Substitution Mapping of Two Closely Linked QTLs on Chromosome 8 Controlling Grain Chalkiness in Rice

Rice varieties are required to have high yield and good grain quality. Grain chalkiness and grain shape are two important traits of rice grain quality. Low chalkiness slender grains are preferred by most rice consumers. Here, we dissected two closely linked quantitative trait loci (QTLs) controlling grain chalkiness and grain shape on rice chromosome 8 by substitution mapping. Two closely linked QTLs controlling grain chalkiness and grain shape were identified using single-segment substitution lines (SSSLs). The two QTLs were then dissected on rice chromosome 8 by secondary substitution mapping. qPGC8.1 was located in an interval of 1382.6 kb and qPGC8.2 was mapped in a 2057.1 kb region. The maximum distance of the two QTLs was 4.37 Mb and the space distance of two QTL intervals was 0.72 Mb. qPGC8.1 controlled grain chalkiness and grain width. qPGC8.2 was responsible for grain chalkiness, grain length and width. The additive effects of qPGC8.1 and qPGC8.2 on grain chalkiness were not affected by higher temperature. Two closely linked QTLs qPGC8.1 and qPGC8.2 were dissected on rice chromosome 8. They controlled the phenotypes of grain chalkiness and grain shape. The two QTLs were insensitive to higher temperature. Supplementary Information The online version contains supplementary material available at 10.1186/s12284-021-00526-4.


Background
Rice is an important food crop in the world. Rice varieties with higher head rice yield, higher transparency and less chalkiness are more popular in the market (Sreenivasulu et al. 2015;Misra et al. 2019). With the increase of living standard, rice varieties are required to have both a higher grain yield and a better grain quality. Grain chalkiness not only affects grain appearance, but also has adverse effects on milling and cooking performance. Chalkiness is a complex quantitative trait, which is easily affected by environments. In the early and middle stages of seed development, the occurrence of temperature stress will cause uneven seed filling and storage biosynthesis obstacles, leading to the formation of chalkiness (Masutomi et al. 2015;Sreenivasulu et al. 2015;Morita et al. 2016). The chalkiness of rice varieties varied greatly. It was found that the grain chalkiness of newly developed varieties was higher than that of old modern varieties, and that of the hybrid varieties was higher than that of other modern varieties (Laborte et al. 2015). Therefore, highyielding varieties usually have higher chalkiness levels (Misra et al. 2019).
Percentage of grain chalkiness (PGC) is a quantitative index of grain chalkiness controlled by quantitative trait loci (QTLs) (Sreenivasulu et al. 2015;Misra et al. 2019). More than one hundred of QTLs controlling chalkiness have been reported in rice genome (Sreenivasulu et al. 2015;Yang et al. 2021). Many QTLs for chalkiness were detected on chromosomes 5, 6 and 8 in different populations and environments (He et al. 1999;Tan et al. 2000;Wan et al. 2005;Hao et al. 2009;Liu et al. 2011Liu et al. , 2012Chen et al. 2011Chen et al. , 2016Guo et al. 2011;Li et al. 2014;Peng et al. 2014;Zhao et al. 2015Zhao et al. , 2016Gao et al. 2016;Yun et al. 2016;Wang et al. 2017;Zhu et al. 2018a;Misra et al. 2019Misra et al. , 2020. On chromosome 8, the hot-spot region of the QTLs for chalkiness was located on the long arm (Li et al. 2003;Wan et al. 2005;Hao et al. 2009;Guo et al. 2011;Liu et al. 2012;Gao et al. 2016;Zhao et al. 2016;Wang et al. 2017). Although many chalk QTLs were reported in the hot-spot chromosome region, only one of the QTLs was usually detected in a population. Therefore, it is not clear how many chalk QTLs there are in the hot-spot chromosome region. It was found that the effect of QTLs for chalkiness is easily affected by high temperature. A set of QTLs for chalkiness were detected under heat stress condition (Kobayashi et al. 2007;Tabata et al. 2007;Wada et al. 2015;Miyahara et al. 2017;Nevame et al. 2018). Recently, we mapped two QTLs for grain chalkiness, qPGC9 and qPGC11, and found that their additive effects on chalkiness significantly decreased under higher temperature (Yang et al. 2021).
The QTL effect of grain chalkiness was usually low, which was easily affected by other genetic factors. Several fine-mapped target genes and GWAS loci have been found to influence chalkiness, but many have either low or moderate effect (Gong et al. 2017;Wang et al. 2017;Quero et al. 2018;Misra et al. 2019). Although 11 GWAS loci for chalky grain rate were identified, the GWAS loci could only explain a small part of the phenotypic variation (Gong et al. 2017). Grain chalkiness and grain shape are two important traits for rice grain quality. It was found that grain width was positively correlated with chalkiness (Zhao et al. 2015). Some QTLs controlling grain width were overlapped with chalk QTLs ). However, Misra et al. (2020) reported that only 3 of 78 QTL regions overlapped with known grain width genes, and the phenotypic variation for chalkiness showed a weak positive correlation with grain width. They believed that the key genes of grain width are not causal factors of chalkiness. Therefore, the genetic architecture of grain chalkiness and grain shape of rice remains unclear.
Substitution mapping is a powerful tool to detect QTLs for complex traits (Tan et al. 2021a, b;Yang et al. 2021). Like near-isogenic lines (NILs), single-segment substitution lines (SSSLs) carry only one substitution segment from donors in the recipient genetic background (Zhang et al. 2004;Keurentjes et al. 2007). We have developed a library of 2360 SSSLs, which were derived from 43 donors of 7 species of rice AA genome in the genetic background of Huajingxian 74 (HJX74), an indica elite variety in southern China (Zhang et al. 2004;Xi et al. 2006;Zhang 2019Zhang , 2021. These SSSLs were widely used to detect QTLs for complex traits, to clone QTLs of agronomic importance and to mine alleles of different functions Teng et al. 2012;Wang et al. 2012;Zhang et al. 2012;Zhu et al. 2014Zhu et al. , 2018bYang et al. 2016Yang et al. , 2021Zhou et al. 2017;Fang et al. 2019;Sui et al. 2019;Tan et al. 2020Tan et al. , 2021a. Recently, the SSSLs were used to detect QTLs controlling stigma exsertion rate (SER) of rice. Two pairs of tightly linked QTLs for SER, qSER-2a and qSER-2b on chromosome 2 and qSER-3a and qSER-3b on chromosome 3, were dissected by substitution mapping (Tan et al. 2021a). In previous study, we finemapped two QTLs qPGC9 and qPGC11 for grain chalkiness on rice chromosomes 9 and 11, which were sensitive to high temperature (Yang et al. 2021). In the present study, two closely linked QTLs for grain chalkiness on chromosome 8, qPGC8.1 and qPGC8.2, were detected. The two QTLs affected on grain chalkiness and grain shape, and were insensitive to high temperature. Dissection of the two closely linked QTLs laid a foundation for revealing the genetic architecture of grain chalkiness in rice.

Grain Chalkiness in SSSLs
Two SSSLs 15-08 and 03-08 with lower grain chalkiness were selected from the HJX74-SSSL library (Fig. 1a). The SSSLs were used to investigate grain chalkiness in consecutive 6 cropping seasons from the first cropping season (FCS) of 2017 to the second cropping season (SCS) of 2019. On average, the PGC of 15-08 and 03-08 was 11.9% and 10.3% respectively, which were significantly lower than 21.2% of recipient HJX74 ( Fig. 1b and Additional file 1: Table S1). It meant that the two SSSLs each carried a QTL for PGC on their substitution segments.
Scanning electron microscopy (SEM) images showed contrasting differences in the microstructure of milled grains between HJX74 and SSSLs. The typical characteristics of chalkiness with the starch granules being small spherical and loosely packed in the opaque endosperm cells were observed in the grains with chalkiness in HJX74. By contrast, in the grains without chalkiness in SSSLs 15-08 and 03-08, compound starch granules were polyhedral and tightly packed, with no air spaces within or between them (Additional file 2: Fig. S1).
The substitution segments of 15-08 and 03-08 were surveyed by densifying molecular markers (Additional file 1: Table S2). The estimated length of substitution segments was 2726.7 kb in 15-08 and 4974.3 kb in 03-08. The two substitution segments overlapped in a 553.7 kb interval ( Fig. 1c and Additional file 1: Table S3).
Eight agronomic traits of 15-08 and 03-08 were investigated. Most traits of the SSSLs were not significantly different from those of HJX74 ( Fig. 1a and Additional file 1: Table S4). However, the grain shapes of 15-08 and 03-08 were significantly different from those of HJX74. For 15-08, the grain width was significantly narrower than that of HJX74, the former was 2.62 mm and 2.45 mm and the later was 2.69 mm and 2.64 mm in SCS of 2018 and in FCS of 2019, respectively. Compared with HJX74, 03-08 showed significantly longer in grain length and significantly narrower in grain width. In SCS of 2018 and in FCS of 2019, 03-08 had 9.14 mm and 8.88 mm in grain length and 2.55 mm and 2.40 mm in grain width, respectively (Additional file 1: Table S4). It is noted that GW8 is outside the substitution segment of 03-08 (Fig. 1c). Therefore, the difference of grain shape between 03-08 and HJX74 was not controlled by GW8 gene.

Substitution Mapping of qPGC8.1
To map the QTL for grain chalkiness on the substitution segment of 15-08, the SSSL was used to develop secondary SSSLs or NILs. Four NILs were developed from an F 2:3 population derived from the cross of HJX74/15-08. The four NILs were then investigated for grain chalkiness. PGC levels of two NILs, NIL15-08-26 and NIL15-08-43, were as low as 15-08, while those of other two NILs, NIL15-08-4 and NIL15-08-14, were as high as HJX74. Substitution segments of the two NILs with low PGC overlapped in the region between markers RM4815 and RM23137, while substitution segments of other two NILs with high PGC located outside the region. These results indicated that the QTL for grain chalkiness, qPGC8.1, was located in the region between markers RM4815 and RM23137 with the estimated interval length of 1382.6 kb ( Fig. 2a, b).

Substitution Mapping of qPGC8.2
To map the QTL for grain chalkiness on the substitution segment of 03-08, the SSSL was used to develop secondary SSSLs or NILs. Five NILs were developed from an F 2:3 population derived from the cross of HJX74/03-08. The five NILs were then investigated for grain chalkiness. PGC levels of three NILs, NIL03-08-9, NIL03-08-15 and NIL03-08-19, were as low as 03-08, while those of other two NILs, NIL03-08-14 and NIL03-08-55, were as high as HJX74. Substitution segments of the three NILs with low PGC overlapped in the region between markers SNP8M54 and SNP8M28, while substitution segments of other two NILs with high PGC located outside the region. These results indicated that the QTL for grain chalkiness, qPGC8.2, was located in the region between markers SNP8M54 and SNP8M28 with the estimated interval length of 2057.1 kb (Fig. 3a, b).
In substitution mapping, NILs segregated in grain length and grain width. Three NILs, NIL03-08-9, NIL03-08-15 and NIL03-08-19, with low chalkiness showed longer and narrower grains as 03-08, while other two NILs, NIL03-08-14 and NIL03-08-55, with high chalkiness had shorter and wider grains as HJX74. The three NILs with low chalkiness were 8.75 mm, 8.91 mm and 9.02 mm in grain length and 2.58 mm, 2.54 mm and 2.62 mm in grain width, while the two NILs with high chalkiness were 8.46 mm and 8.48 mm in grain length and 2.66 mm and 2.65 mm in grain width, respectively (Fig. 3d, e). The results indicated that qPGC8.2 controlled grain shape besides grain chalkiness.
It was noted that qPGC8.2 was closely linked with qPGC8.1 on chromosome 8. The maximum distance of the two QTL location was 4.37 Mb, from markers RM4815 to SNP8M28. The space distance between the two QTL regions was 0.72 Mb, from markers RM23137 to SNP8M54 (Figs. 1c, 2b, 3b).

Influence of Different Cropping Seasons in Grain Chalkiness
The grain chalkiness was tested in two cropping seasons per year. During flowering to harvest of rice, the day and night temperatures of FCS and SCS were very different. In 2017-2019, the average values of maximum, minimum and mean temperatures were 32.1 °C, 25.9 °C and 29.0 °C in FCS, and 28.6 °C, 21.0 °C and 24.9 °C in SCS, respectively. The mean temperature in SCS was 4.1 °C lower than that in FCS (Additional file 1: Table S5).
The grain chalkiness was lower in SCS than that in FCS in all lines. Percentage of chalky grains (PCG) and PGC were significantly different in all lines. Percentage of chalk area (PCA) was significantly different in 15-08, and no significantly different in 03-08 and HJX74. The PGC of HJX74, 15-08 and 03-08 was 25.8%, 15.9% and 13.6% in FCS, and 16.6%, 7.8% and 7.1% in SCS, respectively (Fig. 4). It is obvious that high temperature had great influence in PGC of HJX74 and SSSLs during seed filling period, mainly through the influence in PCG.

Dissection of Two Closely Linkage QTLs qPGC8.1 and qPGC8.2 on Chromosome 8
Grain chalkiness is a complex polygenic quantitative trait (Sreenivasulu et al. 2015). More than one hundred of QTLs for the chalkiness traits have been reported across all 12 chromosomes of rice genome (Sreenivasulu et al. 2015). Many QTLs for chalkiness in different populations were located on the long arm of chromosome 8. In the hot-spot region, however, only one chalk QTL was detected in each population (Li et al. 2003  and qSER-3b on chromosome 3, were dissected by substitution mapping. On chromosome 2, two linkage QTLs, qSER-2a and qSER-2b, were located in the region of 1288.0 kb, and were respectively delimited to the intervals of 234.9 kb and 214.3 kb. On chromosome 3, two QTLs, qSER-3a and qSER-3b, were detected in the region of 3575.5 kb and were narrowed down to 319.1 kb and 637.3 kb, respectively (Tan et al. 2021a). Together, those results indicated that substitution mapping using SSSLs is a powerful tool for dissection of closely linked QTLs for complex traits.

Relationship Between Grain Chalkiness and Grain Shape During Grain Development
Grain chalkiness and grain shape are two important traits for rice grain quality. It was found that grain width had a negative pleiotropic effect on grain chalkiness (Zhao et al. 2015;Gao et al. 2016). GW2 allele significantly increased grain weight and grain width, but also led to increase chalkiness in rice (Song et al. 2007). The gw8, GW7 and gs9 alleles exhibited more slender grain phenotype and reduced chalkiness, which significantly improved the appearance quality of rice (Wang et al. , 2015Zhao et al. 2018). Zhao et al. (2015) identified three QTLs for chalkiness tightly linked to GS3, gw5, and qGL7-2 controlling grain size. However, grain chalkiness was weakly positively correlated with phenotypic variation of grain width, and the key genes affecting grain width were not the cause of chalkiness (Misra et al. 2020). Recently, we detected two QTLs for grain chalkiness, qPGC9 and qPGC11, by substitution mapping, and found that they had no effect on grain shape (Yang et al. 2021). Therefore, the relationship between grain chalkiness and grain shape in rice remains unclear. In the present study, two closely linked QTLs, qPGC8.1 and qPGC8.2, on chromosome 8 showed effects on grain chalkiness and grain shape (Figs. 2, 3). In addition, it is noted that the fgr gene for fragrance (Bradbury et al. 2005) is located not only on the substitution segment of 15-08 but also in the qPGC8.1 mapped region, and the GW8 gene controlling grain width  is located near the qPGC8.2 mapped region although it is outside the substitution segment of 03-08 (Fig. 1c). These results indicated that the long arm of chromosome 8 may be a gene cluster area for grain development. It is unclear that each region of qPGC8.1 and qPGC8.2 carries two different genes, which control grain chalkiness and grain shape respectively, or one gene that controls grain chalkiness and has a pleiotropic effect on grain shape. Therefore, the mapping of qPGC8.1 and qPGC8.2 laid a foundation for revealing the relationship between chalkiness and grain shape during grain development.

Influence of High Temperature in the Effect of Chalk QTLs
It was found that the high temperature during the seed filling stage caused uneven grain filling and resulted in chalk formation (Sreenivasulu et al. 2015;Masutomi et al. 2015;Morita et al. 2016;Ishimaru et al. 2019). In Guandong province of China, rice is planted in two cropping seasons per year. During the seed filling period, the air temperature of FCS is usually higher than that of SCS. In 2017-2019, the mean temperature of FCS was 4.1 °C higher than that of SCS (Additional file 1: Table S5). Due to higher temperature, grain chalkiness of all lines in FCS was significantly higher than that in SCS (Fig. 4). Obviously, the grain chalkiness of HJX74 and SSSLs was greatly affected by higher temperature during the seed filling period.
Some QTLs for chalkiness were detected under high temperature stress (Nevame et al. 2018). Kobayashi et al. (2007) detected three QTLs for chalkiness in japonica varieties under heat stress. Tabata et al. (2007) identified four QTLs for chalkiness in a RIL population derived from a cross between a heat stress-tolerant variety and a heat stress-sensitive variety. Wada et al. (2015) and Miyahara et al. (2017) identified a set of chalk QTLs under heat stress condition using a RIL population derived from a cross between a heat-tolerant variety and a heat stress-sensitive variety. Recently, we mapped two QTLs for grain chalkiness, qPGC9 and qPGC11, and found that the additive effects of qPGC9 and qPGC11 on chalkiness in SCS were almost twice of those in FCS. The additive effects of qPGC9 and qPGC11 on chalkiness greatly decreased by high temperature of FCS. Therefore, the qPGC9 and qPGC11 were sensitive to high temperature (Yang et al. 2021). These results showed that many QTLs could increase chalkiness under high temperature. This may be the reason why most varieties show high chalkiness at high temperature. In the present study, the grain chalkiness of HJX74 and SSSLs was greatly affected by higher temperature during the seed filling period. It is because they have the same genetic background carrying some QTLs increasing chalkiness under high temperature. Interestingly, the additive effects of qPGC8.1 and qPGC8.2 on PCG, PCA and PGC were no significant difference in different cropping seasons (Fig. 5), showing that the additive effects of qPGC8.1 and qPGC8.2 on grain chalkiness were not affected by the high temperature in FCS. It indicates that qPGC8.1 and qPGC8.2 are insensitive to high temperature, which are different from qPGC9 and qPGC11. Therefore, qPGC8.1 and qPGC8.2 can be used for the breeding of rice varieties with lower grain chalkiness under high temperature.

Conclusion
Two closely linked QTLs qPGC8.1 and qPGC8.2 controlling grain chalkiness and grain shape were located on chromosome 8. The effect of qPGC8.1 and qPGC8.2 was incomplete dominance. The additive effects of the two QTLs on grain chalkiness had no significant difference between FCS and SCS. The qPGC8.1 and qPGC8.2 were insensitive to the high temperature in FCS. The mapping of qPGC8.1 and qPGC8.2 laid a foundation for revealing the relationship between grain chalkiness and grain shape during grain development. The two QTLs are favorable for the breeding of rice varieties with low chalkiness under high temperature condition.

Rice Materials and Cropping Seasons
Two SSSLs 03-08 and 15-08 with lower grain chalkiness were selected from the HJX74-SSSL library. The substitution segment of 03-08 was from the donor Zhong4188 and that of 15-08 was from the donor American Jasmine. The donors of both SSSLs are indica varieties. All rice materials were planted at the farm of South China Agricultural University, Guangzhou (23°07′N, 113°15′E) from 2017 to 2019. The materials were planted in two cropping seasons per year, the FCS from late February to middle July and the SCS from late July to middle November. Rice cultivation and controlling of diseases and insect pests were common practices in southern China.

Measurement of Grain Chalkiness
The seeds of each line were harvested after full maturity. The dried seeds of 10 plants of each line were separately processed into milled rice, and 200 head rice grains of each plant were randomly selected for measurement of chalkiness (Yang et al. 2021). Images of head rice grains were captured and the chalkiness parameters were measured by Microtek ScanWizard EZ scanner and rice quality analyzer SC-E software (Hangzhou Wanshen Detection Technology Co., Ltd., Hangzhou, China, www. wseen. com). PGC is the product of PCG in total grains multiplied by PCA per chalky grain. Endosperm microstructure was observed by SEM. Milled grains were transversely cut with a razor blade. The thin and clean fracture was fixed on SEM stubs and sputter-coated with gold under vacuum. The fixed specimens were observed under a scanning electron microscope (ZEISS EVO MA 15).

Phenotyping of Traits and Statistical Analysis
Heading date, plant height and panicle number per plant was investigated in the field. Grain traits were measured by the yield traits scorer (YTS), a rice phenotypic facility (Yang et al. 2014). The percentages were converted to the arcsine square root for statistical analysis. The student's t-test was used for comparison between two groups. Dunnett t-test was used to compare multiple groups with control group. Least significance range (LSR) was used for multiple range test among multiple groups (Duncan 1955). The data analysis and figure making were done by SPSS statistics 23.0 and OriginPro 9.0 (https:// www. origi nlab. com).

Genotyping of Molecular Markers and Substitution Mapping
Molecular markers labeled "RM" were selected from online resources (https:// archi ve. grame ne. org/ marke rs/). New markers used in this study were designed using the software of Primer Premier 5.0 (Lalitha 2000). The DNA samples were amplified by PCR method. The PCR products were separated by gel electrophoresis on 6% denatured PAGE and detected by the silver staining method (Tan et al. 2020). To develop secondary SSSLs or NILs, 03-08 and 15-08 were crossed with the recipient HJX74. The NILs were developed from F 2:3 populations derived from the crosses. Minimum length, maximum length and estimated length of a substitution segment were estimated as described by Tan et al. (2020). When PGC showed significant difference between SSSL genotype and HJX74 genotype, a QTL for PGC was detected on the substitution segment of SSSL. When multiple NILs showed significantly different phenotype from HJX74 and their substitution segments overlapped, the QTL was located in the overlapping region (Eshed and Zamir, 1995;Tan et al. 2020). Additive effect of a QTL was defined as the phenotypic difference between SSSL and HJX74 (Zhou et al. 2020). The naming of QTLs followed the rules proposed by McCouch et al. (1997).

Supplementary Information
The online version contains supplementary material available at https:// doi. org/ 10. 1186/ s12284-021-00526-4. Additional file 1: Table S1. Phenotypes of chalky traits in SSSLs. Table S2. Markers developed to detect the substitution segments of SSSLs. Table S3. Substitution segments of SSSLs. Table S4. Phenotypes of agronomic traits in SSSLs. Table S5. Average temperatures for 30 days after flowering of rice in two cropping seasons.
Additional file 2: Fig. S1. Scanning electron microscopy (SEM) images of endosperm transverse sections from milled grains. In the grain with chalkiness in HJX74 (a), the starch granules loosely pack in the opaque endosperm cells. In the grains without chalkiness in SSSLs 15-08 (b) and 03-08 (c), compound starch granules are tightly packed, with no air spaces within or between them. Scale bar: 10 μm.