Natural variation of WBR7 confers rice high yield and quality by modulating sucrose supply in sink organs

Summary

Grain chalkiness is an undesirable trait that negatively regulates grain yield and quality in rice. However, the regulatory mechanism underlying chalkiness is complex and remains unclear. We identified a positive regulator of white‐belly rate (WBR). The WBR7 gene encodes sucrose synthase 3 (SUS3). A weak functional allele of WBR7 is beneficial in increasing grain yield and quality. During the domestication of indica rice, a functional G/A variation in the coding region of WBR7 resulted in an E541K amino acid substitution in the GT‐4 glycosyltransferase domain, leading to a significant decrease in decomposition activity of WBR7^A (allele in cultivar Jin23B) compared with WBR7^G (allele in cultivar Beilu130). The NIL(J23B) and knockout line NIL(BL130)^KO exhibited lower WBR7 decomposition activity than that of NIL(BL130) and NIL(J23B)^COM, resulting in less sucrose decomposition and metabolism in the conducting organs. This caused more sucrose transportation to the endosperm, enhancing the synthesis of storage components in the endosperm and leading to decreased WBR. More sucrose was also transported to the anthers, providing sufficient substrate and energy supply for pollen maturation and germination, ultimately leading to an increase rate of seed setting and increased grain yield. Our findings elucidate a mechanism for enhancing rice yield and quality by modulating sucrose metabolism and allocation, and provides a valuable allele for improved rice quality.

Introduction

Rice (Oryza sativa L.), one of the most important food crops and an important source of complex carbohydrates, feeds more than half of the world population (Chauhan et al., 2017; Wing et al., 2018). With the continuous improvement of living standards, rice quality has become a decisive factor affecting the commodity value of rice. Rice quality is determined by several factors, including appearance quality, processing quality, eating and cooking quality (ECQ) and nutrition quality (Fitzgerald et al., 2009; Zhou et al., 2020). Chalkiness, the opaque regions in the grain, is a negative indicator of appearance quality. Consumers in different regions prefer different grain shapes, but generally choose rice with low chalkiness and high transparency (Fitzgerald et al., 2009; Siebenmorgen et al., 2013). The presence of chalkiness significantly increases losses during the grain milling and reduces head rice yield (Fitzgerald et al., 2009; Zhao and Fitzgerald, 2013). Since chalkiness is an undesirable trait breeders attempt to minimize it by selection.

Chalkiness formation is closely related to endosperm development and content of storage components and is affected by multiple genes and environmental factors. Starch granules in chalky endosperm usually appear round and loosely arrangement, and the storage components in the chalky regions content are often different (Guo et al., 2011; Li et al., 2014; Wu et al., 2022). Based on distribution of opaque regions, chalkiness is variously described as white‐belly, white‐core, white‐back and floury endosperm. In terms of endosperm structure, the area most distant from the dorsal vascular bundle receives less nutrient supply, which may be a contributing factor to the more frequent occurrence of white‐belly in wider grains (Krishnan and Dayanandan, 2003). In addition, restricted nutrient supply to the endosperm or larger grain size is more likely to lead to increased chalkiness (Pan et al., 2021; TASHIRO et al., 1980; TASHIRO and EBATA, 1975; Wang et al., 2015; Zhao et al., 2018). High temperature is the most important environmental factor for the formation of chalkiness, as it accelerates endosperm development and causes abnormal synthesis of storage components (Nevame et al., 2018; Tabassum et al., 2020). Thus, studies on chalkiness continue to help in understanding its development and consequences.

Chalkiness is a quantitative trait controlled by multiple genes. Quantitative trait loci (QTL) associated with chalkiness have been identified on all 12 rice chromosomes (Gao et al., 2016; Peng et al., 2014b; Yang et al., 2021; Yun et al., 2016), but few genes were cloned in natural populations. Chalk5 was identified as a positively regulator of chalkiness rate through affects the biogenesis of protein bodies and small vesicle‐like structures by disturbing pH homeostasis of the endomembrane trafficking system in developing seeds (Li et al., 2014). WCR1 encodes an F‐box protein, which enhances transcription and inhibits degradation of MT2b, thereby increasing the capacity to clear excessive ROS and ultimately leading to a decreased white‐core rate (Wu et al., 2022). In addition, disruption of genes involved in synthesis of storage components during endosperm development frequently leads to floury endosperm (Fujita et al., 2007; Hu et al., 2018; Long et al., 2017; Peng et al., 2014c; Yu et al., 2021). Therefore, cloning new chalkiness gene will contribute to a better understanding of chalkiness formation and provide insight for rice agronomists and breeders.

Sucrose, the main output from photosynthesis, is transported through the phloem to sink tissues, where it is irreversibly hydrolysed to glucose and fructose by invertases (INV), or reversibly decomposed to fructose and nucleoside diphosphate glucose by sucrose synthase (SUS) (Huang et al., 2016; Ruan, 2014; Ruan et al., 2003). The SUS protein typically consists of two domains, an N‐terminal domain of approximately 250 amino acids, that is responsible for cellular targeting, and a C‐terminal GT‐4 domain of approximately 500 amino acids, that is responsible for glycosyltransferase activity (Stein and Granot, 2019). Seven genes encode SUS in rice (Hirose et al., 2008), among which SUS3 is expressed predominantly in endosperm at the early stage of grain filling, providing precursors for starch synthesis by decomposing sucrose (Huang et al., 1996; Wang et al., 1999). Numerous studies have shown that SUS activity affects starch accumulation in seed. The sh (shrunken) mutant in maize has only 10% of SUS activity compared with the wild type, and starch content in the developing and mature mutant endosperm was reduced by about 40% (Chourey and Nelson, 1976). Overexpression of six SUS genes resulted in significantly increased in grain weight and yield, and seed size and starch content in mature grains exceeded those of wild type (Fan et al., 2019). On the contrary, knockout of maize SUS1 and SUS2 did not cause significant changes in plant growth and endosperm development (Deng et al., 2020). An Arabidopsis sus1/sus2/sus3/sus4 quadruple mutant displayed no significant difference in plant growth and plant morphology relative to the wild type, but the starch content in siliques of the mutant was increased by about 50% (Barratt et al., 2009). Furthermore, SUS was highly expressed in vascular tissues, especially in the phloem, providing substrates for cellulose and complex carbohydrate synthesis as well as energy for cell growth and development (Fujii et al., 2010; Nolte and Koch, 1993). These results indicated that SUS genes regulate sucrose decomposition and has an important effect on metabolite synthesis. However, the mechanism by which SUS genes affect chalkiness and content of storage components in rice endosperm is currently unclear.

Sucrose regulates plant growth by providing energy and substrates. During the development of rice seeds, a decrease in sucrose supply can lead to abnormal seed development and a decline in contents of storage components. Sucrose unloaded in the grain is first decomposed by INV or SUS. Mutation or down‐regulation of CIN2/GIF1, a sucrose hydrolysing invertase gene, caused a significant decrease in CIN activity, soluble sugar content and content of storage components in the grain (Wang et al., 2008). OsSWEET11 and OsSUT1 regulate the transport of sugar, and mutations or down‐regulation of these genes lead to delayed or defective grain filling (Ma et al., 2017). OsNF‐YB1 positively regulates the transcription levels of OsSUT1, and the decrease in sugar supply in the endosperm of a osnf‐yb1 mutant led to a reduction in content of storage components and increased chalkiness (Bai et al., 2016; Xu et al., 2016). The coordinated decrease in sugar content, expression of sugar metabolism genes and content of storage components in floury mutants also suggests that regulation of sugar supply affects the formation of floury endosperm (Chen et al., 2023; You et al., 2019). Nonetheless, the regulation of sucrose metabolism by SUS and its effects on sucrose homeostasis in different organs, as well as its impact on agronomic traits, remains unclear.

In this study, we cloned a white‐belly gene WBR7 from a population derived from a cross of indica cultivar Jin23b (J23B) and japonica cultivar Beilu30 (BL130) by map‐based cloning. A positive effect of WBR7 on the regulation of WBR was confirmed through transgenic assays. Haplotype analysis and phylogenetic analysis of WBR7 found that a mutation occurred in WBR7 ^G during domestication of indica rice gave rise to the WBR7 ^A allele, causing a significant decrease in WBR. Analysis of enzymatic activity showed a significant decrease in the rate of decomposition of WBR7^A compared with WBR7^G. Detection and analysis of sugar contents and sugar metabolite contents in conducting and sink organs revealed less sucrose was consumed in conducting organs as decomposition activity decreased, lead to an increase in supply of sucrose in the sink organs that contributed to improved grain yield and quality.

Results

Map‐based cloning of WBR7

A previous study reported a QTL for WBR flanked by markers RM455 and RM234 on chromosome 7 and named it WBR7 (Yun et al., 2016) (Figure 1b). In this study, we validated the genetic effect of WBR7 in a BC₅F₂ population of Jin23B (recipient parent with lower WBR) with BL130 (donor parent with higher WBR) (Figure 1a), the LOD value of WBR7 was 18.23, with an additive effect of 15.87, explaining 45.5% of the WBR variation (Table S1). The WBR7 allele from BL130 showed incompletely dominant effect (Figure S1a). To fine‐map the WBR7 locus, we developed a segregating BC₅F₄ population of 6000 plants and identified 84 recombinants between markers RM455 and RM234 (Figure 1c). Progeny testing of all recombinants in Lingshui environment in winter 2018 narrowed the WBR7 interval to 5.74 kb flanked by M24 and M27, and there was co‐segregation with M25 and M26 (Figure 1c,d). The isolated WBR7 contained a 1.65 kb promotor region and 4.09 kb coding region of LOC_Os07g42490, which encodes Sucrose Synthase 3 (SUS3) that catalyses decomposition and synthesis of sucrose (Figure 1d). To precisely evaluate the effect of WBR7 on regulation of WBR, we developed a near‐isogenic lines pair, NIL(BL130) and NIL(J23B), in the BC₆F₂ population. NILs and heterozygous individuals were planted in Wuhan environment in summer 2019. Due to the lower temperature at filling stage, the grain WBR of rice planted in Wuhan was lower than that in Lingshui (Figure 1d,f). Although the two NILs showed similar plant architecture (Figure S1b), the WBR value of NIL(BL130) was more than 45% higher than that of NIL(J23B) (Figure 1e,f). Therefore, LOC_Os07g42490 was confirmed as the candidate gene underlying WBR7, and the WBR7 allele from J23B contributed to lower WBR.

Figure 1.

Map‐based cloning of WBR7. (a) Milled rice images of the parents. Scale bar, 10 mm. (b) Genetic linkage analysis of WBR on chromosome 7, arrow points to qWBR7. (c) The 680 kb interval of qWBR7 between M37 and RM234, n represents the number of plants used to screen for recombinants. Number. of Recs. indicates the number of recombinants between WBR7 and flanking molecular markers. (d) Genotypes and phenotypes of recombinants, each of which was confirmed by progeny test. The ‘A’ below WBR represents white‐belly rate phenotype of progeny plants with homozygous BL130 genotype in the heterozygous region, while ‘B’ and ‘H’ refer to the homozygous J23B genotype and heterozygous genotype, respectively. P‐values were based on two‐tailed t‐tests. Progeny test results were defined based on the comparison of WBR differences between two types of homozygous genotypes. When the differences were significant, the candidate gene was considered to be located in the heterozygous region of the recombinant, and the progeny testing results were defined as ‘H’; while the differences were not significant, the candidate gene was considered to be located in the homozygous region of the recombinant, and the progeny testing results were defined as ‘A’ or ‘B’. The progeny testing results of multiple recombinants provide support for the narrowing of the mapping interval. (e) Milled rice images of the NILs. Scale bar, 10 mm. (f) Differences in WBR among different genotypes in a BC₆F₂ population. Significance of differences was determined by Duncan's multiple range test.

WBR7 positively regulates WBR

Considering that the candidate interval covered the promotor and coding region of WBR7, we performed quantitative real‐time PCR (qRT‐PCR) to examine the expression levels of WBR7 in different tissues of the recurrent parent J23B. WBR7 was barely expressed in seedlings, but expression gradually increased in the branch, leaf sheath, stem, flag leaf, husk, panicle and endosperm (Figure 2a). Based on this, we determined the expression levels of WBR7 in different tissues of the NILs. There was no consistent trend in the differences of expression levels between NILs, the expression level in NIL(J23B) was significantly higher in husk at 4 and 12 days after flowering (DAF), branch at 4 and 12 DAF, and stem at 12 DAF, but lower in endosperm at 4 DAF, husk, branch and stem at 8 DAF, and anther than in NIL(BL130) (Figure S2). Transient expression in rice protoplasts to analyse the subcellular localization of WBR7 using WBR7‐YFP (WBR7 fused to yellow fluorescent protein) showed co‐localization with cytoplasm marker pYBA1138‐RFP (Figure 2b), indicating that WBR7 functioned in the cytoplasm.

Figure 2.

Expression patterns, subcellular localization and transgenic verification of WBR7. (a) Spatiotemporal expression patterns of WBR7 in J23B plants (n = 5 biological replicates). Endosperm, husk, branch, stem, leaf sheath and flag leaf were taken from the uppermost part of the main stem at 4 DAF, the panicle represents 5‐cm young panicle, the Anther samples were taken from the field at the day before flowering, the seedling samples were taken from the functional leaf of four‐leaf stage seedlings. (b) WBR7‐YFP and pYBA1138‐RFP (cytoplasm marker) co‐localize in the cytoplasm (n = 3 biological replicates). The empty vector pM999‐YFP was used as a negative control. Scale bars, 20 μm. (c, d) WBR in WBR7 transgenic CO and KO lines (n = 12 biological replicates). Scale bars, 10 mm. Significant differences were based on two‐tailed t‐tests, **, P < 0.01.

To confirm that WBR7 is responsible for the white‐belly phenotype, we transformed NIL(J23B) plants with the complementation vector CO^BL130 (WBR7 coding region from BL130 driven by its native 2 kb promotor; Figure S3a). The T₁ generation transgenic‐positive complementary lines NIL(J23B)^COM‐1 and NIL(J23B)^COM‐2 displayed significantly higher WBR values and higher WBR7 expression levels than transgene‐negative plants CO^BL130(−) (Figure 2c; Figure S4). We also generated knockout (KO) plants in NIL(BL130) background by CRISPR Cas9 genome editing. The nucleotide base deletion in the NIL(BL130)^KO‐1 and NIL(BL130)^KO‐2 mutants of T₁ generation caused premature termination of translation of WBR7 (Figure S3b), and the mutants displayed significantly lower WBR values than the wild‐type (WT, NIL(BL130); Figure 2d). Hence, LOC_Os07g42490 was the causal gene underlying WBR7, which functioned as a positive regulator of WBR.

Natural variation in WBR7

To study functional variation in WBR7, we compared full‐length sequences of the mapped 5.74 kb intervals from BL130 and J23B, and identified 22 polymorphisms between parents, which included six single‐nucleotide polymorphisms (SNPs) and 1 insertion/deletion (InDel) in the promotor region, nine SNPs in the introns and six SNPs in the exons causing four missense mutations (Figure 3a; Figure S5a). We then analysed these variations in 498 cultivated rice accessions. Five haplotypes were identified (Figure 3a,b), with japonica accessions having only haplotype 1 (Hap1) and Hap2. We also investigated the WBR value for all 498 accessions (Data Set S1), the WBR of a large proportion of japonica accessions were relatively low. Moreover, due to the uneven distribution of the japonica accessions in haplotypes (Figure 3b), we restricted an association analysis of WBR and sequence variations to 278 indica accessions. A SNP (A/G) at +3114 bp in WBR7 exon 11 of was predicted to generate a functional variant (Figure 3c), which caused an amino acid change from glutamic (Glu, BL130) to lysine (Lys, J23B) (Figure S5a). In indica rice, accessions H1‐H4 with a G base‐allele showed significantly higher WBR than H5 with an A base‐allele (Figure 3d). According to the Uniprot database (https://www.uniprot.org/), the SNP mutation was located in the GT‐4 glycosyltransferase domain of SUS and could affect glycosyl transfer between uridine diphosphate (UDP) and fructose.

Figure 3.

Natural variation of WBR7. (a) Parental DNA sequence variation within the 5.74 kb mapping interval of WBR7. Red A/G on the 11th exon indicates the functional variation. (b) Haplotype analysis of WBR7 in 498 cultivated rice accessions with reference to the sequences of the parents. Red A/G indicates the functional variation. Frequencies in taxonomic groups are shown on the right. (c) Local association testing of WBR and 22 variations in the 5.74 kb mapping interval of WBR7. The red dotted line and the red A/G represent the threshold and the predicted functional site, respectively. The x‐axis represents the position in the genome. (d) WBR in genotypes with the WBR7 ^G and WBR7 ^A alleles. Significance of differences was based on two‐tailed t‐tests, **, P < 0.01. (e) Sucrose decomposing activity among various prokaryotic expressed WBR7 proteins (n = 6 biological replicates). The empty protein pETsumo was used as a negative control. Anti‐His was used to regulate the concentration of WBR7 proteins to a similar level, the bottom numbers indicate relative grayscale values. Significance of differences was determined by Duncan's multiple range test. (f) Enzymatic properties of three WBR7 proteins under different substrate sucrose concentrations (n = 5 biological replicates). (g) Geographic distributions of 533 cultivated rice accessions. Green and orange sectors in the circles represent the proportions of rice accessions with WBR7 ^G and WBR7 ^A , respectively (n = 3 biological replicates). (h) Phylogenetic relationship of WBR7 generated from 498 cultivated and 442 wild rice accessions. Green and orange sectors in the circles represent the proportions of rice accessions with WBR7 ^G and WBR7 ^A , respectively. Different coloured boxes represent rice subspecies.

To clarify the effect of predicted functional variation on WBR7 activity, segments WBR7 ^J23B (full‐length coding region of J23B), WBR7 ^BL130 (full‐length coding region of BL130) and WBR7 ^BHJ (full‐length coding region of BL130 with replacement of G with A at +3114 bp) were ligated into E. coli expression vector pET28a‐SUMO (Figure S5b), and sucrose decomposition activity was detected after expression and purification of the three WBR7 proteins. WBR7^BL130 displayed significantly higher sucrose decomposition activity than WBR7^J23B at similar protein concentrations, and the sucrose decomposition activity of WBR7^BHJ was decreased to a level slightly lower than WBR7^J23B (Figure 3e). We further analysed the enzymatic properties of the three WBR7 proteins. The maximum reaction rate at a given amount of enzyme, Vmax, of WBR7^BL130 was nearly 1.5 times higher than those of WBR7^J23B and WBR7^BHJ, and the substrate concentrations at half the maximum reaction rate, Km, of WBR7^BL130 were 93% and 82% of those of WBR7^J23B and WBR7^BHJ (Figure 3f; Table S2). Thus, WBR7^BL130 had stronger sucrose decomposition activity, and the predicted functional variation caused the difference in activity between WBR7^BL130 and WBR7^J23B. Due to the inconsistent trends in WBR7 expression levels between NILs, we believe that the +3114 bp variation in the coding region of WBR7 is responsible for the WBR variation of NILs.

The geographical distributions of the WBR7 ^A and WBR7 ^G alleles among 533 accessions varied in different parts of the world (Figure 3g). Phylogenetic analysis of WBR7 in these 533 cultivated rice accessions and 442 wild rice accessions suggested that WBR7 ^G originated from O. rufipogon and underwent mutation to WBR7 ^A during domestication of indica rice (Figure 3h); 31% of indica accessions and all japonica accessions had the WBR7 ^G allele (Figure 3b; Figure 3h), indicating that WBR7 ^A could be a promising breeding target to reduce the WBR in rice.

WBR7 ^A positively regulates sucrose supply and storage components in endosperm

Chalkiness is usually associated with changes in endosperm structure and contents of storage components. Therefore, we investigated endosperm structure, endosperm development and storage components in the NILs. Scanning electron microscopy (SEM) revealed numerous irregularly shaped large starch granules surrounded by small starch granules in the white‐belly region, whereas endosperm structure in transparent endosperm was compact with no loosely packed starch granules (Figure S6a). These results indicated that altered morphology and spatial distribution of starch granules contributed to the formation of white‐belly. We then analysed the grain filling rate in the NILs. NIL(BL130) showed a higher filling rate from 4 DAF to 12 DAF and reached maximum dry weight at 12 DAF, whereas NIL(J23B) had a maximum dry weight at 16 DAF, with no difference in dry weight of mature endosperm between NILs (Figure 4a). Compared to NIL(J23B), NIL(BL130) endosperm showed a distinct opaque area in the belly at 12 DAF (Figure 4b), suggesting that high filling rate was responsible for higher WBR. Transmission electron microscopy (TEM) of 7 DAF endosperm showed that the number of protein bodies in NIL(J23B) endosperm was significantly higher than that in NIL(BL130) (Figure S6b,c). These results suggested that a rapid grain filling rate might affect the arrangement of starch granules and the synthesis of storage proteins.

Figure 4.

Grain filling and sucrose metabolism in endosperm. (a) Grain filling rate of NILs. DAF means days after flowering (n = 3 biological replicates). Significant differences were based on two‐tailed t‐tests, * and **, P < 0.05 and P < 0.01, respectively. (b) Grain morphology of NILs at different grain filling stages. (c, d) Sucrose (c) and fructose (d) contents in 4 DAF and 8 DAF endosperm from different lines. Endo, endosperm (n = 6 biological replicates). (e, f) Sucrose decomposition activity (e) and sucrose synthetic activity (f) in 4 DAF endosperms of different lines (n = 6 biological replicates). Significance of differences was determined by Duncan's multiple range test. (g) qRT‐PCR analysis of genes related to sucrose and starch metabolism in 4 DAF endosperm from NILs (n = 4 biological replicates). NIL(J23B)/NIL(BL130) implies the multiplicative difference in the expression level of genes in NIL(J23B) relative to NIL(BL130). The red dash line represents a ratio of 1 for the relative expression levels, while the black solid line and “**” symbol indicates significant expression differences for these genes. Significant differences were based on two‐tailed t‐tests, **, P < 0.01. (h–j) Amylose (h) total starch (i) and storage protein (j) content in mature seeds of different lines (n = 8 biological replicates). Significance of differences was determined by Duncan's multiple range test.

SUS3 encoded by WBR7 provides energy and substrates for synthesis of endosperm storage components by decomposing sucrose (Chourey and Nelson, 1976; Huang et al., 1996; Li et al., 2013; Wang et al., 1999). Sugar also acts as a signalling molecule affecting metabolic processes (Doll et al., 2017; Liao et al., 2020; Ruan, 2014). The sugar contents, and sucrose decomposition and synthesis activity in the endosperms of NILs, NIL(BL130)^KO and NIL(J23B)^COM lines were measured at different grain filling stages. The sucrose content in NIL(BL130) was significantly lower than in NIL(J23B), but the fructose content did not (Figure 4c,d). Consistently, the sucrose decomposition activity in NIL(BL130) was significantly higher than in NIL(J23B) while the sucrose synthesis activity was significantly lower (Figure 4e,f). However, the sucrose content and fructose content in 12 DAF and 16 DAF endosperm of NIL(BL130) was significantly lower than in NIL(J23B) (Figure S7a,b). Compared to NIL(BL130), NIL(BL130)^KO exhibited a significant increase in sucrose content and significant decrease in fructose content (Figure 4c,d), and a significant decrease in sucrose decomposition and synthesis activity (Figure 4e,f). In contrast, the sucrose content of NIL(J23B)^COM significantly decreased compared to NIL(J23B) (Figure 4c), but the fructose content and sucrose decomposition activity significantly increased (Figure 4d,e). However, there was no significant difference in sucrose synthesis activity between NIL(J23B) and NIL(J23B)^COM at 4 DAF, which might be due to the influence of homologous genes. These results suggesting that lower sucrose decomposition activity might led to a higher sucrose content in endosperms of NIL(J23B) and NIL(BL130)^KO; however, the increased sucrose decomposition activity in endosperms of NIL(BL130) and NIL(J23B)^COM did not lead to the dominance of fructose content.

To investigate the effect of WBR7‐affected synthesis of storage components, RNA‐seq analysis was performed to 4 DAF endosperm of the NILs. A total of 4099 differentially expressed genes (DEGs), of which 3255 (79%) were up‐regulated in NIL(J23B), were identified (Figure S8a). GO analysis indicated that the majority of DEGs were involved in biological processes (BP), among which biological regulation, oxidation and reduction, and carbohydrate metabolic processes were significantly enriched (Figure S8b). KEGG analysis revealed significant effects on metabolism process, followed by processes related to protein processing, folding and degradation, as well as starch and sucrose metabolism (Figure S8c). qRT‐PCR validation performed on 18 starch and sucrose metabolism‐related genes in endosperms at 4 DAF revealed significant differential expression between NILs in all 18 genes, with the majority (13 of 18) up‐regulated in NIL(J23B) (Figure 4g).

To study the impact of differential gene expression on metabolite synthesis, we determined central carbon metabolite (CCM) levels in 8 DAF developing endosperm of the NILs, NIL(BL130)^KO and NIL(J23B)^COM lines. Twenty‐six of 35 CCMs had higher levels in NIL(BL130)^KO and NIL(J23B) than in NIL(BL130) and NIL(J23B)^COM, respectively (Figure S9). We also examined storage component content in mature endosperm of these lines, and the contents of amylose, total starch and storage proteins showed similar increasing trends in mature endosperms of NIL(BL130)^KO and NIL(J23B) compared with NIL(BL130) and NIL(J23B)^COM, respectively (Figure 4h–j). To sum up, increased sucrose decomposition activity in endosperm does not lead to advantages in fructose content and carbohydrate metabolites, which might indicate a relatively insufficient supply of sucrose in endosperm of NIL(BL130) and NIL(J23B)^COM.

WBR7 ^A decreases sucrose metabolism in conducting organs, thereby increasing sucrose supply in endosperm

To analysis the reason for insufficient sucrose supply in endosperm of NIL(BL130) and NIL(J23B)^COM, we analysed the contents of sucrose and fructose in flag leaf sheaths and stems of main panicle in the NILs, NIL(BL130)^KO and NIL(J23B)^COM at 0 DAF and 4 DAF. Compared with NIL(BL130) and NIL(J23B)^COM, the sucrose content significantly increased in the conducting organs of NIL(J23B) and NIL(BL130)^KO at all stages, whereas the fructose content showed no difference except for the stem (Figure 5a,b; Figure S10a,b). We then evaluated sucrose decomposition and synthesis activity in 4 DAF stems in NILs and found that NIL(BL130) and NIL(J23B)^COM had relatively higher sucrose decomposition and synthesis activity in stems (Figure 5c,d). These results indicated that lower SUS activity enhanced the sucrose content in the conducting organs in NIL(J23B) and NIL(BL130)^KO, likely leading to a sufficient supply of sucrose in the endosperm.

Figure 5.

Sucrose metabolism in conducting organs. (a, b) Sucrose (a) and fructose (b) contents in the 4 DAF flag leaf sheaths and stems of different lines (n = 6 biological replicates). (c, d) Sucrose decomposition activity (c) and sucrose synthetic activity (d) in 4 DAF stems of different lines (n = 6 biological replicates). Significance of differences was determined by Duncan's multiple range test. (e) qRT‐PCR analysis of sucrose metabolism‐related genes in conducting organs of NILs at 0 DAF, 4 DAF and 8 DAF (n = 4 biological replicates). (f, g) Starch (f) and cellulose (g) contents in stems of different lines at 0 DAF and 4 DAF (n = 8 biological replicates). (h) ¹³C ratios in stems and endosperms of different lines at 5 DAF (n = 3 biological replicates). (i) Total carbon ratio in stems and endosperms of different lines at 5 DAF (n = 3 biological replicates). Significance of differences was determined by Duncan's test.

A previous study reported that overexpression of SUS3 caused an increase in cellulose content, we wondered whether WBR7 has a role in conducting organs (Fan et al., 2019). To determine whether the decrease in sucrose content in conducting organs was due to an increase in sucrose metabolism. A qRT‐PCR analysis to determine the expression level of sucrose metabolism genes, sugar transport genes and cellulose synthesis‐related genes in the stems and leaf sheaths of the NILs at different grain filling stages showed that a large proportion of the genes had significantly higher expression levels in NIL(BL130) than in NIL(J23B) (Figure 5e; Figure S10c,d). Subsequent analysis of contents of sucrose‐related metabolites in the NILs, NIL(BL130)^KO and NIL(J23B)^COM lines showed that starch contents in conducting organs of NIL(BL130) and NIL(J23B)^COM were higher than, or close to, NIL(J23B), whereas the starch contents in conducting organs of NIL(BL130)^KO were significantly lower than those in NIL(BL130) (Figure 5f). Similarly, the cellulose contents in stems of NIL(BL130) and NIL(J23B)^COM were significantly higher than in NIL(J23B) and NIL(BL130)^KO (Figure 5g). Hence, these results suggested that lower SUS activity resulted in reduced sucrose metabolism in the conducting organs, leading to an increase in sucrose transported to the endosperm.

A ¹³C‐sucrose feeding experiment was performed to determine the regulation of WBR7 on sucrose transport in the conducting organs and endosperm. The main stems of the NILs, NIL(BL130)^KO and NIL(J23B)^COM at 4 DAF, were fed with ¹³C‐sucrose, and the stems and endosperms were sampled for detecting the proportion of ¹³C after 1‐day treatment. Compared with NIL(BL130), the ¹³C proportion was lower in the stems but higher in endosperm of NIL(J23B) and NIL(BL130)^KO (Figure 5g), a similar trend was observed when compared NIL(J23B) to NIL(J23B)^COM (Figure 5g). Measures of total carbon (C) ratio in the stems and endosperms similarly showed that the total C ratios were lower in the stems of NIL(J23B) and NIL(BL130)^KO and higher in the endosperm compared with NIL(BL130) and NIL(J23B)^COM (Figure 5i). In summary, reduced SUS activity in conducting organs of NIL(J23B) and NIL(BL130)^KO led to decreased sucrose metabolism in conducting organs while increasing sucrose supply in the endosperm.

WBR7 ^A positively regulates yield and quality in rice by increasing sucrose supply in sink organs

Sucrose is the main form of assimilate transport, providing energy and substrates for plant growth and development. To investigate the impact of WBR7 on agronomic traits, we investigated some important agronomic traits in the NILs, NIL(BL130)^KO and NIL(J23B)^COM. The plant architectures and heading date were similar (Figure S11a), and there were no significant differences in panicle number per plant and spikelet number per panicle (Figure S11b,c). Compared with NIL(J23B) and NIL(BL130)^KO, NIL(BL130) and NIL(J23B)^COM showed significantly reduced grain length and significantly increased grain width, but the differences were very small, ultimately resulting in similar 1000‐grain weight among different lines (Figure S11b–f). However, NIL (BL130) and NIL(J23B)^COM exhibited a significant decrease in filled grain number per panicle and seed setting rate compared with NIL(BL130)^KO and NIL(J23B), resulting in a significant reduction in yield per plant (Figure 6a,b; Figure S11g). Moreover, NIL(BL130) and NIL(J23B)^COM displayed a lower milled rice rate, likely associated with its higher WBR (Figure 6c). In summary, WBR7 ^J23B allele with lower SUS activity contributes to increased yield and quality.

Figure 6.

WBR7 regulates rice quality and yield by modulating sucrose supply in sink organs. (a–c) Seed setting rate (a), yield per plant (b), and milled rice rate (c) of different lines (n = 48 biological replicates). (d, e) Incompletely stained pollen grain ratios (d) and pollen germination rates (e) of different lines (n = 10 biological replicates). (f, g) KI‐I₂ staining of pollen grains (f) and pollen germination analysis (g) of different lines (n = 10 biological replicates). Yellow and red arrows, respectively, indicate incompletely stained pollen grains and germinated pollen tubes. (h–j) Sucrose (h), fructose (i) and starch (j) contents in anthers of different lines (n = 6 biological replicates). Significance of differences was determined by Duncan's test. (k) Proposed model of the role of WBR7 in regulating rice quality and yield. WBR7^A with lower sucrose decomposition activity leads to decreased sucrose metabolism in the conducting organs and increased supply of sucrose in the sink organs (anther and endosperm), ultimately leading to increased grain quality and yield. The thicker arrows in the figure indicate stronger metabolic processes or increased sucrose transport. The bold words in the figure represent higher performance or content. The red arrows in the brown circles indicate germinated pollen tubes.

Anther, which is a destination for sucrose transport, act as a ‘sink’ for assimilates to provide substance and energy for the subsequent development of the anther and pollen. We speculated that the decrease in seed setting rate of NIL(BL130) and NIL(J23B)^COM may also attributed to insufficient sucrose supply, leading to aberrant pollen development (Figure 6a). Iodine staining of pollen grains from the NILs, NIL(BL130)^KO and NIL(J23B)^COM indicated that almost all pollen grains were stained, but more pollen grains from NIL(BL130) and NIL(J23B)^COM plants exhibited a higher frequency incomplete staining on one side of the outer periphery (Figure 6d,f). Pollen germination assays indicated the percentages of germinated pollen from NIL(J23B) and NIL(BL130)^KO were significantly higher than that of NIL(BL130) and NIL(J23B)^COM (Figure 6e,g). Starch in the pollen grains hydrolysed to provide energy for pollen germination and pollen tube elongation. Compared with NIL(BL130) and NIL(J23B)^COM, the sucrose and starch contents in anthers of NIL(J23B) and NIL(BL130)^KO were significantly higher, whereas no fructose content differences in anthers of the NILs and NIL(J23B)^COM (Figure 6h–j). Taken together, the reduced sucrose metabolism in conducting organs of NIL(J23B) and NIL(BL130)^KO also led to increased sucrose supply and starch content in the anthers, resulting in enhanced pollen viability and seed setting rate, ultimately increasing the yield.

Discussion

Chalkiness, an important trait in assessing rice grain quality, reduces appearance and processing quality, as well as ECQ (Fitzgerald et al., 2009; Siebenmorgen et al., 2013; Zhao and Fitzgerald, 2013; Zhou et al., 2020). In this study, we map‐based cloned WBR7, a positive regulator of WBR, and identified a functional SNP (A/G) in the 11th exon that was responsible for a difference in sucrose decomposition activity between WBR7^BL130 and WBR7^J23B (Figure 3c–f). The WBR7 ^G in BL130 originated from wild rice, and mutated to WBR7 ^A during indica domestication, resulting in lower WBR values (Figure 3h). Since the WBR7 ^A allele does not exist in japonica and a part of indica accessions, there is considerable potential to improve the yield and grain quality by replacing WBR7 ^G with WBR7 ^A (Figure 3b,g,h). Lower SUS activity in NIL(J23B) and NIL(BL130)^KO reduced sucrose metabolism in conducting organs (Figure 5) and increased the sucrose supply in the endosperm, resulting in increased synthesis of storage components and decreased development of WBR (Figure 2c,d; Figure 4). In addition, it increased sucrose supply in anthers, leading to an increased starch content in pollen grains, which ultimately contributed to increased rate of seed setting and yield per plant (Figure 6). Our findings demonstrate that the WBR7 ^A allele from indica reduces sucrose decomposition in conducting organs and increases sucrose transport to sink organs, ultimately contributing to improved yield and quality (Figure 6k).

Chalk5 and WCR1 were the first cloned QTL for chalkiness rate and white‐core rate in natural populations, respectively (Li et al., 2014; Wu et al., 2022). In a previous study, SUS3 was found to affect the level of grain white‐back rather than WBR (Takehara et al., 2018). However, that study did not explore the mechanisms by which SUS3 influenced the white‐back rate, nor did it identify functional variant sites, and no white‐back was observed in the materials used in this study. In present study, we identified that rice SUS3 (WBR7) is a major gene influencing WBR, with the allele from BL130 increasing WBR by approximately 45% in the background of J23B (Figure 1e,f). The SUS gene family is one of two types of glycosyltransferase gene families in plants, and belongs to family 4 of glycosyltransferase (GT4) (Ruan, 2014; Stein and Granot, 2019). The GT4 domain in SUS has a significant impact on glycosyltransferase activity (Ruan, 2014; Stein and Granot, 2019), mutational changes at three conserved amino acid sites in the GT4 domain of SUS3 led to a significant decrease or even loss of enzyme activity by affecting substrate binding (Huang et al., 2016). Here, we identified a functional variation in the GT4 domain of WBR7 (Figure 3c,d) that was conserved in BL130 and homologous genes in other species (Figure S5a). The mutation at this site in J23B significantly decreased sucrose decomposition activity (Figure 3e,f). SUS was reported to preferentially decompose sucrose in plants (Ruan et al., 2003). Here, we observed that the sucrose contents in endosperms and conducting organs in NIL(BL130) and NIL(J23B)^COM were significantly lower than in NIL(J23B) and NIL(BL130)^KO (Figure 4c; Figure 5a), but there was largely no difference in fructose content among these lines (Figure 4d; Figure 5b). Similarly, higher SUS activity was detected in endosperms and conducting organs in NIL(BL130) and NIL(J23B)^COM (Figure 4e,f; Figure 5c,d), suggesting lower SUS activity enhances sucrose content in conducting organs and endosperms. Therefore, the WBR7 ^A allele from indica cultivar J23B with lower SUS activity should be beneficial in improving WBR by enhancing sucrose content in endosperm.

Decomposition of sucrose is the first step in utilization of photosynthetic assimilates, a sufficient sucrose supply is crucial for plant growth and development (Ruan, 2014; Stein and Granot, 2019). In this study, the sucrose content in the endosperm of NIL(BL130) and NIL(J23B)^COM was significantly lower than that in NIL(J23B) and NIL(BL130)^KO (Figure 4c), which could be attributed to a decrease in sucrose supply or an increase in sucrose decomposition. The metabolism of sucrose in endosperm is positively correlate with the content of storage components. In many studies, storage component contents in endosperm, especially starch and storage protein were found to be antagonistic involved in regulation of carbon–nitrogen balance (Li et al., 2014; Peng et al., 2014a). However, the contents of sucrose and storage components in NIL(J23B) and NIL(BL130)^KO endosperm significantly higher than in NIL(BL130) and NIL(J23B)^COM, but the fructose content was not (Figure 4c,d,h–j), indicating that WBR7 affects the synthesis of storage components in the endosperm by regulating sucrose supply rather than sucrose decomposition. Excessive grain filling rate can lead to an increase in chalkiness, in this study, a faster filling rate was observed in NIL(BL130) (Figure 4a), which had noticeable white‐belly in endosperm at 12 DAF (Figure 4b). Sucrose is the source of carbohydrate for grain filling (Ma et al., 2017; Wang et al., 2008). However, the sucrose and fructose content differences in the endosperm of NILs, NIL(J23B)^COM and NIL(BL130)^KO and NIL(J23B) (Figure 4c,d)indicating that the faster filling rate in NIL(BL130) might not determine by sucrose content but was possibly due to higher SUS activity, and therefore promoting grain filling. On the other hand, insufficient sucrose supply in the mid‐grain‐fill stage in NIL(BL130) endosperm might cause developmental obstacles in the grain, resulting in more chalkiness.

SUS was reported provides substrates for cellulose synthesis in conducting organs by breaking down sucrose in many species (Fujii et al., 2010; Ruan et al., 2003), including rice (Fan et al., 2019), but there is little research on the mechanism of sucrose metabolism and regulation in conducting and sink organs. In this study, lower sucrose metabolism in conducting organs of NIL(J23B) and NIL(BL130)^KO resulted in more sucrose transported to the endosperm (Figure 4; Figure 5), leading to an increase in synthesis of storage components in the endosperm (Figure 4), as reported for SUS mutants in Arabidopsis (Barratt et al., 2009). We propose that sucrose decomposition regulated by WBR7 in conducting organs promotes further sugar metabolism, resulting in a decrease in sucrose supply in sink organs, as verified for another sink organ, the anther, in this study (Figure 6). An insufficient sucrose supply in the anther can lead to a decrease in pollen fertility (Chen et al., 2007; Wang et al., 2021). We found that decreased sucrose content in anthers in NIL(BL130) and NIL(J23B)^COM accompanied by lower pollen viability, and the difference in pollen viability between the NILs, NIL(BL130)^KO and NIL(J23B)^COM plants were consistent with the difference in seed setting rate (Figure 6). Therefore, plants with lower SUS activity exhibited decreased sucrose metabolism in conducting organs, resulting in increased sucrose supply in sink organs such as endosperm and anthers, ultimately leading to an increase in grain yield and quality.

In summary, we identified WBR7, a major gene encoding rice sucrose synthase 3, which positively regulated grain WBR. We elucidated the mechanisms by which WBR7 regulated sucrose allocation to conducting and sink organs by modulating sucrose metabolism, and its impact on yield and quality. However, there are still unresolved questions that require further investigation. For example, the impact of functional variation on the activity of WBR7 remains unclear and the reason for the faster filling rate in NIL(BL130) is still unknown. It could be due to the stronger sucrose decomposition activity of WBR7^G or related to sugar signalling. Nevertheless, this study provides new insights and genetic resources for understanding how sugar metabolism regulates yield and grain quality in rice.

Materials and methods

Plant materials

Rice populations were planted under natural field conditions in the experimental stations at Huazhong Agricultural University at Wuhan (N 30.49, E 114.36) in summer and at Lingshui (N 18.51, E 110.04) in winter. Twelve 30‐day‐old seedlings of each line were transplanted in the field with 16.5‐cm spacing in single‐row plots; rows were 26.4 cm apart. Field management followed local practices. Ten plants from the middle of each row were harvested individually for trait measurement.

A BC₅F₂ population consisting of 192 individuals was used for validation of the effect of qWBR7 in summer 2017. To fine‐map qWBR7, a population consisting of 6000 BC₅F₄ plants was grown in summer 2018 for recombinant screening. Progenies of 84 recombinants were grown in winter 2018, and validated in summer 2019.

To precisely evaluate the effect of WBR7 on the regulation of WBR, we developed a BC₆F₂ population in summer 2019. During 2020 and 2021 summer, BC₆F₄ generation NILs and T₂ generation transgenic lines were grown at Wuhan to validate the impact of WBR7 on agronomic traits and investigate the regulatory mechanism of WBR7 on WBR.

Fine‐mapping of qWBR7

The genetic effect of qWBR7 was validated in a segregating BC₅F₂ population, the flanking markers RM455 and RM234 of qWBR7 were used to analysis the genotype of each plant. Local interval mapping was performed, and recombination values were converted to centiMorgans using the Kosambi mapping function in the MapMaker/Exp3.0 software (Lander et al., 1987). QTL effect analysis was conducted using the Windows QTL cartographer software (Wang et al., 2005). To fine‐map qWBR7, 13 newly developed markers in the interval flanked by RM455 and RM234 were used to genotype the recombinants selected from a BC₅F₄ population of 6000 plants.

Markers for QTL mapping were designed based on sequence differences between the parents. The parental sequences within the 5.74 kb interval of WBR7 were confirmed using Sanger sequencing technology.

Investigation of yield and quality‐related traits

Panicles of each plant were used to measure the number of filled grains, seed setting rate, panicle number, panicle length, primary and secondary branch number. Threshed seeds from each plant were air‐dried and stored at room temperature for 3 months before further phenotyping. Fully filled grains on each plant were used to measure grain size (length and width), 1000‐grain weight and yield per plant. Forty‐eight plants in each line were harvested for phenotyping. Milled rice rate was estimated as the percentage weight of milled grains to total grain weight. White‐belly rate (WBR) was scored by the visual assessment of the percentage of white‐belly grains in random samples of more than 100 dehulled grains from each plant. For each line, the milled rice rate and WBR were measured in at least 10 biological replicates.

Detection of storage components

Flour ground from milled grain was used to measure total starch content and amylose content. Total starch and amylose were measured using a K‐AMYL reagent kit (Megazyme, Ireland) following the manufacturer's instructions. Flour glutelin, prolamin, globulin and albumin contents were measured based on the previously published methods (Kumamaru, 1988), and the total storage protein content was the sum of the four measures. All assays were performed with a minimum of eight biological replicates.

Filling rate survey

The flowering spikelets on main panicles on the second day of flowering were marked at noon on the day of flowering. Developing grains were sampled at different stages after flowering for measuring the filling rate, with an average of more than 100 grains per line. The marked panicles were cut, placed in preservation bags and taken to the laboratory. The marked filled grains were sampled and placed in an oven at 80 °C for three days. The endosperm was stripped off with forceps to measure dry weight. Each measurement was conducted on 30 endosperms, with a minimum of three measurements for each line.

RNA extraction and qRT‐PCR

Total RNA was extracted from plant tissues using Invitrogen RNA Extraction Kit 15 596–026 (TRIzol, USA). First‐strand cDNA synthesis was performed in a 20 mL reaction mixture containing 2 mg of RNA and 200 U of M‐MLV reverse transcriptase (Invitrogen C28025‐014, USA). qRT‐PCR was conducted on a QuantStudio 6 Flex instrument using the SYBR Green PCR reagent, following the manufacturer's instructions. Tissues at different developmental stages were obtained based on flowering time. For caryopses samples, more than 30 grains were collected from the upper part of the panicle of each plant, and endosperm and husk were sampled separately. For conducting organs, the mid‐2‐3‐cm segments of flag leaves, flag leaf sheaths and stems were sampled. Data for each sample were based on 3–5 biological replicates from different plants, with three technical replicates for each biological replicate. The rice ACTIN1 gene was used as an internal reference for normalizing gene expression.

Vector construction and transformation

To confirm the function of WBR7, we obtained a 2 kb promoter fragment and a full‐length coding fragment of the WBR7 ^BL130 by PCR. The two fragments were then inserted into the plant binary vector pCAMBIA1301 using a one‐step ligation method, and the positive vectors named CO^BL130. To obtain WBR7 knockout plants, the 20 bp target sequence in the first exon of WBR7 was inserted into the intermediate vector pER8‐Cas9‐U6 and cloned into pCXUN‐Cas9 according to a previously described method (Gao and Zhao, 2014). The sequences of all vectors were confirmed by PCR and Sanger sequencing technology. These constructed vectors were introduced into Agrobacterium tumefaciens strain EHA105 and transformed into the relevant materials by Agrobacterium‐mediated transformation (Toki, 1997).

Subcellular localization

The coding sequence amplified from cDNA obtained through reverse transcription was inserted into pM999‐YFP, generating WBR7‐YFP fusion plasmids. WBR7‐YFP and the cytoplasmic localization marker pYBA1138‐RFP were co‐transformed into rice protoplasts. Fluorescence protein imaging was performed after cultivation for 12 h of darkness using a confocal laser scanning microscopy (Leica TCSSP2, Germany). The fluorescence experiments were repeated at least three times.

Scanning and transmission electron microscopy

Mature grains for scanning electron microscopy (JEOL JSM‐6390LV, Japan) were cut horizontally with a knife and coated with gold under vacuum conditions. The morphologies of starch granules in the endosperm were observed across 30 nm in at least three biological replicates. All procedures were conducted according to the manufacturer's instructions.

Transmission electron microscopy (HITACHI H‐7650, Japan) analysis of the starch granules and protein bodies in the developing endosperm was performed at 2000× magnification on developing seeds at 7 DAF, following a previously described method (Wang et al., 2010). Transmission electron microscopy analysis was performed on at least three biological replicates.

Haplotype assays, phylogenetic analysis and amino acid sequence homology alignment

Haplotypes for the 5.74 kb region of WBR7 of 533 cultivated rice accessions were identified based on sequence variations (frequency >0.05) in BL130 and J23B. The sequence variation data and subspecies information of WBR7 among 533 accessions were obtained from RiceVarMap (http://ricevarmap.ncpgr.cn/). We evaluated WBR in all 533 accessions, after excluding missing and floury grain accessions, the remaining 498 accessions were used for haplotype analysis. Phylogenetic analysis was performed using the neighbour‐joining method in Mega7.0 (Kumar et al., 2016). The amino acid sequence of WBR7 was compared across species using BLASTP alignment in the NCBI database (http://www.ncbi.nlm.nih.gov/), and sequence alignment was performed using ClustalW (Larkin et al., 2007).

Assays of enzyme activity

Conducting organs at 0 DAF and endosperms at 4 DAF and 8 DAF were collected from the field and stored at −80 °C. The decomposition and synthesis activity detection methods for SUS were performed following previously described methods with slight modifications (Li et al., 2018). For each line, enzyme activity assays were performed with a minimum of six biological replicates.

Recombinant enzymes and enzyme activity assays

According to predicted results of functional variation, we amplified three cds fragments of WBR7 to insert into the prokaryotic expression vector pET‐28a‐SUMO containing His‐tag and SUMO‐tag sequences and transformed into the bacterial strain Escherichia coli strain BL21 (DE3), and named them WBR7^J23B, WBR7^BL130 and WBR7^BHJ, respectively. Three proteins were induced for expression in LB medium at a final concentration of 0.1 mM isopropyl β‐D‐1‐thiogalactopyranoside (IPTG) at 18 °C for 8 h. The purification of the three recombinant proteins was performed using Ni‐NTA agarose resin (Thermo Fisher Scientific, USA) according to the manufacturer's protocol. 1 μg of purified protein was incubated with a mixture of 0.1 mL of 100 mM sucrose with 0.3 mL of reaction medium (containing 100 mM HEPES‐NaOH, pH 5.5; 5 mM NaCl; and 1.8 mM UDP) at 30 °C for 15 min, followed by immediate boiling in water for 1 min to inactivate the protein. Next, 0.5 mL of 3,5‐dinitrosalicylic acid (27.6 mM) was added to the assay mixture and incubated in boiling water for 5 min. After cooling, the absorbance of the assay mixture was measured at 540 nm. The unit of enzyme activity was defined as the amount of enzyme that produces 1 μmol of product under the reaction conditions. Each measurement was conducted with a minimum of five biological replicates.

Transcriptome and metabolomics analysis

The endosperms from spikelets of NILs marked on the day of flowering were collected at 4 DAF. Total RNA was extracted using Invitrogen RNA Extraction Kit 15 596–026 (TRIzol, USA). mRNA sequencing library construction, sequencing and differential analysis were performed by Novogene (Tianjin, China). Three biological replicates were conducted for each line.

Content of central carbon metabolites (CCM) in endosperm of NILs, NIL(BL130)^KO and NIL(J23B)^COM plants at 8 DAF was performed by Novogene. Three biological replicates were conducted on each line.

Assays of sugar and cellulose contents

Measurements of sucrose and fructose contents in endosperm and conducting organs at different filling stages were performed using the Megazyme kit K‐RAFGL raffinose/sucrose/D‐glucose (Megazyme, Ireland) following the manufacturer's instructions. Cellulose in conducting organs was measured using the reagent kit CLL‐1‐Y (Cominbio, Suzhou, China) following the manufacturer's instructions. All assays were performed with a minimum of six biological replicates.

Sucrose transport assay

Sucrose transport of was analysed using a ¹³C‐labelled sucrose feeding experiment. Spikelets from main stems of the NILs, NIL(BL130)^KO and NIL(J23B)^COM, plants were marked on the day of flowering. Main stems at 4 DAF cut from second internodes were immediately placed in water and brought to the laboratory. The main stems were cultured in 5 mM ¹³C‐sucrose in darkness at room temperature for 24 h. After drying at 80 °C for 3 days, the stems and endosperms from the same plants were crushed to powder and sieved through a 100‐mesh sieve. The amount of stable carbon isotope ¹³C was determined using an Elementar Vario PYRO cube elemental analyzer interfaced with the Isoprime 100 IRMS system (Isoprime Ltd., UK), and the ¹³C ratios and total carbon ratio were quantified using internal standards. Three biological replicates were conducted on each line.

Pollen vitality analysis

To observe mature pollen, spikelets at the second day of flowering were collected and placed in a centrifuge tube containing 75% ethanol and stored at room temperature. The anthers were removed from the spikelet and placed on a glass slide, followed by the addition of 1–2 drops of 1% iodine‐potassium iodide solution. Anthers were then crushed with forceps to release the pollen grains, and a cover slip was placed on top. Slight pressure was applied with forceps. Observations were made after 2–3 min using a stereo light microscope (Nikon, SMZ25). Ten biological replicates were conducted on each line.

Pollen was germination on an artificial culture medium (250 mM sucrose; 10% (w/v) polyethylene glycol 4000; 0.3 mM calcium nitrate tetrahydrate; 4 mg/L boric acid; 0.3 mg/L vitamin B1; 0.8% (w/v) low‐gelling‐temperature agarose II; pH 5.8) described by previous researchers (Liu et al., 2016). Pollen was dusted on to the medium from flowering spikes by gentle shaking. Pollen germination rate was determined after incubation for 15 min at 37 °C using a stereo light microscope (Nikon SMZ25, Japan). Observations were made on no fewer than 300 pollen grains at 10 times for each line.

Primers

Primers used in this study are listed in Tables S3.

Statistical analysis

Data are presented as means ± standard deviation. Significant differences between paired groups were analysed by two‐tailed t‐tests, and multiple comparisons were assessed for significant differences by Duncan's tests (P < 0.05) using the SPSS software.

Conflict of interest

The authors declare they have no conflict of interest.

Author contributions

Y.H. and A.Y. conceived the project; H.S. designed the research and analysed the data under the supervision of Y.H.; Y.Z. and P.Y. constructed the foundational materials; L.W., Y.W. and S.C. assisted with part of experiments and data analysis; G.G., Q.Z. and J.X. contributed the reagents and experimental support; Y.L. and L.X. provided guidance for this study; H.S. wrote the manuscript and P.L., H.Z., R.L. and Y.L. improved it.

Supporting information

Figure S1 White‐belly rate performance of BC₅F₂ population and plant architecture of NILs.

Figure S2 Differences in WBR7 expression in conducting organs and endosperms of NILs at different grain filling stages.

Figure S3 Construction of WBR7 transgenic lines.

Figure S4 Relative expression level of WBR7 in transgenic complementary lines.

Figure S5 Sequence alignment of homologous WBR7 proteins.

Figure S6 Histology of endosperm in NILs.

Figure S7 Sugar content in developing endosperms of NILs.

Figure S8 Transcriptome analysis of 4 days after flowering endosperm of NILs.

Figure S9 Comparison of central carbon metabolites contents in 8 days after flowering endosperm in different lines.

Figure S10 Sucrose and fructose contents, and expression levels of sucrose metabolism‐related genes in conducting organs.

Figure S11 Yield‐related traits of different lines.

Table S1 Genetic effect of qWBR7.

Table S2 Enzymatic properties of WBR7 proteins.

Table S3 Primers used in this study.

Data S1 WBR of 498 accessions.

Data availability statement

The data that support the findings of this study are available in the supplementary material of this article.