Quantitative Changes in Integrin and Focal Adhesion Signaling Regulate Myoblast Cell Cycle Withdrawal

Primary myoblasts were isolated from pectoralis muscle of nine day Japanese quail embryos as previously described (Konigsberg, 1979). In brief, the breast muscle was dissected from the embryo and myoblasts were dissociated from muscle tissue with 0.1% dispase (Sigma Chemical Co.) in PBS. The cell suspension was filtered through a Sweeney filter; cells were seeded onto gelatin-coated tissue culture plates (0.1% gelatin in PBS). Myoblast cultures were maintained in complete myoblast medium (DMEM [Sigma Chemical Co.] containing 15% horse serum, 5% chick embryo extract, 1% pen/strep, and 1.25 mg/ml fungizone [GIBCO BRL]). Myoblasts were subcultured in trypsin-EDTA (0.06% trypsin, 0.02% EDTA) and used between passages 1 and 10.

The muscle α-actinin–specific mAb, 9A2B8, was kindly provided by D. Fishman (Cornell University, New York, NY) as a hybridoma supernatant. mAb VIF4, which recognizes the human α5 integrin extracellular domain was a gift of R. Isberg (Tufts University, Boston, MA). The chicken α6-specific polyclonal antibody, α6ex (de Curtis et al., 1991), was provided by L. Reichardt (University of California, San Francisco, CA). mAb 2B7 directed against the extracellular domain of the human α6 integrin (Shaw et al., 1993), was a gift of A. Mercurio (Harvard Medical School, Boston, MA). The mAb 165 is directed against paxillin (Turner et al., 1990). mAb 2A7 directed against FAK (Kanner et al., 1990) and the polyclonal Ab BC3 directed against FAK (Shaller et al., 1992) were gifts of T. Parsons (University of Virginia, Charlottesville, VA). The anti-FAK polyclonal antibody C-20, was purchased from Santa Cruz Biotechnology, Inc. (Santa Cruz, CA). The anti-active MAPK polyclonal antibody, which specifically recognizes dually phosphorylated, activated MAPK, was purchased from Promega. The anti-MAPK mAb that recognizes erk-1, as well as the anti-phosphotyrosine antibodies, RC20H and PY20, were purchased from Transduction Laboratories. The anti-hemagglutinin (HA) mAb, 12CA5, was purchased from Boehringer Mannheim. The anti-human CD2 mAb, TS2/18.1.1 was purchased from the Developmental Studies Hybridoma Bank. TS2/16, a mAb against human β1 integrin (Hemler et al., 1984) was from M. Hemler (Dana-Farber Cancer Institute, Boston, MA). Antibody against the human IL2 receptor was purchased from Boehringer Mannheim. Poly-l-lysine was purchased from Sigma Chemical Co. Fibronectin was purified from human plasma by affinity chromatography as previously described (Ruoslahti et al., 1982). Laminin was isolated from murine Englebreth-Holm-Swarm sarcoma as previously described (Kleinman et al., 1982).

The human α5 cDNA in pRSVneo and the chicken α6 cDNA in pRSVneo were described previously (Sastry et al., 1996). The chicken α61044t truncation was constructed by first subcloning a 1.6-kb HindIII-SalI fragment of the chicken α6A DNA into M13 and then introducing an in-frame BclI site at amino acid position 1044 (de Curtis et al, 1991). Mutants were confirmed by restriction digestion of M13 clones with BclI and by single stranded DNA sequencing using the dideoxy-chain termination method according to the Sequenase™ protocol (United States Biochemical Corp.). An 800-bp BstXI-SalI fragment containing the mutation was subcloned into pRSVneoα6 partially digested with SalI and completely with BstXI. The human α6A and α6B cDNAs, in the expression plasmid pRc/ CMV (Shaw et al., 1993), were a generous gift of A. Mercurio (Harvard Medical School, Boston, MA). The pRSVneo-CH8β1 plasmid was constructed by subcloning a 1-kb HindIII fragment, containing the CH8 epitope tag, from the CH8β1 pBJ-1 construct received from Y. Takada (Scripps Research Institute, La Jolla, CA) (Takada and Puzon, 1993) into pRSVneoβ1 expression vector (Reszka et al., 1992). pRSVIL2R-α5 and pRSVIL2R-β1A were constructed by cloning an Nhe1-Xba1 fragment from pCMVIL2R-α5cyto or pCMVIL2R-β1A plasmids received from Susan LaFlamme (Albany Medical College, Albany, NY; Tahiliani et al., 1997) into the Xba1 site of the pRSVneo vector. Clones were screened for orientation by restriction digests. HA-tagged rat MEK1 and HA-tagged rat constitutively active (CA) MEK S218/220D in pCMVneo vector i.e., sodium-deoxycholate, sodium-pyrophosphate, sodium-orthorandate (Catling et al., 1995) were received from M. Weber (University of Virginia, Charlottesville, VA). The chicken paxillin cDNA, the Y118F, and the S188/ 190A mutants in the pcDNA3.0neo vector (Brown et al., 1997) were received from C. Turner. CD2FAK, CD2FAK(Y397F), and CD2FAK(K454R) in CDM8 vector (Chan et al., 1994) were received from A. Aruffo.

Cells were transiently or stably transfected using a liposome-DNA solution as previously described previously (Sastry et al., 1996). In brief, replicating myoblasts (passages 1–3) were plated on 60-mm tissue culture plates coated with 0.1% gelatin in complete myoblast medium for 16–20 h. Cells were incubated for 8–16 h in a solution of 8 μg of plasmid DNA and 50 μg of Lipofectamine (GIBCO BRL) in complete myoblast medium. Transfected myoblasts were either washed with DMEM, refed with myoblast medium, and analyzed for transient expression or were trypsinized and plated into selection medium on gelatin-coated tissue culture plates (myoblast medium containing 0.4 mg/ml G418; GIBCO BRL) for 7–12 d and then into myoblast medium containing 0.2 mg/ml G418 (maintenance medium). For transfections with CD2FAK constructs, in order to generate stable populations, myoblasts were cotransfected with a pRSVneo or a pEGFP-C1neo plasmid (Clontech, Palo Alto, CA) at 1:7 ratio (neo resistance gene:CD2FAKcDNA) and selected in G418 as previously described. The chicken α6, human α6A or α6B, and α61044t transfections were selected and maintained on laminin-coated (20 μg/ml) tissue culture plates. For coexpression of hα6A integrin and CD2-FAK cells were cotransfected with pRc/CMVhα6A and CDM8CD2FAK vectors at a ratio of 1:7, respectively. Transiently transfected cells were sorted by flow cytometry (see below) for hα6A expression and the positive cells were grown in G418 containing medium. CDM8CD2FAK vector does not carry a neo resistance gene, therefore, only cells carrying both neo resistance (hα6A positive) and able to replicate (CD2FAK positive) will survive. Stable populations were analyzed both for hα6 and CD2 expression as described below.

Both transiently and stably transfected (the α5 phenotype was seen in transient as well as stable transfectants) myoblasts were analyzed for surface expression by flow cytometry as previously described (Sastry et al., 1996). Chicken α6A and α61044t transfected cells were stained with a chick α6-specific polyclonal antibody, α6ex, at 20 μg/ml in blocking buffer (Hepes-Hanks PBS-CMF with 2% BSA) and FITC-labeled goat anti–rabbit IgG (Cappel). Human α6A or B transfected myoblasts were analyzed with the human α6-specific mAb, 2B7, at 10 μg/ml in blocking buffer. Human α5 transfected cells were stained with VIF4 mAb. Cells transfected with CD2FAK or its mutants were stained with anti-CD2 mAb TS2/ 18.1.1. Chicken β1 transfected cells were stained with TS2/16 mAb against the human β1 epitope. IL2R-α5 or IL2R-β1A transfected cells were analyzed with an anti-IL2R antibody. The FACS profiles of the IL2R-α5 and IL2R-cyto-transfected cells were not stable, and we were unable to obtain populations greater than 40% positive, which were used for analysis. Flow cytometry was performed on a EPICS cell sorter (Coulter Electronics, Inc.) equipped with Cicero software for data analysis. As shown in Fig. 1, expression levels were assayed by fluorescence activated cell sorting. These profiles reflect enriched surface expression levels of ectopic integrin subunits in transfected myoblasts used in experiments.

For Western blotting and immunoprecipitation experiments, untransfected and transfected myoblasts were plated on FN (UT and hα5 transfected cells), on LM (UT and chicken α6 and chicken α61044t or human α6A transfected cells) or on gelatin (UT, PAX, and CD2FAK transfected cells) for 24 h in complete myoblast medium. Since myoblasts will differentiate in the absence of serum and also secrete their own matrix, we were unable to test the effects of specific matrix ligands on the myoblast response. Therefore, all assays were conducted in the presence of serum and results are presented for steady state conditions. Cells were washed with ice-cold PBS containing 1 mM Na-orthovanadate and lysed in ice-cold modified RIPA extraction buffer (20 mM Tris-HCl, pH 7.4, 150 mM NaCl, 0.5% NP-40, 1.0% Triton X-100, 0.25% sodium-deoxycholate, 2 mM EDTA, and 2 mM EGTA) with protease inhibitors (20 mg/ml leupeptin, 0.7 mg/ml pepstatin, 1 mM phenanthroline, 2 mM phenyl-methyl-sulfonyl-chloride, and 0.05 units aprotinin) and phosphatase inhibitors (30 mM sodium-pyrophosphate, 40 mM NaF, 1 mM sodium-orthovanadate). Protein content of the clarified lysates was determined using the Pierce bicinchoninic acid (BCA) method with bovine serum albumin as the standard.

For phosphotyrosine Western blots, 10–15 μg of lysates were separated on 10% SDS-PAGE gels (Laemmli, 1970) under reducing conditions and transferred to nitrocellulose membranes (Towbin et al., 1979). Membranes were blocked in 1% heat denatured BSA in TST buffer (10 mM Tris-HCl, pH 7.5, 100 mM NaCl, 0.1% Tween-20) overnight at 4°C. Phosphotyrosine containing proteins were detected by incubating the membranes with the anti-phosphotyrosine mAb, PY20, and a secondary horse radish peroxidase (HRP) conjugated anti–mouse antibody (Jackson ImmunoResearch Labs) or with RC20H, a directly conjugated HRP anti-phosphotyrosine Ab. Blots were visualized by chemiluminescence (Pierce Chemical Co.). Membranes were exposed to X-ray film (Kodak, X-OMAT AR) and developed in an automatic film processor. When indicated, membranes were stripped in stripping buffer (62.5 mM Tris-HCl, pH 6.8, 2% SDS, and 100 mM β-mercaptoethanol) for 30 min at 60°C and reprobed with a different antibody.

For anti-MAPK Western blots, cells were trypsinized, washed once with soybean trypsin inhibitor (0.5 mg/ml), washed twice in Puck's Saline G (GIBCO BRL) and resuspended in serum-free medium containing 2% BSA. Cells were held in suspension for 1 h prior to plating on FN or LM in complete myoblast medium for 24 h. Cell extracts were prepared in RIPA buffer as described. 5 μg of cell lysates were separated on 12% SDS-PAGE gels under reducing conditions and the proteins transferred to nitrocellulose membranes. The membranes were blocked in 3% nonfat dry milk in TST overnight at 4°C. Active MAPK was detected by an anti-active MAPK pAb (Promega). Membranes were stripped and reprobed for total MAPK with an anti-erk1 mAb (Transduction Labs) or SC-94 anti-erk1 pAb (Santa Cruz Biotechnologies).

For paxillin, FAK, CD2-FAK, and HA immunoblot analysis, 5–20 μg cell lysates were resolved on 7.5% SDS-PAGE gels under reducing conditions and proteins transferred to nitrocellulose membranes. Membranes were blocked in TST buffer containing 3% nonfat milk and the proteins were detected with 165 mAb (anti-paxillin), BC3 pAb (anti-FAK), TS2/ 18.1.1 mAb (anti-CD2), or 12CA5 mAb (anti-HA).

For FAK immunoprecipitations, 100 μg of RIPA lysate was mixed with 1 μl of anti-FAK mAb, 2A7, 50 μl of packed agarose anti–mouse beads (blocked in 5% BSA; Sigma) in a final volume of 500 μl. The bead-antibody-antigen complex was incubated at 4°C for 2 h with continuous agitation. For paxillin immunoprecipitations, 100 μg of cell lysate and 1 μl of anti-paxillin mAb, 165 were incubated at 4°C with continuous agitation for 1 h. In a separate tube, 50 μl of packed protein A–agarose beads and 30 μg/ml rabbit anti–mouse IgG were incubated in lysis buffer for 1 h. The antigen-antibody mixture was then added to rabbit anti–mouse–protein A beads and incubated at 4°C an additional 2 h. The beads were pelleted gently and washed twice with lysis buffer. Bound protein was released from the beads by boiling in 100 μl Laemmli sample buffer containing 5% β-mercaptoethanol for 5 min. Equal aliquots of the precipitated protein for each antibody were loaded onto 7% SDS-PAGE gels. The FAK IP was blotted for FAK with C-20 and for phosphoFAK with RC20H. The paxillin IP was blotted for paxillin with the 165 mAb or for phosphopaxillin with RC20H. All immunoprecipitations and Western blots were detected by chemiluminescence.

Cells were grown on FN- or LM-coated coverslips. Immunostaining was done at room temperature. Cells were rinsed in PBS and fixed with 3% formaldehyde in PBS for 15 min then permeabilized with 0.4% Triton X-100 in PBS for 10 min, washed and blocked in 5% goat serum in PBS (BB) for 30 min. Cells were incubated with primary Ab in BB for 30 min, washed and incubated with FITC- or rhodamine-conjugated secondary Ab (Cappel) and DAPI (Sigma Chemical Co.) for additional 30 min. Coverslips were washed extensively and mounted in medium containing elvanol and p-phenylenediamine. Fluorescence was observed on a Zeiss Axioplan microscope.

MAP kinase activity was manipulated in hα6A transfected myoblasts by coexpression of constitutively activate (CA) MEK1. Myoblasts were cotransfected with pRc/CMVhα6A and the HA-tagged pCMVneoMEK S218/220D vectors at a ratio of 1:7, respectively. Cells were selected in G418 and stable populations were sorted by flow cytometry for human α6A expression as described. Cell lysates were analyzed for HA expression by Western blotting as described. Stably cotransfected cells were plated on LM-coated plates and observed for 96 h.

To alter MAP kinase activity in hα5 transfected myoblasts, hα5-expressing cells were grown in the presence of the specific MEK inhibitor PD98059 (New England Biolabs) (Alessi et al., 1995). Transfected myoblasts were plated on FN-coated coverslips and on FN-coated TC plates. After 8 h in complete myoblast medium, the first dose of the inhibitor was added to the cells at 1, 10, 25, 50, or 100 μM final concentration. Cells were grown for an additional 24 h and a second dose of the inhibitor was added. After 24 and 48 h in presence of the inhibitor, coverslips were fixed and immunostained for DAPI and muscle α-actinin. At the same time, cells were extracted in RIPA buffer as described and lysates were analyzed by Western blotting for active MAPK and total erk1 expression as described above.

Differentiation was scored using the fusion index, which is the percentage of total nuclei in myotubes as described in Sastry et al. (1996).

We recently reported a specificity for integrin α subunits and their cytoplasmic domains in controlling the proliferative to differentiative transition in primary quail myoblasts (Sastry et al., 1996). Ectopic expression of the human α5 integrin subunit (hα5) enhanced the fraction of myoblasts remaining in the proliferative phase and inhibited the initiation of terminal differentiation. In contrast, ectopic expression of the human α6A subunit of integrin (hα6A) inhibited myoblast proliferation and promoted differentiation. These effects resulted from a three- to fivefold increased surface expression of the α5β1 or the α6Aβ1 integrin (and a two- to threefold increase in total β1 integrin, see below) with little change in the relative expression of other integrin α subunits. These findings suggested that the α5 cytoplasmic domain promotes proliferative signals whereas the α6A cytoplasmic domain inhibits proliferation and enhances the fraction of cells initiating terminal differentiation.

To assess the contribution of these two cytoplasmic domains, we first examined the effect of ectopic α5 and α6A truncation mutants on myoblast proliferation and differentiation. (Fig. 2 and Table I). As we reported previously, ectopic expression of the hα5 truncation, α5GFFKR, which retains only the conserved GFFKR sequence, promoted proliferation and inhibited differentiation similar to the wild-type hα5 subunit (Sastry et al., 1996; Table I). These findings suggest that the majority of the α5 cytoplasmic domain is not required for proliferative signals. On the other hand, ectopic expression of an α6A truncation, α61044t, which deletes the COOH-terminal 11 amino acid residues, restores proliferative signaling and produces a phenotype similar to that of the ectopic α5 subunit. Myoblasts expressing α61044t remain in the proliferative phase and do not differentiate even in high density cultures (Fig. 2). Like hα5 transfected myoblasts (Fig. 2 A), myoblasts expressing α61044t do not express muscle α-actinin (Fig. 2), a myogenic differentiation marker, and exhibit a fusion index of 5% (Fig. 2 B) after 72 h of culture in a rich medium. This contrasts UT controls and hα6A transfected cells (Fig. 2) where a significant fraction of cells express muscle α-actinin and fuse into multinucleated myotubes (Fig. 2 B). Preliminary mapping of the COOH-terminal 11 amino acids points to S1071 (in hα6A) as a key residue, since its mutation to alanine produces a phenotype with enhanced proliferation (data not shown). Furthermore, the proliferation inhibiting effect of hα6 integrin is specific for the α6A cytoplasmic domain isoform. Ectopic expression of the hα6B subunit in myoblasts promotes proliferation and inhibits differentiation (data not shown, Table I). Consistent with these results, the α6A isoform is the predominant α6 integrin expressed in striated muscle (Hogervorst et al., 1993; and our own unpublished observations) and in embryonic cells of determined lineage (Cooper et al., 1991) whereas α6B is highly expressed in proliferating, totipotent or undifferentiated ES cells. Taken together, these observations suggest that the α5 and α6A cytoplasmic domains function differently: the α5 cytoplasmic domains appears permissive whereas a discrete region of the α6A cytoplasmic domain is inhibitory with respect to proliferation.

To determine whether the α5 or α6A cytoplasmic domains act directly or indirectly, we assayed the effects of single-subunit cytoplasmic domain chimeras (LaFlamme et al., 1992), IL2R-α5 or IL2R-α6A, on the ability of myoblasts to proliferate or differentiate. Ectopic expression of either IL2R-α5 or IL2R-α6A had little detectable effect on myoblast proliferation or differentiation (Fig. 2 B). These cells behaved much like control, untransfected (UT) myoblasts. Thus the α subunit cytoplasmic domains do not directly initiate signals for myoblast proliferation or differentiation.

How then do these two integrin α subunits regulate proliferation and differentiation? Our observation that different integrins, e.g., α5, α6B, and α61044t, all produce a similar phenotype suggests an hypothesis in which these α subunits influence the proliferative signaling through the β1 subunit. In this view the α5 cytoplasmic domain (as well as that of the α6B, α61044t, and perhaps others) would permit signaling through the β1 subunit whereas the α6A cytoplasmic domain would inhibit it. Previous studies have shown that the α subunit cytoplasmic domain can regulate β1 integrin localization to focal adhesions (Briesewitz et al., 1993; Ylanne et al., 1993) and integrin activation (O'Toole et al., 1991); both localization and activation, however, are mediated by the β subunit cytoplasmic domain. Furthermore, the β1 cytoplasmic domain alone, when expressed as a single subunit chimera, IL2R-β1, can activate intracellular signals (Akiyama et al., 1994). To test this hypothesis, we first determined whether overexpression of the chicken β1A subunit of integrin would increase the fraction of proliferative myoblasts. We chose the β1A isoform since it is predominant in replicating myoblasts (Belkin et al., 1996). As reported in Sastry et al. (1996) ectopic expression of integrin α subunits also produces a two- to threefold increase in total β1 expression with little change in relative expression of the other endogenous α subunit levels. The increase in total β1 expression maintains myoblasts in the proliferative phase and inhibits terminal differentiation. Myoblasts with enhanced β1A expression grow to confluency but exhibit a fusion index of only ∼10% compared with 60–70% for untransfected cells (Fig. 2 E). Thus, increased β1A expression produces a phenotype resembling that of increased hα5 subunit expression. This finding agrees with a similar result reported previously for the β3 integrin (Blaschuk et al., 1997).

Next, we addressed whether the β1A cytoplasmic domain could independently affect myoblast proliferation or differentiation through ectopic expression of the single subunit chimera, IL2R-β1A. Previous studies show that this chimera localizes in focal adhesions (LaFlamme et al., 1992) and mediates enhanced integrin signaling (Akiyama et al., 1994). Myoblasts expressing IL2R-β1A remain replicative and proliferate until confluent with little detectable fusion into myotubes (Fig. 2 B). Like myoblasts expressing ectopic α5 subunit, they also only exhibit significant cell cycle withdrawal and differentiation if cultured under serum-free conditions (Sastry et al., 1996 and data not shown). These results demonstrate that the β1A cytoplasmic domain is sufficient to transmit proliferative signals and inhibit differentiation and thus modulate the growth factor response. Further, ectopic expression of IL2R-β1A can rescue the hα6A phenotype. Myoblasts that coexpress IL2R-β1A and hα6A integrins proliferate and differentiate like untransfected cells (data not shown; Table I). Taken together, these findings suggest that proliferative signaling through integrins occurs via the β1 subunit and that different α subunit cytoplasmic domains can modulate these signals. The effects of ectopic integrin subunits on myoblast proliferation and differentiation are summarized in Table I.

We next sought to determine the effect of ectopic α5 or α6A expression on β1A integrin signaling pathways. Since integrins stimulate increased tyrosine phosphorylation of several intracellular proteins (Burridge et al., 1992; Kornberg et al., 1992; Bockholdt and Burridge, 1993; Petch et al., 1995; Vuori and Ruoslahti, 1995), we assayed the phosphotyrosine profile of myoblasts expressing different ectopic α subunits. Immunoblotting with an anti-phosphotyrosine antibody shows that myoblasts ectopically expressing the hα5 integrin contain elevated tyrosine phosphorylation of proteins migrating in the molecular mass ranges of 120–130 and 65–70 kD whereas cells transfected with hα6A show a marked, general decrease in tyrosine phosphorylation with no additional bands when compared with UT controls (data not shown). These observations are consistent with the phenotypic effects of the ectopic integrins presented above; i.e., the α subunits do not initiate separate pathways. Thus, ectopic α5 expression permits enhanced signaling through the β1A subunit, whereas the α6A integrin suppresses these signals.

Next, we pursued the identities of the phosphoproteins migrating at 120–130 and 65–70 kD. Focal adhesion kinase (pp125FAK) and paxillin (pp68) are downstream targets of integrin signaling pathways that migrate in these molecular mass ranges (Burridge et al, 1992). Therefore, we assayed the level of FAK and paxillin tyrosine phosphorylation in UT, hα5, and hα6A transfected myoblasts. A Western blot of a FAK immunoprecipitation shows that comparable levels of FAK were precipitated (Fig. 3 A, lanes 1–3). The level of tyrosine phosphorylated FAK decreased in hα6A (Fig. 3 A, lane 6) versus hα5 transfected (Fig. 3 A, lane 5) or UT myoblasts (Fig. 3 A, lane 4). However, the level of FAK tyrosine phosphorylation in hα5 and UT myoblasts does not differ (Fig. 3 A, lanes 4 and 5). Immunodepletion of FAK from the lysate, followed by Western blotting the supernatant for phosphotyrosine reveals an additional 120-kD band in the hα5 transfected cells that could account for the observed increase in phosphotyrosine (data not shown). Thus, whereas FAK phosphorylation decreases in myoblasts expressing ectopic α6A integrin, it is unaffected by ectopic α5 expression.

To determine if paxillin phosphorylation is differentially regulated in UT, hα5, and hα6A transfected myoblasts, paxillin was immunoprecipitated with an anti-paxillin mAb. In contrast to observations with FAK, we observed a major difference in the level of paxillin expression between UT and hα5 transfected myoblasts. As seen in Fig. 3 B, paxillin is significantly upregulated in hα5 transfected myoblasts (Fig. 3 B, lane 2) when compared to UT (lane 1) or hα6A transfected (lane 3) myoblasts. The comparable intensity of a 55-kD band in the paxillin immunoprecipitation corresponds to reduced IgG and serves as a loading control. In addition to elevated levels of paxillin, a phosphotyrosine Western blot of the paxillin immunoprecipitation shows a concomitant increase in tyrosine phosphorylation of paxillin in hα5 transfected myoblasts (Fig. 3 B, lane 5) compared to UT myoblasts (lane 4). Tyrosine phosphorylation of paxillin in hα6A transfected cells (lane 6) is somewhat decreased relative to untransfected cells. The enhanced paxillin expression observed in hα5 transfected myoblasts does not arise as a direct effect of the hα5 integrin. Myoblasts expressing IL2R-β1A also show increased paxillin expression whereas myoblasts expressing IL2R-α5 do not (data not shown). Taken together, these results indicate that enhanced paxillin expression accompanies increased α5 (or β1A) levels, whereas decreased FAK phosphorylation coincides with increased α6A levels. These data also suggest an uncoupling of FAK and paxillin signaling.

The altered expression and activation of paxillin and FAK presented above prompted us to examine the effects of ectopic paxillin or FAK expression on myoblast proliferation and differentiation. As shown in Fig. 4 A (lane 2), the level of paxillin expression increases when compared with controls (Fig. 4 A, lane 1) after transfection of a paxillin cDNA. Ectopic expression of wild-type paxillin inhibits differentiation and results in a proliferative phenotype (Fig. 5, Table II). Myoblasts expressing ectopic paxillin proliferate until confluent but neither fuse into multinucleated myotubes (Fig. 5) nor express muscle α-actinin (Fig. 5). This phenotype is similar to that of hα5 transfected myoblasts (Fig. 2). Thus ectopic paxillin expression alone can recapitulate the effects of the hα5 or IL2R-β1A integrin subunits. Paxillin expression levels in control cells do not differ in replicating myoblasts versus differentiated cultures (data not shown).

Two major sites of phosphorylation in paxillin in response to adhesion to fibronectin are Y118 and S188/190 (Bellis et al., 1997). The tyrosine phosphorylation site, Y118, is also the site phosphorylated by FAK (Bellis et al., 1995). Therefore, we next tested the effect of a Y118F mutation (Fig. 5) or the double mutation S188/190A on the myoblast phenotype. When expressed in myoblasts (Fig. 4 A, lanes 3 and 4), neither of these mutants showed the enhanced proliferation seen when wild-type paxillin was expressed. Instead the paxillin mutants (Fig. 5) exhibited a phenotype characteristic of UT myoblasts (Fig. 5, Table II). This result suggests that these tyrosine and serine phosphorylation sites in paxillin participate in proliferative signaling that regulates myoblast cell cycle withdrawal.

Since FAK phosphorylation decreased in parallel with the inhibition of proliferation in myoblasts expressing hα6A integrin, we next tested the effect of an ectopic FAK mutant, Y397F, which lacks the autophosphorylation site and cannot bind src-family kinases (Schaller and Parsons, 1994), on proliferation and differentiation of myoblasts. Since ectopic expression of soluble FAK often produces short-lived or weak phenotypes (Richardson and Parsons, 1996), we used CD2-FAK, a membrane bound, chimeric FAK construct, which is constitutively active (Chan et al., 1994). Presumably, this results from increased adhesive signaling that arises from its constitutive membrane association and consequent localization in focal adhesions. Ectopic expression of CD2-FAKY397F, inhibits myoblast proliferation while promoting differentiation (Fig. 5). These cells are reminiscent of myoblasts transfected with the hα6A integrin subunit except that their proliferation is not inhibited as completely. They also show extensive fusion into multinucleated myotubes (Fig. 5). Interestingly, myoblasts transfected with wild-type CD2-FAK (Fig. 4 B), remain proliferative and do not initiate terminal differentiation (Fig. 5) when compared to UT controls. Fewer than 5% of CD2-FAK–expressing myoblasts fuse into multinucleated myotubes (Fig. 5) or express muscle α-actinin (Fig. 5). Using FACS analysis of propidium iodide labeled cells to measure G1/S progression, CD2-FAK transfected myoblasts show an increased ratio of G2 to G1 cells compared to untransfected cells (data not shown). Similarly, ectopic expression of CD2-FAK454, which is kinase defective, also inhibits differentiation and promotes proliferation (Fig. 5). Ectopic expression of wild-type CD2-FAK in hα6A transfected myoblasts results in a proliferative phenotype. Myoblasts that coexpress CD2-FAK and hα6A integrin grow to confluency and do not differentiate (Fig. 5). In sum, the hα6A phenotype can be recapitulated by expression of a FAK mutant expressed in control myoblasts or rescued by coexpression of an activated form of FAK. These data suggest that one potential mechanism by which hα6A integrin inhibits myoblast proliferation is through altering FAK phosphorylation. Thus, changes in the level of focal adhesion signaling, through FAK or paxillin, significantly affect the likelihood of myoblasts to proliferate or withdraw from the cell cycle and differentiate. Table II summarizes the effects of ectopic focal adhesion molecules.

Several reports implicate MAP kinase in adhesion dependent regulation of proliferation (Chen et al., 1994; Zhu and Assoian, 1995; Wary et al., 1996; Miyamoto et al., 1996; Lin et al., 1997). In addition, MAP kinase activation plays an important role in muscle differentiation (Bennet and Tonks, 1997). Therefore, we investigated if the ectopic integrins altered MAP kinase activation to control myoblast proliferation and differentiation. MAP kinase activation was assessed by immunoblotting cell lysates with an antibody that specifically recognizes phosphorylated, or active, forms of the 42- and 44-kD MAP kinases. Quail myoblasts express the 44-kD MAP kinase, erk-1, which was detected using an anti–erk-1 mAb (data not shown). Fig. 6 shows Western blots of active MAP kinase in myoblasts expressing ectopic integrins. In all cases, control or transfected cells were cultured for 24 h in complete serum-containing medium before extraction. The results shown reflect differences in steady state levels of MAP kinase activity. Compared with UT myoblasts (Fig. 6, bottom, lane 1), hα5 (lane 2), or IL2R-β1A (lane 3) transfected myoblasts contain elevated levels of active MAP kinase. In contrast, the level of active MAP kinase in hα6A transfected cells is significantly decreased (Fig. 6, top, lane 3) compared with controls (top, lane 1). The different intensities for control levels of active MAP kinase (top versus bottom) is due to different exposure times of the Western blots to film. As presented earlier, the α6A cytoplasmic domain truncation, α61044t, did not produce the proliferation inhibiting effects of hα6A integrin, and instead promoted proliferation and inhibited cell cycle withdrawal. This also altered the level of MAP kinase activation, as assayed by Western blotting. Compared with myoblasts expressing hα6A, those expressing α61044t display enhanced levels of active MAP kinase (Fig. 6, top, lane 2). Stripping and reprobing these membranes for total erk1 levels showed similar expression in all cells tested (data not shown). Therefore, the level of MAP kinase activity depends on the expression of specific integrins and their cytoplasmic domains.

We next investigated whether altering the activation state of MAP kinase could reverse the hα5 or hα6A induced phenotypes. We manipulated the level of active MAP kinase through overexpression of MEK-1, an upstream activator of MAP kinase, or by addition of PD-98059, a specific MEK inhibitor (Alessi et al., 1995). Cotransfection of myoblasts with constitutively active MEK (CA-MEK; Catling et al., 1995) and hα6A integrin restores a proliferative phenotype to myoblasts expressing hα6A. These cells stably express both the hα6A subunit (Fig. 7 B) and CA-MEK (Fig. 7 A) after drug selection and continue to proliferate for the lifetime of the cells in culture. FACS analysis of propidium iodide labeled cells shows an increased ratio of G2 to G1 cells in the hα6A/ CA-MEK cotransfectants (data not shown). As reported previously (Sastry et al., 1996), we were unable to isolate cells stably overexpressing only the α6A integrin. The hα6A-CA-MEK transfected cells are similar to the hα5 transfected myoblasts; i.e., they remain proliferative and do not differentiate appreciably (Fig. 7 E) compared with hα6A transfected (Fig. 7 D) or UT cells (Fig. 7 C). The level of active MAP kinase is enhanced in hα6A-CA-MEK cells (Fig. 6, top, lane 4) when compared to hα6A (Fig. 6, top, lane 3) or UT (Fig. 6, top, lane 1) myoblasts. Interestingly, we were unable to obtain stable expression of CA-MEK in untransfected, control myoblasts. Presumably, excessive levels of activated MEK, or MAP kinase, leads to increased cell death or decreased cell growth.

We next determined whether decreasing the level of active MAP kinase, using the specific MEK inhibitor PD-98059, would reverse the hα5 phenotype. hα5 transfected myoblasts were plated onto FN-coated plates in serum-containing medium and allowed to attach for 8–12 h. Increasing concentrations of the MEK inhibitor, PD-98059 were then added for an additional 24–48 h. With increasing inhibitor concentration, the fraction of differentiated cells increased, whereas at high inhibitor concentrations, the total number of cells decreased, presumably due to inhibited proliferation (Fig. 8 and data not shown). After 48 h, hα5 transfected myoblasts treated with a 25 μM or greater concentrations (Fig. 8 C) of the MEK inhibitor displayed marked differentiation into myotubes compared with untreated hα5 transfected cells (Fig. 8 B) resembling UT controls (Fig. 8 A). A Western blot for the level of active MAP kinase shows that increasing concentrations of PD-98059 reduces MAP kinase activity in hα5 transfected cells (Fig. 8 E). Taken together, our observations demonstrate that quantitative changes in integrins closely parallels changes in MAP kinase activation. Moreover, the level of active MAP kinase appears to be a critical determinant of myoblast proliferation versus differentiation.

In this study we addressed the mechanisms by which integrin α subunits modulate intracellular signal transduction events that lead to phenotypic changes in cell proliferation and differentiation. Skeletal myoblasts are well suited for this kind of study. The probability that a myoblast withdraws from the cell cycle and initiates terminal differentiation is highly sensitive to environmental cues including growth factors and ECM components (Sastry et al., 1996). The sharply contrasting effects of ectopic α5 or α6A integrin on myoblast proliferation and differentiation therefore provide a useful system to address α subunit specific signaling and the mechanisms by which integrins regulate the myoblast decision to withdraw from the cell cycle and initiate terminal differentiation. The data presented here demonstrate that the ectopic α subunits differentially regulate proliferative signaling through the β1 subunit. They also demonstrate that paxillin, FAK, and MAP kinase serve as important regulators of cell cycle withdrawal and the onset of terminal differentiation. Finally, our results show that the decision to withdraw from the cell cycle and initiate terminal differentiation is highly poised and regulated by quantitative changes in the level of MAP kinase activation, which in turn is regulated by quantitative changes in levels of integrin signaling. These observations provide a rationale for: (a) the apparently disparate requirements of different muscle cell lines and primary cultures for proliferation and cell cycle withdrawal, (b) the modulation of integrin subunits during muscle differentiation (Menko and Boettiger, 1987; Muschler and Horwitz, 1991; Bronner-Fraser et al., 1992; Lakonishok et al., 1992; Blaschuk and Holland, 1994; Boettiger et al., 1995) and (c) the hyperproliferative phenotypes observed when integrins are modulated in some other cell types (Carroll et al., 1995).

How does the α5β1 integrin potentiate signaling while the α6β1 integrin attenuates it? Our data show that the β1A subunit (i.e., cytoplasmic domain) is sufficient to transmit proliferative signals. The α5 subunit is permissive and allows β1 signaling while the α6A inhibits it. Previous studies indicate that signaling through integrins requires receptor ligation or clustering into focal adhesions (Kornberg et al., 1991; Miyamoto et al., 1996). In the myoblast system, both the ectopic α5 and α6A subunits form functional receptors with the endogenous β1 subunit for FN or LM, respectively (data not shown, and Sastry et al., 1996). Therefore, the differences in signaling we observe are most likely not due to an inability to bind ligand. Furthermore, since IL2R-β1A alone stimulates proliferation, receptor-ligation does not appear necessary except perhaps to form focal adhesions. Our observation used cells cultured on a complex ECM containing serum and secreted FN, also indicating that receptor ligation may not be a critical factor, for the inhibitory effects of ectopic α6. Interestingly, the α5β1 and the α6β1 integrins exhibit distinct subcellular distributions in muscle when cultured on the appropriate matrix ligand (Sastry, S., J. Muschler, and M. Lakonishok, unpublished observations). The α5β1 integrin localizes in focal adhesions on a FN substrate. In contrast, the α6β1 integrin displays a diffuse cell surface distribution on a laminin substrate. Since the initiating event for signaling through integrins is receptor clustering, the contrasting localizations of α5β1 and α6β1 could reflect a difference in their ability to signal. The clustering of integrins recruits signaling proteins into the focal adhesion signaling complex (Miyamoto et al., 1995). In this view, the α5β1 integrin (or any other β1 integrin with a permissive α subunit) recruits and activates proteins like FAK, paxillin, and MAP kinase to stimulate signaling. The α6β1 integrin, though able to bind to LM, is unable to recruit and/or activate the requisite signaling complex.

In the context of a model in which integrin signaling occurs primarily through the β1 subunit, how might the α subunits regulate signaling? A possible mechanism is that the α5 and α6A cytoplasmic domains differentially regulate the accessibility of binding sites on the β1 cytoplasmic domain. The α5 subunit (and likely several other α subunits) would permit exposure of critical binding sites, whereas the α6A subunit would mask them. A similar mode of regulation is proposed for the α subunit cytoplasmic domain in the ligand-dependent localization of integrins in focal adhesions (Briesewitz et al., 1993; Ylanne et al., 1993). Other models of regulation are also possible. They include a steric masking of the β1 cytoplasmic domain by the α6 cytoplasmic domain through interaction with an inhibitory binding protein, or through an α6-mediated signaling event that results in the activation of a phosphatase. The observation that IL2R-β1A expression rescues the hα6A phenotype, suggests that α6A masks some critical site on β1A cytoplasmic domain. Our observation that ectopic IL2R-α6A does not inhibit proliferation argues against an α6β1 binding protein but does not rule it out.

The critical regulatory site(s) within the α6 cytoplasmic domain appears to reside in the eleven COOH-terminal amino acids of the α6A sequence (Table I). This COOH-terminal region of the α6A cytoplasmic domain contains two putative serine phosphorylation sites (Shaw et al., 1990; Delwel et al., 1993), which could potentially participate in negative regulation of signaling by the α6A integrin. Preliminary mapping of the α6A cytoplasmic domain implicates one of these phosphorylation sites in α6A (S1071) in inhibition of proliferation (unpublished observations). However, the physiological relevance of α6 phosphorylation in this system is not yet clear.

Our observations also identify important roles for intracellular components of integrin signaling pathways in regulating cell cycle withdrawal and the onset of terminal differentiation. FAK and paxillin, stand out as potential integrin proximal mediators of adhesive signaling. These proteins are phosphorylated on tyrosine in response to cell adhesion to ECM in many cell types (Guan et al., 1991; Burridge et al., 1992; Hanks 1992; Kornberg et al., 1992; Schaller et al., 1992; Schwartz et al., 1995). In addition, both FAK and paxillin bind to synthetic peptides derived from the β1A cytoplasmic domain (Schaller et al., 1995). However, their role in cell cycle withdrawal and the onset of terminal differentiation is not well understood. Some evidence implicates FAK in proliferation and cell survival. Displacement of endogenous FAK from focal adhesions in endothelial cells, through microinjection of the focal adhesion targeting domain, interferes with cell cycle progression (Gilmore and Romer, 1996) resulting in apoptosis. Similarly, microinjection of β1A peptides corresponding to the FAK binding site, or anti-FAK antibodies induces apoptosis in cultured fibroblasts (Hungerford et al., 1996). Expression of constitutively active FAK, CD2-FAK, in epithelial cells protects them from apoptosis (Frisch et al., 1996).

Our data support a requirement for FAK activation in basal myoblast proliferation; but it does not contribute to the enhanced proliferation, i.e., the inhibited cell cycle withdrawal, observed in cells expressing ectopic α5. Phosphorylation of FAK on tyrosine is not significantly altered in myoblasts expressing ectopic α5 integrin. In most cell types, FAK phosphorylation peaks within 1 h after cell attachment to the ECM (Guan et al., 1991; Burridge et al., 1992; Hanks et al., 1992; Schaller et al., 1992). A transient increase in FAK phosphorylation is observed in α5 transfected myoblasts after initial attachment on a FN substrate (data not shown). However, we do not observe a sustained increase in FAK activation that coincides with the sustained proliferative phenotype induced by ectopic α5 integrin. Additionally, overexpression of FAK, as well as several FAK mutants and FRNK (Richardson and Parsons, 1996), had no detectable effect on myoblast proliferation or differentiation (unpublished observations). However, ectopic expression of CD2-FAK, a membrane-bound, activated form of FAK, does promote proliferation and inhibit differentiation, thus resembling the effect of ectopic α5 or β1A integrin. Presumably this arises because CD2-FAK is targeted to the membrane and therefore available to recruit additional adhesive signaling complexes. In contrast, the level of FAK phosphorylation is reduced in myoblasts expressing ectopic α6A integrin, correlating with the increased cell cycle withdrawal. Ectopic expression of CD2-FAKY397F, which lacks an autophosphorylation site, acts as a dominant negative and inhibits proliferation and promotes differentiation, thus revealing its role in basal proliferation. This phenotype is similar to but less dramatic than that of ectopic α6A integrin expression. It is interesting to note that Y397 is a binding site for src family kinases (Schaller and Parsons, 1994). Finally, coexpression of CD2-FAK and hα6A results in proliferation. Taken together, these observations are consistent with a role for FAK phosphorylation in the basal level of proliferation seen in control myoblasts and show that it lies on the pathway downstream of the α6A-mediated inhibition of the β1 integrin signaling.

In contrast to FAK, both the expression and subsequent tyrosine phosphorylation of paxillin are significantly enhanced in α5 transfected myoblasts when compared to untransfected or α6A transfected cells. A role for paxillin in proliferation or differentiation has not been demonstrated previously. Paxillin is a multidomain adapter protein that mediates numerous interactions with different signaling proteins, including FAK (Schaller and Parsons, 1994). However, paxillin has no known enzymatic activity. Paxillin does contain multiple LIM domains, which are involved in protein-protein interactions and targeting of paxillin to focal adhesions (Turner and Miller, 1994; Brown et al., 1996). Finally, paxillin has been implicated in growth factor dependent differentiation where an increase in its tyrosine phosphorylation correlates with neuronal differentiation (Khan et al., 1995; Leventhal and Feldman, 1996). In myoblasts, the upregulation of paxillin in response to ectopic α5 integrin coincides with a proliferative phenotype, which can be recapitulated by ectopic expression of paxillin. Our results with paxillin mutants implicate both the Y118F and S188/190A phosphorylation sites as key. Whereas the mechanism by which paxillin inhibits cell cycle withdrawal, i.e., stimulates proliferation, is not known, increased paxillin expression and phosphorylation enhances the formation of adhesive signaling complexes as do CD2-FAK and ectopic α5 expression (unpublished observation). In contrast to the significant effects of ectopic α5 expression on paxillin expression and phosphorylation, ectopic α6A expression had only a very modest effect on paxillin phosphorylation and none on expression. Coexpression of paxillin and hα6A did not relieve the α6A phenotype (data not shown). Thus the level of paxillin expression and phosphorylation appear to contribute the myoblast decision to proliferate or differentiate. However, it may not play a role in the inhibition of proliferation by ectopic α6A expression. It is noteworthy that our results also indicate an uncoupling of FAK and paxillin signaling, since they are not activated in parallel as reported for fibroblasts (Burridge et al., 1992).

Although our results demonstrate an inhibitory role for the α6A integrin in signaling, this integrin is capable of activating intracellular signals in other cell systems. In macrophages, for example, α6Aβ1-mediated attachment to laminin leads to increased tyrosine phosphorylation of paxillin (Shaw et al., 1995). In addition, antibody clustering of the α6 integrin in endothelial cells leads to a distinct profile of tyrosine phosphorylated proteins, which do not include FAK or paxillin (Jewell et al., 1995).

The contrasting effects of ectopic α5 or α6A integrin expression on MAP kinase activation suggests quantitative changes in MAP kinase activation plays a major role in regulating myoblast cell cycle withdrawal. Perturbing the ratio of different integrins in the cell alters the level of MAP kinase activation. In contrast to our findings, the α6A integrin, but not the α6B subunit, activates MAP kinase in macrophages (Wei et al., 1998). Cell type differences likely explain these conflicting observations. In the muscle system, the effect of ectopic α5 or α6A integrins and their mutants on MAP kinase activation parallels their phenotypic effects on myoblast proliferation and differentiation. Likewise, perturbing the activation state of MAP kinase, using a MEK inhibitor or CA-MEK, can overcome the integrin induced effect on both MAP kinase activity and the myoblast phenotype. Consistent with these results, the activation of MAP kinase is apparently an important determinant of myogenic differentiation. Inactivation of MAP kinase by overexpression of MAP kinase phosphatase renders C2C12 myoblasts insensitive to mitogenic stimuli, favoring expression of muscle-specific genes like myoD (Bennet and Tonks, 1997). Finally, our data show that MAP kinase activation in myoblasts requires both adhesive and growth factor signals. We do not observe active MAP kinase in the absence of serum or on a non-specific adhesive substrate like poly-l-lysine (unpublished observations). Taken together, these findings reveal a role for integrins in controlling myoblast proliferation and differentiation through a pathway that regulates the level of MAP kinase activation.

The ability of different integrins to activate MAP kinase is proposed to occur via an association of integrins with the adaptor protein, shc. In fibroblasts or endothelial cells, α5β1 and several other integrins interact with shc and thus can activate MAP kinase whereas the α6 integrin cannot bind shc or activate MAP kinase (Wary et al., 1996). However, the interaction with shc does not require integrin cytoplasmic domains. Our data do not preclude an involvement of shc; the increase in MAP kinase activation seen in response to the α61044t truncation may reflect a difference in the binding of shc, or an unidentified adapter or regulatory protein.

Our observations can be summarized in a working model for adhesive regulation of myoblast withdrawal from the cell cycle and the onset of terminal differentiation. In this model, the myoblast decision to proliferate or withdraw from the cell cycle is regulated, at least in part, by the level of activated MAP kinase. This decision appears highly poised and sensitive to quantitative fluctuations in the level MAP kinase activation: increased MAP kinase activation favoring proliferation and decreased MAP kinase activation favoring cell cycle withdrawal. MAP kinase activation, in turn, is regulated by the signaling emanating from both adhesive and growth factor pathways. Evidently, neither pathway alone is sufficient to sustain proliferation. The absence of either growth factor or adhesive signaling promotes cell cycle withdrawal and the onset of terminal differentiation, which appears to function as a default. At one level, the synergy is compatible with an additive model since increased signaling via either enhanced growth factor or adhesive signaling leads to a decreased probability of cell cycle withdrawal. However, the synergy between the growth factor and adhesive signaling systems likely has interactions that are more consistent with an additive, threshold model, in which signals are required from both the growth factor and adhesive pathways. MAP kinase is not activated and targetted to the nucleus in cells that are either adhering to a nonspecific adhesive substrate like poly-l-lysine or that lack serum growth factors. Recent studies point to RAF as an integration point for adhesive and growth factor signals (Chen et al., 1996; Renshaw et al., 1997; Howe et al., 1998).

This working model provides a rationale for diverse observations on the myoblast decision to proliferate or differentiate. Infection of myoblasts with viruses encoding src family kinases (Alemo and Tato, 1987) or erbB (Olson, 1992), a truncated form of the EGF receptor with intrinsic tyrosine kinase activity, or the ectopic expression of CD2-FAK or paxillin all inhibit cell cycle withdrawal and produce a proliferative phenotype similar to that of myoblasts ectopically expressing the α5 integrin. It is likely that this arises from increased adhesive and/or growth factor signaling resulting in increased MAP kinase activation. Similarly, the general requirement for high confluency and low serum for differentiation of diverse muscle cell lines likely also appears to reflect enhanced mitogenic signaling, i.e., increased MAP kinase activation, which in this case results from immortalization or adaptation to culture. Previous reports show alterations in the expression of integrin subunits during muscle differentiation, our results raises the possibility that they contribute to cell cycle withdrawal and the onset of terminal differentiation. Finally, our studies appear pertinent to keratinocyte proliferation in vivo as well as to myoblast differentiation in vitro. Transgenic mice expressing two- to threefold increased expression of α5β1 or α2β1 integrins exhibit epidermal hyper-proliferation, perturbed keratinocyte differentiation and other features of psoriasis, a skin disease (Carroll et al., 1995). This epidermal hyper-proliferation is similar to our observations on myoblasts expressing ectopic α5 integrin. Our working model provides a possible mechanism. Therefore, it is likely that quantitative changes in mitogenic signaling through alterations in adhesive signaling produces phenotypic alterations that operate through common mechanisms in diverse systems.

We thank J.T. Parsons, M. Weber, Y. Yakada, S. LaFlamme, L. Reichardt, R. Isberg, M. Hemler, D. Fishman, Y. Takada, and A. Mercurio for generous gifts of reagents. We also thank A. Richardson and J.T. Parsons for their advice and M. Berg for her comments on the manuscript.

This work was supported by National Institutes of Health grants GM23244 (A. Horwitz) and GM47607 (C. Turner). C. Turner is an Established Investigator of the American Heart Association.

BCA

bicinchoninic acid

CA-MEK

constitutively active MEK

FAK

focal adhesion kinase

HA

hemagglutinin

MAP

mitogen-activated protein

MEK

MAP/erk kinase

UT

untransfected