Effects of Mechanical Stretching on Trabecular Matrix Metalloproteinases

Abstract

purpose. The homeostatic mechanisms responsible for intraocular pressure (IOP) regulation are not understood. Studies were conducted to evaluate the hypothesis that trabecular meshwork (TM) cells sense increases in IOP as stretching or distortion of their extracellular matrix (ECM) and respond by increasing ECM turnover enzymes.

methods. Flow rates were increased in perfused human anterior segment organ cultures and the matrix metalloproteinase (MMP) levels and IOP were evaluated. Human TMs in stationary anterior segment organ culture were mechanically stretched, and MMP levels were analyzed. TM cells were grown on membranes, which were then stretched, and MMP levels were evaluated. Western immunoblots, zymography, and confocal immunohistochemistry were used to evaluate changes in MMPs and their tissue inhibitors, the TIMPs.

results. Doubling the flow rate in perfused human organ cultures increased gelatinase A levels in the perfusate by 30% to 50% without affecting gelatinase B or stromelysin levels. Immediately after doubling the flow rate, the measured IOP doubled. However, over the next few days the IOP gradually returned to the initial level, although the flow rate was maintained at double the initial value. Stretching stationary organ cultures or stretching TM cells grown on membranes resulted in similar increases in gelatinase A without changes in gelatinase B or stromelysin levels. The gelatinase A increases occurred between 24 and 72 hours and were approximately proportional to the degree of stretching. Although coating the membranes with different ECM molecule affected the gelatinase A response, the optimum response occurred when the cells had been grown long enough to produce their own ECM. By Western immunoblot and confocal immunohistochemistry, the stretch-induced increases in gelatinase A were accompanied by strong decreases in TIMP-2 levels and moderate increases in one membrane type MMP, MT1-MMP. After mechanical stretching of the membrane, gelatinase A, MT1-MMP and TIMP-2 all exhibited a similar punctate immunostaining pattern over the TM cell surface.

conclusions. These results are compatible with the hypothesis that elevations in IOP are sensed by TM cells as ECM stretch/distortion. TM cells respond by increasing gelatinase A and MT1-MMP, while decreasing TIMP-2 levels. This will increase ECM turnover rates, reduce the trabecular resistance to aqueous humor outflow, and restore normal IOP levels. This hypothesis provides a regulatory feedback mechanism for IOP homeostasis.

Elevated IOP is a primary risk factor for glaucomatous optic nerve damage, yet normal intraocular pressure (IOP) regulation is not understood. ^{1 2 3 4 5 6} The ciliary body produces aqueous humor at an average rate of 2.75 μl/min with little regard for the consequent IOP. ^{7 8 9 10} Although glaucoma is a relatively common disease, 90–95% of the population maintain good IOP regulation throughout their lives. ^{6 11} This implies that the normal resistance to aqueous outflow, much of which is thought to reside within the deepest one fourth of the trabecular meshwork (TM), ^{12 13 14} is usually regulated efficiently and reliably by some as yet unidentified mechanism. 

We hypothesized earlier that TM cells could modulate aqueous outflow facility by changing ECM turnover and subsequent extracellular matrix (ECM) replacement rates. ^{1 2 15 16 17 18} We recently reported two observations that support this hypothesis: (1) manipulation of TM matrix metalloproteinase (MMP) activity reversibly modulates outflow facility in perfused human anterior segment organ culture ¹⁹ ; and (2) laser trabeculoplasty, a common treatment for glaucomatous IOP elevations, induces sustained MMP expression specifically within the trabecular juxtacanalicular region. ^{20 21} Thus, a plausible molecular mechanism for trabecular modulation of aqueous outflow facility is present. Interference with endogenous MMP activity causes dramatic reduction in outflow facility; thus, maintenance of the appropriate outflow resistance requires ongoing ECM turnover. ¹⁹  

However, some mechanism of TM cell sensing of IOP or outflow facility would be required for this mechanism to effectively maintain IOP homeostasis. ^{1 2 16 17} A significant portion of the resistance to outflow is thought to reside within the juxtacanalicular ECM. ^{1 2 16 22 23 24 25} Because of the structure of the TM, which might be thought of as a “semi-porous diaphragm” stretched across the outflow pathway and Schlemm’s canal, increases in IOP will tend to selectively stretch or distort the TM. ^{22 26 27} This should be particularly acute in the juxtacanalicular ECM, because a large portion of the resistance to outflow appears to reside there. Thus one possible mechanism that would allow TM cells to sense IOP would be for these cells to be able to recognize mechanical stretch or distortion. ^{1 2 16 17} By now, several reports have appeared showing that TM cells can sense such mechanical stretching forces. ^{28 29 30 31 32 33}  

A variety of cell types have been shown to use integrins, cellular receptors for ECM macromolecules, to detect stretching or distortions of these ECMs and transduce regulatory responses. ^{34 35 36 37} TM cells express a variety of integrins, ³⁸ and may also detect stretching by similar mechanisms. Thus, to test the hypothesis that TM cells can sense ECM stretching or distortion and will respond by adjusting ECM turnover, we used three separate methods of creating mechanical stretching stresses on TM cells and evaluated the effects this had on MMP levels. 

Materials and Methods

Antibodies to individual MMPs and TIMPs were obtained from Triple-Point Biologicals (Portland, OR); secondary antibodies, gelatin, and β-casein were obtained from Sigma (St. Louis, MO); collagen-coated, 24-mm-diameter Transwell-COL membrane culture inserts with 3-μm pores were from Corning Costar (Cambridge, MA); noncoated Falcon 23- mm-diameter PET (polyethylene terephthalate) membrane cell culture inserts with 3-μm pore size were from Becton-Dickinson Labware (Franklin Lakes, NJ); fibronectin-, laminin-, Matrigel-, type IV collagen- and type I collagen-coated 0.45-μm pore size Biocoat inserts were from Collaborative Biomedical (Bedford, MA); Pico Green, Anti-Fade reagent and Live-Dead stain were from Molecular Probes (Eugene, OR); Dulbecco’s modified Eagle’s medium, antibiotics, and antibiotic/antimycotic solutions were from Gibco BRL (Gaithersburg, MD); fetal calf serum was from HyClone (Logan, UT); chondroitinase A was from Seikagaku America (St. Petersburg, FL), and Super Signal chemiluminescent detection kits were from Pierce (Rockford, IL). 

Models to Exert Mechanical Stretch on TM Cells: Perfused Organ Culture

Stationary human anterior segment organ culture was conducted serum-free as previously described. ³⁹ Perfused human anterior segment organ culture was conducted as previously described, ¹⁹ except constant flow rate perfusion was used and IOP was measured. ⁴⁰ Anterior segments were maintained in stationary organ culture for 1 week before being placed in perfusion culture and were perfused for 3 to 5 days until the IOPs had stabilized, before conducting experiments. Flow rates were maintained at 2.5 μl/min, and measured IOPs averaged about 7 to 8 mm Hg. To produce stretch/distortion stresses on TM cells in perfused organ cultures, flow rates were doubled (to 5 μl/min) and maintained for several days, while measuring IOP at various time intervals as indicated. In one set of experiments designed to verify the effects of elevated IOP on gelatinase A levels, perfusion rates were doubled for 24 hours and then returned to 2.5 μl/min. The perfusate was collected in a groove on the rim of the clamping ring at various times as indicated and stored at −20°C for analysis. Probable resultant forces on the TM cells of the juxtacanalicular region of the meshwork are shown in Figure 1A (small arrows). Presumably, the primary impact of doubling the flow rate will be largest at the site of the outflow resistance, putatively within the juxtacanalicular region of the TM (JCT in Fig. 1A ). This flow pressure, the normal IOP of the eye (large arrows, Fig. 1A ), should tend to bow the juxtacanalicular region into Schlemm’s canal. The resultant forces (smaller arrows, not shown to scale) that a cell within or on either side of the juxtacanalicular ECM (gray region in Fig. 1A ) of the TM should experience include the following: (1) a circumferential stretching force running parallel to Schlemm’s canal; (2) a radial stretching force perpendicular to Schlemm’s canal with vectors running toward the scleral spur and toward the cornea; and (3) a force opposite to the IOP due to the venous pressure (not present in the organ culture model). 

Stationary Organ Culture

To apply mechanical stretch/distortion stresses on TM cells in stationary anterior segment organ culture, explants were pinned through the sclera in each quadrant to a paraffin (dental wax) support (Fig. 1B) . Explants to serve as nonstretch controls or as stretch treatments were pinned identically through the sclera without distortion, and pins were exactly vertical (Fig. 1B) . The distance between opposite pins on the explants to be stretched was measured (length 1 and length 2, Fig. 1B ). One of each pair of these pins was then repined at a distance that increased each of these lengths by 10%, while maintaining the pins in an exactly vertical orientation. This provides a consistent, albeit somewhat crude, stretch/distortion stress on the cells of the juxtacanalicular TM, with resultant forces including the same circumferential force parallel to Schlemm’s canal and radial force perpendicular to it that are shown in Figure 1A and detailed above. The cornea, sclera, and the remainder of the TM undoubtedly experience some additional stretch/distortion forces also. 

TM Cell Culture and Mechanical Stretch/Distortion

Porcine and human TM cells were cultured as previously described. ¹⁸ By passages 3 to 5, cells were plated at a density of approximately 90% confluence onto cell culture insert membranes in 6-well culture plates. After 3 to 5 days, serum-free medium was added to the cells for 24 or 48 hours before and during stretching experiments. To apply mechanical stretch/distortion, a glass bead of precise diameter was placed beneath the insert in the center of the membrane, and weight was applied to the lid of the plate to force the lip of the insert down onto the upper lip of plate’s well (Fig. 1C) . This produced a defined upward bowing of the membrane, which increased the surface area by an estimated 10%. The cells should experience radial and circumferential stretch forces, the sum of which are proportional to the reciprocal of the cell’s distance from the center of the membrane. As indicated in one study, a smaller diameter bead was also used to produce a 5% increase. 

Western Immunoblots and Zymograms

Culture medium, perfusate or cellular extracts were analyzed by zymography, using either gelatin or β-casein as substrates for gelatinases A and B (MMP-2 and MMP-9) or stromelysin (MMP-3), respectively, or by Western immunoblot to determine the activity or protein levels of the various MMPs or TIMPs. ⁴¹ For immunoblots, second antibody with conjugated horseradish peroxidase and chemiluminescence detection was used, following the manufacturers’ directions. When cellular/ECM extracts were to be analyzed, proteins were extracted using a modified RIPA buffer with a variety of proteinase inhibitors as previously described. ⁴²  

Confocal Immunohistochemistry

For localization studies, inserts were rinsed with phosphate-buffered saline (PBS), and TM cells and proteins were fixed on the membranes in 4% paraformaldehyde in PBS for 10 minutes and then rinsed twice for 5 minutes each in PBS. Membranes were then washed in room temperature TBS (50 mM Tris, 150 mM NaCl, pH 7.4) and incubated with blocking buffer (TBS with 0.05% Triton X-100, 2% goat serum, and an additional 150 mM NaCl) for 30 minutes. Membranes were then incubated with primary antibody (5 μg/ml) in blocking buffer for 1 hour in a humidified chamber, rinsed in TBS three times more for 3 minutes each time, and incubated with fluorescein-conjugated secondary antibody (17.2 μg/ml) in blocking buffer for 45 minutes in the dark at 100% humidity. Membranes were then rinsed twice for 5 minutes each in TBS, removed from the inserts, and placed on microscope slides. After preequilibration with antifade buffer for at least 5 minutes, antifade reagent was added, and slides were coverslipped, sealed, and stored in the dark at 4°C until analyzed. 

For analysis of the effects of 1× or 2× flow rate on TM cells in perfused anterior segments, 6-μm cryosections were analyzed after 24 hours at the designated flow rate. Explants were rinsed in PBS, and two wedges from opposite sides of the explants were removed and embedded in OCT containing 2.5% glycerol, quick-frozen, and stored at −80°C. Sections (6 μm) were cut with a Leitz Digital 1720 Cryostat (Leica, Germany), using standard methods. Briefly, sections were thaw-mounted onto Superfrost Plus slides (Fischer Scientific, Pittsburgh, PA), immersed in cold acetone for 2 seconds, and stored at −80°C. Slides were warmed to room temperature, fixed in 4% paraformaldehyde in PBS for 10 minutes, and then rinsed briefly in PBS. After pre-equilibration for 5 minutes at 37°C in buffer, 0.1 U of proteinase-free chondroitinase A was added per milliliter, and incubated for 8 minutes at 37°C. Subsequent steps were similar to those detailed above for the membranes. 

Confocal microscopy was conducted as detailed earlier. ⁴³ In addition to nonstretched paired controls, stretched or nonstretched membranes without primary antibody were evaluated. Confocal instrument gain and zero settings were optimized below saturation on an intensely staining sample, and then no setting changes were made for the complete set with that antibody; images were processed together to avoid introducing artificial differences. 

Live–dead staining (Molecular Probes) to evaluate the condition of TM cells was conducted as previously detailed. ¹⁹ DNA analysis using a PicoGreen fluorescence assay (Molecular Probes) following the manufacturer’s instructions was used to correct for differences in cell density on membranes. When perfusion flow rates were doubled, twice the volume of perfusate was applied to gel lanes to correct for the dilution. Gels or autoradiographs were scanned, and relative band density was analyzed ⁴⁴ using a densitometry program (BioImage, Ann Arbor, MI). Student’s t-test or Mann–Whitney ranked sum analysis was used to determine significance when comparing treatment results. 

Results

Effects of Increased Perfusion Flow Rate

After IOPs had stabilized at 2.5 μl/min flow rates, the flow rate was increased to 5 μl/min, and IOP was measured at intervals for several days (solid circles; Fig. 2A ). The initial response to doubling the flow rate was an approximate doubling in measured IOP. However, over the next few days, the IOP gradually returned to the predoubling level, although the higher flow rate was maintained. Analysis of the MMP levels in the perfusate demonstrated a consistent increase of approximately 30% to 50% in gelatinase A (Figs. 2A 2B) , with no change in gelatinase B (Fig. 2B) or stromelysin (not shown). Comparing the latent 72-kDa pro-form and the activated 68-kDa form of gelatinase A (Fig. 2B) suggested that the increase was primarily in the activated form. However, in the several experiments of this type we conducted, cases where the total gelatinase A or primarily the latent pro-form were elevated were observed as well. 

Effects of Stretch/Distortion on Stationary Cultures

When anterior segment explants in stationary organ culture were stretch/distorted and media gelatinase A was evaluated after 24 hours, a similar increase in gelatinase A levels was observed (Fig. 3) . A typical gelatin zymogram is shown with the expected positions of gelatinase A and B labeled. The gelatinase A bands correspond to the pro- and activated forms. Relative band density from densitometric scans of zymograms from three similar experiments are also shown. The increase is significant at P < 0.02. Both the latent and activated forms are increased but the ratio of latent-to-activated forms is increased rather than decreased, as was observed with the perfusion study shown earlier. Minimal or no changes in stromelysin or gelatinase B were observed in response to this stretching (data not shown). 

Effects of Stretch/Distortion of TM Cell Cultures

When TM cells that had been grown on insert membranes and maintained serum-free for 24 hours were subjected to mechanical stretching of the membranes for 24 hours, similar gelatinase A increases were observed (Fig. 4A) without appreciable effects on stromelysin or gelatinase B (not shown). The gelatinase A increase, analyzed by densitometric scans of gelatin zymograms, is approximately 50% above nonstretched controls, with P < 0.01 for n = 9 samples. As with the stationary organ cultures, both the latent and the activated forms of gelatinase A are increased, and the ratio of latent to activated form is increased slightly by stretching. When this same experiment is conducted with ocular fibroblasts from the sclera, no gelatinase A increase is observed (data not shown). 

Time Courses, Dose Responses, and Support Matrix Effects on Stretch Responses

To evaluate the duration of the stretch/distortion that is necessary to initiate these gelatinase A increases, we applied stretch for various times and then removed the glass bead and left the inserts in culture for 24 hours. Stretching for 30 minutes to 2 hours did not initiate a statistically significant response; stretching for 6 hours initiated a modest but significant response, and stretching for the full 24 hours initiated a larger response (Fig. 4B) . 

To evaluate the effect of changing the degree of stretch, 24-hour stretch was conducted with beads of two different sizes. The smaller sized bead, which produces only a 5% surface area increase, results in a gelatinase A increases that is roughly half as large (i.e., in the range of 15%–20%). This is compared with the 30% to 50% increases seen with 10% surface area increases conducted in parallel (data not shown). 

We also compared the efficacy of membranes coated with laminin, Matrigel, fibronectin, type IV or type I collagen, or gelatin with the noncoated membranes. Laminin and Matrigel were the least effective, dramatically attenuating the gelatinase A increases. Fibronectin and type IV collagen were of intermediate efficacy, producing less response than type I collagen or gelatin coating. The time after plating the cells on these membranes, that is, the time the cells were allowed to lay down their own ECM components, appears to be of primary importance. When the cells were allowed several days to establish an ECM before initiating the experiment, the gelatinase A responses were maximized, whether on coated or on uncoated membranes. Thus, all the studies presented herein were conducted on membranes without coatings. 

Possible Involvement of TIMPs and MT-MMPs

Gelatinase A is unique among the MMPs in that its apparent physiologic activation mechanism requires TIMP-2 and the MT-MMPs. ^{45 46} Thus, we investigated the effects of TM cellular stretch/distortion on these proteins. TM cells grown on insert membranes were subjected to 10% stretch for 24, 48, or 72 hours, and media and cellular extracts were analyzed for changes in gelatinase A, TIMP-2, and MT1-MMP levels. Typical zymograms and Western immunoblots are shown (Fig. 5A) . Media gelatinase A is seen as a latent 72-kDa pro-form and activated 68-kDa form; media TIMP-2 is a single band at 20 kDa, and cellular extract MT1-MMP is seen predominantly as a very light latent pro-form at 72 kDa, a dominant activated form at 63 kDa and a smaller breakdown form at 45 kDa (Fig. 5A) . These band sizes are as predicted based on literature values. ^{45 46 47 48} Densitometric scans were analyzed for each (Figs. 5B 5C 5D) after 24, 48, or 72 hours of stretch. Two separate experiments with each time point in triplicate resulted in the indicated significance values, when comparing control versus stretch pairs. Media gelatinase A and cellular MT1-MMP were increased moderately, whereas media TIMP-2 was dramatically reduced by mechanical stretch. The media levels of MT1-MMP were very low with or without treatment (not shown). The cellular extract contained both TIMP-2 and gelatinase A, although at much lower levels than the media. Gelatinase A increased, and TIMP-2 decreased with stretch as observed in the media. 

Effects of Stretch on Cellular Localization of Gelatinase A, TIMP-2, and MT1-MMP

Confocal immunohistochemistry was used to localize these three proteins on TM cells on insert membranes that had or had not been stretched at 10% for 24 hours (Fig. 6) . Typical pairs of control or stretched membranes (as marked above the column) immunostained for gelatinase A, TIMP-2, or MT1-MMP (as marked beside the rows) are shown. Changes in the levels of these proteins, as judged by density of immunostaining, are similar to those observed by zymography and Western immunoblot, that is, gelatinase A and MT1-MMP increase, while TIMP-2 decreases dramatically. The control, (−AB Control) in which the primary antibody was omitted, is shown as indicated. 

A large fraction of the gelatinase A and MT1-MMP immunostaining is particulate, apparently at the cell surface. The increase appears to be both punctate at the surface and cytoplasmic. Some apparent nuclear immunostaining was observed for MT1-MMP. A similar punctate TIMP-2 immunostaining, also apparently at the cell surface, is seen in both the control and to a lesser extent in the stretched cells. Some TIMP-2 immunostaining also appears to be cytoplasmic and somewhat“ fibril-associated.” 

TIMP-2 immunostaining is also shown for cryosections through the outflow pathway of perfusion-cultured explants (Fig. 7) , which had been flowed at 1× (2.5 μl/min) for several days until the measured IOP had stabilized at about 7 to 8 mm Hg. They were then exposed for 24 hours at either 1× or 2× flow (2.5 or 5 μl/min, respectively) before termination of the experiment. The “−AB control” with no primary antibody added is also shown. Reduced TIMP-2 in response to doubling the flow rate is quite apparent. Gelatinase A and MT1-MMP immunostaining in these explants also change with stretch, reflecting the patterns detailed earlier for the membranes (data not shown). 

Discussion

These data support our working hypothesis developed for normal IOP homeostasis. The basic tenants of this hypothesis are that (1) TM cells can sense changes in IOP as mechanical stretch/distortion; (2) they then respond by modulating MMP expression and/or activity; (3) changing MMP activity will adjust the outflow resistance; and (4) this is the, or at least a, homeostatic mechanism used by TM cells to regulate and maintain constant average daily IOP. Although this is still only a working hypothesis, the increases in trabecular gelatinase A levels and/or activity in response to mechanical stretch/distortion in these three separate model systems provide support for tenants (1) and (2). Our previous reports that manipulation of trabecular MMP activity reversibly modulates outflow facility in perfused human anterior segment organ culture ¹⁹ and that laser trabeculoplasty induces sustained trabecular juxtacanalicular MMP expression ^{20 21} provide strong support for portion (3) of this hypothesis. 

Gelatinase A, TIMP-2, and MT1-MMP are normally expressed at moderate levels by TM cells in all three of these model systems. Evidence that has been presented for a complex pattern of gelatinase A regulation includes the following: transcriptional regulation, translational regulation, post-translational processing, secretion, ECM sequestration, recruitment by TIMP-2 to the membrane for activation by MT-MMPs, additional auto-activation steps, proteolytic processing to smaller active and then inactive forms, inhibition by TIMPs particularly TIMP-2 and possibly eventual degradation by TM cells. None of these processes have been unraveled in detail. Considerable evidence exists for a specialized relationship between TIMP-2 and pro-gelatinase A, involving heterodimerization via binding domains on each molecule’s carboxyl terminus. ⁴⁹ TIMP-2 has also been shown to recruit pro-gelatinase A to the cell surface, where TIMP-2 also binds to MT1-MMP and facilitates pro-gelatinase A’s proteolytic activation by MT1-MMP. ^{45 46} The precise details and regulatory nuances of this activation process remain to be resolved. The punctate, apparent cell surface localization pattern of these three macromolecules after stretch is suggestive of colocalization, although we have not yet conducted detailed colocalization studies. In addition, pro-gelatinase A activation at the cell surface exhibits a bell-shaped TIMP-2 concentration dependence. ⁴⁵ When TIMP-2 is at low levels, well below 1:1:1 for these three molecules, the rate of MMP-2 activation is slow; when large excesses of TIMP-2 are present, the rate of MMP-2 activation is also slow, because both gelatinase A and MT1-MMP are also inhibited by TIMP-2. Our data can speculatively be interpreted by this model. Before stretch, TIMP-2 may be in excess thus restricting pro-gelatinase A activation rates; stretch reduces TIMP-2 approaching the optimum 1:1:1 ratio and triggering increases in pro-gelatinase A activation rates. Independently from this event, the observed increases in pro-MMP-2 and MT1-MMP levels will likely also increase active MMP-2 levels. In a number of other tissues, somewhat similar MMP responses to mechanical stretch have been reported. ^{50 51}  

The inability of short-duration stretch to trigger gelatinase A increases in this system is intriguing. One interpretation of this data is that this trabecular homeostatic mechanism would need to discriminate against acute fluctuations in IOP. This would avoid overadjusting in response to transient changes in IOP because of physical activity. It seems reasonable that there would be two separate IOP regulatory mechanisms, one acute and one chronic, necessary to deal with the different types of pressure fluctuations that could be anticipated. 

Although it seems clear that TM cells in the three different model systems that we used will experience mechanical stretching forces, the magnitude of these forces is difficult to estimate. In the perfusion system, which is the closest to the normal physiologic situation, the radial and circumferential forces on the juxtacanalicular TM cells (Fig. 1) seem likely, although their actual magnitudes are unknown. Contributions of other forces on these cells and forces on the remainder of the meshwork and on the remainder of the tissues are difficult to access. When we subjected scleral fibroblasts to mechanical stretching on insert membranes in cell culture, they did not respond with similar MMP/TIMP changes (data not shown). However, this model contains several other cell types. Another possible consideration is fluid shear forces in the perfused explant model. However, the aqueous humor flow rate is quite low. Estimations of the potential shear stress that uveal, corneo-scleral, or juxtacanalicular TM cells should experience at normal flow rates give values one to two orders of magnitude below those normally observed in blood vessels (C. Ross Ethier, University of Toronto, personnel communication). Although not absolutely ruling out any shear stress contributions, it seems unlikely that this would be a major contributor to our studies. 

The stretched stationary culture model is the least well defined of the three that we used. The cornea and sclera will undoubtedly experience some stretch/distortion, and the magnitude of the trabecular stretch is very difficult to access. Although the resultant stretch forces shown in Figure 1A are likely to be important and acting on the juxtacanalicular TM cells, their relative contribution has not been ascertained. In spite of this limitation, the data reinforce that obtained with the other two models. 

Although the most clearly defined of the three models, the membrane insert cell culture model suffers theoretically from the nonlinearity of the stretching forces. The forces on the cells in this model should vary as a function of 1/radius. Our confocal studies, however, argue that a threshold stretching force is achieved throughout the membrane and that the peripheral decline is not a major factor. When the MT1-MMP changes with stretch are evaluated on different regions of the membrane (as in Fig. 6 ), the observed changes in MT1-MMP are uniform across the membrane (data not shown). If a minimal stretch threshold were not reached throughout the membrane, cells on the more peripheral portions of the membrane should show considerably less increase in MT1-MMP than those on the central region. 

Thus, our working hypothesis is supported, although certainly not unequivocally established, by these and earlier data. ^{19 20 21} (1) Elevated IOP appears to be sensed as stretch by TM cells; (2) TM cell stretch triggers gelatinase A and MT1-MMP increases and TIMP-2 decreases; (3) this will increase ECM turnover and reduce the outflow resistance ¹⁹ ; (4) putatively, this results in a return to physiologic IOP levels. This working hypothesis thus provides a plausible self-contained homeostatic mechanism that could explain chronic IOP regulation. 

 Supported in part by National Institutes of Health Grants EY03279, EY08247, and EY10572 and by grants from The Glaucoma Research Foundation, San Francisco, California; Research to Prevent Blindness, New York, New York; and Alcon Laboratories, Fort Worth, Texas.

 Submitted for publication May 17, 2000; revised January 3, 2001; accepted February 6, 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.

 

Diagrammatic representation of stretch models. (A) Perfused anterior segment model showing a radial cross-section through the TM and Schlemm’s canal (SC). The ECM of the juxtacanalicular region of the TM (JCT) is shown in gray, and the ECM within trabecular beams is shown by diagonal lines; TM cell nuclei are solid black ovals or circles. The primary force provided by fluid flow/IOP is indicated by the three large arrows, and the resultant forces on JCT cells are shown by smaller arrows; the anterior chamber is to the lower right. (B) Stretching model using stationary anterior segment organ culture. Two pairs of vertical pins with the two measured lengths as labeled before stretch is applied; stretch is applied by moving one of each pair of pins to increase the respective lengths by 10%. (C) TM cell culture stretch showing one insert from a 6-well culture plate. Components are as labeled.

Figure 2.

 

Effects of doubling perfusion flow rate on gelatinase A levels and IOP. (A) After measured IOP had stabilized while pumping at 2.5μ l/min, the flow rate was doubled (arrow) and the resultant IOP monitored for several more days (•). Perfusate was collected and analyzed at the times indicated. The activated form of gelatinase A is plotted (□) as relative band density from zymograms. (B) The active gelatinase A levels (•) and pro-gelatinase A (○) and pro-gelatinase B (□) levels are also plotted. These data are typical of the five experiments that were conducted.

Figure 3.

 

Effects of stretch/distorting TM cells in stationary anterior segment organ culture. Anterior segments were stretched for 24 hours, and media were analyzed for gelatinase activity by gelatin zymography (A). Positions of gelatinase B (Gel B at ∼92 kDa) and the pro- and activated forms of gelatinase A (Gel A at ∼72 and 68 kDa) are labeled. Both pro-gelatinase A and activated gelatinase A are seen. (B) The zymograms from three similar experiments were subjected to densitometric analysis, and the statistical significance was determined by Student’s t-test (P < 0.02 for n = 3 paired-eye explants; values are means; error bars, ±SD).

Figure 4.

 

Effects of stretching TM cells on membranes. When TM cells are cultured on transwell insert membranes and the membranes stretched for 24 hours, gelatinase A levels in the culture media are increased by approximately 50%. (A) Results of densitometric analysis of gelatinase A bands on gelatin zymograms. The significance from t-tests is P < 0.01 (n = 9), and the zymogram activity is given as relative band density from scans. (B) Effects of short or long duration mechanical stretching on gelatinase A production. Membranes with TM cells at confluence were stretched for different periods of time (as indicated), after which the beads were removed and the culture continued without stretch. Media were then collected at 24 hours, and gelatinase A levels were determined. P values do not reflect significant differences, except as shown for the 6- and 24-hour times (values are means; error bars, ±SD; n = 5 to 7 at each time).

Figure 5.

 

Effects of mechanical stretch on trabecular gelatinase A, TIMP-2, and MT1-MMP levels. Cells were stretched to produce a 10% surface area increase for 24, 48, or 72 hours. Control and stretch media were analyzed for gelatinase A by gelatin zymography or for TIMP-2 by Western immunoblot; cell extract was analyzed for MT1-MMP by Western immunoblot. (A) Typical gels as labeled are shown with three control (C) and three stretched (S) lanes each. This image of the gelatin zymogram was photographically reversed for enhanced viewing contrast. Arrows at 72 and 68 kDa show positions of pro- and activated gelatinase A; arrow at 20 kDa shows the position of TIMP-2; and arrows at 72, 63, and 45 show the pro-, the activated, and the activated and truncated forms, respectively, of MT1-MMP. The results of densitometric scans of both the latent pro-gelatinase and activated gelatinase A bands from zymograms (B); of scans of the TIMP-2 band on Western immunoblots (C); and of the MT1-MMP pro- and activated bands from Western immunoblots (D) are shown. Stretch times are as indicated. Values are means ± SDs for two experiments with each point in triplicate (n = 6 for each data point). ▪, control; , stretch; statistical significance is shown above each pair of bars using Student’s t-test or Mann–Whitney ranked sum analysis.

Figure 6.

 

Confocal immunohistochemical localization of gelatinase A, TIMP-2, and MT1-MMP on control and stretched TM cells. Membranes with nearly confluent TM cells on them were stretched for 24 hours and probed with antibodies against gelatinase A, TIMP-2, or MT1-MMP as indicated on the left. “Controls” were not stretched at all, and“ Stretch” were stretched to give 10% surface area increases. The“− AB Control” was exposed to secondary but not primary antibody. Data shown are typical of several membranes of each in four completely separate experiments. The third column shows higher magnification views of the three respective stretched fields in the adjacent column. White arrows, cell nuclei; white arrowheads, punctate cellar immunostaining. The fields shown in the first two columns are approximately 0.1 × 0.1 mm on a side, and the magnification in the third column is 2.5 times that of the second column.

Figure 7.

 

Confocal immunohistochemical localization of TIMP-2 in the TM of perfused explants at two different flow rates. Cryosections were taken through human TM of paired-eye explants flowed at 1× (2.5 μl/min) or 2× (5 μl/min) for 24 hours after IOPs had stabilized at 1× flow for several days. The “−AB Control” was exposed to secondary but not primary antibody. The anterior chamber is to the upper left, the cornea is below, and the sclera is above, with white arrowheads pointing at Schlemm’s canal.

The authors thank the Lion’s Eye Bank (Portland, OR) for donor eyes; the MMI Confocal Core Facility (Oregon Health Sciences University, Portland, OR) for confocal microscopy; Melissa Radecki, Nathalie Hernandez, and Sylvia Stephens for technical assistance; and C. Ross Ethier for sharing the results of his calculations and for valuable discussions of mechanical forces acting in the three-model systems.