Reduced cell attachment and phosphorylation of focal adhesion kinase associated with expression of a mutant insulin receptor.

Insulin signaling results in rapid changes to the cell cytoskeleton, and it has recently been shown that insulin stimulates the dephosphorylation of the cytoskeletal-associated tyrosine kinase, focal adhesion kinase (pp125FAK). We report here that mutation of two tryptic cleavage sites (Lys164 and Lys582 → Asn; 2N) in the insulin receptor α-subunit results in a cell-line (CHO.2N-10) with altered morphology associated with an increase in cell size, a decrease in cell adhesiveness, and a decrease in pp125FAK tyrosine phosphorylation in the absence of insulin (45.2 ± 9.7% compared to nontransfected Chinese hamster ovary (CHO) cells). In contrast to pp125FAK, paxillin phosphorylation was similar in all cell lines despite lower levels (61.0 ± 10.4% compared to CHO cells) of paxillin protein in CHO.2N-10 cells. We observed comparable protein levels of pp125FAK and the structural focal adhesion protein, vinculin, in all cell lines. Despite underphosphorylation of pp125FAK in the basal state, insulin stimulation of CHO.2N-10 cells still resulted in dephosphorylation of pp125FAK. CHO.2N-10 and CHO.T (overexpress wild-type insulin receptor) cells have similar insulin binding characteristics, insulin-stimulated autokinase and peptide phosphorylation, and insulin-stimulated pp185/IRS-1 phosphorylation. Our results suggest that the insulin receptor may play an important role in cell-matrix interactions, such as modulating cell adhesion and inducing cell architecture changes.

Interactions between growth factors and integrin signaling pathways are probably involved in the regulation of cell proliferation, shape, adhesion and migration (1,2). Integrins, the receptors for extracellular matrix proteins, provide both a physical link to the cytoskeleton (often at sites of focal adhesion) and transduce signals from the extracellular matrix. A connection is observed between integrins and stress fibers of polymerized actin, which is necessary for maintenance of cell integrity and appears to terminate at focal adhesion sites. The assemblies of structural proteins (such as ␣-actinin, talin, vinculin, tensin, and paxillin) that co-localize with some integrins in focal adhesions are thought to play important roles in stabilizing cell adhesion and regulating cell shape, morphology, and mobility (1). These structural proteins may also serve as a framework for the association of signaling molecules that regulate signal transduction pathways leading to integrin-induced changes in cells (1). In addition, there are now many examples of growth factor-induced regulation of components involved in integrin signaling pathways. Growth factors may rapidly stimulate actin polymerization at the plasma membrane of many cell types to produce lamellipodia and edge-ruffles (e.g. PDGF 1 and insulin) (3)(4)(5) and at later times promote actin stress fiber formation (e.g. PDGF) (6). Conversely, components of growth factor signaling pathways (such as the mitogen-activated protein kinase cascade) are probably utilized by integrins as part of their signal transduction pathways (7,8).
A key enzyme in cytoskeletal rearrangements/interactions/ architecture appears to be the phosphorylated, cytoplasmic tyrosine kinase, focal adhesion kinase (pp125 FAK ) which plays a central role in integrin-mediated signal transduction. Apart from activation by integrins, specific tyrosine phosphorylation of pp125 FAK is also induced by several growth factors (9), lysophosphatidic acid (10) and neuropeptides (11,12), suggesting that several diverse signaling pathways may converge at this point. The N-terminal sequence of pp125 FAK can bind synthetic peptides derived from several ␤-integrin cytoplasmic domains, although an in vivo association has not yet been demonstrated (2); indeed, an indirect linkage via talin (13) has recently been proposed. Within the C terminus of pp125 FAK , a focal adhesion targeting sequence is both necessary and sufficient to localize pp125 FAK to focal adhesions (14). The cytoskeletal protein, paxillin, also associates with pp125 FAK through a C-terminal sequence of pp125 FAK distinct from the focal adhesion targeting domain (15,16). Upon integrin clustering, pp125 FAK is autophosphorylated on tyrosine 397 (17), and its tyrosine kinase activity is enhanced, which may then lead to association of pp125 FAK with SH2 domain-containing signaling proteins such as Src, GRB2, phosphatidylinositol 3-kinase (PI 3-kinase), Cterminal Src kinase, and phospholipase C␥ to potentially form large signaling complexes which can, for instance, activate mitogen-activated protein kinase. Growth factors such as PDGF stimulate tyrosine phosphorylation of paxillin and pp125 FAK (9) and induce the association of PI 3-kinase with pp125 FAK in adherent Swiss 3T3 cells (18). Lysophosphatidic acid and neuropeptides such as bombesin also stimulate tyrosine phosphorylation of pp125 FAK and induce formation of focal adhesions and actin stress fibers in Swiss 3T3 cells (6,10,12). However, while associated with cytoskeletal rearrangements, a definitive role for pp125 FAK in signal transduction mechanisms that alter cytoskeletal properties remains to be determined.
As discussed above, most stimuli increase pp125 FAK tyrosine phosphorylation. Surprisingly (and perhaps in contrast to its "PDGF-like" effect on cytoskeletal changes in some cells) (5), insulin stimulates the dephosphorylation of pp125 FAK in rat 1 fibroblasts (19) and in CHO cells (20). Unlike the bell-shaped curve of pp125 FAK phosphorylation in response to PDGF, pp125 FAK dephosphorylation occurs at all insulin concentrations (21). Insulin stimulation was also reported to correlate with a marked decrease in the length and number of actin stress fibers in CHO cells overexpressing insulin receptors (22). We now report that expression, in CHO cells, of an insulin receptor mutated to change lysines 164 and 582 to asparagine leads, in the absence of insulin, to decreased tyrosine phosphorylation of pp125 FAK , altered cell shape, and abrogated cell adhesion. These sites, which are putative tryptic cleavage sites (23,24), were mutated as part of an ongoing study to examine the mechanism of tryptic activation of the insulin receptor (24). The unusual phenotype of the cell line expressing the mutant insulin receptor provides further evidence for the insulin receptor as a potential regulator of cell architecture and adhesion.

EXPERIMENTAL PROCEDURES
Materials-Human insulin receptor containing Lys 164 3 Asn, Lys 582 3 Asn, or K164N/K582N (2N) substitutions were generated using polymerase chain reaction-based mutagenesis by Dr. L. Macaulay (CSIRO, Australia). Rat tail collagen (type I) was kindly provided by Dr. Bob Whitehead (Ludwig Institute, Melbourne, Australia). The mouse myeloid cell line (FDC-P1) was generously provided by Dr. A. Harris (Walter and Eliza Hall Institute, Melbourne, Australia). Monoclonal antibodies against the human insulin receptor, CT-1, and 18-44 were kindly provided by Prof. K. Siddle, University of Cambridge, UK. Antibodies 2A7 (monoclonal) and BC3 (polyclonal) against pp125 FAK were generously provided by Dr. J. T. Parsons, University of Virginia, Charlottesville. The mouse anti-hamster integrin ␣ 5 (clone PB1) and antihamster integrin ␤ 1 (clone 7E2) monoclonal antibodies were generous gifts from Dr. R. Juliano, University of North Carolina, Chapel Hill. Lipofectin® and G418 (Geneticin) were purchased from Life Technologies, Inc. Plasma fibronectin was purchased from Promega, Madison, WI. Insulin kinase substrate "FYF peptide" (RRDIFETDYFRK) was purchased from Auspep, Australia. Radioactive [␥-32 P]ATP (3000 Ci/ mmol) was purchased from Bresatec, Australia. Rhodamine (TRITC)labeled phalloidin was purchased from Sigma. FITC-labeled sheep antimouse IgG used for FACS sorting and sheep anti-rabbit and anti-mouse IgG F(ab) 2 fraction affinity isolated FITC-conjugated antibodies for FACS analysis were purchased from Silenius, Australia. The polyclonal antibody against a peptide corresponding to the amino acids 1223-1242 mapping at the C terminus of IRS-1 (clone C-20) was purchased from Santa Cruz Biotechnology, Santa Cruz, CA. The monoclonal anti-human vinculin (clone h-VIN1) was purchased from Sigma, and monoclonal antibody against paxillin was purchased from Transduction Laboratories, Lexington, KY. The following anti-integrin antibodies were purchased from Chemicon International Inc., Temecula, CA: mouse anti-human ␣ v monoclonal, rabbit anti-␣ 3 affinity purified, and rabbit anti-␤ 5 polyclonals. The monoclonal anti-rat LFA-1 ␤ 2 chain was purchased from Seikagaku Corporation, Toyko, Japan. Anti-phosphotyrosine antibodies conjugated to horse radish peroxidase (HRP) were purchased either from ICN Biomedicals, Cleveland, OH (clone PY20) or Upstate Biotechnology Inc., Lake Placid, NY (clone 4G10). Secondary antibodies used for immunoblotting and conjugated to HRP were purchased from Dako Corp., Carpinteria, CA. Enhanced chemiluminescence reagents were from Dupont NEN. All other reagents were of the highest grade available.
Site-directed Mutagenesis-A Lys 164 3 Asn mutant receptor cDNA was constructed by inserting a mutagenic oligonucleotide (generated by a polymerase chain reaction, encoding the Lys 164 3 Asn mutation between the KpnI and EcoRI restriction sites of the human insulin receptor cDNA) into a KpnI-EcoRI digest of the insulin receptor expression vector pET (25). For mutagenesis of Lys 582 3 Asn, a mutant oligonucleotide was generated by a polymerase chain reaction between the AccI and BamHI sites of the insulin receptor cDNA, and then inserted into the AccI-BamHI digest of a modified pET insulin receptor expression vector. The single mutations were combined to produce the 2N mutant by inserting the EcoRI-XbaI fragment from Asn 582 into the Asn 164 construct. All constructs were sequenced around the insertion, subjected to extensive restriction analysis, and shown to be correct.
Cell Culture-CHO cell lines were maintained in ␣-modified Eagle's medium (␣-MEM) containing 10% v/v fetal calf serum (FCS). For immunofluorescence, cells were cultured on glass coverslips to 80% confluency. When required cells were serum-starved for 18 h with ␣-MEM containing either 0.5% v/v FCS or 0.1% w/v bovine serum albumin (BSA).
Stable Expression of cDNA in CHO Cells-CHO cell lines expressing the 2N (K164N/K582N) insulin receptor (and single point mutants) were obtained by co-transfection of 10 g of plasmid DNA and 2 g of pSVNeo DNA using Lipofectin®. Transfected CHO cell lines were maintained in medium containing 800 g/ml G418 sulfate to select for stable colonies. Positive insulin receptor expressing clones were identified by enzyme-linked immunosorbent assay as described by Clark et al. (24). The only variation to the described method (24) was the addition of 10 mg/ml BSA after coating the microtiter plates with CT-1, to block nonspecific binding. Cell lines expressing high levels of mutant receptors were further selected by fluorescence-activated cell sorting (FACS) as described previously (26), using nonbiotinylated antibody 18 -44 followed by FITC-labeled sheep anti-mouse IgG. In some experiments, cell sorting with simultaneous single cell cloning was performed. Several independently derived clones were analyzed. 125 I-Insulin-binding Assay-CHO cell lines were plated in 24-well plates and incubated in Earle's balanced salt solution, 25 mM Hepes, pH 7.4, and 0.1% w/v BSA containing approximately 20,000 cpm 125 Iinsulin in the absence or presence of increasing concentrations of unlabeled insulin (1-1000 ng/ml) for 4 h at 8°C. Cells were washed and solubilized, and bound radioactivity was measured in an autogamma spectrometer as described previously (27). Nonspecific binding was measured in the presence of 10 g/ml unlabeled insulin and was always less than 2% of total binding. Data were analyzed by the method of Scatchard (28).
Receptor Autophosphorylation and Kinase Activity-CHO cells were treated with or without insulin (172 nM) and lysed, and insulin receptors were immunoprecipitated with monoclonal antibody CT-1 as described previously (27). In vitro autophosphorylation assays were carried out after addition of [␥-32 P]ATP as described previously (27). To measure peptide phosphorylation, receptor immunoprecipates were incubated in 40 l of phosphorylation buffer (50 mM Hepes, pH 7.4, 12 mM MgCl 2 , 100 M Na 3 VO 4 , 2 mM MnCl 2 , 150 mM NaCl, 0.2% v/v Triton X-100, 50 M ATP, and 1 mM EGTA), and the reaction was started by the addition of 200 M FYF peptide (RRDIFETDYFRK) and 0.4 Ci of [␥-32 P]ATP per sample for 5 min at 30°C. At this time, duplicate 10-l aliquots were spotted onto P81 Whatman filters, washed three times in 30% w/v glacial acetic acid, 0.05% w/v phosphoric acid, rinsed in 70% v/v ethanol, and dried, and radioactivity was measured in a liquid scintillation spectrometer. Blank values, in the absence of FYF peptide, were subtracted from each result.
For IRS-1 immunoprecipitation, cells were treated as described by Konstantopoulos and Clark (21). For insulin receptor ␤-subunit immunoprecipitation, cells were treated as described above and immunoprecipitated with monoclonal antibody 18-44 overnight at 4°C. To detect tyrosine phosphorylation of either IRS-1 or insulin receptor ␤-subunit, membranes were probed with HRP-conjugated 4G10 (1:188).
Measurements of Cell Growth Rate and Cell Attachment-CHO cell lines were seeded at 10 ϫ 10 3 cells/ml in 24-well plates and maintained in ␣-MEM containing 10% v/v FCS. Cells from individual wells were washed twice with Versene (phosphate-buffered saline containing 0.2% v/v EDTA), and attached cells were removed by incubation with 0.25% v/v trypsin/Versene for 3 min at 37°C. Cells were counted (in triplicate) in a Coulter counter every 24 h over 5 days, and simultaneously, the Coulter counter measured the mean cell volume. For attachment assays plates were left uncoated or coated with 10 g/ml fibronectin or 15 g/ml rat tail collagen. Treated or untreated plates were blocked with 2 mg/ml BSA for 1 h at 37°C and washed with phosphate-buffered saline prior to the addition of cells. Cells (10 -30 ϫ 10 3 /ml) were allowed to attach for either 2 or 24 h after plating in the presence of ␣-MEM containing 0.1% BSA.
Actin Staining of CHO Cell Lines-For filamentous actin localization, cells grown on coverslips were fixed in 4% formaldehyde in Trisbuffered saline for 8 min at 22°C, rinsed in Tris-buffered saline, and permeabilized for 5 min in Tris-buffered saline containing 0.5% (v/v) Triton X-100. After washing in Tris-buffered saline, cells were incubated with rhodamine-phalloidin (0.1 g/ml) for 20 min at 22°C in a dark humidity chamber. After the final washes, the coverslips were mounted with a drop of Gelvatol. Cells were viewed on a LaborLux D fluorescence microscope and photographed with Kodak TMAX 400ASA film.
Protein Measurements-Cellular protein concentrations were deter-mined using Coomassie Brilliant Blue dye with bovine serum albumin standards (31). Statistical Analyses-Numerical results are expressed as the mean Ϯ S.E. and statistical comparisons were performed by Student's t test for unpaired samples; p Ͻ 0.05 was taken as significant. (2N ϭ K164N/K582N) was initially difficult to obtain, despite production of stable cell lines expressing either of the individual point mutations alone (Lys 164 3 Asn or Lys 582 3 Asn). To avoid overgrowth of 2Nexpressing cells by non-2N-expressing cells, CHO.2N cells were simultaneously FACS sorted and single-cell cloned 10 days after transfection. This method provided eight independently derived clones expressing the 2N insulin receptor (up to 15 passages) which have been studied to varying degrees. One of the highest receptor-expressing clones (2N-10) was used in most of the studies, as its binding and phosphorylation characteristics (see below) were closest to those of CHO.T cells (overexpress wild-type human insulin receptors). The mutations in the external domain of the insulin receptor were of potential trypsin cleavage sites (23,24) and mutated as part of another study (24). However, the unusual phenotype of the cells expressing the double point mutation (see below) led to the analysis presented here. In preliminary experiments, treatment of the cell lines (expressing mutant insulin receptors) with trypsin did not result in significant differences from CHO.T cells. 4 Insulin-binding Characteristics-The ability of two CHO.2N cell lines (clones 10 and 11) to bind insulin is shown in Fig. 1. The insulin binding of CHO.2N-10 and CHO.2N-11 cells was lower than that of CHO.T cells (clone 10 at passage 6, 9.8%, and clone 11 at passage 6, 13.2%, compared to wild-type, 42.0%, of total 125 I-insulin added). In addition, 125 I-insulin binding to CHO.2N-10 and 2N-11 cells decreased with prolonged passage, whereas the ability of CHO.T cells to bind 125 I-insulin did not alter with passage. 125 I-insulin binding always correlated with the level of insulin receptor expression measured by enzymelinked immunosorbent assay (data not shown). CHO.2N-10 cells express fewer insulin receptors per cell compared to CHO.T cells as determined by Scatchard analysis (Table I)  of insulin (0.47% of total 125 I-insulin added, see Fig. 1). The dissociation constants for insulin binding were comparable between the cell lines (Table I).

Expression of 2N cDNA in CHO Cells-Expression of a double point mutant of the insulin receptor
Kinase Activity of CHO.2N Insulin Receptors-The autophosphorylation of insulin-stimulated CHO.2N-10 insulin receptors (passages 6 -10) was similar to wild-type insulin receptors (Fig. 2, lower panel) as measured by immunoblotting with anti-phosphotyrosine antibodies. Autophosphorylation of CHO.2N-10 receptors was stimulated 4.4-fold by insulin compared to 5.6-fold for wild-type receptors (Table I) in an in vitro phosphorylation assay. When the insulin-stimulated kinase activity of CHO.2N-10 and wild-type receptors toward a synthetic peptide (FYF, derived from the major autophosphorylation domain of the insulin receptor tyrosine kinase) was measured, phosphorylation was stimu-lated 20.1-and 19.9-fold, respectively (Table I). Insulin-stimulated phosphorylation of pp185/IRS-1 was also comparable between CHO.T and CHO.2N-10 (passage 6 -8) cells (Fig. 2,  upper panel).
Cellular Morphology and Growth Characteristics-CHO.2N-10 cells (as well as four other 2N clones) showed morphological differences which disappeared upon prolonged passage. These morphological changes were not dependent on cell density. Fig. 3A is a photomicrograph of CHO.2N-10 cells at passage 7; these cells formed a heterogenous monolayer with many large, flat cells containing large, granular nuclei. In contrast, the nontransfected, parental CHO cells (Fig. 3B) or CHO.T cells (not shown) grew as a homogenous monolayer of smaller cells. At late passages (Ͼ15 passages), CHO.2N-10 cells were indistinguishable from CHO cells (not shown). This reversion in morphology strongly correlated with the loss of insulin binding (Fig. 1) and insulin receptor expression of CHO.2N-10 cells at these later passages. Rhodamine-phalloidin staining of CHO.2N-10 cells at passage 9 shows the dense formation of actin stress fibers running the length of the mutant cells (Fig. 3C). In contrast, the nontransfected CHO cells (Fig. 3D) or CHO.T cells (not shown) stain intensely close to the plasma membrane region of the cells, which appears to be absent in the CHO.2N-10 cells (Fig. 3C).
The growth of CHO.2N cell lines, especially clone 10, appeared extremely slow compared to nontransfected CHO and wild-type CHO.T cells. However, the doubling time of CHO.2N-10 cells at all passages was similar to nontransfected CHO cells when corrected for the number of cells attached to dishes after 24 h (Fig. 4). In addition, there was a significant difference in the mean cell volume, with early passage CHO.2N-10 cells having a larger volume compared to CHO cells (1973 Ϯ 42 fl compared to 1231 Ϯ 42 fl, p Ͻ 0.01, n ϭ 5) correlating with the observed differences in cell size (Fig. 3, A compared to B). apparent slower cell growth of CHO.2N-10 cells was explained by their inability to attach efficiently to tissue culture dishes at early passage. After 24 h (reflects both cell attachment and cell growth) CHO.2N-10 (passage 6) cell number was 31% of added cells compared to CHO.2N-10 cells at late passage (150% of added cells) or controls (140% of added cells) (Table II). This reduced ability to adhere to the tissue culture dishes was observed with six out of the eight independently derived clones (data not shown). This was not due to differences in extracellular matrix production by early passage CHO.2N-10 cells as these cells still showed reduced attachment to matrix elaborated by CHO.T cells, and conversely CHO.T cells attached equally well to their own or a CHO.2N-10-elaborated matrix after 2 h (data not shown). In addition, early passage CHO.2N-10 cells attached less well to both fibronectin and type I collagen-coated plates after 2 h when compared to controls (Table II). Viability studies showed that the cells which did not attach to tissue culture plates were still viable (83 Ϯ 4%), as determined by trypan blue exclusion.

FIG. 2. Insulin stimulates IRS-1 and insulin receptor ␤-subunit phosphorylation in CHO (a and b), CHO.T (c and d), and CHO.2N-10 (e and f) cells. Cells were incubated either in the absence
Focal Adhesion Kinase (pp125 FAK ) and Other Focal Adhesion Proteins-Focal adhesions are cellular regions that, in vitro, provide contact points for cells with tissue culture dishes. They contain, among other proteins, a cytoplasmic tyrosine kinase, pp125 FAK , which is phosphorylated upon cell attachment (32).
CHO.2N-10 cells at early passage have a significant reduction in tyrosine phosphorylation of pp125 FAK in the absence of insulin (42.5 Ϯ 9.7% of CHO cells, Figs. 5, panel A, and 6, upper  panel). This difference held whether sample loading was corrected for either cell number (Fig. 5, panel A) or cell protein content (data not shown). It is interesting to note that expression of the wild-type insulin receptor in CHO cells (CHO.T) resulted in a consistent but not significant increase of pp125 FAK tyrosine phosphorylation (168.9 Ϯ 30.0%) compared to nontransfected CHO cells (Fig. 5, panel A). Focal adhesion kinase protein levels were comparable between CHO, CHO.T, and CHO.2N-10 cells (passages 5-8) (100 Ϯ 12%, 77.7 Ϯ 3.4%, and 77.1 Ϯ 4.9%). We also measured pp125 FAK tyrosine phosphorylation as a function of adhesion to fibronectin in all cell lines as cell attachment to fibronectin often leads to an increase in pp125 FAK phosphorylation (32). However, none of the cell lines showed a further increase in pp125 FAK phosphorylation following 2 h of plating onto fibronectin when compared to plating on noncoated dishes (data not shown). Alteration to the basal level of pp125 FAK phosphorylation could result from the action of a phosphatase, we therefore preincubated CHO.2N cells with the tyrosine phosphatase inhibitor pervanadate (1 or 10 M) for 1 h at 37°C. The basal level of pp125 FAK phosphorylation in CHO.2N cells increased significantly toward that of CHO.T cells (Fig. 5, panel B).
Recently, two laboratories have shown that insulin, unlike other growth factors, stimulates dephosphorylation of pp125 FAK in cells overexpressing wild-type insulin receptors (19,20). We have also examined the effect of insulin on pp125 FAK in CHO, CHO.T, and early passage CHO.2N-10 cells. Insulin decreased pp125 FAK tyrosine phosphorylation in both CHO.T and CHO.2N-10 (passages 7 and 8) cells (Fig. 5, panels C and D) when compared to basal pp125 FAK tyrosine phosphorylation (27.9 Ϯ 7.9% and 52.2 Ϯ 10.0% of nonstimulated cells, respectively). CHO cells which express only endogenous hamster insulin receptors showed negligible dephosphorylation of pp125 FAK in response to insulin (91.0 Ϯ 16.5%, Fig. 5, panels C and D). The protein levels of pp125 FAK were not altered in response to insulin stimulation (data not shown). Results were similar whether cells were tested in complete medium or after 18 h of serum starvation.
Paxillin, a structural protein, is also phosphorylated on tyrosine residues, usually (15) but not always (33,34) in parallel to pp125 FAK . In our studies, in contrast to the decrease in pp125 FAK tyrosine phosphorylation, there was no consistent change in paxillin phosphorylation (Fig. 6, middle panel) in CHO.2N-10 cells (passages 7-10). Over a number of experiments, less paxillin protein was immunoprecipitated from CHO.2N-10 cells (61.0 Ϯ 10.4%, p Ͻ 0.01, n ϭ 7) when compared to CHO and CHO.T cells. However, the ratio of paxillin phosphorylation to paxillin protein levels was similar for all  a Cells were seeded at 10 ϫ 10 3 /ml in 24-well plates, and after 24 h cells were counted as described. b Cells were seeded at 30 ϫ 10 3 /ml in 24-well plates coated as described under "Experimental Procedures," and after 2 h attached cells were counted.
c All results are expressed as the mean of triplicate determinations Ϯ standard error of the mean, and are representative of five separate experiments. *p Ͻ 0.05 when compared with CHO, CHO.T, and late passage CHO.2N-10 cells. cell lines. Paxillin binds to the rod domain of vinculin, another structural focal adhesion protein concentrated at the sites of cell-extracellular matrix adhesions involving integrins (35). Vinculin levels were also comparable between CHO, CHO.T, and CHO.2N-10 (passages 7 and 8) cells (Fig. 6, lower panel).
In order to estimate the relative number of focal contacts for each cell line we examined the cellular distribution of vinculin by indirect immunofluorescence. Quantitation was not possible due to the gross differences in cell size between early passage CHO.2N cells and other CHO cell lines (Fig. 3, A compared to  B); however, no overall difference in cellular localization of vinculin was observed (data not shown). We also attempted to estimate the number of focal adhesions present in the adherent cells by immunolocalizing the focal adhesion and membrane ruffle marker, talin. Talin has not only been shown to be indirectly linked with pp125 FAK (13) but also linked directly to the ␤ 1 cytoplasmic portion of integrin (36,37). CHO cell lines did not exhibit well developed focal contacts (also reported by Bauer et al. (38)) but rather showed an intense staining for talin around the periphery of the cell. This pattern of staining was not significantly different between CHO.2N, CHO.T, and CHO cells (data not shown).
Alterations to the expression of integrins (receptors for extracellular matrix proteins) could provide an explanation for the morphological changes in CHO.2N cells. We investigated any gross changes to the level or type of integrin expression in CHO.2N cells by FACS analysis. CHO cells endogenously express significant levels of ␣ 5 and ␤ 1 integrins, and lower levels of ␣ 3 , ␣ v , and ␤ 5 integrins. 3 In general, no change to this pattern of expression was observed in CHO.2N cells compared with the other cell lines; however, there was a trend for an increase in the mean fluorescence of ␣ 3 integrins (319.8 Ϯ 91.5%, n ϭ 5 where FITC alone is corrected to 100%) compared to either CHO.T or CHO cells (181.3 Ϯ 4.5% and 175.4 Ϯ 16.2%, respectively, n ϭ 5) (Fig. 7), but this did not reach statistical significance due to the larger standard error of the CHO.2N cells as a result of their heterogeneity. It was interesting to note the 2-3-fold increase (p Ͻ 0.05, n ϭ 3) in expression of ␤ 5 integrins in cell lines which overexpress insulin receptor constructs (CHO.T and CHO.2N) compared to CHO cells (Fig. 7). DISCUSSION We have shown that mutation of two potential tryptic cleavage sites (23)  the insulin receptor ␣-subunit results in a cell line with altered morphology associated with an increase in cell size, a decrease in cell adhesiveness, a decrease in protein levels of paxillin, and a decrease in tyrosine phosphorylation of focal adhesion kinase. These features correlate with expression of the mutant 2N insulin receptor. The cells revert to normal CHO phenotype at later passages, at which time insulin binding and receptor expression returns to that of nontransfected CHO cells, presumably because the cells lose the capacity to express the mutant insulin receptor. However, decreased CHO.2N cell adhesion was not associated with altered insulin receptor functions, including insulin binding, receptor kinase activity, or IRS-1 phosphorylation, nor was it associated with changes to the structural focal adhesion protein, vinculin, or gross alterations in integrin expression. While a complete survey of focal adhesion and cytoskeletal proteins has not been conducted, the reduction of pp125 FAK phosphorylation and of paxillin protein in CHO.2N cells and the known correlation of these proteins with focal contact formation (14,39,40) suggest changes to focal contacts formed by CHO.2N cells may impair their ability to interact initially with a substratum. Overall, these results support a role for the insulin receptor in modulating cytoskeletal interactions and cell adhesion.
The signals generated as a result of pp125 FAK activation are currently unknown but it is apparent that autophosphorylation of pp125 FAK results in co-association of signaling molecules such as Src, PI 3-kinase, and GRB2, in addition to proteins such as paxillin which, in turn, binds C-terminal Src kinase and vinculin (2,8). Presumably generation of signals via this "facilitator complex" could maintain or change cell architecture via cytoskeletal rearrangements; a defect in this signaling system may result in the altered cell phenotype observed in CHO.2N cells. Cytoskeletal changes have already been reported following overexpression of some signaling proteins, including mitogen-activated protein kinase (41) and C-terminal Src kinase (42). Thus, the underphosphorylation of pp125 FAK in CHO.2N-10 cells may contribute to their altered morphology and adhesion. PI 3-kinase is a major signaling molecule activated following growth factor stimulation and has been implicated in pathways leading to cytoskeletal rearrangements, possibly upstream of Rac (5,43). The interaction of PI 3-kinase with pp125 FAK can be stimulated in vivo by either PDGF (18) or cell adhesion (44) concomitant with pp125 FAK activation; however, no increase in pp125 FAK -associated PI 3-kinase was seen in insulin-stimulated cells (18). Recently, it has been shown that blocking PI 3-kinase activity with wortmannin inhibits PDGF-stimulated pp125 FAK phosphorylation (45). While wortmannin and mutated p85 subunits of PI 3-kinase block insulin-stimulated membrane ruffling (5), our preliminary data show that wortmannin (100 nM) had no effect on the ability of insulin to stimulate dephosphorylation of pp125 FAK . Surprisingly in CHO.2N-10 cells, in the absence of insulin, a consistent increase in pp125 FAK phosphorylation was observed after 10 min of incubation with wortmannin, implying some interaction between PI 3-kinase and pp125 FAK in these cells. 4 Further investigations are under way to examine this linkage.
A decrease in phosphorylation of pp125 FAK could result from activation of a tyrosine phosphatase. A number of tyrosine phosphatases have been proposed to play a role, both positive and negative, in insulin action. Of these, LAR (transmembrane) PTP1B and Syp (cytosolic) appear in the literature most commonly linked with insulin signaling and modulation of their activity, by various means, may either enhance (LAR (46) and PTP1B (47)) or block (Syp (20)) insulin action. Phosphatases have also been linked with cytoskeletal activities and, for instance, changes in cell density (which affect the cytoskeleton) also lead to changes in the level of LAR, PTP1B, and Syp (48). Syp, with two phosphotyrosine SH2 binding domains, has been shown to bind directly to both phosphorylated insulin receptor and IRS-1 (49) and, when overexpressed in CHO cells, a dominant negative Syp mutant induced a significant increase in basal pp125 FAK phosphorylation (20). Furthermore, LAR has been localized to focal adhesions with a possible involvement in focal adhesion disassembly (50). Our initial data which shows that the tyrosine phosphatase inhibitor pervanadate can increase the basal phosphorylation of pp125 FAK in CHO.2N cells is suggestive of phosphatase involvement. A study to isolate putative "pp125 FAK " phosphatases is in progress.
The structural focal adhesion protein, paxillin, is closely linked with pp125 FAK via an association site in the C terminus of pp125 FAK . It is thought that paxillin is an in vivo substrate of pp125 FAK (2), and the phosphorylation of these two proteins usually increases and decreases in concert. However, in CHO.2N-10 cells, despite decreased basal phosphorylation of pp125 FAK , there was no parallel decrease in paxillin phosphorylation demonstrating that the two events are not always linked. A similar result was observed in phorbol 12-myristate 13-acetate down-regulated cells (33) and pp125 FAK -deficient mice (34). Little is known about the overall mechanism for focal contact formation and the contribution of various structural proteins such as paxillin. In this context, our observation that the level of paxillin protein is decreased by 40% in CHO.2N-10 cells could explain (at least in part) the decreased ability of these cells to adhere, despite no alteration to the level of vinculin.
The mechanism whereby a double point mutation (K164N/ K582N) in the external domain of the insulin receptor leads to the observed cell phenotype is unknown. One possibility is that the mutation results in a conformational change in the receptor's cytoplasmic domain such that a signal transduction pathway involved in cell architecture is modulated in the absence of insulin. Truncation of the insulin receptor C terminus and more specifically, mutation of tyrosines 1328 and 1334 in this region abolished the insulin responsive dephosphorylation of pp125 FAK , suggesting that the C terminus and in particular phosphorylation of these two tyrosine residues is critical for insulin effects on pp125 FAK phosphorylation (19). In addition, this region of the receptor and in particular Tyr 1328 and Tyr 1334 has been linked to Syp activation by insulin (51), and also to Shc (52), p85 subunit of PI 3-kinase (53), and Grb10 (54) association. Previous studies have demonstrated that insulin binding to its receptor induces a conformational change in the C terminus of the ␤-subunit prior to and independent of autophosphorylation (55). Again mutation of residues Tyr 1328 and Tyr 1334 of the insulin receptor alters this insulin-induced conformational change (55). Taken together these data suggest that modulation of the insulin receptor ␣-subunit (by binding insulin) can influence the conformational state of the C terminus and that these C-terminal sequences in turn may play a role in signal transmission. Perhaps the 2N mutation has also resulted in a C-terminal conformational change altering a pathway (containing pp125 FAK ) involved with cell morphology. As antibodies are available that recognize different conformational states of the receptor C terminus these hypotheses could be tested.
Finally, the recent demonstration by Richardson and Parsons (40) that overexpression of the pp125 FAK -related non-kinase (pp41/43 FRNK ) blocks pp125 FAK tyrosine phosphorylation and inhibits cell spreading supports the notion that alterations to pp125 FAK can be linked with changes to cytoskeletal interactions. Further studies will be required to elucidate the mechanisms involved in these changes in CHO.2N cells; however, our results linking expression of a mutant insulin receptor with altered cell phenotype add support to a role for the insulin receptor in organization of the cell cytoskeleton.