Introduction

China is a pioneer in the domestication and cultivation of rice, currently holding the position of the world's largest rice producer and consumer (Muthayya et al. 2014). Remarkable advancements in rice productivity have characterized China's trajectory over the past century, witnessing a substantial increase in yields from 1.9 t/hm2 in 1949 to 7.0 t/hm2 in 2018 (http://faostat.fao.org/). This noteworthy growth is primarily attributed to the adoption of semi-dwarf and hybrid rice varieties coupled with the rapid proliferation of high-yielding temperate Geng/Japonica (GJ) cultivars. In national trials evaluating new rice varieties in northeast China, the average yield of emerging temperate GJ varieties demonstrated an upward trend, escalating from 8.3 in 2004 to 9.0 t/hm2 in 2018 (Fei et al. 2020). Despite these impressive achievements in GJ rice production, challenges persist in GJ rice breeding.

GJ rice breeding has proven effective in enhancing grain yield, although the quality of grains often remains at a moderate level. Grain shape is a crucial breeding target because of its influence on both yield potential and appearance quality. Although many GJ rice varieties exhibit short and round grains, consumer preferences lean towards slender shapes given their superior appearance quality. Recent studies have revealed that the determination of rice grain shape involves multiple signaling pathways, including ubiquitination-mediated proteasomal degradation, phytohormones (cytokinin, brassinosteroid, and auxin), and G-protein signaling pathways (Ren et al. 2023). Among these pathways, GS3 is pivotal, representing the first cloned QTL encoding a rice G-protein γ subunit with a negative regulatory role in grain length (Fan et al. 2006; Takano-Kai et al. 2009). The C/A SNP in the 2nd exon of GS3 is a functional mutation that leads to elongated grain shapes, and its prevalence is widespread in rice germplasms (Fan et al. 2009; Wang et al. 2011). Consequently, manipulating GS3 presents a viable avenue for elongating the grains of the GJ varieties.

Another challenge confronting GJ breeding is climate change, particularly global warming, which poses a severe risk to agriculture worldwide. Reports from the Intergovernmental Panel on Climate Change (IPCC) in 2021 and 2022 underscore that global temperatures have surged by 1.1 °C compared to the pre-industrialization era (1850–1900), with projections anticipating a further increase exceeding 1.5 °C in the next two decades (Deliang et al. 2021; Rawshan et al. 2022). Notably, the average temperature in China increased by 0.24 °C per decade from 1951 to 2020, surpassing the global average warming rate. Our recent research revealed a pronounced elevation of 0.48 ℃ in the daily average temperatures over the past two decades in the central rice region of Liaoning Province, a rate significantly surpassing the national average (unpublished data). This increasing temperature trend indicates that the Northeast region is one of the areas most profoundly affected by climate change in China and globally, imposing substantial constraints on rice growth, development, yield, and quality. Therefore, exploring gene resources related to heat tolerance in GJ rice holds both theoretical significance and practical application value, providing a means for GJ rice to adapt to the challenges imposed by climate change. Investigation of the molecular mechanisms underlying thermotolerance in African rice identified a QTL associated with thermotolerance, attributed to natural variations in the G-protein γ subunits GS3/TT2. Upon heat stress, GS3 facilitates cytosolic Ca2+ elevation and fosters downstream interactions between calmodulin and transcription factor SCT1. This interaction suppressed OsWR2 gene expression and diminished cuticular wax content, resulting in increased thermosensitivity. In the absence of GS3, the blockade of Ca2+ accumulation and CaM–SCT1 interaction promotes the deposition of cuticular wax, thereby endowing plants with enhanced thermotolerance (Kan et al. 2022). Hence, the utilization of a knockout mutant of GS3 holds promise for improving the thermotolerance of GJ rice varieties.

The third challenge involves the cultivation of GJ varieties in alkaline salt lands. According to 2015 research conducted by the Food and Agriculture Organization (FAO), more than 1 billion hectares of land worldwide are affected by salt, with approximately 60% of this area categorized as sodic due to elevated pH levels resulting from the presence of sodium bicarbonate (NaHCO3) and sodium carbonate (Na2CO3). In China, saline-alkali land spans approximately 100 million hectares and is primarily situated in the northern provinces along the Yangtze River. Effective utilization of marginal lands is imperative for augmenting annual crop yields, given the limited availability of arable land. In rice plants, GS3 interacts with aquaporin PIP2s and hinders its hydrogen peroxide (H2O2) export function, thereby causing an excessive accumulation of reactive oxygen species (ROS) and rendering plants susceptible to alkalinity. The non-functional gs3 protein fails to inhibit PIP2s, consequently enhancing alkaline tolerance (Zhang et al. 2023). This suggests that the non-functional gs3 allele not only improves grain yield and thermotolerance but also enhances tolerance to alkaline soil. Such enhancements have the potential to expand the suitable areas for cereal cultivation.

Therefore, the concurrent improvement in grain shape, thermotolerance, and alkaline tolerance through the manipulation of Gγ protein in rice is theoretically feasible. To verify this hypothesis, we selected seven GJ varieties with high yield performance, which are widely distributed, as subjects for modification. The GS3 locus was knocked out using CRISPR/Cas9 gene-editing technology. Grain shape, grain quality, thermotolerance, and tolerance to alkali were investigated. Downstream genes of GS3 were identified using transcriptome sequencing.

Materials and methods

GWAS and haplotype assay

The grain length data for 529 diverse O. sativa accessions were obtained from a previous study (Xie et al. 2015). A mixed-model approach was employed, utilizing factored spectrally transformed linear mixed models (FaST-LMM) (Lippert et al. 2011). Comprehensive details regarding GWAS are available elsewhere (Yang et al. 2014). Haplotype analysis of GS3 utilized sequencing data from 4726 accessions accessed through RiceVarMap2 (https://ricevarmap.ncpgr.cn/). Phenotype data for haplotype analysis were sourced from a previous study (Yang et al. 2014).

Plant materials

We chose six interrelated modern Chinese temperate GJ varieties and one related Japanese GJ varieties to be the genetic background of the present study. All seven varieties were in large-scale promotion. The Japanese GJ variety, Sasanishiki (Sasa), developed in 1963, was cultivated in the Tohoku region in Japan and gained status as high-quality rice. Toyonishiki was derived from the cross between Sasa and another Japanese GJ variety Aoyu329, and subsequently introduced to China and served as a backbone parent for a long time. A series of high-yield varieties were bred using Toyonishiki as a parental line, such as Shennong265 (SN265), the first released ‘super rice variety’ by the Chinese Ministry of Agriculture, and Liaoxing1 (LX1), the variety with the largest promotion area in Liaoning Province in China. The other varieties used in the present study, such as Yanfeng47 (YF47), Beigeng1501 (BG1501), Beigeng1 (BG1), and Beigeng (BG1705) were bred using SN265 as a parental line. The pedigree relationship among the seven varieties is demonstrated in Fig. S1.

Vector construction and plant transformation

For CRISPR/Cas9 gene editing, we followed the vector construction protocol described by Li et al. (2017). A targeting sequence, including the PAM sequence (23 bp), was selected within the first exon of the GS3 gene. To ensure the specificity of the targeting sequence, a BLAST search of the rice genome was conducted using the NCBI BLAST tool (http://blast.ncbi.nlm.nih.gov/Blast.cgi). The rice transformation procedures followed the guidelines provided by Hsu et al. (2013). Genomic DNA extracted from the transformants was sequenced to identify the mutants. The resulting PCR products (200–500 bp) were sequenced and analyzed using the Degenerate Sequence Decoding method (Ma et al. 2015).

Yield components and quality investigation

Wild-type varieties and their GS3 mutants were cultivated in the experimental paddy field of Shenyang Agricultural University, Shenyang, China (N41°, E123°). Planting commenced on April 24th in the seedling nursery, with transplantation occurring per hill on May 23rd. The plant materials were arranged in a randomized block design, with three replicates. Each plot, measuring 5.4 m2, accommodated 120 plants at 30 cm × 15 cm intervals. The cultivation methods and field management followed previously established procedures (Li et al. 2018). Field examination plants were harvested 45 d after heading, with a total of 20 plants from the middle rows harvested for each line. The yield components and quality measurements followed the protocols outlined in our previous study (Li et al. 2018).

Thermotolerance and tolerance to alkali soil evaluation

Seeds designated for thermotolerance and alkali-soil tolerance evaluations were harvested simultaneously and stored under identical conditions. To mitigate dehydration stress associated with high temperatures, high relative humidity (RH > 90%) was maintained during the heat treatment. Twelve-day-old seedlings, at the two-leaf stage and grown in a hydroponic culture solution4, underwent a 42 °C treatment for the specified duration, followed by a week of recovery under normal conditions (28 °C). The survival rate was subsequently calculated. To assess alkali-soil tolerance, 75 mM mixed alkali (NaHCO3:Na2CO3 at a molar ratio of 5:1) was applied to the soil. The relative survival rate after exposure to 75 mM alkali stress for three weeks was recorded. Additional details can be found in previous studies (Kan et al. 2022; Zhang et al. 2023).

RNA-seq and bioinformatic analysis

A transcriptome assay involving wild-type varieties and their GS3 mutants was performed to elucidate the downstream genes of GS3, particularly in the mediation of grain size, grain quality, thermotolerance, and tolerance to alkali soil. Total RNA was extracted from the young panicles of wild-type varieties and their GS3 knockout plants, with three biological replicates, each comprising 10 plants. The RNA quality was evaluated using the Agilent 2100 Bioanalyzer (Agilent Technologies, Palo Alto, CA, USA). Paired-end libraries were constructed and subsequently sequenced on an Illumina HiSeq2500 platform at OneMore Tech Co., Wuhan, China. Transcripts exhibiting significant differential expression between wild-type varieties and their GS3 mutants were identified using a False Discovery Rate (FDR) of less than 0.05, and an absolute fold change of at least 1.5. Differentially Expressed Genes (DEGs) were subjected to enrichment analysis based on Gene Ontology (GO) functions and Kyoto Encyclopedia of Genes and Genomes (KEGG).

Results

GS3 is the major locus for grain length

To identify the major locus for grain length, we initially compiled grain length data for 529 diverse Oryza. sativa accessions from previous studies (Yang et al. 2014; Xie et al. 2015). The grain length, ranging from 6.13 mm to 10.97 mm, exhibited a normal distribution (Fig. 1A). Subsequently, we performed GWAS by integrating phenotypic and sequencing data based on the Illumina platform (Yang et al. 2014; Xie et al. 2015). A total of 16 loci exceeding a significant threshold ( − log10P ≥ 5) were detected (Fig. 1B), with a focus on the top peak located on chromosome 3, correlating with the known grain length regulation gene GS3 (Fan et al. 2006). A dataset comprising 4,729 rice accessions, including 595 XI I, 465 XI II, 913 XI III, 786 XI intermediate, 767 GJ-Te, 504 GJ-Tr, 241 GJ intermediate, 269 Aus, 96 VI, and 90 intermediate, was subsequently employed to conduct haplotype network analysis of GS3. Allelic variations in GS3 could be classified into three major haplotypes (Fig. 1C). Among these three haplotypes, the C/A SNP in the 2nd exon of Hap3 generated a stop codon, causing the formation of a truncated GS3 protein. The grain length of Hap3 was significantly longer than those of Hap1 and Hap2 (Fig. 1D). Consequently, we deduced that GS3 serves as the primary regulator of grain length.

Fig. 1
figure 1

GWAS for grain length and identification of the causal gene for the peak on chromosome 3. A Diversity in grain length among 529 O. sativa accessions. B Manhattan plot of grain length with  − log10P ≥ 5. C Haplotype network analysis of GS3 using a dataset of 4729 rice accessions. D Comparison of grain length among the three haplotypes. The grain length data was derived from Fig. 1

Full size image

Gene editing of GS3 in high-yield rice varieties

To validate whether manipulating GS3 could regulate grain length in widely cultivated commercial rice varieties, we selected six Chinese rice varieties and one Japanese GJ variety, all characterized by short and round grains. Single-guide RNAs designed for the target sites on the 1st exon of GS3 were integrated into the CRISPR/Cas9 vector (Fig. 2). These constructs were introduced into seven varieties using Agrobacterium. Sanger sequencing of 20 individuals from each mutant in the T0 generation was conducted to assess the mutation efficiency. Single peaks in Sanger sequencing results were interpreted as homozygous mutations, whereas double peaks signified heterozygous mutations. On an average, mutations were observed in 44.29% of the plants, with 18.57% exhibiting putative homozygous mutations (Table S1), except for Sasa, where at least two types of mutations were identified for each variety (Fig. 2).

Fig. 2
figure 2

Schematic diagram of GS3 and sequence comparison between wild-type varieties and their GS3 mutants. The coding region is depicted in gray, and the untranslated regions are shown in light gray. The lines represent introns. Sequences of the target sites in the wild-type and homozygous mutants were aligned below the gene structure

Full size image

Effect of GS3 on yield related traits

We first observed the plant architecture of GS3-edited plants. The results demonstrated that all GS3-edited plants exhibited significantly reduced height compared to wild-type varieties, affirming the role of GS3 knockout in lowering plant height (Fig. 3A–H). Although GS3 had no discernible effect on panicles number in most varieties, exceptions were noted in SN265 and Sasa. In the SN265 genetic background, GS3 knockout significantly decreased panicle numbers, whereas in the Sasa genetic background, panicle numbers were significantly increased (Fig. 3I–O). Regarding grain number per panicle, GS3 exerted its influence primarily on YF47 and LX1, with GS3 knockout significantly decreasing grain number per panicle (Fig. 4). The GS3 mutant exhibited a disadvantageous effect on the setting rate in most studied varieties; however, the knockout of GS3 under the YF47 genetic background surprisingly improved the setting rate (Fig. 4). Our gene-editing experiments underscored the robust and consistent impact of GS3 on grain length across diverse genetic backgrounds (Fig. 5). The knockout of GS3 led to a notable increase in grain length and 1,000 grain weight, with averages of 6.43% and 5.33%, respectively. Notably, the effect of GS3 was relatively weaker in YF47 and LX1, with modest increases in grain length of 2.28% and 2.99%, respectively (Fig. 5D, F). Conversely, its impact was more pronounced in Sasa, resulting in a substantial 14.29% increase in the grain length (Fig. 5G).

Fig. 3
figure 3

Agronomic trait comparison between wild-type varieties and their GS3 mutants. A Representative images of wild-type varieties and their GS3 mutants. Scale bar = 20 cm. B–O Plant height and panicle number of wild-type varieties and their GS3 mutants. The investigation was conducted using three biological replicates, with each replicate containing 10 plants. Different letters indicate significant differences (P < 0.05) from a Duncan’s multiple range test

Full size image
Fig. 4
figure 4

Panicle architecture in wild-type varieties and their GS3 mutants. A–G Representative images of panicles, grain number per panicle, and setting rate of SN265 (A), BG1705 (B), BG1 (C), YF47 (D), BG1501 (E), LX1 (F), and Sasa (G). Scale bar = 1 cm. Data are presented as mean ± s.e.m. (n = 10). Different letters indicate significant differences (P < 0.05) from a Duncan’s multiple range test

Full size image
Fig. 5
figure 5

Grain shape in wild-type varieties and their GS3 mutants. A–G Representative images of grains, 1,000 grain weight, grain length, and grain width of SN265 (A), BG1705 (B), BG1 (C), YF47 (D), BG1501 (E), LX1 (F), and Sasa (G). Scale bar = 1 cm. Data are presented as mean ± s.e.m. (n = 10). Different letters indicate significant differences (P < 0.05) from a Duncan’s multiple range test

Full size image

Effect of GS3 on quality related traits

Grain length is closely related to milling and appearance quality, with slender grain shapes typically enhancing the appearance quality while influencing milling quality. Our findings revealed that despite the significant increase in grain length resulting from GS3 knockout, there was no discernible effect on the brown rice and head rice ratios (Fig. 6). Interestingly, GS3 exhibited varied effects on chalkiness-related traits across seven varieties. The chalkiness rice level and chalkiness rice ratio significantly improved under the genetic backgrounds of SN265 and YF47 but deteriorated under BG1705, BG1, and Sasa. No significant differences were observed between the gene-edited plants and their wild-type counterparts, such as BG1501 and LX1. To assess the impact of GS3 on protein and amylose content, we observed significant increases in protein content and decreases in amylose content under the genetic backgrounds of SN265 and YF47 (Fig. 6A, D). Notably, GS3 knockout displayed similar effects on chalkiness-related traits, protein content, and amylose content in SN265 and YF47 genetic backgrounds. Both protein and amylose contents increased in the BG1 genetic background (Fig. 6C). The amylose content of the GS3 mutant decreased in the LX1 genetic background, whereas the protein content increased in the Sasa genetic background (Fig. 6F, G). No significant differences in protein and amylose contents were observed among the GS3 mutants of BG1705, BG1501, and their wild-type varieties (Fig. 6B, E). In conclusion, GS3 knockout did not affect milling quality and barely changed protein and amylose content but primarily influenced chalkiness-related traits.

Fig. 6
figure 6

Quality traits of wild-type varieties and their GS3 mutants. A–G Brown rice ratio, head rice ratio, chalkiness rice level, chalkiness ratio, protein contents, and amylose contents of SN265 (A), BG1705 (B), BG1 (C), YF47 (D), BG1501 (E), LX1 (F), and Sasa (G). Data are presented as mean ± s.e.m. (n = 5). Different letters indicate significant differences (P < 0.05) from a Duncan’s multiple range test

Full size image

Effect of GS3 on thermotolerance and tolerance to alkali soil

Recent molecular studies have highlighted the involvement of GS3 in regulating thermotolerance and tolerance to alkali soil in the genetic backgrounds of ZH11 (a GJ variety) and Huajingxian (an XI variety) (Kan et al. 2022). To ascertain whether GS3 knockout could uniformly enhance tolerance to heat and alkali stress, we assessed the thermotolerance and tolerance to alkali soil of GS3 mutants across seven varieties. Although thermotolerance varied among the seven varieties, none exhibited a survival rate greater than 20% after heat treatment. BG1705 displayed the highest survival rate (16.67%), whereas LX1 exhibited the lowest survival rate (2.08%) (Fig. S2). Differential degrees of improvement were observed between the GS3 mutants and their wild-type counterparts during heat treatment. Knockout of GS3 increased the survival rate of BG1705 from 16.67 to 80.21%, which was the highest among the mutants. Conversely, the GS3 mutant of LX1 exhibited the lowest improvement, increasing from 2.08 to 19.79% (Fig. 7). Subsequent assessments of alkali tolerance revealed that YF47 had the highest survival rate (18.60%), significantly surpassing that of the other six varieties (Fig. S3). Comparative analysis of alkali tolerance between mutants and wild-type varieties indicated significant improvement in the survival rates of all mutants. The GS3 mutant of YF47 exhibited further enhancement, reaching a survival rate of 40.40%. Conversely, the lowest survival rate was observed for the Sasa mutant, which increased the survival rate from 1.35 to 6.42% (Fig. 7). Collectively, GS3 knockout indiscriminately improved the tolerance to heat and alkali stress across the GJ varieties used in this study.

Fig. 7
figure 7

Tolerance of wild-type varieties and their GS3 mutants. A Representative images of wild-type varieties and their GS3 mutants under normal and heat treatment conditions. Scale bar = 5 cm. B–H Survival ratio of wild-type varieties and their GS3 mutants after heat treatment. I Representative images of wild-type varieties and their GS3 mutants under 75 nm alkali treatment. Scale bar = 5 cm. J–P Survival ratio of wild-type varieties and their GS3 mutants after alkali treatment. The investigations in B–H and J–P were conducted with three biological replicates, each replicate containing 10 plants. Different letters indicate significant differences (P < 0.05) from a Duncan’s multiple range test

Full size image

Identification of downstream genes regulated by GS3

Given GS3's pleiotropic effects on plant height, grain length, and tolerance to heat and alkali stress, we sought to unveil its downstream genes. To achieve this, we conducted RNA-seq analysis of GS3 mutants and their corresponding wild-type varieties. A total of 14,053 differentially expressed genes (DEGs) were identified, comprising 6599 upregulated and 7454 downregulated genes in GS3 mutants compared with their wild-type counterparts. Gene Ontology (GO) analysis indicated that these DEGs were enriched in biological processes related to carbohydrate metabolic process, polysaccharide metabolic process, and cell wall organization or biogenesis (Fig. S4). Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analyses further revealed enrichment in pathways associated with metabolic pathways, biosynthesis of secondary metabolites, and starch and sucrose metabolism (Fig. S5). Upon closer examination, only 46 genes exhibited differential expression in all seven comparisons between the mutants and their respective varieties (Fig. 8A). Subsequent GO analysis of these 46 DEGs highlighted the enrichment in biological processes related to photosynthesis and photosystem II stabilization (Fig. 8B and Fig. S6). Additionally, KEGG pathway analyses indicated that the 46 DEGs were enriched in pathways related to photosynthesis and cutin, suberine, and wax biosynthesis (Figs. 8 and S7).

Fig. 8
figure 8

Transcriptome analysis of wild-type varieties and their GS3 mutants. A Upset plot represents the differentially expressed genes (DEGs) regulated by GS3. B GO analysis of the 46 DEGs. C KEGG pathway analysis of the 46 DEGs

Full size image

Discussion

In plants, G proteins exhibit structural similarities to their animal counterparts but employ atypical mechanisms and effector proteins to regulate diverse processes such as growth, cell proliferation, defense, stomatal movements, channel regulation, sugar sensing, and hormonal responses (Urano et al. 2013). Within the rice genome, a single subunit of Gα (RGA1) and Gβ (RGB1) is encoded, along with two typical Gγ subunits (RGG1 and RGG2) and three atypical Gγ subunits (GS3, DEP1, and OsGGC2) (Xu et al. 2016a, 2016b). The cloning of two important yield quantitative trait loci (QTLs), GS3 and DEP1, has prompted researchers and breeders to speculate that manipulating G protein signaling could enhance yields not only in rice but also in other crop species (Botella 2012). Subsequent investigations have shown that the manipulation of three Gγ proteins, DEP1, GGC2, and GS3, can achieve the desired grain size (Sun et al. 2018). Additionally, modulation of the C-terminus of DEP1 has been demonstrated to synergistically enhance grain quality and yield in rice (Huang et al. 2022). Recent studies have demonstrated that GS3/TT2 regulates thermotolerance and alkali tolerance through SCT1-dependent alteration of wax biosynthesis and phosphorylation of aquaporins, thereby influencing the distribution of hydrogen peroxide (H2O2) (Kan et al. 2022; Zhang et al. 2023). These findings suggest that manipulating G proteins has the potential to achieve a high and stable yield performance in rice production.

In this study, we selected seven GJ varieties, six of which harbored a truncated DEP1, displaying a dense and erect panicle architecture. Given the opposing effects of truncated DEP1 and truncated GS3 on specific traits, such as reduced grain length with the truncated DEP1 allele and significantly increased grain length with the truncated GS3 allele (Sun et al. 2018; Wu et al. 2022), we introduced a variety, Sasa, with a full-length DEP1. This addition aimed to examine whether the truncated DEP1 would also exhibit contrary effects on thermotolerance and alkali soil tolerance. Our findings indicated that GS3 could perform its function independently of the DEP1 genotype. Knockout of GS3 consistently increased grain length and enhanced tolerance to heat and alkali stress across all genotypes. However, the effects of GS3 on chalkiness-related traits varied depending on the inherent performance of the varieties. For example, SN265 and YF47, the two varieties displaying the poorest chalkiness traits among the seven selected varieties, showed significantly improved chalkiness rice levels and chalkiness rice ratios upon GS3 knockout. In contrast, varieties such as Sasa, BG1705, and BG1, which initially exhibited an advantage in chalkiness-related traits, showed a decline in performance following the GS3 knockout. Our previous study revealed that YF47 carried an XI-type allele at the Chalk5 locus, responsible for encoding a vacuolar H+-translocating pyrophosphatase influencing grain chalkiness in rice (Li et al. 2014; Wang et al. 2023). This indicates that the function of GS3 in chalkiness-related traits in YF47 may be influenced by the distinct genotype of Chalk5. Thus, the strategic combination of a nonfunctional GS3 allele with other chalkiness-regulating genes, such as Chalk5, GS9, OsMADS1, and OsSPL16, has the potential to achieve a balance between yield, quality, and tolerance at a higher level (Wang et al. 2012; Li et al. 2014; Liu et al. 2018; Zhao et al. 2018).

Natural genetic variation plays a pivotal role in advancing crop domestication and evolution. However, optimal variants that simultaneously enhance yield, quality, and tolerance may not exist in natural populations. For example, rice production often faces a trade-off between panicle number and grain number per panicle, as exemplified in IPA1 (Ideal Plant Architecture 1), a typical pleiotropic gene that increases grains per panicle while reducing tillers (Jiao et al. 2010). Addressing this tradeoff, the artificial deletion of a 54-base pair cis-regulatory region in IPA1 resolves the conflict, substantially boosting the grain yield per plant (Song et al. 2022). A base-editing system at the Wx locus was utilized to generate a range of mutants with apparent amylose contents ranging from 1.4 to 11.9%. This approach has successfully achieved the goal of finely tuning rice AC within the range of 0–12%, thereby expanding the diversity of breeding materials accessible to breeders (Xu et al. 2021). Editing the region near the TATA box of the Wxb promoter has led to the creation of six novel Wx alleles, downregulation of Wx expression, and fine tuning of grain amylose content (Huang et al. 2020). Collectively, these studies highlight that targeted gene modification is a widely applicable and reliable approach for generating optimal alleles that may not naturally exist in populations. In this study, we identified three major haplotypes of GS3, with Hap3 carrying a C/A SNP mutation in the exon that generates a stop codon, causing a truncated GS3 protein. This natural variation originated from an ancient GJ lineage and was later introgressed into Xian/Indica (XI) (Takano-Kai et al. 2009). Hap3 not only exhibits an enlarged grain size but also enhanced thermotolerance (Kan et al. 2022; Zhang et al. 2023). However, this truncated GS3 allele displayed increased sensitivity to alkaline conditions (Zhang et al. 2023). In this study, we demonstrated that the knockout of GS3 could simultaneously improve grain shape, thermotolerance, and alkaline tolerance.