Identification of SNP loci and candidate genes related to four important fatty acid composition in Brassica napus using genome wide association study

Rapeseed oil (canola, Brassica napus L.) is an important healthy vegetable oil throughout the world, the nutritional and economical value of which largely depends on its seed fatty acid composition. In this study, based on 201,187 SNP markers developed from the SLAF-seq (specific locus amplified fragment sequencing), a genome wide association study of four important fatty acid content traits (erucic acid, oleic acid, linoleic acid and linolenic acid) in a panel of 300 inbred lines of rapeseed in two environments (JXAU and JXRIS) was carried out. A total of 148 SNP loci significantly associated with these traits were detected by MLM model analysis respectively, and 30 SNP loci on A08 and C03 chromosomes were detected in three traits of erucic acid, oleic acid and linoleic acid contents simultaneously. Furthermore, 108 highly favorable alleles for increasing oleic acid and linoleic acid content, also for decreasing erucic acid content simultaneously were observed. By a basic local alignment search tool (BLAST) search with in a distance of 100 Kb around these significantly SNP-trait associations, we identified 20 orthologs of the functional candidate genes related to fatty acid biosynthesis, including the known vital fatty acid biosynthesis genes of BnaA.FAE1 and BnaC. FAE1 on the A08 and C03 chromosomes, and other potential candidate genes involving in the fatty acid biosynthesis pathway, such as the orthologs genes of FAD2, LACS09, KCS17, CER4, TT16 and ACBP5. This study lays a basis for uncovering the genetic variations and the improvement of fatty acid composition in B. napus.


Introduction
Canola (rapeseed; Brassica napus, AACC genome, 2n = 38) is the world's second largest oil producing crop after soybean (Glycine max L. Merrill), cultivated in temperate regions of many countries worldwide, and accounts for 14% of all edible vegetable oil production [1]. The nutritional and healthy oil qualities of canola seed are mostly determined by the fatty acid compositions [2]. Canola oil is rich in unsaturated fatty acids compared with other vegetable oils, PLOS  comprised primarily of monounsaturated oleic acid and polyunsaturated linoleic and linolenic acid having a 2:1 optimal ratio [3,4]. Of these the oleic acid (C18:1) and linoleic acid (C18:2) are considered to be healthy and nutritious. However, the three double-bonds of extracted linolenic acid (C18:3) are easily oxidized, which leads to a reduced frying thermal stability and storage time of the oil. With erucic acid originally thought to lead to health problems and difficult to digest in humans and livestock due to its long-chain, modern canola-type rapeseed was selected based on 'double-low' seed erucic acid and glucosinolate content [5], Reducing erucic acid and linolenic acid has continued to be an important target for canola/rapeseed production [6]. The genetic basis of seed fatty acid biosynthesis and modification pathways have been well characterised in Arabidopsis thaliana [7][8][9]. Barker et al (2007) established desaturation and elongation pathways well, along with substrate: product relationships assigned to specific enzymes [10]. To better understand the genetic control of seed fatty composition and biosynthesis in rapeseed, in the last few decades, many underlying QTLs of seed quality traits have been detected in bi-parental segregating populations. These QTL have included traits such as oil content [11][12][13][14], protein content [15,16], glucosinolate content [17,18], and fatty acid composition [19][20][21]. In a number of cases, some candidate genes have been identified that coincide with the position of QTL. For example, four loci for B. napus orthologues of FAD2 (BnaFAD2 loci) were mapped to A1, A5, C1 and C5 chromosomes [22], Schierholt et al. (2000) also mapped a locus linked to BnaFAD2 on A05 [23], and Hu et al. (2006) identify a major locus for high oleic acid (C18:1) on A5 chromosome [3], which was proven to be the Fatty acid desaturase-2 (FAD2) gene. In addition, two important FAE1 loci on chromosome A08 and C03 were described by Qiu et al. (2006) [11].
In recent years, genome wide association study (GWAS), also known as association mapping based on linkage disequilibrium, have aimed to identify genetic variants linked to traits. GWAS uncover QTLs or genes from natural populations, and have the advantage of higher resolution and greater cost-effectiveness relative due to screening a larger effective number of recombination events than are accessible in moderate size bi-parental segregating populations. GWAS has successfully been demonstrated to be a powerful tool for dissecting complex traits for crop improvement programs, with the availability of numerous SNPs it has been applied to Zea mays, Triticum aestivum and Oryza sativa [24]. In recent years, many studies have used the Illumina Infinium Brassica 60K SNP array and Dart-seq genotyping approaches to carry out GWAS to detect genetic variation for flowering time, as well as seed quality traits in rapeseed [25][26][27][28][29][30][31][32][33]. The number of robust and well-distributed SNP markers across genome is significant for the efficiency of GWAS in wider germplasm sets. Development of next generation sequencing (NGS) allows identification of a large number of genetic makers for associating with traits of interest based on linkage disequilibrium quickly and efficiently [34]. To identify novel loci and candidate genes that may contribute to variation in fatty acid composition, we carried out an extensive GWAS based on 201,817 SNPs previously developed by SLAF-seq (specific length amplified fragment sequencing) [35,36] and a collection of 300 inbred rapeseed lines. Interactions between the traits were investigated and significantly associated SNP loci were explored along with candidate genes. Favourable allelic variants contributing to an optimal fatty acid composition were identified. This study provides useful information for a more comprehensive understanding of the genetic variation and metabolism mechanism of important fatty acid composition.

Plant materials, growth conditions and field trials
A world-wide collection of 300 diverse rapeseed inbred lines (S4 generation or greater) (S1 Table) was assembled. The provenance and meta-data and genetic relationships for all accessions have been described previously [35,36]. The association population was planted and harvested in the crop field of Jiangxi Agricultural University (JXAU, 115.84E, 28.77N) and Jiangxi Institute of Red Soil (JXIRS, 116.27E, 28.37N) with two biological replications per experimental site in 2014-2015. All seeds were sown on September 29 th 2014 simultaneously in both places. Each variety was planted in a plot with three rows (40 cm line width and 20 cm plant distance), and each row had 12 plants (final seeding time was at the 5-7 leaf phase). Field experiments were arranged and laid out in a randomized complete block design at all sites. All rapeseed inbred lines grown in both environments was cultivated under uniform agronomic practices. Ripe seeds from six plants each accession were harvested and used for seed quality trait analysis after harvests.

Fatty acid compositions evaluation and statistical analysis
The harvested seeds (2g each accession) of 300 rapeseed lines were analysed for estimates of the four fatty acids (erucic acid, oleic acid, linoleic acid and linolenic acid) (%) using DA7200 near infrared spectroscopy (NIRS) (DA 7200, Perten Instruments, Huddinge, Sweden), and the data of four fatty acid content from NIRS were adjusted by the results of gas chromatograph. The value of each fatty acid was expressed as a percentage of the total amount of fatty acids identified. The four fatty acid composition traits of each accession were defined as the mean of the two replicates in the same location. The correlation coefficients between each pair of traits were determined using Student's t-test, and the variance and statistical analysis of four components were obtained using DPS software [37].

SNP genotyping and population structure analysis
Total genomic DNA was extracted from young leave of each rapeseed accession using a modified cetyltrimethylammonium bromide (CTAB) method based on Murray & Thompson (1980) [38], the DNA concentrations and purities of all samples were calculated by a Nanodrop 2000 UV-Vis spectrophotometer (NanoDrop, Wilmington, DE, USA). Quantified DNA samples were used for SLAF sequencing by an Illumina Hiseq TM 2500 [39]. The library construction, paired-end sequencing and SNP calling were conducted as previously described [35,36], a total of 201,817 SNPs with minor allele frequency (MAF) > 0.05 and integrity > 0.8 were selected and used for subsequent analysis, population structure (Q matrix) and relative kinship (K matrix) were analysed by using the Admixture software package [40] and SPAGeDi software [41], respectively, as previously described in our previous research [35,36].

Genome wide association analysis
Based on the 201,817 SNP markers developed for the 300 rapeseed accessions, genome-wide association analysis for the four fatty acid traits was carried out by mixed linear models (MLM) using the Tassel 5.0 software [42]. Fixed effects and random effects in the MLM model were assessed by a Q and K matrix, respectively. The Manhattan plot and Quantile-Quantile plot (Q-Q plot) was drawn by QQman [43] and GGplot2 software [44]. The ideal threshold value was set as 1/201,817 SNPs (-log 10 (p) = 5.3) for identifying the marker-trait associations. Finally, to ensure the accuracy of significant SNPs associated with traits, we removed the unique SNPs associated with a trait in the range of LD decay, others were considered as valid associated-trait SNPs.

Discovery of useful allelic variation for fatty acid composition
The epistatic effect of linked candidate SNPs was evaluated using the epistatic association mapping (EAM) method [45]. When the effect value of SNP is positive, it was taken as increasing effect allele for trait value, conversely, when the effect value is negative as a decreasing effect allele.
SNPs with positive allelic effect values highly associated with oleic acid content and linoleic acid content were analysed, and the SNPs with negative allelic effect value related to erucic acid content and linolenic acid content were counted. Furthermore, the number of varieties with favourable alleles for fatty acid composition were counted.

Prediction of candidate genes for four fatty acid composition
Candidate genes located within the 100 Kb region upstream or downstream of significant associated-trait SNPs were identified based on GO terms (fatty acid biosynthetic process; very long-chain fatty acid metabolic process; fatty acid elongation; fatty acid metabolic; acetyl-CoA metabolic process; fatty-acyl-CoA reductase activity; phosphatidylinositol transporter activity etc.) for fatty acid synthesis, desaturation, elongation and metabolism. Then the identified candidate genes related to fatty acid composition were further confirmed by BLASTX searching against the Arabidopsis protein database.

Candidate gene expression analysis by qRT-PCR
Seven rape varieties from this studied population with significantly different fatty acid contents were selected to carry out the expression analysis of candidate genes (S2 Table), the total RNA from their frozen root, stem, leaf, flower, and seed were extracted using Eastep 1 Super total RNA extraction Kit (Promega, Beijing, China) according to the manufacturer's instructions. RNA quality and quantity were checked using NanoDrop 2000 spectrophotometer (Thermo Fisher Scientific, Wilmington, DE) and RNA integrity was determined by agarose gel electrophoresis.
Three candidate genes (ACP5, FAD2, KCS17) that are significant for fatty acid metabolism were selected to validate the GWAS results by qPCR. The primers for PCR amplification of the three candidate genes were designed with Primer3Plus [46] under strict standards and are provided in supplement file (S3 Table).
A total of 1 μg of total RNA was reverse-transcribed to complementary DNA (cDNA) using the PrimeScript TM RT reagent Kit with gDNA Eraser (Perfect Real Time) (TaKaRa, Japan). according to the manufacturer's instructions. A 20 ul reaction was prepared with 10 ul of SYBR Green Master mix for real time quantitative PCR (Takara, Japan), 1 ul of each primer pair and 1 ul of cDNA templates, 7 ul ddH 2 O was added to the final reaction volume of 20 ul. The PCR amplification of the target genes was performed on a CFX96 Real-Time PCR system (Biorad, Hercules, CA) with the program as follows: 1 cycle of 95˚C for 20s; and 40 cycles of 95˚C for 15s, 60˚C for 30s and 72˚C for 30s; a final melt curve analysis in which the temperature was increased from 55˚C to 95˚C at a rate of 0.5˚C/5s; and a maintenance at 4˚C. In every sample, β-actin was taken as the house-keeping gene, at least 3 technical replicates were performed for each experiment. The relative quantification of gene expression was calculated using the 2 -ΔΔCT method.

Phenotypic variation and correlation analysis for four fatty acid compositions in 300 rapeseed accessions
Continuous and extensive phenotypic variations for each of the four fatty acid composition traits were observed in both environments (JXAU and JXIRS). Linolenic acid content was normally distributed, whilst the contents of erucic acid, oleic acid and linoleic acid had multimodal distributions across the 300 accessions (Fig 1). Descriptive statistical analysis was summarized in Table 1  and 81.90% in JXAU and JXRIS respectively, which was the largest variation of all results; the average content of oleic acid was 48.97% (JXAU) and 49.37% (JXIRS), ranging from10.01% from 81.29% severally; the average linoleic acid content was 15.51% (JXAU) and 14.66% (JXIRS), ranging from 4.35% to 22.64% with the CV of 19.17% and 20.47%, and finally the average content of linolenic acid was 6.89% and 6.67% in two places, ranging from 4.63% to 9.82% with low CV of 9.97% and 11.48%. These data indicated a broad phenotypic variability in four fatty acid compositions within the studied rapeseed population.
There was strong evidence based on ANOVA that traits varied significantly across the 300 genotypes, and had very significant difference under the interaction between genotype and environment (G×E) (P<0.01; Table 2). However, except for oleic acid content, there were no significant effects in other three fatty acids between the two environments. For the correlations between four traits, oleic acid content in both environments had a highly significant negative correlation with erucic acid and linolenic acid content (Table 3). Here, their phenotypic correlation coefficients of -0.8376 �� and -0.5862 �� in JXAU and -0.7942 �� and -0.2775 �� in JXIRS respectively (P<0.01) were observed, but which has a significant positive correlation with linoleic acid content with phenotypic correlation coefficients of 0.4215 �� in JXAU and 0.5824 �� in JXIRS. Moreover, linoleic acid content had a highly significant positive correlation with linolenic acid content, with phenotypic correlation coefficients of 0.6879 �� (JXAU) and 0.5748 �� (JXIRS).

Genome-wide association analysis for the four fatty acid content traits in the 300 rapeseed accessions
To reveal the genetic variations of four fatty acid compositions in B. napus, GWAS for these traits based on MLM models was conducted, The predictive QQ plots show expected distribution agrees of p-values have a high consistency with the observations (Fig 2), and the significantly associated SNPs per traits were displayed on Manhattan plots (Fig 3). The total results of the significant SNP loci associated with the four fatty acids combined under two environments are given in Table 4. GWAS identified 148 SNPs significantly associated with four fatty acid content traits on 8 chromosomes. However, the majority of trait-linked SNPs were mainly distributed on A08 and C03 chromosomes, with 82 and 48 significant SNP on chromosome A08 and C03, respectively (S4-S7 Tables), and all trait-linked closely SNPs explained 6.09% 16.55% of observed phenotypic variation. A total of 78 SNPs of which associated with erucic acid content significantly were detected in GWAS, 12 (15%) were both detected in two environments (S4 Table). In addition, 57 in JXAU and 37 in JXIRS with 22 same SNP loci on 5 chromosomes for oleic acid were detected (S5 Table). Furthermore, we identified 118 SNPs for linoleic acid content trait (80 in JXAU and 38 in JXIRS with 21 uniform SNP loci (S6 Table), but only 2 significant SNP loci in JXAU were found (S7 Table). In addition, 51 peak SNPs (30 of erucic acid, 34 of oleic acid, 37 of linoleic acid and 12 of linolenic acid) were detected on 8 chromosomes (Table 5).
For the SNP locus analysis, there were some SNP loci significantly associated with several traits simultaneously. For example, a total of 30 SNP loci on A08, C03 and C06 chromosomes were detected in three traits of erucic acid content, oleic acid content and linoleic acid content. Apart from what is outlined above, 16 SNP loci linked to erucic acid content and oleic acid content, 25 loci for oleic acid content and linoleic acid content, and 7 SNPs for erucic acid content and linoleic acid content concurrently. These consistent SNPs detected simultaneously in different traits indicated that they might control the different fatty acid composition traits synchronously. What's more, we found two remarkable associated regions on chromosome A08 and C03 were consistent with the QTL results based on bi-parental mapping previously [19,31,47].
To identify highly favourable alleles with for fatty acid composition in B. napus, we investigated the allelic effects of all the significantly associated SNPs. We observed 108 highly favourable alleles for increasing oleic acid and linoleic acid content, also for decreasing erucic acid content simultaneously (S8 Table). Specially, we found 66 highly favourable alleles for reducing the content of erucic acid and linolenic acid (S9 Table), 82 and 105 highly favourable alleles contributed to increase oleic acid and linoleic acid content respectively (S10 Table, S11 Table). Discovery of SNP loci and those favourable alleles in current study will provide insight into the genetic basis of four fatty acid biosynthesis in rapeseed.

Candidate genes identification of four fatty acids in B. napus
To further reveal the molecular function of the SNPs significantly associated with the four traits, we extracted the genes within the 100 Kb upstream or downstream regions of the traitassociated SNPs in the reference genome of B. napus "Darmor v4.1". We found 802 genes were identified located in the candidate regions around 52 SNPs (S12 Table), 29 of these candidate genes were enriched into fatty acid biosynthetic process, fatty acid metabolic process and Lipid transport and metabolism in the GO terms annotation analysis (S13 Table). 20 fatty acid composition candidate genes were homologous to A. thaliana genes involved in metabolic networks of fatty acids. These candidate genes were located in four chromosomes (A08, A09, A10 and C03), 16 of which were distributed on the A subgenome, the chromosome A08 have the most genes (12), and chromosome A09 and A10 have 2 candidate genes, respectively, chromosome C03 have other 4 fatty acid candidate genes (Table 6), which suggested the A subgenome pay a more important role than C subgenome in the metabolic networks of fatty acids of rapeseed. Some genes tagged by associated SNPs in our study represent known fatty acid composition genes, and were enriched in biological processes of FAs biosynthesis. For erucic acid, two Fatty acid elongation 1 (FAE1) orthologous genes (BnaA08g11130D and BnaC03g65980D) were found in the distance of 40.93 Kb from SNP Bn-A08-10146770 and 65.37 Kb form Bn-C03-55618985 respectively, which were well-known Bna.FAE1 homologs controlling erucic acid content found in previous studies [25,31,[48][49][50]. Moreover, another two candidate genes of BnaA08g11140D and BnaC03g66040D orthologous to 3-ketoacyl-CoA synthase 17 (KSC17) in the distance of 47.02 Kb from SNP Bn-A08-10146770 and 98. 16 Kb form Bn-C03-55712102 were detected respectively, which is also a member of the 3-ketoacyl-CoA synthase family involved in the biosynthesis of very long chain fatty acids [51]. For the synthesis of unsaturated fatty acids, which are affected by the fatty acid desaturase activity in plant, the fatty acid desaturase genes have been proved as the major genes for the control of oleic acid content as same as the ratio of polyunsaturated fatty acid, we found the important candidate genes of BnaC03g03500D orthologous to Fatty acid desaturase 2 (FAD2) in the distance of 45.18 from Bn-C03-1664875, which is the major enzyme responsible for the synthesis of 18:2 fatty acids in plant endoplasmic reticulum [3], and FAD2 gene had been mapped in B. napus on chromosomes A1, A5, C1, and C5 in previous studies [4,52]. Furthermore, many other important candidate genes for fatty acid biosynthesis were detected, which were orthologous to A. thaliana, such as Dermatan sulfate epimerase-like (DSEL), Acyl carrier protein 5 (ACP5), Long chain acyl-CoA synthetase 9 (LACS9) and so on, these candidate genes except for BnaA.FAE1 (BnaA08g11130D) and BnaC. FAE1 (BnaC03g65980D) have not been verified in previous genetic analyses, which are likely play important roles in fatty acid synthesis and transport. Therefore, these candidate genes detected in this study should be certified by further analysis in the future.

The expression of three candidate genes in diverse rapeseed accessions and tissues
In order to validate the candidate genes significantly associated fat acids compositions, we selected three key genes involved in fat acid synthesis and measured their gene expression in five different tissues (root, stem, leaf, flower and seed) of seven diverse rapeseed accessions using qRT-PCR. We found the three candidate genes have high expression in seeds or flowers of these selected rapeseed (S1 Fig), and the expression of ACP5 have a strong positive correlation with erucic acid (0.68), have strong negative correlation with oleic acid (-0.67) and linoleic acid (-0.71), suggested the high expression of ACP5 could promote the erucic acid synthesis, Table 3

Correlation coefficient Erucic acid (%) Oleic acid (%) Linoleic acid (%) Linolenic acid (%)
Erucic acid (%)  but decrease the oleic acid and linoleic acid. FAD2 and KCS17 have positive correlation with oleic acid and linoleic acid, indicated they could increase the two fat acids in the studied population (Table 7).

Discussion
Fatty acid compositions in seed of rapeseed play important roles in improving the edible oil nutritional and storage quality. In this study, we investigated the four important fatty acid content traits of erucic, oleic, linoleic and linolenic for 300 rapeseed accessions, the four traits exhibited large variation in two different environments, and there proved to be have the weak or no genetic relationship among 300 accessions in our previous study [35,36], validating the suitability of GWAS for these traits in this study population [53]. In addition, there are high correlations among these fatty acid traits, significant negative correlation between the content of erucic acid and oleic acid was identified, and linoleic acid content trait had high positive correlation with other three traits, they were co-localized in a small region on chromosome A08 and C03 chromosomes, the strong correlation between these traits was also observed in previous studies [19,21,45,54]. Furthermore, in current study, although the materials were grown in two near environments with near locations, but there were very distinct differences in these fatty acid compositions under the interaction between genotype and environment, we speculate the acid soil with low pH value (4.67) in JXIRS was the main reason resulting in very distinct differences of four fatty acids content [55]. Fatty acid composition are typical quantitative traits, based on the genetic markers (SNP, SSR, RFLP and AFLP), bi-parent population and statistical methods, many QTLs for fatty acid composition distributed on most of chromosomes of B. napus were detected in the past few decades [3, 4, 19-21, 25, 30]. Recent efforts have also been made in detecting the genetic loci responsible for fatty acid composition in B. napus based on the Illumina Infinium Brassica 60K SNP array, some genome wide analysis were carried out to identified genomic region and candidate genes associated with fatty acid content in the past few years [26,27,31]. In this study, using 201,187 genome-wide SNP markers developed by SLAF-seq technology [35,36], we carried out the GWAS for four fatty acid traits in 300 rapeseed inbred lines, and hundreds of SNP loci highly associated with the four fatty acid compositions traits were identified on 8  chromosomes of B. napus, many of these SNPs were simultaneously detected in two environments, most of significantly associated SNPs (more than 80%) for the content of erucic acid, oleic acid and linoleic acid were identified on A08 and C03 chromosomes, which was in accordance with above QTL and GWAS studies for fatty acid composition, from this we can infer that the genomic region controlling the fatty acid biosynthesis were mainly distributed on A08 and C03 chromosomes. In addition, we also found a few trait-associated SNPs on other chromosomes, some identical chromosome regions were exist in Qu et al. (2017) [31], for example, the significant SNPs associated with olenic acid on A09 chromosomes and the candidate gene of BnaA09g05410D (orthologous to Transparent testa 16 TT16) was identified both in two studies, but other related gene groups found in A02 and C01 linkages of Qu et al. (2017) were not detected in current study [31], which probably attribute to the different population used in different studies. Under the reference-guided analysis, As the physical position and alleles of SNP markers used in this study are known, therefore, breeders and researchers could easily obtain valuable information for other related rapeseed research based on our results. In addition, by evaluating the allelic effects of trait-associated SNPs, we observed many favourable alleles for decreasing the erucic acid content, which had positive effects for increasing the oleic and linoleic acid content in the same time, these SNPs could be used for low-erucic and higholeic acid rapeseed breeding in the future. The genetic basis of seed fatty acid biosynthesis and modification pathways have been well characterised in Arabidopsis thaliana [9]. Well established desaturation and elongation pathways, along with substrate: product relationships assigned to specific enzymes [10], which was a co-operation of many genes of seed development, energy metabolism, fatty acid and triacylglycerol (TAG) biosynthesis pathways [9,56]. De novo synthesis of fatty acids occur in the seed plastid, where the palmtic acid (16:0), stearic acid (18:0) and oleic acid (18:1) are formed and then released to the endoplasmic reticulum (ER) by two kinds of acyl-ACP thioesterase enzymes (FATA and FATB with high affinity to 18:1-ACP and 16:0-ACP respectively) [57,58]. These fatty acids were modified by desaturation enzymes (FAD2 and FAD3) to produce the polyunsaturated linoleic (18:2) and linolenic acid (18:3) [52,59], the gene expression analysis in this study supported further that the FAD2 gene would promote the linolenic acid synthesis. In addition, the fatty acids could be elongated to erucic acid by FAE1 gene or esterified to glycerol to produce the TAG stored as a major seed oil form in plants [60]. In current study, a total of 14 candidate genes were detected on A08 chromosomes, including DSEL, FAE1, MCCB, CCOAOMT1 and WIN2 etc. In addition, there were 4 fatty acid biosynthesis genes were found on C03 chromosomes, including FAD2, ATCHS, FAE1 and KCS17 etc., Specially, two known homoeologous genes (namely BnaA.FAE1 and BnaC.FAE1) controlling erucic acid content were identified in our study [25,31,50,61], and candidate genes BnaA08g11140D and BnaC03g66040D orthologous to KCS17 of A. thaliana involving in the biosynthesis of very long chain fatty acids were also found on chromosomes A8 and C3. Moreover, the gene of  [62]. These results obviously indicate that there are genomic regions controlling the seed fatty acid biosynthesis on A08 and C03 chromosomes. By the way, some important fatty acid candidate genes were also scanned on other chromosomes. On A09 chromosome, the candidate gene of BnaA09g03610D orthologous to Acyl carrier protein 5 (ACP5) was located in the distance of 71.03 Kb from SNP Bn-A09-1752479 on chromosome A09, which is a small acidic proteins functioning as important cofactors in the de novo synthesis of fatty acids, overexpression of AtACP5 further led to an a decrease of oleic acid (C18:1) and an increase of palmitic acid (C16:0) [63], the gene expression of ACP5 in this study have strong negative relationship with oleic acid, suggested it seem to has the same gene function as AtACP5. In addition, On A10 chromosome, three candidate genes were ascribed to fatty acid, BnaA10g20970D orthologous to A. thaliana Long chain acyl-CoA synthetase 9 (LACS9) encoding major plastid long chain acyl-CoA synthetase with a slight substrate preference of oleic acid over any of the other fatty acids. Protein-tyrosine phosphatase-like (PTPLA) (BnaA10g21780D) with acyl-CoA dehydratase activity has a potential role in synthesis of VLCFAs (very long chain fatty acids) under the interaction with Eceriferum 10 (CER10), which is a component of the microsomal fatty acid elongase complex. On C06 chromosome, FatA acyl-ACP thioesterase (FATA) affects the oil content and fatty acid composition of the seeds in Arabidopsis when reducing expression of FatA thioesterases [64]. Our GWAS analysis led to the identification of promising candidate genes for fatty acid biosynthesis and metabolism efficiently.

Conclusion
In this study, based on GWAS with MLM model analysis, significant association signals for the content of erucic acid, oleic acid, linoleic acid and linolenic acid in seeds of B. napus were found on A06, A08, A09, A10, C02, C03, CO6, C07 and C08 chromosomes in two environments. The genomic regions controlling the fatty acid biosynthesis were inferred to distribute mainly on A08 and C03 chromosomes. 20 orthologs of the functional candidate genes related to fatty acid biosynthesis in a distance of 100 Kb around these significantly SNP-trait associations were identified by BLAST analysis and comparison of previous linkage mappings, including the known vital fatty acid biosynthesis genes of BnaA.FAE1 and BnaC. FAE1 on the A08 and C03 chromosomes, and other potential candidate genes involving in the fatty acid biosynthesis pathway, such as the orthologs genes of FAD2, LACS09, KCS17, CER4, TT16 and ACBP5. This study lays a foundation for uncovering the genetic variations and the improvement of fatty acid composition in B. napus.