A naturally occurring InDel variation in BraA.FLC.b (BrFLC2) associated with flowering time variation in Brassica rapa

Flowering time is an important trait in Brassica rapa crops. FLOWERING LOCUS C (FLC) is a MADS-box transcription factor that acts as a potent repressor of flowering. Expression of FLC is silenced when plants are exposed to low temperature, which activates flowering. There are four copies of FLC in B. rapa. Analyses of different segregating populations have suggested that BraA.FLC.a (BrFLC1) and BraA.FLC.b (BrFLC2) play major roles in controlling flowering time in B. rapa. We analyzed the BrFLC2 sequence in nine B. rapa accessions, and identified a 57-bp insertion/deletion (InDel) across exon 4 and intron 4 resulting in a non-functional allele. In total, three types of transcripts were identified for this mutated BrFLC2 allele. The InDel was used to develop a PCR-based marker, which was used to screen a collection of 159 B. rapa accessions. The deletion genotype was present only in oil-type B. rapa, including ssp. oleifera and ssp. tricolaris, and not in other subspecies. The deletion genotype was significantly correlated with variation in flowering time. In contrast, the reported splicing site variation in BrFLC1, which also leads to a non-functional locus, was detected but not correlated with variation in flowering time in oil-type B. rapa, although it was correlated with variation in flowering time in vegetable-type B. rapa. Our results suggest that the naturally occurring deletion mutation across exon 4 and intron 4 in BrFLC2 gene contributes greatly to variation in flowering time in oil-type B. rapa. The observed different relationship between BrFLC1 or BrFLC2 and flowering time variation indicates that the control of flowering time has evolved separately between oil-type and vegetable-type B. rapa groups.


Background
Brassica rapa is a genus comprising a variety of vegetables such as Chinese cabbage (ssp. pekinensis), pak choi (ssp. chinensis), and turnip (ssp. rapa) as well as oil crops including turnip rape (ssp. oleifera) and sarson (ssp. tricolaris). Flowering time is an important trait in Brassica vegetables because early flowering often leads to low yield and low quality. It is also important for oilseed rape varieties as they are divided into "winter" and "spring" types according to their different flowering times and responsiveness to vernalization. Winter types must be exposed to cold to transition from the vegetative growth stage to the reproductive stage, while this is not necessary for the spring types, which are generally grown in shorter-season areas.
In Arabidopsis, studies of natural variation demonstrated that the vernalization requirement is largely conferred by two dominant genes, FRI and FLC [1][2][3]. FRI acts upstream of FLC to positively regulate FLC expression [4]. FLC encodes a MADS-box transcription factor that functions as a repressor of flowering by inhibiting downstream floral integrator genes [5][6][7][8][9]. Vernalization represses the expression of FLC and induces flowering. The promoter and first exon of FLC are sufficient to initiate the repression of FLC during vernalization, while the maintenance of repression requires additional regions of the gene body [10].
There are four copies of FLC in B. rapa [11][12][13]. They are located on chromosomes A10 (BraA.FLC.a, named BrFLC1), A02 (BraA.FLC.b, named BrFLC2), and A03 (BraA.FLC.c, named BrFLC3 and BraA.FLC.d, named BrFLC5) owing to polyploidy evolution [13,14]. Colinearity analysis indicated that BrFLC1, BrFLC2, and BrFLC3 are located in three R blocks (Xiaowu Wang et al. unpublished data), which is consistent with the three FLC copies that would be expected after a triplication event [12][13][14][15]. BrFLC5 is located between blocks I and J [16]. Multiple gene copies are thought to be responsible for dose-regulated expression, and the mechanism appears to affect variations in flowering time in Brassica crops [17]. These replicated genes may have additive effects. Depending on the specific cross studied, different alleles of the various FLC paralogs may exert different effects on flowering time. In a backcross population derived from two recombinant inbred lines, created from a cross between Per and R500, a quantitative trait locus(QTL) co-located with BrFLC1 explained more of the flowering time variation than BrFLC2 [12]. It was also reported that a naturally occurring splicing variation in BrFLC1 was associated with variation in flowering time in B. rapa, and this locus contributed most of its effect to late flowering [18]. However, Zhao et al. [19] studied a doubled haploid (DH) population derived from a cross between pak choi and yellow sarson, and reported BrFLC2 as a candidate gene for a major QTL for flowering time and the vernalization response in B. rapa. The decreased transcript level of BrFLC2 upon cold treatment provided further evidence for this hypothesis.
Since there are apparent contradictions in the proposed roles of BrFLC1 and BrFLC2, we analyzed sequence variation of BrFLC1 and BrFLC2 in a large collection of B. rapa accessions, and the relationship between sequence variations and flowering time. Our results indicate that among the various B. rapa crop types, there are different genetic controls of flowering time.

Flowering time variation
We determined flowering time for the germplasm collection of 159 B. rapa accessions in two separate experiments; one in an open field in Kunming, South China, and one in a heated greenhouse in Beijing, North China. The data from the two experiments were significantly correlated with each other (R 2 = 0.68, P ≤ 0.01). The flowering time varied from 52 to 155 days from sowing to the opening of the first flower (days to flowering, DTF) in the open field experiment, and 42 to 150 DTF in the greenhouse experiment ( To investigate the allelic variation in BrFLC2, a 1400-bp fragment was amplified from B. rapa genomic DNA with the primer combination of FLC2F8 in exon 4 and FLC2R6 in exon 7 ( Figure 1). Sequencing of the amplified fragments from the nine selected accessions revealed that the amplified region had multiple sequence variations among the nine accessions. All of these polymorphisms were single nucleotide substitutions, with two additional insertion/deletions (InDel) in line L143, a yellow sarson accession. One was a deletion of 57 bp, started at 1851 bp of BrFLC2 (Bra028599, http://brassicadb.org/brad/) and ended at 1914 bp, across exon 4 and intron 4. The deletion was interrupted by 5 nucleotides (TAAAT) that could not be mapped to a certain position of the reference sequence ( Figure 2, Additional file 2). The other one was an insertion of 29 bp at the position of 2430 bp located in intron 6 of BrFLC2. All the single nucleotide substitutions were synonymous. An InDel marker designated as BrFLC2InDel was developed to distinguish the 57 bp deletion across exon 4 and intron 4 and validated in the nine sequenced accessions ( Figure 2).

Relationship between flowering time and nucleotide polymorphisms in BrFLC2
The BrFLC2 InDel was screened over the germplasm collection consisting of 159 accessions. The deletion allele was absent from all of the vegetable-type subspecies, but it was present as a homozygous allele in nine out of 71 B. rapa ssp. oleifera accessions and in all three of the ssp. tricolaris accessions, and as a heterozygous allele in 13 B. rapa ssp. oleifera accessions (Table 3, Additional file 1). The correlation analysis showed that the InDel polymorphism was significantly associated with variation in flowering time among the oil-type B. rapa accessions (Table 3)   flowering time of the accessions with the heterozygous BrFLC2 locus were similar to those of accessions with the homozygously mutated locus, indicating that early flowering was dominant over late flowering. We further analyzed the 159 accessions using the previously reported BrFLC1MvaI CAPS marker, which can distinguish A/G alleles located at the splicing site of exon 4 and intron 4 of BrFLC1. The A allele results in alternative splicing, which is correlated with early flowering in B. rapa [18]. BrFLC1 allelic variation was observed in both vegetable-types and oil-types, showing A, G, and heterozygous alleles ( Table 3) Table 3).
The association between flowering time and the BrFLC2 InDel alleles was also analyzed for DH progenies from a BC 2 population, using a Chinese cabbage accession Z16 as the recurrent parent and yellow sarson accession L143 as the donor. Both parents had "A" alleles for BrFLC1, while Z16 had a BrFLC2 allele without deletions that was possibly functional, and L143 had the deletion allele of BrFLC2. Of the 120 screened DH lines, 3 had the deletion allele. We investigated the flowering phenotype of these three lines, five randomly selected lines with the functional allele, and the two parental lines. The lines with the deletion allele showed significantly shorter flowering times than those of lines with the functional allele ( Table 4). The flowering times of DH lines with the functional BrFLC2 allele ranged from 83 DTF to 92 DTF, while those of lines with the mutated deletion allele ranged from 70 DTF to 80 DTF.

Alternative splicing of BrFLC2
To identify any alternative splicing of the mutated BrFLC2 allele, RT-PCR was conducted with BrFLC2-specific primers (BrCFLC2F and BrCFLC2R) using the yellow sarson accession L143, which had a homozygous deletion allele.  Table 2.  insertion and 22 bp retained at the end of intron 4; and SPD3) 686-bp transcript-same as SPD1 besides intron 3 retained. Of the 36 sequenced cDNA clones, 28 showed SPD1, six showed SPD2, and two showed SPD3. This indicated that the SPD1 was the major splicing pattern for the deletion allele of BrFLC2, while SPD2 and SPD3 were derived from SPD1.

Discussion
An InDel polymorphism across exon 4 and intron 4 of BrFLC2 was discovered in a subset of oil-type B. rapa accessions, including ssp. oleifera and ssp. tricolaris. In this study, we investigated its relationship with flowering time both in a collection of B. rapa natural accessions and in a BC 2 DH population. Plants need to sense their environment and initiate flowering at the appropriate time to ensure successful fertilization and production of abundant seeds. There is considerable variation in the flowering time among, but also within, natural populations, as we observed in the present study of a B. rapa germplasm collection. In A. thaliana, the FRI gene was shown to be a major determinant of flowering time variation in the natural population through its effects on the expression of FLC [21,22]. An InDel variation in the COL1 gene was reported to be correlated with variation in flowering time in B. nigra [23]. In A. thaliana, FLC encodes a MADS-box transcription  factor that acts as a dose-dependent flowering repressor [9,22]. Four copies of the FLC gene in B. rapa increase the potential variation in flowering time [17]. In our previous study, we identified a splicing site polymorphism Pi6 + 1 (G/A) in BrFLC1 that was significantly associated with the naturally occurring variation in flowering time in B. rapa [18]. In that study, we examined 96 lines, six of which were oil-types. In contrast, half of the lines examined in the present study were oil-types. Because there were so few oil-types in our previous study, we did not identify that oil-type and vegetable-type B. rapa showed different relationships between alleles of BrFLC1 and flowering time variation. In the present study, we could not detect any effect of allelic variation in BrFLC1 Pi6 + 1(G/A) on the variation in flowering time for oil-type B. rapa, including ssp. oleifera and ssp. tricolaris. However, we detected an InDel polymorphism across exon 4 and intron 4 of BrFLC2 among the accessions of ssp. oleifera and ssp. tricolaris. This sequence polymorphism was not detected in any of the other vegetable-type B. rapa subspecies. Furthermore, this allelic variation was strongly associated with variations in flowering time in oil-type B. rapa. Zhao et al. [19] suggested that BrFLC2 was a major determinant of flowering time variation in B. rapa. However, there were no firm conclusions from several studies on BrFLC1 [12,18] and BrFLC2 [19]. Since BrFLC1 and BrFLC2 have specific roles in controlling flowering time in different B. rapa groups, we deduced that there was an independent evolution of the control of flowering time, at least the control of the vernalization pathway, between oil-type B. rapa including ssp. oleifera and ssp. tricolaris and the other vegetable-type and turnip B. rapa subspecies. The oil-type B. rapa formed an evolutionary branch that was independent of other B. rapa species in an analysis of molecular phylogeny based on whole genome re-sequencing data generated from 108 accessions (Dr. Xiaowu Wang, unpublished data). This indicates that the evolutionary history of oil-type B. rapa is isolated from that of the vegetable-type subspecies. The fact that the deletion mutation of BrFLC2 was absent from vegetable-type B. rapa indicates that this mutation may have arisen after the division of oil-type from vegetable-type B. rapa, while the splicing site mutation of BrFLC1 may have arisen before this division and been maintained during their respective evolutions. Relationships between naturally occurring alternative splicing variants and flowering time variation have been reported for the FLC gene in A. thaliana [24] and Capsella bursa-pastoris [25], and for BrFLC1 in B. rapa [18]. Alternative splicing variants were also reported for BrFLC5 in a biennial oilseed cultivar, although their relationship with flowering time was not addressed [12]. In the present study, we detected three alternative splicing patterns for BrFLC2 in the yellow sarson accession L143, which has a homozygous deletion allele of BrFLC2. All three alternative splicing variations led to the insertion of premature stop codons in the transcripts. The alternative splicing pattern iii of BrFLC2 has been reported by Zhao et al. [19] using DH lines derived from a cross between the same yellow sarson accession and a pak choi accession, and was deduced to be a regulatory mechanism for the differential expression of BrFLC2 in response to vernalization. In the present study, the transcripts from splicing pattern iii were the minor fraction of transcripts from the deletion allele of BrFLC2. This could be due to the different cultivation conditions in the two studies, as the plants were not cold-treated in this study. A possible reason for the differential expression in response to cold treatment might be that alternative splicing transcripts were eliminated by the mRNA surveillance system. Eukaryotes have an mRNA surveillance system to eliminate the transcripts that are deliberately spliced to contain premature stop codons as a part of their intricate autoregulatory system [26].
B. rapa is a mesohexaploid that has undergone whole genome triplication after divergence from a common ancestor of A. thaliana. During the diploidization process afterwards, which involved considerable gene loss, some gene family showed preferential retention such as circadian clock genes [27], and also many of flowering time genes. It has been speculated that polyploidy and lost of the duplicated genes may have contributed to the evolution of variations in flowering time, a key component of morphological diversity [17]. After the hexaploid process, the three sub-genomes of the ancestor were partitioned into LF, MF1, and MF2 [13]. BrFLC1 is located in LF, BrFLC2 in MF2, and BrFLC3 in MF1 [13], while BrFLC5 is located in the homologous region generated from an α-duplication event that occurred before the diversification of Arabidopsis-Brassica [14]. The fact that a non-functional BrFLC1 mutation introduces early flowering time variation in vegetable-type B. rapa, while the non-functional BrFLC2 introduces early flowering time variation in oil-type B. rapa, indicates that these two loci of FLC in B. rapa play different roles in different groups. It has been proposed that nonfunctionalization of duplicate genes could provide an important source of phenotypic variation [25]. We have shown that the deletion in BrFLC2 also promoted flowering in a genetic background of Chinese cabbage line Z16. However, it remains unknown why different alleles of BrFLC1 show no difference in flowering time in oiltype B. rapa accessions. We need to sequence all of the BrFLC1 alleles in these accessions to determine whether they contain additional mutations. Genetic redundancy provides flexibility for plants growing in changeable environments. It is also possible that the other two homologs of FLC might function to compensate for the loss of function of BrFLC1 or BrFLC2. We sequenced BrFLC3 or BrFLC5 using primers designed from sequences in exon 4 and exon 7, respectively, and we did not identify any functional sequence variation for the nine accessions (unpublished data). However, we can not exclude the possibility that there are sequence variations located in other regions that might affect their functions. Further research on BrFLC3 in subgenome MF1 and BrFLC5 as a relic of the αduplication event and their influence on flowering is underway. We anticipate that they may have differentiated from, and are functionally different from, BrFLC1 and BrFLC2.

Conclusions
Our results suggest that the naturally occurring deletion mutation across exon 4 and intron 4 in BrFLC2 contributes greatly to the variation in flowering time in oiltype B. rapa. The different relationships between BrFLC1 or BrFLC2 and the variation in flowering time of vegetable-type and oil-type B. rapa indicate that control of flowering time undergone separate evolution between these two groups.

Plant materials and flowering time evaluation
To characterize the natural variation of flowering time in B. rapa, we measured 159 accessions belonging to 11 cultivar groups ( A BC 2 DH population with 120 lines derived from a cross between the yellow sarson line L143 (R500 from Wisconsin University) and the Chinese cabbage line Z16, using Z16 as the recurrent parent, were screened for polymorphisms of the BrFLC2Indel marker. We selected five lines with the insertion allele and three lines with the deletion allele and grew them in the greenhouse in Beijing from 20 August 2010 to investigate flowering time. Five replicates were grown for each line. Flowering time was scored as the number of days from sowing to the opening of the first flower (DTF). For the nine B. rapa accessions that were used to sequence BrFLC2, DTF was recorded as described above

BrFLC2 amplification
Genomic DNA was isolated from leaf samples using the CTAB method [28]. Specific primers were designed for BrFLC2 (AY205317S1, AY205317S2). BrFLC2 was amplified by nested PCR. The outside forward primer was FLC012F1 (50-CCTTGATCGATATGGGAAA-CAAC -30) located in exon 2 and the outside reverse primer was FLCR5 (50-TAATTAAGYAGYGGGAGAGTY AC-30) located in exon 7. The inside forward primer was FLC2F8 (50-GGAATCAAATTCTGATGTAAGCGTC -30) located in exon 4 and the inside reverse primer was FLC2R6 (50-TTTGTCCAGGTGACATCTCCATT-30) located in exon 7 ( Figure 1). The amplified fragment covered the region of exons 4-7 and the intervening introns between these exons. PCR was carried out in a total volume of 20 μl containing 50 ng template DNA, 0.5 μM each primer, 200 μM dNTPs, 1× PCR reaction buffer, and 1 U Taq polymerase. PCR was performed under the following conditions: the template was denatured at 94°C for 5 min, followed by 35 cycles of amplification (94°C for 1 min, 56°C for 1 min, 72°C for 1 min 30 s), and a final extension at 72°C for 10 min. PCR products from the nine accessions listed in Table 2 were purified by ethanol and NaAc (3 M, pH = 5.4) and then cloned into the PMD-18 T vectors (Promega, http://www.promega. com) for sequencing.

InDel marker analysis and CAPS marker analysis
The forward primer FLC2IndelF (50-GTCGACTCCCTCG TTCAGC -30) in exon 4 and the reverse primer FLC2In-delR (50-AGGGAAACTAATACAATACGCAA -30) in intron 5 were designed to develop an InDel marker for BrFLC2. PCR was performed under the following conditions: denaturation at 94°C for 3 min, followed by 35 cycles of amplification (94°C for 45 s, 55°C for 45 s, 72°C for 1 min), and a final extension at 72°C for 10 min. The PCR products were fractionated on an 8.0% polyacrylamide gel to determine the genotype of the InDel marker. We used the CAPS marker for BrFLC1, FLC1-MvaI, to screen the 159 accessions as described by Yuan et al. [18].