Application of different color evaluation methods in grain
Grain color is an important quality trait, and Method I is a common phenotypic evaluation method. Method I is a widely used and simple phenotypic evaluation. It has been frequently applied to evaluate the color of seeds or fruits, including foxtail millet(Xie et al. 2021), cucumber(Kishor et al. 2021), wheat(Lin et al. 2016), and other plants. Xie et al. classified the hull color in foxtail millet into two categories, red and golden yellow, using Method I for QTL detection(Xie et al. 2021). Method II is a more detailed identification of hull color relative to Method I. Method I and Method II have the characteristics of possessing simple operation, rapid, and low cost that make them more appropriate for quick evaluation of phenotypes in large populations. However, the accuracy of Method I and Method II is affected by environmental and anthropogenic factors. Method III: LAB color space was used for phenotypic evaluation that it is generating colors as similar to as human visual system(Alam et al. 2014; Philipp and Rath 2002). This method has been widely used in black bean(Cichy et al. 2014), cherry(Sooriyapathirana et al. 2010), grape(Underhill et al. 2020), etc. Method IV: RGB color space was used for phenotypic evaluation. It has been widely used in sesame(Zhang et al. 2013), peanut(Zhang et al. 2021), grape(Underhill et al. 2020), etc. However, compared to LAB color space, RGB color space cannot simulate human perception color perception well(Vanrell et al. 2004). Existing research generally selects only one method for phenotypic evaluation and the application of many methods for phenotypic evaluation simultaneously has not been reported in foxtail millet. The lack of efficient methods for evaluating hull color in foxtail millet has impeded further genetic experiments.
Comparison the QTLs identified in this study with previous ones
To identify QTLs associated with hull color in foxtail millet, four phenotypic evaluation methods were utilized in all four environments. A total of 36 QTLs associated with hull color was detected. Among these QTLs, 13 QTLs were repeatedly mapped in multiple environments, with four major QTLs (qHC1.1, qHC1.2, qHC9.1 and qHC9.5) and nine minor QTLs (qHC1.3, qHC3.3, qHC3.4, qHC3.6, qHC4.4, qHC5.2, qHC8.2, qHC8.3 and qHC9.4) explaining 1.09% to 69.63% of the phenotypic variation. Two major QTLs qHC1.2 and qHC9.5 could be detected repeatedly in all methods. Among them, qH1.2 could explain 8.89% to 69.63% of the phenotypic variations. qHC9.5 could explain 4.43% to 32.08% of the phenotypic variations. These two QTLs explained a relatively high percentage of the phenotypic variation, which suggests that the genetic factors might be the major source of phenotypic variation. In addition, the broad-sense heritability for all the traits ranged from 0.76 to 0.97 in all four environments, indicating that genetic factors played an important role in determining hull color.
Previous studies also identified several QTLs associated with hull color in foxtail millet. In 2013, with multiple associated variants from a genome-wide association study, Jia et al. identified seven QTLs for foxtail millet hull color. Two of them, qHC1.2 and qHC9.5, are consistent with our findings(Jia et al. 2013). Sihc1 was discovered at the end of chromosome 6 based on findings from two preceding studies. However, these specific QTLs did not align with the QTLs identified in this study. In our study, two QTLs namely, qHC1.1 and qHC1.2 were found to be in agreement with qHC1 and qHC2 previously reported by Xie et al.(Xie et al. 2021). Interestingly, both this study and the previous publication used RILs derived from Yugu 18 × Hongjiugu crosses and constructed a high-density bin map for QTL mapping. However, Guo’s research performed QTL mapping for hull color by Method III, while our study performed QTL mapping for hull color by four phenotypic evaluation methods. Notably, six QTLs, namely, qHC1.2, qHC1.3, qHC3.3, qHC9.1, qHC9.4, and qHC9.5, were consistent from previously reported in Guo’s research, however, there are six QTLs identified in this study are novel ones, namely qHC3.4, qHC3.6, qHC4.4, qHC5.2, qHC8.2 and qHC8.3. This not only confirms that our results are reliable but also implies that multiple phenotypic evaluation methods are effective in QTL mapping for hull color.
Candidate genes identified for hull color
To identify candidate genes within the stable QTLs region, we performed RNA-seq on the hull of DY and LY revealing DEGs that may be related to hull color. Previous studies suggested that it was found that lignin metabolism and flavonoids are important factors that affect hull color. Of the two biosynthetic pathway genes, CAD (LOC_Os02g09490) and CHI (LOC_Os03g60509) have been identified as crucial for the biosynthesis of rice hull color. Correspondingly, we found that the expression level of CAD (Seita.1G057300) in the LY hull exceeded that of DY for three periods, while the expression level of CHI (Seita.9G034700) in the LY hull was lower than that of DY at S3. In addition, the cytochrome P450 gene Seita.1G157600 is located in the qHC1.3 candidate interval and is involved in the phenylpropanoid metabolic pathway and flavonoid biosynthesis pathway.
Research on rice hulls has demonstrated that changes in hull color are primarily caused by the accumulation of flavonoids and anthocyanins, as well as defects in lignin synthesis within the organism(Wang et al. 2020). Lignin is one of the important substances in the phenylpropanoid metabolic pathway in plants(Dong and Lin 2021). Thus, we examined the phenylpropanoid and flavonoid metabolic pathways that were significantly enriched by genes expressed differentially. Phenylalanine ammonia-lyase (PAL) catalyzes the formation of cinnamic acid from Phenylalanine, cinnamic acid-4-hydroxylase (C4H) catalyzes the formation of p-Coumaric acid from cinnamic acid(Humphreys and Chapple 2002). The C4H gene (Seita.1G157600) exhibited down-regulation at S1, potentially acting as a regulatory factor at this stage and impacting the synthesis of lignin and flavonoids. GH2 (LOC_Os02g09490) is the only subA subfamily involved in constitutive lignin biosynthesis. It is the main CAD enzyme for the synthesis of coniferyl alcohol and sinapyl alcohol precursors in rice lignification tissues. The hull and stem segments of the gh2 mutant exhibited a golden-yellow coloration attributable to a deficiency in lignin production(Zhang et al. 2006). As a result of the substitution of T with C in exon four of RBH1, the expression of genes regulating lignin and flavonoid metabolism was significantly up-regulated in the hull of the rbh1 mutant. This further confirms that GH2/RBH1 affects hull color by regulating lignin and flavonoid metabolism(Wang et al. 2020). Through phylogenetic analysis and sequence alignment, Seita.1G057300 in foxtail millet showed high homology with GH2 (LOC_Os02g09490) and included a conserved PLN02514 domain. This gene demonstrated significant up-regulation across all three developmental stages (S1-S3) and appears to be an essential player in lignin synthesis and metabolism throughout these stages of development. Another gene, Seita.9G034700, is involved in the flavonoid biosynthesis pathway and is homologous to the GH1 gene (LOC_Os03g60509) in rice. This gene belongs to the Chalcone_3 superfamily. GH1 is a key gene involved in the flavonoid pathway in rice, which encodes a relatively conserved protein of chalcone isomerase. The loss of OsCHI expression in the gh1 mutant resulted in an increase in flavonoid content and a golden yellow hull color(Hong et al. 2012). Seita.9G034700 exhibited significant upregulation during S1, potentially playing a crucial role in flavonoid biosynthesis and consequently affecting anthocyanin synthesis.
Seita.1G057300 sequence analysis in the CDS region revealed four SNPs were detected in DY compared with LY, and one SNP (A to G), located at the third exon, resulted in an amino acid change from isoleucine to valine. Seita.1G057300 in the sequence analysis results of ten foxtail millet varieties are consistent with the sequence analysis between DY and LY. Non-synonymous mutations in the CDS region may result in differential expression of Seita.1G057300 in foxtail millet varieties with distinct hull colors. Therefore, Seita.1G057300 was considered the most possible functional gene underlying qHC1.2.