Celiac disease (CD) is an HLA-associated disorder characterized by a harmful T cell response to dietary gluten. It is not understood why most individuals who carry CD-associated HLA molecules, such as HLA-DQ2.5, do not develop CD despite continuous gluten exposure. In this study, we have used tetramers of HLA-DQ2.5 bound with immunodominant gluten epitopes to explore whether HLA-DQ2.5+ healthy individuals mount a specific CD4+ T cell response to gluten. We found that gluten tetramer-binding memory cells were rare in blood of healthy individuals. These cells showed lower tetramer-binding intensity and no signs of biased TCR usage compared with gluten tetramer-binding memory T cells from patients. After sorting and in vitro expansion, only 18% of the tetramer-binding memory cells from healthy subjects versus 79% in CD patients were gluten-reactive upon tetramer restaining. Further, T cell clones of tetramer-sorted memory cells of healthy individuals showed lower gluten-specific proliferative responses compared with those of CD patients, indicating that tetramer-binding memory cells in healthy control subjects may be cross-reactive T cells. In duodenal biopsy specimens of healthy control subjects, CD4+ T cells were determined not to be gluten reactive. Finally, gluten tetramer-binding cells of healthy individuals did not coexpress regulatory T cell markers (Foxp3+ CD25+) and cultured T cell clones did not express a cytokine profile that indicated immune-dampening properties. The results demonstrate that healthy HLA-DQ2.5+ individuals do not mount a T cell response to immunodominant gluten epitopes of CD.
Celiac disease (CD) is a chronic inflammatory disease of the small intestine, characterized by a harmful immune response to dietary gluten. Gluten is a collective term for grain storage proteins of wheat, rye, and barley (1). In wheat, gluten consists of the subcomponents gliadins and glutenins. CD develops in genetically susceptible individuals, and MHC class II genes, in particular those encoding HLA-DQ2.5, but also HLA-DQ2.2 and HLA-DQ8, are the most important genetic factor (2, 3). These disease-associated HLA-DQ molecules, but not other HLA molecules carried by the patients, present deamidated gluten peptides to CD4+ T cells (4, 5). More than 20 HLA-DQ2.5–restricted T cell epitopes have been reported (6), and some, like the DQ2.5-glia-α1a, DQ2.5-glia-α2, and DQ2.5-glia-ω2 epitopes, are immunodominant (7). The gluten-reactive CD4+ T cells from CD patients produce IFN-γ and IL-21 upon activation (8). Gluten-reactive CD4+ T cells can easily be cultured from duodenal biopsies of CD patients, but not from control subjects (9). Thus, gluten-reactive T cells can be considered the master regulators of immune reactions that create the celiac lesion of the upper small bowel, lesions characterized by leukocyte infiltration, blunting of the intestinal villi, and hyperplasia of the crypts (10).
It is not understood why most HLA-DQ2.5+ individuals do not develop CD despite exposure to dietary gluten. One possibility is that regulatory T cells (TREG) can keep harmful antigluten immune responses in check. Studies in mice have shown that dietary Ags can induce suppressive IL-10–producing T cells (11). In humans, the role of oral tolerance is less well understood. In particular, it is not known whether HLA-DQ2.5+ individuals mount a T cell response to gluten and whether such a response involves the generation of suppressive Foxp3+ TREG and Foxp3− type 1 TREG (TR1). It was reported that CD patients have intestinal, HLA-DQ2.5–restricted TR1 cells that secrete IL-10 and IFN-γ upon stimulation with gluten that can inhibit proliferation of gluten-reactive Th0 cells (12). However, such cells were not observed in healthy subjects. Further, it has been demonstrated that there is increased expression of Foxp3 and increased numbers of CD4+ TREG (CD25+ Foxp3+) in the small-bowel biopsy specimens of CD patients compared with control subjects (13, 14).
Lately, a refined MHC-tetramer–based assay has improved the sensitivity for detection of Ag-specific T cells (15). Using this technique, we recently demonstrated in CD patients a large expansion of circulating, gut-homing, CD4+ effector memory T cells (TEM) that bound tetramers of HLA-DQ2.5 molecules complexed with immunodominant gluten epitopes. We also found circulating gluten tetramer-binding TEM in healthy individuals, albeit at frequencies that were 100-fold lower than in CD patients, and there was no significant increase in the expression of gut-homing markers among these cells (16). In another recent study, we demonstrated that MHC tetramers can be used to identify gluten-reactive CD4+ T cells in lamina propria lymphocytes (LPLs) of CD patients (17).
In this study, we used tetramer-based techniques to explore the T cell response to gluten in the blood and gut of healthy HLA-DQ2.5+ individuals. The results demonstrate that the majority of healthy individuals do not mount any T cell response, including TREG or TR1 cell response, to the immunodominant gluten epitopes involved in CD.
Materials and Methods
We included only individuals who were HLA-DQ2.5 (i.e., DQA1*05 and DQB1*02), of whom 42 were healthy control individuals, 23 untreated CD (UCD) patients, and 17 treated CD (TCD) patients. Nineteen of the control individuals were anonymous blood donors at the blood bank of Oslo University Hospital (OUS), and from these we obtained buffy coat made from 450 ml full blood. Sera from three of the blood bank donors tested negative for IgA antitransglutaminase 2 and IgG antideamidated gluten peptides. For the remaining blood bank donors, we have no information on their clinical state other than that CD is an exclusion criterion for blood donation. The remaining participants donated 50–100 ml blood, and in 14 cases, 12 duodenal biopsy specimens were taken after giving informed written consent. These latter participants were examined at OUS during routine follow-up. CD patients were diagnosed according to guidelines from the American Gastroenterological Association (10). The study was approved by Regional Committee for Medical and Health Research Ethics South-East Norway (S-97201).
Identification of tetramer-positive CD4+ T cells in PBMC and LPLs
Tetramer-binding CD4+ T cells were enriched from PBMCs (16) or identified among LPLs (17) as described elsewhere. We used DQ2.5 tetramers presenting the following gluten epitopes: DQ2.5-glia-α1a (QLQPFPQPELPY, with underlined 9mer core sequence), DQ2.5-glia-α2 (PQPELPYPQPE), DQ2.5-glia-ω1 (QQPFPQPEQPFP), DQ2.5-glia-ω2 (FPQPEQPFPWQP), and DQ2.5-glia-γ2 (QGIIQPEQPAQL). These tetramers were produced and labeled with either allophycocyanin or PE as described previously (18). We used the same Abs, frequency calculations, and gating strategies as previously described (16, 17) to identify or sort cells of interest on a LSR II, FACSAria I, or FACSAria II (BD Biosciences). Vβ6.7+ T cells were identified with FITC OT-145 (Endogen). For further phenotyping of tetramer-binding cells in blood, several different gluten tetramers labeled with the same fluorophore were combined. In six control subjects and four TCD patients, tetramers presenting the DQ2.5-glia-α1a and DQ2.5-glia-α2 epitopes were used. In five other control subjects, tetramers presenting the DQ2.5-glia-α1a, DQ2.5-glia-α2, and DQ2.5-glia-ω2 epitopes were combined, and these were combined with DQ2.5-glia-ω1 in 17 additional controls or DQ2.5-glia-γ2 in two other additional controls. For identification of potential TREG in blood and gut, we stained with FITC anti-CD25, allophycocyanin anti-CD127 (both from Biolegend), and PE anti-Foxp3 (eBioscience). For exclusion of dead cells, we used LIVE/DEAD Viable Violet stain (Invitrogen). Statistical significance was calculated with the Mann–Whitney U test by using GraphPad Prism 5 (GraphPad Software).
Generation and functional testing of T cell lines and T cell clones
Both T cell lines (TCLs) from tetramer-sorted LPLs and T cell clones (TCCs) from tetramer-sorted blood T cells were generated and cultured without exposure to gluten Ags as previously described (19). We validated the specificity of growing TCLs and TCCs by tetramer restaining, and in the case of blood-derived TCCs, also by T cell proliferation assays (16). Specific TCCs were further tested for their proliferative response to titrated amounts of the deamidated and the native form of the same epitopes. We used HLA-DQ2.5–homozygous EBV-transformed cells (HW #9023) to present the following gluten peptides: DQ2.5-glia-α1a [deamidated: QLQPFPQPELPY (underlined 9mer core sequence) and native: QLQPFPQPQLPY], DQ2.5-glia-α2 (deamidated: PQPELPYPQPQL and native: PQPQLPYPQPQL), DQ2.5-glia-ω2 (deamidated: FPQPEQPFPWQP and native: FPQPQQPFPWQP), and DQ2.5-glia-γ2 (deamidated: GIIQPEQPAQL and native: GIIQPQQPAQL). TCCs, at a concentration of 75,000 cells/well, were also tested in duplicates for their proliferative response to titrated amounts of plate-bound anti-CD3 (4-fold dilutions from 5 μg/ml) in the presence of soluble anti-CD28 (1 μg/ml) (both from Biolegend) in a 24-h assay, as described elsewhere (20). We assessed T cell proliferation by [3H]thymidine incorporation. For cytokine analysis, TCCs, at a concentration of 100,000 cells/well, were stimulated with plate-bound anti-CD3 (5 μg/ml) and soluble anti-CD28 (1 μg/ml) in duplicates, as described elsewhere (20). The amount of IL-10, IL-4, and IFN-γ in the supernatant after 24 h was measured using the Bio-Plex Pro Assay (Bio-Rad), according to the manufacturer’s instructions.
We used a SMART-based protocol with template-switch PCR to identify the TRAV and TRBV genes of the 17 TCCs generated from tetramer-sorted memory T cells of healthy control subjects that showed specific tetramer restaining (21, 22). We sequenced the PCR products and analyzed generated sequences as described previously (21). We compared the TCR sequences from memory TCCs of the control individuals with a panel of >200 unique TCR sequences from gluten-reactive, tetramer-sorted memory cells from CD patients (23).
Healthy HLA-DQ2.5+ individuals have few CD4+ gluten tetramer-positive memory T cells in blood
We first investigated the quantity and the phenotype of circulating gluten-reactive CD4+ T cells from healthy HLA-DQ2.5+ individuals. PBMCs were stained with gluten tetramers followed by bead-enrichment and flow-cytometric assessment. The average frequency of cells that stained tetramerized DQ2.5:DQ2.5-glia-α1a and DQ2.5:DQ2.5-glia-α2 molecules was 0.9 per million CD4+ T cells (Fig. 1A). This frequency is 5–12 times lower than the frequency we observed in UCD and TCD patients (on average, 11.0 and 4.7 per million CD4+ T cells, respectively; Fig. 1A). Of note, the frequency is also 5–10 times lower than frequencies of CD4+ T cells stained by tetramers presenting self-Ags or viral Ags in unexposed individuals (24).
To obtain more events for phenotyping of tetramer-binding cells, we combined several different gluten tetramers labeled with the same fluorophore. In control subjects, we found that the fraction of memory cells (CD45RA−) was significantly lower in tetramer-binding cells (35%) than in the tetramer-negative cells (55%) (Fig. 1B). By contrast, in UCD and TCD patients, 86 and 78% of tetramer-binding cells on average were memory cells, respectively. For comparison, among T cells binding tetramers presenting self-Ag or viral Ag epitopes in unexposed individuals, frequencies of memory T cells were found to range between 35 and 85% (24).
Compared with the striking expansion of T cells bearing the TEM phenotype (CD45RA−/CD62L−) among T cells that bind gluten tetramers in CD patients, the fractions of memory T cells being TEM were similar among tetramer-binding and tetramer-negative cells in the control subjects (Fig. 1C). Thus, overall, there are few gluten tetramer-binding T cells among healthy HLA-DQ2.5+ individuals, and among these T cells, there is an underrepresentation of cells with a memory phenotype.
Given the low number of T cells that bound gluten tetramers in healthy control subjects, we wanted to ascertain the Ag specificity of the cells staining with the tetramers. We sorted tetramer-binding memory T cells from seven control subjects and cloned these cells in an Ag-independent manner. Compared with tetramer-sorted memory T cells from CD patients, of which 77% (164/212) showed positive tetramer restaining, only 18% (17/97) of the TCCs derived from memory T cells of healthy control subjects were tetramer positive. In T cell proliferative assays, only 16% (16/97) of TCCs generated from the control subjects showed specific Ag-dependent proliferation, compared with 69% (147/212) of tetramer-sorted TCCs generated from CD patients. In healthy control subjects, tetramer-binding naive T cells (TN) (CD45RA+/CD62L+) were three times more frequent than tetramer-binding memory cells (Fig. 1B). Of tetramer-sorted TN, 59% (101/170) of the cultured TCCs retained tetramer staining and 43% (73/170) showed an Ag-specific proliferative response.
In summary, in blood of healthy HLA-DQ2.5+ subjects, we found few CD4+ memory T cells that bound gluten tetramers, and only one in five of these cells seemed to stain specifically.
Gluten tetramer-binding CD4+ LPLs from healthy HLA-DQ2.5+ individuals are not gluten specific
We next studied whether healthy HLA-DQ2.5+ individuals have gluten tetramer-binding CD4+ T cells in the gut, and to what extent these were gluten reactive. We generated single-cell suspensions from 6–12 duodenal biopsy specimens of seven HLA-DQ2.5+ healthy subjects and three UCD patients. We stained each sample with a combination of the four gluten tetramers DQ2.5:DQ2.5-glia-α1a, DQ2.5:DQ2.5-glia-α2, DQ2.5:DQ2.5-glia-ω1, and DQ2.5:DQ2.5-glia-ω2. In the UCD patients, between 0.3 and 1.3% of CD4+ T cells were stained by the tetramers. In the healthy control subjects, the number of tetramer-binding cells was <0.1–0.2% of total CD4+ T cells (Fig. 2A, 2B).
Tetramer-binding cells from three UCD patients and seven healthy control subjects were sorted and cultured as TCLs in an Ag-independent manner in vitro before restaining with the same tetramers. In the UCD patients, between 81 and 94% of CD4+ cells in the TCL retained tetramer staining. In the healthy control subjects, no tetramer-sorted T cells were tetramer-positive upon retesting (Fig. 2A, 2B). Thus, the absence of tetramer restaining in gut-derived tetramer-sorted TCLs of healthy HLA-DQ2.5+ control subjects indicates an absence of genuine gluten-reactive T cells in the duodenal lamina propria of healthy subjects.
Gluten tetramer-binding memory T cells in blood of CD patients and control subjects are different
In contrast with LPLs of healthy control subjects that did not seem to contain genuine gluten-reactive CD4+ T cells, 18% (17/97) of the tetramer-sorted TCCs from circulating memory T cells of healthy control subjects were tetramer-positive upon restaining. We therefore did an in-depth analysis of the phenotype and molecular characteristics of circulating tetramer-binding memory T cells in healthy control subjects. We wanted to investigate whether these cells were similar to their counterparts found in CD patients, with regard to TCR repertoire and Ag specificity.
In concordance with previous findings (21, 25), we found that TCR Vβ6.7 (encoded by the TRBV7-2 gene segment) was overrepresented among circulating TEM that bound tetramerized DQ2.5:DQ2.5-glia-α2 in CD patients (median: 44%) (Fig. 3). In sharp contrast, we found no biased Vβ6.7 usage among TEM from controls binding the same tetramer (median: 0%). Likewise, we did not observe Vβ6.7 bias in DQ2.5-glia-α2 tetramer-binding TN, neither from healthy control subjects nor from CD patients. Notably, similar TCR bias was not seen among CD4+ TEM staining the DQ2.5:DQ2.5-glia-α1a tetramer or among tetramer-negative cells in CD patients or control subjects (Supplemental Fig. 1), indicating that this TCR bias is an epitope-specific phenomenon restricted to DQ2.5-glia-α2–reactive memory T cells that have been through an Ag-driven clonal expansion. One possibility for the biased TCR usage in the Ag-expanded population could be a selective expansion of TCRs that had higher binding affinity to the cognate Ag. We found some evidence for this notion. Comparing the tetramer-staining intensities of tetramer-binding T cells in UCD and TCD patients, we observed higher staining intensities in TEM and central memory T cells (TCM) (CD62L+ CD45RA−) compared with TN. The ratio of median fluorescence intensity (MFI) of TEM and TCM relative to the MFI of TN was 1.4 and 1.1 in UCD patients and 1.5 and 1.2 in TCD patients, respectively. By contrast, in healthy control subjects, the MFI of TEM and TCM relative to TN was 0.7 and 0.9, respectively (Fig. 4A, 4B). The ratio between MFI of TEM and TN was significantly higher in both UCD and TCD patients compared with control subjects (Fig. 4B).
Next, we compared the proliferative responses to titrated amounts of Ag between TCCs from CD patients and the minority of TCCs from control subjects that were gluten-reactive upon retesting. In general, the TCCs from both CD patients and control subjects were reactive to the deamidated and not the native version of the epitopes (Fig. 5A, 5B). We used the EC50 value (defined as the peptide concentration estimated to give 50% of the total maximum proliferative response) as a marker for T cell sensitivity to the deamidated gluten epitopes (Fig. 5A, 5B). The EC50 values of the 7 TEM TCCs from control subjects were significantly lower than those of 14 TEM TCCs randomly selected from one UCD and three TCD patients (p < 0.01; Fig. 5C). In contrast, there was no significant difference between the EC50 values of these seven TEM TCCs from control subjects and that of nine randomly selected TN TCCs from control subjects. Overall, the T cell sensitivities, judged by EC50 estimates, were in the medium-high range in all groups of TCCs tested, except those of TEM in CD patients. To exclude that general differences in proliferative potential could be a confounder, we tested the seven TEM TCCs from control subjects and eight TEM TCCs from CD patients for their proliferative response to titrated amounts of immobilized anti-CD3 Abs. We observed no significant differences in EC50 values (the anti-CD3 concentration that gave 50% of the total maximum proliferative response) between TEM TCCs of CD patients and control subjects (Fig. 5D).
Among the gluten-reactive memory TCCs from healthy control subjects, we found one single DQ2.5-glia-α2–reactive TEM TCC that expressed a TCR that was prototypic in expanded DQ2.5-glia-α2–reactive T cells in CD patients, namely, the TRBV7-2 gene segment with a conserved arginine in position 5 paired with TRAV26-1 (21, 25). Such cells were not expanded in controls as judged by Vβ6.7 staining that targets the TRBV7-2 gene segment (Fig. 3). The healthy control individual who had this one prototypic DQ2.5-glia-α2–specific TEM TCC did not have CD as far as we could ascertain. The serum from this individual was negative for the CD-specific Abs to TG2 and deamidated gluten peptides. Because of ethical reasons, we were unable to address this further by examination of duodenal biopsies because this was an anonymous blood donor. The other gluten-reactive TCCs generated from the memory T cells of this and other control subjects did not use TCRs typical of gluten-reactive T cells from CD patients (data not shown) (23).
Although a few circulating gluten-reactive memory T cells in CD patients and control subjects may have overlapping TCR usage, we demonstrate that such cells in control subjects are not expanded and show modest proliferative response to gluten epitopes compared with gluten-reactive memory TCCs from blood of CD patients.
Gluten tetramer-binding cells in healthy individuals do not display a TREG or TR1 cell phenotype
Finally, we studied whether tetramer-binding T cells in blood and duodenal biopsies of healthy HLA-DQ2.5+ individuals displayed a TREG phenotype (Fig. 6A). The fractions of CD25+CD127− blood cells, that is, potentially TREG (26), were significantly lower within tetramer-binding than within tetramer-negative cells of the TEM and TCM compartments. Within TN, the fractions of CD25+CD127− cells were similar among tetramer-binding and tetramer-negative cells (Fig. 6B). Further, we stained PBMCs of 11 control subjects and LPLs of six control subjects with CD25 and Foxp3, because TREG coexpress these two markers (27). The tetramer-binding CD4+ memory blood cells (Fig. 6C) and CD4+ LPLs (Fig. 6D) did not coexpress CD25 and Foxp3, and the expression was significantly lower than within the tetramer-negative cells (Fig. 6E).
TR1 cells do not express specific surface markers like TREG, but have a specific cytokine production profile (high levels of IL-10, moderate levels of IFN-γ, and no IL-4) (28). Accordingly, we tested whether seven blood-derived, TEM TCCs that retained tetramer staining and gave epitope-specific proliferation from controls expressed a TR1 cytokine profile, and we compared the cytokine profiles with that of eight blood-derived, gluten-specific TEM TCCs of CD patients. We chose to stimulate the TCCs with anti-CD3 because such stimulation may elicit relatively high amounts of IL-10 (20). The TCCs from control subjects secreted similar amounts of IL-10, IFN-γ, and IL-4 as the TEM TCCs of CD patients (Fig. 7). Thus, both sets of TCCs tested expressed similar cytokine profiles, which were different from the typical TR1 cytokine profile.
Taken together, our data demonstrate that CD4+ T cells in blood and duodenal biopsies of healthy HLA-DQ2.5+ individuals that bind tetramers presenting immunodominant gluten epitopes do not display a TREG or TR1 cell phenotype.
Despite carrying HLA genes associated with high risk for contracting CD, most HLA-DQ2.5+ individuals do not develop this disorder. The observations of our study strongly indicate that the explanation for this disease evasion is not due to immune suppression mediated by CD4+ T cells recognizing immunodominant gluten epitopes in the context of HLA-DQ2.5 molecules. Using HLA-DQ2.5 tetramers presenting gluten epitopes, no genuine gluten-reactive T cells could be found in gut biopsies of healthy subjects. In blood, few gluten-reactive memory T cells binding the tetramers could be identified in healthy subjects. The cells did not express a regulatory phenotype, and TCCs established from tetramer-binding cells showed lower gluten-specific proliferation and generally used a different TCR repertoire compared with TCCs established from CD patients. Importantly, the TCCs did not have a cytokine profile of TREG.
We did identify a rare population of memory T cells in blood from healthy HLA-DQ2.5+ subjects that were reactive to deamidated gluten T cell epitopes. A key issue is whether these cells were primed against gluten Ag in vivo. For several reasons we favor the explanation that they are cross-reactive T cells primed against Ags other than gluten. First, it is known that cross-reactive T cells can be picked up by tetramer staining in human adult blood at frequencies similar to or even higher than what we observed (24). Second, the gluten tetramer-binding memory T cells of healthy subjects did not show signs of a biased TCR repertoire typical of responses to certain gluten epitopes in CD patients (21, 25). Third, the gluten-reactive memory TCCs of healthy subjects showed lower gluten-specific proliferation than those of CD patients. Fourth, the tetramer-binding memory T cells in blood of healthy control subjects did not express markers of gut homing (16). Additionally and importantly, the reactive cells did not display characteristics of TREG.
The gut is the hotbed of TREG. Studies of mice have shown that Foxp3+ TREG specific to orally administrated Ags have a role in inducing oral tolerance (11, 29). A study of HLA-DQ2.5/gliadin-TCR double-transgenic mice that did not develop CD pathology found that TR1-like, IL-10–producing cells were upregulated in response to orally administrated deamidated gliadin, implicating a role of these cells (30). A human study reported that there were HLA-DQ2.5–restricted, gluten-reactive TR1 cells present in gut biopsies of CD patients, but not control subjects (12). Other human studies reported that LPLs expressing Foxp3, CD25, and CD4 were present at higher frequencies in UCD patients than in control subjects (13, 14), and that CD may evolve because of T lymphocytic resistance to suppressive TREG (14). Ag specificity of the presumed TREG was not ascertained in these studies. Interestingly, by looking at gluten-specific cells with MHC-tetramer staining, we did not find any tetramer-binding cells in healthy subjects that costained the TREG markers Foxp3 and CD25, neither within memory cells in blood nor in lamina propria. Further, circulating tetramer-binding TEM and TCM in healthy individuals were not CD25+CD127−, that is, expressing a TREG phenotype. In addition, the gluten-reactive TCCs of healthy control subjects did not secrete high amounts of IL-10 but secreted similar amounts of IFN-γ and IL-4 compared with gluten-reactive TEM from CD patients. Together, these data indicate that active suppression from gluten-reactive TREG or TR1 cells does not play a crucial role in healthy HLA-DQ2.5+ individuals, and this supports the notion that most controls do not develop a T cell response to deamidated gluten.
A weakness of our study is that we have excluded from our analysis T cells that would recognize gluten Ags in the context of HLA molecules other than HLA-DQ2.5. Such T cells could potentially prevent harmful immune responses to gluten in all subjects, also those who do not express HLA-DQ2.5. In contrast, subjects who do not express HLA-DQ2.5 (or HLA-DQ2.2 or HLA-DQ8) are not at risk for developing the disease and would not be in need of gluten-reactive regulatory cells. TREG are most likely to be found in the subjects who express the CD-associated HLA-DQ molecules and who stay free of the disease. Thus, we focused our studies on HLA-DQ2.5–expressing subjects. Ideally, we should also have surveyed T cell responses to native gluten epitopes. We used the powerful MHC-tetramer technology, which, despite its ability to phenotypically characterize and to isolate rare cells, is limited to the analysis of known peptide epitopes that make kinetically stable MHC–peptide complexes. Native gluten peptides that make stable complexes to HLA-DQ2.5 are yet to be identified. Hence we were unable to include native gluten peptides in our analysis.
Many recent studies have performed enrichment of MHC tetramer-binding cells to enumerate rare Ag-specific T cell populations (15, 23, 31). In this study, we have also used this technique. However, when we cultured in vitro memory cells of controls that had been stained and sorted with tetramers, only 18% of the TCCs were Ag specific upon tetramer retesting versus 59% of their more frequent tetramer-positive TN and 77% of the even more numerous tetramer-positive memory T cells in CD patients. This demonstrates how important it is to confirm the specificity when frequencies and phenotypic information are extrapolated from tetramer-staining experiments, especially when the population size is small. This is often not taken into consideration (15, 31).
In summary, our data show that the majority of HLA-DQ2.5+ healthy individuals do not acquire a specific T cell response to deamidated gluten despite regular gluten exposure. The absence of gluten-reactive cells includes absence of TREG and TR1 cells, which conceivably could have played a role to control a harmful antigluten response in these genetically predisposed individuals. Further, our results demonstrate that caution should be exerted on the estimation of the frequencies of rare tetramer-binding cells in the absence of in vitro T cell cloning and specificity testing.
We thank all the subjects who donated biological material to this study, Marie K. Johannesen for technical assistance, and Bjørg Simonsen for tetramer production. We also thank Jorunn Bratlie, Merete Gedde-Dahl, and the blood bank at OUS for supply of blood samples and the flow cytometry core facility and Yan Zhang for assisting with flow cytometry.
This work was supported by Research Council of Norway Grants 179573/V40 (through its Centre of Excellence funding scheme) and 233885, South-Eastern Norway Regional Health Authority Grant 2011050, and the University of Oslo.
The online version of this article contains supplemental material.
Abbreviations used in this article:
The authors have no financial conflicts of interest.