Rice is an important food crop, providing staple food for more than half of the world’s population. Panicle traits are the main factors affecting rice yield. Discover the elite haplotype of the panicle regulation gene, and provide important germplasm and gene resources for pyramiding breeding.

In this study, recombinant inbred lines (RILs) derived from a cross between SN265 and R99 were re-sequenced through high-throughput sequencing. QTL analysis and candidate gene identification were conducted on the grain number on the primary branch, the grain number on the secondary branch, and the grain shape. The sequences of candidate genes were compared using the long-read sequence assemblies of SN265 and R99. The combination of candidate genes that can maximize grain yield was selected among RILs. Finally, the super rice variety SN265 was improved using CRISPR/Cas9 gene editing technology.

The R99 had significantly more grain number per panicle and grain number on the secondary branch, whereas SN265 had significantly more grain number on the primary branch. The grain of R99 is slender, and the grain of SN265 is short and round. The RILs were sequenced with approximately 6.25-fold depth. For parent lines, 30.0-fold depth and 32.0-fold depth data were generated for R99 and SN265, respectively. Subsequently, a bin map was constructed by 1456445 high-quality SNPs. The genetic map containing 3 569 recombinant blocks, with an average length of 58.17 kb. The QTL analysis detected a QTL on Chr.9 for grain number per panicle and grain number on both primary and secondary branch, a QTL on Chr.1 for grain number per panicle and grain number on the secondary branch, a QTL on Chr.5 for grain shape. The candidate gene prediction and sequence comparison showed that DEP1 regulated the grain number on both primary and secondary branches of rice, Gn1a mainly regulated the grain number on secondary branches of rice, and qSW5 mainly regulated the grain shape. The yield of the combination of Gn1a^R99/DEP1^SN265/qSW5^SN265 alleles showed an advantage in yield performance among the RILs. We further conducted a molecular design breeding to SN265 by knocking out the Gn1a locus using CRISPR/Ca9 gene editing technology, and the grain number per panicle of the transgenic plants increased significantly compared to that of SN265.

This study used RILs derived from a XI/GJ cross and high-throughput sequencing technology to conduct QTL analysis of rice panicle traits, revealed the effects of DEP1, Gn1a, and qSW5 on grain number per panicle and grain shape, and clarified that Gn1a^SN265/ DEP1^R99/qSW5^R99 was the best gene combination in RILs. The yield per plant was further improved by knocking out the Gn1a locus of SN265. This study provided important germplasm and gene resources for pyramiding breeding with elite alleles.

The F₁ hybrid sterility between XI/indica and GJ/japonica severely hinders the utilization of hybrid advantage between subspecies. Exploring the genetic mechanism and identifying new regulatory genes for XI/GJ hybrid sterility will provide theoretical basis for promoting genetic improvement of XI/GJ hybrid seed setting rate.

A series of stable genetic recombination inbred lines (RILs) containing 95 plant lines were derived from the cross between XI variety Habataki and GJ variety Sasanishiki after 10 generations inbred using single seed descent method. High throughput sequencing was performed on both parents and RILs on the Illumina platform, and the distribution of Habataki pedigree in RILs was analyzed at the whole genome level. The segregation distortion regions were identified, and hybrid sterile related gene loci were screened within the segregation distortion regions, then identified candidate genes through sequence alignment comparison. The targeted gene was knockout to verify the function using CRISPR gene editing technology.

The hybrid F₁ plants derived from the cross between Habataki and Sasanishiki showed significant heterosis in panicles, grains per panicle, and thousand grain weight, but its seed setting rate significantly decreased. I₂-KI microscopy revealed a significant decrease in F₁ pollen fertility. High throughput sequencing of the entire genome of RILs revealed significant segregation distortion on Chr.1, Chr.3, Chr.5, Chr.6, Chr.7, and Chr.12, indicating that the genotype in this region tends towards the Habataki. Sequence alignment comparison revealed that Sc, S5, and HSA1 are target genes for the segregation distortion on Chr.3, Chr.6, and Chr.12. The CRISPR gene editing mutants with a knock-out Sc-Haba-3 allele in Habataki successfully improved the pollen fertility and seed setting rate of F₁ hybrid with Sasanishiki. A complex structural variation was found between Sasanishiki and Habataki in the segregation distortion of Chr.1. A 24.7 kb segment containing 4 predicted genes in the Sasanishiki genome was replaced by a 64.8 kb segment containing 10 predicted genes in Habataki, the structural variation may involve in controlling the hybrid sterility of XI and GJ cross.

This study detected multiple XI/GJ hybrid infertility related loci, and successfully improved F₁ fertility by using CRISPR gene editing to knock out multiple copies of Sc in Habataki, locking in the target gene in the Sd region of Chr.1.