The identification of grain size genes by RapMap reveals directional selection during rice domestication

Natural variation in crops provides a crucial genetic basis for domestication and improvement. Agronomically important traits are often controlled by multiple quantitative trait loci (QTLs). Identifying the causal genes (QTGs) underlying these QTLs is challenging, with traditional map-based cloning methods being time-consuming and laborious. Bi-parental populations (F2, RIL, DH, BC) are commonly used for QTL mapping, but they often require complex population construction and only explain a fraction of phenotypic variation. Multi-parental populations (like MAGIC and NAM) offer advantages but are also complex and time-intensive to develop. Genome-wide association studies (GWAS) are an alternative, but they require additional experiments to pinpoint causal genes. Grain size in rice, a vital trait for yield and quality, is controlled by multiple QTGs, but the regulatory relationships between them remain unclear. Understanding the molecular basis of natural variation in domestication traits is crucial for bridging molecular analysis and domestication studies. This study aims to overcome the limitations of existing methods by introducing RapMap, a novel method for rapidly cloning QTLs.

The paper reviews existing methods for QTL mapping and cloning, highlighting their limitations. Traditional bi-parental mapping populations are time-consuming and labor-intensive, often failing to capture the full phenotypic variation. Multi-parental populations like MAGIC and NAM offer higher resolution but are complex to develop. GWAS provides a high-throughput approach but identifying causal genes from GWAS results often requires additional experiments. The paper emphasizes the need for a more efficient method, particularly for complex traits like grain size in rice, where numerous QTGs with intricate regulatory relationships are involved. Existing research on rice grain size has identified several QTGs, but the molecular mechanisms and the role of natural variation remain incompletely understood.

RapMap employs F2 gradient populations (F2GPs) constructed from minor-phenotypic-difference accessions, minimizing the number of segregating loci. The phenotypic difference index (PDI), the ratio of phenotypic difference between parents to the low-value parent's phenotype, is used to define the extent of minor phenotypic difference. A series of F2GPs is constructed to increase the chances of discovering more QTGs. Bulked segregant analysis (BSA) is performed using two phenotype-extreme DNA pools to identify candidate QTL regions. The 'co-segregation standard,' a newly defined principle based on 12 single-locus genetic models, ensures simultaneous detection, confirmation, and isolation of near-isogenic line-like lines (NIL-LLs) for each QTL. The co-segregation standard requires that the two homozygous genotypes of a QTL co-segregate with their corresponding phenotypes regardless of the heterozygous genotypes. This three-in-one framework significantly improves efficiency. RICE6K SNP arrays and whole-genome sequencing are used for genotyping. InDel markers are designed for verification and fine-mapping. A rapid DNA extraction technique and KASP genotyping are used for high-throughput screening of recombinants from NIL-LLs. CRISPR/Cas9 and complementation transformations are employed to validate the function of candidate genes. Multiple linear regression analysis assesses the genetic contributions of identified genes to grain size and shape. Population genetic analysis using a large germplasm collection investigates domestication signatures.

RapMap successfully cloned eight rice grain-size genes within three years, including six known genes (GS3, GL7, GS2, GW5, GW8, GW7), two novel GS3 alleles (GS3-5, GS3-6), and two novel genes (GL1, GW5.1). These genes explained 75% of the total grain shape variation. Four grain-length genes (GL1, GS2, GS3, GL7) explained 67.4% of the phenotypic variation in grain length, while four grain-width genes (GW5, GW5.1, GW7, GW8) explained 66.8% of the phenotypic variation in grain width. The eight genes collectively explained 77.2% of the variation in grain shape. The prediction model based on the functional variations of these eight genes showed high predictive power for grain length (r = 0.82), grain width (r = 0.79), and grain shape (r = 0.87). Allele frequency analysis in a large germplasm collection revealed directional selection of long and slender grain alleles in indica rice during domestication, with major genes (GS3, GW5, GL7/GW7) showing stronger selection. Selection sweeps were identified for GS3, GW5, and GW7/GL7 in cultivars, landraces, and wild rice, indicating strong selection pressure. Strong positive correlations were observed between phenotypic effect size, association significance, selection strength, and nucleotide variations. RapMap's efficiency and ability to identify both major and minor effect genes were highlighted.

RapMap provides a significant advancement over existing QTL mapping and cloning methods. Its efficiency stems from the use of F2GPs, the co-segregation standard, and a streamlined three-in-one framework. The identification of both major and minor effect genes demonstrates the method's ability to reveal a more comprehensive picture of genetic architecture compared to traditional bi-parental mapping or GWAS approaches. The findings concerning directional selection during rice domestication highlight the power of RapMap in uncovering the evolutionary history of key traits. The high predictive power of the model suggests the potential of using these identified genes in genomic breeding programs. The limitations of GWAS and the complexity of MAGIC and NAM populations are contrasted with the simplicity and speed of RapMap. The authors emphasize the broader applicability of RapMap to other crops.

RapMap represents a significant improvement in QTL mapping and cloning technology, showcasing its high efficiency and ability to identify both major and minor effect genes. The successful application in rice suggests broad applicability across various crops. Future research should focus on expanding the application of RapMap to other traits and crops and investigating the remaining 25% of heritability not explained by the identified genes, potentially involving epigenetic factors or gene interactions. The findings significantly advance our understanding of rice domestication and provide valuable resources for genomic breeding.

While RapMap significantly improves QTL mapping efficiency, some limitations remain. The sample size, although large for a QTL mapping study, might not fully capture rare variants. Epigenetic modifications and structural variants were not directly investigated, potentially accounting for some unexplained heritability. The study focused on indica rice, limiting the generalizability of the findings to other subspecies. Future research could explore larger populations, incorporate epigenetic analyses, and expand to diverse crop species.