We investigated the role of growth factors and fibronectin on matrix metalloproteinase (MMP) expression and on migration and invasion of mouse skeletal myoblasts in vitro. None of the growth factors tested significantly affected MMP-1 or MMP-2 activity as revealed by gelatin zymography, but both basic FGF (bFGF) and tumor necrosis factor (TNF)-α significantly increased MMP-9 activity (10- and 30-fold, respectively). The increase in secreted MMP-9 activity with TNF-α stimulation was due at least in part to an increase in MMP-9 gene transcription, because an MMP-9 promoter construct was approximately fivefold more active in TNF-α-treated myoblasts than in control myoblasts, as well as an increase in MMP-9 proteolytic activation. However, whereas fibronectin, bFGF, hepatocyte growth factor, and TGF-β1 significantly augmented migration of mouse myoblasts, TNF-α did not, nor did PDGF-BB or IGF-I. Fibronectin and bFGF also significantly augmented invasion of myoblasts across a Matrigel barrier, and plasmin cotreatment potentiated whereas N-acetyl cysteine suppressed the effects of bFGF and fibronectin on myoblast migration and invasion. Finally, transient transfection with an MMP-9 overexpression construct had only minimal effects on myoblast migration/invasion, whereas overexpression of either MMP-2 or MMP-1 significantly augmented myoblast migration and invasion. These observations support the hypothesis that MMP activity is a necessary component of growth factor-mediated myoblast migration but suggest that other consequences of growth factor signaling are also necessary for migration to occur.
- skeletal muscle
- muscle fiber
- basic fibroblast growth factor
- matrix metalloproteinase-9 promoter
migration of myoblasts through connective tissue barriers is an important component of many muscle processes. During embryonic development, myogenic precursor cells migrate out of the somites and into the developing limb buds to form the limb musculature (6, 12) and continue to cross the basal lamina during postnatal development to add nuclei to growing myofibers (26). Several studies have also demonstrated migration of myoblasts both within (35) and between (43) adult muscles during muscle regeneration. More recent work has demonstrated the presence of myogenic stem cells residing outside muscle that have the potential to contribute to muscle growth and repair. Pluripotent cells with myogenic potential have now been isolated from several nonmuscle tissues, including bone marrow (17), embryonic dorsal aorta (14), neural tissue (20), and skin (48). When isolated, amplified in culture, and then reimplanted, these cells are capable of contributing to muscle formation in vivo (14, 20). These cells must migrate through the muscle and cross the basal lamina to contribute to myogenesis.
Recent evidence has strongly supported the idea that matrix degradation is a central component to myoblast migration in vivo. Several elegant studies have shown that activity of the matrix metalloproteinase (MMP) family is both necessary and sufficient for improving the efficiency of myoblast transfer in mice (16, 41). These studies are consistent with the hypothesis that greater myoblast dispersal following implantation in vivo is a consequence of increased migration/invasion due to increased MMP activity. However, because of the complex nature of in vivo studies, a direct causal role for MMPs in augmenting myoblast invasion through connective tissue barriers could not be established, and whereas several studies have shown that MMP-2 (16, 22, 31), MMP-9 (31), and MT1-MMP (16) are all expressed by skeletal myoblasts in vitro, little is currently known about the role of various growth factors in augmenting MMP expression or the role of specific MMPs in the ability of myoblasts to invade across connective tissue barriers.
The present study examines the role of growth factors and fibronectin on MMP expression and on mouse myoblast migration and invasion in vitro. We demonstrate that treatment with either basic FGF (bFGF) or tumor necrosis factor (TNF)-α significantly augmented MMP-9 but not MMP-1 or MMP-2 activity and that the increase in MMP-9 expression in response to TNF-α stimulation was due to both increased MMP-9 promoter activity and increased MMP-9 activation. Using an in vitro migration/invasion assay that allowed us to quantify the response of myoblasts to various growth factors and matrix molecules, we found that bFGF and soluble fibronectin strongly stimulated both migration and invasion across a connective tissue barrier in vitro and that overexpression of either MMP-1 or MMP-2, but not MMP-9, stimulated myoblast migration/invasion.
MATERIALS AND METHODS
Mouse myoblasts were previously isolated from hindlimb muscles of 4- to 6-wk-old severe combined immunodeficient (SCID) mice and clonally purified from contaminating fibroblasts as described (46). More than 99% of the cells were myoblasts, as indicated by positive staining with an anti-desmin antibody (data not shown). Myoblasts (∼1 × 106 cells) were plated on 6-cm tissue culture plates coated with 0.5% gelatin (Sigma, St. Louis, MO) and grown in growth medium consisting of Dulbecco's modified Eagle's medium (DMEM; GIBCO BRL, Gaithersburg MD) supplemented with 20% fetal bovine serum (FBS; GIBCO) and 0.5% chick embryo extract (GIBCO). C2C12 mouse muscle myoblasts were obtained from American Type Culture Collection (ATCC; Rockville, MD) and were grown in DMEM supplemented with 20% FBS.
Gelatin zymography for assaying MMP activity was carried out as described by Guerin and Holland (22) with minor modifications. Briefly, myoblasts were grown in 6-cm tissue culture plates to ∼70% confluence, rinsed three times with serum-free DMEM, and incubated for 3 h in DMEM containing 0.2% BSA to eliminate the effects of serum. The following factors, obtained from R&D (Minneapolis, MN) unless otherwise noted, were used: bovine bFGF, recombinant human TNF-α, purified human transforming growth factor (TGF)-β1, recombinant human platelet-derived growth factor (PDGF)-BB, recombinant human insulin-like growth factor (IGF)-I, recombinant hepatocyte growth factor (HGF), and human serum fibronectin (Sigma). The growth factor concentrations used were those that produced maximal effects as determined by dose-response studies (unpublished observations) and are comparable to the concentrations used by Bischoff (5). Growth factors were added in serum-free medium, and cells were incubated for 24 h. Culture medium was collected, centrifuged to pellet detached cells, and concentrated using the Centricon-10 (Amicon-Millipore, Bedford, MA) system. The protein concentration of the supernatants was determined using the Bio-Rad protein microassay system with BSA as the standard. Samples were stored at −70°C until use. Aliquots (10 μg of total protein per sample) were electrophoresed at constant voltage on a 10% polyacrylamide gel containing 2 mg/ml gelatin. The gel was rinsed three times for 15 min in 2.5% Triton X-100 to remove SDS and renature the proteins and then incubated in MMP activation buffer (0.05 M Tris · HCl, pH 7.5 with 5 mM CaCl2) for 24 h at 37°C with constant shaking. Gels were stained overnight in 0.5% Coomassie blue R-250 and destained for 1 h in 40% methanol:10% acetic acid. Proteinase activity was quantified by densitometric scanning of bands using a Gel Doc 1000 video camera imaging system (Bio-Rad, Hercules, CA).
C2C12 myoblasts were used for these studies and were treated with TNF-α exactly as described for the zymography studies. For Western blotting, both supernatant and cells were collected, and 10 μg of protein were run on a 10% acrylamide gel. Proteins were transferred to polyvinylidene difluoride membrane (Amersham Pharmacia Biotech, Piscataway, NJ) for 6 h at 50 V in transfer buffer (25 mM Tris, 100 mM glycine, and 0.04% SDS). The membrane was dried and then rewetted in 100% methanol, rinsed three times with distilled H2O, and incubated in blocking buffer (5% nonfat dry milk, 1% BSA, and 1% Tween 20 in PBS) for 1 h. Membranes were incubated in primary antibody (mouse anti-MMP-9; Oncogene Research Products, Boston, MA) for 1 h, rinsed three times with PBS and 1% Tween 20 (PBS-T), and incubated in goat anti-mouse secondary antibody (Bio-Rad) in PBS-T for 1 h. After several rinses with PBS-T, antibody reactivity was visualized using the Opt-4CN substrate kit (Bio-Rad). Blots were quantified using densitometry on digitally scanned blots.
For promoter studies, we first isolated the upstream promoter region for the mouse MMP-9 gene from mouse genomic DNA using PCR. Primers flanking ∼1,300 bp upstream and 30 bp downstream of the transcription start site were designed on the basis of the published mouse MMP-9 genomic sequence in the NCBI database (accession no. AF403768) with anMluI site on the 5′-end and an XhoI site on the 3′-end. The resulting fragment was PCR purified, digested withMluI and XhoI overnight at 37°C, and gel purified and ligated with the plasmid vector pGL3, which was also cut with the same restriction enzymes and gel purified. Competent DH5α bacteria were transformed with the resulting transfection mixture, and colonies were selected, grown, and screened by restriction digestion. Positive colonies were confirmed by sequencing of the isolated plasmid DNA and contained the appropriate MMP-9 insert.
For transfection studies using the MMP-9 promoter plasmid, mouse C2C12 cells were grown to 70% confluence and then split 1:4 onto 24-well plates coated with 0.5% gelatin. Transfections were carried out with LipofectAMINE 2000 using the manufacturer's instructions. Briefly, 0.8 μg of plasmid DNA and 5 μl of LipofectAMINE 2000 were mixed with 100 μl of serum-free DMEM per well, allowed to form complexes for 20 min at room temp, and then added to the plates in DMEM with 20% FBS and no penicillin/streptomycin and allowed to incubate overnight. As a control, an equivalent number of wells were transfected with pGL3 vector alone. In addition, a plasmid reporter construct containing 1,200 bp of the human fibronectin promoter in the pGL3 vector (kindly provided by Dr. Jesse Roman, Emory University) was also tested. The following day, the medium was removed and replaced with DMEM and PBS vehicle alone (control), 100 ng/ml TNF-α, or 2 ng/ml TGF-β1. After 24 h, the media were removed, cells were rinsed with PBS, and then extracted with passive lysis buffer (Promega). Luciferase activity was assayed using a firefly luciferase assay kit (Promega). All values for the MMP-9 promoter construct were normalized to the pGL3 control for both untreated and growth factor-treated conditions and were expressed as the degree of activation over the control pGL3 vector.
In vitro migration and invasion assays.
Myoblast migration and invasion were examined by using a commercially available in vitro cell migration and invasion assay kit (Biocoat, Becton Dickinson, Franklin Lakes, NJ) as described by Albini et al. (1). This system consists of a lower chamber that contains the putative chemoattractive factor and an upper chamber containing a membrane with 8-μm pores. Cells are added to the upper chamber, the chemoattractant substance is added to the lower chamber, and the number of cells migrating through the pores to the bottom side of the membrane are counted. Myoblasts were grown to ∼70% confluence and rinsed three times in serum-free DMEM, followed by incubation for 3 h in 0.2% BSA in DMEM to eliminate the effects of serum. Cells were then trypsinized and collected by centrifugation. Cells were resuspended in serum-free DMEM at a density of 1 × 105 cells/ml, 0.5-ml aliquots of cell suspension were added to the top chamber, and growth factors were added to the lower chamber. For invasion studies, the upper membrane was coated with 5 μl of Matrigel diluted to 5 mg/ml in sterile PBS. At the end of the migration/invasion period, the top side of the insert membrane was scrubbed free of cells with a cotton swab and the bottom side was stained using the Leukostat-I system (Fisher Diagnostics, Pittsburgh, PA). The number of cells per field was counted in 10 randomly selected fields and averaged for each condition.
In some studies purified human plasmin (0.05 units in 50 μl of PBS) or 50 mM N-acetyl cysteine (NAC; Sigma) was added with the cells to the top chamber of the migration assay and growth factor was added to the bottom. NAC is a thiol-containing antioxidant with multiple cytoprotective effects (15) that is known to inhibit activity of MMPs, particularly gelatinases MMP-2 and MMP-9 (2), possibly through altering redox states (10). NAC has been extensively used as an inhibitor of gelatinase activity in studies of cell migration and invasion (2,19, 34).
Construction of MMP overexpression vectors.
PBS-92 and PBS-GEL, containing the human MMP-9 and MMP-2 cDNAs, respectively, were kindly provided by Dr. Gregory Goldberg (Washington University School of Medicine), and pcD-X containing the MMP-1 gene was obtained from ATCC. Expression vectors containing human MMP genes were generated with plasmid pNGVL3, which contains the cytomegalovirus immediate-early enhancer, 5′-untranslated region and intron, the rabbit β-globin poly(A) signal sequence, and a kanamycin resistance marker. This plasmid vector was obtained from the Vector Center of the University of Michigan. The MMP-9 coding cDNA insert was excised from the vector PBS-92 with XbaI and ligated into pNGVL3 at the XbaI site with T4 DNA ligase (Boehringer Mannheim, Indianapolis, IN), generating pNGVL3/MMP-9. PNGVL3/MMP-2 was prepared by removing the MMP-2 cDNA from the PBS-GEL plasmid vector byNotI/EcoRI digestion and ligating into pNGVL3 at the NotI/EcoRI sites. Expression vector pNGVL3/MMP-1 was prepared by inserting the MMP-1 cDNA isolated from pcD-X into pNGVL3 at the SalI site. Competent bacteria (Top10; Invitrogen, La Jolla, CA) were transformed, and kanamycin-resistant colonies were selected for all three constructs. All constructs were examined by restriction mapping to confirm the correct structures and orientations.
Transient rather than stable transfection was used because secretion of overexpressed MMPs by transfected cells should be sufficient to allow most, if not all, cells access to increased levels of secreted MMPs. Myoblasts grown in growth medium to ∼50% confluence in six-well plates were transfected overnight by adding growth medium containing 1 μg of expression vector DNA and 3 μl of FuGene 6 reagent according to the manufacturer's instructions (Boehringer Mannheim). Under similar conditions using pCH110 vector DNA (β-galactosidase expression plasmid), ∼20–25% of mouse myoblasts could consistently be transfected (data not shown). For cotransfection with both MMP-1 and MMP-2 vectors, a total of 2 μg of vector DNAs composed of 1 μg of each expression vectors were mixed with 6 μl of FuGene 6 for transfection. The following morning, the transfection mixture was removed and cells were harvested for cell migration/invasion assays as described above, except that 10 μg/ml fibronectin (10% of the regular concentration) was added to the bottom chamber to prime cell migration and invasion.
All migration and invasion data are reported as mean numbers of cells per field counted for 10 fields per experiment; n = 3–5 experiments. Densitometric data for MMP-2 and MMP-9 are reported as the mean degree of change compared with control (BSA treated); n = 3–5 experiments. To determine whether there were significant differences among the various growth factors tested, we used analysis of variance with Fisher's least-square difference post hoc test. A P value of <0.05 was taken as the limit of significance.
MMP expression of mouse myoblasts.
Mouse myoblasts grown in serum-free medium constitutively expressed MMP-2 (zymogen form, 72 kDa), and none of the growth factors tested had a significant effect on MMP-2 activity (Fig.1). Treatment of mouse myoblasts with fibronectin, bFGF, PDGF-BB, TGF-β, or IGF-I also had no significant effect on MMP-9 expression (Fig. 1), whereas TNF-α and bFGF significantly increased MMP-9 activity (∼100-kDa band) to ∼30- and 10-fold over the DMEM control, respectively (Fig. 1, A,C, and D). MMP-1 (57 kDa) was not detected with or without growth factor treatment using gelatin zymography (Fig.1 A), although this does not exclude the possibility of low-level induction that may have been below the limit of detection of this assay.
Treatment of myoblasts with soluble plasma fibronectin induced the proteolytic conversion of MMP-2 to two forms migrating as a doublet at 64 and 62 kDa, respectively (Fig. 1 A, right). When these lower bands were included in the densitometric total, fibronectin treatment resulted in a significant increase in MMP-2 activity, but if only the 72-kDa band was included, fibronectin had no significant effect (Fig. 1 B). This proteolytic conversion was specific for soluble fibronectin, because cells grown on a fibronectin-coated substrate showed only constitutive expression of MMP-2 with no apparent proteolytic activation (Fig. 1 A,right).
Mechanisms of increased MMP-9 activity with TNF-α treatment.
To determine whether increased MMP-9 gene transcription contributed to the increase in MMP-9 gelatinolytic activity observed in the zymograms, we isolated the upstream promoter region of the mouse MMP-9 gene. Analysis of MMP-9 promoter activity using a luciferase reporter construct revealed a significant four- to fivefold increase in MMP-9 promoter activity in TNF-α-treated myoblasts compared with control (Fig. 2 A). In contrast, treatment with TGF-β1, which had no effect on MMP-9 activity in zymogram assays (Fig. 1), also had no significant effect on MMP-9 promoter activity (Fig. 2 A). The promoter for the human fibronectin gene was more active in C2C12myoblasts than was the MMP-9 promoter, but in contrast to MMP-9, its activity was significantly decreased by TNF-α treatment (Fig.2 B). Moreover, fibronectin promoter activity was significantly increased by TGF-β1 treatment of myoblasts (Fig.2 B).
Western blotting of C2C12 myoblast supernatant revealed a corresponding five- to sixfold increase in MMP-9 protein levels between untreated and TNF-α -treated myoblasts (Fig. 2,C and D). Western blotting of C2C12 myoblasts themselves revealed only minimal immunoreactivity, suggesting that all MMP-9 produced by these cells is rapidly secreted (Fig. 2, C and D). Thus the increase in MMP-9 promoter activity in response to TNF-α treatment was translated directly into an equivalent increase in secreted MMP-9 protein levels with no change in intracellular accumulation of MMP-9. However, TNF-α treatment resulted in a 30-fold increase in MMP-9 gelatinolytic activity (Fig. 1), and thus the 5-fold increase in MMP-9 secreted protein levels cannot entirely account for this increase in activity. We therefore examined in more detail the pattern of gelatin zymography in response to TNF-α treatment. When 25 μg of secreted protein were run, there was some MMP-9 activity observed in untreated myoblasts (Fig. 2 C,bottom). Moreover, two bands could be observed in untreated myoblasts, a slower migrating band and a faster migrating band that presumably represent the less active form and the proteolytically activated form, respectively (Fig. 2 C, bottom). The primary difference between untreated and TNF-α-treated myoblast supernatant was an increase in the amount of the lower band, combined with the disappearance of the upper band, suggesting more efficient proteolytic activation of MMP-9 with TNF-α treatment. Together, these data suggest that TNF-α treatment results in an increase in MMP-9 proteolytic activation as well as an increase in MMP-9 gene transcription.
Mouse myoblast migration in vitro.
Migrations in response to bFGF, fibronectin, HGF, and TGF-β1 were all significantly greater than the BSA control, whereas TNF-α, IGF-I, and PDGF-BB did not significantly stimulate myoblast migration compared with control (P < 0.05; Fig.3 A). The largest individual effects were seen in response to fibronectin and bFGF stimulation (14- and 12-fold increase over the BSA control, respectively), whereas HGF and TGF-β1 had smaller but still significant effects on myoblast migration (8- and 5-fold, respectively). The combination of bFGF and fibronectin had an additive effect, stimulating migration >27-fold over the BSA control (Fig. 3 A).
We investigated the effects of plasmin or NAC on MMP activity and myoblast migration/invasion. Treatment of myoblasts with plasmin, which has been previously demonstrated to induce proteolytic activation of MMPs (36), resulted in a consistent increase in MMP-2 activity but not MMP-9 activity in gelatin zymograms (Fig.4 A). Cotreatment of myoblasts with plasmin also increased the migrational response to bFGF nearly twofold, whereas treatment with plasmin alone without a chemotactic growth factor in the lower chamber had a slightly negative effect on myoblast migration (Fig. 3 B). Treatment with NAC almost completely abolished MMP gelatinolytic activity as assessed by zymography (Fig. 4 B), and NAC cotreatment attenuated the effects of bFGF on mouse myoblast migration to a level not significantly different from the BSA control (Fig. 3 B). Moreover, addition of NAC resulted in a dramatic reduction of the migration in response to a combination of bFGF and plasmin, further suggesting that the stimulatory effect of plasmin is upstream from that of MMP activity (Fig. 3 B).
Mouse myoblast invasion in vitro.
Examination of the role of the same set of growth factors on invasion across a Matrigel barrier revealed that only fibronectin and bFGF had significant effects, increasing invasion by approximately four- and sevenfold, respectively (Fig.5 A). Invasion in response to bFGF or fibronectin was significantly greater than the BSA control and all other growth factors tested (P < 0.05). The combination of bFGF and fibronectin gave greater than eightfold invasion activity over the BSA control but was not significantly different from bFGF or fibronectin alone (Fig. 5 A). As observed for migration, plasmin further increased the effects of bFGF, and NAC treatment reduced invasion to a level not significantly different from control (Fig. 5 B).
The relationship between myoblast invasion and migration is shown in Fig. 6 A. There was a strong correlation between whether or not a particular substance stimulated migration and invasion (r 2 = 0.771,P < 0.05). In addition, we examined the relationship between MMP-2 and MMP-9 expression and myoblast migration. There was no correlation between either MMP-2 activity and myoblast migration (r 2 = 0.391, P > 0.05) or MMP-9 activity and myoblast migration (r 2 = 0.056, P > 0.05) in response to the various growth factors examined. These data suggest that growth factor stimulation induces myoblast migration and invasion through mechanisms independent of increased MMP activity or, alternatively, that MMP activity, while a necessary component, is not a sufficient predictor alone of myoblast migrational capability.
MMP overexpression and mouse myoblast migration and invasion.
Transient overexpression of human MMP-1, MMP-2, or MMP-9 was tested in myoblasts to determine whether overexpression of individual MMPs was sufficient to produce increased migration and/or invasion. Successful transfection of MMP-1, MMP-2 and MMP-9 was confirmed by gelatin zymography, showing increased intensity of bands of ∼57, 72 and 92 kDa, respectively (Fig. 7 A). Cells transfected with the MMP-1 vector showed induction of MMP-1 activity from nondetectable levels in control, whereas activity of MMP-2 and MMP-9 was significantly increased by >3.5- and 10-fold in transfected cells over control (Fig. 7 A). Overexpression of each MMP did not significantly affect expression of the other two MMPs.
Transfection of mouse myoblasts with MMP-1 or MMP-2 increased the migration of mouse myoblasts in response to fibronectin by 2.6- and 1.6-fold, respectively, but only migration in response to MMP-1 transfection was statistically significant because of high variance in the MMP-2 group (Fig. 7 B). Myoblast invasion was significantly increased approximately twofold for both MMP-1 and MMP-2 (Fig. 7 C). In contrast, transfection with MMP-9 had only marginal effects on both migration and invasion (Fig. 7, Band C). NAC treatment significantly decreased the migration capability of both MMP-1- and MMP-2-overexpressing myoblasts to 35 and 22% that of non-NAC-treated cells, respectively (Fig. 7 B), and decreased invasion to 40 and 28% of the nontreated cells, respectively (Fig. 7 C), such that it was not significantly different from control. These results suggest that overexpression of MMP-2 and MMP-1, but not MMP-9, can facilitate myoblast migration and invasion in vitro.
The ability of myoblasts to migrate through and invade across connective tissue barriers is central to many biological processes, including muscle development, regeneration, stem cell replenishment of muscle, and myoblast-mediated gene transfer. In the present study we evaluated the role of growth factor signaling in the secretion of MMPs and in migration and invasion of adult skeletal myoblasts in vitro. Our studies show that bFGF, soluble fibronectin or overexpression of MMP-1 or MMP-2 is capable of inducing increased myoblast migration and invasion in vitro. However, our results do not support a direct causal connection among growth factor stimulation, MMP activity, and myoblast migration/invasion, as described below. Indeed, the present work supports a fairly complex model of increased myoblast migration and invasion in response to growth factor treatment in which MMP activity is a necessary component but in which other, as yet undefined mechanisms also play critical roles.
To date there have been no data evaluating the effects of different growth factors on MMP activity or expression in skeletal myoblasts. In the present study, we sought to determine the effects of growth factor stimulation on myoblast MMP expression in vitro. None of the growth factors tested had an effect on MMP-2 activity (Fig. 1, Aand B). However, both TNF-α and bFGF significantly increased MMP-9 gelatinolytic activity (Fig. 1, A andC). We therefore tested the effects of TNF-α on MMP-9 promoter activity to determine whether this increase was a result of increased MMP-9 gene transcription, as has been demonstrated previously for other cell types (8, 21, 32). TNF-α treatment resulted in a fivefold increase in MMP-9 promoter activity but significantly decreased activity of the human fibronectin promoter (Fig. 2, A and B). In contrast, treatment with TGF-β1 did not significantly affect MMP-9 promoter activity but increased human fibronectin promoter activity approximately threefold (Fig. 2, A and B). Together these data suggest that the effect of TNF-α on MMP-9 is specific and reflects neither promiscuous responsiveness of the MMP-9 promoter to all growth factors nor a general effect of TNF-α on transcription. Indeed, these results are consistent with the hypothesis that TNF-α treatment of skeletal myoblasts shifts the gene expression profile toward one favoring increased turnover of the basal lamina, whereas TGF-β1 treatment results in a gene expression pattern favoring increased fibrosis and matrix deposition (42).
The increase in MMP-9 promoter activity was accompanied by a corresponding five- to sixfold increase in secreted MMP-9 protein levels as demonstrated by Western blotting, whereas MMP-9 immunoreactivity was not observed within myoblasts (Fig.2 C). This finding suggests that the increase in MMP-9 gene transcription is accompanied by a commensurate increase in MMP-9 protein levels and that changes in MMP-9 secretion are not a contributing factor to the increase in MMP-9 activity. However, MMP-9 gelatinolytic activity increased >30-fold (Fig. 1), suggesting that other factors contributed to the TNF-α-induced increase in MMP-9 activity. We observed a much greater increase in the activated, faster migrating form of MMP-9 in zymograms of TNF-α-treated myoblasts (Fig.2 C). Consistent with this observation, a recent study (24) also demonstrated an increase in both MMP-9 promoter activity and MMP-9 proteolytic activation in response to TNF-α treatment. Thus the present data suggest that TNF-α has multiple effects on MMP-9 expression and activity in skeletal myoblasts in vitro.
It is also noteworthy that soluble fibronectin, but not substrate-bound fibronectin, induced proteolytic activation of MMP-2 in myoblasts (Fig.1). Fibronectin regulation of migration, invasion, and MMP expression has been demonstrated for other cell types and usually involves stimulation of proteinase expression by substrate-bound fibronectin. Werb et al. (45) reported that plating of rabbit synovial fibroblasts on fragments of fibronectin that interact with the α5β1-integrin induced collagenase (MMP-1) expression, whereas fragments that interact with α4β1-integrin suppressed MMP-1 expression. Attachment to intact fibronectin, which contains both domains, had no effect on MMP-1 expression (27, 45). Because proliferating myoblasts express α5β1- (23) but not α4β1-integrin (38), fibronectin should have inductive effects on MMP-1, but we observed no MMP-1 expression on fibronectin or any other substrate using gelatin zymography (Fig. 1). The lack of an increase in MMP-1 expression in response to fibronectin attachment observed in the present studies suggests the existence of cell type-specific differences between fibroblasts and myoblasts in fibronectin-mediated regulation of MMP expression.
MMPs have been strongly implicated in the process of tumor cell migration and invasion during metastasis (40) and in myoblast migration during development (11). Our results support a role for the MMP family of proteinases in myoblast migration and invasion. Cotreatment of myoblasts with NAC, an antioxidant previously shown to inhibit MMP activity (2), greatly attenuated the migration/invasion ability of myoblasts (Fig. 2). As demonstrated in Fig. 4, NAC treatment successfully inhibited MMP gelatinolytic activity as assayed by zymography. Conversely, treatment with plasmin increased MMP-2 activation (Fig. 4) and also potentiated myoblast migration and invasion (Figs. 3 and 5). The NAC and plasmin studies are consistent with the hypothesis that MMP activity is an essential component of myoblast migration/invasion, though it is possible that other effects of NAC may also have contributed to this effect. The overexpression studies on MMP-1 or MMP-2 also support a role for MMPs in myoblast migration and invasion (Fig. 5), demonstrating that overexpression of either MMP is sufficient to potentiate growth factor-induced myoblast migration and invasion.
However, it is important to note that the present data do not support a direct causal connection between growth factor stimulation and MMP expression with respect to migration/invasion. None of the growth factors tested had an appreciable effect on MMP-1 or MMP-2 activity (Fig. 1), although bFGF, fibronectin, HGF, and TGF-β1 all significantly stimulated myoblast migration and/or invasion (Fig. 3). Conversely, unstimulated myoblasts constitutively express MMP-2 (Fig.4) but undergo minimal migration and invasion (Fig. 1). Finally, TNF-α stimulation resulted in a significant 30-fold increase in MMP-9 expression but did not significantly induce migration or invasion. In summary, there was no relationship between either MMP-2 or MMP-9 activity and migration or invasion in response to the growth factors tested (Fig. 6, B and C). Our results are consistent with the hypothesis that MMP activity is an essential component of fibronectin- or bFGF-mediated myoblast migration and invasion but suggest that fibronectin and bFGF have effects independent of MMP activation that are also necessary for these processes to occur.
The mechanism(s) by which these growth factors and fibronectin stimulate myoblast migration and invasion are currently not defined and were not investigated in the present study. However, some likely candidates include other components of the cell adhesion, matrix degradation, and locomotion pathways, such as cytoskeletal rearrangements, changes in integrin expression and localization, shifts in the expression of metalloproteinase inhibitors, particularly the tissue inhibitors of metalloproteinase (TIMPs), and increased expression of other secreted proteinases such as plasmin. Previous work has established that bFGF stimulates urokinase-plasminogen activator expression (33), which proteolytically activates plasminogen to plasmin (7) and therefore is critical to the initiation of the MMP activation cascade. Consistent with this role is the fact that treatment with plasmin additively increased both myoblast migration and invasion (Fig. 2). Future studies are needed to address which pathway is critical for growth factor-mediated migration and invasion.
PDGF, TGF-β1, and HGF have been individually demonstrated to have chemotactic effects on myoblasts in vitro (3, 5, 30, 44), but to date only Bischoff (5) has directly compared the activity of different growth factors on myoblast migration. We observed that HGF and TGF-β1 had a significant effect on myoblast migration, but the strongest effects on mouse myoblast migration occurred in response to bFGF treatment, which is at odds with the results of Bischoff (5). However, our observations agree with previous work demonstrating that bFGF is a chemotactic agent for various other cell types (4, 9, 25). A role for bFGF in myoblast migration and limb muscle formation in vivo has also recently been demonstrated in developing chicks (18). In addition, treatment of skeletal myoblasts with bFGF substantially increased myoblast-mediated gene transfer in mice (28, 47), although the strong mitogenic and antiapoptotic properties of bFGF likely also contribute to this effect.
To date, there have been no studies examining the ability of myogenic cells to invade across a connective tissue barrier in vitro, although there have been several studies examining the ability of implanted myoblasts to disperse within and incorporate into the injected muscle in vivo (16, 41). In one of the only studies to quantify the invasive ability of myoblasts, Corti et al. (13) examined the ability of different chemotactic growth factors to allow myoblasts to invade through an endothelial cell monolayer in vitro. These authors found that HGF and PDGF had the strongest effects on stimulating this process. However, the ability of myogenic cells to squeeze past endothelial cells, while of obvious importance to their ability to access skeletal muscle from the vasculature, is in all likelihood very different from the ability of myoblasts to invade across a connective tissue layer. In the present study we examined the ability of mouse myoblasts to cross a Matrigel barrier in response to a growth factor gradient. We chose Matrigel because the composition of Matrigel (type IV collagen, laminin, fibronectin) closely matches that of the muscle fiber basal lamina in vivo. Our data suggest that bFGF and fibronectin can stimulate myoblasts to invade across a connective tissue layer in vitro.
Our data suggest that MMP-9 plays only a minimal role in myoblast migration and invasion. TNF-α, which induced MMP-9 expression >30-fold, had negligible effects on myoblast migration and invasion. Similarly, overexpression of MMP-9 had minimal effects on myoblast migration or invasion. However, the fact that MMP-9 activity is strongly induced by either TNF-α or bFGF suggests that it does play some as yet undefined role in skeletal muscle biology. Increased MMP-9 activity has been demonstrated in adult skeletal muscle during muscle injury and in muscle myopathy and is upregulated following downhill running (29, 37, 39), consistent with a possible role for this matrix-degrading enzyme in muscle regeneration and repair.
We thank Sumiko Kurachi for generous assistance in experiments and Dr. Doug Treco and Dr. Mark Caddle at Transkaryotic Therapy, Inc. (TKT), Cambridge, MA, for support and encouragement and for critical reading of the manuscript.
This work was supported by funding from TKT, National Heart, Lung, and Blood Institute Grant HL-53713, the University of Michigan Multipurpose Arthritis and Musculoskeletal Disease Center (NIH-5P60-AR-20557), and the General Clinical Research Center (M01-RR-00042). D. L. Allen was supported by a National Research Service Award through the University of Michigan Cardiovascular Research Training Grant.
Present address of D. L. Allen: Dept. of Molecular, Cellular, and Developmental Biology, University of Colorado, Boulder, CO 80309-0347.
Address for reprint requests and other correspondence: K. Kurachi, Dept. of Human Genetics, Univ. of Michigan Medical School, C570A MSRB II, Ann Arbor, MI 48109-0672 (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 December 4, 2002;10.1152/ajpcell.00215.2002
- Copyright © 2003 the American Physiological Society