Comparative metabolomic and transcriptomic analysis reveals a coexpression network of the carotenoid metabolism pathway in the panicle of Setaria italica

BACKGROUND: The grains of foxtail millet are enriched in carotenoids, which endow this plant with a yellow color and extremely high nutritional value. However, the underlying molecular regulation mechanism and gene coexpression network remain unclear. METHODS: The carotenoid species and content were detected by HPLC for two foxtail millet varieties at three panicle development stages. Based on a homologous sequence BLAST analysis, these genes related to carotenoid metabolism were identified from the foxtail millet genome database. The conserved protein domains, chromosome locations, gene structures and phylogenetic trees were analyzed using bioinformatics tools. RNA-seq was performed for these samples to identify differentially expressed genes (DEGs). A Pearson correlation analysis was performed between the expression of genes related to carotenoid metabolism and the content of carotenoid metabolites. Furthermore, the expression levels of the key DEGs were verified by qRT-PCR. The gene coexpression network was constructed by a weighted gene coexpression network analysis (WGCNA). RESULT: The major carotenoid metabolites in the panicles of DHD and JG21 were lutein and β-carotene. These carotenoid metabolite contents sharply decreased during the panicle development stage. The lutein and β-carotene contents were highest at the S1 stage of DHD, with values of 11.474 μg /100 mg and 12.524 μg /100 mg, respectively. Fifty-four genes related to carotenoid metabolism were identified in the foxtail millet genome. Cis-acting element analysis showed that these gene promoters mainly contain ‘plant hormone’, ‘drought stress resistance’, ‘MYB binding site’, ‘endosperm specific’ and ‘seed specific’ cis-acting elements and especially the ‘light-responsive’ and ‘ABA-responsive’ elements. In the carotenoid metabolic pathways, SiHDS, SiHMGS3, SiPDS and SiNCED1 were more highly expressed in the panicle of foxtail millet. The expression of SiCMT, SiAACT3, SiPSY1, SiZEP1/2, and SiCCD8c/8d was significantly correlated with the lutein content. The expression of SiCMT, SiHDR, SiIDI2, SiAACT3, SiPSY1, and SiZEP1/2 was significantly correlated with the content of β-carotene. WGCNA showed that the coral module was highly correlated with lutein and β-carotene, and 13 structural genes from the carotenoid biosynthetic pathway were identified. Network visualization revealed 25 intramodular hub genes that putatively control carotenoid metabolism. CONCLUSION: Based on the integrative analysis of the transcriptomics and carotenoid metabonomics, we found that DEGs related to carotenoid metabolism had a stronger correlation with the key carotenoid metabolite content. The correlation analysis and WGCNA identified and predicted the gene regulation network related to carotenoid metabolism. These results lay the foundation for exploring the key target genes regulating carotenoid metabolism flux in the panicle of foxtail millet. We hope that these target genes could be used to genetically modify millet to enhance the carotenoid content in the future. SUPPLEMENTARY INFORMATION: The online version contains supplementary material available at 10.1186/s12870-022-03467-2.