Abrogation of MMP-9 Gene Protects Against the Development of Retinopathy in Diabetic Mice by Preventing Mitochondrial Damage

Male, wild-type (WT) C57BL/6 J or MMP-KO (B6.FVB [Cg]-MMP-9tm1Tvu/J) mice were obtained from Jackson Laboratory (Bar Harbor, ME). A group of WT and MMP-KO mice were made diabetic (WT-D and KO-D groups, respectively) by streptozotocin injection (55 mg/kg body wt) for five consecutive days. Mice with blood glucose 250 mg/dL or higher, 3 days after the last injection of streptozotocin, were considered diabetic (WT-N and KO-N groups). Age-matched normal WT and MMP-KO mice served as controls. MMP-KO mice have significantly decreased expression of MMP-9 in their retina, and these mice can be maintained in diabetic conditions (8). Seven months after induction of diabetes, mice were killed by carbon dioxide asphyxiation. One eye was used to remove the retina, and the other eye was incubated in 10% buffered formalin for histopathology. The treatment of the animals conformed to the Association for Research in Vision and Ophthalmology Resolution on the Use of Animals in Research and Institutional Guidelines (Wayne State University).

Human postmortem eyes were obtained from Midwest Eye Banks (Ann Arbor, MI). Diabetic donors (10–12 years of diabetes) with retinopathy were aged 60–74 years, and nondiabetic donors aged 44–75 years. The eyes were enucleated between 6 and 9 h after death (8). A portion of the freshly isolated retina was processed for mitochondrial isolation.

Bovine retinal endothelial cells (BRECs) were grown in Dulbecco’s modified Eagle’s medium containing 15% heat-inactivated FBS, 5% Nu-serum, 50 μg/mL heparin, and 50 μg/mL endothelial cell growth supplement. Cells from third to fourth passage were transfected with MMP-9 small interfering RNA (siRNA) using transfection reagents and siRNA duplex from Santa Cruz Biotechnology (Santa Cruz, CA), as previously described (8,12). For control, the cells were transfected with nontargeting scramble RNA. At the end of the transfection, the medium was replaced with either 5 or 20 mmol/L glucose media, and the cells were incubated for four additional days. This method yields about 50% transfection efficiency, and the transfection process does not alter cell survival or proliferation (8). For control, parallel incubations were run using nontransfected cells, and for osmolarity control, cells were incubated in 20 mmol/L mannitol instead of 20 mmol/L glucose.

The retina from the formalin-fixed eyes was rinsed overnight with running water, and the microvasculature was isolated by incubating the retina at 37°C with 3% crude trypsin (Invitrogen-Gibco, Grand Island, NY) containing 200 mmol/L sodium fluoride for 45 to 70 min. The neuroretinal tissue was gently brushed away, and the apoptotic vascular cells (endothelial cells/pericytes) were detected by incubating the retinal microvasculature with terminal deoxyribonucleotide transferase-mediated dUTP nick-end labeling (TUNEL) stain (In Situ Cell Death kit; Roche Molecular Biochemicals, Indianapolis, IN). After TUNEL staining, the microvessels were stained with periodic acid Schiff–hematoxylin and examined by light microscopy (10,12).

Mitochondria were isolated from the retina or BRECs using mitochondria isolation kit (Pierce, Rockford, IL) according to the manufactures’ instructions, as routinely used in our laboratory (12). The mitochondrial pellet was washed with PBS and suspended in the mitochondria lysis buffer (2% CHAPS in 25 mmol/L Tris, 0.15 mol/L NaCl, pH 7.2). Protein concentration was quantified by the bicinchoninic acid assay (Sigma-Aldrich, St Louis, MO).

Activation of MMP-9 was quantified in the mitochondrial fraction by in situ zymography and confirmed by ELISA. For gelatinase activity, mitochondrial protein (10–20 μg) was separated on nonreducing 8% SDS-PAGE containing 0.1% gelatin. The gel was washed with 2.5% Triton X-100, and incubated for 18 h at 37°C in substrate buffer containing 50 mmol/L Tris-HCl (pH 7.5), 5 mmol/L CaCl₂, 200 mmol/L NaCl₂, and 0.02% Brij-35. The gel was stained with 0.25% Brilliant Blue R-250 (Sigma-Aldrich) and then destained with 10% acetic acid and 30% methanol (8). The intensity of the active band of MMP-9 (85–80 kD) was quantified using Un-Scan-It Gel digitizing software.

Activation of MMP-9 in the mitochondria was also performed using an ELISA kit (Amersham, Buckinghamshire, U.K.). The final color was developed using a specific peptide chromogen, and the resulting color was measured at 405 nm. Diabetes-induced activation of MMP-9 was calculated by quantifying the immunocaptured MMP-9 that was not treated with amino-phenyl mercuric acetate (7,8).

A Cell Death Detection ELISA^PLUS kit (Roche Diagnostics, Indianapolis, IN) was used to detect cell apoptosis. Monoclonal antibodies directed against DNA and histones were used to quantify the relative amounts of mono- and oligonucleosomes generated from apoptotic cells (8).

Retinal mitochondrial superoxide levels were quantified using a cell-permeable probe, MitoTracker Red CM-H₂XROS (Molecular Probes, Eugene, OR). Freshly harvested retina was homogenized in mitochondria isolation buffer, as described previously (10). Mitochondrial protein (2–5 μg in PBS) was supplemented with 400 nmol/L Mitotracker Red CM-H₂XROS. After incubating for 30 min at 37°C, the resultant fluorescence was measured at 599-nm emission wavelength and 579-nm excitation wavelength using LS 55 fluorescence spectrometer (Perkin-Elmer, Waltham, MA).

The collapse of mitochondrial membrane potential was investigated by quantifying the swelling of the isolated mitochondria using a spectrophotometric method routinely used in our laboratory (10,12,15). Decrease in absorbance, induced by calcium chloride, was followed for 3–5 min at 540 nm.

Mitochondrial DNA damage was determined by mitochondrial genome–specific quantitative extended-length PCR with the GeneAmp XL PCR kit (Applied Biosystems, Foster City, CA). In brief, 15 ng DNA was amplified in a reaction mixture containing 1× XL PCR buffer II, 200 μmol/L dNTP, 1.1 mmol/L Mg (OAc)₂, 0.1 μmol/L genome-specific primers (mtDNA-long, forward AAA ATC CCC GCA AAC AAT GAC CAC CCC and reverse GGC AAT TAA GAG TGG GAT GGA GCC AA; mtDNA-short, forward CCT CCC ATT CAT TAT CGC CGC CCT TGC and reverse GTC TGG GTC TCC TAG TAG GTC TGG GAA) and 1 unit rTth DNA polymerase. Mitochondrial DNA was amplified with 24 cycles, and PCR products were resolved on agarose gel. Relative amplification was quantified by normalizing the intensity of the long product to the short product (mtDNA = 13.4 kb/210 bp). These methods are routinely used in our laboratory (16).

Ultrastructutre of the mitochondria in the retinal vasculature region was evaluated by transmission electron microscopy (TEM). The enucleated eyes were dissected along the equators, and the retina was incubated for 24 h in 2.5% glutaraldehyde (in 100 mmol/L phosphate buffer, pH 7.4). The retina was postfixed in 1% buffered osmium tetroxide and dehydrated using graded ethanol solutions. After embedding the samples in 812 resin (Electron Microscope Science, Hatfield, PA), ultrathin transverse sections (70–80 nm) of the areas near the microvasculature region were prepared. The sections were stained with uranyl acetate and lead citrate and viewed by TEM (Zeiss EM 900, Oberkochen, Germany). At least eight random images were recorded from each independent preparation and enlarged to 20,000× magnification. Mitochondria in the endothelial cells were analyzed in a blind manner by ImageJ software to calculate their area (μm²) and average length (μm) as previously described (17).

BRECs grown on 12-mm diameter cover slips coated with 0.1% gelatin were incubated with 5 mmol/L or 20 mmol/L glucose for 4 days. At the end of the incubation, the cells were incubated with 500 nmol/L MitoTracker (green; Molecular Probes) for 30 min (12,16). The cells were washed with PBS, fixed in cold methanol for 20 min, and permeabilized in 0.25% Triton X-100 for 15 min. They were blocked in 5% BSA for 1 h and incubated with anti–MMP-9 antibody (Santa Cruz Biotechnology). After washing the cells with PBS, they were incubated with anti-rabbit secondary antibody (Taxes red conjugated; Molecular Probes), washed with PBS, and mounted on slides using Vecta Shield containing DAPI (Vector Laboratories, Burlingame, CA). The slides were examined under a Zeiss ApoTome using 40× magnification (Carl Zeiss Inc., Chicago, IL).

Retinal mitochondrial protein (25–30 μg) was separated on a 4–20% SDS-PAGE and transferred to nitrocellulose membranes using a semidry transfer. The membranes were incubated with the antibody against the target protein (Bax, TOM34, and TIM44 from Santa Cruz Biotechnology; heat shock protein 70 [Hsp70] and Hsp60 from Abcam, Cambridge, MA). Membranes were reprobed with Cox IV to evaluate the lane-loading control. The band intensities were quantified by Un-Scan-It software.

Retinal or cell homogenate (150–200 μg) was incubated overnight at 4°C with 2 μg of anti–MMP-9 antibody. Agarose beads A/G (Santa Cruz Biotechnology) were mixed with protein antibody for 1 h at 4°C. The immune precipitate was collected and washed three times with the lysis buffer without EDTA or EGTA. Samples were boiled in 2× Laemmli buffer and subjected to electrophoresis on 10% SDS-PAGE (12,15). Western blot analysis was performed using antibodies against Hsp70, Hsp60, or TIM44.

Each measurement was made in duplicate, and the assay was repeated three or more times. Data are expressed as mean ± SD. Statistical analysis was performed using the nonparametric Kruskal-Wallis test followed by Mann-Whitney test, and P < 0.05 was considered statistically significant.

Diabetes of ∼7 months in WT mice, as expected, significantly increased the apoptosis of capillary cells. The number of TUNEL-positive cells in the trypsin-digested retinal microvasculature was about twofold higher compared with the WT-nondiabetic (WT-N) mice (Fig. 1A). In the same retinal vasculature, the number of degenerative capillaries was also significantly increased (Fig. 1B). Consistent with the increase in active MMP-9 in the retina in diabetes (7,8), retinal mitochondria also had an ∼25% increase in active MMP-9, as confirmed by both ELISA and in situ zymography assays (Fig. 2A and B). In the same WT-diabetic (WT-D) mice, mitochondrial superoxide levels were elevated by >100%, mtDNA was damaged (40% reduction in the amplification of the long fragment of mtDNA), mitochondrial permeability was increased, and the translocation of Bax into the mitochondria was increased by ∼50% compared with the values from age-matched WT-N mice (Fig. 2C–F). A significant number of mitochondria in WT-D mice in the endothelium region of the microvasculature were elongated and swollen with electron-lucent matrix and partial cristolysis compared with the intact and tightly packed lamellar cristae in WT-N mice (Fig. 3A). The average area and length of the mitochondria in the endothelial region were ∼40% larger in WT-D mice compared with WT-N mice (Fig. 3B and C). The chaperone proteins (Hsp70 and Hsp60) and the translocators (TOM34 and TIM44) were significantly decreased in the retinal mitochondria (Fig. 4A–D), but the binding of MMP-9 with Hsp70, Hsp60, or TIM44 was increased by 40–80% (Fig. 5).

Retinal microvasculature of MMP-KO mice was protected from diabetes-induced accelerated apoptosis, and the number of degenerative capillaries was also significantly lower compared with the values obtained from WT-D mice (Fig. 1). In the same MMP-KO mice, diabetes had no effect on active MMP-9 in the retinal mitochondria; this was significantly lower compared with WT-D mice (Fig. 2). In addition, mitochondrial dysfunction was protected; the levels of mitochondrial superoxide, damage to the mtDNA, and swelling were significantly lower in MMP-KO diabetic mice compared with WT-D mice (Fig. 2C–F). TEM of the endothelium region of the retinal microvasculature from MMP-KO diabetic mice showed mitochondria with normal ultrastructure; their average area and length were significantly lower compared with the WT-D group, and the cristae were more tightly packed and lamellar compared with the WT-D mice (Fig. 3). Chaperone proteins and the translocase system remained normal in the mitochondria; expressions of Hsp70, Hsp60, TOM34, and TIM44 were significantly higher in the retinal mitochondria from MMP-KO diabetic mice compared with WT diabetic mice, and these values were similar to those observed in MMP-KO mice that are nondiabetic (KO-N) and WT-N mice (Fig. 4). Similarly, an increase in the binding of retinal MMP-9 with Hsp70, Hsp60, or TIM44, as observed in WT diabetic mice, was also prevented in the MMP-KO diabetic mice (Fig. 5).

To confirm our results obtained from the rodent model, major parameters were quantified in the retinal mitochondria prepared from human donors with diabetic retinopathy. Mitochondria from donors with diabetic retinopathy had 60–70% increase in active MMP-9 compared with age-matched, nondiabetic donors, and significantly decreased expression of Hsp70 and Hsp60. The binding of MMP-9 with Hsp70 or with Hsp60 was also significantly increased (Fig. 6).

Exposure of retinal endothelial cells, the site of histopathology characteristic of diabetic retinopathy, to high glucose increased the accumulation of MMP-9 in the mitochondria; cells incubated in high glucose had increased staining of MMP-9 (red) in the mitochondria compared with the cells exposed to normal glucose (Fig. 7A). In the same cell preparations, the active MMP-9 was ∼40% higher in the mitochondria from the cells exposed to high glucose as confirmed by both in situ zymography (Fig. 7B) and ELISA (Fig. 7C), and cell apoptosis was elevated by >75% (Fig. 7D). Consistent with the retina, high glucose decreased mitochondrial Hsp70 and Hsp60 expressions and increased the binding of MMP-9 with Hsp70 or Hsp60 as evidenced by their increased expression in the cells immunoprecipitated for MMP-9 (Fig. 8).

Regulation of MMP-9 by siRNA prevented a glucose-induced increase in mitochondrial MMP-9 (Fig. 7B and C) and cell apoptosis (Fig. 7D). Furthermore, a decrease in mitochondrial Hsp70 and Hsp60, and increase in the binding of MMP-9 with these chaperone proteins, experienced by BRECs in high-glucose conditions, was also ameliorated in the cells transfected with MMP-9 siRNA. The values obtained in high-glucose conditions from the cells transfected with MMP-9 siRNA were significantly different from the cells untransfected or transfected with scramble RNA (Fig. 8).

This is the first report demonstrating a direct role of MMP-9 in the development of diabetic retinopathy; here we show that the genetic ablation of MMP-9 ameliorates diabetic mice from the development of retinopathy. MMP-KO mice, diabetic for ∼7 months, are protected from histopathology characteristic of early signs of retinopathy, and retinal capillary cell apoptosis is not accelerated. Their retinal mitochondria have normal levels of active MMP-9, function, and ultrastructutre, and mtDNA is not damaged. Furthermore, abrogation of MMP-9 gene also protects retinal mitochondrial transport machinery and chaperone proteins from diabetes-induced abnormalities. Retinal mitochondria from donors with diabetic retinopathy have similar alterations in active MMP-9 and chaperone proteins. Silencing of MMP-9 gene using MMP-9 siRNA in isolated retinal endothelial cells, the site of histopathology associated with diabetic retinopathy, prevents damage to the mitochondrial chaperone machinery, experienced by cells in high glucose. Together, these results strongly imply a significant role for mitochondrial MMP-9 in the development of diabetic retinopathy, and the mechanism appears to be through damage of retinal mitochondria.

Matrix metalloproteinases can cleave extracellular matrix constituents and nonmatrix proteins (18). These enzymes exert various roles at different stages of the disease. In diabetes, retinal MMPs are postulated to facilitate an increase in vascular permeability through proteolytic degradation of the tight junction proteins and disruption of the overall tight junction complex (19–21). We have shown that the activation of MMP-9, the most complex member of the MMP family (5,22), is modulated by a small molecular weight G protein, H-Ras, and triggers apoptosis of retinal capillary cells (7). Now we provide convincing data showing a direct role of MMP-9 in the development of diabetic retinopathy. The retinal vasculature of diabetic mice with the MMP-9 gene abrogated is protected from accelerated apoptosis and from the histopathology characteristic of diabetic retinopathy. This implies that the inhibitors of MMP-9 have potential to impede the development of retinopathy in diabetic patients. Optimistically, these inhibitors are being used in clinical trials for other diseases, including chronic obstructive pulmonary disease and multiple sclerosis (23).

The retina and its capillary cells experience increased oxidative stress in diabetes, and increased oxidative stress is considered to play a significant role in the development of diabetic retinopathy (4,24–26). Dysfunctional retinal mitochondria become swollen and their permeability is increased. Cytochrome c leaks into the cytosol, and Bax translocates into the mitochondria (9,10). Data presented here show increased active MMP-9 in the mitochondria and its capillary cells in hyperglycemic conditions, and these results are in agreement with the results from others showing association of active mitochondrial MMP-9 with mechanical dysfunction of cardiac myocytes (27).

The retinae of MMP-KO diabetic mice are protected from increased superoxide radicals and mtDNA damage, and their mitochondrial permeability remains normal. This suggests that the retinal-activated MMP-9 in diabetes damages mitochondria and increases their permeability, Bax moves into the mitochondria, and apoptotic machinery is activated. There remains a possibility that the TUNEL staining used to detect apoptosis of retinal capillary cells is also detecting late necrosis. However, the beneficial effect of MMP-9 gene knockout on diabetes-induced increased translocation of proapoptotic Bax into the mitochondria and on the swelling of the mitochondria further confirms the role of MMP-9 in mitochondrial damage resulting in apoptosis. Furthermore, MMP-9 gene ablation also prevents diabetes-induced ultrastructural changes in the mitochondria in the retinal microvasculature, suggesting a major role of MMP-9 in diabetes-induced mitochondrial dysfunction and disruption. Consistent with these, MMP-9 is considered as a negative regulator of mitochondrial function (11,28), and active MMP in cardiac myocytes is postulated to initiate a negative feedback cycle that degrades mitochondrial membrane potential and impairs function (29). Mitochondrial ultrastructural alterations are observed in cells undergoing apoptosis, and the swollen state is considered to coincide with the beginning stages of cell death (30). In the pathogenesis of diabetic neuropathy, elongated and swollen mitochondria with disrupted cristate are observed in ganglion cells (31,32). Our results demonstrating the prevention of diabetes-induced elongation and partial cristolysis of the retinal endothelium mitochondria by ablation of MMP-9 genes strongly imply that the damage to the mitochondria could be one of the major mechanisms by which active MMP-9 contributes to the development of diabetic retinopathy. The loss of pericyte is also considered as one of the earliest morphological changes in the development of diabetic retinopathy (33), and the possibility that similar MMP-9–mediated mitochondrial damage, as seen in retinal endothelial cells, is also operating in the retinal pericytes cannot be ruled out.

To understand the molecular mechanism by which MMP-9 is increased in the retinal mitochondria in diabetes, we investigated the chaperone and transport system. Mitochondrial protein import and sorting are mediated by the translocase complexes in the membranes and chaperones in the aqueous compartments operating along the import pathways (13,14). The Hsps play an important role in protein–protein interactions and in prevention of unwanted protein aggregation. These molecular chaperones bind and stabilize proteins at intermediate stages of folding, assembly, and aid in the transport of proteins across subcellular membranes to the appropriate cellular compartments. The Hsp60 chaperone system is a prominent component of the mitochondrial protein quality control system, and it helps in the folding of mitochondrial proteins after import has been completed (34). The other chaperone, Hsp70, is mostly antiapoptotic and is shown to inhibit translocation of Bax into mitochondria and release of cytochrome c from mitochondria. Mitochondrial Hsp70 acts as the first chaperone that a newly imported polypeptide chain encounters in the mitochondria (34,35). Our results show that the ablation of MMP-9 gene in mice prevents a diabetes-induced decrease in mitochondrial Hsp70 and Hsp60, suggesting that MMP-9 has a critical role in regulating mitochondrial Hsp70 and Hsp60.

The Hsp70 helps maintain cellular homeostasis, and acts as a carrier of peptides or proteins. Although Hsp60 is mainly a mitochondrial protein, cytosolic Hsp60 prevents translocation of the proapoptotic protein Bax into mitochondria (36,37). Increased binding of retinal MMP-9 with Hsp60 and Hsp70 in diabetes, and regulation of these bindings by inhibition of MMP-9 activation (MMP-KO mice and MMP-9 siRNA–transfected cells) clearly suggests that in diabetic conditions, Hsp70 and Hsp60 are important in chaperoning MMP-9 to the mitochondria.

Mitochondria also have a well-defined translocase system to import nuclear-encoded proteins into mitochondrial membranes (13,14). This system functions as a receptor for recognition of the targeting and/or intramitochondrial sorting signals, and also a driving force for translocating the polypeptide chain. The TOM and TIM complexes translocate proteins across the mitochondrial outer and inner membranes, and both the inner membrane potential and matrix Hsp70 are essential for releasing the preprotein from the TOM complex. The precursor proteins are bound to the molecular chaperones in the cytosol and are delivered to the TOM import machinery of the mitochondrial outer membrane (13,38). TIM44 serves as an adaptor protein for the binding of mitochondrial Hsp70 to the membrane (35). Crossing of proteins into the mitochondria and transporting into the inner membrane are also regulated by the membrane potential (39). Protection of decreased levels of TOM34, TIM44, Hsp70, and Hsp60 in the retinal mitochondria and inhibition of increased mitochondrial permeability, experienced by retina and its capillary cells in diabetes, by abrogation of MMP-9 strongly suggests the role of MMP-9 in modulating mitochondrial transport system.

Recent work from our laboratory has shown that human donors with diabetic retinopathy have increased MMP-9 in their retinal microvessels, and the activation is achieved through the Ras/Raf/mitogen-activated protein kinase/extracellular signal-regulated kinase pathway signaling cascade (8). The results presented here clearly show that the retinal mitochondria of the same donors also have activated MMP-9 and decreased mitochondrial Hsp70 and Hsp60, but the binding of MMP-9 with Hsp70 and with Hsp60 is increased, further strengthening the role of mitochondrial MMP-9 in the development of diabetic retinopathy.

In summary, the activation of MMP-9 in the retinal mitochondria is one of the key events in diabetes that initiates mitochondrial dysfunction, damages mitochondrial structure, and activates apoptotic machinery. MMP-9 is transported to the mitochondria by Hsp70/Hsp60, and activated MMP-9 damages mitochondrial integrity and its DNA. Thus, inhibition of MMP-9 could have potential therapeutic value in preventing the continuation of the vicious cycle of mitochondrial damage that the retina experiences in diabetes (16), ultimately inhibiting the development and continued progression of diabetic retinopathy.

This study was supported in part by grants from the National Institutes of Health, Juvenile Diabetes Research Foundation, Thomas Foundation, and Research to Prevent Blindness.

No potential conflicts of interest relevant to this article were reported.

R.A.K. developed the experimental plan, performed the literature search, and wrote and edited the manuscript. G.M., J.M.d.S., and Q.Z. researched data.

The authors thank Dr. James Hatfield (John D. Dingle VA Medical Center, Detroit, MI) for his help with electron microscopy and Doug Putt and Yakov Shamailov (Wayne State University) for technical assistance.