14-3-3 Protein Binds to Insulin Receptor Substrate-1, One of the Binding Sites of Which Is in the Phosphotyrosine Binding Domain*

Insulin binding to its receptor induces the phosphorylation of cytosolic substrates, insulin receptor substrate (IRS)-1 and IRS-2, which associate with several Src homology-2 domain-containing proteins. To identify unique IRS-1-binding proteins, we screened a human heart cDNA library with32P-labeled recombinant IRS-1 and obtained two isoforms (ε and ζ) of the 14-3-3 protein family. 14-3-3 protein has been shown to associate with IRS-1 in L6 myotubes, HepG2 hepatoma cells, Chinese hamster ovary cells, and bovine brain tissue. IRS-2, a protein structurally similar to IRS-1, was also shown to form a complex with 14-3-3 protein using a baculovirus expression system. The amount of 14-3-3 protein associated with IRS-1 was not affected by insulin stimulation but was increased significantly by treatment with okadaic acid, a potent serine/threonine phosphatase inhibitor. Peptide inhibition experiments using phosphoserine-containing peptides of IRS-1 revealed that IRS-1 contains three putative binding sites for 14-3-3 protein (Ser-270, Ser-374, and Ser-641). Among these three, the motif around Ser-270 is located in the phosphotyrosine binding domain of IRS-1, which is responsible for the interaction with the insulin receptor. Indeed, a truncated mutant of IRS-1 consisting of only the phosphotyrosine binding domain retained the capacity to bind to 14-3-3 protein in vivo. Finally, the effect of 14-3-3 protein binding on the insulin-induced phosphorylation of IRS-1 was investigated. Phosphoamino acid analysis revealed that IRS-1 coimmunoprecipitated with anti-14-3-3 antibody to be weakly phosphorylated after insulin stimulation, on tyrosine as well as serine residues, compared with IRS-1 immunoprecipitated with anti-IRS-1 antibody. Thus, the association with 14-3-3 protein may play a role in the regulation of insulin sensitivity by interrupting the association between the insulin receptor and IRS-1.

Protein tyrosine kinases play key roles in transmitting ex-tracellular signals that induce specific cellular events such as proliferation, differentiation, gene expression, and metabolism. These signals are propagated by sequential protein-protein interactions and the resulting protein phosphorylation cascade. Thus, to identify the molecule(s) associated with the some key molecules mediating these events constitutes a strategy for elucidating the signal transduction network. The CORT (cloning of receptor target) method is a modification of the expression cloning method, using tyrosine-phosphorylated growth factor receptor as a probe, which was originally established by Skolnik et al. (1) for the purpose of isolating cDNAs coding for proteins bound to the tyrosine-phosphorylated epidermal growth factor receptor.
In the case of insulin signaling, the first action exerted by insulin is activation of insulin receptor tyrosine kinase, which leads to the phosphorylation of several cytosolic substrates including insulin receptor substrate (IRS) 1 -1 (2) and IRS-2 (3). IRS-1 possesses 21 potential tyrosine phosphorylation sites and functions as a "docking protein," transmitting insulin signals to several proteins containing Src-homology 2 (SH2) domains (4). Thus, similar methods have been employed successfully to clone cDNAs coding for the proteins that bind to phosphorylated IRS-1. To date, three novel regulatory subunits for phosphatidylinositol 3-kinase (p55␥ (5,6), p55␣ (6,7), and 50␣ (8)) have been isolated by this method and reported. In addition, Fyn tyrosine kinase (9) was demonstrated to bind phosphorylated IRS-1.
In this study, to identify additional IRS-1-binding proteins, we screened a human heart cDNA library and obtained cDNAs coding for the 14-3-3 family proteins, which function as IRS-1binding proteins. Recently, the roles of 14-3-3 proteins have been clarified with respect to the regulation of various signal transductions and enzyme activities (10 -19). These proteins are known to bind to the phosphoserine-containing motifs in several proteins (20) but not to the phosphotyrosine-containing motifs in IRS-1.

EXPERIMENTAL PROCEDURES
Expression Screening with Human [ 32 P]IRS-1 Protein-Baculovirusproduced IRS-1 was labeled by means of incubation with activated insulin receptor in the presence of Mn 2ϩ and [␥-32 P]ATP, as described previously (6). An oligo(dT)-primed human heart cDNA library was prepared in Exlox (Novagen) according to the manufacturer's instructions. A Exlox library was plated at 50,000 phages/plate in Escherichia * This work was supported by Grant-in-aid 08671136 for scientific research (to T. A.) from the Ministry of Education, Science, and Culture of Japan and by a grant from Tanabe Medical Frontier Conference. 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 U.S.C. Section 1734 solely to indicate this fact.
The nucleotide sequence(s) reported in this paper has been submitted to the GenBank TM /EBI Data Bank with accession number(s) AB000732 (human IRS-2).
coli strain BL21(DE3)pLysE. The 15-cm plates containing 2,000,000 plaques were incubated for 8 h at 37°C and then overlaid with nitrocellulose filters (Millipore) that had been impregnated with isopropyl-␤-D-thiogalactopyranoside and incubated for 10 h at 37°C. Hybridization of the filters with the [ 32 P]IRS-1 probe and washing were performed as described previously (6,8). The cDNA inserts in pBluescript were prepared by in vivo excision according to the manufacturer's instructions (Stratagene). The nucleotide sequences were determined using an Applied Biosystems Inc. automatic sequencer.
Preparation of Glutathione S-transferase (GST)-14-3-3 Fusion Protein-The cDNA encoding full-length human 14-3-3⑀, cloned in our laboratory, was subcloned into a pGEX-5X-1 vector (Pharmacia Biotech Inc.) which was used to transform E. coli JM105 (Promega). Transformed cells were grown to an A 600 of 0.6 in LB medium supplemented with 0.1 mg/ml ampicillin and stimulated for 3 h with 1.0 mM isopropyl-␤-D-thiogalactopyranoside. GST fusion proteins were isolated and purified by affinity chromatography on Glutathione-Sepharose 4B (Pharmacia). Glutathione was removed by dialysis against phosphatebuffered saline containing 10 mM dithiothreitol.
Cell Culture-Sf 9 cells were grown in TC100 (Life Technologies, Inc.) medium containing 10% fetal calf serum at 27°C. L6 myoblasts were incubated and fused into myotubes as described (21). HepG2 hepatoma cells and Chinese hamster ovary (CHO) cells were grown in Dulbecco's modified Eagle's medium and Ham's F-12 medium, respectively, containing 10% fetal calf serum at 37°C in 5% (v/v) CO 2 in air.
Preparation of Baculovirus-produced Recombinant Proteins-A fragment of mouse IRS-2 DNA was obtained by polymerase chain reaction using mouse genomic DNA and two degenerated oligonucleotide primers based on the reported amino acid sequence ( The human genomic DNA library was screened using this polymerase chain reaction product as a probe. A cDNA coding for a full-length human IRS-2 protein was subcloned into pBluescript, and the nucleotide sequence was determined using an Applied Biosystems Inc. automatic sequencer. The full-length coding regions of 14-3-3⑀, IRS-1 and IRS-2 were subcloned into pBacPAK9 transfer vector (CLONTECH), and the baculoviruses were prepared according to the manufacturer's instructions. For protein production, Sf9 cells were infected with these baculoviruses and grown for 48 h.
Immunoprecipitation and Immunoblotting-Confluent monolayers of HepG2 cells were incubated for 3 h in serum-free Dulbecco's modified Eagle's medium before insulin stimulation. In some experiments, the HepG2 cells were first incubated with 100 nM okadaic acid for 30 min.
In other experiments, HepG2 cells were stimulated with 10 Ϫ7 M insulin for 5 min or the periods indicated in Fig. 3A at 37°C. The HepG2 cells, L6 cells, and CHO cells were lysed at 4°C with ice-cold phosphate-buffered saline (2 ml/10-cm dish of cells) containing 1% Nonidet P-40, 1 mM phenylmethylsulfonyl fluoride, 10 g/ml aprotinin, 100 mM sodium fluoride, and 2 mM sodium orthovanadate. The Sf9 cells were lysed with phosphate-buffered saline containing 1% Triton X-100 and 0.1 mM phenylmethylsulfonyl fluoride. Insoluble material was removed by centrifugation at 15,000 ϫ g for 10 min at 4°C. The cell lysates (1 ml, approximately 1 mg of total protein) were incubated with antibodies (10 g) (anti-IRS-1, anti-IRS-2, or anti-14-3-3 box I antibodies) for 1 h and precipitated by incubation with 20 l of Protein A Sepharose (Pharmacia) for 1 h. The immunocomplexes were washed five times with the same lysis buffer and boiled in 30 l of Laemmli sample buffer containing 10 mM dithiothreitol. The proteins were resolved on 7.5% or 10% SDS-polyacrylamide gel electrophoresis (PAGE) and transferred to nitrocellulose membranes (Schleicher & Schuell). After blocking with TBS-T (10 mM Tris, pH 7.6, 150 mM NaCl, and 0.05% Tween 20) containing 3% bovine serum albumin for 1 h, the membranes were incubated with the appropriate antibodies. The proteins were visualized by enhanced chemiluminescence using horseradish peroxidaselabeled anti-rabbit or mouse IgG (Amersham). In some experiments, the band intensities were quantified with a Molecular Imager GS-525 using Screen-CH (Bio-Rad).
Two-dimensional Electrophoresis-Bovine brain tissue was homogenized in 5 volumes of 25 mM Tris, pH 7.5, and the extract was obtained by centrifugation for 30 min at 15,000 ϫ g. The extract (250 l, approximately 2 mg of protein) was incubated with anti-14-3-3 box I antibody (20 g) or with control rabbit IgG, and the immunocomplexes were precipitated by incubation with protein A-Sepharose. The immunoprecipitates were dissolved in 8 M urea and 1% Nonidet P-40 and separated by micro two-dimensional PAGE. The electrophoresis was performed by one-dimensional isoelectric focusing on a prefocused 1 mm ϫ 40-mm tube gel containing 8 M urea, 4% acrylamide, 1% Nonidet P-40, and 2.5% carrier ampholytes (Ampholines, pH 3.5-10/pH 5-7 ϭ 1/1, LKB) followed by two-dimensional electrophoresis on a polyacrylamide gradient (8 -17%) slab gel (40 ϫ 40 mm). The proteins were transferred to nitrocellulose membranes and visualized by blotting with anti-IRS-1 antibody and subsequently by enhanced chemiluminescence.

Binding of IRS-1 and Its Truncated Mutants in Cell Lysates with GST-14-3-3 Fusion Protein-The
HepG2 cells overexpressing IRS-1, IRS-1 PHϩPTB , IRS-1 PTB , and LacZ were solubilized with ice-cold phosphate-buffered saline (2 ml/10-cm dish of cells) containing 1% Nonidet P-40, 1 mM phenylmethylsulfonyl fluoride, 100 mM sodium fluoride, and 2 mM sodium orthovanadate. After removing the insoluble materials by centrifugation at 15,000 ϫ g for 10 min at 4°C, 1 ml of the cell lysates was incubated with 500 ng of GST-14-3-3 protein for 1 h. Then, the Glutathione-Sepharose 4B beads were added; 1 h later the beads were washed five times with the solubilizing buffer. The beads were boiled in Laemmli buffer, and the bound proteins were analyzed by SDS-PAGE and subsequent immunoblotting with anti-IRS-1 antibody.
Peptide Competition Experiments-Phosphorylated and nonphosphorylated serine-containing peptides were purchased from Research Genetics, Inc. The peptide sequences were LPKINRSA , and their nonphosphorylated serine-containing counterparts. All peptides were shown to consist of a single peak of the correct molecular weight on mass spectrometry. For peptide competition experiments, the cell lysates of L6 cells were preincubated with one of each of these peptides (1 mM) for 5 min, or no peptide, followed by the addition of GST-14-3-3 protein (500 ng). After a 1-h incubation at room temperature, the Glutathione-Sepharose 4B beads were added, and 1 h later the beads were washed five times with the solubilizing buffer. The beads were boiled in Laemmli buffer, and the bound proteins were analyzed by SDS-PAGE followed by immunoblotting with anti-IRS-1 antibody.
Phosphoamino Acid Analysis of 32 P-Labeled IRS-1-HepG2 cells overexpressing IRS-1 were incubated in phosphate-free Dulbecco's modified Eagle's medium for 6 h, serum deprived for 3 h, and then labeled with [ 32 P]orthophosphate (1 mCi/10 6 cells) for 3 h. Cells were stimulated with 10 Ϫ6 M insulin for 30 min and then lysed in 20 mM Tris, pH 7.4, 150 mM NaCl, 2 mM EDTA, 1% Nonidet P-40, containing 50 mM sodium fluoride, 10 g/ml aprotinin, 10 g/ml leupeptin, 1 mM sodium orthovanadate, and 1 mM phenylmethylsulfonyl fluoride. Lysates were immunoprecipitated with anti-IRS-1 or anti-14-3-3 box I antibodies. Phosphoamino acid analysis was performed as described by Boyle et al. (25). Briefly, the 32 P-labeled proteins were separated by SDS-PAGE and then transferred to polyvinylidine difluoride membranes. The membrane slices containing IRS-1 bands were excised, and IRS-1 proteins were hydrolyzed in 6 N HCl at 110°C for 90 min. The hydrolyzed proteins were dried under a vacuum and resolved in 5 l of H 2 O. Cold phosphoserine, phosphothreonine, and phosphotyrosine were added to 1-l samples that were applied together onto a cellulose-precoated thin layer chromatography plate (Merck) and separated by electrophoresis utilizing an NA-4000 apparatus (Nihon-Eido, Tokyo) with pH 3.5 buffer containing 5% acetic acid and 0.5% pyridine. The location of the phosphoamino acids was determined by ninhydrin staining and autoradiography. The radioactivities were quantified using Bioimage analyzer BAS2000 (Fuji).

Expression Cloning of 14-3-3 Protein and Its Association with
IRS-1-We have employed an expression cloning method using 32 P-labeled recombinant IRS-1 as a probe to isolate cDNAs coding for the proteins that bind to IRS-1. In this study, 21 positive clones were isolated after three or four rounds of screening from a human heart cDNA library. Sequence analysis revealed that 7 of the 21 cDNA clones encode 14-3-3 protein ⑀ or isoforms. To confirm the in vivo binding of 14-3-3 protein with IRS-1, IRS-1 was overexpressed in L6 myotubes, HepG2 hepatoma cells, and CHO cells using an adenovirus expression system (Fig. 1A, upper panel). The cell lysates of these cells overexpressing either IRS-1 or LacZ (control), were immunoprecipitated with anti-IRS-1 antibody and immunoblotted with anti-14-3-3 antibody. As shown in Fig. 1A, lower panel, the amounts of coimmunoprecipitated endogenous 14-3-3 proteins were increased markedly in the IRS-1-overexpressing cells compared with the control cells expressing lacZ, in all cell lines. In L6 and HepG2 cells, two bands were observed and were thought to represent different isoforms of 14-3-3 protein. Based upon the 14-3-3 protein band intensities in the lysates and the anti-IRS-1 immunoprecipitates and the relative amounts of each loaded, approximately 1% of endogenous 14-3-3 protein was immunoprecipitated with the IRS-1.
In addition, 14-3-3 protein either alone or in combination with IRS-1 or IRS-2 was overexpressed in Sf 9 cells using a baculovirus expression system (Fig. 1B, upper and middle panels). The 14-3-3 protein overexpressed was shown to be present in the anti-IRS-1 and anti-IRS-2 antibody immunoprecipitates from the cell lysates overexpressing both 14-3-3 protein and either IRS-1 or IRS-2, but not from those overexpressing only 14-3-3 protein (Fig. 1B, lower panel). IRS-1 and IRS-2 share a similar structure (3), and our results indicate that both IRS-1 and IRS-2 can form a complex with 14-3-3 protein in a variety of cultured cells.
To demonstrate that the association of 14-3-3 protein with IRS-1 is not limited to cultured cells but rather is also present in animal tissue, the bovine brain lysate was immunoprecipitated with the antibody against the 14-3-3 box I region (see "Experimental Procedures"). The immunoprecipitates were separated by micro two-dimensional PAGE and visualized by immunoblotting with anti-IRS-1 antibody. As shown in Fig. 2, the anti-14-3-3 box I antibody immunoprecipitate contained IRS-1 (left panel, indicated by an arrow), whereas the control IgG immunoprecipitate did not (right panel). In addition, in bovine brain tissue, 14-3-3 protein was shown to associate with IRS-1. These results suggest that IRS-1 exists as a complex form with 14-3-3 protein in all and most cell lines or tissues.
The Association of 14-3-3 Protein with IRS-1 Was Unaffected by Insulin Stimulation but Increased by Okadaic Acid Treatment-Next, we investigated which factors regulate the association of 14-3-3 protein with IRS-1. Because 14-3-3 protein contains no SH2 domain (10), we speculated that the tyrosine phosphorylation of IRS-1 is not necessary for the association with these 14-3-3 proteins. Indeed, recent reports have shown that members of the 14-3-3 protein family bind to the specific phosphoserine-containing motif (20,24). It has been shown that IRS-1 is heavily phosphorylated on serine and threonine residues even in the basal condition (26), as demonstrated by treatment with alkaline phosphatase, which reduced the molecular mass of IRS-1 on the SDS-PAGE by 30 -40 kDa (data not shown). In addition to basal phosphorylation, the increase in serine phosphorylation of IRS-1 which is induced by insulin stimulation occurs more slowly and is less marked than the increase in tyrosine phosphorylation (26). First, we investigated the effect of insulin stimulation on the association of 14-3-3 protein with IRS-1 at various periods after the addition . However, no significant difference was observed in the amount of 14-3-3 protein associated with IRS-1 for the initial 30 min after the addition of insulin. These results suggest that the tyrosine phosphorylation, as well as the serine/threonine phosphorylation, of IRS-1 induced by insulin stimulation did not affect the association of 14-3-3 protein with IRS-1.
Treatment with okadaic acid, a potent cell-permeable serine/ threonine phosphatase inhibitor, also increases the serine phosphorylation of IRS-1 and reportedly induces the suppression of tyrosine phosphorylation of IRS-1 by insulin receptor kinase (27). We incubated HepG2 cells overexpressing IRS-1 with 100 nM okadaic acid. In contrast to the case of insulin stimulation, treatment with okadaic acid was shown to increase the association of 14-3-3 protein with IRS-1 by 80% (Fig.  3B, lower panel). The lysates from the cells treated with okadaic acid were revealed to contain larger amounts of IRS-1 capable of binding to GST-14-3-3 compared with control cell lysates (Fig. 3B, middle panel), with no alteration of the total IRS-1 content in the cells (Fig. 3B, upper panel). These results indicate that the serine residues in IRS-1, the phosphorylation level of which is increased by okadaic acid but not by insulin stimulation, play a critical role in the association with 14-3-3 protein.
Binding Sites of IRS-1 with 14-3-3 Protein-A previous study showed that 14-3-3 proteins bind to phosphorylated serine residues within the consensus Arg-Ser-Xaa-pSer-Xaa-Pro (where Xaa is any amino acid) (24). Although this motif has been shown to be present in various proteins bound to 14-3- is the serine residue corresponding to phosphoserine in the 14-3-3 binding motif). To investigate whether these sequences are potential 14-3-3 binding sites, we prepared phosphorylated or nonphosphorylated serine-containing peptides corresponding to the sequences around Ser-270, Ser-374, and Ser-641 in IRS-1, as described under "Experimental Proce-dures." We utilized the peptide around Ser-621 in Raf-1, which was shown to inhibit 14-3-3 binding (24), as a positive control.
We examined whether these peptides inhibit the binding of IRS-1 and GST-14-3-3. As shown in Fig. 4A, all three phosphoserine-containing peptides inhibited the in vitro binding of GST-14-3-3 with IRS-1 to an extent similar to that observed for Raf-1 phosphopeptide (upper panel), whereas none of the corresponding nonphosphorylated peptides affected this association (lower panel). The results of the peptide competition experiments suggested that these three phosphoserinecontaining motifs in IRS-1 can bind to the same COOHterminal acidic portion of 14-3-3 protein to which Raf-1 binds (22,28,29). The inhibition produced by the phosphoserine peptide corresponding to Ser-374 was weaker than those of the other phosphoserine peptides, suggesting that Ser-374 may have a lower affinity for 14-3-3 than Ser-270 and Ser-641. Recently, Zha et al. (20) suggested that RSXSXP and overlapping RXRXXS motifs found in Ser-259 in Raf-1 and Ser-112 and Ser-136 in BAD exhibit a high capacity for binding with 14-3-3 proteins. Among the three 14-3-3 binding motifs of IRS-1, only the motif around Ser-270 possesses the overlapping 14-3-3 binding motifs, although the two proline residues upstream from phosphoserine are replaced with serine.
Ser-270 is within the PTB domain of IRS-1, which plays an important role in the association with the insulin receptor. Reportedly, the NPXY motif including Tyr-960 in the juxtamembrane region of the insulin receptor couples with the PTB domain of IRS-1, resulting in the tyrosine phosphorylation of IRS-1 (30). Thus, binding of the PTB domain of IRS-1 with another molecule may interrupt the association of IRS-1 with the insulin receptor, possibly functioning as a negative regulation mechanism for insulin signaling by suppressing the tyrosine phosphorylation of IRS-1. To confirm the association of the PTB domain of IRS-1 with 14-3-3 in vivo, we overexpressed full-length IRS-1 and its truncated mutants containing the PH and the PTB domains (IRS-1 PHϩPTB ) or only the PTB domain (IRS-1 PTB ) in HepG2 cells, using the adenovirus expression system. Immunoblotting of the cell lysates with the antiserum against the whole IRS-1 molecule showed bands of 180, ϳ55, and ϳ35 kDa corresponding to whole IRS-1, IRS-1 PHϩPTB , and IRS-1 PTB , respectively. Fig. 4B shows that GST-14-3-3 binds to not only IRS-1 but also IRS-1 PHϩPTB and IRS-1 PTB , whereas GST alone binds neither IRS-1 nor its truncated mutants. Based upon the band intensities in the lysates and the GST-14-3-3 precipitates and the relative amount of each loaded, approximately 6% of IRS-1 and IRS-1 PHϩPTB and 30% of IRS-1 PTB were bound to GST-14-3-3, respectively, indicating that the PTB domain may be the main portion binding to 14-3-3 protein. Based on the above results, the following conclusions can be drawn. (i) 14-3-3 protein binds to IRS-1 and IRS-2 in various cell lines and tissues in a phosphoserine-dependent manner, irrespective of tyrosine phosphorylation states; (ii) IRS-1 possesses three putative 14-3-3 binding motifs, one of which is present in the PTB domain and overlaps the 14-3-3 binding motif and associates with 14-3-3 protein in vivo.
Phosphorylation State of IRS-1 Associated with 14-3-3 Protein-Finally, we performed an experiment to investigate the effect of the association of 14-3-3 proteins on the function of IRS-1. To date, numerous biological activities have been attributed to 14-3-3 proteins. They were first suggested to be activa-tors of tyrosine/tryptophan hydroxylase (10,11) and then identified as regulators of protein kinase C (12). In addition, 14-3-3 proteins are required cofactors for the bacterial toxin of Pseudomonas, an ADP-ribosylase known as exoenzyme S (14), and have also been implicated in cell cycle control (15). More recently, 14-3-3 proteins were found to interact with several important molecules that modulate signal transduction pathways, including Raf-1 (16,17), polyoma middle T antigen (18), Bcr (19), and the cell death agonist BAD (20). In these cases, 14-3-3 proteins do not transmit the signal directly, but rather they modulate the functions of associated proteins. The roles of IRS-1 and IRS-2 are thought to involve signal transmission from the insulin receptor downstream to several SH2 proteins via their tyrosine-phosphorylated binding sites. Thus, we investigated the effect of 14-3-3 protein association with respect to the insulin-induced phosphorylation of IRS-1. To do so, we compared the tyrosine and serine phosphorylation levels of IRS-1 immunoprecipitated with anti-IRS-1 antibody with those coimmunoprecipitated with the anti-14-3-3 box I antibody. First, tyrosine phosphorylation levels of the IRS-1 were compared (Fig. 5A). Judging from quantitation of the band inten- , and Ser-641 (peptide 3) in IRS-1 and the corresponding nonphosphorylated serine-containing peptides (indicated by phosphoserine Ϫ) were purchased. For peptide competition experiments, the cell lysates of L6 cells were preincubated with one of these peptides (1 mM) or no peptide for 5 min followed by the incubation with GST-14-3-3 (500 ng) or GST alone and the Glutathione-Sepharose 4B beads. The beads were boiled in Laemmli buffer, and the bound proteins were analyzed by SDS-PAGE and immunoblotting with anti-IRS-1 antibody. Panel B, HepG2 cells overexpressing fulllength IRS-1 and IRS-1 truncated mutants (see "Experimental Procedures") were solubilized, and 1 ml of the cell lysates was incubated with GST-14-3-3 or GST alone and the Glutathione-Sepharose 4B beads. The cell lysates and the proteins bound to these beads were subjected to SDS-PAGE and immunoblotted with anti-IRS-1 antibody. Lanes 1, 4, and 7, full-length IRS-1; lanes 2, 5, and 8, IRS-1 PHϩPTB ; lanes 3, 6, and 9, IRS-1 PTB in cell lysate (2% of a 10-cm plate) (lanes 1-3), bound to GST-14-3-3 (derived from 30% of a 10-cm plate) (lanes 4 -6), and bound to GST alone (lanes 7-9). The data are representative of three or five independent experiments. sities and the ratios of phosphotyrosine to IRS-1 on the blots, insulin-induced tyrosine phosphorylation of IRS-1 was increased approximately 4.5-fold in anti-IRS-1 immunoprecipitates but was increased by only 12% in that with anti-14-3-3 antibody. These results suggest that the IRS-1 coimmunoprecipitated with the anti-14-3-3 box I antibody was more weakly tyrosine phosphorylated with insulin stimulation than the IRS-1 immunoprecipitated with anti-IRS-1 antibody. Based upon the IRS-1 band intensities in the immunoprecipitates with anti-IRS-1 and anti-14-3-3 antibody and the relative amounts of each loaded, the IRS-1 coimmunoprecipitated with anti-14-3-3 antibody amounted to about 5% of the IRS-1 immunoprecipitated with anti-IRS-1 antibody. Because anti-IRS-1 antibody immunoprecipitates about 80% of IRS-1, as assessed by immunoblotting the lysate before and after immunoprecipitation (data not shown), IRS-1 coimmunoprecipitated with anti-14-3-3 antibody amounted to approximately 4% of total IRS-1.
In addition, employing phosphoamino acid analysis, the incorporations of 32 P into serine and tyrosine residues were compared between the IRS-1 in the anti-IRS-1 antibody immunoprecipitate and the IRS-1 in the anti-14-3-3 box I antibody immunoprecipitate. HepG2 cells overexpressing IRS-1 were labeled with [ 32 P]orthophosphate as described under "Experimental Procedures." The cells were solubilized, and the lysate was immunoprecipitated using anti-IRS-1 or anti-14-3-3 box I antibodies. As shown in the upper panel of Fig. 5B, the bands corresponding to 32 P-labeled IRS-1 were clearly observed in the immunoprecipitates obtained with the anti-IRS-1 antibodies and with the anti-14-3-3 box I antibody. The IRS-1 proteins in these bands were hydrolyzed and subjected to phosphoamino acid analysis (Fig. 5B, lower panel). The quantification of the incorporated 32 P revealed insulin-induced tyrosine and serine phosphorylation of IRS-1 immunoprecipitated with anti-IRS-1 antibody and with anti-14-3-3 box I antibody (Fig. 5C). The incorporations of 32 P into the tyrosine and serine residues of IRS-1 in the anti-IRS-1 antibody immunoprecipitate were increased to 2.59-and 1.95-fold with insulin stimulation, respectively, whereas those in the anti-14-3-3 box I antibody immunoprecipitates were increased by only 14 and 2%, respectively. In these experiments, the incorporation of 32 P into the threonine residue was below detectable limits. These results show that the small percentage of IRS-1 (5%) associated with the small percentage of 14-3-3 protein immunoprecipitated (1-2%) is less highly phosphorylated, suggesting that 14-3-3 protein has a higher affinity for less highly phosphorylated IRS-1. If such is the case, then association of 14-3-3 protein with IRS-1 may inhibit its phosphorylation on tyrosine and serine in response to insulin. We speculate that the most plausible explanation for reduced phosphorylation of IRS-1 is that the 14-3-3 protein binds to the PTB domain, thereby interrupting the optimal association between IRS-1 and the insulin receptor. Although the amount of 14-3-3 protein which associated with IRS-1 was unchanged after compared with before insulin stimulation (Fig. 3A), IRS-1 that was not associated with 14-3-3 protein was more highly phosphorylated than that which was associated with 14-3-3 protein, and total IRS-1 phosphorylation was increased, described in Fig. 5C.
In the insulin signaling, 14-3-3 protein may function as a negative regulator. Another reported example of the negative regulatory function of 14-3-3 protein is its effect on spinach leaf nitrate reductase (31). In this case, 14-3-3 proteins bind to a regulatory phosphoserine of nitrate reductase and thereby inactivate the enzyme. DISCUSSION Recently, extensive studies have been done to clarify the mechanism of insulin resistance, which is involved in the pathogenesis of non-insulin-dependent diabetes mellitus, as well as the factors determining insulin sensitivity. A number of reports have demonstrated that serine phosphorylation of IRS-1 affects its tyrosine phosphorylation (27,32,33). When the serine phosphorylation of IRS-1 is augmented by tumor necrosis factor-␣, which has been suggested to be a mediator of insulin resistance in obesity, insulin-induced tyrosine phosphorylation of IRS-1 is impaired (32). Tumor necrosis factor-␣ was also shown to induce serine phosphorylation of IRS-1 and to convert IRS-1 into an inhibitor of the insulin receptor tyrosine kinase activity in vitro (33). The 14-3-3 protein association with the serine-phosphorylated IRS-1, and possibly IRS-2, may contribute to the regulation of insulin sensitivity.
In this study, we have not determined whether IRS-1 is actually phosphorylated on the serine residues in the 14-3-3 protein binding motifs. However, if it is assumed that phosphorylation of these motifs is required for the observed association, then the amount of 14-3-3 protein associated with IRS-1 would be determined by the activity of an as yet unknown serine kinase(s) that phosphorylates the serine residue in the 14-3-3 binding motif of the target protein, rather by the level of 14-3-3 protein expression, because these proteins are extremely abundant (reportedly approximately 1% of total brain tissue-soluble proteins (34), probably more than 0.1% of cytosolic protein in most cells). Thus, to identify the serine kinase involved in the 14-3-3 protein association is essential for clarifying the regulatory mechanism of the 14-3-3 protein association with IRS-1. In the case of tyrosine/tryptophan hydroxylase, it has been demonstrated that calmodulin kinase II phosphorylates the 14-3-3 binding motif of hydroxylase (10, 11). 14-3-3 proteins bind to phosphorylated hydroxylase, thereby activating the enzyme (35). In the case of BAD, heart muscle kinase (a form of protein kinase A) was shown to phosphorylate the serine residue in the 14-3-3 binding motif of BAD in vitro, but the kinase acting in vivo remains unknown (20).
We investigated whether PTB domains found in other proteins share a 14-3-3 protein binding motif similar to that of IRS-1 and IRS-2. Sequential alignments of PTB domains identified in various proteins are shown in Fig. 6. The ␣-helices and ␤-sheet strands based on a structural study (36) and the putative 14-3-3 binding motif (around Ser-270) are boxed. Based on crystallographic study, ␤5, ␤6, ␤7, and ␣2 of IRS-1 are important for recognizing phosphotyrosine in the insulin receptor NPXY motif (36). Because the putative 14-3-3 binding motif is next to the ␣2-helix, 14-3-3 protein binding to IRS-1 would presumably influence the binding of its PTB domain to the insulin receptor. Another interesting point is that IRS-1 and IRS-2 contain the insertion in the corresponding regions of other proteins and that the putative 14-3-3 binding motif is in this insertion sequence. Thus, the binding of 14-3-3 protein to the PTB domain may be specific to IRS-1 and IRS-2.
Further study is necessary to elucidate the physiological role of the 14-3-3 protein association with IRS-1 and IRS-2 in insulin signaling and the regulation of insulin sensitivity, as well as its possible involvement in the insulin resistance observed in non-insulin-dependent diabetes mellitus.