Tumor Necrosis Factor-α in Ocular Mucous Membrane Pemphigoid and Its Effect on Conjunctival Fibroblasts

Abstract

Purpose.: First, to determine whether tumor necrosis factor-(TNF)-α is expressed in the conjunctiva of ocular mucous membrane pemphigoid (MMP) and the consequences of systemic immunosuppressive treatment on this expression. Second, to investigate the in vitro effects of TNFα on human conjunctival fibroblasts.

Methods.: The expression of TNFα in conjunctival tissues of patients with actively inflamed ocular MMP (n = 10), patients with clinically noninflamed ocular MMP after systemic immunosuppressive treatment (n = 10), and normal subjects (n = 10) was studied by immunohistochemistry. The effect of TNFα on functional assays of human conjunctival fibroblast activity were investigated, including migration, collagen lattice contraction, matrix metalloproteinase (mmp), and tissue inhibitor of matrix metalloproteinase (timp) secretion, proliferation, and surface expression of HLA-DR, ICAM, CD80, CD86, CD40, CD40-ligand.

Results.: In active ocular MMP, TNFα is expressed by a large number of stromal infiltrating cells (234 cells/mm²), and although the level of stromal TNFα expression is significantly reduced after immunosuppressive treatment (90 cells/mm²), these levels are still significantly elevated compared with normal conjunctiva (10 cells/mm², P < 0.05). TNFα stimulates increased migration by conjunctival fibroblasts (P < 0.001), increased production of mmp-9 (P = 0.01), decreased production of timp-2 (P = 0.01) and timp-4 (P = 0.04), and upregulated expression of CD40 and ICAM (P = 0.04). No significant effects of TNFα on fibroblast proliferation or collagen lattice contraction were detected.

Conclusions.: Increased conjunctival expression of TNFα in ocular MMP suggests that systemic TNFα antagonists are likely to be effective in controlling severe disease unresponsive to conventional systemic immunosuppression. Residual TNFα expression persists in clinically noninflamed disease. TNFα appears to have profibrotic and proinflammatory effects on human conjunctival fibroblasts.

Ocular mucous membrane pemphigoid (MMP), previously known as ocular cicatricial pemphigoid (OCP), is a blinding systemic autoimmune disease characterized by recurrent episodes of inflammation and progressive fibrosis of the conjunctiva and other mucosal surfaces. It is an immunobullous disease and is associated with autoantibodies directed against basement membrane zone proteins. ¹ While conventional treatment with systemic immunosuppressive agents controls inflammation in most patients with ocular MMP, ² such treatment has limited efficacy in recalcitrant or severely inflamed cases, and side effects related to toxicity restrict the maximum therapy able to be given. 

The success of TNFα antagonists in treating inflammatory autoimmune diseases such as rheumatoid arthritis and inflammatory bowel disease has led to a handful of cases reporting improvement with the use of TNFα antagonists in recalcitrant MMP which had not responded to conventional immunosuppression. ^3–5 However, the scientific rationale for the use of TNFα antagonists in MMP based on current evidence is weak. Although it has been reported that serum levels of TNFα are elevated in eyes with MMP compared with normal control eyes, ⁶ there are limited studies examining TNFα expression in MMP tissue. Tissue expression appears to be important because in rheumatoid arthritis, although there is no evidence that plasma TNFα levels can predict the clinical response to TNFα antagonists, synovial expression of TNFα is a significant predictor of response to TNFα antagonists. ⁷  

The inflammatory lesions in MMP heal with an excessive scarring response. This is most evident in the conjunctiva, where scar tissue formation causes terminal irreversible damage to the ocular surface. ⁸ The pathophysiology of conjunctival fibrosis in ocular MMP is poorly understood. ⁹ Observations from other chronic mucosal fibrotic disorders provide insights into what could be occurring in ocular MMP. Fibrosis typically results from chronic inflammation and repair, which leads to deposition of connective tissue elements that remodel and destroy normal tissue architecture. ¹⁰ Although inflammation typically precedes the development of fibrosis, the situation is complicated further by findings indicating that the mechanisms regulating fibrogenesis are distinct from those controlling inflammation. ¹¹ The role of CD40 signaling in regulation of inflammation and fibrosis has been demonstrated in the lung. ^12,13 During an inflammatory response, fibroblasts increase their expression of CD40, and interactions between tissue fibroblasts and infiltrating T lymphocytes via the CD40/CD40L pathway have been shown to powerfully co-stimulate T-lymphocyte proliferation as well as induce fibroblasts to produce proinflammatory and chemoattractant cytokines. This process in turn may lead to fibroblast proliferation and extracellular matrix synthesis. Chronic stimulation through CD40 may hence lead to fibrosis. Rather than merely acting as structural cells, fibroblasts thus appear to participate actively in the immune response by producing cytokines and chemokines that initiate the recruitment and retention of bone marrow-derived immune effector cells. ¹⁴ Given that conjunctival fibrosis in ocular MMP can still progress despite apparent clinical control of inflammation by conventional immunosuppressive therapy, ^2,15 better understanding of the cellular and molecular changes in treated tissue compared with actively inflamed tissue would assist and guide in developing adjunctive local therapies to target conjunctival fibrosis. 

TNFα is a potent inflammatory cytokine produced by many inflammatory cells, including macrophages, CD4⁺ and CD8⁺ T cells, mast cells, smooth muscle cells, endothelial cells, and fibroblasts. There is controversy about whether TNFα is a profibrosis or antifibrosis cytokine. ^16–19 Results of in vitro and in vivo experiments regarding the role of TNFα in fibrosis are in part contradictory, and fibroblasts from different bodily locations and sources appear to react differently to TNFα in comparison with other fibroblasts. Recent reports suggest that TNFα antagonists may be beneficial in the treatment of fibrotic disorders such as systemic sclerosis. ²⁰  

In this study, we investigated TNFα expression in conjunctival ocular MMP tissue, including changes in TNFα expression after systemic immunosuppressive treatment. To determine what effects TNFα may have on conjunctival fibrosis, we also investigated the effect of TNFα on functional assays of human conjunctival fibroblast activity and co-stimulatory molecule expression. 

Materials and Methods

Conjunctival Tissues

Bulbar conjunctival biopsies obtained from patients with ocular MMP were used to study the expression of TNFα. The patients were classified according to ocular disease activity as having active disease with acute inflammation (n = 10) or chronic disease (n = 10) without clinically apparent inflammation, after immunosuppressive treatment. The diagnosis of ocular MMP was based on clinical presentation and direct immunofluorescence of the conjunctiva showing linear immunoglobulin and/or complement deposition along the basement membrane zone (BMZ). Conjunctivae from 10 normal individuals undergoing routine cataract surgery were used as control samples. Details regarding the patients and normal control subjects are shown in Table 1. The protocol adhered to the tenets of the Declaration of Helsinki, and institutional research and ethics committee approval was granted. Informed consent was obtained from all patients and normal control subjects participating in the study. The presence of positive stromal or epithelial staining was also graded on a scale of 0 to ++++, based on both the number of cells stained and staining intensity. 

Table 1.

 

Details of Patients and Control Subjects

Diagnosis	Case	Age	Sex	Disease Duration (y)	Bulbar Inflammation Grade (0–4)	Tauber Stage 42 (Upper Stage/Lower Stage)	Topical Therapy	Systemic Therapy	Other Eye Disease
Active MMP	1	54	F	0.5	3	IIcIIId/IIcIIId	Hypromellose, acetylcysteine, yellow soft paraffin	Nil
	2	80	M	10	3	IIbIIIc(2)/IIdIIId(2)	Carmellose	Mycophenolate + dapsone
	3	60	F	2	3	I/IIcIIIc(3)	Carmellose	Mycophenolate + dapsone
	4	74	M	2	3.5	IIa/IIcIIIc(2)	Dorzolamide, bimatoprost	Nil	Glaucoma
	5	76	F	2	2.5	IIc/IIbIIIa(2)	Prednisolone, hypromellose, carmellose	Mycophenolate + dapsone	Sicca, blepharitis
	6	51	F	0.1	3	I/IIIa(1)	Nil	Nil
	7	57	M	0.5	3	IIb/IIcIIIb(2)	Dexamethasone, lacrilube	Nil
	8	64	F	1	2	IIa/IIbIIIc(2)	Prednisolone	Nil
	9	83	F	7	3	IIa/IIbIIIb(2)	Brimonidine, latanoprost, timolol, dorzolamide, carmellose	Mycophenolate + doxycycline	Glaucoma
	10	72	F	0.25	4	IIa/IIcIIIc(2)	Ofloxacin, cyclopentolate, carmellose	Nil	Microbial keratitis
Treated uninflamed MMP	1	59	F	10	1	IIbIIIb(2)/IIdIIId(2)	Carbomer 980	Dapsone
	2	86	F	15	1	IIb/IIcIIIb(2)	Nil	Nil
	3	78	F	10	1	IIbIIIb(2)/IIdIIId(2)	Nil	Nil
	4	84	F	2	0	IIb/IIcIIIc(2)	Nil	Mycophenolate
	5	66	F	2	0	IIb/IIcIIIb(2)	Nil	Dapsone
	6	76	F	4	1	IIc/IIcIIIc(3)	Betaxolol	Dapsone	Glaucoma
	7	59	M	4	1	I/IIa	Hypromellose, carmellose, liquid paraffin, chloramphenicol	Cyclophosphamide
	8	76	M	6	1.5	IIdIIId(2)/IIbIIIb(2)	Chloramphenicol, hypromellose, retinoic acid, acetylcysteine	Cyclophosphamide	Blepharitis
	9	60	F	3	1	I/IIaIIIa(1)	Carmellose	Cyclophosphamide + dapsone
	10	62	F	0.75	0.5	IIa/IIcIIIa(1)	Hyaluronate	Prednisolone
Normal control	1	50	F	—	0	—	Nil	Nil	Cataract
	2	76	M	—	0	—	Nil	Nil	Cataract
	3	65	M	—	0	—	Nil	Nil	Cataract
	4	70	F	—	0	—	Nil	Nil	Cataract
	5	65	M	—	0	—	Nil	Nil	Cataract
	6	73	M	—	0	—	Nil	Nil	Cataract
	7	84	F	—	0	—	Nil	Nil	Cataract
	8	57	M	—	0	—	Nil	Nil	Cataract
	9	84	F	—	0	—	Nil	Nil	Cataract
	10	62	M	—	0	—	Nil	Nil	Cataract

Immunohistochemistry

Immunohistochemistry was performed on glycol methacrylate resin-embedded sections of conjunctiva, prepared as described previously, ²¹ using the JB4 kit (TAAB Laboratories, Aldermaston, UK). Serial sections 2-μm thick were cut, mounted on polylysine-coated slides, and air dried. Endogenous peroxidase was inhibited using 0.3% hydrogen peroxide in 0.1% sodium azide, followed by incubation with 10% fetal calf serum to block nonspecific binding. The sections were incubated overnight at room temperature with a mouse antibody against human TNFα (ab9579, 1:50 dilution; AbCam Ltd., Cambridgeshire UK). After the sections were washed with PBS, biotinylated rabbit anti-mouse immunoglobulin (Dako, Cambridgeshire, UK) was applied to the sections, followed by incubation with streptavidin-peroxidase (Dako), and finally amino-ethyl carbazole (AEC; Dako), forming a red AEC reaction product. For TNFα-CD3 double staining, the slides were washed with PBS, and the immunohistochemistry process was repeated with a mouse antibody against human CD3 (Dako) 1:10 dilution, and 3,3′diaminobenzidine (DAB), forming a brown DAB reaction product. The specimens were counterstained with Mayer's hematoxylin (Dako) and mounted (Glycergel mounting medium; Dako). Double staining was identified by a brown or combined red-black color caused by a combined positive reaction to AEC and DAB. Human tonsil sections were used as the positive control, and the two negative controls used were fetal calf serum and an isotype-matched, irrelevant monoclonal antibody (Dako). The number of cells stained in the stroma and the percentage area of the sections stained with intravascular TNFα were measured in a masked fashion in five representative high power-fields per patient (BX51 microscope; Olympus, Tokyo, Japan) with image analysis software. 

Isolation of Conjunctival Fibroblasts

Fibroblasts were grown from the conjunctival biopsies of the normal control subjects as previously described. ²² The explants were placed into tissue culture wells under a coverslip and cultured with fibroblast culture medium (FCM) composed of Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% (vol/vol) heat-inactivated fetal calf serum, 100 IU/mL penicillin, 100 μg/mL streptomycin, and 0.25 μg/mL amphotericin B (all from Gibco-Invitrogen Ltd, Paisley, Scotland, UK) at 37°C with 5% (vol/vol) CO₂ in air. The cells were passaged 1:3 with trypsin/EDTA. Cultures were used between passages 3 and 7 for experiments, and cultures were assessed for typical fibroblast morphology by phase contrast microscopy before every experiment. By passage 3 onward, contaminating epithelial cells had been eliminated in cultures that had been fed only FCM, due to the fastidious nutritional requirements of epithelial cells. 

Serum-free medium (SFM) composed of DMEM supplemented with 0.1% bovine serum albumin (BSA; Sigma-Aldrich, Dorset, UK) was used in all experiments in which the effect of the addition of TNFα was being evaluated. Testing under serum-free conditions eliminates the possibility that the observed responses could be due to the activity of unspecified cytokines and growth factors present in serum-containing medium. The optimal concentration of TNFα used in experiments was initially determined from previous results by our group. ²³ If this level did not result in a significant response, a range of TNFα concentrations (1, 10, and 100 ng/mL) was then evaluated in a preliminary dose–response experiment. The concentration that showed the greatest response in this preliminary experiment was then selected as the optimal concentration for that assay. 3T3 cells (a fibroblast cell line) were used as the control for initial calibration experiments. 

Migration

Cell culture inserts incorporating polyethylene terephthalate membranes with a pore size of 8 μm (BD Falcon; VWR International, Leicestershire UK), which fit into wells in a tissue culture plate, were used to assess fibroblast migration in the presence of TNFα. The cells were seeded into the inserts at a density of 8000 per insert in 200 μL SFM and allowed to attach to the upper surface of the membrane for 4 hours. A volume of 700 μL of 10 ng/mL recombinant human TNFα (R&D Systems, Abingdon, UK) in SFM was added to the well in the tissue culture plate, so that the TNFα-containing medium was in contact with the undersurface of the membrane in the culture insert. SFM was used as a negative control, and 10% bovine serum FCM as a positive control. The cells were incubated for 16 hours to permit migration through the pores, to the undersurface of the membrane. The culture inserts were then washed with PBS to remove excess protein, fixed in 70% (vol/vol) methanol for 5 minutes, then stained with Mayer's hematoxylin (Dako) for 30 minutes and rinsed in tap water. Settled cells on the upper surface of the membrane in the culture inserts were removed with cotton swabs. The number of migrated cells on the undersurface of the membrane was counted per 10× objective field (average of five fields) with an inverted microscope and cell-counting software (Image J, Java image-processing program developed by Wayne Rasband, National Institutes of Health, Bethesda, MD, available at http://rsb.info.nih.gov/ij/). 

Collagen Contraction

To assess matrix contraction in the presence of TNFα, we used free-floating collagen lattice models. Three-dimensional fibroblast-populated type I rat tail collagen (5 mg/mL; Sigma-Aldrich) lattices were prepared ²⁴ with 16.7 × 10⁵ cells/mL of lattice mixture. The collagen lattices were incubated for 2 hours at 37°C to set in the culture wells and then the lattices were detached immediately after feeding with the previously optimized concentration of 10 ng/mL TNFα in SFM. Reduction in lattice area at days 1, 3, and 7, caused by contraction was digitally photographed, and the gel areas calculated using image analysis software (Image J). Conditioned medium was collected from the contracting lattices. 

Matrix Metalloproteinase Protein Levels

Conditioned medium collected from contracting lattices was analyzed for matrix metalloproteinase (mmp) and tissue inhibitor of matrix metalloproteinase protein (timp) levels using an antibody-coated membrane array (Raybiotech Inc, Norcross, GA) in accordance with the manufacturer's instructions. The mmp levels in baseline control medium were subtracted from the conditioned medium results. 

Proliferation

Fibroblast cell division in the presence or absence of TNFα was analyzed by flow cytometry using the fluorescent probe CFSE (carboxyfluorescein diacetate, succinimidyl ester; Vybrant CFDA SE Cell Tracer Kit; Invitrogen). Serum-deprived fibroblasts were incubated with 1 μM CFSE in serum-free medium at 37°C for 10 minutes, then ice-cold FCM was added to stop the reaction. The cell pellet was resuspended in ice-cold FCM for 30 minutes to allow free CFSE to come out of the cells. The cell pellet was then resuspended at 2 × 10⁵ cells/mL in FCM, the cells were seeded into six-well tissue culture plates at a density of 4 × 10⁵ cells per well and stimulated with TNFα 10 ng/mL in FCM. After 72 hours, the cells were detached with trypsin and 10,000 events acquired for flow cytometry (FACScalibur; BD PharMingen, Cowley, UK). Commercial software (Winlist; Verity Software House, Topsham, ME) was used to analyze the number of divided cells per 5000 undivided cells. 

Immunofluorescence Staining and Flow Cytometry

Confluent fibroblasts were cultured in six-well plates in the presence of various concentrations of TNFα (1, 10, or 100 ng/mL) or IFNγ (1 ng/mL; Peprotech, London, UK) in SFM, for 72 hours. For three-color immunofluorescence staining, the fibroblasts were washed with Ca²⁺ and Mg^2+-free Hanks' balanced salt solution (HBSS) and detached with collagenase (1 mg/mL) in HBSS. The cell suspension containing at least 1 × 10⁵ cells was centrifuged at 4000 rpm for 4 minutes, the cell pellet resuspended in 10% mouse serum for 20 minutes to block nonspecific binding, then the cells were stained in suspension with 5 μL each of anti-ICAM-1^PE, anti-CD80^FITC and anti-HLA-DR^PerCP or of anti-CD154^FITC, anti-CD86^PE, and anti-CD40^PECy5 for 30 minutes at 4°C before washing twice and acquiring 10,000 events for flow cytometry (FACScalibur; BD PharMingen). Isotype matched, irrelevant monoclonal antibodies were used as negative controls. The same software (Winlist; Verity Software House) as was used to determine proliferation was used for analysis. Analyses were performed on a population of live cells gated by forward and sidescatter to include the fibroblast population. At least 10,000 events were acquired in the live cell region, and the percentages of positive cells were calculated after subtracting background staining. Mean fluorescence intensity was used to measure intensity of expression. 

Statistical Analysis

Differences between groups were examined for statistical significance using the Mann-Whitney U test, or one-way analysis of variance (ANOVA). Linear regression was used to analyze for a significantly non-0 slope relationship between TNFα concentration and intensity of ICAM expression. P < 0.05 was considered to be statistically significant. 

Results

Expression of TNFα in Ocular MMP Conjunctival Tissue, in Active Disease and in Clinically Noninflamed Treated Disease

Expression of TNFα in the substantia propria of normal control conjunctival sections was confined to intravascular spaces, with very little expression in the stromal tissues (Fig. 1A). In contrast, in active ocular MMP, there were many stromal cells expressing TNFα (234 cells/mm²; Fig. 1B), 32% (76/234) of which were also positive for the T cell marker CD3 (Fig. 2A). It is likely that the remaining TNFα⁺CD3⁻ cells are monocytes/macrophages, because they are, after lymphocytes, the second most abundant stromal inflammatory cells found in ocular MMP. ^21,25,26 After immunosuppressive therapy, in clinically noninflamed ocular MMP, the number of stromal TNFα-expressing cells was significantly reduced to 38% of the number of cells present in active disease (90 cells/mm², P < 0.05), but this was still ninefold greater than the number of stromal cells present in normal control subjects (10 cells/mm², P < 0.05). The proportion of CD3-positive double-stained cells was similar across the 3 groups (32% [76/234], 33% [30/90], and 35% [3.5/10]; Fig. 2A). There was no difference between the three groups with regard to the percentage area of the sections stained with intravascular TNFα (Fig. 2B). Stromal cell TNFα staining was present in all 10 patients with active ocular MMP and in 7 of 10 patients with treated ocular MMP (Table 2). Occasional stromal cells stained positive in 3 of the 10 normal control subject. Epithelial TNFα staining was present in 4 of 10 patients with active ocular MMP, 4 of 10 patients with treated ocular MMP, and 1 of the 10 normal subjects. 

Figure 1.

 

Human conjunctiva of patients with MMP shows TNFα expression. Immunohistochemistry of bulbar conjunctival sections showing (A) intravascular TNFα⁺CD3⁻ cells in a normal subject and (B) positive intravascular and stromal TNFα staining in a patient with actively inflamed ocular MMP. Small arrows: stromal TNFα⁺CD3⁻ cells; arrowhead: stromal TNFα⁺CD3⁺ cell; large arrow: intravascular TNFα⁺CD3⁻ cell. TNFα stained dark red (AEC), CD3 stained brown (DAB). (C) A negative isotype mAb control of a normal subject and (D) a negative isotype mAb control of a patient with actively inflamed ocular MMP. Note the stromal inflammatory cell infiltrate. Bar, 100 μm.

Figure 2.

 

Cell counts of positive stromal staining in conjunctiva of patients with MMP and control subjects. (A) Stromal TNFα/CD3 double-staining cell counts. In patients with active ocular MMP, many stromal TNFα-expressing cells were present (234 cells/mm²), 32% (76/234) of which also expressed the T-cell marker CD3. After immunosuppressive therapy, in clinically noninflamed treated ocular MMP, the number of stromal TNFα-expressing cells was significantly reduced to 38% of the number of cells present in active disease (90 cells/mm², P < 0.05), but the count was still nine times greater than the number of stromal TNFα-expressing cells present in normal control samples (10 cells/mm², P < 0.05). (B) Percentage area of the sections stained with intravascular TNFα. There was no difference in intravascular TNFα staining between active or treated ocular MMP or the control. Data are the mean and SE from results in 10 individuals per group, counting at least nine fields per individual. *P < 0.05.

Table 2.

 

Stromal and Epithelial TNFα Staining Results for Each Patient

Diagnosis	Case Number	Stromal TNFα Staining (Grade N to ++++)	Epithelial TNFα Staining (Grade N to ++++)
Active MMP	1	+++	N
	2	+	+++
	3	+++	++
	4	+++	+++
	5	++	N
	6	++	N
	7	++++	N
	8	++++	++++
	9	++	N
	10	+	N
Treated uninflamed MMP	1	++	+
	2	+/N	N
	3	+/N	N
	4	+	+
	5	N	N
	6	++	++
	7	N	N
	8	+	N
	9	N	N
	10	++	+
Normal control	1	N	N
	2	N	N
	3	N	N
	4	+/N	N
	5	N	N
	6	+/N	N
	7	N	N
	8	+/N	+
	9	N	N
	10	N	N

 N, nil staining.

Fibroblast Migration, Collagen Contraction, and Proliferation in Response to TNFα

We first investigated the effect of TNFα on key fibroblast functions, including cell migration, collagen contraction, and proliferation. Whereas TNFα stimulated conjunctival fibroblast migration under serum-free conditions (P < 0.0001, Fig. 3), no significant contraction of fibroblast-collagen lattices was observed in response to a range of TNFα concentrations (1 to 100 ng/mL) compared with the serum-free negative control (data not shown). There was also no significant difference in fibroblast proliferation after stimulation with 10 ng/mL TNFα, compared with the FCM negative control (data not shown). 

Figure 3.

 

Effect of TNFα on conjunctival fibroblast migration. Conjunctival fibroblasts were seeded into culture inserts incorporating porous membranes and allowed migration overnight toward the lower chamber containing the test substance. Negative control 0.1% BSA (SFM) and positive control 10% FCS (FCM). TNFα stimulated fibroblast migration in serum-free conditions. Data are the mean and SE of results in five individual experiments. *P < 0.0001.

Production of mmp in Response to TNFα

To determine whether the presence of TNFα was associated with altered mmp and timp levels, we analyzed conditioned medium from fibroblast-populated collagen lattices treated with TNFα with an antibody-coated membrane array testing a panel of seven mmps and three timps. Both latent and active forms of the mmps and timps were detected by the mmp antibody array. Figure. 4 shows that the level of mmp-9 increased (P = 0.01) and the levels of timp-2 (P = 0.01) and timp-4 decreased (P = 0.04) in the presence of TNFα, compared with levels in the serum-free negative control. 

Figure 4.

 

Mmp and timp levels in conditioned medium of conjunctival fibroblast-populated collagen gels incubated over 7 days in the presence of TNFα, as detected in an antibody-coated membrane array. Data are relative levels compared with the membrane array negative control. There was a significant increase in mmp-9 (P = 0.01) and a decrease in timp-2 (P = 0.01) and -4 (P = 0.04) levels in the conditioned medium from TNFα-treated collagen gels, in comparison with the untreated control 0.1% BSA (SFM) collagen gels. Data are the mean and SE from the results of five individual donors. *P < 0.05.

Immune Marker Expression by Conjunctival Fibroblasts in Response to TNFα

Interactions between tissue fibroblasts and infiltrating T lymphocytes via the CD40/CD40L pathway have been shown to powerfully co-stimulate T-lymphocyte proliferation and to activate fibroblasts, thus potentially promoting fibrogenesis. ¹³ We investigated surface expression of HLA-DR, ICAM, and T-cell co-stimulatory molecules by conjunctival fibroblasts in response to 72 hours' stimulation with TNFα or IFNγ under serum-free conditions. ICAM was constitutively expressed by a high percentage of fibroblasts (84.3% ± 13.5%) under basal conditions, and there was no significant change in the percentage level of expression after TNFα or IFNγ treatment (Fig. 5A). However, the intensity of ICAM expression by the fibroblasts escalated in response to increasing concentrations of TNFα (mean fluorescence intensity in BSA [bovine serum albumin], 113 ± 29; 1 ng/mL TNFα, 1361 ± 402; 10 ng/mL TNFα, 2117 ± 604; and 100 ng/mL TNFα, 2133 ± 717, P = 0.03 linear regression, Fig. 5B). Fibroblast expression of the co-stimulatory molecule CD40 was significantly upregulated by all concentrations of TNFα, with the maximum percentage expression occurring in response to 10 ng/mL TNFα (BSA, 1.8% ± 1.3%; 1 ng/mL TNFα, 14.3% ± 8.3%; and 10 ng/mL TNFα, 27.7% ± 11.7%, P = 0.04; Fig. 5A). No significant change in either the percentage or intensity of expression of HLA-DR, CD80, CD86, or CD40 ligand was detected in response to TNFα. In comparison, IFNγ upregulated expression of HLA-DR, CD40, and CD154. 

Figure 5.

 

Fibroblast expression of surface markers of activation and co-stimulatory molecules in response to TNFα. The response to IFNγ has been included for comparison. Conjunctival fibroblasts were treated with various concentrations of TNFα (1, 10, and 100 ng/mL) or IFNγ (1 ng/mL) for 72 hours under serum-free conditions, and the surface expression of HLA-DR, ICAM, CD80, CD86, CD40, and CD40-ligand was determined by flow cytometry. Untreated control 0.1% BSA (SFM). (A) Percentage of conjunctival fibroblasts expressing surface markers and co-stimulatory molecules. All concentrations of TNFα significantly upregulated the percentage expression of CD40 (* P < 0.05 in comparison to untreated control), with maximum expression occurring in response to 10 ng/mL. Note the high percentage of fibroblasts expressing ICAM under basal conditions and in the presence of TNFα and IFNγ. (B) Intensity of expression of ICAM (intercellular adhesion molecule) by conjunctival fibroblasts in response to TNFα stimulation. Although the percentage of cells expressing ICAM was high under basal conditions and in the presence of TNFα, the intensity of ICAM expression increased in response to TNFα stimulation (* P = 0.03). Data are the mean ± SE of results from six donors.

Discussion

The importance of TNFα as a proinflammatory cytokine which drives inflammatory disease progression has come to light with the success of TNFα antagonists in treatment of rheumatoid arthritis, spondyloarthropathies, and Crohn's disease. ²⁷ However, the role of TNFα in fibrosis is controversial, with some studies showing antifibrotic effects of TNFα on particular cell types, whereas others indicate a profibrosis role. ^16–19 In this study, we demonstrated increased tissue expression of TNFα in ocular MMP; thus, systemic treatment with TNFα antagonists could be expected to ameliorate inflammatory activity in this disease. Furthermore, we have demonstrated that although systemic immunosuppressive treatment is associated with a significant decrease in tissue TNFα expression, there is residual persistent expression of TNFα, even when inflammation clinically appears to be controlled. We have also demonstrated a possible role for TNFα in the pathogenesis of conjunctival fibrosis in MMP. TNFα is a chemoattractant for conjunctival fibroblasts. Migration of fibroblasts to the site of injury expressing TNFα could lead to a pathologic accumulation of fibroblasts and excessive scarring at this site. TNFα increases mmp-9 production and causes activation of conjunctival fibroblasts by upregulating the intensity of adhesion molecule expression and upregulating CD40 expression. 

We did not detect any difference in cell division in response to 10 ng/mL TNFα in FCM compared with FCM alone, but proliferation of myofibroblasts in response to TNFα has been reported. ^19,28 The fibroblasts used in our experiments had been assessed for myofibroblast characteristics with α-smooth muscle actin staining looking for assembled actin myofilaments, and these were found to be absent. Different TNFα concentrations may have stimulated proliferation in our assay, or it may be that TNFα-stimulated proliferation is a myofibroblast-specific response. 

The profibrosis mediators TGF β (transforming growth factor-β), CTGF (connective tissue growth factor), and IL-4 (interleukin-4) have been shown to be expressed at increased levels in MMP conjunctiva. ^29,30 TNFα has been reported to be present in 6 of 8 patients with ocular MMP in a limited study by Cordero et al. ³¹ It is not clear whether these patients had active inflammation, or clinically noninflammatory disease. Bernauer et al. ³² previously found positive substantia propria expression of TNFα in all active MMP, chronic MMP, and normal conjunctival specimens; this may have been due to intravascular staining, which we similarly found was present in all specimens. We evaluated intravascular staining separately from extravascular stromal TNFα staining, and the latter was found to be virtually absent in normal conjunctiva in our study. It is likely that multiple mediators and cytokines are present during the acute and chronic inflammatory phases of MMP. What is most important is to determine which mediators play key roles in orchestrating the cellular and cytokine cascade of events and to evaluate the molecules present that could have tangible therapeutic applications. In the case of TNFα, the availability of highly successful and safe TNFα antagonists make it a logical choice for investigation. TNFα has also been shown to induce conjunctival fibroblasts to synthesize mRNA for cytokines which play a key role in ocular MMP, such as m-CSF (macrophage colony-stimulating factor) and MIF (macrophage migration inhibitory factor). ^33,34  

Our findings of persistent residual TNFα expression even when the conjunctiva is clinically white and noninflamed after systemic immunosuppressive treatment, is in agreement with work proposing ongoing release of cytokines in the presence of a significant cellular infiltrate (white inflammation) which cannot be seen on the slit lamp. ²⁹ Progressive fibrosis, despite immunosuppressive treatment, may be driven by this underlying chronic inflammation; alternatively, ocular MMP fibroblasts may be transformed into an abnormally active state, ³⁵ in a fashion similar to the profibrosis phenotype observed in scleroderma fibroblasts. ³⁶  

The free-floating fibroblast-populated collagen lattice model is thought to simulate matrix contraction exerted by tractional forces generated by cells migrating through matrix in vivo. ³⁷ Although addition of TNFα did not stimulate matrix contraction, increased mmp-9 levels were detected in the conditioned medium of the fibroblast-populated collagen lattices. These findings are consistent with those in other studies reporting TNFα-induced mmp-9 expression by gingival fibroblasts and other cells. ^28,38 Leonardi et al. ²³ have shown increased mmp-9 and also mmp-1 production, and decreased timp-1 production in response to TNFα stimulation, by conjunctival fibroblasts adherent to a tissue culture plate. They did not examine timp-4 expression. The differences in mmp-1 and timp-1 results between our study and theirs may reflect differences in the in vitro models used. Our in vitro model of fibroblasts populating a three-dimensional collagen gel matrix could be seen as more physiological. 

Mmp-9 (also known as gelatinase B) degrades extracellular matrix collagens such as collagen type IV, the major component of basement membranes, and laminin and plays a role in regulating cell migration. The mmp-9 detected by the antibody array in our study could have been either the latent or active form. Recently, it has been shown that the proteolytic activities of mmps are not necessary for mmp-induced cell migration and that the hemopexin domain of pro-mmp-9 plays an important role in cell migration. ³⁹ The results of our migration assay experiments showing increased fibroblast migration in response to TNFα may well be related to this increased mmp-9 production, noting that this would represent migration without concomitant matrix contraction, given that no contraction of the collagen lattices was observed in response to TNFα. Decreased timp-2 and timp-4 levels were also detected in the conditioned medium of the fibroblast-populated lattices. By virtue of their ability to inhibit mmp activity, all timps are believed to function as inhibitors of angiogenesis. Timp-2 has also been found to directly inhibit endothelial cell proliferation and angiogenesis by an mmp-independent process. ⁴⁰ Timp-4 has a growth promoting activity in vitro, prevents apoptosis, and inhibits TNFα converting enzyme. ⁴¹ It is possible that the ability of TNFα to stimulate angiogenesis is related to this reduction in timp activity. 

T lymphocytes are significantly increased in the subepithelial cellular infiltrate across all phases of ocular MMP disease activity, ²¹ and T-cell–fibroblast cross-talk and feedback loops may be an important mechanism in fibrosis. IFNγ-stimulated lung fibroblasts have been shown to co-stimulate T-lymphocyte proliferation via CD40, but not via the CD80 or CD86 co-stimulatory molecules. ¹³ Interactions between tissue fibroblasts and infiltrating T lymphocytes, via the CD40/CD40 ligand pathway, augment inflammation and may promote fibrogenesis by activating both cell types. We have shown that TNFα selectively upregulates expression of CD40 on conjunctival fibroblasts, which may then interact with T cells via CD40 ligand, inducing the fibroblasts to proliferate, produce cytokines, and lay down extracellular matrix ¹² and stimulating T-cell proliferation and production of cytokines that could promote further fibroblast activation and matrix deposition. The effect of the addition of TNFα on functional fibroblast activity in the presence of T-cell co-culture requires further investigation. Most of the TNFα-expressing cells were not T cells and were likely to be monocytes/macrophages. Their cell type should be confirmed by further experiments. 

In summary, expression of TNFα is increased in the conjunctiva of patients with ocular MMP; hence, systemic TNFα antagonists can be expected to be useful in severe ocular MMP that does not respond to conventional immunosuppressants. TNFα appears to have a profibrotic effect on normal conjunctival fibroblasts. Further studies investigating the effect of TNFα antagonists on conjunctival fibroblast activity would assist in evaluating the role of TNFα and TNFα antagonist therapy in . 

Footnotes

 Supported by an Action UK Medical Research Fellowship (VPJS).

Footnotes

 Disclosure: V.P.J. Saw, None; R.J.C. Dart, None; G. Galatowicz, None; J.T. Daniels, None; J.K.G. Dart, None; V.L. Calder, None

Footnotes

 The publication costs of this article were defrayed in part by page charge payment. This article must therefore be marked “advertisement” in accordance with 18 U.S.C. §1734 solely to indicate this fact.