Suppressor of Cytokine Signaling 6 Associates with KIT and Regulates KIT Receptor Signaling*

Suppressor of cytokine signaling (SOCS) proteins are a family of Src homology 2-containing adaptor proteins. Cytokine-inducible Src homology domain 2-containing protein, SOCS1, SOCS2, and SOCS3 have been implicated in the down-regulation of cytokine signaling. The function of SOCS4, 5, 6, and 7 are not known. KIT receptor signaling is regulated by protein tyrosine phosphatases and adaptor proteins. We previously reported that SOCS1 inhibited cell proliferation in response to stem cell factor (SCF). By screening the other members of SOCS family, we identified SOCS6 as a KIT-binding protein. Using KIT mutants and peptides, we demonstrated that SOCS6 bound directly to KIT tyrosine 567 in the juxtamembrane domain. To investigate the function of this interaction, we constitutively expressed SOCS6 in cell lines. Ectopic expression of SOCS6 in Ba/F3-KIT cell line decreased cell proliferation in response to SCF but not SCF-induced chemotaxis. SOCS6 reduced SCF-induced activation of ERK1/2 and p38 but not activation of AKT or STATs in Ba/F3, murine embryonic fibroblast (MEF), or COS-7 cells. SOCS6 did not impair ERK and p38 activation by other stimuli. These results indicate that SOCS6 binds to KIT juxtamembrane region, which affects upstream signaling components leading to MAPK activation. Our results indicate that KIT signaling is regulated by several SOCS proteins and suggest a putative function for SOCS6 as a negative regulator of receptor tyrosine kinases.

The eight members of the SOCS family, SOCS1-7 and CIS (cytokine-inducible Src homology domain (SH2)-containing protein), are structurally characterized by a SH2 domain followed by a conserved C-terminal motif, the SOCS box (4). The N-terminal region of SOCS proteins is variable both in length and in the primary amino acid sequence. Although many reports including knock-out studies shed light on the function of CIS (5,6), SOCS1 (7,8), SOCS2 (9,10), and SOCS3 (11)(12)(13), very little is known regarding the function of SOCS4, SOCS5, SOCS6, and SOCS7.
The mechanisms whereby CIS, SOCS1, and SOCS3 inhibit signaling by classical cytokine receptor (i.e. receptors without catalytic activity that associate with JAK tyrosine kinases) are the best characterized. All three are involved in the downregulation of the JAK/STAT pathway. SOCS1 has a dual function as a direct potent JAK kinase inhibitor (14 -17) and as a component of an E3 ubiquitin-ligase complex recruiting substrates to the protein degradation machinery (18 -20). SOCS3 also inhibits JAK activity but indirectly through recruitment to the cytokine receptors (1,21). More recently, SOCS3 has been suggested to compete with SHP2 for the same binding sites on glycoprotein 130 (22,23), erythropoietin receptor (21), and leptin receptor (24). CIS binds to cytokine receptors at STAT5docking sites, which impairs recruitment of STAT5 to the receptor signaling complex and results in the down-regulation of STAT5 activation (6,25).
Mice lacking SOCS6 have been generated, and they developed normally with the exception of a 10% reduction in weight compared with wild-type littermates (26). SOCS6 mRNA was induced by erythropoietin in cell lines (27) and was ubiquitously expressed in murine tissues (26). SOCS6 does not interact with JAKs, but the interaction with elongins B and C suggests that, as all SOCS proteins, it might be part of an E3 ubiquitin-ligase complex (28). Yet, there is no evidence so far suggesting that SOCS6 might be involved in the degradation of proteins.
Regulation of receptor tyrosine kinases (RTK) by SOCS proteins is much less understood. In vivo interaction between SOCS proteins and RTK has been previously reported. SOCS1 binds to KIT, FLT3 (29), and FMS (30). SOCS2 binds to the insulin-like growth factor receptor (31). SOCS1 and SOCS3 bind to EGFR and may down-regulate activation of STATs by EGFR (32). SOCS3 interacts with insulin receptor (IR) (33).
More recently, SOCS1 and SOCS6 have also been shown to associate with and inhibit IR downstream signaling events such as the activation of ERK1/2, AKT, and IRS-1 when expressed in hepatoma cells (34).
Binding of stem cell factor (SCF) to KIT RTK activates multiple signal transduction components, leading to the activation of all three MAPKs pathways, phosphatidylinositol 3-kinase, and STAT1, STAT3, and STAT5. We have previously identified SOCS1 protein as downstream component of the KIT receptor signaling pathway (29). We have demonstrated that SOCS1 bound to KIT via its SH2 domain. Constitutive expression of SOCS1 strongly suppressed the proliferative signals transduced by KIT without suppression of KIT catalytic activity. Neither SOCS1 docking site nor the mechanism of SOCS1 inhibition could be determined in these studies. However, we showed that SOCS1 not only interacted with KIT and JAK kinases but also interacted with SH3 domain-containing proteins, other receptor tyrosine kinases, and with VAV proteins.
Here, we have screened the putative interactions of SOCS proteins with the KIT receptor in response to SCF stimulation. Our findings revealed that SOCS6 protein interacts with KIT following SCF-stimulated tyrosine phosphorylation. The SH2 domain of SOCS6 binds directly to tyrosine 567 in KIT juxtamembrane domain. To determine the functional consequences of this interaction, we have studied the effect of SOCS6 ectopic expression on the cell proliferation and migration induced by SCF and investigated which signaling pathways are regulated by SOCS6.

EXPERIMENTAL PROCEDURES
Cell Culture-Ba/F3 were cultured in RPMI 1640 medium with 10% fetal bovine serum (FBS) in the presence of 0.1% of conditioned medium from X63-IL-3 cells (35). EML-C1 cells were grown in RPMI 1640 medium supplemented with 20% FBS and 250 ng/ml murine SCF. COS-7, R4 MEFs, Phoenix A, and GPϩE-86 retrovirus-packaging cell lines were maintained in Dulbecco's modified Eagle's medium supplemented with 10% FBS and 0.5 mg/ml sodium pyruvate. All of the media and sera were purchased from Invitrogen.
Reverse Transcription and PCR-Total RNAs were extracted from cell lines and bone marrow-derived mast cells using TRIzol reagent (Invitrogen). First-strand cDNAs were synthesized with 2 g of RNA using oligo(dT) primers and Superscript II reverse transcriptase (Invitrogen), and one-tenth of the cDNA was then subjected to 35 cycles of PCR amplification using SOCS6 primers (sense, 5Ј-CTCTCACCATT-GCTACCTCCAA-3Ј; antisense, 5Ј-TCTCGCCCCCAGAATAGATGTAG-3Ј, 468-bp product) and 2 units of Taq polymerase (Invitrogen). Amplification of murine and human ␤ 2 -microglobulin RNAs were used as controls.
Expression Constructs-pEF-FLAG-SOCS6 construct was kindly provided by D. Hilton (The Walter and Eliza Hall Institute). To convert tyrosine residues (at codons 544-546-552-567-569-577-702-719-728-745 and 934) to phenylalanine residues in KIT, point mutations were generated by site-directed mutagenesis using Stratagene Chameleon double-stranded mutagenesis kit. The reactions were performed on mouse KIT cDNA in the pECE-KIT vector. All of the cDNAs were entirely sequenced. The KIT-JM construct contained all six Y544F, Y546F, Y552F, Y567F, Y569F, and Y577F mutations in the juxtamembrane domain. The KIT-KI construct contained all four Y702F, Y719F, Y728F, and Y745F mutations in the interkinase domain. Retroviral vectors pMIEV and pMIEV-hemagglutinin-SOCS1 were described previously (29). The FLAG-SOCS6 cDNA was cloned in pMIEV using the Gateway system (Invitrogen). The yeast two-hybrid vectors are described below.
Transfection Procedure-Transfection of COS-7 cells was carried out in 60-mm plates. Cells were transfected with FuGENE 6 (Roche Applied Science) as recommended by the manufacturer's instructions with 1 g of expression vector. Cells were serum-starved overnight 24 h after transfection in Dulbecco's modified Eagle's medium with 0.5% FBS and then stimulated for 5 min with 250 ng/ml murine SCF.
Immunoprecipitation and Immunoblotting-Stimulated cells were washed in ice-cold phosphate-buffered saline prior to lysis, pelleted, and lysed in HNTG buffer (50 mM HEPES, pH 7, 50 mM NaF, 1 mM EGTA, 150 mM NaCl, 1% Triton X-100, 10% glycerol, 1.5 mM MgCl 2 ) containing protease inhibitor mixture (Roche Applied Science) and 100 M Na 3 VO 4 . Clarified whole cell lysates were mixed for 18 h with 2 g of antibodies and a bed volume of 10 l of protein A or protein G-Sepharose (Amersham Biosciences) for immunoprecipitation. The immunoprecipitates were washed three times with HNTG buffer and dissolved in SDS sample buffer. Following SDS-PAGE, proteins were transferred to polyvinylidene difluoride membranes (Immobilon-P, Millipore). Membranes were saturated with 5% bovine serum albumin (Sigma) and probed with different antibodies as specified in the text and figure legends. Blots were revealed using horseradish peroxidase-conjugated secondary antibodies and an ECL detection kit (Amersham Biosciences).
Yeast Two-hybrid System-The entire cytoplasmic domains of wildtype murine KIT, KIT-JM, or KIT-KI mutants (nucleotides 1646 -2988) were fused to LexA DNA-binding domain in the yeast expression vector pBTM116. SOCS6 cDNA coding for SH2 and C-terminal domains (amino acids 310 -533), SOCS6 SH2 domain alone (amino acids 310 -479), and SOCS1 full-length cDNA were cloned in the vector pACT2 using the Gateway system (Invitrogen). The vectors pLexA-lamin and pACT2 were used as controls for each yeast two-hybrid experiment.
GST Pull-down Experiments-The GST-SOCS6 construct in pD-EST15 vector-expressing SOCS6 amino acids 310 -533 was introduced in the Escherichia coli Rosetta (DE3) pLacI strain (Novagen). Bacterial cultures grown to log phase were induced with 0.1 mM isopropyl-␤-Dgalactopyranoside (Invitrogen) for 4 h at 30°C. Bacteria were then lysed in HNTG buffer, and the GST fusion proteins were purified on glutathione-Sepharose beads (Amersham Biosciences). Lysates from non-stimulated and SCF-stimulated Ba/F3-KIT cells were incubated overnight at 4°C with 5 g of the GST fusion construct or with 5 g of GST-␤-catenin control immobilized on gluthatione-Sepharose beads. The beads were washed three times in HNTG buffer, and bound fractions were separated by SDS-PAGE and transferred to polyvinylidene difluoride membranes. Bound KIT protein was visualized by Western blotting using the 4G10 antibody.
Peptide Binding Assay-Forty-four 15-mer peptides corresponding to the 22 tyrosine motifs of KIT intracellular domain either phosphorylated or not were synthesized on cellulose membranes as described previously (36). The membranes were saturated with 10% FBS in TBST buffer (100 mM Tris-HCl, pH 7.5, 150 mM NaCl, 1% Tween 20) for 1 h at room temperature. It was then probed overnight at 4°C with 32 Plabeled GST-SOCS6 (amino acids 310 -533) and washed in TBST buffer. The interactions were analyzed by autoradiography and quantified using a PhosphorImager (Molecular Dynamics). The probe was prepared in a reaction mixture containing 5 g of GST-SOCS6, 20 Ci of [␥-32 P]ATP, 0.2 M ATP, and 2500 units of cAMP-dependent protein kinase catalytic subunit (New England Biolabs) in the manufacturer's kinase buffer (50 mM Tris-HCl, pH 7.5, 10 mM MgCl 2 ). Unincorporated ATP was removed, and the labeled protein was eluted with 20 mM glutathione.
Infection of Cells with Retroviral Vectors-Stable populations of GPϩE-86 cells producing MIEV or MIEV-SOCS6 retroviruses were obtained as follow. Phoenix A cells were transfected using FuGENE 6 reagent with pMIEVSOCS6 or empty pMIEV. In the next 2 days, filtered retroviral culture supernatants were used to infect GPϩE-86 cells. GPϩE-86 populations were then sorted for GFP expression. Infections of Ba/F3 or EML cells were done by coculture in 100-mm plates using 10 6 GPϩE-86 cells stably producing retroviruses and 10 6 Ba/F3-KIT or Ba/F3-EGFR or EML with the target cell culture media and 4 ng/ml Polybrene (Sigma). Non-adherent cells were collected 48 h later and sorted for expression of GFP using a cell sorter. Infected R4 MEFs were obtained through similar procedures but with filtered retroviral culture supernatant instead of coculture.
Cell Proliferation Assay-A total of 5000 cells/well were plated in triplicate into 96-well plates in 100 l of RPMI 1640 medium with 10% FBS and 250 ng/ml SCF. Cells were incubated for 24 h at 37°C and pulsed for 6 h with 0.5 Ci of [methyl-3 H]thymidine (Amersham Biosciences). Cells were then transferred onto glass fiber filters (Packard, Netherlands), and incorporation was measured using a ␤-counter Rackbeta Compact 1212-411 (LKB, Uppsala, Sweden).
In Vitro Two-chamber Migration Assay-Chemotaxis was assayed by a modification of the Boyden chamber assay. 3 ϫ 10 5 serum-starved Ba/F3-KIT cells in 100 l of chemotaxis media (RPMI 1640 with 0.5% bovine serum albumin) were added to the upper chamber of a Costar Transwell 24-well plate (6.5 mm diameter, 5 m pore size, Cambridge, MA). 600 l of chemotaxis medium was added to the lower chamber. SCF was added in the lower chamber to create a positive gradient. Transwell plates were incubated at 37°C for 1 h. SCF-induced cell migration into the lower chamber was measured by counting using FACScan for 20 s at high flow rate. Average cell number and mean Ϯ S.D. were calculated from quadruplet wells.

SOCS6 Protein Associates with KIT Receptor in Response to
SCF-We earlier reported that SOCS1 physically associated with KIT receptor through its SH2 domain in response to KIT-ligand (SCF)-induced activation (29). To determine whether the other SOCS proteins interacted with KIT, we transiently expressed FLAG-tagged SOCS proteins and KIT in COS-7 cells. Transfected cells were serum-starved overnight prior to stimulation with SCF for 5 min. SCF induced a strong association of SOCS6 with KIT (Fig. 1) and a weak association of SOCS4 and SOCS5 with KIT (data not shown). By contrast, CIS bound to KIT constitutively, whereas SOCS2 and SOCS3 did not interact (data not shown).
Transcripts encoding CIS, SOCS1, SOCS2, and SOCS3 are up-regulated following cytokine signaling (1-3). As shown Fig stimulation in primary cultures of bone marrow-derived mast cells. Additionally, socs6 mRNA was present in cells that express KIT, including the cell lines UT-7, TF1, and MO7e and ES cells (Fig. 2B). These data indicate that socs6 is expressed in KIT-positive cells and that KIT signaling induces socs6 mRNA expression.
We set out to analyze the interaction of SOCS6 with KIT. As shown Fig. 1, KIT was detected in SOCS6 immunoprecipitates following KIT activation by SCF (Fig. 1, panel A, lane 7). Conversely, SOCS6 was detected in KIT immunoprecipitates (Fig. 1, panel C, lane 7). The interaction was dependent on SCF stimulation and tyrosine phosphorylation of KIT. Protein expression in the transfected cells was confirmed by immunoblotting of immunoprecipitates with the anti-FLAG (Fig. 1, panel  B) and anti-KIT (Fig. 1, panel D) antibodies, respectively. Tyrosine phosphorylation of KIT was controlled by using an antiphosphotyrosine monoclonal antibody (Fig. 1, panel E). The binding of SOCS6 to KIT was confirmed by two other experimental procedures, the yeast two-hybrid system (Fig. 4A) and GST pull-down experiments (Fig. 5B).
We then studied the kinetics of SOCS6/KIT interaction. As shown in Fig. 3, top panel, KIT coimmunoprecipitated with SOCS6 within 5 min (lane 3). The amount of KIT-SOCS6 interaction increased over time, reached a peak at 30 min, and decreased after 1 h. Moreover, SCF stimulation induced tyrosine phosphorylation of SOCS6 (Fig. 3, middle panel). We concluded that SOCS6 rapidly and stably interacts with KIT and becomes tyrosine-phosphorylated following SCF stimulation.
SOCS6 SH2 Domain Interacts with KIT Phosphorylated at Position Tyr-567-The fact that the KIT/SOCS6 association required the phosphorylation of KIT suggested that the interaction involved SOCS6 SH2 domain and a phosphotyrosine residue in KIT. Using the yeast two-hybrid system, we recapitulated the KIT/SOCS6 interaction and found that SOCS6 SH2 domain alone was sufficient to interact with KIT intracellular domain, although the interaction was weaker than with longer SOCS6 constructs (Fig. 4A).
KIT contains 22 tyrosine residues in the cytoplasmic domain, 11 of which lie outside the kinase domain (Fig. 4B). The docking sites for known KIT interactors have been mapped either in the juxtamembrane region or the kinase insert region of KIT. To determine which tyrosine residue was involved in the interaction with SOCS6 protein, we have constructed a series of mutants with tyrosine residues replaced by phenylalanine. First, we generated two mutants, one with all six juxtamembrane domain tyrosines mutated (JM mutant with substitutions Y544F, Y546F, Y552F, Y567F, Y569F, and Y577F) and the other with all four kinase insert tyrosines mutated (KI mutant with substitutions Y702F, Y719F, Y728F, and Y745F), respectively. By contrast with wild-type KIT and KIT-KI, which coimmunoprecipitated with SOCS6 (Fig. 4, C, lane 3,  and D, lane 6), KIT-JM mutant did not interact with SOCS6 both in transfected COS-7 cells (Fig. 4D, lane 8) and in the yeast two-hybrid system (Fig. 4E). Yet, KIT JM still interacted with other KIT partners such as SOCS1 (Fig. 4E), p85, or GRB2 (data not shown). We concluded that SOCS6-docking site was located in KIT JM region.
We then generated and screened other KIT mutants with various combinations of juxtamembrane tyrosine mutations using transient transfections in COS-7 cells. A mutant with Y544F, Y546F, Y552F, and Y577F mutations retained SOCS6 binding site (Fig. 5A, lane 4). This finding suggested that Tyr-567 and Tyr-569 were sufficient to promote the binding of SOCS6 to the KIT JM region. Accordingly, mutants with both Y567F and Y569F mutations did not bind SOCS6, indicating that Tyr-567 and Tyr-569 were necessary for the interaction (data not shown). To further investigate the requirement for Tyr-567 and Tyr-569, GST pull-down experiments were done using cell lysates from Ba/F3-KIT, Ba/F3-KITY567F, and Ba/ F3-KITY569F cell lines. GST-SOCS6 or GST-␤-catenin control fusion proteins were immobilized on glutathione-Sepharose beads and were mixed with SCF-stimulated cell lysates. The interaction with activated KIT was revealed using the 4G10 antibody. GST-SOCS6 but not GST-␤-catenin associated with KIT WT following SCF stimulation (Fig. 5B, lane 2). The mutation of either Tyr-567 or Tyr-569 abolished the interaction, indicating that both tyrosine residues were indeed required for SOCS6 binding.
Finally, to determine both whether the interaction is direct or indirect and which of the two phosphotyrosine residues is targeted by SOCS6 SH2 domain, we screened 44 peptides corresponding to all 22 motifs with Tyr residues in KIT intracellular domain either phosphorylated or not. The peptides were immobilized on a membrane and subjected to far-Western blotting using radiolabeled GST-SOCS6 protein as a probe. Phosphopeptide INGNNpY 567 VY 569 IDPTQLP containing phosphorylated (p) Tyr-567 and non-phosphorylated Tyr-569 was the major peptide recognized by GST-SOCS6 (Fig. 6). Interestingly, the same peptide sequence phosphorylated at both 567 and 569 positions did not interact with SOCS6. We concluded that SOCS6 interacts directly with KIT and that the binding site is the amino acid sequence surrounding phosphorylated Tyr-567.
SOCS6 Negatively Regulates KIT Receptor Proliferation Signal but Not SCF-induced Chemotaxis-It is hypothesized that SOCS proteins are negative regulators of cytokines. We first tested the negative potential of SOCS6 on KIT-dependent cellular proliferation. Ba/F3-KIT cells were infected by coculture with retrovirus-producing GPϩE-86 cells stably expressing either the empty retroviral vector pMIEV or pMIEV-SOCS6. As a positive control, we used GPϩE-86 cells stably expressing pMIEV-SOCS1, because we have previously shown that SOCS1 abolished KIT-dependent mitogenic signals (29). An IRES sequence in pMIEV-SOCS6 and pMIEV-SOCS1 constructs allowed transcription of a single mRNA encoding for both enhanced green fluorescent protein (EGFP) proteins (Fig.  7A) and SOCS (Fig. 7B). Infected cells were isolated on the basis of EGFP expression by flow-cytometry cell sorting (Fig.  7A). In these and the following experiments, the expression of KIT was not affected by SOCS6 ectopic expression (Fig. 7A). A thymidine incorporation assay was carried out immediately after cell sorting. We repeatedly observed that SOCS6 ectopic expression led to a 40% decrease of cell proliferation compared with control MIEV cells in response to SCF (Fig. 7C). We also consistently observed a moderate decreased of cell proliferation in response to IL-3. To control that SOCS6 is not general inhibitor of cell proliferation, we tested the proliferation of Ba/F3-EGFR cells. Fig. 7D shows that cell proliferation of Ba/ F3-EGFR cells was not inhibited when SOCS6 protein was expressed. Rather, we consistently observed that SOCS6 increased cell proliferation. The inhibition of SCF-mediated cell proliferation was confirmed using another cell line, EML-C1 (50% decrease of cell proliferation), which expresses endogenous KIT and is dependent on SCF for cell proliferation (Fig. 7E).
SCF induced chemotaxis of hematopoietic cells (37). We have examined the effect of ectopic expression SOCS6 on cell migration using an in vitro two-chamber assay. The dose response to SCF chemotactic property was assessed for Ba/F3-KIT cells. The optimal chemotactic response was obtained at 100 ng/ml SCF in the lower chamber (data not shown). We have also evaluated the kinetics of migration in response to SCF within a range from 30 min to 4 h. The best percentage of cells that migrate in the presence of SCF over the nonspecific migration (in the absence of SCF) was obtained after 1 h of migration. Ba/F3-KIT cells expressing SOCS6 showed identical migration induced by SCF to control cells (Fig. 8). We concluded that SOCS6 partially inhibited SCF-induced cell proliferation but not SCF-induced chemotaxis.
ERK and p38 MAPK Activation downstream of KIT but Not AKT nor STATs Are Down-regulated by SOCS6 -We next explored how SOCS6 protein affected SCF-mediated signal transduction, thereby leading to inhibition of cell proliferation. Ba/ F3-KIT cells expressing SOCS6 protein or the empty vector were stimulated or not with SCF for 5 min, and then activation of ERK1/2 and p38 MAPK was examined by Western blotting using phospho-specific antibodies. SCF-mediated activation of ERK1/2 and p38 were reduced in Ba/F3-KIT cells expressing SOCS6 by 25 and 50%, respectively (Fig. 9A). We preferentially observed the phosphorylation of ERK2 over ERK1 in response to KIT activation in these and in the following experiments. By contrast, phosphorylation of STAT5 following KIT activation was not modified (Fig. 9B). We next analyzed these pathways in other cell lines. COS-7 cells were transfected with FLAGtagged SOCS6 and KIT, and SCF-mediated activation of ERK1/2 and p38 MAPK were reduced by 50% in the presence of SOCS6 (Fig. 10A, lanes 4 and 5). These results were confirmed in mouse embryonic fibroblasts (MEFs) stably expressing KIT and SOCS6 protein (Fig. 10B). Again, ectopic expression of SOCS6 inhibited SCF-mediated activation of p38 and ERK1/2 (80 and 75% reduction, respectively) but not AKT activation (Fig. 10C). These results indicate that SOCS6 reduced the activation of MAPKs following the activation of KIT.
We next investigated whether SOCS6 inhibited ERK1/2 and p38 activation by other stimuli. We first tested the stimulation of MEFs with EGF. Fig. 11A shows that SOCS6 protein did not down-regulate the ERK1/2 and p38 MAPK activation by EGF. We also examined the activation of ERK1/2 and p38 proteins by PMA and anisomycin, respectively. The expression of SOCS6 in MEFs did not impair the activation of ERK1/2 and p38 MAPK induced by PMA (Fig. 11B) and anisomycin (Fig. 11C) treatments. We concluded that SOCS6 is not an inhibitor of the MAPK pathway. Rather, SOCS6 protein specifically inhibits an early event in KIT signaling that is upstream of ERK1/2 and p38 activation. DISCUSSION This study was initiated to investigate the implication of members of the SOCS family in the regulation of KIT receptor signaling. We reported that SOCS6 directly interacts with KIT and may down-modulate signaling pathways initiated in KIT juxtamembrane domain, resulting in decreased cell proliferation.
Although multiple studies have been published on SOCS family members CIS and SOCS1, 2, and 3, very little is known regarding the larger members SOCS4 -7. We have previously identified SOCS1 as a KIT-binding protein in a yeast twohybrid screen. We then tested whether the other members of the family were KIT interactors. We found that SOCS4 and SOCS5 weakly interacted, whereas CIS and SOCS6 stably interacted with KIT.
SOCS6 Binds a Multifunction Docking Site-SOCS6 has been recently shown to interact with the cytoplasmic adaptor proteins IRS-2, IRS-4, p85␣ and p85␤ (26), and with insulin receptor (34). The structural requirements for the interactions were not determined in these studies. Using point mutations in KIT and KIT peptides, we mapped SOCS6-docking site to phos- sequence with Val at the ϩ1 position relative to the phosphorylated Tyr and hydrophobic residues at the ϩ2 and ϩ3 positions. Remarkably, the most frequent residues found at ϩ2 and ϩ3 from the peptide library screen were Tyr and Ile, respectively. The sequence around KIT Tyr-567 perfectly fits the consensus sequence and therefore validates the suggested consensus sequence. KIT Tyr-567 is the first in vivo binding site described for SOCS6.
KIT Tyr-567 is located in KIT juxtamembrane region, a domain thought to regulate the catalytic activity of receptor tyrosine kinases. The peptide sequence that includes Tyr-567 and Tyr-569 has been shown to recruit several other SH2 domaincontaining proteins as follows: the tyrosine phosphatases SHP1 and SHP2 (38), members of Src family kinases (39), the Src kinase-negative regulators CSK homologous kinase/megacaryocyte-associated tyrosine kinase (40), and adaptor proteins of the APS (adaptor molecule containing PH and SHÉ domains)/SH2/ LNK family (41). Thus the di-tyrosine motif 567/569 appears to be a major docking site for the formation of protein complexes following KIT activation. Interestingly, the presence of the two sites of tyrosine phosphorylation creates three different docking sites for a unique primary amino acid sequence depending on the phosphorylation status of Tyr-567 and Tyr-569. Peptide binding assays indicated that SOCS6 interacted when Tyr-567 was the unique site phosphorylated and did not interact when both Tyr-567 and Tyr-569 were phosphorylated. Similarly, SHP2 interacted only with the pY567Y569 peptide. By contrast, SRC bound both pY567Y569 and pY567pY569 peptides and SHP1 bound both pY567pY569 and Y567pY569 peptides. Therefore, different qualitative protein complexes may be generated depending on the kinetics of phosphorylation of Tyr-567 and Tyr-569 following KIT activation. Adding to this complexity, competition for the same site could be involved such that the formation of the complexes would depend on the respective amount of KIT interactors.
However, we could not detect the endogenous KIT-SOCS6 interaction. This could be because of the limited sensitivity of the antibodies or the limited expression level of SOCS6 since endogenous SOCS6 detection with the available antibodies has been a difficult task. SOCS6 mRNA is expressed in most mouse tissues. Taking advantage of the LacZ insertion in SOCS6 locus, Krebs et al. (26) show that SOCS6 is expressed in the majority of bone marrow cells and in most hematopoietic progenitors. In addition, the KIT-positive hematopoietic cell lines UT7, Mo7e (27), and TF1 express socs6 mRNA and protein. Therefore, it is established that the expression profiles of SOCS6 and KIT overlap and that most KIT-positive cells in the hematopoietic system also express SOCS6.
In addition, we have shown that transcription of socs6 was induced by SCF in primary cultures of mast cells. This observation is reminiscent of the data showing that other socs genes encoding CIS, SOCS1, and SOCS3 were early response genes induced in response to cytokine signaling.
Modulation of KIT Signaling by SOCS-We had previously shown that SOCS1 bound to KIT and that SOCS1 ectopic expression inhibited KIT ligand-dependent cell proliferation (29). Because SOCS proteins CIS, SOCS2, and SOCS3 have also been implicated in the negative regulation of cytokines or growth factor signaling, we tested whether SOCS6 modulated KIT-mediated cell proliferation and KIT-mediated chemotaxis. SOCS6 partially inhibited cell proliferation, whereas inhibition by SOCS1 was total. In our experiments, SOCS1 was consistently expressed at higher levels than SOCS6, which may account for this difference (Fig. 7B). SOCS6 low expression level compared with smaller SOCS proteins was previously noticed by others (42). Interestingly, SOCS6 did not impair KIT ligandmediated chemotaxis, indicating that some specific signaling pathways were inhibited while others were unaffected.
The MAPK pathways ERK1/2 and p38 were in part activated through Tyr-567 in the juxtamembrane membrane region of KIT (37). SRC and SHP2, two putative Tyr-567 interactors, are likely to be upstream activators of MAPKs in KIT signaling (37,43). We observed that SOCS6 partially inhibited ERK and p38 FIG. 10. Impaired activation of p38 and ERK1/2 MAPK by SCF in COS and MEFs expressing SOCS6 and KIT. A, COS-7 cells transfected with pEF-FLAG-SOCS6 and pECE-KIT vectors were treated with 250 ng/ml SCF for 5 min before lysis. Lysates were separated by SDS-PAGE, and the blots were probed with antibodies that recognize active ERK1/2 (P-ERK1/2), ERK2, active p38 (P-p38), and p38. The membrane was stripped and reblotted with the different antibodies. The expression of SOCS and KIT proteins were controlled on lysates with FLAG and KIT antibodies. B, lysates from MEFs stably expressing KIT and either the control vector MIEV or SOCS6 were subjected to immunoprecipitation using anti-SOCS6 antibody. The blot was revealed by anti-FLAG antibody to control the expression of SOCS6. C, MEFs were starved for 4 h and then stimulated with 250 ng/ml SCF for 5 min before lysis. Total cell lysates were analyzed by Western blotting with the indicated phosphospecific antibodies. The membrane were stripped and reblotted with antibodies detecting total amount of the respective proteins (AKT, ERK2, and p38). both in fibroblasts and in the Ba/F3 hematopoietic cell line but did not affect AKT or STAT5 phosphorylation.
Ueda et al. (37) reported that Tyr-567 in KIT is implicated in the activation of Src family kinases and that the Y567F mutant shows reduction of MEK1/2 and p38 phosphorylation. The reduction they observed with KIT Y567F is very similar to the partial reduction we detected following ectopic expression of SOCS6. These independent observations support our mapping and ectopic expression data. SOCS6 may bind to Tyr-567 and mask this multifunction-docking site, thus mimicking the Y567F mutation.
Activation of ERK and p38 pathways through EGFR, PMA, or anisomycin stimulation were not inhibited by SOCS6, indicating that the inhibition takes place in the initial stages of KIT activation. Two SOCS proteins are thought to downregulate signaling through competition for binding sites on membrane receptors. First, it has been suggested that CIS inhibits STAT5 phosphorylation following growth hormone or erythropoietin activation by binding to STAT5 binding sites on the respective cytokine receptors (6). Second, the SH2 domains of SOCS3 and SHP2 have similar binding specificities in vitro (44). Other studies show that SOCS3 and SHP2 share the same docking sites on glycoprotein 130 (22,23), erythropoietin receptor (21), and leptin receptor (24). In a similar way, SOCS6 might prevent the recruitment of the SH2 domain-containing proteins such as SRC and SHP2 to the receptor and thereby reduce the activation of downstream pathways (ERK and p38).
Despite the generation of SOCS6 knock-out mice (26), the biological action of SOCS6 remains largely unknown. Hematopoiesis in mice deficient for SOCS6 seems normal. Indeed, SOCS6Ϫ/Ϫ mice do not have obvious phenotypes other than a 10% weight deficit. One possible explanation for the lack of phenotype could be that SOCS7 and SOCS6 had overlapping functions. SOCS7 was approximately 55% identical to SOCS6 within the SH2 and SOCS box domains. The binding specificity of both SH2 domains was very similar as determined in peptide binding assays (26). Furthermore, both SH2 domains were shown to coprecipitate with identical phosphorylated proteins including IRS-2, IRS-4, p85a, p85b, and tubulin (26). The SH2 domain of SOCS7 could interact with KIT, 2 suggesting that SOCS7 could also bind Tyr-567 in KIT and be involved in the regulation of KIT signaling. The generation of double deficient mice for both SOCS6 and SOCS7 should shed light on the presumptive overlapping functions of these two proteins.
The IR was the only other receptor that had been shown to interact with SOCS6 (34). In that case, the phosphorylation of ERK1/2, AKT, and IRS-1 was inhibited by SOCS6 ectopic expression. The authors suggested that SOCS6 may impair IR kinase activity, thereby inhibiting the downstream pathways. However, Krebs et al. (26) showed that several IRS family members interact with SOCS6 and suggest that this could account for an indirect binding of SOCS6 to IR. IRS proteins are central docking proteins used by IR to initiate the signaling cascades. The binding of SOCS6 to IRS could therefore abrogate the binding of other signaling molecules and inhibit ERK and AKT activation. In addition to KIT and IR, FLT3 is also inhibited by SOCS6. 2 By contrast, SOCS6 did not impair LIF signaling (42). SOCS6 is so far the only SOCS protein that has been implicated exclusively in the regulation of RTKs. 2 J. Bayle and P. De Sepulveda, unpublished data.
FIG. 11. Activation of p38 and ERK1/2 by EGF, PMA, or anisomycin is not impaired by SOCS6. MEFs stably expressing KIT and SOCS6 or MIEV control vector were starved for 4 h and stimulated or not with 100 ng/ml EGF for 5 min (A), or 50 ng/ml PMA for 2 min (B), or 10 ng/ml anisomycin for 5 min (C) before lysis. Total cell lysates were analyzed by Western blotting with the indicated phosphospecific antibodies. The membranes were then stripped and reblotted with antibodies detecting total amount of the respective proteins (ERK2 and p38).