FLOWERING LOCUS C (FLC) in Arabidopsis encodes a dosage dependent repressor of flowering. We isolated five FLC-related sequences from Brassica napus (BnFLC1–5). Expression of each of the five sequences in Arabidopsis delayed flowering significantly, with the delay in flowering time ranging from 3 weeks to more than 7 months, relative to the flowering time of 3 weeks in untransformed Ler. In the reciprocal experiment, expression of Arabidopsis FLC (AtFLC) in an early flowering B. napus cultivar delayed flowering by 2–6 weeks, confirming the requirement of this gene for floral repression. In B. napus, we show that late flowering and responsiveness to vernalization correlate with the level of BnFLC mRNA expression. The different BnFLC genes show differential expression in leaves, stems and shoot tips, but expression is not detectable in roots. Vernalization dramatically reduces the level of BnFLC transcript and restores early flowering in the winter cultivar Colombus. We conclude that BnFLC genes confer winter requirement in B. napus and account for the major vernalization-responsive flowering time differences in the different cultivars of B. napus in a manner analogous to that of AtFLC in Arabidopsis ecotypes.
In most plants, the transition from vegetative to reproductive development is controlled by both environmental and developmental signals. Environmental factors such as photoperiod, light quality and quantity, temperature, nutrients and water supply affect flowering time (Levy and Dean, 1998), and the plant needs to achieve a certain stage of developmental competence to respond to such cues. Endogenous signals such as phytohormones, notably gibberellins, also influence the vegetative to floral transition (Langridge, 1957; Pharis and King, 1985; Zeevaart, 1976).
Many plants from temperate regions have a distinct winter requirement for flowering. When such plants are grown in the glasshouse under standard conditions, they flower very late, or not at all, unless exposed to an extended period of low temperature, a process called vernalization. Vernalization is an adaptive response whereby plants ensure seed set in the favourable conditions of spring. The integration of these various signals to modulate developmental phase change appears to be complex, however, genetic and molecular analyses of flowering time in Arabidopsis thaliana have defined a sequence of events at the molecular level (Koornneef et al., 1998b; Levy and Dean, 1998; Reeves and Coupland, 2000; Sheldon et al., 2000a).
Induced mutations in Arabidopsis have revealed the existence of about 80 loci that affect flowering time (Levy and Dean, 1998). Genetic analysis with some of these loci show epistatic interactions, while others have additive effects (Koornneef et al., 1998a, b) indicating the existence of more than one pathway for floral induction. However, the majority of flowering time differences in the naturally occurring ecotypes of Arabidopsis can be explained by differences in alleles at just two loci, FRIGIDA (FRI) and FLOWERING LOCUS C (FLC) (Burn et al., 1993a; Clarke and Dean, 1994; Johanson et al., 2000; Koornneef et al.; 1994). FRI and FLC act synergistically to repress flowering.
A recent breakthrough in the understanding of the molecular basis of vernalization has been the cloning of the FLC gene (Michaels and Amasino, 1999; Sheldon et al., 1999). FLC encodes a MADS-box floral repressor whose level of expression correlates with time to flower in the different ecotypes of Arabidopsis. We have shown that vernalization promotes flowering by reducing the level of FLC transcript and protein (Sheldon et al., 1999, 2000b). The fact that flowering time differences between the different ecotypes of Arabidopsis can be explained by the level of FLC expression suggests that FLC is a key repressor of flowering (Sheldon et al., 2000b). FRI seems to exert its main effect on flowering time by upregulating FLC (Michaels and Amasino, 1999; Sheldon et al., 1999).
Mutations that directly affect FLC expression alter flowering time in the absence of altered FRI levels. For example, the flc-11 overexpression mutant takes more than 6 months to flower while the flc-13 knockout mutant flowers in about 16 days compared with 4–5 weeks for the parental C24 (Sheldon et al., 1999; 2000b). Additionally, the autonomous flowering pathway mutants fca, fve, fpa, fd, fld, and fy, are late-flowering because they are unable to down regulate FLC expression (Sheldon et al., 1999; 2000b; D. Rouse, pers. comm.).
In all of these late-flowering mutants, vernalization reduces the FLC transcript to a low level and restores normal flowering time. Thus, FLC is necessary and sufficient for the control of vernalization-responsive flowering time in A. thaliana. Whether FLC plays a similar key role in other species is yet to be shown. If it does, the gene will have important applications in agriculture in the development of crop cultivars with flowering time optimized to the climate for improved harvest index and productivity.
In this study, we investigated the role of FLC in the Brassica crop, canola, which belongs to the same family as Arabidopsis. Canola (Brassica napus L.) includes both spring annuals and winter biennials. Some spring canola varieties are early flowering and do not respond to vernalization, while others respond to cold by flowering earlier, but can flower late without exposure to cold. The winter biennials remain vegetative in the absence of prolonged exposure to cold. Comparative mapping in A. thaliana and B. napus has demonstrated that vernalization-responsive flowering time loci in B. napus are collinear with the tops of chromosome IV and V in Arabidopsis where FRI and FLC, respectively, are located (Osborn et al., 1997). We present evidence that FLC-related sequences are central to the vernalization response of Brassica napus. There are at least five FLC-related genes in B. napus (BnFLC1-BnFLC5), each of which delays flowering when expressed in A. thaliana.
Expression of AtFLC in Brassica napus delays flowering
We evaluated the effect of expressing the Arabidopsis FLC gene in Brassica by transforming the spring annual B. napus cv BLN1239 with AtFLC cDNA, driven by a CaMV 35S promoter. We obtained 15 independent lines that were delayed in flowering. Three of these primary transformants are shown in Figure 1(a). Five of the 15 lines (selected on the basis of flowering time and protein level in the T0 generation) were analyzed further in the next generation (Figure 1b,c). When soluble protein was assayed from the leaves of five hygromycin resistant T1 seedlings in each line, AtFLC protein was detected, with line CS1.8/5 showing the highest and the other four lines (CS1.2/1, CS1.7/3, CS1.15/1, and CS1.18/1) showing slightly lower protein levels (Figure 1b). These T1 lines were delayed in flowering by 2–6 weeks compared with the untransformed wild-type (Figure 1c), the degree of lateness reflecting the level of AtFLC protein accumulation (Figure 1b,c). We conclude that expression of the AtFLC gene delays flowering in the early flowering B. napus cultivar BLN1239. Therefore, it is likely that Brassica has a similar mechanism of floral initiation to that of A. thaliana.
Expression of AtFLC in B. napus also disturbed other components in inflorescence development. In control BLN1239 plants, the vegetative stem elongated (bolted) well before the appearance of floral buds and opening of the flowers. In most of the late-flowering transgenic lines, bolting was inhibited relative to flowering. By the time the first flower opened, the inflorescence stem of control BLN1239 was 2–3 times longer than that of the transgenic lines. The latest flowering transgenic plants had open flowers before any noticeable stem bolting occurred. After opening of the first flower, elongation of the inflorescence stem was as normal, with the transgenic lines elongating at approximately the same rate as the control, and achieving almost the same height. Fertility was also slightly reduced in the transgenic lines, and some plants did not set any seeds. The stamens of the sterile transgenic plants were shorter than the carpel and the anthers remained below the stigma making self-pollination difficult. The flowers were otherwise normal. Hand pollination of these flowers with pollen from the same plant resulted in seed set (data not shown) indicating that the impedance to fertilization is due to the shortening of the anther filaments.
Vernalization of transgenic BLN1239 seeds altered neither the pattern of transgene expression nor the lateness of flowering. This is to be expected because AtFLC was driven by the constitutive 35S promoter, which is not cold-responsive. Transgenic lines took longer to flower and produced about 22 leaf nodes at the time of flowering compared with about 15 leaf nodes for the untransformed BLN1239, whether vernalized or not. The level of AtFLC transcript in the transgenic line immediately after vernalization remained similar to the unvernalized transgenic line (data not shown) suggesting that the effect of vernalization may be primarily on transcription rather than RNA stability or post-transcriptional processing.
Flowering time is controlled by five FLC-related genes in B. napus
To investigate whether FLC-related genes control flowering time in the spring and winter cultivars of B. napus, we isolated five different putative FLC sequences (BnFLC1–5) from a seedling and young leaf cDNA library of the winter cultivar Colombus. These clones encode MADS-box proteins with predicted amino acid sequences showing between 79 and 86% identity to the AtFLC protein, and 80–95% identity with each other (Figure 2a). Although the highest similarity with AtFLC is in the MADS domain, there is also significant similarity outside this region (Figure 2a).
Phylogenetic analysis with selected MADS-box proteins indicated that the predicted B. napus FLC proteins and the Arabidopsis FLC (AtFLC) fall into one clade well segregated from all other MADS-box proteins (Figure 2b). There are five FLC-like (FLCL1–5) genes in the Arabidopsis genome (accession numbers AAK37527, BAB10332, BAA97510, BAA97511and BAB11644), four of the genes (FLCL2–5) are located on chromosome V and FLCL1 is located on chromosome I. These Arabidopsis FLCL sequences are the next most closely related group to BnFLCs amongst all the MADS-box proteins reported to date. AtFLCL1, also called MAF1 or FLM, has been shown recently to repress flowering, but unlike FLC, it has no vernalization response (Ratcliffe et al., 2001; Scortecci et al., 2001). The functions of the other Arabidopsis FLCL sequences have not been reported. Although they lie in close proximity to FLC in the phylogenetic tree, bootstrap neighbour joining statistical analysis predicted that the AtFLCL sequences as a group are separate from FLC with a 100% significance. More importantly, all of the BnFLC sequences are grouped together with FLC 100% of the time (Figure 2b). This suggests that the five BnFLC genes could be orthologues of AtFLC but not of the FLCL genes. We have named the B. napus sequences BnFLC1–5 in which BnFLC1 denotes the sequence with the highest and BnFLC5 with the lowest amino acid identity to AtFLC.
To determine the function of the putative BnFLC genes, we transformed the early flowering ecotype of Arabidopsis, Ler, with each of the five cDNAs under the control of a CaMV 35S promoter. For each construct, 4–14 transformants were analyzed in the T1 generation and the phenotype of one plant representing each construct is shown in Figure 3(a). Untransformed Ler bolted at 20–24 days. All of the five BnFLC constructs delayed flowering significantly, with delays in flowering time varying from 3 weeks to more than 7 months relative to Ler. Some lines never flowered, in particular those containing the BnFLC1 construct. All of the lines containing the BnFLC4 or BnFLC5 construct flowered in less than 3 months. The very-late-flowering lines were characterized by the production of numerous aerial rosettes one above the other resulting in a dome-shaped appearance, reminiscent of the flc-11 overexpression mutant in C24 (Sheldon et al., 1999).
One medium-late-flowering transgenic line representing each gene construct was selected for further testing in the T2 generation. The T2 lines containing the transgene (both heterozygous and homozygous) showed significant delays in bolting, ranging from 39 days with about 18 leaves in the rosette to 140 days with more than 75 leaves in the rosette compared with 20–24 days with 6–7 leaves in the rosette in the untransformed Ler (Figure 3b). Northern blot analyses, using probes for each BnFLC gene, confirmed that the late-flowering phenotype is conferred by expression of the BnFLC transgenes (Figure 3c). The level of BnFLC expression reflected the degree of lateness between lines of the same gene construct; those that never flowered in the T1 generation had the highest expression (data not shown), suggesting that all five B. napus clones could be functionally equivalent to AtFLC. In general, BnFLC1, 2 and 3 caused later flowering than BnFLC4 and 5 in transgenic Arabidopsis both in the T1 and T2 generations, suggesting that some of the genes could be more important than others in controlling flowering time in B. napus. Alternatively, some of the genes could have been more effective at delaying flowering in Arabidopsis because their encoded proteins are more similar in sequence to that of AtFLC and therefore better able to interact with the Arabidopsis floral induction machinery.
BnFLC expression in B. napus cultivars correlates with responsiveness to vernalization and time to flowering
We examined the BnFLC transcript level and response to vernalization of four spring and winter cultivars (BLN1239, Westar, Drakkar, and Colombus) of B. napus. Under long-day glasshouse conditions, BLN1239 and Westar flowered early as measured by the number of days to flowering (Figure 4a) or the number of leaf nodes at flowering (data not shown). Drakkar took longer to flower, while Colombus did not flower in 6 months (Figure 4a). Northern analysis, using a BnFLC1 full-length riboprobe that can detect BnFLC1–5 transcripts, showed that lateness in flowering time is directly related to BnFLC transcript accumulation (Figure 4b). In the early flowering spring cultivars, BLN1239 and Westar, BnFLC transcript could be detected only at low levels, whereas in Colombus, a late-flowering winter cultivar, expression was high; in Drakkar, a medium-late flowering spring cultivar, the level was intermediate (Figure 4b). These experiments confirm the direct relationship between flowering time and FLC levels in canola.
This direct relationship between BnFLC expression and time to flowering was further supported by vernalization experiments. When 2-week-old seedlings were vernalized and their transcript levels analyzed immediately after the cold treatment, we found the BnFLC transcripts were reduced to low levels in all four cultivars (Figure 5a). The greatest change in mRNA level was seen in Colombus where the initial high transcript level was reduced to a level comparable with those of the other cultivars. This reduction was accompanied by a dramatic decrease in flowering time. Vernalized Colombus plants flowered 5–6 weeks post vernalization, having 15–20 leaf nodes on the main axis, while unvernalized plants did not flower in 6 months and had many leaves (Figure 5b). So in Colombus, a large reduction in FLC mRNA level corresponded to a large reduction in flowering time. Vernalization did not significantly alter the flowering time of BLN1239 and Westar, but Drakkar responded to the cold treatment by flowering earlier (Figure 5b). These data demonstrate that FLC activity is the main determinant of vernalization-responsive flowering time in the B. napus cultivars.
Colombus FLC genes are differentially expressed
We analyzed the expression of the BnFLC genes in different tissues of Colombus using gene-specific probes. BnFLC1 is expressed weakly in leaves, moderately in shoot tips, strongly in stems, but not detectably in roots (Figure 6). BnFLC2 showed strong expression only in leaves and not in shoot tips, stems or roots. BnFLC3 is strongly expressed in leaves and the expression level is lower in shoot tips and stems, but is not detected in roots. The BnFLC4 signal was very weak in leaves, weak in shoot tips and stronger in stems. BnFLC5 is the lowest expressed of the five genes. A faint signal could be detected in shoot tips and stems, but not in leaves or roots. Although a very faint signal could be visualized in the root lane of BnFLC1, root expression of all five genes in general was at or below the detection limit of our hybridization conditions (Figure 6). We conclude that, unlike AtFLC, the BnFLC gene family is not highly expressed in roots.
B. napus has at least five FLC genes
Unlike Arabidopsis, which has a single FLC gene, Brassica napus has five closely related genes, all of which can function to repress flowering in transgenic Arabidopsis. Arabidopsis and Brassica are in the same family, Brassicacae. The assumption is that in the genus Brassica there has been a whole-genome triplication, followed by chromosome rearrangements, including fusions, and fissions that led to the formation of Brassica with three copies of a rearranged unit genome. Each unit genome is similar to the genome of Arabidopsis (Lagercrantz, 1998; Lagercrantz and Lydiate, 1996). Comparative genetic analyses fit this assumption, the data showing that a single locus in the Arabidopsis genome is represented by three loci in the genomes of B. rapa, B. oleracea and B. nigra (Kowalski et al., 1994; Lagercrantz, 1998; Osborn et al., 1997). B. napus is derived by interspecific hybridization of B. rapa and B. oleracea (Osborn et al., 1997; Parkin et al., 1995), so it is expected to contain six Arabidopsis genome equivalents. Due to the high rate of chromosomal rearrangement, fusion, and deletion in Brassica (Lagercrantz, 1998), it is not always possible to identify six copies for every gene in Arabidopsis. In fact, on average 2–8 copies have been reported in B. napus for single gene loci in Arabidopsis (Cavell et al., 1998; Osborn et al., 1997; Scheffler et al., 1996). Representation of the FLC gene by five genes in B. napus is consistent with all five being functional orthologues of AtFLC.
Three vernalization responsive flowering time loci VFR1, VFR2 and VFR3 have been identified genetically in B. rapa (Kole et al., 2001; Osborn et al., 1997). These loci have homeologous regions in B. napus, collinear with the top of chromosome V of Arabidopsis, where FLC is located (Osborn et al., 1997). The availability of BnFLC probes should now enable fine scale mapping to determine the chromosomal locations corresponding to the genetic loci.
BnFLCs delay flowering in Ler and AtFLC delays flowering in B. napus
Our finding that the five BnFLC genes delay flowering in Ler (Figure 3) suggests that all of them could be functional in B. napus. Variation in the degree of late-flowering conferred by the different cDNAs in Arabidopsis suggests that some of the genes could be more important than the others in B. napus. Figure 3 shows that BnFLC1, 2 and 3 cause later flowering than BnFLC4 and 5. FLC transcript analysis in Colombus shows that BnFLC4 and 5 have a lower expression compared with the others (Figure 6). It appears that the degree of their relatedness to AtFLC reflects the order of their ability to repress flowering in Arabidopsis. However, whether all of the five genes play a similar role in the control of vernalization responsive flowering in B. napus remains to be shown.
In the reciprocal experiment, constitutive expression of AtFLC in B. napus delays flowering, reduces inflorescence stem elongation before flowering and causes shortening of anther filaments leading to reduced fertility. Gibberellins mediate stem elongation and promote flowering in a number of species, including B. napus, and exogenously applied gibberellins partially substitute for the vernalization requirement. In winter canola, vernalization is shown to increase GA biosynthesis and metabolism (Zanewich and Rood, 1995). The flc-11 overexpression mutant in Arabidopsis requires a greater dose of GA3 to flower than the parental C24 (Sheldon et al., 1999) suggesting that FLC may have a quantitative effect on GA action or biosynthesis.
B. napus FLC genes are differentially expressed
In many ecotypes and autonomous pathway late-flowering mutants of Arabidopsis, FLC is highly expressed in most tissues including roots (Sheldon et al., 1999; 2000b). B. napus FLCs are also expressed in most tissues, but very weakly in roots (Figure 6). In late-flowering autonomous pathway mutants in Ler background such as fca, FLC is highly expressed in the aerial tissue, but expression in the root is below the detection limit similar to the wild-type Ler (D. Rouse, pers. comm.). Thus, expression of BnFLCs in Colombus is similar to that of FLC in Ler mutants. This could be related simply to ecotype differences where very low root expression is sufficient in some cases. It could also be possible that FLC in roots may have another function which is taken-over by another gene(s) in Ler mutants and Colombus. The shoot apical meristem does perceive low temperature and initiate reproductive development. However, a number of experiments showed that other tissue explants including roots from vernalized plants are capable of regenerating a plant which flowers early, as in the vernalized state (Burn et al., 1993b; Metzger, 1988; Wellensiek, 1962; Zeevaart, 1976). This suggests that a number of tissues are capable of sensing and responding to the cold signal.
In Arabidopsis, the detailed mechanism of FLC floral repression is not known. A MADS-box protein, AGAMOUS-LIKE 20 (AGL20) also known as SUPPRESSOR OF OVEREXPRESSION OF CONSTANS 1 (SOC1) has been described recently in Arabidopsis, which functions to promote flowering (Borner et al., 2000; Lee et al., 2000; Onouchi et al., 2000). AGL20 expression is low whenever FLC is high in a number of late-flowering mutants, high in early flowering mutants and ecotypes of Arabidopsis, and upregulated by vernalization in the autonomous pathway late-flowering mutants suggesting that FLC may be a repressor of AGL20 (Lee et al., 2000). However, whether FLC is a direct repressor of AGL20 is not shown, and a canola homologue of AGL20 has not been reported. Dissection of the detailed molecular events in canola will enhance our understanding of the mechanism of cold perception.
Plant material and growth conditions
Brassica napus cv BLN1239 (our own stock), Westar, Drakkar and Colombus (kindly provided by P. Salisbury) were grown in a glasshouse with a 16-h photoperiod and a temperature of 24 ± 2°C during the day and 16 ± 2°C at night. Plants were grown in 15 or 20 cm pots with compost containing gypsum. Flowering time was recorded as the number of days or number of leaf nodes on the main axis when the first flower opened. For RNA extraction, Colombus seeds were aseptically grown in tubes containing a modified Murashige and Skoog medium (Langridge, 1957) for 3 weeks, and roots, stems, leaves and shoot tips were collected separately.
Arabidopsis (Landsberg erecta) seeds were germinated on petri dishes as described (Sheldon et al., 2000b), and grown in soil in cabinet with a 16-h cool fluorescent light at approximately180 µm m−2 s−1 and a constant temperature of 21°C. Flowering time was recorded as the number of days or number of rosette leaves when the floral bolt became clearly visible.
B. napus were either grown in the glasshouse for 14 days and transferred to a cold-room (5 ± 1°C) with a 16-h photoperiod for 8 weeks, or seeds in soil were placed directly into a cold-room for 10 weeks, the first 2 weeks in the dark and the subsequent 8 weeks with a 16-h photoperiod at about 100 µm m−2 s−1. Six to eight plants or seeds per pot were vernalized, and at the end of the cold treatment, samples were either collected for RNA analysis or plants were transferred to a 12°C cabinet for 2 days and then transplanted to one plant per pot and grown in the glasshouse. Flowering time was recorded as the total number of leaf nodes on the main axis when the first flower opened.
cDNA library construction and screening
Total RNA was extracted from unvernalized 14 day-old seedlings and young leaves of B. napus cv. Colombus, and mRNA was isolated using mRNA Purification Kit (Amersham Pharmacia Biotech, NJ, USA). First-strand and second-strand cDNA synthesis were performed using SuperScript Choice System for cDNA synthesis Kit (Life Technologies, Maryland, USA) following the manufacturer's instruction. cDNA inserts were cloned in the EcoRI site of λZipLox vector (Life Technologies), and the library was screened using a 500-bp 3′ (non-MADS box encoding) region of the AtFLC cDNA as probe. Positive plaques were excised in plasmid pZL1 from λZipLox and sequenced. A total of 20 strongly hybridizing plaques were sequenced, which represented five different putative canola FLC genes on the basis of sequence comparison.
Generation of transgenic plants
For canola transformation, the 35S::AtFLC (35S::FLF) NotI cassette from pART27 (Sheldon et al., 1999) was subcloned into the NotI site of pWBVec8 (Wang et al., 1997) and introduced into A. tumefaciens AGL1 by triparental mating. B. napus cv. BLN1239 was transformed with A. tumefaciens and transgenic plants were regenerated from hygromycin-resistant calli with a modified method of Bade and Damm (1995). The T1 seeds of these transgenic lines were germinated on MS medium containing hygromycin (20 mg l−1) for 2 weeks and transplanted to soil. Wild-type BLN1239 was also germinated on MS medium without hygromycin and flowering time was scored as shown in Figure 1(c). BLN1239 seeds containing the hygromycin resistance gene but not the FLC gene flowered the same time as the wild-type after germination on hygromycin medium and transfer to soil (data not shown). For Arabidopsis transformation, first the NotI cassette from pART7 (Gleave, 1992) was subcloned into the NotI site of pART27 (Gleave, 1992) making binary vector pART27 + 7. The EcoRI cassette of each BnFLC cDNA clone from pZL1 (see above) was then subcloned into the EcoRI site of pART27 + 7 and introduced into A. tumefaciens GV3101. Landsberg erecta was transformed with each of the 5 BnFLC constructs using a floral-dip method (Clough and Bent, 1998).
RNA extraction and Northern analysis
Total RNA was extracted from about 100 mg of tissue using the RNeasy Plant Mini Kit (Qiagen, Hilden, Germany) or from larger amounts of tissue according to Logemann et al. (1987). Northern blotting and hybridizations were performed as described (Sheldon et al., 1999). Blots were hybridized at 55°C in formamide hybridization solution (50% formamide, 0.25 m NaPO4 pH 7.2, 0.25 m NaCl, 1 mm EDTA, 7% SDS) washed in 2 × SSC/0.1% SDS twice for 5 min at 55°C, incubated in 2 × SSC containing 500 µg l-1 RNase A for 10 min at room temperature, and washed twice in 0.1 × SSC/0.1% SDS for 10 min at 65°C.
Protein extraction and immunodetection
Total soluble protein extraction and immunodetection of AtFLC protein by chemiluminescence was performed as previously described (Sheldon et al., 2000a). T1 seeds were selected on hygromycin (20 mg l−1) containing MS medium for 2 weeks and leaf samples were combined from five hygromycin-resistant plants in each line. The AtFLC antiserum was kindly provided by D. Rouse, CSIRO, Canberra.
We thank C. Hurlstone for his help with canola transformation, P. Salisbury for providing three canola lines, D. Rouse for the FLC antiserum, J. Finnegan, J. Watson and R. King for critically reading the manuscript. M.T. was supported by the Swiss National Science Foundation postdoctoral fellowship grant and by GrainGene.