Phenotypic variation and correlation analysis
We measure six salt tolerance traits (STT), four alkaline tolerance traits (ATT) and five yield-related traits from the RIL lines and the two parents. All traits performed significant differences between line 1423 and line 2205 in all environments. Compared with other traits, PH showed the lowest coefficient of variance (CV), while SYP exhibited the highest CV (Table1). All traits exhibited continuous normal or near-normal distributions (Fig. S1). Correlation analysis among all traits was carried out. STR was usually described as an effective salt stress indicator. For STT, STR was significantly negatively correlated with all physiological and morphological indicators. For ATT, STR was only significantly negatively correlated with SPAD and RFW. SPAD was significantly positively correlated with RFW and RL (Fig. 1). A correlation was observed between salt tolerance traits and alkaline tolerance traits. For instance, aSPAD was significantly positively correlated with sRL, sRFW, sSFW, and sSDW. These results suggest that alkaline and salt tolerance in B. napus might be linked to these traits and different. Seed yield (SYP) showed significant positive correlations with all yield-associated traits across different years, and especially for NSP and SS with a correlation coefficient of 0.80 and 0.62, respectively. Interestingly, SYP was significantly positively correlated with sSPAD. However, sSPAD was significantly positively correlated with all salt-related traits. Taken together, these results demonstrate that there is a complex association between seed yield and other traits, either directly or indirectly, including salt and alkaline tolerance traits and yield-related traits.
Construction of the linkage map based on SLAF-Seq
In total, 33.35 Gb of raw reads consisting of 166.75 Mb paired-end reads were generated from Illumina sequencing of the SLAF libraries. The average percentage of Q30 bases (bases with a quality score of 30, indicating a 1% chance of an error and thus a 99% confidence level) was 94.83%, and the GC content was 36.43%. A total of 73.82 and 78.45 million reads were generated for the salt-tolerant parent 2205 and the salt-sensitive parent 1423, respectively, while 2.03 million reads were obtained for the 82 RILs. A total of 2,532,319 SNPs were detected, among which, 1,516,733 polymorphic SNPs were successfully sorted into eight segregation patterns (ab × cd, ef × eg, ab × cc, cc × ab, hk × hk, lm × ll, nn × np, and aa × bb), and 673,113 SNPs that belong to the aa × bb segregation pattern were used in linkage analysis. After a three-step filtering process (see Methods), 5,250 SNPs were used for the genetic map construction. In total, 4,159 SNP markers were grouped into 19 linkage groups (LGs) compared with the B. napus reference genome (Fig. S2). The total genetic distance was 1736.59 cM with a mean marker distance of 0.42 cM between adjacent markers. The lengths of LGs ranged from 46.14 cM (LG8) to 136.79 cM (LG9). The number of markers in each LG ranged from 84 to 327, with an average of 219 markers per LG (Table S1).
QTLs for salt-alkali tolerance and yield-related traits
A total of 65 QTLs were detected for salt-alkali tolerance and yield-related traits with a total phenotypic variance explained (PVE) of 7.61–27.84% and a LOD of 2.72–8.40 (Table S2 and Fig. 2). Notably, most QTLs for salt-alkali tolerance traits and each of QTL for SS and NSP showed a PVE of more than 10%. The QTLs for yield-related traits were mainly distributed on chromosomes A01, A09, C08 and C09 (4-6 each), while the QTLs for salt-alkali tolerance traits mapped on A09, A10 and C04. By meta-analysis, 42 of 65 identified QTLs were integrated into 18 unique QTLs controlling two to four traits (Table S3). Six unique QTLs were identified for salt-alkali tolerance traits, of which four formed two adjacent QTL clusters with a genetic distance of 1cM on A10, and the other two controlling SDWs, SFWs and RFWs were located on A09 and C04, respectively, with an average PVE of 15%. Seven unique QTLs were identified for seed yield and yield-related traits. For example, QTL uqA5 and uqC6 for SYP and LMI, uqC9-2 for SYP and NSP, uqA9-2 for SYP, NSP, PH and LMI and uqC3-2 for LMI and PH were identified with positive additive effects, which was in accordance with the significant positive correlations among these traits. Interestingly, five unique QTLs were detected for both salt-alkali tolerance and yield-related traits. For instance, QTL qSYP19-A6-1 overlapped with QTL qRFWa-A6 on A06. QTL uqC8 controlling PH, LMI and RFW and uqC9-1 controlling SS and RL showed consistent additive effects.
Comparative analysis of salt-alkali tolerance and yield-related traits of present and previous QTLs
Among the 18 QTLs for salt tolerance traits, ten salt-related QTLs contained 245 significantly associated SNPs identified previously, of which four were detected for the same traits (Table S4). Specifically, the QTLs qSDWs-C3 and uqC4 for SDW, qRLs-C9 for RL, and uqA10-3 for SFW included nine, one and two SNPs, respectively, indicating their high reliability. In a previous study, some QTLs were detected using the F2 of the same parents, among which five QTLs (qSH12-a, qRDW19-c, qEC8-c, qSOD14-b and qLDW14-b) and one QTL qSH4-b had overlapping CIs with the QTLs qRFWs-C8 and qRLs-C9 in the present study, respectively, indicating the complexity of salt tolerance traits. In addition, nine alkaline-related QTLs contained 163 significantly salt-associated SNPs identified previously, of which qRLa-A10-1 and qRLa-A10-2 for RL possessed five SNPs for the same traits. For yield-related traits, ten of 35 QTLs were colocalized with QTLs identified in the previous study (Table S5). One QTLs for SYP and NSP were identical to qSYP21-A9 and qNSP20-A1 in the present study, respectively. qSS20-A1, qSS20-C1 and qSS20-C9-1 (present study) were identical to 18 QTLs for SS. qLMI19-C6 (present study) was identical to three QTLs identified previously. For PH, four QTLs (qPH19-C8-1, qPH20-C8, qPH19-C4 and qPH19-C8-2) were identical to significantly PH-associated InDels identified by GWAS.
QTLs for salt-alkali tolerance traits co-localized with QTLs for yield-related traits
As shown in Table S3, six unique QTLs for salt-alkali tolerance traits were located on A09, A10 and C04, especially A10. To explore the connection between salt-alkali tolerance traits and yield-related traits, these six unique QTLs plus qRLa-A10-1 and qRLa-A10-2 identified in the current study were compared with the QTLs for yield-related traits previously identified. Since flowering time is significantly associated with plant height, it was added as a yield-related trait. We found that seven QTLs, composed of three identified QTLs and four unique QTLs, overlapped with the QTLs for the six yield-related traits previously identified except for uqC4 (Table 2). qRLa-A10-1 and qRLa-A10-2 for RL overlapped with one SNP for SYP, one SNP for SS, one SNP for LMI, one QTL for LMI and one QTL for FT identified previously. uqA10-1 and uqA10-2 for SPAD and RFW overlapped with one QTL for SYP, one QTL for LMI, two QTLs for PH, one SNP for PH and twelve QTLs for FT identified previously. qRFWs-A9 and uqA10-3 for RFW, SDW and SFW overlapped with one QTL for SS, four QTLs for PH on A09, and one SNP for LMI, ten SNPs for FT and eleven QTLs for FT on A10 identified previously, respectively. Moreover, uqA10-4 for STR and SDW overlapped with five QTLs for FT identified previously. Taken together, seven QTLs that co-control salt-alkali tolerance traits and yield-related traits were identified and named as cQTLs, suggesting that A09 and A10 are important loci for controlling both salt-alkali tolerance traits and yield-related traits.
Identification of differentially expressed genes by transcriptome sequencing
To investigate the potential molecular mechanism underlying the differences in salt tolerance between lines 2205 and 1423 and to explore salt-related candidate genes, RNA-seq analysis was performed using RNA extracted from the roots of the 1423 and 2205 plants under control conditions (mock treated) and salt treatments. We first tested the root length of line 1423 and line 2205 in response to the 200 mM NaCl treatment. Under salt stress, the root length of line 2205 is always higher than that of line 1423 and significant difference was observed at 3h and 24h between line 2205 and line 1423 (Fig. S3). Based on the results, we used the root tissues treated with NaCl for 3h and 24h in the RNA-seq experiment. After trimming off the adapter sequences and removing the low-quality reads, we obtained 484,217,730 and 467,967,990 clean reads for line 2205 and line 1423, respectively, with an average read length of 90 bp and a Q20 percentage (percentage of sequences with sequencing error rates lower than 1%) over 97%. In total, 433,216,368 (89.47 % of the clean reads) and 424,400,341(90.69% of the clean reads) reads were mapped to the B. napus reference genome for line 2205 and line 1423, respectively, and 93,990 transcripts were identified after comparison. Sequencing and assembly statistics are summarized in Table S6. Both Pearson’s correlation and PCA showed a high correlation among the replicas, with the exception of sample T-3h1 (round markers in Fig. S4), which was identified as an outlier and excluded from downstream differentially expressed genes (DEGs) analysis (Fig. S4 and Table S7).
Global expression analysis showed that 15,274 DEGs named as S_SR were found in line 1423 and 12,845 DEGs named as T_SR were found in line 2205 after 3h and 24 h salt stresses (Fig. 3a, b). In total, 19,311 DEGs were identified under salt stress and named as "salt-inducible genes", among which, 6,466 and 4,037 genes showed unique differentially expression in lines 1423 and 2205, respectively. The NaCl treatment led to a less transcriptomic change in line 2205 compared with that in line 1423, indicating that line1423 is more sensitive to salt stress than line 2205 at the transcriptional level. A total of 12,815 DEGs were regulated at 0 h, 3 h and 24 h between lines 1423 and 2205 (Fig. 3c). To further screen salt tolerant genes, 2,909 genes that exhibited unique differential expression between lines 1423 and 2205 under normal conditions were removed, and the remaining 10,423 DEGs were named as V_SR. Among these three groups (S_SR, T_SR and V_SR), a total of 24,852 DEGs were identified and 4,882 DEGs were detected in all three groups, while 4,783 and 2,809 DEGs were only detected in S_SR and T_SR, respectively (Fig. 3d). The expression levels of eight randomly selected genes were measured by using RT-qPCR, which showed consistent results with RNA-seq, suggesting that the RNA-seq data were reliable (Fig. S5).
Functional classification of differentially expressed genes
In order to explore the salt tolerance mechanism of the two parents, three groups of the differentially expressed genes (DEGs), S_SR_ unique (4783), T_SR_ unique (2809) and STV_SR (4882), were functionally annotated using gene ontologies (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) (Fig. 4). GO enrichment analysis revealed that both T_SR_ unique (2809) and STV_SR (4882) DEGs were uniquely involved in secondary metabolic process and response to light stimulus (Fig. 4b, c), while the S_SR_ unique (4783) DEGs were uniquely involved in response to nitrogen compound, response to jasmonic acid and response to fatty acid (Fig. 4a). Moreover, DEGs identified in S_SR_ unique (4783) were enriched with significantly higher GO terms compared with those in T_SR_ unique (2809) and STV_SR (4882). KEGG enrichment analysis also provided similar enrichment results to the GO enrichment. Analysis of the KEGG pathways showed that metabolism pathways, especially for phenylpropanoid biosynthesis, flavonoid biosynthesis and carbohydrate metabolism were uniquely enriched among the T_SR_ unique (2809) and STV_SR (4882) DEGs (Fig. 4e, f), while plant hormone signal transduction, glutathione metabolism, glycosaminoglycan binding proteins and amino acid metabolism were uniquely enriched among the S_SR_ unique (4783) DEGs (Fig. 4d). These results indicated that the two parents had different molecular mechanisms in response to salt stress, and line 1423 was more sensitive to salt stress and elicited more responses in the body. The similar enrichment results of the two groups (T_SR_ unique and STV_SR) also showed that most of the key salt tolerance related genes came from line 2205.
Candidate genes for controlling both salt-alkali tolerance and yield
To further reveal candidate genes controlling both salt-alkali tolerance and yield, these 24,852 DEGs were integrated with seven cQTLs and 1081 genes were found in these QTLs regions. A total of 390 DEGs response to salt stress were co-localized with the cQTLs (Fig. 5a). Since the cQTLs contained both alkaline and salt tolerance traits, we also detected the DEGs using the transcriptome data of the two parents under alkaline stress (unpublished). A total of 400 DEGs were located with the cQTLs in response to alkaline stress (Fig. 5b). Of these, 99 and 162 DEGs were not only "salt-inducible genes" (S_SR/T_SR), but also differentially expressed between lines 1423 and 2205 (V_SR), respectively. To further narrow down the number of candidate genes, the expression profiles of these 99 and 162 DEGs were performed (Fig. S6). The genes that were continuously induced in 1423 or 2205 and differentially expressed between the two parents were selected for further analysis, and, 28 (28S) and 75 (75A) genes were obtained respectively (Fig. 5c). The heat maps of the relative expression profiles are shown in Fig. 5d and 5e. A total of 100 genes were induced by salt or alkaline stress, of which three genes were in response to both salt and alkaline stress (Fig. 5c). Among these 100 genes, 85 were homologous to 83 Arabidopsis genes. GO and KEGG analyses were performed for these 83 Arabidopsis genes, and the result revealed that 64 DEGs were annotated to GO terms and 31 DEGs were mapped to biological pathways, especially the metabolism pathways (Table S8). Based on the functional annotation, 13 genes were screened as the candidate genes (Table 3), of which, four were involved in response to salt, cold and heat abiotic stresses and six genes were annotated with regulation of cell growth, pollen tube growth and gibberellin biosynthetic process. Notably, three genes (BnaA10g16340D, BnaA10g20120D and BnaA10g25000D) were annotated with both abiotic stress and plant organ development, which would serve as the most promising candidates.