Functional Expression of Insulin Receptor Substrate-1 Is Required for Insulin-stimulated Mitogenic Signaling*

To examine the role of the insulin receptor substrate-1 (IRS-1) in mediating insulin biological responsiveness, we generated Chinese hamster ovary cell lines express- ing antisense IRS-1 RNA. These cells displayed morphological alterations as well as markedly reduced growth rates compared to the parental cells. Furthermore, the antisense IRS-1 cell lines had decreased insulin-stimu-lated IRS-1 tyrosine phosphorylation, reduced phospha- tidylinositol 3-kinase activation, and decreased thymidine incorporation relative to the parental cell line. Insulin-dependent transcriptional regulation of a serum response elementfluciferase reporter construct (SRE-Luc) was also reduced in the antisense IRS-1-expressing cell lines. However, co-transfection with a plasmid di-recting the expression of rat IRS-1 fully restored insulin- stimulated SRE-Luc activity in the IRS-1 antisense cell lines. Thus, the inhibition in insulin signaling was a specific effect of decreased IRS-1 tyrosine phosphorylation. Taken together, these data demonstrate that insulin regulation of mitogenic signaling requires the functional expression of IRS-1 and documents its central im- portance in the insulin intracellular signaling pathway.

To examine the role of the insulin receptor substrate-1 (IRS-1) in mediating insulin biological responsiveness, we generated Chinese hamster ovary cell lines expressing antisense IRS-1 RNA. These cells displayed morphological alterations as well as markedly reduced growth rates compared to the parental cells. Furthermore, the antisense IRS-1 cell lines had decreased insulin-stimulated IRS-1 tyrosine phosphorylation, reduced phosphatidylinositol 3-kinase activation, and decreased thymidine incorporation relative to the parental cell line. Insulin-dependent transcriptional regulation of a serum response elementfluciferase reporter construct (SRE-Luc) was also reduced in the antisense IRS-1-expressing cell lines. However, co-transfection with a plasmid directing the expression of rat IRS-1 fully restored insulinstimulated SRE-Luc activity in the IRS-1 antisense cell lines. Thus, the inhibition in insulin signaling was a specific effect of decreased IRS-1 tyrosine phosphorylation. Taken together, these data demonstrate that insulin regulation of mitogenic signaling requires the functional expression of IRS-1 and documents its central importance in the insulin intracellular signaling pathway.
Recently, a major proximal intracellular target for the insulin receptor tyrosine-specific protein kinase has been identified as a 185-kDa protein (11, termed IRS-1,' for insulin receptor substrate-1 (2-6). This molecule contains numerous potential tyrosine phosphorylation sites, six of which are found within a Y" motif (Tyr-Met-Xaa-Met) and three in the related motif YXXM (2). It is generally thought that tyrosine phosphorylation of IRS-1 provides a multimeric docking substrate for specific SH2 (src homology-2) domain-containing proteins. In the case of IRS-1, this signaling complex has been suggested t o include the phosphatidylinositol 3-kinase (PI 3-kinase) (71, the SH2 domain-containing tyrosine-specific phosphatase (8), and the small adaptor molecule Grb2 (9,10). Although Grb2 has been reported to mediate the coupling of tyrosine-phosphorylated receptors with guanylnucleotide exchange factor (11)(12)(13)(14), a functional role for the PI 3-kinase or SH2 domain-containing tyrosine-specific phosphatase in mediating insulin signaling has n o t been established. In addition, the specific role of IRS-1 in mediating insulin signaling has also been difficult to address experimentally due to its ubiquitous tissue and cell type distribution (5,15). Although expression of m a m m a l i a n IRS-1 in unprimed oocytes has been reported to augment insulin stimulation of germinal vesicle breakdown (16), expression of IRS-1 in cells containing high levels of insulin receptors has also been observed to attenuate the insulin-stimulated increases in thymidine incorporation (15). In o r d e r t o f u r t h e r assess the role of IRS-1 in mediating insulin signaling, w e h a v e established two clonal CHO cell lines expressing antisense IRS-1 RNA. These data demonstrate that functional expression of the IRS-1 protein is essential for insulin signaling and the mitogenic potential of these cells.

EXPERIMENTAL PROCEDURES
Isolation of Antisense IRS-1-expressing Cells-The mammalian expression vector (CLDN) used in these studies was provided by Smith-Kline Beecham (King of Prussia, PA). It is a high copy number plasmid (pUC19) containing the dihydrofolate reductase gene for methotrexate selection and the neomycin resistance gene for Geneticin (neomycin analog) selection. The 5' end (-1 kilobase pair) of IRS-1 was cloned in the negative orientation into CLDN and transfected into DG44 cells (a CHO cell line derivative) using Lipofectin (Life Technologies, Inc.). Stable cell lines resistant to 500 pg/ml Geneticin were selected, and the antisense IRS-1 RNA was amplified once with 20 nM methotrexate.
Cell Culture-The CHO cell line used in these studies (DG44/neo) and the two IRS-1 antisense clonal lines SRI-5 and SRI-9 were maintained in a-minimal essential medium (Diabetes and Endocrinology Research Center, University of Iowa) supplemented with 10% fetal bovine serum and 300 pg/ml Geneticin. Prior to stimulation with insulin, the cells were arrested at 80% confluence by overnight incubation with serum-free a-minimal essential medium containing 0.5% bovine serum albumin. Cells were incubated in the presence or absence of 100 nM insulin for 5 min a t 37 "C, followed by ice-cold washes in phosphate buffered saline, and then frozen in liquid N,. Whole cell lysates were prepared by thawing the cells in 1 ml of 1% Triton X-100 lysis buffer (50 mM Hepes, pH 7.8, 1% Triton X-100, 2.5 mM EDTA, 100 mM sodium fluoride, 10 mM sodium pyrophosphate, 2 mw sodium vanadate, 1 mM phenylmethylsulfonyl fluoride, 2 p~ pepstatin, 0.5 trypsin inhibitory units of aprotinin, and 10 VM leupeptin). Triton X-100-insoluble material was removed by centrifugation at 100,000 x g for 60 min, and total supernatant protein was determined by the method of Bradford.
Phosphatidylinositol 3-Kinase Assay-Cell lysates were immunoprecipitated overnight a t 4 "C with either an aIRS-1 antibody (kindly provided by Dr. Alan Saltiel, Parke Davis) or with ap85 (purchased from Upstate Biotechnology Inc.). The amount of primary antibody required to maximally immunoprecipitate the antigen from a given amount of total protein was determined in separate experiments (data not shown). Immune complexes were hound by adding 50 pl of a protein A-Sepharose suspension (1:l with phosphate-buffered saline) for 2 h and centrifuged a t 10,000 x g for 1 min. The immunoprecipitates were assayed for PI 3-kinase activity essentially as previously described (17).
Phosphotyrosine Blotting-The whole cell lysates (400 pg of protein) were separated on reducing SDS-polyacrylamide gels (7.5%) and transferred to nitrocellulose membranes ( 1 A, 90 min, 4 "C) using Towbin's transfer buffer (18) containing 0.075% SDS and 0.5 mM sodium vanadate. The membranes were incubated for 2 h at 23 "C with 0.3 pg/ml aPY (phosphotyrosine antibody provided by Dr. Peter Wilden, University of Missouri). Following extensive washing the nitrocellulose membranes were incubated 30 min at 23 "C with '251-protein A, washed Inhibition of Insulin Signaling by Antisense IRS-1 RNA again, and exposed to Kodak XAR film a t -80 "C. Cell Growth and PHIThymidine Incorporation-Cells were plated a t 1 x lo5 celld35-mm well, and 24 h later the growth media was replaced with serum-free a-minimal essential medium containing 0.5% bovine serum albumin. Cells were incubated with various concentrations of insulin for 18 h a t 37 "C and pulse-labeled with 0.5 pCi/ml [:3Hlthymidine for an additional 3 h. The labeled cells were washed, solubilized in 1 ml of phosphate-buffered saline, 0.1% SDS, and trichloroacetic acid was added to a final concentration of 10%. The trichloroacetic acid pellet was solubilized with 1 N NaOH, neutralized with HCI and the amount of incorporated ["Hlthymidine determined by liquid scintillation counting. In parallel, cell number was determined a t different times after plating by trypsinization and counting the released cells with a hemocytometer.
SRE-Luc Activity-The insulin stimulation of SRE-Luc transcription was determined a s previously described (19). The plasmids used for these experiments were SRE-Luc, RSV-LacZ, and the entire rat IRS-1 cDNA (kindly provided by Dr. Morris White, Joslin Diabetes Center) subcloned into CLDN to generate the mammalian expression plasmid CLDNflRS-1. Briefly, cells were cotransfected with 1 pg of the SRE-Luc reporter gene and 2 pg of RSV-LacZ, either with or without 20 pg of the rat IRS-1 expression plasmid (CLDNIIRS-1). Twelve h later the cells were glycerol-shocked, serum-starved for an additional 12 h, and then incubated with or without 100 nM insulin for 6 h. Cell extracts were prepared and assayed for both luciferase and @-galactosidase activity.

RESULTS AND DISCUSSION
Several studies have demonstrated that IRS-1 is a proximal target for the insulin receptor kinase, which upon tyrosine phosphorylation associates with multiple intracellular signaling molecules including the PI 3-kinase (20-22), Grb2 (9, lo), and SH2 domain-containing tyrosine-specific phosphatase (8).
In order to assess the role of IRS-1 in mediating insulin receptor downstream signaling, we transfected DG44 cells (CHOderived cell line lacking dihydrofolate reductase) with the antisense IRS-l/CLDN construct. Following selection with 500 pg/ml Geneticin and a single round of amplification with 20 nM methotrexate, we isolated 11 individual clones. Of these, two clones had a substantially reduced rate of growth compared to DG44heo cells (CLDN vector without insert, Fig. 1A). These two clonal cell lines were termed SRI-5 and SRI-9. Any further attempts to amplify the antisense RNA by increased methotrexate selection resulted in a complete cessation of cell growth.
Since one of the cellular effects of insulin is the stimulation of DNA synthesis, we also examined the effect of insulin on DNA synthesis ([3H]thymidine incorporation) in the DG44/neo, SRI-5, and SRI-9 cell lines (Fig. 1B). As typically observed, stimulation of the DG44/neo cells with saturating insulin concentrations stimulated a 19-fold increase in [3H]thymidine incorporation with an ECS0 of approximately 10 nM. In contrast, maximal insulin stimulation of L3H1thymidine incorporation was greatly inhibited in both the SRI-5 and SRI-9 antisense cell lines. Despite the large decrease in maximal responsiveness, insulin sensitivity was apparently unchanged between the control and antisense SRI-5 and SRI-9 cell lines. The inhibition of insulin-stimulated L3H1thymidine incorporation observed in the antisense cell lines (Fig. 1 B ) was much greater than the inhibition of growth rate (Fig. IA). However, it should be noted that the growth rate experiments were performed in the presence of serum, whereas the thymidine incorporation experiments were done in the absence of serum. Thus, the relatively smaller inhibition of growth probably reflects the presence of other growth factors in serum that are mediated by intracellular pathways not utilizing IRS-1. Despite the differing magnitudes of inhibition, these data demonstrate that expression of antisense IRS-1 RNA in DG44 cells resulted in a marked reduction in insulin-stimulated DNA synthesis and cell growth.
In addition to alterations in growth properties, these clonal cell lines also displayed several morphological differences (Fig.  2). DG44/neo cells had a normal fibroblastic appearance,

celld35-mm dish and incubated in
( I minimal essential medium supplemented with 10% fetal bovine serum for the times indicated. At the end of each incubation period, the cells were scraped and the total number of cells in each well determined as described under "Experimental Procedures." Error burs indicate the standard error of triplicate determinations. € 3 , control DG44/neo, SRI-5, and SRI-9 cells were plated a t 1 x 10" celld35-mm dish and cultured overnight as described above. Prior to confluence, cells were incubated overnight in serum-free media containing 0.5% bovine serum albumin, followed by the addition of the indicated concentrations of insulin for 18 h. Three h before harvest, the cells were pulse-labeled with 0.5 pCi/ml ["Hlthymidine and its incorporation into DNA was determined by trichloroacetic acid precipitation and liquid scintillation counting. This is a representative experiment independently conducted two times. whereas the SRI-5 and SRI-9 cells were substantially more refractile with a more rounded shape. These differences were more apparent a t lower cell densities and tended to disappear as the cells approached confluence (data not shown).
To determine whether these growth and morphological changes were a consequence of decreased IRS-1 tyrosine phosphorylation, we next performed phosphotyrosine Western blot analysis (Fig. 3, A and B ). In the absence of insulin, one major tyrosine-phosphorylated band occurred a t approximately 120  (lanes 1,3,5, and 7) or presence (lanes 2,4,6, and 8) of 100 nM insulin for 5 min at 37 "C. Cell lysates containing 400 pg of total protein were separated on SDS-polyacrylamide gels (7.5%), transferred to nitrocellulose, and immunoblotted with aPY. E, quantitation of IRS-1 tyrosine phosphorylation by densitometric analysis from four independent experiments. C, comparison of insulin-stimulated PI 3-kinase activity in control DG44/neo, SRI-5, and SRI-9 cell lines. Control and IRS-1 antisense RNA cell lines were incubated with 100 nM insulin for 5 min, and cell lysates were prepared and immunoprecipitated with aIRS-1 and ap85. PI 3-kinase enzyme activity in the aIRS-1 immunoprecipitates were normalized with respect to total cellular PI 3-kinase activity determined in the ap85 immunoprecipitates. The relative PI 3-kinase activity was expressed as a percentage of that found in the DG44/neo cells. These data were from three independent experiments. kDa in the control cells (Fig. 3A, lane 1). This band has been observed in the phosphotyrosine blots of other cell types and was constitutively tyrosine-phosphorylated in an insulin-independent manner (1). In CHO cells overexpressing the human insulin receptor (CHOhIR), insulin stimulation for 5 min resulted in a marked increase in tyrosine phosphorylation of the 185-kDa IRS-1 protein and the 95-kDa P-subunit of the insulin receptor (Fig. 3A, lanes 1 and 2) with no significant change in the 120-kDa protein. Insulin stimulation of the DG44heo cells also increased tyrosine phosphorylation of IRS-1, but phosphorylation of the P-subunit was not observed (Fig. 3A, lanes 3  and 4 ) . The inability to detect autophosphorylation of the insulin receptor P-subunit reflects the relatively low level of insulin receptors present in the DG44heo cells. In contrast to the DG44heo cells, insulin stimulation of the SRI-5 and SRI-9 cell lines resulted in a reduced level of IRS-1 tyrosine phosphorylation (Fig. 3A, lanes 6 and 8). Quantitation of IRS-1 tyrosine phosphorylation determined from multiple Western blots consistently demonstrated a reduction of 30 and 50% compared to control values in the SRI-5 and SRI-9 cells, respectively (Fig.  3B). Although we were unable to directly assess the relative amounts of IRS-1 protein due to the lack of an appropriate immunoblotting IRS-1 antibody, the aPY Western blots demonstrate that the IRS-1 antisense-expressing cell lines have reduced levels of tyrosine-phosphorylated IRS-1 protein in response to insulin.
It has been well established that tyrosine phosphorylation of IRS-1 results in the association of IRS-1 with PI 3-kinase (20)(21)(22). This association is dependent upon tyrosine phosphorylation of IRS-1 Y" motifs (23,24) and their specific recognition by SH2 domains present in the p85 subunit of PI 3-kinase (25,26). We therefore determined the amount of IRS-l-associated PI 3-kinase activity in the DG44/neo, antisense SRI-5, and antisense SRI-9 cell lines (Fig. 3 0 . The amount of insulinstimulated aIRS-1 immunoprecipitated PI 3-kinase activity in the SRI-5 cell line was reduced approximately 20% compared to DG44heo cells, whereas the SRI-9 cell line was reduced to 40% of DG44heo. Thus, insulin-mediated association of IRS-1 with PI 3-kinase was also impaired in the antisense IRS-1 RNAexpressing clones, which is consistent with the reduction in insulin-stimulated tyrosine phosphorylation of IRS-1 protein (Fig. 3, A and B).
Insulin has also been shown to be a potent activator of immediate early gene expression and, in particular, the SRE of the c-fos promoter (19,27). Insulin stimulation of control cells transfected with a SRE fused to the thymidine kinase-luciferase reporter gene resulted in a 4-fold increase in luciferase activity (Fig. 4A). In contrast, insulin-stimulated SRE-Luc activity approximately 2.2-fold in the SRI-5 and 2-fold in the SRI-9 cell lines. In order to demonstrate that the decrease in insulin-stimulated SRE-Luc activity was a specific consequence of the decreased phosphorylation of IRS-1, we co-transfected the cells with an expression plasmid for rat IRS-1 (Fig.  4B). As expected, co-transfection with IRS-1 had only a small effect on insulin stimulation of SRE-Luc activity in DG44heo cells. However, in the SRI-5 and SRI-9 cell lines, expression of IRS-1 fully restored insulin-stimulated SRE-Luc activity. Thus,  (hatched box). Cell extracts were prepared and assayed for luciferase and P-galactosidase activity. B, control DG44/ neo, SRI-5, and SRI-9 cells were co-transfected with an expression plasmid for IRS-1 (CLDN/IRS-1) and the reporter plasmids SRE-Luc plus RSV-LacZ. The cells were then incubated with (hatched box) or without (open box) 100 nM insulin and assayed for luciferase and @palactosidase as indicated above. These data were from an average of three independent experiments each performed in triplicate. In order to directly compare the changes in SRE-Luc activity, the data for the insulin-treated DG44/neo cells were set to 100%. the inhibition of insulin-stimulated c-fos transcription in the IRS-1 antisense RNA-expressing cell lines was a specific consequence of decreased levels of tyrosine-phosphorylated IRS-1.
In summary, these data demonstrate that in CHO cells, expression of IRS-1 was essential for cell growth, insulin-stimulated DNA synthesis, and transcriptional activation of the c-fos serum response element. Although the function of PI 3-kinase in insulin signaling has not yet been established, the reduced associatiodactivation of PI 3-kinase with IRS-1 was also consistent with the inhibition of biological responsiveness in these cell lines. Surprisingly, however, both the decrease in IRS-1 tyrosine phosphorylation and PI 3-kinase association with IRS-1 were modest in comparison to the marked inhibition of growth and DNA synthesis in the antisense-expressing cell lines. This finding, coupled with our inability to amplify or obtain clonal lines exhibiting greater reductions in IRS-1 levels, was consistent with an essential role of IRS-1 phosphorylation in these cells. The requirement for a minimal level of IRS-1 expression was confirmed by complete restoration of insulin-stimulated SRE-Luc activity in IRS-1 antisense cells transfected with the IRS-1 cDNA plasmid. Thus, the impaired biological responsiveness was a direct consequence of reduced IRS-1 levels and was not secondary to nonspecific effects of antisense RNA.