Insulin Resistance Due to Phosphorylation of Insulin Receptor Substrate-1 at Serine 302*

Inhibitory serine phosphorylation is a potential molecular mechanism for insulin resistance. We have developed a new variant of the yeast two-hybrid method, referred to as disruptive yeast tri-hybrid (Y3H), to identify inhibitory kinases and sites of phosphorylation in insulin receptors (IR) and IR substrates, IRS-1. Using IR and IRS-1 as bait and prey, respectively, and c-Jun NH2-terminal kinase (JNK1) as the disruptor, we now show that phosphorylation of IRS-1 Ser-307, a previously identified site, is necessary but not sufficient for JNK1-mediated disruption of IR/IRS-1 binding. We further identify a new phosphorylation site, Ser-302, and show that this too is necessary for JNK1-mediated disruption. Seven additional kinases potentially linked to insulin resistance similarly block IR/IRS-1 binding in the disruptive Y3H, but through distinct Ser-302- and Ser-307-independent mechanisms. Phosphospecific antibodies that recognize sequences surrounding Ser(P)-302 or Ser(P)-307 were used to determine whether the sites were phosphorylated under relevant conditions. Phosphorylation was promoted at both sites in Fao hepatoma cells by reagents known to promote Ser/Thr phosphorylation, including the phorbol ester phorbol 12-myristate 13-acetate, anisomycin, calyculin A, and insulin. The antibodies further showed that Ser(P)-302 and Ser(P)-307 are increased in animal models of obesity and insulin resistance, including genetically obese ob/ob mice, diet-induced obesity, and upon induction of hyperinsulinemia. These findings demonstrate that phosphorylation at both Ser-302 and Ser-307 is necessary for JNK1-mediated inhibition of the IR/IRS-1 interaction and that Ser-302 and Ser-307 are phosphorylated in parallel in cultured cells and in vivo under conditions that lead to insulin resistance.

Insulin resistance is the condition in which target tissues fail to respond appropriately to circulating insulin. Although genetics may play a role in the pathogenesis of type 2 diabetes, it has become increasingly clear that acquired, non-genetic causes of insulin resistance represent a critical link between the rapidly growing national and worldwide epidemics in obesity and type 2 diabetes (1)(2)(3)(4)(5). Obesity, fatty diet, and sedentary lifestyle directly promote insulin resistance, and exercise and weight loss reverse it. Obesity and insulin resistance are also associated with and exacerbate hypertension and hyperlipidemia in addition to predisposing to the development of type 2 diabetes. This constellation of conditions, referred to collectively as either the metabolic or dysmetabolic syndrome, represents a third interrelated epidemic with a prevalence of ϳ24% of adults in the United States (6). Individuals with the metabolic syndrome have a seriously increased risk of developing atherosclerotic cardiovascular disease. Elucidating the molecular pathways that connect obesity to pathogenesis of insulin resistance clearly has great public health importance.
Although IR is a tyrosine kinase, insulin also stimulates Ser/Thr phosphorylation of numerous signaling enzymes and other proteins (21,22). Many are Ser/Thr kinases involved in kinase cascades. Those mediating some of the insulin cellular actions include Raf, MEK (mitogen-activated protein kinase/ extracellular signal-regulated kinase kinase), MAPK, p90RSK, phosphatidylinositol 3-kinase, phosphoinositide-dependent protein kinase 1, protein kinase B/AKT, mTOR, p70 S6 kinase, GSK3␤, PKC␤1, PKC, and PKC. Overexpression of many of these kinases paradoxically inhibits insulin signaling as opposed to activating it, suggesting that the same kinases that mediate insulin signaling might also play roles in negative feedback of it (23). In fact, IR and IRSs are themselves Ser/ Thr-phosphorylated in response to insulin, providing a potential mechanism for negative feedback.
Basal levels of IRS-1 Ser/Thr phosphorylation are increased in cells under various conditions, leading to observable shifts in electrophoretic mobility (15, 24 -27). The magnitudes of the shifts demonstrate that multiple sites are phosphorylated. In * These studies were funded by National Institutes of Health Grants DK63225 (to J. L.), DK43123 (to S. E. S.), DK45943 (to S. E. S.), and DK36836 (Joslin Diabetes Center, Diabetes Endocrinology Research Center), National Institutes of Health Fellowship DK09393 (to E. D. W.), and by the Helen and Morton Adler Chair (to S. E. S.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18  fact, IRS-1 contains 232 serines and threonines, nearly 19% of its 1231 residues, providing great potential for multisite phosphorylation. The rapid accumulation of "hyper-phosphorylated" IRS-1 upon treatment with the Ser/Thr phosphatase inhibitors okadaic acid or calyculin A further indicates that it is constantly being Ser/Thr-phosphorylated. Normally this is countered by phosphatases like PP2A and PP1, such that under steady-state conditions there should be a given "responsiveness" to the system. The balance might be shifted by activating Ser/Thr kinases that phosphorylate IRS or by inhibiting a relevant phosphatase. The net result in these cases would be the same; that is, Ser/Thr phosphorylation of IRS-1, diminished insulin signaling, and the development of insulin resistance. Conversely, insulin signaling should be sensitized either by inhibiting the relevant Ser/Thr kinase(s) or activating the appropriate phosphatase(s).

MATERIALS AND METHODS
Yeast Two-hybrid-Matchmaker LexA two-hybrid reagents were purchased from Clontech. Saccharomyces cerevisiae strain EGY48 (Mat␣ trp1, his3, ura3, 6LexAop-LEU2, LYS2), transformed with p8op-lacZ (kindly provided by T. A. Gustafson, Metabolex), was used as the host for interaction studies. EGY48/p8op-lacZ was sequentially transformed with plasmid constructs using polyethylene glycol/lithium acetate according to manufacturer's protocols. To determine protein-protein interactions as a function of leucine biosynthesis (LEU2), transformants were grown on synthetic dextrose agar plates for 3 days at 30°C. Four independent colonies were streaked on synthetic dextrose glucose agar plates, grown overnight, replica-plated on synthetic galactose/raffinose agar plates, and regrown for 5 days at 30°C to induce expression of B42 fusion proteins.
Disruptive Yeast Tri-hybrid-To create the disruptive yeast tri-hybrid assay, we modified the existing LexA yeast two-hybrid method. S. cerevisiae EGY48 cells were sequentially transformed as above with pLexA expressing human insulin receptor kinase (residues 940 -1343) as bait, pB42AD expressing various human IRS-1 constructs as prey, and in most cases JNK1␣1 in the third pDis plasmid. pDis was derived from the p426:Gal1 plasmid (ATCC) by replacing its multiple cloning region with another having a nuclear localization signal under Gal1 promoter control. In selected experiments alternative kinases were expressed in the pDis plasmid, including GSK3␤(S9A), IKK␣, IKK␤(S177E/S181E), p38 MAPK, PKA, PKC␣, PKC␤2, PKC⑀, and PKC. Transformants were grown on the appropriate synthetic dextrose plates for 3 days at 30°C. Four independent colonies were streaked on synthetic dextrose plates, incubated overnight, and replicaplated on GR plates. The plates were immediately replica-cleaned, incubated overnight, replica-cleaned, and incubated at 30°C for 5 days to induce expression of pDis and B42 fusion proteins.
In Vivo Animal Studies-Fourteen week-old ob/ob (Lep ob/ob ) mice and congenic Lep ϩ/ϩ controls were sacrificed after an overnight fast. For the "diet-induced obesity" study, 8 week-old C57BL/6 (Jackson Laboratories) mice were fed a high fat diet (Research Diet D12451, 45% of calories from fat) for 8 weeks; controls were fed regular chow that derives 17% of calories from fat. After an overnight fast the 16-week-old mice were sacrificed. To determine the effects of acute, high dose insulin, chow-fed 12-week-old C57BL/6 mice were fasted overnight, injected intraperitoneally with 1 units/kg of insulin, and sacrificed after 10 min. Harvested livers, stored until use in liquid N 2 , were pulverized and homogenized with a Polytron for 30 s in lysis buffer. Cleared lysates were used for immunoprecipitations and Western blotting.
Antibody Preparation and Use-Phosphospecific antibodies against IRS-1 Ser(P)-302 were generated in rabbits. Phosphopeptide RRSRTEpSITATSP (p indicates phosphorylated serine) was coupled to keyhole limpet hemocyanin for use as antigen. Rabbit antisera were passed first over immobilized RRSRTESITATSP to remove antibodies that recognized the unphosphorylated sequence. Phosphospecific antibodies were affinity-isolated by passing the precleared sera over immobilized RRSRTEpSITATSP followed by a low pH elution (62). Phosphospecific IRS-1 Ser(P)-307 antibody was either purchased from Cell Signaling or kindly provided by Morris White (Joslin Diabetes Center), anti-insulin receptor and IRS-1 antibodies were described previously (42,43), anti-phosphotyrosine (4G10) antibody was from Upstate Biotechnology Inc., and horseradish peroxidase-conjugated goat anti-rabbit or goat anti-mouse IgG (Amersham Biosciences) was used for Western blotting.
Proteins were immunoprecipitated from cell or tissue lysates for 2 h at 4°C using immobilized anti-IRS-1 and anti-IR antibodies (protein A-Sepharose beads, Amersham Biosciences). After extensive washing, proteins were eluted with Laemmli sample buffer, separated by SDS-PAGE, and transferred to nitrocellulose membranes (Bio-Rad). Blots were probed with the indicated antibodies, and proteins were detected by chemiluminescence (Pierce). For transfection experiments, FLAG-tagged IRS-1 was immunoprecipitated using anti-FLAG antibody (Stratagene).

Identification of Ser-302 Is an Inhibitory Phosphorylation
Site in IRS-1-Gustafson (63) first used a conventional LexA yeast two-hybrid (Y2H) assay to identify direct interactions between IR and IRS-1. We modified this assay to develop a method to identify proteins that block IR-IRS interactions and potential molecular mechanisms for disruption. In our "disruptive yeast tri-hybrid" (Y3H) assay, bait and prey are expressed as usual using pLexA and pB42AD plasmids, respectively, but we removed p8op-lacZ and replaced it with a distinct third plasmid (termed pDis or pDisruptor) for the expression of proteins that might modulate protein-protein interactions.
IR and IRS-1 bind one another under Y2H conditions and in the disruptive Y3H assay with an empty disruptive (pDis) plasmid (Fig. 1B). When JNK1 was expressed in the third plasmid the interaction was disrupted (Fig. 1, A and B) (18). Kinase activity was required for disruption, as kinase deficient (KD) JNK1 having T183A/Y185F substitutions within its kinase activation loop did not disrupt IR/IRS-1 binding. Binding of IR to IRS-1 was retained after removal of the IRS-1 PH domain (110 -1242) or the PH domain plus interdomain (150 -1242), and phosphorylation by JNK1 disrupted these interactions (Fig. 1A). These data are consistent with the suggestion that phosphorylation of Ser-307, a site C-terminal to the PTB domain, is involved in JNK1-mediated disruption (17,18). Also consistent with these previous findings, a region in the C terminus of IRS-1 is required for JNK1-mediated disruption. In the disruptive Y3H assay shorter segments of IRS-1 such as one encompassing its PH and PTB domains (1-330) or its PTB domain alone (140 -260) bound IR, but binding was not disrupted by JNK1 in either of these cases (Fig. 1A). Aguirre and White (17,18) showed that JNK1 disruption requires binding at a C-terminal JNK binding domain (JBD) (residues 857-863) in addition to phosphorylation at Ser-307.
We next substituted Ser-307 with glutamic acid instead of alanine, as glutamate has been found to mimic phosphoserine in other proteins. Unlike S307A-substituted IRS-1, which was resistant to JNK1-mediated disruption, S307E-substituted IRS-1 behaved like the wild-type protein. IRS-1 S307E interacted with IR in the absence of disruptive protein, and binding was blocked by JNK1 phosphorylation. Because glutamate cannot be phosphorylated yet acts as though it were, these findings demonstrated that glutamate is indeed a phosphorylation mimic in this system. Much more importantly, however, these findings indicated that although Ser-307 phosphorylation is necessary for disruption, it is not sufficient. If phosphorylation at Ser-307 were sufficient, then binding would have been disrupted in the absence of kinase or the presence of catalytically inactive kinase, which clearly wasn't the case (Fig. 1, A and B). These findings prompted a search for additional phosphorylation site(s) potentially involved in JNK1 disruption.
We looked first at the interdomain region between the PH and PTB domains because it contains a discrete number of serines. We reasoned that phosphorylation(s) within the Ser-135, Ser-137, Ser-139, and Ser-140 cluster might inhibit IR/ IRS-1 interactions by interacting with the PH or PTB domain. Each serine was substituted with alanine, independently and together. The substituted proteins behaved exactly like wildtype IRS-1 under Y3H conditions (Fig. 1A), clearly indicating that these sites are not involved in JNK1-mediated inhibition. We next hypothesized that the site(s) neighboring Ser-307 might be phosphorylated, and since Ser-302 is closest, this was mutated. The S302A substitution rendered IRS-1 resistant to JNK-1 inhibition (Fig. 1, A and B), exactly as had been seen previously with Ser-307. Perhaps not surprisingly, given that both of the singly substituted S302A or S307A proteins were resistant to JNK1-mediated inhibition, doubly substituted IRS-1 S302/307A was similarly resistant.
We proceeded to ask whether glutamate mimicked phosphoserine in this case as well. Analogous to the situation for IRS-1 S307E, the S302E-substituted protein behaved like wild-type IRS-1; it interacted with IR in the absence of disruptive protein, and binding was blocked by JNK1 phosphorylation (Fig. 1,  A and B). These findings indicated that phosphorylation at both Ser-302 and Ser-307 is necessary for disruption by JNK1 but that phosphorylation at either site alone is insufficient. Doubly substituted IRS-1 S302E/S307E tested whether the two sites together are sufficient for disruption. If phosphorylation at Ser-302 and Ser-307 were sufficient, then binding would have been disrupted in the absence of kinase or the presence of catalytically inactive kinase. This was not the case (Fig. 1, A  and B), suggesting that in addition to Ser-302 and Ser-307, JNK1-mediated disruption of IR/IRS-1 binding requires phosphorylation at additional site(s). We concluded that disruption by JNK1 requires a cluster of three or more phosphorylated serine residues. Preventing phosphorylation at one or both sites (S302A, S307A, or S302/307A) renders IRS-1 resistant to JNK1-mediated inhibition.
Selectivity of the Ser-302 or Ser-307 Sites for JNK1-mediated Disruption-We have subcloned several additional kinases into pDis to ask (a) whether they disrupt IR/IRS-1 binding and (b), if they disrupt, whether Ser-302 or Ser-307 is involved. We examined the effects of GSK3␤, IKK␣, IKK␤, p38 MAPK, PKA, PKC␣, PKC␤I, PKC␤2, PKC⑀, and PKC, as each of these kinases has been proposed to be a potential mediator of insulin resistance (15, 17, 18, 20, 28 -30, 34 -42, 59). IKK␣ and IKK␤ do not block IR/IRS-1 binding in the disruptive Y3H assay. Negative findings with the IKKs have been consistent in our hands under a variety of cellular and biochemical conditions (data not shown) (15), despite a claim to the contrary (59). By contrast, GSK3␤, p38 MAPK, PKA, PKC␣, PKC␤I, PKC␤2, PKC⑀, and PKC all disrupt IR/IRS-1 binding like JNK1 (Table  I). The substituted forms of IRS-1, S302A, S307A, or S302/ 307A, were used to determine whether these sites were in- volved in disruption by each of the kinases. Each substitution was tested with every kinase, and each failed to rescue the interaction, with the exception of JNK1 as the kinase (Table I). These findings demonstrate that whereas Ser-302 and Ser-307 are involved in disruption by JNK1, neither site, alone or in combination, is involved in disruption by GSK3␤, p38 MAPK, PKA, PKC␣, PKC␤I, PKC␤2, PKC⑀, and PKC.
Ser-302 Phosphorylation in Cultured Cells-Phosphospecific antibodies were prepared to determine whether Ser-302 is phosphorylated in cultured cells or in vivo. The antibodies were prepared against a phosphopeptide sequence corresponding to the region of IRS-1 surrounding Ser-302 (RRSRTEpSITATSP) that is identical in rats (Ser-302), mice (Ser-302), and humans (Ser-307) (45)(46)(47). The affinity-purified antibodies were used for Western blotting experiments. Chinese hamster ovary cells transfected with WT or mutated IRS-1 constructs were stimulated with insulin and anisomycin to test antibody efficacy and specificity. The FLAG-tagged IRS-1 proteins were immunoprecipitated with anti-FLAG antibodies and blotted with our new Ser(P)-302 antibodies or antibodies directed against p307, Tyr(P), or IRS-1 (Fig. 2). The Ser(P)-302 antibodies recognized WT IRS-1, indicating that IRS-1 is phosphorylated at Ser-302 in insulin-stimulated cells. The WT protein was recognized as well by anti-Ser(P)-307, anti-Tyr(P), and anti-IRS-1 antibodies. Expression of mutated forms of IRS-1 tested specificity. The anti-Ser(P)-302 antibodies did not recognize IRS-1 S302A, but anti-Ser(P)-307, anti-Tyr(P), and anti-IRS-1 antibodies did recognize IRS-1 S302A. In contrast, IRS-1 S307A was recognized by anti-Ser(P)-302(IRS-1) but not anti-Ser(P)-307(IRS-1) antibodies (Fig. 2). These findings verify that the two antibodies are equivalently specific for the sites they were raised against.
Interestingly, under the conditions used in this assay phosphorylation of IRS-1 S307A was reduced at Ser-302. Under more intensive phosphorylation conditions, for example in cells treated simultaneously with insulin and calyculin A, which promote Ser/Thr phosphorylation and inhibit Ser/Thr dephosphorylation, respectively, there were equivalent amounts of phosphorylation at Ser-302 in WT IRS-1 and IRS-1 S307A. These findings suggest that Ser-307 phosphorylation might be required for efficient Ser-302 phosphorylation. Ordered phosphorylation or "priming" mechanisms are well established in the kinase field but have not been described previously for IRS-1.
Anisomycin, an activator of stress kinases including JNK1, had little or no effect in stimulating phosphorylation further than insulin at either Ser-302 or Ser-307 (Fig. 2). Nevertheless, anisomycin led to a ϳ30% reduction of WT IRS-1 tyrosine phosphorylation. Anisomycin similarly decreased the tyrosine phosphorylation of singly substituted IRS-1 S302A and IRS-1 S307A and doubly substituted IRS-1 S302/307A by 30 -50%. These findings suggested that anisomycin-mediated inhibition involves phosphorylation sites in addition to Ser-302 and Ser-307 and probably kinases other than JNK1, as JNK1-mediated inhibition of IR/IRS-1 binding was reversed by these identical mutations in disruptive Y3H assays (Fig. 1).
Calyculin A is an inhibitor of Ser/Thr phosphatases, which increases Ser/Thr phosphorylation in cells through direct mechanisms as well as through the activation of numerous kinases. Treatment of cells with calyculin A induces a large shift in the electrophoretic mobility of IRS-1, apparently due to the phosphorylation of many Ser/Thr residues (Fig. 3A) (15). Calyculin A stimulated phosphorylation at both Ser-302 and Ser-307, as evidenced by increased intensity of the bands recognized by the sequence-specific anti-Ser(P)-302(IRS-1) and anti-Ser(P)-307(IRS-1) antibodies. The S302A and S307A substitutions abrogated recognition by the anti-Ser(P)-302(IRS-1) and anti-Ser(P)-307(IRS-1) antibodies, respectively, although the substitutions either alone or in combination had no appreciable effect on the magnitude of the shift in electrophoretic mobility. These results demonstrate that although phosphorylation of Ser-302 and Ser-307 is promoted by calyculin A, many additional Ser/Thr sites in IRS-1 are similarly affected.
Cells were subsequently treated with insulin to assess the effects of calyculin A on insulin-stimulated IRS-1 Tyr-phosphorylation (Fig. 3B). Not surprisingly, given the large number of Ser/Thr phosphorylations involved, calyculin A effectively abolished Tyr phosphorylation of WT IRS-1. Neither S302A nor contains an activating S9A substitution. b IKK␤ contains activating S177E/S181E substitutions. S307A alone or in combination noticeably rescued insulin-stimulated Tyr phosphorylation. The findings demonstrate that calyculin A promotes inhibitory phosphorylations at a Ser/Thr site or sites in addition to IRS-1 Ser-302 and Ser-307.

FIG. 2. Antibody specificity and insulin/anisomycin-induced phosphorylation. A, CHO-IR cells expressed FLAG
Ser-302 Phosphorylation of Endogenous IRS-1 in Cultured Cells-Having established the specificity of the antibodies, we went on to ask whether endogenous IRS-1 in cultured cells was phosphorylated at Ser-302 under conditions known to promote Ser/Thr phosphorylation (33)(34)(35)(36)(37). Fao (rat) hepatoma cells were pretreated with PMA (a phorbol ester that activates PKC isoforms), anisomycin, calyculin A (an inhibitor of Ser/Thr phosphatases), or insulin (overnight). The reagents were removed, and cells were stimulated acutely with insulin. PMA and calyculin A induced Ser/Thr phosphorylation of the insulin receptor, as judged by small shifts in electrophoretic mobility (Fig.  4A). All four conditions produced shifts in the electrophoretic mobility of IRS-1, consistent with even greater increases in Ser/Thr phosphorylation (Fig. 4B). Insulin-stimulated tyrosine phosphorylation of IRS-1 was decreased after each treatment (Fig. 4, B and C), consistent with Ser/Thr phosphorylation inhibiting tyrosine phosphorylation and downstream signaling. Calyculin A produced the largest shift in electrophoretic mobility of IRS-1, presumably due to the greatest increase in Ser/Thr phosphorylation, and essentially abolished tyrosine phosphorylation of IRS-1 (Fig. 4C).
Blots with the phosphospecific antibodies showed that each of these conditions promoted phosphorylation at both Ser-302 and Ser-307. Levels of Ser/Thr phosphorylation promoted by PMA, anisomycin, or calyculin A were similar in the absence or presence of insulin stimulation. Insulin itself led to small but definite increases as well. These results indicated that phosphorylation at Ser-302 and Ser-307 occurs in parallel under a wide range of conditions.
Ser-302 Phosphorylation in Genetically Obese Mice-Following the goals of the laboratory to identify in vivo mechanisms of insulin resistance, we next asked whether Ser-302 was phosphorylated in animal models of insulin resistance. Livers were isolated from ob/ob (Lep ob/ob ) mice, a genetic model in which a defect in leptin structure leads to loss of function. Lep ob/ob mice are hyperphagic, profoundly obese, insulin-resistant, and hyperglycemic. Immunoprecipitated IRS-1 was analyzed by Western blotting with Ser(P)-302 and Ser(P)-307 antibodies. Both of the antibodies consistently identified increased IRS-1 Ser phosphorylation in Lep ob/ob mice relative to congenic Lep ϩ/ϩ (C57Bl/6) controls (Fig. 5). Levels of Ser(P)-302 were 5.80-fold (Ϯ1.50) greater in Lep ob/ob mice relative to Lep ϩ/ϩ controls (Ϯ0.41, p ϭ 0.036); comparable levels of Ser(P)-307 were 3.06fold (Ϯ0.47) greater (Ϯ0.31, p ϭ 0.022). Similar increases suggest that in Lep ob/ob mice, phosphorylation at these two sites occurs in parallel.
Ser-302 Phosphorylation in Hyperinsulinemia-Hyperinsulinemia represents a third model in which Ser/Thr phosphorylation might mediate insulin resistance. To test whether Ser-302 and Ser-307 are phosphorylated during acute hyperinsulinemia, chow-fed C57Bl/6 mice were injected intraperitoneally with a single, large 1.0-unit/kg dose of insulin. Tissues were harvested within 10 min before the development of severe hypoglycemia, and liver IRS-1 was analyzed for Ser-302 and Ser-307 phosphorylation. Acute hyperinsulinemia increased Ser(P)-302 by 2.3-fold (Ϯ0.47) relative to saline-injected controls (Ϯ0.19, p ϭ 0.04). This maneuver similarly increased Ser(P)-307 by 1.8-fold (Ϯ0.31) relative to saline-injected controls (Ϯ0.03, p ϭ 0.09) (Fig. 7). We conclude that in vivo Ser/Thr phosphorylation is promoted at both sites by acute hyperinsulinemia. We were, thus, able to see consistent increases in Ser-302 and Ser-307 phosphorylation in three diverse models of murine insulin resistance, obesity, high fat diet, and hyperinsulinemia, suggesting that Ser-302 and Ser-307 might play important roles in related human conditions. DISCUSSION Deletion of IRS-1 in mice causes insulin resistance and diabetes, consistent with a relevant role for this adapter protein in insulin signaling (66 -69). These findings clearly demonstrate that alterations in IRS-1 amount can have serious consequences. Although controversial, the G972R polymorphism in the human irs1 gene has been correlated with both insulin resistance and insulin secretory defects. Most relevant to this study, although patients with type 2 diabetes have near normal amounts of the protein, functional deficiencies in IRS-1 have been identified (70). In particular, tyrosine phosphorylation appears to be reduced (71)(72)(73)(74). Ser/Thr phosphorylation has been increasingly recognized as a negative counterbalance to positive IRS signaling through tyrosine phosphorylation (23). Thus Ser/Thr phosphorylation could be a generalized mechanism for insulin resistance. From a molecular perspective it has been difficult, however, to identify discrete sites that are both phosphorylated in vivo and when phosphorylated, have relevant functional consequences.
Most investigations in this area have relied on the overexpression of candidate kinases in cultured cells and analyses of sites of Ser/Thr phosphorylation through protein sequencing or mass spectrometry. However, overexpression carries with it the potential for spurious phosphorylations outside of either A, fasting C57BL/6 mice (n ϭ 4, each group) were injected intraperitoneally with insulin (Ins, 1 units/kg) or carrier, and after 10 min, the mice were sacrificed, and tissues were harvested. Proteins were immunoprecipitated (IP) from liver lysates with anti-IRS-1 antibody, and separated proteins were immunoblotted (IB) with anti-Ser(P)-302 (pS302), anti-Ser(P)-307 (pS307), or anti-IRS-1 antibodies. B, combined results for four separate experiments (mean Ϯ S.E.; *, p Ͻ 0.05) were plotted relative to controls. usual sites or stoichiometries found physiologically. This could be due to greater than normal concentrations of the transfected candidate kinase or even more insidious in its ability to mislead, the transfected kinase may activate other kinases or inhibit phosphatases that are responsible for the detected phosphorylation. Virtually all of the candidate "insulin resistance" kinases that have been identified function in signaling cascades that involve other kinases and phosphatases. As an alternative approach that circumvents some of these problems, we developed the disruptive Y3H method.
Ser-307 has attracted significant attention as a targeted site for phosphorylation in cultured cells and in vivo (20, 42, 51, 54, 54 -56, 59, 61). In disruptive Y3H experiments phosphorylation of Ser-307 blocks IR/IRS-1 binding (Fig. 1) (18), providing a consistent and coherent mechanism that spans the biological and biochemical findings. We have now used disruptive Y3H to identify a second site in IRS-1 that is phosphorylated in cells and in vivo under equivalent conditions of insulin resistance. Both of these sites were identified as JNK1 targets. Previous studies had shown that Ser-307 was a necessary JNK1-mediated disruption (18), and we now show that Ser-302 is similarly necessary. A host of other kinases disrupt IR/IRS-1 binding in Y3H assays (Table I), including GSK3␤, p38 MAPK, PKA, PKC␣, PKC␤2, PKC⑀, and PKC, yet their molecular mechanisms are distinct as disruption by these other kinases definitely does not involve either Ser-302 or Ser-307. Several kinases associated with inflammatory processes like JNK1 do not disrupt IR/IRS-1 interactions, including the NF-B-activating kinases IKK␣ and IKK␤.
The strengths of our findings lie in showing unequivocally that JNK-mediated phosphorylation at either Ser-302 or Ser-307 (a) abrogates IR/IRS-1 binding and (b) that this occurs in vivo in three distinct animal models of insulin resistance; genetically obese Lep ob/ob mice, diet-induced obesity, and hyperinsulinemia. Our findings are consistent with a role for JNK1 in insulin resistance, as JNK1 phosphorylates Ser-302 and Ser-307 under Y3H conditions, and these sites are phosphorylated in vivo in models of insulin resistance. Nine other Ser/Thr kinases that are potentially associated with insulin resistance do not target Ser-302 and Ser-307 in Y3H experiments (Table  I), making it highly unlikely that these kinases target Ser-302 and Ser-307 in vivo. Yet, with ϳ395 Ser/Thr kinases in the human and mouse genomes (75,76), it should not be concluded that JNK1 alone is responsible for either the in vivo phosphorylation of Ser-302 or Ser-307 or insulin resistance.
During the review of our manuscript, two additional studies have implicated IRS-1 Ser-302 as a potential phosphorylation site. Greene et al. (77) used PKC␦ to radioactively label IRS-1 in vitro and a phosphopeptide-mapping approach to identify phosphorylations at Ser-302, Ser-319, and Ser-570 (in their manuscript they used the equivalent human IRS-1 numbering: Ser-307, Ser-324, and Ser-575). Under the conditions of their assays, PKC␦ did not appreciably phosphorylate Ser-307. Giraud et al. (78) used a phosphospecific antibody/candidate phosphorylation site approach to identify Ser-302. They reported that activation of mTOR signaling induced Ser-302 phosphorylation and that this positively influenced signaling through the IR/IRS-1 axis. The three studies identified Ser-302 using independent methods. Our data and the complementary findings of Greene et al. (77) clearly show that Ser-302 phosphorylation disrupts IR/IRS-1 signaling. Although we cannot speak to reasons that Giraud et al. (78) came to the opposite conclusion, this suggestion is clearly inconsistent with our findings. Phosphorylation at Ser-302 and Ser-307 is equivalently inhibitory in our studies, and it occurs at these two sites in parallel in yeast, in cultured cells, and in vivo under conditions of insulin resistance.
One of our more interesting and provocative findings relates to potential mechanisms of IR/IRS-1 disruption. Although we clearly show that JNK1 phosphorylates IRS-1 at Ser-302 and Ser-307 and both are necessary for disruption, these two phosphorylations either alone or in combination are not sufficient for disruption (Fig. 1). This means that phosphorylation of Ser-302 and Ser-307 and additional site or sites are required. In other words, a cluster of phosphorylated serines and possibly threonines mediates disruption. Although we continue to identify additional sites, the notion of "clustering" lends itself to considering new and testable hypotheses for potential mechanisms of disruption and, by extension, potential mechanisms for insulin resistance.
The disruptive Y3H approach provides a general method for analyzing macromolecular complexes in addition to identifying potential mechanisms of insulin resistance. Related methods have been published, including Vidal's forward and reverse YNH approaches (79), but we were unable to apply these methods to the insulin resistance problem. In this study we specifically used our disruptive Y3H to investigate mechanisms for disruption of a protein-protein interaction. A similar Y3H approach is equally useful for studying the formation of ternary and higher order complexes. As one example, we have used a variant of the approach, termed "bridging" Y3H, to analyze quaternary complexes between JAKs and APS or SH2-B proteins (80). 3 These methods appear to be both robust and of general utility.
In summary, we have developed a new approach for disrupting protein-protein interactions and used the method to analyze mechanisms for inhibiting IR/IRS-1 interactions. We identified Ser-302 as a new phosphorylation site in IRS-1 and showed that Ser-302 phosphorylation blocks IR/IRS-1 binding. Disruptive Y3H was further used to investigate and identify clustering as a potential mechanism for JNK1 mediated disruption. Phosphospecific antibodies were developed and used to show that Ser-302 is phosphorylated in cells and in vivo in a variety of models of obesity, insulin resistance, and type 2 diabetes. These approaches should have general utility in analyzing potential mechanisms of insulin resistance by identifying other disruptive kinases and additional, functionally relevant phosphorylation sites. This should help to provide a more complete understanding of the molecular basis of insulin resistance.