SOCS2 induces neurite outgrowth by regulation of epidermal growth factor receptor activation.

Suppressor of cytokine signaling (SOCS) 2 is a negative regulator of growth hormone (GH) signaling that regulates body growth postnatally and neuronal differentiation during development. SOCS2 binds to the GH receptor and inhibits GH signaling, including attenuation of STAT5 activation. Here we describe a new function and mechanism of action for SOCS2. Overexpression of SOCS2 in central nervous system neurons promoted neurite outgrowth, and in PC12 cells, neurite outgrowth was induced under nondifferentiating conditions, leading to inhibition of the neurite-inhibitory GTPase Rho and activation of the neurite-promoting GTPase Rac1. Addition of the epidermal growth factor receptor (EGFR) inhibitors PP3 or AG490 or the Src kinase inhibitor PP2 blocked the SOCS2-induced neurite outgrowth. The overexpressed SOCS2 bound to the EGFR, which was constitutively phosphorylated at Tyr845, the Src binding site. Overexpression of the phosphatase SHP-2 reduced the constitutive EGFR phosphorylation and subsequent neurite outgrowth. SOCS2 expression also resulted in a modest 30% decrease in phosphorylation of STAT5b at Tyr699, which is the primary site on STAT5 phosphorylated by GH; however, total tyrosine phosphorylation of STAT5 was decreased by 75-80% under basal and epidermal growth factor-stimulated conditions. Our findings suggest that SOCS2 regulates EGFR phosphorylation, leading to regulation of neurite outgrowth through a novel pathway that is distinct from GH.

Neurite outgrowth is regulated by a variety of signaling mechanisms, including growth factors and soluble or membrane-bound guidance cues. Some signals mediate initiation of neurite outgrowth at the cell soma, whereas others direct the growing neurite along a particular pathway by signaling at the growth cone (1). Many of these signals ultimately converge upon the GTPases Rac1, RhoA, and cdc42 that signal to effector molecules to mediate cytoskeletal rearrangement (2,3). Whereas several of these players are now known, how extra-cellular signals are converted into changes in cytoskeletal arrangement and how these signals are regulated are only just beginning to be elucidated. One group of molecules that may play a role in this process is the suppressor of cytokine signaling (SOCS) 1 proteins.
SOCS proteins are primarily known as negative regulators of the Janus-activated kinase/signal transducers and activators of transcription (STAT) signaling pathway. They are induced by cytokine stimulation of the Janus-activated kinase/STAT pathway and are primarily thought to form a negative feedback loop to switch off the cytokine response (4 -7). There are eight known members of the SOCS gene family (SOCS1-7 and CIS), which are expressed in different tissues, induced by different groups of cytokines, and regulate different Janus-activated kinase/STAT pathways (8). However, various members of the SOCS family have also been shown to interact with signal transduction pathways not associated with the Janus-activated kinase/STAT pathway, such as the receptor tyrosine kinases insulin-like growth factor receptor (9,10) and Kit (11). They have also been reported to interact with the GTPase-activating protein p120 RasGAP to enhance Ras activation (12) and the guanine nucleotide exchange factor Vav (11,13). It is therefore possible that SOCS molecules may also regulate other GTPases to modulate cytoskeletal rearrangement and neurite outgrowth.
SOCS2 is expressed in the nervous system during development from embryonic day 14 (E14) to postnatal day 8 (P8) in the mouse (14). In situ hybridization analysis of SOCS2 expression in the developing forebrain showed that SOCS2 was expressed at moderate levels in cortical neuroepithelium and at higher levels in the newly postmitotic neurons of the cortical plate at E14 (14). Expression of SOCS2 in the neural progenitor cells blocked the inhibitory effects of growth hormone (GH) signaling on differentiation of the progenitors into neurons, and as such, SOCS2 is a potent regulator of neuronal differentiation (15). The role of the higher level of SOCS2 expression in new neurons was unclear but suggested that SOCS2 may also contribute to neuronal function. We show here that SOCS2 is a potent inducer of neurite outgrowth in PC12 cells, a commonly used model of neuronal differentiation (16), as well as in SOCS2-overexpressing central nervous system-derived neurons, increasing both primary neurite number and neurite length. Furthermore, we show that SOCS2 binds to and results in phosphorylation of the epidermal growth factor receptor (EGFR) at Tyr 845 , the Src kinase binding site, and that overexpression of the SHP2 phosphatase blocks the SOCS2-induced EGFR activation and consequent neurite outgrowth.

EXPERIMENTAL PROCEDURES
PC12 Culture and Transfection-PC12-Tet-On cells (Clontech) were maintained in Dulbecco's modified Eagle's medium containing 5% fetal calf serum, 10% horse serum, and 100 g of G418 (Invitrogen) and differentiated by removal of serum and addition of 50 ng/ml nerve growth factor (NGF; Calbiochem) for 72 h. For stable transfections, the Tet-On tetracycline inducible system (Clontech) was used, and SOCS2 cDNA was subcloned into the vector pBi-EGFP (Clontech) to generate pBi-SOCS2-GFP, which results in expression of SOCS2 and GFP proteins when induced with 1 g/ml doxycycline (DOX). All transfections were performed using Effectene Transfection Reagent (Qiagen) according to the manufacturer's instructions, and vector pEGFP-c (Clontech) was used as a control. SHP1 and SHP2 expression plasmids and catalytically inactive mutants were a kind gift from Dr. Gen-Sheng Feng (Indiana University School of Medicine) and Dr. Stuart Frank (University of Alabama at Birmingham) (17).
PC12-Tet-On cells were plated on polyornithine (Sigma)-coated chamber slides (Falcon) at 1 ϫ 10 4 cells/well in the presence or absence of serum, with the addition of 50 ng/ml NGF, 50 ng/ml epidermal growth factor (EGF; BD Biosciences), or various signal transduction pathway-modifying reagents such as 50 M mitogen-activated protein kinase (MAPK) pathway inhibitor PD98059 (Calbiochem), 1 M Src inhibitor PP2 (Sigma), 3 M EGFR kinase inhibitor PP3 (Sigma), or 100 nM EGFR kinase inhibitor AG490 (Calbiochem). At 72 h, cells were lysed for immunoprecipitation or Western analysis or fixed and immunostained using mouse anti-␤III-tubulin (1:2000; Promega) and antimouse Cy3 (1:1000; Jackson Immunoresearch) for analysis of neurite outgrowth. The percentage of total cells extending neurites was determined and expressed as mean Ϯ S.E. of 10 fields/well in triplicate wells in at least n ϭ 3 independent experiments. Statistical significance was determined using the t test.
Western Analysis and Immunoprecipitation-Cells were exposed to the indicated stimuli and then lysed essentially as described previously (15). For Western analysis, gel sample reducing buffer was added to cell lysates and boiled. For other experiments, an aliquot of lysates was taken for Western analysis, and the remainder of the lysate was used for immunoprecipitation with mouse anti-phosphotyrosine antibody (1: 100; Cell Signaling) or mouse anti-FLAG antibody (clone M2; 1:500; Sigma) using standard protocols. After electrophoresis and Western transfer, membranes were blocked for 2 h in Tris-buffered saline containing 0.05% Tween 20 and 6% skim milk powder and then incubated with primary antibodies (mouse anti-GAP43 (1:2000; Sigma), mouse anti-␤-actin (1:10,000; Sigma), rabbit anti-MAPK and rabbit anti-phospho-MAPK (each diluted 1:1000; Cell Signaling), rabbit anti-STAT5 (1:1000; Santa Cruz), rabbit anti-EGFR (1:200; Santa Cruz), rabbit anti-phospho-EGFR Tyr 845 (1:1000; Cell Signaling), rabbit anti-EGFR Tyr 1068 (1:1000; Cell Signaling)) followed by anti-mouse (1:40,000) or anti-rabbit (1:20,000) secondary antibody linked to horseradish peroxidase (Cell Signaling). Immunoreactive bands were detected by enhanced chemiluminescence. Autoradiographs were digitally scanned; the brightness/contrast was enhanced, and composite images were made using Adobe Photoshop 6.0 and Macromedia Freehand 10.0 software. Comparison of relative densities of individual bands compared with loading controls was determined using NIH Image software.
Rac1/Rho Activation Assay-Cells were exposed to the indicated stimuli, and after treatment, cells were lysed in lysis buffer as described above. Endogenous Rho-GTP or Rac1-GTP was precipitated from cell lysates at 4°C for 1 h using a glutathione S-transferase-tagged fusion protein corresponding to residues 7-89 of mouse Rhotekin Rho binding domain or a glutathione S-transferase fusion protein containing the p21-binding domain (PBD) of human Pak1, respectively, according to the manufacturer's instructions (Upstate Biotechnology). Beads were collected by centrifugation, washed with lysis buffer, and resuspended in sample buffer. The eluted protein samples were resolved on 12% SDS-PAGE and electrophoretically transferred to nitrocellulose membrane. Rho was detected using mouse anti-Rho antibody (1:500; Pierce), and Rac1 was detected using mouse anti-Rac1 antibody (1:1000; Pierce) overnight, followed by anti-mouse secondary antibody linked to horseradish peroxidase (1:40,000; Cell Signaling). Immunoreactive bands were detected with enhanced chemiluminescence as described above. transfected with tetracycline inducible SOCS2-GFP or GFP expression vectors (pBi-SOCS2-GFP and pBi-GFP). Induction of SOCS2 by the addition of DOX (Fig. 1A) resulted in variable induction of SOCS2-GFP in individual cells, ranging from undetectable to high levels of GFP expression. All results shown are from counts of total cell numbers, so that results were not biased to an arbitrary level of GFP expression and to allow direct comparison with the control GFP-transfected cells. Unlike control cells, under nondifferentiating conditions (high serum, no NGF) neurite formation was observed in ϳ20% of all cells in the SOCS2-GFP-transfected cell lines ( Fig. 1, B, C, and F); however, ϳ100% of cells observed to express high levels of SOCS2-GFP showed neurite outgrowth (data not shown). Under these nondifferentiating conditions, there was no detectable neurite extension observed in the absence of the inducer DOX, and the pBi-SOCS2-GFP cells were indistinguishable from control cells transfected with GFP alone (Fig. 1F).

Overexpression of SOCS2 Induces
When the PC12 cells were induced to differentiate by treatment with NGF in the absence of DOX, the pBi-SOCS2-GFPand pBi-GFP-transfected cells behaved similarly, with ϳ25-30% extending neurites (Fig. 1F). If DOX was added in the presence of NGF, however, ϳ70% of SOCS2-overexpressing cells produced neurites, whereas the neurite outgrowth of the control cells remained at 25% (Fig. 1, D-F). Not only did overexpression of SOCS2 in the presence of NGF show a synergistic induction in the percentage of cells exhibiting neurite outgrowth, the neurites were longer and more numerous (Fig. 1, E and F). The mean neurite length at 72 h of SOCS2-expressing cells under basal conditions was 30 Ϯ 1.02 m, that of GFPexpressing cells with NGF was 53 Ϯ 3.55 m, and that of SOCS2-expressing cells in NGF was 64.4 Ϯ 2.9 m (mean Ϯ S.E.; p ϭ 0.01 for SOCS2 ϩ NGF compared with NGF alone; p Ͻ 0.001 for SOCS2 ϩ NGF compared with SOCS2 alone). SOCS2-expressing cells in NGF also had significantly more primary neurites per cell (3.94 Ϯ 0.09; p Ͻ 0.001) than SOCS2expressing cells under basal conditions (2.61 Ϯ 0.09) or GFPexpressing cells in NGF (2.56 Ϯ 0.11). Furthermore, SOCS2 overexpression also induced and up-regulated expression of GAP43, a neuronal marker associated with neurite outgrowth (Fig. 1G). Overexpression of SOCS2 therefore appeared to directly promote neurite outgrowth of a neural cell line by itself, as well as markedly potentiating the effects of the NGF differentiation stimulus. Overexpression of SOCS1 or SOCS3 in PC12 cells had no effect on neurite outgrowth (data not shown).
SOCS2 Enhances Neurite Outgrowth in Central Nervous System Neurons-To confirm that the enhanced neurite outgrowth observed above was not only a property of PC12 cells, neurite outgrowth in SOCS2-overexpressing central nervous system neurons was measured. These neurons were differentiated from neural stem cells derived from the central nervous system of transgenic mice that overexpress SOCS2 driven off the human ubiquitin C promoter (SOCS2Tg), such that SOCS2 is expressed in most cell types (18). SOCS2Tg and wild-type neurons were immunostained for the neuronal marker ␤IIItubulin (Fig. 2, A and B), and the number of primary neurites, the longest neurite length, and total neurite length were measured. Although the length of the longest neurite on SOCS2Tg neurons was slightly but significantly increased (Fig. 2C), the mean total neurite length was markedly increased (Fig. 2D), and this was reflected in an increased number of primary neurites (Fig. 2E). Wild-type neurons had an average of 2.29 Ϯ 0.08 neurites/cell, whereas SOCS2Tg neurons had an average of 4.32 Ϯ 0.39 neurites/cell (mean Ϯ S.E.; combined results of n ϭ 3 experiments; p Ͻ 0.01). Therefore, SOCS2 is a potent inducer of neurite outgrowth, which is predominantly exhibited as an increased number of neurites, although SOCS2 also enhanced the length of these neurites.
SOCS2 Requires EGFR and Src Kinase but not MAPK Activity to Induce Neurite Outgrowth-Neurite outgrowth is mediated by changes in cytoskeletal dynamics, regulated by activation of small GTPases such as Rho and Rac1. Accordingly, we analyzed lysates of control and SOCS2-PC12 cells for levels of total and activated Rho and Rac1. In SOCS2-overexpressing cells, Rho activation was inhibited, whereas Rac1 was activated (Fig. 3A).
To determine the mechanism by which SOCS2 promoted neurite outgrowth, we examined activation of various signal transduction pathways thought to be involved in neurite outgrowth. Addition of the phosphatidylinositol 3-kinase pathway inhibitors wortmannin or LY294002, the protein kinase C inhibitor Gö6976, or the phospholipase C ␥ inhibitors tyrphostin, AG879, or U73122 did not have any effect on the induction of neurite outgrowth by SOCS2 (data not shown). Neuronal differentiation of PC12 cells is partly through sustained activation of p42/44 MAPK (19); however, in this case, there was no difference in the level of p42/44 MAPK activation between control and SOCS2-PC12 cells. In cells exposed to DOX alone, the level of MAPK activation was low, and in the presence of NGF, p42/44 MAPK was phosphorylated at similar levels in both control and SOCS2-transfected cells (Fig. 3B). Addition of PD98059, an inhibitor of the p42/44 MAPK pathway, did not significantly (p ϭ 0.58) reduce the neurite outgrowth observed due to SOCS2 expression (Fig. 3C), although it inhibited NGFinduced neurite outgrowth in both control and SOCS2-expressing cells. Therefore, SOCS2 can induce primary neurite outgrowth in the absence of MAPK activation, although activation of the MAPK pathway by NGF appears to be required for full differentiation of the PC12 cells as shown above.
Because several major signal transduction pathways did not appear to be involved in the SOCS2-induced neurite outgrowth, but Rac and Rho activation were altered, we examined factors upstream of Rac and Rho activation. Src kinase was a likely candidate because it has been shown to promote neurite outgrowth (20 -23) and regulate GTPase activation (24), and addition of the Src kinase inhibitor PP2 blocked SOCS2-induced neurite outgrowth (Fig. 3D). Interestingly PP3, which we initially used as a control for PP2, also blocked the SOCS2-induced neurite outgrowth (Fig. 3D); however, PP3 is an ATP analogue that is also an EGFR inhibitor (25). Addition of the EGFR inhibitor AG490 also blocked SOCS2-induced neurite outgrowth (data not shown). EGF induces neurite outgrowth through c-Src activation (20); however, in PC12 cells, EGF normally promotes proliferation, not differentiation and neurite outgrowth like NGF. EGF and NGF use many of the same signal transduction pathways; in PC12 cells, EGF elicits a short activation of the Ras/MAPK pathway, whereas NGF induces sustained activation of this pathway (26). However, in SOCS2-PC12 cells, EGF did not promote proliferation and instead further promoted SOCS2-induced neurite outgrowth, which was blocked by the addition of PP2 or PP3 (Fig. 3D).
Activated STAT5 Levels Are Decreased in SOCS2-PC12 Cells-SOCS2 is primarily thought to inhibit STAT5 phosphorylation in response to GH signaling. In hepatocytes from SOCS2Ϫ/Ϫ mice, STAT5b phosphorylation in response to GH was moderately prolonged, and the overgrowth phenotype of the SOCS2Ϫ/Ϫ mice was abrogated in the absence of STAT5b (27). STAT5b has also recently been shown to be activated downstream of EGFR signaling in a c-Src-dependent manner in fibroblasts and breast tumor cell lines (28) and in hepatocytes (29). However, STAT5 phosphorylation and activation by EGF are quantitatively and qualitatively different from that induced by GH, which phosphorylates STAT5b at Tyr 699 . Analysis of STAT5 using anti-phosphotyrosine antibodies has shown that EGF induces higher maximal levels of total STAT5 phosphorylation than GH, but that if an antibody to STAT5b phosphorylated on Tyr 699 was used, then the EGFinduced phosphorylation was lower than GH-induced phosphorylation in hepatocytes (29), indicating that sites other than Tyr 699 were phosphorylated in response to EGF, such as Tyr 725 , Tyr 740 , and Tyr 743 , as shown previously in 293 cells (30). Therefore, to determine whether EGF induced phosphorylation of STAT5 in PC12 cells, whether it was primarily at sites other than Tyr 699 as described above, and whether SOCS2 played any role in regulating such phosphorylation, we examined total and Tyr 699 phosphorylation of STAT5 in control and SOCS2-PC12 cells. Under basal conditions, densitometric analysis revealed that total phosphorylated STAT5 levels in the SOCS2-PC12 cells were 20% that of GFP-expressing control PC12 cells (Fig. 4, top panel), as assessed by immunoprecipitation of phosphotyrosine-containing proteins and then probing for total STAT5 by Western analysis. Re-probing the filters using a Tyr 699 -specific anti-phopho-STAT5b antibody showed a smaller difference (30% decrease) between SOCS2-overexpressing and control cells (Fig. 4, middle panel). GH did not induce phosphorylation of STAT5 in the PC12 cell lines (data not shown). Addition of EGF for 10 min enhanced total STAT5 phosphorylation, particularly in the control cells compared with SOCS2expressing cells, but had little effect on Tyr 699 phosphorylation (Fig. 4). As reported previously (28,29), this effect of EGF on STAT5 activation was at least partly due to Src signaling. Addition of PP2 inhibited STAT5 phosphorylation in control cells, although it did not further decrease STAT5 phosphorylation in SOCS2-PC12 cells (Fig. 4). Interestingly, inhibition of Src also inhibited STAT5 phosphorylation at Tyr 699 , and this effect was observed in both control and SOCS2-PC12 cells. There was no difference in the activation of STAT1 or STAT3 in either the control or SOCS2-PC12 cells under the conditions tested (data not shown). Therefore, EGF phosphorylates STAT5 in PC12 cells in a partially Src-dependent manner, primarily on sites other than the canonical Tyr 699 , and SOCS2 is more potent at inhibiting the phosphorylation at these other sites than at Tyr 699 .
EGFR Phosphorylation Is Constitutive in SOCS2-overexpressing Cells-Given the decreased phosphorylation of STAT5 in SOCS2-PC12 cells, we expected that activation of the EGFR would be decreased. Surprisingly, analysis of EGFR phosphorylation by immunoprecipitation of phosphotyrosine-containing proteins, followed by blotting with anti-EGFR antibodies, showed that the EGFR was phosphorylated in SOCS2-PC12 cells under basal conditions, unlike control cells (Fig. 5A). Addition of EGF for 10 min increased the levels of EGFR phosphorylation in control cells but had little effect in SOCS2-PC12 cells (Fig. 5A). At least part of this phosphorylation, under basal conditions or with EGF, was likely to be due to Src activity (28,29). Src phosphorylates the EGFR at Tyr 845 , and Western analysis of total cell lysates using an anti-phospho-EGFR Tyr 845 antibody showed that the EGFR was phosphorylated at this site in SOCS2-PC12 cells (Fig. 5, A and B), but not in the presence of the Src inhibitor PP2 (Fig. 5B) or the EGFR inhibitor PP3 (data not shown). Other tyrosines on the EGFR (such as Tyr 1068 , which activates the extracellular signal-regulated kinase/MAPK pathway) were not significantly phosphorylated in the absence of EGF and showed no differential phosphorylation between control and SOCS2-PC12 cells (Fig. 5A). Total levels of EGFR were not significantly different between control and SOCS2-PC12 cells under the conditions examined. The NGF receptor (TrkA), which is another receptor tyrosine kinase involved in neurite outgrowth, was not constitutively phosphorylated by SOCS2. TrkA was phosphorylated at low levels in the absence of exogenous NGF and up-regulated by the addition of NGF; however, there was no differential phosphorylation between control and SOCS2-PC12 cells (Fig. 5C).
Overexpression of SHP2 Phosphatase Blocks SOCS2-induced Neurite Outgrowth-One possible mechanism by which SOCS2 overexpression could result in constitutive EGFR phosphorylation was by binding to a phosphatase site on the EGFR and blocking phosphatase activity, as has been shown for SOCS2 and the GH receptor (18). If this was the case, then overexpression of the phosphatase should allow more effective competition with SOCS2 for the binding site on the receptor, allowing dephosphorylation to occur. SOCS2-PC12 cells were therefore transfected with plasmids expressing the phosphatases SHP1, SHP2, and catalytically inactive (CϾS) mutants of these phos- FIG. 4. Analysis of STAT5 phosphorylation. Total STAT5 activation was decreased in SOCS2-PC12 cells (S) after 24 h of induction and was not increased by addition of EGF for 10 min or inhibited by PP2, unlike control GFP-transfected cells (G). Phosphotyrosine-containing proteins (pY) were immunoprecipitated (IP) from cell lysates, and phosphorylated STAT5 was then detected by immunoblotting (IB) using an anti-STAT5 antibody (top panel) and reprobed using an anti-phospho-STAT5 Tyr 699 antibody (STAT5-PY 699 ) (middle panel). Total STAT5 levels were determined in aliquots of cell lysates taken before immunoprecipitation (bottom panel).

FIG. 5. Analysis of EGFR phosphorylation.
A, the EGFR was constitutively phosphorylated in SOCS2-PC12 cells but was only phosphorylated in control cells after the addition of EGF for 10 min. After phosphotyrosine-containing proteins (pY) were immunoprecipitated (IP), phosphorylated EGFR was detected by immunoblotting using an anti-EGFR antibody (top panel). Phosphorylation of the EGFR at Tyr 845 and Tyr 1068 was determined by immunoblotting aliquots of cell lysates taken before immunoprecipitation, using anti-EGFR Tyr 845 or anti-EGFR Tyr 1068 antibodies (middle panels) or total EGFR levels (bottom panel). B, the endogenous activation of EGFR in SOCS2-PC12 cells was inhibited by addition of the Src inhibitor PP2. C, analysis of TrkA activation in PC12 cells showed no differential regulation in GFP-or SOCS2-transfected cells in the presence or absence of 50 g/ml NGF. phatases, SHP1CϾS and SHP2CϾS. Analysis of neurite outgrowth in the transfected cells showed that overexpression of SHP2, but not SHP1 or the catalytically inactive mutants, inhibited the SOCS2-induced neurite outgrowth (Fig. 6, A-G). Furthermore, analysis of EGFR phosphorylation in these cells showed that overexpression of SHP2 blocked the SOCS2-induced constitutive phosphorylation of the EGFR at Tyr 845 (Fig.  6H). Therefore, overexpression of SHP2 blocked the action of SOCS2 at the EGFR and inhibited SOCS2-induced neurite outgrowth. In addition, immunoprecipitation of the overexpressed FLAG-tagged SOCS2 from SOCS2-overexpressing cerebellar neurons using anti-FLAG antibody, followed by Western analysis using an anti-EGFR antibody, showed that SOCS2 bound to the EGFR (Fig. 6I).

DISCUSSION
In this study, we have described a novel function for SOCS2: promotion of neurite outgrowth, as well as a new pathway regulated by SOCS2 expression, the EGFR/c-Src/STAT5 pathway. Furthermore, we show that overexpression of SHP2 abrogates the SOCS2-induced neurite outgrowth and EGFR phosphorylation, suggesting that a possible mechanism by which SOCS2 mediates these effects is by competing with SHP2 for binding to the EGFR.
It is currently thought that the main role of SOCS2 is regulation of STAT5 activation downstream of GH signaling. SOCS2-null mice exhibit an enlarged growth phenotype consistent with hyper-responsiveness to GH signaling (31) that requires STAT5b for expression of the phenotype (27). In addition, activation of both STAT5a and STAT5b is moderately prolonged in cells from these mice in response to GH, due to their inability to effectively down-regulate the activation (27). Regulation of STAT5 activation by SOCS2 has also been demonstrated in overexpression studies, although it is not as potent as other SOCS family members, such as SOCS1 and SOCS3 (32). Although the precise mechanism by which SOCS2 regu-lates STAT5 activation downstream of GH signaling remains to be determined, it appears to involve SOCS2 competitively binding to the STAT5 and SHP2 binding sites on the GH receptor (18,32,33). Here we show that SOCS2 can also inhibit STAT5 activation independently of GH stimulation. GH had no effect on neurite outgrowth or STAT5 activation in the PC12 cells (data not shown). However, EGF has also been shown to activate STAT5 in a number of different systems (29,30,34,35), although there are differences in the mechanism of GH-and EGF-induced STAT5 activation. Both GH and EGF phosphorylate Tyr 699 of STAT5b, but unlike GH, EGF also phosphorylates Tyr 725 , Tyr 740 , and Tyr 743 (30). Activation of STAT5 in response to EGF stimulation involves mediation by Src tyrosine kinases (28,29,34). However, the mechanism by which Src induces STAT5 phosphorylation is not known, and whether it is upstream or downstream of EGFR activation is not clear.
We have shown that there is a basal level of phosphorylation of STAT5 in these cultures that is decreased in SOCS2-overexpressing cells. Because this result was observed when immunoprecipitation of phosphotyrosine-containing proteins was followed by Western blotting for STAT5 but was less marked when using an anti-phospho-STAT5 antibody directed to the canonical Tyr 699 of STAT5b, it is possible that SOCS2 inhibited phosphorylation of STAT5 at Tyr 725 , Tyr 740 , and/or Tyr 743 . Addition of exogenous EGF increased STAT5 phosphorylation in control cells, but STAT5 phosphorylation was still markedly decreased in SOCS2-overexpressing cells, suggesting that SOCS2 inhibited this phosphorylation. Activation of STAT5 has been shown to induce cell proliferation (36,37), and because STAT5 was phosphorylated in PC12 cells under basal conditions, it may be one of the promoters of proliferation, including the EGF-induced proliferation of control cells. Inhibition of this proliferative stimulus by SOCS2 may be one of the initiators of differentiation/neurite outgrowth observed in the SOCS2-PC12 cells. Addition of the Src inhibitor PP2 inhibited FIG. 6. Overexpression of SHP2 blocks SOCS2 activity and SOCS2 binds to the EGFR. A-F, ␤III-tubulin immunostaining (red) of SOCS2-expressing cells (green) (A) transiently transfected with expression plasmids encoding the phosphatases SHP1 (B), SHP2 (D and E), and catalytically inactive mutants of SHP1(CϾS) (C) or SHP2(CϾS) (F). Neurite outgrowth was inhibited only in SHP2-transfected cells (D, E, and G). Bar in F, 20 m. H, EGFR activation at Tyr 845 was also decreased only in SHP2-transfected SOCS2-PC12 cells. I, lysates of wild-type (Wt) and SOCS2-overexpressing (S2) neurons were immunoprecipitated using anti-FLAG antibody to pull down the FLAG-tagged SOCS2 transgene. Precipitates were then immunoblotted and probed for the presence of the EGFR, which was present in SOCS2-overexpressing lysates but not wild-type lysates.
the neurite outgrowth observed in SOCS2 cells, but it did not further inhibit total STAT5 phosphorylation in the SOCS2 cells, indicating that other Src-activated pathways regulated by SOCS2 are also probably involved in the promotion of neurite outgrowth.
EGF can regulate cytoskeletal rearrangement through activation of RhoA (38 -40). Moreover, EGF can induce neurite outgrowth through c-Src activation (20), whereas overexpression of v-Src induces neurite outgrowth in PC12 cells (21), and blocking of c-Src inhibits neurite outgrowth (22,23). Src-induced neurite outgrowth is mediated by activation and recruitment of the adaptor molecule Crk to the Src substrate Sin/Efs (20). Signaling downstream of these molecules in 293T cells has been shown to include activation of Rac1 (24), as we observed in the SOCS2-PC12 cells, whereas v-Crk has been reported to control Rho activation and modulate axonal growth (41,42). Given that RhoA activation in the SOCS2-PC12 cells was significantly inhibited, whereas Rac1 was activated, it appears that SOCS2 is a regulator of Src-mediated neurite outgrowth by promoting EGFR activation.
Sustained EGFR phosphorylation could be due to a lack of dephosphorylation by phosphatases, continuous phosphorylation by activated c-Src at Tyr 845 or Tyr 1101 (43,44), or a lack of degradation of the activated receptor. Given that total EGFR levels were similar in SOCS2-overexpressing and control cells, it does not appear that SOCS2 blocks EGFR degradation, although conversely, SOCS1 and SOCS3 have been proposed to negatively regulate EGFR activation, possibly by promoting EGFR degradation (45). SOCS3 has been shown to interact with the binding sites for the tyrosine phosphatase SHP2 on the gp130 cytokine receptor (46,47) and the leptin receptor (48), and SOCS2 binds to the SHP2 site, Tyr 595 , of the GH receptor (18). We show here that SOCS2 binds to the EGFR and that overexpression of SHP2, but not SHP1, inhibited both the SOCS2-induced neurite outgrowth and the constitutive phosphorylation of the EGFR. Furthermore, we showed that the constitutive phosphorylation of the EGFR is at Tyr 845 , the Src binding site, and that other tyrosine residues on the EGFR, such as Tyr 1068 , which activates the extracellular signal-regulated kinase/MAPK pathway, are not differentially activated in control and SOCS2-overexpressing cells. Therefore, we propose a mechanism whereby SOCS2 may competitively bind to Tyr 845 , the Src kinase site on the EGFR, and block its dephosphorylation by SHP2, resulting in constitutive EGFR phosphorylation and subsequent cytoskeletal rearrangement by Rac and Rho.