Expression of Metalloproteinases from Human Retinal Pigment Epithelial Cells and Their Effects on the Hydraulic Conductivity of Bruch’s Membrane

Abstract

purpose. To evaluate the expression pattern of matrix metalloproteinases (MMPs) from retinal pigment epithelial (RPE) cells in culture, and to determine their ability to alter the transport properties of human Bruch’s membrane.

methods. Human RPE cells from either primary cultures or a cell line were maintained under culture conditions. At different time intervals after subculturing of cells the presence of MMPs in the bathing medium was determined by zymography. Cellular MMP activity was determined in a similar series of experiments where serum was omitted from the culture medium. Cultured cells were introduced onto Bruch’s membrane, mounted in a modified Ussing chamber, to assess entry of MMPs into the membrane. Fluid flow across Bruch’s membrane was determined by hydraulic conductivity for different ages of donor tissue, before and after 24 hours’ incubation with active MMPs from the RPE-conditioned medium or after incubation with purified activated MMPs. Latent (inactive) MMPs from medium containing serum were used in control experiments.

results. Cultured RPE cells expressed both MMP-2 and -9, with active MMP-2 becoming detectable from 4 days after subculture through to confluence and activated MMP-9 becoming abundant up to 24 hours after subculture. Both active MMPs significantly increased hydraulic conductivity of Bruch’s membrane, with the increase after MMP-9 incubation being far greater than that for MMP-2. Both enzymes showed a trend in hydraulic conductivity change with age such that, MMP-2 produced a relatively constant change, whereas MMP-9 showed a greater increase in older eyes.

conclusions. Activation of both MMP-2 and -9 by cultured RPE cells appeared to show a temporal relationship with the growth cycle of the cells. The activated enzymes increased fluid flow of Bruch’s membrane, and the marked effect observed with MMP-9 in older eyes suggests a mechanism that may allow debris removal.

In the aging eye, Bruch’s membrane progressively increases in thickness from 1 to 2 μm at birth to 5 to 6 μm at 70 to 80 years. ¹ This increase in thickness is accompanied by a progressive increase in lipid-rich membranous debris after the first decade and in many individuals, an increase in basal laminar deposits after the sixth decade. ^{1 2 3 4 5} The latter is claimed by some authorities to be a precursor of age-related macular degeneration (AMD). ^{6 7}  

Deposition of lipid-rich material within Bruch’s membrane has been suggested to cause an increase in resistance to the movement of water and waterborne agents across the system. ⁸ This concept has been supported by empiric work determining hydraulic conductivity of isolated Bruch’s membrane–choroid preparations as a function of age, in which it has been shown that there is an exponential increase in resistance to water movement. ^{9 10} However, these studies have clearly demonstrated that the largest increase in resistance occurs before the significant increase in membranous debris or lipids. This observation suggests that Bruch’s membrane, similar to other extracellular matrices (ECMs), undergoes change throughout life and, therefore, a system for modulating this membrane must exist. If such a mechanism can be identified and controlled, it may present an opportunity to influence the remodeling potential of the membrane, particularly in older or AMD-affected eyes. 

Remodeling of ECMs primarily involves two families of enzymes, the serine proteases and the matrix metalloproteinases (MMPs). Preliminary studies on serine proteases have demonstrated them to be extremely aggressive in degrading the entire structure of Bruch’s membrane at low concentrations and short incubation times, rendering measurements of effects on hydraulic conductivity difficult. ¹¹  

Dynamic synthetic and degradative processes control the balance of the ECM structure, and these can be influenced, either directly or indirectly, through the family of MMPs and their inhibitors, the tissue inhibitors of MMPs (TIMPs). Although MMPs cannot be manufactured in the ECM, previous studies have demonstrated their presence within Bruch’s membrane, in particular MMP-1, -2, -3, and -9. ¹² All these agents have the potential, when activated, for reducing the ECM. To date, these ECM-modifying agents have been identified within both RPE ^{13 14 15} and choroidal endothelial cells. ^{14 16}  

Given the presence of MMPs within Bruch’s membrane and the expression of the gelatinases MMP-2 and -9 in RPE cell cultures, ^{12 15} we decided to undertake an investigation of MMPs within retinal pigment epithelial (RPE) cells and to study the temporal expression and activation of these MMPs. Given that latent MMPs are activated by cleavage of a propeptide resulting in a loss of approximately 20 amino acid residues, ^{17 18} the two forms of the enzymes can easily be distinguished on zymography. Further, the levels and activation state of the enzymes should be examined in relation to their effects on the hydraulic conductivity of Bruch’s membrane. 

To facilitate ease of measurement, studies were conducted in vitro. Rather than using primary cultures of human RPE cells, where relatively small numbers of cells could be isolated and all in different phases of cell growth, cell cultures were used to increase the overall defined cell populations, such that initially most cells were migrating followed by a phase of cell division. 

Materials and Methods

Two sources of RPE cells were used in this study: one from freshly isolated primary RPE cells and the other from an immortalized RPE cell line. 

Primary RPE Cell Culture

Primary cell cultures were kindly donated by Jin-Jun Zhang, St. Thomas’ Hospital, King’s College, London. Cells were taken from a 63-year-old donor and maintained in 10% fetal calf serum (FCS) made in a 1:1 solution of Dulbecco’s minimum essential medium (DMEM) and Ham’s F10 medium, containing 100 U/mL penicillin and 100 μg/mL streptomycin (all from Sigma Chemical Co., Poole, UK). Medium samples were collected from cultures 2 and 24 hours and 4, 7, and 8 days after initial plating and therefore were designated the first passage. 

Transformed Cell Line Culture

For subsequent studies, a transformed human RPE cell line was used for studies of MMP expression, in preference to freshly isolated RPE cells to maintain consistency between samples for the various studies. The D407 RPE cell line was selected, because these cells had previously been well characterized with respect to epithelial specific protein expression and were thought to manifest a pattern of MMP expression similar to that of human primary RPE cell cultures. ¹⁹ This cell line has origins in a 12-year-old male donor and spontaneously transforms in the absence of any viral vector. 

To investigate the effects of RPE-derived MMPs on Bruch’s membrane, samples of Bruch’s membrane were dissected from human donor tissue obtained from the Bristol Eye Bank (Bristol, UK), once the corneas had been removed for transplantation surgery. Eyes were transported on ice in saline-moistened sterile containers and dissected immediately on arrival. Twenty-nine eyes were obtained for this study, of which 6 were used in experiments to determine MMP expression in Bruch’s membrane (aged 19–79 years) and 23 were used for flow measurements (aged 13–90 years). 

RPE Cell Culture

Frozen suspensions containing 1 × 10⁸ D407 RPE cells were gently equilibrated to ambient temperature under warm water before subculturing. Cells were cultured in high-glucose DMEM containing 10% FCS (Gibco/BRL; Grand Island, NY), 2 mM l-glutamine, 100 U/mL penicillin, and 100μ g/mL streptomycin. All medium constituents were obtained from Sigma Chemical Co. unless otherwise stated. Briefly, cells were trypsinized with 0.2% trypsin solution for 3 minutes at 37°C, washed with medium containing FCS to inactivate the trypsin, and centrifuged at 2000 rpm for 10 minutes to pellet the cells. After repeating washing and centrifugation, cells were resuspended in 1 mL medium with FCS and a 10-μL aliquot removed to determine cell numbers. Cells from the suspension were then subcultured at a ratio of approximately 1:100 to give a count of at least 1 × 10⁴ cells in each flask. Cells were incubated with medium at 37°C in a humidified atmosphere with 5% CO₂ until confluent. Trypsinization, washing, and centrifugation steps were repeated each time cultures reached confluence, until the fifth passage when aliquots of resuspended cells were used for experimentation. 

Cell counts were determined by diluting the cell suspension with an equal volume of 0.05% trypan blue. A 5-μL sample of this suspension was counted on a hemocytometer, and cell numbers were calculated allowing for each dilution. Total cell numbers at the end of each experiment were expressed as the increase in ratio of surviving cells to the number initially plated. 

MMP Determination

Cell and medium collections from RPE cultures were made at 2 and 24 hours and 4, 7, and 8 days after trypsinization. To standardize medium levels of MMP production, with the exception of the 2-hour sample, all samples were collected 24 hours after renewing the incubation medium. Samples were stored at −30°C before zymographic analysis. Concurrent cell samples were collected by washing and scraping the cells from the flask, followed by centrifugation. The supernatant was discarded and an aliquot of the cell pellet was removed for counting by trypan blue exclusion. The remainder of the cell sample was directly stored at −30°C before MMP extraction for zymography. 

In a further series of experiments designed to eradicate the influence of FCS on MMP expression, cells were harvested and incubated as previously described. However, the medium was replaced with serum-free medium for 24 hours before collection at each time point, after which the cells and medium were collected as before and stored at −30°C before analysis. 

MMP Extraction

MMPs were extracted from RPE cell cultures by homogenization in 20 μL extraction buffer, ¹² followed by centrifugation at 9000 rpm for 30 minutes at 4°C. 

Zymographic Analysis

Sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS-PAGE) ²⁰ was used to detect MMP-2 and -9. A final concentration of 0.1% gelatin was added to 1-mm thick minislab gels (10% acrylamide), to act as substrate for the MMPs. Briefly, 20-μL samples of either RPE-conditioned medium or supernatants from cell extractions were incubated with an equal volume of SDS sample buffer and maintained at room temperature for 1 hour before electrophoresis. The total volume of reaction (40 μL) was then loaded into the sample lanes of the gel. Electrophoresis was performed at 4°C, after which time the gels were washed in 1% Triton X-100 for 1 hour at room temperature to remove SDS. Gelatinase reactions were facilitated by incubation in reaction buffer ²¹ at 37°C for 18 hours. After a rinse in distilled water, the gels were stained with Coomassie brilliant blue (Merck, Ltd., Dorset, UK) for 1 hour. After subsequent destaining with 10% acetic acid/10% methanol (vol/vol), metalloproteinases could be identified by clear bands against a dark blue background. An internal standard of 0.1 ng human MMP-2 and -9 (Chemicon International, Inc., Temecula, CA) was incorporated into each gel, and specific gelatinases identified by their respective molecular weights. A standard was run on each gel and to avoid confusion of MMPs in the sample and MMPs introduced with the FCS, a sample containing 10% FCS was also run on each gel. Gels were photographed using a digital camera (model DC1000; Pretec, Taipei, Taiwan) and imported into a computer (PhotoSuite SE; MGI, Taipei, Taiwan). After gray-scale inversion of the gel images, the images were then imported into an image analysis program (Quantiscan for Windows; Biosoft, Cambridge, UK) for analysis and quantification. The band intensity values for each protein were corrected for background staining for each gel. Pixel analysis determined a graph of intensity and allowed the area under the curve to be calculated. By incorporating the standard in the gel together with the FCS sample and by normalizing staining intensity, MMP bands obtained from different experiments and different gels could be compared. FCS controls were not incorporated in calculating those samples with serum-free medium. 

Human Bruch’s Membrane Preparation

All dissection procedures were performed under sterile conditions within a laminar flow cabinet (Grade II; MDH Ltd., Hampshire, UK). Whole globes were dissected in a Petri dish lined with filter paper (Grade 50; Whatman, Maidstone, UK), moistened with phosphate-buffered saline (PBS) containing 100 U/mL penicillin and 100μ g/mL streptomycin. The anterior portion of the eye was carefully removed by a circumferential incision at the pars plana, and the cornea together with the lens, iris, and vitreous were discarded. The posterior globe was inspected for any gross disease of the retina, and those exhibiting any abnormal appearance were discarded. The neural retina was then gently peeled away from the underlying RPE and cut at the optic nerve head. 

A series of four samples were obtained from the midperiphery of each eye cup with an 8-mm trephine (Steifel Laboratories, Buckinghamshire, UK). The RPE was then carefully brushed away using a fine sable-hair brush. The Bruch’s membrane–choroid complex was gently teased away from the sclera and flattened over a fresh piece of moistened filter paper before being blotted to remove excess moisture. It was then placed in a beaker of cold isopentane, precooled in liquid nitrogen. After 2 to 5 seconds, the sample was removed from the beaker and allowed to reach room temperature for 10 minutes before it was again placed into the beaker for freezing. The filter paper with Bruch’s membrane was then immediately stored at −30°C before experimentation. Pilot studies using zymography demonstrated that this freeze-thawing technique eliminates the production of MMPs by choroidal cells, while preserving the Bruch’s membrane structure. 

Chamber Assembly

A modified Ussing chamber ¹⁰ was used with a 6-mm central aperture, across which the 8-mm Bruch’s–choroid discs could be mounted. The design of the chamber was such that it could be readily detached from the fluid reservoir and manometer. 

Bruch’s membrane was removed from storage and immersed in cold PBS to gently defrost the tissue and allow it to equilibrate to room temperature. The 8-mm tissue disc and filter paper samples were placed over the central aperture of one half of the chamber, while immersed in PBS containing antibiotics. The filter paper prevented the wrinkling of the tissue during mounting and so maintained a flat tissue preparation with the Bruch’s surface exposed. The other half of the chamber was then gently lowered over the tissue using the locating pins and then the two half chambers were secured with two screws. 

Each compartment of the chamber was then washed three times with serum-free medium. The whole procedure was completed in 20 minutes and was conducted under sterile conditions within a laminar flow cabinet. 

MMP Entry into Bruch’s Membrane

Suspensions of 1 × 10³ RPE cells were introduced into the Bruch’s membrane compartment of the chamber and allowed to gravitate and attach to the membrane surface for 2 hours in an incubator. Any unattached cells were washed away with serum-free medium. The Ussing chamber was then rotated through 90° before 500μ L of fresh medium containing 10% FCS was introduced into the Bruch’s membrane–facing compartment. An equal volume of serum-free medium was added into the compartment on the choroidal side and the unit placed in an incubator. After 24 hours’ incubation, the paper-mounted tissue sample was removed from the chamber and a 4-mm trephine of the tissue at the chamber interface was removed. In some of the discs, RPE cells were brushed off before MMP extraction and analysis by zymography, as previously described. 

Hydraulic Conductivity Measurements

Freeze-thawed, paper-mounted Bruch’s-choroid tissue samples were secured in the Ussing chamber, as already described. Both compartments were flushed three times with PBS before being slowly filled to avoid any air bubbles. The compartment appositional to the Bruch’s membrane surface was then coupled to a reservoir of PBS, and the choroidal compartment was linked to a manometer viewed by a traveling microscope. 

This arrangement allowed the hydraulic conductivity of the sample to be determined using the method of Starita et al. ¹⁰ Briefly, pressure was applied to the Bruch’s membrane surface by the PBS reservoir at a fixed height of 220 mm, which applied a constant pressure of 2156 Pa. Therefore, as the flow of buffer through the membrane occurred, the meniscus movement in the manometer could be observed over time, allowing the hydraulic conductivity of the tissue to be calculated. ^{9 10}  

Once the tissue was under pressure, if a hole was present in the membrane, it became readily apparent because of the speed of movement of the meniscus in the manometer, and such samples were discarded. Of the samples collected, fewer than five discs were found to have holes. Intact samples were allowed to equilibrate for 30 to 60 minutes under constant pressure. Measurements of the movement of meniscus were taken every 10 minutes for a period of up to 1 hour, after which time, the Ussing chamber was disconnected and both compartments were washed with PBS. This provided the baseline hydraulic conductivity of the preparation. 

Effect of Isolated RPE Cells

Four tissue cassettes, designed to clamp into a holder to allow hydraulic conductivity measurements, were prepared incorporating Bruch’s–choroid preparations from a 67-year-old donor. Freshly isolated RPE cells from a donor aged 77 years and cultured D407 cells were added to the Bruch’s-facing compartments of two cassettes. The remaining cassettes contained FCS-DMEM to serve as the control. Cassettes with RPE cells were then maintained in an incubator for 2 hours to allow the cells to attach. Thereafter, any unattached cells were removed by two to three gentle rinses with FCS-DMEM, and the cassettes were then placed in an incubator for 24 hours. After incubation, the cells were gently brushed away with a camel’s-hair brush and the hydraulic conductivity status of the four membrane preparations assessed. 

Effect of Conditioned Culture Medium

To determine the effect of active MMPs from RPE cultures on transport characteristics, the compartment of the Ussing chambers exposing Bruch’s membrane was filled with RPE-conditioned medium collected from cultures at either 2 hours or 7 days (active MMP-9 and -2 enriched, respectively) and incubated for 24 hours. The choroidal compartment was filled with an equal volume of freshly prepared medium containing 10% FCS. After incubation, both surfaces of the tissue preparation were thoroughly rinsed with PBS and the hydraulic conductivity measurements repeated. 

Effect of Purified MMPs

Purified samples of MMP-2 and -9 were a kind gift from John Clements (British Biotechnologies, Ltd., Oxford, UK). Active forms were obtained by treatment with p-aminophenylmercuric acetate (APMA), as described elsewhere. ¹² These MMPs were prepared in DMEM and a 1.0-mL aliquot added to the Bruch’s-facing compartment of the Ussing chamber. The choroidal compartment contained DMEM only. After 24 hours’ incubation, both half-chambers were rinsed several times with PBS and hydraulic conductivity determined. Altogether, three adjacent 8-mm trephines were obtained from the peripheral fundus, one served as a control with APMA alone, one was exposed to the inactive form of MMP, and the third was exposed to the APMA-activated form of MMP. 

Results

Cell Division

The graph obtained in Figure 1 shows the growth cycle of the cultured RPE cells. Subsequent to plating cells, there was a latency before cell division commenced, typically lasting 24 hours (Fig. 1) . At 50 hours, the number of cells began to increase, and confluence was achieved by 7 to 8 days. The rate of cell division was maximal at approximately 4 days after subculture, when the cells were observed to be half confluent. 

Gelatinase Expression from Cultured RPE Cells

MMP-2 Expression.

Zymographic analysis of the medium from cultured RPE cells showed expression of latent MMP-2 separating at 65-kDa (Fig. 2A) . The amount of latent MMP-2 gradually reduced until half confluence, when the levels appeared to remain constant up to 8 days. No active form of the MMP-2 enzyme could be detected up to 24 hours, but a 58-kDa enzyme (active form MMP-2) was present from half confluence onward (Fig. 2A , lanes 4–6). This active form of MMP-2 was present only in small quantities (Fig. 3) . No MMPs were detected by zymography in samples derived from RPE cells used in this study (data not shown). By contrast, when the RPE cells were incubated with serum-free medium for 24 hours before analysis, there was expression of only the latent form of MMP-2 within the RPE cells at all time points (Fig. 2B) . In these experiments, no MMP expression in the surrounding medium could be detected by zymography (data not shown). 

MMP-9 Expression.

All medium samples containing serum showed expression of the latent form of MMP-9, separating at 92 kDa (Fig. 2A) . The highest level of this enzyme was observed up to 24 hours after subculture and, although it progressively reduced, was still present up to confluence. By contrast, the active forms of MMP-9, separating at 88 kDa and 84 kDa, were present only up to 24 hours (Fig. 2A , lanes 2, 3). The 88-kDa protein is a transitional intermediate of MMP-9, which retains partial activity, whereas the 84-kDa enzyme is the fully activated form of latent MMP-9. Further, when serum was omitted from the incubation medium, no latent MMP-9 activity was found within the cells at any time point up to confluence (Fig. 2B) . However, the active forms of MMP-9 were detected in the 2-hour and 24-hour samples (Fig. 2B , lanes 1, 2). 

Primary cultures of human RPE cells also produced a cell cycle–dependent expression of active forms of MMP-2 and -9 and a representative zymogram for a 67-year-old donor is presented in Figure 2C . These findings were directly comparable to those from the cell line. However, some extra high-molecular-weight bands were detected in these samples, which were attributed to polymeric forms or posttranslational modifications of the enzymes. 

By subtracting the gray scales obtained from Figures 2A and 2B for the MMPs introduced by the FCS from those obtained from the RPE-conditioned medium, it was possible to acquire a series of values for MMPs within the medium derived from cellular activity alone (Fig. 3) . Figure 3A shows that active MMP-2 was the only MMP expressed above FCS control values in the medium from day 4 onward. In the absence of serum where no MMPs were found in the medium, latent MMP-2 was detected within cells at all time points but progressively diminished (Fig. 3B) . 

By contrast, relatively high levels of both the latent and active forms of MMP-9 were present in the RPE-conditioned medium up to 24 hours, but not thereafter (Fig. 3A) . Serum-free studies also illustrated the presence of active MMP-9 within the cells up to 24 hours only (Fig. 3B) . No postconfluence values were obtained in these experiments, because cells could not survive without FCS under these conditions. Further, the absolute values on the ordinate scale were almost an order of magnitude lower. Using the values obtained for staining intensity of the bands, the gelatinolytic activity of the active form of MMP-2 was calculated as 1.7% of the total activity of MMP-9 in the respective conditioned medium. 

MMP Entry into Bruch’s Membrane

Pilot studies demonstrated that endogenous levels of MMPs within Bruch’s membrane reduced to a point where they could not be detected by zymography after 24 hours’ incubation in serum-free medium. However, detectable levels were obtained from Bruch’s membrane samples 24 hours after plating RPE cells (Fig. 4) . 

In Figure 4 , lane 1 represents MMP activity derived from a sample containing both Bruch’s membrane and RPE cells 24 hours after plating, whereas lane 2 is a similar sample at a similar time, but one in which the RPE cells had been removed before zymography. Both lanes show the presence of latent MMP-2 and active MMP-9. The presence of the enzymes in lane 2 confirms the ability of MMPs secreted by the RPE to enter Bruch’s membrane–choroidal tissue. These findings confirm the observations illustrated in Figure 3 , with the exception that no latent MMP-9 was detected. 

Hydraulic Conductivity Measurements

Incubation with Isolated RPE Cells.

Plating of RPE cells from both the D407 line and freshly isolated from a 77-year-old donor onto Bruch’s–choroid from a 67-year-old donor resulted in an increase in the transport characteristics of the membrane (Fig. 5) . The control samples were collected from adjacent fundus regions of the same eyes. 

Incubation with RPE-Conditioned Medium.

Given that MMP-2 was the only form of MMP present within the medium above control FCS values at 7 days (Fig. 3A) , incubating Bruch’s membrane with medium derived from this period would enable the effects of activated MMP-2 on the substrate be to evaluated in isolation. After 24 hours exposure there was an increase in hydraulic conductivity of three- to fourfold (Fig. 6A) . This level of increase was constant in all ages—that is, the relative increase in fluid flow in young eyes after MMP-2 incubation was equal to that for older eyes. 

At 2 hours, the medium contained both the latent and active forms of MMP-9 and the latent form of MMP-2 (Fig. 3A) . Thus, by incubating Bruch’s membrane in this medium and comparing the results obtained by incubating samples in unconditioned medium containing FCS, which contained latent MMP-2 and -9, the effects on hydraulic conductivity by active MMP-9 alone could be determined (Fig. 6B) . Active MMP-9 induced dramatic changes in hydraulic conductivity, especially in older eyes, where 3 orders of magnitude of change was recorded. 

These changes were so large that it was necessary to confirm the integrity of the membrane after incubation, because the enzyme could have produced holes in the membrane system. To eliminate such artifacts, the integrity of membranes was determined before and after treatment with MMPs by dextran diffusional studies. Holes were not found to be present subsequent to MMP-9 incubation in any experiments. 

Incubation with Purified MMPs.

Exposure of donor Bruch’s membranes to APMA alone or inactive forms of purified MMPs was without effect on hydraulic conductivity (Table 1) . It was expected that APMA would activate endogenous MMPs, but the absence of change in transport suggests that these enzymes were lost during the incubation process. Under the conditions given by Guo et al., ¹² APMA treatment led to conversion of some of the latent MMPs to their active counterparts. After APMA treatment, the active forms of MMP-2 and -9 represented 47% and 21%, respectively, of total activity of the preparation (data not shown). 

Exposure of donor Bruch’s membrane to these purified and activated MMPs produced large increases in the hydraulic conductivity of the membranes (Table 1) . For active MMP-2 incubations, the percentage increase over control levels in donors aged 64, 71, and 87 years was 59%, 505%, and 417%, respectively. In the case of active MMP-9, the percentage increase was 150%, 767%, and 1312%, respectively. 

Discussion

This study demonstrated a complex relationship between the behavior of RPE cells, the activation state of MMPs, and their effects on the passage of fluid through Bruch’s membrane. First, previous observations on the manufacture and release of MMPs by cultured RPE cells ^{13 15} were confirmed and extended to relate enzyme activity to specific phases of cell migration, attachment, and division. Second, by culturing RPE cells on Bruch’s membrane artificially depleted of MMPs, the presence of newly released MMP-2 and -9 within the membrane was demonstrated. Third, active MMPs, obtained from either RPE-conditioned medium or from a purified source can significantly improve the hydraulic conductivity of Bruch’s membrane without damaging the molecular weight exclusion barrier of the membrane. Fourth, freshly isolated or cultured RPE cells plated onto the surface of Bruch’s membrane were also shown to improve its transport characteristics. Finally, empiric studies demonstrated that newly synthesized RPE-derived MMP-9 resulted in a highly significant increase in hydraulic conductivity of Bruch’s membrane. 

This study focused on the role of MMPs in modulating Bruch’s membrane’s permeability, but that of serine proteases cannot be ignored, particularly because urokinase-type plasminogen activator is also expressed by RPE cells, ²² together with its inhibitor, plasminogen activator inhibitor. ²³ Because plasmin is an activator for MMPs, ¹⁸ it is possible that the serine protease may indirectly alter Bruch’s membrane structure by modulating the MMPs. 

Although previous studies had demonstrated MMP synthesis in confluent cultures, ^{13 15} they could not relate the cycle of synthesis and release to specific phases of cell behavior. In the present study, at 2 hours after trypsinization, when cells were migrating and beginning to adhere to the substrate (Fig. 1) , there was a dramatic increase in MMP-9 production, release, and activation (Fig. 2A) . By contrast, as the RPE cells underwent cell division at 4 days after plating, the synthesis and activation of MMP-9 stopped. This period also saw the presence of the active form of MMP-2 in the surrounding media (Fig. 2A) . 

In a surprising finding, at the same time intervals, no detectable levels of MMP-2 or -9 were determined within the cells, which reflects either a lack of sensitivity of zymography or a very brief period between cellular productivity of the enzymes and their release. The differential temporal expression of the enzymes also suggests specific requirements for ECM activity related to cell behavior, before and subsequent to cell division, with MMP-9 having the more significant role during migration and attachment and MMP-2 assuming more importance after mitosis. 

External factors must also play a role in regulation of release of MMPs, because when serum was omitted from the bathing medium, synthesis of MMP-2 and -9 and activation of MMP-9 occurred within the RPE cells (Fig. 2B) , but no enzyme was released into the medium. Further, these experiments demonstrated that production of latent MMP-2 and activation of MMP-9 are not serum-dependent mechanisms. 

In vivo, the RPE cell is bounded by two extensive ECM systems, the interphotoreceptor matrix apically and Bruch’s membrane basally. To date, there is no evidence of selective polarization of MMP release or activity, and such information will not be forthcoming from culture studies unless models are developed with two ECM compartments coupled with demonstrably polarized RPE cells. Zymographic studies of Bruch’s membrane demonstrate the presence of latent forms of both MMP-2 and -9, with MMP-9 having a molecular weight of 92 kDa. ¹² The size of this protein and its presence within Bruch’s membrane suggests that either previously published data giving the exclusion limit of the membrane as between 65 and 75 kDa ²⁴ are in error or that the enzymic action of the molecule facilitates its passage into the matrix. The measurements of hydraulic conductivity (Fig. 5) support the latter active modulation concept. 

Newly synthesized active forms of both MMP-2 and -9 significantly increased hydraulic conductivity in Bruch’s membrane samples of all ages, whereas the serum-introduced latent forms had no effect (Fig. 5) . This latter finding was perhaps not surprising, given that the latent forms are known to be endogenously present. ¹² Purified and activated forms of MMP-2 and -9 also increased the hydraulic conductivity of Bruch’s membrane, but the magnitude of change was much lower than in Figure 5 because the activity of MMP-2 and -9 in this preparation was 44.5% and 0.3%, respectively, of that in the conditioned medium (Table 1) . 

The differential response achieved by the two enzymes as a function of age may give some indication of the senescent behavior of substrate systems. The relative uniform change induced throughout life by the action of MMP-2 suggests that its target substrate does not undergo a significant aging process. By contrast, the dramatic changes induced by MMP-9 after the age of 40 suggest a significant senescent process in its target substrate. There are three possible simplistic explanations for these observations. First, the membrane may become more susceptible to MMP-9 action due to intrinsic age-related structural changes in its ECM composition. Second, the accumulated debris associated with aging of the membrane over 40 may include specific targets for MMP-9 degradation. For example, long-spaced collagen has been suggested as a target substrate of MMP-9 ^{17 18} and is known to accumulate as a function of age. ²⁵ Also, basal laminar deposits in patients with AMD have been shown to contain carbohydrate structures, ²⁶ which may act as a substrate for the MMPs. Third, the action of MMP-9 on Bruch’s membrane may allow the debris to diffuse out of the system and therefore induce a more marked effect in older eyes. 

If these empiric studies on cell cultures have correlates in vivo, then the observations may be helpful in explaining some of the clinically observed phenomena associated with prophylactic laser grid treatment in patients with AMD. ^{27 28 29 30} Previous studies have demonstrated that subsequent to laser irradiation, cells within the RPE display a transient latency before migrating into the irradiated area and recovering Bruch’s membrane. ^{31 32} The migratory phase in vivo commences between 2 and 4 days after irradiation. If the increase in release of activated MMP-9 determined in the present studies occurs in vivo, then such laser-induced mobility within the monolayer would also lead to elevation in extracellular MMPs. If a significant proportion of the MMP-9 were released basally, then enzymic changes within Bruch’s membrane would not be unexpected. In this respect, plating of RPE cells onto donor Bruch’s membrane in the present investigation was associated with improvement in the hydraulic conductivity of underlying Bruch’s membrane. The 24-hour incubation used in the present study favors action by active MMP-9. The amount of improvement in transport characteristics may be dependent on the number of plating cycles, and further investigations are currently being undertaken. 

If the mechanism of change involves a system that allows both debris to diffuse out of the membrane and an increase in hydraulic conductivity, then perhaps the clinical observations of loss of drusen and subepithelial deposits would be expected. The current lack of clear evidence of a therapeutic effect derived from grid treatments in AMD may indicate that the clearance of debris is unimportant in the disease process or that the numbers of lesions used in the current grids are insufficient to induce a significant enzymic effect. As in the present laboratory studies, it may be necessary to repeat the cycle of grid radiation to achieve a measurable benefit in these patients. Finally, this work may have implications for development of pharmaceutical regimens to maintain homeostasis in the outer retina. 

 Supported by PPP Healthcare Medical Trust (AAH).

 Submitted for publication May 30, 2001; revised August 20, 2001; accepted September 14, 2001.

 Commercial relationships policy: N.

 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.

 

Cell population kinetics of the RPE cell line. Total number of cells collected at each time interval was calculated by trypan blue exclusion and corrected for the number of cells initially plated out. The data have been fitted to a sigmoid growth function curve, and individual time points are presented as mean ± SEM (n > 9).

Figure 2.

 

Zymograms illustrating cell cycle–dependent expression of gelatinases by human RPE cells (D407 cell line; A, B) and primary cultures (C). (A) RPE-conditioned medium from D407 cell cultures. At 24 hours before collection, medium was replaced with fresh medium containing 10% FCS (n > 8). (B) Cellular extractions when FCS was omitted from the bathing medium. At 24 hours before collection medium was replaced with serum-free medium (n > 5). (C) RPE-conditioned medium from primary culture of cells from a 63-year-old donor. In all zymograms, lane 1: 10% FCS control; lanes 2–6: medium collected at 2 and 24 hours and at half confluence, confluence, and 24 hours after confluence, respectively.

Figure 3.

 

Quantitation of the relative expression of various forms of MMP-2 and -9 by cultured RPE cells at several phases of the cell cycle. The gel data of Figure 2 were analyzed. (A) Medium samples collected from RPE cell cultures. All data have been corrected for MMPs introduced by incubation in 10% FCS. (B) Cell extracts collected from samples cultured with serum-free medium. (▪) 92-kDa latent MMP-9; ( ) 88-kDa active MMP-9; ( ) 84-kDa active MMP-9; (□) 65-kDa latent MMP-2; ( ) 58-kDa active MMP-2. Results are presented as mean ± SEM (n > 5).

Figure 4.

 

Zymogram illustrating entry of MMP-2 and -9 into Bruch’s membrane. Lane 1: RPE and Bruch’s membrane; lane 2: isolated Bruch’s membrane. The presence of MMP bands in lane 2 indicates MMPs’ ability to penetrate Bruch’s membrane. The inactive gelatinases may have arisen from the FCS, but the active form of MMP-9 was expressed by RPE cells and entered Bruch’s membrane. Note the absence of active forms of MMP-2. These experiments were repeated on samples obtained from six different donors.

Figure 5.

 

The effect of seeding RPE cells on the transport characteristics of donor human Bruch’s membrane. Control samples were incubated with medium containing 10% FCS. Bruch’s membrane from a 67-year-old donor was exposed to RPE cells from primary cultures from a 77-year-old donor, or the D407 cell line for 24 hours before hydraulic conductivity determination.

Figure 6.

 

Change in hydraulic conductivity with increasing age after incubation with active MMP-2 (A, □) or active MMP-9 (B,▵ ). The control hydraulic conductivities are represented in (A) and (B) by (▪) and the aging profile is described by the exponential relationship HC = 100 · exp(−0.396· age), n = 41 donors. (A) The transport properties of Bruch’s membrane were considerably improved after active MMP-2 incubation, shifting the aging curve upward: HC = 463.09 · exp(−0.0259 · age), n = 8. (B) Incubation with active MMP-9–enriched medium caused a dramatic increase in hydraulic conductivity of Bruch’s membrane, and the effect was particularly marked in the elderly donor population, leading to an exponentially increasing function for the aging relationship, HC = 1452.6 · exp(0.0225 · age), n= 12. Note the change in the scale in (B; right axis) applied to the curve of MMP-9 incubated samples.

Table 1.

 

Effect of Purified and Activated MMPs on the Hydraulic Conductivity of Human Bruch’s Membrane

Age (y)	MMP-2					MMP-9
		Control	Latent	Active	Control	Latent	Active
64	1.88 ± 0.094	1.61 ± 0.081	2.78 ± 0.14	1.19 ± 0.06	1.38 ± 0.69	3.22 ± 0.161
71	1.06 ± 0.053	1.95 ± 0.098	9.13 ± 0.46	1.16 ± 0.058	1.53 ± 0.077	11.7 ± 0.59
86	1.56 ± 0.078	1.69 ± 0.085	8.42 ± 0.42	1.67 ± 0.084	1.43 ± 0.072	21.9 ± 1.09

 Three adjacent samples of Bruch’s-choroid were obtained from each eye of a pair and mounted in Ussing chambers. One of the three served as a control and was incubated with DMEM containing APMA, the second was incubated with the latent form of MMP, and the third was incubated with APMA-activated MMP. After a 24-hour incubation, the membrane surfaces were rinsed and hydraulic conductivity determined. Data are expressed as the mean of at least five readings ± SEM (×10⁻¹⁰/m/sec · Pa).