Introduction

Brassica napus. L, an important oil crop, is planted to produce edible oils, biofuels and animal feed. Increasing the oil content (OC) and yield and improving seed quality have been the main breeding objectives in B. napus in recent decades (Abbadi and Leckband 2011). In addition, yellow-seeded rapeseed has many advantages, such as being highly conducive to breeding and producing high-quality seeds requiring less processing during pressing (Wang et al. 2017). To data, most yellow-seeded rapeseed has generally been selected from interspecific hybridization of Brassica species (Rahman et al. 2001; Warwick et al. 2003; Wen et al. 2012).

The seed color trait is a quantitative trait, and studies, including quantitative trait loci (QTL) mapping and candidate gene cloning, have been performed to understand the mechanisms involved in yellow-seed rapeseed. For example, a major QTL for seed coat color with a large effect was identified on chromosome C08 by Badani et al. (2006) and Yan et al. (2011) in multiple environments. In addition, a major seed coat color QTL explaining much of the phenotypic variation (PV) was identified on the homologous fragments of chromosome A09 in different studies (Liu et al. 2012a, 2013; Stein et al. 2013). Wang et al. (2017) also detected two homologous loci on C08 and A09 through genome-wide association mapping (GWAS). In addition, many QTLs with minor effects have also been detected on other chromosomes, such as C02, C05, C06, C07, A01, A04, A07, and A08 (Fu et al. 2007; Gacek et al. 2021; Wang et al. 2017).

Generally, yellow-seeded rapeseed has a higher OC, a higher protein content, fewer pigments and polyphenols, and a lower fiber content than black-seeded rapeseed (Snowdon et al. 2010; Zhou et al. 2016). However, the mechanism behind the association of seed color with other traits, including protein, oil and fiber contents, is rarely reported. Badani et al. (2006) reported that a major QTL for seed color detected on C08 was colocalized with a major QTL controlling acid detergent fiber (ADF). Similarly, a major QTL located on homologous chromosome A09 was also reported to affect seed color and fiber content simultaneously (Liu et al. 2012a, 2013; Stein et al. 2013). Due to the complexity of the B. napus genome (Chalhoub et al. 2014) and yellow-seed trait (YST), which is sensitive to harvesting time, temperature and fertilizers (Jiang et al. 2019; Niu et al. 2020), the molecular mechanisms controlling this phenotype and its connection with other important traits need to be elucidated for further utilization.

The newly developed BSR-Seq technique, which combines bulked segregant analysis (BSA) and transcriptome sequencing, not only provides global gene expression information but can also provide genetic mapping results and has become one of the most promising methods for mapping mono- or multigenic traits (Du et al. 2017; Gu et al. 2017; Liu et al. 2012b). The combination of QTL mapping and BSR-Seq could be used to identify several differentially expressed genes within QTL intervals and refine the gene candidate search; these methods have been used to comprehensively dissect the genetic mechanism controlling the important agricultural traits (Cubillos et al. 2017; Liu et al. 2016).

In this study, the major QTL for the yellow-seed trait was identified by QTL mapping and confirmed by BSR-Seq technology, and the present results are the first to show that a yellow-seed QTL not only controlled seed color and fiber content but also controlled oil content in the seeds. Sufficiently dissecting the major QTL could help to deeply understand the regulatory mechanism controlling seed color and its association with oil and fiber biosynthesis, and breed varieties with high-quality seeds and a high oil content.

Materials and methods

Plant materials and field experiments

The KN double haploid (DH) population derived from the hybridization of yellow-seeded ‘N53-2’ and black-seeded ‘KenC-8’ (Wang et al. 2013) was employed to survey the seed color and detect QTLs for the yellow-seed trait in this study. The female parent ‘N53-2’ has a high oil content (approximately 50%) and a relatively low fiber component content (approximately 3.1% lignin and 4.8% cellulose), and the male parent ‘KenC-8’ has a low oil content (approximately 40%) and relatively high fiber component content (approximately 7.0% lignin and 6.2% cellulose).

A total of 300 DH lines from the KN population and the parents were cultivated in two experimental plots under natural growing conditions in Wuhan (WH, a semi-winter-type B. napus planting area), Hubei Province, and Dali (DL, a winter-type B. napus planting area), Shaanxi Province, for several consecutive years (Chao et al. 2017; Miao et al. 2019). The field trials were implemented using a randomized block design based on Chao et al. (2019) with three replications. Open-pollinated seeds were collected at maturity from five randomly chosen plants for each line for phenotypic analysis.

Seed color measurement and trait analysis

The same amount of seeds from each line was spread on a plastic dish and scanned in real color at 600 dpi. The yellow-seeded degree (YSD) was calculated and used to evaluate the seed color phenotype according to the method described by Fu et al. (2007).

QTL mapping

The KN high-density genetic map, constructed using the Brassica 60 K SNP Array with 3207 markers, was used for QTL mapping of the yellow-seeded phenotype (Chao et al. 2017). QTL mapping was performed using WinQTLCart 2.5 software according to Chao et al. (2017). When the significant QTLs that were identified consistently across different environments had overlapping confidence intervals (CIs), they were integrated into a consensus QTL. This process was performed by BioMercator 4.2 software with default parameters (Arcade et al. 2004). The method described by Mccouch et al. (1997) was modified and applied to the nomenclature of yellow-seeded QTLs, e.g., “qSC-C05-1” represents the first significant QTL identified on linkage group C05, and “cqSC-C05-2” represents the second consensus QTL on C05 after integration by meta-analysis. Consensus QTLs that were detected in at least two trials and had a PV explained of > 10% in each trial were considered as major QTLs.

Correlation of YST with other traits and QTL colocalization analysis

The phenotype data for OC, CC and LC were surveyed and QTL mapping for these traits was performed by Chao et al. (2017) and Miao et al. (2019). Correlations between YST with other traits were analyzed using SPSS software. The QTLs detected for YST, OC, and fiber content were aligned to the “ZS11” reference genome (http://cbi.hzau.edu.cn/cgi-bin/rape/download_ext) based on closely linked markers for QTL colocalization analysis. Alignment was performed according to the method described by Cai et al. (2014).

BSR-Seq for seed color and differentially expressed gene (DEG) analysis

Sixteen extreme yellow- and black-seeded lines from the KN DH population were selected for BSR-Seq. At the seed-coloration stage (45–50 days after flowering), the seed coat from the 32 lines was stripped and immediately placed in liquid nitrogen. Subsequently, total RNA was extracted separately from 16 extreme yellow- and 16 black-seeded lines, and an equal amount of the RNA from each line was mixed into two RNA bulks of extreme yellow and black seeds for RNA sequencing. The resequencing results of ‘N53-2’ and ‘KenC-8’ have been deposited in the NCBI database with the SRA accession SRP156346 (Chao et al. 2017; Li et al. 2018).

Differentially expressed genes (DEGs) identification between the two bulks was performed using the calculation |log2 (fold change)|≥ 1 with a false discovery rate (FDR) < 0.05. The fragments per kilobase of exon model per million mapped reads (FPKM) value was calculated to determine the gene expression level. Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analyses were performed by the online software Blast2GO and KOBAS2.0 (http://kobas.cbi.pku.edu.cn/home.do), respectively.

QTL alignment to the reference genome and candidate gene analysis

QTLs identified in previous reports and in this study were aligned to the “ZS11” reference genome (http://cbi.hzau.edu.cn/cgi-bin/rape/download_ext) according to Raboanatahiry et al. (2017). The closely linked markers and their sequences acquired from corresponding papers were submitted to BLASTn (for SNP markers) or e-PCR (for SSR markers) for physical position identification and projected onto the reference genome. The genes lying in the genomic interval of the QTLs were identified as candidate genes. The interaction network was generated by the String website (https://string-db.org/) and visualized using Cytoscape software (Shannon et al. 2003).

Validating the different expression levels of DEGs using qRT-PCR

RNA samples from yellow and black seed coat bulks were used to synthesize cDNA for qRT-PCR analysis. The qRT-PCR experiment was implemented using SYBR qPCR Mix (Bio-Rad) according to the manufacturer’s specification. Three technical replicates were implemented to analyze the relative expression levels. BnActin7 was used as an internal control. The normalized expression levels were calculated according to the − ΔΔCt method reported by Livak and Schmittgen (2001). The specific primers for BnActin7 and the target genes are listed in Table S1.

Result

QTL mapping for YST

The parental line N53-2 showed a distinct seed coat color compared to the black-seeded Ken-C8, and the color difference varied considerably between the two parents in different environments (Fig. 1a and Table S2). Seed color exhibited a near-normal distribution and obvious transgressive segregation in KN DH population in each environment (Fig. 1b and Table S2), and the differences in seed color of the two parents and frequency distribution in the KN DH population in the different environments suggested that seed color was also influenced by the environments.

Fig. 1
figure 1

Phenotypic analysis of seed color in the parents and KN double haploid population. (a) The seed color difference between the two parents, and (b) the phenotypic distribution of the yellow-seed trait in the KN double haploid population

Full size image

QTL mapping was implemented for YST based on the seed color phenotypic data and KN high-density genetic map. Twenty QTLs were identified with R2 values of 2.60–41.11% in six environments (Fig. 2 and Table S3) and were distributed on chromosomes A09, A10, C01, C03, C05, and C08. Almost all of the QTLs showed a positive additive effect (AE), except for cqSC-C03. Through meta-analysis, ten consensus QTLs were obtained from twenty identified QTLs (Table 1). Four consensus QTLs distributed on A09, C05 and C08, cqSC-A09, cqSC-C05-2, cqSC-C08-1, and cqSC-C08-2, could be detected in two or more environments. The QTL cqSC-A09 was considered a major QTL because it could be detected in six environments and explain R2 of up to 41.11%. Through collinearity analysis between the KN high-density genetic map and the reference genome, cqSC-A09 corresponded to the 58.17–63.18 Mb (5.01 Mb) region at the end of chromosome A09.

Fig. 2
figure 2

Locations of QTLs mapped for the yellow-seed trait in the KN population. The 19 linkage groups are shown on the outermost circle with a scale that represents the genetic position. Six different natural environments are indicated by the background circles. The bars on the background circles represent the significant QTLs identified in the corresponding environments, and the bars near the linkage groups represent consensus QTLs

Full size image
Table 1 Consensus QTL detected for the yellow-seed trait in the KN double haploid population
Full size table

BSR analysis for seed color

Two extreme bulks were prepared using equal amounts of RNA from the seed coats of lines with extreme seed color phenotypes, yellow (YSD: 68–100) and black (YSD: 2–16). Total RNA pools from the two extreme bulks were sequenced based on next-generation high-throughput sequencing technology, and 20.41 Gb and 21.64 Gb of clean bases were obtained for the yellow and black bulks after removing the low-quality data. A total of 108.41 Mb and 113.86 Mb of clean reads from the two extreme bulks were mapped to the ‘ZS11’ reference genome and accounted for 79.66% and 78.91% of the total clean reads, respectively. The high-quality sequencing data ensured that the subsequent analysis was accurate (Table S4). The resequencing results of the two parents were used to obtain credible SNPs for calculating the SNP-index for the two extreme bulks. Associated genomic regions were surveyed by the Δ(SNP-index), which was calculated by subtracting the SNP-index values from the yellow bulk from those of the black bulk (Fig. 3). A significantly associated region (SAR) was detected within a 3.76 Mb genomic region (57.70–61.46 Mb) on chromosome A09 (Δ(SNP-index) > 0.5 and p-value < 0.01). By combining the QTL mapping results, we concluded that the locus cqSC-A09, which controls seed color, was located in the chromosomal region at 58.17–61.64 Mb (3.29 Mb) on chromosome A09.

Fig. 3
figure 3

Manhattan plot of the bulked segregant RNA-Seq (BSR) analysis. The red inverted triangles indicate the location of the significant locus

Full size image

The major YST-QTL controls both oil content and seed fiber

The correlations between YST and other traits, including OC and two fiber components, lignin content (LC) and cellulose content (CC), were analyzed using SPSS software. YST showed a significant correlation with OC, LC and CC (Table 2). The OC, LC and CC traits were evaluated for the two extreme bulks for BSR analysis, and the yellow-seed bulk showed a higher OC and lower LC and CC than the black seed bulk (Fig. 4a).

Table 2 Correlation between the yellow-seed trait (YST) and oil content (OC), YST and lignin content (LC), and YST and cellulose content (CC)
Full size table
Fig. 4
figure 4

The major QTL cqSC-A09 simultaneously controls oil content and fiber components. (a) Seed color (SC), oil content (OC), lignin content (LC) and cellulose content (CC) of two extreme bulks for bulked segregant RNA-Seq (BSR) analysis. Boxes represent the phenotypic variation of the 16 extreme lines used to construct the two extreme bulks. (b) The result of QTL scanning for SC, OC, LC, and CC on A09. QTL scanning for seed color, OC, LC, and CC in different environments is shown by the colored curves. The Δ (SNP-index) plot from the SC-BSR analysis is shown below. The associated regions with statistical significance (red dotted line, P < 0.01) and the region corresponding to the CI of the major QTL cqSC-A09 are connected to the bar in the middle representing part of A09 chromosome. The upper and lower X-axes mark the A09 linkage group and A09 chromosome, respectively

Full size image

QTL mapping for seed fibers and OC has been performed on the KN DH population (Chao et al. 2017; Miao et al. 2019). Compared to the QTLs for fiber components and OC, the major YST-QTL cqSC-A09 was colocalized with the QTLs for OC and two fiber components (Fig. 4b), cqOC-A09 for OC (controlling OC with an AE of 0.89, and an R2 of up to 11.98%), cqLC-A9-1 for LC (controlling seed lignin content with an AE of -1.12, and an R2 of up to 48.50%) and cqCC-A9-2 for CC (controlling seed cellulose content with an AE of -0.32, and an R2 of up to 16.21%), on chromosome A09. Significantly, all three QTLs, cqOC-A9-3, cqLC-A9-1 and cqCC-A9-2, were major QTLs and explained the highest PV for seed OC, LC and CC, respectively. The results from QTL mapping and BSR confirmed that cqSC-A09 was a pleiotropic QTL that increased the oil content and reduced the contents of antinutrients, flavonoids and fiber. Compared to the reported QTLs for seed color, fiber content and OC in previous studies, the major QTL cqSC-A09 was found to colocalize with the major QTL detected for seed color and fiber content on chromosome A09 (Liu et al. 2012a, 2013; Stein et al. 2013; Wang et al. 2015, 2017), a major QTL detected in the TN DH population (Jiang et al. 2014) and a significantly associated locus identified by GWAS (Wang et al. 2018) for OC (Figure S1). However, there are no reports that the locus could simultaneously control oil content, seed color and fiber content.

DEG analysis between the yellow and black seed coat

To identify potential candidates controlling seed color and dissect the mechanism controlling the correlation between OC and seed color, gene expression analysis of the two bulks was performed based on RNA-Seq data from the BSR analysis. DEGs were determined by their |log2 (fold change)| value, and then the differential gene expression from RNA-Seq analysis was validated by qRT-PCR using six randomly selected DEGs (Figure S2). In total, 5643 DEGs were identified in the black bulk vs. yellow bulk, including 889 upregulated and 4754 downregulated genes (Fig. 5a). In addition, 571 differentially expressed transcription factors were detected, including numerous MYB, NAC, bHLH and WRKY family members (Fig. 5b).

Fig. 5
figure 5

Differentially expressed gene (DEG) analysis between the yellow and black seed extreme bulks. (a) Position distribution of DEGs. (b) The categories of TFs show expression differences. (c) and (d) show the top 20 significantly enriched GO biologic processes and KEGG pathways, respectively

Full size image

GO and KEGG analyses of the DEGs were used to explore the potential metabolic mechanism leading to the difference in seed coat color. The changed biological processes were identified by subjecting DEGs to GO enrichment analysis. Secondary metabolic process, phenylpropanoid biosynthetic process, and phenylpropanoid metabolic process were included in the top 20 significantly enriched GO in terms of the biological process category (Fig. 5c), secondary metabolite biosynthetic process. In addition, the lipid metabolic process and fatty acid metabolic process involved in oil accumulation were also significantly enriched. To further determine the metabolic pathways associated with seed coat color, KEGG enrichment analysis was performed (Fig. 5d). Biosynthesis of secondary metabolites, flavonoid biosynthesis, phenylpropanoid biosynthesis, and phenylalanine metabolism were found in the top 20 significantly enriched pathways. In addition, cutin, suberine and wax biosynthesis, fatty acid biosynthesis, fatty acid metabolism, and α-linolenic acid metabolism pathways were significantly enriched. The results indicated that flavonoid metabolism was associated with lipid metabolism in the seed coat.

Candidate gene identification

The 3.29 Mb mapping interval on chromosome A09 contained 648 genes, 55 of which carried a significant SNP associated with seed color within their coding region (Table S5), and 65 genes showed expression differences between yellow and black bulks (Table S6), including seven transcription factors (BnaA09. CMTA3, BnaA09. JAZ5, BnaA09. LRL1, BnaA09. MYB47, BnaA09G0632400ZS, BnaA09. NAC038, and BnaA09. JAZ1). In addition, ten genes with significantly associated SNPs showed a significant expression difference (Table S6). The TFs BnaA09. MYB47 and BnaA09.LRL1 belongs to the R2R3-MYB and bHLH transcript factor families, respectively, which play important roles in flavonoid biosynthesis, thus, these TFs may be important candidates. BnaA09. JAZ1, orthologous to the Arabidopsis JAZ1 gene involved in jasmonate signaling (Huang et al. 2018), and carried three significantly associated SNPs (including a nonsynonymous coding mutation) and showed a significant expression difference.

To further and more comprehensively screen for key candidate genes, an interaction network was constructed based on the candidate genes carrying significantly associated SNPs and/or showing significant expression differences in the 3.29 Mb mapping interval on chromosome A09 and all differentially expressed genes enriched in the terms lipid metabolism, flavonoid biosynthesis and phenylpropanoid biosynthesis process (Fig. 6). Sixteen genes acted as a bridge to connect lipid metabolism and flavonoid and phenylpropanoid pathways. In the network, the JAZ1 gene affects lipid metabolism and flavonoid metabolism by interacting with the genes involved in the corresponding biological processes. BnaA09. JAZ1 was located near the peak of cqSC-A09, carried three significantly associated SNPs (including a nonsynonymous coding mutation) and showed a significant expression difference between the extreme yellow and black bulks (Table S5 and Table S6); hence, it was considered a valuable candidate. BnaA09. GH3.3, orthologous to the Arabidopsis GH3.3 gene that encodes an IAA-amido synthase, was an important candidate because of its connection to both lipid metabolism processes and flavonoid metabolism, and the BSR results showed that BnaA09. GH3.3 has a significant expression difference and three significant SNPs associated with seed color (Table S5 and Table S6). In addition, BnaA09. LOX3, orthologous to the Arabidopsis gene LOX3, which encodes a lipoxygenase catalyzing the oxygenation of fatty acids, interacts with phytochrome-interacting factor 3 (PIF3), which positively regulates anthocyanin metabolism (Shin et al. 2007). BnaA09. LOX3 was located near the peak of cqSC-A09 and showed significantly different expression between the two bulks; it was also considered an important candidate gene.

Fig. 6
figure 6

The interaction network constructed based on differentially expressed genes (DEGs) and candidate genes. DEGs involved in lipid metabolism, flavonoid biosynthesis and phenylpropanoid pathway are grouped into circles, and TFs are represented by an oxblood hexagon. The candidate genes within the interval of cqSC-A09 are represented by ellipses and are shown in the middle; those with differential expression in the two bulks are painted orange, and those with significantly associated SNPs are edged in blue

Full size image

Discussion

Seed color is an important agronomic trait related to seed quality because pigment deposits interfere with industrial processing in B. napus. In recent decades, many attempts have been made to discover genes controlling yellow seeds and conduct parallel genetic research for breeding in rapeseed. However, currently, few genes controlling seed color have been cloned and appropriately applied to production, which might be because the yellow-seed trait is influenced by several factors with minor effects and environment × genotype interactions (G × E). In this study, six of the 10 QTLs detected for YST were identified in only a single environment, which indicated that they could influence seed color by interacting with the environment to a certain extent. Three QTLs, cqSC-C08-1 (detected in two environments), cqSC-C05-2 and cqSC-C08-2 (detected in three environments), showed relatively stable effects on seed color, which suggested that these QTLs were affected by the environment to lesser extent. Although it is difficult to utilize and fine-map environment-specific QTLs under G × E, such findings would be helpful to dissect the complex genetic mechanism controlling seed color in rapeseed.

Although influenced by the environment, YST has been reported to be controlled mainly by a few major QTLs on chromosome A09 (Liu et al. 2012a, 2013; Stein et al. 2013; Wang et al. 2017) and C08 (Badani et al. 2006; Wang et al. 2017). In this study, the QTL cqSC-C08-1 identified on C08 colocalized with the major QTL with an R2 of 51.6% that Badani et al. (2006) identified by linkage mapping and the significantly associated locus Wang et al. (2017) identified by GWAS. The major QTL cqSC-A09 with the highest R2 (41.11%) in this study was found to colocalize with the major QTL on chromosome A09 detected in several other studies (Liu et al. 2012a, 2013; Stein et al. 2013; Wang et al. 2017) (Figure S1). Notably, the significant region containing the major yellow-seed QTL on C08 was homologous to the region containing the major YST-QTL on A09, as previously demonstrated (Liu et al. 2012a; Wang et al. 2017). The major QTL cqSC-A09, which is more stable and has a large effect, is of great value to yellow-seeded rapeseed breeding by molecular marker-assisted selection and has attracted much attention.

Yellow-seeded rapeseed has a lower pigment content, lower fiber content, and higher oil content than black-seeded varieties, which has been widely recognized. In this study, we found the same correlation between the seed color and OC and identified a major QTL, cqSC-A09, by QTL mapping and BSR that not only controls YST with positive AE but also affects oil content with positive AE and a relatively large R2 and fiber (cellulose and lignin) content with negative AE and a large R2. These results indicated that the advantageous allele of the major QTL reduced pigment accumulation, increased oil content and decreased the fiber content in seeds. In previous studies, cqSC-A09 was found to be colocalized with a seed fiber QTL (Badani et al. 2006; Liu et al. 2012a; Stein et al. 2013), but was never reported to be related to OC, and we first uncovered that cqSC-A09 simultaneously controlled seed color and fiber and oil contents herein. Behnke et al. (2018) reported a major pleiotropic QTL on chromosome C05 that reduced acid detergent lignin content and increased oil and protein contents in rapeseed but did not affect seed color. The discovery that cqSC-A09 simultaneously controlled YST and OC partially explained why yellow-seed color could be used as a visible phenotypic marker for high OC selection. In addition, cqSC-A09 colocalized with a major QTL detected in the TN DH population (Jiang et al. 2014) and a significantly associated locus identified by GWAS (Wang et al. 2018) (Figure S1). These findings suggested that the yellow-seed QTL cqSC-A09 is stable and reliable in affecting oil and fiber contents, and would have great application value in developing high OC varieties with high seed quality.

The seed coat color of rapeseed is mainly determined by the content of the phenolic compounds cyanidin and procyanidins (Akhov et al. 2009; Auger et al. 2010), which are synthesized as end products of the flavonoid biosynthesis pathway. Phenolic compounds were correlated with lignin through the phenylpropanoid metabolism, flavonoid biosynthesis and phenylalanine pathways via common substrates, such as coumaroyl CoA and caffeoyl CoA (Lepiniec et al. 2006; Mittasch et al. 2013). However, the reasons for the correlation between flavonoid and oil contents remain unclear, although several studies have provided many clues. For example, Xuan et al. (2018) reported that AtTT4 regulates carbon source redistribution, which affects fatty acid biosynthesis by mediating flavonoids to regulate WRI1 and auxin transport (Xuan et al., 2018). Additionally, the key regulatory factors that regulate flavonoid biosynthetic pathways, such as TT2 and TT8, inhibit the accumulation of fatty acids by targeting TFs related to lipid metabolism in seeds (such as FUS3, LEC1 and LEC2) (Chen et al. 2014; Wang et al. 2014). In addition, the B. napus mutants tt2 and tt8 created by CRISPR/Cas9 recently showed yellow-seed color and higher OC (Xie et al. 2020; Zhai et al. 2020). The pleiotropic major QTL cqSC-A09 identified herein may provide new insights into the correlations between flavonoid and oil metabolism. Through the analysis of DEGs identified between extreme yellow and black seed bulks from BSR analyses, the enrichment of phenylpropanoid metabolic, phenylpropanoid biosynthetic and lipid metabolic processes indicated that flavonoid pathways and oil biosynthesis pathways were simultaneously regulated. A mature seed comprises the seed coat, endosperm and embryo in Arabidopsis and B. napus. The endosperm supplies nutrients to the embryo and degrades into a monolayer in mature seeds, and the embryo is the major organ for the synthesis and accumulation of storage compounds (Buer and Djordjevic 2009; Stone et al. 2008). Unlike the embryo and endosperm, which originated from both parents, the seed coat has a maternal origin (Debeaujon et al. 2000). Sugars are first delivered to the maternal seed coat via the funicular phloem, which is symplistically connected to the outer integument and then reaches the embryo (Chen et al. 2015; Xuan et al. 2018). Rhamnose and glucose are substrates for flavonoid synthesis, and rhamnose is also a substrate for mucilage, which is mainly composed of cell wall polysaccharides and is deposited in the outer integument layer attached to the seed coat (Western et al. 2000). More sugars are intercepted to synthesize flavonols and PAs in the seed coats of dark-colored seeds (starch and sucrose metabolism and flavonoid related pathways were enriched for DEGs between the yellow and black seed coat bulks, as shown in Fig. 5d), reducing the flow to the embryo for matter storage, which might be one of the reasons for the correlation between the seed color and OC.

A knockout mutation in the CCR1 gene that participated in the lignin biosynthesis was considered to explain the major QTL for seed color and lignin content (Liu et al. 2012a); however, Stein et al. (2013) thought that CAD2 was also the cause of the major QTL for both fiber content and seed color. In this study, CCR1 and CAD2 were located within the interval of QTL cqSC-A09, but no sequence variation or expression difference was found between the two parents. In addition, there is no evidence that the variation in CCR1 and CAD2 resulted in changes in the oil and cellulose contents. Therefore, there may be other causal genes underlying the major QTL. In this study, some important candidates were identified by combining sequence variation annotation, expression differences and an interaction network constructed with enriched DEGs related to lipid, flavonoid and phenylpropanoid metabolism and genes within QTL intervals. BnaA09. JAZ1, BnaA09. LOX3 and BnaA09. GH3.3 that was identified to affect lipid and flavonoid metabolism were considered as the most important candidates underlying the QTL cqSC-A09. However, the reason for the link between seed color and oil and fiber contents needs to be determined by further fine mapping to clone the causal gene, which could promote the locus utility for breeding new varieties.

Conclusion

Yellow-seeded B. napus generally has a high oil content and high seed quality with few pigments and a low fiber content; however, the genetic mechanism behind this trait remains unclear. A major yellow-seed QTL on chromosome A09 was identified by QTL mapping and BSR in this study. This QTL not only controlled seed color but also affected seed oil and fiber contents with a large effect, and the advantageous allele increased the oil content and reduced the pigment and fiber content. This is the first QTL that was reported to control seed color, oil content and fiber content simultaneously with a large effect, and it may be of great application value for breeding high oil varieties with high seed quality. The related candidate genes were identified by combining sequence variation annotation, expression differences and an interaction network. This study provides a theoretical basis for the application of yellow-seeded rapeseed resources for breeding high oil varieties.