The Balance between Proteinases and Inhibitors in a Murine Model of Proliferative Retinopathy

Abstract

purpose. To examine the expression of matrix metalloproteinases (MMPs) and their inhibitors during the development of retinal neovascularization (NV) in a mouse model.

methods. A well-characterized murine model of retinal NV was used to study the expression of specific MMPs (MMP-2, MMP-9, and MT1-MMP) and tissue inhibitor of metalloproteinases (TIMPs types 1, 2, and 3). NV of the retina was induced in mice by exposure to 75% O₂ from postnatal day (P)7 to P12, followed by return to room air from P12 to P17. Expression of MMP mRNA was analyzed by reverse transcription–polymerase chain reaction (RT-PCR). In addition, retinal tissue removed from control (without NV) and experimental animals (with NV) was analyzed for the expression of TIMP-1, TIMP-2, and TIMP-3 mRNA and protein using RT-PCR and Western blot analysis.

results. During the angiogenic period from P13 to P17, MMP-2 and -9, and MT1-MMP message expression increased in experimental retinas compared with control samples. The TIMP-2 message and protein levels increased steadily in the retina of control animals until P17. This was in contrast to that seen in the retinas of the experimental animals in which TIMP-2 message and protein remained low and significantly less than in control samples. There were no significant changes in TIMP-3 message levels in retinal tissues, and TIMP-1 message and protein were undetectable.

conclusions. Correlation was made at the mRNA and protein levels of TIMP expression compared with that of MMPs in a murine model of retinal NV, which suggests a temporal role for MMP-2 and -9, MT1-MMP, and TIMP-2 in new vessel formation in response to hypoxic stimulation.

Proliferative retinopathies, such as diabetic retinopathy and retinopathy of prematurity, result from hypoxic conditions due to nonperfusion of the retina or a decrease in oxygen tension, which culminates in the development of new vessels. ^{1 2} These conditions result in complications such as vitreous hemorrhage and tractional retinal detachment, and eventually lead to blindness. Retinal microvascular endothelium participates in the process of neovascularization (NV) in a stepwise fashion. During the initiation phase, endothelial cells respond to locally produced angiogenic factors and upregulate the expression of extracellular proteinases. This is followed by the invasive phase, characterized by the migration of endothelial cells through the basement membrane into the surrounding extracellular space where these cells proliferate and form new capillary tubes. ^{3 4 5}  

Extracellular proteinases and their inhibitors play an important role in the regulation of endothelial cell migration and extracellular matrix remodeling during angiogenesis. ^{3 6} These proteinases include members of the matrix metalloproteinase (MMP) family such as gelatinases (MMP-2 and -9), collagenases (MMP-1, -8, -13, and -18), and stromelysins (MMP-3, -10, and -11). ^{3 7 8 9 10 11 12 13 14} The MMP activity is in part regulated by the tissue inhibitors of metalloproteinases (TIMPs), which bind the proteinases and inhibit their activity. The balance of proteinases and inhibitors has been shown to be a critical determinant of endothelial cell morphology and tube formation in vitro. ⁵ Changes in the proteinase–inhibitor balance are capable of altering morphology of capillary tubes with excessive proteolysis, resulting in saclike noninvasive structures. ⁵  

Three members of the TIMP family have been identified thus far: TIMP-1, -2, and -3. Each TIMP is capable of inhibiting all metalloproteinases; however, preferential binding to specific MMPs has been reported. ^{6 9} TIMP-1 primarily inhibits the activities of MMPs-1, -3, and -9, whereas TIMP-2 inhibits MMP-2. ^{6 15} TIMP-2 has also been shown to bind and stabilize MMP-2 by preventing autolytic degradation and by participating in its activation. ^{15 16 17} TIMP-3 is localized exclusively to the extracellular matrix (ECM) and is relatively insoluble, illustrating its potential to prevent matrix proteolysis and the release of sequestered growth factors stored in the ECM. ⁶ TIMP-3 is present in the Bruch’s membrane of normal human eyes, ¹⁸ and the mRNAs of TIMP-3 have been localized in mouse and human retinal pigment epithelial cells. ^{19 20} The TIMPs inhibit ECM proteolysis and may play an important role in inhibiting new vessel formation. ^{15 21 22 23}  

We have previously shown that human diabetic epiretinal neovascular membranes contain high levels of extracellular proteinases including MMP-2 and -9 and urokinase. ⁴ In addition, the levels of MMP-2 and -9 were elevated in retinal tissue in the mouse model of retinal NV. ²⁴ However, little is known about what other factors, including proteinase inhibitors, regulate the formation of new vessels in the retina. In the present study, we have examined the spatial expression of MMPs in the retina during the development of NV and the relationship and balance between MMPs and TIMPs in this model. 

Methods

Mouse Model of Proliferative Retinopathy and Histologic Analysis

Litters of C57Bl/6J mice were placed with their nursing mothers in an incubator maintained at 75% ± 2% oxygen from postnatal day (P)7 to P12, as described previously. ²⁵ Oxygen levels were continuously monitored using a portable oxygen analyzer. At P12, mice were removed from the incubator to room air (n = 15). Control litters were maintained in room air only (n = 15). Mice were killed with CO₂ at P13, P15, and P17, and both eyes were immediately enucleated and either fixed for histology, or the retinas were frozen for subsequent analysis of RNA expression. Procedures used in all experiments were consistent with the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research. 

For histologic analysis of NV, eyes were removed with care taken to leave the optic nerve intact. The eyes were fixed with 4% paraformaldehyde in 0.1 M PO₄ buffer at 4°C. The whole eye was embedded in paraffin, and serial axial sections (6μ m) were cut parallel to the optic nerve. Sections were mounted on glass slides with mounting medium containing DAPI (diamidinophenylindole; Vectashield; Vector, Burlingame, CA) and examined by fluorescence microscopy. NV was quantitated as previously described. ²⁴ Two eyes for each independent hypoxic (n = 4) or normoxic (n = 4) treatment group were analyzed for neovascular nuclei on the vitreous side of the inner limiting membrane at P17 to confirm a positive or negative hypoxia-induced NV response. To confirm that the cells on the vitreal side were neovascular cells, immunocytochemistry was performed on some of the frozen sections of the retina by using an endothelial cell-specific antibody (1:100; rat anti-mouse CD 31; PharMingen, San Diego, CA) followed by biotinylated goat anti-rat serum, avidin-biotin complex, and diaminobenzidine. 

RNA Isolation and Proteinase–Inhibitor Gene Expression Analysis

RNA was isolated from the murine retinal tissues by using a reagent (Trizol; Gibco, Grand Island, NY) followed by formaldehyde gel analysis to confirm the integrity and quantity of RNA. Both retinas from each of four animals at each age were pooled and analyzed for reverse transcription–polymerase chain reaction (RT-PCR). Comparisons were made between control and experimental animals at P13, P15, and P17. 

First-strand cDNA was prepared from 0.5 μg total RNA using an oligo dT primer and reverse transcriptase (Superscript; Gibco). For semiquantitative PCR, 1 μl of each first-strand reaction was then amplified with primers specific for MT1-MMP, MMP-2, MMP-9, TIMP- 1, TIMP-2, TIMP-3, and 18S RNA. The primer sequences were MT1- MMP: 5′-AGTAAAGCAGTCGCTTGGGT-3′, 5′-TGGGTAGCGATGAAGTCTTC-3′; MMP-2: 5′-TGGGTGGAAATTCAGAAGGTGC-3′, 5′-ATCTACTTGCTGGACATCAGGGGG-3′; MMP-9: 5′-TGCGACCACATCGAACTTCG-3′, 5′-CCAGAGAAGAAGAAAACCCTCTTGG-3′; TIMP-1: 5′-CTTGCATCTCTGGCATCTGG-3′, 5′-AAGTAGACAGTGTTCAGGC-3′; TIMP-2: 5′-GAGATCAAGCAGATAAAGATG-3′, 5′-GACCCAGTCCATCCAGAGGC-3′; TIMP-3: 5′-ATCAGTCAAAGGCAGCAAGC-3′, 5′-AGCATTGAATAGAATTCTGTGTCC-3′; and 18S RNA: 5′-GAGCTCACCGGGTTGGTTTTG-3′, 5′-TACCTGGTTGATCCTGCCAG-3′. 

Standard PCR amplification was performed at 94°C for 1 minute, 60°C for 1 minute, and 72°C for 1 minute for 30 cycles, which has been determined to be within the linear range of product amplification. After completion of PCR, 20 μl of the reactions were analyzed by agarose gel electrophoresis and ethidium bromide staining to determine the presence or absence of specific transcripts, as well as the levels of transcript relative to the control transcript 18S RNA. 

Quantitation of band density was performed using image analysis software (Imager 2200; Alpha Innotech, San Leandro, CA). 

Analysis of TIMP Protein Level

Both retinas from each of four animals at each age were pooled and analyzed. Comparison was made between control and experimental animals at P13, P15, and P17. Western blot analysis was performed using equal amounts of murine retinal tissue extract. The tissue extracts were boiled in sodium dodecyl sulfate (SDS) sample buffer, and 10 μg total protein was fractionated on a 7.5% SDS-polyacrylamide gel and transferred to a nitrocellulose filter. The filters were blocked overnight at 4°C with Tris-buffered saline with 0.1% Tween-20 (TBST) and 1% milk, rinsed with TBST, and incubated for 1 hour at 25°C with 1 μg/ml TIMP-1– or TIMP-2–specific murine antibodies (Oncogene, Manhasset, NY) in TBST. The filters were washed with TBST and incubated for l hour with anti-mouse biotin–conjugated antiserum (1:500, TBST). After a 1-hour rinse in TBST, the filters were developed using 5-bromo-4-chloro-3-inodoyl phosphate–nitroblue tetrazolium (BCIP-NBT) for 2 to 5 minutes. The levels of TIMP protein compared with control samples were then quantitated using image analysis software (Imager 2000; Alpha Innotech). Total protein levels were determined using the BCA assay system (Pierce, Rockford, IL). Control samples for the specificity of antibodies included blots incubated with no primary antibody. TIMP-3–specific antibodies are not available commercially, and therefore Western blot analysis for TIMP-3 could not be performed. 

Statistical Analysis

Statistical analysis was performed by using Student’s t-test, and values were expressed as the mean ± SEM. Tests results were considered statistically significant at P ≤ 0.05. 

Results

Histologic Analysis

Histologic analysis showed a quantifiable neovascular response by P17 in 100% of the animals exposed to 75% oxygen followed by room air. This is consistent with the previously reported results from our laboratory, ²⁴ and other laboratories ²⁵ in this established model of experimentally induced retinal NV (neovascular nuclei count: control, 0.37 ± 0.07; experimental, 43.50 ± 3.50; P < 0.001). Numerous vascular tufts were found to be extending from the retina into the vitreous. Control animals kept in room air for 17 days exhibited little evidence (<1%) of NV extending into the vitreous beyond the inner limiting membrane. Immunocytochemistry of the retinal sections of experimental animals with anti-CD 31 revealed positive staining of the cells on the vitreal side of the inner limiting membrane, indicating these cells to be endothelial cells and thus vascular in origin. 

MMP Expression

We have shown that the expression of MMP-2 and -9 is increased in the retina at the time of maximal NV (P17). ²⁴ In the present study, we examined the earlier stages of the NV process to confirm that the profile of MMP expression remained consistent. Semiquantitative RT-PCR was used to determine the message levels of MT1-MMP, MMP-2, and MMP-9 in experimental animals relative to control samples. The message for MT1-MMP increased significantly from P13 to P15, whereas MMP-9 increased from P13 onward (Fig. 1) . The mRNA for MT1-MMP was significantly higher in the experimental group compared with the control group (P < 0.05 on P13, P < 0.005 on P15, and P < 0.0025 on P17). The message for MMP-2, although significantly higher than the control values (P = 0.005; on P13, P15, and P17), remained relatively constant from P13 to P17. The mRNA for MMP-9 in the retinas of the experimental animals was significantly higher than that in the control animals (P = 0.0025 on P13, P15, and P17). 

TIMP Expression

The levels of TIMP were determined and standardized to 18S RNA message levels. TIMP-1 mRNA was undetectable in the retina of either control or experimental animals at the ages examined. In contrast, the messages for both TIMP-2 and TIMP-3 were detected in both the control and experimental animals. No significant difference was seen in the level of TIMP-2 mRNA between control and experimental animals at P13. However, unlike the control animals (without NV) in which the TIMP-2 message increased over the next 4 days, the level of TIMP-2 mRNA in the oxygen-treated group (with NV) remained relatively constant (Fig. 2) . At P15, the mRNA for TIMP-2 in the retinas of experimental animals was significantly lower than that in the control animals (P = 0.025). By P17 in experimental animals, the age at which NV and MMP expression are maximal, TIMP-2 mRNA levels were still significantly lower than in control retinas (P = 0.02). TIMP-3 message levels in the retinas of both control and experimental animals increased between P13 and P17, although no significant difference was detected between the two groups (Fig. 2) . 

Retinal protein extracts were analyzed by Western blot analysis to determine the levels of TIMP proteins. TIMP-1 protein was undetectable in the retina during this period, which correlated with the absence of TIMP-1 mRNA in the tissue. In control retinas, TIMP-2 protein levels increased from P13 to P15 when they appeared to plateau, remaining unchanged by P17 (Fig. 3) . At P15 and P17, TIMP-2 protein was significantly elevated in the control animals compared with the experimental animals (P < 0.05), consistent with the increase seen in the mRNA expression described before. The TIMP-2 protein levels in experimental retinas did not appear to increase from P13 to P17 and remained significantly lower than in control samples at P17 (Fig. 3) . 

Discussion

Hyperoxia (75% oxygen) followed by room air has been shown to result in retinal NV in 100% of experimental animals at P17. ²⁵ This model is reproducible, quantifiable, and inexpensive and has been used in studies of retinal NV. Previous histologic studies of this model have shown that the vascular development of retinas on P7 have maximal hyaloid vascular regression and minimal retinal vascular development. ²⁵ The NV was maximal at P17 through P21, followed by slow regression of new vessels with the re-establishment of a more normal branching vascular pattern. Furthermore, by using the DAPI stain (a nuclear stain), we identified those cells lying on the vitreal side of the inner limiting membrane as neovascular cells. The stain also allowed us to reproducibly quantitate the neovascular nuclei. Analysis of retinal sections stained with an endothelial cell-specific antibody (anti-CD 31) showed that these neovascular cells are endothelial cells indeed, confirming their origin to new vessels (data not shown). The neovascular tufts extending into the vitreous from the retina in the experimental animals were also found to leak fluorescein-dextran (data not shown). 

Increases in MMP expression and activation along with changes in the levels of the TIMP favoring a net positive proteolytic balance may form the basis for the initiation and progression of the angiogenic events in this model. MT1-MMP, MMP-2, and MMP-9 mRNA levels were significantly increased in the retinas of the experimental animals compared with control retinas. Previous studies in our laboratory have correlated this period of NV in response to a hypoxic stimulus with increased expression and activation of specific MMP proteins—namely, MMP-2 and -9. ²⁴ However, within the experimental animals some differences were seen in the MMP response. MT1-MMP, the primary activator of MMP-2, ²⁶ and MMP-9 mRNA levels increased steadily from P13 to P17. MMP-2 expression at this time increased above that in control retinas but remained relatively constant. This may suggest that activation of MMP-2 by MT1-MMP, along with increased transcription, is critical for the role of this enzyme in response to hypoxic stimulation and may be accounted for by the increased levels of MT1-MMP at P15. 

Previous evaluation of MMP message expression at P12, immediately after the return of mice to room air, revealed no difference in MMP expression between control and experimental mice. By P17 with the onset of relative hypoxia, MMP-2 and -9 activities were found to increase significantly in experimental retinas compared with control retinas. ²⁴ This suggests a response by the retinal cells to the subsequent hypoxic stimulus resulting in the observed changes in MMP and TIMP expression. Hypoxia begins in this model shortly after the return to room air followed by the production of specific angiogenic factors such as vascular endothelial growth factor (VEGF). ²⁷  

MMPs are known to degrade a broad spectrum of ECM components. These proteinases are secreted as latent proenzymes that are cleaved proteolytically to yield the mature active enzymes. MMP-2 and -9 are important during the formation of new vessels, because their substrate specificity includes type IV collagen, a major component of basement membranes. ⁷ Type IV collagen must be degraded to facilitate the migration of vascular endothelial cells. Once the basement membrane is traversed, MMPs may be important in remodeling the ECM and regulating cell–ECM contacts, facilitating cell migration and the formation of new vessels. An important feature in the function of MMP-2 is its ability to interact with a cell surface receptor, theα vβ3 integrin, resulting in localized areas of high proteolytic activity. ²⁸  

In the present study, both the mRNA and protein levels of TIMP-2 in the retina were decreased in the experimental group responding to hypoxic stimuli when compared with control samples during the angiogenic phase (P15–P17). The decrease in TIMP-2 paralleled an increase in MMP expression during the angiogenic phase. The TIMP-2 levels initially increased in control animals from P13 to P15, perhaps because of a specific developmental program, whereas the experimental levels did not significantly change. 

TIMP-1 and -2 are secreted by many cell types, and their expression is regulated oppositely and independently by phorbol esters, transforming growth factor (TGF)-β, and platelet-derived growth factor (PDGF). ^{6 10 12 29} TIMP-MMP complexes are reversible and usually occur in a 1:1 equimolar concentration. ^{6 15} TIMP-2 may bind MMP-2 in a 2:1 ratio, abolishing the complex activity; is also found in complexes with active MMP-2; and is believed to stabilize the enzyme by preventing autocatalytic degradation. ^{16 17} Studies have also shown a role for TIMP-2 in MMP-2 activation by linking the proteinase to its activator membrane type MMP (MT1-MMP). ^{16 17} TIMP-2 has an affinity with MMP-9 but to a lesser extent than with MMP-2. 

It has been speculated that TIMP-3 dysfunction may result in neovascular growth into the Bruch’s membrane because of the potential antiangiogenic activity of the functional protein. ³⁰ Point mutations in TIMP-3 gene have been implicated in patients with Sorsby’s fundus dystrophy, an autosomal dominant macular disease with earlier onset of symptoms similar to those of age-related macular degeneration (ARMD). ^{30 31} The TIMP-3 content in Bruch’s membrane of the macula shows a significant increase in eyes with ARMD compared with age-matched normal eyes. ³² In our study, the TIMP-3 mRNA levels were comparable between normoxic and hypoxic retinas from P13 through P17 which suggests that although this inhibitor may play a role in ECM turnover, its expression is not significantly influenced by the hypoxic stimulus. 

The upregulation of MMP expression and activation as well as a decrease in TIMP-2 levels may represent a final common pathway in the process of hypoxia-mediated retinal NV. Therefore, a more complete understanding of the mechanisms involved in the regulation of this proteinase–inhibitor balance will prove invaluable for the development and evaluation of pharmacologic therapies for retinal angiogenesis. In addition, the determination of hypoxia as either a direct or indirect factor influencing vascular endothelial cell proteinase and inhibitor expression, as opposed to the indirect effects of hypoxia mediated by angiogenic factors (i.e., VEGF) may also be useful in identifying future targets for therapeutic intervention. 

 Supported by Grant RO1 EY 12604-02 (AD) from the National Institutes of Health, Bethesda, Maryland.

 Submitted for publication May 11, 2000; revised August 16, 2000; accepted September 20, 2000.

 Commercial relationships policy: N.

 

Figure 1.

 

(A) RT-PCR analysis of MT1-MMP, MMP-2, and MMP-9 mRNA expression during the development of NV in the murine model of proliferative retinopathy. The mRNA levels in the retina are expressed as a percentage of control values (animals without NV) for P13, P15, and P17. MT1-MMP, MMP-2, and MMP-9 mRNAs were significantly increased on all days in comparison with control animals. Values were expressed as mean ± SEM. *SE too small to be depicted. (B) Representative agarose gel showing the expression of mRNAs for MT1-MMP, MMP-2, and MMP-9 on P13, P15, and P17 from control (13C, 15C, and 17C) and experimental (13E, 15E, and 17E) animals.

Figure 2.

 

RT-PCR analysis of TIMP-2 (A) and TIMP-3 (B) mRNAs in the retinas of experimental animals with NV and control animals without NV. The TIMP-2 mRNA in the experimental group remained low, whereas that in the control group continued to increase from P13 to P17. No significant difference was seen in the level of TIMP-3 mRNA between control and experimental animals. Values are expressed as mean ± SEM. *Significantly less than control. (C) Representative agarose gel showing the expression of mRNAs for TIMP-2, TIMP-3 and 18S RNA on P13, P15, and P17 from control (13C, 15C, and 17C) and experimental (13E, 15E, and 17E) animals.

Figure 3.

 

(A) Western blot analysis of TIMP-2 expression in the retinas of experimental animals with NV compared with control animals without NV. In control animals, the TIMP-2 protein level increased from P13 to P17, whereas the level in the experimental group remained low throughout the period. Values were expressed as mean ± SEM.* Significantly less than control. (B) Representative Western blot of TIMP-2 in the retinas of control (C) and experimental (E) animals on P13, P15, and P17. The density of the TIMP-2 band in the control retinas was significantly increased on P17, whereas that in the experimental group remained low.

The authors thank Susan Alexander, Wanmin Song, and Lin Xu for technical assistance.