Insulin Receptor Substrate-1 (IRSI) and Shc Compete for a Limited Pool of Grb2 in Mediating Insulin Downstream Signaling*

Expression of the insulin receptor substrate-1 (IRS1) or Shc cDNA resulted in both increased protein and in-sulin-stimulated tyrosine phosphorylation of IRSl and Shc proteins, respectively. Although expression of Shc had no significant effect on insulin-stimulated mitogen-activated protein ( M A P ) kinase gel shift or c-fos tran- scriptional activation, expression of IRSl inhibited these responses. The effect of IRSl expression on the formation of multisubunit signaling complexes was determined by a series of indirect co-immunoprecipita- tions. Grb2 immunoprecipitation from IRS1-transfected and insulin-treated cells demonstrated an increased co- immunoprecipitation of Syp and the p85 regulatory subunit of the phosphatidylinositol 3-kinase. Similarly, cell extracts immunoprecipitated with a p85 antibody displayed an increased co-immunoprecipitation of Syp and Grb2. However, expression of IRSl increased the extent of Grb2 associated with IRSl with a concomitant reduc- tion in the amount of Grb2 associated with Shc. Further-more, increased expression of Shc reduced the amount of Grb2 bound to IRSl with a concomitant increase in Grb2 associated with Shc. Together, these data demonstrate that IRSl and Shc compete for a limited cellular pool of Grb2, and insulin activation of MAP kinase and c-fos transcription predominantly occur through the Shc-Grb2 signaling pathway.

More recently a second target, composed of two related proteins (46 and 52 kDa), termed Shc for Src homology S/cr-collagenrelated has been identified (10,11). These proteins contain insulin receptor-specific tyrosine phosphorylation sites responsible for their association with various downstream effector molecules. In the case of IRS1, these include binding sites for the SH2 domains of the phosphatidylinositol 3-kinase, protein tyrosine-specific phosphatase Syp, and the small adapter proteins . In contrast, tyrosine phosphorylation of the Shc proteins has only been shown to directly induce the association with Grb2 (12,16).
In addition to SH2 domains, many proteins involved in intracellular signaling events also contain Src homology 3 (SH3) domains which are responsible for the binding to various proline-rich motifs (17). For example, the SH3 domains of Grb2 bind to proline-rich sequences found in the microtubule-associated protein dynamin and the guanylnucleotide exchange factor for Ras, termed Son of Sevenless or SOS (18)(19)(20)(21). The association of the Grb2-SOS complex with tyrosine-phosphorylated receptors andor Shc have been directly implicated in the activation of the Ras signaling pathway (22)(23)(24)(25)(26). However, the relative insulin signaling properties between Grb2 association with IRSl or with Shc is highly controversial. Studies examining the interactions of SOS with IRSl and Shc were unable to detect any significant insulin-stimulated association of SOS with IRS1, although under identical conditions SOS was co-immunoprecipitated with Shc (27). In addition, the majority of Ras guanylnucleotide exchange activity was found in Shc immunoprecipitates with only a relatively small fraction associated with IRSl immunoprecipitates (28). These data would support a major role for the Shc-Grb2-SOS complex in mediating insulin activation of the Ras pathway. However, several studies have clearly demonstrated the presence of SOS in IRSl immunoprecipitates (29). Furthermore, expression of insulin receptor mutants that poorly tyrosine-phosphorylate Shc but normally tyrosine-phosphorylate IRSl also display complete activation of Ras (30). More recently, expression of IRSl has been shown to either activate or inhibit insulin signaling dependent upon other cell context factors (31,321. Thus at present, the relative role of IRSl and Shc in mediating insulin stimulation of downstream signaling is not at all clear and conflicting as well as overlapping evidence exists for both pathways.
To address this issue, we have developed a rapid and highly efficient transient transfection protocol in which both IRSl and Shc can be routinely expressed in 80-95% of the total cell population. This method has allowed us to examine the relationship between MAP kinase and c-fos transcription, as convenient readouts for insulin action, in comparison with the formation of IRSl and Shc signaling complexes. The data presented in this manuscript demonstrate that IRSl and Shc compete for a limiting pool of Grb2 and that the extent of Grb2 bound to Shc correlates with downstream insulin signaling. Thus, although IRSl and Shc can serve as mediators of insulin biological responsiveness, the specific signaling events ob-IRSl-Grb2 and Shc-Grb2 Signaling Pathways served are dependent upon the relative concentrations and stoichiometries of these multisubunit signaling complexes.
EXPERIMENTAL PROCEDURES Plasmid Constructs-The mammalian expression plasmid, CLDN, was obtained from SmithKline Beecham and utilizes the cytomegalovirus promoter to drive cDNA expression. The Shc cDNA was kindly provided by Dr. Alan Saltiel (Parke-DavisiWarner-Lambert) and was cloned into the CLDN expression vector. The preparation of CLDN-IRSl and the reporter gene constructs SRE-Luc and RSV-PGal have been described previously (32,33).
Cell Culture and Tkansient Tkansfection-CHO cells expressing -3 x 10, human insulin receptors/cell (CHODR) were obtained as described previously (34). These cells were maintained in minimal Eagle's medium containing nucleotides plus 10% fetal bovine serum. Cells were transiently transfected using the calcium phosphate co-precipitation method with CsCl double-banded DNA as described previously (33). Briefly, 16 h prior to use, the cells were plated a t 2 x 10, cells/lOO-mm dish and transfected with various plasmid DNAs totaling 23 pg. Twelve hours following transfection, the cells were glycerol shocked and placed into serum-free Ham's F-12 medium for 12 h.
In order to obtain a high degree of transfection efficiency necessary for immunoprecipitation and Western blotting of whole cell extracts, CHOiIR cells were electroporated with a total of 40 pg of plasmid DNA a t 340 volts and 960 microfarads. Under these conditions approximately 7040% of the electroporated CHOiIR cells were lysed; however, the surviving 20-30% of the cell population were fully viable. Within this cell population, approximately 80-95% of the cells were functionally transfected as determined by in situ staining for the expression of P-galactosidase activity (see Fig. 2). Thirty-six h following transfection the cells were serum-starved for 6 h and either untreated or incubated for 5 min in the presence of 100 rm insulin prior to the preparation of whole cell lysates.
Reporter Gene Actiuities-Whole cell extracts were prepared 5-6 h following 100 rm insulin treatment for the determination of luciferase (Luc) and P-galactosidase (P-gal) activities. All transfections were performed at least three times, in triplicate, using at least two different preparations of plasmid DNA. To correct for differences in transfection efficiencies between plates within an experiment, the luciferase activities in each extract were normalized to P-galactosidase activity.
Immunoprecipitations were performed by a 10-fold dilution of the detergent solubilized cell extracts (lysis buffer without Triton X-100) and incubation with 4 pg of a Shc polyclonal antibody (aShc, Upstate Biotechnology, Inc.), Grb2 polyclonal antibody (aGrb2, Santa Cruz), or p85 polyclonal (ap85, Upstate Biotechnology, Inc.) antibody for 2 h a t 4 "C. Since the carboxyl-terminal IRSl antibody used for Western blotting was not effective for immunoprecipitation, an IRSl rabbit polyclonal antibody prepared against the baculovirus-expressed protein was used (aIRS1, Upstate Biotechnology, Inc.). The samples were then incubated with protein A-Sepharose for 1 h at 4 "C. The resulting immunoprecipitates were subjected to SDS-polyacrylamide gel electrophoresis and Western blotted (enhanced chemiluminescence detection kit, Amersham Corp.) using a Grb2 monoclonal antibody (Transduction Laboratories), Syp monoclonal antibody (Transduction Laboratories), and p85 polyclonal antibody (Upstate Biotechnology, Inc.), as indicated in the individual figure legends. vector. Twenty-four hours following transfection, the cells were serumstarved for 6 h and either untreated (open boxes) or incubated with 100 nM insulin for 6 h (hatched boxes). Cell extracts were prepared and assayed for luciferase and P-galactosidase activities. The data are presented as the amount of SRE-Luc activity divided by the amount of P-galactosidase (@-Gal) activity present in the extract. These data represent the average with the standard error of the mean from four independent experiments each performed in triplicate.

RESULTS
It has previously been observed that expression of IRSl can lead to either an enhancement or inhibition of insulin signaling, depending upon the amount of IRSl present as well as other cell context factors (31, 32). For example, expression of IRSl in parental CHO cells was observed to increase insulin stimulation of DNA synthesis, whereas expression of IRSl in CHO/IR cells reduced this response (31). To further characterize this phenomenon, CHO/IR cells were co-transfection with an insulin-responsive c-fos reporter gene construct (SRE-Luc) and the RSV-PGal reference gene using the CaPO, method as described previously (32, 33). In these transiently transfected cells, insulin treatment typically resulted in an approximate 10-fold specific increase in SRE-Luc reporter gene activity ( Fig.  1). However, co-transfection with the expression plasmid for IRSl (CLDN-IRS1) had no effect on basal reporter gene expression, but attenuated the extent of insulin-stimulated SRE-Luc activity to approximately 40% of control cells. In contrast, expression of Shc had no significant effect on either basal or insulin-stimulated SRE-Luc activity.
Typically, to investigate the molecular basis for this IRS1dependent inhibition of insulin signaling stable cell lines expressing various cDNAs would be prepared and subsequently analyzed. In order to develop a rapid transient transfection method suitable for this type of cell biological analysis, we have adapted electroporation conditions to obtain 80-95% transfection efficiencies in CHOiIR cells (Fig. 2). Electroporation of CHO/IR cells at 960 microfarads and 340 V resulted in a loss of viability in approximately 70-80% of the cells (data not shown). The remaining 20-30% of the surviving cells, however, displayed normal growth characteristics. CHO/IR cells mocktransfected with 40 pg of the empty CLDN expression vector did not stain positive for P-galactosidase activity (Fig. 2 A ) . In contrast, cells transfected with 40 pg of a P-galactosidase reporter plasmid (CLDN-PGal) demonstrated positive in situ staining in 80-95% of cells, depending upon the particular experiment ( Fig. 2B). As controls, CHOiIR cells were also transfected with the empty CLDN vector (Fig. 2C) and CLDN-PGal ( Fig. 2 0 ) using the standard CaPO, method. Under these conditions, only a small percentage of the cells (-1-3%) stained positive for P-galactosidase expression. These data demonstrated that electroporation under relatively high voltage and capacitance conditions resulted in a highly efficient transient transfection of CHOiIR cells. To assess whether this eEcient transient transfection procedure was amenable to the expression of various intracellular effector proteins mediating insulin receptor signaling, we next determined the relative extent of I R S l and Shc protein expression using this method. CHODR cells were electroporated with cDNAs encoding for I R S l and Shc, and whole cell extracts were then immunoblotted with specific antibodies for IRSl and Shc Pig. 3). Mock-transfected CHOmE cells displayed weak immunoblotting with the IRS1-specific antibody which was not readily visualized at this exposure level (Fig. 3A, lunes I and2 1. h l o n g e d exposure of the autaradiogram indicated identical amounts of immunoblotted IRSl in the basal and insulinstimulated states (data not shown). However, following electroporation with the I R S l expression plasmid, there was a large increase in the amount of immunoblotted I R S l protein (Fig. 3A,  lams 3 and 4). Similar to the control cells, identical amounta of IRSl protein were observed in both the basal and insulinstimulated states. In addition, insulin treatment resulted in a somewhat decreased mobility of IRS1, suggesting covalent modification by phosphorylation.
As observed for IRS1, there was also a relatively low level of immunoblotted Shc proteins in control cells which did not significantly change following insulin-stimulation (Fig. 3B, lanes  I and 2). Again, following electroporation with the Shc expression plasmid (CLDN-Shc), there was a large increase in the amount of immunoblotted Shc proteins of 52 and 46 kDa (Fig.  3B, lanes 3 and 4). Similar to the control (mock-transfected) cells, identical amounts of the Shc proteins were observed in both the basal and indin-stimulated states.  4 ) . Thirty-six hours later, the cells were serum-starved for 6 h and subsequently untreated (lanes 1 and 3) or incubated with 100 nM insulin (lanes 2 and 4 ) for 5 min prior to detergent solubilization. The whole cell extracts were then subjected to SDS-polyacrylamide gel electrophoresis and Western blotting with an IRS1-specific antibody (alRS1) as described under "Experimental Procedures." B, CHOiIR cells were electroporated with 40 pg of the empty CLDN vector (lanes 1 and 2 ) or 40 pg of CLDN-Shc (lanes 3 and 4 ) . The cells were then serum-starved and either left untreated (lanes 1 and 3) or stimulated for 5 min with 100 nM insulin (lanes 2 and 4 ) followed by Shc Western blotting (aShc) as indicated in A. These were representative Western blots independently performed four times for IRSl transfected cells and three times for Shc-transfected cells.  (lanes 1 and 2 ) or 40 pg of CLDN-IRS1 (lanes  3 and 4 ) . Thirty-six hours later, the cells were serum-starved for 6 h and subsequently untreated (lanes 1 and 3 ) or incubated with 100 nM insulin (lanes 2 and 4 ) for 5 min prior to detergent solubilization. The whole cell extracts were then subjected to SDS-polyacrylamide gel electrophoresis and Western blotting with a phosphotyrosine-specific antibody (&Y) as described under "Experimental Procedures." B, CHO/IR cells were electroporated with 40 pg of the empty CLDN vector (lanes 1 and 2 ) or 40 pg of CLDN-Shc (lanes 3 and 4 ) . The cells were then serumstarved and either left untreated (lanes 1 and 3 ) or stimulated for 5 min with 100 nM insulin (lanes 2 and 4 ) followed by phosphotyrosine Western blotting as indicated in A. These were representative Western blots independently performed four times for IRS1-transfected cells and three times for Shc-transfected cells.
SRE-Luc reporter plasmid demonstrated an approximate 8-fold specific increase in luciferase activity compared with unstimulated cells. As previously observed for CaPO, transfected cells (Fig. l), expression of IRSl inhibited the extent of insulinstimulated SRE-Luc reporter activity without any effect on basal activity. In addition, expression of Shc again had no significant effect on either basal or insulin-stimulated SRE-Luc activity. These data demonstrate that transfection of these cDNAs by electroporation mirrors the insulin stimulation of SRE-Luc activity determined by the standard CaPO, transfection procedure (Fig. 1). Furthermore, the data presented in Figs. 2-4 demonstrate that the rapid transient transfection obtained by electroporation was essentially equivalent to the preparation of stable cell lines.
Recently, one complete pathway linking the insulin receptor to the transcriptional activation of the c-fos SRE has been identified (35-38). This pathway requires Ras activation as an upstream mediator of the RafMEWMAP kinase pathway, resulting in the phosphorylation and activation of several transcription factors (TCF and SRF) necessary for SRE responsiveness. To determine if the IRS1-mediated inhibition of insulin-stimulated SRE-Luc activity resulted from a blockade of MAP kinase activation, MAP kinase gel shift assays were performed (Fig. 6). The 44-and 42-kDa forms of MAP kinase (ERK1 and ERK2, respectively) have been established to require both tyrosine and threonine phosphorylation for activation (39, 40). These phosphorylation events result in a characteristic decreased mobility of MAP kinase on SDSpolyacrylamide gel electrophoresis (41,42). As can been seen in Fig. 6 A , insulin treatment of control cells induced a nearly complete shift in the mobility of the 42-kDa (ERK2) MAP kinase ( Fig. S A , lanes 1 and 2 ) . In contrast, CHO/IR cells transfected with IRSl displayed a reduction in the proportion of gel shifted MAP kinase (Fig. S A , lanes 3 and 4 ) . Consistent with the expression of Shc having no effect on insulin stimulation of SRE-Luc reporter gene activity (Figs. 1 and 4), Shc expression did not alter the ability of insulin to induce the mobility shift of MAP kinase (Fig. 6B, lanes 1 4 ) .
There are several possible mechanisms that could potentially account for inhibition of insulin signaling by IRSl overexpression, but not Shc. Since tyrosine-phosphorylated IRSl simultaneously associates with multiple effector proteins, high intracellular levels of tyrosine-phosphorylated IRSl could effectively dilute out these proteins such that only a single SH2 domain containing target protein was bound to a given IRSl molecule. To examine this possibility, we performed a series of co-immunoprecipitations between Grb2, Syp, and the p85 regulatory subunit of the phosphatidylinositol (PI) 3-kinase which have been established to form a multisubunit signaling complex with IRSl(43). As expected, immunoprecipitation of Grb2 from control (mock-transfected) cells demonstrated an insulindependent co-immunoprecipitation of Syp (Fig. 7A, lanes 1 and  2 ) and the p85 regulatory subunit of the PI 3-kinase (Fig. 7B, CLDN vector (lanes 1 and 2 ) or 40 pg of CLDN-IRS1 (lanes 3 and 4 ) . Thirty-six hours later, the cells were serum-starved for 6 h and subsequently untreated (lanes 1 and 3 ) or incubated with 100 nM insulin (lanes 2 and 4 ) for 5 min prior to detergent solubilization. The whole cell extracts were then subjected to SDS-polyacrylamide gel electrophoresis and Western blotting with a MAP kinase (ERK1 and -2)-specific antibody as described under "Experimental Procedures." B, CHO/IR cells were electroporated with 40 pg of the empty CLDN vector (lanes 1 and 2 ) or 40 pg of CLDN- Shc  (lanes 3 and 4). The cells were then serum-starved and either left untreated (lanes 1 and 3 ) or stimulated for 5 min with 100 nM insulin (lanes 2 and 4 ) followed by MAP kinase Western blotting as indicated in A. These were representative Western blots independently performed two times each for the IRSl and the Shc-transfected cells.  3 and 4 ). Thirty-six later, the cells were serum-starved for 6 h and incubated in the absence (lanes 1 and 3) or presence (lanes 2 and 4 ) of 100 nM insulin for 5 min. The cells were detergent-solubilized, and the whole cell extracts were immunoprecipitated with a GrbP antibody (crGrb2) as described under "Experimental Procedures." The resultant immunoprecipitates were then subjected to Western blotting for Syp (A ) or the p85 regulatory subunit of the PI 3-kinase (B). These were representative Western blots independently performed two times for Syp and three times for p85.

Shc-Grb2 Signaling
lanes 1 and 2 ) . Expression of IRSl resulted in a substantial increase in the amount of Syp (Fig. 7A, lanes 3 and 4 ) and p85 (Fig. 7B, lanes 3 and 4 ) that was co-immunoprecipitated following insulin treatment.
In a complementary study, we next co-immunoprecipitated Grb2 and Syp using an antibody directed against the p85 regulatory subunit of the PI 3-kinase (Fig. 8). Immunoprecipitation with the p85 antibody from control cells treated with insulin demonstrated the co-immunoprecipitation of Syp (Fig.  8 A , lanes 1 and 2 ) and Grb2 (Fig. 8B, lanes 1 and 2). Similar to Fig. 7, expression of IRSl increased the amount of co-immunoprecipitated Syp (Fig. 8 A , lanes 3 and 4 ) and Grb2 (Fig. 8B,  lanes 3 and 4 ) in the p85 immunoprecipitate from insulinstimulated cells. Taken together, these data demonstrate that a sufficient accessible cellular pool of Syp, Grb2, and p85 effector proteins were present to form a multisubunit complex even in the presence of increased IRSl expression. Thus, high levels of IRSl did not inhibit insulin-dependent signaling by preventing  lanes 3 and 4 ) . Thirty-six hours later, the cells were serum-starved for 6 h and incubated in the absence (lanes 1 and 3 ) or presence (lanes 2 and 4 ) of 100 nM insulin for 5 min. The cells were detergent-solubilized, and the whole cell extracts were immunoprecipitated with a p85 antibody as described under "Experimental Procedures." The resultant immunoprecipitates were then subjected to Western blotting for Syp (A) or the Grb2 (B). These were representative Western blots independently performed two times each for Syp and Grb2.
CLDN-IRS1 --+ + CLDN-IRS1 --+ + the formation of a multisubunit IRSl signaling complex. In addition to IRS1, Grb2 has also been shown to directly associates with tyrosine-phosphorylated Shc (12, 16). Previous studies have also suggested that Grb2 may be rate-limiting for the activation of the Ras pathway (27,32,44). Thus, increased IRSl expression could potentially compete for Grb2 binding to Shc. To further test this hypothesis, we next determined the amount of Grb2 associated with IRSl and Shc (Fig. 9). In the absence of insulin there was no specific Grb2 protein detected in the IRSl immunoprecipitate (Fig. SA, lanes 1 and 3 ) , whereas following insulin stimulation, immunoprecipitation of IRSl resulted in the specific association with Grb2 (Fig. 9A,  lane 2 ) . As expected, increased expression of IRSl resulted in an increase in the amount of Grb2 bound to IRSl following insulin stimulation (Fig. 9A, compare lanes 2 and 4 ) . Similarly, Shc immunoprecipitation demonstrated the presence of co-immunoprecipitated Grb2 from insulin-stimulated cells (Fig. 9B,  lanes 1 and 2 ) . In contrast, increased expression of IRSl reduced the extent of insulin-stimulated Grb2 associated with Shc (Fig. 9B, lanes 3 and 4 ) . Quantitation of these data indicated that increased expressed of IRSl reduced the amount of Grb2 associated with Shc by 67 5% relative to the untransfected control cells. Thus, these data demonstrate that high levels of IRSl expression compete with Shc for a limited pool of Grb2 molecules.
This being the case, then increased expression of Shc should also reduce the amount of Grb2 associated with IRS1. To confirm this prediction, immunoprecipitation of IRSl from control cells demonstrated the co-immunoprecipitation of Grb2 in the insulin-stimulated state (Fig. lOA, lanes 1 and 21, whereas increased expression of Shc reduced the amount of Grb2 coimmunoprecipitated with IRSl (Fig. lOA, lanes 3 and 4). Similarly, Shc immunoprecipitation demonstrated the insulin-dependent association with Grb2 (Fig. 10B, lanes 1 and 2 ) . As expected, increased expression of Shc resulted in an increase in the extent of insulin-stimulated Grb2 bound to Shc (Fig. lOB,  lanes 3 and 4 ) . Thus, these data directly demonstrate that the formation of both IRS1-Grb2 and Shc-Grb2 signaling complexes were in direct competition with each other.  (lanes 1 and 2 ) or 40 pg of CLDN-IRS1 (lanes 3 and 4). Thirty-six hours later, the cells were serum-starved for 6 h and incubated in the absence (lanes 1 and 3 ) or presence (lanes 2 and 4 ) (lanes 1 and 2 ) or 40 pg of CLDN-Shc (lanes 3 and 4 ) . Thirty-six hours later, the cells were serum-starved for 6 h and incubated in the absence (lanes 1 and 3 ) or presence (lanes 2 and 4 ) of 100 n M insulin for 5 min. The cells were detergent-solubilized, and the whole cell extracts were immunoprecipitated with a IRSl (A) or Shc ( B ) antibody as described under "Experimental Procedures." The resultant immunoprecipitates were then subjected to Western blotting for Grb2 (aGrb2). These were representative Western blots independently performed three times each for the immunoprecipitation of IRSl and Shc.

DISCUSSION
Expression of various wild type and mutant cDNAs in tissue culture cells has provided a very powerful tool to dissect the signaling pathways linking receptor tyrosine kinases to downstream biological responses. In the vast majority of cases, this has required the preparation and isolation of stable cell lines expressing cDNAs of interest. In addition to the time required to generate these stable cell lines, there are several concerns over clonal selection and compensatory cell context changes that may occur during isolation. To circumvent these difficul-ties, we examined the transfection efficiency of electroporation under high voltage and capacitance conditions in CHODR cells. Electroporation with an expression vector for the LacZ gene resulted in the expression of P-galactosidase activity in no less than 80%, and in most cases greater than 958, of the viable cell population. This method was applicable to several other proteins as electroporation of expression plasmids for IRSl and Shc also markedly increased their protein levels.
It is also important to recognize that following this procedure only approximately 25% of the total cell population remained viable. However, this surviving cell population displayed the identical characteristics of the total initial cell population, in terms of c-fos transcriptional activation, cell growth, and IRS1but not Shc-dependent inhibition of insulin signaling. This suggests that there was no selection pressure induced by the electroporation procedure itself or via the introduction of these expression plasmids which would result in a nonrepresentative cell population. Thus, these data presented in this manuscript demonstrated that electroporation can be used as an effective procedure to generate a large population of cells expressing a particular gene of interest similar to the preparation of stable cell lines.
It is interesting to note that electroporation with the expression plasmid for Shc resulted in high levels of both the 52-and 46-kDa isoforms of Shc. The presence of these two protein products from this cDNA was consistent with the alternative translation start site usage from a single Shc transcript (10). In addition, previous studies have demonstrated that platelet-derived growth factor and epidermal growth factor stimulation resulted in the tyrosine phosphorylation of both Shc species, presumably at the carboxyl-terminal Wl7 residue (12). However, we have observed that only the 52-kDa Shc protein was tyrosine-phosphorylated in response to insulin in both CHO/IR and CHO/IR cells transfected with the expression plasmid for Shc. Since the amino-terminal 58 amino acids that were deleted in the 46-kDa Shc protein do not contain any tyrosine residues, this differential insulin-dependent tyrosine phosphorylation most likely reflects a regulatory role for this domain in mediating substrate specificity. Alternatively, this domain may be responsible for differential targeting of the 52-and 46-kDa isoforms to distinct subcellular locations. The possibility that this distinctive pattern of Shc tyrosine phosphorylation may underlie tyrosine kinase receptor signaling specificity will require further investigation.
In any case, having demonstrated the utility of electroporation to express these proteins, we wished to examine the relative signaling role of IRSl and Shc in mediating biological actions of insulin. It has been well documented that increased expression of IRSl in several cell types increased insulin sensitivity and responsiveness of DNA synthesis (31,45,46). In addition, expression of antisense IRSl RNAor microinjection of IRSl antibodies also inhibited insulin-stimulated DNA synthesis and growth (47, 48). Surprisingly, however, increased expression of IRSl in a cell line expressing very high levels of the insulin receptor (CHO/IR) resulted in an attenuation of insulinstimulated DNA synthesis as well as in c-fos transcription (31, 32). In the present study, we have confirmed that high levels of IRSl expression inhibited insulin stimulation of SRE-Luc activity which correlated with a blockade of MAP kinase gel shift and, thus, kinase activation.
There are several general mechanisms that can be envisioned to account for this apparent paradoxical effect of IRSl expression on insulin signaling. First, since IRSl functions as a multisite docking protein for several SH2 domain containing effector proteins, an over abundance of tyrosine-phosphorylated IRSl could titrate out these proteins. Under such a cir-cumstance, only a single SH2 domain target protein would be bound to any given IRSl molecule, thereby preventing the formation of a multisubunit signaling complex. This possibility was examined by co-immunoprecipitation of the p85 regulatory subunit of the PI 3-kinase, Syp, and Grb2, all of which have been previously demonstrated to indirectly associate via binding t o tyrosine-phosphorylated IRSl (5, 6). However, the data presented in this manuscript demonstrated that, under identical conditions which inhibited the insulin stimulation of MAP kinase phosphorylation and SRE-Luc expression, there was sufficient amounts of p85, Syp, and Grb2 to generate a multisubunit complex with IRSl as the core molecule.
Having excluded this potential mechanism, it was also possible that increased expression of IRSl resulted in a sequestration of protein components away from other necessary signaling pathways. In this regard, both tyrosine-phosphorylated IRSl and Shc bind to the SH2 domain of the small adapter protein Grb2 (12,16). Grb2 has also been shown to directly associate with the Ras guanylnucleotide exchange factor SOS, thereby linking Ras activation via the stimulation and/or appropriate targeting of SOS (20-26). Consequently, the SOS-mediated increase in GTP bound Ras provides a physical interaction with Raf initiating downstream activation of the Raf7hfEK/MAP kinase cascade (49)(50)(51)(52)(53)(54)(55)(56). Thus, the insulin stimulation of MAP kinase and SRE-Luc transcriptional activation reflect insulin signaling via combinations of either Shc-Grb2-SOS and/or IRS1-Grb2-SOS interactions.
Since the involvement of Grb2 has been established in the MAP kinase pathway leading to SRE-Luc activation, we reasoned that relatively high levels of IRSl expression could have recruited or sequestered Grb2 away from Shc. Alternatively, it was possible that increased IRSl expression could have engaged and/or activated an inhibitory protein to the Grb2-SOS signaling pathway. However, the data presented in this manuscript demonstrate that the amount of Grb2 associated with Shc was decreased in IRSl over expressing cells, whereas the amount of Grb2 bound to IRSl was increased. In addition, increased expression of Shc resulted in a loss of IRSl bound Grb2 with a concomitant increase in Shc associated Grb2. These data demonstrate that Shc and IRSl compete for a limited pool of Grb2. Furthermore, since increased expression of IRSl inhibited insulin responsiveness, whereas Shc had no effect, strongly suggests that the Shc-Grb2 pathway was predominant over the IRS1-Grb2 pathway in mediating insulin signaling for MAP kinase and c-fos transcriptional activation.
In summary, the data present in this manuscript demonstrate the use of electroporation as a highly efficient method to introduce and functionally express various proteins involved in insulin receptor tyrosine kinase signaling. This rapid method for protein expression has several advantages over the preparation of stable cell lines most notably the absence of clonal selection which may introduce significant changes in cell context properties. Utilizing this method, we have determined the basis for IRS1-mediated inhibition of insulin downstream signaling which only occurs in cell expressing high levels of both the insulin receptor and IRS1. The inhibition of insulin signaling under these conditions resulted from a sequestration of a limited pool of Grb2 away from Shc due to the high level of tyrosine-phosphorylated IRSl relative to Shc. Furthermore, these data demonstrate that the formation of the Shc-Grb2 complex rather than an IRS1-Grb2 complex was the major pathway for insulin-stimulated MAP kinase and c-fos transcriptional activation. 14.