Fine mapping and validation of a stable QTL for thousand-kernel weight in wheat (Triticum aestivum L.)

Dyun Mng,Amn Btool,Yzhou Xun,Ruiqing Pn,N Zhng,Wi Zhng,Liy Zhi,g,Xioli Rn,Wnqing Li,Jiji Li,Ynxio Niu,Shuzhi Zhng,f,Jun Ji,,Xioli Shi,Li Wng,,f,Hongqing Ling,g,Chunhu Zho,F Cui,,Xigng Liu,f,Junming Li,,f,,Liqing Song

a Center for Agricultural Resources Research,Institute of Genetics and Developmental Biology,The Innovative Academy of Seed Design,Chinese Academy of Sciences,Shijiazhuang 050022,Hebei,China

b Ministry of Education Key Laboratory of Molecular and Cellular Biology,Hebei Research Center of the Basic Discipline of Cell Biology,Hebei Key Laboratory of Molecular and Cellular Biology,College of Life Sciences,Hebei Normal University,Shijiazhuang 050024,Hebei,China

c State Key Laboratory of North China Crop Improvement and Regulation,College of Agronomy,Hebei Agricultural University,Baoding 071000,Hebei,China

d State Key Laboratory of Plant Cell and Chromosome Engineering,Chinese Academy of Sciences,Beijing 100101,China

e Key Laboratory of Molecular Module-Based Breeding of High Yield and Abiotic Resistant Plants in Universities of Shandong,School of Agriculture,Ludong University,Yantai 264025,Shandong,China

f Hebei Collaboration Innovation Center for Cell Signaling,Shijiazhuang 050024,Hebei,China

g College of Life Science,University of Chinese Academy of Sciences,Beijing 100049,China

Keywords: Wheat Thousand-kernel weight Fine mapping Candidate gene

ABSTRACT Thousand-kernel weight (TKW) is a measure of grain weight,a target of wheat breeding.The object of this study was to fine-map a stable quantitative trait loci (QTL) for TKW and identify its candidate gene in a recombinant inbred line (RIL) population derived from the cross of Kenong 9204 (KN9204) and Jing 411(J411).On a high-density genetic linkage map,24,26 and 25 QTL were associated with TKW,kernel length (KL),and kernel width (KW),respectively.A major and stable QTL, QTkw-2D,was mapped to an 8.3 cM interval on chromosome arm 2DL.By saturation of polymorphic markers in its target region,QTkw-2D was confined to a 9.13 Mb physical interval using a secondary mapping population derived from a residually heterozygous line (F6:7).This interval was further narrowed to 2.52 Mb using QTkw-2D near-isogenic lines (NILs).NILsKN9204 had higher fresh and dry weights than NILsJ411 at various grain-filling stages.The TKW and KW of NILsKN9204 were much higher than those of NILsJ411 in field trials.By comparison of both DNA sequence and expression between KN9204 and J411,TraesCS2D02G460300.1(TraesKN2D01HG49350) was assigned as a candidate gene for QTkw-2D.This was confirmed by RNA sequencing (RNA-seq) of QTkw-2D NILs.These results provide the basis of map-based cloning of QTkw-2D,and DNA markers linked to the candidate gene may be used in marker-assisted selection.

1.Introduction

Bread wheat(Triticum aestivumL.)provides 20%of human dietary calories (https://www.fao.org/faostat/en).Genetic improvement of yield potential remains the primary goal of wheat breeding programs [1,2].Spike number,kernel number per spike(KNPS),and thousand-kernel weight (TKW) are the three components of wheat total yield[3,4],of which TKW has high heritability[5].Exploitation of genetic variation in TKW and related traits is a promising approach to improving wheat yield potential.

Many quantitative trait loci (QTL) analyses have been conducted in wheat for kernel weight and size,based on linkage[4,6-9] and genome-wide association studies (GWAS) [10-14].Numerous QTL for TKW,kernel length (KL),and kernel width(KW) have been identified and characterized [6,9,15,16].Major QTL associated with kernel weight and size have most frequently been found on groups 2 and 5 homoeologous chromosomes[6,8,16-19].

With the availability of high-throughput single-nucleotide polymorphism (SNP) arrays [20-22] and high-quality reference genome sequences[23,24],Kompetitive Allele-Specific PCR(KASP)markers with high accuracy and specificity and low cost have been developed for the fine mapping of QTL for kernel traits.Kernel traits are complex quantitative traits controlled by multiple genes or QTL [8,10].To exclude interference from background QTL without extensive backcrossing,residual heterozygosity and/or nearisogenic lines (NILs) have been used as secondary populations for fine mapping.Using the above stocks and molecular platforms,many major QTL for kernel weight have been identified and finemapped [17,19].A main-effect QTL for grain yield was identified on chromosome 1B and further fine mapped into a genetic interval of 2.9 cM,which corresponded to a 2.2 Mb genomic region[25].On chromosome 5D,the QTLQTKW.caas-5DLwas localized to a 3.9 Mb physical interval and a candidate gene was predicted [26].However,the complexity of the wheat genome impedes validation of causative genes [27].Except forTasg-D1[28] andTaGL1-B1[29],no QTL for wheat kernel weight and size has been characterized by map-based cloning [18].Therefore,it is of great implication to identify genes influencing kernel trait by this approach.

The availability of genome sequences of commercial wheat cultivars [30,31] and multiple sequencing datasets [28-32] provides an entry points for positional cloning of kernel morphological genes in wheat.Simplified genome-sequencing technologies such as RNA sequencing (RNA-seq) associate kernel morphology and gene expression via differentially expressed gene (DEG) analysis[33].Resequencing revealed that the candidate geneTasg-D1contained a SNP in its coding sequence that caused an amino acid substitution from Lys (286K) to Glu (286E) [34].InQTgw.caas-5Bhomozygous lines,the candidate geneTraesCS5B02G044800showed high expression in the 5B-genotype,whereas the corresponding transcript was not detected in the 5B+genotype [19].

In our previous study [35],many QTL for kernel characteristics were identified in a RIL population(KJ-RIL) derived from the cross of twoT.aestivumcultivars,Kenong 9204 (KN9204) and Jing 411(J411),based on a genetic linkage map constructed mainly with SSR and DArT markers.The main object of this study was to finemap a major and stable QTL,QTkw-2D,for TKW by incorporating new SSR/KASP markers and constructing residually heterozygous lines (RHLs)/NILs;and predict its candidate genes from sequence differences and transcriptome sequencing of parent lines andQTkw-2DNILs.Our results offer a starting point for map-based cloning and functional characterization ofQTkw-2D.

2.Materials and methods

2.1.Plant materials and field trials

A set of 188 KJ-RILs derived from the cross between KN9204 and J411 [36] were used for identifying QTL-trait associations.A genetic map of chromosome 2D was also constructed with this RIL population for remapping the TKW QTLQTkw-2D.A residually heterozygous line (RHL),RIL048,was selected using markers closely linked toQTkw-2D(Table S1).RIL048 self-progeny formed a secondary population for QTL fine mapping.Homozygous KN9204 and J411 plants were designatedQTkw-2DKN9204andQTkw-2DJ411,respectively.A total of 1161 homozygous plants(F7:8)composed of 366QTkw-2DKN9204and 795QTkw-2DJ411plants were planted in the 2016-2017 cropping season for evaluating the effects ofQTkw-2D.Heterozygous F7:8plants after marker selection were planted in the next two years and six types of heterozygous recombinants (F8:9) and seven types of homozygous lines (F9:10)were developed from them.Seven near-isogenic line(NIL)populations (NIL1 to NIL7) were developed from the latter for fine mapping ofQTkw-2D.Phenotyping data for QTL analysis were collected from the 2018-2019 to 2021-2022 cropping seasons.NIL1 and NIL7 were also selected for recording the fresh and dry weights of kernels at several stages of development in the 2018-2019 season and to conduct RNA sequencing in the 2019-2020 season.

Field trials were performed at the Luancheng Agro-Ecosystem Experimental Station,Chinese Academy of Sciences (37°53′N,114°41′E,54 m above sea level)in Hebei province from September 2016 to June 2022.A randomized complete block design with three replications was used.Each plot contained three rows 5.0 m long with an inter-row spacing of 25 cm and 65 seeds evenly sown in each row.The plants were managed under standard irrigation and fertilization practices following Cui et al.[36] and Fan et al.[37].

2.2.QTL analysis

On the basis of a high-density genetic linkage map [20] and phenotypic data of the KJ-RIL population in eight environments[36],QTL detection of TKW,KL,and KW was performed using the inclusive composite interval mapping(ICIM)method of IciMapping 4.0 [38] based on stepwise regression.The walking speed was 0.1 cM,and an empirical logarithm of odds (LOD) threshold was calculated based on 1000 permutation tests.A QTL with a mean LOD score ＞ 3 and mean phenotypic variation explained(PVE) ＞10% was defined as a major QTL,and one showing significance in at least four environments,was defined as a stable QTL.QTL were named following Fan et al.[37].

2.3.Genotyping

To reconstruct a new chromosome 2D genetic map,17 SSR markers developed in the Chinese Spring 2D chromosome (IWGSC RefSeq v1.0) [23] spanning theQTkw-2Dphysical region (419.06-595.43 Mb) were used to genotype the 188 KJ-RILs.Six SSR markers were developed in the interval 596.78-618.13 Mb of Chinese Spring chromosome 2D for genotyping the secondary mapping population together with the 11 SSR markers in the 561.05-595.43 Mb interval(Table S1).Primers were developed using DANMAN 7.0 (lynnon Biosoft Corporation,San Ramon,CA,USA) and synthesized by Sangon Bioengineering Co.,Ltd.,Shanghai,China.Genomic DNA was extracted from young leaf tissue of each line by the cetyl trimethyl ammonium bromide (CTAB) method [39].The PCR reaction mixture and touchdown PCR amplification program followed Hao et al.[40].Amplification products were separated by polyacrylamide gel electrophoresis following Singh et al.[41].

For fine mapping,12 KASP markers (Table S2) in the region of 561.69-570.14 Mb on chromosome 2D of Chinese Spring were developed from polymorphic SNPs between the parents on the basis of the KN9204 genome assembly database and the resequencing results of J411 [31].The primers were synthesized by LGC,Biosearch Technologies (Hoddesdon,UK),and KASP assays were performed following the LGC protocol.

2.4.Phenotyping

The phenotypic data of the KJ-RIL population were described by Cui et al.[36].QTkw-2DKN9204,QTkw-2DJ411,NIL1 and NIL7 were phenotyped for the following traits: plant height (PH),spikes per plant(SPP),spike length(SL),spikelet number(SN),kernel number per spike (KNPS),TKW,KL,and KW.NIL2-NIL6 were phenotyped for TKW.At the physiological maturity stage,10 representative plants in the central part of the middle row per plot were selected for recording their agronomic traits.TKW,KL,and KW were evaluated after harvest with an SC-G multifunctional seed analyzer(Wanshen Detection Technology Co.,Ltd.,Hangzhou,Zhejiang,China).

2.5.Reconstruction of the genetic map of the 2D chromosome and QTkw-2D remapping

Using JoinMap 4.0 software(Kyazma Corporation,Wageningen,Netherlands),a high-density genetic map of 218 markers was reconstructed by removal of markers that were not physically located on chromosome 2D in the genetic map of Cui et al.[20]and by incorporating 17 newly designed SSR markers in the physical interval 419.06-595.43 Mb of the Chinese Spring chromosome 2D (Table S1).The best linear unbiased prediction (BLUP) of TKW for each KJ-RIL was calculated across the eight environments using the ‘‘lmer” function implemented in the R package lme4 (v3.6.1)(https://www.r-project.org/).The same phenotypic data of the KJRIL population [36] and BLUP values were used for the highdensity mapping ofQTkw-2D.Genetic maps were drawn with Map-Chart 2.2 (https://www.biometris.nl/uk/Software/MapChart/).

2.6.Fine mapping of QTkw-2D

The secondary mapping population was genotyped with the 17 newly developed SSR markers(561.05-618.13 Mb)(Table S1),and the physical bin containingQTkw-2Dwas confirmed based on the phenotypic data of the kernel traits recorded in the 2016-2017 growing season.

Using the 12 KASP markers (Table S2),QTkw-2DNILs were obtained by genotyping the secondary mapping population.Using the TKW of NILs that were evaluated across four cropping seasons(2018-2019,2019-2020,2020-2021,and 2021-2022),QTkw-2Dwas fine-mapped.

2.7.Grain-filling rate measurement

NIL1 and NIL7 were used to measure the fresh and dry weights of developing kernels in the 2018-2019 season.Five spikes on the main stem of each NIL were tagged for sampling.Twenty kernels from the outer (primary and secondary) florets of spikelets in the middle of each tagged spike were sampled from 9:00-11:00 am at six developmental time points: 5,10,15,20,25,and 30 days post-anthesis (DPA).After the 100-kernel fresh weights were recorded,dry weights were recorded following dehydration of the grain at 105 °C for 10 min and then at 70 °C until constant weight.

2.8.Prediction of QTkw-2D candidate genes

The sequence of theQTkw-2Dphysical interval was compared with the KN9204 genome assembly and the resequencing data of J411 [31].KN9204 and J411 were grown under differing nitrogen fertilization and sampled for RNA-seq analysis,which were described in detail by Shi et al.[31].

NIL1 and NIL7 identified in this study were also used to discover DEGs in theQTkw-2Dphysical interval via RNA-seq analysis.Twenty seeds from the outer florets of spikelets in the main spike of each plant were sampled at 7,14,21,and 28 DPA.Samples were collected in triplicate,frozen in liquid nitrogen,and sent to Shanghai Majorbio Biopharm Technology Co.,Ltd.(Shanghai,China) for cDNA library construction and sequencing on the NovaSeq 6000 platform (Illumina) using paired-end sequencing.The raw RNAseq data were deposited in the Science Data Bank (DOI: 10.57760/ sciencedb.07209).The raw sequence reads were filtered with the FASTQ_Quality_Filter tool from the FASTX-toolkit(https://hannonlab.cshl.edu/fastx_toolkit/z) to obtain clean reads.The cleaned reads were further mapped to the reference genome of Chinese Spring (https://plants.ensembl.org/Triticum_aestivum/Info/Index/) by HISAT2 software (https://daehwankimlab.github.io/hisat2/).To compare gene expression profiles across growth stages,the transcript levels of individual transcripts in each sample were normalized as fragments per kilobase of transcript per million mapped reads (FPKM) by RSEM software (https://deweylab.github.io/RSEM/).DEGs in developing seeds were identified by comparing the NIL pairs at different stages (NIL7S7 vs.NIL1S7,NIL7S14 vs.NIL1S14,NIL7S21 vs.NIL1S21,and NIL7S28 vs.NIL1S28) using DESeq2 (https://bioconductor.org/packages/stats/bioc/DESeq2/)with default parametersP-adjust ＜0.05 and |log2FC| ≥1.

2.9.Statistical analysis

Statistical analyses were based on the phenotypic data described above.The significant differences were identified in TKW,KL,KW,PH,SPP,SL,SN,and KNPS amongQTkw-2Dlines by one-way analysis of variance of Tukey’s test using SPSS Statistics 20.0 software (IBM Corporation,Armonk,NY,USA).

3.Results

3.1.QTL analysis

3.1.1.QTL for TKW

Twenty-four putative QTL associated with TKW were detected on 18 chromosomes (none on 3B,6D,and 7B) (Figs.1,S1;Tables 1,S3).Ten (42%),six (25%) and eight (33%) QTL were located in the A,B and D subgenomes,respectively.Of these,QTkw-2Awas a major QTL,QTkw-4A,QTkw-4B,andQTkw-7A.2were stable QTL,andQTkw-2Dwas a major and stable QTL.QTkw-2Dwas mapped to an 8.3 cM(120.9-129.2 cM)genetic interval between SSR markersXcfd233andXme7em26and was detected in four individual environments,exhibiting PVE ranging from 8.33% to 17.43% and an average LOD score of 8.22.KN9204 provided the beneficial allele to increase TKW by 1.47 g (1.05-1.89 g).

3.1.2.QTL for KL

Twenty-six putative QTL for KL were identified on chromosomes 1A,1B (2),1D (3),2A (3),2B,2D,3A,3B (3),3D,4A (2),4B,5A,6B,6D and 7A (4) (Figs.1,S1;Tables 1,S3).The numbers of QTL in the A,B and D subgenomes were 12 (46%),eight (31%)and six (23%),respectively.Of these,QKl-3B.2was major QTL,QKl-2A.2was stable QTL,andQKl-1B.2was a major and stable QTL.QKl-1B.2was identified in six individual environments,exhibiting PVE of 13.05% and an average LOD score of 5.77.The positive additive effect of the allele from J411 increased KL by 0.10 mm (ranging from 0.08 to 0.12 mm).

3.1.3.QTL for KW

There were 25 putative QTL for KW on 16 chromosomes(1A,1B,2A,2D,3A,3B,3D,4A,4B,5A,5B,5D,6A,6B,7A,and 7D) (Figs.1,S1;Tables 1,S3),including 11 (44%),seven (28%),and seven(28%)in the A,B and D subgenomes,respectively.Of these,QKw-5BandQKw-6B.2were major QTL.No stable QTL was detected.

3.1.4.Co-located QTL

Twelve QTL clusters consisting of 28 QTL(Table 2)were identified on chromosomes 1B,1D,2A,2D,3B,3D,4A(2),4B,5A,6B,and 7A,which simultaneously affected at least two traits with the same additive-effect directions(negative or positive).Three QTL clusters(C3,C7 and C10)for TKW,KL and KW were detected in the A subgenome.QTL clusters (C5,C6,C8 and C11) for KL and KW were detected on chromosomes 3B,3D,4A and 6B,respectively.QTL clusters (C1 and C2) for TKW and KL were detected on chromosomes 1B and 1D,respectively.QTL clusters (C4,C9 and C12) for TKW and KW were detected on chromosomes 2D,4B and 7A,respectively.Among them,C4 (QTkw-2DandQKw-2D.3) was located in the intervalXcfd233-Xmag2956.1(120.9-139.5 cM) on chromosome 2D,with positive additive effects of alleles from KN9204 increasing TKW and KW simultaneously.

3.2.Reconstruction of the chromosome 2D genetic map and QTkw-2D remapping

With the high-density genetic map (170.32 cM) (Fig.S2;Table S4),QTkw-2Dwas located in the 100.73-106.30 cM genetic interval (5.57 cM) between markersXcfd233andwPt-730744on chromosome 2D (Figs.2,S2;Table 3),corresponding to a physical bin from 561.05 to 618.13 Mb in the IWGSC RefSeq v1.0 of Chinese Spring.QTkw-2Dwas detected in six individual environments (E1,E2,E3,E4,E5,and E6) with PVE values of 7.02%-13.22% and an average LOD score ＞3.KN9204 provided the beneficial allele to increase TKW by 1.44 g (1.04-1.93 g).

3.3.Fine mapping of QTkw-2D

To further precisely mapQTkw-2D,17 SSR markers(Table S1)in its target region were used to screen the 188 F6:7KJ-RILs,and a RHL(RIL048) was identified that contained polymorphisms in markersP367(561.87 Mb),P37(566.12 Mb)andP398(568.72 Mb)(Fig.3A).These three polymorphic markers were further used to screenQTkw-2DKN9204andQTkw-2DJ411from the RIL048 selfing offspring.In 2016-2017,the average TKW and KW in theQTkw-2DKN9204lines (F7:8) were significantly higher than those in theQTkw-2DJ411lines by 4.78 g and 0.20 mm,respectively (Fig.S3).Accordingly,theQTkw-2Dtarget region was inferred to lie in theP359-P237(561.28-570.41 Mb) interval,corresponding to a 9.13 Mb physical interval on chromosome 2D of Chinese Spring.

Fig.2.LOD scores of QTkw-2D in the eight environments and BLUP.

Fig.3.Fine mapping of QTkw-2D.(A)Six types of heterozygous recombinants.(B)Seven types of NILs.(C-F)TKW phenotypes of seven types of NILs in 2019,2020,2021 and 2022.Bars indicate means with standard errors,different lowercase letters are statistically significant (Tukey’s test; P ＜0.05),and the number of lines tested over four growing seasons is shown in parentheses.

For the fine mapping ofQTkw-2D,twelve KASP markers betweenP359andP237(561.28-570.41 Mb) were successfully developed on the basis of the KN9204 genome assembly database and the resequencing results of J411.The genotypes of markersK15(569.36 Mb),K16(569.78 Mb) andK18(570.14 Mb) in the secondary mapping population were consistent with the genotypes of KN9204.Thus,the interval was narrowed to 561.28-569.36 Mb.

Nine KASP markers (K1,K4,K6,K7,K9,K10,K11,K12andK13)were used to genotype the secondary mapping population and identified six types of heterozygous recombinants (F8:9) in the 2017-2018 season (Fig.3A).These nine KASP markers were also used to scan the self-progeny of these recombinants,and seven types of homozygous lines (F9:10) were obtained in the 2018-2019 season.NIL1-NIL4 produced the same pattern of fluorescence as KN9204,and NIL5-NIL7 produced the same pattern of fluorescence as J411 in the interval ofK11-K15,which corresponds to a physical interval of 566.84-569.36 Mb (2.52 Mb) in Chinese Spring (Fig.3B).Significant differences in TKW were found between NIL1-NIL4 and NIL5-NIL7 in 2019 (F9:10)(Fig.3C).The TKW phenotypes of the seven types of NILs were also observed in 2020 (F10:11),2021 (F11:12) and 2022 (F12:13).NILsKN9204(NIL1-NIL4) and NILsJ411(NIL5-NIL7) showed significant differences in TKW in all four environments (Fig.3D-F).Thus,QTkw-2Dwas fine-mapped to the genetic interval ofK11-K15encompassing an interval from 566.84 to 569.36 Mb in Chinese Spring,corresponding to an interval of 569.52-572.04 Mb in the KN9204 reference genome.

3.4.Genetic effect of QTkw-2D on grain filling

To evaluate the genetic effects ofQTkw-2Don kernel weight,NIL1 (NILKN9204) and NIL7 (NILJ411) were compared to identify dynamic changes in kernel weight during grain filling in the 2018-2019 season.A difference(P＜0.01)in fresh weight between the contrasting NILs appeared at 15 days post-anthesis (DPA) and continued until maturity (Fig.S4A).A difference (P ＜0.01) in dry weight was observed from 25 DPA to maturity (Fig.S4B).

The major agronomic traits of NIL1 (NILKN9204) and NIL7(NILJ411) were evaluated across four consecutive growing seasons(Fig.S5).No significant differences in PH,SPP,SL,SN,or KNPS were observed between NILKN9204and NILJ411.For kernel traits,significant differences in TKW and KW were identified between the NIL pairs,whereas KL did not show a significant difference.NILKN9204consistently showed higher TKW and KW than the corresponding NILJ411by respectively 5.20%-14.90% and 5.30%-6.95%,showing thatQTkw-2Dcontrols TKW by regulating KW.

3.5.Candidate genes in the QTkw-2D interval

Using the high-quality KN9204 genome assembly in WheatOmics 1.0 (https://202.194.139.32),QTkw-2Dwas predicted based on the parental genome sequences and J411 resequencing data [31].In the corresponding physical interval (569.52-572.04 Mb) of the KN9204 genome assembly,there were 29 annotated genes (Table S5).Among them,three potential candidate genes were identified based on sequence variation in exons(Table S6) and DEG analysis of RNA-seq results from KN9204 and J411 (Table S7).There was one SNP difference in the exon (missense mutation: GTG/ATG;V/M) ofTraesCS2D02G462500.1(TraesKN2D01HG49540) between KN9204 and J411,which was differentially expressed between the two parental lines in the 14 DPA seeds under low-nitrogen conditions. InTraesCS2D02G463100.1(TraesKN2D01HG49600),one SNP difference was found in exon 4 (missense mutation: TTA/TTG;L/F)between the two lines,and its expression was significantly different between KN9204 and J411 in 7 DPA seeds under normal nitrogen conditions and 28 DPA seeds.TraesCS2D02G460300.1(TraesKN2D01HG49350) showed sequence variations of eighteen SNPs and eight Indels in exons 1,2 and 4 between KN9204 and J411,resulting in sequence changes at the N terminus of the protein,and its expression differed between KN9204 and J411 over the entire grain filling stage.

RNA-seq data for NIL1 (NILKN9204) and NIL7 (NILJ411) indicated that onlyTraesCS2D02G460300.1(TraesKN2D01HG49350) was differentially expressed in 7,14 and 21 DPA seeds among the three candidate genes (Fig.4).Given its DNA sequence differences and differential expression between KN9204 and J411,TraesCS2D02G460300.1(TraesKN2D01HG49350) was predicted to be the candidate gene inQTkw-2D.This gene encodes glycosylphosphatidylinositol mannosyltransferase 2,which has eight transmembrane structures and is localized in the endoplasmic reticulum (ER) (https://www.uniprot.org/UniProt/W5BRU7).This protein is involved in GPI-anchor synthesis and glycolipid biosynthesis (https://www.uniprot.org/UniProt/W5BRU7).

Fig.4.Relative expression of TraesCS2D02G460300.1 (TraesKN2D01HG49350) in NILKN9204 and NILJ411 during grain development.**,P ＜0.01;***,P ＜0.001;ns,P ＞0.05.

4.Discussion

4.1.Comparison of clusters with those reported previously

In this study,24,26 and 25 putative QTL were respectively identified for TKW,KL and KW based on the high-density linkage map.the QTL number identified for each kernel trait nearly doubled compared with that in the previous report based on the moderate-density genetic map consisting of 602 loci (e.g.,SSR markers) [35].Twelve clusters involving 28 QTL for TKW,KL and KW were identified on 11 chromosomes (Table 2).We compared these QTL clusters with those previously reported by the rough physical locations via blasting the sequences of the reported markers.QTL clusters C1,C3,C4,C7,and C9 on chromosomes 1B,2A,2D,4A,and 4B,respectively,are likely identical to those identified previously[35]based on the same phenotypic data of the KJ-RIL population.C7 (QTkw-4A,QKl-4A.1,andQKw-4A.1) in the intervalAX-108814240-AX-109342310(640.22-661.97 Mb) is close to the cell wall invertase geneTaCWI-4A,which is associated with TKW and KNPS [42].C9 (QTkw-4B.1,QKw-4B.1,andQKw-4B.2) is flanked by markersAX-111742822andAX-111578547and close to theRht1gene for reduced plant height that was associated with TKW,KL and KW [15,43].

Owing to the use of different genetic linkage maps,more SNP markers were used in this study than in Cui et al.[35],resulting in more QTL and QTL clusters.Seven new QTL clusters (C2,C5,C6,C8,C10,C11,and C12) were identified and compared with those from previous studies.C2 (479.88-487.61 Mb) affecting TKW and KW on chromosome arm 1DL was likely a novel QTL cluster,given that only one QTL cluster for TKW,yield,SN and KNPS on chromosome arm 1DS was previously identified[44].A QTL cluster for KL and KW on chromosome arm 3BS has been previously reported[45],but the C5 detected in this study was ＞30 Mb away.C6 (578.75-587.5 Mb) for KL and KW on chromosome arm 3DL was in a similar position to a QTL cluster for TKW and yield [46].C10 (429.15-548.66 Mb) for TKW,KL and KW on chromosome 5A was located at a similar position to a QTL cluster affecting SL and spike compactness [47];however,its genomic location was different from that of the QTL cluster for TKW,KL,and KW [45].C11 (671.92-673.61 Mb) for KL and KW on chromosome arm 6BL was located near a QTL cluster for SL,PH,and heading date[15].C12 (671.18-680.84 Mb) for TKW and KW was located on chromosome 7A near the ABERRANT PANICLE ORGANIZATION 1 geneTaAPO-A1associated with SN,SL and yield [48].Homoeologous group 7 has been reported [49] to influence wheat yield determination,and loci controlling kernel weight and size have been identified on chromosome 7A [15,43,47,50].

4.2.QTL for TKW on chromosome 2D

Several QTL for kernel weight and size have been mapped on chromosome arm 2DL.Two QTL for KL were reported to lie in the same or neighboring interval asQTkw-2D,of which one was closely linked withXgwm539[51],and the other withrz900a[52].Other QTL associated with KW have been detected in this chromosomal region.A major and stable QTLQTkw.ncl-2D.2was identified [16] on chromosome 2D that was co-located with a QTL for KW in the marker intervalXwmc601-Xgwm349.Another major and stable QTL was reported [49] in the intervalBS0008362351i-kukri_c7605_181that simultaneously controlled TKW and KW (QTKW.ndsu.2DandQKW.ndsu.2D).A QTL cluster for KL and KW was identified on chromosome arm 2DL [8],where a QTL for TKW was also detected via conditional QTL mapping,whose additive effect was derived from the same donor parent KN2007 that donated the beneficial alleles that increased KW.In the present study,based on the high-density linkage map [20],QTkw-2Dwas located in an 8.3 cM genetic interval flanked byXcfd233andXme7em26markers (Table 1),which was narrower than the interval containingQTkw-2D,a 30.2 cM genetic interval flanked by markersXcfd233andXgpw5215.1[35] and co-located withQKw-2D.3(Xbarc228-Xmag2596.1),together constituting a QTL cluster.The genomic region ofQTkw-2Doverlapped QTL regions that were identified in previous studies [8,53-55].Given thatQTkw-2Dwas detected in multiple genetic backgrounds,QTkw-2Dis convincingly of great implication for kernel weight.

Table 1Major or stable QTL for thousand-kernel weight (TKW),kernel length (KL),and kernel width (KW) identified using KJ-RILs in eight environments.

Table 2QTL clusters jointly affecting thousand-kernel weight (TKW),kernel length (KL),and kernel width (KW).

Table 3The putative QTL QTkw-2D for thousand-kernel weight (TKW).

4.3.Fine mapping of QTkw-2D

Thede novoassembly of genomic sequences of Chinese Spring and other Triticeae species make it easier to determine the physical location of a sequence and develop DNA markers.In this study,chromosome 2D high-density genetic map was different from that of Cui et al.[20],markers that were not physically located on the 2D chromosome were removed.Using the new genetic map,QTkw-2Dwas remapped to a 5.57 cM genetic interval flanked byXcfd233andwPt-730744.UsingQTkw-2DNILs,new KASP markers further mapped this QTL toK11-K15,an interval from 566.84 to 569.36 Mb of Chinese Spring [23].High-quality genome sequencing of the commercial wheat variety KN9204 was reported and available from WheatOmics;accordingly,the physical location ofQTkw-2Din the KN9204 reference genome is known to be 569.52-572.04 Mb.In theQTkw-2Dtarget interval,there is 99%sequence similarity but a difference in the number of annotated high-confidence genes between the reference genome sequences of Chinese Spring and KN9204,suggesting that some SNP/Indel differences may lead to the loss of gene function in KN9204.Another explanation of this phenomenon may be chromosomal structure variation;two inversions occurred in the 2D chromosome of KN9204 in comparison with Chinese Spring [31].We suggest that it is better to select the KN9204 reference genome to predict candidate genes than that of Chinese Spring.

4.4.Candidate genes of QTkw-2D

Owing to wheat’s genome complexity,few genes determining wheat kernel weight and size in wheat have been isolated by map-based cloning[34].Alternatively,comparative genomics provides an effective approach to identify functional genes in wheat.To date,33 yield-associated genes have been isolated by homology-based cloning [18].Of these,several associated with kernel weight and size lie on group 2 homoeologous chromosomes,includingTaDA1[56] andTaFlo2[57] on chromosome 2A andTaSus2[58]on chromosome 2B.No such genes have been reported on chromosome 2D.

In this study,theQTkw-2Dcandidate geneTraesCS2D02G460300.1(TraesKN2D01HG49350)was predicted from sequence differences and differential expression in developing seeds between KN9204 and J411,as well as DEGs in developing seeds ofQTkw-2DNILs(Fig.4).GPI anchoring proteins are extracellular proteins that are attached to GPI anchors via posttranslational modification mechanisms and anchored outside of the plasma membrane to participate in physiological processes,such as growth and development,cell wall formation and signal transduction.Loss of function ofArabidopsisGPI-anchored protein(COBRALIKE 10) causes gametophytic male sterility,leading to few seeds[59].These seeds were twice the size of seeds from wild-type plants,owing mostly to the enlarged embryonic cells.The wheat fertility geneTaMs1also encodes a GPI-anchored protein that regulates pollen development [60].A mutation inTaMs1led to reduced fertility or complete male sterility,which further affected seed setting.The maizeroothairless3mutant displayed an aberrant root phenotype and reduced grain yield [61].In this study,theQTkw-2Dcandidate gene is involved in GPI-anchor synthesis necessary for GPI-anchored protein assembly,which increases wheat kernel weight by regulating kernel width without obvious effects on other yield components,such as plant height,spikes per plant,spike length,spikelet number and kernel number per spike.Functional verification of theQTkw-2Dcandidate gene using CRISPR/Cas9 editing and overexpression may shed further light on the genetic control of kernel weight and size.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

CRediT authorship contribution statement

Deyuan Meng:Data curation,Formal analysis,Investigation,Software,Validation,Visualization,Writing -original draft.Aamana Batool:Data curation,Formal analysis,Investigation,Software,Validation,Visualization,Writing -original draft.Yazhou Xuan:Data curation,Investigation,Writing -original draft.Ruiqing Pan:Data curation,Investigation.Na Zhang:Investigation,Writing-review&editing.Wei Zhang:Investigation,Writing-review&editing.Liya Zhi:Data curation,Investigation,Writingreview &editing.Xiaoli Ren:Data curation,Investigation.Wenqing Li:Data curation,Investigation.Jijie Li:Data curation,Investigation.Yanxiao Niu:Data curation,Formal analysis,Investigation,Writing -review &editing.Shuzhi Zheng:Writing-review&editing.Jun Ji:Investigation,Writing-review&editing.Xiaoli Shi:Data curation,Writing -review &editing.Lei Wang:Writing -review &editing.Hongqing Ling:Writing -review &editing.Chunhua Zhao:Writing -review &editing.Fa Cui:Data curation,Project administration,Writing -review &editing.Xigang Liu:Funding acquisition,Writing -review &editing.Junming Li:Project administration,Funding acquisition,Supervision,Writing -review &editing.Liqiang Song:Project administration,Funding acquisition,Supervision,Writing -review &editing.

Acknowledgments

We are grateful to Professor Zhiyong Liu from Institute of Genetics and Developmental Biology,Chinese Academy of Sciences for his critical review of the manuscript.This research was jointly supported by the National Natural Science Foundation of China(32272056,U22A6009,31671673,and 31871612),Hebei Natural Science Foundation (C2021205013,C2022204202),Talents Program of Hebei Agricultural University in China (YJ2021016),and China Agriculture Research System of MOF and MARA (CARS-03).

Appendix A.Supplementary data

Supplementary data for this article can be found online at https://doi.org/10.1016/j.cj.2023.03.007.