Immunologic Activity in the Small Intestinal Mucosa of Pediatric Patients With Type 1 Diabetes

  1. Mia Westerholm-Ormio1,
  2. Outi Vaarala23,
  3. Päivi Pihkala2,
  4. Jorma Ilonen4 and
  5. Erkki Savilahti1
  1. 1Hospital for Children and Adolescents, University of Helsinki, Helsinki, Finland
  2. 2Department of Molecular Medicine, National Public Health Institute, Helsinki, Finland
  3. 3Division of Pediatrics, Department of Molecular and Clinical Medicine, Faculty of Health Sciences, University of Linköping, Linköping, Sweden
  4. 4Department of Virology, University of Turku, Finland
  1. Address correspondence and reprint requests to Mia Westerholm-Ormio, Hospital for Children and Adolescents, Research Laboratory, Biomedicum Helsinki, B433b, P.O. Box 63, FIN-00014 University of Helsinki, Finland. E-mail address: mia.westerholm-ormio{at}hus.fi

Abstract

Involvement of gut immune system has been implicated in the pathogenesis of type 1 diabetes. However, few studies have been performed on the gut mucosa from patients with type 1 diabetes. Thus, we characterized the stage of immune activation in jejunal biopsy samples from 31 children with type 1 diabetes by immunohistochemistry, in situ hybridization, and RT-PCR. We found enhanced expressions of HLA-DR, HLA-DP, and intercellular adhesion molecule-1 by immunohistochemistry even on structurally normal intestine of patients with type 1 diabetes and no signs of celiac disease. In addition, the densities of IL-1α- and IL-4-positive cells detected by immunohistochemistry and IL-4 mRNA-expressing cells evaluated by in situ hybridization were increased in the lamina propria in patients with type 1 diabetes and normal mucosa. Instead, the densities of IL-2, γ-interferon (IFN-γ), and tumor necrosis factor α-positive cells, the density of IFN-γ mRNA positive cells, and the amounts of IFN-γ mRNA detected by RT-PCR correlated with the degree of celiac disease in patients with type 1 diabetes. Our study supports the hypothesis that a link exists between the gut immune system and type 1 diabetes.

Accumulating data suggest that the gut immune system plays a role in the development of type 1 diabetes (1), an autoimmune disease that results from the destruction of insulin-secreting pancreatic islet β-cells by autoreactive T-cells (2). In experimental studies, several indications of the involvement of gut immune system in the development of autoimmune diabetes have been established. First, diet modifies the development of autoimmune diabetes in BB rats and NOD mice (35). Second, the islet infiltrating T-cells express gut-associated homing receptor β7-integrin, and antibodies that block this receptor or its endothelial ligand mucosal addressin cell adhesion molecule-1 inhibit the development of autoimmune diabetes in NOD mice (68). Third, mesenterial lymphocytes from a young NOD mouse transfer autoimmune diabetes to the recipients, indicating that diabetogenic T-cells are present in the gut-associated immune system (9). Finally, feeding autoantigen may induce the development of autoreactive cytotoxic lymphocytes and acceleration of autoimmune diabetes (10,11).

In humans, a link between the gut immune system and type 1 diabetes has also been suggested: enhanced immune responses to several cow milk proteins have been reported in serologic studies on patients with newly diagnosed type 1 diabetes (1214). We have also shown that GAD-specific T-cells in patients with type 1 diabetes express gut-associated homing receptor α4β7-integrin (15). Accordingly, T-cells derived from human diabetic pancreas have been demonstrated to show mucosal homing properties (16).

Furthermore, the association between type 1 diabetes and celiac disease (CD) has been recognized for some years (17,18), the prevalence of CD among children and adults with type 1 diabetes being as high as 2–8% (1924). Both diseases are associated with the HLA class II alleles DQB1*0201 and DQA1*0501 (HLA-DQ2) (25), which may partly explain the association of diseases, but recently it has been proposed that long-term exposure to gluten could induce type 1 diabetes (26). In a previous study, we found markers of immune activation even in structurally normal small intestine of patients with type 1 diabetes (27). The expression of HLA class II antigens in the villous epithelium and the density of α4β7-expressing cells in the lamina propria were increased in patients when compared with control subjects. However, it is not clear how an inflammation in the gut is linked to the process that destroys islet β-cells, and to date very few studies have been performed in the small intestine of patients with type 1 diabetes.

Our aim was to characterize the immune activation in jejunal biopsy specimens from pediatric patients with type 1 diabetes. We investigated the protein expression of interleukin-1α (IL-1α), IL-2, IL-4, γ-interferon (IFN-γ), and tumor necrosis factor-α (TNF-α) in jejunal biopsies from 21 pediatric patients with type 1 diabetes. We also investigated the lymphocytes, measuring their activation as well as the expression of HLA class II antigens on the epithelium. Furthermore, in the same samples, we analyzed the mRNA expression of IL-4 and IFN-γ by in situ hybridization. The specific amounts of IL-2, IL-4, IFN-γ, TNF-α, chemokine receptor-4 (CCR-4), and CCR-5 mRNA were evaluated by RT-PCR from the same mucosal samples and from 12 additional children with type 1 diabetes.

RESEARCH DESIGN AND METHODS

Patients.

Because of positive finding in the annual CD screening test, including the less specific positive gliadin antibodies, or because of gastrointestinal symptoms, 32 small intestinal biopsy specimens were obtained from 31 pediatric patients with type 1 diabetes. These patients were divided into three groups: patients with type 1 diabetes and normal villous structure and without markers of CD; patients with type 1 diabetes and normal villous structure and positive endomysium (EMA) or tissue transglutaminase (tTGA) IgA antibodies, designed as potential CD; and patients with type 1 diabetes and untreated CD. We considered an EMA titer of 1:50 or higher as positive, because a titer of 1:5–1:50 is the range for weak positivity (29). Partial or subtotal villous atrophy with crypt hyperplasia was defined as CD (villous crypt ratio <2).

Of the 32 type 1 diabetes jejunal specimens, 16 showed normal mucosal morphology with negative circulating CD autoantibodies (eight girls; mean age 10.6 years; range 2.9–18.2). Fourteen patients were negative for both EMA and tTGA antibodies, 2 of which were negative and 12 were positive for gliadin IgA or IgG antibodies. None of these 14 patients had had positive EMA or tTGA antibody titers before the biopsy or during follow-up after the biopsy. Two patients showed EMA titers of >1:5<1:50 at time of biopsy (Table 1, patients 5 and 6). Both of these patients had negative tTGA antibodies, a normal jejunal villous architecture, and normal values of intraepithelial lymphocyte counts, and the EMA and tTGA titers have remained negative for 2 years after the biopsy. Mean age at onset of type 1 diabetes was 4.1, and mean duration of type 1 diabetes at time of biopsy was 6.1 years in this group. Eight patients had a diagnosis of CD (five girls; mean age 9.1 years; range 3.9–15.8). Mean age at onset of type 1 diabetes was 5.8, and mean duration of type 1 diabetes at time of biopsy was 3.3 years in patients with CD. Eight patients had signs of potential CD (three girls; mean age 9.4; range 6.0–13.2); the mean age at onset of type 1 diabetes was 7.6, and mean duration of type 1 diabetes at time of biopsy was 1.8 years in this group. At the time of the biopsy, all of the patients were on a normal gluten-containing diet. From one CD patient (Table 1, patient 31), two biopsies were available. In the first one, villous structure was normal but with positive EMA and tTGA IgA antibodies; this specimen was included in the potential CD study group (Table 1, patient 23). The second biopsy specimen, 2.1 years later, showed partial villous atrophy. Clinical data are summarized in Table 1.

Jejunal biopsies on 12 age-matched control pediatric patients (mean age 7.1 years; range 1.6–13.7) were performed because of growth retardation, gastrointestinal symptoms, positive antigliadin antibodies, or any combination of these. The healthy control subjects had no history of CD or dermatitis herpetiformis in the family, they were not on medication, and they did not have any chronic diseases. The morphology of the jejunum was normal, and the EMA and tTGA antibodies were negative in all control children. Clinical data are summarized in Table 2.

Samples.

A Watson biopsy device was used to take a specimen from the proximal jejunum or a gastroscope from the distal duodenum. The specimens were divided for routine histology and immunohistochemical studies as previously described (30). Frozen tissue samples were cut into 8-μm sections, coded, and evaluated without knowledge of the specimen. The part of the mucosal specimen still left was used for RT-PCR. Some of the patients were enrolled later in the study and were included only in the RT-PCR study (Table 1).

HLA genotyping.

HLA genotyping was performed in 11 patients from peripheral venous blood sample and in 19 patients from jejunal paraffin blocks. HLA-DQB1, HLA-DQA1, and HLA-DRB1 analysis was performed by a technique developed for screening type 1 diabetes susceptibility on the basis of the presence of alleles associated with a risk for or with protection against type 1 diabetes. This two-step screening technique is based on the hybridization of relevant PCR products with lanthanide-labeled probes detected by time-resolved fluorometry (31).

Immunohistochemistry.

The avidin-biotin immunoperoxidase system was used on cryostat sections as previously described (32). For immunostaining of cytokines before incubation with monoclonal antibodies (mAbs), permeabilization was performed by incubation of the slides in 0.1% PBS-Tween 20 for 10 min at room temperature and diluted in 1% normal horse sera in 0.1% PBS-Tween20 for 1 h at 37°C.

The mAbs were used at the following dilutions: CD3, 1:200 (T-cell marker; Becton-Dickinson, San Jose, CA); T-cell receptor-γδ (TCRγδ), 1:200 (anti-γδ-TCR; Endogen, Woburn, MA); TCRbF1, 1:40 (anti-αβ-TCR, clone 8A3; Endogen); HLA-DR, 1:500, and HLA-DP, 1:40 (both from Becton-Dickinson); intercellular adhesion molecule (ICAM-1), 1:1,500 (anti-intercellular adhesion molecule-1, clone VF27; Endogen); Ki-67, 1:100 (reacting with nuclear antigens in proliferating cells; DAKO, Glostrup, Denmark); IL-1α, 1:50 (clone 20B8; Biosource International, Camarillo, CA); IL-2, 1:50 (clone 7A3; Biosource International); IL-4, 1:50 (IL-4II, clone 12.1; Mabtech, Nacka, Sweden); IFN-γ, 1:50 (clone 1-D1K; Mabtech); and TNF-α, 1:50 (clone 68B6A3; Biosource International). Nonimmune mouse IgG1 (DAKO) was used as negative primary antibody control and incubation with 1% normal horse sera on additional sections as negative control for reagents in the immunoperoxidase system.

Microscopic evaluation.

The numbers of positively stained cells were counted under a light microscope through a calibrated graticule at ×1,000 magnification, as described previously (27,32). In the same specimen, the positive cells in at least 30 fields either along the epithelium or comprising the lamina propria were counted, and cell densities were expressed as cells/mm or cells/mm2, respectively.

HLA-DR and -DP were used to stain epithelial cells for HLA class II expression. Epithelial staining was graded from 2–6 according to their cellular distribution (13) and area of staining (13) (33). The proportion of the lamina propria that stained positively for ICAM-1 was estimated and graded from 1–3, 1 representing faint and 3 strong staining. Ki-67–positive cells were calculated as percentages of crypt cells, with at least 200 crypt cells counted in each specimen.

Radioactive RNA in situ hybridization.

The sections were subjected to in situ hybridization for human IL-4 and IFN-γ riboprobes obtained from cDNAs described earlier (30). Tissue sections were incubated with 1.2 × 106 cpm of [33P]-labeled (1,000–3,000 Ci/mmol; Amersham, Life Technologies, Arlington Heights, IL) antisense or sense riboprobe in a total volume of 80 μl after the in situ protocol, described previously (34).

Microscopic evaluation of RNA in situ hybridization.

Positive IL-4 or IFN-γ mRNA-expressing cells were counted through a calibrated graticule at ×400 magnification and expressed as the number of positive cells/area of epithelium or lamina propria. A cell was considered positive when expressing seven or more positive cytoplasmic grains, which always corresponds to more than twice the background level. The lamina propria was assessed from the area immediately below the surface epithelium, excluding areas with lymphatic aggregates. For the specific probes, a minimum of two sections were prepared for each patient.

Total RNA isolation and cDNA synthesis.

Total RNA was isolated from the mucosal samples with GenElute Mammalian total RNA kit (Sigma, St. Louis, MO) following the manufacturer’s instructions. cDNA was synthesized using TaqMan Reverse Transcription Reagents (Applied Biosystems, Foster City, CA) in reaction volume of 75 μl. The reaction mixture consisted of 1× TaqMan RT Buffer, 5.5 mmol/l MgCl2, 500 μmol/l of each dNTP, 2.5 μmol/l Random Hexamer primer, 0.4 IU/μl RNase Inhibitor, and 10 ng/μl template RNA. The solution was treated with 2 IU of DNAse at 37°C for 30 min followed by 5 min of inactivation at 75°C. After addition of 25 IU of MultiScribe Reverse Transcriptase, the RT reaction was carried out at 25°C for 10 min, 48°C for 30 min, and 95°C for 5 min, and samples were stored at −20°C until used.

Real-time PCR quantification (TaqMan).

Real-time quantitative PCR was performed using specific TaqMan PDAR Primer/Probes and the ABI-Prism 7700 Sequence Detection System (Applied Biosystems). TaqMan PDARs IL-2 (PN 4309882P), IL-4 (PN 4309883P), IFN-γ (PN 4309890P), TNF-α (PN 4309891P), CCR-4 (PN 4312819P), and CCR-5 (PN 4316064P) were used. For normalizing differences in sample sizes and amount of RNA degradation, PDAR for ribosomal RNA 18S (PN 4310893E) was applied as an endogenous control. A validation experiment described elsewhere (35) was performed to confirm that the amplification efficiencies of the target genes and the endogenous control were approximately equal.

The reaction mixture consisted of 1× TaqMan Universal PCR Master Mix, 1× TaqMan Primer/Probe, and 50 ng (for target gene) or 5 ng (for endogenous control) of template cDNA. PCR was performed under the following conditions: 50°C for 2 min, 95°C for 10 min; 50 cycles of 95°C for 15 s and 60°C for 1 min. Analysis of a homemade calibrator cDNA sample and controls without template was also performed on each PCR plate. Each measurement was set up in triplicate.

Quantitative analysis was done using the comparative CT (ΔCT) method, in which CT value is defined as the cycle number in which the detected fluorescence exceeds the threshold value (35). ΔCT means the difference between CT of a target gene and the endogenous control, whereas ΔΔCT is the difference between ΔCT of the analyzed sample and the calibrator. Calculation of 2−ΔΔCT then gives a relative amount of target compared with the calibrator, both normalized to an endogenous reference (18S).

Statistical analysis.

Cell densities quantified by immunohistochemistry or in situ hybridization and the specific amounts of mRNA within the four subject groups were compared by nonparametric tests (Mann-Whitney U test) because of the nonlinear distribution of the parameters. P < 0.05 was considered significant. Because of the small size of the study groups, values outside the 25–75th percentiles were considered abnormal for the group.

Ethical considerations.

Specimens from pediatric patients were taken for diagnostic purposes. Use of biopsy specimens in this study was approved by the ethics committee of the Hospital for Children and Adolescents, University of Helsinki. In addition, after an oral explanation of the study plan, a written parental consent with a signature was obtained from the parents of all children.

RESULTS

HLA genotyping.

The patients’ type 1 diabetes-associated HLA DQB1*0201 and 0302 risk alleles are shown in Table 1. The heterodimer HLA-DQ2 associated with CD was found in 4 of 8 patients with type 1 diabetes and CD, in 6 of 8 patients with type 1 diabetes and potential CD, and in 3 of 16 patients with type 1 diabetes and normal mucosa. When the patients with type 1 diabetes were divided according to the presence of the HLA-DQ2 heterodimer, the patients carrying HLA-DQ2 showed increased densities of CD3 (median 1,688: 25–75th percentiles 1,250–2,125 vs. 961: 813–1,266; P = 0.041) and γδ-TCR-positive cells (469: 406–491 vs. 113: 81–176; P < 0.0001) in the lamina propria and γδ-TCR-positive intraepithelial lymphocytes (IELs; 46: 20–62 vs. 4.5: 1.3–33; P = 0.023). No other HLA-genotype association was found (data not shown).

Immunohistochemistry

Intraepithelial compartment.

Densities of CD3-, γδ-TCR-, and αβ-TCR-positive cells were higher in the epithelium of patients with type 1 diabetes and CD or potential CD than in patients with type 1 diabetes and normal mucosa and in control subjects. Patients with type 1 diabetes and CD showed only higher density of CD3-positive IELs when compared with patients with type 1 diabetes and potential CD. Furthermore, intraepithelial IFN-γ-positive cells were significantly increased in patients with type 1 diabetes and CD or potential CD when compared with patients with type 1 diabetes and normal mucosa and with control subjects. As a group, patients with type 1 diabetes and normal intestinal mucosa did not differ from control subjects regarding the density of IELs (Table 3).

Lamina propria.

The density of CD3-, γδ-TCR-, and αβ-TCR-positive cells in the lamina propria was higher in patients with type 1 diabetes and CD than in patients with type 1 diabetes and normal mucosa and in control subjects. These cells were also increased in patients with type 1 diabetes and potential CD compared with control subjects but not compared with patients with type 1 diabetes and normal intestine (Table 3).

The densities of IL-1α- and IL-4-positive cells in the lamina propria were greater in all three type 1 diabetes study groups than in control subjects. In addition, higher densities of IL-1α- and IL-4-positive cells could be seen in patients with type 1 diabetes and CD than in the two other type 1 diabetes study groups (Fig. 1). The densities of IL-2 and IFN-γ cells were greater in patients with type 1 diabetes and CD than in patients with type 1 diabetes and normal mucosa and in control subjects (Figs. 1B and 2). TNF-α-positive cells were found in higher density in patients with type 1 diabetes and CD than in patients with type 1 diabetes and potential CD or normal mucosa (Fig. 2B).

Staining with HLA class II antibodies and anti-ICAM-1.

Strong positive staining of epithelial cells with HLA-DR, HLA-DP, and ICAM-1 was present in all patients with type 1 diabetes (medians for the three type 1 diabetes study groups, HLA-DR >5.5/6, HLA-DP >5/6, and ICAM-1 3/3). In control subjects, the staining with these antibodies showed faint to moderate positivity (medians, HLA-DR 2/6, HLA-DP 3/6, and ICAM-1 2/3), a significant difference when compared with the three type 1 diabetes study groups (HLA-DR P < 0.003, HLA-DP P < 0.02, and ICAM-1 P < 0.03 for all three comparisons). In addition, patients with type 1 diabetes and CD expressed stronger staining of HLA-DP than did the other two type 1 diabetes study groups (P = 0.006 and P = 0.001; Fig. 3).

Staining with Ki-67.

The percentage of proliferative cells in the crypts was higher in specimens of patients with type 1 diabetes and CD or potential CD than in those from patients with type 1 diabetes and normal mucosa or from control subjects. The percentages of proliferative cells in patients with type 1 diabetes and normal mucosa did not differ from control subjects (Table 3).

Radioactive RNA in situ hybridization

Intraepithelial compartment.

IL-4 mRNA-positive hybridization signal was located mainly in the lamina propria, with little or no signal in the surface epithelium. IFN-γ mRNA-expressing cells were found in the intraepithelial compartment throughout the surface and the crypt epithelium. These cells were scarcely found in the normal specimens, the median being only 1 cell/10 areas of surface epithelium. In the type 1 diabetes specimens with CD or potential CD, IFN-γ mRNA-positive IELs were higher than in normal type 1 diabetes specimens or controls. In addition, the CD specimens showed higher density of IFN-γ mRNA-positive IELs than the potential CD specimens did (Table 4, Fig. 4).

Lamina propria.

Mononuclear cells expressing IL-4 mRNA were present in all the jejunal samples. The density of IL-4 mRNA-expressing cells was greater in patients with type 1 diabetes and CD or potential CD than in patients with type 1 diabetes and normal mucosa and control subjects. In addition, IL-4 mRNA-positive cells were increased in patients with type 1 diabetes and normal mucosa compared with control subjects.

In situ hybridization with the antisense probe for IFN-γ showed that positively expressed cells were found mainly in the superficial lamina propria and occasionally in the deeper parts. Densities of IFN-γ mRNA-expressing cells in the lamina propria were greater in all type 1 diabetes groups than in control subjects. However, among patients with type 1 diabetes, IFN-γ mRNA positive cells were 5.0 and 2.7 times more abundant in patients with type 1 diabetes and CD or potential CD than in those with normal intestine. Furthermore, patients with type 1 diabetes and CD showed increased densities of IFN-γ mRNA-positive cells compared with the other two type 1 diabetes study groups. The difference between patients with type 1 diabetes and potential CD and patients with type 1 diabetes and normal mucosa did not reach statistical significance (Table 4, Fig. 4). All sections hybridized with the sense probes showed only background signals.

RT-PCR.

Signals of IL-2, IFN-γ, TNF-α (data not shown), CCR-4, and CCR-5 mRNA were detectable by RT-PCR in all jejunal specimens. IL-4 mRNA was not detectable in the controls, whereas it was detected in 3 of 16 of the type 1 diabetes specimens with normal mucosa, in 3 of 8 potential CD, and in 5 of 8 of the patients with type 1 diabetes and CD. The amount of IFN-γ mRNA was greater in patients with type 1 diabetes and CD or potential CD compared with patients with type 1 diabetes and normal mucosa and with control subjects, correlating with the degree of CD. In addition, CCR-4 was expressed in higher amounts in patients with type 1 diabetes and CD than in patients with type 1 diabetes and normal jejunal mucosa (Fig. 5).

DISCUSSION

We found that the small intestine in pediatric patients with type 1 diabetes shows enhanced immune activation. The increased expression of MHC class II antigens and ICAM-1 detected by immunohistochemistry even in patients with structurally normal intestine confirms our earlier findings (27). The expression of HLA-DR and -DP was expanded throughout the villous surface and crypts in addition to the normal expression seen only on the upper villi. In addition, we found a higher density of IL-1α- and IL-4-positive cells in the lamina propria in patients with type 1 diabetes than in control subjects, irrespective of the morphology of the intestine. The findings were not restricted to patients carrying the CD associated HLA-DQB1*0201/HLA-DQA1*0501 risk alleles, suggesting that activation of the gut immune system may be associated with type 1 diabetes and not only with the genetic risk allele shared with CD.

It is interesting that in evaluation by in situ hybridization, IL-4 mRNA expression was higher in type 1 diabetes groups than in control subjects, and IL-4 mRNA was detected by RT-PCR in some biopsy samples from patients with type 1 diabetes but in none of the control subjects. IL-4 is mainly known as the Th2 driving cytokine. IL-4 is spontaneously secreted by gut-derived immune cells, and it enhances epithelial permeability, which is suggested to be increased in type 1 diabetes (36,37). On the other hand, IL-4 has also been shown to be critical for the maturation of dendritic cells and for the upregulation of antigen uptake and presentation by macrophages (38). Recently, IL-4 was shown to trigger type 1 diabetes by activating the autoimmune BDC2.5 T-cells in the pancreas of ins-IL-4/BDC2.5 transgenic mice. It was postulated that IL-4 triggers self-antigen presentation within the pancreatic islets by enhancing the antigen uptake in macrophages (39). In CD, Maiuri et al. (40) demonstrated during an in vitro gliadin challenge of cultured intestinal specimens a rapid release of IL-4 from degranulated mast cells, followed later by production of Th1-type cytokines. As we did not perform double-staining methods, we can only hypothesize on the origin of IL-4-positive cells. On the basis of the size of their cytoplasm, however, some of the IL-4 mRNA-positive cells detected by in situ hybridization showed resemblance of macrophages.

IL-1α is a proinflammatory interleukin secreted by monocyte lineage of immunologic cells and additionally by nonimmunologic cells such as epithelial cells. It has been revealed to participate in the gastrointestinal inflammatory process (41,42). IL-1 contributes to the regulation of dendritic cells and hence in antigen presentation. The increased densities of IL-1α-positive cells in all patients with type 1 diabetes may reflect the inflammation process activated at the level of innate immune system in the intestine of patients with type 1 diabetes.

Increased densities of IFN-γ mRNA-positive cells in lamina propria were also detected by in situ hybridization in patients with type 1 diabetes and normal mucosa, but this increased IFN-γ activation was not confirmed by the other methods used. In the gut epithelium, activation of HLA class II antigens and ICAM-1 was found, indicating enhanced capacity of antigen presentation. Furthermore, our results suggest that lymphocyte activation in the small intestine of patients with type 1 diabetes and no CD is found in the lamina propria and includes activation of Th2 immune cells. Studies performed in NOD mice and in patients with type 1 diabetes suggest that the islet-infiltrating, autoreactive cells express gut-associated homing receptor (69,15). Thus, the lymphocytes may recirculate between the gut and pancreas. The circulating mononuclear cells of the gut immune system home to lamina propria, whereas intraepithelial lymphocytes expressing α4bε receptor do not normally escape to circulation. Transfer experiments in the NOD mice model suggest that the diabetogenic immune cells are activated in the mesenterial immune system already before the infiltration of pancreatic islets, emphasizing the primary role of the gut immune system in type 1 diabetes (9). Dietary prevention of autoimmune diabetes in animal models suggests that changes in the gut immune system may induce the autoimmune destruction of β-cells in pancreas (4,5). Our findings of immune aberrancies in the gut suggest poor development of oral tolerance in children who are prone to type 1 diabetes (1214).

The increased densities of IFN-γ-, TNF-α-, and IL-2-positive cells and amount of IFN-γ mRNA detected by in situ hybridization and RT-PCR correlated with either full-blown or potent CD than that of type 1 diabetes, which is in agreement with previous findings (4346). This was also the case for the densities of IELs. T-cells in the celiac mucosa have been reported to represent for the most part Th1 cells producing IFN-γ (47), and in this respect, patients with type 1 diabetes and CD did not differ from patients with CD. However, in this study, we have also demonstrated an increase in IFN-γ-positive cells in potential CD as previously described (30). Enhanced Th1 cytokine responses in the gut may cause gut inflammation and damage, and high expression of Th1 cytokines is associated with increased permeability in patients with CD, Crohn’s disease, and other chronic gut inflammatory conditions (48).

The CD-associated HLA-DQ2 heterodimer is found in at least 90% of white patients with CD. Surprisingly, only four of eight of our patients with type 1 diabetes and CD expressed this heterodimer; two of them also expressed DR4. Shanahan et al. (49) found no DR4-positive subjects among 17 cases of type 1 diabetes and CD, whereas in our previous study, one of seven patients with type 1 diabetes and CD was positive for DR4 (19). This finding may be explained by the enrichment of DR4 among the Finnish type 1 diabetes population. Of the patients with type 1 diabetes and potential CD, six of eight expressed HLA-DQ2 (one patient of which is shared with the CD group). The follow-up time since the intestinal biopsies for the patients with potential CD is now in average 3.1 years (range 1.3–5.5), during which time only the patient who was already included in the type 1 diabetes patient group with CD has developed overt CD.

As a sign of enhanced immune activation, we found increased expressions of HLA-DR, HLA-DP, and ICAM-1 even on structurally normal intestine of patients with type 1 diabetes when compared with control subjects. Interestingly, patients with type 1 diabetes showed increased densities of IL-1α- and IL-4-positive cells in the lamina propria, whereas the densities of IL-2, IFN-γ, and TNF-α were associated with the degree of CD. The findings were not restricted to CD-associated HLA-DQ2 risk alleles in type 1 diabetes. Our study supports the hypothesis that a link exists between the gut immune system and type 1 diabetes.

FIG. 1.

A: Immunoperoxidase stainings for IL-1α (AD) and IL-4 (EH) in jejunal biopsy specimens from healthy control subjects (A and E) and from patients with type 1 diabetes and normal jejunal mucosa (B and F), potential CD (C and G), or CD (D and H). Positive cells are seen with typical paranuclear staining in higher densities in the type 1 diabetes specimens; also visible is some residual staining, which is easily identifiable, and except for that associated with goblet cells, clearly noncellular in its association. Carbatzole (AEC)-hematoxylin stain, magnification ×200. B: Densities of IL-1α-, IL-2-, and IL-4-positive cells in the lamina propria detected by immunohistochemistry in jejunal specimens of healthy control subjects (▵), type 1 diabetic patients with normal jejunal mucosa (▴), with potential CD (•), or with CD (▪). Horizontal lines show medians for each study group. *P < 0.05, **P < 0.01, ***P < 0.001.

FIG. 2.

Immunoperoxidase staining for IFN-γ in jejunal biopsy specimens from patients with type 1 diabetes and CD (A) or with potential CD (B and C). Positive cells are seen with typical paranuclear staining in the lamina propria (A and B) and are indicated by arrows. One IFN-γ-positive intraepithelial cell is seen in higher magnification in the surface epithelium of a biopsy specimen from a patients with type 1 diabetes and potential CD (indicated by arrowhead in C). AEC-hematoxylin stain, magnification ×200 in A and B, ×400 in C. B: Densities of IFN-γ- and TNF-α-positive cells in the lamina propria detected by immunohistochemistry in jejunal specimens of healthy control subjects (▵) and patients with type 1 diabetes and normal jejunal mucosa (▴), potential CD (•), or CD (▪). Horizontal lines show medians for each study group. *P < 0.05, **P < 0.01.

FIG. 3.

Immunoperoxidase stainings for HLA-DR (A and B) and HLA-DP (C and D) in jejunal biopsy specimens from a healthy control subject (A and C) and from a patient with type 1 diabetes and normal jejunal mucosa (B and D). Intensive, positive HLA-DR staining is seen throughout the epithelial cells of the villi and also in many crypt cells in the specimen from a patients with type 1 diabetes (B), whereas the specimen from a control patient shows only faint HLA-DR staining in the apical parts of the epithelial cells at the tip of the villi (A). Biopsy specimen from a control patient who was treated with HLA-DP antibody shows scattered, positive granules in the apical parts of the epithelial cells at the tip of the villi (C), whereas strong positive staining is seen throughout the villous epithelial cells in the specimen from a patient with type 1 diabetes (D). No positive staining is seen with either HLA-DR or HLA-DP in crypt cells in specimen from control patient (A and C). AEC-hematoxylin stain, magnification ×50.

FIG. 4.

Radioactive in situ hybridization of IFN-γ and IL-4 mRNA in jejunal biopsies; bright (A and C) and dark (B and D) field views. Hybridization with IFN-γ in a specimen from a patient with type 1 diabetes and CD (A and B) shows mRNA-expressing cells in the epithelium and in the lamina propria. Hybridization with IL-4 (C and D) shows a strongly positive cell in the lamina propria of a biopsy specimen from a patient with type 1 diabetes and normal jejunal mucosa. Arrows indicate the strongly IL-4-positive cell, morphologically resembling macrophages (C and D). Magnification ×400.

FIG. 5.

Cytokine and chemokine receptor-specific mRNA detected by RT-PCR in jejunal specimens of healthy control subjects (▵; n = 7) and in patients with type 1 diabetes and normal jejunal mucosa (▴; n = 16), potential CD (• n = 8), or CD (▪; n = 7). Individual results are shown as relative amount (2−ΔΔCT) of target compared with the calibrator, both normalized to an endogenous reference (18S). Horizontal lines show medians for each study group. *P < 0.05, **P < 0.01, ***P < 0.001.

TABLE 1

Clinical findings of the patients with type 1 diabetes

TABLE 2

Clinical findings of the healthy control subjects

TABLE 3

Densities of IFN-γ-, CD3-, TCR γ/δ-, and TCR α/β-positive cells in the surface epithelium (cells/mm) and of CD3-, TCRγ/δ-, and TCRα/β-positive cells in the lamina propria (cells/mm2) in the jejunal mucosa of patients with type 1 diabetes and normal mucosa, potential CD, or CD, and of control subjects

TABLE 4

Radioactive in situ hybridization in the mucosal samples

Acknowledgments

This study was financially supported by the Academy of Finland, the Finnish Medical Association, the Foundation of the Friends of the University Children’s Hospitals in Finland, the Sigrid Jusélius Foundation, and the Juvenile Diabetes Research Foundation.

We are indebted to Ilkka Julkunen, MD, PhD, National Public Health Institute, Helsinki, for the kind gift of IL-4 and IFN-γ cDNAs and to Prof. Markku Heikinheimo and Ilkka Ketola, MD, Hospital for Children and Adolescents, University of Helsinki, for their expertise with radioactive RNA in situ hybridization. The skillful technical assistance of Sirkku Kristiansen and Harry Lybeck is acknowledged.

Footnotes

    • Accepted May 22, 2003.
    • Received October 31, 2002.

REFERENCES

| Table of Contents