Diversity and sub-functionalization of TaGW8 homoeologs hold potential for genetic yield improvement in wheat

Lin M,Chenyng Ho, Hongxi Liu, Jin Hou,Tin Li,*, Xueyong Zhng,*

aKey Laboratory of Crop Gene Resources and Germplasm Enhancement,Ministry of Agriculture/Institute of Crop Sciences,Chinese Academy of Agricultural Sciences,Beijing 100081,China

bInstitute of Animal Sciences,Chinese Academy of Agricultural Sciences,Beijing 100193,China

Keywords:Grain yield Haplotype association Marker assisted breeding Thousand grain weight Triticum aestivum

ABSTRACT Grain Weight 8 (GW8) in rice is a SQUAMOSA Promoter-Binding Protein-Like (SPL) family transcription factor with multiple biological functions.In this study,three GW8 homoeologs were cloned from homoeologous group 7 chromosomes of wheat. Subcellular localization and trans-activation activity assays suggested that TaGW8 is a transcriptional activator.TaGW8 genes were preferentially expressed in young spikes and developing grains.Ectopic expressions of TaGW8 in Arabidopsis and rice resulted in enhanced vegetative growth,earlier flowering and larger seeds. TaGW8-7A was the most highly variable of the three homoeologs with four haplotypes (Hap-1/2/3/4). TaGW8-7B had two haplotypes (Hap-L/H).TaGW8-7A-Hap-2 was associated with high thousand-grain weight (TGW) and large kernel length and showed higher transcriptional levels and binding activity than the other haplotypes. The high frequency of TaGW8-7A-Hap-2 in Chinese wheat populations suggested that it had been extensively selected in breeding. This haplotype showed a high potential for exploitation in global wheat breeding because its frequency was low in exotic germplasm. TaGW8-7B-Hap-H produced abundant transcripts and was associated with early heading and maturity,less tiller number and high TGW.This haplotype might be suitable for multiple cropping systems due to short wheat season. In this study we identified sub-functionalization among TaGW8 homoeologs and obtained functional molecular markers that can be used in breeding for high grain yield.

1. Introduction

Common wheat (Triticum aestivum L.) is an allohexaploid species (2n = 42) that harbors three sub-genomes, named A,B, and D, and feeds ＞35% of the world population [1]. The world wheat harvested area has remained more or less stable since the 1960s but global production and yield have steadily increased (http://www.ers.usda.gov/data-products/wheatdata.aspx). Increased yield is the most important goal in wheat breeding,as worldwide demand for food increases and farmland resources decrease [2]. In recent decades, several yield-related genes have been mapped and isolated through quantitative trait loci (QTL) mapping in rice (Oryza sativa L.),and the genetic and molecular mechanisms of their effects on yield have been explained [3]. Transcription factor OsSPL13 controlling grain size was identified in a diverse worldwide rice collection by genome wide association studies (GWAS)confirming GWAS as a powerful method for identifying and mapping yield-related genes [4-6]. However, identification of yield-related genes in wheat has not been as successful as in rice. One explanation for this might be the slower establishment of genomic resources in wheat compared to rice because of its larger and more complex genome. Although the whole genome sequence of wheat and its relative species are now publicly available,the genome structure and gene annotation still have major deficiencies that continue to complicate efficient QTL mapping [7,8]. So far, few yield-related genes in wheat have been cloned and characterized by QTL-mapping or GWAS [9,10]. Some genes were identified by homologybased methods,showing that it is possible to use this method to isolate yield-related genes in wheat[11-14].

In the case of single copy genes each of three genomes of hexaploid wheat has one homoeolog and due to the conserved nature of homoeologous genes the three homoeologs can have duplicate or compensatory effects, making genetic and functional analyses extremely challenging. The gene series TaGW2-6A/6B/6D is abundantly expressed in developing grains, and genotypes with higher TGW produce significantly lower transcript levels than genotypes with lower TGW. This indicates that all the three TaGW2 homoeologs negatively regulate grain weight [14]. For some other genes there is evidence of functional differentiation among the homoeologs [15,16]. For instance, TaCPK2-A is induced by powdery mildew (caused by Blumeria graminis f. sp. tritici, Bgt)infection,and down-regulation of TaCPK2-A resulted in loss of resistance to powdery mildew in a resistant wheat line,whereas TaCPK2-D mainly responds to cold treatment [17].Reasons for functional differentiation of homoeologous genes remain to be clarified.

As the most widely studied yield-related transcription factors in crops, SQUAMOSA Promoter-Binding Protein-Like (SPL)genes, have multiple roles in regulating yield traits in rice,including tiller number (OsSPL7, OsSPL14, OsSPL16, and OsSPL17), grain number (OsSPL13, OsSPL14, OsSPL16, and OsSPL17), and grain size (OsSPL13 and OsSPL16) [4,18-21].Genetic studies on SPL genes in Arabidopsis thaliana have uncovered multiple functions in plant growth and development, including the timing of vegetative phase change and floral induction, leaf initiation rate, shoot branching,anthocyanin and trichome production on the inflorescence stem,stress responses,carotenoid biosynthesis, shoot regeneration in tissue culture,and lateral root development[22-26].It has been suggested that ancestral SPL function was to regulate vegetative and floral transition； this function has subsequently been partitioned through differential subfunctionalization in both gene duplication and speciation[27].Discovering and exploiting additional SPL genes or useful genetic variation affecting gene function could be a major goal for breeders seeking to improve yield in cereal crops.

Given the important roles of SPL genes in regulation of yield we were interested in elucidating the role of SPL16(GW8), which was previously shown to have multiple functions in rice. GW8 is a positive regulator of cell proliferation and grain filling, and high expression of GW8 results in early flowering and enhanced grain width and yield [20]. GW8 was shown to bind to the promoter of GW7, a grain shape gene that encodes a TON1 RECRUIT MOTIF (TRM)-containing protein that likely recruits the TON1 protein to cortical microtubule arrays, thus influencing cell division patterns resulting in changes to grain size [28,29]. Moreover, the transcript level of downstream MADS-box genes, such as SCO1 and AGL42, was elevated in GW8 over-expression lines,resulting in early flowering.GW8 was recently shown to have a similar function in wheat but the genetic effects of TaGW8 homoeologs remained to be elucidated [30].

In the present study we cloned the TaGW8 homoeologs from wheat group 7 chromosomes and showed them to be transcriptional activators. The expression patterns of TaGW8s were investigated to determine their spatial and temporal expression in wheat growth and development.Ectopic expression of TaGW8 in Arabidopsis and rice was undertaken to determine the biological functions.We analyzed sequence variation in TaGW8s among wheat accessions varying in multiple yield traits and performed association analysis between phenotypes and genotypes.We found that TaGW8-7A was associated with thousandgrain weight(TGW),grain length(GL)and spike length(SPL),and that TaGW8-7A-Hap-2 was associated with favorable traits due to its high transcription level and binding activity. In contrast TaGW8-7B-Hap-H, which was associated with earlier heading(HD) and maturity (MD), higher TGW, and lower tiller number(ETN) was also shown to have more abundant transcript levels.More importantly, by investigating haplotype frequencies, we found that the favorable haplotypes had undergone artificial selection during wheat breeding in China but there was a significant potential for wheat yield improvement internationally.

2. Materials and methods

2.1. Plant materials

Wheat cultivar Chinese Spring was used for gene cloning.Sixty wheat accessions varying in multiple yield traits were used for sequencing to detect TaGW8 polymorphisms. Two Chinese populations, comprising 256 accessions from the Chinese wheat mini-core collection (MCC) and 342 modern cultivars (MC) released since the 1940s, were used for association studies between phenotypic traits and markers(Table S1)[31,32].

A set of nulli-tetrasomic lines of Chinese Spring wheat was used for chromosome location of TaGW8 genes.Four hundred and thirty-three cultivars from North America, 53 from CIMMYT, 364 from Europe, 78 from the former USSR, and 62 from Australia were used to investigate the haplotype distribution (Table S2).

2.2. Phenotypic assessment and statistical analyses

Phenotypic traits of all accessions in the two germplasm populations, including HD, MD, ETN, TGW, GL, grain width(GW), and grain thickness (GT) were assessed in three environments, 2002 and 2005 at Luoyang in Henan province,and in 2010 at Shunyi,Beijing.Each accession was planted in a 2-row 2 m plot with 30 cm between rows in a randomized block design with three replications.Ten random plants from the middle of each plot were chosen to investigate the phenotypic traits.

DNA sequences were analyzed by DNAMAN (http://www.lynnon.com/) software. Variance analyses were performed using SPSS for Windows Version 12.0 (http://www-01.ibm.com/software/analytics/spss/). To determine phenotypic differences between genotypes we used Tukey tests based on analysis of variance (one-way ANOVA) at a significance level of P ＜0.05.

2.3. Nucleic acid extraction and PCR

Genomic DNA was isolated from young leaves using the CTAB method with modifications [32]. Total RNA was extracted from various plant tissues using TIANGENRNA Plant Plus Reagent (Tiangen Biotech (Beijing) Co. Ltd.). cDNA was synthesized with the SuperScript II system (Invitrogen)according to the manufacturer's instructions. All primers are listed in Table S3； they were designed by Primer Premier 5.0 software (http://www.premierbiosoft.com/). PCR were performed in a total volume of 15 μL, containing 50 ng of DNA,1 μL each of 10 mmol L-1forward and reverse primers,0.24 μL of 25 mmol L-1dNTPs,7.5 μL GC buffer I,and 0.15 μL of LA Taq polymerase (TaKaRa). Quantitative real-time PCR (qRT-PCR)was carried out with SYBR Premix Ex Taq (TaKaRa) on a 7500 Real-time PCR System (Applied Biosystems, Foster City, CA,USA). It was performed in total volumes of 20 μL containing 2 μL of cDNA,1 μL of 2 mmol L-1gene-specific primers,0.4 μL of ROX Reference Dye (50×), and 10 μL of 2× SYBR Premix Ex Taq. Relative expressions were presented as fold-change calculated using the comparative CT method.

2.4. Plasmid construction and genetic transformation

The coding sequence of TaGW8-7A was cloned into binary vector pbi121 under control of the CaMV 35S promoter. The recombinant plasmid pbi121::TaGW8-7A was transformed into Arabidopsis (ecotype Col-0) by the floral-dip method using Agrobacterium tumefaciens strain GV3101 [33]. Transgenic plants were selected on solid half-strength MS medium containing 50 mg L-1kanamycin and positive Arabidopsis plants were transferred into potted soil and grown in a controlled growth chamber at 22 °C, with a 16 h light/8 h darkness photoperiod, light intensity of 120 mmol m-2s-1,and 70%relative humidity.

To construct a grain-specific expression vector for genetic transformation, we introduced the full-length coding sequence of TaGW8-7A into a modified pCAMBIA2300 vector under control of the 1Bx7OEpromoter [34]. The resulting p1Bx7OE::TaGW8-7A construct was mobilized into A.tumefaciens EHA105 and transformed into rice (O. sativa ssp.japonica) cv. Kitaake as described previously [35]. Positive T2generation of transgenic lines was grown in the field(Langfang,Hebei,China)for phenotypic analysis.

2.5. SNP detection and functional marker development

Sixty wheat cultivars were initially chosen for analysis of sequence variation (Table S4). Genome-specific fragments were cloned into the pEASY-T1 simple vector (TransGen Biotech Co. Ltd., Beijing, China) and sequenced by 3730XI DNA Analyzer (Applied Biosystems). Sequence alignments were performed by DNAMAN (http://www.lynnon.com/), and SNPs were identified by DNASTAR(http://www.dnastar.com/).

Functional markers were developed based on selected SNPs. Briefly, DNA fragments containing specific endonucleases site at the SNP were amplified by corresponding genomespecific primers and then separated by electrophoresis on agarose gels after digestion by restriction endonucleases(New England Biolabs(Beijing)Ltd.,Beijing,China).

2.6. Subcellular localization and transcriptional activation activity of TaGW8 protein

The full-length coding sequence of TaGW8-7A without a stop codon was fused upstream of the green fluorescent protein gene(GFP)under the control of CaMV 35S promoter in pJIT163-GFP vector, designated 35S::TaGW8-GFP. The recombinant plasmid was confirmed by sequencing and transformed into wheat protoplasts using PEG4000 as described[33].Subcellular localization of TaGW8 protein was observed with a laser scanning confocal microscope (Leika TCS-NT, Wetzlar,Germany).

For transactivation activity assays full-length and truncated TaGW8-7A were cloned into pGBKT7 vector to construct a series of vectors named BD-TaGW8, BD-TaGW8-N, BDTaGW8-SBP, BD-TaGW8-C1, and BD-TaGW8-C2. These constructs were separately transformed into Saccharomyces cerevisiae strain AH109 and grown on synthetic dextrose medium lacking Trp (SD/-Trp). The transformants were then dropped on SD/-Trp/-His/-Ade media for evaluation of transcriptional activation activity based on growth status. β-Galactosidase liquid assays were performed according to the manufacturer's protocols (Yeast Protocols Handbook PT3024-1；Clontech,Mountain View,CA,USA).

2.7. Promoter transcription activity and promotertranscription factor interactions in Nicotiana benthamiana leaves

Analyses of the promoter transcription activity and the promoter transcription factor interaction were performed in N. benthamiana leaves, as described previously [36]. In this pGreenII system a REN gene under the control of the CaMV 35S promoter provided an estimate of the extent of transient expression.Relative activity was expressed as the ratio of LUC to REN activity.

For promoter transcriptional activity assays 2 kb promoter regions of TaGW8-7A-Hap-1/2/3 were amplified by PCR and fused with the luciferase reporter gene into transient expression vector pGreenII 800-LUC via BamH I and Pst I sites. The pGreenII 800-LUC derivatives and pSoup-p19 were cotransformed into Agrobacterium and co-infiltrated into tobacco leaves. The transcriptional activity of promoter was reflected by the relative LUC signals 48 h after infiltration.

For assays of interaction with the promoter transcription factor, approximately 1.8 kb of TaGW7-2D promoter was amplified by PCR and fused with a luciferase reporter gene in the transient expression vector pGreenII 800-LUC via Sal I and Xba I sites to generate reporter construct TaGW7pro::LUC.The 35S::TaGW8-7A-Hap-1/3/4 constructs consisted of the fulllength TaGW8-7A-Hap-1/3/4 coding sequences that were amplified by PCR and cloned into the transient expression vector pGreenII 62-SK via Xba I and BamH I sites to generate the effector constructs. Agrobacteria harboring the reporter and effector constructs were co-infiltrated into N.benthamiana and the relative LUC signals were analyzed 48 h after infiltration.

2.8. Promoter-driven GUS expression in transgenic Arabidopsis

Promoter sequences of 2153 bp from TaGW8-7B-Hap-H and TaGW8-7B-Hap-L were amplified by PCR and fused with the pCAMBIA1391Z vector via EcoR I and BamH I sites. The pCAMBIA1391Z derivatives were co-transformed into Agrobacterium strain GV3101 and transformed into Arabidopsis as described [33]. Transgenic T3plants were used to test GUS activity. GUS staining and GUS activities of seedlings were performed following a published protocol [9]with modifications.

3. Results

3.1. TaGW8, a member of the SPL gene family, is located on wheat group 7 chromosomes

We identified wheat TaGW8 based on rice OsGW8 (OsSPL16)using IWGSC wheat reference genomic sequences and cDNA sequences on Ensembl Plants [37]. We identified six sequences with high similarity and located them on homoeologous group 5 and 7 chromosomes(Table S5).Further analyses showed that all six genes were homologous to OsGW8.Since the genes on group 7 chromosomes had higher similarity to OsGW8 we targeted those genes as TaGW8s. The highly similar sequences of TaGW8-7A/7B/7D were isolated from mixed cDNA and genomic DNA of Chinese Spring wheat.The genomic sequences of TaGW8-7A/7B/7D were 5060, 4826,and 5229 bp, and the CDSs were 1224, 1230, and 1245 bp,respectively.The gene structures of TaGW8s were determined by aligning the genomic sequences and their corresponding CDSs (Fig. S1). Each gene consisted of three exons and two introns, and its chromosomal location was confirmed by using a set of Chinese Spring nulli-tetrasomic lines(Fig. S2).

The predicted TaGW8-7A, TaGW8-7B, and TaGW8-7D proteins have 410, 412, and 417 amino acids, respectively. A sequence alignment with AtSPL13 and OsSPL16 revealed that TaGW8s shared conserved SBP domains with sequence identities of 29.91% and 66.83%, indicating that TaGW8-7A/7B/7D encodes isoform TaSPL16 (Fig. S3). A phylogenetic tree was constructed by aligning TaGW8 proteins and other SPL proteins in rice, maize (Zea mays L.), soybean (Glycine max (L.)Merr.), Brachypodium distachyon L., Brassica napus L., and A.thaliana (Fig. S4). TaGW8s and their homologous genes in monocots clustered into the same clade, whereas those from dicots clustered into a nearby sub-clade, indicating that this gene originated before differentiation between monocots and dicots but might have similar functions throughout the plant kingdom.

3.2. TaGW8, is a transcription activator that is preferentially expressed in young spikes and developing grain

The SPL genes were confirmed to be typical transcription factors. Subcellular localization analysis in wheat protoplasts showed that TaGW8-GFP accumulated only in the nucleus,whereas GFP alone was present throughout the cell (Fig. 1-a).Trans-activation activity assays showed that TaGW8 protein had strong transcriptional activation activity in yeast,whereas deletion analysis showed that neither the N terminus nor the SBP domain alone was required for activity, but the C terminus was essential for its trans-activation activity(Fig. 1-b). These results supported a role for TaGW8 in transcriptional regulation of gene expression in the nucleus.

We investigated the temporal and spatial expression patterns of the TaGW8 homoeologs in Chinese Spring through qRT-PCR using genome-specific primers(Table S3).TaGW8-7A was ubiquitously expressed in various tissues,with relatively high transcript levels in developing spikes and young grains.TaGW8-7D had a similar expression pattern to TaGW8-7A,but with lower expression in stems and leaves during the heading stage. TaGW8-7B was preferentially expressed in developing spikes and young grains, with the highest transcript level in developing spikes 1 cm in length, whereas the transcripts barely accumulated in other tissues (Fig. 1-c). The different expression patterns of TaGW8-7A/7B/7D suggest that TaGW8 homoeologs have diverse functions in wheat.

3.3. TaGW8 contributes to vegetative and reproductive development in transgenic Arabidopsis

Considering that TaGW8-7A, TaGW8-7B, and TaGW8-7D are highly conserved in amino acid sequence (＞98% identity； Fig.S3) we chose TaGW8-7A as a representative, and studied its effect by overexpression in Arabidopsis(Fig.2).Overexpression of TaGW8 increased vegetative and reproductive growth,advanced the flowering time and increased the plant size(Fig. 2-b, c). In addition, the number and length of seed pods were increased (Fig. S5) and seed size was larger than that in the wild-type (Fig. 2-d). These results indicated that TaGW8 positively contributed to vegetative and reproductive development in transgenic Arabidopsis.

3.4. TaGW8 overexpression increased TGW and grain size in transgenic rice

We transformed TaGW8 under control of the wheat endosperm-specific promoter of 1Bx7OEinto rice in order to investigate the effect of TaGW8 overexpression on grain development. The grain size of three independent TaGW8-overexpression lines was much larger than that from wildtype plants (Fig. 3-a). There is a significant increase (～15% on average)in TGW in transgenic lines than in the wild-type(Fig.3-b). These increases in grain size and weight were mainly due to significantly increased grain length (Fig. 3-c) rather than grain width (Fig. 3-d). Apparently, there was minor diversity in control of grain shape between TaGW8 and OsGW8, as TaGW8 clearly affected grain length in this study whereas OsGW8 was reported to affect grain width. These results confirmed that TaGW8 affects grain size and TGW providing the evidence that it could be involved in determination of yield traits in wheat.

3.5. TaGW8-7A shows higher sequence variation than TaGW8-7B or TaGW8-7D

We sequenced the coding and promoter regions of TaGW8s in sixty Chinese wheat accessions (Table S4). Fifty polymorphic sites, including 47 SNPs and 3 InDels, were identified within the 9743 bp genomic sequence containing 4624 bp upstream of the ATG start codon of TaGW8-7A (Table S6). Interestingly,only 46 polymorphic sites were found in accessions‘Aifeng 3'and ‘Jimai 19'； the polymorphic sites were closely linked and formed a third haplotype, TaGW8-7A-Hap-3. A 65-bp InDel in the first exon,245 bp downstream of the ATG start codon,was present in ‘Baimaizi' and named TaGW8-7A-Hap-4. The other 57 accessions were placed in two haplotypes by a SNP (C/G)189 bp downstream of the ATG, and were named TaGW8-7A-Hap-1 and TaGW8-7A-Hap-2 (Fig. 4-a). Three functional molecular markers based on polymorphic sites were developed to discriminate the four haplotypes(Fig.4-c).

Fig.2- Overexpression of TaGW8 promotes vegetative and reproductive development in transgenic Arabidopsis.(a) Relative expression level of AtSPL13 and TaGW8 in transgenic Arabidopsis.Data are mean values ± SD of three biological replicates.(b)Plants at 25 days post sowing. (c)Plants at 50 days post sowing.(d) Seeds of the transgenic Arabidopsis.Scale bars,0.5 mm.

The same set of accessions was used to detect sequence variations in TaGW8-7B. Three SNPs (-337T/G, -683-/A, and-2109G/A) were found in the promoter region, and one SNP(4860G/A) was detected in the 3′UTR. A SNP (3792G/T) and a 276-bp InDel were identified in the coding regions. Two haplotypes were present among the 60 accessions (Fig. 4-b).To test whether these six polymorphisms were also tightly associated in other wheat accessions, we developed three molecular markers based on -2109G/A, 276-bp InDel and 4860G/A, and genotyped the Chinese mini core collections(MCC) (Fig. 4-d). These three polymorphisms were tightly associated in other accessions and the two haplotypes were designated TaGW8-7B-Hap-H and TaGW8-7B-Hap-L(Fig.4-b).

We also sequenced the coding and promoter regions of TaGW8-7D in the same accessions, and detected two polymorphic sites in introns of TaGW8-7D in the MC, which occurred in a very low frequency(1.67%)and could not be used in our association analysis (data not shown). Clearly, sequence variations in the TaGW8s differed and TaGW8-7A was the most variable, and TaGW8-7D the least variable. We did not study the latter in subsequent work.

3.6. Sequence variation resulted in different transcriptional activities for TaGW8-7A and TaGW8-7B haplotypes

We compared TaGW8-7A-Hap-1/2, TaGW8-7A-Hap-3, and TaGW8-7A-Hap-4 to explore the effects of sequence variation in TaGW8-7A.The 121 bp deletion in the first exon of TaGW8-7A-Hap-4 was predicted to cause a frame shift, leading to translation termination and a truncated TaGW8-7A without the complete SBP domain. In contrast, a SNP at 3482 bp (G/A)in TaGW8-7A-Hap-3 caused a missense Arg (R) to His (H)mutation at aa 161, the critical site for SBP binding (Fig. S3).Since GW7 is a direct downstream gene of GW8 in rice,approximately 2 kb of the promoters of TaGW7 were isolated from homoeologous group 2 chromosomes. Only TaGW7-2D contained the consecutive GTAC boxes, the cis-element for the SBP domain (Fig. S6). We assumed that TaGW8-7A haplotypes might have different promoter-binding activities for TaGW7-2D and performed promoter-transcription factor interaction assays based on the yeast one-hybrid method.

Fig.3- Overexpression of TaGW8-7A increases grain size and grain weight in transgenic rice.(a) Mature grains from threeindependent TaGW8-overexpressing rice lines(OE1-OE3)and wild-type(WT).Scale bar,5 mm.(b-d)TGW(b), grain length(c)and grain width(d)among transgenic lines and wild-type.Data are mean values ± SD of three biological replicates.*indicates the significant difference from the wild-type at P ＜0.05.

Binding to the TaGW7-2D promoter, represented by the activities of β-galactosidase in yeast harboring TaGW8-7AHap-4,was significantly lower than that in yeast with TaGW8-7A-Hap-1/2 or TaGW8-7A-Hap-3 (P ＜0.05), suggesting that TaGW8-7A-Hap-4 was a loss-of-function mutant. The relative binding activity of TaGW8-7A-Hap-3 was significantly lower than that of TaGW8-7A-Hap-1/2,demonstrating that TaGW8-7A-Hap-1/2 has the strongest transactivation activities among the haplotypes (Fig. 5-a, b). We also carried out promotertranscription factor interaction assays in N. benthamiana leaves and consistent with the results in yeast, the ratios of LUC to REN in tobacco leaves harboring TaGW8-7A-Hap-1/2 were significantly higher than those for the other two haplotypes(Fig. 5-c,d).

To investigate the difference between TaGW8-7A-Hap-1 and TaGW8-7A-Hap-2 we assessed their promoter activities using the modified promoter transcriptional activity assay.The relative luciferase activity in N. benthamiana leaves harboring the TaGW8-7A-Hap-2 promoter was higher than that with the Hap-1 or Hap-3 promoters (Fig. 5-e). Therefore,variation in the promoter and coding regions altered the gene transcript levels and protein transcriptional activity. We concluded that TaGW8-7A-Hap-2 conferred higher transcript and transcriptional activity levels, leading to the strongest positive regulation among TaGW8-7A haplotypes.

The TaGW8-7B polymorphisms were located in the introns and promoter regions, thus the proteins encoded by the two haplotypes were identical. We predicted the ciselements in the promoter regions of the two haplotypes and noted that SNP (-2109G/A) was located in a GT1-like ciselement (Fig. S7). We speculated that the variation in the respective promoters might cause divergent promoter activity and thus lead to different transcript levels in the two haplotypes. To evaluate the possibility we performed transient promoter activity assays in N.benthamiana leaves,as well as GUS activity assays in transgenic Arabidopsis lines expressing GUS driven by the respective promoters of each haplotype. GUS activity in TaGW8-7B-Hap-H transgenic plants was significantly higher than that in TaGW8-7BHap-L plants (P ＜0.05) consistent with the relative LUC activity in N. benthamiana leaves (Fig. 5-e-g). Therefore, the variation resulted in different transcript levels and transcriptional activities in both the TaGW8-7A and TaGW8-7B haplotypes.

Fig.4-TaGW8 haplotype analysis and molecular marker development.Haplotype analysis of TaGW8-7A(a)and TaGW8-7B(b)；Molecular markers discriminating haplotypes of TaGW8-7A(c)and TaGW8-7B(d).

3.7.TaGW8-7A and TaGW8-7B haplotypes have multiple and divergent genetic effects in wheat natural populations

To evaluate the genetic effects of TaGW8s in the wheat populations we conducted an association analysis of TaGW8 haplotypes and agronomic traits in the MCC and MC.Significant differences in TGW, GL, and SPL among TaGW8-7A haplotypes were detected both in populations (Table S7).TaGW8-7A-Hap-2 was associated with significantly higher TGW and larger GL compared with TaGW8-7A-Hap-1 and TaGW8-7A-Hap-3 in the 2005LY and 2010SY trials (Fig. 6-a, b).The average TGW and GL of TaGW8-7A-Hap-2 in the three years were 43.38 g and 0.7 cm,significantly higher than those of TaGW8-7A-Hap-1 (40.36 g and 0.67 cm), TaGW8-7A-Hap-3(39.33 g and 0.66 cm), and TaGW8-7A-Hap-4 (34.68 g and 0.67 cm) (Table S7). TaGW8-7A-Hap-3 was associated with shorter SPL than TaGW8-7A-Hap-1 or TaGW8-7A-Hap-2 in 2005LY and 2010SY (Fig. 6-c). The average SPL of TaGW8-7AHap-3 in the three years was 9.1 cm, significantly less than that of TaGW8-7A-Hap-1 (9.93 cm), TaGW8-7A-Hap-2(10.23 cm) and TaGW8-7A-Hap-4 (10.1 cm). These results suggest that TaGW8-7A-Hap-2 leads to higher TGW and larger GL, and that TaGW8-7A-Hap-3 is associated with shorter SPL,possibly attributable to different transcriptional activities.

Fig.5- Transcription factor binding activity assay and GUS staining in transgenic Arabidopsis.(a) Schematics diagrams of binding activity assays in yeast.(b) Relative β-galactosidase activity assay in yeast.(c)Schematic diagrams of the dualluciferase reporter and effector constructs.(d) Relative LUC activities from transient expression of reporter/effectors in N.benthamiana leaves.(e)Relative LUC activities from transient expression analyses of different promoters of TaGW8-7A and TaGW8-7B.(f)GUS activities of transgenic Arabidopsis expressing GUS driven by the promoters of TaGW8-7B-Hap-H and TaGW8-7B-Hap-L.(g)GUS staining of transgenic Arabidopsis seedlings at 10 days post sowing.Data are mean values ± SD of three biological replicates.* and**,significant at P ＜0.05 and P ＜0.01,respectively.

We genotyped the MCC and MC with the molecular markers developed for TaGW8-7B and performed an association analysis between TaGW8-7B haplotypes and phenotypes.Significant differences were detected in HD, MD, ETN, and TGW between TaGW8-7B-Hap-H and TaGW8-7B-Hap-L in both the MCC and MC (Table S8). In the MC, TaGW8-7B-Hap-H exhibited significantly earlier HD and MD,less ETN and higher TGW than TaGW8-7B-Hap-L； mean differences between the two haplotypes were 2.55, 2.09, and 1.15 days for HD, 2.70,1.52, and 1.30 days for MD, 1.05, 2.13, and 0.81 for ETN, and 0.59, 2.48, and 1.92 g for TGW in 2002, 2005, and 2010,respectively (Fig. 7). The phenotypic difference in TGW between TaGW8-7B-Hap-H and TaGW8-7B-Hap-L in the MCC were 2.89,5.14,and 4.51 g in 2002,2005 and 2010,respectively,consistent with that in MC(Table S8).The different haplotype effects of TaGW8-7B were attributed to different transcriptional effects due to the SNPs in the promoters.

Fig.6- Differences among the four TaGW8-7A haplotypes for TGW(a),GL (b),and SPL(c)in the MC. *and **indicate the significant difference at the level of P ＜0.05 and P ＜0.01, respectively.

Fig.7-Phenotypic difference between TaGW8-7B-Hap-H and TaGW8-7B-Hap-L on HD(a),MD(b),ETN(c),and TGW(d)in MC.*and** indicate the significant difference at P ＜0.05 and P ＜0.01,respectively.

Because TaGW8 haplotypes were associated with multiple and divergent agronomic traits except for TGW,we performed an association analysis between the combined TaGW8-7A/7B haplotypes and phenotypes to explore whether they had an additive effect on TGW. Four combined haplotypes Hap-1H/3H/4H, Hap-2L, Hap-2H, and Hap-1L/3L represented cultivar groups possessing one favorable haplotype(TaGW8-7A-Hap-1/3/4/TaGW8-7B-H, TaGW8-7A-Hap-2/TaGW8-7B-L), two favorable haplotypes(TaGW8-7A-Hap-2/TaGW8-7B-H)and no favorable haplotype (TaGW8-7A-Hap-1/3/TaGW8-7B-L). Hap-2H was associated with higher TGW and Hap-1L/3L was associated with lower TGW than others even though the differences were not significant in all years(Fig.S8).Our data indicated additive effects of TaGW8-7A and TaGW8-7B haplotypes on TGW,although the effect of TaGW8-7B on TGW was not as strong as that of TaGW8-7A.

3.8. Favorable TaGW8 haplotypes have potential for yield improvement at the global level

Favorable alleles or haplotypes associated with high yield traits are subject to selection in breeding.We investigated the geographic distribution of TaGW8-7A haplotypes in Chinese wheat landraces and modern cultivars.The favorable TaGW8-7A-Hap-2 haplotype had obviously been positively selected during breeding in all production zones, except X and VI,consistent with the fact that breeders have targeted larger grain size and higher TGW over many years (Fig. 8-a, b). The frequency of TaGW8-7A-Hap-3, corresponding to shorter SPL,was positively selected in zones I, II, III, IV, and VIII,suggesting that compact spike type also was a breeding target in those zones. Correspondingly, the frequencies of TaGW8-7A-Hap-1/4 decreased as a result of their relationship with poor yield traits. Interestingly, TaGW8-7A-Hap-4, a local rare allele in zone IV, appeared in currently grown modern cultivars, maybe a reflection of the wide geographic sources of germplasm used in wheat breeding(Fig. 8-b).

We further evaluated the global distribution frequencies of TaGW8-7A haplotypes in cultivar collections from North America, CIMMYT, Europe, the former USSR, and Australia(Table S2). TaGW8-7A-Hap-4 was present as a minor variant only in China (Fig. S9-a). The frequency of TaGW8-7A-Hap-2 was 15.80% in North America, 18.18% in CIMMYT, 4.15% in Europe, 2.57% in former USSR, 7.81% in Australia, and 14.04%in China (Fig. S9-a). The low frequency of TaGW8-7A-Hap-2 in the global wheat collections suggest that it holds potential worldwide for its selection to improve yield based on higher TGW and larger grain size.

We also investigated the frequency changes of TaGW8-7BHap-H and TaGW8-7B-Hap-L in China. The frequency of TaGW8-7B-Hap-L was 35.07% in landraces, and 12.36% in modern cultivars, whereas the frequency of TaGW8-7B-Hap-H was 64.93% in landraces and 87.64% in modern cultivars,indicating positive selection in the breeding program. This selection had occurred in all wheat production zones,especially in multiple cropping system regions with short wheat season in South China (Fig. 8-c, d). The frequencies of TaGW8-7B-Hap-H in the global populations were 91.70%,86.72%, 100.00%, 93.82%, 100.00%, and 87.64% in North America, CIMMYT, Europe, former USSR, Australia, and China,respectively(Fig.S9-b).Thus TaGW8-7B-Hap-H predominates worldwide and likely underwent selection in breeding.

4. Discussion

4.1. TaGW8 facilitates vegetative and reproductive development

In the past two decades, SPL genes have emerged as pivotal regulators of diverse biological processes in plants, including the timing of vegetative and reproductive phase change, leaf development, tillering/branching, plastochron, panicle/tassel architecture, fruit ripening, fertility, and response to stresses[25,38]. OsGW8 (OsSPL16), a typical transcriptional activator,has a role in affecting rice yield by regulating genes involved in the cell cycle and SPL-targeted MADS-box genes[20].OsGW8 represses expression of OsGW7, up-regulation of which results in increased cell length versus width, representing a strategy to simultaneously improve yield and grain quality in rice [28,29]. In Arabidopsis, SPL13, the orthologous gene of OsSPL16, promotes vegetative phase changes, inhibits adventitious root development, and participates in floral induction with other SPL genes by promoting expression of flowering genes such as SOC1,LFY,AP1,and FUL[25,39,40].

In this study, we cloned TaGW8 homoeologs and characterized the molecular functions. Like OsGW8, TaGW8 is a transcriptional activator and activates transcription of TaGW7-2D (Figs. 1, 5). TaGW8 overexpression in Arabidopsis and rice promoted vegetative and reproductive development,and increased seed size and weight, providing evidence that TaGW8 has a conserved function in facilitating plant development (Figs. 2, 3). Similarly, a recent study revealed that ectopic expression of TaSPL16 in Arabidopsis caused early flowering, increased organ size, and affected yield-related traits, demonstrating regulatory roles of TaSPL16 in plant growth and development as well as seed yield[30].Combining these results with the expression patterns of TaGW8s, we conclude that TaGW8s might be positive regulators in determination of yield in wheat, especially by regulating grain size and flowering time, characteristics that are consistent with the function of OsGW8 in rice[25,28,29].

4.2. Sub-functionalization of TaGW8 homoeologs results in different selections in wheat breeding

During evolution, orthologous genes in different species evolve from a single gene in an ancient species. Due to their highly conserved sequence and structure, orthologs tend to have similar functions[41-43].The most conspicuous domestication transition, from shattering to non-shattering, is attributable to the domestication gene Shattering 1, which encodes a YABBY transcription factor that was first identified in sorghum (Sorghum bicolor L.) [44]. Variation in Shattering 1 resulted in non-shattering phenotypes, and its orthologs in rice (OsSh1), maize (ZmSh1-1 and ZmSh1-5.1 + ZmSh1-5.2) and foxtail millet (Setaria italica (L.) Beauv.) (FmSh1) have been verified as the key regulators in shattering [44-47]. Another critical gene that underwent parallel domestication in cereals is Heading Date 1(HD1).Reshaping the HD1 locus has played a major role in the adaptation of rice, sorghum and foxtail millet due to its critical regulation of photoperiod response[48].

Fig.8-Geographic distributions of TaGW8 haplotypes in Chinese wheat landraces and modern cultivars across the 10 Chinese wheat ecological production zones.Distributions of TaGW8-7A haplotypes in landraces (a) and modern cultivars(b)；distributions of TaGW8-7B haplotypes in landraces(c)and modern cultivars(d).I,Northern Winter Wheat Zone；II,Yellow-Huai River Valleys Facultative Wheat Zone；III, Middle and Lower Yangtze Valleys Autumn-Sown Spring Wheat Zone；IV,Southwestern Autumn-Sown Spring Wheat Zone；V,Southern Autumn-Sown Spring Wheat Zone；VI,Northeastern Spring Wheat Zone；VII, Northern Spring Wheat Zone；VIII,Northwestern Spring Wheat Zone；IX,Qinghai-Tibetan Plateau Spring-Winter Wheat Zone；X,Xinjiang Winter-Spring Wheat Zone.

Many genes in common wheat perform similar functions to their orthologs in rice； for example TaGW2, TaGS5, TaBT1,and TaAGP-L/S1 [12-14]. The three TaGW2 homoeologs play parallel roles in the development of grain size, which negatively regulates grain weight [14]. In contrast, TaGW8s are multi-functional transcription factors that affect yield in multiple ways； for example, TaGW8-7A mainly controls grain traits (TGW and GL), whereas TaGW8-7B not only regulates grain weight,but also influences other yield-related traits(HD,MD, and ETN) (Figs. 6, 7). The different genetic effects of TaGW8-7A and TaGW8-7B indicate sub-functionalization,which has been reported for gene sets such as LEAFY HULL STERILE1 (WLHS1) and TaCPK2 [16,17]. Gene editing and mutagenesis methods have been applied in functional analysis of wheat homoeologous genes [49]. These methods will be used to confirm phenotypic effects and subfunctionalization of TaGW8 homoeologs in future work.

A recent GWAS identified TaGW8-B1 as a candidate gene for regulation of grain traits, and cultivars with TaGW8-B1a possessed significantly wider grain, longer grain, higher grain number per spike,higher thousand-grain weight,and higher spikelet number per spike than those with TaGW8-B1b[50].We compared the two haplotypes of TaGW8-7B with the reported TaGW8-B1a/b and found that the favorable TaGW8-7B-Hap-H haplotype in our study was not the favorable TaGW8-B1a, but was in fact the less favorable allele in that research[50].This inconsistency might be due to the different populations used in the association analyses. We found that TaGW8-7A had more significant effects on grain traits than TaGW8-7B, whereas TaGW8-7B was associated with more agronomic traits,such as HD,MD,and ETN (Figs. 6, 7). Therefore, this functional divergence between TaGW8-7A and TaGW8-7B resulted in stronger positive selection for TaGW8-7B-Hap-H than TaGW8-7AHap-2 in Chinese and global wheat cultivars, providing new evidence for the sub-functionalization of homoeologs(Figs.8,S9).

4.3. TaGW8 has the potential for high yield in global wheat breeding regions

Breeders always select alleles associated with superior phenotypes including high adaptation for the local conditions,leading to clear footprints in the genome[51-53].Alleles Wxaand Wx in rice are associated with different amounts of amylase. In the so-called Glutinous Rice Zone of southern Asia where waxy rice is preferred over non-waxy rice, the Wxb-derived allele that confers more amylopectin is predominant indicating preferential selection for Wxbin rice breeding in that region[54,55].

Marker-assisted selection is a cost-effective practice to improve genetic composition and it allows progress toward the concept of‘breeding by design'[56,57].The rice cultivar Zhongke 804 was recently popularized in northern China.This cultivar combined several favorable alleles (IPA4,Qsw5, GS3, Ghd8, TAC1, SSII-1, DEP1, and SBE1) along with ideal plant architecture, high yield, and superior quality[58]. This approach was an example of the successful application of molecular breeding and demonstrated that rational design is a powerful strategy for meeting the challenges of future crop breeding, particularly for pyramiding multiple complex traits [59]. Here, we developed molecular markers based on variation in TaGW8s,and these markers offer the potential for use in various wheat breeding regions. Markers for increasing grain size and weight based on TaGW8-7A-Hap-2 can be applied worldwide with the expectation of considerable yield gains in view of its current low frequency among leading cultivars,whereas markers based on TaGW8-7B-Hap-H could be used to improve overall yield per unit area in multiple croppingsystem regions with short wheat season (Figs. 8, S9).Previous studies have demonstrated that marker-assisted selection based on multiple markers is more effective than selection based on a single marker in yield improvement[13,14,49]. Hence, pyramiding the favorable haplotypes of TaGW8s and other yield-related genes should be beneficial for future yield improvement in wheat breeding.

5. Conclusions

TaGW8 is a conserved transcriptional activator in affecting plant vegetative and reproductive development. TaGW8-7AHap-2 and TaGW8-7B-Hap-H were identified as the favorable haplotypes associated with superior yield-related traits due to their higher transcription levels. Sub-functionalization of TaGW8 provides multiple candidate genes to meet the variable demands of breeding wheat for high yield.

Supplementary data for this article can be found online at https://doi.org/10.1016/j.cj.2019.09.006.

Acknowledgments

We thank Dr.Jianmin Wan and Mrs.Xiuping Guo,Institute of Crop Sciences, CAAS, for help with rice transformation. This work was supported by the National Key Research and Development Program of China (2016YFD0100402), the National Natural Science Foundation of China (31671687), the National Major Project for Developing New GM Crops(2016ZX08009001), and Agricultural Science and Technology Innovation Program of CAAS.