Regulation of Insulin Receptor Substrate 1 Pleckstrin Homology Domain by Protein Kinase C: Role of Serine 24 Phosphorylation

Abstract

Phosphorylation of insulin receptor substrate (IRS) proteins on serine residues is an important post-translational modification that is linked to insulin resistance. Several phosphoserine sites on IRS1 have been identified; the majority are located proximal to the phosphotryosine-binding domain or near key receptor tyrosine kinase substrate- and/or Src-homology 2 domain-binding sites. Here we report on the characterization of a serine phosphorylation site in the N-terminal pleckstrin homology (PH) domain of IRS1. Bioinformatic tools identify serine 24 (Ser24) as a putative substrate site for the protein kinase C (PKC) family of serine kinases. We demonstrate that this site is indeed a bona fide substrate for conventional PKC. In vivo, IRS-1 is also phosphorylated on Ser24 after phorbol 12-myristate 13-acetate treatment of cells, and isoform-selective inhibitor studies suggest the involvement of PKCα. By comparing the pharmacological characteristics of phorbol 12-myristate 13-acetate-stimulated Ser24 phosphorylation with phosphorylation at two other sites previously linked to PKC activity (Ser307 and Ser612), we show that PKCα is likely to be directly involved in Ser24 phosphorylation, but indirectly involved in Ser307 and Ser612 phosphorylation. Using Ser24Asp IRS-1 mutants to mimic the phosphorylated residue, we demonstrate that the phosphorylation status of Ser24 does play an important role in regulating phosphoinositide binding to, and the intracellular localization of, the IRS1-PH domain, which can ultimately impinge on insulin-stimulated glucose uptake. Hence we provide evidence that IRS1-PH domain function is important for normal insulin signaling and is regulated by serine phosphorylation in a manner that could contribute to insulin resistance.

Insulin resistance is a common pathological state wherein target tissues produce a less than normal response to circulating insulin. This plays a central role in the development of type 2 diabetes and is commonly associated with conditions such as obesity, polycystic ovary syndrome, sepsis, cachexia, cancer, hypertension, and cardiovascular disease. Understanding the molecular mechanisms involved in the regulation of both normal and deregulated insulin signaling may therefore be key to identifying novel diagnostic and therapeutic strategies.

Uncoupling of insulin signaling can occur at numerous points along the pathway and by numerous mechanisms such as decreased expression and/or activity of key signaling components (1). In particular, insulin receptor substrates (IRS) have been implicated as major targets in insulin-resistant states (2). The IRS family members are defined by a characteristic tandem arrangement of N-terminal pleckstrin homology (PH) and phosphotyrosine-binding (PTB) domains. Both these domains are well conserved, particularly between IRS1 and IRS2, and are required for recruitment to, and association with, the insulin receptor (IR) (3, 4). The remainder of the proteins are poorly conserved across family members but do contain key receptor tyrosine kinase phosphorylation- and Src-homology (SH2) domain-binding sites. IRS family members play a pivotal role in transmitting insulin signals and are absolutely required for normal insulin-stimulated glucose uptake. Indeed, mice lacking either IRS1 or IRS2 exhibit peripheral insulin resistance and reduced neonatal growth (5). Despite the similarities between IRS1 and IRS2, some functional differences have also been reported. Whereas IRS2 is crucial for β-cell growth and function, IRS1 is important in regulating metabolism in muscle and adipose tissue (5). Reductions in IRS1 protein levels and/or active tyrosine-phosphorylated IRS1 in these tissues are well documented in models of insulin resistance. Inducers of insulin resistance, such as FFA and TNFα, can also cause post-translational changes in IRS1, as can activators of serine kinases and/or inhibitors of serine phosphatases (1, 5, 6). Indeed, serine phosphorylation of IRS1 is currently the best-substantiated post-translational modification of IRS proteins, in addition to tyrosine phosphorylation.

Many serine kinases have now been implicated in mediating serine phosphorylation of IRS1 either directly or indirectly. These include casein kinase II, protein kinase C (PKC)α, PKCβI/II, PKCδ, PKCε, PKCθ, PKCζ, mitogen-activated Erk kinase (MEK), p38MAPK, cJun NH₂-terminal kinase (JNK), cRafK (Raf-1 kinase or v-raf-leukemia viral oncogene 1), PI3K (phosphatidylinositol 3-kinase), AKT/PKB (AKR mouse thymoma viral proto-oncogene/protein kinase B), glycogen synthase kinase (GSK)3β, mammalian target of rapamycin (mTOR), inhibitor of κB kinase β (IKKβ), salt-inducible kinase 2, AMP-activated protein kinase, and Rho-dependent kinase α (reviewed in Refs. 1, 6, and 7). Most of these kinases appear to negatively regulate insulin signaling, although a few may play a role in positive feedback regulation (8). There are, however, some serine kinases, such as casein kinase II, whose role in insulin signaling remains to be defined.

The mechanistic basis by which serine phosphorylation may uncouple proximal insulin signaling is beginning to be revealed by mapping the specific residues that are targeted by these serine kinases (1). Recent estimates suggest that IRS1 may contain more than 70 potential serine/threonine phosphorylation sites. However, only 16 have been demonstrated to become phosphorylated in response to agonist stimulation in vivo (9). Most of these are located proximal to/or downstream of the PTB domain. Phosphorylation of sites that lie close to the PTB domain (e.g. Ser307) can disrupt IR-IRS interactions, thereby reducing tyrosine phosphorylation of IRS1 (7, 9, 10). However, serine phosphorylation in this region can also disrupt IRS-protein tyrosine phosphatase interactions, leading to sustained tyrosine phosphorylation of IRS1 (8). Further downstream, residues such as Ser612, Ser632, Ser662, and Ser570 occur near Src-homology 2-binding motifs where they can prevent tyrosine phosphorylation and/or p85 recruitment (11, 12). Some serine phosphorylation sites have been reported to impinge on intracellular trafficking of IR and/or signaling complex formation (13, 14). Finally, serine phosphorylation may also play a role in recruiting chaperones such as 14–3-3 that target IRS1 molecules for proteosomal degradation, thereby reducing IR-IRS signaling (15–18).

Many of these mechanisms [especially those involving AKT, atypical PKC (aPKC), MAPK] are potentially targeted by insulin signaling itself and may represent physiologically relevant pathways involved in negative feedback. It is postulated that the perturbations induced during pathological states such as insulin resistance involve inappropriate activation of these inhibitory pathways. However, additional mechanisms may also exist that are not targeted by insulin but solely by agents that induce insulin resistance. Given the number of potential phosphoserine (pSer) sites it is also likely that multiple mechanisms are present by which serine phosphorylation of IRS1 can regulate insulin signaling. In this study, we report on the identification of a serine phosphorylation site located in the PH domain of IRS1. This site has recently been identified in an independent study and implicated as a target for IL receptor-associated kinase (19). Here we show that Ser24 is located in a substrate motif commonly targeted by a conventional PKC family of serine kinases and is indeed directly phosphorylated by PKCα in vitro and in vivo. Unlike other serine phosphorylation sites that are linked to PKC activity, Ser24 is not phosphorylated after chronic insulin stimulation and exhibits a pharmacological profile that is distinct from that of Ser307 and Ser612. We also demonstrate that the phosphorylation status of Ser24 plays an important role in regulating lipid binding and intracellular localization of the IRS1-PH domain. Ultimately this can also impair insulin-stimulated glucose uptake. Hence, serine phosphorylation of the IRS1 PH domain may represent a novel regulatory mechanism that could be important in insulin resistance.

RESULTS

Identification of a Putative Phosphorylation Site in the PH Domain of IRS1

The average occurrence of tyrosine, serine, and threonine residues in proteins is estimated to be about, 3%, 7%, and 6%, respectively (20). However, in IRS1 almost 15% of the amino acids present are serines, whereas tyrosines and threonines represent 3% and 5%. This is consistent with the major role of serine phosphorylation in regulating IRS1 function. To identify which residues could represent putative phosphorylation sites in IRS1, we analyzed protein sequences with two web-based bioinformatics tools, NetPhos 2.0 and Motif Scan. NetPhos allowed both sequence- and structure-based prediction of protein phosphorylation sites whereas Motif Scan predicted protein motifs that confer kinase-specific substrate sites. Screening for potential phosphorylation sites using NetPhos 2.0, revealed that all IRS molecules have many more putative serine phosphorylation sites than previously estimated: 113 in human (h) IRS1 [109 in rat IRS1 (rIRS1)]. This was substantially greater than the putative phosphothreonine (13 in hIRS1; 15 in rIRS1) and phosphotyrosine (20 in hIRS1; 21 in rIRS1) sites. However, the number of putative serine phosphorylation sites represents 62% (for hIRS1) of all serine residues present, a proportion that was similar for the putative phosphotyrosines (63% for hIRS1). In contrast, only 22% of all threonines in hIRS1 scored highly as putative phosphorylation sites. These comparative estimates appear largely conserved across the six species of IRS1 tested (data not shown). Using Motif Scan, a different proportion of putative phosphoserine and phosphotyrosine sites was predicted: this tool predicted eight phosphotyrosines, six phosphoserines, and one phosphothreonine (under high-stringency screen of hIRS1). However, this tool offered additional information: predicting specific kinases or family of kinases that may target specific phosphorylation sites located in known substrate motifs.

Despite the limitations of each bioinformatics tool, when used in combination and in conjunction with structural alignment analyses, they prove to be very powerful predictors of both potential phosphorylation sites as well as the kinases that may be involved. Indeed, motif scanning of all IRS1 proteins did highlight one serine residue, Ser24, as a putative PKC substrate site (Fig. 1A). This residue also scored highly (>0.5) as a putative phosphorylation site in the Net-Phos analysis. Sequence-structure alignment further revealed that Ser24 is highly conserved throughout all species of IRS1. However, it is present only in IRS1 and IRS3 and not in IRS2, IRS4, IRS5, or IRS6 (Fig. 1A). Chico, the Drosophila homolog for IRS1, also appeared to have a putative PKC phosphorylation site, albeit a threonine, in the same location. Serine 24 is located within the N-terminal PH domain of IRS1, and analysis of the crystal structure of IRS1 PH domain (db:1qqga) confirmed that it is located in the exposed variable loop 1 (VL1) region of this domain, a key area implicated in phosphoinositide binding and PH domain function (Fig. 1, A and B). Collectively, these data suggest that Ser24 is a candidate phosphorylation site for PKCs and may play a role in regulating PH domain function of IRS1.

Fig. 1.

Phosphorylation Potential, Conservation, and Structural Location of IRS-1 Serine 24

A, Sequence and structural alignment of N-terminal sequences of PH domains from all known IRS proteins and related phosphopotential scores. Sequence alignments were generated by ClustalW and compared with the known structure of hIRS1 (PDB:1qqga) using Fugue. Structural information for 1qqga is represented as follows: β-strand (blue), solvent accessible (lowercase), solvent inaccessible (uppercase), hydrogen bond to main-chain amide (bold), hydrogen bond to main-chain carbonyl (underline), and positive ϕ-torsion angle (italic). All numbering (except xIRS1) is according to Swissprot/Trembl. Sequence numbering for the xIRS1 sequence is taken from Ref. 57. Partial sequence availability is indicated by *. Arrows indicate key phospholipid-binding residues (26). Phosphopotential of each sequence was determined, using NetPhos and MotifScan (high stringency). Only one residue in the PH domain scored highly by both platforms (uppercase, bold, and red). This was Ser24 in rmhIRS1 (analogous to Ser14 in sIRS1, Ser27 in xIRS1, Ser44 in rIRS3, and Thr20 in dIRS1). Scores corresponding to each putative phospho residue are presented on the right. B, Ribbon diagrams of hIRS1 PH domain. β-Sheets are in green, α-helices in blue, and intervening coils or loops in yellow. The location of basic residues implicated in lipid binding (i.e. Lys 21, Lys 23, His 26, Arg 28, Lys 61, and Arg 62) are represented by red sticks and Ser24 by spheres. nd, Not detected.

PKC Phosphorylate IRS-1 on Ser24 in Vitro

To facilitate investigations of site-specific phosphorylation events on serine 24 of IRS1, an antibody was generated against the phosphopeptide [C]YLRKPKS(p)-MHKRFF (see Materials and Methods). After affinity purification and ELISA-based characterization, this antibody was used to test whether IRS1 Ser24 is a substrate site for PKC family members. In vitro kinase assays were first performed on recombinant peptides corresponding to the N-terminal PH domain (first 113 amino acids) of IRS1. Kinase reactions were performed in the presence and absence of ATP using recombinant enzymes representing different PKC family members [a conventional PKC (PKCα), a novel PKC (PKCδ), and an atypical PKC (PKCζ)]. The proteins were then separated by SDS-PAGE and immunoblotted with the anti-pSer24 antibody. A separate gel was run in parallel and stained for total protein. Figure 2A shows that incubation with all three active PKCs led to phosphorylation of IRS1^PH at Ser24. Although equal amounts of kinase and substrate were used, PKCα appeared consistently to produce the greatest amount of phosphorylation at Ser24, followed by PKCδ and then PKCζ. Similar reactions were performed to investigate whether casein kinase II would phosphorylate Ser24 in vitro. This enzyme has been reported to induce serine phosphorylation of IRS1 PH domain, albeit at Ser99 (21). Phosphorylation of Ser24 by casein kinase II was not observed (Fig. 2A), indicating that phosphorylation at this site is kinase selective. To confirm that the anti-pSer24 antibody was specific for this site, in vitro kinase assays were repeated in parallel with an IRS1^PH peptide harboring a S24A mutation (Fig. 2B). The anti-pSer24 antibody detected phosphorylation of wild-type (wt) PH domain but not of the S24A mutant peptide.

Fig. 2.

In Vitro Phosphorylation of IRS1 PH Domain on Ser24

A and B, wt (WT) or S24A (A) recombinant His6-IRS-PH domains were incubated with recombinant PKC isoforms or casein kinase II (CKII) as described in Materials and Methods. The kinase reactions were terminated and proteins separated by SDS PAGE followed by Western blot analysis with an antibody against pSer24. Parallel gels were also run and stained with either Coomassie (lower panels in A and B) or silver stain (SS in panel B) to confirm equal loading of kinase and recombinant IRS-PH domains. PKC-mediated anti-pSer24 immunoreactivity was ATP dependent (panel A) and Ser24 specific (panel B). C, In vitro kinase reactions were performed as described for panel A except that substrate used here was full-length rIRS1 (Myc-tagged) immunoprecipitated from serum-starved NIH/IR-rIRS1wt-MycHis6 cell lysates. Proteins were separated by SDS-PAGE followed by sequential Western blot analysis with antibodies against pSer24, pSer307, pSer612, or myc. The membranes were stripped clean of primary antibodies between each Western blot. Data are representative of at least three independent experiments. CKII, Casein kinase II.

To investigate whether PKCs phosphorylate Ser24 in the context of full-length IRS1, in vitro kinase assays were repeated on immunoprecipitates of IRS1 wt isolated from NIH/hIR/rIRS1wt cells. Figure 2C illustrates the resulting immunoblots wherein full-length IRS1 is also phosphorylated in an ATP-dependent manner on Ser24 by PKCα, -δ, and -ζ but not by casein kinase II. Again the magnitude of phosphorylation was greatest with PKCα and least with PKCζ. The same membrane was sequentially stripped and reprobed with antibodies to pSer307 and pSer612. This allowed us to investigate the action of the kinases on two additional serine phosphorylation sites in IRS1. Figure 2C shows that casein kinase II was indeed active and that all four enzymes were able to directly phosphorylate Ser307 and Ser612 in vitro. Interestingly, each site exhibited a different profile with respect to the relative levels of phosphorylation achieved by the panel of kinases. For example, PKCα most effectively mediated phosphorylation of Ser24, whereas casein kinase II was most effective at phosphorylating Ser612, and PKCδ was most effective for Ser307 phosphorylation. This suggests that site-specific differences may exist and can be reflected in vitro. Incubation with each of these serine kinases was also sufficient to induce a noticeable reduction in the electrophoretic mobility of rIRS1. This confirms that serine phosphorylation of IRS-1 is sufficient to account for the mobility shift reported for IRS-1 from insulin-resistant cells and tissues.

Phorbol 12-Myristate 13-Acetate (PMA), But Not C2 Ceramide or Chronic Insulin Treatment, Stimulates IRS1 Ser24 Phosphorylation in Vivo

To test whether insulin resistance-inducing agents could stimulate endogenous PKCs to phosphorylated IRS1 on Ser24 in vivo, NIH/hIR cells overexpressing either myc-rIRS1 wt or myc-rIRS1 S24A were stimulated with either PMA, C2 ceramide, or chronic insulin treatment. Anti-myc immunoprecipitates were then collected and analyzed for phospho-Ser24 immunoreactivity (Fig. 3). PMA induced detectable Ser24 phosphorylation of wt rIRS1 but not of the S24A mutant. PMA treatment also stimulated phosphorylation of IRS1 at Ser307 and Ser612 in both wt and S24A mutant. In contrast, neither ceramide nor insulin treatments stimulated detectable Ser24 phosphorylation in vivo but both induced phosphorylation at Ser307 and Ser612. Taken together, these data confirm that Ser24 can be phosphorylated in vivo. Furthermore, it is likely that PMA, C2 ceramide, and insulin activate different serine kinases and/or PKC isoforms, resulting in differential phosphorylation of Ser24 compared with Ser307 and Ser612.

Fig. 3.

In Vivo Phosphorylation of Full-Length rIRS1 on Serines 24, 307, and 612

NIH/hIR cells ectopically expressing either rIRS1wt or rIRS1 S24A were serum starved for 4 h and stimulated with either PMA (1 μm), C2 ceramide (100 μm), insulin (1 μm) or dimethylsulfoxide (vehicle control) for 1.5 h. Monolayers were washed with ice-cold PBS, and total protein was extracted and quantified. IRS1 immunoprecipitates were collected from 2.5 mg protein lysates and separated by SDS-PAGE. Proteins were transferred to polyvinylidine difluoride membranes and sequentially immunoblotted using antibodies against pSer24, pSer307, pSer612, or myc. The membranes were stripped clean of primary antibodies between each Western blot. Cntrl, Control; Ins, insulin; C2, C2 ceramide.

Conventional PKC Mediates PMA-Stimulated Phosphorylation of Ser24, Ser307, and Ser612

PMA activates diacylgycerol (DAG)-dependent PKC isoforms whereas C2 ceramide and chronic insulin treatment activate atypical PKCs (22, 23). In addition, numerous downstream serine kinases are also activated in response to all three stimuli, and many of these are implicated in insulin resistance. We sought to determine 1) which PKC family member(s) may be responsible for mediating PMA-stimulated Ser24 phosphorylation and 2) whether the same kinase was responsible for PMA-stimulated phosphorylation at Ser307 and Ser612. The three families of PKCs are distinguishable by their differential requirement for co-factors. Both conventional PKC (cPKC: α, β, γ) and novel PKCs (nPKC: δ, ε, η, θ) are DAG dependent but can be distinguished by their requirement for calcium. In contrast, atypical PKCs (aPKC: ζ, λ, μ) are calcium and DAG independent. We took advantage of these functional differences to characterize the PMA-stimulated phosphorylation of Ser24, Ser307, and Ser612. We reasoned that whereas 1,2-bis(o-aminophenoxy)ethane-N,N,N′,N′-tetraacetic acid tetra(acetoxymethyl) ester (BAPTA-AM), a chelator of cytosolic free calcium, would inhibit only cPKCs, a DAG antagonist such as calphostin would inhibit both cPKC and nPKCs. PMA-induced phosphorylation of Ser24, Ser307, and Ser612 was sensitive to BAPTA-AM pre-treatment (Fig. 4A). This suggests that calcium-dependent PKCs are activated by PMA and are required to induce phosphorylation of IRS-1 at these three serine residues. Consistent with this, we observed that serine phosphorylation induced by PMA was completely prevented by pretreatment with Gö6976, an inhibitor of cPKCα, -β, and -μ and Gö6983, an inhibitor of PKCα and -β but not -μ (Fig. 4A). In contrast, pretreatment with rottlerin, a selective inhibitor of nPKCδ and -θ was ineffective (Fig. 4, A and C). Unfortunately, calphostin treatment promoted significant degradation of IRS-1, making it difficult to analyze its impact on PMA-stimulated serine phosphorylation (Fig. 4A). Significant cytotoxicity was also observed using this concentration of calphostin (data not shown).

Fig. 4.

Effect of Selective PKC Inhibitors on PMA-Stimulated IRS1 Phosphorylation on Ser24, Ser307, and Ser612

A, Serum-starved NIH/hIR cells ectopically expressing rIRS1-wt were pretreated with indicated inhibitors for 30 min (or 24 h for chronic PMA) before stimulation with or without PMA (1 μm). IRS1 immunoprecipitates were collected and analyzed as described for Fig. 3. Selectivities of inhibitors are as follows: Gö6976 (10 μm) inhibits PKC α and β but not δ, ε, or ζ; Rottlerin (10 μm) inhibits nPKCs δ and θ to a significantly greater extent than cPKCs α, β, and γ and has the least effect on PKCs ∊, η, and ζ; BAPTA-AM (10 μm) inhibits all calcium-dependent cPKCs; calphostin (100 nm) inhibits all DAG-dependent PKCs (cPKC and nPKC); and Go6983 (100 nm) inhibits cPKC α and β but not PKC μ, δ, ∊, or ζ. Chronic PMA pretreatment down-regulated DAG-dependent PKCs α and δ but not μ and ζ. PKC β and γ were not detected (data not shown). Equal loading was confirmed by blotting against the p85 subunit of PI3K. Data are representative of at least three independent experiments. C, Rottlerin treatment reduced PMA-stimulated PKCδ activation. Whole-cell lysates from experiments described in panel A were probed for active PKCδ using anti-pSer643-PKCδ (upper panel) or PKCδ protein (lower panel).

To confirm the involvement of DAG-sensitive PKCs, cells were pretreated with PMA for 24 h, a condition known to down-regulate these enzymes. Figure 4A shows that chronic treatment with PMA alone is sufficient to prevent the PMA-induced phosphorylation of Ser24, Ser307, and Ser612. The selective depletion of endogenous PKCs was confirmed in the same samples (Fig. 4B). PKCα and -δ were significantly decreased whereas PKCμ and -ζ levels were unaltered. Neither PKCβ nor PKCλ was detected in these cells (data not shown). Taken together, these data suggest that PMA-stimulated phosphorylation at all three serine residues requires active conventional PKCs. Of these, PKCα represents the most likely PKC isoform that is required for PMA-stimulated serine phosphorylation of IRS1 on Ser24, Ser307, and Ser612 in NIH3T3 cells.

PMA-Stimulated Phosphorylation of Ser24, Ser307, and Ser612 Exhibit Different Pharmacological Inhibitor Profiles

Having established that all three serine residues can be directly phosphorylated by PKCs in vitro (Fig. 3) and by PKCα in vivo (Fig. 4), we next investigated whether PMA-stimulated PKCα activation was acting directly on IRS1 or indirectly by activating other downstream serine kinase cascades. To address this, pharmacological inhibitor profiles were constructed for each of the three serine residues using a range of chemical inhibitors that selectively target kinases previously implicated in IRS1 serine phosphorylation. Figure 5A demonstrates that PMA-stimulated phosphorylation of Ser24 was insensitive to the entire panel of six inhibitors used. This argues against the involvement of MEK, JNK, p38, IKKβ, mTOR, GSK3, and PI3K in Ser24 phosphorylation. In contrast, PMA-stimulated phosphorylation of Ser307 and Ser612 was sensitive to inhibition by all the kinase inhibitors (except SB203580) to varying degrees. Both Ser307 and Ser612 phosphorylation exhibited similar profiles but varied in the magnitude of response to individual kinase inhibition. These data suggest that although Ser307 and Ser612 are likely to be phosphorylated by similar kinases, they may vary in their relative affinities to individual kinases. In contrast, Ser24 is likely to be phosphorylated by a distinct kinase that is insensitive to the entire panel of inhibitors used in this experiment. Because PKCα is insensitive to these inhibitors, it remains a good candidate for the direct phosphorylation of Ser24 and it is likely to play an indirect role in PMA-stimulated phosphorylation of Ser307 and Ser612 (Fig. 5B).

Fig. 5.

Effect of Serine Kinase Inhibitors on PMA-Stimulated IRS1 Phosphorylation on Ser24, Ser307, and Ser612

A, Serum-starved NIH/hIR cells ectopically expressing rIRS1-wt were pretreated with indicated inhibitors for 30 min and then stimulated with or without PMA (1 μm). IRS1 immunoprecipitates were collected and analyzed as described for Fig. 3. The inhibitors used selectively target MEK1 (PD098059, 50 μm), JNK (SP600125, 20 μm), p38 (SB203580, 10 μm), IKKβ (salicylic Acid, 5 mm), mTOR (rapamycin, 200 nm), GSK3 (LiCl₂, 25 mm), and P13K (LY294002, 20 μm). B, Schematic diagram illustrating PMA-stimulated serine kinases involved in site-specific serine phosphorylation of IRS1. Data are representative of at least three independent experiments.

PKCα Interacts with the IRS-1 PH Domain

The direct interaction between full length IRS-1 and PKCα has been reported elsewhere (24). To test whether the IRS1-PH domain alone could mediate this interaction, glutathione-S-transferase (GST)-pull down assays were performed using recombinant PKCα incubated with either GST alone or GST fused to the IRS1-PH domain. Figure 6 confirms that PKCα can be identified only in GST immunoprecipitates that contain the IRS-1 PH domain. Subsequent in vitro kinase assays confirmed that this kinase-substrate complex contained sufficient levels of PKC to induce detectable phosphorylation on Ser24. These findings are in agreement with other studies suggesting that PKCα can directly interact with IRS1 (24) and that the PH domain is sufficient for this interaction (25).

Fig. 6.

Coprecipitation of IRS1 PH Domain with PKCα

Recombinant PKCα and glutathione beads were incubated with either GST peptide alone or GST-IRS1-PH domain. In vitro kinase assays were performed on GST precipitates followed by SDS-PAGE and immunoblotting against pser24, active PKCα, GST, and PKCα.

Functional Role of Serine 24 Phosphorylation

Recombinant IRS-1 PH-PTB domain peptides have been reported to bind phosphoinositides (26). To confirm that the IRS1-PH domain alone was sufficient to bind phospholipids, the affinity of recombinant GST-IRS-PH domain to various phosphoinositides was determined using a time-resolved fluorescence resonance energy transfer (FRET) assay (27). Under our assay conditions, PtdIns(4,5)P₂ (phosphatidylinositol 4,5-bisphosphate) was the only phosphoinositide to bind with any appreciable affinity (Fig. 7A). To determine whether phosphorylation of Ser24 can impact on PtdIns(4,5)P₂ binding to the IRS1-PH domain, we compared the binding properties of wt IRS1 PH domain with pseudophosphorylated (S24D) and non-phosphorylated (S24A) mutant peptides. As illustrated in Fig. 7B, binding by the S24D mutant is almost undetectable whereas S24A exhibits impaired PtdIns(4,5)P₂ binding affinity. To confirm that loss of PtdIns(4,5)P₂ binding affinity is not a result of mis-folded mutants, but can be induced by phosphorylation-dependent events, the assay was conducted using wt peptide after in vitro kinase treatment in the presence and absence of PKC (Fig. 7C) or ATP (Fig. 7D). Both these treatments are sufficient to impair the lipid-binding capabilities of the IRS1 PH domain.

Fig. 7.

Role of Serine 24 in Phosphoinositide Binding to IRS1 PH Domain

A, Lipid binding to recombinant wt GST-IRS1-PH domain was determined as described in Materials and Methods using increasing concentrations of the indicated biotinylated phosphoinositide and determining binding by generation of a FRET signal. B, PtdIns(4,5)P₂ binding to wt or mutant GST-IRS1-PH domains was determined as in panel A. FRET signals were normalized for absolute peptide concentration. C, Recombinant wt GST-IRS1-PH domain was incubated with or without recombinant PKCζ (minus cofactors) for 2 h at 30 C before determination of PtdIns(4,5)P₂ binding. D, Recombinant wt GST-IRS1-PH domain was incubated with recombinant PKCζ (minus cofactors) in the presence or absence of ATP for 2 h at 30C before determination of PtdIns(4,5)P₂ binding. Recombinant PKCζ was itself not tagged with GST. PI(3,4,5)P₃, Phosphatidylinositol 3,4,5-triphosphate; PI(3,4)P₂, phosphatidylinositol 3,4-bisphosphate; PI(4,5)P₂, phosphatidylinositol 4,5-bisphosphate; PI(3)P, phosphatidylinositol 3-monophosphate; PI(4)P, phosphatidylinositol 4-monophosphate.

The ability of PH domains to bind PtdIns(4,5)P₂ is often also reflected in their ability to associate with PtdIns(4,5)P₂-rich membranes such as plasma membrane. Indeed, the IRS-1 PH domain has been reported to localize to the cell periphery (28, 29). To determine whether phosphorylation of Ser24 could impact on this function of the IRS-1 PH domain, confocal fluorescent images were taken of human embryonic kidney (HEK)293 cells transfected with either green fluorescent protein (GFP) alone, GFP-tagged IRS-1 PH wt, or GFP-tagged IRS-1 PH Ser24 mutant peptides. As illustrated in Fig. 8, GFP alone is uniformly distributed throughout the cell cytoplasm and nucleus. However, the wt GFP-IRS-1PH domain is found in the nucleus but is also localized to the cell periphery. This is consistent with its ability to bind PtdIns(4,5)P₂, which is particularly abundant in the plasma membrane of these cells. In contrast, the S24D mutant does not exhibit any discernable peripheral localization whereas the S24A mutant can be found at the cell periphery (Fig. 8).

Fig. 8.

Intracellular Localization of GFP-Tagged PH Domain Mutants

HEK293 cells were grown on poly-l-lysine-coated cover-slips and transiently transfected with either pNEGFP, pNEGFP-IRS1-PH wt, pNEGFP-IRS1-PH S24A, or pNEGFP-IRS1-PH S24D. Monolayers were serum starved (overnight) 48 h post transfection, washed with PBS, fixed in 2% formalin, and mounted in aqueous mount. Confocal fluorescent images were captured using a Zeiss imaging microscope set to capture GFP and DAPI blue fluorescence from a slice thickness of 0.8 nm using a ×60 oil emersion objective. Representative pictures are shown from at least three independent experiments performed in duplicate.

Loss of PH domain function has the potential to negatively impact on insulin sensitivity. To investigate the impact of Ser24 phosphorylation on insulin sensitivity of adipocytes, we established stable preadipocyte cell lines constitutively expressing either empty vector, wt IRS1, IRS1 S24A, or IRS1 S24D. Initial examination of adipocyte differentiation potential confirmed that the Ser24 mutant did not significantly alter the ability of preadipocytes to accumulate lipids under various adipogenic conditions (Fig. 9A). Western blot analysis demonstrates that similar levels of each S24 mutant were expressed in mature adipocytes and that the level of ectopic IRS-1 expression approximately matches that of endogenous IRS-1 because there appears to be a 2-fold increase in total IRS1 expression in the transfected cells relative to the empty vector controls (Fig. 9B). In dose-response assays for insulin-stimulated glucose uptake, those adipocytes expressing the pseudophosphorylated IRS1-Ser24D mutant consistently produced a significantly lower response to insulin than their wt counterparts (Fig. 9C), and the S24A mutant produced an intermediate response. This suggests that Ser24 phosphorylation may indeed impair insulin-stimulated glucose uptake in adipocytes.

Fig. 9.

3T3-L1 Adipocytes Expressing rIRS-1 Ser24 Mutants Show Similar Adipogenic Profiles and Levels of Mutant Expression but Respond Differently to Insulin-Stimulated Glucose Uptake

A, 3T3-L1 preadipocytes were induced to differentiate with range of adipogenic treatments; adipocyte media only (AM), AM supplemented with insulin (INS), AM supplemented with isobutylmethylxanthine only (IBMX), AM supplemented with dexamethasone only (Dex), and AM supplemented with full induction cocktail of isobutylmethylxanthine, dexamthasone, and insulin (MDI). On d 8, these were fixed and accumulated lipids stained with oil red O. B, Equal amounts of total protein extract from mature adipocytes were separated on SDS-PAGE and immunoblotted for myc-tagged mutants, total IRS-1 protein, and IRβ as a marker of differentiation. C, Mature adipocytes expressing IRS-1 Ser24 mutants were serum starved (24 h) and 2-deoxy-d-[2,6 ³H]-glucose uptake was determined in triplicate. Data represent mean ± sd and is representative of at least four independent experiments. Asterisk indicates statistical significance (P < 0.01) in comparison with wt cells with the same treatment. EV, Empty vector; DPM, disintegrations per min.

DISCUSSION

Serine Phosphorylation Sites in IRS-1

Serine phosphorylation of IRS1 is an important mechanism for attenuating insulin signaling. IRS1 is also unusually rich in serine residues, and recent reports suggest that more than 70 serines appear in consensus phosphorylation sites for kinases (9). However, given that knowledge of preferred substrate motifs in the entire kinome remains limited, the number of sites that can potentially be phosphorylated is likely to be significantly greater. Indeed, NetPhos analysis predicts that as many 113 serines represent putative phosphorylation sites in hIRS1. To date, approximately 16 serine phosphorylation sites have been observed in vivo and in response to agonist stimulation. This makes it unlikely that a single serine site is responsible for inactivating IRS1 especially in vivo in pathological circumstances such as insulin resistance. However, the location of individual sites and/or clusters can reveal the molecular basis for negative regulation. Here we describe the identification and confirmation of a phosphorylation site that lies in the N-terminal PH domain of IRS1. The location of this residue, in a putative phosphoinositide-binding pocket, suggests a novel mechanism for regulating IRS1 signaling via its PH domain.

Role of Serine 24 Phosphorylation in PH Domain Function

By analogy to the crystal structure of human IRS1 PH-PTB peptide, Ser24 of rIRS1 lies in a large exposed cationic patch that is present at the base of the PH domain. This region includes the putative phosphoinositide-interacting residues Lys 21, Lys 23 His 26 and Arg 28, Lys 61 and Arg 62 (highlighted in Fig. 1B) (26). Substitution of one of these, R28C, is sufficient to disrupt lipid binding to PH domain (30) as well as interactions with the pleckstrin homology domain interacting protein PHIP (31). Our findings suggest that the addition of a charged phosphate group on Ser24 can also induce significant changes to cationic patch and alter lipid interactions (Fig. 7, B and C). This modification is also sufficient to alter both intracellular localization (Fig. 8) and subsequent participation in insulin-stimulated glucose uptake (Fig. 9C). These observations are entirely consistent with a recent independent study demonstrating that the IRS1-S24D mutant shows impaired insulin-stimulated IR-IRS-1 interactions, tyrosine phosphorylation of IRS-1, recruitment/activation of PI 3-Kinase, and insulin-stimulated Glut4 translocation (19).

A surprising observation in our studies was that an alanine substitution of Serine 24 did not behave in a fashion akin to a nonphosphorylated peptide (Figs. 7B, 8, and 9C). Indeed, this mutation consistently produced actions that were intermediate between wt and the pseudophosphorylated Ser24. One possible explanation for these observations is that serine 24 is also involved in hydrogen bonding to a main-chain amide (Ref. 26 and Fig. 1), and substitution with an alanine is sufficient to disrupt this bond and hence alter the functionality of the PH domain. It is intriguing to speculate that, in addition to altering the electrostatic properties of the cationic patch, phosphorylation of Ser24 may also significantly alter PH domain structure by disrupting this hydrogen bond.

The PH domains of IRS1 and IRS2 are more similar to each other (58% identical and 78% similar) than to IRS3 (IRS1-PH is 37% identical and 63% similar to IRS3, and IRS2-PH is 33% identical and 57% similar to IRS3) (32). Hence, it is surprising that Ser24 is not conserved between IRS1 and IRS2, but rather that it is present in IRS3 and IRS_CHICO. It is interesting to note that functional differences between the PH domains of IRS1 and IRS2 have been reported (33, 34). Whether this is accomplished by differential regulation of PH domains by Ser24 phosphorylation remains a possibility.

Recently, two other proteins have been reported to be phosphorylated in their PH domains after PKC activation (30, 31). As with IRS1-Ser24, these phosphorylation events are also sufficient to disrupt phosphoinositide binding (35) and membrane translocation (36). Intriguingly, DGKδ1-PH is also phosphorylated by cPKCs, and the targeted serine site is also located in an analogous region. Both these serine sites are predicted as putative PKC phosphorylation sites in our bioinformatics analysis (data not shown). This further validates the use of bioinformatic software in hypothesis generation.

Role of PKCα in Ser24 Phosphorylation

In this study, we show that in vitro, IRS1-Ser24 is a substrate for representatives of all three PKC subfamilies (α, β, and ζ). However, this relative lack of specificity is not recapitulated in vivo. In intact NIH/IR/IRS1 cells, agonist-induced phosphorylation of Ser24 was selective for specific PKC isoforms. Ser24 is phosphorylated after activation of phorbol ester-dependent kinases and not by C2-ceramide or chronic insulin treatment. This suggests an involvement of DAG-dependent PKCs. PMA-stimulated phosphorylation of Ser24 is likely to be mediated by conventional PKCα, as it is sensitive to inhibition by the calcium chelator, BAPTA-AM, inhibitors of cPKC (i.e. Gö6976 and Gö6983), and chronic PMA treatment, and neither PKCβ nor PKCγ is present in NIH3T3 cells (22). This conclusion is also consistent with the in vitro observations wherein PKCα consistently produced the most robust phosphorylation at this site. It is therefore likely that PKCα is a preferred kinase to phosphorylate Ser24.

It is interesting to note that Motif scan predicted Ser24 to be preferred by PKCζ as a substrate site. Also the amino acid sequence flanking Ser24 does not conform to two commonly cited consensus sequences for substrates of classic/generic PKCs, i.e. X-S/T-X-R/K (Prosite pattern and Ref. 37) or R-X-X-S/T (9). However, it is in absolute agreement with PKC isoform-specific sequence motifs reported by Cantley and colleagues (38). They report that all PKCs prefer substrates that have basic residues in the −3 position and a hydrophobic amino acid must occupy position +1. Additionally, cPKCs prefer substrates with basic residues particularly at positions +2 and +3, whereas nPKC and aPKC tend to prefer hydrophobic residues at these positions. In IRS1 Ser24 is flanked by such residues in the positions that confer this as a preferred phosphorylation site for cPKCs. Therefore our experimental evidence is entirely consistent with this particular motif-based prediction.¹ This consensus motif, together with our in vivo and in vitro data, suggests that PKCα is likely to directly phosphorylate Ser24. The finding that PMA-stimulated Ser24 phosphorylation is not affected by any of the non-PKC inhibitors (Fig. 5) further supports this notion, as does the findings that IRS1 and PKCα are constitutively associated in vivo (39, 40) and that the PH domain itself is sufficient for this interaction (Fig. 6).

One possibility that cannot be excluded is that differences in endogenous levels, relative activities, and/or intracellular location of different PKC isoforms may also determine Ser24 phosphorylation in vivo. In this regard, PKCα, -δμ, and -ζ are all expressed in NIH3T3 fibroblasts (Fig. 4B), but only PKCα is involved in PMA-stimulated Ser24 phosphorylation. Whether this also occurs in other cell types that express cPKC β- and/or γ-isoforms warrants further investigation.

Role of PKCα in Insulin Resistance

Attempts to address the role of cPKCs in insulin signaling have produced conflicting data. Studies using PKC inhibitors and phorbol ester-induced PKC down-regulation have suggested that DAG-activated PKCs are not required for normal insulin-stimulated glucose uptake in adipocyte and muscle cells. In contrast, the targeted disruption of murine PKCα does enhance insulin signaling through IRS1, leading to increased insulin-stimulated glucose transport in adipocytes and skeletal muscle (41). It is therefore possible that PKCα may serve as a tonic endogenous inhibitor of IRS1-dependent pathways (23).

Although the role of cPKCs in physiological responses to insulin remains unresolved, it is clear that overexpression or activation of DAG-dependent PKCs can significantly impair insulin signaling (42, 43). Indeed, elevated PKCα, -δ, and θ activities are implicated in insulin resistance in vivo (44, 45). Our observation that IRS1 Ser24 is not phosphorylated in response to insulin but only after PMA stimulation suggests that this site may represent a novel diagnostic marker of elevated cPKC activity that may occur in pathophysiological states such as insulin resistance and hyperglycemia (23).

pSer24 Exhibits Characteristics that Are Distinct from pSer307 and pSer612

As with Ser24, we found that PKCs could directly phosphorylate Ser307 and Ser612 in vitro, but PMA-stimulated phosphorylation required active PKCα in vivo. This could suggest that these sites are also the direct targets of cPKC activity. However, neither site lies within a consensus motif for PKC substrates and, unlike PMA-stimulated pSer24, both pSer307 (312 in hIRS1) and pSer612 (616 in hIRS1) were sensitive to other serine kinase inhibitors. This implicates other downstream serine kinases in the direct interaction and phosphorylation of Ser307 and Ser612. Indeed, Ser612 was identified as a target for active PKCs and Erk (46), whereas Ser307 is targeted by numerous kinases including JNK, IKKβ, AKT, and mTOR (47–49). Figure 5B summarizes the residue selective roles of PMA-responsive serine kinases in IRS1 phosphorylation. Future studies should help delineate the specific serine kinase and upstream kinase pathways that directly target these sites.

An additional novel observation made during our studies was that although casein kinase II has little effect on Ser24 phosphorylation, it does directly phosphorylate Ser307 and Ser612 in vitro. This activity on full-length IRS1 was also sufficient to cause the most significant mobility shift, suggesting that there may be a number of other sites. Indeed, IRS1 has 19 consensus sites for casein kinase II phosphorylation (S/T-X1-X2-E/D, where X1 is any amino acid except proline). This consensus motif is based on the requirement of acidic rather than basic residues (37). However, neither Ser307 nor Ser612 complies with this consensus and, as such, is unlikely to be directly phosphorylated by casein kinase II in vivo. This remains to be experimentally confirmed.

In summary, an increasing array of agonists, including insulin, IGF, TNF, FFA, C2 ceramide, and PMA, are being reported to stimulate serine phosphorylation of IRS1 at multiple sites including Ser307 and Ser612. It appears that many of these sites are promiscuous and open to regulation by multiple kinases (and phosphatases) and may be physiologically important in mediating feedback regulation of insulin signaling. In contrast, Ser24 is phosphorylated in vivo in response to a limited number of agonists and serine kinases. Once phosphorylated it can significantly alter PH domain function and impair insulin sensitivity. Future studies will address whether phosphorylated Ser24 has potential for use as a diagnostic marker for detection of pathological signaling events that occur during insulin resistance and hyperglycemia.

MATERIALS AND METHODS

Materials

Recombinant PKC isoforms were purchased from Calbiochem (La Jolla, CA). The specific primary antibodies used were anti-myc [9E10-T (Siddle laboratory), or 4A6 (Upstate Biotechnology, Inc., Lake Placid, NY)], anti-pSer307-rIRS1 (Upstate no. 07–247), anti-pSer616-hIRS1 (equivalent to pSer612 in rIRS1; no. 44–550; Biosource International Ltd., Camarillo, CA), anti-IRS1 (Upstate), anti-IRβ (Santa Cruz Biotechnology, Inc., Santa Cruz, CA) anti-PKCζ (Santa Cruz), anti-PKCα, -β, -δ, and -μ (BD Biosciences, Palo Alto, CA), anti-pSer643-PKCδ (New England Biolabs, Beverly, MA), anti-His6 (Santa Cruz), anti-GST (GST103, Siddle laboratory), and anti-p85 (Siddle laboratory). Unless otherwise stated, all other reagents were purchased from Sigma Chemical Co. (St. Louis, MO) or Calbiochem.

Bioinformatic Analysis

Protein sequences for IRS family members were systematically analyzed with two computational tools that predict putative phosphorylation sites. NetPhos 2.0 (www.cbs.dtu.dk/services/NetPhos/) is based on neural network predictions for serine, threonine, and tyrosine phosphorylation sites (50) whereas Motif Scan in Scansite (scansite.mit.edu/) identified experimentally determined protein binding or kinase substrate motifs in proteins (51). Structure-based homology alignment was performed using Fugue, (www-cryst.bioc.cam.ac.uk/~fugue/) (52) to compare IRS-PH domains with the crystal structure of hIRS1 PH domain (PDB:1qqga). Also used was Homstrad, a homologous structure alignment database (www-cryst.bioc.cam.ac.uk/homstrad/) (53), the output of which incorporates three-dimensional structural features that are annotated with JOY (www-cryst.bioc.cam.ac.uk/~joy/) (54).

Generation of Anti-pSer24 IRS1 Antibody

The generation of anti-pSer24-IRS1 was commissioned from Cambridge Research Biochemicals Ltd. (Cambridge, UK). Briefly, rabbit antisera were generated against the following N-terminally, keyhole limpet hemocyanin-conjugated phosphopeptide: [C]YLRKPKS(p)MHKRFF. The phosphoreactive serum was then affinity purified on a Thiopropyl Sepharose 6B column derivatized with nonphosphorylated antigen. The unbound antiserum was then passed down a Thiopropyl Sepharose 6B column derivatized with the phosphorylated peptide. The resulting triethylamine eluate retained significant phosphospecific immunoreactivity on peptide-coated ELISA plates and was used in Western blot analysis as described below.

Plasmid Construction

C-terminally tagged full-length rIRS1-Myc-(His)6 in pcDNA3.1 was sequenced in full, and all nonconserved variations were corrected by using QuikChange XL (Stratagene, La Jolla, CA) such that rIRS1wt cDNA encoded the protein sequence NP_037101. This then was used as the template in further site-directed mutagenesis to construct the corresponding Ser24Ala and Ser24Asp mutants. Successful mutagenesis was confirmed by direct sequencing. The coding sequence for full-length myc-his tagged rIRS1 wt or S24 mutants was released from pcDNA3.1 with an NheI/AflII digestion and blunt ended with Klenow. They were then ligated into the SnaB1 site of the retroviral vector, pBabePuro. Diagnostic digests with BamHI confirmed both presence and correct orientation of the inserts.

All expression plasmids encoding IRS1-PH domains corresponded to amino acids 1–113 of rIRS1 (IRS1^PH). The tagged constructs were generated by PCR amplification using appropriate mutant templates and primers also encoding compatible restriction sites to facilitate in-frame, directional ligation into either the N-terminal His6-tagging vector pMwHis6 or C-terminal GST-tagging vector pMwGST and C-terminal GFP tagging vector pEGFP-N1 (BD Biosciences; CLONTECH Laboratories, Inc., Palo Alto, CA). Correct orientation and in-frame positioning were verified by direct sequencing. Primer sequences are available on request.

Generation of Recombinant rIRS1^PH Peptides

Competent cBL21DE3pLys cells were transformed with appropriate expression plasmids and peptide expressing, isopropyl-β-d-thiogalactopyranoside (IPTG)-inducible clones were identified. For HIS6-tagged peptides, overnight starter cultures were used to inoculate small-scale (100 ml) expression cultures. These were grown at 37 C to 0.6 OD₆₀₀ before being induced with 1 mm IPTG and left to express peptide during an overnight incubation at 22 C. The resulting bacterial cultures were pelleted (at 5000 × g for 10 min) and resuspended in ice-cold, sterile sonication buffer (10 mm HEPES, 200 mm NaCl, 25 mm MgCl₂, DNAse, 4-(2-aminoethyl)benzenesulphonyl fluoride, 1% Triton X). After sonication these were clarified by centrifugation (at 10,000 × g for 10 min), and resulting supernatant was incubated with activated Ni²⁺ Sepharose beads (at 4 C for 1 h). Beads were washed twice with cold wash buffer (10 mm HEPES, 200 mm NaCl, 20 mm imidazole) and His-tagged peptides eluted in elution buffer (10 mm HEPES, 200 mm NaCl, 300 mm imidazole). For GST-tagged peptides, large-scale (1 liter) expression cultures were induced with IPTG and grown for a further 6 h at 37 C. Each was pelleted and lysed in Bug Buster (Novagen, Madison, WI). After centrifugation, clarified supernatants were incubated with reduced glutathione-sepharose beads (at room temperature for 1 h). Beads were washed three times with wash buffer (PBS and 1% Triton-X100) followed by a fourth wash with equilibration buffer (100 mm Tris-HCl, pH 8.8; 200 mm NaCl; 20% glycerol). GST-tagged peptides were then eluted in elution buffer (100 mm Tris-HCl, pH 8.8; 200 mm NaCl; 20% glycerol; and 15 mm glutathione). Peptide purification was confirmed by SDS-PAGE, and eluates were subsequently dialysed using Slide-A-Lyser cassettes (Pierce Chemical Co., Rockford, IL) with dialysis buffer (100 mm Tris-HCl, pH 8; 200 mm NaCl; and 20% glycerol) as per manufacturer’s instructions.

In Vitro Kinase Assays

Recombinant rIRS1 PH domains (5 μg) or immunoprecipitated full-length IRS1 was incubated with 0.2 μg of recombinant casein kinase II or PKC isoforms α, δ, and ζ in the appropriate kinase buffer. Casein kinase buffer comprised 20 mm Tris-HCl (pH 7.5), 50 mm KCl, 10 mm MgCl₂, 100 μm Mg-ATP, whereas PKCα required Ca²⁺-dependent kinase buffer (20 mm HEPES, 10 mm MgCl₂, 0.5 mm CaCl₂, 0.5 mm dithiothreitol, 100 μm Mg/ATP), and PKCδ and PKCζ required Ca²⁺-independent buffer (20 mm HEPES, 10 mm MgCl₂, 0.5 mm EGTA, 0.5 mm dithiothreitol, 100 μm Mg/ATP). Additional lipid supplements included 10 μg/ml diolein and 100 μg/ml phosphatidylserine (for PKCα and PKCδ) or 100 μg/ml phosphatidylserine (for PKCζ). Reactions were initiated with the addition of either kinase or ATP, incubated at 30 C for 2 h and terminated by the addition of reducing Laemmli loading buffer and heat inactivation.

In Vitro Lipid-Binding Assay

Lipid binding was determined using a time-resolved FRET assay as described elsewhere (27) in which increasing concentrations of a biotinylated phosphoinositide were bound to a FRET acceptor, streptavidin allophycocyanin conjugate, the FRET donor being a Europium chelate-labeled anti-GST antibody complexed with a GST fusion of the IRS1 PH domains. Binding of PH domain to lipid results in formation of a complex generating a FRET signal. The binding assays were carried out in 50 mm HEPES (pH 7.4), 150 mm NaCl, 2 mm dithiothreitol, and 0.02% Cholate with allophycocyanin conjugate-Streptavidin (Prozyme Ltd., San Leandro, CA) 32 nm, 0–20 pmol (as required) biotinylated, short-chain (diC8) phosphoinositides (Cell Signaling Technology, Danvers, MA) and 20–30 nm IRS1 PH domain with 21 nm Lance chelate-labeled anti-GST antibody in a final volume of 50 μl. For all assays, the samples were mixed in 96-well plates in Lumitrax 200 and the plates read in an LJL Analyst (Molecular Devices, Sunnyvale, CA) with the following settings: excitation filter, 360 –35 nm; emission filter, 665 nm; dichroic filter, 505 nm; digital sensitivity of PMT 1000 V, set to 2; flashes per well, 100; interval between flashes, 10 msec; read time after flash, 50 msec; and integration time, 1000 msec. Data are expressed as a FRET Ratio of the FRET signal divided by the total Europium fluorescence.

Cell Culture

BOSC23 cells and 3T3-L1 preadipocytes were routinely propagated in DMEM containing 4.5 g/liter glucose, 10% bovine calf serum (Hyclone Laboratories, Inc., Logan, UT), 50 U/ml penicillin, and 50 μg/ml streptomycin at 10% CO₂. HEK293 cells were propagated in DMEM containing 4.5 g/liter glucose, 10% newborn calf serum, 50 U/ml penicillin, and 50 μg/ml streptomycin at 5% CO₂. NIH/hIR cells, mouse fibroblasts overexpressing the human IR (and G418 resistant), were grown in the same medium as HEK293 cells but with the addition of 600 μg/ml G418.

Generation of Stable Cell Lines Expressing rIRS1 Mutants

3T3-L1 and NIH/hIR cells stably expressing myc-his-tagged IRS1 were generated by retroviral mediated gene transfer (55). Briefly, subconfluent BOSC23 packaging cells were transfected with pBabePuro-rIRS1mychis plasmids, and conditioned media was collected 48 h later. This was passed through a 0.45-μm filter, supplemented with 16 μg/ml of polybrene, and used to infect proliferating 3T3-L1 preadipocytes or NIH/hIR cells. Drug selection with puromycin (4 μg/ml for 3T3-L1 and 16 μg/ml for NIH/hIR) was initiated 2 d later. After a week in drug selection, stable cell lines were confirmed by Western blot analysis and expanded. Stocks were frozen in liquid nitrogen until required.

Adipocyte Differentiation and Glucose Uptake

3T3-L1 preadipocytes (2 d postconfluent) expressing either empty vector or full-length IRS1 mutants were maintained at 10% CO₂ in adipocyte media AM [high-glucose DMEM containing antibiotics, 10% Cosmic Calf Serum (Hyclone), 4 μg/ml puromycin]. Adipocyte differentiation was induced by incubation for 2 d in AM supplemented with induction cocktail (5 μg/ml insulin, 0.5 mm isobutylmethlyxanthine, and 1 μm dexamethasone). This was followed by a further 2-d incubation in AM supplemented with insulin only. Thereafter, cells were fed unsupplemented AM every 2 d. Glucose-uptake assays were performed on fully differentiated adipocytes using 2-deoxy-d-[2,6 ³H]-glucose uptake as described previously (56).

Confocal Fluorescence Microscopy

HEK293 cells were grown on poly-l-lysine-coated coverslips and transiently transfected with either pNEGFP, pNEGFP-IRS1-PH wt, pNEGFP-IRS1-PH S24A, or pNEGFP-IRS1-PH S24D. Monolayers were serum starved (overnight) 24–48 h after transfection, washed with PBS, fixed in 2% formalin, and mounted in 4′,6-diamidino-2-phenylindole (DAPI) containing Vectorshield mount medium (Vector Laboratories, Inc., Burlingame, CA). Confocal fluorescence microscopy was performed using a LSM510 laser confocal microscope system (Carl Zeiss, Thornwood, NY) and a ×60 oil emersion objective. Fluorescent images were captured using Argon 488-nm (GFP) and Krypton 413-nm (DAPI) lasers and a confocal slice thickness of less than 0.8μm.

Phosphoprotein Extraction and Analysis

Two-day postconfluent cells were washed and incubated with serum-free medium and starved for 4 h before stimulation with agonists. After stimulation, culture medium was removed and monolayers were transferred to 4 C. They were then washed with ice-cold PBS and lysed in modified RIPA buffer (50 mm Tris-HCl, pH 7.4; 1% NP-40; 0.25% sodium deoxycholate; 150 mm sodium chloride; and 1 mm EDTA) freshly supplemented with 50 mm β-glycerol-2-phosphate, 5 μm 4-(2-aminoethyl)benzenesulphonyl fluoride, 1 μg/ml aprotinin, 1 μg/ml leupeptin, 1 μg/ml pepstatin A, 1 mm sodium pervanadate, 1 mm sodium fluoride, and 50 nm okadaic acid. Cell lysates were scraped and clarified by centrifugation and the supernatant was collected. Protein quantification was determined using the Bio-Rad DC protein assay kit (Bio-Rad Laboratories, Inc., Hercules, CA).

For immunoprecipitation with anti-myc antibody (9E10), protein lysates were precleared with protein G beads and incubated overnight (with rotation, at 4 C) with prebound T9E10-protein G beads. Immunoprecipitates were subsequently washed three times with ice-cold PBS supplemented with 50 mm β-glycerol-2-phosphate, 1 mm sodium pervanadate, 1 mm sodium fluoride, and 50 nm okadaic acid and resuspended in 1× Laemmli loading buffer and boiled for 5 min. Immunoprecipitated proteins were separated by SDS-PAGE, transferred onto polyvinylidine difluoride membranes (Amersham Pharmacia Biotech, Arlington Heights, IL), and used for immunodetection with the following primary antibody dilutions: anti-pSer24 (1:100), anti-pSer307 (1:500), anti-pSer616 (equivalent to pSer612 in rIRS1; 1:1000), and antirabbit (1:10,000). Sequential immunoblotting was performed only after blots had been stripped and complete removal of primary antibody had been confirmed.

Abbreviations

AKT/PKB

AKR mouse thymoma viral proto-oncogene/protein kinase B

aPKC

atypical PKC

BAPTA-AM

1,2-bis(o-aminophenoxy)ethane-N,N,N′,N′-tetraacetic acid tetra(acetoxymethyl) ester

cPKC

conventional PKC

DAG

diacylglycerol

DAPI

4′,6-diamidino2-phenylindole

FRET

fluorescence resonance energy transfer

GFP

green fluorescent protein

GSK

glycogen synthase kinase

GST

glutathione-S-transferase

HEK

human embryonic kidney

IKK

inhibitor of κB kinase

IPTG

isopropyl-β-d-thiogalactopyranoside

IR

insulin receptor

IRS

insulin receptor substrate

JNK

cJun NH2-terminal kinase

MEK

mitogen-activated Erk kinase

mTOR

mammalian target of rapamycin

nPKC

novel PKC

PH

pleckstrin homology

PI3K

phosphatidylinositol 3-kinase

PKC

protein kinase C

PMA

phorbol 12-myristate 13-acetate

pSer

phosphoserine

PTB

phosphotyrosine binding

PtdIns(4,5)P2

phosphatidylinositol 4,5-bisphosphate

wt

wild type