QTL Analysis Based on A Rice Short-Wide Grain CSSL-Z414 and Substitution Mapping of qGL11 and qGW5

Background: Most of rice agronomic traits as grain length etc. are complex traits controlled by multiple genes. Chromosome segment substitution lines (CSSLs) are ideal materials for dissecting and studying of these complex traits. Results: We developed a novel rice short-wide grain CSSL, Z414, deriving from progeny of the recipient parent Xihui 18 (an indica restorer line) and the donor parent Huhan 3 (a japonica cultivar). Z414 contained 4 substitution segments (average length was 3.04 Mb). Compared with Xihui 18, Z414 displayed seven signicantly different traits as grain length, width and weight, chalkiness degree, brown rice rate etc. Then, 8 quantitative trait loci (QTLs) were found responding these difference traits by F 2 population from Xihui 18/Z414. Among them, 6 QTLs (qPL3, qGW5, qGL11, qRLW5, qRLW11, qGWT5) could be veried by novel developed single segment substitution lines (SSSLs, S1-S6). In addition, 4 QTLs (qGL3, qGL5, qCD3 and qCD5) were novel detected by S1 and S5. Thus, the short–wide grain of Z414 was responded by qGL11, qGL3, qGL5, and qGW5. Then, qGL11 and qGW5 were delimited within intervals of 0.405 and 1.14 Mb on chromosomes 11 and 5, respectively, by substitution mapping. Again by sequencing, qRT-PCR and cell morphology analysis, qGW5 should be a novel allele of GS5 and qGL11 is novel QTL encoding CycT1;3, whose specic function of regulating grain length was still unknown. Finally, pyramid of qGL3 (a=0.22) and qGL11 (a=-0.19) displayed qGL11 epistatic to qGL3. In addition, novel S1 and D2 exhibited different grain sizes and lower chalkiness degree. They are potential to be directly used in breeding hybrid rice varieties. Conclusions: We constructed a novel rice short–wide grain CSSL-Z414 with 4

Grain size and weight are important target traits determining rice yield and quality in rice (Feng et al. 2021). At present, more than 400 QTLs related to grain size have been mapped (Huang et al. 2013). Some of them have been cloned. Several signaling pathways that determine grain size have been identi ed by these cloned genes, including phytohormones pathway, G protein signal pathway, ubiquitin-proteasome pathway, mitogen-activated protein kinase (MAPK) signal transduction pathway, and transcription factor regulation (Li and Li. 2016). Several phytohormones such as brassinosteroids (BRs) and auxins (IAA) have been suggested to play an important role in rice grain size. In rice, Three QTLs (GS5, qGL3/GL3.1 and GW5) for grain size might be involved in BR signaling. GS5 encodes a putative serine carboxypeptidase, and competitively inhibits the interaction between OsBAK1-7 and OsMSBP1 by in uencing BR signaling. Higher expression of GS5 results in wide and heavy grains as a result of increased cell proliferation and expansion in spikelet hulls (Li et al. 2011;Xu et al. 2015a). GL3.1 suppresses BR signaling by regulating the phosphorylation and stability of OsGSK3 (Gao et al. 2019). GW5, as a novel positive regulator of BR signaling, can repress the kinase activity of GSK2, resulting in accumulation of unphosphorylated OsBZR1 and DLT proteins in the nucleus to mediate BR responsive gene expression and growth responses, thus affect the grain width (Liu et al. 2017). Two QTLs (TGW6 and BG1) for rice grain size might participate in IAA signaling. TGW6 encodes indole-3-acetic acid (IAA)glucose hydrolase. In sink organs, the Nipponbare tgw6 allele limits cell number and grain length by controlling IAA supply. Loss of function of the Kasalath allele enhances grain weight and yield signi cantly (Ishimaru et al. 2013). BG1 controls rice grain size as a positive regulator of IAA response and transport ). G protein signaling is involved in regulating grain size in rice. In the pathway, heterotrimeric G protein is usually composed of ve subunits. Gα subunits of RGA1 provides a foundation for grain size expansion, Gβ subunits of RGB1 is essential for plant survival and growth, Gγ subunits of GS3 act as a brake in this pathway, GS3 reduces grain length by competing for binding to RGB1 to inhibit the downstream signal of DEP1/GGC2 (Sun et al. 2018). Ubiquitin can directly or indirectly regulate grain size by affecting protein transport, signal transduction and degradation. GW2, encoding a RING-type E3 ubiquitin ligase, negatively regulates cell division by targeting its substrate to proteasomes for regulated proteolysis. 1 bp missing in the GW2 allele from WY3 causes the terminated translation prematurely during the transcription process, which makes the substrates that should be degraded cannot be recognition, and the grain length becomes longer (Song et al. 2007). LG1 encodes OsUBP15 that possesses de-ubiquitination activity in vitro, loss-of-function and down-regulated expression of OsUBP15 produced smaller grains (Shi et al. 2019). MAPK cascades transmit developmental signals to target molecules through sequential phosphorylation (Xu et al. 2015b). OsMKKK10, OsMKK4 and OsMAPK6 act in the same MAPK pathway to regulate grain size as a positive regulator (Xu et al. 2018). GSN1 is a negative regulator of the OsMKKK10-OsMKK4-OsMPK6 cascade, and gsn1 mutant has larger grain size (Guo et al. 2018). Transcription regulators are considered to be important factors in controlling plant seed size. GS2 affects grain size by encoding a transcription factor OsGRF4 (Duan et al. 2015). GW8 as a SBP-domain positive transcription factor regulates grain width, and can bound directly to the GW7 promoter and repress its expression .
Although several genes for rice grain size have been cloned, it is still fragmentary for our understanding of the underlying mechanisms that regulate rice grain size. Thus, it is necessary to identify more QTLs for grain size. Here, we identi ed a rice short-wide grain chromosome segment substitution line (CSSL) Z414 with 4 substitution segments from indica restorer line Xihui 18 as recipient parent and Huhan 3 as donor parent. Chromosome segment substitution lines (CSSLs) have become important materials for QTL mapping (Li et al. 2019). In this study, we will clarify the following theoretical problems: Since Z414 contained 4 substitution segments, how many QTLs will respond the difference traits in Z414 and how are they distributing in these substitution segments? If QTLs controlling the same trait are more than one, will they display independent inheritance or epistatic interaction? Which of these QTLs are reported or novel? Hence, we characterize Z414 systematically and map quantitative trait loci (QTLs) for associated traits by a secondary F 2 population derived from a cross between Xihui18 and Z414. And then validate these QTLs and analyze the inheritance model and pyramid effect of target QTLs using the singlesegment substitution lines (SSSLs) and dual-segment substitution lines (DSSLs) developed in the F 3 generation. Finally, we also analyze the candidate genes of major qGL11 and qGW5. The ndings will be important for our design breeding plan.

Materials
Z414 was developed from Xihui 18 as the recipient parent and Huhan 3 as the donor parent. Xihui 18 was an elite indica rice restorer line bred by Southwest University, with the characteristics of good combining ability, large panicle and multiple grains, and narrow-long grain. Huhan 3 as a japonica variety had the characteristics of strong stress resistance and short-wide grain. Firstly, 429 simple sequence repeats (SSR) markers covering the whole rice genome were used to screen the polymorphisms between Xihui 18 and Huhan 3. Then, 241 polymorphic markers were selected to develop CSSLs beginning from BC 2 F 1 , 20 plants for each line in each generation were taken. Until in BC 3 F 7 , a short-wide grain CSSL-Z414 with 4 substitution segments was identi ed. The identi cation of chromosome substitution segments referred to the method described by Ma et al (2019), and the estimated length of chromosome substitution segments was calculated according to the method of Paterson et al (1991

Assessment of agronomic traits and quality parameters
In the maturity period, 10 plants of Xihui 18 and Z414 and 6 SSSLs and 2 DSSLs together with 150 individuals of F 2 were harvested. Eleven yield-related traits were investigated including plant height, panicle number per plant, panicle length per panicle, spikelets number per panicle, grains number per panicle, seed-set rate, grain length, grain width, ratio of length to width, 1000-grain weight, and the yield per plant. The speci c method followed Wang et al (2020).
Five quality parameters were analyzed referring to the national standard GB/T5495-2008. Firstly 10 g grains of Xihui 18, Z414,150 F 2 individuals, 6 SSSLs and 2 DSSLs were ground into brown rice. The brown rice was milled into polished rice using the CLS.JNM-1 rice husker. And then brown rice rate and head rice rate were calculated. The chalky rice rate and chalkiness degree were measured using all head rice for each sample by Wanshen SC-E. And the gel consistency was measured refer to the method described in Tang et al (1991).
The mean values of each trait were used for further according analysis.
Scanning electron microscope analysis of glumes in Z414 and Xihui 18 At the completion of the booting stage and before the heading period, the inner and outer epidermal cells of the glume in Xihui18 and Z414 were investigated using a Hitachi SU3500 scanning electron microscope (Hitachi, Tokyo, Japan) with a frozen stage (−40℃) under a low-vacuum environment.
QTL mapping QTL mapping population was a secondary F 2 population consisted of 150 individuals derived from crosses between Xihui18 and Z414.The improved CTAB method described by McCouch et al(1988) was used to extract DNA from the parents and 150 F 2 individuals. PCR ampli cation, polyacrylamide gel electrophoresis and rapid silver staining were carried out by the method described by Zhao et al (Zhao et al. 2016). Xihui 18 lanes were scored as "−1", Z414 lanes were scored as "1", heterozygous lanes were scored as "0", and the absence of marker lanes was scored as ".". Lanes of each marker located on the substitution segment of Z414, together with the phenotypic value of each individual of the F 2 population were used to identify putative QTL using the restricted maximum likelihood (REML) method implemented in the HPMIXED procedure in SAS 9.3 (SAS Institute Inc., Cary, NC, USA). The P-value <0.05 was used as the threshold to determine whether a QTL linked to a certain marker on the substitution segment of Z414.
Development of SSSLs and DSSLs, and validation and epistatic interaction analysis of target QTLs

Development of SSSLs and DSSLs
According to QTL mapping in 2019, eight individuals with target segment and less heterozygous markers selected from the F 2 were used to develop SSSL and DSSL by molecular marker-assisted selection (MAS) method. Each was planted as a line (Z728-Z735) in 2020. Then, the leaves of 20 individuals for each line were taken to extract DNA to conduct molecular detection using both the target substitution markers and the residual heterozygous markers by MAS. Development of SSSL and DSSL obeyed the rule that each substitution line carried only the homozygous target substitution segment while the lanes of other markers were same with Xihui 18.
Ten plants of 6 SSSLs, 2 DSSLs and Xihui 18 in each plot were harvested after maturity in August of 2021. All the involved traits were measured again with the same methods in 2019.

Validation of target QTLs by SSSLs
Concerning each SSSL i (S1-S6), given that the theoretical hypothesis H0: no any QTL was existed on the substitution segment of SSSL i . Then when P-value was less than 0.05 by student t-test, we denied the hypothesis and considered that a QTL for a certain trait existed in SSSL i . According to the genetic model under certain environment (same year and same experimental eld and no replicate plot designed), P 0 =μ+ε for Xihui18 and P i =μ+a i +ε for SSSL carrying a speci c QTL, where, P 0 and P i represented the phenotype value of any plant in plot of Xihui 18 and the SSSL i . μ represented the mean value of Xihui18 population, a i represented additive effect of the QTL, ε represented random error. We took the half of the difference of phenotype value into account as caused by inheritance. Thus, additive effect of the QTL was calculated as half the difference between the mean phenotypic values of SSSL and Xihui18 . All calculations were conducted in Microsoft Excel 2016.

Epistatic interaction analysis between target QTLs by DSSLs
Regarding each DSSL ij , given that the theoretical hypothesis H0: two loci for a certain trait located in "i" and "j" substitution segment belonged to independent inheritance, showed as "2+0=1+1". Then when Pvalue was more than 0.05 by comparing (Xihui18 + DSSL ij ) and (SSSL i + SSSL j ) using student t-test, we accepted the hypothesis that two loci belonged to independent inheritance. At this time, the phenotype value of (Xihui18 + DSSL ij ) was the same with (SSSL i + SSSL j ). In contrast, when P-value was less than 0.05 by student t-test, we denied the hypothesis and considered that epistatic interaction occurred between the two allelic loci, namely "2+0 ≠1+1". According to the genetic models, P 0 =μ+ε for Xihui18, P i =μ+a i +ε for SSSL i, P j =μ+ a j +ε for SSSL j and P ij =μ+a i + a j + I ij +ε for DSSL ij , where P ij represented the phenotype value of any plant in plot of the DSSL ij , μ represented the mean value of Xihui18 population, a i and a j represented the additive effect of QTL in substitution segment i and j, respectively. I ij represented the a i a j epistatic effect between QTLs in substitution segment i and j. We also took the half of the difference of phenotype value into account as caused by inheritance. Thus, the epistatic effects between non-allele QTLs were estimated as half of the mean phenotypic values of (Xihui18+DSSL ij )-(SSSL i +SSSL j ) ). Finally, we used IBM SPSS Statistics 25.0 to conduct multiple comparison for all SSSLs and DSSLs as well as Xihui18.
Overlapping substitution mapping and candidate gene analysis of qGL11 and qGW5 For qGW5, we developed three secondary SSSLs (S3-S5) with overlapping substitution segments in the F 3 population in 2020. For qGL11, we constructed 5 SSSLs (S7-S11) with overlapping substitution segments in the F 2 population from Xihui 18 /S6 in 2021. The maximum and estimated substitution length of the according secondary SSSLs were estimated by the marker positions (Paterson et al,1991)). The QTLs were located by substitution mapping . When GL or GW showed signi cant difference between secondary SSSL genotype and Xihui 18 genotype, a QTL for GL or GW was detected on the substitution segment of SSSL. When multiple substitution segments in SSSLs with target trait overlapped, the QTL was located in the overlapping region ). The additive effect of the QTL was calculated as half the difference between the mean phenotypic values of SSSL and Xihui18 . Within the maximum substitution intervals, we predicted the candidate gene information and combined with gene annotations to screen the possible candidate genes of qGL11 and qGW5 by the Gramene (http:/www.gramene.org/) and the Rice Annotation Project Data-base (https:/rapdb.dna.affrc.go.jp/) and the China National Rice Database Center (http:/www.ricedata.cn/).
The candidate gene sequence, including 3000 bp before the start codon ATG and 1500 bp after the stop codon was downloaded, and the primers were designed on Vector NTI to amplify the target fragments using DNA of Xihui 18 and corresponding SSSL as templates, respectively. The amplicons were sequenced by Tsingke Biological Technology Co., Ltd. (Chongqing, China).
Total RNA extraction and qRT-PCR analysis Total RNA was extracted from root, stem, leaf, sheath and panicle of Xihui 18 and SSSLs using the RNAprep Pure Plant RNA Puri cation Kit (Tiangen, Binjing, China) in the booting stage, and reverse transcribed using the GoScript Reverse Transcription System, and then analyzed quantitatively on a Bio-Rad CYF96 using real-time PCR Master Mix (TaKaRa Biotechnology (Dalian, China) Co. Ltd.).

Identi cation of substitution segments and phenotype analysis of Z414
In the study, 8 polymorphic SSR markers on the substitution segments and 233 polymorphic SSR markers outside the substitution segments of Z414 were used to detect the molecular backgrounds of Z414. The results showed that the substitution segments of 10 plants of Z414 were consistent and no other residual segments from Huhan 3 were detected. Z414 contained 4 substitution segments from Huhan 3, which were distributed on the chromosome 3, 5 and 11. The total estimated length of substitution segments was 12.17 Mb, and the average length was 3.04 Mb (Fig. 1).
Compared with Xihui 18, Z414 displayed signi cant increase in grain width (Fig. 2c), 1000-grain weight, brown rice rate and chalkiness degree by 23.5%, 16.4%, 8.8% and 26.1%, respectively (Fig. 2f, h, i, j). While there was signi cant decrease in panicle length, grain length and ratio of length to width of Z414 (Fig. 2b,  c), reducing by 17.5%, 7.7% and 25.1%, respectively (Fig. 2d, e, g). There were no signi cant differences in the other traits (Fig. 2a), such as plant height, panicle number per plant, spikelet number per panicle, grain number per panicle, seed-set rate, yield per plant, head rice rate, chalky rice rate and gel consistency (no difference data not shown).
Cytological analysis of the glumes in Z414 and Xihui 18 Since the grain length and width of Xihui 18 was different from Z414 (Fig. 3a, d), Scanning electron microscopy was used to analyze the cell morphology of glumes in Xihui 18 and Z414 at the heading stage. The cell width in the inner epidermis of the glumes of Z414 was increased signi cantly than that of Xihui 18 by 22.23% (Fig. 3c, f, h). The total cell number in the outer epidermis of the glume along the longitudinal axis of Z414 was reduced signi cantly than that of Xihui 18 by 13.52% (Fig. 3b, e, i). The cell length in the inner epidermis of the glumes of Z414 exhibited no signi cant difference compared with that of Xihui 18 (Fig. 3c, f, g). The results indicated that the short-wide grain of Z414 was mainly caused by decrease of glume cell number and increase of glume cell width.
Identi cation of QTL using the secondary F 2 population from Xihui18/Z414 A total of 8 QTLs were detected on 3 substitution segments of Z414. They explained the phenotypic variation from 3.86-50.39% (Table 1). The allele qGW5 from Huhan 3 increased the grain width of Z414 by 0.12 mm, explaining 50.39% of the phenotypic variation. The additive effect of qGWT5 from Huhan 3 increased 1000-grain weight of Z414 by 0.63 g. Furthermore, qGW5, qRLW5 and qGWT5 all linked to the same marker RM5874. The allele qGL11 from Huhan 3 reduced the grain length of Z414 by 0.08 mm, explaining 9.96% of the variation in grain length. Similarly, qGL11, qRLW11, qGWT11 and qBRR11 all linked to the same marker RM1812. However, the additive effects of qGL11, qRLW11and qGWT11 decreased the values of corresponding traits, while qBRR11 increased the value (Fig. 1, Table 1). In addition, the additive effect of qPL3 reduced panicle length of Z414 by 0.58 cm per panicle (Table 1). On the basis of primary QTL mapping, 6 single segment substitution lines (SSSLs, S1-S6) and 2 dualsegment substitution lines (DSSLs, D1, D2) were developed in F 3 by MAS method. Among them, S3, S4, and S5 belonged to SSSLs with overlapping substitution segments (Fig. 4a).
6 QTLs (qPL3, qGW5, qGL11, qRLW5, qRLW11, qGWT5) could be veri ed by SSSLs (Fig. 4a-f), which indicated that these QTLs could be inherited stably. 2 QTLs (qGWT11 and qBRR11) were not validated by S6, suggesting that the expression of some minor QTLs might be easily in uenced by the environment, whose contribution rates to phenotypic variation were only 3.86% for qGWT11 and 5.92% for qBRR11. In addition, 4 QTLs (qGL3, qGL5, qCD3 and qCD5) for grain length and chalkiness degree were detected by S1 and S5 (Fig. 4c, g), which were not detected in the secondary F 2 population (Table1). The results showed that SSSLs had a higher e ciency of QTL detection.
Pyramid of qGL3 (a= 0.22) and qGL11 (a=-0.19) yielded an epistatic effect of -0.31, which resulted in reduction 0.28 mm of the grain length in D2. The result suggested that pyramid of qGL3 and qGL11 resulted in shorter grains than S6 (containing qGL11) (Fig. 4c), indicating that qGL11 displayed epistatic to qGL3. However, qGL3 (a= 0.22) and a substitution locus without GL on chromosome 3 in D1 belonged to independent inheritance. The grain length (10.25 mm) of D1 exhibited no signi cant difference with that (10.35 mm) of S1, while signi cantly longer than that of Xihui18 and S2 (Fig. 4a, c). Pyramid two substitution loci without QTL for 1000-grain weight on chromosomes 3 and 11 in D2 produced an epistatic effect of -2.94, resulting in decrease 2.94 g of 1000-grain weight genetically in D2. Thus, 1000grain weight (28.99 g) of D2 displayed signi cantly lower than that (32.65, 31.98, and 30.75 g) of S1, S6 and Xihui18 (Fig. 4a, e). As for the other QTLs in D1 and D2, they all belonged to independent inheritance (Fig. 4b, d, f, g).
Substitution mapping and candidate gene analysis of qGL11 and qGW5 candidate gene analysis of qGL11 According to the above results, we rstly dissected qGL11 into S6 whose estimated and maximum substitution length was 1.42 Mb and 1.66 Mb, respectively (Fig. 5a). In order to ne mapping of qGL11, we developed 5 novel secondary SSSLs (S7-S11) by a cross of Xihui 18 and S6. By theory of substitution mapping, qGL11 was nally delimited into 405 Kb of the estimated substitution interval (Fig. 5a). There were 44 genes in total were found in the estimated substitution interval, only 18 genes with speci c functional description except the others annotated as 16 expressed protein, 6 transposon protein and 2 retrotransposon protein and 2 carrier & putative protein. Then, we found only LOC_Os11g05850 (CycT1;3) might be the candidate gene of qGL11 according to the possible signaling pathway regulating grain size (Li and Li. 2016), By DNA sequencing between Xihui 18 and S6, there were 6 SNP differences and a 25base insertion in the 5'UTR and 1 SNP difference in the 3'UTR, and 1 SNP difference in the CDS which did not cause amino acid change (Fig. 5b). Furthermore, the protein structure displayed no difference between S6 and Xihui18 (Fig. 5c). Especially, expression levels of LOC_Os11g05850 was signi cantly higher in sheath and panicle in S6 than in Xihui18 (Fig. 5d). Thus, LOC_Os11g05850 (CycT1;3) should be the candidate gene for qGL11.
Moreover, the protein structure also displayed some differences between S5 and Xihui18 (Fig. 6c). For LOC_Os05g07720 (OsTAR1), as an IAA biosynthesis gene, there were 4 SNP differences in the CDS between Xihui18 and S5, of which 3 caused amino acid mutations and 1 nonsense mutations (Fig. 6d). Furthermore, qRT-PCR analysis showed that expression levels of the LOC_Os05g06660 (GS5) were signi cantly higher in stem, leaf, sheath and panicle in S5 than in Xihui18 (Fig. 6e), while no signi cant expression differences of the LOC_Os05g07720 (OsTAR1) were found in all organs between Xihui 18 and S5 (Fig. 6f). Thus, LOC_Os05g06660 (GS5) should be a different allele of qGW5 and LOC_Os05g07720 (OsTAR1) were potential one for qGW5.

Discussion
We construct a novel short-wide CSSL-Z414 and novel series of secondary SSSLs for different grain size, which are valuable for both genetic study for grain quality and hybrid rice breeding wide and larger grain, shorter panicle, and higher brown rice rate and chalkiness degree when compared with Xihui18. Thus, Z414 did not suit to be directly acted as rice restorer line due to its high chalkiness degree. While it is an ideal material for genetic study due to its near isogenic habit. Luckily, by genetic dissection, we obtained 6 novel single segment substitution lines (S1-S6) and 2 dual-segment substitution lines (D1-D2) harboring target QTLs. Intriguingly, compared with Xihui 18, S1 carried qGL3 and qCD3 and exhibited both long grain and lower chalkiness degree. D2 harbored qGL3, qGL11 and qCD3 where qGL11 was epistatic to qGL3 and thus displayed shorter grain and lower chalkiness degree.
S4 contained qGW5, qRLW5 and qGWT5 and exhibited long-wide large grain and same chalkiness degree with xihui18. S6 carried qGL11 and qRLW11 and showed short-narrow grain and same chalkiness degree with xihui18. Lower chalkiness degree is a very important factor affecting rice grain quality ). Again, different grain sizes are import to meet the requirement of various people's preferences (Feng et al. 2021;Liang et al. 2021). Thus, these four SSSLs are potential to be directly used as novel rice restorer lines to breed novel hybrid rice varieties. As for S5 with short-wide grain (qGL5, qGW5, qGWT5, qRLW5) and high chalkiness degree (qCD5), it can be used in researching the formation mechanism of these traits. Zhang (2021) argued that SSSLs is helpful for rapid screening of traits hidden in genomes of different donors, make a large amount of previously unexplored genetic variation rapidly available to plant breeders and geneticists, and make the genetic variation directly usable for breeding. SSSLs represents a new resource that could greatly enrich conventional rice breeding (Zhang. 2021). Obviously the novel SSSLs is an unusual gene pool for genetic research for grain quality and rice breeding. They will play important part in alleles discovery and implement the strategy for research on rice breeding by design.
qGW5 should be a different novel allele of GS5 and qGL11, qRLW11, qCD3 and qBRR11 should be novel QTLs compared with the reported genes To explore the problems proposed in introduction, we developed the secondary F 2 segregation population from Xihui 18/Z414 and 6 SSSLs (S1-S6) and 2 DSSLs (D1 and D2), Finally a total of 12 QTLs were found to be responsible for seven difference traits of Z414. Based on the analysis of these QTLs, the short grain of Z414 was controlled by qGL11, qGL3 and qGL5, which was dissected into S6, S1 and S5, respectively. The wide grain of Z414 was responded by qGW5, which was then validated by S4 and S5. Large grain of Z414 was in charge of qGWT5 (veri ed by S4 and S5) and qGWT11 (veri ed by S6). The short panicle of Z414 was harbored by qPL3 (validated by S1). Higher brown rice rate of Z414 was explained by qBRR1. The qCD5 and qCD3 (detected by S5 and S1) answered for the higher chalkiness degree of Z414.
Compared with the reported genes, OsAPC6 and OspPLAIIIα can be as the candidate genes for qPL3 according to its physical distance and biological function. OsAPC6 interferes with the gibberellin signal pathway and leads to cell reduction (Awasthi et al. 2012). OspPLAIIIα encodes glycoprotein-related phospholipase A and reduces panicle length (Liu et al. 2015b). According to the results in the study, qGW5 for grain width, qGL5 for grain length, qRLW5 for ratio of length to width and qGWT5 for 1000grain weight were within the interval of RM405 to RM17984 of chromosome 5. At the substitution interval, GS5 and OsTAR1 were potential candidate genes of these QTLs. GS5 is a positive regulator of grain size such that grain width and weight are correlated with its expression level. Polymorphisms of GS5 in the promoter region are responsible for the variation in grain size. Xu et al (2015b) showed that two SNPs between the wide-grain allele GS5-1 and the narrow-grain allele GS5-2 in the upstream region of the gene that were responsible for the differential expression in developing young panicles. By DNA sequencing of GS5, we found that two GGC repeat encoding Glycine were added in CDS sequence of S5 compared with that of Xihui18, and the protein structure also existed some differences in S5 compared with Xihui18. While there were no differences found in the promoter region between Xihui18 and S5. Again, qRT-PCR analysis showed that the expression level of GS5 was signi cantly higher in stem, leaf, sheath and panicle in S5 than in Xihui18. Thus, qGW5 should be a different allele of GS5, which is important for studying biodiversity of grain size. OsTAR1 encoding tryptophan aminotransferase, a IAA biosynthesis gene, regulate the production of IAA in the developing rice grain together with OsYUC9 and OsYUC11 (Abu-Zaitoon et al, 2012). And OsYUC11-mediated auxin biosynthesis is essential for endosperm development in rice (Xu et al, 2021b). By DNA sequencing, there was 4 SNP differences in the CDS between Xihui18 and S5. While the expression of OsTAR1 displayed no signi cant differences between Xihui 18 and S5. Thus, OsTAR1 also can act as the potential candidate gene of qGW5. Of course, whether GS5 and OsTAR1 are target genes, the genetic complementary experiments of two genes are necessary and they are ongoing. Although GS5 and OsTAR1 have been cloned, we have found the different alleles from Huhan 3 and developed the S5 based on the genetic backgrounds of Xihui 18, which can be directly used in our design breeding plan. qGL3 was in the similar region with PGL1, OsLG3, TUD1 and OsOFP19. PGL1 encodes an atypical basic helix-loop-helix protein (bHLH) that does not bind DNA. Its overexpression can increase grain length (Heang et al. 2012). OsLG3 positively regulates rice grain length and has no effect on grain quality ). TUD1 encodes an E3 ubiquitin ligase of the U-box family, which participates in the brassinolide response and interacts with the heterotrimeric G protein subunit D1 to regulate brassinolide-mediated rice growth (Hu et al. 2013). OsOFP19, OSH1 and DLT might form a functional complex that regulates cell proliferation and cell growth (Yang et al. 2018). qCD5 was in the similar region with Chalk5. Chalk5 encodes a vacuolar H+-translocating pyrophosphatase (V-PPase), which increases the chalkiness of the endosperm by disturbing the PH homeostasis of the endomembrane tra cking system in developing seeds (Li et al. 2014). Although some of these genes have been cloned, when compared with their identi cation in mutant, they are more usable in breeding practice by identi cation in SSSLs. Consequently, these alleles are still important for both biodiversity research and pyramid breeding based on genetic background of Xihui 18. In addition, qGL11, qRLW11, qCD3 and qBRR11 have not been previously reported, our knowledge. They can be further used for ne mapping, cloning and functional analysis to explore the genetic mechanisms of these quality trait.
qGL11 is a novel QTL encoding CycT1;3 whose function in regulating grain length was still unknown Elucidation of the molecular mechanism underlying grain size is important for rice design breeding. By substitution mapping of qGL11 using 6 SSSLs (S6 -S11) with overlapping substitution segments each other, qGL11 was nally delimited within 405 Kb of estimated substitution interval on chromosome 11. At this interval, there were 44 genes in total, in which 18 genes have speci c functional description except 26 genes for expressed protein, transposon protein, retrotransposon protein and carrier & putative protein.
Considering genes associated with grain size that have been cloned, the majority involved in phytohormones pathway, G protein signal pathway, ubiquitin-proteasome pathway, Mitogen-activated protein kinase (MAPK) signal transduction pathway, and transcription factor regulation (Li et al, 2016). Among the 18 candidate genes, only CycT1;3 might be the candidate gene for qGL11. By DNA sequencing and qRT-PCR analysis between Xihui 18 and S6, there were differences in both DNA sequences and gene expression levels in sheath and panicle for CycT1;3 between Xihui 18 and S6. Thus, CycT1;3 should be the candidate gene for qGL11. CycT1;3 encodes cyclin protein, involving in biological progress of the cell Mitosis cycle. Intriguingly, cytological analysis showed that the shorter grain of Z414 depended on the decrease of cell number of glume rather than cell length. The results suggest that qGL11 is related to cell mitosis cycle. There were many studies on cell cycle regulation in yeast and animals, but few studies in plants. Cell proliferation in plants is mainly controlled by a super-family of cyclindependent kinases (CDKs). There are more reports of A, B, D and E-type family cyclins in plants (Nieduszynski et al. 2002), but still few reports on T-type family cyclins. Therefore, T-type cyclin in plant is worthy to further study. Qi et al. (2012) found that GL3.1 encoding OsPPKL1 can directly use Cyclin-T1;3 as a substrate to phosphorylate Cyclin-T1;3, which makes Cyclin-T1; 3 down-regulated in rice and resulting in shorter grains. However, the molecular mechanism of how CycT1;3 affects grain development remains unknown. In this study, although we detected qGL3 in S1 and found that qGL11 was epistatic to qGL3, however, OsPPKL1 was not in the institution interval of S1. Thus, qGL11 is a novel QTL. On all accounts, these results are important for the in-depth study of qGL11.

Conclusions
We constructed a novel rice short-wide grain CSSL-Z414 carrying 4 substitution segments based on the genetic backgrounds of indica restorer line Xihui18. Z414 displayed seven different traits from Xihui 18, including grain length, grain width, ratio of length to width, 1000-grain weight, brown rice rate, chalkiness degree and panicle length. In total, 12 QTLs responded seven different traits of Z414, and they were dissected into 6 novel SSSLs (S1-S6) and 2 novel DSSLs (D1 and D2). The short grain of Z414 was controlled by qGL11, qGL3 and qGL5. By Cytological analysis, DNA sequencing and qRT-PCR analysis, qGL11 should be CycT1;3, whose speci c function regulating grain length was still unknown and is a novel QTL. The grain width of Z414 was controlled by qGW5, and should be a different novel allele of GS5. In particular, S1 carried qGL3 and qCD3 and exhibited both long grain and lower chalkiness degree. D2 harbored qGL3, qGL11 and qCD3 where qGL11 was epistatic to qGL3 and thus displayed shorter grain and lower chalkiness degree. They are potential to be directly used as novel rice restorer lines to breed novel hybrid rice varieties.

Funding
The study was supported by National natural science foundation of China (31871593), the Chongqing technical innovation and application development Project (cstc2019jscx-msxmX0392).

Availability of Data and Material
The datasets supporting the conclusions of this article are included within the article.

Ethics Approval and Consent to Participate
This study complied with the ethical standards of China, where this research work was conducted.

Consent for Publication
All authors provide their consent for publication.     Substitution mapping, sequence analysis and relative expression level of qGL11 between Xihui 18 and S6. a Substitution mapping of qGL11. Black regions indicate the estimated length of substitution segment. b The DNA sequence of CycT1;3 in S6 compared with Xihui 18. In candidate gene sequence, the red box represents the coding sequences region, the white box represents 5'UTR and 3'UTR, the solid red line represents introns, the black line in the gene sequence presents the mutation site, and arrow represents sequence change from Xihui18 to S6. c Protein structure of CycT1;3 predicted by SWISS-MODEL. d Relative expression level of candidate genes CycT1;3 in root, stem, leaf, sheath and panicle between Xihui 18 and S6. Figure 6 Substitution mapping, sequence analysis and relative expression level of candidate genes for qGW5 between Xihui 18 and S5. a QTL mapping of qGW5. Black regions indicate the estimated length of substitution segment. b The DNA sequence of GS5. c Protein structure of GS5 predicted by SWISS-MODEL.d The DNA sequence of OsTAR1 in S5 compared with Xihui 18. In candidate gene sequence, the red box represents the coding sequences region, the white box represents 5'UTR and 3'UTR, the solid red