α2 Integrin Subunit Cytoplasmic Domain-dependent Cellular Migration Requires p38 MAPK*

The α2 integrin subunit cytoplasmic domain uniquely supported epidermal growth factor (EGF)-stimulated migration on type I collagen. p38 MAP kinase- and phosphatidylinositol 3-kinase-specific inhibitors, but not a MEK-specific inhibitor, eliminated EGF-stimulated and unstimulated α2-cytoplasmic domain-dependent migration. Following adhesion to collagenous matrices, cells expressing the full-length α2 integrin subunit, but not cells expressing a chimeric α2 integrin subunit in which the α2-cytoplasmic domain was replaced by the cytoplasmic domain of the α1-subunit, exhibited sustained and robust phosphorylation of p38 MAP kinase. Expression of dominant negative p38 MAP kinase inhibited α2-cytoplasmic domain-dependent, EGF-stimulated migration as well as unstimulated migration on collagen. Expression of constitutively active Rac1(Val-12) augmented p38 MAP kinase activation and α2-cytoplasmic domain-dependent migration. It also rescued the ability of cells expressing the α1-cytoplasmic domain to activate p38 MAPK and to migrate. These results suggest that the α2 integrin cytoplasmic domain uniquely stimulates the p38 MAP kinase pathway that is required for unstimulated and EGF-stimulated migration on type I collagen.

The ␣ 2 integrin subunit cytoplasmic domain uniquely supported epidermal growth factor (EGF)-stimulated migration on type I collagen. p38 MAP kinase-and phosphatidylinositol 3-kinase-specific inhibitors, but not a MEK-specific inhibitor, eliminated EGF-stimulated and unstimulated ␣ 2 -cytoplasmic domain-dependent migration. Following adhesion to collagenous matrices, cells expressing the full-length ␣ 2 integrin subunit, but not cells expressing a chimeric ␣ 2 integrin subunit in which the ␣ 2 -cytoplasmic domain was replaced by the cytoplasmic domain of the ␣ 1 -subunit, exhibited sustained and robust phosphorylation of p38 MAP kinase. Expression of dominant negative p38 MAP kinase inhibited ␣ 2 -cytoplasmic domain-dependent, EGF-stimulated migration as well as unstimulated migration on collagen. Expression of constitutively active Rac1(Val-12) augmented p38 MAP kinase activation and ␣ 2 -cytoplasmic domaindependent migration. It also rescued the ability of cells expressing the ␣ 1 -cytoplasmic domain to activate p38 MAPK and to migrate. These results suggest that the ␣ 2 integrin cytoplasmic domain uniquely stimulates the p38 MAP kinase pathway that is required for unstimulated and EGF-stimulated migration on type I collagen.
The integrin family of heterodimeric cell surface adhesion receptors mediates not only adhesion to the extracellular matrix or other cells but also serves to integrate signals from the outside of the cell to the inside of the cell (1)(2)(3)(4)(5). Although both the ␣and ␤-subunits are composed of extracellular and transmembrane domains and short cytoplasmic domains, only the role of the ␤-subunit cytoplasmic domain in interacting with cytoskeletal proteins and signaling molecules, such as focal adhesion kinase and integrin-linked kinase, has been well characterized (6 -10). The role of the many different ␣-subunit cytoplasmic domains is not well understood. Although both the ␣ 1 ␤ 1 and ␣ 2 ␤ 1 integrins mediate cellular adhesion to collagens and/or laminins (11)(12)(13)(14), studies from our laboratory as well as a number of other laboratories (15)(16)(17)(18)(19)(20)(21)(22) have suggested that the ␣ 1 ␤ 1 and ␣ 2 ␤ 1 integrin receptors are not simply redundant receptors but serve to mediate different downstream events.
To begin to explore the mechanisms underlying the distinct phenotypes mediated by the two collagen/laminin receptors, we re-expressed either the full-length ␣ 2 integrin subunit (X2C2) or a chimeric integrin ␣-chain composed of the extracellular and transmembrane domains of the ␣ 2 -subunit fused to the cytoplasmic domain of the ␣ 1 -subunit (X2C1) in a variant subclone of the NMuMG cell line, which lacks endogenous expression of either the ␣ 1 ␤ 1 or the ␣ 2 ␤ 1 integrin. As reported earlier, the X2C2 and X2C1 transfectants both effectively adhered, spread, and formed focal adhesion complexes on type I collagen matrices (16). However, the X2C2 transfectants, but not the X2C1 transfectants, developed elongated branches and tubules in three-dimensional collagen gels and migrated on type I collagen in response to a chemotactic gradient of epidermal growth factor (EGF). 1 In this study we demonstrate that ␣ 2cytoplasmic domain-dependent migration required activation of the p38 mitogen-activated protein kinase (MAPK) cascade. The ␣ 2 -cytoplasmic domain, but not the ␣ 1 -cytoplasmic domain, led to p38 MAPK activation. Additionally, expression of constitutively active Rac1(Val-12) rescued p38 MAPK activation and EGF-stimulated chemotactic migration in the X2C1 transfectants.

EXPERIMENTAL PROCEDURES
Cell Culture and Transfection-The murine NMuMG cell line was maintained in Dulbecco's modified Eagle's medium (Life Technologies, Inc.) supplemented with 10% fetal bovine serum and insulin (5 g/ml). The variant NMuMG subclone NMuMG-3, lacking ␣ 1 ␤ 1 and ␣ 2 ␤ 1 integrin receptors, was derived by limiting dilution techniques and has been described in detail elsewhere (16). The full-length human ␣ 2 integrin (X2C2) cDNA, in the expression vector pFneo, was a generous gift from Dr. Martin E. Hemler (Harvard Medical School, Boston) (23,24). The chimeric integrin cDNA containing the extracellular domain of ␣ 2 and the cytoplasmic domain of ␣ 1 was constructed, as previously described (16). The full-length and chimeric cDNA constructs were subcloned into the expression vector pSR␣ (a gift from Dr. Andrey S. Shaw, Washington University School of Medicine, St. Louis, MO) that contains a cytomegalovirus promoter. All constructs were transfected into the NMuMG-3 clonal cell line using calcium phosphate transfection methodology. Clonal cell lines were selected and maintained in geneticin (850 g/ml) and evaluated by Southern blot analysis for integration site determination to ensure that distinct clones were evaluated.
The X2C2-and X2C1-expressing clonal cell lines were cotransfected with FLAG-tagged wild-type or dominant negative (DN) p38␣ MAPK cDNA clones, or DN c-Jun N-terminal kinase cDNA clones (gifts from Dr. Aubrey R. Morrison, Washington University School of Medicine, St. Louis, MO), or HA-tagged constitutively active Rac1(Val-12) or dominant negative Rac1(Asn-17) cDNA clones (gifts from Dr. Margaret Chou, University of Pennsylvania, Philadelphia) and the Selecta Vecta-Hyg plasmid (Novagen) using Lipofectin (Life Technologies, Inc.), according to the manufacturer's instructions. Clonal and nonclonal cell lines were selected and maintained in media containing only geneticin (850 g/ml) or geneticin plus hygromycin (448 g/ml). Selected cell lines were evaluated for the expression of the epitope tag using either the goat polyclonal anti-FLAG tag (OctA-probe) (Santa Cruz Biotechnology) * This work was supported in part by National Institutes of Health Grants CA70275 and CA83690. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
‡ To whom correspondence should be addressed: or the mouse monoclonal anti-HA antibody (12 CA5) (Roche Molecular Biochemicals).
Cell Migration on Type I Collagen-Cell migration assays were performed using a modification of the protocol described previously (25). Cells were serum-starved in media containing 0.4% serum and insulin (5 g/ml) for 48 h and then in media containing insulin (5 g/ml) for 24 h. Cells were removed from flasks with trypsin/EDTA and washed twice before being replated in transwell chambers. Briefly, 12-mm transwell chambers (Corning Costar Corp.) containing polycarbonate membrane with 12-m pores were coated overnight at 4°C with type I collagen (25 g/ml) (Collaborative Biomedical Products) or fibronectin (25 g/ml) (Sigma Diagnostics Inc.). The filters were washed with phosphate-buffered saline and air-dried. The bottom chamber was filled with Dulbecco's modified Eagle's medium containing 1% bovine serum albumin (Sigma) and Mg 2ϩ (2 mM). Cells were placed in the top chamber at 1.5 ϫ 10 5 cells/ml in Dulbecco's modified Eagle's medium containing 1% bovine serum albumin and Mg 2ϩ (2 mM) and allowed to migrate for 5 h at 37°in a humidified CO 2 incubator. In experiments in which recombinant human EGF was included, EGF (10 ng/ml) was placed in either the lower chamber only or in both the upper and lower chambers as a control for chemotaxis. In experiments where inhibitors were used, the cells were incubated with inhibitors for 15 min. Inhibitors were included in both the upper and lower chambers of the transwell device. After the 5-h incubation, cells remaining on the upper surface of the transwell filter were removed by mechanical scraping. Cells migrating to the lower surface of the filter were detected by fixation and staining with Gill's hematoxylin and eosin solution (Sigma). The number of cells migrating to the lower surface was determined by counting the number of cells in 10 random high power (ϫ 40) fields. Data presented represent the mean Ϯ S.E. of at least three separate experiments. Statistical analyses were carried out by unpaired t tests using GraphPad Prism version 2.01.
Immunoblot and Immunoprecipitation Analyses-Cells were serumstarved and treated as described for the migration assays but were plated onto Petri dishes coated with either type I collagen (25 g/ml) or fibronectin (25 g/ml) and lysed at defined time points in lysis buffer (50 mM HEPES (pH 7.2), 250 mM NaCl, 2 mM EDTA, 0.1% Nonidet P-40, 100 g/ml aprotinin, 50 g/ml leupeptin, 40 mM NaF, 0.5 mM phenylmethylsulfonyl fluoride, 0.5 mM o-vanadate, and 1 mM dithiothreitol). Total protein concentration was determined by the Pierce protein assay (Fisher).
For immunoblot analyses, equivalent amounts of protein lysate were subjected to SDS-PAGE and electroblotted onto Immobilon-P transfer membrane (Fisher). Immunoblots were blocked in 5% bovine serum albumin or 5% dried milk in Tris-buffered saline containing 0.5% Tween 20 and incubated overnight with an appropriate dilution of primary antibody at 4°C. Secondary antibody incubation was performed with horseradish peroxidase-conjugated sheep anti-mouse IgG or anti-rabbit IgG (Amersham Pharmacia Biotech) for 2 h at room temperature. The ECL Chemiluminescence System (Amersham Pharmacia Biotech) was used for visualization.
For immunoprecipitation analyses, equivalent amounts of protein lysate were pre-cleared with rabbit anti-mouse IgG (Jackson Immu-noResearch Laboratories, Inc.) and immunoprecipitated with polyclonal anti-EGF receptor antibody (Santa Cruz Biotechnology) and protein A-Sepharose beads (Sigma). Immunoprecipitated protein was subjected to SDS-PAGE, electroblotted onto Immunobilon-P transfer membrane (Fisher), and immunoblotted with the appropriate dilution of either monoclonal anti-PY99 (Santa Cruz Biotechnology) and subsequently with polyclonal anti-EGF receptor (Santa Cruz Biotechnology).
Antibodies and Reagents-Polyclonal anti-total p38 MAP kinase antibody, polyclonal anti-phospho-p38 MAP kinase antibody (New England Biolabs), monoclonal anti-HA antibody (Roche Molecular Biochemicals), monoclonal anti-FLAG antibody (Sigma), and polyclonal anti-Jun kinase antibody (New England Biolabs) were used for immunoblot analysis.

RESULTS
We recently reported that an integrin collagen receptor containing the ␣ 2 integrin cytoplasmic domain, but not the ␣ 1 integrin cytoplasmic domain, supported EGF-stimulated chemotaxis on a matrix of type I collagen (16). In agreement with these recent observations, NMuMG-3 subclones, lacking endogenous ␣ 1 ␤ 1 and ␣ 2 ␤ 1 integrins, but expressing the full-length ␣ 2 integrin subunit cDNA construct (X2C2), migrated on type I collagen when stimulated by a chemotactic gradient of EGF (10 ng/ml). In contrast, NMuMG-3 transfectants expressing a chimeric integrin subunit consisting of the extracellular and transmembrane domains of the ␣ 2 integrin subunit fused to the cytoplasmic domain of the ␣ 1 integrin subunit (X2C1) migrated poorly on type I collagen in response to EGF (Fig. 1A), a finding also in agreement with our previous study. Since the vectoronly control transfectants did not adhere to collagen substrates, their migration could not be evaluated. To determine whether EGF-stimulated migration was unique to ␣ 2 ␤ 1 integrin-mediated adhesion to collagen substrates or was a more general accompaniment of integrin-mediated adhesion, we assessed the ability of EGF to stimulate migration of the transfectants on fibronectin. The NMuMG-3 cell line and the transfectants expressed similar levels of the ␣ 5 ␤ 1 integrin. The level of expression of the ␣ 5 ␤ 1 integrin on these cells was comparable to the level of the ␣ 2 ␤ 1 expression by the transfectants (16). Although the X2C2 and X2C1 transfectants adhered to fibronectin, both transfectants failed to migrate on fibronectin in response to EGF (Fig. 1, B and C). The differences in EGFstimulated migration were not due to differences in the expression or the extent of phosphorylation of the EGF receptor on the two different clonal cell lines (Fig. 1D). Similar results were obtained with multiple clones (see Ref. 16 and data not shown).
To dissect the pathways responsible for EGF-stimulated ␣ 2 integrin subunit cytoplasmic domain-dependent chemotaxis, we determined the ability of inhibitors of different signaling pathways to inhibit chemotactic migration. As shown in Fig.  2A, inhibition of either the phosphatidylinositol 3-kinase (PI3K) or the p38 MAPK pathways completely inhibited ␣ 2 integrin-dependent EGF-stimulated chemotaxis on type I collagen. Inhibition of chemotactic migration by the p38 MAPKspecific inhibitor SB203580 was concentration-dependent, as shown in Fig. 2B. Surprisingly, the MEK-specific inhibitor of the MAPK/extracellular signal-regulated kinase pathway PD98059 (50 M) failed to inhibit migration. Although rapamycin (5 pM), the p70 S6 kinase-specific inhibitor, has been shown to inhibit migration of endothelial cells and leukocytes (26 -28), it failed to inhibit ␣ 2 -cytoplasmic domain-dependent migration. Since the PI3K pathway is important for stimulating migration through a number of different integrins (29,30), we focused on the role of the p38 MAPK pathway in mediating signals downstream of the ␣ 2 ␤ 1 integrin that may lead to chemotactic migration.
To evaluate the role of p38 MAPK activation by the X2C2 and X2C1 transfectants, we evaluated the time course of activation and phosphorylation of p38 MAPK resulting from adhesion to type I collagen. The X2C2 and X2C1 transfectants were serum-starved in a manner identical to that described for the chemotactic migration assays, plated on type I collagen (25 g/ml) in the absence or presence of EGF (10 ng/ml), and lysed at defined time points. Adhesion of the X2C2 transfectants to type I collagen resulted in phosphorylation of p38 MAPK within 15 min (Fig. 3). Low level p38 MAPK phosphorylation was maintained at 1 h and then declined to base-line levels by 3 and 6 h. The presence of EGF slightly augmented p38 MAPK phosphorylation at 15 min and 1 h. Adhesion to collagen by the X2C1 transfectants stimulated only minimal p38 MAPK phosphorylation at 15 min; phosphorylated p38 MAPK was undetectable after 1 h. EGF slightly augmented phosphorylation of p38 MAPK by the X2C1 transfectant at 15 min but not at later time points (Fig. 3). These findings demonstrate that in a cell line that expressed no endogenous ␣ 1 ␤ 1 or ␣ 2 ␤ 1 integrin, so that adhesion to collagen was entirely dependent upon the transfected collagen receptor, adhesion to type I collagen activated p38 MAPK in a manner that was dependent upon the presence of an ␣ 2 -cytoplasmic domain. In contrast, EGF only slightly augmented phosphorylation of p38 MAPK when the X2C1 transfectants were adherent to type I collagen.
To examine further the requirement of the p38 MAPK pathway in ␣ 2 -cytoplasmic domain-dependent EGF-stimulated chemotaxis, the X2C2 transfectants were cotransfected with either epitope-tagged wild-type or dominant negative (DN) p38 MAPK cDNA constructs and the Selecta Vecta-Hyg plasmid. Clonal cell lines were selected in geneticin plus hygromycin, and expression of the epitope tag was evaluated by immunoblot analysis (Fig. 4A). Only clonal cell lines expressing low levels of FLAG-tagged DN p38 MAPK were obtained (Fig. 4A), possibly due to toxicity at higher levels of DN p38 MAPK expression. As shown in Fig. 4B, expression of wild-type p38 MAPK failed to augment unstimulated or EGF-stimulated chemotactic migration. In contrast, expression of the DN p38 MAPK cDNA markedly reduced migration on type I collagen, when cells were either unstimulated (Control) or stimulated by a gradient of EGF (EGF) (Fig. 4B). Unstimulated migration of the X2C1 transfectants and the X2C2 DN p38 MAPK cotransfectants was significantly less than the unstimulated migration of the X2C2 transfectants (p Ͻ 0.008). EGF-stimulated chemotaxis of the X2C1 transfectants and the X2C2 DN p38 MAPK cotransfectants was significantly less than the EGF-stimulated chemotaxis of the X2C2 transfectants (p Ͻ 0.01). These results suggest that the p38 MAPK pathway was necessary for ␣ 2 integrin cytoplasmic domain-dependent migration on type I collagen in the absence and presence of EGF. However, increased levels of wild-type p38 MAPK protein alone failed to augment migration. The failure of wild-type p38 MAPK protein to increase migration suggests that activation of the p38 MAPK but not simply increased protein expression was necessary to stimulate cell migration.
The upstream signals required for p38 MAP kinase activation include activation and phosphorylation of a cascade of protein kinases including SEK (31)(32)(33). In some systems, SEK is activated by the small G protein Rac1 (34). Since Rac1 is Migration through the pores of the filter was either unstimulated (Control) or stimulated with EGF (10 ng/ml) (EGF). Cell migration proceeded for 5 h in a 5% CO 2 humidified chamber at 37°C. The number of cells attached to the lower surface of the transwell filter was quantitated microscopically. Results are presented as the mean Ϯ S.E.M. of at least three separate experiments. B, the X2C2 and (C) X2C1 transfectants were treated as described above and plated on transwell filters coated with either type I collagen or with fibronectin (25 g/ml), and migration was either unstimulated (Control) or stimulated with EGF (10 ng/ml) (EGF). Results are presented as mean Ϯ S.E.M. of at least three separate experiments. D, the X2C2 and X2C1 transfectants expressed similar levels of the EGF receptor and phosphorylated EGF receptor to a similar extent. The X2C2 and X2C1 transfectants were serum-starved for 72 h, either treated or not with EGF (10 ng/ml) for 5 min, and lysed. Cell lysates were immunoprecipitated using the polyclonal anti-EGF antibody. Immunoprecipitated proteins were evaluated by SDS-PAGE and immunoblotted using the anti-phosphotyrosine antibody PY99 followed by the anti-EGF receptor antibody (EGF-R). One of three replicate experiments is shown. responsible for stimulating invasion of malignant breast cancer and colon cancer cell lines, the role of Rac1 in ␣ 2 -cytoplasmic domain-mediated migration was evaluated (29,30). The X2C2 transfectants were cotransfected with either HA-tagged constitutively active Rac1(Val-12) or dominant negative Rac1(Asn-17) cDNA constructs and the Selecta Vecta-Hyg plasmid, and clonal cell lines were selected in geneticin plus hygromycin (Fig. 5A). The unstimulated and EGF-stimulated migration of clonal cell lines expressing constitutively active or dominant negative Rac1 was evaluated. Constitutively active Rac1(Val-12) significantly augmented unstimulated migration (p Ͻ 0.05) and increased EGF-stimulated chemotaxis (although not statistically significant) of the X2C2 transfectants on type I collagen (Fig. 5B). In contrast, dominant negative Rac1(Asn-17) failed to inhibit unstimulated migration but significantly decreased (p Ͻ 0.05) chemotactic migration in response to EGF when compared with the X2C2 transfectants (Fig. 5B). Expression of dominant negative Rac1(Asn-17) reduced p38 MAPK phosphorylation following adhesion to type I collagen (Fig. 5C). The lack of a significant diminution in unstimulated migration is likely due to the very low level of migration in the absence of EGF. Constitutively active Rac1(Val-12) and dominant negative Rac1(Asn-17) failed to alter migration on fibronectin (data not shown).
To address the ability of Rac1(Val-12) to rescue X2C1 migration, the X2C1 transfectants were cotransfected with HAtagged, constitutively active Rac1(Val-12) and Selecta Vecta-Hyg plasmid, and cells expressing high levels of HA-Rac1(Val-12) were selected in geneticin and hygromycin (Fig. 5A). Expression of Rac1(Val-12) rescued the ability of the X2C1 transfectants to migrate on type I collagen in the presence or absence of EGF, as shown in Fig. 6A. Unstimulated and EGFstimulated migration of the X2C2 transfectants and the X2C1Rac1(Val-12) cotransfectants was significantly greater than unstimulated or EGF-stimulated migration of the X2C1 transfectants (p Ͻ 0.005 or p Ͻ 0.02, respectively). In fact, the X2C1 Rac1(Val-12) cotransfectants migrated in a manner similar to the X2C2 transfectants in the presence or absence of EGF. Migration of the X2C1 Rac1(Val-12) cotransfectants, as well as the X2C2 transfectants was completely inhibited by addition of SB203580, the p38 MAPK inhibitor (Fig. 6B). The X2C1 Rac1(Val-12) cotransfectants were evaluated for their ability to phosphorylate p38 MAPK following adhesion to type I collagen. Expression of Rac1(Val-12) augmented p38 MAPK phosphorylation by the X2C1 transfectants (Fig. 6C). These findings suggest that signals downstream of the ␣ 2 , but not the ␣ 1 , integrin cytoplasmic domain stimulated migration via the p38 MAPK pathway. Rac1(Val-12) rescued the ability of the X2C1 transfectants to migrate in both an unstimulated and EGF-stimulated manner and to phosphorylate p38 MAPK. Rac1(Val-12)-stimulated migration was inhibited by the p38 MAPK-specific inhibitor.
Since Rac1 activates both p38 MAPK and the c-Jun N-terminal kinase (JNK), we evaluated the short term time course of phosphorylation of JNK resulting from adhesion to type I collagen. Following serum starvation, the X2C2 and X2C1 transfectants were plated on type I collagen (25 g/ml) in the absence and presence of EGF (10 ng/ml) and lysed at defined time points. Adhesion of the X2C2 transfectants to type I collagen resulted in rapid and robust phosphorylation of JNK (Fig. 7A). EGF augmented the phosphorylation of JNK at all time points. Adhesion of the X2C1 transfectants stimulated weak phosphorylation of JNK that was only slightly augmented by EGF. To discern whether JNK activation was in part responsible for ␣ 2 -cytoplasmic domain-dependent migration, the X2C2 transfectants were cotransfected with the dominant negative (DN) JNK cDNA construct, and the Selecta Vecta-Hyg plasmid and cells overexpressing DN JNK were selected in geneticin and hygromycin. X2C2 cotransfectants expressing DN JNK at levels (Fig. 7B) comparable to the level of DN p38 MAPK kinase expressed by the X2C2 DN p38 MAPK cotransfectants migrated at a rate similar to the X2C2 transfectants (Fig. 7C). Expression of DN JNK failed to alter significantly EGF-stimulated ␣ 2 -cytoplasmic domain-dependent migration, indicating that Rac1 activation of p38 MAPK, but not JNK, is required for EGF-stimulated chemotaxis of the X2C2 transfectants. DISCUSSION Data from a number of laboratories have suggested that although the ␣ 1 ␤ 1 and ␣ 2 ␤ 1 integrins are structurally similar and both function as adhesion receptors for collagens and/or laminins, the two receptors mediate profound differences on cell phenotype (11)(12)(13)(14)(15)(16)(17)(18)(19)(20)(21)(22)(35)(36)(37). The results reported here demonstrate that the ␣ 2 , but not the ␣ 1 , integrin cytoplasmic domain mediated signals, via the p38 MAP kinase pathway, that lead to a migratory phenotype. Expression of a constitutively active small G protein, Rac1, augmented p38 MAP kinase phosphorylation and migration of the X2C2 transfectants in the absence and presence of EGF. In addition, expression of constitutively active Rac1 by the X2C1 transfectants restored activation of the p38 MAP kinase pathway and a migratory phenotype. Thus, the activation of the p38 MAP kinase pathway that occurs upon adhesion to collagen is uniquely dependent on the ␣ 2 domain, but not the ␣ 1 -cytoplasmic domain, and is required for cell migration, either unstimulated or stimulated by EGF.
Our earlier work, using complementary "gain-of-function" and "loss-of-function" models indicated that the ␣ 2 ␤ 1 integrin was required for epithelial morphogenesis and branching in three-dimensional collagen gels (15, 38 -40). The ability of epithelial cells to branch and to form glandular structures was not supported by the ␣ 1 ␤ 1 integrin. These results were the first to suggest that these two receptors mediate distinctly different cell phenotypes when expressed in epithelial cells. Other laboratories (17,18) demonstrated that, when cultured in threedimensional collagen gels, fibroblasts utilized the ␣ 1 ␤ 1 integrin to down-regulate collagen gene expression but utilized the ␣ 2 ␤ 1 integrin to up-regulate matrix metalloproteinase gene expression. The ␣ 1 ␤ 1 and ␣ 2 ␤ 1 integrins have also been shown to mediate distinctly different signals that promote cell cycle progression (20,21). 2 Wary et al. (20,21) demonstrated that the ␣ 1 ␤ 1 integrin was among a subset of integrin receptors, includ- FIG. 4. Requirement of p38 MAPK for ␣ 2 -cytoplasmic domain-dependent migration. A, the X2C2 transfectants were cotransfected with either the FLAG-tagged wild-type (WT) or the FLAG-tagged dominant negative p38 MAP kinase cDNA construct and the Selecta Vecta-Hyg plasmid. Following clonal selection in geneticin plus hygromycin, expression of the FLAG-tagged epitope was evaluated by immunoblot analysis. Clonal cell lines expressing increased levels of either wild-type or DN p38 MAPK were selected for study. B, migration assays of X2C2, X2C1 transfectants, or wild-type or DN p38 MAPK-expressing X2C2 cotransfectants were conducted as described above. Results are presented as the mean Ϯ S.E.M. of at least three separate experiments. Unstimulated migration of the X2C1 transfectants and the X2C2 DN p38 MAPK-expressing clonal cell lines 4, 5, and 12 were statistically reduced in comparison to the X2C2 transfectants, with a p Ͻ 0.008 (*). EGF-stimulated chemotaxis of the X2C1 transfectants and the X2C2 DN p38 MAPKexpressing clonal cell lines 4, 5, and 12 were statistically reduced in comparison to the X2C2 transfectants with a p Ͻ 0.01 (**).
ing ␣ 5 ␤ 1 , ␣ v ␤ 3 , and ␣ 6 ␤ 4 , that associated with shc to activate the Ras/MAP kinase pathway. Caveolin-1 functions as a membrane adaptor to couple this subset of integrins via the transmembrane domain of the ␣ integrin subunit to the shc pathway. In the studies of Wary et al. (20,21) and in our own unpublished studies, 3 adhesion of cells expressing the ␣ 2 ␤ 1 integrin does not result in shc phosphorylation. Under similar conditions, adhesion of cells expressing the ␣ 1 ␤ 1 integrin produced shc phosphorylation. 3 These results indicate that both the ␣ 1 ␤ 1 and the ␣ 2 ␤ 1 integrins mediate downstream signals but do so by distinctly different mechanisms. Our results suggest that the ␣ 2 integrin subunit utilizes the cytoplasmic domain for signaling specificity that results in activation of the p38 MAP kinase pathway.
The ability of different integrin ␣and ␤-subunit cytoplasmic domains to respond to different inside-out or outside-in signaling pathways is poorly understood. Here we demonstrate that EGF-stimulated migration on type I collagen requires ␣ 2 -cytoplasmic domain-dependent p38 MAP kinase activation and not p42/44 MAP kinase activation. EGF failed to stimulate ␣ 5 ␤ 1 -  5. Rac1(Val-12)-stimulated ␣ 2cytoplasmic domain-dependent migration. A, the X2C2 and X2C1 transfectants were cotransfected with either HAtagged constitutively active Rac1(Val-12) or HA-tagged dominant negative Rac1(Asn-17) cDNA construct, and the Selecta Vecta-Hyg construct and clonal cell lines of the X2C2 transfectants were selected in geneticin plus hygromycin. Expression of the HA-tagged epitope was evaluated by immunoblot analysis. Clonal cell lines expressing elevated levels of Rac1(Val-12) or Rac1(Asn-17) were selected for further study. B, migration of the X2C2 and X2C1 transfectants and X2C2 Rac1(Val-12) 1, X2C2 Rac1(Val-12) 10, X2C2 Rac1(Asn-17) 3, and X2C2 Rac1(Asn-17) 4 cotransfectants was stimulated with either no additional growth factor (Control) or EGF (EGF) (10 ng/ml). Results are presented as the mean Ϯ S.E.M. of at least three separate experiments. Unstimulated migration of the X2C2 Rac1(Val-12) cotransfectants 1 and 10 was significantly greater than that of the X2C2 transfectants with a p Ͻ 0.05 (*). EGF-stimulated chemotaxis of the X2C1 transfectants and X2C2 Rac1(Asn-17) cotransfectants 3 and 4 was significantly less than that of the X2C2 transfectants with a p Ͻ 0.05 (*). C, following serum starvation for 72 h, the X2C2 transfectants and the X2C2 Rac1(Asn-17) cotransfectants were either lysed (time ϭ 0) or plated on type I collagen in the absence (Ϫ) or presence (ϩ) of EGF (10 ng/ ml). Following adhesion to type I collagen, the cells were lysed at defined time points, including 0.25, 1, 3, and 6 h. Immunoblot analysis of cell lysates was carried out using the polyclonal anti-phospho-p38 MAP kinase antibody (P-p38) followed by polyclonal anti-total p38 MAP kinase antibody (Total p38). One of greater than three replicate experiments is shown. mediated migration on fibronectin. R-Ras-stimulated migration has a number of similarities with EGF-stimulated migration (42). The effect of R-Ras was dependent on the ␣ 2 -and not the ␣ 5 -cytoplasmic domain. In addition, R-Ras-stimulated migration was independent of MEK/MAPK kinase activation. MEK and MAPK kinase activation has been implicated in cell migration of a number of cell types (43), and EGF stimulation of the EGF receptor activates the p42/44 MAP kinase (44,45). Initially, we were somewhat surprised that EGF-stimulated ␣ 2 -cytoplasmic domain-dependent migration was independent of the p42/44 MAP kinase pathway. These results suggest the potential cross-talk between the R-Ras pathway and the EGFstimulated ␣ 2 -cytoplasmic domain-dependent migration.
During completion of the study reported here, Ivaska et al. (22) reported that ␣ 2 ␤ 1 integrin-mediated up-regulation of collagen gene expression in three-dimensional collagen gels was mediated by p38 MAP kinase and required the ␣ 2 -cytoplasmic domain. Our findings support an important and unique role for the ␣ 2 -cytoplasmic domain in mediating activation of p38 MAP kinase activation. Our findings extend the earlier studies to include a role for p38 MAP kinase in mediating cellular migra-tion on two-dimensional collagen substrates. Our results demonstrate that the collagen-␣ 2 ␤ 1 integrin interaction is sufficient to mediate p38 MAP kinase phosphorylation. Changes in cell shape are not sufficient for p38 MAP kinase phosphorylation and activation since both the X2C2 and X2C1 transfectants spread in a similar manner.
One attractive model by which the ␣ 2 -tail may activate p38 MAPK places the ␣ 2 -cytoplasmic domain upstream of the small G proteins Cdc42 and/or Rac1. Ivaska et al. (22) suggested that Cdc42 was responsible for ␣ 2 -cytoplasmic domain-stimulated p38 MAP kinase phosphorylation. Since both Cdc42 and Rac1 have been shown to activate the p38 MAP kinase pathway and stimulate carcinoma cell invasion and migration (29,30), we evaluated the role of both Cdc42 and Rac1 in ␣ 2 ␤ 1 integrin-dependent migration on collagen. Migration studies using a series of stable cell lines expressing either constitutively active or dominant negative Cdc42 or Rac1 identified Rac1 as an important mediator of cell migration. In contrast to the findings of Ivaska et al. (22) expression of either constitutively active or dominant negative Cdc42 failed to significantly alter ␣ 2 -cytoplasmic domain-dependent unstimulated or EGF-stimulated FIG. 6. Migration and p38 MAPK phosphorylation of X2C1 Rac1(Val-12) cotransfectants. A, migration of the X2C2 and X2C1 transfectants and X2C1 cotransfectants expressing constitutively active Rac1(Val-12) (X2C1 Rac1(V12)) was evaluated. Results are presented as the mean Ϯ S.E.M. of at least three separate experiments. Unstimulated migration and EGF-stimulated migration of the X2C2 transfectants and X2C1 Rac1(Val-12) cotransfectants was significantly greater than unstimulated and EGFstimulated migration of the X2C1 transfectants with a p Ͻ 0.005 (*) and p Ͻ 0.02 (**), respectively. B, following serum starvation, the X2C2, X2C1, and X2C1 Rac1(Val-12) transfectants were incubated for 15 min at 37°C in either media alone or media containing SB203580 (SB) (10 M) (p38 MAP kinase-specific inhibitor). Following incubation, the cells were plated in either media alone or media containing inhibitor on the upper surface of a transwell filter coated with type I collagen (25 g/ml). Migration through the transwell filter was either unstimulated, stimulated with EGF, or stimulated with EGF (10 ng/ml) plus inhibitor. Following migration through the transwell filter the number of cells attached to the lower surface of the transwell filter was quantitated microscopically. Results are presented as the mean Ϯ S.E.M. of at least three separate experiments. C, following serum starvation for 72 h, the X2C2 and X2C1 transfectants and X2C1 Rac1(Val-12) cotransfectants were either lysed (time ϭ 0) or plated on type I collagen in the absence (Ϫ) or presence (ϩ) of EGF (10 ng/ml). Following adhesion to type I collagen the cells were lysed at defined time points, including 0.25, 1, 3, and 6 h. Immunoblot analysis of cell lysates was carried out using the polyclonal antiphospho-p38 MAP kinase antibody (P-p38) followed by polyclonal anti-total p38 MAP kinase antibody (Total p38). One of three replicate experiments is demonstrated.
migration (data not shown). On the other hand, expression of constitutively active Rac1 augmented migration and activation of p38 MAP kinase. Furthermore, expression of dominant negative Rac1 greatly diminished the EGF-stimulated chemotaxis. Dominant negative Rac1 had no effect on low level, unstimulated ␣ 2 -cytoplasmic domain-dependent migration. In the al-ternative model to that proposed by Ivaska et al. (22), Rac1 would act, not as a downstream effector of the ␣ 2 -cytoplasmic domain to p38 MAPK, but in a parallel pathway that also activates p38 MAPK. In this case, Rac1 synergizes with the ␣ 2 -cytoplasmic domain to stimulate migration of epithelial cells in a manner similar to that by which lysophosphatidic acid FIG. 7. The ␣ 2 -cytoplasmic domain and activation of JNK. A, following serum starvation for 72 h, the X2C2 and X2C1 transfectants were either lysed (time ϭ 0) or plated on type I collagen in the absence (Ϫ) or presence (ϩ) of EGF (10 ng/ml). Following adhesion to type I collagen, the cells were lysed at defined time points, including 0.25, 1, and 3 h. Immunoblot analysis of cell lysates was carried out using the polyclonal antiphospho-JNK antibody (P-JNK) followed by polyclonal anti-total JNK antibody (Total JNK). One of three replicate experiments is demonstrated. B, the X2C2 transfectants were cotransfected with DN JNK cDNA construct and the Selecta Vecta-Hyg plasmid. Following selection in hygromycin and geneticin, expression of DN JNK by a nonclonal cell line was evaluated by immunoblot analysis, as shown. C, migration assays of X2C2 and X2C1 transfectants and the DN JNK-expressing X2C2 cotransfectants were conducted as described above. Results are presented as the mean Ϯ S.E.M. of at least three separate experiments. Unstimulated migration of the X2C1 transfectants but not the X2C2 DN-JNK-expressing cell lines was reduced in comparison to the X2C2 transfectants. EGF-stimulated chemotaxis of the X2C1 transfectants but not the X2C2 DN-JNK-expressing cotransfectants was reduced in comparison to the X2C2 transfectants. activation of Rho synergizes with EGF and platelet-derived growth factor to stimulate migration of fibroblasts (46).
Cooperative signals from the extracellular matrix and growth factors are required for cells to progress through the cell cycle and to migrate (47,48). The pathways leading from integrins and growth factor receptors converge at several points in the G 1 phase of the cell cycle (3)(4)(5). 2 In this report, we have described a synergy between signals initiated by EGF and signals generated by the ␣ 2 -subunit cytoplasmic domain that lead to cell migration. Unstimulated and EGF-stimulated chemotactic migration on type I collagen are supported by the ␣ 2 integrin cytoplasmic domain. Signals initiated by EGF synergized with the ␣ 2 -cytoplasmic domain-dependent phosphorylation of p38 MAP kinase to stimulate chemotactic migration. Cooperative synergy between specific integrin receptors and/or integrin cytoplasmic domains and growth factors has been reported by other investigators (41, 48 -51). In cultures of primary skeletal muscle, overexpression of the ␣ 6 -subunit or the ␣ 5 -extracellular/␣ 6 -cytoplasmic domain subunit (X5C6) inhibited proliferation and supported differentiation in the presence of growth factors or serum. In contrast myoblasts expressing the ␣ 5 integrin subunit or the ␣ 6 -extracellular domain/␣ 5 -cytoplasmic domain subunit (X6C5) proliferated but failed to differentiate in response to serum or growth factors (49,50). In the chorioallantoic membrane model, basic fibroblast growth factor or tumor necrosis factor-␣-induced angiogenesis depends on the ␣ v ␤ 3 integrin, but vascular endothelial cell growth factor, transforming growth factor-␣, or phorbol ester-induced angiogenesis depend on the ␣ v ␤ 5 integrin (51). Vascular endothelial cell growth factor-induced angiogenesis requires members of the src kinase family, whereas basic fibroblast growth factor signals do not (41). Therefore, in angiogenesis, responses to specific growth factors are linked to and are dependent upon the expression of specific integrin receptors.
In summary, the results reported here demonstrate that the ability of the ␣ 2 ␤ 1 integrin to mediate a migratory phenotype requires the cytoplasmic domain of the ␣ 2 -subunit. Signals uniquely mediated by the ␣ 2 -, but not the ␣ 1 , cytoplasmic domain lead to phosphorylation of p38 MAP kinase and Jun kinase. This study is the first to demonstrate the activation of the p38 MAP kinase and not the Jun kinase pathway is necessary for ␣ 2 ␤ 1 integrin-mediated cell migration and EGF-stimulated chemotaxis. The ␣ 1 integrin cytoplasmic domain did not mediate these signals, but migratory activity and phosphorylation of p38 MAPK was rescued by expression of constitutively active Rac1 in the X2C1 transfectants. These results begin to dissect the complex mechanisms by which the ␣ 1 and ␣ 2 integrin cytoplasmic domains, in concert with signals from growth factors, exert profound influences on cell phenotype.