Bifurcation of Cell Migratory and Proliferative Signaling by the Adaptor Protein Shc

Rabbit polyclonal antibodies to Shc were purchased from Transduction Laboratories. Mouse monoclonal Shc antibodies were from Santa Cruz Biotechnology. Anti-phosphotyrosine antibodies (4G10) were from Upstate Biotechnology. Shc cDNAs subcloned into pCDNA3.1HisC were previously described (Ricketts et al. 1999). TRITC-conjugated phalloidin was obtained from Sigma. 6F1 was a gift of Dr. Virgil Woods (UCSD, La Jolla, CA) and mAb 1973Z was obtained from Chemicon.

FG-M cells were maintained in RPMI supplemented with 10% FBS, 1 μg/ml gentamicin, and 0.4 mM glutamine (gentamycin/glutamine; Sigma). Cells were used from passage 8 to passage 13. Cos-7 cells were maintained in DME supplemented with 10% FBS, 0.4 mM glutamine, and 1 μg/ml gentamicin. Cells were used from passage 6 to passage 20 and were mycoplasma free during the course of these studies.

Cells were transfected with lac z (to identify transfected cells) and the indicated constructs on 10-cm tissue culture plates using LipofectAMINE (GIBCO BRL) per the manufacturer's instructions. Cos-7 cells were incubated in the transfection mixture for 5 to 7 h and FG-M cells were allowed to incubate for 15 h. At the end of this time, cells were returned to full growth medium for either 15 h (Cos-7 cells) or 8 h (FG-M) cells, and serum deprived for 20–24 h before use in experiments.

Cos-7 cells were transfected with wild-type, murine p52 Shc cDNA subcloned into pCDNA3.1HisC using LipofectAMINE per the manufacturer's instructions. 3 d later, cells were passaged into growth medium supplemented with 500 μg/ml G418 to select for p52 Shc-expressing cells. After selection, cells were seeded at low density into 96-well plates, single colonies were isolated with a cloning ring, and expanded for two passages before experimental use.

Acid-washed coverslips were coated with collagen (10 μg/ml) and nonspecific binding was blocked with 0.5% RIA grade BSA (Sigma). Serum-deprived cells were harvested as described below and seeded onto collagen-coated coverslips. Cells were allowed to spread for 2 h, then were fixed in 3.7% formaldehyde in PBS, permeabilized for 1 min in 0.01% Triton X-100, and actin was visualized by incubating the coverslips for 1 h with 2 μg/ml rhodamine conjugated phalloidin (Sigma). Coverslips were washed three times in PBS and three times in deionized water then mounted with gelvatol. Images were acquired with Bio-Rad MRC1024 confocal system using Bio-Rad Lasersharp software and a 63× objective on a Zeiss Axiovert 100. Images were processed with Adobe Photoshop.

Motility experiments were performed as previously described (Klemke et al. 1998). In brief, Boyden chambers (Millipore Millicells, 8-mm pore size) were coated on either their lower surfaces with 10 μg/ml of the indicated extracellular matrix proteins for 2 h at 37°C. Chambers were then placed in 24-well culture dishes containing either serum-free DME supplemented with gentamicin, glutamine, and 0.5% BSA (RIA grade, “migration buffer”; Sigma) or migration buffer supplemented with 100 ng/ml EGF (mouse receptor grade; UBI) or 25 mg/ml insulin (human recombinant insulin; Boehringer Mannheim Biochemica) as indicated. Cells were harvested in buffer consisting of Hanks' balanced salt solution, 25 mM Hepes, 5 mM EDTA, and 0.01% trypsin. Next, cells were washed twice in serum-free DME containing 0.5% BSA and 150,000 cells were loaded into the upper portion of the Boyden chambers. Migrations were allowed to proceed for 4–6 h. At the end of this time, nonmigratory cells were removed with a cotton swab, and chambers were stained for β-galactosidase activity (to identify transfected cells) or stained with crystal violet. Migration was quantitated by blind counting the number of migratory cells on the lower surface of the membrane of at least two fields per chamber using a 20× objective. Statistical analysis was performed using InStat (Graph Pad) by pooling the data from multiple independent experiments.

Integrin antibody blocking experiments were performed as described above except that Cos-7 cells were mixed with the indicated anti-integrin antibodies immediately before loading into Boyden chambers; experiments were then allowed to proceed as above for 4 h. Random migration experiments were performed using chambers that had been coated with extracellular matrix protein on either their lower surfaces (haptotaxis) or their upper and lower surfaces (random migration). Video time-lapse microscopy was also attempted to assess random migration in serum-deprived cells on extracellular matrix protein; however, during the course of the experiment (6 h), Cos-7 cells did not migrate using this method.

Petri dishes were coated with extracellular matrix proteins as indicated. Nonspecific binding of cells to plates were blocked by incubating with heat-denatured BSA for 30 min. Next, serum-deprived cells were harvested as described above and either left in suspension for 20 min or plated onto matrix-coated, blocked plates for the indicated times and harvested in a lysis buffer containing 10 mM Tris, pH 7.6, 150 mM sodium chloride, 0.1% SDS, 1% sodium deoxycholate, 1% Triton X-100, and 5 mM EDTA. The following inhibitors were added immediately before use: complete protease inhibitor (1 tablet/20 ml; Boehringer Mannheim), 1 mM PMSF, 2.5 mM sodium orthovanadate. Shc was either immunoprecipitated using a polyclonal Shc antibody (Transduction Labs) or His-tagged Shc was isolated using nickel agarose beads (Ni-NTA agarose; Qiagen). Phosphotyrosine was detected using the monoclonal anti-phosphotyrosine antibody, 4G10. Shc was detected using a either a monoclonal or a polyclonal Shc antibody. Bands were visualized by chemiluminescence (Super Signal; Pierce). Expression levels of Shc mutants in migration assays was determined by reserving an aliquot of cell lysate from the migration assay and analyzing it by Western blot with a polyclonal Shc antibody.

Cells were transfected as described above on coverslips. 24 h later, cells were serum starved overnight and stimulated with either serum-free DME or DME containing 100 ng/ml EGF. 12 h later, cells were incubated for an additional 6 h with bromodeoxyuridine (BrdU) to identify newly synthesized DNA. Coverslips were then washed with PBS and fixed in 3.7% formaldehyde. BrdU was quantitated in transfected cells by staining with a rat anti-BrdU antibody (Amersham) and a mouse antiExpress antibody (Invitrogen, to identify proteins expressed from pCDNA3.1HisC).

Recent reports have identified a role for p52 Shc in integrin signaling leading to cell cycle progression (Wary et al. 1996). Since many of the signals that regulate mitogenesis also promote cell migration (Pages et al. 1993; Cowley et al. 1994; Olson et al. 1995; Anand-Apte et al. 1997; Klemke et al. 1997), we reasoned that Shc might also play a role in cell movement. To address this question, Cos-7 cells were stably transfected with wild-type murine p52 Shc cDNA. The resulting cells displayed an altered morphology from the parental cell line. Phalloidin staining of p52 Shc cells and parental Cos-7 cells revealed a decreased number of actin-containing stress fibers in Shc-expressing cells (Fig. 1), suggestive of a motile phenotype. The migratory capacity of the parent and stable p52 Shc cells was then assessed. Serum-deprived cells were allowed to migrate in the absence of added cytokines using Boyden chambers that had been coated on their lower surface with collagen. As shown in Fig. 1, p52 Shc stable cells were approximately three times more migratory than the parental cells. When the data from four independent experiments were analyzed this difference was found to be statistically significant (P < 0.05 by Student's t test), suggesting that Shc expression is associated with the haptotactic migratory phenotype of these cells.

To extend these findings, we assessed the motility of Cos-7 cells transiently expressing p52 Shc. Cells were serum-deprived and allowed to migrate on either collagen or vitronectin. As shown in Fig. 2, expression of p52 Shc was sufficient to promote cell migration to an extent similar to that seen with EGF, a strong migratory stimulus for these cells (Klemke et al. 1998). In addition, Shc promoted a similar migration response on collagen and vitronectin. Data from several experiments demonstrated that both of these responses were statistically significant (P < 0.01 by Student's t test). Cos-7 cells can use α2β1 to migrate on collagen (Klemke et al. 1997) suggesting that Shc's ability to transmit integrin-mediated signals for migration may be more permissive than previously observed for cell cycle progression (Wary et al. 1996). In agreement with the motility results, ligation of integrins with either collagen or vitronectin resulted in tyrosine phosphorylation of wild-type p52 Shc indicating that both of these ECM proteins support Shc phosphorylation (Fig. 2).

Specific anti-integrin antibodies were used to determine the integrin dependence of Shc's effect upon motility. To determine the role of the collagen receptors α2β1 and α1β1 in this response, we first established concentrations of α1 (mAb 1973Z) and α2β1 (6F1) monoclonal antibodies that inhibited adhesion of Cos-7 cells to collagen (5 μg/ml of anti-α1 and 10 μg/ml of anti-α2β1, data not shown). Next, these concentrations of monoclonal antibodies were added to the cells immediately before placement in Boyden chambers. Both anti-α1 antibodies and α2β1 antibodies inhibited Shc's ability to promote cell migration; α2β1 antibodies had a more pronounced effect (Fig. 3). These data establish that α2β1 and Shc can signal along a common pathway resulting in cell movement. P4C10, an antibody directed against the shared β1 subunit, completely inhibited haptotaxis of mock and Shc transfected cells on collagen (data not shown). LM609, a function blocking monoclonal antibody to αvβ3, prevented Shc-stimulated migration towards vitronectin, but not on collagen (data not shown), further confirming that Shc requires integrins to effect cell movement.

To determine the nature of Shc-induced cell migration, cells were allowed to migrate in Boyden chambers under two experimental conditions. Chambers were either coated with collagen on their lower surface to enable cells to engage in haptotactic migration, or they were coated on both their upper and lower surfaces to evaluate the random motility of these cells. Shc was only able to significantly enhance motility when chambers were coated on the lower surface, demonstrating that Shc promotes haptotaxis and not random migration (Fig. 3 b).

Having established a role for Shc in haptotaxis, mutational analysis was performed to identify regions of Shc that mediate this response. Several point mutants were employed for these studies (Fig. 4). To address the role of Shc tyrosine phosphorylation in cell migration, increasing amounts of cDNA encoding mutant forms of Shc bearing tyrosine to phenylalanine point mutations at residues 239/240 (Shc Y239F/Y240F), residue 317 (Shc Y317F) or both residues 239/240 and residue 317 (ShcY239F/Y240F/Y317F, Shc 3YF) were expressed in Cos-7 cells. With the exception of Shc 3YF, these mutants were expressed to similar extents as shown in Fig. 5 A. Next, cells expressing these mutants were allowed to migrate on a collagen or vitronectin substrate. Only wt Shc was found to significantly enhance cell migration (P < 0.01 by analysis of variance [ANOVA] of six independent experiments). Neither Y239/240F nor Y317F mutants stimulated motility substantially; although stimulation was observed by both of these mutants in certain experiments, it was not found to be statistically significant when all experiments were analyzed (P > 0.1 by ANOVA). The triple mutant, Shc 3YF, was unable to induce cell motility on either matrix (Fig. 5 a) and appeared to suppress migration somewhat. Therefore, tyrosine phosphorylation of both residues 239/240 and 317 appear to play a critical role in this response.

Shc's phosphotyrosine binding domains, the PTB and the SH2 domains, are known to be crucial for mitogenic signaling through growth factor receptors (Gotoh et al. 1995; Ricketts et al. 1996; Thomas and Bradshaw 1997). Phosphotyrosines on activated tyrosine kinase receptors can be bound by these domains, thereby localizing Shc to the receptor (Bonfini et al. 1996). To determine whether these domains might also be important for integrins to signal through Shc, cells were transfected with varying amounts Shc cDNAs containing point mutations selectively designed to inhibit function of these domains. Shc S154P contains a loss of function mutation in the PTB domain that prevents interaction with activated insulin receptors (Ricketts, 1999, and Gustafson, T., personal communication). Shc R401L contains a mutation in the conserved phosphotyrosine binding FLVR motif of the Shc SH2 domain, rendering its SH2 domain non-functional (Mayer et al. 1992). Mutation of the PTB domain (ShcS154P) blocked Shc's ability to promote haptotaxis (Fig. 5 A). In contrast, the Shc SH2 domain appeared to be dispensable for migration, since ShcR401L consistently promoted migration as efficiently as wild-type Shc (Fig. 5 A). Analysis of six independent experiments confirmed that the migration stimulated by Shc R401L was statistically significant (P < 0.01 by ANOVA). The integrin-stimulated tyrosine phosphorylation pattern of these Shc constructs correlated with their ability to promote cell motility. For example, wild-type and Shc R401L were phosphorylated in response to cell adhesion while Shc 3YF and Shc S154P were not. Once again, no differences were observed between the migratory or tyrosine phosphorylation responses on collagen and vitronectin in these studies (Fig. 5 B). In summary, expression of p52 Shc promotes motility of Cos-7 cells, and tyrosine phosphorylation and the PTB domain are required for this response.

To establish whether Shc might be required for cell migration in general, experiments were performed using the constitutively migratory pancreatic carcinoma cell, FG-M. These cells, selected for their migratory properties in vitro, acquired a metastatic phenotype in vivo (Klemke et al. 1998). Shc was constitutively tyrosine phosphorylated in serum-deprived FG-M cells, even when cells are held in suspension (Fig. 6). This is in contrast to Cos-7 cells, which displayed adhesion-dependent Shc phosphorylation. Therefore, we expressed migration incompetent forms of Shc (Shc 3YF and Shc S154P) in FG-M cells to determine if Shc was required for the migration of these cells. As shown in Fig. 6, haptotaxis in FG-M cells was highly Shc-dependent, since expression of either Shc 3YF or Shc S154P dramatically inhibited haptotaxis towards vitronectin (by 76 and 83%, respectively). These results suggest that Shc is required for the spontaneous, haptotactic phenotype of these cells.

Based on the findings that Shc appeared to be critical for haptotaxis, studies were performed to investigate the role of Shc in chemotaxis. In these experiments, cells expressing wild-type or mutant forms of Shc were allowed to migrate in collagen coated Boyden chambers towards a gradient of EGF. The EGF response was quantitated by dividing the number of migratory cells observed in the presence vs the absence of EGF for each experimental condition. In mock transfected cells, addition of EGF-stimulated a 2.6-fold increase in migration relative to that observed in its absence. Both wt Shc and Shc R401L substantially increased the number of migratory cells observed in the absence of growth factor, as discussed above. Stimulation of the EGF receptor increased this number further, generating an EGF response of approximately twofold. Surprisingly, Shc was not required for EGF-stimulated motility since expression of either Shc 3YF or Shc S154P failed to block EGF-induced increases in cell migration (Fig. 7). Similar results were observed when IGF-1 was used to stimulate migration (data not shown). Therefore, Shc appears to play a preferential role in haptotaxis as it is not required for the chemotactic response of these cells.

The role of Shc in mitogenic signaling by growth factor receptors is well established (Bonfini et al. 1996). In nontransformed cells, integrin ligation and Shc phosphorylation is required for ERK activation and cell cycle progression even in the presence of growth factors (Wary et al. 1996). That Shc was able to mimic growth factor effects on cell migration suggests it may also regulate DNA synthesis under conditions of serum deprivation. To test this possibility, Cos-7 cells were transfected with either wild-type or mutant forms of Shc and then assessed for BrdU incorporation in the absence of exogenous growth factor addition. As shown in Fig. 8, Shc did not regulate DNA synthesis under these growth conditions as wild-type or mutant forms Shc failed to influence BrdU incorporation. Thus, in the absence of exogenous growth factors, Shc expression was not sufficient to promote proliferation even though it was able to induce haptotactic cell migration under these conditions.

Whereas Shc does not influence anchorage-dependent DNA synthesis, previous studies have documented that Shc is crucial for proliferation (Bonfini et al. 1996). Interestingly, the Shc requirements for mitogenesis and haptotaxis are not identical. As shown in Fig. 8, mutation of either tyrosines 239/240 or 317 alone or in combination completely inhibited EGF's ability to increase DNA synthesis, suggesting non-redundant function of these sites for cell cycle progression. Expression of Shc R401L, which induced haptotaxis to the same extent as wild-type Shc, inhibited EGF-stimulated DNA synthesis. In contrast to its effect on haptotaxis, Shc S154P did not affect EGF-stimulated proliferation. Thus, the Shc SH2 domain selectively transduces mitogenic signals in response to EGF, while the PTB domain is critical for haptotaxis.

Cell migration and proliferation are required for wound healing, embryonic development, and angiogenesis. The observation that, during these processes, cells can either migrate in the absence of proliferation or proliferate without migrating necessitates the existence of signaling pathways that distinguish between these two responses (Henderson and Copp 1997; Martin 1997; Perris 1997; Risau 1997). In this report evidence is provided that the adaptor protein, p52 Shc, participates in this decision-making process. Under growth factor–limiting conditions, Shc was required for haptotaxis, but did not influence DNA synthesis. In contrast, when growth factors were present, Shc was crucial for proliferation, but failed to impact cell migration. Shc's phosphotyrosine binding domains were differentially required for these responses; the PTB transduced migratory signals, while the SH2 was required for mitogenesis.

Shc's signaling capacity is regulated by tyrosine phosphorylation. After activation of growth factor receptors and cytosolic tyrosine kinases, Shc becomes phosphorylated (McGlade et al. 1992; Pelicci et al. 1992) on residues 239/240 and 317 within the CH domain and binds Grb2 (Rozakis-Adcock et al. 1992; Salcini et al. 1994; Gotoh et al. 1996; van der Geer et al. 1996), ultimately resulting in the activation of Ras and ERK. However, the biological role of these residues in integrin signaling remains unclear. Recent reports have demonstrated a requirement for Shc Y317 for integrin-mediated ERK activity (Wary et al. 1996), while another report suggested that integrins can stimulate the ERK pathway independently of Y317 (Schlaepfer et al. 1998). In the current report, mutation of either of these sites alone resulted in a loss of Shc's ability to stimulate migration, suggesting that integrins can utilize both of these sites. Tyrosines 239/240 and 317 were crucial for EGF-stimulated mitogenesis in Cos-7 cells, in agreement with findings of others that these sites couple to distinct effectors in mitogenic signaling (Gotoh et al. 1996, Gotoh et al. 1997). In summary, Shc tyrosine phosphorylation does not distinguish between proliferative and cell migratory signaling pathways.

The observation that cytoskeletally associated kinases phosphorylate Shc (McGlade et al. 1992; Schlaepfer et al. 1998; Wary et al. 1998) suggests that recruitment of Shc to the actin-associated cytoskeleton is an important step in integrin signaling. Recent studies defined a role for the SH3 domain of the Src family member Fyn in this process (Wary et al. 1998). Presumably, the Fyn SH3 domain interacts with proline-rich regions in the CH domain of Shc. The PTB domain may also contribute to Shc recruitment. The amino-terminal domain of Shc was reported to mediate the association of this adaptor protein to an actin-rich cellular fraction (Thomas et al. 1995). In agreement with this report, we find that mutation of the PTB domain abolishes integrin-stimulated Shc tyrosine phosphorylation; the SH2 domain is dispensable for this process. The PTB can bind phospholipids (Zhou et al. 1995) which enhances Shc phosphorylation by cSrc (Sato et al. 1997). Taken together, these observations suggest a model in which the PTB domain localizes Shc to the membrane where it becomes phosphorylated by cytoskeletally associated tyrosine kinases, such as Src, Fyn and/or FAK after integrin ligation, which ultimately results in cell migration.

While Shc-dependent cell migration and proliferation depend upon tyrosine phosphorylation, these processes can be distinguished by their differential requirement for Shc's phosphotyrosine binding domains. The SH2 domain is selectively required for proliferation, as mutation of this domain dramatically inhibited EGF-stimulated DNA synthesis but had no effect on motility, or integrin-induced Shc phosphorylation. The PTB domain, in contrast, mediated tyrosine phosphorylation of Shc in response to integrin ligation and haptotactic cell migration. These findings may be biologically relevant for a number of processes. For example, Shc may play an important role during neural crest cell migration, when cells migrate along a haptotactic gradient without proliferating (Henderson and Copp 1997) or during wound healing when fibroblasts migrate in a fibrin-rich ECM (Martin 1997). However, when cells reach areas of high growth factor concentration, Shc may facilitate the cell proliferation response. Activation of the EGF receptor promotes the redistribution of Shc from a perinuclear to a plasma membrane localization (Lotti et al. 1996). EGF-R phosphorylation is required for this effect, presumably because receptor activation creates consensus binding sites for Shc's phosphotyrosine binding domains (Lotti et al. 1996). Thus, when growth factors are present, Shc may be recruited from areas where it transduces haptotactic signals, such as integrin containing focal contacts, to regions of the membrane containing growth factor receptors where it participates in mitogenic signaling.

In summary, proliferation and motility are controlled by adhesion proteins and growth factors associated with the ECM (Martin 1997). Integrin and growth factor receptor signals signal through the adaptor protein Shc to regulate these responses (Bonfini et al. 1996; Wary et al. 1996, Wary et al. 1998). We find that migratory and proliferative signals also bifurcate at the level of Shc. This divergence is characterized by differential requirements of the Shc PTB domain and the Shc SH2 domain.

Note added in proof. While this manuscript was being reviewed, Shc was reported to decreased focal adhesion organization and to increase cell migration in Gu, J., M. Tamura, R. Pankov, E.H.J. Danen, T. Takino, K. Matsumoto, and K.M. Yamada. 1999. Shc and FAK differentially regulate cell motility and directionality modulated by PTEN. J. Cell Biol. 146:389–404.

The authors would like to thank Drs. Richard Klemke, Jie Leng, Dwayne Stupack, and David Schlaepfer for helpful advice and insightful discussions. We also thank Dr. Kathy Spencer for her expert assistance with confocal microscopy and time-lapse video microscopy.

This work was supported by National Institutes of Health grants CA50286, CA45726, HL54444, and P01CA78045 to D.A. Cheresh. L.R. Collins was supported by National Cancer Institute minority supplement CA45726-12S1. This is manuscript number 12586-IMM from The Scripps Research Institute.