Human bronchial intraepithelial T cells produce interferon-γ and stimulate epithelial cells

Abstract

Intraepithelial lymphocytes (IELs) can be identified among epithelial cells in systemic mucosal tissues. Although intestinal IELs play a crucial role in mucosal immunity, their bronchial counterparts have not been well studied. The purpose of this study was to determine the immunological functions of human bronchial IELs, which interact directly with epithelial cells, unlike lamina propria lymphocytes (LPLs). We isolated successfully bronchial IELs and LPLs using a magnetic cell separation system from the T cell suspensions extracted from bronchial specimens far from the tumours of resected lungs. Human bronchial IELs showed an apparent type 1 cytokine profile and proliferated more actively in response to CD2 signalling than did bronchial LPLs. CD8⁺ IELs were identified as the most significant sources of interferon (IFN)-γ. Human bronchial epithelial cells constitutively produced the T cell growth factors interleukin (IL)-7 and IL-15, and levels of those factors increased when cells were stimulated by IFN-γ. Bronchial epithelial cells expressed cell surface proteins CD58 and E-cadherin, possibly enabling adhesion to IELs. In summary, human bronchial IELs have immunological functions distinct from bronchial LPLs and may interact with epithelial cells to maintain mucosal homeostasis.

Introduction

Intraepithelial lymphocytes (IELs) are a unique lymphocyte population found throughout the epithelium of systemic mucosal tissues. Among them, intestinal IELs play an important role in mucosal immunity and exhibit functional and ontogenic characteristics distinct from other circulating lymphocytes [1] in terms of extrathymic differentiation [2], cytotoxicity [3,4], immunity to infection [5] and influence on turnover of epithelial cells [6]. In addition, increased numbers of intestinal IELs are seen in coeliac disease [7]. Unlike intestinal IELs, however, there has been little investigation of bronchial IELs [8].

Previously we have demonstrated that human bronchial IELs survive longer than do bronchial lamina propria lymphocytes (LPLs) in human bronchial xenografts transplanted into severe combined immune deficiency mice [9], suggesting that bronchial IELs and LPLs represent independent cell populations. That study did not address functional differences between these two cell types.

In previous studies, sputum [10–12] or bronchoalveolar lavage [13,14] has been used to examine the function of lung T cells. These studies show that they consist of CD4⁺ and CD8⁺ cells and produce interferon (IFN)-γ, interleukin (IL)-4, IL-5, IL-12 and other cytokines. However, it cannot be determined precisely whether these T cells originated from the bronchial intraepithelium, subepithelium or lung parenchyma. In other studies, endobronchial biopsy has been performed and T cells were analysed immunohistochemically [15,16]. Those investigators examined only subepithelial T cells (perhaps LPLs), probably because the epithelial layers of small specimens obtained by biopsy are often highly damaged. Some studies show that IELs can be obtained by bronchial brushing [17], but LPLs were not obtained by this method.

In most studies of intestinal IELs, adequate numbers of cells have been obtained from resected gut by conventional separation methods using chemicals, dithiothreitol, ethylenediamine tetraacetic acid (EDTA) or proteases [18]. By contrast, isolation of bronchial IELs from removed bronchi is challenging because of the tissue's complex branching structure. Here, we successfully isolated human bronchial IELs and LPLs separately from human bronchial tissues using new methods. We then defined their cytokine profile, assayed their proliferation activity, and determined that cross-talk may occur between these T cells and neighbouring epithelial cells.

Materials and methods

Tissue preparation

Human bronchial tissue specimens were obtained from 19 patients (mean age 63·4 ± 10·5 years, range 40–75 years, 11 males and eight females, nine smokers and 10 non-smokers) undergoing lung resection for cancer. Patients with airway or pulmonary disease other than lung cancer were excluded. Informed consent was obtained from all patients and their families. Lung specimens were maintained on ice in Krebs–Henseleit solution containing 120 mM NaCl, 4·6 mM KCl, 1·03 mM KH₂PO₄, 25 mM NaHCO₃, 1·0 mM CaCl₂, 1·1 mM MgCl₂ and 11 mM glucose, and the lung parenchyma was removed carefully with tweezers. Normal bronchi of the third to fifth generations were selected. All selected specimens were confirmed to show no sign of tumour invasion, inflammation or other pathological change based on microscopic examination (Fig. 1a).

Fig. 1.

(a) Subsegmental bronchial wall section prior to removal of the epithelial layer. There are no indications of epithelial damage, inflammation and tumour invasion. (b) Subsegmental bronchial wall section after removal of epithelial layer. The basement membrane and lamina propria are maintained, and very few epithelial cells remain. Bronchi in (a) and (b) were obtained from one donor. Bars indicate 100 µm.

Isolation of human bronchial IELs and LPLs

Human bronchial specimens free from tumour lesions were opened longitudinally. Epithelial layers of bronchial tissues were separated by gently scraping the mucosal side with 18-gauge needles. Bronchial walls were confirmed microscopically to maintain basement membranes and avoid epithelial cells (Fig. 1b). Suspensions of epithelial layers were pipetted vigorously, and bronchial walls without epithelial layers were cut into small pieces. Both tissues were incubated in RPMI-1640 medium containing 10% fetal calf serum and 50 U/ml recombinant IL-2 for 24 h. After treatment, T cells migrated into culture medium from each tissue [19]. IELs and LPLs were isolated from each cell suspension using a magnetic cell sorting (MACS) system with anti-human CD3 microbeads according to the manufacturer's instructions (Miltenyi Biotec, Bergisch-Gladbach, Germany). More than 95% of these isolated cells were determined to be viable using the trypan blue staining technique.

Detection of CD103-expressing cells in IELs and LPLs by flow cytometry

To confirm the reliability of the isolation method, CD103 expression of both T cell populations separated was analysed by flow cytometry, as we have shown previously that CD103 is expressed more robustly on IELs than on LPLs in the human bronchus [9]. Isolated cells were stained with phycoerythrin (PE)-conjugated mouse anti-human CD3 monoclonal antibody (mAb) (BD Pharmingen, San Diego, CA, USA) and fluorescein isothiocyanate (FITC)-conjugated mouse anti-human CD103 mAb (Beckman Coulter, Fullerton, CA, USA). Cells were analysed with a fluorescence activated cell sorter (FACScan) flow cytometer using CellQuest software (Becton Dickinson, Mountain View, CA, USA). Double-stained cells were gated by the PE-positive area and examined by FITC fluorescence.

Selection of CD4⁺ or CD8⁺ IELs and LPLs

Mouse anti-human CD4 or CD8 mAbs were incubated for 30 min at 4°C with Dynabeads [anti-mouse immunoglobulin G (IgG)] (Dynal, Oslo, Norway) and washed three times. Beads were incubated with Isolated IELs or LPLs with rotation for 60 min at 4°C. CD4- or CD8-positive IELs or LPLs were selected using a Dynal magnetic particle concentrator (MPC-1). More than 95% of these selected cells were determined to be viable by trypan blue staining technique.

Detection of cytokine-producing cells in IELs and LPLs by an enzyme-linked immunospot assay

Cells producing IFN-γ, IL-4, IL-2 or IL-10 were evaluated using an enzyme-linked immunospot (ELISPOT) kit (R&D Systems, Minneapolis, MN, USA), according to the manufacturer's instructions. In brief, isolated cells were stimulated by phorbol 12 myristate 13 acetate (PMA) and the calcium ionophore A23187 (both from Sigma, St Louis, MO, USA) in 96-well microplates precoated with mAbs against IFN-γ, IL-4, IL-2 and IL-10. Twenty hours later, cells were washed away and optimally diluted biotinylated mAbs were added to the wells and incubated at 2–8°C overnight. After washing with a ELISPOT kit ‘Wash Buffer’, we incubated wells with alkaline phosphatase-conjugated streptavidin for 2 h at room temperature. Positive signals were visualized using 5-bromo-4-chlor-3-indolyl-phosphate/nitroblue tetrazolium (BCIP/NBT). Assay plates were observed under a microscope (Olympus, Tokyo, Japan) and photographed, and the number of blue immunospots were determined.

Production of T cell growth factors from the human bronchial epithelial cell (HBEC) line BEAS-2B and from primary HBECs

The HBEC line BEAS-2B (American Type Culture Collection, Manassas, VA, USA) was cultured in small airway cell basal medium (Biowhittaker/Clonetics, San Diego, CA, USA) in 10-cm polystyrene dishes coated with collagen type IV (Sigma) from human placenta. Cells were passaged for four generations. Primary HBECs were isolated from human bronchi of lung tissue resected for cancer and cultured for two passages. Cells were then detached from dishes using trypsin/EDTA (Gibco-Invitrogen, Grand Island, NY, USA), seeded into six-well dishes, and incubated with IFN-γ, IL-4, IL-2, IL-10 or tumour necrosis factor (TNF)-α (R&D Systems). Medium was collected 1, 6, 12 or 24 h after start of incubation and IL-7 and IL-15 were measured using an enzyme-linked immunosorbent assay (ELISA) kit (R&D Systems).

Assay of T cell proliferation

The IELs and LPLs were stimulated using anti-human CD2, CD3 and CD28 mAbs. Peripheral blood (PB) memory (CD45RO⁺) T cells were isolated with a MACS system from PB mononuclear cells separated using the Ficoll-Paque method (Amersham Pharmacia Biotech, Uppsala, Sweden) from heparinized blood samples from healthy volunteers. Dynabeads (conjugated with anti-mouse IgG) were incubated with anti-human CD2 mAb (mouse IgG), anti-human CD3 mAb (mouse IgG) or anti-human CD28 mAb (mouse IgG) (all from R&D Systems). Beads were used as polyclonal T cell stimulators. Isolated IELs, LPLs or PB memory T cells (10⁴/well) were incubated in 96-well round-bottomed microtitre plates with the Dynabeads. Assays performed in triplicate were cultured for 4 days and pulsed with [³H]-thymidine (1 µmCi/well). Sixteen hours later, cells were harvested onto glass fibre filters. [³H]-thymidine uptake was determined using a Microbeta 1450 liquid scintillation counter (Wallac, Gaithersburg, MD, USA).

Immunohistochemistry of adhesion molecule expression in HBECs

Immunohistochemistry was undertaken to detect CD58 and E-cadherin in HBECs. Bronchial specimens frozen immediately after resection and cut into 6 µm-thick sections using a cryostat, placed on glass slides, air-dried for 1 h, fixed in acetone at 4°C for 10 min and air-dried again for 30 min. Circles were drawn around sections using a Dako pen (Dako, Glostrup, Denmark). Non-specific binding sites were blocked with 5% goat serum in phosphate-buffered saline (PBS). Mouse anti-human CD58 mAb (BD Pharmingen) and anti-human E-cadherin mAbs (Dako) diluted appropriately in PBS were applied for 45 min at room temperature. Preparations were washed and incubated with alkaline phosphatase-conjugated goat anti-mouse Ig (Dako) (1:100) for 30 min. Enzymatic activity was visualized with a vector red alkaline phosphatase substrate kit (Vector Laboratories, Burlingame, CA, USA), according to the manufacturer's instructions. Endogenous alkaline phosphatase was blocked using levamisole, and nuclei were counterstained with haematoxylin.

Western blot analysis of adhesion molecules

Western blotting was undertaken to detect CD58 and E-cadherin in HBECs. Confluent cultures of BEAS-2B and primary HBECs (Clonetics, San Diego, CA, USA) grown on 10-cm collagen-coated dishes were washed twice with 1× PBS. The cells were lysed in a buffer containing 20 mM tris-HCl (pH 7·8), 50 mM NaCl, 50 mM NaF, 30 mM Na₄P₂O₇, 5 mM ethylene glycol tetraacetic acid, 1 mM Na₃VO₄, 1% Triton X-100 and protease inhibitors (1 mg/ml leupeptin, 1 mM phenylmethylsulphonyl fluoride, 1 mg/ml aprotinin and 1 mg/ml pepstatin A) on ice for 60 min. After centrifugation at 12 000 g for 20 min on 4°C, supernatants were collected and the proteins were denatured by incubation on 37°C for 40 min or 95°C for 4 min. Samples were electrophoresed through 8% Tris-glycine gels (Invitrogen, Carlsbad, CA, USA) and transferred to nitrocellulose membranes (Amersham, Little Chalfont, Buckinghamshire, UK). Membranes were blocked with NET/gelatin (150 mM NaCl, 5 mM EDTA, 50 mM Tris, 0·05% Triton X-100, and 0·25% gelatin), and reacted with mouse anti-human E-cadherin mAb (BD Transduction Laboratories, San Diego, CA, USA) (1:1000) or mouse anti-human CD58 mAb (R&D systems) (1:1000) overnight at 4°C, then incubated with horseradish peroxidase-conjugated goat anti-mouse IgG (Santa Cruz Biotechnology, Santa Cruz, CA, USA) (1:5000) for 40 min at room temperature. The signals were developed using the Western Lightning Chemiluminescence Reagent (Perkin-Elmer Life Sciences, Boston, MA, USA) and exposed to Kodak BioMax films (Eastman Kodak, Rochester, NY, USA).

Statistical analysis

Values were expressed as means ± standard error of the mean. Statistical significance (P < 0·05) of the number of cytokine-producing cells per 10⁴ cells was analysed using the Wilcoxon's signed rank test, and the statistical significance (P < 0·05) of ELISAs or proliferation assays was analysed using the Mann–Whitney U-test.

Results

CD103 expression in isolated human bronchial IELs and LPLs

To confirm that the separation of IELs and LPLs was successful we analysed CD103 expression of the isolated these two T cell populations. The percentage of CD103-expressing cells in isolated IELs was 63·9% and that in isolated LPLs was 18·3% (Fig. 2). These ratios are similar to those we reported previously [9], suggesting that we have isolated populations of IELs and LPLs.

Fig. 2.

CD103 expression in isolated human bronchial intraepithelial lymphocytes (IELs) and lamina propria lymphocytes (LPLs). Freshly purified T cells were stained directly by fluorescein isothiocyanate (FITC)-labelled anti-human CD103 monoclonal antibody and expressions were determined by flow cytometry. The percentage of CD103⁺ cells increased in bronchial IELs compared with LPLs.

Cytokine production of human bronchial IELs and LPLs

To investigate types of cytokines produced by bronchial IELs and LPLs, the ratio of cytokine-producing cells was determined. The number of IFN-γ spot-forming cells (SFCs) in bronchial IELs was significantly greater than in LPLs (1256·0 ± 131·0 SFCs versus 895·5 ± 91·6 SFCs/10⁴ cells; P = 0·02; Fig. 3a). Overall, 12·6% of bronchial IELs and 9·0% of LPLs secreted IFN-γ. In contrast, the number of IL-2, IL-4 and IL-10 SFCs did not differ significantly between IELs and LPLs (Fig. 3b–d). The ratio of IFN-γ SFCs/IL-4 SFCs in IELs was significantly greater than that in LPLs (14·6 ± 3·2 versus 7·8 ± 1·2; P = 0·003; Fig. 3e). These results suggest that a larger proportion of IELs produces type 1 cytokines than that seen in LPLs.

Fig. 3.

Cytokine-secreting cells in human bronchial intraepithelial lymphocytes (IELs) and lamina propria lymphocytes (LPLs) examined by enzyme-linked immunospot assay. The number of spot-forming cells (SFCs) (open circles) of interferon (IFN)-γ (a), interleukin (IL)-4 (b), IL-2 (c), IL-10 (d) per 10⁴ of IELs and LPLs is shown. (e) The ratio of the number of IFN-γ SFCs to the number of IL-4 SFCs. Solid circles and bars indicate mean values ± standard error of the mean. Lines connect IEL and LPL values from the same donor. The number of IFN-γ-producing cells in bronchial IELs was significantly greater than in LPLs (a), and the IFN-γ SFCs/IL-4 SFCs ratio in IELs was also significantly greater than that in LPLs (e). *P < 0·05.

To examine a potential influence of smoking, we compared results in samples from smokers and non-smokers. Neither the number of IFN-γ SFCs in bronchial IELs (smokers versus non-smokers: 1324·4 ± 190·6 SFCs versus 1194·4 ± 187·8 SFCs/10⁴ cells; P = 0·74) nor the ratio of IFN-γ SFCs/IL-4 SFCs in IELs (smokers versus non-smokers: 12·9 ± 3·4 versus 16·2 ± 5·3; P = 0·87) differed significantly in smokers versus non-smokers. Similarly, the number of IL-2, IL-4, and IL-10 SFCs in bronchial IELs and LPLs was not significantly different between smokers and non-smokers.

Cytokine production by CD4⁺ and CD8⁺ cells in IELs and LPLs

Next, we asked the origin of the cytokines, IFN-γ, IL-4, IL-2 and IL-10 in the isolated bronchial IELs and LPLs. We examined the production of these cytokines of the CD4⁺ cells and CD8⁺ cells isolated individually from bronchial IELs or LPLs by ELISPOT assay. CD8⁺ IELs produced IFN-γ more robustly than did CD4⁺ IELs (1259 ± 262 SFCs versus 796 ± 146 SFCs/10⁴ cells; P = 0·01; Fig. 4a), suggesting that the most significant source of type 1 cytokine in bronchial mucosa is CD8⁺ IELs. The number of IL-10-producing cells in CD4⁺ IELs was greater than that seen in CD4⁺ LPLs (467 ± 99 SFCs versus 312 ± 80 SFCs/10⁴ cells; P = 0·01; Fig. 4d), whereas other cytokine productions were not significantly different between the subpopulations of IELs and LPLs.

Fig. 4.

Cytokine-secreting cells of human bronchial intraepithelial lymphocytes (IELs) and lamina propria lymphocytes (LPLs) purified to the CD4⁺ or CD8⁺ compartment. Shown are the number of spot-forming cells (SFCs) (open circles) of interferon (IFN)-γ (a), interleukin (IL)-4 (b), IL-2 (c), IL-10 and (d) per 10⁴ of CD4⁺ or CD8⁺ IELs and CD4⁺ or CD8⁺ LPLs, and the ratio of IFN-γ SFCs to IL-4 SFCs (e). Solid circles and bars indicate values ± standard error of the mean. Lines connect IEL and LPL values from the same donor. Larger numbers of CD8⁺ IELs produced IFN-γ compared with CD4⁺ IELs (a), and the ratio of IFN-γ SFCs to IL-4 SFCs was greater in CD8⁺ IELs than in CD4⁺ IELs (e). *P < 0·05.

Production of T cell growth factors from HBECs stimulated by cytokines

We hypothesized that HBECs may secrete T cell growth factors to maintain bronchial IELs and that the production of those factors from epithelial cells might be regulated by bronchial IELs. Thus, we analysed cytokine production from the HBEC line BEAS-2B. IL-7 and IL-15 were produced constitutively by BEAS-2B, and cytokine production was accelerated by IFN-γ stimulation (Fig. 5a, b). Production of IL-15 was dependent on the concentration of IFN-γ (Fig. 5b), and production of both IL-7 and IL-15 was time-dependent (Fig. 5c, d). In addition to TNF-α, neither 0·05 nor up to 50 µg of IL-4, IL-2, IL-10 stimulation for 1–24 h induced IL-7 and IL-15 production over background (data not shown).

Fig. 5.

Interleukin (IL)-7 (a) and IL-15 (b) production from bronchial epithelial cell line/BEAS-2B cells treated with various concentrations of interferon (IFN)-γ, IL-4 or tumour necrosis factor (TNF)-α for 24 h; 0 indicates untreated cells assayed in the same manner. Shown is IL-7 (c) and IL-15 (d) production from BEAS-2B cells grown for 1, 6, 12 or 24 h without cytokine, with IFN-γ or with TNF-α. IFN-γ stimulates IL-7 and IL-15 production. *P < 0·05, different from untreated cells; **P < 0·05, different from untreated cells incubated for the same period.

Primary bronchial epithelial cells produced IL-7 and IL-15, especially by IFN-γ stimulation (Fig. 6a, b) but not by IL-4, IL-2, IL-10 stimulation (data not shown), similar to BEAS-2B.

Fig. 6.

Interleukin (IL)-7 (a) and IL-15 (b) production from primary human bronchial epithelial cells (HBECs) grown in the presence of interferon (IFN)-γ IL-4 and tumour necrosis factor (TNF)-α for 24 h; 0 indicates untreated cells assayed in the same manner. IFN-γ stimulates production of both IL-7 and IL-15 in primary HBECs.

Proliferation of human bronchial IELs, LPLs and PB memory T cells

To investigate the ability of proliferation of IELs and LPLs, we stimulated these cells by anti-human CD2, CD3 and CD28 mAbs. Human bronchial IELs proliferated more actively following cross-linking of CD2 molecules than did LPLs or PB memory T cells. Furthermore, when CD2 or CD3 was cross-linked simultaneously with CD28 (a co-stimulatory molecule), proliferation of human bronchial IELs was increased significantly compared with cross-linking of CD2 or CD3 alone (Fig. 7).

Fig. 7.

Proliferation of human bronchial intraepithelial lymphocytes (IELs) and lamina propria lymphocytes (LPLs) by stimulation with monoclonal antibodies (mAbs) to T cell surface molecules. Stimulation index = [³H]-thymidine incorporation stimulated by mAb/[³H]-thymidine incorporation stimulated by control immunoglobulin G₁. IEL proliferation was increased by CD2 or CD3 stimulation, and additional CD28 stimulation increased proliferation further. Increased proliferation resulting from CD2 stimulation of IELs was significantly greater than that seen in LPLs and peripheral blood memory T cells (PB CD45⁺ T cells). **P < 0·05.

Adhesion molecule expression on HBECs

To investigate the adhesive potential of human bronchial IELs and LPLs to epithelial cells, we analysed expression of surface molecules on epithelial cells that may adhere to bronchial T cells. Immunohistochemical analysis showed that HBECs express constitutively the adhesion molecules CD58 (a CD2 ligand) and E-cadherin (a CD103 ligand) (Fig. 8b, c respectively). The proteins of these molecules were also detected in cultured HBECs by Western blotting regardless of denatured temperature (Fig. 8d).

Fig. 8.

Immunohistochemistry of a human bronchial specimen using mouse immunoglobulin G₁ (a), mouse anti-human CD58 monoclonal antibody (mAb) (b) and mouse anti-human E-cadherin mAb (c). Human bronchial epithelium was CD58- and E-cadherin-positive. Bars indicate 100 µm. CD58 and E-cadherin were also detected on bronchial epithelial cell line/BEAS-2B and primary human bronchial epithelial cells (HBECs) HBECs by Western blotting (d).

Discussion

In this study, we show that: (i) human bronchial IELs exhibit an apparent type 1 cytokine profile and proliferate more actively in response to CD2, CD3 or CD28 signalling than do bronchial LPLs; (ii) the most significant source of IFN-γ is CD8⁺ IELs; (iii) HBECs produce the T cell growth factors IL-7 and IL-15 when stimulated by IFN-γ; and (iv) CD58 (a CD2 ligand) and E-cadherin (a CD103 ligand) are expressed constitutively on the epithelial surface.

The success of methods that we used to isolate bronchial IELs and LPLs was confirmed by microscopic observation of the basement membrane preserved after scraping the epithelial layer, and by observation of ratios of CD103 expression by flow cytometry in IELs and LPLs similar to our previous immunohistochemistry analysis of human bronchi [9].

The characteristics of human bronchial IELs and LPLs have not been investigated thoroughly. In fact, some studies focusing on immunohistochemical analysis of T cells obtained by endobronchial biopsy have not distinguished between IELs and LPLs [15,16], and some investigators who obtained T cells by bronchial brushing did not identify LPLs [17].

Although, to our knowledge, we are the first to analyse cytokine production by human bronchial IELs and LPLs using an ELISPOT assay, human intestinal IELs and LPLs have been investigated previously. Using the ELISPOT technique, Carol et al. showed that 3·6% of duodenal IELs and 1·9% of LPLs secrete constitutively IFN-γ and 1·3% of duodenal IELs, and 0·7% of LPLs secrete constitutively IL-4, and indicate that IFN-γ and IL-4 were produced mainly by CD4⁺ T cell population [20]. Lundqvist et al. demonstrated that about 10% of intestinal IELs produced IFN-γ, but no type 2 cytokines were produced following PMA and ionomycin stimulation [21]. As noted, the cytokine profile of bronchial IELs and LPLs was not entirely consistent with human intestinal counterparts. We speculate that these differences originate from organ-specific mucosal immunity. Unlike the airway tract, the intestine is exposed to large amounts of food antigens and bacterial flora and have gut-associated lymphoid tissue in which some IELs develop and differentiate [22,23].

The IFN-γ, which was the most highly secreted cytokine by bronchial CD8⁺ IELs, has anti-viral activity, induces bronchial epithelial cells to express major histocompatibility complex class II [24] and inhibits epithelial growth [25]. Recently, Hodge et al. showed that a larger number of CD8⁺ cells produced IFN-γ than did CD4⁺ cells [17] using bronchial brushing methods (in which most separated T cells will be IELs), a finding comparable to the results of bronchial IELs reported here. Intestinal CD8⁺ IELs have cytotoxic activity towards intestinal epithelial cells [3,21,26]. Therefore, bronchial IELs may contribute to elimination of damaged bronchial epithelial cells by these mechanisms. Because, in contrast to the type 2 immune response, the type 1 immune response occurs immediately without B cell activation, the type 1 response in bronchial epithelium is a possible candidate for mediating mucosal immunity described above. The percentage of CD4⁺ and CD8⁺ cells in human bronchial IELs was about 50% and 50% respectively, and in LPLs was 65% and 35% (unpublished data). We hypothesize that CD8⁺ cells act in the first line of bronchial mucosal immune system.

The present study showed that IL-7 and IL-15 are secreted constitutively from bronchial epithelial cells. IL-7 is a stromal cell-derived pleiotropic cytokine regulating lymphopoiesis. IL-15 is an IL-2-like T cell growth factor produced by several cell types in different tissues, whereas IL-2 is produced mainly by activated T cells. IL-7 and IL-15 are synthesized by human intestinal epithelial cells and can stimulate growth of intestinal mucosal T cells [27,28]. Therefore, IL-7 and IL-15 secreted constitutively from HBECs may maintain bronchial IEL homeostasis. IFN-γ is produced primarily from human bronchial IELs and stimulates bronchial epithelial cells to produce IL-7 and IL-15, and airway inflammation activates IELs to secrete higher levels of IFN-γ, which stimulate bronchial epithelial cells to increase IL-7 and IL-15 production, further activating IELs. We speculate that in human bronchial tissue there is cell–cell positive interaction between epithelial cells and IELs.

Proliferation of bronchial IELs was more sensitive to CD2 and CD3 signalling than was proliferation of LPLs in this study, whereas studies of intestine indicate that IELs and LPLs are much more reactive to CD2 than to CD3 signalling [29,30], and that no significant proliferative differences exist between IELs and LPLs [30]. Another study indicates that intestinal LPLs are more reactive to CD3 plus the co-stimulatory molecule CD28 signalling than are IELs [31]. Because, unlike intestine, bronchial mucosa is generally aseptic, high bronchial IEL reactivity to CD2 and CD3 with or without CD28 signalling may contribute to the response to low levels of antigens derived from bronchial mucosal pathogens.

We found constitutive expression of CD58 (LFA-3), a CD2 ligand, and E-cadherin, a CD103 ligand on HBECs. It was shown previously that CD58 on intestinal epithelial cells stimulates intestinal mucosal T cells [32], and that E-cadherin expressed on various epithelial tissues adheres preferentially to CD103 [33]. Therefore, CD58 and E-cadherin expressed on the bronchial epithelial cells may provide an important physiological ligand for bronchial T cells, as in intestine, and function in adhesion and signalling between human bronchial IELs and epithelial cells.

The reason that human bronchial IELs survive for longer periods than did LPLs in our previous study [9] is that bronchial epithelial cells maintain preferably IELs through these cell interactions by mediators or adhesion molecules that were shown in this study.

In summary, we isolated human bronchial IELs and LPLs successfully, and found that (i) human bronchial IELs, particularly CD8⁺ cells, produce IFN-γ and proliferate more actively following surface antigen stimulation than did bronchial LPLs, (ii) HBECs produced T cell growth factors IL-7 and IL-15 by stimulation with IFN-γ, and expressed constitutively CD58 and E-cadherin, ligands for CD2 and CD103 respectively. These results suggest that human bronchial IELs have immunological functions distinct from those of bronchial LPLs and that IELs interact with epithelial cells to maintain mucosal homeostasis. Differences between bronchial IELs and LPLs may be due to ontogeny, the frequency of stimulation by airborne antigens or the degree of their association with bronchial epithelial cells. Further analysis of the role of human bronchial IELs, LPLs and epithelial cells will provide an understanding of the human airway mucosal system and the pathogenesis of various airway diseases.