Background

Cabbage (Brassica oleracea L. var. capitata L.) is one of the most cultivated vegetables worldwide. In 2010, more than two million hectares were estimated to be under production, with an average yield of 27.8 tonnes per hectare (FAOSTAT 2010). Cabbage Fusarium wilt (CFW) is a destructive fungal disease caused by Fusarium oxysporum f. sp. conglutinans, which for many years has caused severe losses to cabbage yields all over the world. CFW was first discovered in the USA in 1899, and then in Japan and several other countries in the following decades [1, 2]. In recent years, CFW was discovered in several provinces in China and has spread rapidly [3–16], but, until now, no reports have focused on gene analysis or on functional studies of candidate genes. In a previous study, we located the resistance gene on chromosome C06 of cabbage, with two InDel markers flanking the gene at 1.2 and 0.6 cM [16]. A total of 18 markers were anchored to the genetic map using JoinMap 4.0, which matched to the physical map according to the positions of the markers (Figure 1a). Finally, using these markers and the DH population, we mapped the FOC1 gene to a 382-kb genomic region between the S3 and A1 markers (Figure 1b). Moreover, the SSR markers indicated that this was the same region as the region reported in the previous study (Figure 1b) [16].

Table 1 Primers used for fine map** of FOC1
Full size table
Figure 1
figure 1

Fine map** of FOC1 using the DH and F 2 populations. (a) The genetic map of 18 markers obtained from JoinMap 4.0. (b) Preliminary map** of FOC1 using the DH population. The FOC1 gene was mapped to an interval of about 382 kb with S3 and A1 as the flanking markers according to the reference sequence of B. oleracea. (c) Fine map** of FOC1 using the F2 population with 3976 plants. The FOC1 gene was delimited to an interval between V17 and S9, with an estimated length of 84 kb. (d) Ten genes were annotated between markers V17 and S9 based on the reference genome sequence. The genetic structure for each recombinant type is depicted as black rectangles for homozygous susceptible alleles, white for homozygous susceptible alleles and grey for heterozygous alleles, respectively. The relative positions of markers on C06 were determined according to the B. oleracea genome sequence.

Full size image

The flanking markers A1 and S3 obtained from the DH population were used to screen the F2 population for recombinant individuals, which were then evaluated for their resistance to Fusarium wilt. Next, more markers between A1 and S3 were used to genotype these individuals. As a result, the FOC1 gene was narrowed to an 84-kb genomic region with V17 and S9 as the flanking InDel markers (Figure 1c). Both InDel markers had two recombinants with FOC1.

Candidate gene analysis

From BRAD (http://brassicadb.org) [21], the genomic database for B. oleracea where the reference genome sequence for B. oleracea is now available [18], we obtained the reference genome sequence and found that there were ten annotated genes in the 84-kb genomic region (Figure 1d, Table 2). However, we noticed that there was no gene numbered Bol037164 in the database, and the prediction of the sequence between Bol037163 and Bol037165 showed that no gene existed here. The best hits of the ten annotated B. oleracea genes to the Brassica rapa and Arabidopsis thaliana genome are shown in Table 2. Annotation analysis of these genes from BRAD indicated that four of them were most likely related to disease resistance [22]: Bol037161 was predicted to have transcription factor activity; Bol037156, Bol037157 and Bol037158 contained the most commonly conserved domain of R proteins: leucine-rich repeat (LRR), toll-interleukin receptor (TIR) domain and LRR domain respectively (Table 2), based on the InterPro domain annotation and Gene Ontology annotation for B. olerecea and A. thaliana obtained from BRAD.

Table 2 Annotation of the B. oleracea genes in the candidate region
Full size table

Primers were designed to amplify the full length of the four candidate genes in ten resistant inbred lines (including the resistant parental line) and ten susceptible inbred lines (including the susceptible parental line) (Table 3). However, no common sequence variation was found between the R_bulk and the S_bulk for all the four genes.

Table 3 Primers used for cloning full length gene of four candidate genes
Full size table

Nevertheless, all the three genes seemed too short with only one R protein domain based on a common sense of TIR-NBS-LRR or NBS-LRR type resistance gene (R gene) except Bol037161 as a transcription factor (Table 2) [22]. Thus comparative genomics analysis was made between B. rapa genes and B. oleracea ones in the target region (Figure 2a, b and c). The results showed that the homologous gene for Bol037156 in B. rapa was a putative TIR-NBS-LRR type R gene Bra012688, or BrNL17 [GenBank: FJ842771.1] (http://blast.ncbi.nlm.nih.gov) [23], while the homologous gene for Bol037157 and Bol037158 was also a putative TIR-NBS-LRR type R gene Bra012689, or BrTNL32 [GenBank: FJ842772.1] (Figure 2b and c). To further analyze these genes, the target genomic sequence of B. olerecea containing the three genes was submitted to the FGENSH program (http://linux1.softberry.com/berry.phtml) [24] to conduct gene prediction. Because the reference line was highly susceptible in our inoculation test which might affect the prediction result, the re-prediction program used the target sequence of the resistant parental line P1.The results indicated that the re-predicted gene Bol037156, designated as re-Bol037156, was a putative TIR-NBS-LRR gene with high identity to Bra012688, while the other two genes Bol037157 and Bol037158 were re-predicted to be another putative TIR-NBS-LRR gene, named re-Bol0371578 which showed high identity to Bra012689 (Figure 2d).

Figure 2
figure 2

Re-prediction of the genes in the candidate region. (a) Reference genome sequence of B. oleracea; (b) Map** results of B. rapa homologous genes in the candidate region of B. oleracea; The exons are depicted as rectangles. (c) The original positions of the B. oleracea genes in the candidate region; (d) Re-prediction of the B. oleracea genes in the candidate region using FGENSH program; (e) The map** results of the cDNA sequence of the two re-predicted genes; (f) Reads map** results of the B. oleracea transcriptome.

Full size image

The cDNA sequences showed high identity to the re-predicted genes’ sequences (Figure 2e). The B. olerecea transcriptome reads map** results further confirmed the re-prediction result (Figure 2f). Thus, we obtained two re-predicted B. oleracea genes as the candidates, both of which were putative TIR-NBS-LRR type R genes.

To further identify which one was the most likely candidate gene for CFW resistance, primers were designed to obtain both full length DNA and cDNA of the two genes from ten resistant cabbage inbred lines including the resistant parent (R_bulk) and ten susceptible lines including the susceptible parent (S_bulk) in order to conduct sequence alignment and mutation loci analysis (Table 4). The results showed that no consistent variation was found between the R_bulk and the S_bulk for the gene re-Bol0371578. While for re-Bol037156, the cDNA from the ten R_bulk lines showed a high identity of 98.97% and no InDel was found. The cDNA from the ten S_bulk lines also showed high identity. When the R_bulk and S_bulk lines were compared, two types of InDels in the S_bulk were found; one was a 10-bp deletion in two sequences and the other was a 1-bp insertion in the other eight sequences (Figure 3a). Both InDels caused frameshifts in the open reading frames that resulted in terminating mutations in the susceptible lines (Figure 3b).

Table 4 Primers used for cloning of re-Bol037156 and re-Bol0371578 in the R_bulk and S_bulk lines
Full size table
Figure 3
figure 3

Alignment of CDS sequences of candidate genes in ten R_bulk lines and ten S_bulk lines. (a) Alignment of the CDS sequences. R_bulk: ten resistant lines, the coding sequences of ten R_bulk lines showed high identity and no InDels were found; S_bulk-1: all the susceptible lines except S1 and S3; one type of 1-bp insertion was found in S_bulk-1 to cause frame-shift and termination mutation; S_bulk-2: the susceptible lines S1 and S3; another type of 10-bp deletion was found to cause frameshift and termination mutation. (b) Frame-shift and termination mutation in the open reading frames of the susceptible lines caused by two types of InDels in the S_bulk lines.

Full size image

Thus, the newly predicted gene re-Bol037156 was considered as a candidate gene for CFW resistance, with a total length of 6.4 kb, CDS length of 4.1 kb, eight exons and seven introns (Figure 2e). The candidate gene was predicted to encode a TIR-NBS-LRR type protein containing five conserved domains including one TIR domain, one NBS domain and three LRR domains according to a Pfam sequence search result (Figure 2e) (http://pfam.sanger.ac.uk/) [25].

Validation of the FOC1gene

To verify re-Bol037156 as the candidate gene for cabbage Fusarium wilt resistance gene FOC1, primers were designed to obtain the two InDel loci sequence from 30 resistant cabbage inbred lines (R_bulk) and 30 susceptible lines (S_bulk) (Table 5). Together with the sequences of the former 20 lines, we conducted sequence alignment.

Table 5 Primers used for two InDel loci analysis in 80 lines
Full size table

The sequences from the 40 R_bulk lines showed high identity and no InDel was found at the two loci. The 40 sequences from the ten S_bulk lines also showed high identity. When the R_bulk and S_bulk lines were compared, two types of InDels in the S_bulk were found as was described above: sequences of 29 lines matched the 1-bp insertion and sequences of other 10 lines matched the 10-bp deletion. However, sequence of one line (S12) in the S_bulk matched neither of the two types of InDels. As we didn’t obtain the full length cDNA of the gene in these lines, we concluded that there might be other types of variation that caused loss of gene function. The result of 80 lines’ sequences alignment further validated that re-Bol037156 was a most likely candidate for FOC1.

Discussion

Fusarium wilt and cabbage resistance breeding

CFW is a soil-borne disease that is hard to control by traditional physical or chemical methods. Therefore, CFW has spread quickly across China from Bei**g in 2003 to the Hebei, Gansu, Shanxi, and Shaanxi provinces where it is now present. Thus, the application of resistant varieties is of great significance [6–8]. Indeed, their application has often been cited as a good example of using a plant’s own power to overcome fungal disease. Two factors account for this: firstly, the inheritance of cabbage resistance to Race 1 is single dominant [11, 12], as was also shown in our prior study [16] located the R gene on a linkage group, with two SSR markers flanking the gene at 4.6 and 1.2 cM respectively without chromosome information, and in a previous study, we located the R gene on chromosome C06, with two InDel markers flanking the gene at 1.2 and 0.6 cM [26]. Here also, the candidate region harbored two TIR-NBS-LRR genes: re-Bol037156 and re-Bol0371578.

Plant resistance gene to disease caused by Fusarium oxysporum

As far as we know, more than 70 R genes have been cloned in the past decades [27]. The main categories of these genes include the coiled coil-nucleotide binding site-leucine rich repeat (CC-NB-LRR) class, the TIR-NBS-LRR class, the kinase class, the LRR kinase class and the extracellular LRRs [22]. The most significant TIR-NBS-LRR class R gene may be the tobacco N gene conferring resistance to tobacco mosaic virus [28]. For N gene, all the three conserved domains are indispensable for a normal function [29].

Until now, the Fusarium wilt R genes have been cloned in three plant species: Arabidopsis, tomato, and melon [30–32]. The function of the melon R gene, FOM-2, has not yet been established [31]. RFO1 in Arabidopsis was reported to be a wall-associated kinase, which was neither a representative R gene nor race-specific [32]. On the other hand, the CFW R gene was race-specific and can be overcome by Race 2 [13]. In tomato, I-2 was shown to be a typical R gene, with the encoded protein belonging to the CC-NBS-LRR type [30]. A low identity of 41% was found among the three cloned R genes, indicating that although all three were R genes to Fusarium wilt, different mechanisms may be involved. We also performed whole-genome blast searches with a cut-off value of < E-10, using the RFO1, I-2 and FOM-2 sequences as the queries. No matches to these sequences were found in the B. oleracea genome. Thus, we can infer that FOC1 may be a novel R gene. The results of our study showed that the candidate gene for FOC1 was a putative TIR-NBS-LRR type R gene. The cDNA sequences of the ten resistant and ten susceptible lines revealed two types of InDel mutations in the cDNA of the susceptible lines that resulted in frameshifts and translation termination. Further sequence analysis of the two InDel loci from 80 lines confirmed the two types of InDels. Thus, we inferred that the re-modeled gene re-Bol037156 was the most likely candidate gene for FOC1. However, further work is needed to verify the function of FOC1 by transformation.

Conclusions

In this study we fine mapped a novel cabbage resistance gene to Fusarium wilt using one DH population and one F2 population, analyzed the genes that fell into the candidate region and identified the re-modeled gene re-Bol037156, a putative TIR-NBS-LRR type R gene as the candidate for the cabbage Fusarium wilt resistance gene FOC1. This work may lay the foundation for marker-assisted selection in cabbage resistance breeding as well as for further function analysis of the FOC1 gene.

Methods

Plant materials and pathogen strain

To generate F1 plants, the inbred-line 99–77 (P1, resistant parent) with immunity to CFW was crossed to the highly susceptible line 99–91 (P2, susceptible parent). An isolated microspore culture of F1 flower buds was adopted to obtain 160 DH lines [20]. At the same time, F1 plants were self-pollinated to generate about 4000 individual F2 plants. All the plant materials were provided by the Cabbage and Broccoli Research Group, Institute of Vegetables and Flowers (IVF), Chinese Academy of Agricultural Sciences (CAAS). The entire inoculation test was performed in the greenhouse of the Plant Protection Research Group at IVF, CAAS.

The strain FGL3-6 of F. oxysporum f. sp. conglutinans used in this study was isolated from diseased plants from Yanqing County, Bei**g. This strain had been shown to be Race 1 [9, 33, 38]. The Kosambi function was used to convert the recombinant value to genetic distance [39]. In this step, the gene was preliminarily mapped to an interval with two flanking markers.

The flanking markers obtained from the DH population were used to screen the F2 population for recombinants. The recombinants were further genotyped using the markers that mapped between the flanking ones. Except for the ‘a’ and ‘b’ alleles, individuals that had both the alleles from the two parents were scored as ‘h’. In this step, we obtained a narrowed candidate interval with new flanking markers.

To identify the probable genes associated with disease resistance, genes that fell into the candidate interval were analyzed based on the annotations for the B. oleracea reference chromosomes from BRAD.

Analysis of the candidate gene

To identify which one was the candidate gene, primers were designed and used to amplify the genomic DNA and cDNA of ten resistant cabbage inbred-lines (R_bulk) including the resistant parent P1 and ten susceptible lines (S_bulk) including the susceptible parent P2. The obtained sequences were aligned to detect common differences between the R_bulk and S_bulk lines. Comparative genomics analysis between B. rapa and B. oleracea in the target region was performed in order to help identify candidate genes. RT-PCR and transcriptome reads (RNA sequencing was adopted to obtain the transcriptome using RNA extracted from the roots of the seedling six days after inoculation) map** were also conducted to help confirm the candidate gene for FOC1.

Validation of the candidate gene

To further validate the candidate gene for FOC1, we chose 40 different highly resistant cabbage inbred-lines including the former ten resistant lines to make the R_bulk and 40 highly susceptible lines including the former ten susceptible lines to make the S_bulk. Primers were designed to amplify the mutation loci of the candidate gene in these 80 cabbage inbred lines. The obtained sequences were aligned to detect common variation between the R_bulk and S_bulk lines.

Availability of supporting data

The reference genome sequence of B. oleracea was retrieved from BRAD, the genomic database for B. oleracea (available at http://brassicadb.org). As one of the data owners, we confirm that we have full permission to use the genome sequence of Brassica oleracea, including downloading and using it for the purposes of this study (i.e. candidate gene identification and characterization). The use of the reference genome for these purposes was performed with the agreement of all the data owners.