Protein-tyrosine Phosphatase 1B Mediates the Effects of Insulin on the Actin Cytoskeleton in Immortalized Fibroblasts*

Insulin regulates diverse cellular responses including actin reorganization. The mechanism by which insulin induces formation of lamellipodia in cultured cells is not known but is likely to involve activation of Src family protein-tyrosine kinases. Here we show that protein-tyrosine phosphatase 1B (PTPIB) activates Src, thereby initiating the activation of a Rac-dependent pathway leading to cytoskeletal remodeling. Conversely, expression of a proline to alanine (P309,310A) PTP1B mutant, which cannot activate Src, fails to activate Rho GTPases or cause changes in actin organization. Rat fibroblasts lacking PTP1B expression do not activate Src or Rac in response to insulin and cannot reorganize actin. These results show that PTP1B, best known as a negative regulator of the metabolic effects of insulin, is required for the effects of insulin on actin organization in immortalized fibroblasts.

Insulin is a potent signaling molecule that affects glucose metabolism, gene transcription, cytoskeletal structure, and cell motility (1,2). Upon insulin treatment, membrane ruffling is stimulated and stress fibers dissolve, a phenotype associated with motile cells. The molecular signaling pathways by which insulin induces these cytoskeletal alterations are largely unknown, although phosphatidylinositol 3-kinase has been implicated as playing an essential role (3). Such cytoskeletal changes are essential for the effects of insulin on glucose transport and on transcription (1).
The Rho family of GTPases are key regulators of the actin and tubulin cytoskeleton (4). Certain Rho family GTPases, such as Rac, are activated downstream of phosphatidylinositol 3-kinase, and it is likely that the gain of membrane ruffles and the loss of stress fibers that occur upon insulin treatment of cells is related to changes in the activity of these GTPases. Indeed, Nobes et al. (5) have shown that expression of a dominant interfering form of Rac inhibits insulin-mediated membrane ruffling. These findings suggest that insulin regulates the activity of Rho GTPases. However, a direct demonstration of such regulation and a description of the mechanism(s) by which this might occur have not been described.
Protein-tyrosine phosphatase (PTP) 1 1B is an abundant, cytoplasmic protein-tyrosine phosphatase that is a key regulator of insulin signaling. Overexpression of PTP1B impairs insulin signals (6 -10), whereas loss of PTP1B is associated with increased sensitivity to insulin (11)(12)(13). PTP1B binds to and catalyzes the dephosphorylation of the insulin receptor and many of the effects of PTP1B on insulin signaling can be explained on the basis of this interaction (9,14). However, PTP1B also has additional substrates, including the cytosolic proteintyrosine kinases Bcr-Abl (15), Tyk2, and Jak2 (16), the transcription factors STAT5a and 5b (17), and the cytoskeletal protein p130 Cas (6,18,19). The role, if any, of these or other PTP1B substrates in insulin signaling is not known. In addition, the activity of PTP1B is regulated by the insulin receptor (20 -22) indicating that this enzyme may play a more complicated role than that of a simple off-switch for insulin signaling.
We and others have previously shown that overexpression of PTP1B affects the actin cytoskeleton (23,24). When overexpressed in rat 3Y1 cells, PTP1B impedes integrin-stimulated p130 Cas tyrosine phosphorylation, the formation of p130 Cas /Crk complexes, extracellular signal-regulated kinase activation, and cell motility (18,23). Paradoxically, SV40-transformed mouse embryo fibroblasts lacking PTP1B also display defects in integrin signaling; specifically, these cells do not activate Src properly and show delayed p130 Cas phosphorylation, extracellular signal-regulated kinase activation, and cell spreading in response to plating to fibronectin-coated surfaces (25). These findings suggest that the effects of PTP1B on integrin signaling are complex, perhaps reflecting the modulation of the activity of c-Src family kinases (25)(26)(27), p130 Cas (19,23,28), and perhaps other key integrin-regulated proteins such as paxillin (18,23). Based on the phenotype of fibroblasts either over or underexpressing PTP1B, it is apparent that changes in the expression level and/or the activity of this enzyme have profound effects on the cytoskeleton. Because changes in PTP1B expression have been documented in various disease states (29,30), and the catalytic activity of this enzyme has been identified as a target for therapeutics (31,32), it is important to understand the mechanisms by which PTP1B regulates the cytoskeleton.
Here, we assess the role of PTP1B in mediating the cytoskeletal effects of insulin. We show that overexpression of PTP1B stimulates membrane ruffling through activation of Rac1 GTPase and that this activation of Rac1 GTPases by PTP1B is mediated via activation of Src family kinases. Loss of PTP1B expression compromises the ability of insulin to activate Src and Rac1 and to reorganize the actin cytoskeleton. As the insulin receptor can phosphorylate and activate PTP1B (9, 20, 33) these data suggest a spatial and temporal mechanism by which PTP1B regulates insulin-mediated actin reorganization.
Anti-phospho-Src Tyr-416 and Tyr-527 antibodies were purchased from Cell Signaling. Monoclonal anti-PTP1B (FG6) antibody was obtained from Oncogene Research Products. PP1 was obtained from Biomol. Rhodamine-phalloidin was purchased from Molecular Probes.
Cell Culture-Rat 3Y1 cells were grown to 40 -60% confluence in Dulbecco's modified Eagle's medium plus 10% fetal bovine serum. Cells were transfected by a calcium phosphate method with either pJ3H or pJ3H-PTP1B-WT (wild type), denoted as 3Y1-WT, or pJ3H-PTP1B-PA (P309,310A) mutant of PTP1B, denoted as 3Y1-PA, together with a plasmid encoding a puromycin resistance marker (18,34). The cells were selected by growth in 2 g/ml puromycin, and colonies were isolated with cloning cylinders.
F-actin Visualization-Control rat 3Y1, 3Y1-WT, or 3Y1-PA cells were grown on cover slips. Cells were serum-starved followed by treatment with either vehicle or insulin. Cells were then washed with phosphate-buffered saline, fixed in 2% paraformaldehyde for 20 min, permeablized with 0.1% Triton X-100 for 10 min, and stained with rhodamine-phalloidin for 20 min followed by washing with phosphatebuffered saline containing 0.05% Tween 20.
GTPase Pull-down Assay-GST fusion proteins, PBD, and rhotekin were made and purified as described previously (35,36). Serum-starved control rat 3Y1, 3Y1-WT, or 3Y1-PA cells were either treated with vehicle or stimulated with insulin and lysed in 1% Nonidet P-40 lysis buffer (50 mM Tris-HCl, pH 8.0, 137 mM NaCl, 10% glycerol, 1% Nonidet P-40, 50 mM NaF, and 10 mM ␤-glycerol phosphate) containing 1 mM sodium vanadate, 1 mM phenylmethylsulfonyl fluoride, and 10 g/ml aprotinin. Lysate protein concentrations were measured using BCA (Pierce). 500 g of total cell lysate was mixed with 20 l of either GST-PBD or GST-rhotekin and incubated at 4°C for 2 h. The beads were washed three times with Nonidet P-40 lysis buffer and then boiled in SDS sample buffer. The samples were fractionated by SDS-PAGE, transferred to nitrocellulose membranes, and probed with the indicated antibodies.
In Vitro Binding Assay-GST fusion proteins of SH3 domains of p130 Cas and c-Src were expressed in Escherichia coli and isolated by affinity chromatography with glutathione-Sepharose (Amersham Biosciences). 500 g of total cell lysate from either 3Y1-WT or 3Y1-PA cells was incubated with 20 l of GST alone or GST fusion proteins beads at 4°C for 2 h. The beads were washed three times with Nonidet P-40 lysis buffer and then boiled in SDS sample buffer. The samples were fractionated by SDS-PAGE, transferred to nitrocellulose membranes, and probed with indicated antibodies.
Immunoblotting-Cells were lysed in 1% Nonidet P-40 lysis buffer. Lysate protein concentrations were measured using BCA (Pierce). Protein samples were fractionated by SDS-PAGE and transferred to nitrocellulose membranes. Immunoblots were developed by a chemiluminescence method (Pierce) using alkaline phosphatase conjugated secondary antibodies.

Effect of PTP1B on Cell Morphology and F-actin
Organization-The signaling pathways by which insulin induces cytoskeletal alterations are largely unknown. Quiescent or serum-starved cells show presence of abundant stress fibers but fewer membrane ruffles. Insulin treatment has been shown to dissolve stress fibers and induce membrane ruffling in cells (37). Because we and others (23,24) have previously shown that overexpression of PTP1B affects the actin cytoskeleton, we asked if PTP1B could mediate any of the cytoskeletal effects of insulin in immortalized cells. For these experiments we used rat 3Y1 cells that stably express about 3-5-fold excess HAtagged PTP1B-wild type (3Y1-WT) and a proline to alanine (P309,310A) mutant of PTP1B (3Y1-PA), which we have previously shown is incapable of associating with SH3-containing proteins such as p130 Cas (19,27) (Fig. 1A). Serum-starved control rat 3Y1 cells have few membrane ruffles but abundant stress fibers (Fig. 1B, panel a). Upon insulin stimulation, control rat 3Y1 cells formed membrane ruffles and showed dissolution of stress fibers (Fig. 1B, panel b). In contrast, 3Y1-WT cells showed simultaneous presence of membrane ruffles and stress fibers under serum-starved as well as insulin-stimulated conditions (Fig. 1B, panels c and d). 3Y1-PA cells resembled control cells under serum-starved conditions. They showed abundant stress fibers but fewer membrane ruffles (Fig. 1B, panel e) when treated with insulin they formed membrane ruffles and showed partial loss of stress fibers (Fig. 1B, panel f). These results suggest that the majority of the cytoskeletal effects of PTP1B are mediated by interactions with one or more SH3 domain-containing proteins.
PTP1B Activates Rho Family GTPases-The distinctive morphological features of 3Y1-WT cells, such as the presence of membrane ruffles even under serum-starved conditions, prompted us to examine the activity of the Rho family GT-Pases. Rac1 and RhoA GTPases are known to regulate cytoskeletal phenotypes such as membrane ruffling and stress fibers, respectively (4). GST fusion proteins with the p21-binding domain of Pak (GST-PBD) and the Rho-binding domain of Rhotekin (GST-Rhotekin) were used to assess GTPase activity (35,36). GST-PBD binds only to the activated form of Rac1, whereas GST-Rhotekin binds only to the activated form of RhoA. Quiescent cells show the presence of abundant stress fibers and few membrane ruffles (Fig 1B, panel a) suggesting increased RhoA and diminished Rac1 activity. Upon stimulation with insulin, the control cells gain membrane ruffles whereas stress fibers dissolve (Fig. 1B, panel b), suggesting increased Rac1 and decreased RhoA activity respectively. Measurements of GTPase activity bear out these predictions. In control 3Y1 cells, basal Rho activity is high, whereas Rac is low; and upon insulin stimulation, Rho activity decreases whereas Rac activity increases (Fig. 2). In 3Y1-WT cells Rac1 activity was elevated even under serum-starved conditions, and RhoA activity was not inhibited even after insulin stimulation (Fig. 2). GTPase activity profiles in 3Y1-PA cells were similar to those of control cells (Fig. 2). Thus, the morphological changes induced by overexpression of wild type PTP1B are accompanied by appropriate biochemical changes in GTPase activity. (39 -44), affect the activity of Rho family GTPases and because Src has been suggested to be activated by PTP1B (by dephosphorylating the negative regulatory Tyr-527 site in Src) (25,26,45), we examined the effect of PTP1B expression on the activity of c-Src family kinases. In control rat 3Y1 cells c-Src family kinase was not active under serum-starved conditions as assessed by immunoblotting using anti-phospho-Src (Tyr(P)-416) antibody (Tyr-416 is phosphorylated only when Src is active) (Fig. 3A). Insulin stimulated c-Src family kinase activity in control 3Y1 cells. 3Y1-WT cells showed elevated levels of c-Src family kinase activity even under serum-starved conditions (Fig. 3A). On the other hand, under serum-starved conditions, 3Y1-PA cells displayed low levels of c-Src family kinase activity. These results suggest that PTP1B activates c-Src kinases and that this activation is mediated by SH3/proline interaction.

PTP1B Regulates Rho GTPases via Src Family Kinases-Because c-Src family kinases, via phosphorylation of GTD exchange factors and GTPase activating proteins
Because PTP1B contains two proline-rich motifs that fit the consensus sequence for SH3 binding and is known to associate with selected SH3-containing proteins (19), we asked if PTP1B interacts with Src family kinases in vitro via an SH3/proline interaction. We found that WT-PTP1B binds to the SH3 domains of a variety of proteins, including p130 Cas , Src (Fig. 3B), but does not bind to the SH3 domains of Nck or Fyn (data not shown). Consistent with our previous studies, the PTP1B-PA mutant was severely impaired in its ability to bind SH3 domain-containing proteins (19) (Fig. 3B). Binding of the PA mutant of PTP1B to the SH3 domain of p130 Cas was completely abolished, whereas binding to the SH3 domain of Src, though not completely abolished, was greatly reduced (Fig. 3B). However, despite the robust interaction of the proline-rich domain of PTP1B and the SH3 domain of c-Src, we could not demonstrate an in vivo association between full-length PTP1B and c-Src in cells (data not shown).
If PTP1B acts primarily through Src family kinases to activate Rho family GTPases, then these effects should be abolished in cells treated with a chemical inhibitor of Src family kinase, such as PP1 (46). We treated control 3Y1 cells and

3Y1-WT cells with PP1 and then checked for activation of Rac1
GTPase. Inhibition of c-Src family kinases by PP1 abolished Rac1 activation under serum-starved and insulin-stimulated conditions in control 3Y1 cells as well as in 3Y1-WT cells (Fig.  3C). These data strongly support the idea that PTP1B affects actin organization via activation of one or more Src family kinases.
Cumulatively, our data indicated that overexpression of wild type PTP1B induced activation of c-Src family kinase in quiescent 3Y1-WT cells, leading to the activation of Rac1 GTPase further leading to the gain of membrane ruffles.

Loss of PTP1B Function/Expression Suppresses Src and Rho
Family GTPases Activation-If PTP1B mediates insulin stimulation of Rho family GTPases via Src, then loss of PTP1B function/expression should impair the ability of insulin to activate the Rho family GTPases. We used antisense methodology to reduce the endogenous PTP1B expression in rat 3Y1 cells (24,47) (Fig. 4A). Insulin treatment of control 3Y1 cells versus antisense-treated rat 3Y1 cells showed that c-Src family kinase was not activated in antisense-treated cells as compared with the control cells, indicating that PTP1B is required for the Lysates from 3Y1-WT and 3Y1-PA cells were incubated with the glutathione-Sepharose beads bound to the GST fusion proteins. Following extensive washes, the proteins were eluted into SDS-PAGE sample buffer and analyzed by immunoblot (upper panel). Whole cell lysates were also analyzed for PTP1B expression (lower panel). C, control 3Y1 or 3Y1-WT cells were serum-starved and either were left untreated (N.T.) or pretreated with 10 M PP1 for 10 min, followed by stimulation with insulin and Rac1 activity was examined by GTPase pull-down assay as described under "Experimental Procedures." activation of Src, as assessed by monitoring the phosphorylation of Tyr-416 (Fig. 4A). Similarly, in the absence of PTP1B, insulin failed to activate Rac1 (Fig. 4A). These biochemical changes mirror the cytoskeletal response of cells with reduced PTP1B expression. Cells treated with antisense of PTP1B fail to form lamellipodia upon insulin stimulation (Fig. 4B). These results indicate that PTP1B is required for insulin-mediated activation of c-Src family kinase and Rac1, as well as the formation of membrane ruffles (Fig. 4B).
Based on the findings reported in the present study and the recent report that growth factor receptors must be internalized from their site of activation at the plasma membrane before they encounter PTP1B at the external face of the endoplasmic reticulum (48), we wanted to address the spatio-temporal regulation of the effects of insulin on cytoskeleton. Indeed, when cells were treated with insulin and analyzed for insulin receptor autophosphorylation and Rac1 activation, we found that insulin receptor autophosphorylation was maximal at 1-2 min postinsulin treatment (Fig. 5), which is consistent with published reports (2), whereas Rac activation began about 4 min of insulin treatment (Fig. 5). DISCUSSION In the present study we show that PTP1B is required for insulin to exert its effects on the actin cytoskeleton. Overexpression of PTP1B induces a biochemical and morphologic phenotype in cells similar to that seen when cells are treated with insulin. Loss of PTP1B function/expression inhibits the ability of insulin to signal to c-Src family kinase or Rac and to induce reorganization of the actin cytoskeleton. Although previous studies have suggested that PTP1B can activate c-Src family kinase via dephosphorylation of Tyr-527 (25,26,45), our results are the first to suggest a mechanism by which this event occurs. However, c-Src family kinase and PTP1B do not appear to interact directly in cells, as they do not co-immunoprecipitate, but the activation of c-Src family kinase by PTP1B clearly depends on the proline-rich region in PTP1B, which binds the SH3 domain of p130 Cas in vitro and in vivo (19) and the SH3 domain of Src in vitro. In 3Y1-PA cells, Src Tyr-527 remains phosphorylated and Src remains inactive.
The notion that PTP1B plays a positive role in cytoskeletal signal transduction is consistent with recent data from Cheng et al. (25), who showed that transformed PTP1B Ϫ/Ϫ fibroblasts fail to activate c-Src family kinases, or to tyrosine phosphorylate p130 Cas , in response to integrin engagement by fibronectin. As c-Src family kinases and p130 Cas play important roles in cell motility (49), it is likely that PTP1B is required for haptotactic cell locomotion. However, the role of PTP1B in growth factormediated motility is less certain. Buckley et al. (50) recently reported that insulin-like growth factor-1-mediated motility is enhanced in the absence of PTP1B, suggesting that this phosphatase acts as a negative regulator of growth factor-induced cell locomotion. The activity of Rho family GTPases was not examined in this work nor was the cytoskeleton imaged. Indeed, to our knowledge, our data are the first to document the effects of insulin on Rho family GTPases.
Because growth factor receptors must be internalized from their site of activation at the plasma membrane before they encounter PTP1B at the external face of the endoplasmic reticulum (48), our findings predicted that the cytoskeletal effects of insulin would lag behind its metabolic effects. Indeed, whereas insulin receptor autophosphorylation in rat fibroblasts is maximal at 1-2 min post insulin treatment (2), ruffling is maximal at 4 -8 min (38). Consistent with this temporal pattern of insulin signaling, we noted that maximal Rac activation occurs 6 -10 min post insulin treatment. Thus, we theorize that the temporal sequence of insulin signaling is governed, at least in part, by access to PTP1B. In this model, PTP1B is activated by the insulin receptor at the endoplasmic reticulum surface, whereupon it activates Src by dephosphorylating the inhibitory Tyr-527 site probably mediated by an SH3/ proline interaction either with Src or with a docking protein such as p130 Cas . This model would explain why point mutations within the non-catalytic, C-terminal proline-rich region of PTP1B completely abolish its cytoskeletal activity (18,23). This model also suggests that it might be possible to selectively modulate the actions of PTP1B on insulin signaling by interfering with its interactions with Src.
In summary, our findings indicate that PTP1B plays a sig- FIG. 5. Temporal profile of insulin receptor and Rac1 activation by insulin. 3Y1 cells were treated with insulin for the indicated time intervals. Insulin receptor autophosphorylation was determined by probing the blot with anti-phosphotyrosine antibody, and Rac1 activation was examined by GTPase pull-down assay as described under "Experimental Procedures." nificant role in transmitting the signals of insulin to the cytoskeleton. This signaling involves, but may not be limited to, activation of Src family kinases. It is possible that other PTP1B substrates, such as p130 Cas or unknown proteins, also mediate these effects. We are currently examining the rapidly expanding number of known or suspected PTP1B substrates with respect to the regulation of the actin cytoskeleton.