Variation in allelic frequencies at loci associated with kernel weight and their effects on kernel weight-related traits in winter wheat

Teng Li,Hongwei Liu,Chunyn Mi,Gungjun Yu,Huili Li,Lingzhi Meng,Dwei Jin,Li Yng,Yng Zhou,Hongjun Zhng,*,Hongjie Li,*

aNational Key Facility for Crop Gene Resources and Genetic Improvement,Institute of Crop Sciences,Chinese Academy of Agricultural Sciences,Beijing 100081,China

bXinxiang Innovation Center for Breeding Technology of Dwarf-Male-Sterile Wheat,Xinxiang 453731,Henan,China

cZhaoxian Experiment Station,Shijiazhuang Academy of Agricultural and Forestry Sciences,Zhaoxian 051530,Hebei,China

dInstitute of Agricultural Science,The Fourth Division of the Xinjiang Production and Construction Corps,Yining 835000,Xinjiang,China

ABSTRACT Article history:Received 12 March 2018 Received in revised form 8 August 2018 Accepted 11 August 2018 Available online 28 September 2018

Keywords:Allelic frequency Allelic variation Functional markers Triticum aestivum Knowledge of allelic frequencies at loci associated with kernel weight and effects on kernel weight-related traits is crucial for yield improvement in wheat.Kernel weight-related traits were evaluated in 200 Chinese winter wheat cultivars(lines)grown at the Xinxiang Experimental Station,Chinese Academy of Agricultural Sciences,Xinxiang in Henan Province,for three consecutive years from 2014 to 2016.Alleles associated with kernel weight at nine loci,TaCKX6-D1,TaCwi-A1,TaCWI-4A,TaGS1a,TaGS5-A1,TaGS3-3A,TaGW2-6A,TaSus2-2B,and TaTGW6-A1,were determined for all cultivars(lines).ANOVA showed that genotypes,years and their interactions had significant effects on thousand-kernel weight(TKW),kernel length(KL)and kernel width(KW).The overall mean frequencies of alleles conferring high and low TKW at the nine loci were 65.9%and 33.4%,with the ranges of 37.0%-85.0%and 13.5%-63.0%for single loci.The frequencies of high-TKW alleles were over 50.0%at eight of the loci.Genotypes at each locus with the high-TKW allele had higher TKW than those with the low-TKW allele.The high-TKW allele Hap-H at the TaSus2-2B locus can be preferably used to increase grain yield due to its high TKW(49.32 g).A total of 18 main allelic combinations(ACs)at nine loci were detected.Three ACs(AC1-AC3)had significantly higher TKW than AC6 with high-TKW alleles at all nine loci even though they contained some low-TKW alleles.This indicated that other loci controlling kernel weight were present in the high-TKW cultivars.This work provides important information for parental selection and marker-assisted selection for breeding.

1.Introduction

World wheat(Triticum aestivum L.)production in 2017 was approximately735 million tonnes(FAOSTAT data,http://www.faolorg/).According to FAO estimates,an increase of 70%in production will be required by 2050 to protect food security for an increasing global population[1].Improvement in grain yield is the primary aim of most wheat breeding programs.Grain yield of wheat is determined by spike number,kernel number per spike and kernel weight[2].Kernel weight,as the component with highest heritability,contributes about 20%of the genetic variation in wheat yield[3].During the past six decades,increased thousand-kernel weight(TKW)has been a major objective of many Chinese wheat breeding programs.Zhou et al.[4]reported that TKW had been significantly increased with an annual genetic gain of 1.3%in the Northern China Plain over the four decades from 1960 to 2000,and 0.87%in the Southern China Winter Wheat Zone during the period of 1949-2000[5].Therefore,increasing kernel weight is an important way to improve wheat yield.

Kernel weight is tightly associated with the kernel dimensions(i.e.,width,length and thickness)[6]and isconditioned by polygenes[7-15].Although a large number of QTL associated with kernel weight have been identified,most of them were not used for marker-assisted selection(MAS)in wheat breeding programs.Using comparative genomics approaches[16],the main genes conferring kernel weight in common wheat,including cytokinin oxidase/dehydrogenase(TaCKX6-D1),cell wall invertase(TaCwi-A1 and TaCWI-4A),glutamine synthetase(TaGS1a),grain size(TaGS5-A1 and TaGS5-3A),grain width(TaGW2-6A),sucrose synthase(TaSus2-2B),and IAA-glucose hydrolase(TaTGW6-A1),have been cloned and corresponding functional markers have been developed[7-15].These functional markers are useful in breeding for high yield by MAS.

A principal gene affecting grain yield in rice(Oryza sativa L.)produces cytokinin oxidase/dehydrogenase(CKX)[17].Wheat ortholog TaCKX6-D1 was isolated and an inDel marker was developed for identifying the TaCKX6-D1a allele for high TKW and the TaCKX6-D1b allele for low TKW[7].Cell wall invertase(CWI)plays a decisive role in sink development and carbon partitioning,and has a close association with kernel weight[8].Two orthologous genes,TaCwi-A1 on chromosome 2A and TaCWI-4A on chromosome 4A,were cloned from common wheat.Association analysis indicated that the alleles TaCwi-A1a at the TaCwi-A1 locus and Hap-4A-T at the TaCWI-4A locus had positive effects on TKW in Chinese wheat cultivars[8,9].At the TaGS1 alocus,haplotypehap 2 was significantly associated with higher TKW and wider kernels[10].Gene OsGS5,which codes a putative serine carboxy peptidase,affects cell division and can lead to larger grain size in rice[18].Two favorable alleles TaGS5-A1b and TaGS5-3A-T at the TaGS5 locus conferred higher TKW and wider KW than the alternative alleles TaGS5-A1a and TaGS5-3A-G in wheat[11,12].Haplotypes Hap-6A-A and Hap-6A-G at the TaGW2-6A locus were distinguished by a cleaved amplified polymorphic sequence(CAPS)marker.The Hap-6A-A allele was associated with higher TKW and wider kernels[13].Two alleles,Hap-H and Hap-L,were identified at the TaSus2-2B locus;Hap-H had a positive effect on TKW in Chinese wheat cultivars[14].TaTGW6-A1 was tightly associated with TKW in a set of 242 wheat cultivars,and the TaTGW6-A1a allele conferred an increased TKW[15].

Although the genes conferring kernel weight have been cloned and functional markers have been developed,the frequencies of the alleles and their effects on kernel weight related traits in winter wheat cultivars and breeding lines have not been critically analyzed.The aims of the present study were to 1)determine the allelic frequencies at nine loci associated with kernel weight in a panel of widely grown past and present winter wheat cultivars and breeding lines,and 2)compare the mean phenotypic effects of alleles at each locus and of allelic combinations at the nine loci on kernel weight-related traits.

2.Materials and methods

2.1.Plant materials

Two hundred winter wheat accessions,including current and past,widely grown cultivars and breeding lines were examined in this study;they included 11 cultivars(lines)from the Northern Winter Wheat Zone(Beijing,Shanxiand Ningxia),170 from the Yellow and Huai River Valley Winter Wheat Zone(Anhui,Hebei,Henan,Shaanxi,Shandong and Jiangsu provinces,and Italy),5 from the Middle and Low Yangtze River Valley Winter Wheat Zone(Jiangsu)and 14 from the Southwestern Winter Wheat Zone(Sichuan).Detailed information for each accession is provided in Table S1.

2.2.Field trials

The materials for phenotypic evaluation were grown at the Chinese Academy of Agricultural Sciences(CAAS),Xinxiang Experimental Station(35°31′N,113°85′E)in Henan province,representing the southern part of the Yellow and Huai River Valley Winter Wheat Zone.All genotypes were arranged usinga 20×10 lattice design with two replications for each year.Each plot consisted of two 2 m rows spaced 0.2 m apart,with 40 plants per row.Field trials were sown on Oct.10,2014,Oct.2,2015and Oct.4,2016.Prior to sowing in each year,191 kg ha-1of ammonium phosphate(N-P2O5,12%-42%),200 kg ha-1of urea(N,42%)and 41 kg ha-1of potassium chloride(K2O,60%)were applied.Additional urea was applied at 104 kg ha-1at the shooting stage(Zadoks growth stage(GS)31)[19].Irrigations were carried out at the tillering(GS 25),shooting(GS 31)and grain filling(GS 71)stages.Field management followed local agricultural practices.All plots were hand-harvested and kernels were naturally dried to a 13%moisture content.Kernel weight was obtained by averaging the values measured from three samples of 500 kernels and converted to TKW(g).Twenty randomly sampled kernels were used to measure kernel length(KL,mm)and kernel width(KW,mm)using a vernier gauge.

2.3.Analysis of genotypes for loci associated with kernel weight

DNA was extracted from each genotype by the CTAB method[20].The nine loci,TaCKX6-D1,TaCwi-A1,TaCWI-4A,TaGS1a,TaGS5-A1,TaGS3-3A,TaGW2-6A,TaSus2-2B,and TaTGW6-A1,associated with kernel weight were genotyped.Alleles at each locus were identified using gene-specific functional markers[7-15].Detailed information for each locus is provided in Table S2.All PCR primers were synthesized by the Shanghai Invitrogen Biological Technology&Services Co.Ltd.,Shanghai.The restriction endonucleases were obtained from New England Biolabs Ltd.,Beijing(www.net-china.com).

DNA amplifications were performed in reaction volumes of 20 μL,and comprised 1 μL of 50-100 ng μL-1DNA,1 μL 10 μmol L-1of each primer,10 μL 2 × Taq PCR Master Mix(Sangon Biotech Col.Ltd.,Shanghai),and 7 μL of sterilized ddH2O.The thermo-cycling program was:initial denaturation at 94 °C for 5 min,followed by 35 cycles at 94 °C for 30 s,annealing at 56-60 °C for 30 s,extension at 72 °C for 1-3 min,and a final extension step at 72°C for 10 min.PCR products were separated by electrophoresis in 1%-2%agarose gels at 120 V for 40 min and stained with ethidium bromide,or in 10%polyacrylamide gels and stained with GeneFinder(Bio-V,Xiamen,Fujian).

2.4.Statistical analysis

The META-R(Multi Environment Trail Analysis with R for Windows)software(version 6.03)was used for the best linear unbiased estimators(BLUE)for kernel weight-related traits[21].The BLUE values were used in subsequent analyses.Analysis of variance(ANOVA)for each trait was performed using PROC GLM in the Statistical Analysis System(SAS Institute,2000).Significant differences in phenotype between alleles at each locus for each trait were determined by t-tests.Broad-sense heritabilities(h2)of all traits were estimated by:where,,andare mean squares for genotype,genotype×year interaction and residual error,and y and r represent the numbers of years and replications,respectively[22].Linear regression was used to determine correlations between kernel weight-related traits and between years for the same trait based on the regression equation y=a+bx,where y is the dependent variable,a is the regression intercept,and b is the regression coefficient.

3.Results

3.1.Phenotypic analyses of kernel weight-related traits

ANOVA showed that genotypes,years and their interactions had significant effects on TKW,KL,and KW(Table 1).The mean values of TKW,KL,and KW for the 200 genotypes over the 3 years were 48.60 g,6.93 mm,and 3.59 mm,respectively(Table 2).There was wide variation in all three traits,with a range of 37.24-57.40 g,6.01-7.68 mm,and 3.30-3.97 mm for TKW,KL and KM,respectively(Table 2).These traits exhibited high broad-sense heritabilities(h2),of 0.95 for TKW,0.96 for KL and 0.89 for KW(Table 2).In all 3 years,TKW was significantly correlated with KL and KW(Fig.1-B),whereas no significant correlation was observed between KL and KW(Fig.1-C).Positive correlations(P≤0.01)between years for the same trait were observed for TKW(Fig.1-D),KL(Fig.1-E),and KW(Fig.1-F).Based on the positive correlations between years,the mean values for each trait across the 3 years were used in all subsequent analyses.

3.2.Frequency distributions of alleles in 200 winter wheat accessions

The alleles at each locus were identified using genespecific functional markers;amplification profiles are shown in Fig.2.Detailed information on alleles(genotypes)for each accession is given in Table S3.Heterogeneous genotypes were treated as missing data and allelic frequencies of only homozygous genotypes were used in estimation of allelic frequencies.The mean frequencies of alleles conferring high and low TKW across all nine loci were 65.9%and 33.4%,with ranges of 37.0%(Hap-4A-T)-85.0%(TaTGW6-A1a)and 13.5%(TaTGW6-A1b)-63.0%(Hap-4A-C)at single loci,respectively(Table 3).The frequencies of alleles associated with high TKW were higher than those associated with low TKW,except for the TaCWI-4A locus(Table 3).Among them,the frequencies of alleles TaCKX6-D1a,TaCwi-A1a,Hap2,TaGS5-A1b,TaGS5-3A-T,Hap-6A-A,and TaTGW6-A1a were over 60%.

3.3.Effects of contrasting alleles at each locus on kernel weight-related traits

Genotypes with high-TKW alleles had higher TKW than those with low-TKW alleles at each of the nine loci(Table 4).Genotypes with the TaCKX6-D1a,TaCwi-A1a,TaGS5-A1b,TaGS5-3A-T,and Hap-H alleles had significantly higher TKW than those with corresponding low-TKW alleles at TaCKX6-D1,TaCwi-A1,TaGS5-A1,TaGS5-3A,and TaSus2-2B loci,respectively.Compared to the genotypes with low-TKW alleles,those with high-TKW alleles had longer KL at TaCwi-A1,TaCWI-4A,and TaSus2-2B loci,and wider KW at all nine loci except TaGS1a.Among them,genotypes with the TaCwi-A1a allele(6.96mm)had significantly longer KL than those with the TaCwi-A1b allele(6.87mm).Loci with alleles conferring higher TKW and significantly wider KW included TaCKX6-D1,TaCWI-4A,TaGS5-A1,TaGS5-3A,TaSus2-2B,and TaTGW6-A1.

Table 2-Means,standard deviations(SD),ranges and broad-sense heritabilities(h2)for kernel weight-related traits in 200 winter wheat accessions.

Fig.1-Correlation analyses of kernel weight-related traits in each year(A-C)and between years(D-E).TKW,thousand-kernel weight;KL,kernel length;KW,kernel width.*and**,significant at P≤0.05 and P≤0.01,respectively.

Fig.2-Amplification profiles of alleles at the TaCKX6-D1,TaCwi-A1,TaCWI-4A,TaGS1a,TaGS5-A1,TaGS5-3A,TaGW2-6A,TaSus2-6B,and TaTGW6-A1 loci in selected accessions.M,marker;1,Jimai 19;2,Jimai 20;3,Jimai 21;4,Jimai 22;5,Jinan 17;6,Liangxing 99;7,Lumai 7.

Table 3-Allelic frequencies at loci associated with thousand-kernel weight(TKW)among 200 winter wheat accessions.

3.4.Effects of allelic combinations at nine loci on kernel weight related traits

Eighteen main allelic combinations(ACs)involving all nine loci were detected(Fig.3),and the detailed information for AC classes is shown in Table S3.There were significant phenotypic differences among ACs for three kernel weight-related traits.Allelic combination 6(AC6),including genotypes with high-TKW alleles at all nine loci,had higher TKW(50.42 g)than AC7-AC18(with TKW values ranging from 44.00 to49.93 g),and had significantly higher TKW than AC13-AC18(with the TKW values ranging from 44.00 to 48.34 g).On the other hand,the AC6 genotypes also had longer KL than other AC types except AC2 and AC3.AC1-AC5 had higher TKW than AC6,and AC1-AC3 had significantly higher TKW and KL or KW.Genotypes at the TaGS5-A1 and TaGS5-3A loci gave similar data for the TKW-related traits.

Table 4-Comparison of thousand-kernel weights(TKW),kernel lengths(KL),and kernel weights(KW)between contrasting alleles at each locus in the 200 winter wheat accessions based on mean values for 3 years.

Fig.3-Allelic combinations(ACs)at nine loci associated with kernel traits and mean values for thousand-kernel weight(TKW),kernel length(KL),and kernel width(KW)over 3 years.Black and white boxes represent the alleles associated with high and low TKW,respectively.Different letters after means indicate significant differences(P≤0.05)in the kernel weight-related traits among ACs.

4.Discussion

China is the largest wheat producing country in the world,and wheat yields have steadily increased with an annual genetic gain of 1.0%during the last 65 years[23].Among the three major yield components,kernel weight and spike weight have stably increased over the past several decades[24].Phenotypic evaluation of 1800 commercial cultivars demonstrated that TKW increased from a mean 31.5 g in the 1940s to 44.64 g in the 2000s,representing a 2.19 g genetic gain on a decade basis[25].Therefore,increasing TKW is an important strategy for the improvement of wheat yield.Correlation analysis showed that KW(P≤0.01)and KL(P≤0.05)were positively correlated with TKW,and correlation coefficients between KW and TKW were higher than those between KL and TKW,demonstrating that KW had a more important role in affecting TKW than KL.However,TKW is easier to measure than KL and KW especially when a breeding program involves large number of lines.

Functional markers are derived from known functional alleles[26].They are ideal for MAS because they are fully diagnostic of the target allele[27].To date,more than 100 functional markers associated with genes for processing quality,agronomic traits and disease resistance have been developed in wheat[26],and many of them have been converted to Kompetitive Allele Specific PCR(KASP)markers[28].In this study,nine loci associated with kernel weight were detected using functional markers.Among them,variations at the TaGS5-A1 and TaGS5-3A loci present almost identical genotypic data and phenotypic effects(Tables 4 and S3,and Fig.3).Both loci are orthologs of rice gene GS5 and located on chromosome 3A[11,12].They represent the same differently named locus,identified by different diagnostic markers from different studies.

In the genotypes examined,TKW ranged from 38.24 to 57.40 g(Table 2),and the highest TKW for genotypes with the high-TKW allele at a single locus and high-TKW allelic combination at nine loci was 49.32 and 50.42 g,respectively.Three ACs(AC1-AC3)had significantly higher TKW than AC6 with high-TKW alleles at all nine loci examined even though those ACs carried some low-TKW alleles(Fig.3).Thus the loci included in this study accounted for only part the genetic variation in TKW.Other loci associated with TKW must be present in high-TKW cultivars such as Zhumai 4(57.40 g),Zhengyumai 9987(57.29 g),and Zhongmai 875(56.33 g).New mapping populations need to be developed from these cultivars for identification of additional QTL associated with TKW.

Alleles associated with high TKW had higher distribution frequencies than those with low TKW except for TaCWI-4A,indicating that artificial selection for those allele shad occurred in the past[29].The variable frequencies of high-TKW alleles among loci indicated that those loci had undergone different selection pressures.Furthermore,the high-TKW alleles at most loci had higher frequencies in wheat accessions from the southern wheat zones than the northern zones(data not shown).This was in agreement with Chinese wheat breeding strategies requiring higher TKW in cultivars grown in southern areas,in contrast to greater emphasis on spike numbers in northern production regions[23].

Genotypes with high-TKW alleles had higher TKW than those with low-TKW alleles at each locus,in agreement with previous studies[7-15].However,the extent of actual phenotypic differences in TKW between two alleles at each locus varied in different genotypes.For example,the phenotypic difference in TKW between the Hap2 and Hap1 alleles at the TaGS1a locus was 4.10 g in a Chuan 35050×Shannong 483 recombinant inbred line(RIL)population[10],but in the present study the corresponding value was 0.11 g.Similarly,the phenotypic difference in TKW between Hap-4A-T and Hap-4A-C was 4.90 g in a previous study[28],but only 0.45 g in this study.This infers that the discrepancies in the phenotypic performances in different studies might be caused by differences in genetic backgrounds,especially when data from biparental cross populations and more unrelated germplasm sets are compared.

In conclusion,the frequencies of high-TKW alleles were greater than 50%at eight of the nine loci examined as revealed by analysis of functional markers for target alleles for wheat kernel weight-associated traits.The high-TKW allele Hap-H at the TaSus2-2B locus can be used preferentially in increasing wheat yield due to its high TKW(49.32 g)and relatively low frequency(52.5%)in current Chinese wheat accessions.Three ACs(AC1-AC3)had significantly higher TKW than AC6 which carried high-TKW alleles at all nine loci even though those three ACs contained some low-TKW alleles.Thus further study should detect variation at additional loci affecting TKW among current high-TKW genotypes.

Supplementary data for this article can be found online at https://doi.org/10.1016/j.cj.2018.08.002.

Acknowledgments

This study was supported by the National Key Research and Development Program of China (2017YFD0101000,2016YFD0101004),the National Natural Science Foundation of China(31771881,31401468),and the CAAS Innovation Team and the National Engineering Laboratory of Crop Molecular Breeding.