Development of a new set of molecular markers for examining Glu-A1 variants in common wheat and ancestral species

In common wheat (Triticum aestivum L.), allelic variations of Glu-A1 locus have important influences on grain end-use quality. Among the three Glu-A1 alleles, Glu-A1a and -A1b encode the high-molecular-weight glutenin subunits (HMW-GSs) 1Ax1 and 1Ax2*, respectively, whereas Glu-A1c does not specify any subunit. Here, we detected a total of 11 Glu-A1 locus haplotypes (H1 to H11) in three wheat species, by developing and using a new set of DNA markers (Xrj5, Xid3, Xrj6, Xid4 and Xrj7). The main haplotypes found in the diploid wheat T. urartu were H1, H4, H5 and H6, with H1 and H4 expressing both 1Ax and 1Ay subunits. The major haplotypes revealed for tetraploid wheat (T. turgidum) were H1, H8 and H9, with the lines expressing both 1Ax and 1Ay belonging to H1, H4 or H7. Four major haplotypes (H1, H9, H10 and H11) were discovered in common wheat, with Glu-A1a associated with H1 and H8, Glu-A1b with H10 or H11, and Glu-A1c with H9. The Glu-A1 locus haplotypes and the new set of DNA markers have potential to be used for more effectively studying and utilizing the molecular variations of Glu-A1 to improve the end-use quality of common wheat are discussed.


Introduction
Common wheat (Triticum aestivum, 2n = 42, AABBDD) is one of the most important staple food crops in the world, providing 20% of dietary energy and protein sources for over 60% of the world population [1,2]. T. aestivum was originated about 10,000 years ago due to natural hybridization between tetraploid wheat (T. turgidum, 2n = 28, AABB) and the D genome donor Aegilops tauschii (2n = 14, DD) [3,4]. T. turgidum was derived from the hybridization between T. urartu (2n = 14, AA) and an unknown species of the Sitopsis section about 0.5 a1111111111 a1111111111 a1111111111 a1111111111 a1111111111 million years ago [5]. Both T. urartu and T. turgidum contain the A genome, which is closely related to the A genome present in bread wheat.
Compared with other major cereal crops (such as rice and maize), wheat is unique in that its flour can be processed into multiple types of food products (e.g., bread, noodles and cakes). This is owing to the formation of wheat doughs with different viscoelastic properties [6]. It is now well known that wheat dough properties are mainly determined by three families of gluten proteins, high-molecular-weight glutenin subunits (HMW-GSs), low-molecular-weight glutenin subunits (LMW-GSs) and gliadins, present in the flour [6,7]. HMW-GSs are the major determinant of dough elasticity, whereas LMW-GSs tend to contribute to both dough elasticity and extensibility [7,8]. Gliadins may contribute to dough extensibility [9]. In common wheat, the genes encoding HMW-GSs are contained in three homoeologous Glu-1 loci (Glu-A1, -B1 and -D1) located on homoeologous group 1 chromosomes [8]. In each locus, there are two duplicated HMW-GS genes, designating as x-and y-type subunits, respectively. The HMW-GS proteins that have been characterized to date all share a conserved primary structure composed of a signal peptide (removed from mature subunit), a N-terminal domain, a central repetitive domain, and a C-terminal domain [8]. The N-, C-and central repetitive domains all contribute to the function of HMW-GSs in end-use quality control [8,10].
Within each Glu-1 locus, the two HMW-GS genes are usually separated by a large distance (54-190 kb), with multiple additional genes and transposable elements (TEs) distributed in the vicinity of x-and y-type genes [11][12][13]. Comparisons of allelic Glu-1 loci have revealed nucleotide sequence variations in not only HMW-GS genes but also the surrounding genes and TEs [11][12][13]. However, at present, only nucleotide sequence variations in HMW-GS genes are considered in differentiating different Glu-1 alleles. These variations frequently lead to the expression of allelic x-and y-type subunits differing in electrophoretic mobility on SDS-PAGE, or the silencing of one or more HMW-GS genes in certain genotypes. Consequently, Glu-A1, -B1 and -D1 have all been found to have multiple alleles, and Glu-B1 has the most alleles, followed by Glu-D1 and Glu-A1 [14]. Importantly, Glu-1 alleles have been found to differ considerably in their effects on dough functionality and end-use quality. For example, in Glu-D1, the Glu-D1d allele (encoding 1Dx5 and 1Dy10 subunits) is functionally superior to Glu-D1a (specifying 1Dx2 and 1Dy12) [15][16][17].
To date, three Glu-A1 alleles have been defined in bread wheat; Glu-A1a (expressing 1Ax1 subunit) and Glu-A1b (Specifying 1Ax2 Ã ) alleles have positive effects on breadmaking quality whereas Glu-A1c (not expressing any HMW-GS because of gene silencing) has a low quality score [9,15]. One common feature shared by the three alleles is that the gene encoding y-type subunit is silenced [11,[18][19][20]. Interestingly, this gene is active in some T. urartu and T. turgidum accessions [21][22][23][24][25][26]. However, in these species, the Glu-A1 genomic region is still poorly understood, and the 1Ax and 1Ay genes in this species are largely uncharacterized at the molecular level. Moreover, the allelic differentiation of Glu-A1 in T. urartu and T. turgidum populations has not been systematically studied using molecular markers, although SDS-PAGE data have shown that the expression of 1Ax and 1Ay genes varied extensively within and between the two species [24,26].
Because allelic variation of Glu-1 loci affects their function in end-use quality control, substantial efforts have been devoted to mine functionally superior Glu-1 alleles in wheat and related species [27][28][29][30]. In a previous study, we found that DNA marker-mediated haplotype analysis was highly efficient for revealing the molecular variations of Glu-D1 genomic region, tracing the origin of the superior Glu-D1d allele, and guiding the mining of more Glu-D1d variants in specific wheat species [31]. Therefore, the major objectives of this study were to develop a new set of DNA markers for Glu-A1 genomic region and to use them for investigating the haplotype variants of Glu-A1 locus. The relationships between Glu-A1 locus haplotype variants and the expression of 1Ax and/or 1Ay subunits were also explored. Our data should facilitate more effective study and utilization of Glu-A1 variants in improving the end-use quality of common wheat in further research.

Plant materials
A total of 99 T. urartu accessions (S1 Table), 95 T. turgidum lines (S2 Table) and 215 common wheat varieties (S3 Table) were employed in this study. For validating the specificity of newly developed Glu-A1 DNA markers, the nulli-tetrasomic line N1AT1D and the ditelosomic line Dt1AS of Chinese spring (CS) [32,33] were applied.

Marker development and haplotype analysis
The BAC sequences of DQ537335 (from common wheat variety Renan, T. aestivum), AY494981 (from durum wheat variety Langdon, T. turgidum) and JQ240472 (from T. urartu accession G1812), each carrying a completely sequenced Glu-A1 locus, were retrieved from the NCBI database (http://www.ncbi.nlm.nih.gov/). TE insertions and deletions and short nucleotide indels in the three sequences were considered for marker development. The software primer premier 5.0 (PREMIER Biosoft International, USA) was used for designing the primer pairs specific for each of the five Glu-A1 markers (S4 Table). To verify the A genome specificity of the designed primers, the BAC sequences from Renan Glu-B1 (DQ537336) and Glu-D1 (DQ537337) loci were also used for sequence alignment. Genomic DNA samples were extracted from desired plant lines grown in the greenhouse using a cetyltrimethyl ammonium bromide (CTAB) method [34], and used for PCR amplifications. Glu-A1 locus haplotypes were based on differences in the alleles of the five diagnostic markers.
PCR and electrophoresis conditions for five Glu-A1 markers PCR was carried out in 20 μl volume containing 50 ng genomic DNA template, 10 mM dNTPs, 5 pmol of each primer, and 1 U Taq polymerase (Transgen Biotech, China). The cycling parameters were 94˚C for 5 min, followed by 35 cycles of 94˚C for 30 s, 60˚C for 30s and 72˚C for 1 min, and a final extension at 72˚C for 5 min. PCR products were separated in 1.5% agarose gels.
Cloning the nucleotide sequences of the 1Ay gene of T. urartu accession PI428339 The primers and PCR conditions described by Bai et al. 2004 [39] were used to amplify the complete coding sequence of 1Ay. The target PCR fragment was isolated from agarose gel, and then cloned into the pGEM-T Easy vector (Promega, USA). Three independent clones were sequenced. For multiple sequence alignment among the 1Ay cloned here and those published previously, two 1Ay pseudogenes (GenBank accessions AY245579 and EU984505) and five 1Ay active genes (AY245578, EU984503, EU984504, FJ404595 and JQ240472) isolated from different T. urartu materials were retrieved from the NCBI database (https://www.ncbi.nlm. nih.gov/).

Phylogenetic network analysis
The alleles of the five Glu-A1 markers were employed for phylogenetic network analysis of Glu-A1 locus haplotypes using the software network 4.6.1.3 (Fluxus Technology Ltd., UK).

Development of a new set of DNA markers for Glu-A1 locus
The genomic sequences, available for Glu-A1 regions in the T. urartu accession G1812, the durum wheat variety Langdon and the common wheat variety Renan [11][12][13], permitted a detailed comparison of the structural variations of this locus among A genomes from three species. The Glu-A1 allele of Renan encoded 1Ax2 Ã [40] and thus belonged to Glu-A1b; the Glu-A1 allele of Langdon was Glu-A1c that encodes a null allele; the Glu-A1 allele of G1812 expressed both 1Ax and 1Ay proteins, and thus differed from the three Glu-A1 alleles defined previously [9,15]. The patterns of TE insertion and deletion varied considerably among the three Glu-A1 regions [11][12][13]. These variations permitted us to develop a new set of DNA markers for investigating the haplotype variants of Glu-A1 locus ( Fig 1A).
Xid3 and Xid4 were two indel markers. Xid3 was developed based on the polymorphism in the coding sequences of the three 1Ay genes. A 63 bp indel was found between the 1Ay coding regions of Renan and G1812 (S1 Fig), while a LTR retrotransposon (Wis-3, 8633 bp) was observed in the 1Ay open reading frame (ORF) of Langdon ( Fig 1A). Xid4 was developed based on a 18 bp deletion in the 1Ax ORF of Renan (S2 Fig, Fig 1A). This deletion did not occur in the 1Ax gene of G1812 and Langdon. Moreover, the forward and reverse primers used for amplifying Xid3 or Xid4 did not have identical binding sites in either Glu-B1 or Glu-D1 regions (S1 and S2 Figs), indicating Xid3 and Xid4 should be Glu-A1 specific.
Xrj5, Xrj6 and Xrj7 were three repeat DNA insertion site based polymorphism (ISBP) markers [41,42]. Xrj5 was developed to diagnose the presence of an intact miniature inverted repeat transposable element (Mite-5) in the ORF of the receptor kinase b gene in Renan, which was truncated in Langdon and absent in G1812 ( Fig 1A, S3 Fig). Xrj6 resided between 1Ay and 1Ax genes, and was designed based on the polymorphisms exhibited by the retroelement Hawi-2; this element was relatively intact in Langdon, but was partially deleted (resulting in Hawi-2p) in Renan and completely absent in G1812 ( Fig 1A). Xrj7 was located downstream of 1Ax, and designed based on another retrotransposon Nobude-1 ( Fig 1A). This marker should not amplify PCR products from Langdon owing to the insertion of Erika-2 (13651 bp) upstream of Nobude-1 ( Fig 1A). Mite-5, Hawi-2 and Nobude-1 were absent in the Glu-B1 or Glu-D1 regions, indicating that the three ISBP markers were likely Glu-A1 specific.
The A genome specificity and ability of the five developed markers to reveal potential haplotype variations of Glu-A1 region was tested using three common wheat varieties carrying Glu-A1a (Xiaoyan 54), -A1b (Renan) and -A1c (Chinese spring) alleles, respectively. In addition, three tetraploid wheat lines (Langdon, Stewart and WEW270) and three T. urartu accessions (PI428281, G1812 and PI428204) were also included in the test. From Fig 1B, it is clear that the amplicons of the five markers were all polymorphic. No products were amplified in the line lacking chromosome 1A (N1AT1D) or 1AL (Dt1AS) (Fig 1B), which validated the location and specificity of the five markers on 1AL chromosome arm.
Xrj5 yielded three types of amplicons, i.e., 882 bp from Renan, 719 bp from Langdon and 594 bp from G1812 ( Fig 1B, S3 Fig). The 719 bp amplicon was also amplified in Chinese Spring, and the 594 bp fragment was also detected in Xiaoyan 54 and WEW 270 ( Fig 1B). Xid3 amplified a null allele in Langdon, a 718 bp fragment in Renan, and a 665 bp fragment in G1812; the null allele was also found in Chinese Spring and Stewart, but an unexpected allele of 808 bp was detected in PI428204 ( Fig 1B). This 808 allele was caused by an insertion in the 1Ay gene of PI428204 (S1 Fig). Xrj6 amplified the expected fragment (797 bp) in Langdon and a null allele in Renan and G1812; the 797 bp fragment was also detected in Chinese Spring, Xiaoyan 54, Stewart and WEW270 ( Fig 1B). The amplicon of Xid4 was either 158 or 176 bp New molecular markers for examining Glu-A1 variants due to a 18 bp deletion in the 1Ax gene of Renan (Fig 1B, S2 Fig). Lastly, Xrj7 amplified the anticipated fragment (961 bp) in Renan and G1812 and a null allele in Langdon, with the 961 bp fragment being also amplified in Xiaoyan 54, WEW270 and two other T. urartu accessions ( Fig 1B). Together, these results suggested that the five markers could reveal the polymorphisms existed in Glu-A1 genomic region in diploid, tetraploid and hexaploid wheat species, and were thus useful for analyzing the haplotype variations of Glu-A1 locus.

Analysis of Glu-A1 locus haplotypes in T. urartu
Ninety-nine T. urartu accessions from Lebanon and Turkey were genotyped and seven Glu-A1 locus haplotypes were identified. These haplotypes were designated as H1 to H7, respectively (Table 1, S1 Table). In the examined T. urartu lines, both Xrj5 and Xrj7 were monomorphic (with the amplicon sizes of 594 and 961 bp, respectively), whereas Xid3 derived from 1Ay gene (Fig 1A) showed the highest level of polymorphism with three different amplicons (Table 1). Xrj6 and Xid4 each yielded two alleles (Table 1).
Among the seven haplotypes found in T. urartu, H1 was the most common that was found in 30% of the lines, followed by H4 (detected in 27% of the lines). The only difference between H1 and H4 was the presence of Xrj6 but absence in H4 ( Table 1). The amplicon size of Xid4 was 158 bp in H7 but 176 bp in H1 to H6 (Table 1). Regarding the three amplicons of Xid3, the 718 bp fragment was shared by H1, H4 and H7, the 808 bp fragment by H2 and H5, and the 655 bp fragment by H3 and H6 (Table 1). H1, H4, H5 and H6 were considered as major haplotypes because each was found in more than 15 different T. urartu accessions (Table 1).
HMW-GS composition was analyzed in the 99 T. urartu accessions carrying different Glu-A1 locus haplotypes using SDS-PAGE (Fig 2). 1Ax protein was expressed in all 99 lines, whereas 1Ay was detected in 77 lines (S1 Table). Both 1Ax and 1Ay exhibited variations in electrophoretic mobility, with the magnitude of the variations being considerably larger for New molecular markers for examining Glu-A1 variants 1Ay (Fig 2). Among the four major haplotypes (H1, H4, H5 and H6), H1 and H4 contained the accessions that expressed both 1Ax and 1Ay, with 1Ay silenced most frequently in the accessions of H5 (S1 Table).
Interestingly, the 1Ay protein detected in the accession PI428339 (tentatively designated as 1Ay-PI428339) displayed the slowest electrophoretic mobility (Fig 2, arrowed). We thus cloned the coding region of 1Ay-PI428339 (GenBank accession number MF098417) by PCR, and found that the amplicon was 1899 bp. Analysis of translated amino acid sequence showed

Glu-A1 locus haplotypes in common wheat
Five haplotypes (H1, H8, H9, H10 and H11) were identified among 215 common wheat varieties after genotyping with the five Glu-A1 markers ( Table 1, S3 Table). H1 (in 54.4% of the varieties) was the predominant haplotype in the common wheat lines examined, followed by H9 (in 36.7% of the varieties). H10 (in 6.3% of the varieties) and H11 (in 2.3% of the varieties) were new haplotypes found in only common wheat, and the only difference between the two haplotypes was the amplicon size of Xrj5, which was 719 bp in H10, but 882 bp in H11 (Table 1). Together, the lines carrying H1 or H9 accounted for 91.2% of the common wheat varieties examined ( Table 1).
The Glu-A1 alleles and composition of HMW-GSs were investigated in the 215 common wheat lines. The Glu-A1 alleles (i.e., Glu-A1a, -A1b and -A1c) in these lines were readily identified following the standard established previously (Payne and Lawrence 1983) (S3 Table, Table, (Figs 2 and 3). However, strong correspondence was found between Glu-A1 allele and Glu-A1 locus haplotype (S3 Table). The varieties carrying Glu-A1a generally had H1, with only one of them belonging to H8; the varieties with Glu-A1b had either H10 or H11, whereas those possessing Glu-A1c all had H9 (S3 Table).

Network analysis of Glu-A1 locus haplotypes
Based on similarities and differences in the alleles of the five DNA markers, phylogenetic network analysis was conducted to investigate possible relationships among the 11 haplotypes. As displayed in Fig 4, complex network connections were found among H1, H2, H3, H4, H5 and H6, but the connections among H1, H8 and H9, as well as those among H7, H10 and H11, were relatively simple. Interestingly, H1, H8, H9, H10 and H11, but except for H7, were all detected in common wheat. The connection among H1, H8 and H9 established the connection between Glu-A1a and Glu-A1c (Fig 4). The connection between H10 and H11 paralleled with the finding that both haplotypes carried Glu-A1c (Fig 4, Table 1).

Molecular variations of orthologous Glu-A1 loci in T. urartu, T. turgidum and common wheat
Previous knowledge on the sequence variations among orthologous Glu-A1 loci in T. urartu, T. turgidum and common wheat was very limited because it was based on the information from only three genotypes (i.e., G1812, Langdon and Renan) [11][12][13]. Judging from the genotyping data obtained in this work for multiple T. urartu, T. turgidum and common wheat lines, we suggest that the sequence variations of Glu-A1 locus are much more extensive than previously revealed through BAC sequencing at all three ploidy levels. For example, the generation of three different amplicons by Xid3 in T. urartu (655, 718 and 808 bp, Table 1) indicates the presence of at least three 1Ay alleles in this species. The presence of premature stop codon in these alleles confers additional variations in 1Ay expression. The positive amplification of Xrj6 in many T. urartu accessions (Table 1) indicates the occurrence of Hawi-2 element in this species although it was not found in the Glu-A1 locus of G1812 [13]. In T. turgidum, although the amplification of a 719 bp fragment by Xrj5 is in agreement with the presence of a truncated Mite-5 insertion in Langdon (Fig 1A), the production of a 594 bp fragment by the same marker indicates the absence of Mite-5 insertion in some other accessions of this species (Table 1). The lack of amplification by Xid3 in many T. turgidum lines is consistent with the insertion of Wis-3 element in the 1Ay gene of Langdon (Fig 1A, Table 1). However, this marker also amplified a 718 bp fragment in T. turgidum (Table 1), indicating the existence of the 1Ay gene without Wis-3 insertion in this species. Similarly, the lack of amplification by Xrj7 in 48.4% of T. turgidum accessions is in line with the insertion of Erika-2 element in Langdon (Fig 1A, Table 1). But this marker also amplified a 961 bp fragment in 51.6% of the T. turgidum lines ( Table 1), indicating that the lack of Erika-2 insertion is prevalent in T. turgidum. In common wheat, the amplification of the 882 bp fragment by Xrj5 is in agreement with the insertion of an intact Mite-5 element in the receptor kinase b gene of Renan (Fig 1A, Table 1). However, this marker also yielded the 594 and 719 fragments (Table 1), which indicate the lack of Mite-5 insertion, or the presence of an internally deleted Mite-5, in the receptor kinase b gene in common wheat.
The sequence variations of Glu-A1 in T. turgidum and common wheat may come mainly from two sources. One source of variations is derived from the ancestral species, and the other is newly evolved. For instance, the 594 bp allele of Xrj5 is present in all three wheat species examined here, suggesting the transmission of this allele from T. urartu to T. turgidum and from T. turgidum to common wheat. On the other hand, it is likely that the 719 bp and 882 bp alleles of Xrj5 were formed in T. turgidum and common wheat, respectively. This may not be surprising as it is now generally accepted that polyploidization promotes the differentiation of new alleles in the evolution of many crop plants including common wheat [43,44].
Clearly, the use of multiple markers enables rapid assessment of the molecular variations of Glu-A1 genomic region within and among different wheat species. The rich sequence variations of orthologous Glu-A1 loci facilitated our haplotype analysis. They may also aid further basic and applied studies of these loci in the future (see below).

Relationships among Glu-A1 locus haplotypes in three wheat species
A total of 11 haplotypes were identified for Glu-A1 locus in the T. urartu, T. turgidum and common wheat accessions analyzed in this study (Table 1). Of the seven Glu-A1 haplotypes of T. urartu, H1 was shared with T. turgidum and common wheat, and H4 and H7 were also found in T. turgidum. These data suggest that T. urartu haplotypes H1, H4 and H7 may play an important role in shaping the Glu-A1 haplotypes of T. turgidum and common wheat. For both T. turgidum and common wheat, their H1 haplotype was likely derived from T. urartu. This is consistent with the finding of H1 as a major Glu-A1 haplotype in all three species (Table 1). Moreover, H1 might also give rise to H8 (Fig 4). T. urartu haplotype H4 was involved in the differentiation of H7, and H7 contributed to the differentiation of H10 (Fig 4). Because H8 and H9 first appeared in T. turgidum, they were likely formed in this species, and transmitted to common wheat via hexaploidization. Since H10 and H11 were unique to common wheat, they were probably differentiated at the hexaploid stage. Finally, H2, H3, H5 and H6, detected in only T. urartu (Table 1, Fig 4), might be more ancient, and contributed to the evolution of H1 and H4.
It is interesting to note that four T. urartu (H2, H3, H5 and H6) and two T. turgidum (H4 and H7) Glu-A1 locus haplotypes were not present in common wheat (Table 1, Fig 4). The six haplotypes might be lost during the hexaploidization process leading to common wheat. The reduction of Glu-A1 haplotype diversity in common wheat is paralleled with a severe decrease in the expression of 1Ax and 1Ay alleles in this species (Figs 2 and 3, S1 Fig). These reductions may be caused by two main reasons: polyploidy diversity bottlenecks that have been found to act in the evolution of diverse polyploid plants and artificial selections during domestication and modern breeding [45][46][47].
Association between Glu-A1 locus haplotypes and the expression of 1Ax and 1Ay subunits or the three Glu-A1 alleles of common wheat In common wheat, 1Ay subunit is always silent, and 1Ax sometimes appear silent [11,[18][19][20]. Although the expressed 1Ax and 1Ay subunits in some T. urartu and T. turgidum accessions have been reported [21][22][23][24][25][26], expression of 1Ax or 1Ay was unpredictable up to now. In this study, we found strong association between Glu-A1 locus haplotypes and the expression of 1Ax and 1Ay subunits or the three Glu-A1 alleles of common wheat. When considering the data from all three species, the association between Glu-A1 locus haplotypes and the expression of 1Ax and 1Ay subunits appears very complex (S5 Table). For example, among the lines with haplotype H1, some expressed both 1Ax and 1Ay, whereas others specified only 1Ax or none (S5 Table). Nevertheless, within each species, there are some clear associations between specific Glu-A1 locus haplotypes and subunit expression patterns. In T. urartu, three main haplotypes showed exclusive (H1 and H4) or predominant (H6) associations with the expression of both 1Ax and 1Ay subunits. In common wheat, H1 and H8 were both associated with the expression of 1Ax1, while H10 and H11 were both linked with 1Ax2 Ã expression; on the other hand, H9 was tightly associated with the silencing of both 1Ax and 1Ay subunits (S5 Table). Because the number of T. turgidum lines examined for HMW-GS composition in this work was limited, further study is needed to investigate potential associations between specific Glu-A1 locus haplotypes and subunit expression patterns in this species.
There also exist strong correspondences between Glu-A1 locus haplotypes and the three Glu-A1 alleles of common wheat, i.e., H1 and H8 to Glu-A1a, H9 to Glu-A1c, and H10 and H11 to Glu-A1b. We were unable to assess such correspondences for T. urartu and T. turgidum because systematic schemes for designating the Glu-A1 alleles in these two species are still not available. This may be due to the fact that the molecular variations of 1Ax and 1Ay subunits in T. urartu and T. turgidum are considerably more complex than those observed in common wheat.
Because of the complexity in 1Ax and 1Ay expression in T. urartu and T. turgidum, it is difficult to systematically assign alleles to the Glu-A1 variants in different materials of the two species based on electrophoretic mobility variations of 1Ax and 1Ay subunits. In this situation, Glu-A1 locus haplotypes, as revealed by this work, may provide a convenient way to describe the Glu-A1 variants in specific T. urartu and T. turgidum genotypes.

Implications for further research
First, the Glu-A1 locus haplotypes may be useful for studying the evolutionary origins of the three Glu-A1 alleles of common wheat. Although Glu-A1a, -A1b and -A1c have been extensively used in contemporary wheat quality breeding programs, their evolutionary origins are still unclear. This is not conducive for mining additional elite Glu-A1 alleles to aid wheat quality improvement. Based on network relationships of the 11 Glu-A1 locus haplotypes (Fig 4), and the known evolutionary relationships among T. urartu, T. turgidum and common wheat [3,4], we hypothesize that Glu-A1a was originally formed in T. urartu, and then transmitted to common via two rounds of allopolyploidization. This is supported by the finding of H1 in all three wheat species and the strong association of Glu-A1a with H1 in common wheat (Table 1, Fig 4). Glu-A1b might also be initially formed in T. urartu, and then passed to T. turgidum through tetraploidization and common wheat via hexaploidization. This is supported by 1) the presence of the 158 bp amplicon of Xid4 diagnostic for Glu-A1b in all three wheat species, and 2) the Glu-A1 locus haplotypes carrying Glu-A1b were found in T. urartu (H7), T. turgidum (H7) and common wheat (H10 and H11). In contrast to Glu-A1a and -A1b, Glu-A1c might be differentiated in T. turgidum and transmitted to common wheat through hexaploidization. This is because H9, which carried Glu-A1c in common wheat, was also detected in T. turgidum. Furthermore, we speculate that Glu-A1c was probably evolved from Glu-A1a since H8, which directly linked with H9, carried Glu-A1a (Fig 4). Further studies are needed to verify the above hypothesis and speculation.
Second, of the 11 haplotypes, three (i.e., H1, H4 and H7) are useful for introducing active 1Ay alleles into common wheat. It is well known that the 1Ay gene is silenced in common wheat [25,48], but several studies have shown that introducing diploid or tetraploid wheat Glu-A1 alleles expressing both 1Ax and 1Ay subunits into common wheat actually conferred improved end-use qualities [22,23,27]. Here we found that in T. urartu the accessions belonged to H1, H4 and H7 all carried the Glu-A1 alleles with both 1Ax and 1Ay subunits expressed (S1 Table). Notably, in these accessions, the 718 bp allele of Xid3 was strictly linked with actively expressed 1Ay alleles (Table 1), including the 1Ay-428339 allele displaying very slow electrophoretic mobility (Fig 2). The 1Ay-428339 subunit, although resembling its allelic counterparts in the primary structure, displays novel variations in the central repetitive domain and C-terminal region (S4 Fig). Notably, its central repetitive domain was considerably longer than that of other active 1Ay alleles (S4 Fig). Because it has been shown that a longer repetitive domain enhances the function of HMW-GSs in gluten and dough functionalities [49,50], it will be interesting to test the usefulness of 1Ay-428339 for improving wheat end-use quality in the future.
Like in T. urartu, many of the T. turgidum accessions with haplotypes H1, H4 or H7 also expressed both 1Ax and 1Ay subunits (Fig 3). Therefore, the active 1Ay alleles in the T. urartu and T. turgidum accessions with haplotypes H1, H4 or H7 may be transferred to common wheat to improve the end-use traits through marker assisted wide hybridization, an approach that has been found valuable in the molecular breeding programs of diverse crops [51,52].
Finally, based on amplification patterns produced by the five DNA markers (Table 1), we suggest that Xid4 and Xrj7 allow efficient detection and distinction of three Glu-A1 alleles in common wheat. Specifically, Xid4 alone is sufficient for detecting Glu-A1b based on the production of an amplicon of 158 bp. The combined use of Xid4 and Xrj7 permits the distinction of Glu-A1a from Glu-A1c. For Glu-A1a the amplicon lengths of the two markers were 176 and 961 bp, respectively, whereas for Glu-A1c the two markers yielded a 176 bp amplicon and a null allele, respectively (Table 1). Xid4 can be considered as a perfect marker [53], because it resides in the coding region of 1Ax gene (Fig 1A). Xrj7 is also physically close to 1Ax and 1Ay genes ( Fig 1A). Thus the use of Xid4 and Xrj7 can facilitate efficient and accurate selection of three Glu-A1 alleles in wheat quality breeding practice.
In summary, the new set of DNA markers developed in this work is useful for studying the molecular variations and evolutionary mechanisms of orthologous Glu-A1 regions in common wheat and ancestral species. These markers, together with the haplotype variants revealed here, may also aid further research in improving the end-use traits of common wheat.