Cytolytic mechanisms of intraepithelial lymphocytes in coeliac disease (CoD)

Abstract

The effector arm of the mucosal immune system comprises lymphocytes scattered at intraepithelial and lamina propria levels. Intraepithelial lymphocytes (IEL) are a large population of oligoclonal resting cells which exhibit phenotypic and functional characteristics of cytolytic T cells when activated. Several mechanisms have been demonstrated to account for their cytotoxicity. Among them, one is mediated by perforin and granzyme molecules, another is mediated by Fas ligand (FasL) which delivers apoptotic signals through Fas receptor on target cells. There is good evidence that a flat intestinal mucosa may be produced by activated T cells. The aim of our study was to evaluate FasL and perforin expression by IEL, and its possible correlation with the increased enterocyte apoptosis in coeliac mucosa. Endoscopic duodenal biopsy specimens from 10 untreated coeliac patients, 10 treated coeliac patients, and 10 biopsied controls were evaluated for enterocyte apoptosis by terminal deoxynucleotidyl transferase-mediated digoxigenin-deoxyuridine triphosphate nick end label method, for perforin expression by immunohistochemistry, and for FasL expression by immunocytochemistry. In untreated CoD there was a significant increase of percentage of both FasL⁺ and perforin⁺ IEL which positively correlated with enterocyte apoptosis in comparison with controls. All these parameters were significantly lower in treated CoD, even though they did not normalize. Our study demonstrates that in untreated CoD FasL and perforin expression by IEL is increased, and significantly correlates with the level of enterocyte apoptosis.

INTRODUCTION

Activated T lymphocytes mediate cytotoxicity through several mechanisms. Two of them are Fas/Fas ligand (FasL) interaction and exocytosis of lytic granules [1]. FasL, a type II transmembrane protein which belongs to the tumour necrosis factor (TNF) family, induces apoptosis after the binding with Fas, a type I transmembrane protein member of the TNF receptor family, expressed on the target cell [2]. Perforin is a molecule contained, together with granzymes, in the cytoplasmic granules of cytolytic T cells. Its release causes pore formation on cellular membrane that permits penetration of granzymes into cytoplasm, leading to apoptosis of target cells [3].

CoD is an immune-mediated enteropathy, caused in genetically susceptible individuals by an abnormal T cell response to the gliadin component of wheat protein [4]. More recently it has been suggested that tissue transglutaminase mediates this process through a specific deamidation of gliadin which creates a new epitope that binds to HLA-DQ2 and is recognized by gut-derived T cells [5,6]. CoD is characterized on pathological grounds by an increased density of intraepithelial lymphocytes (IEL) and by villous flattening [7] as a consequence of an exaggerated enterocyte apoptosis [8]. In CoD, IEL increase represents the earliest morphological change [9], which can also be recognized in the latent form of this condition [10]. Gut IEL are oligoclonal, resting, cytolytic T cells [11], and although they do not appear to contribute to enterocyte apoptosis in normal human small intestine [12], their role in the generation of mucosal damage in CoD is more than conceivable [13]. In this study we have attempted to determine whether perforin⁺ and FasL⁺ IEL are increased in coeliac mucosa.

PATIENTS AND METHODS

Patients and tissues

Multiple size appropriate and well-orientated endoscopic biopsy specimens were obtained from the second part of the duodenum from 10 untreated coeliac patients (F/M ratio 6/4, mean age 34·6 years, range 21–62 years), 10 coeliac patients on a gluten-free diet (F/M ratio 5/5, mean age 37·7 years, range 18–68 years) for at least 12 months, and 10 consenting subjects (F/M ratio 6/4, mean age 36·2 years, range 19–69 years) undergoing upper gastrointestinal endoscopy for functional dyspepsia. The consenting subjects using steroidal or non-steroidal anti-inflammatory drugs or presenting an inflamed mucosa at histology were excluded from the study.

Some of the biopsy specimens obtained were processed according to standard methods for traditional histology and immunohistochemistry. Some other specimens were used to obtain purified IEL and enterocyte suspensions.

Preparation of cell suspensions

The biopsy specimens were placed in calcium- and magnesium-free Hanks' balanced salt solution (HBSS; Gibco, Life Technologies Ltd, Paisley, UK) containing 1 mm DDT (Sigma Chemical Co., St Louis, MO) to remove mucus and supplemented with antibiotics (100 U/ml penicillin and 100 μg/ml streptomycin) and 5% fetal calf serum (FCS). The epithelial layer was removed with 1 mm EDTA (Sigma) and 1 mm DDT. After continuous agitation for 1 h at 37°C, the single-cell suspension was pelleted from the supernatant and washed once with 5 ml RPMI 1640 medium (Gibco) supplemented with antibiotics and 10% FCS. IEL and enterocytes were then separated on a Percoll density gradient (Pharmacia, Uppsala, Sweden). A discontinuous density gradient (25%, 40% and 75%) was used. The cells that layered between the 40% and 75% fractions were collected as IEL, whereas the cells that layered between the 40% and 25% interface were collected as enterocytes. Single-cell suspensions were resuspended in 1 ml PBS and kept on ice until use.

Immunocytochemistry for FasL and Fas

The isolated IEL and enterocytes were placed through cytocentrifugation on electrostatic slides, fixed by incubation for 3 min in cold acetone and air dried. Afterwards the cells were incubated for 5 min in 3% hydrogen peroxide to quench endogenous peroxidase activity, rinsed in PBS and incubated with bovine serum albumin (BSA) 2% in PBS for 30 min. Successively IEL were incubated for 60 min at room temperature with a mouse anti-human FasL MoAb (clone NOK-1; PharMingen, San Diego, CA) at 1:500 dilution, whereas enterocytes were incubated with a mouse anti-human Fas MoAb (Upstate Biotechnology, Lake Placid, NY) at 1:100 dilution. Afterwards, the cells were rinsed in PBS and treated with a secondary biotinylated antibody and peroxidase-conjugated streptavidin. Finally, diaminobenzidine reaction and nuclear counterstaining with Harris' haematoxylin were performed. As positive control, the TM4 cell line (immortalized mouse Sertoli cells) and Jurkat cell line were used, respectively. For negative control, slides of IEL and enterocytes were incubated without the primary antibody.

Immunohistochemistry for perforin

Formalin-fixed and paraffin-embedded biopsies were pretreated in a microwave oven in 0·05 m etilen-glycolaminoetyl-tetracetic acid solution, four times for 5 min each at 700 W. After endogenous peroxidase blocking with hydrogen peroxide 3% in PBS, sections were incubated overnight at 4°C with an anti-perforin MoAb at 1:1000 dilution (clone KM 585; Kamiya, Thousand Oaks, LA). As positive control a section of lymph node involved by Kikuchi disease was used, whereas as negative control a sequential section was incubated without the primary antibody.

Immunohistochemistry for CD3 detection

To determine whether perforin⁺ cells were T lymphocytes, seriate sections were processed with a polyclonal anti-CD3 antibody (Dako, Glostrup, Denmark) at 1:2000 dilution. Pretreatment was performed by incubation for 30 min at room temperature with Pronase 0·01 g % (Calbiochem, La Jolla, CA).

Terminal deoxynucleotidyl transferase-mediated digoxigenin-deoxyuridine triphosphate nick end labelling detection

For the ‘in situ’ detection of apoptotic cells, terminal deoxynucleotidyl transferase-mediated digoxigenin-deoxyuridine triphosphate nick end labelling (TUNEL) was used [14], using the peroxidase ApopTag Kit (Oncor, Gaithersburg, MD). The sections were deparaffinized, rehydrated and digested with proteinase K 20 μg/ml (Sigma) for 15 min at room temperature, then washed in tap water. Endogenous peroxidase was quenched by 3% hydrogen peroxidase for 30 min and then washed in PBS. After equilibration, the sections were incubated in a humidified chamber with terminal deoxynucleotidil transferase enzyme for 1 h at 37°C. Afterwards, sections were soaked in stop-wash buffer for 30 min and then rinsed in PBS and incubated with anti-digoxigenin (DIG) peroxidase in a humidified chamber for 30 min, followed by addition of diaminobenzidine and hydrogen peroxide for 1–3 min to allow colour development on apoptotic nuclei. Finally, sections were counterstained with Harris' haematoxylin. As positive control, Apoptag control slide was used, whereas negative control was performed by omission of the terminal deoxynucleotidyl transferase enzyme.

Quantification of positive cells

Sections were examined, using conventional light microscopy, in a blind fashion by an expert observer. Counts were performed, at a constant magnification (× 1000) by a differential count of at least 500 cells in the epithelium and the results expressed as percentage of apoptotic enterocytes and perforin⁺ IEL per 100 cells. FasL⁺ lymphocytes were expressed as mean percentage by a differential count of at least 500 isolated IEL.

Statistical analysis

Data are expressed as mean percentage. Statistical comparisons between mean values were carried out using the Mann–Whitney U-test for non-parametric data. Correlations were studied by Spearman's rank correlation test. P < 0·05 was considered statistically significant.

RESULTS

Immunocytochemical analysis of FasL and Fas expression

Figure 1 shows FasL⁺ IEL isolated from a biopsy specimen obtained from a patient with untreated CoD (a), and FasL⁻ IEL isolated from a biopsy specimen obtained from a control (b). The percentages of FasL⁺ IEL in the three groups studied are shown in Fig. 2. In untreated CoD almost all the IEL were FasL⁺ and their percentages (mean 86·6%, range 72·2–95·3%) were significantly higher than in treated patients (mean 33·6%, range 15·6–62·4%) and controls (mean 6·0%, range 1·4–11·7%). In treated CoD FasL⁺ IEL, although significantly lower than in untreated CoD, were in all cases above the control range. In untreated CoD the percentages of Fas⁺ enterocytes were significantly higher (mean 88·4%, range 72·2–93·1%) than in treated CoD (mean 16·2%, range 10·6–19·0%; P < 0·001) and controls (mean 12·7%, range 8·5–15·3%; P < 0·001). No significant difference was found between treated CoD and controls.

Fig. 1.

Fas ligand (FasL) expression by immunocytochemistry on intraepithelial lymphocytes (IEL) isolated from a duodenal biopsy specimen of an untreated coeliac patient (a), and a biopsied control (b). (Original mag. × 1000.)

Fig. 2.

Percentages of Fas ligand (FasL)⁺ intraepithelial lymphocytes (IEL) (% lymphocytes) in 10 untreated coeliac patients compared with those of 10 treated coeliac patients and 10 biopsied controls.

Immunohistochemical analysis of perforin expression

Figure 3 shows that there were many perforin⁺ IEL within the epithelial layer and crypts of untreated CoD. The percentages of perforin⁺ IEL in the three groups studied are shown in Fig. 4. In untreated CoD this percentage (mean 25·0%, range 15–35%) was significantly higher than in treated CoD (mean 11·3%, range 5–25%) and controls in which only a few IEL (mean 0·8%, range 0–2%) were perforin⁺. In treated CoD the percentage of perforin⁺ IEL was still significantly higher than in controls.

Fig. 3.

Perforin expression in untreated duodenal mucosa by immunohistochemistry. Most intraepithelial lymphocytes (IEL) both at the mucosal surface and along the crypts are perforin⁺. (Original mag. × 400.)

Fig. 4.

Percentages of perforin⁺ intraepithelial lymphocytes (IEL) (% epithelial cells) in 10 untreated coeliac patients compared with those of 10 treated coeliac patients and 10 biopsied controls.

Figure 5 shows that the percentages of FasL⁺ and perforin⁺ IEL were positively correlated with the percentages of TUNEL⁺ enterocytes (r_s = 0·75, P < 0·01 and r_s = 0·72, P < 0·01, respectively).

Fig. 5.

Significant direct correlations in untreated coeliac mucosa between the percentage of Fas ligand (FasL)⁺ intraepithelial lymphocytes (IEL) and the percentage of terminal deoxynucleotidyl transferase-mediated digoxigenin-deoxyuridine triphosphate nick end labelling (TUNEL)⁺ enterocytes (a) and between the percentage of perforin⁺ IEL and the percentage of TUNEL⁺ enterocytes (b).

Immunohistochemical analysis of CD3 expression

Qualitative comparisons between perforin⁺ IEL and CD3⁺ IEL showed a good overlap. The few perforin⁺ CD3⁻ IEL were regarded as natural killer (NK) cells.

DISCUSSION

The function of IEL remains controversial, but their position in intimate contact with epithelial cells, and in close proximity to the antigens contained in the gut lumen, suggests that they play an important role in mucosal immunity both in healthy and disease states, in many of which their density is increased [15]. IEL have been implicated in immune surveillance of the intestinal epithelium [16], in maintenance of epithelial integrity [17] and in regulation of immune response to foreign antigens [18]. It is also likely that IEL participate in the elusive mechanisms of oral tolerance [19] as well as in tumour surveillance [20]. The vast majority of them are cytotoxic lymphocytes [21] and murine studies suggest that activated αβ-CD8⁺, but not αα-CD8⁺ TCRαβ⁺ IEL can mediate perforin-based cytotoxicity, whereas both the subsets are active in FasL-based cytotoxicity [22]. Finally, an important role of IEL in the induction of enteropathy in animal models has been demonstrated [23].

In CoD, a T cell-mediated reaction against an epitope, generated by the transglutaminase-mediated deamidation of dietary gliadin, has been suggested as being involved in the generation of mucosal lesions in susceptible individuals [5]. For many years it has been known that coeliac small bowel epithelium is heavily infiltrated by lymphocytes [24,25], but only very recently has it been shown that increased enterocyte apoptosis is the mechanism responsible for villous flattening in this condition [8]. Therefore, we proceeded to investigate FasL and perforin expression by IEL in CoD. Our results show that the percentages of both FasL⁺ and perforin⁺ IEL are significantly increased in untreated CoD compared with treated CoD and controls. Furthermore, we found a positive significant correlation between both these percentages and the proportion of TUNEL⁺ enterocytes, suggesting a possible involvement of these two cytolytic mechanisms in the generation of mucosal damage in this condition.

Recently, Morimoto et al. [26] demonstrated that isolated colonic IEL constitutively express functional Fas–FasL, thus contributing to maintain lymphocyte homeostasis and to regulate immune response to a variety of dietary antigens in the normal intestine. Moreover, in mouse model of graft-versus-host disease [27,28], a condition which shares many pathological and immunological similarities with CoD, an involvement of FasL-mediated IEL cytotoxicity has been described. In our study we tried at first to detect FasL expression on IEL by immunohistochemistry using the same MoAb employed in the present work. Unfortunately, the resulting staining was unsatisfactory and to circumvent this pitfall we chose to assess FasL expression on isolated IEL. Although the isolation procedure might by itself cause an activation of T cells [12], the strong over-expression of FasL found on coeliac IEL, in comparison with the rare and weak staining on IEL isolated from normal biopsy specimens, allows us to conclude that this artefact, if any, does not significantly affect our results. The pathogenic relevance of this receptor-triggered lytic mechanism is emphasized by our demonstration that Fas expression is up-regulated on coeliac enterocytes.

Unlike FasL-mediated cytotoxicity, two previous studies showed that IEL intracytoplasmic granules may be responsible for IEL cytotoxic activity in untreated CoD [29,30]. Oberhuber et al. [29] showed that the neutral serine protease granzyme B is contained in coeliac IEL, whereas Shiner et al. [30] identified BLT⁺ intracytoplasmic granules of IEL in biopsy specimens from three coeliac children. Our results, showing an increased proportion of perforin⁺ IEL in untreated coeliac mucosa, confirm and extend these earlier observations on quantitative grounds.

In allegedly treated CoD, the proportions of both FasL⁺ and perforin⁺ IEL, although significantly lower than in untreated patients, were significantly higher than in controls. Since all our treated patients showed a clear improvement of duodenal histology, a possible explanation for this finding is a persistent activation of IEL despite gluten-free diet. However, it should be noted that ingestion of very small amounts of gliadin has been reported to expand the IEL population and to activate local intestinal immunity, even when the mucosal structure looks otherwise normal [31,32]. These findings are particularly intriguing when considering that enteropathy-associated T cell lymphoma is believed to arise from IEL [33] and that the vast majority of neoplastic lymphocytes express cytolytic granules [34,35].

In conclusion, although it is conceivable that cytotoxic IEL can kill their target by several mechanisms, our results enable us to hypothesize an involvement of FasL and perforin molecules in inducing enterocyte apoptosis in CoD.