Identiﬁcation of Major Loci and Candidate Genes for Anthocyanin Biosynthesis in Broccoli Using QTL-Seq

: Anthcyanins determine the colors of ﬂowers, fruits, and purple vegetables and act as important health-promoting antioxidants. BT 126 represents a broccoli variety with a high content of anthocyanins (5.72 mg/g FW). Through QTL-seq bulk segregant analysis, the present study aimed to determine the quantitative trait loci (QTLs) involved in anthocyanin biosynthesis in the F2 population (n = 302), which was obtained by crossing BT 126 with a non-anthocyanin-containing SN 60. The whole-genome resequencing of purple (n = 30) and green (n = 30) bulk segregates detected ~1,117,709 single nucleotide polymorphisms (SNPs) in the B. oleracea genome. Two QTLs, tightly correlated with anthocyanin biosynthesis ( p < 0.05), were detected on chromosomes 7 ( BoPur7.1 ) and 9 ( BoPur9.1 ). The subsequent high-resolution mapping of BoPur9.1 in the F2 population (n = 280) and F3 population (n = 580), with high-throughput genotyping of SNPs technology, narrowed the major anthocyanin biosynthesis QTL region to a physical distance of 73 kb, containing 14 genes. Among these genes, Bo9g174880 , Bo9g174890, and Bo9g174900 showed high homology with AT5G07990 (gene encoding ﬂavonoid 3 (cid:48) hydroxylase), which was identiﬁed as a candidate gene for BoPur9.1 . The expression of BoF3’H in BT 126 was signiﬁcantly higher than that in SN60. Multiple biomarkers, related to these QTLs, represented potential targets of marker-assisted selection (MAS) forantho-cyanin biosynthesis in broccoli. The present study provided genetic insights into the development of novel crop varieties with augmented health-promoting features and improved appearance. Chinese kale [34], only few have focused on broccoli [20]. The purple broccoli mutant represents an interesting mutation, in which the accumulation of anthocyanins in flower buds causes the mutant heads to exhibit purple coloration. The broccoli purple mutation was found to be controlled by QTLs. We designate the symbol Pur for the Purple allele. Through QTL-seq, this work aimed to identify QTLs involved in anthocyanin biosynthesis in the F2 population, which was obtained by crossing the purple broccoli line BT 126 with SN 60 (a cultivar with green heads).


Introduction
Broccoli (Brassica oleracea var. italic) is a popular vegetable of B. oleracea that differs from most Brassica species, including Chinese cabbage, turnip, cabbage, broccoli, cauliflower, and oilseed rape. Most varieties of broccoli are domesticated from crop wild relatives in the Mediterranean Basin and grow as annuals, producing a large head with florets, buds, leaves, stalks, and stems for consumption. Both of broccoli and cauliflower cultivar groups are members of the CC genome B. oleracea (2n = 18) coenospecies. High-quality reference genomes of cauliflower have been reported, and the assembled cauliflower genome was 584.60 Mb in size [1]. As a great food source of essential vitamins and minerals, broccoli contains antioxidant phytochemicals, such as glucoraphanin, which may help prevent cancer [2]. Purple broccoli attracts increasing attention as a functional food, owing to its pleasing appearance and high level of health-promoting effects [3]. The purple coloration has been identified as one of the signs of anthocyanin accumulation [4].
Anthocyanins belong to a class of flavonoid compounds that impart color to plants and play an important role in plant protection against a variety of biotic and abiotic stresses [5][6][7]. Anthocyanin biosynthetic pathway genes have been extensively characterized in Arabidopsis (Arabidopsis thaliana), maize (Zea mays), tomato (Solanum lycopersicum), eggplant (Solanum melongena), and other plant species [8][9][10][11]. The induction of structural genes and transcription factors is considered to be an important mechanism for the regulation of anthocyanin biosynthesis in the Brassica species [12][13][14]. A cauliflower purple mutation exhibited a tissue-specific pattern of anthocyanin overproduction [15]. Due to the insertion of Harbinger transposon, the upregulation of BoMYB2 specifically activates BobHLHs and some downstream anthocyanin structural genes to generate the ectopic accumulation of anthocyanin. Similarly, the accumulation of anthocyanin is caused by the activation of TT8 and MYB2 genes in red cabbages [14]; of BrMYB2 and downstream genes, such as BrTT8, BrF3'H, BrDFR1, BrANS1, BrUGTs, BrATs, and BrGSTs, under the control of BrMYB2 in a purple head of Chinese cabbage cultivar 11S91 [16]. In purple cabbages, deleting or replacing nucleotides in the exon of BoMYBL2-1 is solely responsible for the purple coloration [17]. In addition, temperature and light are the major environmental factors that affect anthocyanin accumulation. In purple head Chinese cabbage, BrMYB2 and BrTT8 activated anthocyanin structural genes after low temperature induction [18]. Elevated temperature could suppress anthocyanin accumulation via COP1-HY5 signaling, and MYBL2 down-regulation partially modulated the high-temperature-associated suppression of anthocyanin production [19]. Some efforts have also led to the identification of candidate genes that regulate the coloration of the Brassica species. In broccoli, three QTLs have been mapped to the purple sepal trait of the flower head on chromosome C01 [20]. In ornamental kale, the genes that individually conferred pink and purple leaf colorations have been mapped to chromosomes C3 [21] and C9 [22], respectively. In Zicaitai, an important locus on chromosome 7 highly controlled the stalk color trait, which was significantly correlated with bHLH49 expression in the F2 population [23]. In purple-heading Chinese cabbage, the purple inner leaf trait was solely regulated by the dominant gene BrPur, which was mapped to A07, between SSR markers A710 and A714, with a genetic distance of 3.1 and 3.5 cM, respectively [24].
The majority of agronomic traits are controlled by QTLs, which are critical for improvement in crop breeding, through marker-assisted selection (MAS). The classical method of QTL mapping is linkage mapping, which is laborious and requires a great amount of time. Nextgeneration sequencing (NGS) has become the new strategy for establishing associations between agronomic traits and biomarkers or genes. It has been demonstrated that the QTLseq method, which combines bulk segregant analysis (BSA) with NGS, is an effective tool for mapping and isolating QTLs [25]. QTL-seq is performed with two groups of individual plants, with a contrasting phenotype on a trait of interest, from segregating population-either F2 recombinant inbred lines, double haploid, or backcross populations. Through high-throughput SNP (Hi-SNP) technology, the genotype analysis of two mixed pools identifies the genomic position of the polymorphic molecular markers, and the major QTL region with significant segregation of genotypes is identified. Hi-SNP is a technique for large-scale SNP genotyping, based on multiplex-PCR, combined with the next generation sequencing and bioinformatics tools. Amplicon sequencing, combined multiplex-PCR and NGS with higher depth and low false discovery rate (and was more accurate than the whole-genome resequencing (WGS)), has been used for known SNPs genotyping in diploid but was not reported in allopolyploid crops. QTL-seq used for SNP genotyping had many advantages, such as simple primer design, single short reads sequences, high-throughput, high-depth sequencing, and easy to automate and process, e.g., recently, the QTL-seq method has been utilized for mapping QTLs related to resistance to rice blast disease and seedling vigor [26], leaf spot resistance in peanuts [27,28], heat tolerance in broccoli [29], resistance to Fusarium oxysporum f. sp. niveum race 1 in watermelon [30], cucumber early flowering [31], and tomato fruit weight and lobule amounts [32].
Mutant analyses, leading to purple or red organs, has been extensively studied in flowers, fruits, and model plants [15]. Although there is much research on the underlying mechanisms of anthocyanin biosynthesis in cauliflower [15], cabbage [17], kohlrabi [33], kale [22], and Chinese kale [34], only few have focused on broccoli [20]. The purple broccoli mutant represents an interesting mutation, in which the accumulation of anthocyanins in flower buds causes the mutant heads to exhibit purple coloration. The broccoli purple mutation was found to be controlled by QTLs. We designate the symbol Pur for the Purple allele.
Through QTL-seq, this work aimed to identify QTLs involved in anthocyanin biosynthesis in the F2 population, which was obtained by crossing the purple broccoli line BT 126 with SN 60 (a cultivar with green heads).

Plant Materials
Two broccoli inbred lines, including BT 126 (with purple heads) and SN 60 (with green heads), were used as parental lines in this study. To develop segregating populations for anthocyanin biosynthesis, the purple head plant BT 126 was crossed with the green head plant SN 60 ( Figure 1). The F1 plants were self-pollinated to produce the two F2 populations with 302 and 280 individuals, respectively.
Mutant analyses, leading to purple or red organs, has been extensively studied in flowers, fruits, and model plants [15]. Although there is much research on the underlying mechanisms of anthocyanin biosynthesis in cauliflower [15], cabbage [17], kohlrabi [33], kale [22], and Chinese kale [34], only few have focused on broccoli [20]. The purple broccoli mutant represents an interesting mutation, in which the accumulation of anthocyanins in flower buds causes the mutant heads to exhibit purple coloration. The broccoli purple mutation was found to be controlled by QTLs. We designate the symbol Pur for the Purple allele.
Through QTL-seq, this work aimed to identify QTLs involved in anthocyanin biosynthesis in the F2 population, which was obtained by crossing the purple broccoli line BT 126 with SN 60 (a cultivar with green heads).

Plant Materials
Two broccoli inbred lines, including BT 126 (with purple heads) and SN 60 (with green heads), were used as parental lines in this study. To develop segregating populations for anthocyanin biosynthesis, the purple head plant BT 126 was crossed with the green head plant SN 60 ( Figure 1). The F1 plants were self-pollinated to produce the two F2 populations with 302 and 280 individuals, respectively. Initially, a population of 302 F2 individuals, along with 10 plants from each parental line, were grown at Zhuanghang Experimental Base of Shanghai Academy of Agricultural Sciences, Shanghai in 2018. The phenotyping of each F2 individual was carried out on at least three separate days, after the flower head reached maturity in the field. Curd color, which showed a color distribution from green to purple, was visually scored from 1 to 5: 1-green, 2-slight purple, 3-light purple, 4-slightly darker purple, and 5-purple ( Figure 1, e-i). The total anthocyanin content has been determined by the high-performance liquid chromatography (HPLC) method at six biological replicates. The BT 126 is a purple broccoli variety, with an average content of anthocyanin 5.72 mg/g FW. It was crossed to SN 60, with very low anthocyanin content (0.64 mg/g FW), to generate F1 progeny. The average contents of anthocyanin, for the 5 phenotypic categories, were 5.38, 3.26, 2.29, 1.27, and 0.78 mg/g FW. All the above information was increased in the revised manuscript. Initially, a population of 302 F2 individuals, along with 10 plants from each parental line, were grown at Zhuanghang Experimental Base of Shanghai Academy of Agricultural Sciences, Shanghai in 2018. The phenotyping of each F2 individual was carried out on at least three separate days, after the flower head reached maturity in the field. Curd color, which showed a color distribution from green to purple, was visually scored from 1 to 5: 1-green, 2-slight purple, 3-light purple, 4-slightly darker purple, and 5-purple ( Figure 1E-I). The total anthocyanin content has been determined by the high-performance liquid chromatography (HPLC) method at six biological replicates. The BT 126 is a purple broccoli variety, with an average content of anthocyanin 5.72 mg/g FW. It was crossed to SN 60, with very low anthocyanin content (0.64 mg/g FW), to generate F1 progeny. The average contents of anthocyanin, for the 5 phenotypic categories, were 5.38, 3.26, 2.29, 1.27, and 0.78 mg/g FW. All the above information was increased in the revised manuscript. Subsequently, two populations of 280 F2 and 580 F3 individuals were grown under routine management at Zhuanghang Experimental Base of Shanghai Academy of Agricultural Sciences, Shanghai in 2019 and 2020, respectively. We started the broccoli seeds in the plastic tunnel in August. Four weeks after germination, we transplanted them to field plots. The head color of each F2 and F3 plants was visually phenotyped in 60 days, when grown from transplants, when the heads usually had a diameter larger than 2 1 4 inches. To avoid the sunlight and temperature effects, all the plants were grown in the same plastic tunnel. Leaf tissues were harvested and stored at −20 • C for DNA extraction. The head tissues of both parents (BT 126 and SN 60), at the full-size, mature stage, were collected from the same site of the top head, at three biological replicates, and used for RT-qPCR analysis. The anthocyanins in the broccoli heads were isolated and assessed, according to the method proposed by Liu C. et al. [4].
All materials were obtained from the Institute of Horticulture, Shanghai Academy of Agricultural Sciences.

DNA Purification, Library Generation, and Whole-Genome ReSequencing
Total DNA was extracted from fresh leaves using the CTAB method [35]. A total of 1.5 µg DNA, per specimen, was utilized for DNA sample preparations. The DNA quality was assessed by 1% agarose gel electrophoresis. Among the F2 segregating population, composed of 302 plants, 30 with purple heads (high anthocyanin biosynthesis, HAB) and 30 with green heads (low anthocyanin biosynthesis, LAB) were selected to extract equal amount of DNA, pooled to construct the DNA bulks. The DNA bulks, together with the two parental DNA, were used for whole-genome resequencing.
By ultrasonication, 350-bp fragments were obtained from the tested DNA sample. A sequencing library was generated using the Truseq Nano DNA HT Sample Preparation Kit (Illumina, San Diego, CA, USA), according to the manufacturer's instructions. The constructed library was sequenced on the Illumina HiSeq4000 platform (Illumina, CA, USA), and 150-bp paired-end reads were produced with approximately 350-bp inserts. Stringent quality control (QC) steps were applied to ensure the reliability and accuracy of the reads. Then, the filtered, high-quality sequences from the DNA bulks and parental genotypes were aligned and mapped to the public B. oleracea genome database (TO1000) (http://plants.ensembl.org/Brassica_oleracea) (accessed on 5 May 2021) with the Burrows-Wheeler alignment (BWA) tool [36,37]. Alignment files were converted into BAM files with SAMtools (Wellcome Trust Genome Campus, Cambridge, UK) [38]. SNP detection was carried out for each specimen, utilizing the UnifiedGenotyper function in the GATK3.8 software (The Broad Institute of Harvard and MIT, Cambridge, MA, USA) [39]. ANNOVAR (Children's Hospital of Philadelphia, Philadelphia, PA, USA) was employed for SNP annotation, based on the GFF3 file of the reference genome [40].

QTL-Seq Analysis
By applying an established QTL-seq method, based on SNP-index and ∆( SNP-index) estimates, candidate genomic regions harboring the main QTL(s) involved in anthocyanin biosynthesis in broccoli were identified. Only the SNPs homozygous in either parent and polymorphic between the parents were prepared for the further analysis. Then 1,117,709 SNPs, between both parents, were selected. The read depth for the above-mentioned homozygous SNPs, in LAB and HAB bulks, was obtained to evaluate the SNP-index. The SNP-index, calculated according to the reads order-checking depth information, utilized short reads covering the given nucleotide position, calculated the differs reads bar number, and accounted for the ratio of the total number [26]. The genotype of one parent was utilized as a reference to determine the number of reads for the parental genotype, or other genotypes, in the LAB and HAB libraries. Then, the number of the various reads was divided by the total read number, and the ratio constituted the SNP-index of the base sites. The SNP-index points below 0.3, in both libraries, were removed. The sliding window method was utilized to present the SNP-indexes for the entire genome. All SNP-indexes in a given window were averaged to obtain the SNP-index for that particular window. The window size of 1Mb and the step size of 1Kb were routinely utilized. The ∆ (SNPindex) was then calculated using the following formula: [SNP index (HAB bulk)-SNP index (LAB bulk)].

High-Throughput Genotyping of SNPs in Intra-Specific Mapping Individuals
In order to validate the BoPur QTL, obtained by QTL-seq, classical QTL mapping was performed, via the selection of the SNP, with a polymorphism between the genotypes of the parents (BT 126 and SN 60). Through the Hi-SNP method, a total of 33 SNPs, differentiating BT 126 and SN60, were utilized for validation and high-throughput genotyping in the F2 segregating population with 280 individuals and parents. The HI-SNP technology was developed by Shanghai BioWing Applied Biotechnology Company (http://www.biowing.com.cn/) (accessed on 9 June 2021) the genotyping of 33 SNPs was carried out by multiplex PCR with NGS on Illumina X-10 (Illumina, CA, USA) [41]. The genotyping primers were shown in the Supplementary Table S1. Based on the phenotypic and genotypic data, linkage maps were constructed by QTL IciMapMaker (Chinese Academy of Agricultural Sciences, Beijing, China) with a logarithm of the odds (LOD) value [42]. The LOD score is a measure of the strength of the evidence for the presence of a QTL at a particular location (the LOD score = log10 likelihood ratio, comparing single-QTL model to the "no QTL anywhere" mode).

Analysis of the Candidate Gene
To infer the gene regulation patterns of BoF3'H during anthocyanin biosynthesis, a qRT-PCR assay was performed. The RNA was isolated from BT 126 and SN60 at three head developmental stages, with both low and high anthocyanin contents. Reverse transcription was conducted using oligo-dT primers and PrimeScript™ RT Master Mix (Takara BIO, Inc., Shiga, Japan). The primer pair sequences for BoF3'H, BoANS2, and BoTTG1 used in this study were previously designed for the RT-qPCR analysis. BoLBD38.3-specific primers were designed based on the sequences in NCBI (accession number: XM_013739916). The qPCR primer pair sequences for BoLBD38.3 were 5 -GCCCAAACGGAGACGATTAG-3 (forward) and 5 -AACCGTTCACTGGCGATGTG-3 (reverse), gene-specific primers that were confirmed to produce specific gene products by sequencing [15]. Real-time PCR was performed using TB Green ® Premix Ex Taq™ (Tli RNaseH Plus) (Takara, Dalian, China) on an ABI QuantStudio 5 real-time PCR system(Thermo Fisher Scientific, MA, USA) according to the manufacturer's instructions. Actin was used as a reference gene [43]. The data were processed using the 2 −∆∆Ct method [44]. Sequence data of BoF3'H can be found in the GenBank data libraries, under accession number XM_013751545.

Inheritance of Head Color of BT 126 with SN 60
Significant differences in head color were observed among the BT 126 (parent 1, P1, purple head), SN 60 (parent 2, P2, green head), and F1 hybrids planted in 2018 and 2019 ( Figure 1). The BT 126 purple line exhibited a high anthocyanin accumulation in the heads (5.72 mg/g FW) and developed normally when compared to green broccoli, under normal conditions, in the field and greenhouse [4]. It was crossed to SN 60, with very low anthocyanin content (0.241 mg/g FW), to generate the F1 progeny. F1 individuals displayed third grade color. As expected, the F2 population was divided into purple and green classes, with varying degrees of continuous distribution in 1st-5th color grades ( Figure 1E-I). No transgressive segregation was observed in either direction of the parental genotype in the mapping population. We could infer that the purple trait was controlled by QTL.

QTL-Seq Analysis
The high-throughput, whole-genome resequencing of both parents, as well as the LAB and HAB libraries, yielded 158.49 to 261.57 million reads per sample. Based on the B. oleracea reference genome (estimated genome size~630 Mb), the read mapping ratio was >93% ( Table 1). The mean coverage of the reference genome, across all samples, was 33×. As depicted in Table 1, the 4× coverage of each specimen ranged from 86.85% to 92.43%. Meanwhile, the value of Q20 was above 93% in all specimens (data not shown). The com-parative genome sequence analysis of the parental DNA and DNA bulks with the reference genome revealed 1,117,709 polymorphic SNPs. The SNP-index of each SNP, that distinguished LAB from HAB, was determined. The mean SNP-index, covering 1 Mb genomic sequence, was assessed separately in LAB and HAB, by the 1 kb sliding window method, and mapped against all B. oleracea reference chromosomes (9 in total). The ∆ (SNP-index) was determined by combining the SNPindex data of LAB and HAB and was plotted against the positions (Mb) on the B. oleracea genome. The QTL-seq analysis detected two major genomic regions on chromosomes 7 (36,784,791,849) (BoPur7.1) and 9 (46,217,406-52,600,419) (BoPur9.1), with significant associations with anthocyanin biosynthesis in broccoli ( Figure 2). Moreover, LAB and HAB mapping plants with reduced or elevated anthocyanin biosynthesis had the majority of SNP alleles from SNP 1 to SNP 26 (Table S1). An important genomic region [BoSNP5 (46,217,406 bp) to BoSNP 26 (52,600,419 bp)] harboring the QTL Pur on chromosome 9 showed a ∆ (SNP-index) that was markedly different from 0 (p < 0.05) (Figures 2A and 3B).

Validation of QTL-Seq-Derived Anthocyanin Biosynthesis QTL through High-Throughput SNP
To assess BoPur7.1 and BoPur9.1, detected by QTL-seq, a high-throughput SNP was carried out. The genotyping data of 26 SNP markers were mapped to a highly dense intraspecific genetic region of chromosomes 7 and 9, depicting polymorphisms between pa-   To further delineate the BoPur9.1 QTL, the genotyping data of 17 selected SNP markers, the most tightly linked to BoPur9.1 (SNP10-SNP26), were analyzed in 580 F3 mapping individuals. Loci mapping analysis identified a major QTL for anthocyanin biosynthesis, designed by two SNP markers, including SNP22 and SNP23. This result was consistent with the QTL analysis supporting a major anthocyanin biosynthesis QTL BoPur9.1, which was located in the genomic interval of 51.71-51.79 Mb on chromosome 9 (Table S2).

Validation of QTL-Seq-Derived Anthocyanin Biosynthesis QTL through High-Throughput SNP
To assess BoPur7.1 and BoPur9.1, detected by QTL-seq, a high-throughput SNP was carried out. The genotyping data of 26 SNP markers were mapped to a highly dense intraspecific genetic region of chromosomes 7 and 9, depicting polymorphisms between parental genotypes and between LAB and HAB, and were combined with phenotypic features of a second F2 mapping population with 280 individuals. Classical QTL analysis, based on interval mapping and composite interval mapping, revealed two highly impor-  (Table 2). To further delineate the BoPur9.1 QTL, the genotyping data of 17 selected SNP markers, the most tightly linked to BoPur9.1 (SNP10-SNP26), were analyzed in 580 F3 mapping individuals. Loci mapping analysis identified a major QTL for anthocyanin biosynthesis, designed by two SNP markers, including SNP22 and SNP23. This result was consistent with the QTL analysis supporting a major anthocyanin biosynthesis QTL BoPur9.1, which was located in the genomic interval of 51.71-51.79 Mb on chromosome 9 (Table S2).
BoF3'H was annotated as a homologue of the anthocyanin biosynthesis gene. Table 3. Annotation of B. oleracea genes in the candidate region. A primer pair, spanning the full-length CDS of BoF3'H was designed, and PCR was performed using the cDNA of BT 126 and SN 60 as a template. DNA sequencing revealed that the full-length of BoF3'H in purple-head SN 60 is 1669 bp, whereas it was 1600 bp in green-head SN 60. Compared with BoF3'H in BT 126, a deletion of 68 bp and 1bp was found at nucleotide 1080 and 1501, respectively, and nine SNPs were present in SN60 (sequences of BT 126 and SN60 were illustrated in Table S3). The polymorphism of the candidate gene BoF3'H was further confirmed in the segregating population. To delineate the potential candidate genes regulating the anthocyanin biosynthesis in broccoli, differ-ential expression profiling of BoF3'H was performed in two parental lines, at three

Discussion
Increasing the anthocyanin content in Brassica vegetables represents an important goal for nutriment breeding. This work discovered two important genomic regions harboring anthocyanin biosynthesis QTLs BoPur7.1 and Bopur9.1 on chromosomes 7 and 9, based on intra-specific broccoli mapping individuals, through whole-genome NGS-based, high-throughput QTL-seq. The two QTL-seq-derived Pur QTLs were subsequently verified by Hi-SNP analysis. In previous studies, a major locus was reported to control anthocyanin pigmentation. In purple cauliflower (B. oleracea var botrytis), a Pur gene, encoding the transcription factor R2R3 MYB, was isolated [43]. Broccoli flower head's purple sepal trait is regulated by a major loci and two minor loci on chromosome C01 [20]. The purple gene of non-heading Chinese cabbage is subject to a single dominant inheritance mode but does not follow the Mendel law [33]. Due to the considerable variations in anthocyanin content, a novel locus was mapped in the linkage group R07 in purple turnip plants [46]. Moreover, a single dominant gene on C09, named BoPr, was reported to control anthocyanin pigmentation in leaves. The physical region of the C09 QTL was from 19,018,694 to 24,359,626 (5.3 Mb in internal length), which was different from our major QTLs for head anthocyanin contents. The major locus on C09 from 51,716,244 to 51,788,927 was included in this study. This indicates a population-specific inheritance modality for certain QTLs controlling anthocyanin biosynthesis in the Brassica species. Hence, the integrated approach developed here could be used for rapidly identifying the target QTLs, important genes, and/or alleles involved in the qualitative and quantitative traits of various crops. Diverse models of purple traits, related to anthocyanins, may come from different genetic

Discussion
Increasing the anthocyanin content in Brassica vegetables represents an important goal for nutriment breeding. This work discovered two important genomic regions harboring anthocyanin biosynthesis QTLs BoPur7.1 and Bopur9.1 on chromosomes 7 and 9, based on intra-specific broccoli mapping individuals, through whole-genome NGS-based, high-throughput QTL-seq. The two QTL-seq-derived Pur QTLs were subsequently verified by Hi-SNP analysis. In previous studies, a major locus was reported to control anthocyanin pigmentation. In purple cauliflower (B. oleracea var botrytis), a Pur gene, encoding the transcription factor R2R3 MYB, was isolated [43]. Broccoli flower head's purple sepal trait is regulated by a major loci and two minor loci on chromosome C01 [20]. The purple gene of non-heading Chinese cabbage is subject to a single dominant inheritance mode but does not follow the Mendel law [33]. Due to the considerable variations in anthocyanin content, a novel locus was mapped in the linkage group R07 in purple turnip plants [46]. Moreover, a single dominant gene on C09, named BoPr, was reported to control anthocyanin pigmentation in leaves. The physical region of the C09 QTL was from 19,018,694 to 24,359,626 (5.3 Mb in internal length), which was different from our major QTLs for head anthocyanin contents. The major locus on C09 from 51,716,244 to 51,788,927 was included in this study. This indicates a population-specific inheritance modality for certain QTLs controlling anthocyanin biosynthesis in the Brassica species. Hence, the integrated approach developed here could be used for rapidly identifying the target QTLs, important genes, and/or alleles involved in the qualitative and quantitative traits of various crops. Diverse models of purple traits, related to anthocyanins, may come from different genetic backgrounds. The purple gene in ornamental kale (B. oleracea L. var. acephala) shows a single dominant inheritance pattern on chromosome C09, and Bo9g058630 encoding dihydroflavonol 4-reductase (DFR) was considered to be a candidate gene [22].
Combining QTL-seq with Hi-SNP analysis, the 0.73 Mb QTL Bopur9.1 region [BoSNP37 (51,716,244 bp) to BoSNP38 (51,788,927 bp)], encompassing 14 genes, was detected on chromosome 9, which accounted for ∼28.19% of all phenotypic variations in anthocyanin biosynthesis. Among these genes, according to their respective annotations (Table 3), three genes, Bo9g174880, Bo9g174890, and Bo9g174900 were homologues of Arabidopsis F3'H, encoding flavanone 3 -hydroxylase. The BoF3'H gene has been suggested to have played a critical role in modifying plant coloration [45], which contained a coding sequence length of 1536 bp and encoded a protein of 511 amino acids with four exons. All of the 3 candidate genes became aligned to the different regions of BoF3'H. An increased expression of F3'H was found to be responsible for anthocyanin production in a number of anthocyaninaccumulating mutants. For example, the ectopic expression of apple MdF3'H can increase the production of flavonols and cyanidin-based pigments in the Arabidopsis tt7 mutant, under nitrogen pressure [47]. The VvF3'H gene was identified from grapevine, and its ectopic expression results in high accumulation levels of anthocyanin and flavonols in the petunia ht1 mutant line [48]. OsF3'H editing in the Heugseonchal or Sinmyungheugchal variety, using the CRISPR/Cas9 system, might be responsible for ocher seeds with lower anthocyanin contents than wild-type black rice plants [49]. Two copies of the F3'H gene were isolated in the barley genome and exhibited a tissue-specific expression pattern of anthocyanin synthesis [50]. Furthermore, nine SNPs and a 68-bp insertion, between the 3rd and 4rd exon of BoF3'H CDS sequences, were found, which may cause a gain-of-function mutation ( Figure 4A). The 68 bp InDel was in the open reading frame (between E3 and E4), as well as the T InDel in the 3 end. Therefore, a frame shift is expected, resulting in a putative truncated or extended protein. In addition, since most enzymes had conserved regions, anthocyanin discoloration might occur, as a result of BoF3'H enzyme inactivity, due to one of the two isoforms. In addition, the 6.92 Mb QTL Bopur7.1 region [BoSNP1 (36,784,249 bp) to BoSNP2 (43,705,152 bp)], encompassing 3 anthocyanin-related genes, was detected on chromosome 7, which accounted for ∼38.12% of all phenotypic variations in anthocyanin biosynthesis. Among these genes, BoANS2 and BoLED38 were activated at three head developmental stages of anthocyanin accumulation [12,13]. However, further experiments involving transformation are needed to verify whether the function of this gene is responsible for the purple head in broccoli.
Most of the agronomic traits are controlled by multiple QTLs. The traditional QTL finemapping method usually requires several generations of backcrossing, screening of a large population for recombinants and exhaustive field phenotyping, which is both time-consuming and labor-intensive [51]. Although the genotyping-by-sequencing (GBS) approach has been shown to be an efficient way to develop high-resolution genetic mapping [52], the genotyping of a large population with genome-wide markers is expensive. It is important to note that the QTL-seq approach has been successfully used to conduct a BSA analysis, as well as deployed for the mapping of a segregated population of homozygous parental lines with opposite phenotypes. Using QTL-seq, only two samples need to be sequenced. The sequence coverage is decided by genome size and complexity, but as an example for broccoli, sufficient data could be generated by sequencing each bulk to 30× coverage. Further, it's an effective method to identify markers most tightly linked to the trait.
In this study, two identified QTLs were distributed on chromosome 7 and 9, respectively. The purple intensity and size of the heading leaves had great variations among F2 individuals. The anthocyanin content of purple head BT 126 was also affected by the temperature condition. Therefore, environmental factors might have an important effect on anthocyanin synthesis. The genetic mechanism of anthocyanin biosynthesis is complex.
Candidate genes that control the purple trait may be different in diverse genetic backgrounds and may have different spatial and temporal expression patterns.
In summary, the above findings indicated that combining QTL-seq, Hi-SNP analysis and differential gene expression profiling might help identify candidate genes manipulating anthocyanin biosynthesis at major QTL intervals in broccoli. Bo9g174880, Bo9g174890, and Bo9g174900 were first selected as the strong candidate genes at the BoPur locus. The functional assessment of the candidate gene homologs, identified within the confidence intervals of both QTLs, would provide novel insights into the detailed mechanisms of anthocyanin biosynthesis in broccoli.