Dissection of Genetic Basis Underpinning Kernel Weight-Related Traits in Common Wheat

Genetic dissection kernel weight-related traits is of great significance for improving wheat yield potential. As one of the three major yield components of wheat, thousand kernel weight (TKW) was mainly affected by grain length (GL) and grain width (GW). To uncover the key loci for these traits, we carried out a quantitative trait loci (QTL) analysis of an F6 recombinant inbred lines (RILs) population derived from a cross of Henong 5290 (small grain) and 06Dn23 (big grain) with a 50 K single nucleotide polymorphism (SNP) array. A total of 17 stable and big effect QTL, including 5 for TKW, 8 for GL and 4 for GW, were detected on the chromosomes 1B, 2A, 2B, 2D, 4B, 5A, 6A and 6D, respectively. Among these, there were two co-located loci for three traits that were mapped on the chromosome 4BS and 6AL. The QTL on 6AL was the most stable locus and explained 15.4–24.8%, 4.1–8.8% and 15.7–24.4% of TKW, GW and GL variance, respectively. In addition, two more major QTL of GL were located on chromosome arm 2BL and 2DL, accounting for 9.7–17.8% and 13.6–19.8% of phenotypic variance, respectively. In this study, we found one novel co-located QTL associated with GL and TKW in 2DL, QGl.haaf-2DL.2/QTkw.haaf-2DL.2, which could explain 13.6–19.8% and 9.8–10.7% phenotypic variance, respectively. Genetic regions and linked markers of these stable QTL will help to further refine mapping of the corresponding loci and marker-assisted selection (MAS) breeding for wheat grain yield potential improvement.


Introduction
Wheat is one of the most important food crops, which has the largest cultivated area worldwide (http://faostat.fao.org, (accessed on 1 December 2020)). It is the staple food of more than 35% of the world's population, which provides one fifth of the daily intake of calories and protein. [1]. In order to meet the higher demand of wheat due to the increasing human population, rapid urbanization and sharp climate change, it has been estimated that global production is necessary to grow at a rate of 1% per year in the future [2]. Therefore, wheat yield improvement is still the top priority in wheat breeding programs to guarantee global food security.
Wheat yield is a complex trait consisting of three main components-the number of spikes per area, kernel number per spikes, and thousand kernel weight (TKW). Among them, TKW has relatively high heritability [3] and is a quantitative trait. As one key component of grain yield, TKW is mainly influenced by grain size and grain filling [4]. Grain size provides the room for grain filling and can be broken into grain length (GL), grain width (GW), grain thickness (GT) and grain surface area. Compared with GL, the other three components are more sensitive to the environment as a result of their establishment in the later stage of grain development [5][6][7]. Grain filling directly affects dry matter accumulation in grains and can be split into grain filling rate and duration, which could be severely hindered under adversity stress [8][9][10][11][12][13]. Although these components are susceptible to environmental impact, their inheritance are relatively stable compared with the total yield of wheat. Therefore, TKW and its related traits are often used in genetic analysis of wheat grain yield.
In this study, we selected HeNong 5290 (hereafter HN5290) with small grain and 06Dn23 with big grain to make an F 6 recombinant inbred lines (RILs) population for understanding the genetic basis of controlling TKW, GL and GW. The purposes were to detect the stable and co-located loci for TKW, GL and GW and analyze the relationships between detected co-located loci on the three traits.

Phenotypic Statistics
Phenotypic data were obtained from the field trials at three different locations over two continuous years (as such, four environments). The mean value of TKW, GL and GW for 06Dn23 ranged from 46.3-70.8 g, 7.52-8.33 and 3.37-3.89 mm, respectively, while it was 25.9-45.8 g, 6.08-6.67 and 2.79-3.39 mm for HN5290, respectively ( Table 1). The TKW, GL and GW of the RILs ranged from 17.6-67.0 g, 6.85-7.60 and 3.09-3.65 mm, respectively (Table 1). Based on a t-test, 06Dn23 showed significantly higher values for three traits over all environments than HN5290 ( Table 1). The coefficient of variation of TKW ranged from 7.7-15.8%. In addition, the highly significant difference interactions between genotypes and environments for grain weight related-traits among RILs were detected by ANOVA analysis ( Table 2). The frequency distribution of phenotypic values of TKW, GL and GW showed a continuous and normal distribution (Figure 1) with an exception of the 2018 crop cycle, while there was a significant distorted distribution of the three grain-related traits in the 2018 crop cycle, indicating that this might be due to the high temperature during the grain filling period in 2018 (Supplementary Table S3).

Linkage Map Construction
A total of 18,357 SNP loci were polymorphic between two parents. Among them, 4655 SNPs with a high missing rate (>10%) or low distortion p value (<0.001) were removed and 13,702 markers were used to construct linkage map for subsequent QTL mapping. Finally, 70 linkage groups consisting of 12,533 SNPs were established on 21 chromosomes (Supplementary Table S1). The entire linkage map spanned 8181.2 cM in total length, with an average distance of 0.7 cM per marker.

QTL Analyses of Three Grain Related-Traits
A total of 39 additive effect QTLs were identified for TKW, GL and GW in the F6 HN5290/06Dn23 population with a 50 K SNP array by using IciMapping software. These In addition, the correlation coefficients of grain-related traits were extremely significant under four environments, while it was 0.78-0.87 for GL among different environments and it was also significantly correlated with TKW and GW, which were 0.42-0.71 and 0.21-0.57, respectively. Interestingly, TKW was significantly positively correlated with GW (r = 0.54-0.94), while the correlation between GL and GW was weak (r = 0.21-0.57) ( Table 3). The broad-sense heritability of TKW, GL and GW was 0.80, 0.89 and 0.75, respectively (Table 2), indicating that GL was more insensitive to the environmental variation than TKW and GW in the present study.

Linkage Map Construction
A total of 18,357 SNP loci were polymorphic between two parents. Among them, 4655 SNPs with a high missing rate (>10%) or low distortion p value (<0.001) were removed and 13,702 markers were used to construct linkage map for subsequent QTL mapping. Finally, 70 linkage groups consisting of 12,533 SNPs were established on 21 chromosomes (Supplementary Table S1). The entire linkage map spanned 8181.2 cM in total length, with an average distance of 0.7 cM per marker.

QTL Analyses of Three Grain Related-Traits
A total of 39 additive effect QTLs were identified for TKW, GL and GW in the F 6 HN5290/06Dn23 population with a 50 K SNP array by using IciMapping software. These QTLs were distributed on almost all 21 chromosomes with exceptions of chromosomes 1D, 3A, 7A and 7D. There were 12 QTLs corresponding to TKW, while 19 QTLs were for GL and 8 QTLs for GW (Supplementary Table S2). Among these loci, 17 loci showed stability in multiple environments and were considered as stable QTLs in below (Table 4 and Figure 2).

Co-Located Loci for All Three Traits
In this study, two co-located loci for the three yield related-traits were identified on chromosome 6AL and 4BS, respectively. These loci were derived from the large grain parent 06Dn23 (Table 4).
QTL QTkw.haaf-6AL/QGl.haaf-6AL/QGw.haaf-6AL was mapped on the long arm of chromosome 6A. It was the most stable locus with the largest phenotypic effect across all of four environments. This co-located locus explained 15.4-24.8% for TKW, 4.1-8.8% for GL and 15.7-24.4% for GW (Table 4). It was flanked by AX-111501610 and AX-110680682 markers and located on the physical position at 78.1 Mb-545.7 Mb based on the Chinese Spring wheat reference genome.
The second co-located locus was QTkw.haaf-4BS/Gl.haaf-4BS/QGw.haaf-4BS. It was mapped on wheat chromosome 4BS and flanked by AX-109389480 and AX-108850477 markers. This co-located locus explained 8.4-10.2%, 8.8% and 9.3-11.2% phonotypical variation for TKW, GL and GW, respectively (Table 4). It was stably detected in three and two environments for both TKW and GW, but the effect on GL was only detected in one environment. The physical position of this co-located locus was placed at 172.4-345.0 Mb based on the Chinese Spring reference genome (Table 4). −0.05 a Logarithm of odds (LOD) score of QTL peak. b Proportion of phenotypic variance explained by each QTL. c Additive effect for the QTL; positive values indicate the effect of HN5290 alleles, whereas negative values indicate that the alleles came from 06Dn23.

Co-Located Loci for Two Yield Related-Traits
There were 4 loci that were mapped for two yield related-traits among the three and they were identified on the chromosomes 2BL, 2DL (2DL.1 and 2DL.2) and 6DL, respectively. We mapped the effect on TKW at all the four loci, while GL effect was only identified on 2BL, 2DL (2DL.2) and 6DL and GW effect was on 2DL (2DL.1) chromosomes. These loci were also derived from the large grain parent 06dn23 ( Table 4). The QTkw.haaf-2BL/QGl.haaf-2BL was flanked by AX-86163179 and AX-110369359 SNP markers and it explained phenotypic variation of 8.5-13.5% for TWK and 9.7-17.8% for GL across all of the four environments (Table 4).

Co-Located Loci for Two Yield Related-Traits
There were 4 loci that were mapped for two yield related-traits among the three and they were identified on the chromosomes 2BL, 2DL (2DL.1 and 2DL.2) and 6DL, respectively. We mapped the effect on TKW at all the four loci, while GL effect was only identified on 2BL, 2DL (2DL.2) and 6DL and GW effect was on 2DL (2DL.1) chromosomes. These loci were also derived from the large grain parent 06dn23 ( Table 4). The QTkw.haaf-2BL/QGl.haaf-2BL was flanked by AX-86163179 and AX-110369359 SNP markers and it explained phenotypic variation of 8.5-13.5% for TWK and 9.7-17.8% for GL across all of the four environments (Table 4).

The QTL Only Mapped for One Trait Effect
We also mapped one stable locus for TKW, four loci for GL and one locus for GW in the present study and they were distributed on wheat chromosomes 1BS, 2AS, 2AL, 2DS and 5AL, respectively. All of these loci were contributed to by the big grain parent 06dn23 with the exceptions of QGl.haaf-1BS.1, QGl.haaf-2DS and QTkw.haaf-5AL from HN5290.
For the TKW, there was a stable QTL, QTkw.haaf-5AL, which was present over three environments. It was flanked by AX-86165895 and AX-110976589 SNP markers and explained 5.9-10.7% for TKW.
According to GL, three QTL QGl.haaf-2AL, QGl.haaf-2BL and QGl.haaf-2DL.2 were observed on the wheat homoeologous group 2 under all tested environments in the present study and their physical position was co-linear, indicating that there might present three homologous genes controlling GL on group 2 in 06Dn23. They explained 1.8-19.8% phenotype variation of GL under different tested environments. Interestingly, two GL-related loci were located on chromosome 1BS, QGl.haaf-1BS.1 and QGl.haaf-1BS.2, derived from HN5290 and 06Dn23, respectively. The genetic distance between the two QTLs was 4 cM and explained 4.3-6.8% and 6.9-11.3% of phenotypic variance, respectively (Table 4 and Figure 2). Besides, QGl.haaf-2DS was detected in over three environments and had minor effects on increasing GL, which can contribute to relatively low phenotypic variance ranging from 3.8 to 6.4% (Table 4).
Only one stable QTL associated with GW was discovered on the chromosome arm 2AS and it was detected in all four environments. It explained phenotypic variance ranging from 8.9 to 14.3%.

Factor ANOVA Analysis between Two Co-Located Loci for Three Traits
In this study, only QTkw.haaf-4BS/Gl.haaf-4BS/QGw.haaf-4BS and QTkw.haaf-6AL/QGl.haaf-6AL/QGw.haaf-6AL were shown to have a co-located effect on three traits. We did factor ANOVA analysis for the gene interaction between these two loci and found that the effect of the single locus was significant (p < 0.0001). However, the variation contributed to by the interaction between two loci was 0 (Table 5), and we did not find a significant interaction between them that might due to the additive effect between them. In addition, QTkw.haaf-4BS and QTkw.haaf-6AL contributed more variation on TKW than the other two traits, which explained 8.4-10.2% and 15.4-24.8%, respectively (Table 4). Table 5. Factor ANOVA analysis of two co-located loci for thousand kernel weight (TKW), grain length (GL) and grain width (GW) on the chromosomes 4BS and 6AL.

Phenotypic Variation Caused by Environments
With global warming and climate change, high temperature poses a great threat to further improve the yield potential for wheat [50][51][52]. Heat stress can reduce grain weight during the grain filling period [12,53,54], resulting in a significant difference for TKW, GL and GW between 2018 and 2019 in the present work that might due to the late planting resulting in the high temperature at the middle and late stage of grain filling in 2018 (Table 1; Figure 1; Supplementary Table S3). Interesting, more QTLs for GL and more stability across different environments were detected than in those of TKW and GW in the present study, indicating that GL was less influenced by environment conditions, as this is determined in the early stage of grain development [8]. In addition, according to Pearson correlation analysis (Table 3), the correlation between TKW and GW is the highest, indicating that GW can reflect grain weight better than GL. Therefore, TKW and GW, with higher coefficient of variation, were more susceptible to the environment in this work. Furthermore, QGw.haaf-2DL.1 and QGw.haaf-4BS were only identified in 2018, which might contribute to the heat tolerance for wheat yield improvement [10].

Comparison of Stable QTL for Grain-Related Traits with Previous Studies
As it is known, grain weight and size are essential components determining wheat yield. Many QTL or homology-based cloned genes for grain-related traits (TKW, GL and GW) have been discovered on all 21 chromosomes [44,55,56]. Based on the genetic analysis of the 194 RIL population of HN5290/06Dn23, 17 stable QTLs were found in chromosome arms 1BS, 2AS, 2AL, 2BL, 2DS, 2DL, 4BS, 5AL, 6AL and 6DL (Supplementary Table S2,  Table 4). To identify the relationship of these loci with the previously reported QTLs, the rough physical intervals of QTL regions were obtained by aligning the sequences of markers of the corresponding QTL with the Chinese Spring wheat reference genome [45].
TKW is one of the most direct indicators for wheat yield. Our study showed five stable QTLs for TKW (Table 4 and Figure 2). As expected, most QTLs associated with TKW coincided with GL and/or GW with an exception of the locus on chromosome 5AL, which was consistent with the common agreement that TKW was the core of other grain traits. According to the location of the latest marker AX-110680682, the physical location was at 413 Mb on chromosome 6A, which was located in the same position as the previously reported QTL region (QTKW-6A-AN and QTKW.caas-6AL) of TKW [23,37,44,57,58]. Likewise, TaTPP-6AL1 was 13-48 Mb away from these loci and might be the causal gene for this QTL region [44]. However, the effect of this locus on GW and GL has not been reported yet.
The effect of QTkw.haaf-4BS on TKW, GL and GW was smaller than the 6A locus. It was located in the physical region at the interval of 172-195 Mb, as this QTL affected TKW by controlling the grain size. There were two major genes for wheat yield, Rht-B1 and TB-B1, which have been reported on the chromosome 4BS [24,59,60]. However, they could be different from QTkw.haaf-4BS/QGl.haaf-4BS/QGw.haaf-4BS, at a physical distance of 140 Mb away based on the Chinese Spring reference genome [45]. Chen et al. delimited a QTL for TKW, GL and GW at the 245-432 Mb position of 4BS chromosome [20], which was similar to the QTkw.haaf-4BS/QGl.haaf-4BS/QGw.haaf-4BS in present study. Therefore, we speculated that the two QTLs might be the same locus.
QTkw.haaf-5AL, the unique locus related to TKW with its additive from Henong5290, was identified across three environments in the present study and it was almost located in the same position as QTkw-5A2 and QTkw.caas-5AL, as reported by Liu et al. [28,61], respectively. TaGL3-5A, a gene controlling grain length in wheat [62], was also located on chromosome 5AL, and was at least 35 Mb away from the closest marker of QTkw.haaf-5AL.
QTkw.haaf-2DL.2 was a novel and stable locus and its effect on TKW and collocation with QGl.haaf-2DL.2 related to grain length. Xie et al. [8] reported that grain length was determined in the early stage of grain development, while grain width was associated with grain filling in the middle and late stage. In present study, QTkw.haaf-2DL.2/ QGl.haaf-2DL.2 was not associated with grain width. So, this QTL might only be expressed in the early Plants 2021, 10, 713 9 of 14 stage of grain development. It was reported that some QTLs related to kernel number per spike and spike number per unit area were located in this region, which were QSnpp-2D.2, QKnps-2D.2, QKN-2D-AN, QTSS.sicua-2D.2 and KNS-IWB74, respectively [44]. These loci are probably controlled by the same gene or genes and further fine mapping is needed.
Grain size was an important part to improving grain yield, which provided space for grain filling [63]. QTL mapping revealed that 12 QTLs associated with grain size were dispersed on chromosomes 1B, 2A, 2B, 2D, 4B, 6A and 6D (Table 4) in present study. The previously reported QTLs or genes were at the similar positions to them. For instance, three stable loci for grain size identified across all environments were separately located on the chromosomes of homologous group 2. Among them, QGl.haaf-2AL corresponds to the locus detected in a genome-wide study by Li et al. [29], and within the interval of a cloned gene, TaFlo2-A1 for TKW [64]. Su et al. [34] and Chen et al. [20] mapped the QTLs of GL and TKW, respectively, which were located in the same interval of QGl.haaf-2BL. QGw.haaf-2DL, a stable major locus, was located the same position as qKW-2D.1 reported by Su et al. [34]. In addition, rice gene Flo2 played a crucial role in regulating grain size [65]. To dissect the relationship between Flo2 and the latter two loci (QGl.haaf-2BL and QGw.haaf-2DL), we queried Flo2 on the online website EnsemblPlants (http://plants.ensembl.org/index.html, (accessed on 1 October 2020)) and obtained the physical positions of orthologs of Flo2 in wheat (Supplementary Table S4). Noteworthily, the homologous genes TaFlo2-B1 and TaFlo2-D1 were also close to QGl.haaf-2BL and QGw.haaf-2DL, respectively. Furthermore, the expression of three orthologs of Flo2 was dynamic during the early spike and grain development of bread wheat [66][67][68]. Hence, we speculated that the orthologs of rice Flo2 controlling grain size are most likely to be the candidate genes for GL-or GW-corresponding QTLs on the chromosomes of wheat homologous group 2 based on the transcriptomic analysis (WheatOmics, http://202.194.139.32/expression/index.html, (accessed on 1 October 2020)). It was reported that WFZP-2D has significant effects on spike morphogenesis and yield components [44,69], which was located in the confidence interval of QGl.haaf-2DS.

Plant Materials
One hundred and ninety-four F 6 recombinant inbred lines (RILs) derived from a cross of HN5290 and 06Dn23 were used to do genetic analysis. HN5290, a half-winter wheat cultivar derived from a cross of Shaan 160 × Laizhou 95021, showed an excellent resistance to lodging and fungal disease (yellow rust and powdery mildew). While 06Dn23 showed a higher kernel weight, larger grain size and taller plant height than HN5290. The RIL population was constructed by using single seed descent.

Field Trials and Phenotype Evaluation
The RILs and parents were grown over two cropping seasons at three locations, viz. GaoYi (2018Y), Shijiazhuang (2018S and 2019S) and Xinxiang (2019X). Field trials were carried out around the 10th of October during the cropping season of 2018-2019, while planting was postponed on the 30th of October during the cropping season of 2017-2018 due to the continuous rainfall around the middle of October at Shijiazhuang city. The trials were carried out in a randomized complete block design with two replicates over all environments. Each plot consisted of two 1-m rows with 30 cm spacing.
We harvested the single row of each RIL around the middle of June and mixed threshing. The harvested seeds were dried at room temperature and we measured the TKW, GL and GW every year when the moisture content of the seeds was stabilized around 11-12%. The grain size related-traits were measured using an automatic seed analyzer (SC-G, Wanshen Detection Technology Co., Ltd., Hangzhou, China). GL and GW were calculated by image analysis software after scanning around 500 full-blown grains evenly dispersed on a seed plate, whereas TKW was recorded by weighing above grains using an electronic balance. Finally, the mean of the two replicates in each environment was used for subsequent QTL analysis.

DNA Extraction and Genotyping
Genomic DNA of parents and each RIL was isolated from their leaf using the modified cetyltrimethyl ammonium bromide (CTAB) method [70], and the quality of DNA was appraised using 1% agarose gel electrophoresis. Wheat 50 K SNP array contained over 50,000 SNPs evenly distributed on 21 chromosomes. Some of them were functional markers of cloned genes or closely linked markers to the major loci responsible for agronomic traits and disease resistance [29]. It was designed for breeding selection and developed by the Institute of Crop Science, Chinese Academy of Agricultural Sciences (CAAS). In the present study, the 50 K SNP array was used to genotype the two parents as well as the RIL population. The polymorphic SNPs between parents were used to construct the linkage maps.

Genetic Linkage Maps Construction and QTL Analysis
We selected the polymorphic markers between two parents from the raw SNP data and used BIN functionality of QTL IciMapping 4.2 software (Chinese Academy of Agricultural Sciences, Beijing, China) [71] in order to discard the markers with high missing rate (>10%) and low distortion p value (<0.001). Linkage maps were constructed by the same software with the nnTwoOpt method to determine the order of anchors. Based on the anchor results of the marker's sequence on wheat genome, chromosome location of the remaining linkage groups was identified.
QTL analysis was carried out following the BIP functionality of QTL IciMapping 4.2 software by the inclusive composite interval mapping with additive effect (ICIM-ADD) method. The mapping parameters were set as step = 1.0 cM, PIN = 0.001, and LOD threshold was 3.0 by manual input. QTLs were named according to McCouch's method [72].

Statistical Analysis
Fundamental data analyses, including descriptive statistics, phenotypic correlation analysis, analysis of variance (ANOVA), normality test and t-test, were conducted using SPSS 20.0 software (SPSS, Chicago, IL, USA). The formula H 2 = σ 2 g /(σ 2 g +σ 2 ge /m+σ 2 gy /n+σ 2 /nr) was used to estimate the broad-sense heritability. H 2 represents each trait with the Ime4 package in R 3.5 software. σ 2 g , σ 2 ge and σ 2 gy represent the variance of genotype, genotype × environment interaction and genotype × year, respectively, whereas m, n and r represent the number of environments (locations), cropping seasons and replicates, σ 2 is the residual error. The linkage map and LOD profile of QTLs were drawn using MapChart 2.32 software. SAS version 9.3 was used for ANOVA and correlation analysis with TKW data in each environment.

Conclusions
HN5290 was a new wheat variety bred by the Wheat Breeding Laboratory of Hebei Agricultural University. It was highly resistant to stripe rust and moderately resistant to leaf rust and was a high-yield and high-quality wheat variety. 06dn23 was a wheat line which has not been released in Hebei Province. It had large grain, high TKW and higher plant height than HN5290. The two parents had good agronomic characters and had big potential to be used in wheat breeding. Maintaining high and stable yield of wheat had always been the goal of breeders. The grain length of wheat played an important role in increasing TKW. The new locus QGl.haaf-2DL.2 in this study could significantly increase the grain length of wheat, and indirectly increase the yield of wheat. The tightly linked marker of QTkw.haaf-2DL.2/QGl.haaf-2DL.2 could be used in wheat molecular breeding.
Supplementary Materials: The following are available online at https://www.mdpi.com/article/ 10.3390/plants10040713/s1, Table S1: Descriptive statistics of 70 linkage groups across 21 wheat chromosomes, Table S2: QTL for thousand kernel weight (TKW), grain length (GL) and grain width (GW) over all tested environments by using Inclusive Composite Interval Mapping software, Table S3: Comparison of TGW, GL and GW between the RILs with 'HN5290' genotype and '06Dn23' genotype for two reliable QTL on chromosome 2BL and 6AL in the F 6 RILs population derived from HN5290×06Dn23, Table S4: Approximate physical positions of three QTL for grain size on homologous chromosome group 2 and three orthologs of rice Flo2 in wheat, Table S5: Highest and lowest temperature of one day during wheat grain filling period in different environments.

Data Availability Statement:
The data presented in this study are available on request from the corresponding author. The data are not publicly available due to the related research is under going for other traits.

Conflicts of Interest:
The authors declare no conflict of interest.