Genome-wide association study and gene network analyses reveal potential candidate genes for high night temperature tolerance in rice

Abstract

High night temperatures (HNT) are shown to significantly reduce rice (Oryza sativa L.) yield and quality. A better understanding of the genetic architecture of HNT tolerance will help rice breeders to develop varieties adapted to future warmer climates. In this study, a diverse indica rice panel displayed a wide range of phenotypic variability in yield and quality traits under control night (24 °C) and higher night (29 °C) temperatures. Genome-wide association analysis revealed 38 genetic loci associated across treatments (18 for control and 20 for HNT). Nineteen loci were detected with the relative changes in the traits between control and HNT. Positive phenotypic correlations and co-located genetic loci with previously cloned grain size genes revealed common genetic regulation between control and HNT, particularly grain size. Network-based predictive models prioritized 20 causal genes at the genetic loci based on known gene/s expression under HNT in rice. Our study provides important insights for future candidate gene validation and molecular marker development to enhance HNT tolerance in rice. Integrated physiological, genomic, and gene network-informed approaches indicate that the candidate genes for stay-green trait may be relevant to minimizing HNT-induced yield and quality losses during grain filling in rice by optimizing source-sink relationships.

Figure 1

Synchronization of the reproductive stage by stagger sowing (A) for phenotyping indica panel under HNT stress (B) in walk-in chambers (Exp. 1) and field-based heat tents (Exp. 2). DFF days to 50% flowering, S seeding, T transplanting, PI panicle initiation, FL flowering, GF grain filling, PM physiological maturity. Amplitude is calculated as a difference between average day temperature and average night temperature for each treatment separately. VPD vapor pressure deficit, RH relative humidity and SD standard deviation. Schematic representation of materials and methods followed to impose HNT stress is visualized in Supplementary Fig. S1.

Figure 2

Boxplots showing the phenotypic responses of traits (A, D yield; B, E 100-seed weight; C, F harvest index) under control and high night temperature stress in Exp. 1 (n = 173) and Exp. 2 (n = 206). In Exp. 1 (walk-in chambers; A-C), temperature treatments included control night temperature (CNT) and high night temperature (HNT) starting from panicle initiation to physiological maturity. In Exp. 2 (D-F), genotypes were exposed to control night (CNT) and high night temperature (HNT) with common day-time ambient temperature across treatments. The solid and dotted line represents the median and means of the data. The closed circles at the boundary show outliers at 90th percentile. The marker-based heritability (h²) was obtained in the Genomic Association and Prediction Integrated Tool. Different letters indicate statistically significant least square difference (LSD) for treatments at the level of p < 0.05.

Figure 3

Pearson’s correlation coefficients between the traits in walk-in chambers (173 genotypes in Exp. 1; A) field-based heat tents (206 genotypes in Exp. 2; B) and between experiments (Exp. 1 vs. Exp. 2, C). In experiment 1 (walk-in chambers), temperature treatments included control night temperature (CNT) and high night temperature (HNT), and in experiment 2, two different night temperature treatments i.e., control night (CNT) and high night temperature (HNT) were imposed with common day-time ambient temperature across treatments. YPP- grain yield (g per plant); HWT-100-seed weight (g); HI Harvest index and CHALK chalkiness (%). *p < 0.05, **p < 0.01, ***p < 0.001 indicate significant correlations among traits.

Figure 4

Rice grain width (mm, A) and grain length (mm, B) in response to control night (CNT; 24 °C, covered tents) and high night temperatures (HNT; 29 °C, covered and heated) under common day-time temperature under field conditions (Exp. 2; Supplementary Fig. S1). Manhattan plots of genome-wide association results for rice mature grain width (C–E) and length (F–H) in the study. The marker-based heritability (h²) was obtained in the Genomic Association and Prediction Integrated Tool. The previously characterized major genes controlling grain shape traits (yellow bar) such as the grain width (GW5, C,D) on chromosome 5 and grain length (GS3, F,G) genes on chromosome 3 are labeled. r—Pearson’s correlation coefficient between treatments. ***p < 0.001 indicate significant correlation of traits between CNT and HNT.

Figure 5

Gene regulatory networks are made up of top candidate genes from the GWAS of absolute (A) and relative trait values (B) and candidate genes with strong transcription factor interaction levels (> 200). A list of transcription factors for each candidate is given in Supplementary Table S5.