Introduction

Gluten is a macro polymer that confers to wheat dough its unique viscoelastic properties. The main components of gluten are gliadins and glutenins, two classes of wheat seed storage proteins that, depending on their quantity and quality, greatly influence gluten technological properties (Shewry et al. 2002). Gliadins are monomeric proteins divided into α/β, γ, and ω gliadins depending on their mobility in acid-polyacrylamide gel electrophoresis (Barak et al. 2015). Glutenins are polymeric proteins crosslinked with each other through disulfide bonds. Under reducing conditions, glutenins can be separated based on their molecular weight into high-molecular-weight glutenin subunits (HMW-GSs, 70–90 KDa) and low-molecular-weight glutenin subunits (LMW-GSs, 20–45 KDa) (D’Ovidio and Masci 2004).

Compared to gliadins and LMW-GSs, the HMW-GSs are less abundant, representing only 10% of the total wheat seed storage proteins (Wang et al. 2010). However, their fundamental role in wheat end-use quality has been extensively reported in the literature (He et al. 2005; Jin et al. 2013; Liu et al. 2005; Lukow et al. 1989; Payne et al. 1987) and the HMW-GS allelic profile is often used as a predictor of gluten quality. The LMW-GSs, in contrast, represent around one third of the total wheat seed storage proteins and around 60% of the glutenins (Bietz and Wall 1973). Their importance in different wheat end-use quality parameters has been well-established (Bonafede et al. 2015; Liu et al. 2005; Nieto-Taladriz et al. 1994; Payne 1987; Zhang et al. 2012); however, they have received much less attention compared to the HMW-GSs mainly because of their greater complexity at both the gene and protein level.

The typical LMW-GSs are usually divided into m-type (methionine), s-type (serine), or i-type (isoleucine) based on the first amino acid present at the N-terminal region of the mature protein. They all possess a highly similar structure consisting of a signal peptide, an N-terminal region (absent in the i-type LMW-GSs), a repetitive region, and a C-terminal region. Differences in the repetitive domain are primarily responsible for LMW-GS gene and protein size variation (D’Ovidio and Masci 2004).

The LMW-GS proteins are encoded by multigene families located in the short arms of the group 1 chromosomes at the Glu-A3, Glu-B3, and Glu-D3 loci. The number of LMW-GS genes varies depending on the locus and the variety, but generally more than 15 LMW-GS genes are present in individual bread wheat varieties (Zhang et al. 2013). The genes at each Glu-3 locus are separated by large intergenic regions of up to 748 Kb between two neighboring genes at the Glu-B3 locus (Ibba et al. 2017a). Also, recombination was detected at the Glu-A3, Glu-B3, and Glu-D3 loci (Dong et al. 2010; Ibba et al. 2017a) and, based on this evidence, each Glu-3 locus was divided into multiple haplotypes (Ibba et al. 2017b).

The most common methods for the analysis of the LMW-GS profile are sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE) and reversed-phase high-performance liquid chromatography (RP-HPLC). However, the complexity of the protein bands/peaks obtained using these methods has limited their routine use for the selection of specific LMW-GS alleles (Zhang et al. 2011). Other protein-based methods with greater resolution have been adopted for the LMW-GS analysis, like two-dimensional gel electrophoresis (2-DE, IEF × SDS-PAGE) and matrix-assisted laser desorption/ionization time-of-flight mass spectrometry (MALDI-TOF-MS), but also these methods possess several limitations that have restricted their use: time consuming (2-DE), requirement of skilled personnel, and high cost (Liu et al. 2010).

PCR-based markers have several advantages over biochemical markers, namely, easy to perform, low cost, fast, require a minimum amount of DNA, and can be applied at any stage of the plant life cycle (Kiszonas and Morris 2017; Kumar 1999). For these reasons, PCR-based markers are usually preferred for the identification of favorable alleles in breeding programs. Currently, several PCR-based molecular markers have been developed for the identification of different Glu-3 allelic variants (Appelbee et al. 2009; Wang et al. 2009, 2010; Zhang et al. 2004; Zhao et al. 2006, 2007). However, even though these markers have been efficiently used for the discrimination of Glu-A3 and Glu-B3 alleles (Liu et al. 2010), they were not able to capture the complete LMW-GS genic profile, but only the polymorphism of a few of the LMW-GS genes.

In contrast, the PCR-based molecular marker system developed by Zhang et al. (2011) took advantage of the genetic structure of the LMW-GS genes allowing their differentiation based on the length polymorphism related to the repetitive region of each LMW-GS gene. With this marker system, most of the LMW-GS genes of a given bread wheat variety could be amplified and differentiated by using as few as two PCR reactions (Ibba et al. 2017b). This marker system has been effectively used in numerous studies (Espí et al. 2014; Ibba et al. 2017a, b, c; Zhang et al. 2012, 2013) allowing the sequencing of a greater number of LMW-GS genes, a better understanding of the Glu-3 loci organization, and the connection between specific LMW-GS genic profiles and the observed Glu-3 allelic variants. However, this method has some disadvantages, for example, the complexity of the PCR profiles, presence of overlapping PCR amplicons, impossibility to selectively amplify LMW-GS genes known to be involved in wheat end-use quality, and impossibility to easily differentiate genes coming from the three Glu-3 loci.

Recently, in the study of Ibba et al. (2017b), the molecular marker system developed by Zhang et al. (2011) was adopted to investigate the role that each haplotype has in the variation of gluten strength and wheat end-use quality parameters and specific haplotypes with greater impact on these phenotypes were identified. Results of that study highlighted the need for developing new molecular markers that specifically target single LMW-GS haplotypes in order to be able to restrict the selection to only the LMW-GS genes mainly involved in wheat end-use quality.

For these reasons, the objective of the present study was to develop a set of haplotype-specific PCR-based molecular markers in order to effectively differentiate specific Glu-3 haplotype polymorphisms and resolve the ambiguities generated from the molecular marker system developed by Zhang et al. (2011). These markers could greatly improve marker-assisted breeding for wheat quality.

Materials and methods

Plant material

A total of 44 bread wheat (Triticum aestivum L.) varieties were selected for the development and validation of the LMW-GS haplotype-specific molecular markers. The LMW-GS genic profile of these varieties was previously determined by Ibba et al. (2017b, c) using the LMW-GS molecular marker system developed by Zhang et al. (2011). The location of the DNA markers was confirmed using the variety Chinese Spring and its chromosome 1 long arm ditelosomic lines (Dt1AL, TA#3102; Dt1BL, TA#3113; Dt1DL, TA#3131) obtained from the Kansas State University Wheat Genetics Resource Center.

Primer design

For primer design, a total of 90 LMW-GS gene sequences (AY453158, EU369701, EU369704- EU369710, EU369715-EU369721, EU369724, EU369729, FJ549929, FJ549932, FJ549933, FJ549939, JF271919, JF339156, JF339184, JF339186, JF339201, JX877780-JX878230, KJ666154, and KX879102) were selected and used for nucleotide alignment through Clustal Omega (https://www.ebi.ac.uk/Tools/msa/clustalo/) using the default parameters.

From the alignment, specific SNPs unique for the LMW-GS genes at each haplotype located before and/or after the repetitive region of each sequence were identified. Based on the identified SNPs, primer pairs specific for each LMW-GS haplotype were manually designed, and when possible, by including the mismatched nucleotides at the 3′ end of each primer. Their sequences were successively uploaded in the software Primer3 v0.4 (Untergasser et al. 2012), and the relative sequence lengths were adjusted to meet the default primer standards.

The primer sets developed for the Glu-A3-3 haplotype were refined by including the forward primer GluA3eF (5′-CAATGAAAACCTTCCTCGTCTG-3′) (Zhang et al. 2004). This additional primer identified the differences between the two variants of the gene A3-646, associated with alleles Glu-A3e and Glu-A3g. Similarly, the primer set developed for the haplotype Glu-B3-3 was modified by including the primer GluB3hF (5′-CATTACCACCACAACAAACATTAA-3′) (adapted from Wang et al. 2009). This primer discriminated the gene B3-688-2 associated with allele Glu-B3h.

DNA extraction and PCR amplification

Genomic DNA was extracted from the leaves of 10–14-day-old glasshouse-grown seedlings using the DNeasy 96 Plant Kit (Qiagen, Redwood City, CA, USA). All PCR reactions were performed using the M13 tail fluorescent labeling method. The sequence of the labeled universal primer was 5′-CACGACGTTGTAAAACGAC-3′. The same sequence was added at the 5′ end of each forward primer.

PCR reactions for the Glu-A3-1, Glu-A3-2, Glu-B3-1, Glu-D3-1/5, Glu-D3-4, Glu-D3-5, and Glu-D3-7 markers were performed each in 25-μl reaction volumes containing 100 ng of genomic DNA, 1× Standard Taq Buffer (10 mM Tris-HCl, 50 mM KCl, 2.5 mM MgCl2, pH 8.3), 200 mM of dNTPs, 0.2 mM of reverse primer, 0.04 mM of unlabeled forward primer, 0.16 mM of labeled universal primer, and 1 unit of Taq DNA Polymerase. PCR reactions for the Glu-A3-3, Glu-B3-2, Glu-B3-3, Glu-D3-2, and Glu-D3-6 molecular markers were performed using the same abovementioned conditions but with different primer concentrations: 0.28 mM of reverse primer, 0.07 mM of unlabeled forward primer(s), and 0.21 mM of labeled universal primer.

Amplifications were performed in a Peltier Thermal Cycler PTC-200 using the same conditions for all the markers: 5 min at 94 °C, 40 cycles of 30 s at 94 °C, 50 s at 63 °C, and 1 min at 72 °C, and a final extension step of 10 min at 72 °C.

PCR product analysis

PCR products labeled with either FAM or VIC fluorescent dyes were diluted 1:8 in water. Three microliters of the diluted product was added to 9 μl of HiDi-formamide together with either 1 μl of 685-LIZ internal size standard or 0.1 μl of 1200-LIZ internal size standard (Applied Biosystems, Foster City, CA, USA). The mixtures were denatured in a Peltier Thermal Cycler PTC-200 at 95 °C for 5 min and successively analyzed on a 3130xl DNA analyzer (Applied Biosystems, Foster City, CA, USA). The size of each PCR amplicon was determined using the software GeneMarker v3.7 (SoftGenetics, LLC, USA).

Results

Marker development

From the analysis of the LMW-GS gene alignment, SNPs specific for each Glu-3 haplotype were identified and used to develop a primer set specific for each of these linkage blocks (Supplementary File 1). The only exceptions were the haplotypes Glu-D3-1 and Glu-D3-5 that were amplified by the same marker. No markers were developed for the haplotype Glu-D3-6 due to the absence of polymorphism at this group as reported by Zhang et al. (2013) (Table 1). All the markers were designed in order to amplify the repetitive region of the LMW-GS genes and, consequently, to distinguish the allelic variants of the genes at each haplotype based on length polymorphism.

Table 1 Haplotype-specific LMW-GS markers
Full size table

In addition to the haplotype-specific primers, the GluA3eF and the GluB3hF primers were included in the marker sets specific for the haplotypes Glu-A3-3 and Glu-B3-3, respectively. The addition of these markers facilitated the discrimination of the genes A3-646 and B3-688 based on their sequence polymorphism.

To ensure the correct annealing location of each primer set, the markers were applied to the common wheat variety Chinese Spring and its ditelosomic lines Dt1AL, Dt1BL, and Dt1DL. As shown in Supplementary Fig. 1, the molecular markers specific for the haplotypes at the Glu-A3, Glu-B3, and Glu-D3 loci did not produce any PCR amplicon in the Dt1AL, Dt1BL, and Dt1DL lines, respectively. The exception was the marker Glu-D3-1/5 that produced two tightly close unspecific bands in the Dt1DL line. These bands, however, were absent in Chinese Spring and the Dt1AL and Dt1BL lines.

Marker validation

To validate marker efficiency, each haplotype-specific LMW-GS molecular marker was applied to a set of 44 bread wheat varieties of known LMW-GS genic profile. In order to effectively distinguish each allelic variant at all of the Glu-3 haplotypes, the PCR results were separated by capillary electrophoresis, producing electropherograms as the ones shown in Supplementary Fig. 2.

According to previous studies, only one LMW-GS gene is located at the haplotype Glu-A3-1 (Ibba et al. 2017b). Five different allelic variants of this gene were previously identified among the bread wheat variety validation set (Ibba et al. 2017b,c), namely, A3-353, A3-370, A3-373, A3-391, and a Null allele. These alleles were effectively identified by the Glu-A3-1 molecular marker and corresponded to amplicons with a length of 246, 264, 267, and 286 nt, and no amplicon, respectively (Tables 2 and 3).

Table 2 List of the LMW-GS genes and their relative fragment size as identified by the haplotype-specific LMW-GS molecular markers
Full size table
Table 3 Glu-A3 genic profile of the 44 bread wheat varieties used to validate the Glu-A3 haplotype-specific molecular markers
Full size table

The molecular marker Glu-A3-2 was also able to correctly identify the allelic variation at the Glu-A3-2 haplotype. Among the 44 bread wheat varieties analyzed, six different alleles were identified. These alleles were named A3-374, A3-394, A3-400, A3-402, A3-408, A3-411, and Null allele (Zhang et al. 2013) and produced, respectively, PCR products with a length of 196, 215, 224, 227, 233, and 236 nt, and no amplicon (Table 2). For all the analyzed varieties, the results previously reported in the literature matched the results obtained with the Glu-A3-2 molecular marker (Table 3).

In contrast to the previous two Glu-A3 haplotypes, at Glu-A3-3 there are typically present two to four i-type LMW-GS genes. Among the varieties analyzed were identified seven different variants of this haplotype. By using the molecular marker Glu-A3-3, each allelic variant of the genes constituting the Glu-A3-3 haplotype was identified by a specific amplicon (Table 2). As expected, the results obtained with the molecular marker Glu-A3-3 matched with previously reported results, with the exception of variety India 115-2 (Table 3). This variety in fact exhibited both alleles A3-620 (651 nt) and A3-646 (678 nt) whereas in previous studies, it exhibited only allele A3-620. Also, by including in the Glu-A3-3 marker set the primer GluA3eF, it was possible to distinguish the two different variants of the gene A3-646 previously reported (Zhang et al. 2013), and associated with the Glu-A3 alleles Glu-A3e and Glu-A3g. When the A3-646 variant associated with allele Glu-A3e was present, two amplicons were obtained (678 and 729 nt), whereas when only a single amplicon was present (the smaller with 678 nt), the A3-646 variant allele Glu-A3g was identified (Supplementary Fig. 3).

The marker Glu-B3-1 was able to discriminate the three allelic variants detected at the haplotype Glu-B3-1: B3-548 (440 nt), B3-557 (452 nt), and a Null allele (absence of an amplicon). For all the varieties, the Glu-B3-1 profile identified here was in accordance with previous results (Tables 2 and 4).

Table 4 Glu-B3 genic profile of the 44 bread wheat varieties used to validate the Glu-B3 haplotype-specific molecular markers
Full size table

The marker Glu-B3-2 was developed for the haplotype Glu-B3-2. The analyzed varieties exhibited three different allelic variants at this group: B3-510, B3-530, and a Null allele. All these alleles were correctly identified across the varieties by the Glu-B3-2 molecular marker and produced amplicons with lengths of 456 and 477 nt, and no amplicon, respectively (Tables 2 and 4).

Similar to the haplotype Glu-A3-3, at the haplotype Glu-B3-3 there are generally multiple s-type LMW-GS genes (1 to 3) tightly linked together. According to previous studies (Ibba et al. 2017b, c), the set of bread wheat varieties analyzed here exhibited 12 different variants of the Glu-B3-3 haplotype. The molecular marker Glu-B3-3, developed to amplify the s-type LMW-GS genes from the Glu-B3 locus, was able to successfully detect the gene polymorphisms associated with the Glu-B3-3 variants previously identified. The only exception was the allele B3-813, which did not produce any PCR product (Table 2). From the results of this marker, there were also detected some alleles that were not previously characterized like B3-613 (676 nt) and B3-642 (704 nt). Few discrepancies between the present and previous results were identified. Specifically, according to the results obtained from the marker Glu-B3-3, the variety Courtot exhibited allele B3-613 (676 nt), whereas according to previous results, this variety did not exhibit any allelic variants related to that gene. Similarly, varieties JD, WA8236, WA8237, and WA8238 all exhibited allele B3-587 (650 nt) according to this study. In contrast, according to previous studies, these varieties did not possess any of the allelic variants of that gene (Table 4). In order to better characterize the genes at this haplotype, one additional primer was included in the marker set specific for Glu-B3-3. This marker allowed the discrimination of the B3-688-3 variant (opposed to B3-688-1, B3-688-2, and B3-688-4) (Supplementary Fig. 4). When the allele B3-688-3 was present, two amplicons were produced with lengths of 729 and 751 nt, whereas the other B3-688 allelic variants produced only one amplicon of 751 nt (Supplementary Fig. 4).

For the haplotypes Glu-D3-1 and Glu-D3-5, one molecular marker (Glu-D3-1/5) was developed due to the high sequence similarity of the genes at these haplotypes. This marker produced a 303-nt amplicon, corresponding to the Glu-D3-1 gene D3-575, and an amplicon with a length of 335, 339, 342, 345, or 348 nt, or no amplicon, corresponding to the allelic variants of the gene of the haplotype Glu-D3-5 (Table 2). The genic profile of these haplotypes matched the results previously reported. The exception was the variety Cappelle-Desprez that, according to Ibba et al. (2017c), exhibited allele D3-597 whereas here it exhibited allele D3-594 (348 nt) (Table 5).

Table 5 Glu-D3 genic profile of the 44 bread wheat varieties used to validate the Glu-D3 haplotype-specific molecular markers
Full size table

The marker Glu-D3-2 was developed to capture the variation at the haplotype Glu-D3-2. This marker generated an amplicon of 637, 646, or 648 nt, or no amplicon, corresponding to alleles D3-432, D3-441, D3-444, and a Null allele, respectively (Table 2). The present results obtained with this molecular marker were consistent with previous results (Table 5).

To detect polymorphism at the haplotype Glu-D3-4, the molecular marker Glu-D3-4 was developed. This marker produced PCR amplicons of 292 or 299 nt, or no amplicon corresponding, respectively, to alleles D3-385′, D3-393, and a Null allele (Table 2). The results obtained here (Table 5) were consistent with results previously reported with the exception of variety NSA10_2196. According to Ibba et al. (2017b), this variety possessed allele D3-385′ whereas from the present analysis, the variety had allele D3-393 (Table 5).

The marker Glu-D3-6 was developed to discriminate the alleles at the haplotype Glu-D3-6. Among the varieties analyzed, three different alleles were detected, namely D3-525, D3-528, and a Null allele. They respectively amplified a 513- or 516-nt amplicon, or no amplicon (Table 2). The Glu-D3-6 genic profile previously reported was consistent with the results obtained here apart from varieties HE9817_1_2 and WA8207. These varieties had previously been shown to carry allele D3-375 (Ibba et al. 2017b), but here no amplicons were observed using Glu-D3-6 (Table 5).

Polymorphism at the haplotype Glu-D3-7 was detected by the molecular marker Glu-D3-7. The set of bread wheat varieties chosen for the analysis exhibited either allele D3-394 or a Null allele. Both of the alleles were correctly detected by the marker and were associated with a PCR product of 292 nt or with no amplicon (Table 2). Exceptions were varieties Farnum and WA8207 that in this study exhibited a Null allele at the haplotype Glu-D3-7 whereas, according to previous studies (Ibba et al. 2017b), possessed a previously uncharacterized allele (D3-382) (Table 5).

Discussion

The LMW-GSs are one of the major components of gluten and both their quantity and quality strongly influence wheat dough functionality (Shewry et al. 2002). For these reasons, numerous methods have been developed for the identification and selection of favorable LMW-GS alleles. These methods can be divided into two main groups: protein-based methods (SDS-PAGE, RP-HPLC, 2-DE, MALDI/TOF) and DNA-based methods (PCR-based molecular markers) (Kiszonas and Morris 2017; Liu et al. 2010).

Typically, PCR-based molecular markers are the preferred method in breeding programs for the selection of desired alleles because they are easy to use, fast, accurate, and independent of the plant stage of development (Kumar 1999). Based on the available LMW-GS genes, allele specific markers were developed for the Glu-A3 and Glu-B3 loci (Wang et al. 2009, 2010). However, the relatively high number of primer pairs required for the Glu-A3 and Glu-B3 allelic characterization (7 and 10, respectively) and the impossibility of these markers to capture the complexity of the Glu-3 loci limited their use. In contrast, the PCR-based molecular marker system developed by Zhang et al. (2011) was able to characterize most of the LMW-GS genes present in a given bread wheat variety through their length polymorphism. The development of this marker system facilitated a deeper understanding of the Glu-3 loci structure, the LMW-GS gene sequences, and the connection between different LMW-GS genic profiles and their relative LMW-GS alleles (Ibba et al. 2017a, b; Zhang et al. 2013).

In the present study, we developed a set of PCR-based haplotype-specific LMW-GS molecular markers. This set, different from the molecular markers developed by Wang et al. (2009, 2010), is able to identify and differentiate most of the LMW-GS genes present in each bread wheat variety. Also, different from the marker system developed by Zhang et al. (2011), these markers are easier to score, less ambiguous, take into account the presence of recombination at each Glu-3 locus and can be used to selectively amplify only the LMW-GS haplotypes more highly involved in wheat end-use quality. The molecular markers developed by Zhang et al. (2011), in fact, produced complex electropherograms displaying multiple DNA fragments with lengths ranging from 353 to 813 nt (for the LMWGS1 marker). The interpretation of this output could be challenging since (1) the molecular marker amplifies indiscriminately LMW-GS genes from the Glu-A3, Glu-B3, and Glu-D3 loci and (2) DNA fragments with the same length but representing different alleles could be present.

The primer sets used in this study were developed based on numerous LMW-GS genes and were found to be specific for each Glu-3 locus and for the haplotypes at each Glu-3 locus. In contrast to the molecular markers developed by Zhang et al. (2011), the main purpose here was to selectively detect polymorphism in specific Glu-3 haplotypes known to have an effect on wheat end-use quality rather than indistinctly amplify all the LMW-GS genes of a given bread wheat variety. According to previous studies (Dong et al. 2010; Ibba et al. 2017a), there is recombination at the Glu-A3, Glu-B3, and Glu-D3 loci. These findings required a re-evaluation of the Glu-3 loci as sets of multiple haplotypes rather than as unique linkage blocks with “allele” status. Consequently, the methods of analysis to assess LMW-GS polymorphism also required reconsideration. On the one hand, the selection of specific LMW-GS alleles based on the polymorphism of a unique gene (Wang et al. 2009, 2010) could no longer be considered as representative of the Glu-3 loci polymorphism but, on the other hand, the analysis of the entire LMW-GS genic profile of a given bread wheat variety could be excessively complex, especially because many of the LMW-GS gene are pseudogenes and/or do not strongly affect wheat end-use quality (Ibba et al. 2017b).

In general, the molecular markers developed in this study were equally or more efficient than the molecular marker system of Zhang et al. (2011), and the results obtained in this study almost always corresponded with previously reported results (Ibba et al. 2017b, c). This result is notable given the large genetic diversity of the included varieties, specifically chosen to represent the genetic variation at these three LMW-GS loci.

Among the three sets of markers developed for the haplotypes at the Glu-A3 locus (Glu-A3-1, Glu-A3-2, and Glu-A3-3), discrepancies were identified only for the Glu-A3-3 profile of the variety India 115-1. Here, India 115-1 exhibited both genes A3-620 (651 nt) and A3-646 (678 nt), whereas according to Ibba et al. (2017c), it possessed only the gene A3-646. Zhang et al. (2013) reported that these two genes are allelic variants of the same gene so it is unlikely that they are both present in a single variety. It is possible that these discrepancies are the result of seed/DNA mislabeling or seed contamination. According to Ibba et al. (2017c) the LMW-GS genic profile of the haplotype Glu-A3-3 was representative of the Glu-A3 alleles that were established by SDS-PAGE. However, by using the molecular markers of Zhang et al. (2011), it was not possible to distinguish alleles Glu-A3e and Glu-A3g. The varieties exhibiting these two alleles, in fact, possessed the same Glu-A3-3 genic profile represented by alleles A3-502 and A3-646, but differed in the sequence of allele A3-646. By including in the Glu-A3-3 marker set the primer GluA3eF developed by Zhang et al. (2004), it was possible to differentiate the two A3-646 gene sequences indicating the possibility to use this marker set to directly identify the Glu-A3 alleles present in a single bread wheat variety.

Among the three sets of markers developed for the Glu-B3 haplotypes, differences between the results obtained here and previous results were identified only for the haplotype Glu-B3-3. Interestingly, all of the varieties that according to the studies of Ibba et al. (2017b, c) exhibited only alleles B3-578 and B3-621, in this study, they produced a third amplicon. Specifically, varieties JD, WA8236, WA8237, and WA8238 all possessed allele B3-587 (650 nt), whereas varieties Courtot and Bungulla exhibited two novel alleles B3-613 (677 nt) and B3-642 (705 nt), respectively. In the case of allele B3-587, it is possible that this allele was not identified through the marker system of Zhang et al. (2011) because its amplicon overlapped with the Glu-D3 gene D3-589, present in all the four abovementioned varieties. The two other alleles (B3-613 and B3-642), in contrast, were not previously detected, likely because of their relatively low amplification efficiency. Similar to the haplotype Glu-A3-3, the genic profile of the haplotype Glu-B3-3 was associated with different Glu-B3 allelic variants (Ibba et al. 2017c). By including in the Glu-B3-3 marker set the primer GluB3hF modified from Wang et al. (2009), it was possible to discriminate the B3-688-3 variant from the other B3-688 alleles. According to Zhang et al. (2012), allele B3-688-3 is associated with allele Glu-B3h. Results obtained here suggest that the Glu-B3-3 marker set could be effectively used to discriminate most of the Glu-B3 alleles with the exceptions of the alleles Glu-B3c and Glu-B3d, which exhibit the same Glu-B3-3 profile, and the alleles Glu-B3ac and Glu-B3ad, which appear to be determined by both the Glu-B3-2 and the Glu-B3-3 genic profile (Ibba et al. 2017c).

As found in the Glu-A3 and Glu-B3 genic profiles, the Glu-D3 genic profile obtained here was consistent with the one previously reported (Ibba et al. 2017b, c) with few exceptions. Specifically, from the analysis of the haplotype Glu-D3-4, variety NSA10_2196 exhibited allele D3-393 (299 nt) whereas according to the previous study (Ibba et al. 2017b), it possessed allele D3-385′. In this case, it is possible that the allele D3-385′ was incorrectly identified. This allele in fact produces an amplicon with the same length as another Glu-D3 gene (D3-385), present in the variety NSA10_2196, making its identification through the molecular marker system of Zhang et al. (2011) problematic. In contrast, at the haplotype Glu-D3-5, varieties Cappelle-Desprez and Evina both exhibited allele D3-594 (348 nt) whereas according to Ibba et al. (2017b, c), they possessed allele D3-597. From these results, it appears that either the molecular marker developed here is unable to distinguish the allele D3-597 or that this allele was previously incorrectly identified. More varieties exhibiting either of the two alleles should be analyzed in order to reach a conclusion. Finally, according to the study of Ibba et al. (2017b), varieties HE9817_1_2 and WA8207, and varieties Farnum and HE9817_1_2 hypothetically possessed the previously uncharacterized alleles D3-375 (Glu-D3-6) and D3-382 (Glu-D3-7), respectively. The molecular markers Glu-D3-6 and Glu-D3-7 did not detect these novel alleles. These differences could be determined by both an incorrect gene assignment by Ibba et al. (2017b), but also by a possible limitation of the Glu-D3-6 and Glu-D3-7 molecular markers. More studies will be needed in order to characterize the alleles D3-375 and D3-382 previously identified and to determine if they are actually allelic variants of the genes at the haplotypes Glu-D3-6 and Glu-D3-7 or represent unique genes.

Conclusion

In the present study, three, three, and five PCR-based molecular marker sets specific for the haplotypes at the Glu-A3, Glu-B3, Glu-D3 locus, respectively, were developed. Each of these molecular markers was found to be specific for the different Glu-3 loci and for each haplotype. The LMW-GS genic profile detected using these molecular markers corresponded to the results obtained in previous studies confirming the reliability of these haplotype-specific molecular markers. Also, in many cases, the markers developed here were more efficient than those previously reported. Specifically, compared to the molecular markers developed for the Glu-A3 and Glu-B3 allele discrimination (Wang et al. 2009, 2010; Zhang et al. 2004), the markers developed here were more accurate since they were able to detect most of the LMW-GS genes rather than the polymorphism of single genes. Also, they could discriminate different Glu-A3 alleles, and most of the Glu-B3 alleles by performing a single PCR reaction as opposed to seven (Zhang et al. 2004), ten (Wang et al. 2009), or four (Wang et al. 2010). In contrast, compared to the molecular marker system developed by Zhang et al. (2011), the haplotype-specific molecular markers developed here are easier to interpret, less ambiguous, and selectively amplify only specific LMW-GS haplotypes rather than a relatively large number of amplicons. Even if more studies are needed to confirm the efficiency of this set of markers, the primers developed here could be effectively used for the discrimination of the known Glu-A3 and Glu-B3 alleles, to solve the ambiguities from the molecular marker of Zhang et al. (2011) and, in the future, could be employed for the selection of specific LMW-GS haplotypes more highly involved in wheat end-use quality.