Creation of high‐resistant starch rice through systematic editing of amylopectin biosynthetic genes in rs4

Summary

Resistant starch (RS) is a special kind of starch with beneficial effects on obesity, type 2 diabetes and other chronic complications. Breeding high‐RS rice varieties is considered a valuable way to improve public health. However, most rice cultivars only contain an RS level lower than 2% in cooked rice, and cloning of RS genes is critical to improve RS levels in rice. The loss of function of Starch Synthases IIIa (SSIIIa) and SSIIIb, two amylopectin biosynthetic genes, could elevate RS levels up to 10%. Here, we performed a systematic genetic study of 14 amylopectin biosynthetic genes in the ssIIIa ssIIIb double mutant via genome editing, and investigated their effects on RS formation, the eating quality and grain yield. The results showed that deficiency in SSIIa, SSIVb or ISA2 under the ssIIIa ssIIIb background could each elevate RS content to above 14%, and the quadruple mutants of sbeI sbeIIb ssIIIa ssIIIb and sbeI ssIVb ssIIIa ssIIIb could further increase RS levels to over 18%. Furthermore, the eating quality of cooked rice and grain yield decreased along with the elevated RS contents, showing a trade‐off among these traits. In these mutants, ssIIIa ssIIIb showed the balanced performance of RS and grain yield. This study provides insights into RS biosynthesis with a series of RS genes in the amylopectin biosynthesis pathway and practical strategy to breed high‐RS rice varieties with balanced performance.

Introduction

Starch provides ~50% energy source for human (Miao and Hamaker, 2021). Starch granules with semi‐crystalline structure are the primary form of starch, which are composed of two basic glucose polymers: amylose and amylopectin. Amylose is synthesized by granule bound starch synthase (GBSS), whereas amylopectin is synthesized by three classes of enzymes: starch synthases (SS), starch branching enzymes (SBE) and debranching enzymes (DBE) (Huang et al., 2021). For nutritional purposes, starch can be classified as rapidly digestible, slowly digestible and resistant starch (RS). RS is a special kind of starch that cannot be digested in the small intestine and passes to the colon to be fermented by microbiota (Shen et al., 2022). Thus, RS in food is beneficial to type 2 diabetes, weight management and inflammatory bowel disease (Birt et al., 2013; Li et al., 2024). However, most staple food contain a small amount of RS, for example, nearly all rice natural varieties only contain less than 2% RS, far below the recommended level (Patterson et al., 2020). Therefore, breeding of high‐RS rice is of great demand for public health improvement.

Recently, several high‐RS rice mutants have been generated and key functional genes for RS synthesis have been cloned, which are all amylopectin biosynthetic genes. Deficiency in SSIIIa could increase RS contents in cooked rice, which was dependent on the strong allele of Waxy (Wx) encoding GBSSI (Zhou et al., 2016). Deficiency in SBEIIb could also increase RS contents (Zhu et al., 2012). Meanwhile, although the single mutant of sbeI or ssIIIb showed no significant effect on RS levels, the double mutant sbeI sbeIIb or ssIIIa ssIIIb could further elevate RS contents of the sbeIIb or ssIIIa single mutant, respectively (Huang et al., 2023; Miura et al., 2021; Wang et al., 2023; Zhu et al., 2012). Therefore, it is considered that the highly branched amylopectin could be conducive to being catalyzed by β‐amylases, whereas the amylose with rare branches is hard to be digested and critical to RS formation (Ahuja et al., 2013). However, the function of other amylopectin biosynthetic genes on RS formation remains largely unknown.

In previous study, we identified a high‐RS mutant rs4 with RS levels up to 10%, which was an ssIIIa ssIIIb double mutant in the indica rice cultivar R7954 background harbouring the strong Wx ^a allele (Wang et al., 2023). In this study, we expected that the mutation of amylopectin biosynthetic genes may further increase the RS contents and performed a systematic genetic study of SSs, SBEs and DBEs via genome editing in the rs4 background. We found that the mutation of specific amylopectin biosynthetic gene could elevate RS levels to above 14% and double mutations of certain amylopectin biosynthetic genes showed increased RS levels over 18% in the ssIIIa ssIIIb background, however, the performance on eating and cooking quality of cooked rice and yield decreased along with the increased RS contents. Among these mutants, ssIIIa ssIIIb showed the balanced performance of RS and grain yields. This study provided insights into RS biosynthesis with a series of RS genes in amylopectin biosynthesis pathway.

Results

Loss of function of specific amylopectin biosynthetic gene elevates RS contents in the ssIIIa ssIIIb background

We designed sgRNAs targeting 14 amylopectin biosynthetic genes, including 7 SSs of SSI, SSIIa, SSIIb, SSIIc, SSIVa, SSIVb, SSV, 3 SBEs of SBEI, SBEIIa, SBEIIb, and 4 DBEs of isoamylase 1 (ISA1), ISA2, ISA3, and pullulanase (PUL), to generate mutants in the rs4 background. From the genotyping results of T₂ plants, we obtained the homozygous frameshift and transgenic‐free mutants of 13 genes with at least two independent lines except SSI (Figures S1‐S3). For SSI, we only obtained one homozygous in‐frame deletion mutant rs4 ssI ^−6bp (Figure S1a), similar to the previous research that failed to obtain homozygous ssI ssIIIa null mutant (Hayashi et al., 2015), indicating that ssI ssIIIa ssIIIb triple mutants may be lethal. The measurement of RS contents in cooked rice showed that the deficiency in SSIIa, SSIVb or ISA2 in rs4 could increase RS levels to over 14% (Figure 1a), and the genome editing of other seven amylopectin biosynthetic genes could also slightly increase RS levels. Meanwhile, although rs4 ssI ^−6bp was an in‐frame mutant, its RS content was also significantly increased to 15% (Figure S4a), indicating that rs4 ssI ^−6bp was a leaky mutant and SSI was a functional RS gene. We then tested whether the accumulation of mutations in amylopectin biosynthetic genes could further increase RS contents. We chose SBEI, SBEIIb and SSIVb with divergent functions, and generated all combinations of their double mutants in the rs4 background via CRISPR/Cas9 (Figure S5). Interestingly, the RS contents of homozygous rs4 sbeI sbeIIb ^cr and rs4 sbeI ssIVb ^cr mutants were further increased to over 18% (Figure 1b), whereas rs4 ssIVb sbeIIb ^cr showed a similar RS content as rs4 ssIVb ^cr , demonstrating that double mutations of certain amylopectin biosynthetic genes could further increase RS contents to over 18% in the ssIIIa ssIIIb background. All these results demonstrated that the mutation of specific amylopectin biosynthetic gene could further increase the RS contents in the ssIIIa ssIIIb background (Figure 1c).

Figure 1.

The mutation of specific amylopectin biosynthetic genes further increased the RS content in the ssIIIa ssIIIb background. (a, b) RS contents in cooked rice of single (a) and double (b) amylopectin biosynthetic gene mutants generated by genome editing in the rs4 background. (c) Summary of RS genes in amylopectin biosynthetic genes. RS genes were labelled in red. In (a), ** P < 0.01, *** P < 0.001, ns, not significant; based on two‐tailed Student's t test. Values are means ± s.d. (n = 3 biological replicates) and different letters at the top of each column in (b) indicate a significant difference at P < 0.05 determined by Tukey's HSD test.

High‐RS mutants exhibit altered appearance quality and starch granule morphology

The increased RS levels could lead to significant decrease in grain quality and grain yield (Chen et al., 2022; Miura et al., 2021), thus the balance between RS levels and agronomical traits is critical to breeding high‐RS varieties. To further investigate the effect of RS genes on eating and cooking quality, we selected eight lines with different RS levels, including wild‐type R7954 (RS < 2%), single mutants with RS 5%–10% of two previously reported RS genes SSIIIa and SBEIIb (Figure S4b) (Zhou et al., 2016; Zhu et al., 2012), double mutant rs4 (ssIIIa ssIIIb) (RS 10%–14%), and above generated triple mutants rs4 ssIIa ^cr , rs4 ssIVb ^cr (RS 14%–18%) and quadruple mutants rs4 sbeI sbeIIb ^cr and rs4 sbeI ssIVb ^cr (RS > 18%) for further study.

We first observed the appearance quality of grains, and found that all examined mutants displayed floury seed (Figure 2; Figure S6). Then, we examined the morphology of starch granule via scanning electron microscope (SEM). In wild type, the starch granules displayed a regular and uniform polyhedral structure, whereas the examined mutants exhibited various morphologies in shape and size. Starch granules of ssIIIa, sbeIIb ^cr and rs4 were variable in size with rounded surfaces, consistent with the previous reports (Wang et al., 2023; Zhu et al., 2012). Similarly, starch granules of rs4 ssIIa ^cr were also variable in size. However, rs4 ssIVb ^cr , rs4 sbeI sbeIIb ^cr and rs4 sbeI ssIVb ^cr had small starch granules with spherical structures, especially the granule size of rs4 ssIVb ^cr was even smaller (Figure 2; Figure S6). The severely altered starch granule of rs4 ssIVb ^cr may be due to the crucial role of SSIIIa and SSIVb in determining starch granule morphology and maintaining the amyloplast envelope structure (Toyosawa et al., 2016). The loosely packed structure of starch granule in these mutants may lead to abnormal appearance quality and grain filling.

Figure 2.

Morphology of seeds of eight selected RS lines. Intact seeds and transverse sections revealed by light microscopy (first column), and scanning electron micrographs of the endosperm in transverse sections with different magnifications (second to the fourth column) are shown.

Eating and cooking quality is deteriorated in high‐RS mutants

To study the starch composition of these eight lines, we analysed the apparent amylose content (AAC) and the total starch contents. We found that in these lines, AAC gradually elevated in accordance with the increase of RS contents in cooked rice and the linear correlation was very high (Pearson Correlation Coefficient, PCC = 0.98) (Figure 3a), which is consistent with the previous findings that the AAC positively correlated with RS content (Ding et al., 2021). The total starch contents were slightly decreased in these RS mutants compared with the wild type (Figure 3b). We further investigated the RS contents in raw rice and found that RS levels in raw rice were also positively correlated with the AAC in these RS mutants (PCC = 0.84) (Figure 3c), but the correlation was lower than that between the AAC and RS levels in cooked rice. Interestingly, we found that sbeIIb ^cr and rs4 ssIVb ^cr showed higher RS contents in raw rice than the mutants with higher AAC, suggesting that SBEIIb and SSIVb may have specific functions in the formation of different types of RS. The independent lines with the same mutation combinations showed consistent results (Figure S7). These results demonstrated that the high level of AAC was critical to both RS in cooked rice and raw rice in these mutants.

Figure 3.

Physical–chemical properties of eight selected RS lines. (a–c) Apparent amylose content (AAC) (a), total starch content (b) and RS contents in raw rice (c) of eight selected RS lines. (d–i) rapid viscosity analysis (RVA) profiles (d), Pasting temperature (PT) (e), peak viscosity (PV) (f), trough viscosity (TV) (g), breakdown viscosity (BDV) (h), and final viscosity (FV) (i) of eight selected RS lines. In (a–c), (e–i), the lines with RS <5%, 5%–10%, 10%–14%, 14–18% or > 18% are shown in grey, blue, green, yellow, and red bars, respectively. In (a–c), (e–i), values are means ± s.d. (n = 3 biological replicates) and different letters at the top of each column indicate a significant difference at P < 0.05 determined by Tukey's HSD test.

To test the changes in starch quality of these mutants, we analysed their starch pasting properties (Figure 3d; Figure S8a). Pasting temperature (PT) gradually increased along with the RS levels when RS < 14%, and reached high levels above 90 °C in all mutants with RS > 14% (Figure 3e; Figure S8b). Likewise, the peak viscosity (Figure 3f; Figure S8c), trough viscosity (Figure 3g; Figure S8d) and breakdown viscosity (Figure 3h; Figure S8e) dramatically decreased in all mutants and reached low level less than 50 cP in all mutants with RS > 10%, and final viscosity significantly declined to less than 100 cP in mutants with RS > 14% (Figure 3i; Figure S8f). Hence, the eating and cooking quality of high‐RS mutants were inevitably deteriorated especially for the mutants with RS > 14%.

Thousand‐seed weight and grain yield are impaired in high‐RS mutants

Furthermore, we grew these mutants in field and measured the yield‐related traits (Figure 4a). The grain yield per plant was significantly decreased compared with R7954 and showed strong anti‐correlation with RS contents (PCC = −0.94) in all these mutants (Figure 4b). We then investigated the key attributes contributing to this yield loss and found that the reduced 1000‐seed weight was the major reason (PCC = −0.93 with RS contents) for the yield loss in these mutants (Figure 4c). Different with the strong changes in grain yield per plant and 1000‐seed weight, these mutants showed subtle changes in grain number per main panicle, seed setting rate, tiller number, and plant height compared to R7954 (Figure S9). The independent lines with the same mutation combinations showed consistent results in decreased grain yield per plant and 1000‐seed weight (Figure S10a,b). Taken together, the increased RS contents was anti‐correlated with the reduced 1000‐seed weight and yield.

Figure 4.

Yield‐related traits were significantly affected in eight selected RS lines. (a) Plant morphologies of eight selected RS lines, bars = 20 cm. (b–c) Grain yield per plant (b) and 1000‐seed weight (c) of eight selected RS lines in Beijing. (d–e) Total RS (d) and AAC yield (e) per plant of eight selected RS lines. (f) Model of relative RS contents, AAC, PT, and grain yield in the wild type (WT), ssIIIa, ssIIIa ssIIIb (rs4), rs4 ssIIa, and rs4 sbeI ssIVb mutants. In (b‐c) values are means ± s.d. (n = 10 plants), in (d–e) values are means ± s.d. (n = 3 biological replicates) and different letters at the top of each column indicate a significant difference at P < 0.05 determined by Tukey's HSD test.

ssIIIa ssIIIb showed balanced performance in RS and grain yield

To balance the trade‐off between RS levels and grain yield, we calculated the theoretical RS yield per plant through multiplying RS contents with grain yield per plant of individual mutants (Figure 4d). We found that the RS yield per plant were elevated along with the RS content before 10%, and then the RS yield reached a ceiling level for high‐RS mutants with RS content over 10%.

Amylose also has valuable industrial application potential in food and non‐food usages, such as thicker, stabilizer, biodegraded package, drug delivery and so on (Obadi et al., 2023). Since these RS mutants also exhibited greatly increased AAC levels, we then calculate the AAC yield per plant (Figure 4e). The mutants with RS content less than 10% showed higher AAC yield per plant than mutants with more than 10% RS. Among them, rs4 had the highest AAC yield per plant. Taken together, rs4 with 10% RS contents showed the balanced performance in RS, AAC and grain yields per plant among these RS mutants.

Discussion

In previous studies, only a few RS genes have been cloned and all of them functioned in amylopectin biosynthesis pathway. Deficiency in both SBEI and SBEIIb could further elevate the RS contents in the sbeIIb mutant to over 15% (Miura et al., 2021; Zhu et al., 2012). The RS levels of SBE family gene multiplex edited heterozygous mutants reached 15% (Biswas et al., 2023). Recently, multiplex genome editing of starch synthetic genes in barley generated sbeIIa sbeIIb and ssIIa sbeIIa sbeIIb mutants which elevated RS contents to 12% and 14%, respectively (Yang et al., 2024). The function of other amylopectin biosynthetic genes on RS formation remains largely unknown. Here, through systematic study on amylopectin biosynthetic gene, we identified a series of uncovered RS genes. Loss of function of 10 amylopectin biosynthetic genes could each significantly increase RS levels under the rs4 background, among which the deficiency in SSIIa, SSIVb or ISA2 could elevate RS contents to above 14% (Figure 1a). Furthermore, we found that the accumulation of multiple RS genes could further increase RS levels. The quadruple mutants of sbeI sbeIIb ssIIIa ssIIIb and sbeI ssIVb ssIIIa ssIIIb could each increase RS levels to over 18% (Figure 1b). Meanwhile, with the elevation of RS contents, the 1000‐seed weight and grain yield of these mutants were inevitably decreased (Figure 4b,c). To balance the RS content and yield, rs4 with 10% RS contents showed the optimized performance of RS and grain yield among these mutants. In summary, we propose a model showing the relative levels of RS, AAC, PT, and grain yield among the RS mutants of amylopectin biosynthetic genes (Figure 4f; Table S1).

According to the physical structure, RS can be divided into five types of RS1–RS5. RS3 and RS5 mainly exist in cooked rice, whereas RS2 mainly exists in raw rice (Shen et al., 2022; Yi and Li, 2022). The enzymatic hydrolysis of native rice starch can be affected by amylose content, starch hierarchical structures and physical barriers of starch granules to form RS2 (Li et al., 2020). During gelatinization process, the crystalline structure of starch is destroyed (Wang and Copeland, 2013), and the disordered amylose and amylopectin are aggregated to form a recrystallized gel network with high retrogradation rate to generate RS3 (Wang et al., 2015). Increased gelatinized temperature is associated with increased RS3 content, however, the casual relationship between them is still obscure. Amylose content is considered the main controller in determining the starch digestibility of gelatinized and retrograded rice (Ding et al., 2021; Yi and Li, 2022). In accordance with the previous reports, our results showed that the AAC exhibited a strong linear correlation to RS3 but a weak linear correlation to RS2 (Figure 3a,c). Interestingly, both sbeIIb ^cr and rs4 ssIVb ^cr exhibited higher RS2 levels than mutants with similar or higher RS3, indicating that SBEIIb and SSIVb may have specific functions in RS2 formation. These mutants could also be used to study the different mechanism in RS2 and RS3 formation and their relationship with other factors such as gelatinized temperature. Moreover, we obtained quadruple mutants with RS contents higher than triple mutants (Figure 1b), which supported the idea that pyramiding more mutations of amylopectin biosynthetic genes could further elevate RS contents (Fujita et al., 2022). In future, multiplex genome editing or cross with existed high RS mutants may allow us to generate quintuple mutants or sextuple mutants that may achieve even higher RS levels.

Inhibition of the amylopectin biosynthesis is an efficient way to increase RS contents, however, it also causes an avoidable loss of grain yield. Starch biosynthesis genes may also affect starch in leaf, stem and other tissues to influence yield‐related traits (Dong et al., 2024; MacNeill et al., 2017). The level of yield loss may vary in different rice varieties, for example, the ssIIIa ssIIIb mutants in japonica background showed lower levels of 1000‐seed weight loss compared with the ssIIIa ssIIIb mutants in indica background (Huang et al., 2023; Wang et al., 2023). Previous study also found that backcrossing high RS mutant with elite rice cultivars could elevate the seed weight without compromising the RS content (Miura et al., 2022). In future, it is of great importance to identify the genes other than amylopectin biosynthetic genes to increase RS content without yield penalty. In addition, digestion rate of rice may also be affected by other attributes rather than RS, which could be used to optimize the trade‐off between eating and cooking quality of cooked rice and digestibility in rice breeding (Buenafe et al., 2021; Parween et al., 2020; Zhu et al., 2023). In conclusion, this study identified a series of RS genes and provided new insights into RS biosynthesis and new strategy to breed high‐RS rice varieties with balanced overall performance.

Accession numbers

The name and ID for the genes mentioned in this study are as follows: SSI (LOC_Os06g06560, Os06g0160700), SSIIa (LOC_Os06g12450, Os06g0229800), SSIIb (LOC_Os02g51070, Os02g0744700), SSIIc (LOC_Os10g30156, Os10g0437600), SSIIIa (LOC_Os08g09230, Os08g0191433), SSIIIb (LOC_Os04g53310, Os04g0624600), SSIVa (LOC_Os01g52250, Os01g0720600), SSIVb (LOC_Os05g45720, Os05g0533600), SSV (LOC_Os02g56320, Os02g0807100), SBEI (LOC_Os06g51084, Os06g0726400), SBEIIa (LOC_Os04g33460, Os04g0409200), SBEIIb (LOC_Os02g32660, Os02g0528200), ISA1 (LOC_Os08g40930, Os08g0520900), ISA2 (LOC_Os05g32710, Os05g0393700), ISA3 (LOC_Os09g29404, Os09g0469400), PUL (Os04g0164900).

Materials and methods

Plasmid construction

Single‐guiding RNAs (sgRNAs) were designed through the CRISPR‐GE tool (http://skl.scau.edu.cn/) (Xie et al., 2017). For single gene editing in R7954 or the mutant rs4, two sgRNAs were designed for each gene targeting coding region. 28 sgRNAs were cloned into the CRISPR/Cas9 vector VK005‐1 (Beijing Viewsold Biotech) by enzyme digestion and ligation methods. For multiplex gene editing, three pYLCRISPR/Cas9P_ubi‐H‐based constructs were prepared, each carrying two sgRNAs expression cassettes driven by the OsU3 and OsU6a promoters for coding region of combination of SSIVb, BEIIb and BEI. CRISPR/Cas9 vector pYLCRISPR/Cas9P_ubi‐H were constructed as previously reported (Ma et al., 2015). Primers of designed sgRNAs were listed in Table S2.

Generation of edited plants

Oryza sativa L. ssp. indica cv. R7954 was used as the wild type to generated the sbeIIb edited line. The rs4 mutant was used to generate ssI, ssIIa, ssIIb, ssIIc, ssIVa, ssIVb, ssV, sbeI, sbeIIa, sbeIIb, isa1, isa2, isa3, pul, ssIVb sbeIIb, sbeI sbeIIb, and sbeI ssIVb edited rice lines. The vectors were separately transformed into the Agrobacterium tumefaciens strain EHA105, the strains were mixed with equal amounts for same targeted genes and used to transform the rs4 or R7954 callus. Agrobacterium‐mediated transformation was performed as previously described (Hiei et al., 1994; Zhang et al., 2016).

Plant materials and growth conditions

R7954, sbeIIb ^cr , rs4, rs4 ssI ^−6bp , rs4 ssIIa ^cr , rs4 ssIIb ^cr , rs4 ssIIc ^cr , rs4 ssIVa ^cr , rs4 ssIVb ^cr , rs4 ssV ^cr , rs4 sbeI ^cr , rs4 sbeIIa ^cr , rs4 sbeIIb ^cr , rs4 isa1 ^cr , rs4 isa2 ^cr , rs4 isa3 ^cr , rs4 pul ^cr , rs4 ssIVb sbeIIb ^cr , rs4 sbeI sbeIIb ^cr , and rs4 sbeI ssIVb ^cr plants were grown under field conditions at the experimental field of the Institute of Genetics and Developmental Biology, Chinese Academy of Sciences in Beijing and Hainan, China, from 2021 to 2024.

Genotyping of edited plants

For genotyping of gene‐edited plants, genomic DNA was isolated by the CTAB (cetyltrimethylammonium bromide) method (Murray and Thompson, 1980). PCR was performed to amplify the genomic region surrounding the CRISPR target sites using the corresponding primers (Table S2). The PCR product was sequenced by the Sanger method, and the sequencing chromatograms were decoded by the Degenerate Sequence Decoding method (Liu et al., 2015).

Preparation of rice flour

Rice seed sample preparation was conducted as previously described with minor modifications (Zhou et al., 2016). Harvested rice grains were air‐dried to achieve the moisture content about 12% for measurements of RS contents and grain quality traits. Rice samples were dehulled by a small rice hulling machine (TOP Cloud‐agri, TP‐JLG‐2018), milled to white rice using a grain polisher (Kett, TP‐3000) and subsequently grounded to flour in the ultra‐centrifugal mill (Retsch, ZM200) with a 1.0 mm sieve for measuring RS contents and pasting properties. The rice flour was passed through a 0.5 mm and 0.2 mm sieve for total starch analyses and apparent amylose content (AAC) analyses, respectively.

RS, AAC and total starch analysis

RS contents were analysed as previously described using a resistant starch assay kit (Megazyme, K‐RSTAR) according to the manufacturer's assay procedure (Wang et al., 2023). The total starch content was analysed by using a total starch measurement kit (Megazyme, K‐TSTA) through RTS‐NaOH procedure. AAC was measured by using the method of ISO 6647 (International Organization for Standardization) with standard samples of four levels (1.5%, 10.4%, 16.2% and 26.5%) provided by China National Rice Research Institute.

Starch pasting properties

A total of 3 g rice flour was dispersed in 25 mL distilled water and characterized using the rapid visco analyser (Perten, RVA4500) according to the methods described previously (Wang et al., 2023). Pasting viscosity, pasting temperature, trough viscosity, breakdown viscosity and final viscosity were recorded.

Seed morphology and SEM

Dehulled whole mature seeds and cross‐sections were taken by stereomicroscope (OLYMPUS, SZX16). Scanning electron micrographs of starch granules in endosperm cross‐section was performed as previously described (Fujita et al., 2007). Rice seeds were fractured across the short axis and adhered to double‐sided adhesive tape mounted on an aluminium stub. The surface was sputter‐coated with gold of ion sputtering equipment under vacuum (HITACHI, E‐1010) and observed by SEM (ZEISS, Crossbeam 340).

Phenotypic evaluation

Twenty‐four plants of each edited mutants and the wild type were grown in three rows with eight plants per row in the experimental field in the summer in Beijing (39°54′ N 116°23′ E), and 10 plants were randomly selected for the phenotypic analysis of the plant heights, tiller numbers, grain yield per plant, grain number per main panicle and seed setting rates. Grain length,  grain width and 1000‐seed weight were measured using automatic seed counting and analysing instrument (Model SC‐G, Wanshen). Mutants and the wild type were grown in the experimental field in the winter in Hainan (19°12′ N 109°42′ E), and three plants were randomly selected for the phenotypic analysis of the grain yield per plant and 1000‐seed weight.

Statistics

All numerical values were presented as means ± s.d. Different letters at top of each column indicate a significant difference at P < 0.05 determined by Tukey's HSD test. The two‐tailed Student's t‐test was performed by GraphPad Prism version 9.0. All information on replication, statistical test and presentation is given in the respective figure legends.

Conflict of interest

The authors have declared no conflict of interest.

Author contributions

A.W., H.Y. and J.L. designed the experiments. A.W., Q.C., W.L., M.K., Y.Z., X.M., H.Y., Y.J., M.C. and G.L. performed the experiments. A.W. and H.Y. analysed the results. H.Y., G.L., D.W., and J.L. supervised the project. A.W., H.Y. and J.L. wrote the paper. All authors approved the final manuscript.

Supporting information

Figure S1. CRISPR/Cas9‐induced mutations in SS‐related mutants.

Figure S2. CRISPR/Cas9‐induced mutations in SBE‐related mutants.

Figure S3. CRISPR/Cas9‐induced mutations in DBE‐related mutants.

Figure S4. RS contents in cooked rice of rs4 ssI ^−6bp , ssIIIa and beIIb single mutants.

Figure S5. CRISPR/Cas9‐induced mutations in double mutants.

Figure S6. Morphology of seeds of independent RS mutants with same mutation combinations.

Figure S7. Physical–chemical properties of independent RS mutants with same mutation combinations.

Figure S8. Rapid viscosity analysis (RVA) profiles and pasting properties of independent RS mutants with same mutation combinations.

Figure S9. Other yield‐related traits were significantly affected in eight selected RS lines.

Figure S10. Grain yield per plant and 1000‐seed weight of independent RS mutants with same mutation combinations.

Table S1. Summary table of key data of RS, AAC, PT and grain yield in the wild‐type (WT, R7954), ssIIIa, ssIIIa ssIIIb (rs4), rs4 ssIIa and rs4 sbeI ssIVb mutants.

Table S2. Primers used in this study.

Data availability statement

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