Possible Function of Salivary Gland Epithelial Cells as ...1884 The Journal of Rheumatology 2002;...

1884 The Journal of Rheumatology 2002; 29:9

Possible Function of Salivary Gland Epithelial Cells asNonprofessional Antigen-Presenting Cells in theDevelopment of Sjögren's Syndrome SHIZUKA TSUNAWAKI, SEIJI NAKAMURA, YUKIKO OHYAMA, MASANORI SASAKI, AKIKO IKEBE-HIROKI,AKIMITSU HIRAKI, TSUTOMU KADENA, EIJI KAWAMURA, WATARU KUMAMARU, MASANORI SHINOHARA,and KANEMITSU SHIRASUNA

ABSTRACT. Objective. To explore the potential of salivary gland epithelial cells to act as nonprofessional anti-gen-presenting cells (APC) in the development of Sjögren's syndrome (SS). Methods. Expression of HLA-DR antigens, costimulatory molecules, and adhesion molecules onepithelial cells was immunohistochemically examined in labial salivary glands from patients withSS. An association with the expression of T cell derived cytokine messenger RNA (mRNA) wasobserved. The expression of these molecules was confirmed using cultured salivary gland epithelialcells. The ability of the salivary gland epithelial cells as nonprofessional APC was examined in amixed culture system using the salivary gland epithelial cells and allogeneic lymphocytes. Results. Expression of HLA-DR antigens, CD80, intercellular adhesion molecule-1 (ICAM-1),vascular cell adhesion molecule (VCAM), and E-selectin was immunohistochemically detected onduct cells from all patients; however, the expression of CD86 was limited to only some patients.Concomitant expression of CD80 on duct cells and Th1 cytokine mRNA, and CD86 on duct cellsand Th2 cytokine mRNA, was observed. Interferon-γ (IFN-γ) induced the cultured salivary glandepithelial cells to express HLA class I antigens, HLA-DR antigens, CD80, and ICAM-1, while tumor necrosis factor-α (TNF-α) induced the expression of HLA class I antigens, CD80, CD86, andVCAM. Cultured salivary gland epithelial cells treated with either IFN-γ or TNF-α also caused allo-geneic lymphocytes to proliferate. Conclusion. The ability of salivary gland epithelial cells to express HLA-DR antigens, costimula-tory molecules, and adhesion molecules and thus to act as nonprofessional APC was suggested.CD80 and CD86 expression of these cells was also suggested to be involved in the activation of Th1and Th2, respectively. (J Rheumatol 2002;29:1884-96)

Key Indexing Terms: SJÖGREN'S SYNDROME ANTIGEN-PRESENTING CELLS COSTIMULATORY MOLECULES ADHESION MOLECULES SALIVARY GLANDS EPITHELIAL CELLS

From the Department of Oral and Maxillofacial Surgery, Graduate Schoolof Dental Science, Kyushu University, Fukuoka; and the Department ofOral and Maxillofacial Surgery, School of Medicine, KumamotoUniversity, Kumamoto, Japan. Supported in part by grants from the Ministry of Education, Science, andCulture of Japan. S. Tsunawaki, DDS; S. Nakamura, DDS, PhD; Y. Ohyama, DDS, PhD; M. Sasaki, DDS, PhD; A. Ikebe-Hiroki, DDS, PhD; A. Hiraki, DDS; T. Kadena, DDS, PhD; E. Kawamura, DDS; W. Kumamaru, DDS; K. Shirasuna, DDS, PhD, Department of Oral and Maxillofacial Surgery,Graduate School of Dental Science, Kyushu University; M. Shinohara,DDS, PhD, Department of Oral and Maxillofacial Surgery, School ofMedicine, Kumamoto University.Address reprint requests to Dr. S. Nakamura, Department of Oral andMaxillofacial Surgery, Graduate School of Dental Science, KyushuUniversity, 3-1-1 Maidashi, Higashi-ku, Fukuoka 812-8582, Japan. E-mail: [email protected] Submitted February 16, 2001; revision accepted March 8, 2002.

Several lines of study have suggested that HLA-DR anti-gens are aberrantly expressed on target cells and presentantigenic peptides to T cells, and thus play an important role

in the immunopathogenesis of autoimmune diseases1.Recent studies have indicated that the ectopic expression ofHLA antigens on nonprofessional antigen-presenting cells(APC) by itself is not enough to induce primary T cellresponses; a second signal via costimulatory molecules isthus needed2. In brief, the CD28 pathway is regarded as anessential costimulus of T cell activation3; the known naturalligands for CD28 are CD80 and CD86, and the expressionof these ligands is usually limited to professional APC, suchas dendritic cells, macrophages, and activated B cells4,5.These costimulatory signals have also been reported to playa determining role in the development of the T cell subsetsTh1 and Th26,7. In addition, the involvement of adhesionmolecules, such as intercellular adhesion molecule-1(ICAM-1), vascular cell adhesion molecule (VCAM), andE-selectin, in the activation of T cells, has been reported8-10.In various autoimmune diseases1,11-15, aberrant expression ofthese costimulatory and adhesion molecules on target cellshas also been reported, suggesting the possible function of

www.jrheum.orgDownloaded on June 17, 2021 from

http://www.jrheum.org/

1885Tsunawaki, et al: APC in SS

these cells as nonprofessional APC in the activation of path-ogenic T cells. In Sjögren's syndrome (SS), we16 andothers17-20 have reported that HLA-DR antigens, adhesionmolecules, and costimulatory molecules were aberrantlyexpressed on salivary gland epithelial cells. Since SS hasbeen described as autoimmune epithelitis21, the aberrantactivation status of these cells may thus indicate that theyplay a central role in the immunopathogenesis of SS.

We investigated the potential of salivary gland epithelialcells to function as nonprofessional APC in the activation ofpathogenic T cells in SS. We first immunohistochemicallyexamined the expression of HLA-DR antigens, costimula-tory molecules, and adhesion molecules on epithelial cells inlabial salivary glands (LSG) from patients with SS. Then weexamined the relationship between the expression of costim-ulatory molecules on epithelial cells and T cell derivedcytokines in the LSG. Expression of HLA antigens, costim-ulatory molecules, and adhesion molecules was examinedusing cultured salivary gland epithelial cells. Finally, thecapacity of salivary gland epithelial cells to function as non-professional APC was examined using mixed cultures ofsalivary gland epithelial cells and allogeneic lymphocytes.

MATERIALS AND METHODS Patients. Forty patients with SS referred to the Department of Oral andMaxillofacial Surgery, Kyushu University DentalHospital, were studied.All patients were women between the ages of 18 and 78 years (mean age57.5 yrs). All fulfilled the diagnostic criteria for definite SS proposed by theResearch Committee on Sjögren's Syndrome of the Ministry of Health andWelfare of the Japanese Government22, and the diagnosis was also based onthe diagnostic criteria proposed by the European Community Study Groupon Diagnostic Criteria for Sjögren's Syndrome23. Each patient exhibitedobjective evidence of salivary gland involvement based on the presence ofsubjective xerostomia and a decreased salivary flow rate, abnormal find-ings on parotid sialography, and focal lymphocytic infiltrates in the LSG.The histologic findings in the LSG were graded based on the Chisholm andMason scale24. Twenty patients had primary SS with no other autoimmunedisease, while the remaining 20 patients had secondary SS with otherautoimmune diseases. As controls, LSG biopsy specimens were alsoobtained from 5 patients with mucoceles who had no clinical or laboratoryevidence of systemic autoimmune disease. These LSG were all histologi-cally normal. Cultures of salivary gland epithelial cells. Primary cultures of salivarygland epithelial cells were established from the LSG, parotid glands, orsubmandibular glands as described25. The parotid and submandibularglands were obtained during total neck dissection from patients with squa-mous cell carcinoma of the oral cavity who had neither radiotherapy norchemotherapy before the surgical treatment, and were all histologicallynormal. Parenchymal tissue specimens of the salivary glands were mincedwith scissors into 1 mm fragments and placed on a 60 mm plastic dish(Becton Dickinson, Mountain View, CA, USA). The basal nutrient mediumwas serum-free, and was a 1:9 mixture of Dulbecco's modified Eagle'smedium (Sigma, St. Louis, MO, USA) and MCDB 153 (Kyokuto, Tokyo,Japan) (Ca++ concentration 0.227 mM). The complete medium was a basalnutrient medium supplemented with 25 mM N-2-hydroxy-ethylpiperazine-N'-2-ethanesulfolic acid (Wako Pure Chemical, Osaka, Japan), 2.0 g/lNaHCO3 (pH 7.4), 1 µg/ml insulin (Sigma), 10 mM dexamethasone(Sigma), and 10 ng/ml human epidermal growth factor (Genzyme,Cambridge, MA, USA). The fragments were cultured at 37°C in a 5% CO2atmosphere. After fragments were removed from the dish, only prolifer-

ating adhered cells were passaged using 0.5% trypsin in phosphate bufferedsaline (PBS) containing 5.2 nM EDTA. To stimulate salivary gland epithe-lial cells with interferon-γ (IFN-γ) or tumor necrosis factor-α (TNF-α), thecultured cells were incubated in the presence of 50 to 5000 U/ml of recom-binant IFN-γ (supplied by Shionogi Co., Osaka, Japan) or 1 to 100 ng/ml ofrecombinant TNF-α (supplied by Dainippon Pharmaceutical Co., Tokyo,Japan) at 37°C for 6 to 72 h in the same way. Immunohistochemical and immunocytochemical analyses. Staining wasperformed by the conventional avidin-biotin complex method16,26,27. Forimmunohistochemical studies, the LSG were immediately placed in anembedding medium (OCT compound; Miles, Elkhart, IN, USA) andfrozen, and thereafter the frozen sections were prepared. Quantification ofstained lymphocytes was performed by counting the positive cells alongwith the total number of mononuclear cells in 4 mm2 sections from 3different areas. To count the total number of mononuclear cells, serialsections stained with hematoxylin and eosin were used. The percentage ofstained cells was then calculated as the number of stainedlymphocytes/total number of mononuclear cells. The expression of HLA-DR antigens, CD80, CD86, ICAM-1, VCAM-1, and E-selectin on the sali-vary gland epithelial cells was quantified by counting the percentage ofpositive cells per duct in 3 different 4 mm2 areas. A cell was considered tobe positive when it stained more intensely than those stained with isotypematched monoclonal antibodies (Mab).

They were scored as follows: – = none, ± = less than 10%, + = 10–50%,and ++ = more than 50%. Serial sections stained with the anti-cytokeratinantibodies KL-1 and K8.12 were also examined in order to differentiateduct cells from endothelial cells, since cytokeratins are expressed on ductcells not on endothelial cells. Immunocytochemical studies of cultured cellswere carried out on glass coverslides. The cells were considered positivewhen they stained more intensely than those stained with the isotypematched Mab. The Mab used were T3-II (anti-CD3, IgG1; supplied by Dr.S.M. Fu, University of Virginia, USA), B1 (anti-CD20, IgG1; Dr. S.M. Fu),L293 (anti-CD28, IgG1; Becton Dickinson), B9.12.1 (anti-HLA class Iantigens, IgG2a; Cosmo Bio, Tokyo, Japan), OKDR (anti-HLA-DR anti-gens, IgG1; Becton Dickinson), G46-6 (anti-HLA-DR antigens, IgG2a;Becton Dickinson), BB1 (anti-CD80, IgM; Ancell, Bayport, MN, USA),L307.4 (anti-CD80, IgG1; Becton Dickinson), BU63 (anti-CD86, IgG1;Ancell), IT2.2 (anti-CD86, IgG2b; PharMingen, San Diego, CA, USA),BBIG-I1 (anti-ICAM-1, IgG1; R&D Systems, Minneapolis, MN, USA),BBIG-V1 (anti-VCAM-1, IgG1; R&D Systems), BBIG-E6 (anti-E-selectin, IgG1; R&D Systems), KL-1 (anti-cytokeratin, IgG1; ImmunotechSA, Marseille, France), and K8.12 (anti-cytokeratin, IgG1; BioMakor,Rehovot, Israel). As control Mab we used HDP-1 (anti-DNP, IgG1), SS1(anti-sheep erythrocyte, IgG2a), NS8.1 (anti-sheep erythrocyte, IgG2b),and NS4.1 (anti-sheep erythrocyte, IgM), all of which were generous giftsfrom Dr. S.M. Fu. Flow cytometric analysis. Cultured salivary gland epithelial cells wereharvested using 0.5% trypsin in PBS containing 5.2 nM EDTA, and wereanalyzed by flow cytometry as described28. Briefly, the cells were firststained with Mab, washed, and then further incubated with FITC conjugat-ed F(ab')2 goat anti-mouse Ig antibodies. After further washing, 104 cellswere analyzed with a FACSort (Becton-Dickinson). The Mab used were thesame as those used in the immunohistochemical and immunocytochemicalanalyses. RNA extraction and complementary DNA (cDNA) synthesis. Total RNAfrom the LSG and cultured salivary gland epithelial cells was prepared bythe acid guanidinium-phenol-chloroform method26,27. Three micrograms ofthe total RNA preparation were used for the synthesis of cDNA. Briefly,RNA was incubated 1 h at 37°C with 20 units of RNasin in ribonucleaseinhibitor (Promega, Madison, WI, USA), 0.5 µg of oligo-(dT)12-18(Pharmacia, Uppsala, Sweden), 0.5 mM of each dNTP (Pharmacia), 10 mMdithiothreitol, and 100 units RNase H– reverse transciptase (BRL,Gaithersburg, MD, USA). Polymerase chain reaction amplification of cDNA. For the cytokines,




amplification was performed as reported27. Briefly, the PCR reactionmixture consisted of 10 mM Tris-HCl (pH 9.0), 50 mM KCl, 0.1% Triton X-100, 2.5 mM MgCl2, 0.2 mM dNTP (Pharmacia), 400 nM 5' and 3'oligonucleotide primers, and 2.5 units Taq DNA polymerase (Perkin-ElmerCetus, Emeryville, CA, USA). Amplification was performed with a DNAthermal cycler (Perkin-Elmer Cetus) under the following conditions: dena-turing at 94°C for 5 min for the first cycle and 40 s for the subsequentcycles, and annealing/extension at 55°C [for interleukin (IL) 2, IFN-γ,TNF-α] or 65°C (for ß-actin, CD3d, IL-4, and IL-5) for 90 s. The PCRproducts were electrophoresed through 1.8% agarose gel, transferred to aNytran nylon membrane (Schleicher & Schuell, Dassel, Germany), andthen hybridized with 32P labeled oligonucleotide internal probes. Afterhybridization for 18 h at 50°C, the filters were washed in 2× saline-sodiumcitrate with 1% sodium dodecyl sulfate at 50°C and then exposed to x-rayfilm. To quantify the radioactivity of the hybridized bands, autoradiogramswere generated using a Fuji BA100 Bioimage analyzer (Fuji Photo Film,Tokyo, Japan). The sequences of the specific oligonucleotide primers andinternal probes have all been described27. For HLA-DR antigens andcostimulatory and adhesion molecules, the sequences of the specificoligonucleotide primers were based on those previously published19,20,29,30,and the amplification was performed under the same conditions as thoseused for the cytokines with slight modifications, as shown in Table 1. PCRproducts were electrophoresed through 1.8% agarose gel and visualized bystaining with ethidium bromide. Estimation of messenger RNA expression. To provide meaningful compar-ison between different individuals or samples, we first normalized theamount of cDNA in each sample to yield equivalent amounts of CD3δ PCRproducts for standardization of T cell mRNA or ß-actin PCR products forstandardization of total cellular mRNA, as described27. Briefly, serial 5-folddilutions of sample cDNA were amplified using a set of CD3δ or ß-actinprimers. After 27, 30, and 33 cycle amplifications, the PCR products wereelectrophoresed, transferred, and hybridized, and the radioactivities of theproducts were measured using a Fuji BA100 Bioimage. From these results,the amounts of cDNA were then normalized to yield equivalent amounts ofCD3δ or ß-actin PCR products between samples. For the T cell derivedcytokines, the sample cDNA were normalized to yield equivalent amountsof CD3δ PCR product, serially diluted 5-fold, and then amplified for 33cycles using each set of cytokine-specific primers. After hybridization witha 32P labeled oligonucleotide internal probe, the amount of cytokine PCRproducts was graded as follows: – = no band detected, even in the undilutedsample cDNA; + = band detected in the undiluted sample cDNA but not inthe 1:5 diluted cDNA; ++ = band detected in the 1:5 diluted cDNA but notin the 1:25 diluted cDNA; +++ = band detected in the 1:25 diluted cDNA.For HLA-DR antigens and costimulatory and adhesion molecules, thesample cDNA were normalized to yield equivalentamounts of ß-actin PCRproduct, amplified for 35 cycles using each set of specific primers, elec-trophoresed, and visualized by staining with ethidium bromide.

Mixed culture of salivary gland epithelial cells and allogeneic lymphocytes.Cultured salivary gland epithelial cells (4 × 104 cells/well) were cultured intriplicate in the presence of 2000 U/ml of IFN-γ and/or 20 ng/ml of TNF-αin flat bottom 96 well plates at 37°C for 3 days in a 5% CO2 atmosphere.The cells were then irradiated with 10,000 rad, washed, and incubated with2 × 105 cells/well of allogeneic peripheral blood mononuclear cells(PBMC) in RPMI 1640 medium supplemented with 2.5% human AB serumat 37°C for 5 days in a 5% CO2 atmosphere. During the last 8 h, the cellswere pulsed with 1 µCi/well of [3H]thymidine and the incorporatedradioactivity was determined.

Blocking of HLA antigens and costimulatory molecules on the salivarygland epithelial cells was performed with Mab. Irradiated cells (1 × 106cells/ml) were incubated with different Mab (anti-HLA class I, anti-HLA-DR, anti-CD80, and anti-CD86) at a final concentration of 5 µg/ml at 4°Cfor 1 h, washed, and then incubated with PBMC as described above.

RESULTS Expression of costimulatory molecules on salivary glandepithelial cells in LSG. The expression of HLA-DR anti-gens, costimulatory molecules, and adhesion molecules onsalivary gland epithelial cells was immunohistochemicallyexamined in the LSG from the 40 patients with SS, and rep-resentative findings are shown in Figure 1. Expression ofHLA-DR antigens was detected in all the SS patients and itwas also frequently observed on acinar and duct cells inclose association with high lymphocytic infiltration. Theexpression was stronger on intercalated duct cells than onintralobular and interlobular duct cells. Expression of CD80was detected in all SS patients and observed on duct cells,especially intercalated and intralobular duct cells, but not onacinar cells. The expression was generally associated withheavy lymphocytic infiltration; however, this associationwas not absolutely observed. In 5 cases, the expression ofCD80 was weaker in areas with heavy lymphocytic infiltra-tion than in areas without lymphocytic infiltration. Theexpression of CD86 was detected in 13 of the patients, andeven in those patients it was limited to a small number ofduct cells with heavy lymphocytic infiltration. Expressionof ICAM-1, VCAM, and E-selectin was observed on ductcells, especially intercalated duct cells associated withheavy lymphocytic infiltration, in 40, 27, and 36 patients,

Table 1. Primers and conditions for PCR amplification.

Target Messenger RNA Sequence of Primer PCR Conditions temperature (˚C), time (s)(size of amplified fragment) (5′-3′) Denaturing Annealing Extension

HLA-DR antigens Sense CCCCACAGCACGTTTCTTG 94, 30 56, 60 72, 60(274 bp) Antisense CCGCTGCACTGTGAAGCTCT

CD80 Sense AACTCGCATCTACTGGCAAAAGGAGAA 94, 60 60, 60 72, 90(401 bp) Antisense TTCAGGATCTTGGGAAACTGTTGTGTT

CD86 Sense GTATTTTGGCAGGACCAGGA 94, 60 55, 120 72, 180(663 bp) Antisense GCCGCTTCTTCTTGTTCCAT

ICAM-1 Sense TGACCATCTACAGCTTTCCGGC 94, 30 56, 30 72, 60(351 BP) Antisense AGCCTGGCACATTGGAGTCTG

VCAM Sense CCAAGAATACAGTTATTTCTGTG 94, 30 55, 30 72, 30(270 BP) Antisense TAGGGAATGCTTGAACAATTAATTC




Figure 1. Immunohistochemical staining of HLA-DR antigens, costimulatory molecules, and adhesion molecules on duct cells from labial salivary glands(LSG) from (A) a patient with SS and (B) a healthy control subject. Serial sections were stained with anti-HLA-DR antigens (OKDR), anti-CD80 (BB1), anti-CD86 (BU63), anti-ICAM-1 (BBIG-I1), anti-VCAM (BBIG-V1), anti-E-selectin (BBIG-E6), and control antibodies (a mixture of HDP-1, SS1, NS8.1, andNS4.1). Expression of these molecules was seen on duct cells surrounded by lymphocytic infiltration in the LSG from a patient with SS, but not in controlsamples. LSG from 40 SS patients and 5 controls were examined; results of a representative case are shown here.




respectively, but not on acinar cells. In normal salivaryglands, none of these molecules were expressed on ducts.Two Mab for each of HLA-DR antigens CD80 and CD86were used in all staining experiments. As the 2 Mab consis-tently gave almost identical results, only the results of eachantibody are reported. Relationship between expression of costimulatory moleculeson duct cells and T cell derived cytokine mRNA in LSG fromSS patients. As costimulatory signals have been suggested toplay a determining role in the development of T cellsubsets6,7, the relationship between the expression ofcostimulatory molecules on duct cells and cytokine mRNAin the LSG was examined in 17 of the 40 SS patients (Table2). As reported27,31, Th1 cytokines such as IL-2 and IFN-γwere detected in most of the patients (IL-2 in 14 patients andIFN-γ in 15). In contrast, Th2 cytokines such as IL-4 and IL-5 were less frequently detected (IL-4 in 8 patients and IL-5in 7) than Th1 cytokines, and the expression of Th2cytokines was closely associated with B cell accumulationin the LSG. Expression of CD80 on duct cells was observedin all the patients, while expression of CD86 on duct cellswas found in 6 patients. Concomitant detection of IL-4 andIL-5 was more frequent in the patients with CD86 expres-sion on duct cells (5 of the 6 patients) than in those patientswithout CD86 expression on duct cells (2 of the 11 patients),and this association was statistically significant (p < 0.05,chi-square test). Expression of HLA antigens, costimulatory molecules, andadhesion molecules on cultured salivary gland epithelialcells. To confirm the expression of HLA-DR antigens,costimulatory molecules, and adhesion molecules on sali-

vary gland epithelial cells, epithelial cells of the LSG werecultured and passaged 3 times. At this time, this culture wasusually contaminated with lymphocytes. As shown in Figure2, the cultured cytokeratin positive epithelial cells from SSpatients were spindle shaped and expressed HLA class Iantigens, HLA-DR antigens, CD80, and ICAM-1, but notCD86, VCAM, and E-selectin. Expression of HLA-DR anti-gens and ICAM-1 gradually diminished after further pas-sages, while the expression of CD80 decreased but wasmaintained for up to 12 serial passages (data not shown). Incontrast, the cultured cytokeratin positive epithelial cells ofnormal LSG were cuboidal shaped and expressed HLA classI antigens but not HLA-DR antigens, CD80, CD86, ICAM-1, VCAM, or E-selectin (data not shown).

The effects of IFN-γ and TNF-α on the induction ofHLA-DR antigens, costimulatory molecules, and adhesionmolecules were then examined, because IFN-γ and TNF-αhave been reported to induce various molecules to expresson vascular and parenchymal epithelial cells32-34 and to beconsistently produced in the LSG from patients with SS27,31.In particular, IFN-γ has been suggested to induce HLA-DRantigens, CD80, and CD86 on salivary gland epithelial-cells20,35,36. To obtain a large number of cells, salivary glandepithelial cells were cultured from normal major salivaryglands and then were passaged more than 5 times: almost allthe cultured cells were cytokeratin positive. As shown inFigure 3, except for the HLA class I antigens, none of theHLA-DR antigens, costimulatory molecules, or adhesionmolecules was expressed on the control cells cultured withmedium alone. When the cells were treated with IFN-γ for 3days, HLA-DR antigens, CD80, and ICAM-1 were appar-

Table 2. Expression of costimulatory molecules on ductal cells and cytokines in labial salivary glands from patients with SS.

Patient Lymphocyte Lymphocyte Subset %**, Expression of Costimulatory Molecules† Expression of Cytokines††Infiltration * CD3 CD20 CD80 CD86 IL-2 IFN−γ IL-4 IL-5

1 4 53.6 42.1 ++ + ++ +++ + +2 3 49.3 45.2 + + + +++ + +3 4 51.2 46.8 + + + +++ + +4 4 47.4 49.6 + + _ +++ + +5 4 48.6 48.3 + + _ +++ + +6 3 78.1 14.3 + + ++ +++ – –7 3 54.3 45.6 + – +++ +++ + +8 3 57.6 40.6 + – ++ +++ + +9 4 65.3 28.5 ++ – ++ +++ + –10 4 57.4 40.6 ++ – ++ +++ – –11 3 78.2 19.8 + – + +++ – –12 4 78.3 20.8 + – + +++ – ––13 2 76.2 23.6 + – + +++ – –14 3 78.2 20.3 + – + ++ – –15 4 86.6 4.2 + – +++ – – –16 4 73.1 23.8 + – + – – –17 2 72.6 14.3 + – – ++++ – –

* Degree of lymphocytic infiltration was graded on the Chisholm and Mason scale23. ** Percentage of stained lymphocytes was calculated as described inMaterials and Methods. † Degree of CD80 and CD86 expression on duct cells was graded as described in Materials and Methods. †† Degree of cytokinemessenger RNA expression was graded as described in Materials and Methods.




ently expressed and the expression of HLA class I antigenswas thus augmented. The expression of CD86, VCAM, andE-selectin was scarcely detected after treatment with IFN-γ.On the other hand, when the cells were treated with TNF-αfor 3 days, expression of CD80, CD86, ICAM-1, andVCAM was induced, although CD80 and CD86 wereweakly expressed. Expression of HLA-DR antigens and E-selectin was not induced and the expression of HLA class Iantigens was slightly augmented. The effects of IFN-γ andTNF-α were observed in a dose dependent manner aftertreatment with 50, 100, 500, 1000, and 5000 U/ml of IFN-γor 1, 5, 10, 50, and 100 ng/ml TNF-α, and these effects weredetectable 24 h after the treatment, reached their peak 3 to 5days after treatment, and thereafter were maintained for upto 7 days (data not shown). Further, synergistic effects ofIFN-γ and TNF-α were not observed (data not shown). It isinteresting that the treatment with IFN-γ and TNF-αchanged cytokeratin positive cuboid cells to spindle shapedcells, which were similar to cultured epithelial cells of LSGfrom SS patients in Figure 2.

Expression of HLA antigens, costimulatory molecules,and adhesion molecules on the surface of the culturedepithelial cells was examined by flow cytometry, as shownin Figure 4. When the cells were treated with IFN-γ for 3days, HLA-DR antigens and ICAM-1 were stronglyexpressed on the cell surface and the expression of HLAclass I antigens was apparently augmented. The expressionof CD80 was weakly but reproducibly induced (thepercentage of positive cells was 22.0 ± 6.8% in 6 experi-ments). The expression of CD86, VCAM, and E-selectinwas scarce after treatment with IFN-γ. On the other hand,treatment with TNF-α for 3 days, as well as that with IFN-γ, induced the expression of ICAM-1. The expression ofHLA class I antigens was augmented, although the effects ofTNF-α were weaker than those of IFN-γ. The expression ofCD80, CD86, and VCAM was faint but reproduciblyinduced (percentage of positive cells 8.4 ± 3.1%, 12.8 ±3.8%, and 16.3 ± 5.7%, respectively, in 6 experiments). Theexpression of HLA-DR antigens and E-selectin was notdetectable.

Figure 2. Immunocytochemical staining of HLA antigens, costimulatory molecules, and adhesion molecules oncultured epithelial cells of LSG from a patient with SS. Epithelial cells of LSG were cultured, passaged 3 times,and stained with anti-HLA class I antigens (B9.12.1), anti-cytokeratins (KL-1), and the same Mab as those usedin samples shown in Figure 1. The cultured cytokeratin positive epithelial cells were spindle shaped and positivefor HLA class I antigens, HLA-DR antigens, CD80, ICAM-1. Epithelial cells from 3 SS patients were examinedby the same protocol and the results of a representative case are shown.




Figure 3. Effects of IFN-γ and TNF-α on the expression of HLA antigens, costimulatory molecules, and adhesion molecules on cultured salivary gland epithe-lial cells. Epithelial cells were cultured from a normal major salivary gland, passaged 6 times, treated with 2000 U/ml IFN-γ or 20 ng/ml TNF-α for 3 days,and then immunocytochemically stained with the same Mab as those used in Figure 2. Compared with nontreated cells, IFN-γ stimulated these cells to expressHLA-DR antigens, CD80, and ICAM-1, while expression of HLA class I antigens was also augmented. TNF-α stimulated these cells to express CD80, CD86,ICAM-1, and VCAM, although expression of CD80 and CD86 was weak. Expression of HLA class I antigens was slightly augmented. Epithelial cells ofmajor salivary glands from 4 different donors were examined by the same protocol and the results of a representative case are shown.

Figure 4. Effects of IFN-γ and TNF-α on the surface expression of HLA antigens, costimulatory molecules, and adhesion molecules on cultured salivary glandepithelial cells. Epithelial cells were cultured from a normal major salivary gland, passaged 6 times, treated with 2000 U/ml IFN-γ or 20 ng/ml TNF-α for 3days, stained by indirect immunofluorescence with the same Mab as those used in Figure 2, and then were analyzed by flow cytometry. As negative controls,the results of nontreated cells stained with isotype matched control antibodies (HDP-1 or NS4.1) are shown. IFN-γ or TNF-α stimulated cells were similarlystained with these isotype matched control antibodies (data not shown). HLA class I antigens were only positive on nontreated cells in comparison with neg-ative controls. Compared with nontreated cells, IFN-g stimulated these cells to express HLA-DR antigens, CD80, and ICAM-1, and expression of HLA classI antigens was augmented. TNF-α stimulated the cells to express CD80, CD86, ICAM-1, and VCAM, although expression of CD80, CD86, and VCAM wasweak. Expression of HLA class I antigens was augmented. Expression of E-selectin was not induced with either IFN-γ or TNF-α (data not shown). Epithelialcells of major salivary glands from 6 different donors were examined by the same protocol and the results of a representative case are shown.




The activation of cultured epithelial cells with IFN-γ andTNF-α was further examined at the transcript level, asshown in Figure 5. Treatment with IFN-γ apparentlyinduced mRNA expression of HLA-DR antigens andICAM-1, 12 to 48 h and 6 to 24 h, respectively, after thetreatment. In contrast, treatment with TNF-α inducedmRNA expression of HLA-DR antigens, ICAM-1, andVCAM 48 h, 6 to 24 h, and 6 to 48 h, respectively, aftertreatment. Expression of CD80 mRNA was induced bytreatment with either IFN-γ or TNF-α, and it was very weakbut reproducible. The expression of CD86 mRNA wasdetectable without treatment, and reproducible effects ofIFN-γ and TNF-α were not obtained. Induction of allogeneic lymphocyte reaction with culturedsalivary gland epithelial cells. As HLA antigens, costimula-

tory molecules, and adhesion molecules are involved in Tcell activation to alloantigens37-39, we examined the abilityof salivary gland epithelial cells to stimulate alloreactive Tcells in a mixed culture of salivary gland epithelial cells andallogeneic lymphocytes. Figure 6 illustrates representativedata obtained with PBMC from 2 healthy donors. As deter-mined by the uptake of [3H]thymidine, nontreated epithelialcells induced very low levels of proliferation of the allo-geneic PBMC. In contrast, epithelial cells treated with eitherIFN-γ or TNF-α significantly stimulated the proliferation ofallogeneic PBMC. The effects of IFN-γ were alwaysstronger than those of TNF-α. However, the proliferationachieved after treatment with IFN-γ or TNF-α was stilllower than that in the conventional mixed lymphocyte reac-tion.

Figure 5. Effects of IFN-γ and TNF-α on mRNA expression of HLA antigens, costimulatory molecules, andadhesion molecules on the cultured salivary gland epithelial cells. Epithelial cells were cultured from a normalmajor salivary gland, passaged 5 times, and treated with 2000 U/ml IFN-γ or 20 ng/ml TNF-α for 6, 12, 24, or48 h. Total RNA was prepared from the epithelial cells and used for synthesis of cDNA. The sample cDNA werenormalized to yield equivalent amount of ß-actin PCR products, amplified for 35 cycles using each set of spe-cific primers, electrophoresed, and visualized by staining with ethidium bromide. IFN-γ induced the mRNAexpression of HLA-DR antigens and ICAM-1, while TNF-α induced mRNA expression of ICAM-1 and VCAM.Expression of CD80 mRNA was induced by either IFN-γ or TNF-α, but it was very weak. Expression of CD86mRNA was detectable with no treatment, and no reproducible effects of IFN-γ and TNF-α were obtained.Epithelial cells of major salivary glands from 6 different donors were examined by the same protocol and theresults of a representative case are shown.




As shown in Figure 7, blocking of HLA antigens andcostimulatory molecules on the salivary gland epithelialcells was performed with Mab. The IFN-γ induced prolifer-ation of the allogeneic PBMC was inhibited by the antibodyblocking of HLA class I antigens, HLA-DR antigens, andCD80. In contrast, the effects of TNF-α were inhibited bythe antibody blocking of HLA class I antigens and CD80,but not by HLA-DR antigens. Antibody blocking of CD86did not result in significant inhibition of either IFN-γ orTNF-α induced proliferation of allogeneic PBMC.

DISCUSSION Aberrant expression of HLA-DR antigens and adhesionmolecules on salivary gland epithelial cells was immunohis-tochemically observed, as in previous reports16-20. In regardto CD80 and CD86, ectopic expression on acinar and ductcells has recently been reported by Manoussakis, et al20;however, there are some differences in our results. In theirreport, both CD80 and CD86 were frequently expressed onacinar and duct cells. In contrast, we observed the expres-sion of CD80 and CD86 only on duct cells, especially inter-

calated and intralobular duct cells. Furthermore, CD80 wasexpressed in samples from all of the patients, while CD86was only expressed in some patients. Regarding infiltratinglymphocytes, both CD80 and CD86 were observed byManoussakis, et al20, while only CD86 was detected in ourstudy (data not shown). The discrepancies in these resultsmay be due to differences in either the antibodies used or thestaining procedures: paraffin embedded specimens wereused by Manoussakis, et al20 and frozen specimens in ourstudy.

Some reports suggest that CD80 and CD86 play a deter-mining role in the development of the T cell subsets Th1 andTh2; however, there are conflicting results regarding the roleof these molecules6,7. The discovery of the CD28 homo-logue, cytotoxic T lymphocyte associated antigen-4 (CTLA-4), has increased the complexity of costimulatoryinteractions. CTLA-4 has been reported to bind with higheraffinity to CD80 and CD86 and thereby induce an inhibitoryeffect on T cell activation40. Thus regulation of Th1/Th2balance is relatively complicated, and involvement ofvarious other molecules including antigenic peptides,

Figure 6. Effects of IFN-γ and TNF-α on the induction of allogeneic response in mixed cultures of cultured salivary gland epithelial cells and allogeneic lym-phocytes. Epithelial cells were cultured from a normal major salivary gland, passaged 8 times, and treated in triplicate with 2000 U/ml IFN-γ or 20 ng/mlTNF-α in 96 well plates for 3 days. Cells were then irradiated, washed, and incubated with allogeneic PBMC from 2 healthy donors for 5 days. Epithelialcells were thus used as stimulator cells, while PBMC were used as responder cells. For conventional mixed lymphocyte reactions, irradiated PBMC from adonor of the epithelial cells were used as stimulator cells. During the last 8 h, cells were pulsed with [3H]thymidine and the incorporated radioactivity wasdetermined. The SD of each mean radioactivity were within 15%. Three sets of experiments were performed by using the salivary gland epithelial cells from3 different donors and the results of a representative case are shown. In all 3 experiments, IFN-γ or TNF-α stimulated epithelial cells achieved significantlyhigher levels of proliferation compared with nontreated epithelial cells (p < 0.05, Student t test). As epithelial cells were extensively washed after treatmentwith IFN-γ and TNF-α, the remaining IFN-γ and TNF-α did not significantly affect the proliferation of the PBMC (data not shown).




cytokines, and chemokines has been reported41-44. Similarto reports on other autoimmune diseases6,14,15,45-50, the consis-tent expression of CD80 as well as HLA-DR antigens on theduct cells that we observed suggests that the costimulationpathway via CD80 plays an important role in the activationof pathogenic T cells in SS. The concomitant expression ofCD80 on duct cells and Th1 cytokine mRNA in LSG wasalso observed in this study, suggesting that the costimulationpathway via CD80 might be involved in Th1 activation. Incontrast, Th2 cytokine mRNA was frequently detected inLSG from patients with CD86 expression on ductal cells,suggesting that the costimulation pathway via CD86 mightbe involved in Th2 activation. We previously described amodel for the pathogenesis of SS27. Th1 activation plays akey role in the induction and/or maintenance of the disease,giving rise to the eventual destruction of the target organ.Additional Th2 activation induces B cells to differentiate, toproliferate, and to produce immunoglobulins, and is thusconsidered to be responsible for the lymphoaggressivenessof the disease. As a result, CD80 may play an essential rolein the initiation and/or maintenance of the disease, whileCD86 is active in the progression of the disease. However,in a study of autoimmune exocrinopathy resembling SS inNFS/sld mutant mice, CD86, rather than CD80, has beensuggested to play a crucial role in the initiation and subse-quent progression of Th1 mediated autoimmunity51. Further

studies, for example on the in situ association ofCD80/CD86 expression on duct cells and Th1/Th2 activa-tion, are needed to determine the exact role of CD80 andCD86 in the development of SS.

Cultured epithelioid cells were used in this study to con-firm the expression of HLA-DR antigens, costimulatorymolecules, and adhesion molecules on salivary glandepithelial cells, especially ductal cells, which were observedin situ, and to examine the effects of cytokines on theexpression of these molecules. Our previous study25 indi-cates that epithelioid cells, which grew in the same serum-free medium as in the present study, represent the undiffer-entiated ultrastructure of ductal cells, express antigens spe-cific to basal cells of the intra and interlobular ducts in situ,and differentiate into luminal duct cells after a confluent cul-ture. We thus consider that the cultured epithelioid cellsused in this study are suitable as the target cells in SS. Clark,et al35 and Brookes, et al36 examined the expression ofHLA-DR antigens on cultured epithelial cells from salivaryglands. Manoussakis, et al20 also used these cells to confirmthe expression of HLA-DR antigens, CD80, and CD86.However, the epithelial cells they used were obtained underculture conditions different from ours, and they were alsonot fully characterized.

Clark, et al35, Brookes, et al36, and Manoussakis, et al20immunocytochemically examined the effects of IFN-γ on

Figure 7. The effect of antibody blocking of HLA antigens and costimulatory molecules on IFN-γ and TNF-α induced allogeneic response in mixed culturesof salivary gland epithelial cells and allogeneic lymphocytes. Epithelial cells were treated with IFN-γ or TNF-α and irradiated as described in legend forFigure 6. Cells were then incubated with different Mab (anti-HLA class I, B9.12.1; anti-HLA-DR, G46-6; anti-CD80, L307.4; and anti-CD86, BU63; con-trols, SS1 and HDP-1) at a final concentration of 5 µg/ml at 4˚C for 1 h, washed, and incubated with allogeneic PBMC from a healthy donor for 5 days.During the last 8 h, the cells were pulsed with [3H]thymidine and the incorporated radioactivity was determined. The SD of each mean radioactivity werewithin 15%. Four sets of experiments were performed using salivary gland epithelial cells from 4 different donors, and the results of a representative case areshown. *p < 0.05 relative to controls, Student t test.




the expression of HLA-DR antigens and found resultsalmost identical to ours. Further, in this study, induction ofthe surface expression of HLA antigens on salivary glandepithelial cells was confirmed by flow cytometry. At themRNA level we examined by PCR, the expression of HLA-DR antigens was inducible and the results were compatibleto those achieved with immunocytochemical staining andflow cytometry. The effects of TNF-α were also examinedin this study. Under the conditions we examined, TNF-α didnot induce the expression of HLA-DR antigens at eitherprotein or mRNA levels. It is important to note thataugmented expression of HLA class I antigens by eitherIFN-γ or TNF-α was observed by immunocytochemicalstaining and flow cytometry.

Regarding CD80 and CD86, we obtained different resultsfrom those reported by Manoussakis, et al20. At the proteinand mRNA levels, Manoussakis, et al observed that IFN-γinduced both CD80 and CD86 expression, while only CD80was expressed in our study. These discrepancies may be dueto differences in the cultured epithelial cells. The surfaceexpression, observed only in this study, and the mRNAexpression of CD80 were relatively weak, compared withthe results from immunocytochemical staining. AlthoughIFN-γ might not fully induce the expression of CD80, suchinduction was confirmed at the protein and mRNA levels.TNF-α, examined only in this study, weakly induced CD80and CD86 expression in immunocytochemical staining, andonly a faint induction result was observed by flow cyto-metry. Induction of CD80 was confirmed at the mRNAlevel, but the induction of CD86 was not fully confirmed,most likely because the induction was too weak to be repro-ducibly observed by the PCR based analysis we performed.As a result, full expression of CD80 and CD86 was notachieved by stimulation with IFN-γ and TNF-α. Based onthe strong staining of CD80 and CD86 in situ, either IFN-γor TNF-α may not be sufficient and additional stimulatoryconditions may thus be needed to induce full expression,especially the surface expression of these molecules. Inaddition, it is interesting that the spontaneous and sustainedCD80 expression by longterm cultured epithelial cells fromSS patients, as reported by Manoussakis, et al20, was alsoobserved in our study. As Manoussakis, et al discussed, theaberrant CD80 expression by salivary gland epithelial cellsin SS might be partially constitutive or induced in anautocrine manner by cytokines produced by them, in addi-tion to upregulation by such T cell derived cytokines as IFN-γ.

In this study, the expression of adhesion molecules onsalivary gland epithelial cells was also confirmed using cul-tured salivary gland epithelial cells. IFN-γ activated thesecells to express ICAM-1, while TNF-α activated expressionof ICAM-1 and VCAM. The effects of IFN-γ and TNF-α onthe expression of adhesion molecules were confirmed byflow cytometry and at the mRNA level. Neither IFN-γ nor

TNF-α induced the expression of E-selectin. Taken togetherwith previous findings that the majority of infiltrating Tcells in LSG from SS patients express lymphocyte functionassociated antigen-1 (LFA-1), which is a ligand of ICAM-1,and very late activation antigen-4 (VLA-4), a ligand ofVCAM16, our data suggest the interaction of ICAM-1/LFA-1 and/or VCAM/VLA-4 might be involved in periductalinfiltration and activation of T cells in SS.

The ability of salivary gland epithelial cells to stimulatealloreactive T cells was examined in a mixed culture systemusing salivary gland epithelial cells and allogeneic lympho-cytes. The epithelial cells treated with IFN-γ or TNF-α sig-nificantly stimulated the proliferation of the allogeneicPBMC, compared with untreated epithelial cells. Blockingexperiments against the interaction of HLA antigens andcostimulatory molecules with their ligands revealed whichmolecules stimulate the alloreactive responses observed inthis system. This effect of IFN-γ depended on the upregulat-ed HLA class I antigens, HLA-DR antigens, and CD80,while that of TNF-α depended on the upregulated HLAclass I antigens and CD80. The involvement of CD86 wasnot observed in either the IFN-γ or TNF-α induced alloreac-tive responses, probably because the expression of CD86was not fully induced and therefore the costimulationinduced by CD86 is not considered to be involved in thisculture system. These negative results do not necessarilyrule out an in vivo role of CD86. In addition, the antibodyblocking of adhesion molecules also showed no significantinhibition (data not shown). The proliferation achieved byIFN-γ or TNF-α was relatively lower than that in conven-tional mixed lymphocyte reaction. The low levels of prolif-eration can probably be explained because neither IFN-γ norTNF-α sufficiently induced the surface expression of CD80or CD86 on salivary gland epithelial cells and thereafter theactivity of these cells as APC. Alternatively, although ourresults suggest the potential of these cells to function as non-professional APC, the activity of these cells might not be asproficient as professional APC.

Taken together, our findings offer several insights. In thedevelopment of SS, salivary gland epithelial cells may con-tribute to their own destruction as a consequence of theirability to activate autoreactive T cell responses as nonpro-fessional APC. Further, these cells may also play a deter-mining role in the activation of Th1 and Th2 via the expres-sion of costimulatory molecules. A more thorough under-standing of the interaction of T cells and the target organmight lead to therapy to inhibit the initiation and/or pro-gression of SS.

ACKNOWLEDGMENTWe express our gratitude to Dr. S.M. Fu for his valuable discussions, andDr. B.T. Quinn for his critical review of the manuscript.

REFERENCES 1. Bottazzo GF, Todd I, Mirakian R, Belfiore A, Pujol-Borrell R.




Organ-specific autoimmunity. A 1986 overview. Immunol Rev1986;94:137-69.

2. van Lier RA, Brouwer M, Arden LA. Signals involved in T cellactivation. T cell proliferation induced through the synergisticaction of anti-CD28 and anti-CD2 monoclonal antibodies. Eur JImmunol 1988;18:167-72.

3. Azuma M, Cayabyab M, Phillips JH, Lanier LL. Requirements forCD28-dependent T cell-mediated cytotoxicity. J Immunol1993;150:2091-101.

4. Linsley PS, Clark EA, Ledbetter JA. T cell antigen CD28 mediatesadhesion with B cells by interacting with the activation antigenB7/BB-1. Proc Natl Acad Sci USA 1990;87:5031-5.

5. Freedman AS, Freeman G, Horowitz JC, Daley J, Nadler LM. B7, aB cell-restricted antigen that identifies preactivated B cells. J Immunol 1987;139:3260-7.

6. Kuchroo VK, Das MP, Brown JA, et al. B7-1 and B7-2costimulatory molecules activate differentially the Th1/Th2development pathways: application to autoimmune therapy. Cell1995;80:707-18.

7. Lenschow DJ, Ho SC, Sattar H, et al. Differential effects of anti-B7-1 and anti-B7-2 monoclonal antibody treatment on thedevelopment of diabetes in the nonobese diabetic mouse. J ExpMed 1995;181:1145-55.

8. Springer TA. Adhesion receptor of the immune system. Nature1990;346:425-34.

9. Dubey C, Croft M, Swain LS. Costimulatory requirements of naiveCD4+ T cells: ICAM-1 or B7-1 can costimulate naive CD4 T cellsactivation but both are required for optimum response. J Immunol1995;155:45-57.

10. Cavallo F, Martin-Fontecha A, Bellone M, et al. Co-expression ofB7-1 and ICAM-1 on tumors is required for rejection and theestablishment of a memory response. Eur J Immunol 1995;25:1154-62.

11. Nickoloff BJ, Mitra RJ, Lee K, et al. Discordant expression ofCD28 ligands, BB-1 and B7 on keratinocytes in vitro and psoriaticcells in vivo. Am J Pathol 1993;142:1029-40.

12. Denfeld RW, Kind P, Sontheimer RD, Schöpf E, Simon JC. In situexpression of B7 and CD28 receptor families in skin lesions ofpatients with lupus erythematosus. Arthritis Rheum 1997;40:814-21.

13. Sfikakis PP, Anagnostopoulos J, Karameris A, et al. Expression ofthe CD80 (B7/BB1) T cell costimulatory molecule in the skin ofpatients with systemic sclerosis [abstract]. Arthritis Rheum 1994;37Suppl:S265.

14. Garcia-Cozar FJ, Lopez-Beltran A, Molina IJ, Benito P, Pena J,Santamaria M. Aberrant expression of the B7/BB1 antigen onthyroid cells of Graves-Basedow disease (GB): induction of B7 inthyroid cells which are not autoimmune [abstract]. Immunologia1996;12:32.

15. Sugiura T, Murakawa Y, Takeuchi H, Imaoka K, Kondo M,Kobayashi S. Expression of adhesion molecules in inflammatorymyopathies [abstract]. Arthritis Rheum 1996;39 Suppl:S98.

16. Hiroki A, Nakamura S, Shinohara M, et al. A comparison ofglandular involvement between chronic graft-versus-host diseaseand Sjögren's syndrome. Int J Oral Maxillofac Surg 1996;25:298-307.

17. Lindahl G, Hedfors E, Klareskog L, Forsum U. Epithelial HLA-DRexpression and T-cell subsets in salivary glands in Sjögren'ssyndrome. Clin Exp Immunol 1985;61:475-82.

18. Fox RI, Bumol T, Fantozzi R, Bone R, Schreiber R. Expression ofhistocompatibility antigen HLA-DR by salivary gland epithelialcells in Sjögren's syndrome. Arthritis Rheum 1986;29:1105-11.

19. Saito I, Terauchi K, Shimuta M, et al. Expression of cell adhesionmolecules in the salivary and lacrimal glands of Sjögren'ssyndrome. J Clin Lab Anal 1993;7:180-7.

20. Manoussakis MN, Dimitriou ID, Kapsogeorgou EK, et al.Expression of B7 costimulatory molecules by salivary glandepithelial cells in patients with Sjögren's syndrome. ArthritisRheum 1999;42:229-39.

21. Moutsopoulos HM. Sjögren's syndrome: autoimmune epithelitis.Clin Immunol Immunopathol 1994;72:162-5.

22. Homma M, Tojo T, Akizuki M, Yamagata H. Criteria for Sjögren'ssyndrome in Japan. Scand J Rheumatol 1986; Suppl 61:26-7.

23. Vitali C, Bombardieri S, Moutsopoulos HM, et al. Preliminarycriteria for the classification of Sjögren's syndrome: results of aprospective concerted action supported by the EuropeanCommunity. Arthritis Rheum 1993;36:340-7.

24. Chisholm DM, Mason DK. Labial salivary gland biopsy inSjögren's disease. J Clin Pathol 1986;21:656-60.

25. Ohkura M, Shirasuna K, Hisanuma T, et al. Characterization ofgrowth and differential of normal human submandibular glandepithelial cells in a serum-free medium. Differentiation1993;54:143-53.

26. Ohyama Y, Nakamura S, Matsuzaki G, et al. T cell receptor Vα andVß gene use by infiltrating T cells in labial glands of patients withSjögren's syndrome. Oral Surg Oral Med Oral Pathol 1995;79:730-7.

27. Ohyama Y, Nakamura S, Matsuzaki G, et al. Cytokine messengerRNA expression in the labial salivary glands of patients withSjögren's syndrome. Arthritis Rheum 1996;39:1376-84.

28. Nakamura S, Sung S-SJ, Bjorndahl JM, Fu SM. Human T cellactivation IV: T cell activation and proliferation via the earlyactivation antigen EA 1. J Exp Med 1989;169:677-89.

29. Navarrete SA, Riemann D, Thiele K, Kehlen A, Navarrete SA,Langner J. Constitutive expression of HLA class II mRNA insynovial fibroblast-like cells from patients with rheumatoidarthritis. Immunol Lett 1997;58:53-8.

30. Windhagen A, Newcombe J, Dangond F, et al. Expression ofcostimulatory molecules B7-1 (CD80), B7-2 (CD86), andinterleukin 12 cytokine in multiple sclerosis lesions. J Exp Med1995;182:1985-96.

31. Ohyama Y, Nakamura S, Hara M, et al. Accumulation of human Tlymphotropic virus type I-infected T cells in the salivary glands ofpatients with human T lymphotropic virus type I-associatedSjögren's syndrome. Arthritis Rheum 1998;41:1972-8.

32. Pober JS, Gimbone MA Jr, Lappirre LA, et al. Overlapping patternsof activation of human endothelial cells by interleukin 1, tumornecrosis factor, and immune interferon. J Immunol 1986;137:1893-6.

33. Dustin ML, Springer TA. Lymphocyte function-associated antigen-1 interaction with intercellular adhesion molecule-1 is one of atleast three mechanisms for lymphocyte adhesion to culturedendothelial cells. J Cell Biol 1988;107:321-31.

34. Cotran RS, Pober JS. Endothelial activation: its role ininflammatory and immune reactions. In: Simionescu N, SimionescuM, editors. Endothelial cell biology. New York: Plenum Press;1988:335-47.

35. Clark DA, Lamey PJ, Jarrett RF, Onions DE. A model to study viraland cytokine involvement in Sjögren's syndrome. Autoimmunity1994;18:7-14.

36. Brookes SM, Price EJ, Venables PJW, Maini RN. Interferon-gammaand epithelial cell activation in Sjögren's syndrome. Br JRheumatol 1995;34:226-31.

37. Azuma M, Cayabyab M, Buck D, Phillips JH, Lanier LL. CD28interaction with B7 costimulates primary allogeneic proliferativeresponses and cytotoxicity mediated by small, resting Tlymphocytes. J Exp Med 1992;175:353-60.

38. Gang Y, Barrera C, Fan X, et al. Expression of B7-1 and B7-2costimulatory molecules by human gastric epithelial cells. J ClinInvest 1997;99:1628-36.




39. Pechhold K, Patterson NB, Craighead N, Lee KP, June CH, HarlanDM. Inflammatory cytokines IFN-γ plus TNF-α induce regulatedexpression of CD80 (B7-1) but not CD86 (B7-2) on murinefibroblasts. J Immunol 1997;158:4921-9.

40. Chambers CA, Kuhns MS, Egen JG, Allison JP. CTLA-4-mediatedinhibition in regulation of T cell responses: Mechanisms andmanipulation in tumor immunotherapy. Annu Rev Immunol2001;19:565-94.

41. Abbas AK, Murphy KM, Sher A. Functional diversity of helper Tlymphocytes. Nature 1996;383:787-93.

42. Mosmann TR, Sad S. The expanding universe of T-cell subsets:Th1, Th2 and more. Immunol Today 1996;17:138-46.

43. Nakanishi K, Yoshimoto T, Tsutsui H, Okamura H. Interleukin-18regulates both Th1 and Th2 responses. Annu Rev Immunol2001;19:423-74.

44. Ishikawa S, Sato T, Abe M, et al. Aberrant high expression of Blymphocyte chemokine (BLC/CXCL13) by C11b+CD11c+dendritic cells in murine lupus and preferential chemotaxis of B1cells towards BLC. J Exp Med 2001;193:1393-402.

45. Harlan DM, Hengarner H, Huang ML, et al. Mice expressing bothB7-1 and viral glycoprotein on pancreatic beta cells along withglycoprotein-specific transgenic T cells develop diabetes due to a

breakdown of T-lymphocyte unresponsiveness. Proc Natl Acad SciUSA 1994;91:3137-41.

46. Wong S, Guerder S, Visintin I, et al. Expression of thecostimulatory molecule B7-1 in pancreatic b-cells acceleratesdiabetes in the NOD mouse. Diabetes 1995;44:326-9.

47. Tandon N, Metcalfe RA, Barnett D, Weetman AP. Expression of thecostimulatory molecule B7/BB1 in autoimmune thyroid disease. Q J Med 1994;87:231-6.

48. Racke MK, Scott DE, Quigley L, et al. Distinct roles for B7-1(CD80) and B7-2 (CD86) in the initiation of experimental allergicencephalomyelitis. J Clin Invest 1995;96:2195-203.

49. Finck BK, Linsley PS, Wofsy D. Treatment of murine lupus withCTLA4Ig. Science 1994;265:1225-7.

50. Cross AH, Girard TJ, Giacoletto KS, et al. Long-term inhibition ofmurine experimental autoimmune encephalomyelitis using CTLA-4-Fc supports a key role for CD28 costimulation. J Clin Invest1995;95:2783-9.

51. Saegusa K, Ishimaru N, Yanagi K, et al. Treatment with anti-CD86costimulatory molecule prevents the autoimmune lesions in murineSjögren's syndrome through upregulated Th2 response. Clin ExpImmunol 2000;119:354-60.


Possible Function of Salivary Gland Epithelial Cells as ...1884 The Journal of Rheumatology 2002;...

Documents

Transcript of Possible Function of Salivary Gland Epithelial Cells as ...1884 The Journal of Rheumatology 2002;...