Tolerance to Ingested Deamidated Gliadin in Mice is Maintained by Splenic, Type 1 Regulatory T Cells

GASTROENTEROLOGY 2011;141:610 – 620

Tolerance to Ingested Deamidated Gliadin in Mice is Maintained by Splenic, Type 1 Regulatory T Cells M. FLEUR DU PRÉ,* ANNE E. KOZIJN,* LISETTE A. VAN BERKEL,* MARIËTTE N. D. TER BORG,* DICKY LINDENBERGH–KORTLEVE,* LISE TORP JENSEN,‡ YVONNE KOOY–WINKELAAR,§ FRITS KONING,§ LOUIS BOON,储 EDWARD E. S. NIEUWENHUIS,** LUDVIG M. SOLLID,¶ LARS FUGGER,‡,# and JANNEKE N. SAMSOM* *Department of Pediatrics, Erasmus Medical Center – Sophia Children’s Hospital, Rotterdam, The Netherlands; ‡Clinical Institute, Aarhus University Hospital, Skejby Sygehus, Aarhus, Denmark; §Department of Blood Transfusion and Immunohematology, Leiden University Medical Center, Leiden, The Netherlands; 储Bioceros BV, Utrecht, The Netherlands; ¶Centre for Immune Regulation, Institute of Immunology, University of Oslo and Oslo University Hospital – Rikshospitalet, Oslo, Norway; **Department of Pediatric Gastroenterology, Wilhelmina Children’s Hospital, University Medical Centre Utrecht, Utrecht, The Netherlands; and #Department of Clinical Neurology and MRC Human Immunology Unit, Weatherall Institute of Molecular Medicine, John Radcliffe Hospital, University of Oxford, Oxford, England

BASIC AND TRANSLATIONAL AT

BACKGROUND & AIMS: Patients with celiac disease have permanent intolerance to gluten. Because of the high frequency of this disorder (approximately 1 in 100 individuals), we investigated whether oral tolerance to gluten differs from that to other food proteins. METHODS: Using transgenic mice that express human HLA-DQ2 and a gliadin-specific, humanized T-cell receptor, we compared gluten-specific T-cell responses with tolerogenic mucosal T-cell responses to the model food protein ovalbumin. RESULTS: Consistent with previous findings, the ovalbumin-specific response occurred in the mesenteric lymph nodes and induced Foxp3⫹ regulatory T cells. In contrast, ingestion of deamidated gliadin induced Tcell proliferation predominantly in the spleen but little in mesenteric lymph nodes. The gliadin-reactive T cells had an effector-like phenotype and secreted large amounts of interferon gamma but also secreted interleukin-10. Despite their effector-like phenotype, gliadin-reactive T cells had regulatory functions, because transfer of the cells suppressed a gliadin-induced, delayed-type hypersensitivity response. CONCLUSIONS: Ingestion of deamidated gliadin induces differentiation of tolerogenic, type 1 regulatory T cells in spleens of HLA-DQ2 transgenic mice. These data indicate that under homeostatic conditions, the T-cell response to deamidated gliadin is tolerance, which is not conditioned by the mucosal immune system but instead requires interleukin-10 induction by antigen presentation in the spleen. Keywords: Food Allergy; Mucosal Tolerance Induction; Treg Cells; Immune Regulation.

C

eliac disease is one of the most common food intolerances, affecting approximately 1% of the Western world. Celiac disease is induced by the ingestion of gluten from wheat, rye, and barley and can be treated by avoiding gluten in the diet. A central role for CD4⫹ T cells in the pathogenesis of celiac disease is well established. Inflammatory gluten-specific CD4⫹ T cells can be isolated from the small intestinal mucosa of patients with celiac disease but not of healthy individuals.1 These interferon (IFN)-␥– secreting gluten-specific T cells2 recognize a diverse array of gluten peptides that are presented exclusively in the

context of HLA-DQ2 or HLA-DQ8.3 In agreement, the majority of patients with celiac disease carry HLA-DQ2, whereas HLA-DQ2–negative patients usually express HLA-DQ8.3 A crucial event in the binding of gluten peptides to these HLA molecules is peptide deamidation by the enzyme transglutaminase 2 (TG2). Gluten peptides generally have a low affinity for HLA-DQ2 and HLA-DQ8, but TG2 strongly facilitates this interaction by the introduction of negative charges in gluten peptides.4,5 Despite our understanding of the role of HLA-DQ2 (and HLA-DQ8) in the development of celiac disease, it is still unclear why oral tolerance to complex gluten is permanently broken in a significant proportion of the HLADQ2– carrying population whereas other HLA-DQ2–positive individuals remain tolerant. Oral tolerance to soluble proteins is strictly regulated by the microenvironment in mucosa-draining lymphoid tissue. Within mesenteric lymph nodes (MLN) and Peyer’s patches (PP), food protein–specific regulatory T (Treg) cells differentiate that acquire forkhead box P3 (Foxp3) under the control of the mucosal environment.6 –9 Because oral tolerance to soluble protein is such a controlled mechanism, it can be questioned why permanent intolerance to gluten is relatively frequent. We hypothesize that tolerance to gluten may be induced by different mechanisms compared with other food proteins. To test this hypothesis, we have generated mice that transgenically express HLA-DQ2 and a humanized HLADQ2–restricted gliadin-specific T-cell receptor (TCR) derived from a T cell of the celiac lesion and determined the response to orally administered deamidated gliadin in these mice.

Abbreviations used in this paper: BM-DC, bone marrow– derived dendritic cells; DC, dendritic cell; DTH, delayed-type hypersensitivity; FACS, fluorescence-activated cell sorter; IFN, interferon; IL, interleukin; MLN, mesenteric lymph nodes; MTX, methotrexate; OVA, ovalbumin; PP, Peyer’s patches; TG2, transglutaminase 2; TCR, T-cell receptor; Treg cell, regulatory T cell. © 2011 by the AGA Institute 0016-5085/$36.00 doi:10.1053/j.gastro.2011.04.048

Materials and Methods Generation of HLA-DQ2-Transgenic and Gliadin-TCR-Transgenic Mice For HLA-DQ2-transgenic mice, a 68-kilobase fragment containing the DQA1*0501 and DQB1*0201 (DQ2.5) genes, including promoters and regulatory elements, was purified from bacteriophage clone p797a11 and used for microinjection. For generation of TCR chimeric constructs, genomic DNA was purified from the T-cell clone 4.32, which was generated from a small intestinal biopsy specimen of a patient with celiac disease challenged in vitro with pepsin and trypsin-digested gliadin.10 V(D)J regions of rearranged TCR␣ and TCR␤ chains were polymerase chain reaction amplified from genomic DNA and cloned into a pair of cassette expression vectors containing the murine TCR constant sequences as well as natural mouse TCR promoter/enhancer elements.11 Transgenic mice were generated by microinjecting the constructs without vector sequences into fertilized eggs from (DBA/2⫻C57BL/6)F1 matings. Viable embryos were transferred to the oviducts of pseudo-pregnant females for development to term. Transgenic offspring were backcrossed twice to the MHCII⌬/⌬ mice.12

Adoptive Transfer BALB/c acceptor mice received 1 ⫻ 107 CD4⫹KJ1.26⫹ cells in 100 ␮L saline by intravenous injection6; DQ2 acceptor mice received 7 ⫻ 106 CD4⫹V␤1⫹ cells. One day after transfer, DQ2 mice received 75 to 100 mg of nondeamidated or deamidated gliadin by 3 to 4 oral gavages with intervals of at least 4 hours. In BALB/c acceptor mice, tolerance was induced by giving a single dose of 70 mg ovalbumin (OVA) intragastrically or immunity by giving 400 ␮g OVA intramuscularly. Spleens and draining lymph nodes (MLN for intragastric, popliteal lymph node for intramuscular) were isolated 72 hours after antigen administration for subsequent experiments.

In Vitro Restimulation For in vitro restimulation experiments, CFSE⫹ CD4⫹V␤1⫹ cells were purified by flow cytometric cell sorting from spleens of DQ2 acceptor mice. Next, 1 ⫻ 105 sorted CD4⫹V␤1⫹ cells were restimulated in vitro with 1 ⫻ 104 DQ2⫹ bone marrow– derived dendritic cells (BM-DC) and 0.5 mg/mL deamidated gliadin for 48 hours. For intracellular detection of cytokines, monensin (GolgiStop; BD Pharmingen, Woerden, The Netherlands) was added during the last 4 hours of culture. For quantitative polymerase chain reaction analysis, sorted CFSE⫹CD4⫹V␤1⫹ cells were restimulated with phorbol myristate acetate/CaI (50 ng/mL and 0.1 ␮g/mL, respectively) for 14 hours.

Delayed-Type Hypersensitivity Response Dividing CD4⫹ gliadin-TCRtg T cells were fluorescenceactivated cell sorted (FACS) from spleens of DQ2 mice that were enriched with CFSE⫹ CD4⫹ gliadin-TCRtg T cells and were fed 75 mg TG2-gliadin 1 day after cell transfer. A total of 2.5 ⫻ 105 dividing CD4⫹V␤1⫹ cells were adoptively transferred to DQ2.gliadinTCR acceptor mice by intravenous injection in 100 ␮L saline. Gliadin-specific Foxp3⫹ Treg cells were generated in vitro by culturing 5 ⫻ 105 CD4⫹V␤1⫹ cells with 2 ⫻ 104 splenic DQ2⫹CD11c⫹ dendritic cells (DC) and 0.5 mg/mL nondeamidated gliadin. At t ⫽ 96 hours, dividing CD4⫹CD25⫹V␤1⫹ cells were sorted and 2.5 ⫻ 105 cells were transferred to DQ2.gliadinTCR acceptor mice. Expression of Foxp3 in sorted

GLIADIN INGESTION ELICITS SPLENIC TR1–LIKE CELLS

611

in vitro– generated Treg cells was 83%, as determined by flow cytometry. One day after adoptive transfer, recipient mice were sensitized subcutaneously in the tail base with 100 ␮g TG2gliadin in 50 ␮L of a 1:1 incomplete Freund’s adjuvant/saline solution (Difco; BD, Alphen a/d Rijn, The Netherlands). Five days later, mice were challenged with 10 ␮g TG2-gliadin in 10 ␮L saline in both ears; after 24 hours, increases in ear thickness were determined and compared with values before challenge. As a control, a DTH response was induced in DQ2.gliadinTCR mice that did not receive transferred cells. See Supplementary Materials and Methods for additional information.

Statistics Data are expressed as mean ⫾ SD and analyzed using either Student t test or analysis of variance. P ⬍ .05 was considered significant.

Results Generation of HLA-DQ2-Transgenic and Gliadin-TCR-Transgenic Mice HLA-DQ2 transgenic mice were generated on the basis of a genomic fragment containing the DQA1*0501, DQB1*0201 (DQ2.5) genes as well as promoter and regulatory elements. One founder animal was selected, of which all offspring exhibited a similar expression of the transgene product. Gliadin-TCR transgenic mice were generated with a chimeric TCR construct in which the variable domains were derived from a human gliadin-specific T-cell clone, whereas the constant regions and regulatory elements were of mouse origin. HLA-DQ2 transgenic and gliadin-TCR transgenic mice were backcrossed twice to mice lacking all conventional MHCII genes.12 HLA-DQ2.MHCII⌬/⌬ transgenic mice (DQ2 mice) were crossed with gliadinTCR.MHCII⌬/⌬ transgenic mice (gliadinTCR mice) to obtain HLA-DQ2.gliadin-TCR.MHCII⌬/⌬ double transgenic mice (DQ2.gliadinTCR mice). Expression of the transgene products was assessed by flow cytometry on splenocytes of DQ2.gliadinTCR mice. HLA-DQ2 was expressed on all B cells (Figure 1A) and DCs (Figure 1B), whereas the humanized gluten-specific TCR was present on CD3⫹ T cells (Figure 1C). Analysis of the peripheral CD4/CD8 T-cell ratio confirmed proper thymic selection of CD4⫹ gliadinTCRtg T-cells in DQ2.gliadinTCR mice (Figure 1D). In contrast, ⬎95% of CD3⫹ T cells in DQ2-negative gliadinTCR.MHCII⌬/⌬ transgenic mice are CD8⫹CD4⫺ T cells and only very small numbers of CD4⫹CD8⫺, doublepositive and double-negative CD3⫹ T cells are detected in the periphery (Figure 1E). All mice are healthy and have normal intestinal morphology (data not shown).

Gliadin-TCRtg T Cells Only Respond to Deamidated Gliadin Splenocytes from DQ2.gliadinTCR mice and nontransgenic MHCII⌬/⌬ control mice were cultured in the presence or absence of deamidated gliadin (TG2-treated chymotrypsin-digested gliadin, denoted as TG2-gliadin) to determine cellular responsiveness. At 72 hours, prolif-

BASIC AND TRANSLATIONAL AT

August 2011

612

DU PRÉ ET AL

GASTROENTEROLOGY Vol. 141, No. 2

Figure 1. Transgenic mice express HLA-DQ2 on APC and gliadin-TCR on CD4⫹ T cells. FACS analysis: (A and B) DQ2.gliadinTCR splenocytes stained for HLA-DQ2 (SPV-L3) and (A) B220 or (B) CD11c. (C) Murine CD3 and human TCR V␤1 expression on DQ2.gliadinTCR splenocytes. (D) CD4 and CD8 expression, gated on CD3⫹ splenocytes from DQ2.gliadinTCR mice. (E) CD4 and CD8 expression, gated on CD3⫹ splenocytes from gliadinTCR mice. BASIC AND TRANSLATIONAL AT

eration in response to deamidated gliadin was only detected with splenocytes from DQ2.gliadinTCR mice, not from MHCII⌬/⌬ mice, indicating that there is functional presentation of the antigen in cells from DQ2.gliadinTCR mice (Figure 2A). Culture of splenic DQ2⫹ DC and CD4⫹ gliadin-TCRtg T cells with deamidated gliadin induced high levels of interleukin (IL)-2 (Figure 2B), whereas responses to nondeamidated gliadin, irrelevant protein OVA, or medium control were very low (Figure 2B). FACS analysis of CFSE-labeled CD4⫹ gliadin-TCRtg T cells in the same experimental setup revealed up to 7 cycles of division for the majority of CD4⫹ gliadin-TCRtg T cells, whereas a very small fraction of T cells proliferated to nondeamidated gliadin (Supplementary Figure 1A). Stimulation with deamidated gliadin predominantly induced secretion of IFN-␥ and moderate amounts of IL-4, IL-17, and IL-21 (Figure 2C). For reactivity to synthetic peptides, see Supplementary Figure 1.

Oral Deamidated Gliadin Induces a Dominant T-Cell Proliferation in the Spleen But Not in the MLN Using a DO11.10 transfer model, we and others have previously reported that OVA feed exclusively induces T-cell differentiation in the intestinal immune system, in particular in the gut-draining MLN and PP.8,9 To elucidate whether deamidated gliadin feed induces a comparable response, a DQ2.gliadinTCR T-cell transfer model was set up. Thereto, CFSE-labeled CD4⫹ gliadin-TCRtg T

cells from DQ2.gliadinTCR mice were intravenously injected in DQ2 acceptor mice. Starting the day after T-cell transfer, DQ2 mice received 75 to 100 mg deamidated gliadin. In parallel, BALB/c mice received OVA-specific CD4⫹ DO11.10tg T cells and 24 hours later were either fed 70 mg OVA to induce a tolerogenic T-cell response or given 400 ␮g OVA intramuscularly to induce an inflammatory T-cell response.13 At 72 hours after feed, lymphoid organs and spleens were analyzed for the division of transferred protein-specific T cells. Confirming earlier data (Hauet-Broere8 and du Pré et al14), oral OVA administration resulted in proliferation of OVA-specific T cells in the gut-draining MLN at 72 hours after ingestion (Figure 3A, middle panel). Intramuscular injection of OVA induced proliferation of OVA-specific T cells in the draining inguinal and popliteal lymph nodes (peripheral lymph nodes) and the spleen (Figure 3A, right panel, and Unger et al13). After deamidated gliadin feed, virtually no proliferation was detected in the MLN of DQ2 acceptor mice, whereas proliferation predominantly occurred in the spleen (Figure 3A, left panel). To exclude that division of gliadin-specific T cells in MLN may have occurred earlier, different time points after protein feed were investigated. Also at 48 hours, very few dividing CD4⫹ gliadin-TCRtg T cells were observed in MLN (Supplementary Figure 2), whereas dividing gliadin-specific T cells in the spleen had already undergone 3 divisions (Supplementary Figure 2). It should be noted that at 72 hours after OVA feed,

GLIADIN INGESTION ELICITS SPLENIC TR1–LIKE CELLS

613

BASIC AND TRANSLATIONAL AT

August 2011

Figure 2. CD4⫹ gliadin-TCRtg T cells respond to deamidated but not to nondeamidated gliadin. (A) A total of 5 ⫻ 105 splenocytes from DQ2.gliadinTCR mice and MHCII⌬/⌬ mice were cultured with or without deamidated gliadin (TG2-gliadin), and 3H-thymidine incorporation was measured. (B and C) Gliadin-specific CD4⫹V␤1⫹ T cells were stimulated with DQ2⫹ SPL-DCs loaded with gliadin, TG2-gliadin, OVA, or medium. (B) Release of IL-2 at 48 hours. (C) Release of IFN-␥, IL-4, IL-17, and IL-21 at 96 hours of culture. n.d., nondetectable. *P ⬍ .05.

614

DU PRÉ ET AL

GASTROENTEROLOGY Vol. 141, No. 2

BASIC AND TRANSLATIONAL AT

Figure 3. Oral deamidated gliadin induces a dominant T-cell proliferation in the spleen but not in the MLN. (A) DQ2 mice received 7 ⫻ 106 CFSE-labeled CD4⫹V␤1⫹ cells intravenously and 100 mg TG2-gliadin orally the next day. BALB/c mice received 1 ⫻ 107 CFSE-labeled CD4⫹KJ126⫹ cells intravenously and the next day OVA intragastrically (tolerance) or 400 ␮g OVA intramuscularly (immunity). Spleens and draining lymph nodes were isolated 72 hours after antigen administration and analyzed by FACS. For each group, a representative histogram plot and dot plot showing CFSE dilution (gated on transgenic CD4⫹ V␤1⫹ or CD4⫹KJ1-26⫹ cells) is depicted. The percentages of CD4⫹V␤1⫹ or CD4⫹KJ1-26⫹ cells in each peak of division were calculated and are represented as the mean ⫾ SD for 5 mice (TG2-gliadin intragastrically) and 3 mice (OVA intragastrically/ intramuscularly). (B) DQ2 mice enriched with CFSE-labeled CD4⫹V␤1⫹ T cells were given 100 mg of nondeamidated gliadin or deamidated gliadin orally or 500 ␮g intramuscularly. At 72 hours after gliadin administration, spleens (for gliadin intragastrically) or draining popliteal lymph nodes (for gliadin intramuscularly) were isolated and CD4⫹V␤1⫹ T cells were analyzed by flow cytometry for CFSE dilution.

proliferating OVA-specific T cells can also be detected in the spleen (Figure 3A, middle panel) but are still absent at 48 hours (Supplementary Figure 2). This agrees with earlier reports that the cells observed at 72 hours have migrated to the spleen after initial activation and division in the MLN.9 It was excluded that the unexpected localization of the gliadin-specific T-cell response in the spleen was due to an effect of the digestion/deamidation process, because neither mock deamidation of OVA nor cofeeding of OVA and deamidated gliadin altered the localization of the OVA-specific T cells in the MLN (data not shown). To finally establish that the gliadin-specific T-cell response was specific for deamidated gliadin only, DQ2 mice received multiple oral gavages of nondeamidated or deamidated gliadin. At 72 hours after gliadin feed, spleens were analyzed for the division of transferred gliadin-specific T cells. As depicted in Figure 3B, the proliferation of CD4⫹ gliadin-TCRtg T cells to nondeamidated gliadin was insignificant, whereas gliadin-specific T cells specifically proliferated in response to oral deamidated gliadin (Figure 3B). These results indicate that under homeostatic

conditions gliadin is not deamidated in vivo by murine TG2. It has been shown that TG2 in the murine intestine is inactive during intestinal homeostasis and is transiently activated on tissue damage.15 Indeed, we observed that inducing tissue damage by intramuscular injection of nondeamidated gliadin was sufficient for the proliferation of gliadin-TCRtg T cells in the draining peripheral lymph nodes (Figure 3B). These data show that, in the DQ2.gliadinTCR mouse model, the CD4⫹ T-cell response to deamidated gliadin occurs predominantly in the spleen.

Dividing Gliadin-TCRtg T Cells Have an Inflammatory T-Cell Phenotype Previously, we have shown that the mucosal administration of a harmless food protein such as OVA results in the differentiation of functionally suppressive mucosal Treg cells in the MLN that are characterized by a rapid down-regulation of CD62L and the acquisition of Foxp3 expression.14 As a consequence, the difference in localization between the T-cell responses induced by oral

GLIADIN INGESTION ELICITS SPLENIC TR1–LIKE CELLS

615

BASIC AND TRANSLATIONAL AT

August 2011

Figure 4. Dividing gliadin-TCRtg T cells have an inflammatory phenotype. (A and B) DQ2 mice and BALB/c mice were enriched with CFSE-labeled gliadin-specific or OVA-specific T cells and challenged as described in Figure 3. At 72 hours after antigen administration, the phenotype of dividing transferred T cells was analyzed in draining lymph nodes or spleen. (A) Representative dot plot for CD62L and Foxp3. (B) Percentage of CD62Lhi and Foxp3⫹ cells calculated and expressed as mean ⫾ SD for at least 3 mice. (C) DQ2 mice were enriched with CFSE-labeled gliadin-TCRtg T cells and, the next day, received gliadin or TG2-gliadin intragastrically. At 72 hours, gliadin-TCRtg T cells were FACS sorted from the spleens and restimulated in vitro with TG2-gliadin–loaded DQ2⫹ BM-DC. After 48 hours of restimulation, supernatants were analyzed for secreted cytokines. n.d., nondetectable. *P ⬍ .05.

OVA and deamidated gliadin may affect the type of immune response that is initiated. Therefore, we determined the phenotype of the gliadin-specific T cells that differentiate in the spleen and compared it with the phenotype of OVA-specific T cells that are induced after oral or intramuscular OVA administration. The majority of gliadinspecific T cells expressed high levels of CD62L, the homing receptor that allows migration into lymph nodes (Figure 4A, upper left plot, and B). In addition, deamidated gliadin feed did not induce expression of Foxp3 (Figure 4A, lower left plot, and B). This CD62LhiFoxp3⫺ phenotype was opposite to the tolerogenic T-cell response that was seen in the MLN after OVA feed (du Pré et al,14 Figure 4A, middle plots, and B). By comparison, the response was

similar to that after encounter of OVA via a nonmucosal route (ie, through injection in the thigh muscle), which resulted in a nontolerogenic T-cell response that was characterized by minor Foxp3 expression and a CD62Lhi phenotype (du Pré et al,14 Figure 4A, right plots, and B). To determine the cytokine profile of responding gliadin-specific T cells, DQ2 mice were enriched with CD4⫹ gliadin-TCRtg T cells and were fed either deamidated gliadin or nondeamidated gliadin. After 72 hours, both responding gliadin-TCRtg T cells and nonresponding cells were isolated from the spleen and restimulated in vitro with DQ2⫹ BM-DC loaded with deamidated gliadin. At 48 hours of restimulation, the secreted cytokines were measured. As shown in Figure 4C, the reactive T cells from

616

DU PRÉ ET AL

GASTROENTEROLOGY Vol. 141, No. 2

Figure 5. Gliadin-TCRtg T cells secrete IL-10. (A) Dividing gliadin-TCRtg T cells were FACS sorted and restimulated as described in Figure 4C. (B) Restimulation of splenic gliadin-TCRtg T cells with phorbol myristate acetate/CaI for 14 hours. Relative expression of IL-10 messenger RNA by real-time polymerase chain reaction. (C) Splenic gliadin-TCRtg T cells were obtained and restimulated with TG2-gliadin as described in Figure 4C. After 48 hours of restimulation, cells were harvested and stained intracellularly for V␤1, IFN-␥, and IL-10. One representative dot plot is shown (gated on V␤1⫹ cells, n ⫽ 3). (D) Naive gliadin-TCRtg T cells (5 ⫻ 105) were stimulated with deamidated gliadin-loaded DQ2⫹ SPL-DCs (2 ⫻ 104) for 72 hours, harvested, and restimulated with fresh deamidated gliadin-loaded DQ2⫹ BM-DC in the presence or absence of a neutralizing ␣IL-10R antibody (1B1.2) or isotype control (GL113). Release of IFN-␥ at 48 hours of culture was measured by enzyme-linked immunosorbent assay. *P ⬍ .05. BASIC AND TRANSLATIONAL AT

mice that were fed deamidated gliadin had a Th1-like effector phenotype, as indicated by prominent secretion of IFN-␥ and the presence of IL-6. This response was specific for restimulated cells because the nonresponding cells isolated from mice that received nondeamidated gliadin feed produced lower amounts of these cytokines. The cells in the culture also secreted tumor necrosis factor ␣, MCP-1, and IL-12p70 (Figure 4C). However, the latter cytokines were not specifically associated with restimulation because the cells isolated from mice that received nondeamidated gliadin feed produced equal amounts. These findings show that deamidated gliadin feed induces the differentiation of gliadin-specific T cells with a Th1like phenotype.

Gliadin-TCRtg T Cells Also Secrete IL-10 Next, we determined whether differentiating splenic gliadin-specific T cells also secreted anti-inflammatory cytokines, because it has previously been reported that suppressive gliadin-specific IL-10 –producing Tr1 cells can be detected in patients with established celiac disease.16 Similar to the experiment described in Figure 4C, gliadin-specific T cells were purified from the spleens of DQ2 mice by flow-cytometric cell sorting and restimulated in vitro with deamidated gliadin loaded DQ2⫹ BM-DC. Gliadin-specific T cells that had not encountered deamidated gliadin in vivo (ie, that were obtained from mice that were

fed nondeamidated gliadin) secreted very low amounts of IL-10 protein on stimulation in vitro with deamidated gliadin (Figure 5A). However, T cells that had been activated in vivo by oral administration of deamidated gliadin and that received a second stimulation in vitro were found to secrete very high levels of the immunosuppressive cytokine (Figure 5A). Moreover, these gliadin-specific T cells produced significantly lower amounts of IL-2 when stimulated for the second time (Figure 5A). Quantitative polymerase chain reaction analysis revealed that IL-10 messenger RNA levels were elevated as well in proliferating gliadin-specific T cells from mice that were fed deamidated gliadin (Figure 5B). Both IL-10 –secreting Th1 cells as well as Tr1 cells that secrete variable amounts of IFN-␥ have been reported to exist.17,18 Because gliadin-specific T cells in our model secreted high amounts of IFN-␥ concurrent to IL-10 (Figure 4C), we determined whether both cytokines were produced by the same cell subset. After in vitro restimulation and intracellular cytokine staining, splenic gliadin-specific T cells that were induced by deamidated gliadin feed could be subdivided into 3 different subsets: IFN-␥–producing cells (4.85% ⫾ 0.63%, n ⫽ 3), IL-10 –producing cells (13.18% ⫾ 0.55%, n ⫽ 3), and cells that produced both cytokines (4.57% ⫾ 0.59%, n ⫽ 3) (Figure 5C). These data infer that multiple subsets of gliadin-specific T cells contribute to the high amounts of IFN-␥ and IL-10.

August 2011

GLIADIN INGESTION ELICITS SPLENIC TR1–LIKE CELLS

617

To assess whether IL-10 release had functional effects on the T-cell response, gliadin-TCRtg T cells were stimulated in vitro with splenic DQ2⫹ DC and deamidated gliadin. At day 3, T cells were harvested and restimulated with fresh DQ2⫹ BM-DC and deamidated gliadin in the presence of neutralizing ␣IL-10R antibodies. As shown in Figure 5D, the interference with IL-10R signaling resulted in an augmented IFN-␥ release. Together these data show that deamidated gliadin feed induces differentiation of a splenic suppressive Tr1-like T-cell population.

Transfer of Dividing Gliadin-TCRtg T Cells Induces Tolerance

Modulation of the T-Cell Response in DQ2.gliadinTCR Mice Because these data established that gliadin feed elicited oral tolerance while systemic sensitization with deamidated gliadin dissolved in adjuvant induced an inflammatory DTH response, we determined whether the presence of such an inflammatory gliadin-specific T-cell response was sufficient to induce intestinal pathology (Supplementary Figure 3). In agreement with other reports using HLA-DQ8 and HLA-DQ2 transgenic models,19,20 the immunization and generation of a systemic inflammatory T-cell response was not sufficient to induce intestinal pathology (Supplementary Figure 3).

Figure 6. Transfer of gliadin-TCRtg T cells induces tolerance. Dividing CD4⫹ gliadin-TCRtg T cells were FACS sorted as described in Figure 4C. A total of 2.5 ⫻ 105 cells were adoptively transferred to DQ2.gliadinTCR mice. Two mice received 2.5 ⫻ 105 purified in vitro– generated gliadinspecific Foxp3⫹ Treg cells. Control mice did not receive any transferred cells. One day after adoptive transfer, both groups were sensitized subcutaneously with 100 ␮g TG2-gliadin. Five days later, mice were challenged with 10 ␮g TG2-gliadin in both ears and, after 24 hours, the increase in ear thickness was determined and compared with values before challenge. *p ⬍ .05.

Next, we assessed the contribution of innate immune activation to the adaptive immune response. Innate inflammation can induce TG2 activation, which is required for T-cell activation. Siegel et al have shown that TG2 becomes activated by treating mice with polyinosinic/ polycytidylic acid, a synthetic double-stranded RNA that elicits transient innate inflammation and villous atrophy.15 Treatment of mice with polyinosinic/polycytidylic acid and subsequent nondeamidated gliadin did not induce T-cell proliferation in either spleen or MLN at 72 hours after feed (Supplementary Figure 4A). Because intestinal damage induced by polyinosinic/polycytidylic acid was very transient, the cytostatic drug methotrexate (MTX), which induces a more prolonged and severe intestinal damage,21 was used in parallel experiments (Supplementary Figure 4B). However, despite severe inflammation, feeding nondeamidated gluten did not induce a T-cell response (Supplementary Figure 4A). Immunohistochemical staining after 5-biotinamidopentyl-amine injection in mice confirmed TG2 activation in the small intestine at 72 hours after MTX treatment (Supplementary Figure 5). These data suggest that, at these regimens, innate immune activation induced by polyinosinic/polycytidylic acid or MTX did not induce sufficient deamidation to induce proliferation. This is in great contrast to nondeamidated gliadin intramuscular injection, which elicited TG2 activation (Supplementary Figure 5) and readily induced T-cell proliferation in peripheral lymph nodes (Figure 3B).

BASIC AND TRANSLATIONAL AT

Dividing OVA-specific T cells from the MLN of BALB/c mice that are orally tolerized with OVA can induce tolerance on adoptive transfer to naive acceptor mice, as shown by a suppression of an OVA-specific delayed-type hypersensitivity (DTH) response in the ears.8 To assess the function of the differentiating gliadinTCRtg Tr1-like cells in vivo, a similar approach was taken. Dividing gliadin-specific CD4⫹ T cells from the spleen were isolated by flow cytometric sorting and transferred to naive DQ2.gliadinTCR mice. As a control, 2 DQ2. gliadinTCR mice received Foxp3⫹ gliadin-specific Treg cells (83% Foxp3⫹), which were generated in vitro by culture of naive gliadin-TCRtg T cells with gliadin-loaded splenic DQ2⫹ DCs. One day after cell transfer, the DQ2. gliadinTCR acceptor mice were subjected to a DTH response consisting of sensitization with deamidated gliadin in the tail base and a subsequent challenge with deamidated gliadin in the ears. As seen in Figure 6, DQ2.gliadinTCR mice that had received no cell transfer developed a gliadin-specific DTH response after injection of deamidated gliadin in the ears following a sensitization in the tail base. However, transfer of the dividing gliadinspecific CD4⫹ splenic T cells to naive DQ2.gliadinTCR mice suppressed the gliadin-specific DTH response, as shown by a significantly lower increase in ear thickness that was comparable to that of mice that received in vitro– generated Foxp3⫹ gliadin-specific T cells (Figure 6). These data show that, despite their inflammatory phenotype and ample secretion of effector cytokines, the gliadin-specific T cells that differentiate in the spleen after gliadin feed are tolerogenic.

618

DU PRÉ ET AL

Using the MTX treatment, we next assessed whether destruction of the epithelial barrier would induce a more predominant T-cell proliferation in the MLN and PP. Thereto, mice were treated with MTX and received deamidated gliadin orally at 72 hours after MTX treatment. Intriguingly, no predominant T-cell proliferation occurred in MLN and PP, whereas splenic proliferation was maintained, implicating that increased translocation does not change the location of presentation (Supplementary Figure 4C). Previously, we have established that MLN cells from MTX-treated mice release increased amounts of inflammatory cytokines such as IFN-␥ on anti-CD3 and antiCD28 stimulation.21 Therefore, we assessed whether MTX treatment altered the phenotype of the dividing T cells in the spleen. However, in vitro restimulated gliadin-specific T cells from the spleen of MTX-pretreated mice did not secrete enhanced amounts of IFN-␥ or reduced IL-10 when compared with cells from saline-treated mice (Supplementary Figure 4D), suggesting that MTX-induced inflammation is not sufficient to breach the tolerogenic splenic response.

Discussion

BASIC AND TRANSLATIONAL AT

Here we show that tolerance to deamidated gliadin ingestion is regulated by systemic Tr1-like cells without evidence for a role for Foxp3⫹ T cells. Even though mechanistically different from other harmless food proteins, the net outcome of deamidated gliadin ingestion is tolerance. However, this tolerance is predominantly mounted outside of the intestinal immune system, namely in the spleen, where IL-10 –secreting Tr1-like cells are formed but virtually no adaptive Foxp3⫹ T cells. In our mouse model of double transgenic mice that express HLA-DQ2 and a chimeric TCR of a patient with celiac disease specific for deamidated gliadin peptide, feeding nondeamidated gliadin induced no signs of celiac disease pathology or T-cell division. Feeding of deamidated gliadin to DQ2.gliadinTCR mice resulted in a gliadin-specific T-cell response in the spleen, but not in the MLN. In addition to the MLN, oral antigens such as OVA induce proliferation in the PP.8 However, in our model, the response to deamidated gliadin in PP was mostly absent and, when detected, highly variable in the percentage of proliferating cells. This new model allowed us to show that deamidated gliadin, when given by the oral route, induces tolerance. The mechanisms that underlie oral tolerance to deamidated gliadin, however, appear strikingly different from that seen for other food proteins thus far. One difference is the localization of the T-cell response in the spleen. Another is that within 48 hours after antigen intake, most gliadin-specific T cells have an activated phenotype and lack Foxp3 expression. This is exceptional considering that feeding harmless soluble proteins to naive mice normally induces oral tolerance that is associated with the induction of antigen-specific Foxp3⫹ mucosal Treg cells

GASTROENTEROLOGY Vol. 141, No. 2

in mucosa-draining lymphoid tissue.6 –9,22 Thus, because oral tolerance to deamidated gliadin occurs in the spleen, it does not occur under the strict control of mucosal resident DCs and the mucosal microenvironment. Splenic presentation of deamidated gliadin may be related to its molecular structure, which is very different from OVA and most other food proteins. Gluten proteins are proline-rich molecules that are insoluble in water and contain disulfide and hydrogen bonds, resulting in the assembly of stable aggregates. As a result, gliadin may reach the small intestine in the form of relatively large immunogenic peptides, as has been reported for the 33-mer from gliadin-␣2.23 It could be hypothesized that in our model little T-cell proliferation occurs in the MLN and PP because deamidated gliadin does not effectively pass the intact epithelial barrier. From our initial experiments in which we treated DQ2 acceptor mice with MTX before transfer of gluten-specific T cells and deamidated gliadin feed, it can be inferred that enhanced translocation due to intestinal damage does not lead to increased T-cell proliferation in the MLN and PP. This suggests that productive antigen presentation of gliadin-␥1 from complex gluten may require an additional event. TG2 binding may influence antigen presentation and trafficking. Our initial MTX experiments did not show evidence for such an effect despite detectable TG2 activation. Possibly, the level of TG2 activation was not sufficient to provide this effect. Alternatively, a particular subtype of antigen-presenting cell is involved in this process. As such, in biopsy specimens from patients with celiac disease, a unique subset of DCs has been detected that was absent in controls without celiac disease.24 It is expected that feeding whole digested deamidated gliadin could differ from feeding deamidated peptide, which may be more readily taken up by M cells. In a DQ2-␣-II epitope-dependent mouse model, De Kauwe et al20 showed that a 3-day feeding regimen with deamidated gluten peptide in peanut oil resulted in a limited but detectable division in the PP and MLN. However, because the deamidated gluten peptide was emulsified in peanut oil, this may have affected its uptake. Whether gliadin presentation in MLN or PP would lead to mucosal control and impose the typical Foxp3⫹ CD62Llo Treg differentiation is questionable, because at the times that proliferation was detected in PP in our study, we also observed the development of Foxp3⫺CD62Lhi cells. The latter agrees with De Kauwe et al, who also reported no increased Foxp3 induction in mucosally differentiating cells.20 To fully resolve which properties of deamidated gliadin drive splenic presentation, detailed future analysis is required. It is difficult to establish whether gliadin-specific responses also predominantly occur in the spleen in humans. However, it is clear that inflammatory immune responses in celiac disease are not always restricted to the intestine. Functional analysis of the dividing gliadin-specific T cells in the spleen revealed that these cells have a Tr1 phenotype characterized by suppressive activity in a

August 2011

GLIADIN INGESTION ELICITS SPLENIC TR1–LIKE CELLS

DTH model, low IL-2 release, and IL-10 – dependent inhibition of IFN-␥ production. However, simultaneously to the production of IL-10, the cells do secrete relatively large amounts of IFN-␥, which would argue denoting them as having a Th1 phenotype. Intracellular cytokine staining revealed that despite originating from the same HLA-DQ2-TCR interaction, the differentiating T cells consist of 3 separate subpopulations staining positive for IFN-␥, double positive for IFN-␥ and IL-10, or positive for IL-10. This heterogeneity could either imply that the dividing T cells differentiate into distinct subpopulations or denote that these cells are Th1 cells that develop into IL-10 –secreting cells in a specific phase of differentiation.17,18 For this reason, we have chosen to denote the cells as “Tr1-like.” Although future experiments need to reveal which pathways determine gliadin-specific Tr1-like cell differentiation, our in vitro T-cell proliferation assays revealed that IL-10 –producing T cells can be induced by both splenic CD11c⫹ DCs and B cells (data not shown), suggesting that the induction of these cells can be mediated by different subsets of antigen-presenting cells. It should be noted that the cytokine profile of differentiating gliadin-specific T cells was dependent on the nature of the antigen. In comparison to whole protein stimulation, peptide stimulation resulted in a reduced secretion of IFN-␥ but an increased production of the Th2cytokine IL-4. This may be related to possible adjuvant effects of complex gliadin that are lost when using the gliadin-␥1 peptide. The finding that oral tolerance to deamidated gliadin is mediated by Tr-1–like cells suggests that impaired tolerance in patients with celiac disease may be related to dysregulation in the Tr1-like T-cell response. This agrees with a recently developed mouse model for celiac disease– like enteropathy, which was established by transferring presensitized gliadin-reactive T cells to mice that had no tolerance-inducing cells due to a lack of T cells or T and B cells.25 In celiac disease, IL-10 release has been observed in isolated intestinal T-cell subsets from pediatric patients with celiac disease that had been challenged with a glutencontaining diet.26 Moreover, it has previously been shown that IL-10 –producing gliadin-specific T-cell clones can be isolated from the mucosa of treated patients with celiac disease. Importantly, these Tr1-cell clones contain suppressive capacity because they could inhibit the proliferation of gliadin-specific effector T-cell clones.13 Future research will reveal whether particular defects in the induction of Tr1-like cells can be observed in patients with celiac disease. In sum, this study provides a new mouse model that allows further dissection of the mechanisms that underlie the induction of tolerogenic gluten-specific Tr1-like cells and may provide new strategies to unravel how these mechanisms are disturbed, leading to the inflammatory T cells that are seen in active celiac disease.

Supplementary Material

619

Note: To access the supplementary material accompanying this article, visit the online version of Gastroenterology at www.gastrojournal.org, and at doi: 10.1053/j.gastro.2011.04.048.

BASIC AND TRANSLATIONAL AT

References 1. Lundin KE, Scott H, Hansen T, et al. Gliadin-specific, HLA-DQ(alpha 1*0501,beta 1*0201) restricted T cells isolated from the small intestinal mucosa of celiac disease patients. J Exp Med 1993; 178:187–196. 2. Nilsen EM, Lundin KE, Krajci P, et al. Gluten specific, HLA-DQ restricted T cells from coeliac mucosa produce cytokines with Th1 or Th0 profile dominated by interferon gamma. Gut 1995;37:766 – 776. 3. Tollefsen S, Arentz-Hansen H, Fleckenstein B, et al. HLA-DQ2 and -DQ8 signatures of gluten T cell epitopes in celiac disease. J Clin Invest 2006;116:2226 –2236. 4. Molberg O, McAdam SN, Korner R, et al. Tissue transglutaminase selectively modifies gliadin peptides that are recognized by gutderived T cells in celiac disease. Nat Med 1998;4:713–717. 5. van de Wal Y, Kooy Y, van Veelen P, et al. Selective deamidation by tissue transglutaminase strongly enhances gliadin-specific T cell reactivity. J Immunol 1998;161:1585–1588. 6. Broere F, du Pre MF, van Berkel LA, et al. Cyclooxygenase-2 in mucosal DC mediates induction of regulatory T cells in the intestine through suppression of IL-4. Mucosal Immunol 2009;2:254 – 264. 7. Coombes JL, Siddiqui KR, Arancibia-Carcamo CV, et al. A functionally specialized population of mucosal CD103⫹ DCs induces Foxp3⫹ regulatory T cells via a TGF-beta and retinoic acid-dependent mechanism. J Exp Med 2007;204:1757–1764. 8. Hauet-Broere F, Unger WW, Garssen J, et al. Functional CD25- and CD25⫹ mucosal regulatory T cells are induced in gut-draining lymphoid tissue within 48 h after oral antigen application. Eur J Immunol 2003;33:2801–2810. 9. Worbs T, Bode U, Yan S, et al. Oral tolerance originates in the intestinal immune system and relies on antigen carriage by dendritic cells. J Exp Med 2006;203:519 –527. 10. Sjostrom H, Lundin KE, Molberg O, et al. Identification of a gliadin T-cell epitope in coeliac disease: general importance of gliadin deamidation for intestinal T-cell recognition. Scand J Immunol 1998;48:111–115. 11. Kouskoff V, Signorelli K, Benoist C, et al. Cassette vectors directing expression of T cell receptor genes in transgenic mice. J Immunol Methods 1995;180:273–280. 12. Madsen L, Labrecque N, Engberg J, et al. Mice lacking all conventional MHC class II genes. Proc Natl Acad Sci U S A 1999;96: 10338 –10343. 13. Unger WW, Hauet-Broere F, Jansen W, et al. Early events in peripheral regulatory T cell induction via the nasal mucosa. J Immunol 2003;171:4592– 4603. 14. du Pré MF, van Berkel LA, Ráki M, et al. CD62L(neg)CD38(⫹) expression on circulating CD4(⫹) T Cells identifies mucosally differentiated cells in protein fed mice and in human celiac disease patients and controls. Am J Gastroenterol 2011;106:1147– 1159. 15. Siegel M, Strnad P, Watts RE, et al. Extracellular transglutaminase 2 is catalytically inactive, but is transiently activated upon tissue injury. PLoS One 2008;3:e1861. 16. Gianfrani C, Levings MK, Sartirana C, et al. Gliadin-specific type 1 regulatory T cells from the intestinal mucosa of treated celiac patients inhibit pathogenic T cells. J Immunol 2006;177: 4178 – 4186. 17. Groux H, O’Garra A, Bigler M, et al. A CD4⫹ T-cell subset inhibits antigen-specific T-cell responses and prevents colitis. Nature 1997;389:737–742.

GASTROENTEROLOGY 2011;141:610 – 620

Tolerance to Ingested Deamidated Gliadin in Mice is Maintained by Splenic, Type 1 Regulatory T Cells M. FLEUR DU PRÉ,* ANNE E. KOZIJN,* LISETTE A. VAN BERKEL,* MARIËTTE N. D. TER BORG,* DICKY LINDENBERGH–KORTLEVE,* LISE TORP JENSEN,‡ YVONNE KOOY–WINKELAAR,§ FRITS KONING,§ LOUIS BOON,储 EDWARD E. S. NIEUWENHUIS,** LUDVIG M. SOLLID,¶ LARS FUGGER,‡,# and JANNEKE N. SAMSOM* *Department of Pediatrics, Erasmus Medical Center – Sophia Children’s Hospital, Rotterdam, The Netherlands; ‡Clinical Institute, Aarhus University Hospital, Skejby Sygehus, Aarhus, Denmark; §Department of Blood Transfusion and Immunohematology, Leiden University Medical Center, Leiden, The Netherlands; 储Bioceros BV, Utrecht, The Netherlands; ¶Centre for Immune Regulation, Institute of Immunology, University of Oslo and Oslo University Hospital – Rikshospitalet, Oslo, Norway; **Department of Pediatric Gastroenterology, Wilhelmina Children’s Hospital, University Medical Centre Utrecht, Utrecht, The Netherlands; and #Department of Clinical Neurology and MRC Human Immunology Unit, Weatherall Institute of Molecular Medicine, John Radcliffe Hospital, University of Oxford, Oxford, England

BASIC AND TRANSLATIONAL AT

BACKGROUND & AIMS: Patients with celiac disease have permanent intolerance to gluten. Because of the high frequency of this disorder (approximately 1 in 100 individuals), we investigated whether oral tolerance to gluten differs from that to other food proteins. METHODS: Using transgenic mice that express human HLA-DQ2 and a gliadin-specific, humanized T-cell receptor, we compared gluten-specific T-cell responses with tolerogenic mucosal T-cell responses to the model food protein ovalbumin. RESULTS: Consistent with previous findings, the ovalbumin-specific response occurred in the mesenteric lymph nodes and induced Foxp3⫹ regulatory T cells. In contrast, ingestion of deamidated gliadin induced Tcell proliferation predominantly in the spleen but little in mesenteric lymph nodes. The gliadin-reactive T cells had an effector-like phenotype and secreted large amounts of interferon gamma but also secreted interleukin-10. Despite their effector-like phenotype, gliadin-reactive T cells had regulatory functions, because transfer of the cells suppressed a gliadin-induced, delayed-type hypersensitivity response. CONCLUSIONS: Ingestion of deamidated gliadin induces differentiation of tolerogenic, type 1 regulatory T cells in spleens of HLA-DQ2 transgenic mice. These data indicate that under homeostatic conditions, the T-cell response to deamidated gliadin is tolerance, which is not conditioned by the mucosal immune system but instead requires interleukin-10 induction by antigen presentation in the spleen. Keywords: Food Allergy; Mucosal Tolerance Induction; Treg Cells; Immune Regulation.

C

eliac disease is one of the most common food intolerances, affecting approximately 1% of the Western world. Celiac disease is induced by the ingestion of gluten from wheat, rye, and barley and can be treated by avoiding gluten in the diet. A central role for CD4⫹ T cells in the pathogenesis of celiac disease is well established. Inflammatory gluten-specific CD4⫹ T cells can be isolated from the small intestinal mucosa of patients with celiac disease but not of healthy individuals.1 These interferon (IFN)-␥– secreting gluten-specific T cells2 recognize a diverse array of gluten peptides that are presented exclusively in the

context of HLA-DQ2 or HLA-DQ8.3 In agreement, the majority of patients with celiac disease carry HLA-DQ2, whereas HLA-DQ2–negative patients usually express HLA-DQ8.3 A crucial event in the binding of gluten peptides to these HLA molecules is peptide deamidation by the enzyme transglutaminase 2 (TG2). Gluten peptides generally have a low affinity for HLA-DQ2 and HLA-DQ8, but TG2 strongly facilitates this interaction by the introduction of negative charges in gluten peptides.4,5 Despite our understanding of the role of HLA-DQ2 (and HLA-DQ8) in the development of celiac disease, it is still unclear why oral tolerance to complex gluten is permanently broken in a significant proportion of the HLADQ2– carrying population whereas other HLA-DQ2–positive individuals remain tolerant. Oral tolerance to soluble proteins is strictly regulated by the microenvironment in mucosa-draining lymphoid tissue. Within mesenteric lymph nodes (MLN) and Peyer’s patches (PP), food protein–specific regulatory T (Treg) cells differentiate that acquire forkhead box P3 (Foxp3) under the control of the mucosal environment.6 –9 Because oral tolerance to soluble protein is such a controlled mechanism, it can be questioned why permanent intolerance to gluten is relatively frequent. We hypothesize that tolerance to gluten may be induced by different mechanisms compared with other food proteins. To test this hypothesis, we have generated mice that transgenically express HLA-DQ2 and a humanized HLADQ2–restricted gliadin-specific T-cell receptor (TCR) derived from a T cell of the celiac lesion and determined the response to orally administered deamidated gliadin in these mice.

Abbreviations used in this paper: BM-DC, bone marrow– derived dendritic cells; DC, dendritic cell; DTH, delayed-type hypersensitivity; FACS, fluorescence-activated cell sorter; IFN, interferon; IL, interleukin; MLN, mesenteric lymph nodes; MTX, methotrexate; OVA, ovalbumin; PP, Peyer’s patches; TG2, transglutaminase 2; TCR, T-cell receptor; Treg cell, regulatory T cell. © 2011 by the AGA Institute 0016-5085/$36.00 doi:10.1053/j.gastro.2011.04.048

August 2011

Supplementary Materials and Methods Mice HLA-DQ2(DQA1*0501,DQB1*0201⫽HLADQ2.5). MHCII⌬/⌬ transgenic mice, gliadin-TCR.MHCII⌬/⌬ transgenic mice, which have a transgenic TCR specific for the DQ2-␥-I epitope,27 and DO11.10 transgenic mice, which have a transgenic TCR specific for the OVA 323–339 peptide, were bred at the Erasmus Medical Center. Singletransgenic DQ2 and gliadinTCR mice were maintained on a C57BL/6 –DBA/2 mixed background. HLA-DQ2.gliadinTCR.MHCII⌬/⌬ double transgenic (DQ2.gliadinTCR) mice were obtained by crossing DQ2.MHCII⌬/⌬ (DQ2) transgenic mice with gliadin-TCR.MHCII⌬/⌬ (gliadinTCR) transgenic mice. BALB/c mice were purchased from Charles River (Maastricht, The Netherlands). All mice were kept under specific pathogen-free housing conditions, and experiments were approved by the Animal Experimental Committee of the Erasmus Medical Center. DQ2 mice, gliadinTCR mice, and DQ2.gliadinTCR mice were bred and maintained on a gluten-free chow (Arie Blok BV, Woerden, The Netherlands).

Proteins and Peptides Gliadin. Crude gliadin from wheat (Sigma-Al-

drich, Zwijndrecht, The Netherlands) was dissolved in a 0.1 mol/L NH4HCO3 2 mol/L urea buffer (100 mg/mL) and digested with 50 ␮g/mL ␣-chymotrypsin (Sigma) at room temperature for 24 hours (gliadin). Digestion was stopped by heating to 98°C for 10 minutes. The chymotrypsin-treated gliadin was centrifuged (5000g, 45 minutes), filter sterilized (0.45 ␮m), and dialyzed against sterile phosphate-buffered saline. Protein concentration was determined using a bicinchoninic acid assay (Perbio Science, Etten-Leur, The Netherlands). To obtain deamidated gliadin, chymotrypsin-digested gliadin was treated with guinea pig liver transglutaminase (TG2; 0.08 U/mg) (Zedira GmbH, Darmstadt, Germany) for 16 hours at 37°C. For in vivo experiments, mice received 75 to 100 mg gliadin (here termed nondeamidated gliadin) or TG2treated gliadin (here termed deamidated gliadin) by oral gavage or 500 ␮g intramuscularly by injection into the thigh muscle of each hind limb. For in vitro experiments, the synthetic native gliadin-␥1-Q peptide (QPQQPQQSFPQQQRPF) and the synthetic deamidated gliadin-␥1-E peptide (QPEQPQQSFPEQERPF) were used (5 ␮g/mL). OVA. In all in vivo experiments, intact 98% pure OVA (either from Sigma-Aldrich or from Calbiochem, San Diego, CA) was used. In in vitro experiments, either intact OVA (0.5 mg/mL; Calbiochem) or OVA323–339 peptide (0.2 ␮g/mL) was used.

In Vitro Proliferation Assays incorporation. Splenocytes from DQ2.gliadinTCR mice and MHCII⌬/⌬ control mice were cultured in the presence or absence of 0.5 mg/mL deami3H-Thymidine

GLIADIN INGESTION ELICITS SPLENIC TR1–LIKE CELLS

620.e1

dated gliadin for 72 hours. Proliferation was assessed by incorporation of 3H-thymidine. In vitro T-cell differentiation. Splenic DCs were isolated by digesting the tissue with Liberase (Roche, Woerden, The Netherlands). CD11c⫹ cells were isolated using anti-CD11c MACS beads (Miltenyi Biotec, Bergisch Gladbach, Germany). To obtain CD4⫹ gluten-specific T cells, lymph nodes and spleens from DQ2.gliadinTCR mice were enriched for CD4⫹ T cells using Dynabeads as described previously.6 The cells were labeled with CFSE (Molecular Probes, Invitrogen, Leiden, The Netherlands). Next, 5 ⫻ 105 CFSE-labeled CD4⫹V␤1⫹ gliadin-specific T cells were incubated with 2 ⫻ 104 DQ2⫹ DCs for 96 hours in the presence of nondeamidated or deamidated gliadin (0.5 mg/mL), OVA (0.5 mg/mL), gliadin-␥1 peptide (0.5 ␮g/mL), or OVA323–-339 peptide (0.2 ␮g/mL). Anti–IL-10R (purified from 1B1.2 hybridoma, 10 ␮g/mL, gift from Schering-Plough Biopharma) and appropriate isotype control (GL113) were used to neutralize IL-10R signaling.

Polyinosinic/Polycytidylic Acid and MTX Treatments To study the contribution of innate immune activation to the adaptive immune response in our new mouse model, either polyinosinic/polycytidylic acid or MTX was used. Polyinosinic/polycytidylic acid. DQ2 acceptor mice received 7 ⫻ 106 CFSE⫹CD4⫹V␤1⫹ cells in 150 ␮L saline by intravenous injection. The next day, mice were injected intraperitoneally with 30 mg/kg polyinosinic/ polycytidylic acid (Sigma) or equal volumes of saline as a control. After 30 minutes, DQ2 mice received 75 mg of nondeamidated or deamidated gliadin distributed in 3 oral gavages with intervals of at least 4 hours. MTX. At t ⫽ 0 and 24 hours, mice were injected intraperitoneally with 120 mg/kg and 60 mg/kg MTX (Teva Pharmachemie BV, Haarlem, The Netherlands) or equal volumes of saline. At t ⫽ 72 hours, DQ2 acceptor mice received 7 ⫻ 106 CFSE⫹CD4⫹V␤1⫹ cells in 150 ␮L saline by intravenous injection. At t ⫽ 96 hours, DQ2 mice received 75 mg of nondeamidated or deamidated gliadin distributed in 3 oral gavages with intervals of at least 4 hours. After either treatment, representative segments of the small intestine (duodenum, jejunum, and ileum) were isolated for immunohistochemical analysis. Spleens, PP, and MLN were isolated at 72 hours after antigen administration for analysis of T-cell proliferation and function.

Flow Cytometry The following antibodies were used for flow cytometry: anti-CD3 (145-2C11), anti-CD4 (GK1.5), antiCD8 (53-6.7), anti-CD62L (MEL-14), anti-CD25 (PC61), anti-MHCII (M5/114; all BD, Franklin Lakes, NJ), DO11.10 Tg TCR (KJ1.26; Invitrogen, Breda, The Netherlands),

620.e2

DU PRÉ ET AL

anti-Foxp3 (JFK-16S), anti–IL-10 (JES5-16E3), anti–IFN-␥ (XMG1.2; all EMELCA Bioscience, Bergen op Zoom, The Netherlands), and anti-TCR V␤1 (BL37.2) (Beckman Coulter, Woerden, The Netherlands). Anti–HLA-DQ2 (XIII.358.4) was a kind gift of Dr M. C. Mazilli (Rome, Italy). Anti–HLA-DQ (SPV-L3) was kindly provided by Dr H. Spits (Amsterdam, The Netherlands). Phenotype and cell division (based on fluorescence intensity of single CFSE peaks) were measured on a FACSCanto or FACSCalibur flow cytometer (BD).

Detection of Cytokines Concentrations of IL-10, IFN-␥, IL-12, IL-6, tumor necrosis factor ␣, and MCP-1 in supernatants were determined by BD Cytometric Bead Array according to the manufacturer’s instructions (BD). Quantitative enzymelinked immunosorbent assays for IL-2, IFN-␥, IL-4, IL-17, and IL-10 were performed using the following antibody pairs and recombinant cytokines: recIL-2, anti–IL-2, anti– IL-2-bio (all BD), recIFN-␥, anti–IFN-␥ (Biolegend, Uithoorn, The Netherlands), anti–IFN-␥-bio (cultured from R46A2 hybridoma), recIL-4 (BD), anti–IL-4 (cultured from 11B11 hybridoma), anti–IL-4-bio (BVD6-24G2), recIL17, anti–IL-17 (TC11-18H10.1), anti–IL-17-bio (TC11-8H4), recIL-10, anti–IL-10 (SXC1.1), and anti–IL-10-bio (JES5.2A5; all BD).

Real-Time Polymerase Chain Reaction Total RNA was purified from purified CFSE⫹CD4⫹ V␤1⫹ cells using the Nucleospin RNA XS kit (MachereyNagel, Düren, Germany). RNA was reverse transcribed to single-stranded complementary DNA using a mix of random hexamers (2.5 ␮mol/L) and oligo dT primers (20 nmol/L). The real-time reaction was performed in a total volume of 25 ␮L containing 0.2 mmol/L of each deoxynucleoside triphosphate (Amersham Pharmacia BioTech, Piscataway, NJ), 200 U Moloney murine leukemia virus reverse transcriptase (Promega, Madison, WI), and 25 U RNasin (Promega). Conditions for the real-time reaction were 37°C for 30 minutes, 42°C for 15 minutes, and 94°C for 5 minutes. Real-time quantitative polymerase chain reaction was performed using an ABI Prism 7900 Sequence Detection System (PE Applied Biosystems, Foster City, CA) based on specific primers and general fluorescence detection with SYBR green. Cyclophilin was used to control for sample loading and to allow normalization between samples. The expression levels relative to cyclophilin were calculated using the following equation: Relative Expression Level ⫽ 2⫺⌬Ct, whereby ⌬Ct ⫽ Cttarget – Cthousekeeping. Specific primers were designed across different constant region exons resulting in these primers: cyclo: forward 5=-AACCCCACCGTGTTCT-3=, reverse 5=-CATTATGGCGTGTAAAGTCA-3=; IL-10: forward 5=-CAAGCCTTATCGGAAATG-3=, reverse 5=-CATGGCCTTGTAGACACC-3=.

GASTROENTEROLOGY Vol. 141, No. 2

Histology Segments of approximately 1 cm were collected from the small intestine of DQ2.gliadinTCR mice, fixed in 4% formalin solution, and embedded in paraffin. Sections of 4-␮m thickness were stained with hematoxylin (Vector Laboratories, Burlingame, CA) and eosin (Sigma) and analyzed by microscopy. The length of 5 representative villi and adjacent crypts was measured to calculate villous/crypt ratios. For immunohistochemical detection of CD3 (rabbit anti-CD3; Dako, Heverlee, Belgium), sections were deparaffinized and endogenous peroxidases were quenched with 3% H2O2 in methanol for 20 minutes. Antigen retrieval was achieved by microwave treatment in citrate buffer (10 mmol/L, pH 6.0). Sections were blocked for 1 hour in Teng-T (10 mmol/L Tris-HCl, 5 mmol/L EDTA, 0.15 mol/L NaCl, 0.25% [wt/vol] gelatin, and 0.05% [vol/vol] Tween 20, pH 8.0) with 10% normal mouse serum. Antibody incubation was performed overnight at 4°C. Immunoreactions were detected using biotinylated secondary goat anti-rabbit serum with the Vectastain ABC Elite Kit (Vector Laboratories) and 3,3=diaminobenzidine tetrahydrochloride (Sigma-Aldrich). Sections were counterstained with hematoxylin.

TG2 Activity Assay MHCII⌬/⌬ mice were injected intraperitoneally with 180 mg/kg MTX as described previously or equal volumes of saline as a control. As a positive control, saline-treated mice were injected intramuscularly in the hind leg with 15 ␮L saline. Noninjected thigh muscle served as a negative control. The TG2 substrate 5-biotinamidopentylamine was dissolved to 25 mg/mL in saline. One hour before the mice were killed, they were injected intraperitoneally with 100 mg/kg 5-biotinamidopentylamine or equal volumes of saline. Representative segments of the small intestine (duodenum, jejunum, and ileum) and injection site of the thigh muscles were isolated and embedded in a cryopreservative solution (OCT, Tissue-Tek, Miles, Elkhart, IN). Samples were stored at ⫺80°C until assayed for TG2 activity.

Immunohistochemical Analysis of TG2 Activation Cryostat sections (6 ␮m) were fixed in ice-cold acetone for 10 minutes, allowed to air dry, and rinsed with phosphate-buffered saline. Sections were blocked in Teng-T plus 10% NMS for 15 minutes. After rinsing, TG2 activity was visualized by incubation with streptavidin/ Alexa Fluor 594 (Molecular Probes) at a dilution of 1:100 in phosphate-buffered saline plus 2% NMS for 1 hour. Nuclei were stained with 4=,6-diamidino-2-phenylindole, and slides were mounted with a Mowiol mounting solution (Sigma-Aldrich). Images were acquired and analyzed using a Leica DM5500B upright microscope and LAS-AF image acquisition software (Leica Microsystems, Rijswijk, The Netherlands).