Involvement of the interferon-γ–induced T cell–attracting chemokines, interferon-γ–inducible 10-kd protein (CXCL10) and monokine induced by interferon-γ (CXCL9), in the salivary gland lesions of patients with Sjögren's syndrome

Authors


Abstract

Objective

To elucidate the mechanism of the development of T cell infiltrates in the salivary glands of patients with Sjögren's syndrome (SS), we studied T cell–attracting chemokines and their receptors.

Methods

The expression of the T cell–attracting chemokines, interferon-γ (IFNγ)–inducible 10-kd protein (IP-10; also called CXCL10), monokine induced by IFNγ (Mig; also called CXCL9), and stromal cell–derived factor 1 (SDF-1; also called CXCL12), in salivary glands from SS patients was investigated by polymerase chain reaction–enzyme-linked immunosorbent assay (ELISA). Cells that produce chemokines and lymphocytes that express chemokine receptors were identified by immunohistochemistry. The production of IP-10 and Mig proteins by salivary epithelial cells in response to IFNγ was determined by ELISA.

Results

Expression of IP-10 and Mig messenger RNA (mRNA) was significantly up-regulated in SS salivary glands compared with normal salivary glands (both P < 0.01). There was no significant difference in SDF-1 mRNA expression between the SS and normal salivary glands. IP-10 and Mig proteins were predominantly expressed in the ductal epithelium adjacent to lymphoid infiltrates. Most of the CD3+ infiltrating lymphocytes in dense periductal foci expressed CXCR3, the receptor for IP-10 and Mig. IFNγ induced the production of high levels of IP-10 and Mig proteins from cultured SS salivary epithelial cells.

Conclusion

These findings suggest that IFNγ stimulates the production of IP-10 and Mig in the SS ductal epithelium, and that IP-10 and Mig are involved in the accumulation of T cell infiltrates in the SS salivary gland. Chemokines or chemokine receptors could be a rational new therapeutic target in SS.

Sjögren's syndrome (SS) is a unique autoimmune and lymphoproliferative disorder in which the salivary and lacrimal glands are the major target for autoimmune tissue damage. Malignant lymphomas develop in less than 10% of patients (1). One of the features of the SS salivary gland is an elevated expression of HLA–DR antigens on the salivary epithelial cells (2, 3). The HLA–DR molecules are thought to allow epithelial cells to present putative autoantigens to autoreactive T lymphocytes. Interferon-γ (IFNγ) is a major cytokine that is present in the SS salivary gland (4–7), and it induces HLA–DR expression on salivary epithelial cells (2, 7, 8). It is believed that mononuclear cell infiltrates containing a predominance of activated T cells develop around the duct in the SS salivary gland as a result of autoimmune reactions between epithelial cells and lymphocytes (9, 10). B cells are found less frequently than T cells in SS salivary gland lesions, and macrophages and natural killer cells are rarely found (3). These cell infiltrates lead to salivary gland destruction and, ultimately, to salivary hypofunction through such mechanisms as Fas and Fas ligand interactions between lymphoid infiltrates and glandular cells (11, 12). However, it is not clear how cell infiltrates develop in the SS salivary gland.

Chemokines have recently been cloned and have chemotactic activities toward a variety of cells (13). Each chemokine has specificity for different target cells and binds to its receptors on the cell surface. The responsiveness of a target cell to a chemokine depends on receptor signaling as well as receptor expression (14, 15). Chemokines are involved in various aspects of immune responses, such as atherosclerosis (16), antiviral defense (17), rejection of allografts (18), and experimental autoimmune encephalomyelitis (19). Cytokines such as IFNγ are important regulators of chemokine expression (20, 21). In many tissues, including the heart, liver, lung, ovary, spleen, and uterus, IFNγ induces two CXC chemokines: IFNγ-inducible 10-kd protein (IP-10; also called CXCL10) and monokine induced by IFNγ (Mig; also called CXCL9) (22). IP-10 and Mig bind to a chemokine receptor, CXCR3, that is highly expressed on activated T lymphocytes but not on resting T cells, B cells, monocytes, or granulocytes (23, 24). Stromal cell–derived factor 1 (SDF-1; also called CXCL12), on the other hand, attracts CD45RA+ naive T lymphocytes and progenitor B cells that express CXCR4 (25–27). Two isoforms of SDF-1, SDF-1α and SDF-1β, are generated through alternative splicing and only differ by a 4–amino acid extension in the carboxy-terminus of SDF-1β. IP-10, Mig, and SDF-1 are important chemokines in the development of cell infiltrates that consist predominantly of T cells because all other chemokines which act on lymphocytes also act on monocytes and granulocytes (28).

To study the involvement of IP-10, Mig, and SDF-1 in the development of the SS salivary gland lesion, the expression of these chemokines in minor salivary glands (MSGs) was determined using polymerase chain reaction–enzyme-linked immunosorbent assay (PCR-ELISA) and immunohistochemistry techniques. Expression of CXCR3 and CXCR4 in SS salivary glands was examined by immunohistochemistry. Using cultured epithelial cells from salivary glands, the effects of IFNγ on the production of IP-10 and Mig messenger RNA (mRNA) were studied by reverse transcription (RT)–PCR, and the effects of IFNγ on the production of IP-10 and Mig proteins were assessed by ELISA. Our data suggest that IP-10 and Mig induced by IFNγ in salivary epithelial cells play an important role in the accumulation of activated T cells in the SS salivary gland.

MATERIALS AND METHODS

Patients and controls.

Twenty-seven patients with SS (23 women and 4 men; mean ± SD age 60.9 ± 13.8 years, age range 31–80 years) and 16 normal healthy control subjects (11 women and 5 men; mean ± SD age 56.1 ± 5.0 years, age range 31–91 years) were recruited for the study. All the patients were from Kanazawa Medical University Hospital (Ishikawa-ken, Japan). The SS patients fulfilled the diagnostic criteria for SS as previously described (29). All experimental protocols were approved by the Institutional Review Board of the Kanazawa Medical University Hospital.

The SS patient group consisted of 17 patients who had primary SS (16 women and 1 man; mean ± SD age 59.9 ± 14.6 years, age range 31–80 years), and 10 patients who had secondary SS (7 women and 3 men; mean ± SD age 59.8 ± 11.7 years, age range 43–74 years). Of the 10 patients with secondary SS, 4 had rheumatoid arthritis, 2 had systemic lupus erythematosus, 1 had dermatomyositis, 1 had polymyositis, and 2 had scleroderma. After obtaining informed consent, labial minor salivary gland biopsy was performed on all patients and in 12 of the healthy controls for diagnostic evaluation of SS. Normal parotid gland tissue was also obtained from a patient who was undergoing surgery for a benign pleomorphic tumor of the parotid gland. MSG and parotid gland tissues were processed for RT-PCR, immunohistochemistry, and production of primary cultures of epithelial cells.

Reagents.

Recombinant human IFNγ was generously provided by Shionogi Corporation (Osaka, Japan). Recombinant human interleukin-1β (IL-1β), IL-4, IL-6, IL-10, tumor necrosis factor α (TNFα), and transforming growth factor β (TGFβ) were purchased from Wako (Osaka, Japan). As for the antibodies used for immunohistochemistry, goat polyclonal antibody to human IP-10, mouse monoclonal antibody (mAb) to human Mig (49106.11), mouse mAb to CXCR3 (49801.111), and mouse mAb to CXCR4 (44716.111) were purchased from R&D Systems (Minneapolis, MN). Mouse mAb to human CD3 and biotinylated rabbit anti-goat Ig were purchased from Dako (Glostrup, Denmark). Mouse IgG1 (DAK-GO1) and IgG2b (DAK-GO9) were purchased from Dako and used as negative controls. Fluorescein isothiocyanate–conjugated mouse mAb to human CXCR3 (49801.111) was purchased from Dako and used for flow cytometry. Recombinant human IP-10 and Mig were purchased from R&D Systems and used for chemotaxis.

Primary cultures of epithelial cells from salivary gland tissues.

Epithelial cells from MSG or from parotid gland tissues were cultured immediately after biopsy or surgery as previously described (5). Briefly, tissues were rinsed with cold sterile phosphate buffered saline (PBS) containing 100 units/ml of penicillin and 100 μg/ml of streptomycin, then minced into fragments of ∼1 mm3. One piece of the tissue was placed in a 24-well plate coated with type I collagen (Sigma, St. Louis, MO). The culture medium was a mixture (2:1) of calcium-free Dulbecco's modified Eagle's medium (Invitrogen/Life Technologies, Carlsbad, CA) and Ham's F-12 (Invitrogen) containing 2.5% fetal calf serum (FCS), 2 mML-glutamine (Invitrogen), 5 μg/ml of insulin (Wako), 0.4 μg/ml of hydrocortisone (Sigma), 10 ng/ml of epidermal growth factor (Sigma), and 25 μg/ml of bovine pituitary extract (Sigma).

Epithelial cell outgrowth from the explant was observed after 1–2 weeks of explantation. The cultured epithelial cells were subjected to experiments when they reached 70–80% confluency. Fibroblasts were routinely removed from the culture by treating the cells with 0.02% EDTA (Invitrogen).

RT-PCR.

RNA was extracted from the cells or tissues using TRIzol reagent (Invitrogen). The concentration of the RNA was estimated by spectrophotometry. The RNA was treated with DNase I (Invitrogen) to remove any genomic DNA that might be contaminating the RNA preparations. Complementary DNA (cDNA) was synthesized from 2 μg of total RNA using a cDNA synthesis kit (SuperScript II DNA Preamplification System; Invitrogen). A cDNA reaction mixture from 0.1 μg of RNA was used for DNA amplification by PCR. A typical amplification reaction included 2 units of Taq polymerase (Takara, Shiga, Japan), 20 pmoles of sense and antisense oligonucleotide primers, and 200 μM each of dATP, dCTP, dGTP, and dTTP. Amplification was carried out for 30 cycles of 1 minute at 92°C, 1 minute at 55°C, and 1 minute at 72°C. The amplified DNA was electrophoresed on a 2% agarose gel (Invitrogen), stained with ethidium bromide, visualized under ultraviolet light, and photographed.

The primer sequences used were as follows: for GAPDH, 5′-TCC-ATG-ACA-ACT-TTG-GTA-TCG-3′ (sense) and 5′-GTC-GCT-GTT-GAA-GTC-AGA-GGA-3′ (antisense); for IP-10, 5′-GGA-ACC-TCC-AGT-CTC-AGC-ACC-3′ (sense) and 5′-CAG-CCT-CTG-TGT-GGT-CCA-TCC-3′ (antisense); for Mig, 5′-AGA-AAG-GGA-ACG-GTG-AAG-TA-3′ (sense) and 5′-CAG-CAG-TGT-GAG-CAG-TGA-TT-3′ (antisense); for SDF-1α, 5′-CAT-TAA-TCT-TGC-TTC-TGC-TT-3′ (sense) and 5′-TCA-CTC-AAA-GCG-AGC-TCT-3′ (antisense); for SDF-1β, 5′-CCA-TGG-GAG-AAA-ATA-GAT-AA-3′ (sense) and 5′-CAA-GGG-AGT-GTC-AGG-TAG-AG-3′ (antisense); for CXCR3, 5′-TTG-ACC-GCT-ACC-TGA-ACA-TA-3′ (sense) and 5′-ACG-TCT-ACC-CTG-CTT-TCT-CG-3′ (antisense); and for CXCR4, 5′-ACG-TCA-GTG-AGG-CAG-ATG-3′ (sense) and 5′-GAT-GAC-TGT-GGT-CTT-GAG-3′ (antisense). The expected sizes of the cDNA amplicons were as follows: 376 bp for GAPDH, 377 bp for IP-10, 205 bp for Mig, 272 bp for SDF-1α, 272 bp for SDF-1β, 456 bp for CXCR3, and 251 bp for CXCR4.

Semiquantitative analysis of chemokine mRNA expression by PCR-ELISA.

PCR-ELISA was used to semiquantify the levels of mRNA for IP-10, Mig, SDF-1α, SDF-1β, and GAPDH as previously described (30). Briefly, cDNA was synthesized from 0.1 μg of RNA using 2 units of Taq polymerase, 20 pmoles of sense and antisense oligonucleotide primers, 200 μM each of dATP, dCTP, and dGTP, 190 μM dTTP, and 10 μM digoxigenin-conjugated dUTP. Amplification was carried out for 30 cycles of 1 minute at 92°C, 1 minute at 55°C, and 1 minute at 72°C. The digoxigenin-labeled PCR products were then analyzed by ELISA.

Briefly, 10 μl of a PCR reaction mixture was denatured and mixed with a biotin-labeled probe. The reaction mixture was pipetted into the wells of streptavidin-coated 96-well micro-ELISA plates. The biotin-labeled capture probe allowed immobilization of the specific hybrid of the probe PCR product to a streptavidin-coated surface. Microtiter plates were incubated for 3 hours at 55°C on a microplate shaker. After incubation, unbound nonspecific PCR products were washed from the wells and discarded. Then the bound hybrid was detected by an antidigoxigenin–peroxidase conjugate and by the substrate, and the color was developed in proportion to the amount of bound specific PCR product. The ratio of each gene message to the GAPDH gene message was calculated.

The sequences of the biotin-labeled capture probes for the PCR-ELISA were as follows: for GAPDH, 5′-CGG-CAG-GTC-AGG-TCC-ACC-AC-3′; for IP-10, 5′-TGG-CCT-TCG-ATT-CTG-GAT-TC-3′; for Mig, 5′-CAG-AGG-CTA-ACT-GGG-CAC-CA-3′; for SDF-1α, 5′-GCG-GGT-AAG-CAG-GGG-GAC-CA-3′; and for SDF-1β, 5′-CAG-CAC-CAG-GTC-CCG-GAG-GG-3′.

Immunohistochemistry.

Immunohistochemical studies were performed on 0.5 μm–thick frozen sections of MSGs using an avidin–biotin–immunoperoxidase system, LSAB kit (Dako). Briefly, sections were fixed in cold acetone for 10 minutes, placed on silanized slides (Dako), quenched with 0.3% H2O2 in methanol, and blocked with normal goat serum at room temperature. Slides were incubated overnight with the primary antibody to IP-10, Mig, CXCR3, CXCR4, or CD3 at a concentration of 1 μg/ml at 4°C. A biotinylated rabbit anti-mouse IgG immunoconjugate (Dako) was used for the second antibody, according to the manufacturer's instructions. To detect bound goat antibody to IP-10, sections were incubated with biotinylated rabbit anti-goat Ig antibody (1:500 dilution; Dako). Incubations with the second antibodies were performed for 30 minutes in a humidified chamber at room temperature, followed by 3 changes of Tris buffered saline for 5 minutes. Complexes were stained with streptavidin and developed with chromogenic peroxidase substrate for 10–20 minutes. Mayer's hematoxylin was used for counterstaining. Control slides were incubated with Tris buffered saline containing isotype-matched antibodies instead of the primary antibody; they were invariably negative (data not shown).

Determination of chemokine levels in culture supernatants by ELISA.

To quantify the amount of chemokines released into the culture supernatants of the salivary epithelial cells, chemokine levels were analyzed by ELISA. IP-10 was measured using a human IP-10 ELISA test kit (Hycult Biotechnology, Uden, The Netherlands), and Mig was measured using a mAb for human Mig (49106.11) and a biotinylated anti-human Mig mAb (BJG01) according to the manufacturer's instructions (R&D Systems).

Isolation of peripheral blood mononuclear cells (PBMCs).

Heparinized peripheral blood was obtained from some of the SS patients and normal controls. Mononuclear cells were separated by Ficoll-Hypaque (Amersham Pharmacia Biotech, Uppsala, Sweden) centrifugation as previously described (31). The resultant mononuclear cells were resuspended in RPMI 1640 with 10% FCS, 100 units/ml of penicillin, and 100 μg/ml of streptomycin, and used for RT-PCR, flow cytometry, and chemotaxis assay.

Flow cytometry.

A total of 1 × 106 cells in PBS (pH 7.2) containing 1 mg/ml of human IgG and 0.02% NaN3 were incubated with a fluorescein isothiocyanate–conjugated anti-human CXCR3 mAb (49801.111) for 30 minutes at 4°C. After 3 washes with PBS, cells were analyzed on a FACScan flow cytometer (BD PharMingen, San Diego, CA). Negative controls were cells incubated with the same amount of isotype-matched mouse Ig.

Chemotaxis assay.

Cells were washed twice with PBS, then cell pellets were made by centrifugation. Viable cells were resuspended in 200 μl of RPMI 1640 containing 10% FCS in a 6.5 mm–diameter polycarbonate Transwell with 5.0-μm pores (Corning Life Sciences, Acton, MA). The lower chamber was filled with 800 μl of medium containing an optimum concentration of a recombinant chemokine. Transwells were incubated for up to 3 hours at 37°C in an atmosphere of 5% CO2 and 95% air. Cells that migrated to the bottom chamber were transferred to a tube. The number of transmigrated cells was determined at 1-hour intervals using a Coulter counter (Beckman Coulter, Fullerton, CA). All experiments were performed in triplicate.

Statistical analysis.

Results are expressed as the mean ± SD and were compared by the Mann-Whitney U test. P values less than 0.05 were considered significant.

RESULTS

Expression of IP-10 and Mig in the ductal epithelium of SS salivary glands.

First, we examined the expression of chemokine mRNA in MSGs. Total RNA was extracted from MSGs, reverse-transcribed, and analyzed for the expression of mRNA for IP-10, Mig, SDF-1α, and SDF-1β by PCR. A representative result is shown in Figure 1A.

Figure 1.

Expression of mRNA for interferon-γ (IFNγ)–inducible 10-kd protein (IP-10), monokine induced by IFNγ (Mig), stromal cell–derived factor 1α (SDF-1α), and SDF-1β, as analyzed by reverse transcription–polymerase chain reaction (PCR) in minor salivary glands (MSGs) from 2 Sjögren's syndrome (SS) patients, one with a focus score of 2.3 (lane 1) and the other with a focus score of 8.0 (lane 2), and from 2 normal controls (lanes 3 and 4). GAPDH was used as a housekeeping gene. A normal lymph node (LN) was used as a positive control for chemokine mRNA expression (lane 5). B, Relative levels of mRNA for IP-10, Mig, SDF-1α, and SDF-1β in MSGs from 9 SS patients and 7 normal controls. Expression of IP-10, Mig, SDF-1α, and SDF-1β mRNA in SS and normal MSGs was semiquantified by PCR–enzyme-linked immunosorbent assay. The ratio of each gene message to the GAPDH gene message was calculated, and the relative message levels in the SS MSGs were standardized against the mean in the 7 normal controls. Bars show the mean and SD. IP-10 and Mig mRNA expression was significantly enhanced in SS MSGs compared with normal MSGs (∗ = P < 0.01).

SS patients showed stronger expression of IP-10 and Mig mRNA compared with normal controls. In contrast, SDF-1α was expressed in all SS and normal MSGs examined. We further analyzed this by semiquantification of the chemokine mRNA levels in MSGs by PCR-ELISA (Figure 1B). The expression of IP-10 mRNA and Mig mRNA was significantly enhanced in SS MSGs compared with normal control MSGs (mean ± SD arbitrary units for IP-10 12.3 ± 8.64 versus 1.00 ± 1.44 [P < 0.01]; for Mig 6.77 ± 2.56 versus 1.00 ± 0.80; P < 0.01). The SS patients who showed high expression of IP-10 mRNA also showed high expression of Mig mRNA. There was no significant difference in SDF-1α or SDF-1β mRNA expression between the two groups (mean ± SD arbitrary units for SDF-1α 1.08 ± 0.32 versus 1.00 ± 0.47; for SDF-1β 2.56 ± 2.33 versus 1.00 ± 0.87).

Next, we identified cells producing IP-10 or Mig proteins in MSG samples by immunohistochemistry (Figure 2). Frozen sections of MSGs from SS patients and normal controls were stained with specific antibodies to IP-10 or Mig. In all the SS salivary glands studied, IP-10 and Mig proteins were expressed in the ductal epithelium and, to a lesser extent, in the acinar cells. Most notably, IP-10 and Mig proteins were predominantly expressed in the ductal epithelium adjacent to lymphoid infiltrates. IP-10 and Mig proteins were not detected in normal MSGs.

Figure 2.

Expression of IP-10 and Mig proteins in situ in MSGs from SS patients and normal subjects. Frozen sections of MSGs from 8 SS patients and 2 normal controls were stained with specific antibodies to IP-10 (a, c, and e) and Mig (b, d, andf). Ductal epithelium adjacent to lymphoid infiltrates stained predominantly for IP-10 (a and c) and Mig (b and d) in SS MSGs. In normal MSGs, IP-10 and Mig proteins were not detected (e and f). (Magnification × 100 in a, b, e, and f; × 400 in c and d.) See Figure 1 for definitions.

Expression of the CXCR3 chemokine receptor by T lymphocytes in dense periductal foci in SS salivary glands.

Because IP-10 and Mig were expressed in the ductal epithelium of SS salivary glands, we sought to identify cells expressing their receptor, CXCR3, in the SS salivary gland lesions. We therefore analyzed the expression of CXCR3 in MSGs by immunohistochemistry (Figure 3). Most infiltrating lymphocytes in dense periductal foci in SS MSGs were positive for CXCR3. The CXCR3+ cells were virtually all CD3+ lymphocytes. In contrast, CXCR4, the receptor for SDF-1, was not expressed on lymphocytes in dense periductal foci. CXCR4 was expressed on lymphocytes in the area of less intensive infiltration in serial sections of SS salivary glands. Because CXCR3 are exclusively expressed on activated T cells (23, 24), these findings suggest that CXCR3+ activated T cells mainly consist of intensive lymphocytic aggregates of dense periductal foci. Cells expressing CXCR3 or CXCR4 were not detected in normal MSGs.

Figure 3.

Expression of CXCR3 and CXCR4 chemokine receptor proteins in situ in SS and normal MSGs. Frozen sections of MSGs from 5 SS patients and 2 normal controls were studied. Adjacent sections of an SS MSG were stained with anti-CXCR3 (a and c) or CXCR4 (b and d). Most lymphocytes in dense periductal foci were positive for CXCR3 (a and c). In contrast, CXCR4 was expressed on lymphocytes that showed less intensive infiltration around dense periductal foci, with some extracellular staining (b and d). Adjacent sections of an SS MSG were stained with anti-CXCR3 (e) or anti-CD3 antibody (f). The distribution of CXCR3+ cells is similar to that of CD3+ cells. In normal MSGs, cells expressing CXCR3 (g) and CXCR4 (h) were not found. (Original magnification × 200 in a, b, e, and f; × 400 in c and d; × 100 in g and h.) See Figure 1 for definitions.

Induction of IP-10 and Mig production in SS salivary epithelial cells by IFNγ.

The fact that ductal epithelium adjacent to lymphoid infiltrates predominantly expressed IP-10 and Mig prompted us to attempt to identify the factor produced by infiltrating lymphocytes that induces the expression of IP-10 and Mig in the SS ductal epithelium. We chose to study IFNγ because IFNγ is a primary inducer of IP-10 and Mig (28), and it is one of the major cytokines expressed in the SS salivary gland (4). To examine the effect of IFNγ on chemokine expression in the SS salivary gland, we used cultured epithelial cells from salivary glands as previously described (5). Epithelial cells were grown from MSG explants, and cells were cultured in the presence or absence of IFNγ (1,000 units/ml for 12 hours). RNA was isolated from the cells, and the expression of IP-10 and Mig mRNA was examined by RT-PCR.

Quiescent cells from some SS patients showed low levels of IP-10 and/or Mig mRNA, whereas epithelial cells from normal control subjects did not (Figure 4). The culture period for each of the samples was as follows: 74 days for patient SS1, 81 days for patient SS2, 121 days for patient SS3, 25 days for patient SS4, 38 days for patient SS5, 51 days for normal subject 6, and 87 days for normal subject 7. The expression of chemokine mRNA in quiescent cells from some SS patients may be related to the difference in the culture period, because SS4 and SS5 cells, which showed spontaneous expression of IP-10 and/or Mig mRNA, were cultured for shorter periods than were the other SS samples. IFNγ induced the expression of IP-10 and Mig mRNA in SS epithelial cells, but it had no effect on the expression of SDF-1α and SDF-1β mRNA.

Figure 4.

Effects of IFNγ on the expression of IP-10, Mig, SDF-1α, and SDF-1β mRNA in cultured epithelial cells from 5 SS (lanes 1–5) and 2 normal (lanes 6 and 7) MSGs. Epithelial cells were incubated with or without 1,000 units/ml of IFNγ for 12 hours, and isolated RNA was analyzed by reverse transcription–PCR. Quiescent cells from some SS patients showed IP-10 and/or Mig mRNA, whereas epithelial cells from normal controls did not. IFNγ induced IP-10 and Mig mRNA expression, while IFNγ had no effect on SDF-1α and SDF-1β expression, in SS and normal MSG epithelial cells. See Figure 1 for definitions.

To confirm the IFNγ-induced chemokine production by salivary epithelial cells, we measured the amount of IP-10 and Mig proteins that were produced. Salivary epithelial cells grown to confluency were incubated in the presence of IFNγ (1,000 units/ml), and culture supernatants were collected on days 2, 4, and 6. The amount of IP-10 or Mig proteins secreted into the supernatants was determined by ELISA.

SS salivary epithelial cells produced greater amounts of IP-10 and Mig proteins in response to IFNγ compared with normal controls (Figures 5A and B). The culture period for each of the samples was as follows: 44 days for patient SS1, 85 days for patient SS2, 29 days for patient SS3, 22 days for patient SS4, 53 days for patient SS5, and 26 days for the normal controls. The relationship between the culture periods and the production of chemokine proteins was less clear than it had been for the production of mRNA. This suggests that the regulation of chemokine protein production in epithelial cells is more complicated than the expression of mRNA. Only little amounts of chemokines were produced in the absence of IFNγ both in SS and in normal salivary epithelial cells. These results indicate the importance of IFNγ in the production of IP-10 and Mig in the SS salivary gland.

Figure 5.

Levels of A, IP-10 and B, Mig proteins in culture supernatants from IFNγ-stimulated salivary epithelial cells from 5 SS patients (SS1–SS5) and a normal control subject. Confluent salivary epithelial cells were stimulated with 1,000 units/ml of IFNγ for up to 6 days in 6-well plates coated with type I collagen. The supernatants were collected on days 2, 4, and 6, and IP-10 and Mig secreted into the supernatants were determined by enzyme-linked immunosorbent assay. The supernatant collected on day 2 of culture in the absence of IFNγ was used as day 0 supernatants. See Figure 1 for definitions.

To determine whether cytokines other than IFNγ were involved in the production of IP-10 and Mig in SS salivary glands, the effects of a panel of cytokines on SS salivary epithelial cells were examined. IL-1β, IL-4, IL-6, IL-10, TNFα, and TGFβ were chosen because these cytokines are known to be present in the SS salivary gland (5, 6, 32, 33). Stimulation of SS salivary epithelial cells with these cytokines demonstrated that TNFα could induce the expression of IP-10 mRNA but not Mig mRNA (Figure 6). No other cytokines examined were able to induce the production of either IP-10 or Mig mRNA in SS salivary epithelial cells. These data suggest that IFNγ is the key cytokine that induces the production of IP-10 and Mig in the SS salivary gland.

Figure 6.

Effects of cytokines on the expression of IP-10 and Mig mRNA in SS MSG epithelial cells. MSG epithelial cells from an SS patient were incubated for 12 hours with interleukin-1β (IL-1β; 10 ng/ml), IL-4 (10 ng/ml), IL-6 (10 ng/ml), IL-10 (10 ng/ml), IFNγ (100 units/ml), tumor necrosis factor α (TNFα; 10 ng/ml), or transforming growth factor β (TGFβ; 10 ng/ml). Cells were harvested, and isolated RNA was analyzed for the expression of IP-10 and Mig mRNA by reverse transcription–PCR. See Figure 1 for other definitions.

Induction of the expression of functional CXCR3 on freshly isolated SS PBMCs by IFNγ.

Our final goal for this study was to determine whether IFNγ could affect the expression of chemokine receptors on lymphocytes. The effects of IFNγ on the expression of CXCR3 and CXCR4 mRNA in SS and normal PBMCs were therefore examined by RT-PCR. Freshly isolated SS and normal PBMCs were incubated in the absence or presence of 1,000 units/ml of IFNγ for 12 hours. Total RNA was extracted, reverse-transcribed, and analyzed for the expression of CXCR3 and CXCR4 mRNA by PCR. We examined the chemokine receptor message at 12 hours of incubation with IFNγ because the induction of the message occurs rapidly in response to exogenous stimulations. For example, anti-CD3 mAb stimulation induces the cell surface expression of CXCR3 on T cells within a few days (14). IFNγ induced CXCR3 mRNA expression in SS and normal PBMCs, but had little effect on CXCR4 mRNA expression (Figure 7).

Figure 7.

Effects of IFNγ on the expression of CXCR3 and CXCR4 chemokine receptor mRNA in peripheral blood mononuclear cells (PBMCs) from 4 SS patients (lanes 1–4) and 4 normal controls (lanes 5–8). Freshly isolated PBMCs were incubated with or without 1,000 units/ml of IFNγ for 12 hours, and isolated RNA was analyzed by reverse transcription–PCR. IFNγ induced CXCR3 mRNA expression in SS and normal PBMCs, whereas it had little effect on CXCR4 mRNA expression. See Figure 1 for other definitions.

We then examined whether the enhancement of CXCR3 mRNA by IFNγ is in fact related to the up-regulation of functional CXCR3 on lymphocytes. To address this, CXCR3 expression on freshly isolated PBMCs and PBMCs stimulated with IFNγ was examined by flow cytometry. In this experiment, we stimulated the PBMCs with IFNγ for 7 days because the maximum expression of functional CXCR3 on T cells requires a longer period of stimulation compared with mRNA expression. Maximum expression of functional CXCR3 is observed on day 15 in T cells stimulated with IL-2 and phytohemagglutinin (24). As shown in Figure 8, CXCR3 expression was induced on SS and normal lymphocytes incubated with IFNγ. This up-regulation of CXCR3 was found to be related to the enhanced chemotactic activity of SS and normal lymphocytes toward recombinant IP-10 and Mig (Figure 9). These data indicate that IFNγ induces the expression of functional CXCR3 on SS and normal lymphocytes.

Figure 8.

CXCR3 receptor expression on A, freshly isolated and B, IFNγ-stimulated (1,000 units/ml for 7 days) peripheral blood lymphocytes (PBLs) from SS patients. Cells were stained with fluorescein isothiocyanate (FITC)–conjugated anti-human CXCR3 monoclonal antibody (solid histograms) or with a relevant isotype control (open histograms), gated on lymphocytes, and analyzed by flow cytometry. The percentages of CXCR3+ cells are shown. Results are representative of 3 separate experiments. Normal PBLs showed similar CXCR3 expression in response to IFNγ (data not shown). See Figure 1 for other definitions.

Figure 9.

Chemotaxis of peripheral blood mononuclear cells (PBMCs) from 6 SS patients and 6 normal controls against recombinant chemokines. Cells were incubated with A, IP-10 (100 ng/ml) and B, Mig (100 ng/ml) for 3 hours in freshly isolated PBMCs (open bars) and for 1 hour in IFNγ-stimulated PBMCs (1,000 units/ml for 7 days) (shaded bars) in a Transwell. The numbers of transmigrated cells were counted with a Coulter counter. Values are the mean and SD ratio of transmigrated cells with chemokine to transmigrated cells in nonchemokine controls. No significant difference in the chemotactic activities of IFNγ-stimulated and unstimulated cells was noted between SS patients and normal controls. See Figure 1 for other definitions.

Collectively, our findings suggest that IFNγ acts upstream of the events in the development of T cell infiltrates in the SS salivary gland compared with IP-10 and Mig. Locally produced IFNγ is involved in the expression of IP-10 and Mig in the ductal epithelium and in the expression of CXCR3 on T cells, and it contributes to the accumulation of activated T cells in the SS salivary gland.

DISCUSSION

T cells are predominant in infiltrating lymphocytes in the SS salivary gland (3). Among a variety of infiltrating cells, activated T cells are thought to play an important role in the destruction of the salivary gland tissue (11). In the present study, we focused on 3 T cell–attracting chemokines: IP-10 and Mig, which are chemoattractants for activated T cells, and SDF-1, which is a chemoattractant for naive T cells. We demonstrated that IP-10 and Mig proteins were predominantly expressed in the ductal epithelium adjacent to lymphoid infiltrates of the SS salivary gland. IP-10 and Mig were not expressed in the normal salivary gland. Furthermore, most CD3+ infiltrating lymphocytes in dense periductal foci expressed CXCR3, the receptor for IP-10 and Mig. These data suggest that IP-10 and Mig produced from the SS ductal epithelium stimulated by a certain factor from infiltrating lymphocytes attract activated T cells around the duct.

In our search for the stimulating factor for IP-10 and Mig from the infiltrating lymphocytes, we chose to study IFNγ because of its importance as an inducer of IP-10 and Mig (28) and because of its major involvement in the SS salivary gland lesion (4). In order to study the role of IFNγ on the expression of IP-10 and Mig, we used cultured epithelial cells from salivary glands. This culture system provided us with a very feasible method to study the function of salivary epithelial cells in vitro. By using this system, we clearly demonstrated that IFNγ induced the production of IP-10 and Mig mRNA and proteins in cultured SS epithelial cells. Moreover, we found that IFNγ is the only cytokine for the simultaneous induction of IP-10 and Mig mRNA in SS salivary epithelial cells. IFNγ also induced the expression of functional CXCR3 on freshly isolated SS peripheral blood lymphocytes.

Taken together, our findings indicate that IFNγ produced by infiltrating lymphocytes plays a major role in the development of cell infiltrates by inducing the production of IP-10 and Mig in the ductal epithelium and the expression of CXCR3 on T cells. Inhibition of the expression of other chemokines, such as IL-8 and eotaxin, by IFNγ (21, 34) might be the reason for the development of mononuclear cell infiltrates of T cell predominance in the SS salivary gland. This is the first report that shows a link between a cytokine and chemokines in the SS salivary gland.

SDF-1 mRNA, in contrast, was similarly expressed in SS and normal salivary glands. Our findings are consistent with those of a recent study showing the existence of SDF-1 in SS and normal salivary glands (35). In that study, Amft et al found by immunohistochemical analysis that SDF-1 was expressed more strongly on SS ductal epithelial cells compared with normal ductal epithelial cells. We could not find the presence of SDF-1 message in the SS salivary epithelial cells, although high levels of SDF-1 mRNA were shown in SS MSGs by RT-PCR. It is possible that the cells behave differently in the artificial environment than in vivo, such that epithelial cells lose their ability to produce SDF-1 during culture.

Interestingly, CXCR4+ lymphocytes showed a different distribution compared with CXCR3+ lymphocytes in the SS salivary gland. CXCR4+ lymphocytes were located in the area of less intensive infiltration around dense periductal foci, where CXCR3+ lymphocytes were located. SDF-1 is possibly involved in the recruitment of naive T cells and B cells from the circulation into tissues for immune surveillance (36). Our findings suggest that CXCR4+ naive T lymphocytes recruited from the circulation differentiate into CXCR3+ activated T lymphocytes that infiltrate around the duct in the SS salivary gland. Furthermore, SDF-1 has been shown to act as a costimulator for proliferation and cytokine production, such as IFNγ, in anti–CD3-activated CD4+ T cells (37) and has been shown to inhibit activation-induced apoptosis of T cells (35). SDF-1 may help promote and perpetuate chronic inflammation in the SS salivary gland.

Other chemokines, such as macrophage inflammatory protein 1β (CCL3), RANTES (CCL5), B cell–attracting chemokine 1 (CXCL13), Epstein-Barr virus–induced gene 1 ligand chemokine (CCL19), and pulmonary and activation-regulated chemokine (CCL18), have been identified in the SS salivary gland (35, 38, 39). It is no wonder that various chemokines are involved in autoimmune inflammatory lesions because differential regulation of chemokine expression by cytokines is common in these disorders (40, 41). For example, B cell–attracting chemokine 1 is an important chemokine in the formation of ectopic lymphoid follicles in the SS salivary gland (35). Temporal and spatial regulation of expression of chemokines should be clarified in order to know the exact mechanism of the development of cell infiltrates in the SS salivary gland.

In addition to the above, we considered two findings in this study. One finding is the spontaneous expression of IP-10 and Mig mRNA in the absence of exogenous IFNγ in some SS patients but not in normal controls. Spontaneous expression of IP-10 and Mig mRNA was shown in SS epithelial cells cultured for shorter periods. Cultured epithelial cells might lose some of their in vivo properties with time because of the change in environment. Another finding is the overproduction of IP-10 and Mig proteins from SS salivary epithelial cells, although SS and normal salivary epithelial cells expressed similar levels of IP-10 and Mig mRNA in response to IFNγ. The regulation of protein production by chemokines in epithelial cells is more complex than the expression of mRNA because the difference in culture periods could not explain the overproduction of IP-10 and Mig proteins from the SS salivary epithelial cells.

These findings suggest that SS salivary epithelial cells may be somewhat different from normal salivary epithelial cells even in a quiescent state. One possible difference is the cell activation state, because SS epithelial cells are capable of spontaneous cytokine production, HLA–DR expression, and expression of costimulatory molecules (42). Spontaneous mRNA expression and overproduction of IP-10 and Mig proteins could be additional evidence of an abnormal cell activation state of the SS epithelial cells.

Apoptotic cells produce various target molecules for autoantibody production in human autoimmune diseases (43, 44). Autoreactive T cells may initiate autoimmune responses against autoantigens, such as α-fodrin, presented by apoptotic salivary epithelial cells in SS (45, 46). A few autoreactive cells are considered to control a majority of nonspecific cells in inflammatory infiltrates in tissue-specific autoimmune diseases (47). One of the tools of autoreactive T cells that orchestrate many bystander cells may be chemokines, such as IP-10 and Mig, induced by IFNγ. In fact, RANTES was found to be increased in the serum and joints in a rat model of adjuvant-induced arthritis, and polyclonal antibody to RANTES was able to make the arthritis less severe (48). Treatment with anti-TNFα mAb was shown to reduce inflammatory cell migration into joints, and the blockade of IL-8 and monocyte chemoattractant protein 1 was found to be responsible for the antiinflammatory effects of the antibody (49). Interferon-inducible T cell α chemoattractant (I-TAC, or CXCL11) is another potent ligand for CXCR3 that is induced in a wide range of human tissues by IFNγ (50). I-TAC is expressed together with IP-10 and Mig in some epithelial cells, such as gastric epithelial cells (51) and bronchial epithelial cells (52). It is possible that I-TAC, along with IP-10 and Mig, is also involved in the SS salivary gland lesions. Due to the redundancy in ligands for CXCR3, it would be easier to block CXCR3 rather than its ligands to attenuate chronic inflammation in SS (53).

In conclusion, two IFNγ-induced chemokines, IP-10 and Mig, are involved in the development of cell infiltrates consisting of activated T cells that express CXCR3 in the salivary glands of patients with SS. A strategy for blocking chemokines or chemokine receptors could be a rational new approach for the treatment of SS in addition to cytokine-targeted therapy.

Acknowledgements

We thank Professor Ko-ichi Tomoda and Dr. Yuzo Shimode (Department of Otolaryngology, Kanazawa Medical University) for providing us with normal parotid gland tissue. We are grateful to Ms Naomi Nitta and Mr. Ken-ichi Yoshida for assistance with immunohistochemistry and flow cytometry and Ms Yoko Tokuno and Ms Rika Shodo for secretarial assistance.

Ancillary