Interstitial flow promotes vascular fibroblast, myofibroblast, and smooth muscle cell motility in 3-D collagen I via upregulation of MMP-1

Zhong-Dong Shi, Xin-Ying Ji, Henry Qazi, John M. Tarbell

This article has a correction. Please see:


Neointima formation often occurs in regions where the endothelium has been damaged and the transmural interstitial flow is elevated. Vascular smooth muscle cells (SMCs) and fibroblasts/myofibroblasts (FBs/MFBs) contribute to intimal thickening by migrating from the media and adventitia into the site of injury. In this study, for the first time, the direct effects of interstitial flow on SMC and FB/MFB migration were investigated in an in vitro three-dimensional system. Collagen I gels were used to mimic three-dimensional extracellular matrix (ECM) for rat aortic SMCs and FBs/MFBs. Exposure to interstitial flow induced by 1 cmH2O pressure differential (shear stress, ∼0.05 dyn/cm2; flow velocity, ∼0.5 μm/s; and Darcy permeability, ∼10−11 cm2) substantially enhanced cell motility. Matrix metalloproteinase (MMP) inhibitor (GM-6001) abolished flow-induced migration augmentation, which suggested that the enhanced motility was MMP dependent. The upregulation of MMP-1 played a critical role for the flow-enhanced motility, which was further confirmed by silencing MMP-1 gene expression. Longer exposures to higher flows suppressed the number of migrated cells, although MMP-1 gene expression remained high. This suppression was a result of both flow-induced tissue inhibitor of metalloproteinase-1 upregulation and increased apoptotic and necrotic cell death. Interstitial flow did not affect MMP-2 gene expression or activity in the collagen I gel for any cell type. Our findings shed light on the mechanism by which vascular SMCs and FBs/MFBs contribute to intimal thickening in regions of vascular injury where interstitial flow is elevated.

  • matrix metalloproteinase
  • neointima formation
  • three-dimensional migration

neointima formation is often induced in regions where the endothelium has been damaged by mechanical injury, increased wall stress (as in hypertension), or chemical insult (11, 19, 21, 25, 34). The migration of smooth muscle cells (SMCs) from the media to the intima and their subsequent proliferation and new extracellular matrix (ECM) secretion are well-recognized characteristics of neointima formation (19, 25). In addition to medial SMCs, adventitial fibroblasts (FBs) and their activated counterpart myofibroblasts (MFBs) are also capable of migrating from the adventitia to the media and even to the intima where they contribute to the thickening of the neointima in response to endothelial injury (33, 35, 38, 42).

It has been shown that matrix metalloproteinases (MMPs) are involved in intimal lesion formation. MMPs are a family of zinc-dependent enzymes responsible for the degradation of ECM. The ECM degradation initially involves active collagenases (especially MMP-1), which digest intact collagens and then gelatinases (mainly MMP-2 and MMP-9) for further digesting collagen fragments and gelatins (14). The excessive proteolytic ECM degradation by MMPs can be reduced by endogenous tissue inhibitors of metalloproteinases (TIMPs). The balance between MMPs and TIMPs plays a major role in vascular remodeling and diseases (31). Increased MMP expression and activity are associated with neointimal lesion development and SMC migration after arterial balloon injury (2, 54). Upregulated MMP matrix-degrading activity may also be necessary for FBs to move through the adventitial matrix into the media and intima (39). MMP expression and SMC migration were promoted at vein-to-graft anastomoses in hemodialysis grafts, leading to intimal thickening and stenosis (23). However, the mechanisms by which SMC and FB “sense” vascular injury and then contribute to neointima formation still remain unclear.

Interstitial flow, the movement of fluid through the ECM, is present in all tissues (10a). It is an important coupling factor between mechanical stress and signaling in the three-dimensional (3-D) matrix (10). Modeling studies have shown that transmural interstitial flow (driven by transmural pressure) passes through the oriented SMC layers to the adventitia and imposes shear stresses on SMCs that are of the order of 0.1 to 1 dyn/cm2, depending on the location of the cells in the media (46, 47, 51). Shear stress on FBs would be expected to be lower than on SMCs since the permeability of “loose” adventitia is higher than that of “dense” media. In the case of vascular injury or hypertension, interstitial flow and flow-induced shear stress on SMCs and FBs will be elevated (22, 46, 48). For example, when the endothelium is denuded, the hydraulic conductance is increased 2.5-fold (1) in rabbit aortas and 1.75-fold (41) in rat aortas, which leads to 2.5- and 1.75-fold increases in shear stress, respectively (see Eq. 2 for shear stress estimation). Previous studies have demonstrated that shear stress on cell monolayers [two dimensional (2-D)] affects vascular SMCs and FBs migration in vitro (7, 8). Changes in shear stress in venous grafts may also cause SMC migration and intimal thickening (23). However, there has been no assessment of whether interstitial flow directly affects vascular SMC and FB/MFB migration in three dimensions.

Vascular SMCs and FBs/MFBs reside in a 3-D ECM, which is mostly collagen I. The migration of SMCs and FBs/MFBs initially occurs in collagen I; cells then penetrate the external/internal elastic lamina and basement membrane as they move into the intima. To model the in vivo environment more faithfully, therefore, we used 3-D collagen I as the ECM to investigate the direct effects of interstitial flow on the motilities of vascular SMCs and FBs/MFBs. We show that interstitial flow can promote SMC and FB/MFB migration rates in 3-D collagen I gels, and this enhanced motility is controlled by the upregulation of MMP-1, not MMP-2.


The animal procedures were approved by the City College of New York's Institutional Animal Care and Use Committee and in accord with the American Physiological Society's “Guiding Principles for the Care and Use of Animals.”

3-D gel preparation.

Rat aortic SMCs and adventitial FBs and MFBs were obtained as previously described (7, 8). Cells (passages 3–5) were first suspended in rat-tail collagen I (BD Science) gels (cell density, 2.5 × 105 cells/ml; and final gel concentration, 4 mg/ml). Then, 200 μl of gel were loaded into each 12-well cell culture insert with 8-μm pore-size membrane (BD Science), the edge of which had been precoated with 50 μl of 4 mg/ml of cell-free collagen without serum to prevent the gel from later contracting and detaching during cell spreading (Fig. 1A). One milliliter of growth medium (DMEM containing 10% FBS and 1% penicillin-streptomycin) was added to each well, and the gels were continuously incubated for 24 h to allow cell spreading.

Fig. 1.

Schematic of the three-dimensional (3-D) interstitial flow and cell migration system. A: 200 μl of cell/gel mixture were loaded into a 12-well insert with a 50-μl cell-free collagen-precoating layer along the edge. Gel concentration was 4 mg/ml, and cell density was 2.5 × 105 cells/ml. B: the gel was exposed to interstitial flow driven by a pressure drop (ΔP). C: after exposure to flow, to check cell motility, the gel was precompacted with a 24-well insert on top and then incubated with 20 ng/ml of PDGF-BB to initiate migration.

Interstitial flow experiment and cell migration assay.

The gels were subjected to interstitial flow driven by 1, 3, or 10 cmH2O pressure drop for 1, 3, or 6 h (Fig. 1B). The flow medium was the same as the growth medium. Other gels not exposed to flow served as no-flow controls. Immediately after the flow exposure period, the undersides of the insert membranes were swabbed by wet and sterile cotton-tipped applicators to remove any cells that had already migrated to the undersides. In this way, all the inserts indicated zero cell migration at the start of the migration period.

Because the compaction of the gels due to different flow rates could affect cell migration rates, all the gels were precompacted to a defined state just before the migration phase of the experiment. To accomplish the precompaction, a 24-well format insert was placed on the top of each gel (Fig. 1C) and compressed to a fixed depth by placing the lid over the 12-well plate. Then by confocal microscopy, we did not see any change in cell distribution between flow-treated gels and controls, which suggests that the cells did not migrate in the gels during 6 h of the flow period (data not shown). Thus, to check the effect of flow on cell motility, we added 1 ml of 20 ng/ml PDGF-BB (Sigma) to each plate well to initiate migration to the bottom of the insert membrane. This way we were able to count migrated cells easily as a modified Boyden chamber. After 48 h of chemotactic incubation, the cells that had migrated to the undersides of the insert membranes were stained with Diff-Quik (Dade Behring), and five fields (100×) (1 center, 4 edges) were randomly picked for counting. We observed that there was no difference in cell migration whether the gel was precompacted right before flow or after flow; however, the level of compaction significantly affected the cell migration. Preliminary experiments with 4 mg/ml collagen gels showed that the compaction reduced the cell migration by ∼30% (data not shown).

MMP activity inhibition experiments.

A potent, broad-spectrum MMP inhibitor, GM-6001 (Calbiochem), and its structural analog, GM-6001-NC, as a negative control were used in some experiments to block MMP activity. One milliliter of 10 μM GM-6001 or GM-6001-NC was mixed with 1 ml of 20 ng/ml PDGF-BB and then added to some plate wells at the start of the migration period.

MMP activity assay.

The MMP-1 and MMP-2 activities in the conditioned media of control and 1-h flow- and 6-h flow-treated gels were determined using the SensoLyte Plus 520 MMP-1 Assay Kit and SensoLyte 520 MMP-2 Assay Kit from AnaSpec (San Jose, CA). Briefly, the conditioned media was collected from the companion plate well after the 48-h migration period. To measure MMP activity, the conditioned media was concentrated to the same fold by an YM-10 Microcon centrifugal filter device (Millipore). In the MMP-1 assay, 100 μl of conditioned media were added to each microplate well, which was precoated with MMP-1 antibody. After MMP-1 was pulled down, the MMP-1 substrate 5-FAM/QXL520 FRET (fluorescence resonance energy transfer) peptide was added to the well. To measure the total MMP-1 level, 1 mM p-aminophenylmercuric acetate (APMA) was added to the microplate wells to activate pro-MMP-1 for 1 h at 37°C. The fluorescence intensity representing the MMP-1 activity was measured at 490/520-nm wavelength. In the MMP-2 assay, 50 μl either with or without 1 mM APMA-treated medium were added to each microplate well, followed by an addition of MMP-2 substrate 5-FAM/QXL520 FRET peptide. The fluorescence intensity representing the MMP-2 activity was measured at 490-/520-nm wavelength.

Collagen/gelatin zymography and reverse collagen zymography.

Collagen/gelatin zymography was used to reinforce MMP activity assay data. The collagen and gelatin proteolytic activities of MMPs in the conditioned media of control, 1-h flow-, and 6-h flow-treated gels were assessed by collagen or gelatin substrate zymography as previously described (9). Electrophoresis was carried out using a Mini-PROTEAN 3 Cell system (Bio-Rad). Conditioned media (20 μl) was mixed in nonreducing 5× sample buffer containing SDS, glycerol, and bromophenol blue (without 2-mercaptoethanol) and subjected to electrophoresis on 10% polyacrylamide SDS gels (1.5 mm thick) containing 0.5 mg/ml of calf skin collagen type I (C9791, Sigma) or 1 mg/ml of porcine skin gelatin (G2500, Sigma). After electrophoresis, the polyacrylamide gels were washed three times (20 min each) in 2.5% Triton X-100 for 1 h to remove all traces of SDS. The gels were rinsed in distilled water for 5 min and finally incubated at 37°C in developing buffer containing 100 mM Tris·HCl, 5 mM CaCl2, 0.005% Brij-35, and 0.001% NaN3 (pH 8.0) for 20 h. The gels were stained with 0.25% Coomassie brilliant blue G-250 in 50% methanol and 10% acetic acid solution and destained with 40% methanol and 10% acetic acid solution. The MMP level was identified as clear zones of lysis against a blue background. Finally, the gels were incubated in 5% methanol and 7.5% acetic acid. To check the TIMP-1 levels, 40 μl of conditioned media from each well were subjected to reverse zymography, which was performed as previously described (43). Reverse collagen zymography was very similar to collagen zymography except that the polyacrylamide SDS gels were not only incorporated with calf skin collagen I but also contained day 4 rat aortic FB-conditioned medium. The level of TIMP-1 was identified by the blue bands. To quantify the MMP levels, the images were acquired using ChemiDoc XRS (Bio-Rad) and processed using Quantity One software (Bio-Rad).

Releasing cells from collagen gels.

To collect the cells from the collagen gels, the gels were detached from cell culture inserts and transferred into centrifuge tubes. Each gel was then incubated with 100 μl of 0.2% of collagenase I (Worthington Biochemical) in complete growth media at 37°C for about 1 h with gentle agitation. Once the collagen gels were completely digested, the media with collagenase was removed by centrifugation. The cell pellets were then washed once with ice-cold PBS and collected.

Reverse transcription-polymerase chain reaction.

RT-PCR assays were performed to assess the relative mRNA levels. Briefly, the cells were lysed and reverse transcription was performed following the procedures for the Cells-to-cDNA II kit (Ambion). PCR reactions were performed using designed primers and Multiplex PCR Master Mix (New England Biolabs). A 383-bp rat MMP-1 product was amplified using the sense primer 5′-GAC CTC ATG TTC ATC TTT AG-3′ and the antisense primer 5′-CAC CAC AAT AAG GAA TTC GT-3′ (Genebank accession no. M60616). A 200-bp rat MMP-2 product was amplified using the sense primer 5′-GAT GGA TAC CCA TTT GAC GG-3′ and the antisense primer 5′-CTG CTG TAT TCC CGA CCA TT-3′ (Genebank accession no. NM_031054). A 203-bp TIMP-1 product was amplified using the sense primer 5′- GTG TTT CCC TGT TCA GCC AT-3′ and the antisense primer 5′- GTT CAG GCT TCA GCT TTT GC-3′ (Genebank accession no. NM_053819). A 232-bp GAPDH product was amplified using the sense primer 5′-TCT TCA CCA CCA TGG AGA A-3′ and the antisense primer 5′-ACT GTG GTC ATG AGC CCT T-3′ (Genebank accession no. NM_017008). PCR amplification was conducted using the following protocol: a predenaturation at 95°C for 5 min and then either 40 cycles (for MMP-1) or 30 cycles (for all other genes) of denaturation at 94°C for 35 s, annealing at 52°C for 35 s, and extension at 72°C for 35 s, followed by a final extension at 72°C for 10 min. The amplified mRNA products were separated by electrophoresis in 2.5% agarose gels and photographed under UV light in the presence of ethidium bromide. The products were confirmed not only by DNA loading marker but also by sequencing.

Hairpin small-interfering RNA plasmid.

The hairpin small interfering (si)RNA oligonucleotides (MMP-1 sense, 5′-GATCC GTGAACAAGCTTCAGGTAA TTCAAGAGA TTACCTGAAGCTTGTTCAC AGA-3′; MMP-1 antisense, 5′-AGCTTCT GTGAACAAGCTTCAGGTAA TCTCTTGAA TTACCTGAAGCTTGTTCAC G-3′) contained a region specific to bases 1501 to 1519 of MMP-1 mRNA (underlined), a hairpin loop region (italicized), and 5′ and 3′ linker sequences for subcloning into the BamHI and HindIII sites of the pSilencer 4.1 CMV-neo vectors (Ambion). After the cloning, the plasmid was then confirmed by both BamHI and HindIII digestion and sequencing. The plasmid that had no siRNA insert was used as a vector sham.

MMP-1 siRNA transfection.

MMP-1 siRNA plasmid was transfected into rat aortic SMCs and MFBs (both at passage 4) using Lipofectamine (Invitrogen), following the manufacturer's instruction. The cells were then split once. After reaching 80% confluence, a fraction of the cells was used to verify MMP-1 gene knockdown by RT-PCR and the other cells were directly suspended into collagen gels for flow-migration experiments. After the flow experiments, both RT-PCR and MMP-1 activity assays were performed to check MMP-1 gene knockdown and activity reduction.

Cell apoptosis and viability assay.

After 48 h of chemotatic incubation, the cells in the flow-treated collagen gels were stained using the Vybran apoptosis assay kit no. 2 (Invitrogen). Live cells in different gels with the same flow treatment were stained with calcein AM (Invitrogen) to check viability. Both assays were carried out according to the supplier's instructions. The images were taken under an inverted fluorescent microscope by focusing on the layer of cells right above the cell culture insert membrane.

Cell apoptosis induction and wound healing assay.

Rat aortic MFBs and SMCs were grown in small dishes until 80% confluency. The growth media was then replaced by tissue necrosis factor-α (TNF-α; 20 ng/ml) and cycloheximide (CHX; 3 μg/ml) (Sigma) in complete growth media for 2.5 h to induce apoptosis. Wounds in the monolayer cells were scratched using a small pipette tip. TNF-α/CHX was replaced by PDGF-BB (20 ng/ml) in DMEM for cell migration for 24 h.

Data analysis.

The results are presented as means ± SE. The data sets were analyzed for statistical significance using a Student's t-test with a two-tailed distribution, and P < 0.05 was considered statistically significant. For comparison of more than two groups, we used one-way ANOVA followed by the t-test with Bonferroni correction, and P < 0.05/number of comparisons (N) was considered statistically significant.


Flow velocity, Darcy permeability, and shear-stress estimation.

The Darcy permeability (Kp) of the gel/cell layer is defined by Kp=μ(Jv/A)/(ΔP/L) and Wang and Tarbell (52) developed a theory for the interstitial flow-induced shear stress (τ) on cells imbedded in a porous matrix that can be expressed approximately as τμ(Jv/A)/Kp where μ is the viscosity of the flow medium, Jv is the volumetric flow rate, A is the cross section area of the gel, ΔP is the pressure drop across the gel, and L is the thickness of the gel. Interstitial flows driven by 1, 3, and 10 cmH2O pressure drops induced shear stresses on cells in collagen gels of 0.05, 0.10, and 0.36 dyn/cm2, respectively. The Kp was on the order of 10−11 cm2, and the interstitial flow velocity was in the range of 0.5–2.4 μm/s. Gel compaction also occurred as the gel thickness decreased with increasing flow. The Kp, flow velocity, and shear stress for the experimental conditions are summarized in Table 1.

View this table:
Table 1.

Flow velocity, Darcy permeability, and shear-stress estimation

Interstitial flow affects cell motility in collagen gels.

The migration rates of MFBs, FBs, and SMCs (Fig. 2) were significantly elevated after exposure to 1–6 h of 1 cmH2O flow compared with no-flow controls. One hour of 3 and 10 cmH2O flow also promoted MFB and FB motility but did not affect that of SMCs. However, 6 h of 10 cmH2O flow suppressed the migration for all three types of cells. Overall, the migration rates were higher in lower flow cases compared with corresponding time points in higher flow cases. The trends of migration rates in response to flow were very similar for FBs and MFBs. The difference between SMCs and FBs/MFBs was that there was no enhancement of SMC migration in any of the 3 and 10 cmH2O flow cases.

Fig. 2.

Effects of interstitial flow and matrix metalloproteinase (MMP) inhibitor GM-6001 on vascular myofibroblast (MFB), fibroblast (FB), and smooth muscle cell (SMC) migration rates in 3-D collagen gels. Cells in 3-D collagen gels were exposed to up to 6 h of flow driven by 10, 3, or 1 cmH2O pressure drop and then incubated with 1 ml of 20 ng/ml PDGF-BB to examine cell migration. In the cases of MMP inhibition, GM-6001 (1 ml of 10 μM) was added together with PDGF-BB. In x-axis, 0 stands for no-flow control and values denoting flow intensity (in cmH2O)-exposure time (in h) are shown. Data are means ± SE; n = 4–19 experiments. *P < 0.05 vs. no-flow control; #P < 0.015 vs. their corresponding flow cases without GM-6001 treatment.

The effect of flow on cell migration rates in collagen gels was also visualized using confocal microscopy. The migration rate can be inferred from the cell spatial distribution, which can be represented more concisely by the intensity profile of fluorescently labeled cells (see online data supplement of Fig. IA, IB, IC, and ID). This observation is in agreement with the above migration counting data.

MMP inhibitor abolishes interstitial flow-induced migration.

A broad spectrum MMP inhibitor (GM-6001) was used to investigate the potential role of MMPs in interstitial flow-enhanced vascular FB, MFB, and SMC motility. GM-6001 had very similar inhibitory effects on both FB and MFB migration (Fig. 2): it attenuated migration in no-flow control gels to 70–80% and completely abolished the elevated migration in 1 cmH2O flow-treated gels. Surprisingly, GM-6001 had no effect on SMC migration in no-flow control gels, but 1 cmH2O flow-enhanced migration was attenuated nearly to control levels by GM-6001.

Interstitial flow affects MMP-1 and TIMP-1 levels but not MMP-2.

The AnaSpec MMP-1 activity assay showed that active MMP-1 levels in all flow-promoted cell migration cases were significantly increased compared with no-flow controls (compare Figs. 2 and 3). There were no statistically significant changes in MMP-1 activity in 6-h flow-treated gels for MFBs and FBs at 3 and 10 cmH2O. For SMCs, there were no significant changes in MMP-1 activity in any of the 3 and 10 cmH2O flow cases. To check whether there was a relationship between cell migration and MMP-1 activity, cell migration was plotted versus MMP-1 activity (Fig. 4), and this clearly showed a good correlation between MMP-1 activity trends and cell migration trends. Because of the strong similarities in cell migration and MMP-1 activity between FBs and MFBs, the rest of this study focused on the MFB and SMC cell types.

Fig. 3.

Interstitial flow affects MMP-1 activity. MMP-1 activity in the conditioned medium was determined by AnaSpec MMP-1 activity assay kits. In x-axis, 0 stands for no-flow control and values denoting flow intensity (in cmH2O)-exposure time (in h) are shown. Data are means ± SE; n = 4 experiments. *P < 0.05 vs. no-flow control.

Fig. 4.

The correlation between MMP-1 activity and cell migration. Cell migration data were plotted as a function of MMP-1 activity data. The lowest MMP-1 activity data points for both cell types were from the MMP-1 small interfering (si)RNA experiment (Fig. 7). A good correlation is shown between migration and MMP-1 activity. Data also suggest that the baseline migrations (∼75% for MFBs and ∼100% for SMCs) are not MMP-1 dependent. In addition, there is no essential difference in migration between MFBs and SMCs.

Collagen zymography was also used to measure MMP-1 levels (Fig. 5). 1 After exposure to flow, active MMP-1 levels for MFB were markedly increased. In SMCs, active MMP-1 levels were only increased in lower flow cases and no significant changes were observed in the higher flow (10 cmH2O) cases. These data are consistent with the AnaSpec MMP-1 activity assay results (Fig. 3) and also with cell migration data (Fig. 2). The only exceptions are for the 3 and 10 cmH2O flow cases at 6 h in MFB, where collagen zymography showed enhanced MMP-1 activity. The apparent discrepancy between these methods of determining MMP-1 levels was due to the fact that zymography detected all active MMP-1, not only free active MMP-1 but also MMP-1 in MMP-1/TIMP complexes that were dissociated in zymography (43). These data suggest that high flow induced more TIMP expression. To check the TIMP-1 level, reverse collagen zymography was performed, which clearly showed that longer exposure to higher flow induced more TIMP-1 expression in both MFBs and SMCs (Fig. 5). By gelatin zymography, we observed, surprisingly, that for both SMCs and MFBs, the active MMP-2 levels displayed no significant changes in response to flow at any time point (Fig. 5). AnaSpec MMP-2 activity assays confirmed this finding (data not shown). In addition, the MMP-1 protein level was much lower than that of MMP-2 in the present study (data not shown).

Fig. 5.

Collagen and gelatin zymography for MMP-1 and MMP-2 and reverse collagen zymography for tissue inhibitor of metalloproteinase-1 (TIMP-1). For MMP-1 and MMP-2, an identical volume (20 μl) of conditioned medium was loaded into each well. The first bands in the zymograms are the pro-MMPs, and second bands are the active MMPs. For reverse collagen zymography, 40 μl of conditioned medium were loaded into each well. In the panels, 0 stands for no-flow control and x1-x2 stands for flow intensity (in cmH2O)-exposure time (in h). Note: in some cases, in order to keep the same format, the zymograms were spliced from 2 distinct gels or from the same gel but with some unwanted bands removed or the orders of some bands switched. The vertical white lines on the figures indicate where the splices or rearrangements were made. The quantification of each band was accomplished with its own control for 2 distinct gels or with the same control if they were originally from that same gel. Local background subtraction was used for the quantification of the zymograms.

Interstitial flow upregulates MMP-1 and TIMP-1 gene expression but not MMP-2.

Interstitial flow significantly promoted MMP-1 gene expression for both MFB and SMC (Fig. 6). Overall, the longer the exposure, the higher the MMP-1 expression, and the higher the flow, the greater the expression. There was no detectable MMP-1 gene expression in SMCs for the no-flow control case. By comparison, SMCs expressed less MMP-1 than MFBs. The MMP-2 gene expression was much higher than MMP-1 expression in controls, but there was no change in any flow case compared with controls for either SMCs or MFBs. TIMP-1 gene expression was also significantly upregulated in MFBs by longer exposure to a higher intensity of flow and slightly upregulated in SMCs.

Fig. 6.

Interstitial flow upregulates MMP-1 and TIMP-1 gene expression but not MMP-2. Rat vascular MFBs and SMCs were released from collagen gels by collagenase right after flow. Reverse-transcription polymerase chain reaction (RT-PCR) was conducted to detect rat MMP-1, MMP-2, and TIMP-1 mRNA expression with GAPDH as a reference gene. The exposure times for taking images were different for different genes. In the panels, 0 stands for no-flow control and values denoting flow intensity (in cmH2O)-exposure time (in h) are shown.

Silencing MMP-1 expression eliminates interstitial flow-promoted cell motility.

To further verify that MMP-1 was required for flow-induced migration, we silenced MMP-1 expression by siRNA. MMP-1 siRNA completely abolished the flow-enhanced migration for both MFBs and SMCs (Fig. 7A), whereas there was no effect on control cases. MMP-1 RNA interference data were highly consistent with the general MMP inhibitor results (Fig. 2). MMP-1 gene expression knockdown and activity reduction were confirmed by RT-PCR and MMP-1 activity assay (Fig. 7, B and C).

Fig. 7.

MMP-1 siRNA abolishes flow-enhanced cell migration. A: cell migration data. B: RT-PCR was used to confirm MMP-1 gene silencing. C: MMP-1 activity assay was used to check MMP-1 activity reduction. In the panels, 0 stands for no-flow control and values denoting flow intensity (in cmH2O)-exposure time (in h) are shown. Data are means ± SE; n = 3 to 4 experiments. *P < 0.05 vs. no flow control; #P < 0.008 vs. their corresponding flow cases with sham transfection.

High-intensity interstitial flow induces cell apoptosis in 3-D collagen.

Longer exposure to higher intensity flow induced SMC and MFB apoptosis and necrosis (Fig. 8). Six hours of 10 cmH2O flow induced apoptosis in many more cells compared with 3 h of 1 cmH2O flow. In addition, many more dead cells and fewer normal live cells were observed due to a longer exposure to a higher intensity of flow. A wound healing assay showed that the cells in the dish containing more apoptotic MFB induced by TNF-α/CHX displayed much lower motility than the dish that had not been induced (Fig. 9). Similar phenomena were also observed for SMCs (data not shown).

Fig. 8.

Longer exposure to higher flow induces cell apoptosis and necrosis. Cells in collagen gel were exposed to 1 cmH2O flow for 3 h (A, C, and C′) and 10 cmH2O flow for 6 h (B, D, and D′) and then incubated for 48 h with PDGF-BB as a chemoattractant. SMCs (A and B) and MFBs (C, C′, D, and D′) are shown. Cells in gels were stained either by Vybrant apoptosis assay kit no. 2 or by calcein AM. Images were taken by focusing on the layer of cells right above the cell culture insert membrane. In A–D, apoptotic cells were stained with Alexa Fluor 488 annexin V in green and dead cells (necrosis) were stained with propidium iodide (PI) in red. In C′ and D′, live cells in gels were stained with calcein AM in green.

Fig. 9.

Apoptosis impairs cell motility. In the wound healing assay, a monolayer of cells in a dish was treated with (A) or without (B) tissue necrosis factor (TNF)-α (20 ng/ml) and cycloheximide (CHX) (3 μg/ml) for 2.5 h to induce apoptosis, followed by wound scratching shown as the space between 2 broken white lines; media was then replaced by DMEM with PDGF-BB (20 ng/ml) for cell migration for 24 h. Representative cell type was MFB.


This was the first study to investigate the influence of interstitial flow on the migration of vascular wall cells (FBs, MFBs, and SMCs) in a 3-D model and to determine the underlying mechanisms. The primary finding is that interstitial flow can promote vascular FB/MFB and SMC motility via the upregulation of MMP-1. Separation of the flow period from the migration period ensured that the effects of flow on the migration rates could not be interpreted as resulting from the convection of chemoattractant or other molecules produced by the suspended cells. We also eliminated the effect of gel compaction due to flow, because gel compaction affects gel fiber density and thus gel permeability and stiffness, which in turn affect cell migration rates (17, 53). It is also clear in Fig. 1B that the direction of interstitial flow corresponds to the direction of cell migration. However, one cannot argue that there was a direct mechanical enhancement of migration in the direction of flow (drag of flow on the suspended cells) during the initial 1–6-h flow period, because cell distributions in the gels, determined by confocal microscopy, were not different for no-flow controls and gels exposed to flows (data not shown).

The permeability of the collagen gel is very important since it controls the mass transport to cells by diffusion and convection and the shear stress on cells. The permeability depends strongly on the gel concentration. In the present study, the Kp of 4 mg/ml collagen gels was on the order of 10−11 cm2, which is two to three orders higher than in the layers of the rabbit aortic wall (10−14 cm2) (16). However, our Kp is consistent with data in the literature: 10−8-10−12 cm2 for 2.5–45 mg/ml collagen gels (27, 32, 52). The interstitial flow velocity in our experiments was in the range of 0.5–2.4 μm/s, which is substantially higher than the interstitial flow velocity in the normal aorta (0.01–0.1 μm/s) (49, 51). But the estimated shear stresses on suspended cells in our experiments (0.05–0.36 dyn/cm2) are in the expected range for the aorta based on Eqs. 1 and 2. In addition, the porous membrane of the culture inserts with 8-μm pore size (area fraction of the pores is 0.03) mimics the external/internal elastic lamina, whose fenestral pore area fraction is 0.001–0.036 (47). The pore size of an uncompacted 2.5 mg/ml collagen gel was observed by electron microscopy to be no more than 1 μm, which is much smaller than the cell diameter (∼10 μm) (52). The gel pore size in the present study that used compacted 4 mg/ml collagen gels should be somewhat smaller but of the same order of magnitude. The cells would not be able to crawl through the pores of such small dimensions and must use enzymes (MMPs) to create accessible pathways through the matrix.

The basic phenomena of flow versus migration displayed in Fig. 2 suggests that the lower levels of interstitial flow (1 cmH2O differential pressure) enhance the motility of all cell types studied (FBs, MFBs, and SMCs), whereas the higher levels of interstitial flow (10 cmH2O differential pressure) at longer exposure times (6 h) suppress migration rate of all cell types. The roles of MMPs, TIMPs, and apoptosis in mediating these diverse phenomena have been investigated and all are discussed below.

In an earlier study using a 2-D flow system, we observed that much higher shear stress (∼10 dyn/cm2) enhanced FB migration but inhibited MFB and SMC migration in Matrigel (7, 8). Differences in shear-stress level, matrix material (collagen vs. Matrigel), and system dimension (2-D vs. 3-D) undoubtedly contributed to these differences. In the more physiological 3-D system, the cells exhibit matrix adhesions all over their surface, thus the interstitial flow not only acts directly on the cell surface but also affects the matrix structure, cell-matrix adhesion and tethering that could result in amplified mechanosignaling (17, 29, 45).

Vascular SMC, FB, and MFB migration and MMP secretion are typical features of neointima formation after endothelial injury (2, 33). The proteolytic effects of MMPs play an important role in vascular remodeling and cellular migration (6), and there is abundant evidence of MMP upregulation in animal models of neointima formation (25). The levels of MMP-1 and other MMPs are increased after femoral artery injury in mice (18), and MMP-1 upregulation occurs in injured monolayers of vascular SMCs (12). In the present study, a broad-spectrum MMP inhibitor GM-6001 completely abolished flow-promoted cell motility (Fig. 2), suggesting that MMPs play a major role in flow-enhanced cell motility in collagen matrixes. We further investigated whether MMP-1 and/or MMP-2 play a specific role in the flow-enhanced migration rate. We found that the flow induced MMP-1 protein and gene expression (Figs. 3, 5, and 6) and that the trends in MMP-1 activity showed good agreement with the trends in cell migratory activity (Fig. 4); however, there was no effect of flow on MMP-2 at either the gene or protein level (Figs. 5 and 6). By using MMP-1 siRNA, we further confirmed the critical role of MMP-1 in flow-regulated migration rate (Fig. 7). MMP-1 siRNA almost completely abolished flow-induced migration (Fig. 7A). In addition, the knockdown of MMP-1 did not affect membrane-type MMP-1 (MT-MMP-1; MMP-14) and MMP-13 gene expression (data not shown). Because the knockdown of MMP-1 was not 100%, the flow activation of MMP-1 expression was still present. However, the MMP-1 activity in MMP-1 siRNA samples was significantly lower than its corresponding time point in sham, and only slightly higher than that in the no-flow sham control (Fig. 7C). Combining the migration data with the MMP inhibitor, MMP-1 siRNA, and MMP-1 activity data (Figs. 2, 4, and 7), we also noted that migration in no-flow controls was partly MMP dependent for FBs/MFBs and totally independent of MMP for SMCs.

We have shown that the upregulation of MMP-1 but not MMP-2 by interstitial flow substantially mediated vascular FB, MFB, and SMC migration in collagen I matrixes. Consistent with this observation, it has also been reported that MMP-1 is essential for hepatocyte growth factor-mediated human corneal epithelial cell migration on collagen I (4). However, other studies have shown that MMP-2 plays an important role in vascular SMC and FB migration both in vivo and in vitro. Changes in MMP-2 activity affect SMC and FB migration, which contribute to neointima formation (2, 26, 39, 54). Two-dimensional shear stress suppresses SMC migration in Matrigel via the downregulation of MMP-2 activity (8, 28). The upregulation of MMP-1 and -2 are responsible for chlorotyrosine-induced human aortic SMC migration (24). Other recent studies reported that fluid shear stress modulated endothelial cell invasion into 3-D collagen matrixes through MMP-2 activation and that collagen I but not Matrigel matrixes form an MMP-dependent barrier to ovarian cancer cell penetration (13, 44). However, it has been shown that MMP-2 activation is delayed in rat and mouse studies (25). In vivo, the accumulation of new ECM into the extravascular space of the damage site could alter the properties of the matrix that cells migrate within, which might affect MMP expression and activation. Therefore, in our 6-h flow experiments, MMP-2 might have also been delayed. All of these findings indicate that MMP activation is dependent on many factors including substrate materials, cell types, and the nature and duration of stimuli.

TIMPs play a crucial role in maintaining vascular wall homeostasis (23, 31, 39). In addition to MMP-1, the TIMP-1 gene and protein expression were also elevated by higher flow and longer exposure (Figs. 5 and 6). The TIMP-1 response may have been the attempt of the cell to eliminate excessive MMP formation to strike a balance. The elevated TIMP-1 expression by higher flow certainly could partially reduce MMP-1 activity, which in turn would reduce the cell migration rates.

We also observed that at the highest flow rate and longest exposure time, migration rates were suppressed for all cell types (Fig. 2). This was most likely related to both the elevation of TIMP-1 expression and the appearance of apoptotic and necrotic cells under these conditions (Fig. 8). The apoptotic cell populations displayed a reduced migration rate as expected (Fig. 9). The more apoptotic and necrotic cells induced by higher flow together could significantly reduce cell migration rates. It has been shown that enhanced SMC apoptosis and reduced migration are likely involved in the inhibition of neointima formation by a Rho-kinase inhibitor (40). Vascular SMC apoptosis is one characteristic of vascular remodeling that occurs in atherosclerosis, hypertension, and restenosis following angioplasty (20, 37). The apoptosis of SMCs often occurs rapidly right after vascular injury (30). It has been reported that fluid shear stress can induce tumor cell-cycle arrest and vascular SMC apoptosis in 2-D in vitro and that changes in wall tension can also cause SMC apoptosis in vivo (3, 5, 36). Apoptosis can also be induced in the ECM with a low stiffness (15), and the collagen matrix stiffness can be reduced when MMP expression is overabundant. However, in the present study, the largest number of apoptotic cells appeared in the longest exposure to the highest flow case, whereas the MMP-1 activity was at a lower level. These data suggest that cell apoptosis was mainly induced by interstitial flow, not reduced matrix stiffness.

The precise physiological relevance of the results of the present study can only be hypothesized at the present time. One plausible scenario that incorporates the basic findings on interstitial flow and migration (Fig. 2) is that after vessel injury, the intima is exposed to elevated concentrations of chemoattractant originating in blood elements that initiate cell migration to the intima. Elevated interstitial flow increases shear stress on all cells in the wall, but it induces a lower shear stress on cells like FBs and MFBs that are embedded in the loose connective tissue of the adventitia (with higher Kp than the medial connective tissue) than on the SMC in the media (recall Eq. 2). The migration of these adventitial cells is therefore elevated early by the enhanced interstitial flow, whereas the SMC migration is suppressed. As injury healing proceeds, the interstitial flow shear stresses are reduced and a period in which the SMC migration is enhanced and the FB/MFB migration is reduced ensues. The net result is an enhanced migration of all cell types by interstitial flow during some phase of the injury healing process.

Taken together, our results indicate that interstitial flow may be one of the direct links between vessel injury and vascular FB and SMC migration. Interstitial flow can promote vascular cell motility by stimulating MMP-1 expression and activation, whereas if the flow is too high, it might suppress cell motility. However, to determine exactly how the vascular cells “sense” interstitial flow and then promote MMP-1 expression requires additional investigation. That longer exposure to higher interstitial flow induces cell apoptosis and necrosis also needs to be further clarified. Certainly, neointima formation from vascular injury in vivo is very complicated, involving not only elevated flow but also inflammatory cells and other factors. However, using a 3-D collagen I gel system as a model of physiological interstitial flow, we, for the first time, have been able to observe significant influences of interstitial flow on cell migration. Our results suggest a possible mechanism whereby vessel injury enhances interstitial flow that activates medial SMCs and adventitial FBs and thus further contributes to neointima formation.


This work was supported by National Heart, Lung, and Blood Institute Grants HL-35549 and HL-57093.


We thank Dr. Herb Sun and Jeff Garanich for valuable discussions, Danielle E. Berardi for excellent technical assistance, and Giya Abraham for cell culture.


  • 1The final published versions of Figs. 5, 6, and 8 in this article differ from the figures originally published in the Articles-in-PresS version. The background intensity of several gel panels was adjusted for clarity in Figs. 6 and 8 in the original version, but the final published versions are the original gels without adjustments. The gel panels in Fig. 5 were spliced together from several gels. The final published version of Fig. 5 indicates where the splices were made. None of these alterations has any effect on the conclusions or interpretations presented in the paper.


  1. 1.
  2. 2.
  3. 3.
  4. 4.
  5. 5.
  6. 6.
  7. 7.
  8. 8.
  9. 9.
  10. 10.
  11. 10a.
  12. 11.
  13. 12.
  14. 13.
  15. 14.
  16. 15.
  17. 16.
  18. 17.
  19. 18.
  20. 19.
  21. 20.
  22. 21.
  23. 22.
  24. 23.
  25. 24.
  26. 25.
  27. 26.
  28. 27.
  29. 28.
  30. 29.
  31. 30.
  32. 31.
  33. 32.
  34. 33.
  35. 34.
  36. 35.
  37. 36.
  38. 37.
  39. 38.
  40. 39.
  41. 40.
  42. 41.
  43. 42.
  44. 43.
  45. 44.
  46. 45.
  47. 46.
  48. 47.
  49. 48.
  50. 49.
  51. 51.
  52. 52.
  53. 53.
  54. 54.
View Abstract