Production and Activation of Matrix Metalloproteinase-2 in Proliferative Diabetic Retinopathy

Abstract

purpose. To investigate the matrix metalloproteinase (MMP) species and their activation associated with the pathogenesis of proliferative diabetic retinopathy (PDR).

methods. Sandwich enzyme immunoassays were used to measure concentrations of MMP-1, -2, -3, -7, -8, -9, and -13 in vitreous samples from patients with PDR and nondiabetic vitreoretinal diseases. To evaluate activation ratios of the zymogen of MMP-2 (proMMP-2) and -9 (proMMP-9) in the vitreous samples and fibrovascular tissues, gelatin zymography was performed. Production and tissue localization of MMP-2, membrane type 1-MMP (MT1-MMP), tissue inhibitor of metalloproteinases (TIMP)-2, and MMP-9 in the fibrovascular tissues were examined by immunohistochemistry. mRNA expression of MT1-MMP in the tissues was determined by reverse transcription–polymerase chain reaction (RT-PCR).

results. Among the seven different MMPs examined in the vitreous samples, only the levels of MMP-2 and -9 were significantly higher in the PDR samples than in the control. However, activation ratios of proMMP-2 (10.6% ± 11.8%) and proMMP-9 (2.5% ± 5.1%) in PDR vitreous samples were low and not significantly different from those of the control. In contrast, high activation ratios of proMMP-2 (54.3% ± 13.6%) and notable activation of proMMP-9 (19.5% ± 7.8%) were observed in the fibrovascular tissues. Immunohistochemical study demonstrated the localization of MMP-2 and -9 in the endothelial cells and glial cells of the fibrovascular tissues. MMP-2 was colocalized with MT1-MMP and TIMP-2, which are an activator and an activation-enhancing factor, respectively, for proMMP-2. RT-PCR analysis indicated the gene expression of MT1-MMP in the tissues.

conclusions. These data demonstrate that proMMP-2 is efficiently activated in the fibrovascular tissues of PDR, probably through interaction with MT1-MMP and TIMP-2, and suggest the possibility that the activity of MMP-2 and MT1-MMP is involved in the formation of the fibrovascular tissues.

Proliferative diabetic retinopathy (PDR) is still a main cause of visual impairment, despite the recent progress in vitreoretinal surgery. It is characterized by proliferation of fibrovascular tissue formed by the extension of retinal angiogenesis into the vitreous cavity, and formation of the fibrovascular tissue results in severe complications, such as vitreous hemorrhage and tractional retinal detachment. Many studies have been conducted to investigate the relevant factors in the progression of PDR. Vascular endothelial growth factor (VEGF) is considered to be a major angiogenic factor in the neovascularization in retinopathy. ^{1 2 3} We have proposed the possibility that the coexpression of VEGF₁₆₅, VEGF receptor (VEGFR)-2, and neuropilin-1 in the fibrovascular tissue contributes to its rapid growth in patients with PDR. ⁴ In addition to these factors, degradation of extracellular matrix (ECM) components is thought to be essential to the development of the fibrovascular tissues in PDR. Actually, some species of matrix metalloproteinases (MMPs), which are the family of zinc-dependent endopeptidases capable of degrading various ECM macromolecules, have been detected in both vitreous and fibrovascular tissues in eyes with PDR. ^{5 6 7 8 9 10 11}  

In humans, the MMP family comprises 22 members that can be classified into five subgroups, based on domain structures and substrate specificity: interstitial collagenases (MMP-1, -8, and -13), gelatinases (MMP-2 and -9), stromelysins (MMP-3 and -10), membrane type-MMPs (MT1, -2, -3, -4, -5, and -6-MMPs) and other MMPs. ¹² Because MMPs are produced in zymogen form (proMMP), they must be activated by the intracellular, extracellular, or pericellular pathway to exhibit their proteolytic activities in the tissues. ¹³ Among the activation systems, proMMP-2 activation mediated by MT1-MMP is well established. ^{14 15 16} The activity of MMPs is inhibited by tissue inhibitors of metalloproteinases (TIMPs), which comprise four different molecules (TIMP-1, -2, -3, and -4). ¹² However, accumulated lines of evidence have indicated that TIMP-2 accelerates activation of proMMP-2 by functioning as a link protein for the interaction between proMMP-2 and MT1-MMP on the cell membranes. ^{17 18} Recent studies have also shown the activation of proMMP-2 through the trimolecular complex proMMP-2/TIMP-2/MT1-MMP in human cancer tissue ^{19 20} and rheumatoid synovium. ²¹  

Previous studies by gelatin zymography and/or immunoblot have reported that proMMP-9, but not proMMP-2, is increased in the vitreous samples from the eyes of patients with PDR, compared with those of nondiabetic patients. ^{5 6 7 9 11} However, these results were based on semiquantitative methods (gelatin zymography or immunoblot analysis) and were standardized by the protein content of vitreous samples. Therefore, there is no information on the concentrations of these MMPs in the vitreous fluid samples. In addition, little is known about MMP species other than MMP-2 and -9 in the vitreous of patients with PDR. Although the activated form of MMP-2 was reported to be present in the fibrovascular tissues of PDR, ⁸ no information so far has been available on the expression of proMMP-2 activation-associated molecules (i.e., MT1-MMP and TIMP-2) in the tissues. 

In the present study, we measured concentrations of MMP-1 (tissue collagenase), MMP-2 (gelatinase A), MMP-3 (stromelysin 1), MMP-7 (matrilysin 1), MMP-8 (neutrophil collagenase), MMP-9 (gelatinase B), and MMP-13 (collagenase 3) in vitreous samples of patients with PDR and nondiabetic patients by using corresponding sandwich enzyme immunoassay (EIA) systems. The activation ratio of proMMP-2 and -9, and tissue localization of MMP-2 and -9, MT1-MMP, and TIMP-2 were further studied by gelatin zymography and immunohistochemistry, respectively. Our data suggest that proMMP-2 is produced and activated by MT1-MMP in fibrovascular tissues in PDR. 

Methods

Vitreous Samples and Fibrovascular Tissues

Vitreous and fibrovascular tissue samples were collected from 24 eyes of 24 patients with PDR when they were operated on for prolonged vitreous hemorrhage and tractional retinal detachment involving macular lesions. The age of the patients ranged from 30 to 80 years (mean, 58.8). For a control, vitreous samples were obtained from 13 eyes of patients with nondiabetic ocular diseases: 9 eyes with epiretinal membrane and 4 with idiopathic macular hole. The age of the control patients ranged from 56 to 77 years (mean, 66.5). The clinical characteristics of the patients are summarized in Table 1 . All patients gave their informed consent to our study, which adhered to the tenets of the Declaration of Helsinki. The vitreous samples were collected without dilution at the start of the pars plana vitrectomy. After centrifugation at 10,000g for 20 minutes at 4°C, supernatants of the samples were stored at −20°C until use in enzyme immunoassay (EIA) and gelatin zymography analyses. 

Fibrovascular tissues were also obtained from patients with PDR. Of 24 fibrovascular tissues, 12 samples were embedded in paraffin after fixation in 4% paraformaldehyde, and serial sections were prepared for immunohistochemical staining. Eight samples, which were relatively large, were divided into two pieces. Half of the tissue was processed for immunohistochemical staining, and the remainder was used for reverse transcription–polymerase chain reaction (RT-PCR). Four samples were stored at −20°C for gelatin zymography. 

Sandwich Enzyme Immunoassay for MMPs

Concentrations of MMP-1, -2, -3, -7, -8, -9, and -13 in the vitreous samples (24 PDR and 13 control samples) were determined according to our method, using EIA systems, as described previously. ^{22 23 24 25 26 27 28} The systems for MMP-1, -3, -8, and -13 detect both latent and active forms of each MMP, whereas those for MMP-2, -7, and -9 recognize only latent forms. In case of MMP-1, -3, and -13, the molecules complexed with TIMPs are also detected. Detection limits of these systems for MMP-1, -2, -3, -7, -8, -9, and -13 are 1.0, 6.3, 12.5, 0.63, 1.9, 3.1, and 0.25 ng/mL, respectively. The results of EIA were compared between patients with PDR and nondiabetic patients by the Mann-Whitney test, and P < 0.05 was considered to be statistically significant. 

Gelatin Zymography

The activation ratios of proMMP-2 and -9 in undiluted supernatants of the vitreous samples (24 PDR and 13 control samples), which had been stored after centrifugation, were determined by gelatin zymography. ^{29 30} Twenty microliters of each sample was mixed with the same amount of sample buffer containing 4% sodium dodecyl sulfate (SDS) without reducing agent, and incubated at 37°C for 20 minutes. Then the samples were electrophoresed at 4°C in 8.5% SDS-polyacrylamide gel containing 0.2% gelatin. After electrophoresis, the gels were soaked in 2.5% Triton X-100 to remove SDS and incubated for 24 hours at 37°C in TNC buffer (50 mM Tris-HCl [pH 7.5], 150 mM NaCl, 10 mM CaCl₂, and 0.02% NaN₃). Subsequently, they were stained for 40 minutes with 0.1% Coomassie brilliant blue. As for the positive control, a mixture of latent and active forms of MMP-2 and -9 partially purified from the culture media of HT1080 cells ³¹ and supernatants of lung carcinoma tissue homogenates (data not shown) ^{14 32} were subjected to gelatin zymography. The density of negatively stained bands was measured with NIH Image 1.41 software (available by ftp from zippy.nimh.nih.gov/or from http://rsb.info.nih.gov/nih-image; developed by Wayne Rasband, National Institutes of Health, Bethesda, MD). Gelatinolytic bands of 92, 83, 68, and 62 kDa corresponded to proMMP-9, active MMP-9, proMMP-2, and active MMP-2, respectively. ³³ The activation ratios of each MMP were calculated by dividing the density of the band for the active form by the sum of the density of the bands for both the latent and active forms. ^{29 33}  

Activation of proMMP-2 and -9 in the fibrovascular tissues (four samples) was also analyzed by gelatin zymography. After tissues were homogenized in TNC buffer containing 1% SDS, supernatants were obtained by centrifugation. Protein concentrations of the supernatants were measured by a protein quantification kit (Proteostain; Dojindo Laboratories, Kumamoto, Japan) and adjusted to approximately 1.5 mg/mL. The samples were then subjected to gelatin zymography, and the activation ratios were calculated as described earlier. To determine whether the difference in preparation of the vitreous and fibrovascular tissue samples affects the activation ratios of proMMPs, the vitreous samples were subjected to simple centrifugation or homogenization according to the method used for the fibrovascular tissues and analyzed by gelatin zymography. The results showed no difference in the activation of proMMP-2 and -9 in these samples, indicating that the methods used for tissue processing cause no artificial activation of proMMPs (data not shown). 

Immunohistochemistry and Morphometric Analysis

Production and tissue localization of MMP-2, MT1-MMP, TIMP-2, and MMP-9 in fibrovascular tissues were examined by immunohistochemical staining on serial paraffin-embedded sections. Immunohistochemistry for glial fibrillary acid protein (GFAP) was also performed to identify glial cells in the tissues. Paraffin-embedded sections were incubated with 0.3% hydrogen peroxide in methanol and subsequently with 10% normal goat serum to block endogenous peroxidase activity and nonspecific binding, respectively. Then, the sections were reacted overnight at 4°C with either mouse monoclonal antibodies against MMP-2 (2 μg/mL; clone 75-7F7), MT1-MMP (2 μg/mL; clone 222-1D8), TIMP-2 (10 μg/mL; clone 67-4H11), and MMP-9 (8 μg/mL; clone 56-2A4), or rabbit polyclonal antibodies against GFAP (1/400 dilution; Dako A/S, Glostrup, Denmark). The specificity of these mouse monoclonal antibodies and their suitability for immunohistochemical study have been verified. ^{30 34 35} In the present study, the immunoreactivity of the antibodies to MMP-2, MT1-MMP, TIMP-2, and MMP-9 was further confirmed by using the tissue sections of lung carcinomas, which are known to express these proteins (data not shown). ³⁶ After incubation of the sections with the antibodies, they were reacted for 30 minutes at room temperature with either goat antibodies against mouse immunoglobulins conjugated to a peroxidase-labeled dextran polymer (En Vision+ mouse; Dako Corp., Carpinteria, CA) for MMP-2, MT1-MMP, TIMP-2, and MMP-9 or goat antibodies against rabbit immunoglobulins conjugated to a peroxidase-labeled dextran polymer (En Vision+ rabbit; DAKO) for GFAP. As the negative control for staining, the first antibodies were replaced with either nonimmune mouse IgG (Dako A/S) for MMP-2, MT1-MMP, TIMP-2, and MMP-9 or nonimmune rabbit immunoglobulins (Dako A/S) for GFAP. Color was developed with 3,3′-diaminobenzidine tetrahydrochloride (0.2 mg/mL; Dojindo Laboratories) in 0.05 M Tris-HCl (pH 7.6) containing 0.003% hydrogen peroxide, and the sections were counterstained with hematoxylin. Serial sections were examined under light microscopy and the intensity of the immunostaining was semiquantitatively evaluated by two pathologists into three grades: −, +, or ++. Tissue localization was evaluated mainly by focusing on its distribution to vascular endothelial cells and glial cells. Morphometric analysis was performed to evaluate the degree of angiogenesis in the fibrovascular tissue samples, as described previously. ⁴ The vascular density (the number of vessels per square millimeter) of each sample was calculated, and the correlation between vascular density and immunoreactivity of each MMP in the endothelial cells was statistically analyzed by Spearman rank correlation. 

RNA Extraction and RT-PCR

The expression of MT1-MMP mRNA in fibrovascular tissues was examined by RT-PCR. Total RNA was prepared using extraction reagent (Isogen; Nippon Gene, Toyama, Japan), according to the manufacturer’s protocol, with a modification. In brief, half of the divided fibrovascular tissues (eight samples) were homogenized in l mL of the extraction reagent and 200 μL chloroform was added. After centrifugation at 4°C, the aqueous phase was collected, and 600 μL acid phenol (pH 4.3) and 210 μL chloroform were added. The aqueous phase was collected after centrifugation, and total RNA was precipitated with an equal volume of isopropanol. Extracted total RNA was reverse-transcribed with a cDNA synthesis kit (First-Strand; Pharmacia Biotech, Uppsala, Sweden) at 37°C for 1 hour in a 15-μL reaction volume containing random hexadeoxynucleotides and Moloney murine leukemia virus reverse transcriptase. Two microliters of the reaction product was subjected to 30 cycles of PCR for amplification of either MT1-MMP or β-actin cDNA. PCR was performed in a 50-μL reaction volume containing 800 nM of each primer, 250 nM of dNTPs, and 5 U Taq DNA polymerase (Toyobo, Tokyo, Japan) in a thermal controller (MiniCycler; MJ Research, Inc., Watertown, MA). The thermal cycle was 1 minute at 94°C, 2 minutes at either 61°C for MT1-MMP or 67°C for β-actin, and 3 minutes at 72°C, followed by 3 minutes at 72°C. The nucleotide sequences of the PCR primers were 5′-TCG GCC CAA AGC AGC AGC TTC-3′ (forward) and 5′-CTT CAT GGT GTC TGC ATC AGC3′ (reverse) for MT1-MMP; and 5′-TGA CGG GGT CAC CCA CAC TGT GCC CAT CTA-3′ (forward) and 5′-CTA GAA GCA TTT GCG GTG GAC GAT GGA GGG-3′ (reverse) for β-actin. ^{4 30} The expected sizes of the amplified cDNA fragments of MT1-MMP and β-actin were 0.18 and 0.66 kb, respectively. To confirm the specific amplification from the target mRNAs, we subcloned the products into a vector (pBluescript KS; Stratagene, La Jolla, CA) and analyzed them by sequencing with fluorescent T7 primer (Amersham Pharmacia Biotech, Buckinghamshire, UK), using a fluorescence-labeled primer cycle sequencing kit (Thermo Sequenase; Amersham Pharmacia Biotech) and a DNA sequencer (ALF II; Amersham Pharmacia Biotech). 

Results

Concentrations of MMPs in Vitreous Samples

The levels of seven different MMPs in the vitreous samples were determined by corresponding EIA systems. Among the MMPs examined, MMP-2 was detected in all the samples of PDR (Fig. 1B) , whereas MMP-1, -3, and -9 were detectable in 2, 2, and 12 of 24 PDR samples, respectively (Figs. 1A 1C 1D) . MMP-7, -8, and -13 were not measurable in the samples (data not shown). In control vitreous samples from nondiabetic eyes, only MMP-2 was detectable (Fig. 1B) , whereas other MMPs were below detection levels (Fig. 1 for MMP-1, -3, and -9 and data not shown for MMP-7, -8, and -13). The level of MMP-2 in the eyes with PDR (192.9 ± 93.2 ng/mL, mean ± SD) was significantly higher than that in the control eyes (85.2 ± 37.2 ng/mL; Mann-Whitney test, P < 0.05; Fig. 1B ). The level of MMP-9 in PDR (12.3 ± 18.9 ng/mL) was also significantly higher than that in the control (P < 0.05; Fig. 1D ). 

Activation of ProMMP-2 and -9 in Vitreous Samples and Fibrovascular Tissues

The activation ratios of proMMP-2 and -9 in the vitreous samples and fibrovascular tissues were evaluated by gelatin zymography. A gelatinolytic 68-kDa band of proMMP-2 was detected in all the PDR and control vitreous samples, whereas the 62-kDa band of active MMP-2 was present in 63% (15/24) of the PDR samples and 38% (5/13) of the control samples, respectively (Fig. 2A) . As for MMP-9, a 92-kDa band of proMMP-9 was detected in 83% (20.24) of PDR samples, and only a faint band of proMMP-9 was observed in 23% (3/13) of the control samples. The active form of MMP-9 (83 kDa) was detectable in 33% (8/24) of the PDR samples, but was not present in the control samples (Fig. 2A) . Densitometric analysis of the bands indicated that activation ratios of proMMP-2 in the PDR and control vitreous samples were 10.6% ± 11.8% and 7.7% ± 12.0%, respectively (not significant; P = 0.324). The activation ratio of proMMP-9 in the PCR vitreous samples was 2.5% ± 5.1%. 

In the fibrovascular tissues in patients with PDR, both zymogen and active forms of MMP-2 and -9 were observed, and the activation ratios of proMMP-2 and -9 were 54.3% ± 13.6% and 19.5% ± 7.8%, respectively (Fig. 2B) . Both ratios were significantly higher, than those of the vitreous samples (P < 0.01). 

Immunohistochemistry of MMP-2, MT1-MMP, TIMP-2, and MMP-9 and mRNA Expression of MT1-MMP in Fibrovascular Tissues

The vascular density and immunohistochemical data on MMP-2, MT1-MMP, TIMP-2, and MMP-9 in the fibrovascular tissues are summarized in Table 2 . As we have reported, ⁴ the fibrovascular tissues are composed of blood vessels and stromal cells in a fibrous interstitium. Endothelial cells of the blood vessels were positively immunostained with the anti-CD34 antibody (data not shown) and some of the stromal cells were stained by the anti-GFAP antibody (Fig. 3D) . As shown in Figures 3A and 3B and Table 2 , MMP-2 and MT1-MMP, respectively, were colocalized in endothelial cells of the blood vessels in all the samples (19/20) except for case 17, which showed no immunostaining for MT-1-MMP. TIMP-2 was also immunostained in the endothelial cells in 50% (10/20) of the samples (Fig. 3C , Table 2 ). GFAP-positive glial cells were present in 75% (15/20) of the fibrovascular tissue samples(Table 2) . MMP-2, MT1-MMP, and TIMP-2 were colocalized in the glial cells in more than 80% of the samples (Fig. 3 , Table 2 ). In contrast, MMP-9 was immunolocalized in endothelial cells and glial cells in 45% (9/20) and 27% (4/15) of the samples, respectively (Table 2) . No staining was observed with nonimmune mouse IgG (Fig. 3E) or rabbit immunoglobulins (data not shown). There was no correlation between the immunoreactivity of MMP-2, MMP-9, or MT1-MMP in the endothelial cells and vascular density (data not shown). 

To ascertain the expression of MT1-MMP in the fibrovascular tissues, RT-PCR was performed. As shown in Figure 4 , MT1-MMP was expressed in seven (88%) of eight samples of the tissues. The specific amplification from the target mRNA was confirmed by sequencing the amplified DNA fragments (data not shown). 

Discussion

In the present study, we demonstrated for the first time that among seven different MMPs examined, concentrations of only MMP-2 and -9 are significantly increased in vitreous samples from PDR-affected eyes compared with those from nondiabetic eyes. Previous studies have shown that MMP-2 and -9 are present in the vitreous samples of PDR, and those investigators concluded that MMP-9, but not MMP-2, is enhanced in PDR. ^{5 6 7 9 11} In most of these studies, the data were obtained by semiquantitative analyses, such as gelatin zymography or immunoblot, and the results were analyzed by standardization of the band intensity with the protein levels in vitreous samples. ^{5 9 11} We determined concentrations of each MMP in undiluted vitreous samples by using corresponding EIA systems. Although we do not exclude the possibility that increased levels of MMP-2 and -9 were partially derived from serum, our data indicate that in addition to proMMP-9, concentrations of MMP-2 was significantly (2.3-fold) higher in the vitreous samples of PDR eyes than in samples from the nondiabetic eyes. In addition, we demonstrated that the amounts of MMP-1, -3, -7, -8, and 13 were negligible in the 24 PDR and 13 control vitreous samples, although 8% (2/24) of the PDR samples showed very low levels of MMP-1 and -3. Thus, it seems likely that these MMPs are present at extremely low levels in the vitreous, if present at all. 

Because most MMPs except MT-MMPs and MMP-11 are secreted as inactive zymogens, ¹³ overexpression is not sufficient for the in vivo action of MMPs, and activation is a prerequisite to their functioning in the tissues. ¹³ Thus, we examined the activation of proMMP-2 and -9 by gelatin zymography, and found that their activation ratios in the vitreous of patients with PDR were confined to low levels, only approximately 10% or less. In contrast, the fibrovascular tissue samples of PDR showed much higher activation ratios of proMMP-2 (54.3% ± 13.6%) and proMMP-9 (19.5% ± 7.8%). These data, therefore, demonstrate that both MMP-2 and -9 are present mainly in their latent forms in the vitreous of PDR and suggest that the vitreous is not a place for the efficient activation of these proteinases. Furthermore, high activation ratios of proMMP-2 in the fibrovascular tissues of PDR suggest the presence of activator(s) for proMMP-2 in the tissues. 

Unlike other MMPs, proMMP-2 is special, in that it is not activated by serine proteinases such as plasmin, but is activated by MT1-MMP. ¹⁴ The zymogen of MT1-MMP is activated intracellularly by furin, and the activated form is expressed on the cell membranes. ^{14 37} Thus, MT1-MMP functions as an activator of proMMP-2, once it is expressed by the cells in the local tissues. ^{14 37} Although six different MT-MMPs (MT1-, MT2-, MT3-, MT4-, MT5- and MT6-MMP) have been sequenced, among which MT1-, MT2-, and MT3-MMPs are capable of activating proMMP-2 in vitro, ^{29 30 38} MT1-MMP is believed to be the main activator of proMMP-2 in various pathologic tissues in humans. ³⁸ Because each fibrovascular tissue sample was extremely small, the expression of other MT-MMP species and correlations of MT1-MMP expression levels (protein and/or mRNA) with proMMP-2 activation ratios could not be studied in the present study. However, data from immunohistochemistry and RT-PCR in the present study demonstrated the definite expression of MT1-MMP in the fibrovascular tissues where proMMP-2 was efficiently activated. In addition, our immunohistochemistry showed the colocalization of MMP-2 and MT1-MMP in the endothelial cells and glial cells in almost all the fibrovascular tissues. In contrast, MT1-MMP is reported to be undetectable in the vitreous, ³⁹ in which proMMP-2 activation is minimal, as shown in the present study. Thus, these strongly suggest that proMMP-2 activation in the fibrovascular tissues of PDR is mainly attributable to MT1-MMP. 

Recent studies ^{17 18} have demonstrated that TIMP-2 is required for the efficient activation of proMMP-2 by MT1-MMP on the cell membranes, where TIMP-2 acts as a link protein to form the ternary complex of proMMP-2/TIMP-2/MT1-MMP. ¹⁷ In the present study, colocalization of MMP-2, TIMP-2, and MT1-MMP was shown in some of the endothelial cells and glial cells in more than 50% of the fibrovascular tissue samples, suggesting the possibility that such interaction may occur in the cells of the fibrovascular tissues of PDR. Compared with proMMP-2, the activation ratio of proMMP-9, analyzed by gelatin zymography, was much lower in the fibrovascular tissues. However, a recent study has shown that proMMP-9 bound to its substrates has some proteolytic activity without changing its molecular weight, although the activity of proMMP-9 is 10 times lower than that of active MMP-9. ⁴⁰ Thus, the data in our study do not exclude the possibility that MMP-9 is also proteolytically active in the fibrovascular tissue. 

Fibrovascular tissues of PDR, which contain cellular constituents including endothelial cells, glial cells, lymphocytes, monocytes, and fibroblasts, ⁴¹ are formed along the interface between the posterior hyaloid membrane and internal limiting membrane (ILM). Thus, both endothelial cells and glial cells must migrate from the retina to the vitreoretinal interface by degrading the ILM components, comprising type IV collagen, laminin, fibronectin, proteoglycans, and type I collagen. ^{42 43 44} Both MMP-2 and -9 are capable of degrading type IV collagen and gelatins that are formed after the cleavage of the triple helical portion of the native collagens by collagenolytic MMPs, such as MMP-1 ¹³ and MT1-MMP. ⁴⁵ Moreover, MMP-2, but not MMP-9, can digest laminin, fibronectin, and proteoglycans. ^{31 46} Thus, our present immunohistochemical and gelatin zymographic data suggest that MMP-2 activity in fibrovascular tissues is implicated in the degradation of the basement membrane components of the retinal vessels or ILM during formation of the fibrovascular tissues. In addition, MT1-MMP itself is known to degrade the fibrin matrix that forms around the leaky vessels during the angiogenic process. ^{47 48} Therefore, it is conceivable that MT1-MMP expressed by the endothelial cells may be directly involved in the blood vessel formation in the fibrovascular tissues. Further studies are needed to substantiate these hypotheses. 

Production and Activation of MMP-2 in PDR

In PDR, degradation of basement membrane of vasculature and internal limiting membrane is crucial in the formation of the fibrovascular membrane (FVM). In this study, to define the responsible proteinases for formation of the FVM, MMPs in PDR eyes were analyzed with sandwich EIA, gelatin zymography, and immunohistochemistry. The resultant data showed that, in addition to MMP-9, which is reportedly abundant in PDR eyes, the level of MMP-2 is significantly high in PDR. Furthermore, the activation of MMP-2 in FVM itself is suggested to play an important role in the pathogenesis of PDR. 

 Supported by Grant-in-Aid 13670189 for Scientific Research (C) from the Japan Society for the Promotion of Science (EI) and the Keio Gijuku Academic Development Funds (EI).

 Submitted for publication July 3, 2002; revised October 28, 2002; accepted November 20, 2002.

 Disclosure: K. Noda, None; S. Ishida, None; M. Inoue, None; K.-I. Obata, Daiichi Fine Chemical Co. (E); Y. Oguchi, None; Y. Okada, None; E. Ikeda, None

 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.

 

Table 1.

 

Clinical Information on the Pateints

Case	Age (ys)	Gender	Diagnosis	HbA1c (%)
PDR cases
1	75	F	PDR	10.2
2	30	M	PDR	5.3
3	65	M	PDR	6.0
4	47	M	PDR	6.8
5	80	M	PDR	7.8
6	63	M	PDR	8.6
7	68	F	PDR	6.8
8	38	F	PDR	8.2
9	66	M	PDR	5.0
10	49	M	PDR	6.2
11	66	F	PDR	6.1
12	56	M	PDR	6.4
13	65	F	PDR	10.5
14	55	F	PDR	7.4
15	60	F	PDR	6.8
16	59	M	PDR	9.1
17	66	F	PDR	5.8
18	39	F	PDR	5.6
19	57	M	PDR	7.0
20	66	M	PDR	6.1
21	66	M	PDR	9.2
22	62	M	PDR	7.0
23	59	F	PDR	9.4
24	53	M	PDR	6.7
Non-PDR cases
1	77	M	ERM	ND
2	71	M	MH	ND
3	68	M	MH	ND
4	72	M	ERM	ND
5	67	M	ERM	ND
6	68	F	ERM	ND
7	56	F	ERM	ND
8	56	F	MH	ND
9	62	M	ERM	ND
10	70	F	ERM	ND
11	67	M	MH	ND
12	73	F	ERM	ND
13	57	M	ERM	ND

Figure 1.

 

Levels of (A) MMP-1, (B) -2, (C) -3, and (D) -9 in the PDR and control vitreous samples. The samples were analyzed by the corresponding EIA systems. Levels of MMP-2 and -9 in the PDR vitreous samples were significantly higher than those in control samples. Bars, mean results; dotted lines: detection levels. MMP-7, -8, and -13 were not detectable in all the samples examined (data not shown). *P < 0.05; Mann-Whitney test.

Figure 2.

 

Gelatin zymography of the vitreous and fibrovascular tissue samples. (A) Representative zymography of the vitreous samples from three control patients (lanes 1–3) and five patients with PDR (lanes 4–8). (B) Zymography of homogenates of the fibrovascular tissues from four patients with PDR (lanes 1–4). Gelatinolytic bands of 92, 83, 68, and 62 kDa correspond to proMMP-9, active MMP-9, proMMP-2, and active MMP-2, respectively. Note that both proMMP-2 and proMMP-9 were highly activated in the fibrovascular tissues (B) compared with the vitreous fluid samples (A). Molecular masses of the protein standards are as follows: phosphorylase b (94 kDa), transferrin (77 kDa), bovine serum albumin (68 kDa), heavy chain of IgG (55 kDa), ovalbumin (43 kDa), and carbonic anhydrase (29 kDa). St, authentic MMP-2 and -9 species purified from HT1080 cells.

Table 2.

 

Summary of Immunohistochemical Study on Fibrovascular Tissues

Case	Vascular Density (vessels/mm2)	MMP-2			MT1-MMP			TIMP-2			MMP-9			GFAP
				Endothelial Cell	Glial Cell	Endothelial Cell	Glial Cell	Endothelial Cell	Glial Cell	Endothelial Cell	Glial Cell
1	42.0	+	−	+	−	−	−	+	−	+
2	149.4	++	++	+	+	+	+	++	+	+
3	40.5	+	+	+	+	+	−	−	−	+
4	229.1	+	++	+	+	−	+	+	−	+
5	21.9	+	ND	+	ND	−	ND	−	ND	−
6	189.6	+	−	+	+	−	−	−	+	+
7	51.7	+	ND	+	ND	−	ND	+	ND	−
8	102.8	+	+	+	+	+	+	−	−	+
9	34.6	+	ND	+	ND	+	ND	+	ND	−
10	65.1	+	++	+	+	+	+	+	+	+
11	169.9	++	ND	+	ND	−	ND	+	ND	−
12	45.7	+	+	+	+	−	+	+	−	+
13	70.6	++	ND	+	ND	+	ND	−	ND	−
14	40.2	+	+	+	+	+	+	+	−	+
15	138.9	+	+	+	+	−	+	−	−	+
16	42.5	+	+	+	+	+	+	−	−	+
17	17.5	+	+	−	−	+	+	−	+	+
18	54.9	+	+	+	+	−	+	−	−	+
19	54.7	+	+	+	+	−	+	−	−	+
20	136.7	+	+	+	+	+	+	−	−	+
Positive ratio		100%	87%*	95%	87%*	50%	80%*	45%	27%*	75%

Figure 3.

 

Immunohistochemistryof MMP-2, MT1-MMP, TIMP-2, and GFAP in the fibrovascular tissues from eyes with PDR. A representative sample (case 10) is presented. Serial sections were immunostained with monoclonal antibodies against MMP-2 (A), MT1-MMP (B), and TIMP-2 (C); polyclonal antibodies against GFAP (D); or nonimmune mouse Ig (E). MMP-2 (A), MT1-MMP (B) and TIMP-2 (C) were colocalized in some vascular endothelial cells (arrowheads) and glial cells (arrows) in the fibrovascular tissue. Hematoxylin counterstaining. Scale bar, 50 μm.

Figure 4.

 

RT-PCR for the expression of MT1-MMP in the fibrovascular tissues. The specific primers for human MT1-MMP and β-actin amplified 180- and 660-bp fragments of the cDNAs, respectively. The band corresponding to the MT1-MMP cDNA fragment was detected in seven of eight samples (lanes 1–8).

The authors thank Hitoshi Abe and Michiko Uchiyama for skillful technical assistance; Shizuaki Kitamura, Tadahiko Eshita, and Hajime Shinoda for help in obtaining surgical specimens; Toru Takebayashi in the Department of Preventive Medicine and Public Health, Keio University School of Medicine, for advice on the statistical analyses; and Yukiko Kishida in the Department of Pathology, Tokyo Postal Services Agency Hospital, for a critical reading of the manuscript.