Neoplastic transformation induced by insulin receptor substrate-1 overexpression requires an interaction with both Grb2 and Syp signaling molecules.

The insulin receptor substrate-1 (IRS-1) is the major intracellular substrate of insulin and insulin-like growth factor-I (IGF-I) receptor tyrosine kinase activity, and this protein has been found to be overexpressed in human hepatocellular carcinomas. IRS-1 contains several src homology 2 (SH2) binding motifs that interact following tyrosyl phosphorylation with SH2-containing proteins, and this interaction may be essential for transmitting the growth signal from the cell surface to the nucleus. We have previously reported that overexpression of IRS-1 may induce neoplastic transformation of NIH 3T3 cells. This study examines the role of two SH2-containing molecules, namely the Grb2 adapter and Syp tyrosine phosphatase proteins as important components of the cellular transforming activity of IRS-1. Mutations of tyrosine 897 in the YVNI motif (Y897F) and of tyrosine 1180 in the YIDL motif (Y1180F) reduced the intracellular interaction of IRS-1 with Grb2 and Syp proteins, respectively. Furthermore, a single mutation at either Phe-897 or Phe-1180 substantially but not completely reduced IGF-I-dependent transforming activity of IRS-1, whereas creation of a double mutation of both tyrosine residues (Y897F/Y1180F) strikingly attenuated the transforming activity of IRS-1. Stable expression of the IRS-1 mutant constructs in NIH 3T3 cells was associated with a lower level of activation of the mitogen-activated protein kinase kinase (MAPKK)/MAPK cascade following IGF-I stimulation compared with cells stably transfected with the "wild-type" IRS-1 gene. These results suggest that IRS-1-induced cellular transformation requires an interaction with both Grb2 and Syp signal transduction molecules since neither interaction alone appears to be required, and this event subsequently leads to activation of the MAPKK/MAPK cascade.

The insulin receptor substrate-1 (IRS-1) is the major intracellular substrate of insulin and insulin-like growth factor-I (IGF-I) receptor tyrosine kinase activity, and this protein has been found to be overexpressed in human hepatocellular carcinomas. IRS-1 contains several src homology 2 (SH2) binding motifs that interact following tyrosyl phosphorylation with SH2-containing proteins, and this interaction may be essential for transmitting the growth signal from the cell surface to the nucleus. We have previously reported that overexpression of IRS-1 may induce neoplastic transformation of NIH 3T3 cells. This study examines the role of two SH2containing molecules, namely the Grb2 adapter and Syp tyrosine phosphatase proteins as important components of the cellular transforming activity of IRS-1. Mutations of tyrosine 897 in the YVNI motif (Y897F) and of tyrosine 1180 in the YIDL motif (Y1180F) reduced the intracellular interaction of IRS-1 with Grb2 and Syp proteins, respectively. Furthermore, a single mutation at either Phe-897 or Phe-1180 substantially but not completely reduced IGF-I-dependent transforming activity of IRS-1, whereas creation of a double mutation of both tyrosine residues (Y897F/Y1180F) strikingly attenuated the transforming activity of IRS-1. Stable expression of the IRS-1 mutant constructs in NIH 3T3 cells was associated with a lower level of activation of the mitogenactivated protein kinase kinase (MAPKK)/MAPK cascade following IGF-I stimulation compared with cells stably transfected with the "wild-type" IRS-1 gene. These results suggest that IRS-1-induced cellular transformation requires an interaction with both Grb2 and Syp signal transduction molecules since neither interaction alone appears to be required, and this event subsequently leads to activation of the MAPKK/MAPK cascade.
The overexpression of IGF-I and/or IGF-II proteins has been associated with several types of malignancies and, therefore, may contribute to carcinogenesis by autocrine and/or paracrine mechanisms (4). Several reports have demonstrated that the IGF-I receptor, a molecule that may be stimulated by IGF-I, IGF-II, and weakly by insulin, was overexpressed in several human malignancies including breast cancer and HCC (4). Indeed, overexpression of "wild-type" IGF-I receptors in vitro may be sufficient to induce neoplastic transformation of NIH 3T3 cells (5). Furthermore, disruption of the IGF-I receptor blocks transformation induced by SV40 or v-ras oncogenes (6,7). However, subsequent characterization of the signal transduction pathways following IGF-I binding to its receptor that eventually leads to transformation of mammalian cells has not been clarified.
We have found that one of the molecules up-regulated in HCC tumors at the protein and RNA level is the insulin receptor substrate-1 (IRS-1), the main intracellular substrate of IGF-I/insulin receptors (8,9). The human IRS-1 (hIRS-1) gene has been found to be overexpressed 3-5-fold in HCC tumors and was initially cloned from an HCC cell line (10). This finding has recently been confirmed by others (11). Although hIRS-1 is expressed at very low levels in normal hepatocytes (12,13), overexpression of the hIRS-1 gene has been detected in multiple HCC cell lines and clinical samples of hepatoma tumor tissue (10,11). Furthermore, there is direct evidence suggesting that tyrosyl phosphorylation of IRS-1 plays an important role in liver regeneration with respect to hepatocyte proliferation and growth (13). It is now well established that IRS-1 is tyrosyl-phosphorylated by cell surface receptor tyrosine kinases and subsequently interacts with SH2-containing molecules on phosphorylated tyrosine residues located in specific binding motifs as a mechanism to send the IGF-I/insulin signals from the cell surface to the nucleus. For example, the rat IRS-1 has a 895 YVNI motif that binds the Grb2 adapter protein (or Ash) (14,15). In addition, the 1172 YIDL interacts with the Syp phosphatase molecule (also known as PTP1D, PTP2C, and SH-PTP2) (16) and has several other YMxM/YxxM motifs that bind the p85 subunits of phosphatidylinositol-3 kinase (PI3K) (17,18). Although the IRS-1 protein has no catalytic domain, it is apparent that following tyrosine phosphorylation IRS-1 may initiate the signal transduction process through its interaction with such SH2-containing molecules (8,9). Indeed, previous reports have established that IRS-1-mediated mitogenic signals were regulated directly by this interaction with Grb2 and Syp, and such interactions were followed by the activation of the mitogen-activated protein kinase kinase (MAPKK) and MAPK cascade (18 -22). It has also been suggested that constitutive activation of the MAPKK/MAPK cascade may be required for neoplastic transformation of mammalian cells to occur (23,24).
Even though there is strong in vivo and in vitro evidence that indicates a close association between IRS-1 overexpression and hepatocarcinogenesis, the precise role of IRS-1 in the transformation process is unknown. It is of interest, however, that expression of IRS-1 antisense mRNA diminishes the mitogenic response of cells to insulin (25), and overexpression of IRS-1 enhances the proliferative response in Chinese hamster ovary cells but not in a background of insulin receptor overexpression (26). We have previously reported that hIRS-1 overexpression may promote neoplastic transformation of NIH 3T3 cells, and clones stably transfected with hIRS-1 exhibited the phenotypic properties of increased transformed foci formation, anchorageindependent growth in soft agar, and production of large tumors in nude mice (27). Similar findings have been reported in NIH 3T3 cells with IGF-I receptor overexpression (5). To explore the potential molecular mechanisms of IRS-1 role in hepatocarcinogenesis, we examined the importance of interactions with the SH2-containing proteins, Grb2 and Syp, on cellular transforming activity and studied activation of the downstream MAPKK/MAPK cascade in response to IGF-I stimulation.

EXPERIMENTAL PROCEDURES
Materials-We have isolated a hIRS-1 cDNA from a human HCC cell line (FOCUS), and this construct was used in the study (10). Anti-IRS-1 rabbit polyclonal antibody (Upstate Biotechnology Inc., Lake Placid, NY), anti-Grb2 rabbit polyclonal antibody (Santa Cruz Biotechnology, Inc., CA), anti-Syp/PTP2C mouse monoclonal antibody (Transduction Laboratories), and anti-p85 subunit of PI3K rabbit polyclonal antibody (Upstate Biotechnology Inc.) were used for immunoblot and immunoprecipitation analysis. IGF-I was purchased from Sigma. For assays of MAPKK and MAPK enzymatic activity, we used recombinant MAPK protein conjugated to agarose (Upstate Biotechnology Inc.) and myelin basic proteins (MBP, Sigma) as the specific substrates, respectively.
Site-directed Mutagenesis and Plasmid Construction-The predicted structure of the hIRS-1 protein contains the same Grb2 and Syp binding motifs of 897 YVNI and 1180 YIDL, respectively, as the rat IRS-1 protein.
Cell Cultures and Transfection Assays-NIH 3T3 cells were cultured in Dulbecco's modified Eagle's media (DMEM) (Mediatech, Washington, DC) supplemented with 10% heat-inactivated fetal bovine serum (FBS) (Sigma). In brief, 1 ϫ 10 6 cells in 3.5-cm dish were transfected with 1 g of linearized hIRS-1-wt, Phe-897, Phe-1180, Y897F/Y1180F, and pBK-CMV (mock) plasmids using Lipofectamine (Life Technologies, Inc.) according to manufacturer's instructions. Cells were split 1:10 at 2 days and grown confluently in 10% FBS/DMEM to characterize the transient transfectants. Stable transfected cells were established in the presence of G418 (400 g/ml; Life Technologies, Inc.), and 10 colonies overexpressing hIRS-1 were pooled for the studies outlined below. For characterization of the phenotype, 1 ϫ 10 6 pooled stable transfectants were seeded on 10-cm plastic dishes in DMEM with 10% FBS, and the cell growth rate was determined every day for 2 weeks. To examine exponential growth of the transfectants, 1 ϫ 10 4 cells were seeded on 10-cm plastic dishes with DMEM and 10% FBS followed by replacement with DMEM containing various concentrations of FBS 24 h later. The growth of the hIRS-1 stable transfectants was then assessed by counting the number of cells. To determine transforming potential, the ability of established cells to exhibit anchorage-independent growth in soft agar was analyzed. For this assay 1 ϫ 10 3 cells of each transfectant were seeded in DMEM containing 10% FBS and 0.4% soft agar (SeaPlaque GTG Agarose, FMC Bioproducts, Rockland, ME) over a bottom agar containing medium consisting of DMEM containing 10% FBS and 0.53% agar. IGF-1 was added at 10 and 100 g/ml. The numbers of colonies formed in soft agar was determined at 2 weeks after seeding as another assay for transforming activity.
Immunoblot and Immunoprecipitation Analysis-Cell lysates from the stable transfectants treated with 10% FBS for 10 min after 6 h of serum starvation were prepared in cold Triton-lysis buffer (50 mM Tris-HCl, pH 7.5, containing 1% Triton, 2 mM EGTA, 10 mM EDTA, 100 mM NaF, 1 mM Na 4 P 2 O 7 , 2 mM NaVO 4 , 1 mM phenylmethylsulfonyl fluoride, 25 mg/ml aprotinin, 3.5 mg/ml pepstatin A, and 25 mg/ml leupeptin). For immunoblot analysis, cell lysates containing 100 g of protein were loaded onto SDS-polyacrylamide gels, and proteins were transferred onto Immobilon-P membranes (Millipore Corp., Bedford, MA). The immunodetection was developed with the specific antibodies for proteins using the ECL system (Amersham Corp.). To determine the interaction of IRS-1 with SH2-containing proteins, immunoprecipitation was performed using an antibody for the C terminus of hIRS-1 (amino acids 1230 -43), and this antibody therefore does not affect the amino acid-897 and -1180 binding motifs. Cell lysates containing 500 g of proteins were incubated with 1 g of the IRS-1 antibody and immunoprecipitated with protein A-agarose. The presence of SH2-containing proteins in the immunoprecipitants was immunodetected by antibodies specific for the proteins of interest using Western blot analysis.
MAPKK and MAPK Enzymatic Assays-IGF-I-induced MAPKK and MAPK activation were analyzed using lysates of stable transfectants that were treated with or without 100 ng/ml IGF-I for 5 min after a 6-h serum starvation period. MAPKK activity was evaluated by measuring 32 P incorporation into recombinant-human full-length p42 MAPK, as described previously (28,29). Briefly, 25 g of protein of each whole cell lysate in 50 mM ␤-glycerophosphate, pH 7.3, 1.5 mM Na-EGTA, 100 M NaVO 4 , 400 M phenylmethylsulfonyl fluoride, 25 mg/ml aprotinin, 3.5 mg/ml pepstatin A, 25 mg/ml leupeptin, and 1 mM dithiothreitol was applied for detection of MAPKK activity. A 10-l sample was mixed with 6 l of 20 mM Hepes, pH 7.4, 1 mM EGTA, 1 mM dithiothreitol, and 0.4 mg/ml bovine serum albumin containing 50 g/ml recombinant MAPK-glutathione-S-transferase protein conjugated to agarose beads plus 4 l of 50 mM MgCl 2 and 0.5 mM [␥-32 P]ATP. Samples were incubated for 30 min at 30°C and washed first with lysis buffer, next with 0.5 M LiCl, 100 mM Tris-HCl, pH 7.6, and then with 150 mM Tris-HCl, pH 7.4, 3 mM dithiothreitol, and 30 mM MgCl 2 . Following SDS-polyacrylamide gel electrophoresis, MAPK-glutathione-S-transferase bands were excised from the stained gels and counted by a liquid scintillation counter (Beckman). To exclude the activity exhibited by autophosphorylated recombinant enzyme, the lysis buffer was used as a control.
Enzymatic activity of MAPK was measured according to previously described methods (8,18). Cell lysates with or without IGF-I stimulation were electrophoresed on an SDS-polyacrylamide gel containing 0.5 mg/ml MBP. Next, SDS was removed from the gels by washing with 20% 2-propanol in 50 mM Tris-HCl, pH 8.0, for 1 h and then 50 mM Tris-HCl, pH 8.0, containing 5 mM 2-mercaptoethanol for an additional 1 h at 20°C. After the enzyme in the gels was denatured by treatment with 6 M guanidine HCl and 50 mM Tris-HCl, pH 8.0, for 1 h at 20°C and renatured in 50 mM Tris-HCl, pH 8.0, containing 0.04% Tween 40 and 5 mM 2-mercaptoethanol, the gel was preincubated in kinase buffer (40 mM Hepes, pH 8.0, 2 mM dithiothreitol, 0.1 mM EGTA, 20 mM MgCl 2 ) for 1 h at 25°C. Phosphorylation of MBP was performed by incubation of the gel with kinase buffer containing 25 Ci of [␥-32 P]ATP (DuPont NEN) for 1 h at 25°C. Then the gel was washed in 5% (w/v) trichloroacetic acid solution containing 1% sodium pyrophosphate. The locations corresponding to MAPK-phosphorylated MBP were determined using the autoradiographs, and relevant areas were cut and counted by a liquid scintillation counter.
Transfection Studies on HepG2 Cells-To clarify the effect of hIRS-1 overexpression in a well differentiated human hepatocyte-derived cell line, transfection experiments with wild-type or mutant hIRS-1 cDNAs were performed using the human hepatoblastoma cell line HepG2. In brief, 1 ϫ 10 6 HepG2 cells were transfected with 1 g of hIRS-1-wt, Y897F/Y1180F, and pBK-CMV (mock) circular plasmids. Two days later and after 6 h of serum starvation, transfected HepG2 cells were incubated in the presence or absence of 100 ng/ml IGF-I for 5 min. Cell lysates were prepared and immunoblot analysis and MAPK enzymatic assays were performed as described above.

RESULTS
Interaction of hIRS-1 with SH2-containing Proteins-The hIRS-1 protein is 90.5% homologous to the rat protein (10) and contains the SH2-binding motifs of 897 YVNI and 1180 YIDL analogous to the 895 YVNI and 1172 YIDL motifs in rat IRS-1 that specifically binds to Grb2 and Syp, respectively. A diagram of the mutant hIRS-1 constructs is shown in Fig. 1A. Transfection of wild-type hIRS-1 (hIRS-1-wt) enhanced the interaction of both Grb2 and Syp in NIH 3T3 cells (Fig. 2, B and D), and the levels of interaction correlated with the overexpression of IRS-1 as compared with mock-transfected cells (Fig. 1B). While there were no differences in the levels of expression of Grb2 among the transfectants (Fig. 2A), the mutant constructs Phe-897 and Y897F/Y1180F demonstrated a strikingly reduced interaction of IRS-1 with Grb2 protein (Fig. 2B). Similarly, the mutant hIRS-1 constructs Phe-1180 and Y897F/Y1180F exhibited reduced binding to Syp protein (Fig. 2D). In contrast, the same level of expression of Syp protein was found in all these hIRS-1 wild-type and mutant stable transfectants (Fig. 2C). As shown in Fig. 1B, the levels of overexpression of IRS-1 in the cells transfected with the wild-type and mutant hIRS-1 constructs were equally increased compared with mock transfectants containing low levels of endogenous murine IRS-1. The p85 subunit of PI-3 kinase was expressed (Fig. 2E) and equally associated with wild-type and mutant IRS-1 proteins in all transfectants (Fig. 2F). The interaction of Grb2 and Syp with hIRS-1 was not affected by mutations at Phe-1180 and Phe-897, respectively (Fig. 2), and demonstrates that these binding motifs were specific for their target proteins.
Growth Characteristics of hIRS-1 Transfectants-These NIH 3T3 cells overexpressing hIRS-1-wt, Phe-897, Phe-1180, and Y897F/Y1180F (Fig. 1) were characterized phenotypically and compared with mock transfectants. Observing confluent growth in 10% FBS/DMEM revealed that cells transiently transfected with hIRS-1-wt piled up into multi-layered cell structures and formed transformed loci within 2 weeks (84.8 Ϯ 10.4 colonies/g DNA). This aggregate formation of transformed foci was also recognized in cells transfected with the hIRS-1 Phe-897 or Phe-1187 mutant constructs, but the number of transformed foci tended to be less frequent (48.5 Ϯ 7.7 and 44.3 Ϯ 13.3 colonies/g DNA, respectively) and smaller in size than hIRS-1-wt transfectants. In contrast, transient transfection of the double hIRS-1 mutant construct (Y897F/Y1180F) exhibited no multilayered and aggregated growth characteristics, and such cells were identical in appearance to the mocktransfected NIH 3T3 cells (0 colonies/g DNA). The exponential growth of the stable transfectants were further characterized by changing the FBS concentrations. There was no difference in the number of G418-resistant colonies among the transfectants, and this finding indicates approximately the same DNA transfection efficiency in each experiment. Although the various hIRS-1 transfectants showed similar growth rates in DMEM with 20, 10, or 2% FBS additions (data not shown), cell growth varied significantly in 1% FBS/DMEM. NIH 3T3 cell lines stably transfected with hIRS-1-wt and mutant constructs had markedly different growth ability in 1% FBS as shown in Fig. 3. hIRS-1-wt transfectants displayed high growth rates during 8 days of culture, whereas the mutant hIRS-1 Phe-897-or Phe-1180-transfected cells showed reduced growth rates compared with hIRS-1-wt cells; mock transfectant cells could not grow under these low serum conditions. More important, the hIRS-1 Y897F/Y1180F double mutant construct exhibited a strikingly reduced growth rate and was similar to that observed with mock-transfected cells.
Anchorage-independent Cell Growth and IGF-I-dependent Transforming Activity of hIRS-1 Clones-For analysis of anchorage-independent growth of these stable transfectants, cells were plated on soft agar with 10% FBS, and colony formation FIG. 1. A, schematic representation of wild-type and mutant hIRS-1 proteins. PH indicates a pleckstrin homology domain, and PTB indicates a phosphotyrosine binding domain that may play a role in receptor binding (8,37). B, overexpression of the hIRS-1 in wt and mutant proteins in NIH 3T3 cells as detected by a polyclonal antibody specific for the amino acid 1230 -1243 region of IRS-1 (␣IRS-1). The levels of expression are comparable between the clones. In mock-transfected cells, low levels of endogenous murine IRS-1 were observed.

FIG. 2. Interaction of IRS-1 with Grb2 and Syp in the stably transfected hIRS-1 clones.
Grb2 p25 expression was detected using rabbit anti-Grb2 antibody (␣Grb2) (A), and Grb2 interaction with IRS-1 was detected by immunoprecipitation with ␣IRS-1 and followed by immunoblotting with ␣Grb2 (B). Syp p72 expression was detected using mouse anti-Syp antibody (␣Syp) (C), and Syp interaction with IRS-1 was detected by immunoprecipitation with ␣IRS-1 and immunoblotting with ␣Syp (D). PI3-K p85 subunit expression was detected using rabbit anti-p85 subunit of PI3-K antibody (␣PI3K) (E), and PI3K interaction with IRS-1 was detected by immunoprecipitation with ␣IRS-1 and immunoblotting with ␣PI3K (F). Characteristics of the interaction of PI3-K with hIRS-1-wt and mutant proteins are shown in F. Note that the levels of interaction are approximately the same. The lower bands in B, D, and F indicate IgG chains.
was assessed at 2 weeks. Mock-transfected cells had no colony formation. In contrast, cells transfected with hIRS-1-wt, Phe-897, and Phe-1180 formed colonies in soft agar with 10% FBS at different levels. As shown in Fig. 4, hIRS-1-wt transfectants formed the highest levels at 16.2 Ϯ 1.9 colonies per 1 ϫ 10 3 cells; Phe-897 and Phe-1180 formed lower levels at 10.5 Ϯ 1.5 and 7.5 Ϯ 1 colonies, respectively. Addition of IGF-I resulted in approximately 2-3-fold increase in colony number indicating that hIRS-1-induced colony formation was mediated in part by IGF-I signals. In contrast, cells transfected with the hIRS-1 Y897F/Y1180F double mutant construct did not exhibit anchorage-independent growth in soft agar, and only a few colonies were observed following IGF-I stimulation. These studies suggest that hIRS-1 overexpression induces cellular transformation, and this effect is augmented by IGF-I stimulation. A change in the cell phenotype, however, requires a hIRS-1 interaction with both Grb2 and Syp molecules.
MAPKK/MAPK Cascade Is Activated by IGF-I Stimulation in the hIRS-1 Transfectants-To analyze the potential signal transduction pathways promoting cellular transformation by hIRS-1 overexpression, the possible involvement of an IGF-Idependent activation of MAPKK/MAPK cascade was examined. According to recent reports, constitutive activation of MAPKK/MAPK cascade was required for neoplastic transformation of mammalian cells to occur (23,24). As shown in Fig.  5, all serum-starved cells exhibited equally low MAPKK enzymatic activity (40 -45 cpm). IGF-I additions (100 ng/ml) induced significant MAPKK activation in hIRS-1-wt-transfected cells (330 Ϯ 16 cpm); lower levels were observed in the Phe-897 (259 Ϯ 12 cpm) and Phe-1180 (247 Ϯ 11 cpm) mutant constructs. Such a significant enhancement of MAPKK activity by IGF-I stimulation, however, was not observed in the hIRS-1 Y897F/Y1180F double mutant-transfected cells (137 Ϯ 12 cpm) or in the mock-transfected control cells (82 Ϯ 6 cpm). Similar results were obtained with respect to MAPK activation, the downstream effector molecule of MAPKK. Fig. 6 demonstrates the MAPK activity following IGF-I stimulation. All unstimulated cells have equally low MAPK activity (72-81 cpm). In contrast, IGF-I stimulation enhanced MAPK activity in hIRS-1-wt transfectants (695 Ϯ 21 cpm) but not in mock-transfected control cells (124 Ϯ 16 cpm). In cells transfected with the mutant Phe-897 and Phe-1180 hIRS-1 constructs, the MAPK activity was reduced compared with hIRS-1-wt-transfected cells (578 Ϯ 16 cpm and 498 Ϯ 15 cpm, respectively). In con-trast, the hIRS-1 double mutant Y897F/Y1180F transfectants demonstrated a limited capability to be stimulated by IGF-1 with respect to MAPK activity (210 Ϯ 27 cpm). Our studies indicate that hIRS-1 overexpression will enhance IGF-I signaling effects. Maximal activation of the downstream MAPKK/ MAPK cascade, however, requires hIRS-1 to interact with both Grb2 and Syp molecules.
hIRS-1 Overexpression in Human Hepatocyte-derived Cell Line Enhances IGF-I-dependent MAPK Activation-To determine the effect of hIRS-1 overexpression in a well differentiated human hepatocyte-derived cell line, the HepG2 human hepatoblastoma cell line which has a low level expression of endogenous hIRS-1 similar to that found in adult human hepa- tocytes was studied (10,11). HepG2 cells were transfected with hIRS-1-wt, Y897F/Y1180F, and mock DNA and then incubated with or without IGF-I addition (100 ng/ml). As shown in Fig.  7A, immunoblot analysis revealed that the levels of hIRS-1 protein were equal in HepG2 cells transfected with either the hIRS-1-wt or Y897F/Y1180F mutant construct compared with the low level found in mock DNA-transfected cells. IGF-I stimulation induced MAPK activation in mock DNA-transfected HepG2 cells (277 Ϯ 15 to 658 Ϯ 23 cpm). In contrast, transfection with hIRS-1-wt strikingly enhanced IGF-I-dependent MAPK activity (240 Ϯ 20 to 1541 Ϯ 76 cpm). Transfection with the Y897F/Y1180F mutant construct reduced this IGF-1-stimulated MAPK activity in HepG2 cells (263 Ϯ 17 to 703 Ϯ 26 cpm) to levels observed in mock DNA-transfected control cells. This study indicates that IGF-I-dependent MAPK activity was enhanced by hIRS-1 overexpression in a well differentiated human hepatocyte-derived cell line, and this activation was mediated through the Grb2/Syp signaling pathway. Therefore, one of the possible cellular consequences of hIRS-1 overexpression during hepatocarcinogenesis may be an enhancement of IGF-I-dependent MAPK activation.

DISCUSSION
Some of the general molecular mechanisms associated with neoplastic transformation has been linked to mitogenic signaling pathways coupled with constitutive activation of serinethreonine kinases such as MAPK (23,24). MAPK is activated by the tyrosine/threonine kinase MAPKK, and it in turn has been previously activated by Raf-1-induced phosphorylation of serine/threonine residues (30). Raf-1 has been defined as an effector molecule for the Ras-GTP complex (31). This cascade is initiated by upstream growth factor receptors that subse-quently undergo autophosphorylation upon binding to ligand by activation of intrinsic tyrosine kinase activity (32). Indeed, overexpression of wild-type growth factor receptors such as epidermal growth factor (33) and IGF-I (5) will produce liganddependent cellular transformation. In addition, disruption of IGF-I receptor function has been found to subsequently inhibit v-ras-induced cellular transformation (7). Therefore, in certain circumstances neoplastic transformation requires intact signaling pathways for growth factor receptors such as IGF-I (2). It has been observed that following ligand stimulation, the activated IGF-I receptor subsequently tyrosyl phosphorylates IRS-1 as its principal intracellular substrate (8,9). As shown here, hIRS-1 overexpression induces cellular transformation in an IGF-I-dependent manner and suggests that IRS-1 signaling pathways may be involved as a general mechanism of cellular transformation in the liver. IRS-1 lacks a catalytic domain but possesses several SH2-binding motifs that interact with downstream signal transduction molecules such as Grb2, Syp, PI3K, and Nck (8). To address the putative role of these signal transduction molecules in hIRS-1-induced cellular transformation, the interaction of Grb, Syp, and PI3K with hIRS-1 was analyzed in NIH 3T3 cells.
In this regard, Grb2, a 23-kDa SH2-SH3-containing protein, binds to a phosphotyrosine residue in the YVNI motif on IRS-1 via its SH2 domain (14,15). The SH3 domain of Grb2 binds to a proline-rich region of son-of-sevenless (Sos) protein recently characterized as a Ras-specific GDP/GTP exchange factor (23,24). The Grb2-Sos complex is brought to the cell surface membrane where Sos interacts with Ras-GDP and subsequently A, following transfection with hIRS-1-wt, Y897F/Y1180F double mutant construct, and mock plasmid DNA, cell lysates with (ϩ) or without (Ϫ) IGF-I (100 ng/ml) addition were immunoblotted with ␣IRS-1 antisera. Levels of IRS-1 were comparable in HepG2 cells transfected with wild type and mutant proteins. In mock DNA-transfected cells, low levels of endogenous hIRS-1 were observed. These levels are comparable with normal human hepatocytes (13). B, p42/p44 MAPK activity using MBP-containing gels. Cell lysates with (ϩ) or without (Ϫ) IGF-I stimulation were electrophoresed through an SDS-polyacrylamide gel containing MBP followed by reaction with [ 32 P]ATP in a kinase buffer and then autoradiographed. C, enzymatic activity of MAPK with (ϩ) or without (Ϫ) IGF-I stimulation was determined by measuring the amount of 32 P incorporated into MBP with a scintillation counter. Results are expressed as mean Ϯ S.E. obtained from two independent experiments. catalyzes a GDP/GTP exchange on Ras, followed by activation of the downstream Raf/MAPKK/MAPK cascade (32). Our study confirms that the 897 YVNI motif in hIRS-1 was essential for the interaction with Grb2 (Fig. 2B), but the functional consequence of a single mutation of 897 tyrosine (Phe-897) resulted in a limited effect on either hIRS-1-induced cellular transformation or IGF-1-stimulated MAPKK/MAPK activation as compared with the hIRS-1-wt molecule. This observation is consistent with a previous report using a rat IRS-1 mutant construct that alters the Grb2-binding motif (34,35). Recently, the SH2-containing phosphotyrosine phosphatase Syp has been identified as a positive effector for the Ras-signaling pathway (19,20,22). As shown in the present report a single mutation of 1180 YIDL in the Syp-binding domain (Phe-1180) partially but not completely inhibited the transforming potential of hIRS-1-wt and IGF-I-stimulated activation of the MAPKK/MAPK cascade. Li et al. (36) have revealed that phosphorylation of Syp on a tyrosine residue allows it to act as an adapter protein between the platelet-derived growth factor receptor and the Grb2-Sos complex. IGF-I and insulin receptors, however, do not tyrosyl phosphorylate Syp as a direct substrate, and phosphorylated Syp has not been detected after IGF-I or insulin stimulation of cells (21,36). 2 As demonstrated in Fig. 2, the Phe-1180 hIRS-1 mutant decreases the specific interaction with Syp (D) but does not affect the interaction with Grb2 (B), suggesting that other possible proteins contribute to the mitogenic signaling pathways via a hIRS-1/Syp interaction, rather than a complex formation through Syp-Grb2-Sos. In addition, a recent report has shown that a catalytically inactive Syp (21) inhibits MAPK activity. Thus the catalytic substrate(s) of Syp phosphatase could play an important role in regulating mitogenic signals although the target substrate(s) have yet to be identified.
IRS-1 is one of the major intracellular substrates of IGF-I and insulin receptors, and such receptors probably directly interact with the phosphotyrosine binding (PTB) domain of IRS-1 (38). The Shc protein has recently been identified as another substrate of these receptor tyrosine kinases (38,39). Shc also has a PTB (N terminus) and SH2 (C terminus) domains, and, using either or both of these regions, Shc may interact with various growth factor receptors (38,40,41). However, Shc contains only one SH2 binding motif, namely 317 YVNV that could interact with Grb2 (42). Interestingly, Shc overexpression also induces cellular transformation similar to what we have found with hIRS-1 (40), possibly due to amplification of cellular growth signals through Shc or IRS-1. Salcini et al. (42) suggests that the interaction of Shc with Grb2 was necessary for Shc-induced transformation to occur. In the present report, we found that the interaction of IRS-1 with Grb2 was insufficient to explain hIRS-1-induced cellular transformation. Indeed, the double mutation of both Grb2 and Syp binding sites (Y897F/Y1180F) was required to substantially reduce IGF-I-induced colony formation in a soft agar as a measurement of the transformed phenotype. In addition, the Y897F/ Y1180F mutant construct clearly reduced IGF-I-stimulated MAPKK/MAPK activation in both NIH 3T3 and HepG2 cells compared with hIRS-1-wt-transfected cells. Cellular transformation induced by IRS-1 overexpression therefore requires an interaction with Grb2 and Syp molecules, and a maximal level of cellular transformation requires both interactions.
IGF-I signals have been found necessary for neoplastic transformation to occur in vitro (5) as well as to promote tumorgenicity in vivo (43). We are led to believe that this cellular transformation pathway may be specifically mediated through IRS-1. Indeed, hIRS-1 overexpression in human hepatocytes may induce significant enhancement of IGF-I-dependent MAPK activation through Grb2/Syp signaling pathways as shown here by studies with the well differentiated HepG2 human hepatocyte-derived cell line. There is the possibility that hIRS-1-induced MAPK/MAPKK activation is not always related to an interaction with Grb2 and Syp proteins. IRS-1 is known to contain other SH2 binding motifs that interact with downstream effector molecules such as PI3K and Nck (17,18,44). It is of interest that Nck overexpression also causes neoplastic transformation of NIH 3T3 cells (45,46), and a recent report demonstrated that Nck SH3 domains bind to Sos and may induce Ras-signaling pathways (29). Such studies begin to address the possible molecular mechanisms of hIRS-1-induced cellular transformation and its potential role in hepatocarcinogenesis.