Insulin Induces SOCS-6 Expression and Its Binding to the p85 Monomer of Phosphoinositide 3-Kinase, Resulting in Improvement in Glucose Metabolism*

The suppressors of cytokine signaling (SOCS) family is thought to act largely as a negative regulator of signaling by cytokines and some growth factors. Surpris-ingly, the SOCS-6 transgenics had no significant defects in the cytokine signaling and hematopoietic system but displayed significant improvements in glucose metabolism. Insulin stimulation of Akt/protein kinase B was also potentiated. Biochemical analysis showed that, after insulin stimulation, SOCS-6 interacted with the monomeric p85 subunit of class-Ia phosphoinositide (PI) 3-kinase but not with p85/p110 dimers. Furthermore, SOCS-6 expression is transiently increased by serum and insulin in normal fibroblasts. However, both the mRNA and protein of SOCS-6 were rapidly degraded after induction by insulin. The degradation of the SOCS-6 protein

The suppressors of cytokine signaling family (SOCS) 1 consists of eight proteins (CIS and SOCS-1-7) that were first characterized as negative feedback regulators for cytokine receptor signaling. Members of the SOCS family share a common structure containing a central SH2 domain and a C-terminal SOCS box, and variability in their sequences at the N terminus probably confers specificity of function (1). The SH2 domain allows the SOCS proteins to interact with tyrosine-phosphorylated target proteins such as Janus kinases (JAKs), and the SOCS box is suggested to act as a common binding domain for a large family of E3 ubiquitin protein ligases (2)(3)(4). In this manner, molecules containing the SOCS domain may regulate protein degradation by targeting proteins for ubiquitination and, subsequently, for proteasome-mediated degradation (5). CIS, SOCS-1, SOCS-2, and SOCS-3 are the most widely studied members of SOCS family, and it is known that the expression of these genes is rapidly induced by an array of stimuli such as cytokines (6 -8) and growth factors (9 -11), including insulin (12,13). The increased levels of CIS, SOCS-1, SOCS-2, and SOCS-3 result in feedback inhibition of a wide variety of agonist-induced signaling pathways (14). However, this finding is not just restricted to cytokine signaling. For example, in the case of insulin signaling there is evidence that the overexpression of SOCS-1 and SOCS-3 attenuates insulin signaling by binding to the insulin receptor (IR) directly and targeting the insulin receptor substrate molecules (IRS-1 and IRS-2) for proteasome-mediated degradation (12,15,16). Thus, SOCS molecules appear to act as part of a classical negative feedback loop, inhibiting the signaling pathways that initially led to the expression of SOCS genes. However, an important feature of this system is that it allows for regulatory cross-talk between signaling pathways. For example, the increase in SOCS-1 and SOCS-3 expression by cytokines has been suggested as the mechanism by which they bring about the attenuation in insulin signaling that causes insulin resistance associated with the development of type 2 diabetes (15).
Despite structural similarities in SOCS proteins, evidence is now emerging to show that different members of the SOCS family have distinct roles. For example, the regulation and function of SOCS-1 and SOCS-3 in cytokine signaling appears to be very similar to both binding to JAKs and the inhibition of kinase activity. However, mice lacking SOCS-1 or SOCS-3 showed completely different phenotypes (17)(18)(19)(20), suggesting that the roles of SOCS-1 and SOCS-3 in signaling outside the cytokine pathways might differ. Furthermore, we observed previously that CIS negatively regulates IL-2 signaling but promotes T-cell receptor stimulation of NF-B and mitogen-activated protein kinase pathways in T-cells (21). This is suggestive of a dual function of CIS in the negative regulation of cytokine signaling and the positive regulation of T-cell receptor signaling, which may be essential for the maintenance of homeostasis of the immune system.
Unlike CIS and SOCS-1-3, SOCS-4 -7 have not been extensively studied (22). SOCS-6 and SOCS-7 have longer N-terminal domains than do CIS, SOCS-1, SOCS-2, and SOCS-3. Like other SOCS proteins, SOCS-6 binds to elongin B/C via the SOCS box domain (23). In contrast to the other SOCS molecules studied, SOCS-6 does not interact with JAK and does not have any inhibitory effect on the signaling induced by the growth hormone, the leukemia inhibitory factor, or prolactin (24). However, a recent report indicates that SOCS-6 has direct effects on IR signaling in regard to reported interactions with the IR- (16) and SH2 domain-mediated interactions with the p85 subunit of phosphoinositide (PI) 3-kinase as well as IRS-2 and IRS-4 (23). This finding would suggest that SOCS-6 might be involved in the regulation of glucose metabolism. However, mice with targeted disruption of the SOCS-6 gene do not display significant defects in glucose metabolism although they do display growth defects, weighing 8 -10% less than wild type mice (23).
In the present study, we established transgenic mice overexpressing SOCS-6 to examine its function in vivo. SOCS-6 transgenics developed normally but displayed improved glucose and insulin tolerance. An insulin-induced interaction of SOCS-6 with the monomer form of the p85 subunit of PI 3-kinase was found and was associated with increased Akt/PKB activation induced by insulin. In addition, we found that SOCS-6 was induced by insulin but rapidly degraded by proteasome. Together, these findings indicate that SOCS-6 could play an important role in regulating insulin action in vivo.

EXPERIMENTAL PROCEDURES
Antibodies-Mouse anti-human c-Myc monoclonal antibody (9E10) was purchased from BD Biosciences. Mouse monoclonal anti-phosphotyrosine (PY99) antibody, goat polyclonal anti-Akt/PKB antibody, and goat polyclonal anti-ERK antibody were purchased from Santa Cruz Biotechnology. Mouse monoclonal anti-phospho-ERK antibody, rabbit polyclonal anti-phospho-Akt/PKB antibody, and rabbit polyclonal antiphospho-IR antibody were purchased from New England Biolabs. Rabbit polyclonal anti-p85 (␣p85pan) antibody and anti-phospho-c-Jun Nterminal kinase were purchased from Upstate Biotechnology, Inc. The anti-phospho-IB kinase was obtained from Cell Signaling Technology. The anti-SOCS-6 antiserum was generated by immunizing rabbits with a fusion protein encoding the SOCS-6 SH2 domain and the SOCS box (Fusion Antibodies, Belfast, Northern Ireland, UK).
Generation of SOCS-6 Transgenic Mice-A murine full-length SOCS-6 cDNA was isolated from a mouse liver cDNA library (Stratagene). SOCS-6 cDNA tagged with the human c-Myc tag at its C terminus was cloned into the EcoRI and NotI sites of vector pEFneo under an elongation factor I promoter. Mice constitutively overexpressing the SOCS-6 transgene were established through an embryonic stem cell approach. Briefly, the vector-free pEF-SOCS-6-pSV40-neo fragment was transfected into E14 cells by electroporation. The Geneticinresistant E14 clones were screened for SOCS-6 transgene expression by Western blotting with anti-c-Myc, and two of the positive clones were injected into blastocysts of C57BL/6. Both of the two clones successfully transmitted to the germ line, and the SOCS-6 transgenics was backcrossed five times to C57/BL6 background before phenotyping was performed.
Metabolic Studies-To determine fasting blood glucose concentrations, two-month-old mice were starved overnight, and total blood was obtained by tail bleeding. All blood glucose concentrations were determined with a MediSense Optium glucometer and MediSense Optium electrodes (Abbott Laboratories). Glucose tolerance tests were performed after overnight fasting by intraperitoneal injection of 2g/kg D-glucose at time 0. Blood samples were taken at the indicated time points, and glucose levels were determined as described (Fig. 3b). For the insulin tolerance test, blood samples were obtained at indicated time points after intraperitoneal injection of 0.75 IU/kg human insulin (Sigma) (Fig. 3c). Plasma insulin levels in the serum from fasting animals were measured by an enzyme-linked immunosorbent assay using mouse insulin as a standard (Crystal Chem Inc., Chicago, IL).
Hyperglycemic Clamps-Under anesthesia, a PE-10 tubing catheter was inserted into the left jugular vein and externalized through an incision in the skin flap behind the head. The mice were allowed to recover for 2 days. Before starting the experiments, mice fasted for 5 h. Then, through the implanted catheter, a 30% glucose solution was infused into the mice at the rates adjusted to maintain the blood glucose concentration at 10 mM, as determined from tail vein blood. Blood was collected after 2 h of performing the glucose clamp technique, and plasma insulin was determined by enzyme-linked immunosorbent assay.
In Vivo Insulin Stimulation and Analysis of Insulin-signaling Proteins-Two-month-old mice were starved overnight, anesthetized with Hypnorm and Hypnovel, and 5 IU of human insulin (Sigma) was injected into the portal vein. Liver and muscles were removed at 3 and 5 min respectively, and instantly frozen in liquid nitrogen. Immunoprecipitation and immunoblot analysis of insulin-signaling molecules were performed on tissue homogenates extracted with homogenization buffer containing 1% Triton X-100, 50 mM HEPES (pH 7.4), 150 mM NaCl, 10 mM EDTA, 50 mM sodium pyrophosphate, 100 mM sodium fluoride, 10 mM sodium orthovanadate, 10 g/ml aprotinin, 5 g/ml leupeptin, 2 mM benzamidine, and 1 mM phenylmethylsulfonyl fluoride.
Northern Blotting-For the study of the regulation of SOCS-6 expression by insulin stimulation, mouse embryo fibroblasts were isolated from embryonic day 13.5 embryos of C57BL/6 mice and maintained in Dulbecco's modified Eagle's medium containing 10% fetal bovine serum. After overnight or 4-h starvation of cells in Dulbecco's modified Eagle's medium without fetal bovine serum, the cells were stimulated with 1 M human insulin at the indicated time points (Fig. 5), and total RNA was extracted with TRIzol (Invitrogen). To confirm SOCS-6 transgene expression in SOCS-6 transgenic mice, liver and muscles were isolated from 6 -8 week-old wild type or SOCS-6 transgenics, and total RNA was extracted as described above. 30 g of total RNA from each sample was fractionated on formaldehyde agarose gel and blotted onto Hybond Nϩ membrane (Amersham Biosciences). The SOCS-6 cDNA probe was a 1-kb fragment that spanned nucleotides 420 to 1434 in the SOCS-6 coding sequence. The glyceraldehyde-3-phosphate dehydrogenase cDNA probe was a 560-bp PCR fragment spanning nucleotides 40 -600 in the glyceraldehyde-3-phosphate dehydrogenase coding sequence and served as the loading control.
PI 3-Kinase Assay-Cleared cell lysates were immunoprecipitated using either a polyclonal anti-sera that recognized all forms of p85 (p85pan) or a monoclonal anti-phosphotyrosine antibody (PY99), washed three times with the tissue homogenization buffer and twice with PI 3-kinase reaction buffer (50 mM HEPES (pH 7.4), 100 mM NaCl, 0.5 mM EDTA, and 1 mM dithiothreitol), and then suspended in 35 l of PI 3-kinase reaction buffer containing 0.2 mg/ml phosphatidylinositol (Sigma). The reactions were initiated by adding 15 l of a MgCl 2 -ATP mixture (5 mM MgCl 2 and100 M ATP) containing 1 Ci of [␥-32 P]ATP to the immunoprecipitates and incubated at room temperature for 20 min. The reaction was stopped by acidification. After the phosphorylated lipids were extracted and evaporated, the pellets were resuspended in 30 l of chloroform/methanol (4:1), The phosphorylated lipids were separated by TLC and visualized by a phosphorimaging device.
Proteasome Degradation Assay-For a pulse-chase assay with CHO cells, the cells were pulsed with [ 35 S]methionine and [ 35 S]cysteine master mix (Amersham Biosciences) after 4 h of starvation. The labeled cells were then incubated with insulin-containing medium, proteasome inhibitor LLnL (Sigma), or Me 2 SO. Cells were harvested and lysed at different time points (Fig. 6). The precipitates with different antibodies, as indicated ( Fig. 6), were resolved by SDS-PAGE, dried, and visualized.
Proteasome Degradation Assay with Cardiomyocyte-Primary neonatal cardiomyocytes were generated from SOCS-6 transgenics and cultured as described previously (25). Briefly, hearts were collected from 1-2-day-old SOCS-6 transgenics, the atrial tissue was removed, and the ventricles were cut into small pieces and digested with 0.05% pancreatin (Sigma) and 100 units/ml collagenase (Sigma) in phosphatebuffered saline. The cells were centrifuged at 1000 rpm for 5 min and cultured in M199 media (Invitrogen) with 15% fetal bovine serum. The non-myocytes were removed by incubation for 1 h in an uncoated dish. The enriched cardiomyocyte fractions were then plated on gelatincoated plates and cultured for 16 h in serum-free M199 media. Cells were then treated with cycloheximide (1 M) for 6 h to inhibit de novo syntheses of proteins in the cardiomyocytes. The treated cells were then incubated with insulin-containing medium, the proteasome inhibitor LLnL, or Me 2 SO. Cells were harvested and lysed at different time points (Fig. 6). The anti-p85 precipitates were resolved by SDS-PAGE and immunoblotted with anti-bodies against c-Myc for the detection of Myc-tagged SOCS-6, p85, and IRS-1, respectively.

Generation of SOCS-6 Transgenic
Mice-Two founder lines of transgenic mice were produced, each overexpressing the Myc-tagged SOCS-6 transgene at different levels ( Fig. 1). Both lines were used for characterization of the phenotype. The expression levels of SOCS-6 in line 1 and line 2 were 5-and 10-fold higher, respectively, than those for wild type litter mates in liver, muscle, and T lymphocytes at either the mRNA ( Fig. 1b) or protein level (Fig. 1a). All transgenic mice appeared to be healthy and grossly normal at baseline. They exhibited a normal reproductive rate, and the mortality rate over 12 months was the same as that for the wild type controls.
Insulin Induces the Interaction of SOCS-6 and p85 Subunit of PI 3-Kinase-SOCS proteins were initially discovered to be the negative regulators of cytokine signaling, with SOCS-1, SOCS-2, SOCS-3, and CIS being associated with the downregulation of JAK-STAT pathways (14). However, SOCS-6 and SOCS-7 have longer N-terminal domains than SOCS-1, -2, -3, and CIS (22), and, despite the structural homology in SH2 and SOCS domains to other SOCS proteins, SOCS-6 does not interact with JAK2 or inhibit signaling by growth hormone, leukemia inhibitory factor, or prolactin (24). This finding suggests that SOCS-6 may not be acting on cytokine-signaling pathways and, consistent with this possibility, we could not detect any suppressive function of SOCS-6 in lymphocytes stimulated by interleukin-2, interferon-␥, interleukin-4, granulocyte/macrophage colony-stimulating factor, and interleukin-12. 2 The phosphorylation of STAT-1, -2, -4, -5, and -6 induced by cytokines was similar between the SOCS-6 transgenics and the wild type counterparts. 2 While the mice were being generated it was reported that SOCS-6 could interact with IRS proteins and the p85 regula-tory subunit of the class-Ia PI 3-kinase (23). To investigate the possibility that these interactions could be occurring in our mice, we examined whether these proteins associated with the overexpressed SOCS-6. Transgenic mice were challenged with insulin, and anti-Myc antibodies were used to immunoprecipitate the SOCS-6 transgene product from liver extracts. The precipitates were Western blotted with antibodies specifically recognizing IRS-2, IRS-4, the IR, and p85, respectively. We could not detect neither the IR nor IRS-2 and IRS-4 in the co-precipitates (data not shown). However, the p85␣ subunit of PI 3-kinase was readily detected in the SOCS-6 co-precipitates (Fig. 2a). Moreover, the interaction of SOCS-6-Myc and p85 was only detected after stimulation with insulin (Fig. 2a). To confirm this interaction in another system, we transfected the SOCS-6-expressing construct into CHO-IR cells. Again, the interaction between p85␣ and SOCS-6 was readily detected, but only after insulin stimulation (Fig. 2b). However, no interaction between SOCS-6 and the IRS proteins was found (data not shown). Immunoprecipitation using an antibody that recognizes all isoforms of p85 showed that the wild type and transfected cells contained similar levels of p85 and the p110␤ catalytic subunit (Fig. 2b). In contrast, SOCS-6 immunoprecipitates from insulin-stimulated CHO-IR cells overexpressing SOCS-6 contained p85␣ but did not contain any p110␤ (Fig. 2b) which indicates the following: (i) that the p85␣ associated with SOCS-6 is a free monomer not bound to the catalytic subunit; and (ii) that the p85␣ associated with SOCS-6 is not recruited to the IRS proteins. The finding that insulin specifically recruits SOCS-6 to p85 monomers is very interesting, as there have been a number of recent studies indicating that p85 can have both positive and negative effects on insulin signaling and that the negative effects overcome the positive effects when high levels of a p85 monomer are present in cells (26,27). Therefore, SOCS-6 could act as a physiologically relevant modulator of the effects of the p85 monomer, either mediating or attenuating the negative effects on insulin signaling.
Improved Glucose Metabolism in SOCS-6 Transgenics-To test the effect of SOCS-6 on insulin action, we investigated glucose metabolism by insulin in SOCS-6 transgenic mice. We found lower fasting glucose levels in both lines of transgenic mice (84 Ϯ 15 mg/dl for transgenic line 1 and 80 Ϯ 18 mg/dl for transgenic line 2 versus 105 Ϯ 33 mg/dl) as compared with the levels for wild type. (Fig. 3a). Both lines of mice also exhibited significant reductions in plasma glucose levels at all points in a glucose tolerance test as compared with the levels for wild type mice, and both lines also showed a significantly better clearance of glucose in an insulin tolerance test (Fig. 3, b and c). Furthermore, significantly higher levels of glucose transport stimulated by insulin were observed in mouse embryonic fibroblasts (MEFs) prepared from the SOCS-6 transgenics as com- After stimulation, cultures were pulsed with 2-deoxy-D(2,6 -3H)glucose for 10 min. The cells were washed three times with cold phosphate-buffered saline and lysed in SDS-phosphate-buffered saline. The lysates at the same protein concentration were submitted to measure radioactivity on a ␤-counter. The results presented are the percentages of increased glucose-uptake in comparison to basal levels from wild type MEF cells. *, p Ͻ 0.05. e, serum insulin concentration was measured during a glucose tolerance test at the indicated time points. Results are presented as mean Ϯ S.D. of n ϭ 5 mice per group. *, p Ͻ 0.05 versus wild type. f, hyperglycemic clamps were preformed with wild type and SOCS-6 transgenic lines 1 and 2. Plasma insulin levels were determined before and after the clamp assay. Data are the means Ϯ S.D. of n ϭ 5 mice per group. The plasma insulin levels induced by the glucose clamp in SOCS-6 and normal control mice were not statistically different. pared with MEFs from control animals (Fig. 3d). The increased rate of glucose disposal could be due to increased insulin sensitivity in peripheral tissues, decreased hepatic glucose output, or increases in insulin secretion. The data from the insulin tolerance tests supports a role for increased insulin sensitivity, and this possibility is further supported by the finding that insulin levels were significantly lower in transgenic animals, both in the fasting state and 30 min after a glucose challenge (Fig. 3e). Furthermore, insulin levels remained lower in transgenic animals even when glucose levels had been clamped at 10 mM glucose for 120 min (Fig 3f).

SOCS-6 Increases the Activity of Akt in Response to
Insulin Stimulation-To examine the effect of SOCS-6 overexpression on insulin-mediated signaling, we investigated the activation of the IR, Akt/PKB, and the p42/p44 ERK. Overexpression of SOCS-6 in either the transgenic mice or CHO-IR cells did not affect the insulin-stimulated phosphorylation of the IR, the IRS, or the ERK (Fig. 4, a and b). In addition, the phosphorylation of IB kinase and c-Jun N-terminal kinase in insulinstimulated liver was also similar to that in wild type and SOCS-6 transgenic mice (data not shown). The recruitment of p85/p110 to insulin-signaling complexes was also unaffected by the overexpression of SOCS-6 in CHO-IR cells (Fig. 4b). However, in both cell types there was a significant increase in the Ser-473 phosphorylation of Akt/PKB, which is indicative of increased activity (Fig. 4, a and b). These findings suggest that SOCS-6 is attenuating the negative effects of p85 monomers, and the effect on potentiating Akt/PKB suggests that SOCS-6 is directly potentiating signaling through PI 3-kinase. However, we do not find any increase in the lipid kinase activity present in p85/p110 dimers (Fig. 4c), and there is in fact a reduced activation of PI 3-kinase in tyrosine-phosphorylated signaling complexes after insulin stimulation (Fig. 4d), indicating that the negative signal of p85 attenuated by SOCS-6 is independent of PI 3-kinase, which is consistent with the similar findings in p85 knock-out (26 -30).
Induction of SOCS-6 Expression by Insulin-The experiments described above suggest a role for SOCS-6 in the regulation of PI 3-kinase signaling, but this role can only have physiological relevance if SOCS6 is normally present in cells. We went on to investigate the regulation of SOCS-6 gene expression in MEF cells from C57BL/6 and found that, upon serum starvation, SOCS-6 mRNA levels fell to Ͻ15% of that observed in cells maintained in fetal bovine serum (Fig. 5), a finding consistent with previous reports (23). The level of SOCS-6 expression was significantly but transiently increased by insulin (Fig. 5). SOCS-7, a close homologue to SOCS-6, was also induced by insulin (Fig. 5), although to a much lesser extent. This is suggestive of a possible overlapping function between these two molecules in insulin signaling.
SOCS-6, But Not the p85 Subunit, Is Degraded by Proteasome-The SOCS members share sequence similarity within a 40-residue C-terminal motif termed the SOCS box. It has been discovered that the SOCS box interacts with elongin B/C, which is suggested to target the SOCS-associated molecules to the proteasome degradation pathway (1). Recent findings revealed that SOCS-6 binds to elongin B/C and may target associated proteins to the proteasome for degradation (23), but the finding that SOCS-6 overexpression did not alter p85␣ levels suggests that SOCS-6 is not targeting the regulatory subunit for degradation (Fig 4a). To address whether SOCS-6 targets p85 to the proteasome pathway for degradation, we performed pulse-chase experiments in which the degradation of newly synthesized SOCS-6 and SOCS-6-associated p85 was moni- FIG. 4. SOCS-6 increases Akt/PKB activation in response to insulin stimulation. a, protein lysates from livers of wild type (WT) or SOCS-6 transgenic mice stimulated with saline or insulin (Ins) were blotted (IB) with anti-pY99 and p85 antibodies. MEF cells isolated from wild type and two SOCS-6 transgenic mice lines (Tg1 and Tg2) were starved overnight and stimulated with 1 M insulin for 15 min. Protein lysates were blotted with anti-phospho-Akt/PKB (pAkt), Akt/PKB (Akt), phospho-ERK (pErk), and ERK2 (Erk2) antibodies. b, protein lysates from CHO cells overexpressing IR alone or both IR and SOCS-6 were blotted with anti-phospho-IR (pIR), anti-phospho-Akt/PKB (pAkt), and ␤-actin antibodies or immunoprecipitated (IP) with pY99 antibody followed by immunoblotting with anti-SOCS-6 (SOC-6), p85, and p110␤ (p110) antibodies. c and d, CHO cells overexpressing IR alone or both IR and SOCS-6 were stimulated with 1 M insulin for 10 min. Protein lysates were submitted for immunoprecipitation by pY99 antibody (c) or p85 antibody (d), and PI3 kinase activity was measured as described under "Experimental Procedures." The radioactivity was quantitated in each PI3P spot (indicated by arrow) and graphed. PBS, phosphate-buffered saline. tored in CHO-IR cells in the presence and absence of the proteasome inhibitor LLnL. The newly synthesized SOCS-6 was rapidly degraded within 2 h (Fig. 6a, lower panel). As shown in Fig. 6a, LLnL partially inhibited the degradation of SOCS-6, which supports the finding of the interaction between the SOCS domain and elongin B/C (23). The level of the reduction of SOCS-6 degradation by LLnL is similar to the reported results of SOCS-1 (31). The interaction between p85 and SOCS-6 was detected again after insulin stimulation (Fig. 6a,  upper panel, lanes 2 and 6). However, the SOCS-6-associated p85 was not degraded with SOCS-6 ( Fig. 6a). After 2 h of chase, pulse-labeled SOCS-6 was completely degraded, but the labeled p85 was still detected at a similar level in the co-precipitates of anti-SOCS-6 ( Fig. 6a, lanes 4 and 8). To confirm these findings in CHO cells, cardiomyocytes isolated from SOCS-6 transgenics were treated with cycloheximide and then stimulated with insulin in the presence or absence of LLnL. An analysis of anti-p85 immunoprecipitates showed similar results for both the insulin-mediated interaction of SOCS-6 with p85 and the selective proteasome degradation of SOCS-6 (Fig.  6c). These results suggest that SOCS-6 functions to regulate the p85 monomer rather than to target p85 to the proteasome degradation pathway.
IRS-associated p85 Does Not Bind to SOCS-6 -Our data would suggest that the SOCS-6-p85 complex is not recruited into the insulin-signaling complex (Fig. 4b). Therefore, the p85-p110 complex, which does not interact with SOCS-6, could interact optimally with IRS. To examine whether the recruitment of the p85-p110 dimer to IRS is affected by SOCS-6, the SOCS-6-and IR-expressing CHO cells were pulse-labeled with [ 35 S]methionine and then stimulated with insulin during different chase periods. The cleared cell lysates were first precipitated with excess anti-SOCS-6 antibodies to deplete SOCS-6 and the SOCS-6-p85 complex. The depleted cell lysates were then precipitated with anti-pan-p85 antibodies. The results showed that anti-p85 co-precipitated p110 before and after insulin stimulation (Fig. 6b, lanes 1-8). The interaction between p85 and IRS was detected after insulin stimulation and maintained throughout the chase periods (Fig 4b, lanes 2 and  6). This interaction was stable and not affected by the degra-dation of SOCS-6. This result shows that SOCS-6 binds preferentially to the p85 monomer. Moreover, the p85-p110 complex, but not the SOCS-6-p85 complex, was recruited by IRS to the insulin-signaling complex. DISCUSSION A surprising finding of the current study was that SOCS-6 overexpression results in improvements in the rate at which a glucose load is cleared by insulin and in the potentiation of insulin stimulation of Akt/PKB. This finding indicated that the overexpression of SOCS-6 was having a positive effect on the PI 3-kinase-dependent signaling pathways in the cell. However, our studies show that this is clearly not a direct effect on the lipid kinase activity of class-Ia PI 3-kinase. In fact, we did not detect interaction with the catalytically active p85/p110 forms of PI 3-kinase, and SOCS-6 interacted only with the monomeric form of the class-Ia PI 3-kinase regulatory subunit. This find- After pulse, the cells were divided into two parts. The proteasome inhibitor LLnL in Me 2 SO was added into one part, and Me 2 SO (DMSO) was added into other part as control. Both were chased in normal medium with or without 1 M insulin (Ins) for different periods of time. After chase, the cells were lysed, and the lysates of each time point were divided into two; one was for immunoprecipitation (IP) with anti-SOCS-6 antibody (a), and the other was first precipitated with an excessive anti-SOCS-6 antibody to deplete SOCS-6 and SOCS-6 associated p85 and then re-precipitated with anti-pan p85 antibody (b). The precipitates were then resolved with 10% SDS-PAGE and visualized by autoradiography. To identify the p85, SOCS-6, p110, and IRS proteins, a total lysate of IR-and SOCS-6-expressing CHO cells stimulated by insulin was precipitated with either SOCS-6 or the p85 antibody. The precipitates were separated with 10% SDS-PAGE and subsequently blotted either with anti-SOCS-6, anti-pan p85, p110, or the IRS-1 antibody. The migrating position of these molecules on both gels was identical. SOCS-6, p85, p110, and the IRS were indicated. To confirm the results obtained from CHO cells, cardiomyocytes from wild type and transgenic mice were treated with cycloheximide for 6 h before stimulation with insulin. A similar pattern of SOCS-6 degradation was revealed by quantitative analysis of SOCS-6, IRS-1, and p85 from anti-p85 co-precipitates (c).
ing, at face value, seems to be a paradox, but it has recently become apparent that the p85 protein is more than just an regulatory subunit for PI 3-kinase and has functions that are independent of its association with the PI 3-kinase catalytic subunit (26,27,32). Some reports have suggested that, under normal circumstances, significant levels of the p85 monomer can exist in cells and that these levels can have a negative effect on insulin signaling (26,27,33). Indeed, when levels of either p85␣ or p85␤ are reduced in gene knock-out models, there is a significant improvement in glucose homeostasis similar to that observed in our SOCS-6 overexpression model (28 -30, 33). Our three findings that the interaction between monomeric p85 and SOCS-6 is insulin regulated, that insulin causes a rapid but transient increase in SOCS-6 mRNA, and that SOCS-6 is rapidly degraded suggest that SOCS-6 could provide a physiologically relevant mechanism for modulating the cellular effects of monomeric p85 proteins in insulin action. In turn, this possibility suggests that monomeric p85 is not necessarily an undesirable artifact of differences in the rate of p85 and p110 turnover but that monomeric p85 might represent a deliberate mechanism by which the cell is able to tightly control the activity of the PI 3-kinase axis in glucose homeostasis.
The finding that SOCS-6 is capable of interacting with IRS-2, IRS-4, the p85 subunit of PI 3-kinase (23), and the IR (16) indicates that SOCS-6 may have a role in regulating insulin signaling. Here, we show that the interaction of SOCS-6 with the p85 monomer is dependent on insulin stimulation. The mechanism by which this association occurs is via the SOCS-6 SH2 domain (23). An analysis of binding specificity to phosphopeptides showed that the SOCS-6 SH2 domain bound preferentially to motifs containing a valine in the Tyr(P) ϩ 1 position and a hydrophobic residue in the Tyr(P) ϩ 2 or Tyr(P) ϩ3 position (23). Several insulin-induced tyrosine phosphorylation sites have been reported on p85, including tyrosines 368, 580, and 607 (34,35). The latter two are particularly interesting, as they fit the binding consensus requirements for the SOCS-6 SH2 domains (23), are in the region of p85 involved in binding to p110, and would thus, presumably, be inaccessible in the dimeric p85-p110, which could explain specific binding to the monomeric form (36). Furthermore, there is clear evidence that structural changes or phosphorylation in this region regulate the function of PI 3-kinase (37)(38)(39). Tyrosine-phosphorylated p85 has been shown not to bind to insulin-stimulated signaling complexes (40), which is consistent with our finding that SOCS-6-associated p85 does not associate with IRS proteins. Whether the phosphorylations are the binding sites for SOCS-6 is yet to be clarified.
The mechanism by which SOCS-6 increases insulin signaling is not clear, but because excess free p85 is deleterious, it is possible that the binding provides a mechanism for reducing levels of p85 monomers. In support of this possibility, the SOCS box of SOCS-6 has been found to interact with both elongins B and C (23). The complex of elongin B/C could induce the activation of E3 ubiquitin ligase activity by forming a multiprotein complex with cullin2 or cullin5 and Rbx1 (1). Thus, the SOCSassociated proteins could be targeted for polyubiquitination and be subsequently degraded by proteasome. SOCS-6 was indeed rapidly degraded by proteasome after insulin stimulation, which is consistent with the findings of proteasome-mediated degradation of SOCS-1 (31). However, the degradation only occurred to SOCS-6 itself but not to the p85 subunit associated with it, indicating that the mechanism for SOCS-6 to regulate the function of p85 is based on the interaction of two molecules during insulin signaling. The resistance of p85 to proteasome degradation could be an essential feature that enables a stable pool of p85 monomers to be maintained, which is important for the stability and function of p110 (26). Moreover, rapid induction and quick degradation of SOCS-6 by insulin also suggest that regulation of the p85 function by SOCS-6 is transient and that it regulates the pools of p85 monomers and p85-p110 dimers in order to quickly balance the glucose level during and after insulin stimulation.
p85 is the regulatory subunit of PI 3-kinase and consists of different isoforms, including p85␣ and its splicing variants (p55␣ and p50␣) as well as p85␤ and p55 (41). p85␣ is the most dominant isoform and represents nearly 70% of the total regulatory subunit (27,29). Evidence from PI 3-kinase inhibitors and dominant negative mutants has clearly shown that PI 3-kinase activity is essential for the metabolic and mitogenic effects of insulin (41). It was therefore surprising that mice lacking p85␣ and its splicing variants or p85␤ show much improved sensitivity to insulin (28 -30, 33). Mice heterozygous for deletion of the p85␣ gene have reduced levels of p85␣ and its splicing variants. However these mice are protected from the diabetes that develops in mice heterozygous for both IR and IRS-1 deletions (33). An apparently paradoxical occurrence of hypoglycemia associated with the decreased activity of IRSassociated PI 3-kinase induced by insulin is, in fact, a common phenotype across all the p85 knock-out mice models (28 -30). Biochemical studies on cells derived from p85␣Ϫ/Ϫ showed a 70% reduction of regulatory subunits combined with a 50% decrease in PI 3-kinase activity (28,30). However, cells from p85␤Ϫ/Ϫ cells showed only a 25% reduction in regulatory subunits but with normal PI 3-kinase activity (29). These results suggest that Ͼ30% of the regulatory subunits exist as monomers in the normal cells. The preserved PI 3-kinase activity and increased Akt/PKB activation in the p85␤ knock-out model reveal that the excessive p85 monomers have a negative function in the regulation of insulin action (29). The mechanism for this negative function is still not clear. One of the possible mechanisms is that excessive p85 monomers block the recruitment of p85-p110 heterodimers to IRS. However, the increasing expression of p85 does not decrease the recruitment of the p85-p110 dimer to IRS nor reduce the PI 3-kinase activity induced by insulin but results in a decrease in Akt/PKB activity, suggesting that the negative regulation of insulin action by p85 monomers occurs downstream of PI 3-kinase (26). Compelling evidence from the study of PIP 3 levels in cells from the p85␣ knock-out confirms this hypothesis, as the p85␣Ϫ/Ϫfibroblasts have a higher level of PIP 3 than do wild type cells (27). The same result is also found from adipocytes derived from knock-outs of p85␣ (28). This may well be due to an observed reduction of lipid phosphatase PTEN activity in p85Ϫ/Ϫ cells (26). These results suggest that the negative signal of p85 monomer for reducing the level of PIP 3 and PIP 3 -dependent downstream signaling events could be essential for balancing glucose homeostasis after insulin activation is terminated. Therefore, it is inevitably important to control this negative signal during insulin action. Moreover, this negative regulation is independent of the insulin-activated PI 3-kinase pathway (26).
In support of a role for SOCS-6 as a regulator of insulin signaling, we find that SOCS-6 overexpression in mice resulted in significant improvements in both glucose clearance and insulin tolerance, indicating that SOCS-6 is positively regulating insulin action. However, overexpression of SOCS-6 did not increase the activity of PI 3-kinase and, in contrast, the PI 3-kinase activation induced by insulin was slightly decreased. The phosphorylation of IR and IRS-1 induced by insulin in SOCS-6-overexpressing cells was normal. The amount of the p85-p110-IRS complex was similar between SOCS-6-overexpressing and wild type cells, suggesting that the interaction of SOCS-6 with the p85 monomer does not affect the direct signaling events of an insulin-signaling complex but does effect downstream signaling by a mechanism independent of the regulation of PI 3-kinase by the complex. Indeed, although the increased insulin action was not accompanied by increased activation of PI 3-kinase, augmented activation of Akt/PKB was observed in SOCS-6-overexpressing cells. The similarities in phenotype between SOCS-6 transgenics and p85 knock-outs imply that the interaction of SOCS-6 with p85 is one of the mechanisms for inhibiting the negative signal of a p85 monomer and that SOCS-6 may be a dynamically regulated balancing mechanism for ensuring that insulin signaling proceeds even in the face of some degree of p85 monomer.
The induction of SOCS-6 by fetal calf serum suggests that insulin is not the only stimulus for the induction of SOCS-6. Therefore, SOCS-6 may be involved in different signaling pathways that are yet to be investigated. However, not only does insulin induce the expression of SOCS-6, but the binding of SOCS-6 to p85 is dependent on insulin, indicating that blocking the negative signal of p85 in insulin action is insulin-specific. SOCS-7, a close homologue of SOCS-6, has also been reported to interact with p85 (23). The overlapping function may explain the lack of a significant phenotype in SOCS-6 knock-out mice (23). Indeed, we found that SOCS-7 was also induced by insulin and that the kinetics of SOCS-7 expression in insulin-stimulated fibroblasts was similar to that of SOCS-6, indicating an overlapping function of SOCS-7 and SOCS-6 in the regulation of insulin signaling. These data propose that the modulation of SOCS-6 and/or SOCS-7 could be a potential therapeutic strategy for the treatment of type II diabetes and the diseases associated with insulin resistance