads to the prevention of bolting and flowering, even after a prolonged vernalization period (Fig. 2C and fig. S5C), it may have interesting applications in breeding. Our BvFT1-ox plants may fulfill the desire for a “winter beet” that can be planted in the fall without bolting during the following spring to provide a prolonged growing season and harvesting campaign with increased yields. This demonstrates how extending our understanding of the molecular determinants of vernalization has direct applications in agriculture.

Supporting Online Material

Materials and Methods

SOM Text

Figs. S1 to S10

Tables S1 to S3


References and Notes

  1. Materials and methods are available as supporting material on Science Online.
  2. We thank F. Domène, J. Bensefelt, Å. Karlsson, R. Sant, P. van Roggen, and A.-M. Nilsson for technical assistance, and Y. Fischer for help in preparing the manuscript. We also thank C. Bellini for critical reading of the manuscript. This work was supported by the Swedish Research Council and the Swedish Governmental Agency for Innovation Systems to O.N. and T.K., as well as by Südzucker. Phylogenetic data have been deposited with Dryad Digital Repository (DOI: 10.5061/dryad.4930). Nucleotide sequences have been deposited with GenBank under accession numbers HM448909 to HM448918 and HQ148107 to HQ148132. The use of BvFT1 and BvFT2 in control of sugar beet flowering is the subject of the patent application WO2010/025888.
View Abstract

Navigate This Article