Changes in Matrix Metalloproteinase Network in a Spontaneous Autoimmune Uveitis Model

Abstract

Purpose.: Autoimmune uveitis is a sight-threatening disease in which autoreactive T cells cross the blood-retinal barrier. Molecular mechanisms contributing to the loss of eye immune privilege in this autoimmune disease are not well understood. In this study, the authors investigated the changes in the matrix metalloproteinase network in spontaneous uveitis.

Methods.: Matrix metalloproteinase (MMP) MMP2, MMP9, and MMP14 expression and tissue inhibitor of metalloproteinase (TIMP)-2 and lipocalin 2 (LCN2) expression were analyzed using Western blot quantification. Enzyme activities were examined with zymography. Expression patterns of network candidates were revealed with immunohistochemistry, comparing physiological appearance and changes in a spontaneous recurrent uveitis model.

Results.: TIMP2 protein expression was found to be decreased in both the vitreous and the retina of a spontaneous model for autoimmune uveitis (equine recurrent uveitis [ERU]), and TIMP2 activity was significantly reduced in ERU vitreous. Functionally associated MMPs such as MMP2, MMP14, and MMP9 were found to show altered or shifted expression and activity. Although MMP2 decreased in ERU vitreous, MMP9 expression and activity were found to be increased. These changes were reflected by profound changes within uveitic target tissue, where TIMP2, MMP9, and MMP14 decreased in expression, whereas MMP2 displayed a shifted expression pattern. LCN2, a potential stabilizer of MMP9, was found prominently expressed in equine healthy retina and displayed notable changes in expression patterns accompanied by significant upregulation in autoimmune conditions. Invading cells expressed MMP9 and LCN2.

Conclusions.: This study implicates a dysregulation or a change in functional protein-protein interactions in this TIMP2-associated protein network, together with altered expression of functionally related MMPs.

Equine recurrent uveitis (ERU) is a spontaneous autoimmune disease in horses that immunopathologically and clinically resembles human autoimmune uveitis. ^1,2 Characterized by repeated episodes of ocular inflammation mediated by CD4⁺ T cells targeting retinal antigens, it eventually leads to blindness. ^{2 –5} Most important, ERU is the only spontaneous animal model for human autoimmune uveitis ^{1,2,5 –7} and thus provides the advantage of studying underlying pathomechanisms of a recurrent autoimmune disease on a variable genetic background, ^3,8 as is the case in human autoimmune uveitis. The transfer value of ERU as a model for human autoimmune uveitis was already proven because CRALBP, an autoantigen first discovered in the horse model, ⁵ was found to be a frequently targeted autoantigen among humans with uveitis. ⁹  

In one of our previous studies comparing the vitreous of healthy and ERU-affected eyes with the use of two-dimensional gel electrophoresis, we identified tissue inhibitor of metalloproteinases (TIMP)-2 in healthy vitreous samples. ⁷  

In general, TIMPs are known as modulators of matrix metalloproteinase (MMP) activity. ¹⁰ This is important because MMPs play a major role in the degradation of extracellular matrix components and expand their function to modify cytokines, protease inhibitors, and cell surface signaling systems. ^{11 –14} Further, migrating immune cells express MMP2 and MM9 to overcome blood tissue barriers. ^{15 –17} In autoimmune diseases such as experimental autoimmune neuritis, MMP2- and MMP9-mediated protein cleavage is involved in disease. ¹⁸ Because of this modulation of cell-cell or cell-ECM interactions, MMPs and their inhibitors, most notably the TIMPs, inherit crucial relevance in physiological and pathologic mechanisms. ^11,13  

Representing a unique member of the TIMP family, TIMP2 is not only required for MMP2 inhibition, it also mediates pro-MMP2 activation. ^{19 –21} Besides its interaction with MMPs, TIMP2 is known to function in the inhibition of angiogenesis ²² and in neuroprotection. ²³  

Given that we found TIMP2 to be downregulated in ERU vitreous through differential proteome analysis, the objective of the present study was to evaluate TIMP2 expression in the retina, the target tissue of uveitis. Interactions of TIMP2 and MMP14 in the process of modulating MMP2 activity ²⁴ and amelioration of experimental autoimmune uveitis by selective MMP2 and MMP9 inhibition ^{15 –17} were reported. This prompted us to investigate these TIMP2 interactors in the MMP network to gain further insight into TIMP2 involvement in the pathophysiology of spontaneous recurrent autoimmune uveitis. 

Methods

Vitreous Specimen

For this study, a total of 85 vitreous specimens were sampled and processed. Forty-six were derived from healthy eyes, and 39 were derived from eyes diagnosed with ERU. ⁴ Vitreous samples of control horses were obtained from horses euthanatized because of incurable and ERU-unrelated disease; vitreous samples of ERU cases were obtained during therapeutic pars plana vitrectomy. Vitreous samples were stabilized with EDTA-free protease inhibitor (Roche, Mannheim, Germany), then lyophilized, solubilized in ultrapure water, and dialyzed against a 50-mM phosphate buffer (ph 7.6). Protein content was determined using the Bradford assay (Sigma, Deisenhofen, Germany). All animals were treated according to the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research. All horses were presented as patients to the Equine Clinic of LMU Munich; no experimental animals were used in this study. 

Western Blot Analysis

For Western blot analysis, the sample input onto the gels was normalized to equal total protein loading. Vitreous samples were resolved by 8% or 12% SDS-PAGE and blotted semidry on polyvinylidene difluoride membranes (GE Healthcare, Freiburg, Germany). Unspecific binding was blocked for 1 hour with 1% polyvinylpyrrolidone (PVP) in PBS-T. Blots were incubated with primary antibodies at 4°C overnight (monoclonal mouse antibody specific for human MMP2 with reported cross-reactivity to horse MMP2 [Millipore, Schwalbach, Germany; 1:500]), rabbit antibody specific for human MMP9 (Enzo Life Sciences, Lörrach, Germany; 1:1000) or monoclonal mouse antibody specific for TIMP2 (Biozol, Erding, Germany; 1:500), washed, and incubated with horseradish peroxidase-coupled secondary anti-mouse IgG antibody (Sigma-Aldrich, Deisenhofen, Germany; 1:5000) or anti rabbit IgG (Sigma-Aldrich; 1:3000), respectively. Signals were detected with enhanced chemoluminescence on x-ray film (Fuji; Christiansen, Planegg, Germany). 

Films were scanned on a transmission scanner, and quantification of Western blot signals by densitometry was performed (ImageQuantTL software; GE Healthcare). TIMP2 and MMP2 abundances between ERU cases and controls were statistically analyzed using the Mann–Whitney U test. The differences were considered significant if P < 0.05. 

Gelatin Zymography

Gelatinolytic activity of MMP2 and MMP9 in vitreous samples was determined using gelatin zymography. Ten micrograms of vitreous protein per specimen were resolved by 8% SDS-PAGE containing gelatin (1 mg/mL; Sigma-Aldrich) under denaturing, but nonreducing, conditions. Gels were washed in 2.5% Triton X-100 (Sigma-Aldrich) for 45 minutes at room temperature, then transferred to zymogram developing buffer, pH 7.6, containing 50 mM Tris, 0.2 M NaCl, 5 mM CaCl₂ (all from Applichem, Darmstadt, Germany), and 0.02% surfactant (Brij 35; Sigma-Aldrich) and incubated for 30 minutes. The developing buffer was renewed, and gels were incubated at 37°C for 18 hours. Gels were stained with Coomassie dye (Coomassie Blue R250; Sigma-Aldrich). ²⁵ Gelatinolytic activity was detected as transparent bands on a blue background of undegraded substrate. Control gels were incubated in the presence of 50 mM EDTA or 2 mM phenylmethylsulfonyl fluoride (PMSF) to ensure the MMP specificity of the observed gelatinolytic bands. Recombinant human pro-MMP2 and pro-MMP9 (both from ProteaImmune GmbH, Berlin, Germany) were used as positive controls. 

To compare MMP activity between controls and ERU samples, gels were scanned and inverted to give black bands on a light background. Bands were analyzed by densitometry, as already described for Western blot. The mean densitometry software value of all control samples on one gel was set at 100% and put into relation to the measured values of the ERU samples run on the same gel. Pro-MMP9 and pro-MMP2 abundances between ERU cases and controls were statistically analyzed using the Mann-Whitney U test. The differences were considered significant if P < 0.05. 

Detection of TIMP2 by Reverse Zymography

Vitreous samples were separated on a 12% SDS-gel containing 1 mg/mL gelatin and 0.8 μg recombinant MMP2. Ten micrograms of protein per sample was loaded onto the gel. Gels were washed in 2.5% Triton X-100 for 45 minutes at room temperature and afterward incubated in a zymogram developing buffer for 30 minutes. The developing buffer was renewed, and the gel was incubated at 37°C for 18 hours, followed by staining with Coomassie dye. TIMP inhibitory activity appeared as dark bands against a lighter background. TIMP2 was identified at a molecular weight previously described for equine TIMP2 ²⁶ by comparison with a molecular weight marker (PageRuler Prestained Protein Ladder; Fermentas, St. Leon-Rot, Germany). To ensure the specificity of the reverse zymography for TIMP activity detection, vitreous samples were electrophoresed on a 12% SDS-gel without substrate. 

Immunohistochemistry

Paraffin-embedded ERU (n = 15) and healthy retinal tissue (n = 10) samples were sectioned at 8 μm. Histopathologic changes were evaluated on sections stained with hematoxylin and eosin. For immunohistochemistry, heat antigen retrieval was performed at 99°C for 15 minutes in 0.1 M EDTA-NaOH buffer (ph 8.8). Monoclonal mouse antibody specific for MMP2 (Millipore, Schwalbach, Germany; 1:100) and specific for TIMP2 (Biozol, Erding, Germany; 1:25), polyclonal rabbit antibody specific for MMP-9 (Enzo Life Sciences, Lörrach, Germany; 1:100), polyclonal rabbit antibody specific for human lipocalin 2 (LCN2; Abcam, Cambridge, UK; 1:500), and polyclonal rabbit antibody for human MMP-14 (MT-MMP1; Santa Cruz Biotechnology, Heidelberg, Germany; 1:100) were used for candidate detection in tissue. For fluorescence labeling, anti-mouse IgG antibody coupled to Alexa488 or Alexa568, as well as anti-rabbit IgG coupled to Alexa488 or Alexa568 or Alexa 647 (all from Invitrogen, Karlsruhe, Germany; 1:500), were used, respectively. Cell nuclei were counterstained with DAPI (Invitrogen, Karlsruhe, Germany; 1:1000). Multiple labeling was performed consecutively, with blocking steps (ProteinBlock; Dako, Hamburg, Germany) between single antibody incubations. Fluorescence staining was photographed (Axio Imager M1; Zeiss, Göttingen, Germany) and visualized with digital microscopy software (Axio Vision 4.6; Zeiss). 

Results

TIMP2 Activity Is Significantly Reduced in ERU Vitreous

MMPs are inhibited by specific endogenous TIMPs, a group of peptidases involved in degradation of the extracellular matrix. TIMP2 is constitutively expressed in healthy horse vitreous. ⁷ TIMP2 has a unique role among TIMP family members because of its ability to directly suppress the proliferation of endothelial cells. As a result, the encoded protein may be critical to the maintenance of tissue homeostasis by suppressing the proliferation of quiescent tissues in response to angiogenic factors and by inhibiting protease activity in tissues undergoing remodeling of the extracellular matrix. Because of these unique functions, we were interested in changes in TIMP2 expression in association with spontaneous uveitis. Expression of TIMP2 was determined in healthy controls (Fig. 1A, left) and compared with uveitic vitreous specimens (Fig. 1A, right). A uniform, significant, downregulation of TIMP2 protein expression to 5.7% (±6.3%) was determined in ERU cases. Next, we analyzed TIMP activity by reverse zymography (Fig. 1B). Gelatin degradation was inhibited only at 21 kDa in all vitreous samples tested, indicating only TIMP2 to be active in vitreous (Fig. 1B, blue band). In uveitis cases, TIMP2 activity was reduced, which was evident in a markedly decreased area of substrate protected by vitreous TIMP2 activity in uveitic samples when compared with controls (Fig. 1B). In addition, in the upper part of the gel vitreous-derived MMP activity appeared. Gelatin digestion was obvious at only one band size in control vitreous (Fig. 1B, white band at 70 kDa) but at one additional band exclusively in ERU cases (Fig. 1B, bands at 70 kDa in ERU cases and additional band at 95 kDa). 

Figure 1.

 

Detection of TIMP2 in vitreous samples. (A) TIMP2-specific Western blot signal, quantified by densitometry (***P < 0.001). 94.3% decrease in TIMP2 expression in ERU vitreous (n = 16) compared with control samples (n = 19). Insets above columns: representative Western blot bands; left: controls; right: ERU. (B) Reverse zymography gel demonstrating decreased TIMP2 activity in uveitic vitreous compared with control samples.

MMP Activity Changes in Autoimmune Condition

Because TIMP2 inhibits MMPs, especially MMP2, we next evaluated MMP activity in the vitreous of controls and ERU cases using gelatin zymography (Fig. 2A, representative zymogram). Pro-MMP2 activity was detected in 78 of 79 vitreous specimens but with higher activity in controls (Fig. 2A, lanes 1–4) than in uveitic specimens (Fig. 2A, lanes 5–8). Gelatinolytic activity of MMP2 was reduced in ERU cases to 70% (±43%) of physiological pro-MMP2 values (Fig. 2B). Compared with healthy vitreous, only 11% (±4%) of MMP2 protein was still expressed in autoimmune uveitis, as revealed by Western blot analysis (Fig. 2C). Further, MMP9 gelatinolytic activity was evident in 76% of ERU vitreous samples, but was mostly undetectable in controls (Fig. 2D). MMP9 activity was 12.65- (±20.48)-fold increased in diseased eyes (Fig. 2D). Accordingly, MMP9 protein expression increased 17-fold (16.96 ± 14.66) in ERU vitreous (Fig. 2E). 

Figure 2.

 

Analysis of MMP activity in vitreous samples. (A) Representative gelatin zymography of vitreous samples. Lanes 5–8: detection of pro-MMP9 gelatinolytic activity as clear band against blue background at 95 kDa in ERU. Lane 7: active MMP9 at 85 kDa. Lane 10: pro-MMP9 standard. Lanes 1–8: detected pro-MMP2 gelatinolytic activity at 70 kDa in ERU and control samples. Lanes 1, 3, 4, 7: active MMP2 at 62 kDa. Lane 9: pro-MMP2 standard. Proof of MMP-dependent gelatinase activity in vitreous samples. Activity inhibited by EDTA (lanes 11–13) but not by PMSF (lanes 14–16). (B) 30% decline in observed pro-MMP2 gelatinolytic activity in ERU vitreous samples compared with controls and quantification of gelatinolytic bands by densitometry (**P < 0.01). (C) Reduced pro-MMP2 expression to an average of 11% in ERU vitreous samples and quantification of Western blot signal by densitometry. Insets: representative Western blot signals. (D) Average 12.65-fold increase of pro-MMP9 gelatinolytic activity in ERU vitreous samples and quantification of gelatinolytic band by densitometry (***P < 0.001). (E) Average 16.96-fold increase of MMP9 protein expression in ERU vitreous as detected by Western blot analysis (*P < 0.05).

MMP Network Changes Expression in Target Tissue during Autoimmune Uveitis

Because TIMPs have the capacity to inhibit protease activity in tissues undergoing remodeling of the extracellular matrix, we next evaluated the expression patterns of TIMP2 and MMP2 in uveitis target tissue, the retina itself. Differential interference contrast (DIC) image of the normal equine retina (Fig. 3A) compared with the uveitic retina (Fig. 3B) revealed destruction of the retinal structure in uveitic eyes. TIMP2 (Fig. 3C, red) and MMP2 (Fig. 3E, green) are expressed at photoreceptor inner and outer segments of normal equine retinas, which is consistent with the expression described in other species. ¹² TIMP2 expression almost disappears in the ERU condition (Fig. 3D), whereas MMP2 is strongly upregulated in ERU (Fig. 3F). Further, MMP2 expression pattern changes from one restricted to photoreceptor inner segments and weak expression in the outer limiting membrane (ONL) to strong expression at the ONL and additional dispersed expression throughout retinal tissue (Fig. 3F). MMP2 seems also to be coexpressed at cell nuclei in the ERU diseased retina (Fig. 3F; turquoise staining pattern at DAPI (blue)-stained cell nuclei, MMP2 expression is green). 

Figure 3.

 

Shift in retinal expression of TIMP2 and functional associated proteins in uveitis. DIC image of a healthy retina (A) compared with a uveitic retina (B). TIMP2 expression (red) in healthy retina showing an accumulation at the photoreceptor inner and outer segments (C) compared with reduced TIMP2 expression (red) in ERU-affected retinal tissue (D). Expression of MMP2 (green) in healthy retina, where the main immunofluorescence signal is visible at photoreceptor inner and outer segments (E). MMP2 expression pattern (green) is shifted in the uveitic state toward an enhanced signal at the outer limiting membrane and in the outer nuclear layer (F). MMP14 (magenta) is predominantly detected at the inner photoreceptor segments of healthy eyes (G). Decrease in MM14 expression in ERU (H). TIMP2 (red), MMP2 (green), and MMP14 (magenta) are coexpressed at the inner photoreceptor segments of healthy eyes (I, overlay, white), whereas in the uveitic state MMP2 is predominant (J; TIMP2, red; MMP2, green; MMP14, magenta; overlay, white). Cell nuclei are stained with DAPI (C–J, blue).

TIMP2 has been also described as a positive regulator of MMP14 by promoting the availability of the enzyme at the cell surface and supporting pericellular proteolysis after forming the trimolecular complex of MMP14, TIMP2, and pro-MMP2. Hence, we also examined MMP14 expression in healthy and uveitic retinal tissue. MMP14 is similar to MMP2 expressed at photoreceptor inner segments in physiological conditions (Fig. 3G, magenta) but is undetectable in the autoimmune diseased retina (Fig. 3H). Therefore, in ERU, only an upregulated and changed expression pattern of MMP2 remains, whereas expression of the other interactors of this network, TIMP2 and MMP14, is no longer evident in ERU target tissue (Fig. 3I; triple staining, overlay of expression of TIMP2, MMP2, and MMP14 results in white staining; Fig. 3J, only MMP2 (green) expression remains in ERU). 

Coexpression of MMP9 and LCN2 Is Limited to Infiltrating Cells in ERU

Considerable MMP9 activity was detected in vitreous of spontaneous uveitis cases; therefore, we next evaluated MMP9 expression pattern in the equine retina (Figs. 4A, 4C, hematoxylin and eosin staining and DIC image of healthy control retina; 4B, 4D, ERU retina). Among the MMP family, MMP9 is the most important angiogenic factor. ²⁷ MMP9 was found to be expressed in ganglion cell layer, inner plexiform layer, and photoreceptor segments in physiological conditions of the equine retina (Fig. 4E, red). In contrast, MMP9 was no longer expressed in uveitic retinal tissue itself (Fig. 4F); rather, infiltrating cells stained positive for MMP9 (Fig. 4F). 

Figure 4.

 

Changes in MMP9 and LCN2 expression patterns in uveitic retina. Hematoxylin and eosin staining of a healthy retina (A) compared with a retina affected by uveitis (B). The same specimen but different sections (DIC images: C, healthy; D, uveitis) is shown stained with antibodies against MMP9 and LCN2. Predominant expression of MMP9 (red) in photoreceptor segments of healthy retina (E). Uveitic retina showing reduced MMP9 expression (red) confined to well-defined areas (F). LCN2 expression (green) in healthy retina (G) compared with shifted LCN2 expression pattern in ERU-affected retinal tissue (H). Double staining of LCN2 (green) and MMP9 (red), detection of overlay in the inner plexiform layer and at the outer limiting membrane of healthy retina (I, overlay, yellow). Coexpression of both proteins at the site of the remaining MMP9 expression in the retina affected by uveitis (J, overlay, yellow). Cell nuclei are stained with DAPI (E–J, blue). Insets: enlargements of representative cells reflecting the staining pattern of large infiltrating cells in ERU.

We added analysis of LCN2 expression because both proteins strongly interact and in vitro data suggest a possible role for LCN2 in the protection of MMP9 against autolysis. ²⁸ Further, we anticipated a role of LCN2 in autoimmune uveitis because in mouse models of retinal degeneration Müller glia cells were reported to express LCN2 in response to photoreceptor damage. ²⁹ Immunohistochemistry revealed LCN2 expression in healthy retina in the nerve fiber layer, the inner plexiform layer, the outer nuclear layer, and the outer limiting membrane (Fig. 4G, green). In contrast, LCN2 was markedly upregulated in ERU (Fig. 4H), and expression patterns shifted toward diffuse LCN2 expression throughout whole retinal tissue in the uveitic state and infiltrating cells (Fig. 4H). MMP9 and LCN2 were coexpressed only in the inner plexiform layer and at the outer limiting membrane (Fig. 4I) in physiological conditions. Only LCN2 expression was left in ERU retinas (Fig. 4H), but MMP9 coexpression occurred in infiltrating cells (Fig. 4J; LCN2, green; overlay of MMP9 and LCN2, yellow). 

Discussion

TIMPs play an important role in physiological and pathologic mechanisms because they influence matrix metalloproteinases, which are known to modulate cell-cell or cell-ECM interactions. ^11,13,20 The results of this study demonstrate a notable decrease in TIMP2 expression and an altered expression of several members of the MMP family in target tissues of spontaneous recurrent uveitis. 

TIMP2 was first identified in healthy horse vitreous in an initial proteomic experiment, analyzing vitreous proteins changed in ERU by two-dimensional gel electrophoresis. ⁷ In the present study, we further investigated the expression of TIMP2 in eyes from horses affected by spontaneous recurrent uveitis compared with controls. Reduced TIMP2 expression in ERU could be quantified to 94.3% (±6.31%), revealing a significant decrease in uveitic vitreous compared with controls (Fig. 1A). Subsequently, decreased TIMP2 activity in an ERU vitreous specimen was confirmed through reverse zymography (Fig. 1B). Additionally, significant reduction of TIMP2 expression could be confirmed for the target tissue of recurrent uveitis, the retina (Fig. 3D). Despite a report of increased TIMP2 expression in the aqueous humor of uveitis-related secondary glaucoma, ³⁰ to our knowledge no one has yet monitored changes in ocular TIMP2 expression in autoimmune uveitis, though TIMP2 occurrence has previously been reported in healthy human retina ^31,32 and vitreous. ³³  

In ERU, activated peripheral T lymphocytes cross the blood retinal barrier, targeting autoantigens in the retina. ² TIMP2 has not only been assigned neuroprotective properties, ²³ it was also reported to be capable of preventing in vitro migration of CD3⁺ T cells derived from a diabetic mouse model through an endothelial cell layer. ³⁴ In the same study, TIMP2 treatment also restored insulin levels in an in vitro organ culture model for type 1 diabetes mellitus, which the authors suggested was caused by the potential TIMP2 capability of preventing autoreactive T cells from destroying insulin-producing β cells. ³⁴ Therefore, the significant reduction of TIMP2 in spontaneous recurrent uveitis target tissue is an important finding because it might contribute to disease pathogenesis, either by loss of the eye's immune privilege or by lack of neuroprotection. 

One major effect of TIMP2 is its ability to modulate MMP2 activity, either supporting MMP2 activation or inhibiting this protease in a dose-dependant manner. ^19,35 This might be an important function in the context of autoimmune uveitis because MMP inhibition by the synthetic inhibitor BB-1101 reduced the incidence of S-Ag–induced experimental uveoretinitis in Lewis rats, ³⁶ and the specific inhibition of MMP2 and MMP9 ameliorated IRBP-induced experimental autoimmune uveitis in mice. ¹⁷ Interestingly, elevated MMP2 and MMP9 levels have been described in aqueous humor samples of patients with active uveitis. ^{37 –39} Despite these reports focusing on aqueous humor expression, little is known about the expression of MMP2 or MMP9 in uveitic vitreous. 

In this context, the observed decrease of pro-MMP2 gelatinolytic activity and protein expression, combined with a very notable increase in MMP9 gelatinolytic activity in the vitreous of horses with spontaneous uveitis (Fig. 2), is an important finding. Given that MMP2 is known to be a modulator of chemokines, as is the case for monocyte chemoattractant protein D3, which, after cleavage by MMP2, acts as a general chemokine antagonist, ⁴⁰ MMP2 might inherit an anti-inflammatory function in healthy vitreous. Therefore, decreases in MMP2 might favor chemotactic attraction of T cells or macrophages, the predominant cells infiltrating the eyes in ERU. ^2,41  

MMP9 is the most important angiogenic factor of the MMP family, ²⁷ switching vessels from vascular quiescence to angiogenesis by rendering VEGF more available to its receptors. Further, in vitro chemical hypoxia of cultured human RPE cells led to increases in VEGF gene expression and secretion of MMP9. ⁴² This hypoxic expression of MMP9 was mediated by autocrine VEGF signaling. We have already demonstrated a role for VEGF in ERU retinas ⁷ ; therefore, we were interested in unraveling the meaning of MMP9 upregulation in ERU vitreous. Surprisingly, MMP9 expression in the uveitic retina was decreased (Fig. 4F), indicating that this might not have been the source of the elevated MMP9 level in uveitic vitreous. T cells, preferentially Th1 cells, express MMP9 and MMP2 to facilitate cell migration, as is known from in vitro studies on human T cells derived from patients with multiple sclerosis. ⁴³ Further, MMP9 is secreted by macrophages, ¹⁵ and inflammatory cells were thought to be a major provider of MMP9 in experimental choroidal neovascularization of mice. ⁴⁴ As mentioned earlier, both cell types are the major infiltrating cells in ERU retinas. ^2,41 They are, therefore, likely to contribute to the alteration of MMP9 expression during ERU, especially because the only remaining MMP9 expression in the uveitic retina was detected on infiltrating cells (Fig. 4F). In a mouse model of choroidal neovascularization, neutrophil granulocytes were demonstrated to contain MMP9 in intracellular granules that were recruited to the neovascular area. ⁴⁴ This could have been the case in ERU retinas (Fig. 4F insert; high resolution of invading cell shows MMP9-positive granules); hence, these cells expressing MMP9 in ERU should be further characterized in additional studies. 

In the healthy human retina, MMP2 is reported to be expressed in ganglion cell bodies and their axons. ³² An increase in retinal MMP2 expression is known from mouse models of diabetic retinopathy, ⁴⁵ but to date there are no reports on MMP2 expression changes in the retina caused by autoimmune uveitis. Retinas of horses affected by ERU showed notably elevated MMP2 expression and altered expression patterns (Fig. 3F), possibly because of the active involvement of MMP2 in the destruction of retinal tissue or at least in the inflammatory process, especially considering the different modulatory capabilities of MMPs. ^11,13,14 Another member of the MMP family, MMP14, which has been detected by Western blot analysis in healthy human retinas, ⁴⁶ was expressed at the inner photoreceptor segments of healthy horse eyes (Fig. 3G), but expression decreased notably in retinas affected by ERU (Fig. 3H). 

Additionally, we could for the first time demonstrate the coexpression of TIMP2, MMP2, and matrix metalloproteinase-14 (MMP14) in retinas under physiological conditions (Fig. 3I). Because TIMP2 is an important modulator of MMP2 activity ^19,20 and this effect is also reported to involve interaction with MMP14, ²⁴ coexpression of these three proteins in healthy retinas might point to an involvement in maintaining tissue integrity by controlling MMP2 activation. In all ERU cases investigated, TIMP2 and MMP14 expression was almost undetectable, whereas MMP2 expression increased and presented changes in protein expression patterns (Fig. 3J). It has been shown in cell culture experiments that the phosphorylation status of MMP2 significantly affects its activity and probably also its substrate specificity. ⁴⁷ A possible role of caveolin 1 in the modification of MMP2 at the cell membrane was discussed, because MMP2 was phosphorylated by protein kinase C (PKC) in vitro. ⁴⁷ PKC localizes to caveolin-1 at the cell membrane, where MMP2 has been found to associate with caveolin-1, ⁴⁸ suggesting a possible role for this kinase in the phosphorylation of secreted MMP2. Interestingly, we were able recently to demonstrate the increased expression of caveolin-1 at retinal membranes of ERU cases. ⁴⁹ Therefore, we suggest further in-depth investigations of MMP pathway and protein modifications of participating candidates in the future for greater understanding of the interaction within the network. 

With respect to MMP9 activity, we believe LCN2 warrants further exploration. LCN2 is a protein that protects MMP9 from degradation, thereby preserving MMP9 activity. ^28,50 Interestingly, we could demonstrate LCN2 expression in healthy retinas (Fig. 4G) and, for the first time, evaluate notable changes in its expression pattern in spontaneous autoimmune uveitis retinas (Fig. 4H). This is an important finding, because in autoimmune uveitis little is known about LCN2 function in the retina. Thus far, LCN2 has been identified only in thee aqueous humor of patients with active uveitis, forming a complex with MMP9. ³⁷ Additionally, LCN2 was found to be linked to other autoimmune diseases such as autoimmune myocarditis, and to be expressed in cardiomyocytes, fibroblasts, and neutrophils of an induced rat model and in the myocardium of human patients with myocarditis. ⁵¹ The authors suspected an induction of LCN2 by proinflammatory cytokines such as IL-1 and suggested a cytoprotective role for LCN2. ⁵¹ Double staining of the uveitic retina with MMP9 revealed areas of LCN2/MMP9 coexpression restricted to infiltrating cells (Fig. 4J). It has been described that LCN2 forms a complex with MMP9 in human neutrophil granulocytes, ⁵² but the heterogeneous cell types expressing LCN2/MMP9 in ERU (Fig. 4J) might point to an additional involvement of other cell populations and warrant further characterization. As mentioned earlier, not only was LCN2 expression restricted to infiltrating cells, it was also strongly expressed in retinal tissue. In line with our observations, LCN2 expression was evident in retinal Müller glial cells in a rat model of diabetes ⁵³ and has been interpreted as a reaction of Müller glial cells to photoreceptor damage in mouse models of retinal degeneration. ²⁹ The obvious changes in the retinal LCN2 expression pattern of horses affected by ERU are thus an interesting finding and may be linked to the activation of Müller glia cells, which we can already identify as key players in ERU. ⁵⁴  

This study provides the first report on the simultaneous decrease of TIMP2 protein expression in the vitreous and retina in a spontaneous model of autoimmune uveitis, whereas functionally associated matrix metalloproteinases showed altered or shifted expression. These findings implicate a fundamental change in functional protein-protein interactions in the MMP-associated protein network, highlighting the importance of further studies targeting the identification of other functional protein aspects in the pathophysiology of autoimmune uveitis. 

Footnotes

 Supported by Deutsche Forschungsgemeinschaft Grant DE 719/2-1 and 2-2.

Footnotes

 Disclosure: F. Hofmaier, None; S.M. Hauck, None; B. Amann, None; R.L. Degroote, None; C.A. Deeg, None

The authors thank Sieglinde Hirmer for excellent technical assistance.