Celiac disease is a chronic inflammatory disease of the small intestine elicited by T-cell mediated immune response to dietary gluten. Almost all patients express the HLA-DQ2.5, HLA-DQ2.2 or HLA-DQ8 allotypes with HLA-DQ2.5 being expressed by more than 90% of the patients. The disease-associated HLA-molecules present gluten to CD4+ T cells. Gluten is a complex family of proteins found in wheat. Currently, 27 distinct HLA-DQ2.5-restricted T-cell epitopes have been characterized from gluten and related proteins such as hordein (barley), secalin (rye) and avenin (oat) (1). CD4+ T cells reactive to these epitopes can only be found in the intestinal tissue of celiac disease patients, and not in controls including non-celiac gluten-intolerant patients (2, 3). All HLA-DQ2.5-restricted gluten T-cell epitopes contain one or several glutamic residues within the 9-mer core HLA-binding region. Native gluten has few glutamic acid residues, but contains over 35% glutamine residues (4). Certain glutamine residues in the native gluten can be post-translationally modified into glutamic acid residues by the action of transglutaminase 2 (TG2), in a reaction known as deamidation (5). The negatively charged glutamic acid residues increase the binding affinity of gluten peptides to HLA-DQ2.5 (6).

Among the multitude of HLA-DQ2.5-restricted gluten T-cell epitopes, only a few are known to be recognized by nearly all celiac disease patients. These immunodominant epitopes are the epitopes DQ2.5-glia-α1, DQ2.5-glia-α2, DQ2.5-glia-ω1, DQ2.5-glia-ω2, and DQ2.5-hor-3 (7, 8). T cells that recognize the other HLA-DQ2.5-restricted gluten epitopes are found occasionally in a minority of celiac disease patients. Despite decades of efforts in characterizing T-cell epitopes from gluten (reviewed in 1), there are still undiscovered T-cell epitopes in gluten, a fact manifested by T-cell clones that recognize deamidated peptic digest of gluten, but not any of the known epitope sequences. The proportion of gluten-reactive T cells that recognize hitherto uncharacterized gluten epitopes is largely unknown since few systematic efforts have been made in the past few years.

The fact that an unknown fraction of T cells from celiac disease patients recognize yet unknown gluten epitopes, impacts the estimation of the frequency of gluten-reactive T cells in the celiac intestinal tissue. Most studies use ELISPOT with overlapping peptides from a few well-characterized gluten proteins (8, 9) or tetramers presenting a limited number of the known gluten epitopes (10, 11) in their estimation of this frequency. Only one single study has reported the frequency of gluten-reactive T cells by direct cloning of T cells from the lamina propria of duodenal biopsies (10). This study, however, included only three treated patients, and two untreated patients. The average frequency of gluten-reactive T cells in this study was 1.0% (range 0.5% - 1.8%, four patients) in celiac lesion with Marsh 3, a histological score that indicates the most severe tissue damage. Yet other approaches such as cloning from polyclonal T-cell line can give some indication of the specificity distribution, including the proportion of cells that recognize unknown gluten epitopes. However, due to in vitro expansion of the polyclonal lines, this approach does not give the precursor frequency of gluten-reactive T cells in the celiac lesion.

In the current study, we have cloned T cells directly from the lamina propria of total nine subjects, of whom five were untreated patients with Marsh 3 lesion, one refractory celiac patient with Marsh 3, and three control subjects that do not have celiac disease. From the six celiac disease patients with Marsh 3 in the duodenum, 24 T-cell clones were found to be gluten-reactive of total 1,652 clones tested, giving an average frequency of 1.5% of gluten-reactive T cells. We also found that around one third of gluten-reactive T-cell clones from these patients only responded to deamidated gluten, but not to peptides containing any of the known HLA-DQ2.5-restricted gluten epitopes.

Gluten-reactive T cells are a hallmark of celiac disease. These cells recognize TG2-deamidated gluten presented on disease-associated HLA-DQ molecules and play pivotal roles in the pathogenesis of celiac disease. In the current study, we have generated T-cell clones directly from lamina propria tissues and found that all six samples taken from tissues in active disease contained gluten-reactive T cells. In average, 1.4% (range 0.7% - 1.9%) of T cells from inflamed celiac tissue were gluten-reactive, compared to zero reactive clones of total 984 TCCs generated from non-celiac tissues. We found that almost half of the gluten-reactive TCC were specific to the four immunodominant DQ2.5-glia-α1/DQ2.5-glia-α2 and DQ2.5-glia-ω1/DQ2.5-glia-ω2 epitopes, whereas combined, only a quarter of the TCC were specific to any of the other 23 HLA-DQ2.5 restricted gluten epitopes. More surprisingly, we found that one third of the gluten-reactive TCC were specific to some not yet discovered epitopes in gluten.

To our knowledge, only one previously published study has estimated the frequency of gluten-reactive T cells by direct cloning (10). Of the four Marsh 3 patients investigated in this study from 2013, the average frequency was 1.0% (0.5% - 1.8%) of all TCC tested. Our results are not statistically significantly different from this estimate. However, we could speculate that since the intestinal biopsies used in Bodd et al. (10) were mostly Marsh 3A, including from two subjects that were treated patients on gluten-free diet, less inflammation could be the cause of the somewhat lower frequency estimate found in that study.

Interestingly, a refractory celiac disease patient who had massive inflammation in the duodenum despite gluten-free diet, had gluten-reactive T cells with the same frequency as other untreated celiac disease patients with similar Marsh 3 changes. This finding is in accordance with Bodd et al. (10), where the frequency of gluten-reactive T cells estimated by direct cloning was correlated with the Marsh score, rather than with the gluten consumption status. Overall, our results further corroborate the notion that tissue injuries in celiac disease are driven by gluten-reactive T cells.

An important assumption for our estimate of gluten-reactive T cells based on in vitro cloning, is that all T cells have equal potential to grow under the in vitro culture conditions used. However, it is conceivable that recently in vivo activated gluten-reactive T cells could be less prone to proliferate in vitro due to activation-induced refractoriness. If that is the case, the frequency we have calculated would under-estimate the true frequency of gluten-reactive T cells. Similarly, our estimate would not include gluten-reactive cells that do not proliferate well under the in vitro culture and testing conditions used, including regulatory T cells. On the other hand, since 8%-29% of the TCC could in reality be oligoclonal as estimated by the growth in Terasaki wells, the frequency we have shown may slightly over-estimate the true frequency of gluten-reactive T cells. This last notion is supported by flow cytometry data where 7 of 37 gluten-specific TCC showed > 5% contamination with CD8+ T cells. Since the staining was performed after many rounds of expansion, the CD8+ T cells may have originated from unintended out-growth of feeder cells that is known to occur occasionally despite irradiation.

We have previously reported that gluten-reactive memory T cells persist for decades in celiac disease patient on gluten-free diet (11). A definite cure from the disease is therefore dependent on the eradication or ‘re-education’ of existing gluten-reactive memory T cells that were primed and formed during the active disease phase. It is clearly a shortcoming of the current study that we did not assess the frequency of gluten-reactive memory T cells in treated celiac disease patients on gluten-free diet. As a method for assessing the frequency of gluten-reactive T cells, direct cloning is labor-intensive. Other methods such as tetramer staining is much less labor-intensive assuming the availability of the key reagents. The current study and Bodd et al. (10) have both found that around half of all gluten-reactive T cells are specific to one of the four immunodominant DQ2.5-glia-α or DQ2.5-glia-ω epitopes. By using this knowledge, the total frequency of gluten-reactive T cells can thus be extrapolated from tetramer-staining data.

It is a weakness of this study that TG2-deamidated gluten was used as the sole antigen during the initial screening. It is therefore likely that T cells that recognize hordein, secalin and avenin specific sequences did not pass this screening and were thus not included in the downstream analysis. This may explain the low prevalence of hordein, secalin and avenin specific T cells found in this study, where only one single DQ2.5-sec-3-specific TCC was found. T cells that recognize the DQ2.5-hor-3a epitope have been reported to be relatively prevalent in patients that consume barley (8). However, we did not find any DQ2.5-hor-3a TCC in our study possibly due to the limited variety of antigen that was used during the initial screening.

It was a surprise that up to one third of the gluten-reactive TCC apparently have unknown epitope specificity. In comparison, Bodd et al. found two TCC with unknown epitope specificity among 15 gluten-reactive in total (10). Decades of epitope hunting has resulted in an extensive list of total 27 HLA-DQ2.5-restricted gluten/secalin/hordein/avenin epitopes (1). Nevertheless, it is clear that there are still yet uncharacterized HLA-DQ2.5-restricted gluten epitopes. In our study, ‘gluten-only’ TCC was found in five of the six subjects, and was absent only in the subject that had the lowest number of gluten-reactive TCC (n=4). Future studies will show whether these TCC respond to a few commonly recognized novel epitopes, or a multitude of private epitopes.

The original contributions presented in the study are included in the article/ Supplementary Material . Further inquiries can be directed to the corresponding author.

The studies involving human participants were reviewed and approved by The regional committee for medical research ethics. The patients/participants provided their written informed consent to participate in this study.

S-WQ — Study design, data collection and analysis, and writing of paper. SD-K, LE — Data collection. KL — Procurement of materials, LS — Intellectual contribution to direction of study. All authors contributed to the article and approved the submitted version.

This work received support from Centre for Immune Regulation (Research Council of Norway, project 179573/V40 through the Centre of Excellence funding scheme and project 233885), and grants from the Stiftelsen Kristian Gerhard Jebsen for Coeliac Disease Research Centre (SKGJ-MED-017).

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.