Phenotypic characteristics of the dwarf mutant bnd2
A B. napus mutant, bnd2, was isolated and screened from the EMS-mutagenized seeds of the cultivar “2B” (wild type, WT) [19]. At the seedling stage, bnd2 showed reduced hypocotyl length and shorter petiole leaves compared to WT, respectively (Fig. 1a-e). At the flowering stage, the bnd2 mutant exhibited an extremely dwarf and compact stature, and the flowering period of bnd2 was slightly longer than that of WT (Fig. 1f and g). At maturity stage, the plant height of bnd2 was 100.65 ± 8.09 cm (n = 10), which is only 59.8% of that of WT (168.2 ± 7.61 cm, n = 10) (Fig. 1h and i, Additional file 1: Table S1). In addition, the first branch height, internode length, internode number and main inflorescence length of bnd2 were 41%, 76.7%, 69%, and 85.2% of that of WT, respectively. These results suggested that the dwarf traits were associated with lower position of first branch, shorter internode length, less internode number and reduced main inflorescence length (Fig. 1h-j, Additional file 1: Table S1). Accordingly, bnd2 produced fewer yield per plant (YPP) (48.4% of WT) due to shorter silique length (83.1% of WT), fewer seeds per silique (SPS) (92.1% of WT) and less thousand-seed weight (TSW) (90% of WT) compared to WT, although similar siliques per plant (SPP) were observed both in bnd2 and WT (Fig. 1k-n, Additional file 1: Table S1, Additional file 2: Figure S1).
Cell elongation and expansion in stem is decreased in bnd2
To look into the underlying cellular basis of the dwarf phenotype in bnd2, we performed paraffin section observation on the cross section and longitudinal section of the stem of bnd2 and WT at the early bolting stage. As shown in Fig. 2, the parenchyma cells of bnd2 were closely arranged with irregular shapes and different sizes compared to WT (Fig. 2a and b). The cell area and length were significantly reduced in both cross and longitudinal sections in bnd2 plant (Fig. 2c-e). Indeed, cell area in both cross and longitudinal sections were decreased by more than 48.2% and 50.5%, and cell length were decreased by more than 31.6% and 16.6%, respectively. These results suggest that the reduction of parenchyma cell area and length in stem were likely to be the main causes for the dwarfism of the mutant bnd2.
Inheritance of the dwarf phenotype in the mutant bnd2
To analyze the inheritance of the dwarf mutant bnd2, bnd2 was used to make crosses with its original parent WT and another commercial cultivar L329. The resulting heterozygous BC1F1 plants (bnd2/WT) displayed intermediate plant height between that of WT and the mid-parent value, suggesting that the allele BND2 is semi-dominant to the allele bnd2 (Fig. 3a-c). In addition, according to the plant height of BC1F2 generation crossed by 2B and bnd2, the 236 BC1F2 individuals could be classified into two groups: one has the dwarf phenotype of bnd2 (dwarf plants, n = 49) and another has high plant height similar or close to WT (tall plants, n = 187) group. The BC1F2 generation was in line with an expected Mendelian inheritance ratio of 1:3 (dwarf plants: tall plants, χ2 = 2.04 < χ20.05,1 = 3.84) (Fig. 3d). Another F2 population was conducted from the cross between bnd2 and another commercial cultivar L329 which possessed a normal plant height of ~ 159 cm. There were 75 plants with dwarf phenotype and 188 plants with plant height similar or close to that of L329 in F2 population, also showing a Mendelian segregation ratio of 3:1 (tall plants: dwarf plants, χ2 = 1.46 < χ20.05,1 = 3.84) (Additional file 3: Figure S2). Taken together, these results suggested that the dwarfism phenotype of bnd2 was controlled by a single recessive gene.
Genetic mapping of the dwarf mutant bnd2 by BSA-seq
To map the gene conferring for bnd2, the F2:3 population derived from cross between bnd2 and L329 was used to perform bulked segregant analysis (BSA) resequencing. In the F2:3 population (n = 157), 25 extremely dwarf and 23 extremely tall homozygous lines were selected to make two bulks, such as a short bulk, and a high bulk. Through sequencing in two bulks and their parents, 105,361,953, 89,416,611, 99,097,181 and 109,214,266 clean reads were harvested for the L329 parent, the mutant bnd2 parent, the high bulk and the short bulk, respectively (Additional file 4: Table S2). The sequencing data showed that the percentage of bases with a quality score of more than 30 in two pools and two parents (Q30) reached more than 92.99%, and Q20 reached more than 97.78% (Additional file 4: Table S2). In addition, the average GC content was 37.35% and the average genome coverage was 74.57% with an average coverage depth of 21.66 X (Additional file 4: Table S2). Therefore, we consider that the quality of the sequencing data is consistent with expectations and can be used for further analysis. According to aligning with the ‘Darmor-bzh’ reference genome [26], 1,157,351 polymorphisms (containing 948,896 single nucleotide polymorphisms (SNPs) and 208,455 insertions/deletions (InDels)) were identified between the two pools. The G’ value and SNP-index were calculated from the short bulk and the high bulk, the △(SNP-index) was drawn based on the physical positions of the reference genome (Fig. 4a and b). And only one significant △(SNP-index) peak was identified and located into the 4.31 Mb region from 13.77 Mb to 18.08 Mb on chromosome A08 (Fig. 4c), suggesting that it was the candidate locus harboring the BND2 gene.
Fine mapping and candidate gene analysis
To fine mapping the BND2 locus, six insertion/deletion (InDel) markers (ID1421, ID1470, ID1482, ID1530, ID1656, ID1667) were developed from the 4.31-Mb region harboring bnd2 based on the BSA-seq result. Then, in the F2:3 population (bnd2/L329) with 543 lines, the six markers were used to genotype 107 recessive lines with dwarf statue, as well as two controls, such as 25 wild type lines with tall stature and 25 heterozygous lines with segregation in plant height (Fig. 5). According to fine genotypes of these lines, bnd2 was furtherly fine mapped into the 1.26-Mb interval flanked by two InDel markers ID1530 and ID1656 (Fig. 6a). In order to further narrow the candidate interval, six pairs of new polymorphic markers were developed in the region of bnd2, such as the single nucleotide polymorphism (SNP) markers SNP1540, SNP1552, SNP1553, SNP1557 and SNP1562 and the InDel marker ID1576 (Fig. 6b). Subsequently, BND2 was narrowed down to an interval from 15.62 Mb to 15.76 Mb, and the physical distance was 140.0 Kb (Fig. 6b). After fine mapping and the annotation information of reference genome ‘Darmor-bzh’, there are 27 genes in the 140 kb candidate interval, 14 of which were not cloned or had unknown functions (Fig. 6c). By analyzing the annotation results of all mutations in the candidate interval, one SNP occurred in the candidate gene, BnaA08g20960D (Fig. 6d), which encodes an Inositol-pentakisphosphate 2-kinase family protein, where a single nucleotide substitution from C to T occurs in the fifth intron region. Therefore, we take this gene as a key candidate gene.
The potential application of bnd2 in hybrid rapeseed breeding
Due to the low yield of bnd2, it cannot be used in inbreed rapeseed breeding. In order to test its potential application in hybrid breeding, we crossed the bnd2 mutant (bnd2/bnd2) with a commercial cultivar L329 (BND2/BND2) to get their hybrid line F1 (BND2/bnd2). The plant height of the F1 hybrid was similar to L329 (Fig. 7a and b, Additional file 5: Table S3). While the yield per plant (YPP) of F1 was significantly higher than both of bnd2 and L329, showing an increase of 32.7% than L329 due to more seeds per silique (SPS), and three times as much as bnd2 (Fig. 7c, Additional file 5: Table S3). This result suggested that the introduce of bnd2 in the hybrid line can produce a hybrid of no increase on plant height but higher grain yield due to the semi-dominant effect of BND2 to bnd2 and the heterosis between two lines.