Evolution of the Wx genotype combination in hybrid rice

Background: With increased interest in the quality characteristics of hybrid rice, its quality traits have received a certain extent of improvement. However, comparisons to high-quality conventional rice, especially its eating quality, have revealed that more work is still required. Amylose content is a key determinant of the eating quality of rice, and Wx , the major gene controlling amylose content, has many allelic variants. A complete understanding of the Wx allelic variations in cultivated rice and the evolution process of the Wx genotype in hybrid rice are the premises for the rational utilization of rich genetic resources in the quality breeding of hybrid rice. Based on the 3K RGP re-sequencing data, we sought to analyse the allelic variation at the Wx locus and develop a set of Kompetitive Allele-Specific PCR markers. Furthermore, we used the markers to analyse the evolution of a combination of the Wx genotype in hybrid rice. Results: Eight known alleles existing globally were identified, and their evident regional preferences and Indica - japonica background differences were revealed. An additional five non-synonymous mutations were identified in the coding region of the Wx gene for the first time, including a new functional site located in the active centre of OsGBSS1. By genotyping the basic hybrid parents obtained from 1976 to 2018, we found that only three Wx allelic variations existed. Wx lv was widely used in the female parents of the three-line hybrid rice, and Wx a was used in the female parents of early two-line hybrid rice; however, they were gradually replaced by Wx b . Conclusions: In this study, the Wx lv allele was found to be the cause of the poor eating quality of early hybrid rice. With the elimination of the Wx lv allele and

introduction of Wx b in both parents, the eating quality of hybrid rice was generally improved. Only three allelic variations were present in the previous hybrid rice. For further eating quality improvement of hybrid rice, more Wx alleles should be introduced.

Background
Rice is one of the most important crops. In fact, its greatest success has been achieved in the application of heterosis. For many years, researchers have devoted much attention to high-yield hybrid rice. However, in China, research on the quality of rice began fairly late, especially in the southern areas. With continued improvement in the standard of living, the quality of rice is now of greater concern than ever before. In the last 15 years, the quality traits of indica hybrid rice was improved to a certain extent; however, further improvements are still critical compared to the high-quality conventional rice, especially its eating quality (Chen et al. 2015; Tang et al. 2016). As many hybrid parents widely used from the same basic materials, the genetic polymorphism of hybrid rice may be relatively low (Li et al. 2000). Understanding the diversity and evolution of genes associated with eating quality is thus the foundation needed to fully utilize the rich genetic resources to achieve improvements in the quality of hybrid rice.
The quality of rice is a complex characteristic, with amylose content (AC) serving as a key determinant of its eating and cooking qualities (Tian et al. 2009;Qian et al. 2016). The Wx gene, encoding a granule bound starch synthase (GBSS), is a major gene in the control of AC. Its RNA expression and the activity of the OsGBSS1 protein positively correlate with AC (Liu et al. 2014). To date, at least eight Wx alleles have been identified in rice-Wx a, Wx b, Wx in, Wx op, Wx m p, Wx m w, Wx lv , and wx. The AC of the eight alleles decreases in sequence from Wx a (>25%), Wx lv (20-25%), Wx in (18-22%), Wx b (15-18%), Wx m w (10-14%), Wx m p (8-12%), and Wx op (5-10%) to wx (AC<2%) (Zhang et al. 2019;Yang et al. 2013;Liu et al. 2009; Sato et al. 1996;Mikami et al. 1999Mikami et al. , 2000Mikami et al. , 2008. Presently, eight Wx alleles have been identified. However, whether there are new functional alleles in the cultivated rice resources worldwide are yet to be elucidated. Nonetheless, we hope to develop a set of molecular markers for high-throughput genotyping in the Wx loci but whether the known alleles can cover most of the Wx allelic variations and explain the AC variances remain unknown. A complete understanding of the Wx allelic variations in cultivated rice worldwide is thus the premise for Wx genotyping and the efficiency of utilizing hybrid rice. In 2015, a total of 3,000 germplasm accessions from 89 different countries/regions were selected for genome-wide resequencing, and a comprehensive SNP and InDel sub-database was established for the Rice Functional Genomics-based Breeding (RFGB) Database (Li et al. 2014). From the samples, 2,466 accessions were retrieved from a core collection established from more than 101,000 rice accessions in the IRGC while 534 accessions were selected from a core collection established from 61,470 rice accessions preserved in the CNCGB (Zhang et al., 2011). Thus, the 3000 germplasm accessions represent a panel with abundant genetic diversity in the global cultivation of rice. By utilizing the currently available genome information from the 3 K RGP sequencing data, different types of allelic variations in the Wx loci can be comprehensively analysed, and new allelic variants that are responsible for different AC classes can be discovered to improve future quality of hybrid rice.
A prior research revealed that a significant correlation exists between the number of (CT)n and AC (Larkin et al. 2003) and related microsatellite markers have also been developed. However, some of the AC variations cannot be accurately explained by these markers (Dobo et al. 2010). A series of Wx alleles has been recently identified in rice (Zhu et al. 2015), and numerous SNP genotyping methods have been used for polymorphism analysis of the Wx loci. Kompetitive Allele-Specific PCR (KASP) is a high throughput method that can type SNPs and InDels at specific sites.
Based on terminal fluorescence reading, different genotypes at a single site can be detected by two-colour fluorescence (He et al. 2014) Therefore, owing to advantages such as efficiency, accuracy, and high throughput, establishing KASP to distinguish functional SNP would greatly accelerate the molecular design for breeding good quality rice.
Herein, to make better use of Wx allele resources for improving the eating-quality breeding of hybrid rice, we sought to understand the diversity of Wx alleles in cultivated rice and the evolution of the Wx genotype combination in hybrid rice. By using the 3K RGP re-sequencing data, we could identify not only the eight known alleles that widely existed in the germplasm resources, but also a potential novel

AC measurement
The AC of rice seeds was measured using the iodine colorimetry assay described previously (Williams et al. 1970).

Results
Analysis of the Wx allelic variation among 3 K RGP By using the CDS of Nipponbare as a reference sequence in this study, allelic variations in the coding region of the Wx gene were analysed according to the 3 K RGP sequencing data. After eliminating the alleles with heterozygous sequences in the coding region and the possible deletion caused by insufficient sequencing coverage, a total of 2,752 Wx alleles were obtained for variation analysis. Based on the G/T SNP at the splice site of the first intron (In1G/T SNP), two important alleles, Wx a and Wx b , were originally identified. Although this functional SNP does not belong to the coding region, it was considered for the variation analysis.  Fig. 2-F). By analysing the crystal structure of OsGBSS1, we found that Glu410Gly (Ex10+101) is located in the active centre of the enzyme, beside the ligand ADP of OsGBSS1, while the remaining 4 mutation sites were located at a distant from the active centres (Fig. 3-B). In addition, Ala54Thr did not exist in the truncated body of OsGBSS1.

Construction of the Wx haplotype flowchart
To comprehensively understand the genetic background of each haplotype and the relationships among them, a haplotype flowchart was constructed to describe the mutational steps of the 30 haplotypes. In Fig. 4, Hap1, Hap3 Altogether, these findings indicate that each subpopulation contained significantly dominant Wx alleles, which suggest that Wx alleles might have regional preference and a corresponding environmental adaptability.

Worldwide distribution of different Wx alleles
Based on the geographic distribution of the eight alleles (Fig. 5), Wx a, Wx b, Wx in, and Wx lv are the four predominant alleles that display a wide distribution in most rice producing regions of the world. However, evident regional differences exist among them; Wx lv was among the most widely distributed in the world, Wx b was widely distributed in higher latitudes, Wx in was widely distributed in the tropical areas of low latitudes, wx was commonly distributed in the Southeast and East Asia,

Correlation analysis between the Wx alleles and CT repeats
In most prior studies, GBSS alleles were primarily defined according to the CT repeat in exon 1. Generally, varieties with 17 or 18 CT repeats are considered to be "low amylose" types, while those with 10 or 11 CT repeats are generally classified as "high amylose" types (Dobo et al. 2010). Although previous studies found a significant correlation between the number of CT repeats and AC, some related molecular markers have been developed. Nonetheless, a considerable part of AC variation could not be explained by the CT repeats. By analysing the number of CT repeats that corresponds to different Wx alleles in the 3K data, we found that most low amylose varieties had long CT repeats while high amylose varieties had short CT repeats, a result that aligns with that of a previous study. However, some high amylose alleles, such as Wx a and Wx lv, also contain long CT repeats, and some low amylose alleles, such as Wx op, only have short CT repeats (Fig. 6). These exceptions help to explain the AC variation, which cannot be explained by microsatellite markers.

KASP genotyping of the Wx alleles in hybrid parents
Owing to the importance of the eight Wx alleles that account for 99.9% of the Wx allele resources, a set of KASP markers was developed based on these Wx alleles (Fig. 7). Thereafter, 36 basic parent lines, including 18 female parents and 18 male parents of hybrid rice from 1976 to 2018 (Hu et al. 2016), were selected for genotyping (Fig. 8). Fig. 9 indicates that Wx lv is the major allele in the male sterile line of the three-line hybrid rice, such as zhenshan97A, which is the female parent of many hybrids that are widely used in the early stages of developing three-line hybrid rice. Conversely, for the two-line hybrid rice, Wx a is the major allele of the early sterile line, such as Annong S-1. Additionally, its appearance indicates the beginning of two-line hybrid rice, and Peiai 64S, which is the most influential sterile line, has the largest two-line hybrid combination planting area. Owing to the development of the two-line hybrid rice, Wx a was replaced by Wx b. Wx b is a major allele for nearly all restorer lines, except Ce64-7 and R402. The genotyping results revealed that the major allelic combinations changed from Wx lv/ Wx b to Wx a/ Wx b and then to Wx b/ Wx b, ultimately indicating the evolution process, at the molecular level, for the quality improvement of hybrid rice in China.

Discussion
Rice quality, especially its eating quality, is a complex characteristic, with Wx serving as one of the major genes affecting its eating quality and consisting of many allelic variations. In the present study, resequencing data from 3,000 germplasm resources from 89 countries worldwide were used to comprehensively analyse the Wx alleles. Through further analyses, 99.9% of the alleles were recognized to be the eight known functional alleles. Thus, a set of KASP markers was developed, and the basic hybrid parent varieties from 1976 to 2018 were selected for Wx genotyping. Based on the typing results, the allelic combinations changed from Wx lv/ Wx b to Wx a/ Wx b then to Wx b/ Wx b as quality improved.
Previously, the Wx a allele was widely assumed to be the main reason for the poor taste of the early three-line hybrid rice. However, the findings of this study demonstrate that Wx lv, an ancestral allele of Wx (Zhang et al. 2019), is responsible for the considerable effects of the mouthfeel of rice grains and was thus gradually selected in the subsequent breeding process. Wx a was used in male parents of the early two-line hybrid rice and could gradually be replaced with Wx b. As Wx b was introduced in both parents, the quality of the two-line hybrid rice was generally improved. Quality improvement of hybrid rice might thus be caused by Wx lv elimination. In addition, the homozygosis of the Wx b allele in hybrids may serve as another important factor. Genotype segregation is known to exist in the endosperm characters of hybrid rice. With homozygosis of the Wx allele in hybrids, the AC of each grain in the hybrid combination tend to be more homogeneous, and the eating quality was identified to be better. For the future breeding of hybrid rice with highquality, more favourable Wx allelic variations should be identified, introduced, and effectively used. By evaluating the genetic effects of different allelic combinations, we can derive the most appropriate genotype combination to achieve high-quality.
By analysing the molecular evolution of the Wx allele in the indica hybrid rice, we found that only three Wx allelic variations existed in the main parents of hybrids.
Thus, expanding the genetic background of hybrid parents might be key for improving the quality character. Our analysis showed that rich allelic variations of Wx exist in cultivated rice. However, the mechanism whereby these alleles affect the eating quality of hybrids has not been comprehensively evaluated, and most are not well utilized in quality breeding. Nonetheless, we hope to introduce the remaining four non-glutinous alleles to accommodate different consumption demands in the quality breeding of indica hybrid rice. With the introduction of Wx b in both parents in recent years, the AC of indica hybrid rice was found to generally reduce to 15% or lower, with potential improvements in its quality traits. Some studies have confirmed that a high temperature at the milky stage of grain filling has the greatest influence on rice quality, with the panicle identified as the organ most sensitive to high temperature (Yamakawa et al. 2007). As high temperatures are frequent in the southern regions of China, the AC of hybrids in these regions was significantly reduced and their appearance and eating qualities might be greatly affected. Wx in is a major allele with intermediate AC (18-20%). Thus, when it is compared to Wx b, the effect of its response to abnormally high temperature on the quality of AC may not be evident. Based on our analysis, Wx in is the predominant allele in the circum-Basmati group (cB) that many high-quality rice varieties, such as Basmati and Sadri aromatic varieties. Most of the Wx in alleles were distributed in tropical areas with low latitudes compared to the Wx b alleles distributed in high latitudes. Thus, we concluded that Wx in may be the preferred choice for improving indica-type rice quality in southern China.
In the present study, five new alleles were identified for the first time. By determining the AC in four resources, including the allelic variations, we found that with the exception of the significant decline in the AC of IRIS_313-10892 compared to that of the Wx a genetic background, the AC of the other three resources was generally unchanged. By analysing the crystal structure of OsGBSS1, we found that Lys413 and Arg408 are two important amino acids that directly interact with ADP. In addition, Glu410 was found to be linked to both amino acids via simultaneous hydrogen bonds. Previously, the effects of amino acid substitutions on OsGBSS1 activity were revealed (Liu et al. 2014). Glu410Gly is a natural allelic variation where the acidic amino acid is replaced by a non-polar amino acid without any R group; this type of change must have a greater effect on the conformation of Lys413 and Arg408. With the exception of wx in glutinous rice, Wx op is presently the allelic variation with the lowest AC in non-glutinous rice. In this study, we found a novel allele that has an AC between wx and Wx op, which ultimately cause more diversity in the variation of AC in cultivated rice. Thus, in future rice breeding, varieties with different ACs could be employed to meet different consumer demands and industrial applications.

Conclusions
Previously, replacing Wx b with Wx a in hybrid rice was widely believed to improve its eating quality. However, in the present study, the Wx a allele in early hybrid rice was in fact Wx lv, which may have been responsible for the poor taste of early hybrid rice. By eliminating the Wx lv allele and introducing Wx b in both parents, the eatingquality of hybrid rice can be generally improved. As only three allelic variations were present in the previous hybrid rice, for further eating quality improvement of hybrid rice, more Wx alleles should be introduced. Availability of data and material All data generated or analysed during this study are included in this published article and its supplementary information files Competing interests responsible for material selection and planting. All authors read and approved the final manuscript.   Worldwide distribution of different Wx alleles. Each colour represents a unique Wx allele. Red Primer sequences for KASP genotyping of different Wx alleles. Primer X and Primer Y are two