Interaction of Human Suppressor of Cytokine Signaling (SOCS)-2 with the Insulin-like Growth Factor-I Receptor*

SOCS (suppressor of cytokine signaling) proteins have been shown to be negative regulators of cytokine receptor signaling via the Janus kinase (JAK)/signal transducer and activator of transcription (STAT) pathway. We have cloned a member of this family (hSOCS-2) by utilizing the insulin-like growth factor I receptor (IGF-IR) cytoplasmic domain as bait in a yeast two-hybrid screen of a human fetal brain library. The hSOCS-2 protein interacted strongly with the activated IGF-IR and not with a kinase negative mutant receptor in the two-hybrid assay. Mutation of receptor tyrosines 950, 1250, 1251, and 1316 to phenylalanine or deletion of the COOH-terminal 93 amino acids did not result in decreased interaction of the receptor with hSOCS-2 protein. hSOCS-1 protein also interacted strongly with IGF-IR in the two-hybrid assay. Glutathione S-transferase-hSOCS-2 associated with activated IGF-IR in lysates of mouse fibroblasts overexpressing IGF-IR. Human embryonic kidney cells (293) were transiently transfected with vectors containing IGF-IR and FLAG epitope-tagged hSOCS-2. After IGF-I stimulation, activated IGF-IR was found in anti-FLAG immunoprecipitates and, conversely, FLAG-hSOCS-2 was found in anti IGF-IR immunoprecipitates. Thus, hSOCS-2 interacted with IGF-IR both in vitro and in vivo. HSOCS-2 mRNA was expressed in many human fetal and adult tissues with particularly high abundance in fetal kidney and adult heart, skeletal muscle, pancreas, and liver. These results raise the possibility that SOCS proteins may also play a regulatory role in IGF-I receptor signaling.

and activator of transcription) pathway (1)(2)(3)(4)(5)(6)(7)(8)(9). The first member of this family to be reported was mouse CIS (cytokine inducible SH2-containing protein) (1). Upon binding of ligand to cytokine receptors, receptor-associated JAKs become activated and phosphorylate tyrosine residues on the membrane distal portion of the receptor (10). Signaling molecules which subsequently bind to these phosphotyrosine containing motifs on the receptor include members of the STAT family. STATs are phosphorylated by cytokine receptor-associated JAKs, form dimers, and travel to the nucleus where they activate transcription. CIS was isolated as a cytokine responsive immediateearly gene in mouse hematopoietic cells (1,2). CIS mRNA encodes a polypeptide of 257 amino acids that contains an SH2 domain in the middle of the molecule. Expression of CIS in IL-3-dependent hematopoietic cell lines reduced the growth rate of the transformants, suggesting a negative role of CIS in signal transduction. The CIS protein associated with tyrosinephosphorylated erythropoietin (EPO) receptor and the tyrosine-phosphorylated ␤ chain of the IL-3 receptor, presumably by binding of the CIS SH2 domain to phosphotyrosine containing motifs in the receptors. A mutant IL-2 receptor that failed to activate STAT5 could not induce CIS, suggesting that STAT5 was important for cytokine induction of CIS. Indeed, upstream of the transcription initiation site in the CIS promoter are four potential STAT5-binding sites. Expression of STAT5 and the EPO receptor in HEK293 cells conferred EPOdependent activation of the CIS promoter. In these cells, EPOdependent tyrosine phosphorylation of STAT5 was suppressed when CIS was coexpressed. Taken together, these results provide evidence for a negative feedback loop in which CIS is induced by the cytokine and then binds to the cytokine receptor, preventing the activation of STAT by JAKs (1,2). Subsequently, three CIS-related proteins were described and these proteins have been designated as SSI (STAT-induced STAT inhibitor) or SOCS (suppressor of cytokine signaling) proteins 1-3 (3)(4)(5)(6)(7)9). Together with CIS, SSI/SOCS proteins share a common domain structure consisting of an NH 2 -terminal region of variable length, a central SH2 domain and a COOHterminal motif, termed the SOCS box, of unknown function.
In contrast to cytokine receptors which do not have intrinsic tyrosine kinase activity but utilize JAKs for receptor phosphorylation and phosphorylation of downstream signaling molecules such as STATs, the insulin-like growth factor I (IGF-I) receptor is a member of the tyrosine kinase family of growth factor receptors (11,12). The IGF-I receptor is important for cellular growth, differentiation, and inhibition of apoptosis. Binding of IGF-I or IGF-II to the IGF-I receptor results in receptor autophosphorylation. Receptor autophosphorylation amplifies the tyrosine kinase activity of the receptor and creates binding motifs for downstream signaling molecules. The IGF-I receptor and the closely related insulin receptor utilize a family of large docking proteins (insulin receptor substrate, IRS) (13). IRS proteins bind to phosphotyrosine motifs on the receptor and, in turn, are phosphorylated on multiple tyrosine residues, creating binding motifs for the regulatory subunit of phosphoinositide 3-kinase (p85), the adapter proteins Grb-2 and Nck, and the tyrosine phosphatase, Syp. The adapter protein Shc also binds directly to the IGF-I receptor (14) and insulin receptor (15), providing an alternative pathway for the activation of Ras via Grb-2 and SOS. Thus, the major signaling pathways utilized by the IGF-I receptor are those in which Ras and phosphoinositide 3-kinase play an important role. We now report that a member of the SOCS family (SOCS-2) also interacts with the IGF-I receptor. SOCS-2 was cloned by a yeast two-hybrid screen of a human fetal brain library using the IGF-I receptor as bait. GST-hSOCS-2 binds to the activated IGF-I receptor from mouse fibroblasts and hSOCS-2 associates with the activated IGF-I receptor in vivo after transient transfection of human embryonic kidney 293 cells with IGF-I receptor and hSOCS-2 plasmids. These results raise the possibility that SOCS proteins may also play a regulatory role in IGF-I receptor signaling.

EXPERIMENTAL PROCEDURES
Materials-The human fetal brain activation domain fusion cDNA library, yeast strain EGY48, and two-hybrid expression plasmids were obtained from Dr. Roger Brent and have been previously described (14,16). NIH 3T3 cells overexpressing the IGF-IR (NWTc43 cells) were obtained from Dr. Derek LeRoith. Human transformed primary embryonal kidney cells (293) were obtained from American Type Culture Collection. The antibody to the IGF-IR (number 4803) used for immunoblotting has been previously described (14,16). The FLAG fusion protein expression system and anti-FLAG M2 monoclonal antibody were purchased from Eastman Kodak. The LexA monoclonal antibody was purchased from CLONTECH. The HA (12CA5) monoclonal antibody was obtained from Boehringer Mannheim. The monoclonal antibody to the IGF-I receptor (␣IR3) was purified from ascites fluid by protein G affinity chromatography and coupled to Reacti-Gel (Pierce). All oligonucleotides and primers were synthesized using an ABI DNA synthesizer. Other reagents were purchased from commercial sources as indicated in the text or figure legends.
Plasmid Constructions-The LexA-IGF-IR and LexA-IGF-IR (KR) (kinase negative receptor) bait hybrid plasmid constructs used in these studies have been previously described (14,16). All mutants derived from the LexA-IGF-IR fusion protein were generated either by truncation or site-directed mutagenesis using PCR. The sequences of the mutant PCR fragments were verified by manual dideoxy or automatic ABI prism DNA sequencing. The full-length SOCS-2 cDNA was constructed by PCR using the overlapping clones 63 and 7-10 as templates. The SOCS-1 cDNA was obtained by PCR using cosmid clone 356d7 (accession number AC002286) as template. The GEX-hSOCS-2 plasmid was constructed by introducing the coding sequence for amino acids 29 -198 of hSOCS-2 into the vector GEX-4T-1 (Pharmacia Biotech Inc.). The GST fusion protein was expressed in the BL21 Escherichia coli strain and purified according to the manufacturer's protocol. The cDNA residues 29 -198 of hSOCS-2 were also subcloned in the pFLAG-CMV-2 mammalian expression vector (Eastman Kodak) to generate FLAG-hSOCS-2. The full-length human IGF-IR cDNA (provided by Dr. Derek LeRoith) was subcloned into pcDNA mammalian expression vector (Invitrogen) for the construct pcDNA-IGF-IR.
Two-hybrid Screening and Cloning of hSOCS-2-Routine yeast culture, preparation of various yeast selection media, and yeast transformations were carried out as described (17). The two-hybrid library screen was performed as described (16). In the first step, 0.5 g of the library was transformed into EGY48 yeast containing LexA-IGF-IR plasmid and two reporter genes, lacZ and LEU2. The transformants were selected for growth on media lacking tryptophan, uracil, and histidine, and containing glucose as the carbon source. In the second step, interactors were selected by plating approximately 10 7 primary transformants on the same medium containing galactose and 5-bromo-4-chloro-3-indoyl ␤-D-galactoside and lacking leucine. About 80 -100 clones having galactose-dependent, lacZϩ, LEU2ϩ phenotypes were sorted by PCR and restriction digestion. This analysis identified eight distinct cDNA inserts. One of those cDNA fragments (clone 7-10) encoded amino acids 29 -198 of human SOCS-2.
Subsequently, 5Ј-RACE PCR was used to clone the full-length hSOCS-2 cDNA from a human fetal brain cDNA library (Marathon ready, CLONTECH). Two gene specific primers, 5Ј-CCTTGCACATCT-GAACATAGTAGTCGATCAG-3Ј (hSOCS-2,-720/-689, accession number AF037989) and 5Ј-GATTTGACACATATGATAGACTCCAATCT-G-3Ј (hSOCS-2, -658/-629), designed from the complementary sequence of clone 7-10, were used in this amplification. The 5Ј primers were provided by CLONTECH. Twelve RACE PCR fragments were cloned into pCR2.1 vector (Invitrogen) and sequenced. The largest clone (number 63), contains the 5Ј sequence of hSOCS-2 cDNA. Northern Hybridization-Multiple tissue human poly(A) RNA blots were obtained from CLONTECH and hybridized to 32 P-labeled clone number 63 (nucleotides 1-658; accession number AF037989). The probe was labeled by random priming and separated from the free nucleotides by G-50 Sephadex chromatography. The blots were hybridized at 68°C for 1 h after adding the labeled DNA probe at a concentration of 2 ϫ 10 6 cpm/ml of hybridization buffer. Hybridization was followed by three to five high stringency washes and autoradiography.
In Vitro Binding Studies-GST-hSOCS-2(29 -198) or glutathione Stransferase (GST) in bacterial lysates was bound to glutathione-Sepharose beads and the washed beads were incubated overnight at 4°C with cell lysates derived from NWTc43 cells (NIH 3T3 cells overexpressing human IGF-IR). Subconfluent monolayers of these cells were serum starved for 24 h in Dulbecco's modified Eagle's media and lysed prior to or after IGF-I (20 nM) stimulation. Lysates were prepared as described previously (16). After incubation with the lysates at 4°C overnight the beads were washed 4 times with cold lysis buffer, boiled in Laemmli SDS sample buffer containing 100 mM dithiothreitol, and resolved by SDS-PAGE. The proteins were transferred to nitrocellulose and probed with primary and secondary antibodies described in the text and figure legends. Detection was with ECL (Amersham).

Isolation of hSOCS-2 as an IGF-IR Interacting Protein-We
have used the yeast two-hybrid system to identify proteins which interact with the cytoplasmic domain of the IGF-IR. A LexA DNA-binding vector containing the entire coding sequence of the cytoplasmic domain of the IGF-IR was used as bait to screen a human fetal brain cDNA library fused to the B42 activation domain. Using this system we detected positive interactors by galactose-dependent activation of two reporter genes, lacZ and LEU2. Clone 7-10 was one of eight distinct cDNAs, which included Grb10 (14), 14-3-3␤ and (16), and p55␥ (18). Sequencing of clone 7-10 showed that it contains an open reading frame encoding a protein of 170 amino acids. A search of the data base at the time of initial cloning of 7-10 showed that it had weak homology to the SH2 domain of p85 subunit of phosphoinositide 3-kinase but no good match was found. We therefore used 5Ј-RACE to identify the full-length protein. Using a human fetal brain cDNA as template, we obtained 12 overlapping clones. The nucleotide sequence of the longest clone (number 63) overlapped 258 base pairs of the 5Ј-end of clone 7-10 (Fig. 1A). These overlapping clones (numbers 63 and 7-10) represent a 1947-base pair cDNA that contains an open reading frame spanning nucleotides 318 to 914 and encodes a protein of 198 amino acids (Fig. 1B). The presumed initiation codon is similar, but not identical to, a Kozak consensus sequence (19). Data base analysis done at that time indicated that this protein is identical to hSOCS-2 (6). The hSOCS-2 protein contains a 47-residue amino-terminal region, a central SH2 domain, and a COOH-terminal SOCS box (20).
Characterization of the Interaction of the IGF-IR with SOCS Proteins in the Yeast Two-hybrid System-We used the yeast two-hybrid system to characterize the interaction of SOCS proteins with the IGF-IR. To determine if the full-length SOCS-2 protein interacts with the IGF-IR, the cDNA encoding the full-length protein was constructed using a primer overlapping a sequence common to clones numbers 63 and 7-10 and the product was inserted into the activation domain hybrid plasmid (AD-SOCS-2(FL)). Co-expression of this plasmid with the IGF-IR bait plasmid resulted in galactose-dependent activation of both the lacZ and LEU2 reporter genes to levels similar to or greater than those observed with the SOCS-2(29 -198) fragment isolated in the library screen (i.e. 7-10) or with the AD-IRS-1(2-516) hybrid, a relatively strong indicator (14) (Table I).
In contrast, neither reporter gene was activated when the AD-SOCS-2(FL)hybrid was coexpressed with a kinase negative receptor bait in which lysine 1003 is changed to arginine (Table  I). These results indicate that the full-length SOCS-2 protein also interacts with the IGF-IR and that receptor activation is necessary for this interaction.
To determine if other members of the SOCS family interact with the IGF-IR, an activation domain hybrid containing the complete coding sequence of SOCS-1 was constructed (3). When this construct (AD-SOCS-1(FL)) was co-expressed with the wild-type IGF-IR bait, the reporter genes were expressed at high levels. Again, co-expression with the kinase negative bait did not result in reporter gene activation (Table I).
We have previously shown that the yeast two-hybrid system can be used to map the sites of interaction of the IGF-IR with IRS-1, Shc, and 14-3-3 proteins (14, 16). To identify the site of interaction of the IGF-IR with SOCS-2, we coexpressed the full-length protein with a series of mutant receptor baits. These included mutants in which tyrosines 950, 1250, 1251, and 1316 were mutated to phenylalanine, either alone or in combination, and constructs in which the carboxyl-terminal portions of the receptor was deleted. Mutation of the tyrosine residues to phenylalanine, either alone (data not shown) or in combination (i.e. 4F), had little effect on binding of the receptor (Table I). Deletion of the carboxyl-terminal 93 amino acids (i.e. 1244Y) increased binding slightly (Table I). These data indicate that tyrosine 950 and the carboxyl-terminal portion of the receptor containing tyrosines 1250, 1251, and 1316 are not required for interaction with SOCS-2 and suggest that the interaction occurs through either the kinase domain or the juxtamembrane region of the receptor.
Expression of hSOCS-2 mRNA in Fetal and Adult Tissues-We examined the tissue distribution of hSOCS-2 mRNAs by Northern blot analysis using human multiple tissue poly(A) RNA blots and probing with the 32 P-labeled amino-terminal fragment number 63 (nucleotides 1-658; accession number AF037989) (Fig. 1A). A 5.0-kb mRNA species was present in all fetal and adult tissues examined (Fig. 2, A-C). This mRNA was most abundant in fetal kidney and adult heart, skeletal muscle, pancreas, and liver. A prominent 5.0-kb band also seen in adult kidney, thymus, prostate, testis, small intestine, and colon. Interestingly, additional smaller mRNA species (the most prominent being 2.0 and 2.8 kb) were also observed in adult tissues but not in the fetal tissues examined. Thus hSOCS-2 mRNA expression appears to be under both developmental and tissue specific controls.
In Vitro Binding of hSOCS-2 to the IGF-IR-We investigated the association of IGF-IR with hSOCS-2 in vitro using a GST-hSOCS-2 (amino acids 29 -198) fusion protein and the IGF-IR derived from mouse fibroblasts overexpressing this receptor (NWTc43). Subconfluent monolayer cultures of NWTc43 cells were serum starved for 24 h and lysed prior to or after IGF-I stimulation for different times, as indicated in the legend to Fig. 3. Extracts were incubated with either GST-Sepharose beads or GST-hSOCS-2-Sepharose beads. Material remaining bound to the washed beads was resolved by SDS-PAGE and analyzed by immunoblotting with an antiserum directed against the carboxyl terminus of the human IGF-IR (number 4803) (Fig. 3A) or an antiserum to phosphotyrosine (Fig. 3B). GST-hSOCS-2 beads bound to IGF-IR from NWTc43 in an IGF-I dependent pattern. No binding was observed when the lysates were incubated with resin containing GST alone. These data show that hSOCS-2 binds the IGF-IR in an in vitro system and that the association is dependent on receptor activation.
In vivo Association of IGF-IR and hSOCS-2 in 293 Cells-To demonstrate that the interaction of the IGF-IR and hSOCS-2 occurs in vivo we utilized transient transfection of human embryonic kidney cells (293). An expression vector that directed the synthesis of hSOCS-2 (amino acids 29 -198) fused to an amino-terminal FLAG epitope tag was co-transfected with pcDNA-IGF-IR expression plasmid into 293 cells. After 48 h in growth medium, transfectants were serum starved for 32 h and then lysed either prior to or after IGF-I stimulation for 40 min. Extracts were immunoprecipitated with IGF-IR antibody (␣IR3) beads; coimmunoprecipitated proteins were resolved by SDS-PAGE and immunoblotted with either FLAG M2 monoclonal antibody or IGF-IR antibody (number 4803). Receptor expression in the 293 cells transfected with receptor was confirmed by the immunoblotting the lysates with receptor antibody (Fig. 4A, top panel, lanes 3, 4, 6, and 7). A low level of endogenous receptor expression was seen in the untransfected cells (lane 1). IGF-IR antibody coimmunoprecipitated a protein hSOCS-2 Is Not Phosphorylated on Tyrosines as a Consequence of Its Interaction with IGF-IR-We have also analyzed the phosphorylation status of hSOCS-2 in 293 cells. Sequence analysis of hSOCS-2 indicated it has a YVQM motif (amino acids 129 -132, Fig. 1B), a possible target of IGF-IR tyrosine kinase. IGF-IR substrates which contain this motif include IRS-1 and -2 and Shc (11,14). Using immunoblot analysis with a phosphotyrosine antibody (Fig. 4A, bottom panel), we failed to detect hSOCS-2 in cell lysates or in immunoprecipitates from 293 cells that were stimulated with IGF-I in multiple experiments. IGF-IR autophosphorylation served as a positive control in these experiments (Fig. 4A, third panel). Thus hSOCS-2 does not appear to be a substrate for IGF-IR in vivo.

DISCUSSION
In this study, we cloned hSOCS-2 in a yeast two-hybrid screen of a human fetal brain library using the IGF-I receptor as bait. Human SOCS-2 has also recently been cloned from an activated Jurkat cDNA library (6) and by a combination of EST data base search and RACE-PCR using poly(A)(ϩ) RNA from Mo7e cells (9). Besides two-hybrid interaction, we have demonstrated in vitro and in vivo interaction of the IGF-I receptor and hSOCS-2. This interaction depends upon IGF-I receptor activation in fibroblasts and human embryonic kidney 293 cells overexpressing the IGF-I receptor. In the yeast two-hybrid assay the interaction is also dependent upon receptor autophosphorylation because kinase inactive receptor did not inter-act with the hSOCS-2. No other binding partner of SOCS-2 has been reported; specifically, SOCS-2 does not bind to JAK-2 (9).
Several earlier reports indicated that expression of some but not all SOCS mRNAs are inducible by cytokines and STATs (3)(4)(5)(6). A wide range of the cytokine superfamily members, including interleukin-3, interleukin-4, interleukin-6, leukemia inhibitory factor, erythropoietin, gramulocyte macrophage colony stimulatory factor (1, 3-5, 8, 9), and growth hormone (21), induce transcriptional activation of one or more of the SOCS or CIS genes in hematopoietic cells or murine liver through activation of the JAK/STAT signaling pathway. Thus, SOCS genes may function as part of an intracellular negative feedback loop, inhibiting either JAK activity or STAT phosphorylation and thereby suppressing cytokine signal transduction. Peripheral leptin administration rapidly induced SOCS-3 mRNA in the hypothalamus but had no effect on CIS, SOCS-1, or SOCS-2 (22). In mammalian cell lines, SOCS-3, but not SOCS-2 or CIS, blocked leptin-induced signal transduction. The suppression of cytokine signal transduction by SOCS-2 is not clear. In a preliminary study, hSOCS-2 was shown to inhibit leukemia inhibitory factor-mediated differentiation and growth arrest of myloid leukemia M1 cells (6). However, in a similar experiment utilizing M1 cells, Masuhara et al. (9) observed inhibition of leukemia inhibitory factor-induced differentiation and growth arrest by SOCS-3 but not SOCS-2.
Although SOCS family members exhibit a similar domain structure with a central SH2 domain and a COOH-terminal SOCS box, the comparison of the amino acid sequences among family members shows that CIS, SOCS-1, SOCS-2, and SOCS-3 are only distantly related (3). A search of DNA data bases for amino acid sequences corresponding to conserved residues in the SOCS box identified four additional members of the family (SOCS-4 to SOCS-7) (20). Full-length cDNAs have not been isolated for all of these new members of the SOCS family. The fact that SOCS family members are not closely related at the level of protein sequence may point to different functions among family members.
Tissue expression of mRNA is different among SOCS family members. SOCS-1 and SOCS-3 mRNA expression is most prominent in thymus, spleen, and lung whereas expression was more widespread for CIS and SOCS-2 (3). We found that hSOCS-2 mRNA was most abundant in adult heart, skeletal muscle, pancreas, and liver with intermediate amounts in kidney, thymus, prostate, testes, small intestine, and colon. Although Minamoto et al. (1996) also found that hSOCS-2 mRNA was expressed in many tissues their results differed from ours for a number of tissues. Compared with our results, Minamoto et al. (6) found relatively lower levels of SOCS-2 mRNA in liver, skeletal muscle, pancreas, and thymus, and higher levels of TABLE I Interaction of the IGF-IR with SOCS proteins in the yeast two-hybrid system IGF-IR lexA baits are given in the left column and various SOCS and IRS-1 preys are given in the second column. The KR IGF-IR mutant is a kinase negative construct; 4F IGF-IR mutant has tyrosines 950, 1250, 1251, and 1316 changed to phenylalanine; 1244Y IGF-IR has a carboxyterminal deletion of 93 amino acids. The data are of triplicate samples. The plates were read at 72 h and the scoring system was as described in Dey et al. (14). Immunoblotting with HA antibody indicated that all of the prey constructs were expressed at approximately equal levels. It seems likely that the binding of CIS to the EPO receptor and IL-3 receptor ␤ subunit and the binding of SOCS-1 to JAK2 is explained by the SOCS SH2 domain binding to a phosphotyrosine motif in the receptors (1,2,5). However, these phosphotyrosine motifs have not been identified. Our finding that hSOCS-2 binds only to the autophosphorylated IGF-I receptor is consistent with hSOCS-2 SH2 domain binding to a phosphotyrosine containing motif in the receptor. However, mutation of tyrosines 950, 1250, 1251, and 1316 to phenylalanine did not result in a decrease in the interaction of hSOCS-2 with the IGF-I receptor. In the case of binding of SOCS-1 to JAK2 there is evidence for additional involvement of other regions of the SOCS-1 protein (4). Although it could be demonstrated that SOCS-1 SH2 domain bound to JAK2, full inhibition of JAK2 kinase was not achieved by a SOCS-1 construct lacking both the COOH-and NH 2 -terminal regions. In the case of the binding of SOCS-1 to Tec, a cytoplasmic tyrosine kinase, the COOHterminal region, and the SH2 domain were not required for interaction, nor was kinase activity of Tec required (7). These results suggest that the interaction of SOCS-1 and Tec does not utilize either the SH2 domain of SOCS-1 or phosphotyrosine motifs in Tec.
Although both CIS and SOCS-1 have been shown to inhibit the JAK/STAT pathway, the mechanism of inhibition appears to be different for each of the regulatory proteins. CIS associates with the tyrosine-phosphorylated EPO receptor and the phosphorylated ␤ subunit of the IL-3 receptor (1). Binding of CIS does not inhibit the phosphorylation of the EPO receptor or the IL-3 receptor ␤ subunit. CIS does not interact with JAK-2 (4). In Ba/F3 lymphoid cells expressing the EPO receptor and CIS under the control of dexamethasone, induction of CIS expression resulted in a decrease in the phosphorylation of STAT5 in response to EPO (2). Therefore, one model for the inhibitory action of CIS is that the binding of CIS to the phosphorylated EPO receptor prevents the binding of STAT5 to the receptor, resulting in decreased tyrosine phosphorylation of STAT5 (1,2). In contrast, SOCS-1 binds to the kinase domain of JAK2, inhibiting autophosphorylation of JAK and the phosphorylation of substrates of JAK, including the gp130 component of the IL-6 receptor and STATs (4,5). Thus, SOCS family members do not display a uniform mechanism of action, making it difficult to predict how SOCS-2 might regulate IGF-I receptor signaling.
Activation of the JAK/STAT pathway is not confined to cytokine receptors (10). For example, the tyrosine kinase JAK1 and the transcription factors STAT1 and STAT3 are phosphorylated in response to epidermal growth factor (EGF) (23). Although JAK1 is phosphorylated in response to EGF it is not required for STAT activation. Similarly, STAT activation in JAK2-and Tyk2-defective cells is also normal. The kinase function of the EGF receptor is required for the activation of STAT. Since immunoprecipitation experiments have suggested that STAT1 can bind directly to the EGF receptor (24), it is likely that JAKs can be bypassed in the activation of STATs by the EGF receptor. JAK and STAT proteins are also activated by platelet-derived growth factor acting through the platelet-derived growth factor receptor, a receptor with intrinsic tyrosine kinase activity (25). All three ubiquitously expressed JAKs (JAK1, JAK2, and Tyk2) are phosphorylated on tyrosine residues in fibroblasts overexpressing the platelet-derived growth factor receptor, and all three proteins are associated with the activated receptor. However, none of the JAKs is individually required for the activation of STAT1 and STAT3 by plateletderived growth factor, suggesting that STATs are activated directly by the receptor or that JAKs can substitute for each other in the activation of STATs. The insulin receptor is closely related to the IGF-I receptor. Recently, two laboratories have reported the identification of STAT5b in yeast two-hybrid library screens using the insulin receptor as bait (26,27). Insulin promotes the rapid tyrosine phoshorylation of endogenous STAT5b in cells overexpressing the insulin receptor and STAT5b was tyrosine phoshorylated by the purified insulin receptor kinase domain in vitro. Moreover, perfusion of mouse liver with insulin resulted in the rapid tyrosine phosphorylation of STAT5. Insulin stimulated tyrosine phosphorylation of STAT5 in insulin receptor expressing cells and in mouse liver occurred in the absence of detectable JAK tyrosine phosphorylation, suggesting direct activation of STAT5 by the insulin receptor. In contrast, Gaul et al. (28) reported that insulin and IGF-I stimulated the phosphorylation of JAK1 and JAK2 in fibroblasts overexpressing the insulin receptor or the IGF-I receptor, and provided evidence for the association of JAK1 with activated receptors. These results demonstrating the activation of STATs and JAKs by tyrosine kinase receptors raise the possibility that SOCS proteins could also function in regulating receptors not belonging to the cytokine receptor class.
In summary, we have provided evidence for interaction of hSOCS-2 with the activated IGF-I receptor. We are conducting experiments to investigate the role of SOCS-2 in IGF-I receptor signaling. The distant relatedness of SOCS family members, variation in tissue mRNA expression, and differences in mechanism of action of two family members (CIS and SOCS-1), suggest that SOCS proteins may have diverse functions.