MMP-14 is necessary but not sufficient for invasion of three-dimensional collagen by human muscle satellite cells

Abstract

The twenty-five known matrix metalloproteases (MMPs) and their endogenous inhibitors, tissue inhibitors of metalloproteases (TIMPs), mediate cell invasion through the extracellular matrix (ECM). In a comparative three-dimensional assay, we analyzed human and mouse satellite cells' competence to invade an artificial ECM (collagen I). We identified a single MMP that 1) is expressed by human muscle satellite cells; 2) is induced at the mRNA/protein level by adhesion to collagen I; and 3) is necessary for invasion into a collagen I matrix. Interestingly, murine satellite cells neither express this MMP, nor invade the collagen matrix. However, exogenous human MMP-14 is not sufficient to induce invasion of a collagen matrix by murine cells, emphasizing species differences.

mammalian adult skeletal muscle is composed of parallel, syncytial myofibers derived from the fusion of differentiated myocytes during fetal and postnatal development. Because the resulting myonuclei are terminally postmitotic, muscle growth, repair, or regeneration requires muscle satellite cells, a population of muscle-specific precursor cells located between the sarcolemma and the basal lamina in uninjured muscle (35, 51, 53, 71). Satellite cells are the obligate stem cell of skeletal muscle (45, 54, 68): after activation by damage-induced local or systemic signals they will expand to generate a pool of differentiation-competent myocytes, which will then fuse to each other or to existing myofibers to replace or repair damaged muscle tissue. Although the soluble factors influencing satellite cell activation, proliferation, and (to a lesser extent) motility are increasingly well described (18), our understanding of the roles of signals derived from cell-matrix interactions and matrix remodeling in vivo is limited. The extracellular matrix (ECM) in skeletal muscle is made up of both interstitial matrix (composed of different multiple collagen isoforms, fibronectin, hyaluronic acid, and proteoglycans such as perlecan) and myofiber-associated basement membrane (composed of organized layers of collagen IV, collagen VI, and laminin) (29, 33, 69). Collagen type I in particular is dynamically expressed in muscle by multiple different resident cell types, including fibroblasts, satellite cells, and differentiated myofibers (30, 32). During skeletal muscle regeneration, satellite cells first exit their niche between the lamina and the sarcolemma, then transit on the myofiber lamina as they reposition to the site of injury. This would suggest that satellite cells have the potential to traverse the interstitial matrix, and indeed satellite cells have been observed protruding from the lamina into the interstitial matrix as well as relocating between myofibers (39).

Cell-matrix interactions are frequently bidirectional: binding of ECM proteins by cellular adhesion receptors such as integrins activates intracellular signaling pathways, leading to changes in cell proliferation, cell shape, and cell motility (40), but the ECM is also actively remodeled by secreted and cell surface proteases during processes such as wound healing, fibrotic diseases, and scar formation. This remodeling not only changes the physical properties of the ECM, it liberates cryptic cleavage products that are then capable of signaling to local cells, including satellite cells (1). In addition to degradation and remodeling, deposition of new matrix proteins by local cells will modify the local signaling environment, and thus change the kinetics of tissue regeneration and/or remodeling. For example, recent work suggests that active production of fibronectin by satellite cells modulates their activity and stem cell status by altering their local niche signaling (8).

Matrix metalloproteases (MMPs) are zinc-dependent endopepsidases that play a major role in the constructive proteolysis of the ECM, including collagen type I (13, 85), and are required for cellular invasion through the ECM (77, 88). Twenty-five members of the MMP family have been identified; most are secreted, but six are membrane-type MMPs (MT-MMPs). MMPs are historically grouped as collagenases, gelatinases, stromelysins and matrilysins based on their respective specificity for ECM substrates (22). MMPs are responsible for the activation and processing of secreted molecules during ECM remodeling, including chemokines, cytokines, and growth factors (65, 83), which can in turn reciprocally activate MMPs (2, 37, 75, 89). In general, pathological activation of MMPs is implicated in inflammation, angiogenesis, and cell death, as well as in kidney disease, tumor proliferation and metastasis (14, 89). In skeletal muscle, MMP activity is implicated in both homeostasis and regeneration, and impinges on myofiber integrity, satellite cell activation and the adult as well as myoblast proliferation, migration and fusion during development (16, 24, 26, 42, 57, 84, 91).Therefore, as would be expected, overall MMP activity increases upon injury, correlating with a decrease in intact ECM proteins (16), and inhibition of MMPs also inhibits skeletal muscle regeneration (84). In cell transplant studies, MMP activity has been implicated in migration of exogenous myoblasts, and is therefore a potential therapeutic target (24, 81).

In this study we show that human, but not mouse, satellite cells are competent to invade a three-dimensional (3D) collagen type I matrix in vitro. We also show that expression of the membrane-type metalloprotease MMP-14 (also known as MT1-MMP) is necessary for this invasion. Interestingly, primary murine satellite cells neither express MMP-14 nor invade a collagen I matrix, highlighting the importance of MMP-14 activity for ECM invasion and myoblast migration in vivo. These data highlight one of several differences between species, and may identify an area of potential concern for the translation of murine results into human therapeutic perspective. In particular, these data might be useful in interpretation of satellite cell transplant experiments in cases in which human and murine grafts have given differing results.

METHODS

All methods involving live animals were performed under approval from the University of Missouri Animal Care and Use Committee.

Murine cell culture.

C2C12 cells were cultured in DMEM (GIBCO) supplemented with 15% fetal bovine serum. Mouse satellite cells were harvested from hindlimb muscles of adult (90–120 days) females using our established protocol (19). Mouse muscle fibroblasts were harvested simultaneously and enriched by differential plating 24 h after harvest. Primary cells and MM14s were cultured on gelatin-coated plates (Nunc) in Ham's F-12 media (GIBCO) supplemented with 10% horse serum (HS) and 5 nM fibroblast growth factor 2 (FGF2) (58).

Human cell culture.

Immortalized human satellite cells were originally isolated from a muscle biopsy taken from a 25-yr-old male donor, obtained from Myobank, affiliated with EuroBioBank, in accordance with the French legislation and EU regulation (10). They were grown in a one-fourth volume of 199-DMEM media respectively supplemented with 2.5 ng/ml hepatocyte growth factor (HGF; Miltenyi), 10⁻⁷ M dexamethasone (Invitrogen), and 20% fetal calf serum (GIBCO) as previously described (50). Primary human satellite cells were grown under the same conditions, except for the 5-yr-5-day-old clone, which was not supplemented with HGF or dexamethasone. All human cultures were also challenged in an invasion assay in which they were preconditioned for 24 h then plated on the 3D collagen matrix in murine growth medium (Ham's F-12 + 10% HS + FGF2); this did not affect their invasion.

3D collagen culture.

One hundred microliters of acid-extracted rat-tail type I collagen (Sigma) (2 mg/ml in growth medium) was added per well in 96-well plates (Corning) and allowed to rapidly polymerize at 37°C. When used, recombinant tissue inhibitors of metalloproteases (TIMPs) 1, 2, or 3 (R&D Systems) at a final concentration of 3.5 nM (IC₅₀ provided by the supplier is 2.2, 2.5, and 3.0 nM, respectively) or GM6001 (Calbiochem) at a final concentration of 10 μM was added to the collagen prior to polymerization. Satellite cells (50,000) suspended in 100 μl growth medium were added and cultured for 90 h. All conditions in each experiment were repeated in triplicate, in at least five independent runs.

The matrices were fixed in 4% paraformaldehyde, equilibrated into 50% sucrose, and snap frozen. Cryosections (40 μm) were labeled with 1 μg/ml phalloidin-594 or -635 (Invitrogen) and Vectashield containing DAPI (Vector) for analysis. Invading cells were quantified using the cell counting function of μManager (www.micro-manager.org); criteria to score a cell in later statistical analysis required the entire nucleus and cytoplasm to be present in the section.

Live-cell imaging.

C2C12 myoblasts adhered to a 48-well plate (Corning) and overlaid with 3D collagen type I (previously described) copolymerized with DQ collagen type I (1 mg/ml; Life Technologies) were assayed via time-lapse microscopy. Images were automatically collected from each field every 15 min for 6 h using IPLab (Scanalytics) and analyzed via μManager. Images are representative of triplicate wells.

Immunostaining.

Cells were immunostained as described previously (79). Rabbit monoclonal (EP1264Y) anti-MMP-14 (Abcam) was used at 1:250; this antibody recognizes both the inactive and active forms of both human and mouse MMP-14. Mouse monoclonal anti-Pax7 (Developmental Studies Hybridoma Bank) was used neat. Samples were stained with fluorescently labeled secondary antibodies (Invitrogen) at 1:500 and/or fluorescently labeled phalloidin (Invitrogen) at 1 μg/ml concentration. Nuclei were visualized with DAPI (Vector.)

Slides were imaged with an Olympus BX61 microscope using SlideBook (Intelligent Imaging Innovations) and μManager software. Z-stacks were generated for each image, and Z-projections were used for analysis and quantification of cell invasion.

Injury.

Adult mice were anesthetized with 2,2,2-tribromoethanol (Avertin; Sigma) and injected intramuscularly with 50 μl of 1.2% barium chloride in sterile solution. Images shown are of muscles harvested 5 days after injury.

RT-PCR.

Total RNA from human (25-yr-old male clone) or primary mouse satellite cells cultured on collagen-coated 10-cm plates was reverse-transcribed into cDNA (SuperScriptIII, Invitrogen). Four hundred nanograms of each cDNA sample was used as template; primer sequences used were as follows: GAPDH forward (F), 5′-CAAGGTCATCCATGACAACTTTG-3′, reverse (R), 5′-GGGCCATCCACAGTCTTCTG-3′; mouse MMP-2 F 5′-GGAGAAGGCTGTGTTCTTCG-3′, R 5′-GCATCTACTTGCTGGACATCAG-3′; mouse MMP-9 F 5′-CAGAGGTAACCCACGTCAGC-3′ R 5′-GGGATCCACCTTCTGAGACTT-3′; mouse MMP-14 F 5′-GGACTGAGATCAAGGCCAAT-3′, R 5′-GCCCACCTTAGGGGTGTAAT-3′; mouse TIMP-1 F 5′-TACGCCTACACCCCAGTCAT-3′, R 5′-ATGTGCAAATTTCCGTTCCT-3′; mouse TIMP-2 F 5′-AGGTACCAGATGGGCTGTGA-3′, R 5′-GTCCATCCAGAGGCACTCAT-3′; mouse TIMP-3 F 5′-CCACGTGCAGTACATTCACAC-3′, R 5′-TGTACATCTTGCCTTCATACACG-3′; human MMP-2 F 5′-TATGGCTTCTGCCCTGAGAC-3′, R 5′-CACACCACATCTTTCCGTCA-3′; human MMP-9 F 5′-TCGTCATCCAGTTTGGTGTC-3′, R 5′-ATGGGCGTCTCCCTGAAT-3′; human MMP-14 F 5′-GGCAAATTCGTCTTCTTCAAA-3′, R 5′-GAGCAGCATCAATCTTGTCG-3′; human TIMP-1 F 5′-CTGTTGTTGCTGTGGCTGAT-3′, R 5′-AACTTGGCCCTGATGACG-3′; TIMP-2 F 5′-AGAAGAGCCTGAACCACAGG-3′, R 5′-TGACCCAGTCCATCCAGAG-3′; TIMP-3 F 5′-CCCATGTGCAGTACATCCATAC-3′, R 5′-CCATCATAGACGCGACCTG-3′.

siRNA transfection.

MMP-14 siRNA (15, 25, or 50 μM/well; Invitrogen) or control siRNA was transfected into human satellite cells cultured on gelatin-coated six-well plates (Nunc) using Lipofectamine RNAiMAX (Invitrogen) according to the manufacturer's protocols prior to seeding on 3D collagen I matrices. All samples were cotransfected with Alexa Fluor 555-labeled Red Fluorescent Oligo (Invitrogen) to identify transfected cells. Ninety hours after transfection, cells were fixed with 4% paraformaldehyde for analysis.

Immunoblotting.

Ninety hours after siRNA transfection, total cell lysates were collected in Allen buffer (50 mM Tris, 10 mM EDTA, 5 mM EGTA pH 7.4 with 1 × Roche Protease Inhibitor, 1 mM sodium orthovanadate, 20 mM sodium fluoride, 1 μg/μl pepstatin A, and 1% Triton X-100.) Ten micrograms of each lysate were loaded onto 4–8% gradient polyacrylamide gel (Invitrogen), transferred to polyvinylidene difluoride membranes, and blocked in StartingBlock (TBS) blocking buffer (Fisher). Membranes were incubated overnight at 4°C with primary antibodies to MMP-14 (EP1264Y, Abcam) at 1:2,000 and IP90 (AB10286, Abcam) at 1:2,000 in StartingBlock, washed, and incubated with horseradish peroxidase-conjugated secondary antibody (Santa Cruz) for 1 h at room temperature. Chemiluminescent substrate (Pierce SuperSignal West) was detected with a LAS3000 imager (Fujifilm).

MMP-14 overexpression.

Ninety-six hours after isolation, primary mouse satellite cells on gelatin-coated 10-cm plates (Nunc) were transfected with 5 μg of human MMP-14 expression vector (GenBank accession no. BC064803 cloned into pCMV-SPORT6, Open BioSystems) and 5 μg pCMV-Tomato reporter using 15 μl Fugene HD (Promega); control cells were transfected with 5 μg pCDNA4 and 5 μg pCMV-Tomato reporter. After 8 h the cells were seeded onto the 3D collagen type I matrices and cultured for an additional 90 h. Collagen matrices were then fixed, sectioned, immunostained, and imaged for analysis as above.

Zymography.

Forty micrograms of protein lysate were separated on 8% acrylamide gels polymerized with 2 mg/ml gelatin. After electrophoresis, gels were renatured in 2.5% Triton X-100 for 1 h at room temperature then incubated in developing buffer (50 mM Tris·HCl, 0.2 M NaCl, 5 mM CaCl₂, 0.02% Brij 35) at 37°C for 12 h. Gels were rinsed in distilled H₂O and stained with Coomassie R-250 for 30 min, then destained in methanol:acetic acid:water (50:10:40) and imaged on a flatbed scanner.

RESULTS

Human, but not mouse, satellite cells invade a 3D collagen type I matrix.

We compared the potential of primary adult murine satellite cells and adult human satellite cells immortalized by expression of telomerase and cdk4 (93) to invade a 3D collagen matrix, as an in vitro model of cellular movement through muscle tissue in vivo. Murine and human satellite cells were seeded on the top of a 2% collagen type I matrix and cultured for 90 h, after which the matrices were fixed, sectioned, and stained with Alexa 594-phalloidin to visualize satellite cell cytoarchitecture. Under these conditions, we observed that human satellite cells invade the collagen matrix an average of 87 μm (Fig. 1A) based on at least five sets of independent runs (see methods). In contrast, primary murine satellite cells failed to invade any detectable distance (Fig. 1B). The same result was seen in both cell types when the cells were first adhered to the culture plate then overlaid with 3D collagen I and challenged to invade “up” (data not shown).

Fig. 1.

Human, but not mouse, satellite cells invade a three-dimensional (3D) collagen type I matrix. A and B: human (A) and mouse (B) satellite cells after 90 h on a 3D collagen type I matrix; F-actin (Alexa 594-phalloidin, red) and nuclei (DAPI, blue). Human satellite cells invade the 3D collagen type I matrix, while mouse cells do not. Dotted line indicates the upper boundary of the 3D collagen. C: immunocytochemistry of primary murine satellite cells for phospho-focal adhesion kinase (green), phalloidin 594 (red), and DAPI (blue) on laminin, gelatin, and 2D collagen I. The negative no primary antibody control was performed on laminin.

To test whether poor adhesion to the collagen I prevented murine cells from initiating invasion, we compared adhesion of human and murine cells to both collagen and laminin. We have previously shown that murine satellite cells detectably express mRNAs for all integrin monomers except αE and αL, and that while laminin is the preferred substrate for primary mouse satellite cells as well as immortalized muscle cell lines, they adhere to and migrate over collagen I as well (72). Consistent with these data, murine satellite cells plated on collagen have a rounder morphology and fewer focal adhesions on collagen or gelatin than on laminin (Fig. 1C), but are nonetheless adhered. Murine cells also failed to invade matrices composed either of collagen I copolymerized with 50% or 75% laminin, or of Matrigel, which is ∼60% laminin (data not shown).

Invasion of collagen I by human satellite cells is MMP dependent.

While the primary mechanism of mesenchymal cell invasion is via MMPs, other proteases are expressed in muscle and could be active in remodeling the ECM and in the process of invasion of the matrix. To determine the specific participation of MMPs in satellite cell invasion, we treated human satellite cell cultures as above with a global MMP inhibitor, GM6001. This treatment completely prevented their invasion into the collagen matrix (Fig. 2), indicating that one or more MMPs are essential for invasion. To validate that the invasion is MMP dependent, we treated human satellite cells in 3D cultures with recombinant tissue inhibitors of MMPs (TIMP-1, TIMP-2, and TIMP-3), the endogenous inhibitors of MMPs (80). Each TIMP binds to specific MMPs at a different rate and affinity and blocks their catalytic activity: in general, TIMP-1 inhibits soluble MMPs, while TIMP-2 and TIMP-3 inhibit both soluble MMPs and membrane-type MMPs. TIMP-1 did not significantly inhibit invasion, but TIMP-2 and TIMP-3 both did, implicating membrane-type rather than soluble MMPs in this process.

Fig. 2.

Matrix metalloprotease (MMP) inhibitors differentially inhibit human satellite cell invasion. A–D: human satellite cell invasion in the presence of tissue inhibitors of metalloprotease-1 (TIMP-1; A), TIMP-2 (B), TIMP-3 (C), or GM6001 (D); all but TIMP-1 significantly inhibit invasion. E: quantitative comparison of human and mouse satellite cell invasion based on the fraction of cells invading beneath the surface of the collagen matrix, compared with all cells in the section (n ≥ 3). *P < 0.05. Statistically significant P values were based on comparison of protease inhibitors to the no inhibitor positive control.

Human, but not mouse, satellite cells express MMP-14 when adhered to collagen I.

The primary MMPs expressed in skeletal muscle are MMP-2, MMP-9, and MMP-14 (6, 16). In other systems, MMP-2 and MMP-9 are required for invasion of angiogenic, endothelial, and metastatic cells through the ECM (17, 23, 24, 92). MMP-14 (also known as MT1-MMP) is more broadly associated with invasion by fibroblasts, endothelial cells, cancer cells, inflammatory cells, and leukocytes in addition to myogenic cells (23, 27, 44, 76, 79) and acts not only to degrade the ECM but also to activate other MMPs including MMP-2 (87) and MMP-9 (82) in trans. When we screened human and murine satellite cells for their expression of MMP-2, MMP-9, MMP-13, and MMP-14 mRNA using primers designed to amplify sequences from both species, we found that while both express MMP-2, only mouse cells express MMP-9 and MMP-13, and only human satellite cells express MMP-14 mRNA (Fig. 3A).

Fig. 3.

Human, but not mouse, satellite cells express MMP-14; MMP-14 expression in human satellite cells is dependent on contact with collagen I. A: RT-PCR for candidate MMP mRNAs. GAPDH was used as a positive control. B and C: cross sections of collagen I matrices on which human (B) or mouse (C) satellite cells were seeded and allowed to invade for 90 h. Sections were stained for MMP-14 (green) and then labeled with phalloidin 594 (red) and DAPI (blue); note the differential MMP-14 expression in human cells based on location adjacent to the collagen matrix. D: cross section of uninjured mouse tibialis anterior (TA) stained for Pax7 (red), MMP-14 (green), and DAPI (blue). E: cross section of mouse TA 5 days after BaCl injury showing high local expression between nascent myofibers in the area of injury. MMP-14 (green), DAPI (blue). F: coculture of ROSA^mTmG mouse satellite cells (red) and unlabeled mouse fibroblasts on a collagen I matrix; all cells were stained for MMP-14 expression (green.) G: human satellite cells were seeded onto gelatin-coated coverslips and analyzed for MMP-14 (green) and integrin-β₁ (red) colocalization.

Because of this exclusive expression of MMP-14 by human cells, we considered it a potential mediator of their invasive phenotype. To confirm protein expression and test for its localization, we stained sectioned matrices as above for MMP-14 by immunohistochemistry. Consistent with the mRNA results, human cells on 3D collagen I show robust expression of MMP-14 protein (Fig. 3B) but murine satellite cells do not (Fig. 3C). We confirmed that the antibody detects murine MMP-14 on uninjured muscle sections (Fig. 3D) costained with antibodies to MMP-14 (green) and the satellite cell marker Pax7 (red), as well as on sections of hindlimb muscle five days after barium chloride-induced injury (Fig. 3E).

We noted Pax7-negative, MMP-14-positive interstitial cells, possibly muscle fibroblasts, in the uninjured sections, raising the possibility that nonmyogenic cells might act as a local source of MMP-14 in regenerating mouse muscle to compensate for the lack of expression by satellite cells. In vitro, primary mouse fibroblasts express MMP-14 protein (64), so we asked whether coculture with fibroblasts would enhance murine satellite cells' ability to invade collagen I. We isolated primary adult mouse muscle fibroblasts (54) from a wild-type mouse and cocultured them with genetically labeled satellite cells isolated from a ROSA^mTmG mouse (55) constitutively expressing membrane-localized tdTomato, to differentiate between the two cell types. The mixed cell population was plated as above on a collagen I matrix, and while the muscle fibroblasts both express MMP-14 and invade, the satellite cells do neither (Fig. 3F). This suggests that primary mouse satellite cells do not invade collagen type I even in the presence of secreted MMPs and cleaved collagen fibers generated by other cell types.

Interestingly, the MMP-14 protein expression in human cells is not homogeneous in that only cells in direct contact with the collagen matrix show significant protein expression. 3D collagen-dependent MMP-14 expression has previously been described in vivo and in vitro in lung fibroblasts, endothelial cells, and metastatic cells (66, 67, 82); in at least one cell type where it was tested this expression is independent of both integrin activity and substrate stiffness (67). Consistent with those data, we found that MMP-14 and β₁-integrin do not colocalize on human satellite cells (Fig. 3G). In addition, neither antibody neutralization of β₁-integrin nor treatment with cyclic arginine-glycine-aspartic acid (RGD) peptides significantly inhibited invasion (data not shown). Thus, our data support an integrin-independent mechanism for invasion of 3D collagen by human satellite cells.

To ensure that the difference we observed between human and mouse satellite cells is not an artifact of the immortalized human satellite cell line, we repeated invasion assays using two different samples of nonclonal, nontransformed human primary myoblasts, as well as two immortalized murine myoblast cell lines. Primary human myoblasts generated at the Institute of Myology (http://www.institut-myologie.org/), one from a 5-yr-5-day-old male (Fig. 4A) the other from a 73-yr-old female (Fig. 4B), both express MMP-14 and invade into the collagen matrix. C2C12 myoblasts, the most commonly used murine myoblast cell line (90), also express MMP-14 and display a remarkable capacity for invasion into collagen (Fig. 4C). MM14 myoblasts (47), which on the basis of morphology, gene expression, and myogenic capacity are the most similar to primary mouse satellite cells, did not invade although they do express low levels of MMP-14 (Fig. 4D). C2C12 cells rapidly proteolyze 3D collagen I, as indicated by cleavage-induced activation of fluorescent collagen I (Fig. 4E and Supplemental Video S1; Supplemental Material for this article is available at the Journal website). These characteristics may correlate with the tendency of C2C12s to form tumors when engrafted into host muscle (61).

Fig. 4.

All human satellite cell populations tested express MMP-14 and invade collagen I, while murine cell lines differ from primary cells and each other. Two primary (nonclonal, nontransformed) human myoblast samples from a 5-yr, 5-day-old male (A) and a 73-yr-old female (B) and two immortalized murine myoblast cell lines, C2C12 (C) and MM14 (D), were seeded on a 3D collagen type I matrix for 90 h and stained for MMP-14 (green) with phalloidin 635 (red) and DAPI (blue). E: images are reference stills from Supplemental Video S1. The panel illustrates collagen proteolysis (green) by C2C12 myoblasts over 4 h. A composite (top) of the FITC (bottom) and brightfield channels shows localized proteolysis respective to the cell. C2C12 satellite cells adhered to the bottom of a 48-well plate were challenged with 3D collagen I copolymerized with DQ Collagen I (1 mg/ml; green.)

MMP-14 is necessary but not sufficient for invasion of collagen I.

To determine whether MMP-14 activity is necessary for human satellite cell invasion, we knocked down MMP-14 expression in human cells using targeted siRNA (Invitrogen). Knockdown was confirmed by immunocytochemistry and Western blotting, and transfection efficiency (98%) was verified through BLOCK-iT Alexa Fluor Red Fluorescent Control Oligo. Transfection with MMP-14-specific (Fig. 5, A and B) but not scrambled (Fig. 5, C and D) siRNA significantly reduced expression of MMP-14 protein as well as human satellite cell invasion through collagen I, and this reduction is dose dependent (Fig. 5, E-G). These data indicate that MMP-14 is necessary for invasion of collagen I by human satellite cells, which could be either direct cleavage of matrix substrates or proteolytic activation of other collagenases, or both. Reduction of MMP-14 activity also reduces the level of active MMP-2 when assayed by gelatin zymography (Fig. 6A), raising the possibility that the requirement for MMP-14 is through MMP-2. Human satellite cells also displayed an elevated amount of TIMP-2 mRNA when compared with primary murine satellite cells (Fig. 6B), and it is important to note that TIMP-2 is required for activation of MMP-2 in conjunction with MMP-14 dimerization (9).

Fig. 5.

MMP-14 is necessary for invasion by human satellite cells. A-D: transfection with MMP-14 siRNA (A and B; B is inset from A), but not scrambled control siRNA (C and D; D is inset from C), reduced invasion by human satellite cells. E: this decrease in invasion was dose dependent. F: knockdown of MMP-14 protein was confirmed by Western blotting; IP90 was used as a loading control. Protein lysates from three separate experiments are shown. G: the ratios of invading cells for control siRNA and MMP-14 siRNA were quantified as described above in 3D collagen culture. **P < 0.005.

Fig. 6.

MMP-2 activity in human satellite cells is correlated with MMP-14 expression. Conditioned medium from human satellite cells transfected with either MMP-14 or scrambled control siRNA was run in triplicate onto a gelatin zymogram. Medium from cells treated with MMP-14 siRNA contained reduced MMP-2 activity (A). On the basis of RT-PCR for TIMP-1, TIMP-2, and TIMP-3, human cells express higher levels of TIMP-2 mRNA compared with primary murine satellite cells (B).

To test whether MMP-14 is sufficient to permit invasion of 3D collagen, we asked whether ectopic expression of MMP-14 in primary murine satellite cells would confer an invasive phenotype. Interestingly, even primary murine satellite cells expressing human MMP-14 did not invade (Fig. 7). Coupled with the lack of invasion by MMP-14-expressing MM14 myoblasts noted above, we conclude that MMP-14 is necessary but not sufficient in this system. It is possible that the decreased levels of TIMP-2 we noted in murine cells may contribute to the lack of invasion.

Fig. 7.

Human MMP-14 is not sufficient to confer an invasive phenotype on primary murine satellite cells. Primary murine satellite cells were transfected with human MMP-14 or empty vector (control) with pCMV-TOMATO (red) as a transfection marker. Phalloidin 633 (white; B) and DAPI (blue; A and B) were used to visualize whole cells seeded onto the matrix.

DISCUSSION

Surprisingly, given the key role played by satellite cell-ECM interactions during muscle homeostasis and repair (48), analysis of the suite of proteins secreted by satellite cells for matrix modification in vivo is not yet well characterized. To date, the only published secretome for murine myoblasts was made using the C2C12 cell line (36), which are not an ideal model for primary myoblasts (31, 46). The secretome from differentiating human myocytes was recently published (43): of the 253 soluble secreted proteins identified by gel-free nano-flow LC-MS/MS analysis, 72 are matrix-modifying enzymes or structural matrix proteins; for comparison, in the same set, there are only 25 secreted growth factors represented. The presence of both constructive and destructive modulators of the ECM suggests a dynamic interplay between the local matrix and satellite cells, which we interpret to be remodeling and not simply proteolysis.

Matrix remodeling by MMPs contributes to adult tissue homeostasis and stem cell-mediated regeneration (3, 16, 21, 25, 38, 85); MMP-14 in particular is required for regeneration in multiple tissues (4, 15, 21, 57, 86). In the context of skeletal muscle, other membrane-type MMPs (MMP-2 and MMP-9 as well as MMP-14) have been implicated in muscle homeostasis (57) as well as in satellite activation after damage (26, 42) and motility in vivo (24). MMP-14 is directly or indirectly required for activation of membrane-type proteases (i.e., MMP-2 and MMP-9) and secreted proteases (i.e., MMP-13) (66, 70, 90) and also activates multiple chemokines, cytokines, and growth factors that would be expected to impinge on muscle regeneration (37, 75). Thus, MMP-14 has the potential to affect muscle regeneration by multiple distinct mechanisms and would not be limited to directly promoting matrix invasion.

It is important to note that these in vitro experiments do not fully represent multiple aspects of muscle regeneration in vivo, and accordingly care should be taken in their interpretation. We did not assess satellite cell activity on or in response to a bona fide muscle ECM in vivo, which is not simply a more complex mix of proteins but is also organized, cross-linked, and polarized in patterns that are difficult to replicate in vitro (48). Attempts to modify the experimental system by using Matrigel (a more complex ECM mixture) instead of collagen I produced the same results (data not shown.) However, our own data suggest that murine satellite cells that have been continually in contact with a native matrix or their host myofiber may transit through 3D collagen (72). This may be a contributing factor to the dramatically enhanced engraftment and spread of satellite cells when single fibers, rather than monoculture satellite cells, are injected into host muscles (34).

Muscle connective tissue fibroblasts, which are a local source of both extracellular matrix components and matrix-modifying enzymes in vivo and are necessary for robust satellite cell-mediated muscle regeneration (54), were included in coculture experiments to ask whether they provided a key component of the in vivo environment. Although they did not enhance myoblast invasion in our experimental system, a role for fibroblast-specific effects on the matrix in vivo is still strongly suggested by the work of other groups. Inflammatory cells including leukocytes and macrophages express MMPs including MMP-14 (7, 56), and pericytes, which may participate in muscle regeneration independently of satellite cells (20), require MMP-14 for key processes during tissue growth and remodeling (78); these cell types were not tested in our in vitro model. Our inability to induce invasion by murine cells with ectopic expression of MMP-14 also does not rule out the possibility that other membrane proteins are necessary in an MMP-14 containing proteolytic complex, and thus MMP-14 alone would not be sufficient.

Our results also serve to reinforce the need for caution in generalizing from animal models to human physiology. It is well established that muscle regeneration differs between mice and humans, particularly in the context of disease (reviewed in Ref. 11) or therapy (5, 12, 28, 59). Potential cell-based therapies are frequently tested either by engrafting murine satellite cells into immunodeficient mice (24, 52, 63) or by doing the same with human cells (62, 73, 74); our data suggest caution both in interpreting the results of these experiments and in generalizing mouse-mouse or human-mouse grafts to human cells engrafted in human patients. One of the confounding difficulties in developing such therapies to date has been failure of engrafted cells to spread away from the point of injection (60); thus a key area of interest is in enhancing migration and invasion. In published studies of simple 2D motility, murine satellite cells are significantly more motile than human (41, 72); however, in the current study murine cells are also significantly less competent to invade a three-dimensional matrix; neither set of results adequately explains the differences observed between species in satellite cell transplantation studies.

We have not tested for species-specific differences in MMP-14 expression and activity in nonmuscle cell types. Because both its matrix remodeling and intracellular signaling activities play key roles in multiple physiological and pathological processes including angiogenesis and tumor metastasis (14, 89), it would be important to know whether any such differences exist in other contexts. If the phenotype is widespread, it may necessitate a reexamination of existing research outside the muscle field as well.

GRANTS

This work was supported by National Institutes of Health/National Institute of Arthritis and Musculoskeletal and Skin Diseases Grant AR-062836 (to DDW Cornelison).

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the author(s).

AUTHOR CONTRIBUTIONS

D.K.L., V.M., D.C. conception and design of research; D.K.L., performed experiments; D.K.L., D.C. analyzed data; D.K.L., V.M., D.C. interpreted results of experiments; D.K.L., D.C. prepared figures; D.K.L., drafted manuscript; D.K.L., V.M., D.C. edited and revised manuscript; D.C. approved final version of manuscript.