Wheat gluten is the source of a large repertoire of T-cell epitopes involved in the pathogenesis of CD. It is clear that most patients react to epitopes derived from α-gliadins. However, there is increasing evidence that more T-cell epitopes from other gluten genes are involved in CD [24, 57]. It was shown that next to α-gliadins also T-cell epitopes from γ-gliadins play a significant role in the development of CD [2, 17–19, 22, 24]. Tye-Din et al.  found that gluten specific polyclonal T-cells in the peripheral blood of CD patients were specific for the same set of gluten peptides after feeding the patients with the same cereal but after the patients ingested respectively wheat, barley and rye different T-cells, recognizing different sets of gluten peptides, were found. Also, it was hypothesized that a T-cell response to gluten can be initiated against a relatively large number of peptides, and when the response evolves it focuses on the most immuno-dominant sequences . So, there are some indications that the composition of different cereals and wheat varieties in the diet [24, 57] as well as the disease status  may play a role in the defining the pattern of T cell reactivity towards different gluten peptides in a celiac patient.
Using polyclonal, gliadin reactive T-cell lines, Camarca et al.  showed that T-cell reactivity towards the γ-gliadin derived peptides is more heterogeneous compared to the reaction to α-gliadin derived peptides; reaction to the latter was focused towards a few immunodominant peptides whereas a larger repertoire of different γ-gliadin derived peptides was found and patients were recognizing varying sets of peptides. They suggested that this less focussed behaviour of γ-gliadin peptides in celiac disease may reflect the high genetic diversity of γ-gliadins.
To obtain insight in CD epitopes derived from γ-gliadins we performed a detailed study of the presence and natural variation of CD epitopes derived from the γ-gliadin gluten fraction of bread wheat (T. aestivum L.) which is an allo-hexaploid (2n = 6x = 42) carrying three homoeologous genomes A, B and D. This study included two datasets: (1) 69 novel genomic γ-gliadin clones from the diploid wheat species T. monococcum, Ae. speltoides and Ae. tauschii. These diploid wheat species are carrying respectively genome Ab, S and D which are ancestrally related to the three respective homoeologous genomes of bread wheat, A, B and D; and (2) γ-gliadin transcripts from bread wheat (T. aestivum) derived from the public database (Genbank, NCBI). By similarity analysis the transcripts of bread wheat could be assigned to locus Gli-A1, Gli-B1, or Gli-D1 (on the respective homoeologous chromosomes 1A, 1B and 1D). To reduce the complexity of the multigene γ-gliadin family, transcripts of bread wheat were analyzed (instead of genomic sequences). As such, the γ-gliadin transcriptome is a minimal estimate for the γ-gliadin variants in gluten of bread wheat. The γ-gliadin transcripts of bread wheat (N = 717) encode for 26 different γ-gliadin protein isoforms of which ten represented full length γ-gliadins (contigs >300 bp, ≥ four transcripts, 98% to 99% sequence identity). This figure is in keeping with the 15–40 estimated γ-gliadin genes in hexaploid wheat estimated before [35, 36, 41]. Half of the γ-gliadin transcripts from bread wheat grouped with Ae. tauschii sequences (D genome, 6 contigs, 354 transcripts) and were assigned to the γ-gliadin locus on chromosome 1D (Gli-D1), indicating that the majority of the γ-gliadins are expressed from the D genome. The remainder of the transcripts were equally assigned to Gli-A1 (9 contigs, 178 transcripts) and Gli-B1 (11 contigs, 185 transcripts). Similarly, for the α-gliadins it was found that the highest expression occurred from the D genome (Gli-D2) .
With regard to CD epitope content, our analysis shows that γ-gliadins from bread wheat contain on average between 4 and 10 potential CD epitopes in the first variable domain (Table 3, range 4.6 to10.9 epitope cores per transcript). The 26-mer γ-gliadin peptide that harbours four distinct CD epitopes  is only present in D-genome γ-gliadins. It is present in 9% of the γ-gliadin transcripts of bread wheat. Thus, the γ-gliadins from all three genomes encode a large number of different immunogenic peptides which exceeds the number of identified immunogenic peptides in the α-gliadins. Perhaps this is the basis of the observation that immuno-dominant responses to α-gliadin derived CD epitopes are generally found in patients, while responses to the γ-gliadins appear less consistent as was previously suggested by Camarca et al. . If patients make frequently T-cell responses to a limited number α-gliadin peptides but can respond to a large number of different γ-gliadin derived peptides, T-cells for some specific α-gliadin epitopes may prevail numerically but collectively γ-gliadin derived T-cell epitopes are probably similarly effective in causing CD.
The DQ2-γ-I epitope (PQQSFPQQQ) is regarded as a major CD epitope as it is frequently recognized by CD patients [22, 24]. We observed a high level of natural genetic variation in the C-terminal flanking sequence of this CD epitope. According to the rules for deamidation by TG2 [13, 14] the variation in the C-terminal flanking sequences determines the deamidation pattern of this epitope core. Among diploid and hexaploid wheat six variants of the 9-mer DQ2-γ-I epitope core were evident with respect to variation in amino acid composition and deamidation pattern. Testing of these DQ2-γ-I variants for their capacity to trigger proliferation of DQ2-γ-I specific T-cells confirmed that TG2 deamidation of glutamine (Q) at position 9 is essential for T-cell stimulation, most likely because it is providing a negative charge which is essential for HLA-DQ2 binding. Variant peptides that displayed a serine (S) at position 5 instead of an phenylalaninine (F) were found to have lost T-cell stimulatory capacity. Moreover, the presence of a positively charged arginine (R) residue at position +1 diminished T-cell proliferation (Additional file 3), most likely because it influenced the deamidation of Q9 as previously observed by Dørum et al. . Similarly, a tryptophan (W) at position +2 inhibited the T-cell response (Additional file 3). Thus, both substitutions within and outside the epitope core can influence the T-cell stimulatory properties of the DQ2-γ-I epitope.
Taken together, the γ-gliadins from all three wheat genomes appear to be a significant source of CD epitopes. Similar to the α-gliadins, the highest number of potential immunogenic γ-gliadin peptides are encoded by the D genome of bread wheat which is thus the most critical genome with regard to CD toxicity. However, gluten derived from tetraploid wheat varieties lacking the D genome is not tolerated by CD patients as well, indicating that the mere elimination of the D genome is not sufficient for the generation of safe wheat. Regarding the A-genome of wheat, there are indications for the existence of T. monococcum spp. with a level or type of gliadin that is unable to induce IFN-γ production and histologic damage as was observed for instance in duodenal biopsy specimens from patients with celiac disease . So, the high level of genetic variation among wheat lines and the presence of genetic variation influencing the immunogenicity of the major CD epitopes may offer possibilities to generate wheat varieties with a reduced CD-immunogenicity, tailored to major CD epitopes (i.e. DQ2-α-I, DQ2-α-II;  and/or DQ2-γ-I, this study). Such varieties would help to reduce the presence of immunogenic CD epitopes in wheat flour and, while not safe for consumption by patients, might help to prevent the onset of CD in people that carry genetic risk factors [44, 45]. However, introgression from the diploid to the hexaploid level is a time-consuming process, and consequently it will take many years before the product of such a synthetic hexaploidisation has been bred to sufficient agronomic quality. Alternatively, one could screen a large number of existing varieties for differences in their toxicity for CD patients as the analysis of CD epitope regions in transcript sequences does provide an accurate and quick method for screening [46, 58]. Consequently, we are currently analyzing CD epitope regions in the gluten transcriptome of tetraploid wheat cultivars in a medium-throughput way by employing next-generation sequencing technology.