Genome-Wide Association Study (GWAS) analysis was used in current project to identify regulators for fiber (FC), starch (SC), and amylose (AC) contents in rice grains using 293 rice accessions, including both Indica and Japonica subpopulations, harvested from 2020 to 2022. The GWASs were performed using three different models, including the General Linear Model (GLM), Mixed Linear Model (MLM) and Fixed and random model Circulating Probability Unification (FarmCPU). The phenotype results revealed that Japonica exhibited higher FC but lower SC and AC compared to Indica. Negative Pearson correlations between SC and FC across populations and years indicated a potential “switch” mechanism governing the allocation of sugars between starch and cell wall metabolic pathways. The model comparison suggested that GLM and FarmCPU outperformed MLM for FC and SC, while MLM and FarmCPU were more suitable for AC. GWASs identified multiple loci associated with these three interested traits. Interestingly, two stable associations on Chromosome 6 were identified for AC across three years, whereas one common association on Chr4 were identified for both FC and SC in Indica in 2020, and three associations (one on Chr1 and two on Chr8) were detected for these two carbohydrates in Total population in 2021 or 2022. Through gene-based haplotype and bioinformatic analysis, two promising candidate genes were selected for AC, including a STARCH SYNTHASE (Wx) for synthesizing amylose and a GLYCOSOL HYDROLASE FAMILY 17 (GH17) potentially involved in β-glucan degradation, and three candidates were chosen for both FC and SC, including ETHYLENE-RESPONSIVE TRANSCRIPTION FACTOR 23 (OsERF23), PLASTIDIC GLUCOSE TRANSPORTER 1 (OspGlcT1), and α-CARBONIC ANHYDRASE 4 (OsαCA4).
To validate the functionality of these promising candidates in regulation the FC and SC, overexpression lines of OsERF23 were generated, and Crispr/Cas9 mutants were obtained for OspGlcT1 and OsαCA4. Preliminary phenotype results indicated that OsERF23 suppressed plant height, panicle length, and tiller numbers. Notably, when the leaf OsERF23 expression level exceeded 100-fold, plant height, panicle length, and tiller numbers plateaued, while grain length and width decreased, and FC increased at the cost of SC. In some severe cases, we observed empty grains or reduced mature grain, suggesting possible roles of OsERF23 on grain development and carbon allocation. For OspGlcT1 and OsαCA4, no visible changes in plant and grain morphology were found in the corresponding Crispr/cas9 mutants. However, variable FC and SC were observed in individual mutants. Next generation lines, with a larger sample size of both mutants and wild type, are needed to understand the role of the OspGlcT1 and OsαCA4 genes in carbon allocation in rice grains.
Overall, these findings in our study provide valuable insights into the genetic and molecular mechanisms underlying carbon allocation in rice grains, offering potential applications for crop improvement and agricultural practices.