Cross-talk between Insulin Receptor and Integrin α5β1 Signaling Pathways*

The ligation and clustering of cell surface αβ heterodimeric integrins enhances cell adhesion and initiates signaling pathways that regulate such processes as cell spreading, migration, differentiation, proliferation and apoptosis. Here we show that insulin treatment of Chinese hamster ovary cells expressing insulin receptors (CHO-T) markedly promotes cell adhesion onto a fibronectin matrix, but not onto bovine serum albumin or poly-lysine. Incubation of cells with a GRGDSP peptide that specifically binds integrins (but not the nonspecific GRADSP peptide) abolishes this insulin effect, as does the potent phosphoinositide 3-kinase (PI 3-kinase) inhibitor wortmannin. Moreover, a specific blocking monoclonal anti-α5β1 integrin antibody, PB-1, blocks insulin-stimulated cell adhesion onto fibronectin. Conversely, activating α5β1 integrins on CHO-T cells by adherence onto fibronectin markedly potentiates the action of insulin to enhance insulin receptor and insulin receptor substrate (IRS)-1 tyrosine phosphorylation. Activation of α5β1 integrin also markedly potentiates the recruitment of p85-associated PI 3-kinase activity to IRS-1 in response to submaximal levels of insulin in CHO-T cells. These data indicate that insulin potently activates integrin α5β1 mediated CHO-T cell adhesion, while integrin α5β1 signaling in turn enhances insulin receptor kinase activity and formation of complexes containing IRS-1 and PI 3-kinase. These findings raise the hypothesis that insulin receptor and α5β1 integrin signaling act synergistically to enhance cell adhesion.

The ligation and clustering of cell surface ␣␤ heterodimeric integrins enhances cell adhesion and initiates signaling pathways that regulate such processes as cell spreading, migration, differentiation, proliferation and apoptosis. Here we show that insulin treatment of Chinese hamster ovary cells expressing insulin receptors (CHO-T) markedly promotes cell adhesion onto a fibronectin matrix, but not onto bovine serum albumin or poly-lysine. Incubation of cells with a GRGDSP peptide that specifically binds integrins (but not the nonspecific GRADSP peptide) abolishes this insulin effect, as does the potent phosphoinositide 3-kinase (PI 3-kinase) inhibitor wortmannin. Moreover, a specific blocking monoclonal anti-␣ 5 ␤ 1 integrin antibody, PB-1, blocks insulin-stimulated cell adhesion onto fibronectin. Conversely, activating ␣ 5 ␤ 1 integrins on CHO-T cells by adherence onto fibronectin markedly potentiates the action of insulin to enhance insulin receptor and insulin receptor substrate (IRS)-1 tyrosine phosphorylation. Activation of ␣ 5 ␤ 1 integrin also markedly potentiates the recruitment of p85-associated PI 3-kinase activity to IRS-1 in response to submaximal levels of insulin in CHO-T cells. These data indicate that insulin potently activates integrin ␣ 5 ␤ 1 mediated CHO-T cell adhesion, while integrin ␣ 5 ␤ 1 signaling in turn enhances insulin receptor kinase activity and formation of complexes containing IRS-1 and PI 3-kinase. These findings raise the hypothesis that insulin receptor and ␣ 5 ␤ 1 integrin signaling act synergistically to enhance cell adhesion.
The tyrosine kinase activity of the cell surface insulin receptor is required to mediate its many biological actions. Insulin receptor activation promotes the rapid autophosphorylation of its ␤ subunit, as well as tyrosine phosphorylation of proteins involved in insulin signaling, such as insulin receptor substrate (IRS) 1 and Shc proteins (1)(2)(3)(4). Insulin-mediated phosphorylation of these proteins is thought to provide tyrosine phosphate docking sites for the recruitment of signaling proteins containing Src homology 2 domains (SH2) (1,5,6). One SH2 domaincontaining family of proteins which associates with IRS proteins in response to insulin are isoforms of the p85 regulatory subunit of the p110-type PI 3-kinases (1)(2)(3)(4)7). PI 3-kinase activity in such signaling complexes appears to be required for insulin action on many cellular processes, including glucose transport (8,9), glycogen synthesis (9, 10), stress fiber breakdown (11), and membrane ruffling (12). Thus, inhibition of PI 3-kinase activity by wortmannin or disruption of PI 3-kinase recruitment to IRS proteins by dominant inhibitory constructs of p85 subunits ablate the actions of insulin on these processes (11)(12)(13). Our understanding of the downstream elements that mediate the action of the 3Ј-phosphoinositide products of the PI 3-kinases is incomplete, but appear to include protein kinases such as PDK1 and Akt/protein kinase B (PKB) (14), a family of proteins containing Sec7 homology domains (15), and the zinc finger containing protein EEA1 (16).
Insulin action also promotes dephosphorylation of tyrosine phosphates on such proteins as focal adhesion kinase (FAK) and paxillin (17)(18)(19), thought to be involved in cell regulation by integrins. Integrins are ␣␤ heterodimeric transmembrane receptors that mediate interactions between the cell surface and the extracellular matrix and also initiate signaling events (20 -22), including tyrosine phosphorylation of FAK and paxillin (23)(24)(25), cytoskeletal reorganization (20 -22, 25), activation of mitogen-activated protein kinase cascades (26), and regulation of gene expression (22,27). These biological actions of integrins can overlap with those of growth factors, and there is evidence that signaling pathways initiated by integrins synergize functionally with those triggered by growth factors (22,28). Thus, cell adhesion has been shown to greatly enhance autophosphorylation of epidermal growth factor and plateletderived growth factor (PDGF) receptors (25,29). This in turn potentiates the action of these growth factors in activating mitogen-activated protein kinases, PI 3-kinase, and the downstream protein kinases PDK1 and Akt/PKB (29,30). The mitogenic effects of insulin have also been shown to be enhanced by interactions of extracellular vitronectin with cell surface ␣v␤3 integrins, which associate with insulin receptor and IRS-1 in response to insulin (31,32). However, the mechanism of such synergism in the actions of insulin and integrins is still unclear.
Conversely, it also has been shown that growth factor receptors, such as KIT (33) and PDGF (34) receptors, stimulate integrin-mediated cell adhesion onto fibronectin through a PI 3-kinase-dependent pathway. Thus, in the present studies, we addressed the questions whether insulin may directly activate integrin-mediated cell adhesion and if integrin activation modulates insulin signaling. Here, we show that insulin markedly promotes CHO-T cell adhesion onto a fibronectin matrix by a mechanism that is mediated by ␣ 5 ␤ 1 integrin. Activation of this integrin in turn enhances insulin receptor and IRS-1 tyrosine phosphorylation, as well as recruitment of PI 3-kinase activity to IRS-1 in response to insulin. This cross-talk between insulin and integrin receptor pathways is likely to play an important role in biological processes regulated by insulin and which depend on integrin engagement.

EXPERIMENTAL PROCEDURES
Materials-Anti-phosphotyrosine (anti-Tyr(p)) mouse monoclonal 4G10, anti-p85 polyclonal, and laminin were purchased from Upstate Biotechnology. Rabbit polyclonal anti-IRS-1 immunoglobulin used for immunoprecipitation was prepared as described previously (35). Antiinsulin receptor monoclonal (CT-1) and polyclonal antibodies were from mouse ascites and from Santa Cruz Biotechnology, respectively. Plasma human fibronectin was from Boehringer. GRGDSP and GRADSP peptides for the adhesion competition assay were from Life Technologies, Inc. Specific anti ␣ 5 ␤ 1 integrin receptor monoclonal antibody, used in CHO-T cell adhesion blocking assay, was a gift from Dr. Rudolph Juliano. [␥-32 P]ATP and 35 S-labeled protein labeling mix were from NEN Life Science Products. Phosphatidylinositol (PI) was from Avanti Polar Lipids.
Cell Culture-CHO-T cells were maintained in Ham's F-12 medium, 10% fetal bovine serum, and 50 g/ml streptomycin/penicillin and grown to confluence before use.
Cell Adhesion Assay-To assay cell adhesion on different matrices, cell culture dishes (12-well plate) were coated with fibronectin (0.5 g/ml), laminin (10 g/ml), and poly-lysine (0.5 g/ml) at 4°C overnight and blocked with 0.1% bovine serum albumin (BSA) in phosphatebuffered saline (PBS) for 1 h, prior to plating the cells. Confluent CHO-T cells were serum starved for 12 h in F-12 serum-free medium containing 0.5% BSA and labeled with 35 S-labeled protein labeling mix (1 Ci/150 cm 2 plate) at 37°C and washed twice with PBS. Cells labeled with 35 S were detached by adding EDTA-trypsin (0.05 mM, 0.025%). Cells were held in suspension for 5 min, and additions were made for the times indicated in the figure legends. Cells were plated in dishes coated with different substrata, as indicated. Nonadhered cells were removed, and attached cells were washed twice with PBS containing 25 mM MgCl 2 . Adhered cells were quantified by adding 0.5 ml SDS (0.1%), and samples were counted by a ␤-counter.
Cell Lysis, Immunoprecipitation, and Immunodetection-Cells held in suspension or attached on a substrata were lysed by adding detergent lysis buffer (50 mM HEPES, pH 7.5, 100 mM NaF, 10 mM NaPP i , 2 mM Na 3 VO 4 , 1% Nonidet P-40, 1 mM phenylmethylsulfonyl fluoride, 10 g/ml leupeptin, 10 g/ml aprotinin) and kept on ice for 20 min. Detergent lysates were precleared by centrifugation at 14,000 ϫ g for 15 min in the cold, and protein concentration was determined by the Bicinchoninic Acid (BCA) protein assay protocol from Pierce. Appropriate antibodies were then added to the cleared cell lysate standardized for total cell protein, and the lysates were incubated overnight at 4°C by endover-end mixing. Protein A-Sepharose was added and the samples incubated for 2 h. Immunoprecipitated proteins were then washed four times in PBS with 1% Nonidet P-40, separated by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE), and transferred to nitrocellulose membranes for immunoblotting with appropriate antibody. Bound primary monoclonal or polyclonal antibodies were visualized by the Renaissance Chemiluminescence Detection System (NEN Life Science Products) with horseradish peroxidase-conjugated detection antibody at 1:15,000 dilution. The tyrosine phosphorylation levels of immunoprecipitated insulin receptor and IRS-1 were quantified, where indicated, by scanning densitometry.
Assay of IRS-1-associated PI 3-Kinase Activity-IRS-1-associated PI 3-kinase activity assays were performed in vitro by immunoprecipitation of IRS-1 from total cell lysates as described above. The IRS-1 immunoprecipitates were washed three times in PBS, 1% Nonidet P-40, twice in buffer containing 10 mM Tris-HCl, pH 7.5, 100 mM NaCl, 1 mM EDTA, and once in PI 3-kinase assay buffer (20 mM Tris-HCl, pH 7.5, 100 mM NaCl, 0.1 mM EGTA, 10 mM MgCl 2 ). The IRS-1 immunoprecipitates were resuspended in 0.2 ml of PI 3-kinase assay buffer containing 50 g of PI and [␥-32 P]ATP (100 Ci at final concentration of 0.05 mM), and the reaction was incubated for 20 min at room temperature and quenched with HCl 1 N. Phospholipids were extracted with chloroform:methanol (1:1), washed twice with methanol:HCl 1 N (1:1), and resolved by thin layer chromatography (TLC), and the radioactive products were quantified as described previously (35).

RESULTS AND DISCUSSION
Insulin Stimulates Cell Adhesion Mediated by Integrins-To determine whether insulin signaling can activate integrin-mediated cell adhesion, CHO-T cells were detached from culture dishes, treated with 100 nM insulin for 10 min at 37°C, and plated onto dishes coated with bovine serum albumin, polylysine, or fibronectin. As shown in Fig. 1A, adhesion of CHO-T  35 S-labeled and serum-free CHO-T cells were detached and incubated at 4°C with 10 g/ml of anti-␣ 5 ␤ 1 integrin blocking monoclonal antibody (PB-1). After 1 h of incubation, cells were stimulated (ϩ) or not (Ϫ) with 100 nM insulin for 10 min at 37°C and replated on fibronectin (0.5 g/ml) or laminincoated (5 g/ml) dishes, as indicated, and allowed to adhere for 10 or 20 min, respectively. Data presented are the average values of three independent experiments Ϯ S.E. cells onto a fibronectin matrix was stimulated 2-3-fold by prior treatment with insulin. A similar result was observed when a laminin matrix was used (Fig. 2, right). Nonspecific adhesion processes were not found to be affected by insulin since insulin treatment did not have any effect on CHO-T cell adhesion to dishes coated with bovine serum albumin or poly-lysine. Thus, insulin can modulate CHO-T cell adhesion, similar to the effects described for other growth factor receptors, such as those that bind PDGF, and KIT (33,34).
We next investigated whether the action of insulin to increase cell adhesion to fibronectin was mediated by integrins. Thus, the effect of a specific GRGDSP sequence peptide, which binds to integrins (36), on insulin-stimulated CHO-T cell adhesion to fibronectin was examined. As seen in Fig. 1B, the addition of 0.5 mM GRGDSP peptide to assay media abolished basal and insulin-stimulated cell adhesion to fibronectincoated dishes, whereas an inactive GRADSP peptide at the same concentration did not. Thus, these data suggest that insulin action modulates integrin-mediated CHO-T cell adhesion.
It has previously been established that CHO cells adhere to fibronectin by ␣ 5 ␤ 1 integrin receptors (37). We thus examined the effect of a monoclonal anti-␣ 5 ␤ 1 blocking antibody, PB-1 (37-39), on CHO-T cell adhesion to fibronectin in the presence and absence of insulin. As seen in Fig. 2 (right), addition of PB-1 to assay media markedly inhibited the adhesion of CHO-T cells to a fibronectin matrix under both basal and insulinstimulated conditions. In contrast, no inhibition of CHO-T cell adhesion onto laminin could be detected by adding PB-1 to the assay media (Fig. 2, left). These data are consistent with a previous report showing PB-1 inhibition of CHO cell adhesion onto fibronectin, but not onto laminin (37). Taken together, the data in Figs. 1 and 2 demonstrate a marked effect of insulin to enhance CHO-T cell adhesion onto fibronectin through ␣ 5 ␤ 1 integrin.
Insulin-stimulated Cell Adhesion Is Inhibited by Wortmannin-To determine whether insulin-stimulated and integrinmediated cell adhesion is dependent on activated PI-3-kinase, CHO-T cells were detached from the culture dishes, treated with or without 50 nM of the PI 3-kinase inhibitor wortmannin for 15 min prior to incubation with or without insulin for 10 min, and then replated onto fibronectin or laminin-coated plates. As depicted in Fig. 3, wortmannin treatment of the cells markedly inhibited insulin-stimulated, fibronectin-dependent CHO-T cell adhesion. Similar results were observed when cells were plated onto laminin-coated dishes (Fig. 3). These results suggest that insulin-stimulated cell adhesion in this system is dependent, at least in part, on PI 3-kinase activity. The mechanism of this insulin action on cell adhesion may thus share common elements with the actions of growth factors such as PDGF on adhesion, which are also blocked by wortmannin (33,34,40).

Integrin-mediated CHO-T Cell Adhesion Increases Insulin Receptor and IRS-1 Tyrosine Phosphorylation in
Response to Insulin-Attachment of cells to fibronectin results in increased tyrosine phosphorylation of FAK (M r ϭ 125,000) and paxillin (M r ϭ 68,000 -75,000) proteins that are associated with focal adhesion complexes (23)(24)(25). Also, it has been shown that integrin-mediated cell anchorage promotes increases in tyrosine phosphorylation of epidermal growth factor and PDGF receptors (28,29). To examine whether the ␣ 5 ␤ 1 integrin engagement modulates insulin receptor and IRS-1 tyrosine phosphorylation in response to insulin, we performed experiments in which serum-starved CHO-T cells were detached from the culture dish, held in suspension for 30 min at 37°C, treated with or without 100 nM insulin for 10 min, and then either kept

FIG. 4. Adhesion onto fibronectin induces tyrosine phosphorylation of proteins in CHO-T cells.
Serum-starved CHO-T cells were detached, held in suspension for 30 min at 37°C and then stimulated (ϩ) or not (Ϫ) with 1 M insulin for 10 min, replated on fibronectin-coated (2.5 g/ml) dishes or held in suspension for 20 min. Suspension and adherent cells were lysed, total proteins (25 g) were resolved by SDS-PAGE on 7.0% gels, eletrophoretically transferred to nitrocellulose, blocked, and subsequently incubated with anti-p-Tyr antibody, as described under "Experimental Procedures." A, anti-p-Tyr immunoblotting of total proteins from cells in suspension or adhered onto fibronectin. Arrowheads indicate bands corresponding to insulin receptor (IR, 95 kDa), IRS-1 (175 kDa), and focal adhesion kinase (FAK, 125 kDa). B, the data shown in panel A for insulin receptor, IRS-1 and FAK tyrosine phosphorylation from suspension cells (Ϫ) or adhered on fibronectin (FN, ϩ) were quantified using a scanning densitometer. 35 S-labeled and serum-free CHO-T cells were detached, treated (ϩ) or not (Ϫ) with 50 nM wortmannin (WT) for 15 min at 37°C and stimulated (ϩ) or not (Ϫ) with 100 nM insulin for 10 min at 37°C. Cells were than replated on fibronectin or laminin-coated dishes and allowed to adhere as described in the legend of Fig. 2. Data presented are average values from five independent experiments Ϯ S.E. in suspension or plated onto fibronectin-coated plates for 20 min. As shown in Fig. 4A and B, CHO-T cell attachment to fibronectin markedly increased tyrosine phosphorylation of the 125-kDa protein that corresponds to FAK as well as 68-kDa proteins. Surprisingly, cell attachment to fibronectin also markedly increased insulin receptor and IRS-1 tyrosine phosphorylation under basal conditions and in response to insulin by 2-3-fold (Fig. 4, A and B). Fig. 4 also shows that insulin treatment of these cells caused dephosphorylation of FAK, consistent with previous reports that insulin induces FAK tyrosine dephosphorylation (17)(18)(19).

FIG. 3. Wortmannin blocks insulin-stimulated CHO-T cell adhesion onto fibronectin and laminin.
To examine whether ␣ 5 ␤ 1 integrin engagement potentiates signaling by insulin at submaximal doses, the effect of different concentrations of insulin on insulin receptor and IRS-1 tyrosine phosphorylation was assessed in cells held in suspension or allowed to attach onto fibronectin (Fig. 5). CHO-T cell attachment onto a fibronectin matrix markedly potentiated the ability of insulin to enhance tyrosine phosphorylation of the insulin receptor at all concentrations employed. Similar to this effect observed on insulin receptor phosphorylation, IRS-1 tyrosine phosphorylation in response to all doses of insulin tested was enhanced by CHO-T cell attachment onto fibronectin (Fig. 6A).
␣ 5 ␤ 1 Integrin Engagement Potentiates Insulin Stimulation of IRS-1-associated PI-3-Kinase Activity-The increase in insulinstimulated tyrosine phosphorylation of IRS-1 by CHO-T cell adhesion onto fibronectin suggests increased binding of signaling proteins such as PI 3-kinase to IRS-1 may be elicited (1)(2)(3)(4)7). Thus, we tested the effect of ␣ 5 ␤ 1 integrin engagement on the association of the p85 subunit of PI 3-kinase with IRS-1. As seen in Fig. 6A, CHO-T cell adhesion onto fibronectin promotes the increased association of p85 with IRS-1 at both submaximal and maximal concentrations of insulin.
Experiments were also conducted to determine the PI 3-kinase activity associated with IRS-1 after treatment of adherent and nonadherent CHO-T cells with or without various concentrations of insulin. Fibronectin engagement of ␣ 5 ␤ 1 integrin receptor was associated with a significant, severalfold increase in basal CHO-T cell PI 3-kinase activity detected in IRS-1  1, 1, 10, 100, 1000 nM) for 10 min, replated on fibronectin-coated dishes, or held in suspension for 20 min. Suspension and adherent cells were lysed, and samples were incubated with anti-IRS-1 for immunoprecipitation and subsequent immunoblotting using anti-p-Tyr and anti-p85 antibodies or assayed for PI 3-kinase activity. A, the upper panel shows anti-p-Tyr immunoblots of IRS-1 immunoprecipitates from suspension or adherent cells treated with different doses of insulin. The lower panel shows anti-p85 immunoblots from the same membranes after stripping. Arrowheads indicate IRS-1 and p85 in upper and lower panels, respectively. B, IRS-1 immunoprecipitates were assayed for PI 3-kinase activity as described under "Experimental Procedures," and PI 3-kinase products were resolved by TLC. Spots corresponding to phosphatidylinositol 3-P (PI 3-P) upon TLC are indicated by arrowhead, and were cut out and quantified using a ␤-counter (C). Open and full bars corresponding to IRS-1-associated PI 3-kinase activity from suspension and adherent cells, respectively. The data presented are the average values from three independent experiments Ϯ S.E. immunoprecipitates (Fig. 6, B and C). Elevated PI 3-kinase activity associated with IRS-1 was also observed when submaximal or maximal doses of insulin were incubated with adherent cells compared with nonadherent cells. These data are consistent with the hypothesis that activation of ␣ 5 ␤ 1 integrin through cell adhesion onto fibronectin causes enhanced recruitment of p85/p110-type PI 3-kinases to IRS-1, resulting in enhanced catalytic activity of these enzymes.
The findings presented here demonstrating the potentiation of insulin receptor modulation of IRS-1 by ␣ 5 ␤ 1 integrin engagement suggests the concommittant enhancement of signaling events downstream of IRS-1. One function proposed to be regulated by IRS-1/PI 3-kinase signaling complexes is cell proliferation (41), and recent evidence indeed indicates that ␣V␤3 integrin-mediated cell adhesion enhances this insulin action (31). As PI 3-kinase activation appears to also be required for the metabolic actions of insulin (8 -10, 13), it will be important in future studies to determine whether integrin ligation may influence such insulin effects. The present work shows for the first time that insulin stimulates cell adhesion, and this effect is dependent upon PI 3-kinase activity (Fig. 3). Cell adhesion is also enhanced upon ligation and clustering of integrins on the cell surface. Thus, our findings are consistent with the hypothesis that potentiation of the insulin signaling pathway through IRS-1/PI 3-kinase by integrins reflects an indirect mechanism to further enhance adhesion over that caused directly by the integrins. Testing this hypothesis will require determining whether IRS-1/PI 3-kinase complexes formed in response to integrin activation actually mediate increased cell adhesion. Such studies will also be important in ultimately probing the full physiological implications of insulin regulation of cell adhesion.