Genome-Wide Association Study and Candidate Gene Identification for Grain Number per Spike in Dryland Wheat (Triticum aestivum L.)

Wheat (Triticum aestivum L.) ranks among the most vital cereal crops globally, making the enhancement of its yield crucial for ensuring food security. The grain number per spike (GNS) serves as a critical determinant of wheat yield. Understanding the genetic architecture of GNS and identifying key candidate genes will promote genetic improvement and enable marker-assisted selection for this trait.

A panel of 123 wheat accessions was phenotyped for GNS across multiple drought-prone environments. Genome-wide association studies (GWAS) were conducted utilizing a mixed linear model (MLM) based on genotyping data derived from a 35K SNP array. For marker-trait associations (MTAs) significantly correlated with GNS, haplotype analysis was performed, Kompetitive Allele-Specific PCR (KASP) markers were developed, and multi-environment phenotypic validation was executed. Candidate genes were identified through public RNA-seq data to screen for genes with high expression levels in spike tissues. This was followed by validation using quantitative real-time reverse transcription PCR (qRT-PCR) and functional annotation based on homology and previously reported functional information.

GNS exhibited coefficients of variation between 17.18% and 22.48% across different environments, indicating a significant environmental impact. The GWAS identified 47 significant MTAs, with the highest concentration of 9 MTAs located on chromosome 5A. Notably, the marker AX-94445381 on chromosome 5A was consistently detected in two environments, demonstrating a stable genetic association. Haplotype analysis categorized this marker into two distinct haplotypes (HapⅠ and HapⅡ). Validation of the KASP marker indicated that accessions carrying HapⅡ displayed significantly higher GNS across three environments compared to those with HapⅠ. An analysis of the favorable haplotype AX-94445381-HapⅡ, utilizing publicly available resequencing data, revealed that its frequency was lower in modern cultivars (30.1%) than in landraces (40.0%), suggesting that this haplotype has not been fully exploited in wheat breeding programs. Candidate gene screening within the 1 Mb region surrounding AX-94445381, in conjunction with RNA-seq data, identified genes that are highly expressed specifically in spike tissue. Validation through qRT-PCR and homology analysis with rice orthologs identified three candidate genes directly involved in spike development, which encode a UDP-glycosyltransferase, a MADS-box transcription factor, and a WD repeat-containing protein, respectively.

A major stable locus, AX-94445381, associated with GNS has been identified. The advantageous haplotype AX-94445381-HapⅡ was shown to markedly enhance GNS, underscoring its potential for further exploitation and application in wheat breeding.