QTL Mapping of Kernel Traits and Validation of a Major QTL for Kernel Length-Width Ratio Using SNP and Bulked Segregant Analysis in Wheat

One RIL population derived from the cross between Dalibao and BYL8 was used to examine the phenotypes of kernel-related traits in four different environments. Six important kernel traits, kernel length (KL), kernel width (KW), kernel perimeter (KP), kernel area (KA), kernel length/width ratio (KLW), and thousand-kernel weight (TKW) were evaluated in Yangling, Shaanxi Province, China (2016 and 2017), Nanyang, Henan Province, China (2017) and Suqian, Jiangsu Province, China (2017). A genetic linkage map was constructed using 205 SSR markers, and a total of 21 significant QTLs for KL, KW, KP, KA, KLW and TKW were located on 10 of the 21 wheat chromosomes, including 1A, 1B, 2A, 2B, 2D, 3D, 4D, 5A, 5B, and 7D, with a single QTL in different environments explaining 3.495–30.130% of the phenotypic variation. There were four loci for KLW, five for KA, five for KL, three for KP, two for KW, and two for TKW among the detected QTLs. We used BSA + 660 K gene chip technology to reveal the positions of major novel QTLs for KLW. A total of 670 out of 5285 polymorphic SNPs were detected on chromosome 2A. The SNPs in 2A are most likely related to the major QTL, and there may be minor QTLs on 5B, 7A, 3A and 4B. SSR markers were developed to verify the chromosome region associated with KLW. A linkage map was constructed with 7 SSR markers, and a major effect QTL was identified within a 21.55 cM interval, corresponding to a physical interval of 10.8 Mb in the Chinese Spring RefSeq v1.0 sequence. This study can provide useful information for subsequent construction of fine mapping and marker-assisted selection breeding.

Among all the QTLs detected by the ICIM and GCIM methods, there were 22 common QTLs, and six additional QTLs were also identified by the GCIM method. Among these 6 QTLs, including 3, 2, and 1 QTLs for KL, TKW, and KP had been detected, respectively. Therefore, we combined the results of these two modules and identified 45 QTLs for six kernel traits. The 45 QTLs contributed to 12 chromosomes. A total of 4, 8, 17, 11, and 5 QTLs were identified for TKW, KA, KP, KLW, KL, and KP, respectively. Among the 45 QTLs, we performed stepwise regression analysis of all the QTLs on each trait. And then we introduced in details all the QTLs remained in the best regression equation of the stepwise regression analysis. At this time, the number of significant QTLs is 21.
Among the above 21 significant QTLs, two were found to be associated with TKW. One QTL (qTKW3D-1) was identified commonly by the ICIM and GCIM methods and located between markers Xgwm191 and Xgwm71 on chromosome 3D. Another was located between markers Xwmc728 and Xwmc367 on chromosome 1B. Each of the two QTLs accounted for approximately 11.0% of the PV in E4 (2017-2018, Suqian) and E1 (2016-2017, Yangling). Their novel alleles were derived from BYL8.
Three QTLs associated with KP were mapped onto chromosomes 2A, and 5A, explaining 8.115%-27.718% of PV. They were identified commonly by the ICIM and GCIM methods and their novel alleles were derived from Dalibao. The QTL, qKP5A-1 was located between marker Xbarc56 and Xbarc180 in E1. Another two QTLs, qKP2A-1 and qKP5B-1 were within marker intervals Xwmc658-Xwmc181 and Xbarc232-Xbarc500 in E3.
Four significant QTLs for KLW were mapped onto chromosomes 2A, 5B, and 7D. In particular, two QTLs (qKLW2A-1and qKLW2A-3) were detected in the marker interval between Xwmc658 and Xwmc181 on chromosome 2A in two different environments, explaining 6.262%-28.540% of PV, this locus can be considered as a major QTL for KLW. They were identified commonly by the ICIM and GCIM methods and their novel alleles were derived from BYL8. One QTL, qKLW5B-1 within the marker interval between Xbarc340 and Xwmc705 detected in E2 (2017-2018, Yangling), with positive effects coming from also BYL8. And one QTL, qKLW7D-1, was within the marker intervals between Xbrac76 and Xwmc698 in E4. It was identified commonly by the ICIM and GCIM methods and novel allele was derived from Dalibao.
Five significant QTLs associated with KL were detected on chromosomes 2B, 2D, and 7D, explaining 9.685-19.618% of the PV. Of these QTLs, two QTLs (qKL7D-1and qKL7D-2) were detected in the marker interval between Xbrac76 and Xwmc698 on chromosome 7D in two different environments. Their novel allele was derived from Dalibao but only qKL7D-2 was identified commonly by ICIM and GCIM methods. Another two QTLs (qKL2B-1 and qKL2B-2) were located within marker interval Xwmc501-Xwmc154 and Xwmc154-Xwmc25.1 on the chromosome 2B detected in E1 and E4, respectively. They were identified commonly by ICIM and GCIM methods and their novel allele were derived from Dalibao. Another QTL, qKL2D-1 in the marker intervals between Xwmc154 and Xwmc25.1detected in E4, had positive effects from BYL8.
Two QTLs (qKW1A-1 and qKW2A-1) for KW were considered as significant QTL. They were identified commonly by the ICIM and GCIM methods and located within markers intervals Xwmc24-Xwmc278 and Xwmc658-Xwmc181, accounted for 15.367% and 29.455% of the PV in E2 and E3. Their novel alleles were derived from BYL8.
Pleiotropic QTLs for kernel traits. QTLs related to different kernel traits could be found in the same marker intervals, possibly due to the pleiotropic effect of a single gene or a set of tightly linked genes. In this study, four QTL clusters related to KA, KP, KLW, KL, and KW kernel traits were located on chromosomes 2A, 2B 5A, and 7D (Table 5). A QTL cluster for KA on chromosome 2A was found in two environments, while other QTL clusters were detected in one environment. BSA+ 660 K result analysis. After genotyping with 660 K SNP array, the number of homozygous polymorphisms SNP between DNA bulks and their parental lines were counted, resulting in a total of 5285 SNPs. The chromosome 2A had the highest number of SNPs (670). The others were located on the rest of chromosomes (Fig. 3). Furthermore, for chromosome 2A, the proportion of SNPs that overlapped between the bulk and the parent was the highest. The numbers and the proportions of SNPs on chromosomes 5B, 7A, 3A, and 4B were also higher. These results indicated that SNPs in 2A were extremely likely to be associated with the major locus and there may be minor QTLs on 5B, 7A, 3A, and 4B. Most of the SNPs on 2A were within the interval Ax-94815470-Ax-109351630 spanning 28.8 Mb (2A:751004612-2A:779844131) in the 660 K map. We employed a BLAST search to obtain physical positions of polymorphic SSR markers using the SSR forward primer, which is the newly released Chinese Spring sequence (Reference Sequence v1.0, the International Wheat Genome Consortium (IWGSC), http://www.wheatgenome.org/). The markers Xwmc658 and Xwmc181 spanned 42.6 Mb (2A:728609541-2A:771166682). The QTL mapped for KLW was located between the SSR markers Xwmc658 and Xwmc181, with a genetic distance of 28.9 cM, corresponding to a physical interval of 42.6 Mb in the Chinese Spring RefSeq v1.0 sequence. By contrast, this physical location contains the segment of most of the SNPs in the 660 K results. BSA combined with the 660 K gene chip has located a major QTL for KLW in a smaller region. Subsequently, a genetic map constructed using the 7 newly developed SSR markers (Table 6) and data from the 547 F 7 RILs spanned 93.65 cM. A major QTL for KLW located in a 21.55 cM interval spanned by SSR3 and SSR4 corresponding to a physical interval of 10.8 Mb in the Chinese Spring RefSeq v1.0 sequence.

Discussion
QTLs for kernel traits. The present study investigated six kernel traits using a RIL population consisting of 142 individuals from the cross Dalibao/BYL8. Plant materials were grown under four environmental conditions. A total of 45 QTLs were identified for six kernel-related traits on the chromosomes 1A, 2A, 5A, 7A, 1B, 2B, 4B, 5B, 2D, 3D, 4D, and 7D across four environments. Among these 45 QTLs, there were 21 QTLs were considered as significant QTLs for six kernel traits.
With respect to KL, our study detected five QTLs on chromosomes 2B, 2D, and 7D in three environments. Many previous studies have found QTLs for KL on 2B and 2D 1,2,5,8,[11][12][13][14]16 , indicating that these two chromosomes have an important influence on wheat kernel traits. Okamoto 10 detected a QTL for KL on chromosome 7D.
Significant QTLs for KW on chromosome 1A and 2A were detected in this study. Campbell 1 and Li 14 detected QTLs for TW located on chromosome 1A. In addition, there are many reports of QTLs for KW on chromosome 2A 1,8,12,13 . Among these, Sun 8 detected a QTL associated with marker Xwmc181 on chromosome 2A, the same marker found in our study. Previous studies have also found QTLs on all of the other chromosomes (except for 3A), for which none were found in our study 2,5,6,10,[15][16][17]27,36,37 .
We detected some consistent QTLs for TKW, KL, KW, KA, KP, and KLW in various environments. It is believed that these QTLs can be used for marker-assisted selection breeding and fine mapping.
Pleiotropic QTLs for kernel traits. Among QTLs for kernel traits detected in this study, 4 regions controlled two or more kernel traits at the same time forming overlapping QTLs. There is a significant correlation between kernel traits at the QTL level. In earlier studies, some pleiotropic QTLs related to kernel traits were rep orted 8,11,[13][14][15]28,35,36,[38][39][40] . In this study, there were seven pleiotropic QTLs related to KA, KP, KLW, KL, and KW located on chromosomes 2A, 2B, 5A, and 7D.  www.nature.com/scientificreports www.nature.com/scientificreports/ Five QTLs, one each for KLW (qKLW2A-1), KA (qKA2A-2), KP (qKP2A-1) and KW (qKW2A-1) were located on chromosome 2A within the marker interval between Xwmc658 and Xwmc181. The marker Xwmc181 closest to this pleiotropic QTL was earlier reported by Sun 8 for KW and TKW. Kumari 15 detected three pleiotropic QTLs for KLW and KW, KLW and KL, and KW and KA on chromosomes 6A, 7B, and 7D. The pleiotropic QTL on the 2A chromosome associated with KA was found in two environments, while QTLs for KLW, KP and KW were found in one environment. The QTL effects between KA, KP, KW and KLW were negative, and the relationships between KA, KP and KW were positive, in accordance with the simple correlation analysis results.      www.nature.com/scientificreports www.nature.com/scientificreports/ Two QTLs, one each for KA (qKA2B-1) and KL (qKL2B-2) were co-localized on chromosome 2B in the marker interval between Xbarc200 and Xbarc55. Each of these QTLs was detected in only one environment. There was positive relationship between these three traits, consistent with the simple correlation analysis results.
Two QTLs, one each for KA (qKA5A-1) and KP (qKP5A-1), were mapped onto chromosome 5A within the marker interval between Xbarc56 and Xbarc180. Each of these QTLs was detected in only one environment. The parental genotype Dalibao displayed positive effects for the above QTLs, and the relationship between these two traits was consistent with the results of the simple correlation analysis.
A QTL each for KLW (qKLW7D-1) and KL (qKL7D-1) were mapped in the marker interval between Xbarc76 and Xwmc698 on chromosome 7D. Each of these QTLs was detected in only one environment. The positive relationships between the three traits were in agreement with the simple correlation analysis results. Ramya 11 previously found a pleiotropic QTL for KLW and KL on chromosome 4B. Li 14 have detected a pleiotropic QTL for KLW and KL on chromosomes 1A and 2D.
In this study, QTLs for KL and KW were detected on chromosomes 2B, 2D, 7D and 1A, 2A, respectively. There were no pleiotropic QTLs between these two traits, but the correlation between KL and KW was high. This is not consistent with previous research results that KL and KW were largely under independent genetic control 41 . As for TKW, the correlations between TKW and KL, KW, KLW, KA and KP were significant. Campbell 1 previously reported that KL and KW influenced the QTL for TKW, but no pleiotropic QTLs were found in our study. Previous studies on KLW traits are relatively rare, so no results similar to ours are available for comparison.
These common or overlapping QTL regions need to be fine mapped to determine whether the regions are pleiotropic in different QTLs or are closely linked QTLs that control individual traits. comparison of BSA+ 660 K and SSR results. In our study, we combined BSA with the 660 K SNP array to search for major QTLs for KLW that have not been reported previously. The results suggest that the major QTLs were extremely likely to be related with the SNPs on chromosome 2A, and there may be minor QTLs on 5B, 7A, 3A, and 4B. These results can be verified using SSR markers. However, no genetic positions in relation to the wheat 660 K SNPs have been documented 35 . We can use SSR marker forward primers and SNP probes to obtain physical locations by BLAST using the Chinese Spring RefSeq v1.0 sequence. In subsequent work, we developed more SSR markers based on the results from the wheat 660 K to validate the accuracy. These results should be valuable for the wheat breeder to improve the kernel traits via marker-assisted selection breeding. Phenotype analysis. At maturity, wheat plants were bulk-harvested and sun-dried for phenotypic evaluation. Three individuals from each family were randomly selected and threshed separately. Then, at least 200 fully filled kernels of each line were used to measure six kernel-related traits, (KL, KW, KP, KA, KLW and TKW) using the Wanshen kernel testing equipment developed by Wanshen Science and Technology Ltd. (Hangzhou, China; http://hzwseen.foodmate.net/). All the above traits from the four environments were measured as the average of each family from three replicates. Statistical analysis. The calculations of descriptive statistics, Student's t-test, correlation analysis, and analysis of variance (ANOVA) were performed using SPSS 21.0 software (http://www.spss.com). The phenotypes of six kernel traits used for statistical analysis in the four environments were from the averages of three replicates per family. Additionally, the GLM model in SPSS was used to compute the broad-sense heritability (h 2 ). Frequency distributions for six kernel traits were drawn using the Microsoft Office Excel 2007 software (https://www.microsoft.com).

Materials and Methods
Genotype analysis and QTL mapping. Genomic DNA was extracted from young leaves of both parents and families by the hexadecyl trimethyl ammonium bromide (CTAB) method. TE buffer was used to dissolve extracted DNA, and the DNA quality was detected by 1% agarose electrophoresis. DNA was stored at −20 °C. A total of 142 individuals were genotyped using 205 SSR markers. These 205 SSR markers were evenly selected from the 21 chromosomes of the whole genome using the GrainGenes database (http://wheat.pw.usda.gov/GG2). A 6% denaturing polyacrylamide gel electrophoresis (PAGE) was used to separate PCR conducts. Using the Biparental Populations (BIP) module of the inclusive composite interval mapping (ICIM) in the software QTL IciMapping 4.1 (http://www.isbreeding.net) for QTL analysis 42 . The critical LOD scores for a significant QTL were set at 3.0. The GCIM method in the software QTL.gCIMapping from the R website (https://cran.r-project.org/web/ packages/QTL.gCIMapping/index.html) was also used to identify QTLs for the above traits, with the purpose of identifying the results from the ICIM method 43,44 ; the critical LOD scores for a significant QTL was also set at 3.0, and the walking speed for the genome-wide scan was set at 1 cM.