Transcriptome-wide association analyses reveal the impact of regulatory variants on rice panicle architecture and causal gene regulatory networks

Abstract

Abstract Panicle architecture is a key determinant of rice grain yield and is mainly determined at the 1-2 mm young panicle stage. Here, we investigated the transcriptome of the 1-2 mm young panicles from 275 rice varieties and identified thousands of genes whose expression levels were associated with panicle traits. Multimodel association studies suggested that many small-effect genetic loci determine spikelet per panicle (SPP) by regulating the expression of genes associated with panicle traits. We found that alleles at cis-expression quantitative trait loci of SPP-associated genes underwent positive selection, with a strong preference for alleles increasing SPP. We further developed a method that integrates the associations of cis- and trans-expression components of genes with traits to identify causal genes at even small-effect loci and construct regulatory networks. We identified 36 putative causal genes of SPP, including SDT (MIR156j) and OsMADS17, and inferred that OsMADS17 regulates SDT expression, which was experimentally validated. Our study reveals the impact of regulatory variants on rice panicle architecture and provides new insights into the gene regulatory networks of panicle traits.