Reduced Retinal Angiogenesis in MMP-2–Deficient Mice

Abstract

purpose. To study the putative role of endogenous matrix metalloproteinases (MMPs) in retinal neovascularization, an established mouse model was used to compare the retinal neovascularization observed in wild-type mice with that in mice without the MMP-2 or -9 genes.

methods. C57Bl/6 (MMP-2^+/+ and -9^+/+), MMP-2–deficient (MMP-2^−/−), and MMP-9–deficient (MMP-9^−/−) mice were used. After oxygen-induced retinopathy was induced in the mice, their eyes were rapidly removed and frozen in optimal cutting temperature embedding compound. Sections were histochemically stained with specific markers for vascular cells and angiogenesis-related factors. The area of new retinal vessels was measured using image-analysis software and compared between groups.

results. Retinal neovascularization was not significantly different between wild-type and MMP-9^−/− mice. The MMP-2^−/− mice had significantly less extraretinal neovascularization than did wild-type mice. The mean number of extraretinal neovascular buds per cross section was significantly lower in MMP-2^−/− mice than in wild-type mice (P < 0.05). The expression of other angiogenesis-related factors, vascular endothelial growth factor and pigment epithelium-derived factor, was not different between wild-type and MMP-2^−/− mice.

conclusions. MMP-2 may be essential in the regulation of retinal neovascularization. Pharmacologic intervention using MMP inhibitors may be a future therapeutic approach for angiogenic retinal diseases.

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 results 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. The retinal microvascular endothelium participates in neovascularization in a stepwise manner. During the initiation phase, endothelial cells respond to locally produced angiogenic factors and upregulate the expression of extracellular proteinases. This phase is followed by the invasive phase, which is 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}  

Extracellular proteinases have important roles in the regulation of endothelial cell migration and extracellular matrix remodeling during angiogenesis. Matrix metalloproteinases (MMPs) are a highly regulated family of at least 14 structurally related enzymes capable of degrading most, if not all, of the components of the extracellular matrix. ^{5 6} MMPs are grouped into collagenases, gelatinases, stromelysins, and matrilysin, which are secreted as inactive proenzymes and activated by proteolysis. In angiogenesis, new blood vessels originate from postcapillary venules through a series of events, beginning with the disruption of the endothelial cell basement membrane and extracellular matrix by proteolytic enzymes. Several MMPs are believed to be important in this process, but particular interest has been focused on MMP-2 and -9 because they preferentially degrade basement membrane components such as type IV collagen. ⁷  

MMP inhibitors have been the focus of anticancer research and have been used in clinical trials to block angiogenesis in tumors as well as tumor metastasis. ⁸ The results with the broad-spectrum MMP inhibitor marimastat (British Biotech, Oxford, UK) and its analogue batimastat (British Biotech) have been disappointing, mainly because unexpected side effects (such as musculoskeletal pain and inflammation) limited the MMP inhibitor dosages administered in subsequent trials. In addition, the results of human clinical trials of a broad-spectrum oral MMP inhibitor (protocol AG3340; Agouron Pharmaceuticals, Inc., San Diego, CA) for controlling choroidal neovascularization (CNV) secondary to age-related macular degeneration have been disappointing (Blodi BA, et al. IOVS 2001;42:ARVO Abstract 1673). Moreover, it caused severe and intolerable side effects. The results of clinical trials of broad-spectrum MMP inhibitors point to the need to determine the role of specific MMPs in specific stages of disease progression, to design more selective inhibitors that are devoid of adverse reactions induced by broad-spectrum inhibitors, and to identify a more effective drug delivery route. 

Although several investigators have described a possible role of MMPs in retinal angiogenesis, ^{4 9 10} there has been no conclusive evidence regarding a critical role of MMPs in retinal angiogenesis. There is also no evidence regarding which of the MMPs is a valid target for the treatment of retinal angiogenesis. To study the putative role of endogenous MMPs in retinal neovascularization, we used an established mouse model and compared the retinal neovascularization observed in wild-type mice to that in mice without the MMP-2 or -9 gene. Our findings indicate that MMP-2 deficiency significantly reduced retinal neovascularization. 

Materials and Methods

Animals

The experiments were performed using C57Bl/6 (MMP-2^+/+, MMP-9^+/+), MMP-2-deficient (MMP-2^−/−), and MMP-9–deficient (MMP-9^−/−) mice. C57BL/6 mice with targeted disruption of the MMP-2 (knockout MMP-2) or the MMP-9 (knockout MMP-9) gene were generated as described previously. ^{11 12} Genotyping to verify the absence or presence of the MMP-2 or -9 gene or of the targeting vector was accomplished using polymerase chain reaction of DNA from tail biopsy specimens. All experiments were approved by the Ethics Committee for Animal Use in Research and Education of the Tokyo Medical and Dental University and conformed to the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research. 

Murine Models of Oxygen-Induced Retinopathy

Oxygen-induced retinopathy was induced in newborn mice according to the protocol of Smith et al. ¹³ On postnatal day (P)7, mice were placed along with their dam into a custom-built chamber. They were maintained in 75% oxygen for up to 5 days (P12), after which they were transferred back to cages in room air (normoxia). Room temperature was maintained at 68°F, and the rooms were illuminated with standard fluorescent lighting on a 12-hour light–dark cycle. Newborn mice were nursed by the dam and given food (standard mouse chow) and water ad libitum. At P17, the pups were killed with an overdose of pentobarbital sodium, and their eyes were rapidly removed and frozen in optimal cutting temperature embedding compound (OCT; Miles, Elkhart, IN). 

Staining of Neovascularization with a Vascular Cell–Specific Marker

Serial sections (10 μm) were cut through the entire extent of each eye. The entire eye was sampled by staining sections roughly 50 to 60 μm apart, which provided 13 sections per eye for analysis. Sections were histochemically stained as previously described ^{14 15} with biotinylated Griffonia simplicifolia lectin B4 (GSA; Vector Laboratories, Burlingame, CA), which selectively binds to vascular cells. Slides were incubated in 4% paraformaldehyde for 30 minutes, washed with 0.05 M Tris buffer (TB; pH 7.4), incubated in methanol-H₂O₂ for 10 minutes at 4°C, washed with 0.05 M TB, and incubated for 30 minutes in 10% normal swine serum. Slides were rinsed with 0.05 M TB and incubated 2 hours at 37°C with 1:20 GSA, rinsed again with 0.05 M TB, and incubated with undiluted streptavidin-phosphatase (Dako, Glostrup, Denmark) for 30 minutes at room temperature. After a 10-minute wash in 0.05 M TB (pH 7.6), the slides were developed with diaminobenzidine (Dako). 

Quantitation of Extraretinal Neovascularization

Lectin-stained sections were examined by microscope (at 400 × magnification; model Q 600 HR; Leica, Heidelberg, Germany), and images were digitized with a three-color charge-coupled-device video camera and a frame grabber. The accompanying software (Quantimet; Leica) was used to delineate lectin-stained cells, and the area was measured. 

Statistical Analysis

The mean of the 13 measurements from each eye was used to generate a single experimental value for each animal group. Group comparisons were performed with the nonparametric Kruskal-Wallis test. Pair-wise comparisons were performed with the Mann-Whitney post hoc test when the results of the Kruskal-Wallis test were significant. The level of significance was set at P < 0.05. 

Retinal Flatmounts

Retinal flatmounts were prepared by modification of a previously described technique. ^{13 14} After mice were maintained in 75% oxygen for up to 5 days (P12), they were anesthetized, the descending aorta was clamped, and the right atrium was cut and perfused through the right ventricle with 1 mL phosphate-buffered saline containing 50 mg/mL fluorescein-labeled dextran (2 × 10⁶ average MW; Sigma-Aldrich Co., St. Louis, MO). The eyes were removed and fixed for 1 hour in 10% phosphate-buffered formalin. The cornea and lens were removed, and the entire retina was then carefully dissected from the eye cup, radially cut from the edge of the retina to the equator in all four quadrants, and flatmounted in aqueous medium (Aquamount; BDH, Poole, UK) with photoreceptors facing downward. Flatmounts were carefully examined by fluorescence microscopy, and photographed (T-64 film; Eastman Kodak, Rochester, NY). 

Immunohistochemical Analysis of the Angiogenic Factors

Reports have suggested an interaction between the expression of angiogenic factors and the expression of MMPs. ^{16 17 18 19} To examine the role of other angiogenesis-related factors involved in hypoxia-induced extraretinal neovascularization, we performed immunohistochemical analysis of the angiogenesis-related factors vascular endothelial growth factor (VEGF) and a recently discovered potent antiangiogenic factor in the eye, pigment epithelium-derived factor (PEDF). ²⁰ At P12 and P17, eyes were enucleated, and 8-μm cryostat sections were prepared for immunohistochemical analysis. Sections from the periodate-lysine-paraformaldehyde–fixed samples were treated with 0.3% H₂O₂ and 10% normal horse serum to block endogenous peroxidase and nonspecific binding, respectively. The sections were then treated with rabbit polyclonal antibodies against human VEGF (1:3000 dilution; Santa Cruz Biotechnology, Santa Cruz, CA) or rabbit polyclonal antibodies against human PEDF (1:3000 dilution; Transgenic, Kumamoto, Japan) at room temperature for 90 minutes. After reacting with goat antibodies against rabbit IgG conjugated to peroxidase labeled-dextran polymer (no dilution; EnVision+ Rabbit; Dako) at room temperature for 30 minutes, the color was developed with aminoethyl carbazole (AEC; Zymed Laboratories, San Francisco, CA) in 50 mM Tris-HCl (pH 7.6), containing 0.006% H₂O₂. Counterstaining was performed with hematoxylin. As a negative control, primary antibodies were replaced with nonimmune rabbit IgG (Dako), and the sections were incubated at room temperature for 2 hours before immunostaining. 

Results

Reduction of Extraretinal Neovascularization in MMP-2^−/− Mice

Newborn mice that were raised in 75% oxygen for 5 days (P7–P12) and returned to room air showing development of extraretinal neovascularization by P17. In our studies, the MMP-2^−/− mice exhibited significantly less extraretinal neovascularization than did wild-type mice (Fig. 1) . The mean number of extraretinal neovascular buds per cross section was significantly lower in MMP-2^−/− mice than in wild-type mice (P < 0.05; Fig. 2 ). Extraretinal neovascularization, however, was not significantly different between wild-type and MMP-9^−/− mice. Neovascularization was most prominent at the border zones between vascular and avascular tissue. Neovascularization did not occur in adult mice exposed to identical hyperoxic conditions or in room-air–raised control animals (data not shown). There was no difference in the amount of extraretinal neovascularization in mice without ischemic retinopathy between wild-type, MMP-2^−/−, and MMP-9^−/− mice (Fig. 2) . 

There was no difference in the size of the capillary-free area in retinal flatmounts at P12 before the development of extraretinal neovascularization in either wild-type or MMP-2 null mice (Fig. 3A 3B , respectively). 

VEGF and PEDF Immunostaining in Wild-Type and MMP-2^−/− Mice

In hypoxia-induced extraretinal neovascularization, other angiogenic factors are also involved in the development of neovascularization. Some reports have suggested an interaction between the expression of these angiogenic factors and the expression of MMPs. ^{16 17 18 19} Therefore, to examine the possibility that the observed difference in the angiogenic response between wild-type and MMP-2^−/− mice was due to modification of other angiogenesis factors, we investigated VEGF (a potent angiogenic factor) and PEDF (a potent antiangiogenic factor in the eye) expression in hyperoxia-treated wild-type and MMP-2^−/− mice. The results of a typical immunohistochemical analysis are shown in Figures 4 and 5 . VEGF and PEDF protein expression was consistent with their characteristics as hypoxia-regulated angiogenic and antiangiogenic factors, respectively. At P12, there was VEGF protein expression in the ganglion cell layer and in the inner nuclear layer in room-air–raised wild-type and MMP-2^−/− mice (Figs. 4A 4B) . The VEGF protein expression in oxygen-incubated P12 wild-type and MMP-2^−/− mice was reduced compared with the age-matched control (Figs. 4C 4D) . At P17, VEGF expression increased in oxygen-incubated wild-type and MMP-2^−/− mice (Figs. 4H 4I) , and there were higher VEGF levels than in age-matched room-air–raised P17 mice (Figs. 4F 4G) . There was no difference in VEGF expression between wild-type and MMP-2^−/− mice from P12 to P17. At P12, there was faint PEDF protein expression in Müller glial cell process and in the ganglion cell layer in room-air–raised wild-type and MMP-2^−/− mice (Figs. 5A 5B) . The PEDF protein expression in oxygen-incubated P12 wild-type and MMP-2^−/− mice increased compared with that in the age-matched control (Figs. 5C 5D) . At P17, the PEDF protein levels decreased in oxygen-incubated wild-type and MMP-2^−/− mice, and the levels were lower than that of age-matched room-air–raised P17 mice (Fig. 5F 5G 5H 5I) . Also, there was no difference in PEDF expression between wild-type and MMP-2^−/− mice throughout this period. 

Discussion

Several research groups have examined the expression pattern and possible roles of MMPs in retinal neovascularization. Das et al. ²¹ reported that human diabetic epiretinal neovascular membranes contain high levels of extracellular proteinases including MMP-2 and -9 and urokinase. Majka et al. ¹⁰ reported increased MMP-2, MMP-9, and human membrane Type 1-MMP expression in the retinas of a murine model of proliferative retinopathy. Das et al. ⁴ reported that an intraperitoneal injection of MMP inhibitor significantly inhibited retinal neovascularization in a murine model of oxygen retinopathy. Although these studies suggest a possible link between MMP expression and ocular angiogenesis, there has been no direct evidence of a relation between retinal angiogenesis and MMPs. In the present study, we examined the hypothesis that MMPs have an important role in ocular angiogenesis, by studying retinal neovascularization in genetically engineered mice without functional expression of two MMP genes. The use of MMP-deficient mice is advantageous because the specific activities of the particular MMP gene that was knocked out are eliminated. In addition, nonspecific inhibition of extracellular matrix components and other MMPs is minimized. Our data using MMP-deficient mice provide more striking evidence for a critical role of these enzymes in ocular angiogenesis. 

In this report, we demonstrated that the angiogenic response induced in the oxygen-induced retinopathy model is markedly reduced in mice that do not have a functional MMP-2 gene, compared with that in wild-type mice. To the best of our knowledge, this is the first study to show direct involvement of MMPs in inducing extraretinal neovascularization. The precise mechanism by which MMPs influence angiogenesis is not clear. Kato et al. ²² have recently demonstrated, however, that endothelial cells from gelatinase-A–deficient mice fail to display normal outgrowth in the presence of 5 ng/mL basic fibroblast growth factor (bFGF) using aortic ring assays in synthetic matrix (Matrigel; BD Biosciences, Bedford, MA). They suggest that endothelial cells that have no functional gelatinase-A may fail to traverse the basement membrane. Therefore, degradation of the basement membrane by gelatinase may be an important event in angiogenesis. 

The present study demonstrated a critical role for MMP-2, but not for MMP-9, in the development of extraretinal neovascularization. Several important experimental observations suggest the importance of MMP-2 among the various MMPs in the development of angiogenesis. Fang et al. ²³ has reported that suppression of MMP-2 alone inhibits the transition from the prevascular to the vascular stage during tumor development in an experimental tumor model. Also, endothelial cells produce MMP-2 during differentiation into capillary tube–like structures, and exogenous addition of MMP-2 enhances this process. ²⁴ It has been demonstrated recently that MMP-2 hydrolyzes the ectodomain of FGF receptor type 1 and yields a soluble FGF receptor that may modulate the mitogenic and angiogenic activities of FGF. ²⁵ These data do not preclude the possibility that other MMPs also have an important role in the development of extraretinal neovascularization. Others have suggested an important role for MMP-9 in extraretinal neovascularization. ^{8 26 27} The present study, however, suggests that MMP-2 has a more important role in extraretinal neovascularization than does MMP-9, and given the redundancy of this enzyme family, it is particularly striking that the suppression of only one MMP gene has such a profound effect on the development of extraretinal neovascularization. 

Berglin et al. ²⁸ and Lambert et al. ²⁹ have reported a significant reduction in laser-photocoagulation–induced development of CNV in MMP-2- or -9–deficient mice. It is unclear why extraretinal neovascularization in MMP-9–deficient mice was not reduced in the present study. One explanation is the difference in the experimental models used to induce ocular neovascularization. The laser-induced CNV model uses intense laser beams and causes massive destruction of Bruch’s membrane, which secondarily causes an intensive wound-healing response that does not occur in the oxygen-induced retinopathy model. Another explanation is that the angiogenic response may not be the same between different vascular beds in the posterior segment of the eye (retina versus choroid). The cause of the differences in our and others’ results requires further investigation. 

The expression of many angiogenesis-related factors, including VEGF and PEDF, is also influenced by hypoxia-induced extraretinal neovascularization. PEDF is a potent antiangiogenic factor recently identified in the retina. ²⁰ A migration assay of bovine capillary endothelial cells indicated that PEDF is the most potent natural angiogenesis inhibitor, and PEDF is considered to be the key factor associated with avascularity of the eye. ²⁰ In addition, PEDF is downregulated by hypoxia. ²⁰ In the eye, it appears that a balance between VEGF and PEDF is essential for new vessel formation. ³⁰ Observations that PEDF levels are lower in the aqueous humor or in the vitreous of active diabetic retinopathy further support this theory. ^{31 32 33} Also, some reports suggest that there is an interaction between VEGF expression and MMP expression. Although there are some reports that MMPs regulate VEGF expression, ^{16 18 19} there are other reports of VEGF-induced upregulation of MMPs. ¹⁷ Therefore, to examine the possibility that the observed difference in the angiogenic response between wild-type and MMP-2^−/− mice is due to the activation of other angiogenesis-related factors including VEGF, we performed immunohistochemical analysis for the representative angiogenesis factors in the retina, VEGF and PEDF. VEGF and PEDF protein expression was consistent with their characteristics as hypoxia-regulated angiogenic and antiangiogenic factors, respectively, as shown in the previous studies. ^{34 35} At P17, the time at which the amount of extraretinal neovascularization is maximal, there was a significant reduction in retinal PEDF and a substantial increase in VEGF in the hyperoxia-treated wild-type and MMP-2^−/− mice compared with the age-matched room-air–raised control (Figs. 4 5) . There was, however, no significant difference in the expression of these angiogenesis factors between wild-type and MMP-2^−/− mice from P12 to P17. This finding suggests that MMP itself has a critical role in the regulation of extraretinal angiogenesis in MMP-2^−/− mice. 

In summary, experiments in knockout mouse models suggest that MMP-2 may be of particular importance in the regulation of extraretinal neovascularization. Pharmacologic intervention with MMP-2 inhibitors may be a future therapeutic approach for angiogenic eye diseases. 

 Supported by Grant 14571659 from the Japan Society for the Promotion of Science.

 Submitted for publication March 11, 2003; revised July 3, 2003; accepted August 4, 2003.

 Disclosure: K. Ohno-Matsui, None; T. Uetama, None; T. Yoshida, None; M. Hayano, None; T. Itoh, Discovery Research Laboratories (E); I. Morita, None; M. Mochizuki, 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.

 

Figure 1.

 

Assessment of retinal blood vessels and neovascularization in wild-type, MMP-2^−/−, and MMP-9^−/− mice with ischemic retinopathy. Retinal frozen sections were histochemically stained with the endothelial cell–selective lectin Griffonia simplicifolia, by the peroxidase-antiperoxidase technique. Retinal blood vessels within the retina and neovascularization on the surface of the retinas were stained with reaction product. (A, B) Wild-type mice with ischemic retinopathy had clumps of endothelial cells on the surface of the retina due to neovascularization. (C, D) MMP-9^−/− mice with ischemic retinopathy also had numerous endothelial cells on the surface of the retina. (E, F) MMP-2^−/− mice with ischemic retinopathy had fewer endothelial cells on the surface of the retina.

Figure 2.

 

Quantification of the total area of endothelial cell staining in retinal sections of wild-type, MMP-2^−/−, and MMP-9^−/− mice, with and without ischemic retinopathy. Neovascularization after hyperoxia was reduced in MMP-2^−/− mice compared with wild-type. There was no difference among nonischemic wild-type, MMP-9^−/−, and MMP-2^−/− mice, nor was there a difference between ischemic wild-type and MMP-9^−/− mice. Mice were raised either in room air from birth to P17 (room air) or placed in 75% oxygen from P7 until P12 and returned to room air until P17 (ischemic). Group differences were analyzed with the Kruskal-Wallis and Mann-Whitney tests. P < 0.05 was considered statistically significant.

Figure 3.

 

Representative photograph of flatmounted retinas perfused with fluorescein-dextran from wild-type and MMP-2^−/− mice. Age-matched animals were exposed to 75% oxygen from P7 to P12, and the retinal vasculature was examined by fluorescein angiography on P12. On P12, there was no difference in the size of the capillary-free area in wild-type (A) and MMP-2^−/− mice (B).

Figure 4.

 

VEGF staining in the retinas of wild-type and MMP-2^−/− mice, with and without ischemic retinopathy at P12 and P17. P12 (A) wild-type mouse and (B) MMP-2^−/− mouse without ischemic retinopathy. P12 (C) wild-type mouse and (D) MMP-2^−/− mouse with ischemic retinopathy. (E) VEGF staining was absent on an adjacent section when nonimmune serum was substituted for primary antibody. P17 (F) wild-type mouse and (G) MMP-2^−/− mouse without ischemic retinopathy. P17 (H) wild-type mouse and (I) MMP-2^−/− mice with ischemic retinopathy. GCL, ganglion cell layer; INL, inner nuclear layer; ONL, outer nuclear layer.

Figure 5.

 

PEDF staining in the retinas of wild-type and MMP-2^−/− mice with and without ischemic retinopathy at P12 and P17. P12 (A) wild-type mouse and (B) MMP-2^−/− mouse without ischemic retinopathy. P12 (C) wild-type mouse and (D) MMP-2^−/− mouse with ischemic retinopathy. (E) PEDF staining was absent on an adjacent section when nonimmune serum was substituted for primary antibody. P17 (F) wild-type mouse and (G) MMP-2^−/− mouse without ischemic retinopathy. P17 (H) wild-type mouse and (I) MMP-2^−/− mice with ischemic retinopathy. Abbreviations as in Figure 4 .

The authors thank Tomoko Yoshida for excellent technical support.