MATERIALS AND METHODS

Patient selection

Conjunctival tissue was biopsied from glaucoma patients at the time of their filtration surgery as described before.¹³ The tenets of the Declaration of Helsinki were adhered to, with ethical approval obtained from the Moorfields Eye Hospital ethics committee and informed consent obtained from the patients before surgery. The aetiologies of our glaucoma patients consisted of primary open angle glaucoma, angle closure glaucoma, or pigment dispersion glaucoma. One subgroup of patients was about to undergo filtration surgery for the first time, a second subgroup of patients was considered higher risk for conjunctival scarring since these patients had already failed a previous trabeculectomy because of aggressive conjunctival scarring. All our patients had previously received eyedrop treatment for their glaucoma.

Immunohistology

The indirect immnuoperoxidase technique was used to detect T cells and fibroblasts in the conjunctival sections. Frozen sections of 6 μm were air dried for 2 hours and then fixed in acetone:chloroform 1:1 for 5 minutes. The sections were incubated for 10 minutes with normal rabbit serum (NRS) to minimise background staining by blocking non-specific binding sites. They were then incubated with 50 μl of the primary antibodies—a pan-anti-T IgG monoclonal antibody (mAb) mix (RFT mix) or the AS02 mAb (Dianova, Hamburg, Germany) previously used by our group as a specific fibroblast mAb at pretitrated concentrations for 45 minutes at room temperature.¹⁰ After a wash in phosphate buffered saline (PBS), 50 μl of the secondary peroxidase conjugated goat antimouse IgG Ab (P161, Dako, Bucks) diluted 1/100 in PBS supplemented with 4% normal human serum (NHS) was applied for 45 minutes at room temperature. The reaction was developed using diaminobenzidine (D-5905, Sigma, Dorset, 10 mg of DAB dissolved in 16.6 ml of PBS, followed by 55.3 μl of 3% hydrogen peroxide) and then counterstained with haematoxylin.

The positive controls used were sections of human tonsil for T cells and skin for fibroblasts. Two negative controls were included, omitting the first layer to detect background staining and using an isotype mAb (a mouse IgG2a, M9144, Sigma, Dorset) as an isotype control for the RFT mix. The isotype control for the fibroblast staining was the RFT mix.

The biotin/streptavidin/alkaline phosphatase technique is a more sensitive technique and was selected to identify IFNβ in conjunctival tissue. Sections were incubated for 18 hours with 50 μl of the mAb against IFNβ (MAS291, Harlan Sera-Lab, Loughborough) at a pretitrated concentration diluted in PBS+0.1% BSA. After a wash in TRIS buffered saline (TBS), the sections were incubated with 50 μl of the horse anti-mouse biotinylated second layer (BA-2000, Vector Laboratories, Peterborough) diluted 1/100 in PBS+0.1%BSA for 1 hour at room temperature. The sections were then incubated with 50 μl of the streptavidin-alkaline phosphatase third layer (SA-5100, Vector Laboratories) diluted 1/100 in PBS+0.1% BSA for 1 hour at room temperature. The colour reaction was developed by the addition of 50 μl of the substrate for 15 minutes (a combination of 0.005 g naphthol ASBI phosphate, 10 ml TRIS-HCl (pH 8.2), 200 μl dimethylformamide, 0.01g fast red and 10 drops levamisole). The sections were counterstained with Mayer's haematoxylin (Sigma, Dorset), before being finally mounted in PBS:glycerol (9:1). The positive control used was sections of eczematous skin. The negative controls were omitting the first layer to detect background staining and an isotype mAb (IgG2a, M9144, Sigma, Dorset).

Quantification of immunohistology

An image analysis system (Seescan Imaging, Cambridge, magnification ×320) was used to count the number of T cells per defined frame, in three random areas per section. This allowed the number of cells per unit area (UA) to be calculated. A semiquantitative grading system was used to measure the fibroblast and IFNβ staining. This was based on a grading system ranging from 1+ to 4+. An experienced masked observer performed the grading to reduce bias. Fibroblast grading 1+ represented scattered cells, 2+ represented more fibroblasts located especially beneath the epithelium, 3+ represented extensive cells staining positive. IFNβ grading 1+ represented interepithelial staining, 2+ greater interepithelial staining, 3+ represented fibroblast-like cells staining in the stroma, 4+ represented more cells staining in the stroma. The number of IFNβ and AS02 positive fibroblasts was counted using a light microscope with a counting grid (×40 magnification) in three separate fields of view. The Mann-Whitney U test was used to analyse the numeric data and Fisher's exact test was used to analyse the quantitative data.

RESULTS

A total of 17 glaucoma patients were recruited of whom six underwent repeat filtration surgery. All had a diagnosis of primary open angle glaucoma, except for two with chronic angle closure glaucoma and three with pigment dispersion glaucoma. All were white except for five who were of Afro-Caribbean ethnicity and one of Asian ethnicity. The mean age was 68.3 years (range 45–87). The mean cumulative number of months of glaucoma eyedrop use was 154.5 months (range 9–356). Table 1 summarises the demographics of our patients.

Fibroblast and T cell staining

The conjunctiva of all glaucoma patients, except for two, contained spindle-shaped cells staining positive for the fibroblast marker AS02 (Fig 1A and B). These cells were identified as Tenon's fibroblasts on the basis of their cell shape and positive AS02 staining. It was noted that many of them were located beneath the conjunctival epithelium in Tenon's layer. Again, the conjunctiva of all of our glaucoma patients, except for one, showed RFT+ T cell staining (Fig 1C and D). Positive staining cells were located immediately beneath the conjunctival epithelium and in Tenon's layer. The biopsy sample of one patient was crushed and therefore was discounted from the analysis.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

Human conjunctiva of glaucoma patients contains IFNβ producing fibroblasts and T lymphocytes. The indirect immunoperoxidase technique was used to stain human conjunctiva for fibroblasts and T cells using the fibroblast specific mAb AS02 and a pan-anti-T IgG mAb mix. Numerous spindle-shaped cells staining positive for AS02 were identified in Tenon's layer (A, red arrow). The same conjunctival biopsy also showed RFT+ T cell staining (C, blue arrow). The negative isotype mAb controls did not stain positive (B and D). The biotin/streptavidin/alkaline phosphatase technique was used to stain human conjunctiva for IFNβ using the mAb MAS291. Spindle-shaped fibroblast-like cells staining positive for IFNβ were identified in Tenon's layer, although not all AS02 positive fibroblasts stained positive for IFNβ (E, black arrow). The negative isotype mAb control did not stain for IFNβ (F). Scale bar = 150 μm.

Of the two patients described as negative for fibroblast staining, one patient showed diffuse AS02 positive staining, which was not fibroblast-like in distribution and therefore, was not included in the analysis. Other cells that stained positive but were not spindle-shaped were also not included in the quantitative analysis because these cells could not be definitely identified as fibroblasts. In addition, AS02 staining was noted in the endothelial cells of blood vessels in the conjunctiva. The positive controls tonsil and skin stained for T cells and fibroblasts (data not shown).

IFNβ staining

Spindle-shaped fibroblast cells located in Tenon's layer beneath the epithelium stained positive for IFNβ in the conjunctiva of three of our patients (two high risk and one low risk) (Fig 1E and F). It was noted that not all the AS02 positive fibroblasts stained positive for IFNβ. In addition, IFNβ staining was found in the conjunctival epithelium of all of our glaucoma patients. This staining was especially strong in the outer layers and less strong in the inner layers of the epithelium. There were large mononuclear cells, which stained positive for IFNβ. However, these cells were not included in the grading assessment because they were not fibroblast-like in shape. There were two technical failures in the low risk group that were not included in the analysis. Also, the blood vessel endothelial cells that had stained positive for the AS02 fibroblast marker did not stain positive for IFNβ.

Quantitative assessment of immunohistology

An image analysis system was used to count the number of T cells per defined frame. Three random areas per section were selected to enable calculation of the number of cells per unit area (UA). The median RFT cell number per unit area (RFT/UA) was 3.15. Our grading system indicated that the median fibroblast grade was 1 and the median IFNβ staining grade was 2. The mean IFNβ positive cell count was 1.97 (SEM 0.20) compared to the mean AS02 positive cell count of 30.53 (SEM 9.22). Therefore, 6.45% of the fibroblast cell population in these samples appeared to be producing IFNβ. Table 1 summarises the results of the RFT staining and the grading for fibroblast and IFNβ staining.

We divided our patients into those patients who underwent filtration for the first time (low risk group) and those who underwent repeat surgery (high risk group). The median RFT cell number per unit area (RFT/UA) for the high risk for scarring group was 1.93. The median RFT cell number per unit area for the low risk for scarring group was 4.28. After taking into account differences between the individual patients, there was no significant difference between the RFT cell number per unit area between the two groups (p=0.60). Again, there were no statistical differences in fibroblast and IFNβ grading between the two groups (p = 0.44 and p = 0.3 respectively).

DISCUSSION

The aim of this study was to investigate whether the components for fibroblast mediated prevention of T cell apoptosis were present in vivo in human conjunctiva. This is the first report of the in vivo production of IFNβ by HTF, staining positive in the conjunctiva of three of our glaucoma patients. In addition, the conjunctiva of our glaucoma patients contained T lymphocytes. Therefore, our study has that all the components for this particular fibroblast/T cell interaction were present in human conjunctiva.

T cell/fibroblast interactions appear to have an important role in the development of abnormal scarring.² Hitchings and Grierson showed that early trabeculectomy failures were associated with excessive inflammation and increased conjunctival fibroblast numbers.¹⁴ Nuzzi et al demonstrated that the preoperative conjunctiva of repeat filtration surgery patients contained an increased number of activated T lymphocytes and fibroblasts.¹⁵ In our study, T lymphocytes and IFNβ producing fibroblasts were present in the conjunctiva of two of our repeat trabeculectomy patients, who were a mean of 51.6 months (range 4–90) following their first failed trabeculectomy. This could suggest that conjunctival fibroblast mediated prevention of T cell apoptosis could continue for a prolonged period of time and thereby, contribute to the development of late as well as early failures.

Broadway et al showed that topical treatment altered the conjunctival cellular profile of glaucoma patients, with multiple therapy or treatment for more than 3 years resulting in an increase in lymphocytes, macrophages, and fibroblasts.¹⁶ Baudouin et al demonstrated that the preservative used in glaucoma drops was enough to stimulate conjunctival inflammation.¹⁷ All the patients in our group had received glaucoma eye drops as part of their medical treatment before undergoing filtration surgery. At present, the factors influencing IFNβ production are not fully known, although our previous research did indicate that mitomycin C treatment (using a concentration of 0.4 mg/ml) of in vitro HTF increased HTF IFNβ production.¹² Therefore, the medical treatment of our patients may have had some effect on HTF IFNβ production. In our study, another control group would have been to investigate for the presence of IFNβ producing fibroblasts in untreated conjunctiva; however, with current medical practice, we were unable to obtain ethical approval for biopsies of this type of tissue.

The interferons are a heterogenous group of cytokines consisting of the type I interferons (interferon α and β) and type II interferon (interferon γ).^18,19 The fibroblast is a common source of IFNβ, and high levels are produced when fibroblasts are stimulated by viruses or agents such as polyinosinic:polycytidylic acid that mimic viral infections.²⁰ Our previous research showed that several types of fibroblasts including in vitro HTF produced IFNβ constitutively.¹² IFNβ is structurally dissimilar to IFNα but signals through the same receptor.

The interferons have long been recognised for their antiproliferative effects during viral infections.²¹ More recently, they have been shown to be able to reduce the wound healing response; Latina et al and Nguyen et al showed that IFNγ inhibited HTF collagen synthesis in vitro.^22,23 Gillies et al showed that intraoperative and postoperative subconjunctival injections of IFNγ reduced HTF proliferation after glaucoma filtration surgery, producing a success rate of 79% after 2 years follow up, similar to the success rate of 89% when the antimetabolite 5-fluorouracil was used.^24,25

Our study has demonstrated the presence of the components of the interaction fibroblast-mediated prevention of T cell apoptosis in human conjunctiva; however, we did not demonstrate that this interaction was actually occurring. Our study demonstrated the presence of IFNβ producing HTF in only three of our glaucoma patients. In addition, only 7.4% of the fibroblast population appeared to be producing IFNβ. This was not unexpected because IFNβ would not necessarily be produced by all fibroblasts at the same time or at the time of collection. Also, we were not able to demonstrate differences in the staining between glaucoma patients of different risk for scarring. This may have been because the number of patients studied was quite small. Furthermore, the presence of different T cell subgroups was not investigated, although our in vitro research showed that fibroblast-mediated prevention of T cell apoptosis could be produced in both CD4+ and CD8+ T lymphocytes.²⁶

In conclusion, this study has identified the in vivo presence of the components of the interaction fibroblast mediated prevention of T cell apoptosis, indicating the possibility of a new potential role for HTF produced IFNβ in the development of an aggressive wound healing reaction. The environmental milieu is crucial in directing the progression of a wound healing reaction and given the appropriate conditions, this particular interaction could promote the development of chronic conjunctival inflammation. This could subsequently promote greater fibroblast activity and ultimately excessive scar tissue formation. HTF IFNβ mediated prevention of T cell apoptosis could have an important role in the pathogenesis of late surgical failures. Therefore, it will be important to further our understanding of this particular interaction with the aim of developing further strategies to control the wound healing reaction and improve our surgical success rate.