Arteries remodel in response to environmental changes. We investigated whether mechanical strain modulates production of matrix metalloproteinase (MMP)-2 and -9 by cultured vascular smooth muscle cells (SMC). MMP-2 and MMP-9 expression were tested using human saphenous vein SMC cultured on silicone membranes at rest or subjected to physiological levels (5%) of stationary or cyclical (1 Hz) uniaxial strain. Compared with control, stationary strain significantly increased MMP-2 mRNA levels at all time points, whereas cyclic strain decreased it after 48 h. Both secreted and cell-associated pro-MMP-2 levels were increased by stationary strain at all times (P < 0.01), whereas cyclic strain decreased secreted levels after 48 h (P < 0.02). MMP-9 mRNA levels and pro-MMP-9 protein were increased after 48 h of stationary stretch (P< 0.01) compared with both no strain and cyclic strain. Our study indicates that vascular SMC show a selective response to different types of strain. We suggest that local increases in stationary mechanical strain resulting from stenting, hypertension, or atherosclerosis may lead to enhanced matrix degradation by SMC.
- matrix metalloproteinase
- mechanical stretch
- vascular remodeling
blood vessels undergo physiological and pathological remodeling. This process necessitates the breakdown and reorganization of the extracellular matrix scaffold by vascular cells. Mounting evidence, obtained mainly from the study of remodeling during the development and complication of naturally occurring atherosclerotic lesions (5, 12, 17,30), and that of vascular lesions triggered by interventions (2, 31, 35), implicates the contribution of specialized enzymes called matrix metalloproteinases (MMPs) (9). MMPs are secreted, both constitutively and inducibly, as inactive zymogens by human vascular (10, 15) and inflammatory (33) cells. In vitro and in vivo studies have shown that certain factors associated with vascular pathology, such as cytokines and mechanical injury, induce the expression and trigger the activation of MMPs (2, 10). MMPs have been found to be capable of degradation of collagen in the vulnerable shoulders of atherosclerotic plaques (30) and may be the cause of vascular tissue weakening and plaque destabilization. Because plaque failure was discovered to represent the major trigger for clinical events such as myocardial infarction and stroke (8), the potential action of MMPs has recently received a lot of attention. Similarly, increased expression of MMPs was demonstrated in several experimental models of arterial mechanical injury and graft failure (2, 31, 35), whereas MMP inhibitors decrease such lesions (1). Thus MMPs are believed to facilitate a variety of vascular pathologies.
One potentially relevant association is the observation that MMPs are overexpressed and activated in the vulnerable shoulders of atherosclerotic plaques (12), known to be exposed to the highest mechanical stress (6, 27). Colocalization of increased MMP expression and matrix degrading activity in the vulnerable shoulders (13) with areas of high tensile stress (21) may simply be an unfortunate coincidence, in which the weakest spot is subject to the highest tensile stress, but could also be indicative of a response to a specific mechanical environment. The cells and the extracellular matrix of the arterial vessel wall are likely responsive not only to biochemical but also to mechanical cues from the environment.
Hemodynamic forces have been demonstrated to play an important role in the development and evolution of arterial lesions in atherosclerosis and hypertension (29). Previous studies (6,27) have shown that the distribution of forces that act on the vessel wall is affected by the development of an atherosclerotic lesion in the intimal layer of artery and that plaques tend to rupture in those areas exposed to the highest mechanical stresses. The same vulnerable areas present MMP overexpression and MMP activity (12), and colocalization of increased circumferential stress and overexpression of MMP-1 was demonstrated by Lee et al. (21). However, the cyclic deformation of human smooth muscle cells (SMC) in vitro was found to actually inhibit MMP-1 expression (34). Other work (20) has shown that axial strain in vivo upregulates MMP-2 in rabbit carotid arteries. We undertook the present study based on the same general assumption that mechanical stress could contribute to the increased MMP expression in atherosclerotic lesions, but we decided to investigate the possibility that this modulation depends on the nature of the mechanical stress. We thus compared the in vitro effects of cyclic versus stationary uniaxial strain on the production of MMP-2, the main MMP constitutively produced by human SMC, and of MMP-9, an MMP shown to be inducible by cytokines in human SMC in vitro (10).
Cell culture and stretching experiments.
Experiments were performed with the use of cultured human SMC obtained from explants of excess saphenous veins obtained after bypass surgery (10). Primary SMC grown from vein explants were passaged and used between passages 3 and 6. Culture purity was verified with the use of anti-SMC α-actin with immunostaining (Sigma A2547, 1:200). Each passage was monitored to ensure the lack of any potential morphological or phenotypical changes or change in the growth rate. SMC were grown to confluency in Dulbecco's modified Eagle medium (DMEM) supplemented with 10% fetal calf serum (Cellgro/Fisher). The silicon membranes were prepared beforehand by being washed with 10 N sulfuric acid for 1.5 h and then by being rinsed extensively with autoclaved water. Each membrane was coated with a sterile solution that was mixed with (1:1) 1 mg/ml collagen type I (Collaborative Biomedical Products) and 100 μg/ml chitosan (USB) in 0.2 N acetic acid through five spraying/drying cycles, UV sterilized (30 min), rehydrated with warm (37°C) DMEM containing an antibiotic-antimycotic mixture (100 U/ml penicillin; 100 mg/ml streptomycin), and supplemented with 10% fetal calf serum. SMC were seeded on the coated membranes and allowed to reach confluency (typically overnight). Membranes carrying confluent cells were washed twice with Hanks' balanced salt solution and then transferred into serum-free medium DMEM/F12 (1:1) supplemented with 1 mmol/l insulin and 5 mg/ml transferrin 24 h before the stretch experiments.
The strain device consists of a silicone membrane suspended in cell culture medium, between an immobile steel rod and a mobile rod attached to a computer-controlled stepping motor. The entire apparatus is maintained in a cell culture incubator (37°C, humidified atmosphere). The amplitude and frequency of strain were controlled through the motion of the mobile steel rod (19). The volume of culture medium required to fill the system is 210 ml. Cell confluence was assessed by phase contrast microscopy before stretching.
In each experiment, the three different conditions (control and stationary or cyclic stretch) were run in parallel. The nonstrained control was obtained by mounting a membrane cultured with SMC and by holding it at resting length (0% strain). The other two membranes were simultaneously mounted in stretching devices and subjected to two different types of uniaxial strain, using the same initial rate in all experiments. The membrane subjected to stationary strain was held at 105% of the resting length. The third membrane was cyclically stretched to 110% of the resting length at 1 Hz. The strain experiments were performed for 24, 48, or 72 h. Cell viability was assessed using the Live-Dead Cell Assay Kit (Molecular Probes).
Sample collection and analysis.
Culture medium was collected from the reservoir of the stretching device at the end of 24, 48, or 72 h. In some experiments, culture medium was concentrated by precipitation with trichloroacetic acid (TCA). The SMC were lysed using 10 mM phosphate buffer, pH 7.2, and 150 mM NaCl (PBS) containing 1% Triton X-100, 0.1% SDS, 0.5% sodium deoxycholate, and 0.2% sodium azide, as described previously (10). Protein concentration was measured using DC Protein Assay (Bio-Rad; Hercules, CA) as recommended for detergent-solubilized samples.
Total SMC RNA was extracted by using TRI reagent (Sigma) following the manufacturer's instructions with slight modifications. These included sample extraction on ice, longer centrifugation time (25 min), and precipitation at −20°C for 60 min. RNA was measured from the optical absorbency at 260 nm.
Quantitative competitive RT-PCR.
Quantitative competitive RT-PCR was performed essentially as described previously (32). This method uses the competitive amplification of mRNA sequences from the sample against a similar but shorter fragment of standard probe. Double-stranded DNA sequences that contained full-length inserts of MMP-2 or MMP-9 were generously provided by Dr. G. Wells (Neures).
To generate the MMP-2 standard from the MMP-2 cDNA, we used a sense primer corresponding to the T7 promoter and the reverse primer: 5′-ACATTGACCTTGGCACCGG∧TGTCACTCCTGAGATCTGCAA-3′ (where the “∧” indicates a 21 base pair deleted sequence of the cDNA). For the MMP-9 standard, we used the same sense T7 promoter primer and the reverse primer 5′-AAGATGCTGCTGTTCAGCG∧ACGTGAAGGCGCAGATGGT-3′. The double-stranded DNA PCR product was gel purified and used to generate a 340-nt synthetic RNA sequence for MMP-2 and a 302-nt synthetic RNA sequence for MMP-9 with the use of T7 polymerase. These exogenous RNA standards were identical to the endogenous cDNAs of MMP-2 and MMP-9 of human SMC except that 21 base pairs near to the 3′ regions had been deleted. Exogenous standards were treated with RNase-free DNase (10 U/μl) for 15 min at 37°C and extracted using phenol-chloroform. To determine the range of mRNA levels in the samples, we ran simultaneous reactions containing 0.1μg of sample RNA and up to 100 pg of MMP standard RNAs. Reverse transcription was performed using a commercially available kit (Gene Amp RNA PCR, Perkin-Elmer Applied Biosystems; Foster City, CA). PCR was then performed with AmpliTaq (Applied Biosystems), using the sense primer 5′-GCAAAGAACACAGCCTTCTC-3′ and the antisense primer 5′-ACATTGACCTTGGCACCGG-3′ for MMP-2. For MMP-9, we used the sense primer 5′-ACCTGGTTCAACTCACTCCG-3′ and the antisense primer 5′-AAGATGCTGCTGTTCAGCG-3′. These PCR reactions resulted in amplification of a 257-bp sequence from the sample RNA, a 236-bp sequence from the exogenous standard of MMP-2, a 205-bp sequence from the sample RNA, and a 184-bp sequence from the exogenous standard of MMP-9. The PCR reactions were run on a 2.0% agarose gel and visualized using ethidium bromide staining. The level of mRNA was estimated on the basis of densitometric analysis of the signal for two competing RNA products (Molecular Analyst, Bio-Rad). After an initial approximation of the concentration, we selected several concentrations closely distributed within that range to more precisely determine mRNA abundance. The mRNA abundance (per μg total RNA) was calculated as the amount of competitor required to produce a band of equal intensity to the mRNA product. For assay of MMP-2 mRNA in nonstretch or cyclic samples, the range used for template RNA was 0–0.4 pg per 0.1 μg of sample RNA. For stationary conditions, the range was 0–20 pg. For MMP-9 mRNA, the range was 0–0.5 pg for nonstretch, cyclic, or stationary.
Proteins were identified by immunoblotting of cell lysates and of cell culture media. Cell lysates were used directly, whereas culture media were concentrated by TCA precipitation. Equal amounts of protein (30 μg) from each sample were loaded on 10% SDS-PAGE mini gels. After electrophoresis, proteins were transferred onto nitrocellulose membrane with the use of a Trans-Blot system (Bio-Rad). Nonspecific binding was blocked with the use of 0.1% Tween and 5% dry milk in PBS. The blots were incubated with monoclonal anti-human MMP-2 (1 ng/μl Ab-3) (Oncogene), anti-human MMP-9 (2 ng/μl Ab-1), or polyclonal anti-human TIMP-2 (1:5,000 Ab-801) (Chemicon) or TIMP-1 (1:1,000, Ab-8122) in PBS containing 0.1% Tween and 5% milk. The blots were then washed and incubated in horseradish peroxidase-linked secondary antibodies (1:1,000, Amersham). Positive bands were revealed by using the ECL chemiluminiscence kit (Amersham) and quantified by laser densitometry and image analysis with Molecular Analyst software (Bio-Rad).
Proteins with gelatinolytic activity in cell lysates or culture media were identified by electrophoresis in the presence of SDS in 10% polyacrylamide gels containing 1 mg/ml gelatin, as described before (10). Equal amounts (30 μg) were loaded on each lane. Gels were stained with colloidal Coomassie blue (Sigma) and bands were quantified by laser densitometry and image analysis.
For each variable, data from at least four independent experiments were quantified and analyzed. Statistical analysis was performed with the use of ANOVA and Student's t-test. Comparisons were made as two-tailed Student's t-tests. P values <0.05 were considered to indicate statistical significance.
Under conventional static cell culture conditions, human SMC constitutively produced pro-MMP-2. The second gelatinase, pro-MMP-9, is also produced by human SMC after stimulation with cytokines (10). In the present study, we investigated the effect of stretch upon expression of these two SMC gelatinases. We compared SMC maintained in the classic, nonstrained, cell culture conditions with those subjected to static or cyclic uniaxial strain. We investigated mRNA and protein levels, including both cell-associated protein and protein secreted in the culture medium. The methods we used allowed detection of both pro-enzyme and activated forms of MMP-2 and MMP-9.
Expression of MMP-2.
Levels of MMP-2 mRNA, as assayed by RT-PCR, extracted from SMC subjected to nonstretch and cyclic stretch did not vary significantly with time and were comparable at all time points (Fig.1 A). In contrast, stationary stretch increased the MMP-2 mRNA level such that after 24 h this was ∼50-fold higher than the level of MMP-2 mRNA of both nonstretched and cyclically stretched SMC (Fig. 1 B).
We investigated MMP-2 protein expression and activity in both SMC lysates and in SMC-conditioned culture medium. Cell-associated MMP-2 was identified in cell extracts by Western blotting (Fig.2) and SDS-PAGE gelatin zymography (not illustrated). The protein detected in cell extracts included both intracellular protein and cell surface-associated protein. The amount of cell-associated pro-MMP-2 detected in cells subjected to stationary strain was significantly higher after 48 h (Fig. 2 B), whereas the level of cell-associated MMP-2 protein in cyclically strained SMC was not significantly different from nonstrained conditions. In addition, we found that the active form of MMP-2, the one capable of digesting matrix substrates in vivo, was significantly increased by stationary stretch compared with both nonstretch and cyclic stretch conditions after 48 h (Fig. 2 C).
Secreted MMP-2 was identified in the culture medium collected from the stretch chambers (Fig. 3 A). The culture medium of SMC maintained under stationary conditions contained significantly higher levels of pro-MMP-2 compared with both nonstrained and cyclically strained cells after 48 and 72 h (Fig.3 B). Analysis of results indicated that the level of MMP-2 secreted by SMC under cyclic strain was lower than in the other conditions. This decrease was significant beginning at 48 h compared with stationary stretch and after 72 h compared with both stationary strain and nonstrained conditions, suggesting a mechanism to restrict the amount of MMP-2 secreted by SMC in the normal vascular environment. We detected the activated form of MMP-2 by immunoblotting only in the samples harvested after 72 h of stationary strain (data not quantified). Secreted MMP-2 zymogen and the active form were also analyzed with the use of SDS-PAGE zymography (Fig.4 A). We found that the amount of pro-MMP-2 secreted by SMC increased with time in both nonstrained and stationary strain conditions. Again, the cells subject to stationary strain exhibited higher levels of pro-MMP-2 at 48- and 72-h time points. A highly significant increase compared with both nonstrained and cyclic strain was detected starting at 48 h (Fig.4 B). Furthermore, cyclic stretching of SMC significantly reduced the amount of the secreted pro-MMP-2 beginning at 48 h. The activated form of MMP-2 was undetectable in most samples with the exception of stationary conditions. Quantification of the active form showed that this increase was highly significant after 48 and 72 h (Fig. 4 C). Because MMP activity is also modulated by the level of its natural inhibitors, we also explored the potential effects of stretch on MMP tissue inhibitor (TIMP)-1 and -2. We found that neither type of strain had a statistically significant effect on levels of TIMP-1 or TIMP-2, although TIMP-2 showed a trend toward increased levels under cyclic or stationary stretch (data not shown).
Expression of MMP-9.
Levels of MMP-9 mRNA detected under the three different conditions were very low (0–4 pg/0.1 μg total RNA). We did not detect significant differences between different conditions at various time points (data not shown).
Levels of secreted pro-MMP-9 protein, measured by Western blotting (Fig. 5 A), were close to the lower detection limit in most samples with the exception of samples obtained after stationary stretch at 48 and 72 h. These values were significantly higher that those detected in both other conditions (Fig. 5 B). Levels of pro-MMP-9 secreted in the culture medium were also very low by SDS-PAGE zymography (data not shown).
Vascular tissue remodeling (13) underlies the development, evolution, and complication of vascular lesions in atherosclerosis, hypertension, aneurysmal dilatation, graft failure, and after vascular interventions such as balloon angioplasty or stenting. The naturally occurring vascular pathological conditions are largely limited to the arterial side of circulation, which is exposed to pulsatile blood flow and frequent changes of the hemodynamic environment. These are not seen in the venous circulation, but occur in veins grafted into the arterial circulation, further supporting a role for physical forces in the natural history of arterial lesions (13, 24, 29). Similarly, aortic aneurysms occur in segments exposed to high pressure. Inside the vessel wall, the residing SMC are exposed to pressure, which manifests at the cellular level as mechanical stress. Mechanical stretching may occur chronically in hypertension, and prolonged stationary stretch results from the placement of a stent to open an occluded artery. Previous studies (21) showed that increased tensile stress colocalizes with increased MMP expression in the vulnerable shoulders of human atheroma. We have also found that increasing the ex vivo perfusion pressure enhances the expression and activity of MMPs within the wall of porcine arteries (7). We also showed that longitudinally stretched human saphenous veins had increased MMP expression and activity (23). These findings suggest that a combination of high mechanical stresses and areas with weakened matrix scaffold could endanger the mechanical strength of vascular tissue. At the same time, these observations have also raised the possibility that high mechanical stress may be a cause for increased matrix degradation via the action of MMPs. We therefore decided to test the hypothesis that uniaxial stress modulates the expression of the major SMC-derived MMPs, MMP-2, and MMP-9.
Previously reported (4) in vitro effects of mechanical strain on SMC include increased cell proliferation and deposition of a matrix necessary for organization and development of arteries, and later for vascular remodeling during atherogenesis and hypertension. This reaction was suspected as contributing to increased intimal thickening triggered by overstretching of arteries during balloon angioplasty. By using a similar simplified in vitro model, we now find that stationary strain of SMC increases their capacity to break down matrix. This could enhance arterial remodeling where SMC are exposed to such forces in situ and may enable compensatory remodeling and repair.
By examining MMP-2 and MMP-9, we found that compared with conventional static cell culture conditions, cyclic strain decreases the expression of MMP-2 in human SMC, as previously reported for MMP-1 (34). In contrast, stationary strain significantly increased the level of mRNA as well as latent and active MMP-2. Similarly, although much attenuated, static strain had enhancing effects upon expression of MMP-9. Thus we found that the in vitro response of SMC is dependent on the strain regimen. Furthermore, we found that stationary strain increased not only the expression of latent MMP-2, but also the level of cell-associated and secreted active MMP-2. This finding has important functional implications because degradation of matrix substrates by MMPs requires posttranslational processing of the latent forms secreted by cells. Conditions that trigger the activation of MMP zymogens are still poorly understood. A cell-associated activator of MMP-2 was described on the surface of transformed cells (28). This type of molecule is also expressed by activated SMC (26) and it may also be upregulated by mechanical strain.
Another potential mechanism for in situ MMP-2 activation may involve the generation of reactive oxygen species. We have shown that reactive oxygen radicals can activate in vitro latent forms of gelatinases secreted by cultured human SMC (25). Production of reactive oxygen species known to be present in atheroma is commonly attributed to inflammatory cells and may activate MMPs in the vulnerable shoulders of plaques (11); however, reactive oxygen species can also be produced by vascular cells, especially on stimulation (14). The fact that mechanical strain may act as a stimulus for vascular cells is supported by evidence that stretching of cultured endothelial cells leads to release of hydrogen peroxide (19). In addition, we showed that oxidative stress might contribute to the remodeling of human saphenous vein grafts experiencing arterial hemodynamic conditions through activation of latent MMPs (22). Thus an increased level of available latent MMP-2 in areas where SMC are subjected to stationary stretch will likely generate increased levels of enzymatically active MMP-2 in situ, through several potential pathways of activation that may be at work in the vessel wall.
Observed differences between the magnitude of effects at the level of mRNA and that of MMP-2 protein may be related to the translational and posttranslational distribution of mRNA. MMP-2 mRNA is translated into pro-MMP-2 protein, which is then distributed between cell-associated and secreted pro-MMP-2. This is in turn posttranslationally processed into the active forms of the enzyme and ultimately degraded by autolysis (3).
The results of our current in vitro study also support the notion that cyclic strain downregulates production of MMPs. This suggests that under normal hemodynamic conditions the matrix-degrading activity in healthy arteries is low. However, the increased stiffening due to the development of advanced calcified lesions in the arterial wall (18) and the increased oxidative stress impair the normal cyclic expansion and recoil of diseased arteries (16), while the wall is subjected to increased circumferential tensile stress (27). Similarly, diseased arteries are subjected to acute increases in stationary stretch during angioplasty, and chronic increases after stenting. We propose that increased stationary strain acts as a stimulus to increase production of matrix-degrading enzymes by the SMC, possibly as a natural reaction meant to allow the expansion of the arterial wall subjected to stretch. However, this reaction, triggered by vascular interventions or possibly mismatched compliance of an engineered vascular substitute to the adjacent tissue, will eventually facilitate development of lesions. In the shoulder areas of atherosclerotic plaques, where the stress is maximal (6), such process would contribute to the weakening of the plaque's fibrous cap and ultimately to its failure. Thus our in vitro results provide a likely mechanistic support to observations made in human and experimental vascular lesions, which suggest a relation between mechanical stretch and increased remodeling. A better understanding of the connection between mechanical stimulation and remodeling of vascular tissue in general, and of the matrix scaffold in special, should also aid in the design of cell seeded vascular grafts.
The authors thank Dr. Karen Schentzer for valuable advice regarding the stretching device and Xiaoping Meng for assistance with the cell culture.
This study was supported in part by Georgia Tech-Emory Biotechnology Research Center, the Whitaker Foundation, and National Heart, Lung, and Blood Institute Grant R01 HL-64689-01A1.
Address for reprint requests and other correspondence: Z. S. Galis, Emory Univ. School of Medicine, Div. of Cardiology, 1639 Pierce Dr., WMB 319, Atlanta, GA 30322 (E-mail:).
The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
First published January 23, 2003;10.1152/ajpheart.00494.2002
- Copyright © 2003 the American Physiological Society