Songyue Chi, Qin Yo, Rui Liu, Wenhui Xing, Xue Xio, Xing Fn, Jin Zeng, Lin Sh,Houyng Kng, Hiqin Zhng, Dn Long, Dndn Wu, Yonghong Zhou, Yi Wng,*

a Triticeae Research Institute, Sichuan Agricultural University, Wenjiang 611130, Sichuan, China

b College of Resources, Sichuan Agricultural University, Wenjiang 611130, Sichuan, China

Keywords:Polish wheat MADS-box gene Alternative splicing Kernel length Breeding

ABSTRACT Kernel size,one of the traits that determine wheat yield,is controlled by multiple quantitative trait loci.Polish wheat(Triticum polonicum)has elongated and plump kernel and is a valuable material for breeding high-yielding wheat cultivars.However,genes or loci determining kernel length(KL)in Polish wheat are unknown.We identified and validated a major KL gene,KL-PW,at the P1 locus in Polish wheat.KL-PW is VRT-A2, which encodes a MIKC-type MADS-box protein (MADS55). An insertion/deletion mutation in intron 1 of VRT-A2PW led to an alternatively spliced transcript, VRT-A2PW2. Quantitative PCR analysis showed that VRT-A2PW was more highly expressed in developing seeds than was VRT-A2Ailanmai.Brassinosteroid (BR) sensitivity experiment and the expression of BR-related genes indicated that VRTA2PW functions as a positive regulator of BR responses. VRT-A2PW significantly increased KL of wheat.These findings not only reveal the molecular basis of KL-PW in controlling KL,but also provide a valuable genetic resource for increasing kernel size in wheat.

1. Introduction

Spike number,kernel number per spike,and kernel size are key components that determine yield of wheat (Triticum aestivum L.,AABBDD, 2n = 6x = 42) [1]. During the history of domestication and breeding, large kernel size has always been selected to improve wheat yield[2–4].Kernel size is influenced mainly by kernel length (KL), width, and thickness and is positively correlated with kernel weight and quality(protein and nutrient element contents) [5–8]. Several genes that contribute to KL of wheat, including TaGS1 [8], TaGS5 [9,10], TaCYP78A3 [11], TtGRF4 [12], TaGS3[13], TaGL3 [4], and TaDA1 [14], have been identified by comparative genomics approaches. However, these genes, except TtGRF4[12], have all been identified in hexaploid wheat. As valuable genetic resources for hexaploid wheat improvement, tetraploid wheat species, such as T. jakubzineri (AABB), T. polonicum (AABB),T. timopheevi (AAGG), and T. ispahanicum (AAGG), carry many genes that control KL, but few are known [15].

Polish wheat(T.polonicum L.,AABB,2n=4x=28)has elongated and plump kernels, and a high thousand-kernel weight of approximately 80 g,far greater than the 40–60 g of hexaploid wheat,and has been recommended as a valuable material for breeding highyielding wheat cultivars [16,17]. A quantitative trait locus (QTL)controlling the elongated kernel of Polish wheat was mapped on chromosome 7AS between simple sequence repeat (SSR) markers Xkupg174 and Xgpw2119, a region that also contained the P1 locus underlying long glumes [18]. However, a genomic alignment against the genome of T. aestivum ‘Chinese Spring’ (IWGSC RefSeq v1.0) [19] indicated that this region has approximate 60 million nucleotides and 800 candidate genes.It is desirable to further narrow the candidate genetic region, performing map-based cloning of the KL gene in Polish wheat.

Although many wheat KL genes are predicted in multiple signaling pathways,including phytohormone signaling(TaGS5,TaGL3,and TaCYP78A3)[4,9–11],transcriptional regulatory factor(TtGRF4)[12], G-protein signaling (TaGS3) [13], and ubiquitin-proteasomal degradation(TaDA1)[14],it is unknown whether alternative splicing of genes is responsible for determining KL in wheat. In rice(Oryza sativa L.), alternative splicing of genes was reported[20,21] to be responsible for the elongated-kernel trait. For example, a splice-site mutation in the eighth exon of OsMADS1 leads to an alternatively spliced form of OsMADS1,resulting in increased KL[21]. A C-to-A point mutation in the ninth intron of OsSSH1 alters splicing of its mRNA, causing reduced shattering, larger seeds,and higher kernel weight [22]. Alternative splicing occurs extensively in eukaryotes as a mechanism that increases transcriptome plasticity and proteome diversity [23]. Alternative splicing is also found extensively in wheat since its polyploidization and domestication [24]. We hypothesized that alternative splicing of genes is also responsible for determining KL in Polish wheat.

The objective of this study was to clone the candidate gene(KLPW) controlling KL in Polish wheat, using QTL mapping, nearisogenic-line(NIL)development,a survey of haplotype distribution among wheat accessions, and introduction of the candidate gene into hexaploid wheat for investigation of its effects on KL.

2. Materials and methods

2.1. Plant materials and growth conditions

Dwarf Polish wheat (DPW) and tall Polish wheat (TPW) were originally collected from Turpan, Xinjiang, China, by professors Chi Yen and Junliang Yang, Sichuan Agricultural University,Chengdu, China in the 1980s. Both DPW and TPW have elongated kernels, and DPW carries Rht-B1b on chromosome 4BS while TPW does not [25]. Ailanmai (T. turgidum L., AABB, also named Jianyangailanmai) collected from Sichuan has short kernels, and also carries Rht22 on chromosome 7AS [26]. The original recombinant inbred line(RIL)population RIL_DJ from a cross between DPW and Ailanmai comprised 207 lines.A selected set of 174 lines from F6, F7and F8RIL_DJ populations were genotyped and phenotyped for QTL analysis; and the full set of 207 F8lines from the RIL_DJ population was used to further narrow the genomic region of KLPW.The RIL population RIL_TJ comprising 205 F8lines was derived from a cross between TPW and Ailanmai. Because it is unknown whether Rht22 affects KL, two pairs of NILs, NIL-KL-PWPW-1and NIL-KL-PWAilanmai-1lacking Rht22, and NIL-KL-PWPW-2and NIL-KLPWAilanmai-2carrying Rht22, were obtained from two heterozygous lines of F6RIL_TJ. Two secondary F2populations (NIL_F2) comprisong 279 and 123 plants were derived from crosses between NIL-KL-PWPW-1and NIL-KL-PWAilanmai-1and between NIL-KLPWPW-2and NIL-KL-PWAilanmai-2, respectively. The F7and F8RIL_DJ,F8RIL_TJ and NIL_F2populations were used to confirm the candidate genomic region.

Hexaploid wheat(HW)F1populations(HW_F1)were developed from crosses between F6RIL_DJ lines with elongated kernel and several hexaploid wheat cultivars, including Chuanmai 64(CM64), CM104, Shumai 126 (SM126), SM133, Mianmai 1501(MM1501), MM1618, and Nanmai 660 (NM660). The HW_F2populations with 243 lines, including 34 plants of 2197 (RIL_DJ/MM1501), 30 plants of 2202 (RIL_DJ/CM64), 31 plants of 2203(RIL_DJ//K1041/CM64), 32 plants of 2209 (RIL_DJ/CM104), 23 plants of 2212 (RIL_DJ/SM126), 26 plants of 2213 (RIL_DJ/SM133), 28 plants of 2214 (RIL_DJ/MM1618), and 39 plants of 2237 (RIL_DJ/NM660), were then used to validate the candidate gene and investigate the effect of KL-PW on KL. A set of 82 tetraploid and hexaploid wheat accessions were used to estimate the allele frequency of KL-PW (Table S1).

All populations were grown on the Wenjiang and Chongzhou experimental farms of Sichuan Agricultural University. The F6RIL_DJ population was grown in the 2017–2018 wheat growing season from October 2017 to June 2018 in Wenjiang (2018WJ).The F7RIL_DJ population and HW_F1population were grown in the 2018–2019 season from October 2018 to June 2019 in Wenjiang (2019WJ) and Chongzhou (2019CZ). The F8RIL_DJ, F8RIL_TJ,NIL_F2, and HW_F2populations and tetraploid and hexaploid wheat accessions were grown in the 2019–2020 season from October 2019 to June 2020 in Wenjiang (2020WJ) and Chongzhou(2020CZ). Twenty seeds of each line were planted in each row,with rows 2 m long and 30 cm apart.

2.2. Phenotypic measurements

Twenty-five kernels of the parents and each RIL, NIL_F2, and HW_F2plant were scanned with an Epson Expression 10,000 XL scanner (Seiko Epson, Suwa, Nagano, Japan). Kernel length was evaluated by WinSEEDLE (Regent Instruments Inc., Quebec City,Quebec, Canada) based on the output images. Kernel pericarp cell lengths of two pairs of NIL were observed and photographed with a Quanta 450FEG scanning electron microscope(FEI,Hillsboro,OR,USA). Statistical analysis, including KL, gene expression, root length,and cell length of kernel pericarp,were performed via analysis of variance (ANOVA) using SPSS 18.0 (SPSS, Chicago, IL, USA)with default parameters. Significant differences were determined by ANOVA at P ＜0.05 or P ＜0.01.Figures were drawn with Sigma-Plot 12.0 (Systat Software Inc, Point Richmond, CA, USA).

2.3. QTL analysis

Linkage mapping was performed with JoinMap 4.0 software(Kyazma BV, Wageningen, Netherlands); a logarithm of odds(LOD) threshold of 3.0 was recommended for candidate regions.Genetic distances(cM)were calculated with the Kosambi mapping function[27].The physical position and reference sequence of candidate regions were obtained from the genomic sequence of T.aestivum ‘Chinese Spring’ (IWGSC RefSeq v1.0) using the position of each marker.

A high-density genetic map of F6RIL_DJ population with 174 lines (unpublished data) was developed using genotyping by sequencing (GBS) [28]. The KL phenotype data from F6, F7and F8RIL_DJ in five environments were used for QTL analysis with IciMapping 4.1 software (http://www.isbreeding.net/) with the composite interval mapping function, with the following parameters: LOD threshold of 3.0, 1 cM step, and 1000 permutations at P ＜0.05.

2.4. Gene mapping

Genomic DNA was extracted from parents and each RIL,NIL_F2,HW_F1and HW_F2plant using a plant genomic DNA kit(TIANGEN,Beijing, China). According to the genomic sequence of the QTL region in T. aestivum cv. ‘Chinese Spring’ (IWGSC RefSeq v1.0)(http://plants.ensembl.org/), microsatellites were predicted using the MIcroSAtellite identification tool (https://webblast.ipk-gatersleben.de/misa/)[29,30]and simple sequence repeat(SSR)primers were designed with Beacon Designer 7.0 software (http://www.premierbiosoft.com/) (Table S2). PCR amplification of SSR markers was performed with 2× Taq Master Mix for Page P114 (Vazyme,Nanjing,Jiangsu,China)with a profile of 95°C for 5 min,35 cycles of 95 °C for 45 s, 58 °C for 45 s, and 72 °C for 45 s, with a final extension at 72 °C for 7 min. PCR products were separated on 8%polyacrylamide gels. First, the F7RIL_DJ population was used to narrow the QTL region of KL-PW using six new SSR markers. Second, the F8RIL_DJ, F8RIL_TJ and NIL_F2populations were used to further narrow the genomic region of KL-PW using six other new SSR markers.

2.5. Gene cloning and sequence analysis

PCR primers used for amplification of the genomic sequences and the coding sequences(CDS)of candidate genes were designed from the Chinese Spring reference genome sequence (Table S2).PCR amplification was performed with Phanta Max Super-Fidelity DNA Polymerase P505 (Vazyme, Nanjing, Jiangsu, China) under the following conditions: 95 °C for 5 min; 35 cycles of 95 °C for 30 s, 58 °C for 30 s, and 72 °C for 2 min; and a final extension at 72 °C for 10 min. Each amplified fragment was cloned into the pMD19-T vector (TaKaRa, Dalian, Liaoning, China) prior to sequencing. Vector NTI 11.5.1 software (Invitrogen, Carlsbad, CA,USA) was used to align the nucleotide and amino acid sequences.Expasy (https://web.expasy.org/translate/) was used to translate amino acid sequences. Prosite (https://prosite.expasy.org/prosite.html) was used to predict protein domains.

2.6. Exploitation and detection of VRT-A2PW-InDel marker

Using the sequence differences in VRT-A2 between Polish wheat and Ailanmai, a pair of VRT-A2PW-InDel primers (InDel-F:GTACCTTCGCCTCTCCTCTC; InDel-R: GCACAAACCGTTCCCTAAAAC)was designed using Beacon Designer and amplified in F7and F8RIL_DJ, F8RIL_TJ, NIL_F2and HW_F2populations and tetraploid and hexaploid wheat accessions following the protocol described above.

2.7. Phylogenetic analysis

The amino acid sequences of VRT-A2 homologs downloaded from Ensembl Plants (http://plants.ensembl.org/) were aligned with Clustal W (https://www.genome.jp/tools-bin/clustalw/). A phylogenetic tree was constructed with MEGA5 software(https://www.megasoftware.net/)using the neighbor-joining algorithm and default parameters.The Poisson model was used to calculate the evolutionary distances. The phylogeny was tested by bootstrapping with 1000 replications.

2.8. RT-qPCR

Total RNA was extracted with a Plant RNA Kit (Omega Bio-Tek,Norcross, GA, USA). cDNA was synthesized using the M-MLV First Strand cDNA Synthesis kit (Invitrogen). Two pairs of specific qPCR primers were designed from the CDS sequence of VRT-A2(Table S2). RT-qPCR was performed on the CFX-96 system (Bio-Rad, Hercules, CA, USA) as described by Wang et al. [31]. To normalize gene expression levels, the Actin gene was used as a reference gene [31]. Relative expression levels were calculated with CFX Manager 3.1 software (Bio-Rad) by to the 2ΔΔCT method. To investigate whether the candidate gene participates in BR responses,the expression levels of genes involved in BR biosynthesis and BR signaling except for GSK2 (Table S2) were performed as described by Cheng et al. [32]. Each experiment was performed with three biological replicates.

2.9. Epi-brassinolide (BL) treatment of seedlings

Seeds of NIL-KL-PWPW-1and NIL-KL-PWAilanmai-1were sterilized with 2% NaClO for 5 min and then washed with distilled water for 1 min. Germinated seeds were incubated at 4 °C for 2 days in the dark and in a 20 °C light incubator for 2 days. Seedlings were cultivated in hydroponic boxes containing four concentrations (0,0.001, 0.01, and 0.1 μmol L-1) of BL (Aladdin, Hangzhou,Zhejiang,China) and grown in a greenhouse(20 °C, 16 h light and 8 h dark)for 6 days.The length of primary roots was measured with a ruler.The experiment was repeated three times, with at least 10 plants per treatment.

3. Results

3.1. QTL for elongated kernel

The KLs of TPW and DPW were similar and greater than that of Ailanmai(P ＜0.05)(Fig.1A).The KLs of three generations of RIL_DJ population for all the five environments were positively correlated with one another, with a range of correlation coefficients from 0.940 to 0.972 (Table S3). A stable QTL on chromosome 7AS for elongated kernel of DPW, named KL-PW, was detected in all three generations of the RIL_DJ population in five environments(Table S4). The candidate genomic region of KL-PW was narrowed to a 9.09-Mb region on chromosome 7AS between two GBS SNP markers: 7A_134 (124.22 Mb) and 7A_140 (133.31 Mb), which overlapped with the P1 locus (Fig. 1B).

3.2. Further narrowing of the candidate genomic region of KL-PW

Six pairs of new SSR markers located between 7A_134 and 7A_140 were linked to KL-PW after genotyping the F7RIL_DJ population.The candidate genomic region of KL-PW was re-localized in a 2.03 Mb genetic interval between SSR markers XP9 and XP14(Fig. 1C; Table S5). Six further newly designed SSR markers were used to genotype the F8RIL_DJ and the F8RIL_TJ populations. The candidate genomic region of KL-PW was further narrowed to lie between SSR markers XP85 and XP87,with a cosegregating marker XP90 (Fig. 1D–E; Table S5).

To confirm the genomic region of KL-PW,two pairs of NILs were used for further genetic analysis. The kernel and glume lengths of NIL-KL-PWPWwere longer(P ＜0.05)than those of NIL-KL-PWAilanmai(Fig.1F–G).The flowering time of NIL-KL-PWPWwas also later than that of NIL-KL-PWAilanmai(Fig. 1H). However, the KLs of NIL-KLPWPW-1and NIL-KL-PWAilanmai-1without Rht22 were longer(P ＜0.05)than those of NIL-KL-PWPW-2and NIL-KL-PWAilanmai-2with Rht22 (Fig. 1F), but the flowering times were similar (Fig. 1H).These results indicated that Rht22 negatively affected KL, but did not affect flowering time. The candidate genomic region of KLPW was also located between SSR markers XP85 and XP87, and marker XP90 cosegregated with KL-PW in the NIL_F2populations(Fig. 1I; Table S5).

According to the reference genomic information and gene annotation of wheat chromosome 7AS between XP85 (128.79 Mb) and XP87(128.92 Mb),the region contained six high-confidence genes:TraesCS7A02G175200 (MADS-box transcription factor 55), TraesCS7A02G175300 (peptidyl-prolyl cis–trans isomerase CYP40),TraesCS7A02G175400 (blue copper protein), TraesCS7A02G175500(pentatricopeptide repeat-containing protein), and two genes,TraesCS7A02G175600 and TraesCS7A02G175700, lacking functional descriptions (Fig. 1J).

3.3. Candidate gene cloning of KL-PW

First, TraesCS7A02G175600 and TraesCS7A02G175700 were excluded from the candidate genes, because there were no differences in coding sequences between parents (Fig. S1A). Second,TraesCS7A02G175200 encodes a MADS-box transcription factor,which was named VEGETATIVE TO REPRODUCTIVE TRANSITION 2(VRT2) in wheat [33]. As orthologs of VRT-A2 (Fig. S2A), AtAGL24 in Arabidopsis thaliana and OsMADS55 in rice mediate flowering time[34–36].Given that long kernel was linked to delayed flowering time(Fig.1H)in the NIL-KL-PWPW-1and NIL-KL-PWPW-2,VRT-A2 appeared to be the leading candidate gene for KL-PW.

Fig.1. QTL analysis and map-based cloning of KL-PW.(A)Kernel morphology and KL of TPW,DPW and Ailanmai.(B)QTL location for KL-PW.(C)High-resolution linkage map of KL-PW in F7 RIL_DJ population. (D) High-resolution linkage map of KL-PW in F8 RIL_DJ population. (E) High-resolution linkage map of KL-PW in F8 RIL_TJ population. (F)Comparison of kernels length between NIL-KL-PWPW and NIL-KL-PWAilanmai.(G)Glume morphology of NIL-KL-PWPW and NIL-KL-PWAilanmai.Scale bars,1 cm.(H)Comparison of flowering time between NIL-KL-PWPW and NIL-KL-PWAilanmai. (I)Linkage map of KL-PW in NIL_F2 population.(J) Predicted high-confidence genes between SSR markers XP85 and XP87 according to IWGSC RefSeq v1.0.There was full colinearity between the linkage map and the physical map of all SSR markers.NS,not significant;**,significant at P ＜0.01.

Sequencing of genomic DNA of VRT-A2(165 bp upstream of the ATG to 21 bp downstream of the terminator)revealed 36 SNPs and nine insertions/deletions in VRT-A2PWof DPW in comparison with VRT-A2Ailanmaiof Ailanmai (Fig. S1B). Sequencing of cDNA showed that VRT-A2PWproduced two forms of transcripts, VRT-A2PW1and VRT-A2PW2, while VRT-A2Ailanmaihad only one. Sequence alignment showed that VRT-A2 comprised eight exons and seven introns(Fig. 2A). The cDNA sequences of VRT-A2PW1and VRT-A2Ailanmaishowed 98.83% identity. However, the cDNA sequence of VRTA2PW2had 3291 bp(intron 1 was retained).These results indicated that the insertion/deletion mutation in intron 1 caused alternative splicing of VRT-A2PWin Polish wheat. A VRT-A2PW-InDel marker developed from the insertion/deletion in intron 1 cosegregated completely with KL-PW in the F7and F8RIL_DJ, F8RIL_TJ, and NIL_F2populations (Fig. 1C–E, I). These results further indicated VRT-A2 as the candidate gene for KL-PW.

VRT-A2PW1and VRT-A2Ailanmaiencode a MIKC-type MADS-box protein,containing MADS-box,I-region,K-box,and C-box domains(Fig. 2B). The alternatively spliced protein VRT-A2PW2contained a MADS-box domain with an additional 60 residues,and potentially a polypeptide that contained the predicted domains of the I-region,K-box, and C-box (Fig. 2B).

Fig. 2. Allelic variation of VRT-A2 between PW and Ailanmai. (A) Nucleotide mutations of VRT-A2 distinguishing PW and Ailanmai. (B) Amino acid mutations of VRT-A2 distinguishing PW and Ailanmai.

3.4. Effect of VRT-A2PW on KL in the background of hexaploid wheat

VRT-A2PWwas introduced into hexaploid wheat by hybridization. The KLs of HW_F1lines were longer (P ＜0.05) than those of their hexaploid wheat parents(Fig.3A;Table S6).Genotype analysis using marker VRT-A2PW-InDel indicated that all HW_F1lines were heterozygous. Genotyping and phenotyping confirmed that marker VRT-A2PW-InDel cosegregated with KL in the HW_F2population(Fig.3B;Table S5).These results further indicated VRT-A2 as the candidate gene of KL-PW.

3.5. Relative expression of VRT-A2 in kernel

The expression of VRT-A2 in various tissues was predicted using an RNA-seq expression database of wheat (Wheat Exp; http://wheat.pw.usda.gov/WheatExp/).VRT-A2 was predicted to be highly expressed in leaves and stems,but rarely or even not expressed in kernel (Fig. S2B).

Relative expression of VRT-A2PW2was observed in NIL-KL-PWPW-1, but not in NIL-KL-PWAilanmai-1(Fig.4A), confirming that the transcript of VRT-A2PW2was present only in NIL-KL-PWPW-1. The total expression of VRT-A2PW1and VRT-A2PW2in NIL-KL-PWPW-1was over 50-fold higher than that of VRT-A2Ailanmaiin NIL-KL-PWAilanmai-1(Fig.4B).Thus,the insertion/deletion mutation in intron 1 not only caused the alternative splicing,but also increased the expression of VRT-A2PWin NIL-KL-PWPW-1.

Given that OsMADS55 can directly interact with OsDEP1 to determine KL in rice [20,37], we suspected that the expression of DEP1 in developing kernels would also differ between NIL-KLPWPW-1and NIL-KL-PWAilanmai-1. As expected, the expression of DEP1 in developing kernels was lower (P ＜0.05) in NIL-KL-PWPW-1 than in NIL-KL-PWAilanmai-1(Fig. 4C). These results further confirmed VRT-A2PWas the candidate gene for the elongated kernel in Polish wheat.

3.6. Distribution of VRT-A2PW-InDel in tetraploid and hexaploid wheats

To investigate the distribution of the insertion/deletion mutation in intron 1 of VRT-A2, the VRT-A2PW-InDel marker was used to genotype 82 tetraploid and hexaploid accessions. The insertion/deletion mutation in intron 1 of VRT-A2 was found only in Polish wheat and Xinjiang rice wheat (T. petropavlovskyi) (Fig. 5).

3.7. Effect of BL on seedlings in NILs

Fig.4. Expression of target genes. (A,B)Expression of VRT-A2 in developing kernels of NIL-KL-PWPW-1 and NIL-KL-PWAilanmai-1.(C)Expression analysis of DEP1 in developing kernels of NIL-KL-PWPW-1 and NIL-KL-PWAilanmai-1. **, significant at P ＜0.01.

Because rice OsMADS55 acts as a negative regulator of BR response[34],we hypothesized that VRT-A2 in wheat is responsive to BR. The lengths of primary roots and leaves did not differ between NIL-KL-PWAilanmai-1and NIL-KL-PWPW-1under 0 μmol L-1BL,but their growths were inhibited(P ＜0.05)under both 0.01 and 0.1 μmol L-1BL (Fig. 6A–B). Crucially, the inhibition of primary root length and the growth of the second leaf were higher(P ＜0.05) in NIL-KL-PWAilanmai-1than in NIL-KL-PWPW-1seedlings(Fig. 6A–B). These results indicated that VRT-A2 was responsive to BR and that VRT-A2PWlowered sensitivity to high concentrations of BL in NIL-KL-PWPW-1.

3.8. Expression of genes involved in BR biosynthesis and BR signaling

Given that VRT-A2PWreduced the sensitivity of NIL-KL-PWPW-1to BL,it was desirable to investigate whether VRT-A2PWinfluenced the expression of genes involved in BR biosynthesis(D11,Brd2,D2,and Dwarf4)and signaling(GSK2,BRI1,BZR1,DLT,TUD1,and RAVL1)in developing seeds. Compared with NIL-KL-PWAilanmai-1, the relative expression levels of BZR1 and TUD1 were higher (P ＜0.05) in NIL-KL-PWPW-1, whereas those of D11, Brd2, D2, Dwarf4, BRI1, and GSK2 were lower (Fig. 6C). The relative expression levels of DLT and RAVL1 were comparable between the two lines (Fig. 6C).

3.9. Cell size of kernel pericarp

The cell lengths of kernel pericarp were longer (P ＜0.05) in NIL-KL-PWPW-1and NIL-KL-PWPW-2than in NIL-KL-PWAilanmai-1and NIL-KL-PWAilanmai-2(Fig. 6D–G).

4. Discussion

4.1. The candidate gene of KL-PW is VRT-A2, a novel gene controlling KL in wheat

In this study, we found that KL-PW localized on chromosome 7AS between SSR markers XP85 and XP87 is a novel gene associated with elongated KL in Polish wheat, because its physical position differs from the known kernel size-associated genes on chromosome 7AS, including QTgw.crc-7A [38], QTkw-7A.2 [39], TaSUS1-7A[40], TaGASR7-A1 [41], TaTGW-7A [42], and 6-SFT-A2 [43].

Fig.5. Haplotype analysis of VRT-A2PW-InDel in 82 tetraploid and hexaploid wheat accessions.Red arrow indicates wheat accession and white arrow indicates the insertion/deletion mutation in intron 1 of VRT-A2 in Polish wheat and Xinjiang rice wheat.

The candidate gene of KL-PW was assigned as VRT-A2 for the following reasons. First, the genomic region of KL-PW between XP85 and XP87 contained six annotated genes, one of which was VRTA2. Second, KL-PW was linked with not only KL, but also glume length and flowering time, like VRT-A2 and its homologs [33–36,45,46]. Third, an insertion/deletion mutation in intron I of VRT-A2PWfrom Polish wheat caused alternative splicing and increased the expression of VRT-A2PWin developing kernels of NIL-KL-PWPW-1. The glume length of Polish wheat depends on the dosage of VRT-A2 [44]. Fourth, the increased expression of VRTA2PWsignificantly repressed the expression of DEP1 in the same way that OsMADS55 directly interacts with OsDEP1 to determine rice KL[20,37].Last,the VRT-A2PW-InDel marker cosegregated completely with KL-PW in the RIL_DJ,RIL_TJ,NIL_F2,and HW_F2populations. Given that the insertion/deletion in intron 1 of VRT-A2PWwas found only in Polish wheat and Xinjiang rice wheat, Polish wheat may have contributed as a parent to the origin of Xinjiang rice wheat [46].

4.2.Elongated kernels of Polish wheat result from increased expression of VRT-A2PW caused by alternative splicing

MIKC-type MADS-box genes play a critical role in flower development, flowering time control, inflorescence architecture, pollen,leaf and root development,and kernel,seed,and fruit development[47,48]. Although overexpression of TaVRT2 led to no phenotypic change among TaVRT2 overexpression lines, wild-type line, and transgenic null lines apart from flowering time [36], overexpression of the genomic Polish wheat VRT-A2 allele elongated glumes and kernels in a dosage-dependent manner [44]. The insertion/deletion mutation in intron 1 of VRT-A2PWresulted in an alternatively spliced transcript VRT-A2PW2, and led to a total expression of VRT-A2PWover 50 times higher than that of VRT-A2Ailanmai(Fig. 4A and B), thus elongating kernels of Polish wheat.

Fig.6. Comparison of BL effects on seedlings,expression of BR biosynthesis-and signaling-associated genes in developing kernels,and cell length of mature kernel pericarp between NIL-KL-PWPW-1 and NIL-KL-PWAilanmai-1.(A,B)Effects of BL concentration on seedling root length and leaf morphology.(C)Relative expression of BR biosynthesis-and signaling-associated genes in developing kernels.(D,F)Scanning electron microscopy of mature kernel pericarp;(E,G)Statistical analysis of cell length of kernel pericarp.*,significant at P ＜0.05; **, significant at P ＜0.01.

It is well known that the MADS-box domain of MIKC-type MADS-box protein has DNA binding activity [49]; the I-region and K-box domain promote dimerization and higher-order complex formation of two or more MADS-box proteins [50–52]; and the C-box domain is involved in transcriptional activation [49]. A unique conformational regulation of the transcriptional activation domain affects the transcriptional activity of MIKC-type MADS-box genes, in which transcriptional activation is inhibited in the fulllength native form but they show strong transcriptional function following the deletion of the MADS-box domain [21,53]. The retained intron 1 in the alternatively spliced transcript VRT-A2PW2may cause the deletion of the MADS-box domain in VRT-A2PW2(Fig. 2B), promoting transcriptional activation to increase the expression of VRT-A2PWin Polish wheat.MIKC-type MADS-box proteins usually regulate their downstream target genes by formation of homodimers or heterodimers [54,55]. Thus, VRT-A2PW2might form new heterodimers with VRT-A2PW1and interact with target genes, changing their expression level and causing elongated kernel.

VRT-A2 belongs to the SVP/StMADS11 class in wheat [56]. In rice, the SVP/StMADS11 class MADS-box genes, including OsMADS22, OsMADS47 and OsMADS55, act as negative regulators of BR responses [34,57]. However, our results revealed that VRTA2PWacts as a positive regulator of BR responses. First, inhibition of root and second-leaf growth was lower in NIL-KL-PWPW-1than in NIL-KL-PWAilanmai-1seedlings under 0.01 and 0.1 μmol L-1BL treatments (Fig. 6A–B), and the cell size of the kernel pericarp was greater in NIL-KL-PWPWthan in NIL-KL-PWAilanmai(Fig. 6D–G),in agreement with the effects of BR on cell size [58,59]. Second,compared with NIL-KL-PWAilanmai-1,the expressions of BR upstream genes (D11, Brd2, D2, Dwarf4, and BRI1) and BR negative signaling gene (GSK2) in developing kernels of NIL-KL-PWPW-1were lower,whereas the expressions of two BR positive signaling genes (BZR1 and TUD1) were higher (Fig. 6C).

A model summarizing how VRT-A2PWcauses elongated kernel in Polish wheat is presented in Fig. 7. Higher expression of VRT-A2PWinhibits TpGSK2 expression and directly or indirectly promotes TpBZR1 and TpTUD1 expression; the BR biosynthesis genes are down-regulated and inhibit the high BR signal by feedback. At the same time, up-regulated BR signaling genes induce the expression of downstream target genes and promote cell elongation, thereby causing elongated kernels.

Fig.7. Proposed model of the positive regulation of KL in Polish wheat by VRT-A2PW.

4.3. Application of VRT-A2PW in wheat breeding

Introducing the newly-described KL gene VRT-A2PWincreased KL in wheat.The discovery of VRT-A2PWprovides a valuable genetic resource for increasing kernel size in wheat breeding.

5. Conclusions

We identified KL-PW as VRT-A2, which encodes a MIKC-type MADS-box protein and exerts pleiotropic effects on KL, glume length, and flowering time. An alternatively spliced transcript VRT-A2PW2up-regulates the expression of VRT-A2PWand positively regulates BR response. KL-PW introduced into hexaploid wheat increased KL.

CRediT authorship contribution statement

Songyue Chai:Data curation,Formal analysis,Validation,Visualization, Writing - original draft, Writing - review & editing.Qin Yao:Formal analysis, Validation, Visualization, Writing - original draft, Writing - review & editing.Rui Liu:Investigation, Formal analysis,Resources.Wenhui Xiang:Investigation,Formal analysis,Resources.Xue Xiao:Investigation, Formal analysis.Xing Fan:Software, Writing - review & editing.Jian Zeng:Writing - review& editing.Lina Sha:Writing - review & editing.Houyang Kang:Writing-review&editing.Haiqin Zhang:Writing-review&editing.Dan Long:Investigation, Formal analysis.Dandan Wu:Writing - review & editing.Yonghong Zhou:Conceptualization,Supervision,Resources, Writing- review& editing.Yi Wang:Conceptualization, Data curation, Funding acquisition, Project administration, Resources, Writing - original draft, Writing - review &editing.

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.

Acknowledgments

This work was supported by the National Natural Science Foundation of China(31671688)and the Bureau of Science and Technology of Sichuan Province (2020YJ0141).

Appendix A. Supplementary data

Supplementary data for this article can be found online at https://doi.org/10.1016/j.cj.2021.07.006.