Index Introduction Materials and Methods Antibodies Site-directed mutagenesis and construction of expression vectors Expression of wild-type and mutant IRS proteins, dominant-negative proteins, and the human insulin receptor Immunoblotting of phosphotyrosine-containing proteins Detection of IRS proteins associated with the plasma membrane by confocal microscopy Statistical analysis Results The effects of dominant-negative IRS proteins on insulin-induced IRS tyrosine phosphorylation in Cos-1 cells overexpressing the human insulin receptor The effects of dominant-negative IRS proteins on insulin-induced tyrosine phosphorylation of IRS-1 with the mutations disrupting the phosphoinositide recognition motifs in its PH and PTB domains The effects of N-terminal fragment of PHIP on insulin-induced tyrosine phosphorylation of IRS proteins in Cos-1 cells The effects of N-terminal fragments of IRS proteins or PHIP on insulin-induced targeting of IRS-1 to the plasma membrane Discussion Acknowledgements References

Introduction

Insulin initiates its biological effects by binding the cell surface insulin receptor and activating its endogenous tyrosine kinase, which causes tyrosine phosphorylation of endogenous substrates including insulin receptor substrate (IRS) proteins, growth factor receptor-bound protein 2-associated binder 1, p60dok, Cbl, Src homology and collagen, and adaptor protein containing pleckstrin and Src homology 2 domains (White, 2002). Subsequent to tyrosine phosphorylation, these proteins bind various Src homology (SH) 2 domain-containing proteins, such as the regulatory subunits of phosphoinositide (PI) 3-kinase, growth factor receptor-bound protein 2, and SH2 domain-containing tyrosine phosphatase 2, which mediate various biological processes (White, 2002). Four members of the IRS protein family have been cloned, which have a similar structure and contain a pleckstrin homology (PH) domain, which interacts with phospholipids and protein ligands resulting in the localization of IRS to the plasma membrane, a phosphotyrosine binding (PTB) domain which directly binds to the tyrosine-phosphorylated NPXY motif containing Tyr⁹⁶⁰ in the juxtamembrane region of the insulin receptor, and a C-terminal domain with a number of tyrosine phosphorylation sites (White, 2002). Despite the structural similarities between these IRS proteins, analyses of the IRS knockout mice demonstrated that IRS proteins have different function in development and metabolism (Kaburagi et al., 1997; White, 2002; Yamauchi et al., 1996), although the molecular mechanisms of these functional differences among IRS molecules have not been analyzed in detail.

The analyses of the deletion of the PH or PTB domain of IRS proteins have demonstrated that both of these domains are essential for insulin-stimulated tyrosine phosphorylation of IRS proteins (Craparo et al., 1995; Gustafson et al., 1995; He et al., 1995; Myers et al., 1995; Voliovitch et al., 1995; Xu et al., 1999; Yenush et al., 1996). In our previous study, we transiently expressed rat IRS-1, IRS-3, or chimeric proteins of these two IRS proteins in Cos-1 cells and showed that the PH domain of IRS-1 is involved in the wortmannin-sensitive accumulation of IRS-1 to the plasma membrane in response to insulin, while IRS-3 was localized to the plasma membrane via its PH domain in an insulin-independent and wortmannin-resistant manner causing the tyrosine phosphorylation of IRS-3 in a receptor NPXpY motif-independent fashion (Kaburagi et al., 2001). From these data, we proposed a model providing a mechanism for the interaction between the insulin receptor and IRS molecules in which both the PH and PTB domains of IRS proteins cooperatively take part in transmitting the insulin signal from the receptor (Kaburagi et al., 2001).

Phospholipids including PI(3)P, PI(4,5)P₂, and PI(3,4,5)P₃ (Dhe-Paganon et al., 1999; Razzini et al., 2000) have been recently proposed as potential ligands for IRS PH domains. In their studies, the PH domain of IRS-1 preferentially binds to both PI(3,4,5)P₃ and PI(4,5)P₂ (Dhe-Paganon et al., 1999) or only PI(3,4,5)P₃ (Razzini et al., 2000), that of IRS-2 to PI(3,4)P₃ (Razzini et al., 2000) and that of IRS-3 to PI(3)P₃ (Razzini et al., 2000). Moreover, several proteins such as pleckstrin homology domain-interacting protein (PHIP) (Farhang-Fallah et al., 2000) have been reported to be the in vitro ligands for the PH domains of IRS proteins. However, the contributions of these candidate PH targets in insulin-induced cell signaling mediated by each IRS protein in intact cells have not been critically evaluated. In this study, we expressed each IRS protein in the presence of the N-terminal truncated proteins containing the PH/PTB domains of IRS proteins or the N-terminal PH domain-interacting region of PHIP, and analyzed the functional differences in transducing the insulin signals among the members of the IRS family. In addition, to examine the relationship between the subcellular distribution and signaling efficiency of IRS proteins, we evaluated the effects of these N-terminal truncated proteins on the targeting of IRS molecules to the plasma membrane in response to insulin.

Materials and Methods

Antibodies

Rabbit polyclonal antibodies against hemagglutinin (αHA, Y-11) and rabbit polyclonal antibodies against the β subunit of human insulin receptor (αIRβ, C-19) were purchased from Santa Cruz Biotechnology (Santa Cruz, CA, USA), mouse monoclonal anti-FLAG antibody (αFLAG, M2) from Sigma (St Lois, MO, USA) and mouse monoclonal anti-phosphotyrosine antibody (αPY, 4G10) from Upstate Biotechnology (Lake Placid, NY, USA). Tetramethyl rhodamine isothiocyanate (TRITC)-conjugated antibody against rabbit IgG and fluorescein isothiocyanate (FITC)-conjugated antibody against mouse IgG was obtained from Molecular Probes (Eugen, OR, USA).

Site-directed mutagenesis and construction of expression vectors

The pcDNA3 vector containing the rat IRS-1 or IRS-3 cDNA with the C-terminal HA tag was previously described (Lavan et al., 1997; Sun et al., 1991; Yamamoto-Honda et al., 1996; Yamauchi et al., 1998). Using polymerase chain reaction (PCR) (Higuchi et al., 1989; Kadowaki et al., 1989) of mouse genomic DNA containing the IRS-2 gene (a generous gift provided by Dr. Naoto Kubota) as a template (Kubota et al., 2000), cDNA of mouse IRS-2 (Sun et al., 1995) with the C-terminal HA tag was synthesized. Each IRS cDNA containing a 3'-terminal HA tag sequence was inserted into pEF4/V5-His (Invitrogen, Carlsbad, CA, USA) (pEF4-IRS1, pEF4-IRS2, and pEF4-IRS3, respectively). In addition, the mutations disrupting the affinity for phosphoinositides of the PH domain (R28C) and the PTB domain (K169Q/K171Q/K177Q) of IRS-1 (Jacobs et al., 2001) were introduced into the full length of IRS-1 cDNA by PCR and inserted into the IRS-1 expression vector (pEF4-IRS1-R28C/3KQ) (Fig. 1). The cDNAs encoding the C-terminal FLAG tagged PH and PTB domain-containing region of IRS-1 (amino acid 1–270), IRS-2 (amino acid 1–293), or IRS-3 (amino acid 1–284) were also synthesized using PCR, and inserted into pcDNA3.1 (+) (Invitrogen) (pcDNA3.1-IRS1-N, pcDNA3.1-IRS2-N, and pcDNA3.1-IRS3-N, respectively). Human PHIP cDNA (Farhang-Fallah et al., 2000) was generated from human muscle cDNA (Clontech, Palo Alto, CA, USA) by PCR. Triple tandem FLAG tag sequence was inserted into the N-terminus of human PHIP cDNA containing the region interacting with the PH domain of IRS-1 (amino acid 1–209) (Kaburagi et al., 2001) and the expression vector (pcDNA3.1 (–), Invitrogen) containing the cDNA was constructed (pcDNA3.1-PHIP-N).

View Details

Fig. 1. Effects of the N-terminal IRS proteins on insulin-induced IRS tyrosine phosphorylation and Akt activation phosphorylation in Cos-1 cells. Panel A, tyrosine phosphorylation of IRS-1. Panel B, tyrosine phosphorylation of IRS-2. Panel C, tyrosine phosphorylation of IRS-3. After treatment with insulin, total lysates from Cos-1 cells expressing intact IRS, the IRS-N proteins, and the insulin receptor were immunoprecipitated with αHA, subjected to SDS-PAGE followed by immunoblotting with αPY. Cell lysates were also subjected to SDS-PAGE followed by immunoblotting with αHA, αFLAG or αIRβ. The bands corresponding to IRS proteins, IRS-N proteins and the β subunit of insulin receptor (IRβ) are indicated. Tyrosine-phosphorylated IRS proteins were quantified by scanning densitometry. Data are the means±S.E. from three independent experiments. The statistical significance was assessed in comparison with the data from insulin-treated cells expressing each IRS without IRS-N proteins. *, p<0.05 versus control; **, p<0.01 versus control; ***, p<0.001 versus control.

Expression of wild-type and mutant IRS proteins, dominant-negative proteins, and the human insulin receptor

Subconfluent Cos-1 cells in 6-cm dishes were transfected with 1 μg/dish of a plasmid containing the full length IRS cDNA, 1 μg/dish of the wild-type human insulin receptor (A-type) expression vector (pSV2-IR) (Kaburagi et al., 1993; Ullrich et al., 1985), and the indicated amount of a plasmid containing the cDNA of either the PH/PTB domain of IRS proteins or the full length or the PH domain-interacting region of PHIP using lipofectamine plus reagent (Invitrogen). After 24 h, the transfected cells were incubated in serum-free medium for 12–16 h, and subjected to the experiments described below.

Immunoblotting of phosphotyrosine-containing proteins

Assays were performed as described (Kaburagi et al., 1995). Briefly, Cos-1 cells treated or untreated with 10^–7 M insulin for 1 min at 37°C were lysed in buffer A (25 mM Tris-HCl; pH 7.4, 2 mM sodium orthovanadate, 10 mM NaF, 10 mM sodium pyrophosphate, 1 mM phenylmethylsulfonyl fluoride, 1 mM EGTA, and 1 mM EDTA) containing 1% Triton X-100. Cell lysates containing equal amounts of protein were immunoprecipitated with αHA, subjected to SDS-PAGE followed by Western blotting with αPY, and visualized using ECL plus Western blotting detection system (Amersham Biosciences, Piscataway, NJ, USA). In some experiments, to evaluate in vivo binding of IRS proteins with the PHIP fragment, the αHA-immunoprecipitates were also subjected to SDS-PAGE, and followed by immunoblotting with αFLAG. The bands derived from tyrosine-phosphorylated IRS proteins were evaluated by densitometry. To check the expression of overexpressed proteins, total lysates were also subjected to SDS-PAGE followed by Western blotting with αHA, αFLAG or αIRβ.

Detection of IRS proteins associated with the plasma membrane by confocal microscopy

For immunostaining, Cos-1 cells plated onto 35-mm diameter glass base dishes (Iwaki, Tokyo, Japan) were transfected with indicated expression vectors. After 24 h, transfected cells were serum-starved for 12–16 h, and treated or untreated with 10^–7 M insulin for 5 min at 37°C. Then, the cells were washed with ice-cold phosphate-buffered saline (PBS), followed by treatment with PBS containing 4% paraformaldehyde for fixation. The cells were permeabilized by treating with ChemMate^TM antibody diluent (DakoCytomation, Glostrup, Denmark), and were incubated with αHA (1:100) and αFLAG (1:100) at 4°C overnight, washed and incubated with TRITC-conjugated anti-rabbit IgG (1:100) and FITC-conjugated anti- mouse IgG (1:100) at 37°C for 1 h. The stained sections were mounted in Gel/Mount (Biomeda, Foster City, CA) and were visualized and imaged using a scanning laser confocal immunofluorescence microscope (Zeiss, Esslingen, Germany). Transfected cells were selected by staining with αHA and αFLAG, and the plasma membrane targeting of overexpressed IRS were evaluated using confocal microscopy, as previously described (Jacobs et al., 2001). Briefly, cells with no plasma membrane staining with αHA received a score of 0, whereas cells with plasma membrane association of overexpressed IRS received a score of 1 (Fig. 4A). The mean score of 150 cells was calculated, and the data were statistically evaluated from three independent experiments.

Statistical analysis

Where appropriate, data were expressed as the means and standard errors of three independent experiments. Statistical significance was assessed using Student’s t test. A p<0.05 was considered statistically significant.

Results

The effects of dominant-negative IRS proteins on insulin-induced IRS tyrosine phosphorylation in Cos-1 cells overexpressing the human insulin receptor

It has been recently suggested using in vitro binding assays that phosphoinositides are the ligands for the PH domains of IRS proteins (Dhe-Paganon et al., 1999; Razzini et al., 2000), while two-hybrid analyses revealed that several proteins such as nucleolin and PHIP bind IRS PH domains (Burks et al., 1998; Farhang-Fallah et al., 2000). To test the roles of these in vitro ligands for the PH domains of IRS proteins in insulin-induced signal transduction, we firstly assessed the effects of the PH domains on tyrosine phosphorylation of IRS-1, -2, or -3, which revealed that none of the PH domains affected tyrosine phosphorylation of IRS-1~3 (data not shown). As other group previously reported the dominant-negative effects of the N-terminal fragment containing the PH and PTB domains of IRS-1 and IRS-2 on insulin-induced tyrosine phosphorylation of IRS-1 and IRS-2 (Matsumoto et al., 2002), we generated the expression vectors containing cDNAs of these N-terminal fragments around the PH and PTB domains of IRS proteins (IRS-1~3N), expressed these IRS fragments with intact IRS proteins and the human insulin receptor in Cos-1 cells, and evaluated the effects of these fragments on insulin-induced tyrosine phosphorylation of each IRS (Fig. 1). Tyrosine phosphorylation of IRS-1 was significantly impaired by the coexpression of IRS-1N in a dose-dependent fashion (p<0.001) as previously reported (Matsumoto et al., 2002) (Fig. 1A). In addition, we observed that IRS-2N inhibited tyrosine phosphorylation of IRS-1 but the effect of IRS-2N was significantly less than that of IRS-1N (p<0.01), while IRS-3N most efficiently blocked tyrosine phosphorylation of IRS-1 (Fig. 1A). We also assessed the effects of these IRS-N proteins on tyrosine phosphorylation of IRS-2 and IRS-3 in response to insulin. Although in vitro binding studies verified that the PH domain of each IRS interacts with its peculiar phosphoinositide (Razzini et al., 2000), every dominant-negative IRS fragment containing the PH domain significantly inhibited tyrosine phosphorylation of IRS-2 and IRS-3. Among the dominant-negative IRS fragments, IRS3-N most effectively suppressed tyrosine phosphorylation of both IRS-2 and IRS-3, whereas IRS1-N and IRS-2-N were less effective (Fig. 1B & 1C). Different from IRS-1, the inhibition of IRS-2 phosphorylation via IRS-2N appeared to be significantly more potent than that via IRS-1N at least at a 3 μg-vector dosage (Fig. 1B). These findings demonstrate that the signal transduction of each IRS protein is inhibited not only by its own N-terminal truncated fragment but also by other IRS-N fragments, indicating that IRS-mediated signaling may not be specifically regulated by the affinities of the PH domains to phosphoinositides.

The effects of dominant-negative IRS proteins on insulin-induced tyrosine phosphorylation of IRS-1 with the mutations disrupting the phosphoinositide recognition motifs in its PH and PTB domains

To examine whether the affinity to phosphoinositides is essential for tyrosine phosphorylation of IRS proteins, we compared tyrosine phosphorylation of mutant IRS-1 disrupting the phosphoinositide recognition motifs (IRS-1 R28C/3KQ) (Jacobs et al., 2001) with that of normal IRS-1 in the presence or absence of IRS-N proteins (Fig. 2). Even in the absence of IRS-N proteins, the R28C/3KQ mutations of IRS-1 significantly reduced but did not totally abrogate the tyrosine phosphorylation of IRS-1, indicating that the association with phosphoinositides is not essential for IRS-1-mediated signaling (Fig. 2A). When IRS-N proteins were coexpressed with IRS-1 R28C/3KQ, these dominant-negative IRS proteins further inhibited insulin-induced tyrosine phosphorylation of the mutant IRS-1 in a manner similar to that of wild-type IRS-1 (Fig. 2A & 2B). These results suggest that the interactions with phosphoinositides is not indispensable for tyrosine phosphorylation of IRS proteins and that there may exist molecular targets of the PH domains other than phosphoinositides, which are common to all the IRS proteins analyzed.

View Details

Fig. 2. Effects of the phosphoinositide affinity-disrupting mutations (R28C/3KQ) of IRS-1 on insulin-induced IRS tyrosine phosphorylation in Cos-1 cells expressing the N-terminal IRS proteins. Panel A, the effects of IRS-1N. Panel B, the effects of IRS-2N and -3N. After treatment with insulin, total lysates from Cos-1 cells expressing intact IRS with or without the phosphoinositide affinity-disrupting mutations (R28C/3KQ), the IRS-N proteins and the insulin receptors were immunoprecipitated with αHA, subjected to SDS-PAGE, followed by immunoblotting with αPY. Cell lysates were also subjected to SDS-PAGE followed by Western blotting with αHA, αFLAG or αIRβ. The bands corresponding to IRS-1, IRS-N proteins and the β subunit of insulin receptor (IRβ) are indicated. Tyrosine-phosphorylated IRS proteins were quantified by scanning densitometry. Data are the means±S.E. from three independent experiments. When unindicated, the statistical significance was assessed in comparison with the data from insulin-treated cells expressing normal IRS-1 without IRS-N proteins. *, p<0.05 versus control; **, p<0.01 versus control; ***, p<0.001 versus control.

The effects of N-terminal fragment of PHIP on insulin-induced tyrosine phosphorylation of IRS proteins in Cos-1 cells

Two-hybrid analysis recently revealed that PHIP binds the PH domains of IRS-1 and IRS-2, and that the expression of the PH-binding region of PHIP impairs tyrosine phosphorylation of IRS-1 leading to a marked attenuation of insulin actions such as mitogenesis and Glut4 translocation (Farhang-Fallah et al., 2000; Farhang-Fallah et al., 2002). To test whether the expression of the PH-binding region of PHIP affects tyrosine phosphorylation of each IRS molecule, we expressed IRS proteins, the mutant PHIP containing the N-terminal region interacting with the PH domain of IRS-1 (PHIP-N) and the human insulin receptor in Cos-1 cells, and evaluated the dominant-negative effects of this truncated proteins on insulin-induced tyrosine phosphorylation of each IRS (Fig. 3). Whereas previous studies reported that the expression of the PH-binding region of PHIP suppressed tyrosine phosphorylation of IRS-1 and IRS-2 (Farhang-Fallah et al., 2000; Farhang-Fallah et al., 2002), tyrosine phosphorylation of either IRS molecule was not significantly affected by the PHIP-N expression in our analysis (Fig. 3). In addition, tyrosine phosphorylation of IRS-3 in response to insulin was not impaired by the PHIP fragment expression (Fig. 3). These findings suggest that PHIP may not play a major role in IRS-mediated signaling through either IRS proteins, raising the possibility that there may exist membrane ligands of IRS molecules other than PHIP.

View Details

Fig. 3. Effects of the PHIP N-terminus on insulin-induced IRS tyrosine phosphorylation in Cos-1 cells. After treatment with insulin, total lysates from Cos-1 cells expressing IRS and the PH-interacting region of PHIP with the insulin receptors were immunoprecipitated with αHA, subjected to SDS-PAGE, and followed by immunoblotting with αPY. Cell lysates were also subjected to SDS-PAGE followed by Western blotting with αHA, αFLAG or αIRβ. The bands corresponding to IRS-1, PHIP and the β subunit of insulin receptor (IRβ) are indicated. The amount of tyrosine-phosphorylated IRS proteins was evaluated by scanning densitometry. Data are the means±S.E. from three independent experiments.

The effects of N-terminal fragments of IRS proteins or PHIP on insulin-induced targeting of IRS-1 to the plasma membrane

To evaluate the effects of the N-terminal fragments of IRS proteins or PHIP on the plasma membrane targeting of IRS-1 in response to insulin, IRS-1, dominant-negative proteins and the insulin receptor were expressed in Cos-1 cells. IRS-1 targeted to the plasma membrane was detected using immunostaining with αHA (Fig. 4A), and was evaluated by confocal microscopy as previously described (Jacobs et al., 2001) (Fig. 4B). In the absence of dominant-negative proteins, IRS-1 localized to the plasma membrane was significantly increased from 12.2±2.4% to 35.5±1.7% in response to insulin (p<0.001) (Fig. 4B). When IRS-1N was coexpressed, the percentage of IRS-1 localized to the plasma membrane in insulin-treated cells was significantly reduced to 23.3±2.3% (p<0.001) (Fig. 4B). Likewise, IRS-2N significantly reduced the percentage of IRS-1 localized to the plasma membrane in insulin-treated cells comparable to IRS-1N (25.5±2.5%, p<0.01) (Fig. 4B). However, the effect of IRS-3N expression was maximal among these IRS-N proteins, because IRS-1 localized to the plasma membrane in insulin-treated cells was significantly reduced to 15.0±0.9% (p<0.001) (Fig. 4B). The introduction of the mutations disrupting the affinity for phosphoinositides into IRS-1 (IRS-1 R28C/3KQ) significantly decreased the percentage of IRS-1 localized to the plasma membrane in insulin-treated cells (29.3±4.7%, p<0.05), although insulin-induced translocation to the plasma membrane was preserved (Fig. 4B). Even in the presence of insulin, coexpression of IRS-1N significantly decreased the cell surface distribution of IRS-1 R28C/3KQ to 16.5±1.7% (p<0.01), indicating that there may be ligands other than phosphoinositides playing a role in the insulin-induced redistribution of IRS-1 (Fig. 4B). We also showed that the PHIP-N did not affect insulin-induced localization of IRS-1 to the plasma membrane (31.3±4.0%) (Fig. 4B) indicating that PHIP is not essential for the insulin-induced membrane targeting of IRS-1, although the possibility that PHIP may play other functional roles in insulin signaling cannot be excluded. These results suggest that insulin-induced tyrosine phosphorylation and the cell surface targeting of IRS proteins are regulated in a similar manner, and that there may exist molecules other than phosphoinositides or PHIP causing the translocation of IRS proteins to the plasma membrane thereby transducing insulin-induced signals.

View Details

Fig. 4. Effects of IRS-N and PHIP-N on insulin-induced targeting of IRS-1 targeted to the plasma membrane in Cos-1 cells expressing the human insulin receptor. Panel A, The typical pattern of cells with or without the plasma membrane association of overexpressed IRS-1. Panel B, The score of plasma membrane association of overexpressed IRS-1. Cos-1 cells transiently expressing wild-type or R28C/3KQ IRS-1 and the insulin receptors with IRS-1N, IRS-2N, IRS-3-N, or PHIP-N were stimulated with insulin. Overexpressed IRS proteins localized to the plasma membrane were detected using confocal microscopy as described under “Materials and Methods”. One hundred and fifty cells per condition were scored for targeting to the plasma membrane using confocal microscopy as described under “Materials and Methods”. The data were statistically evaluated from three independent experiments, and shown as the means±S.E. from three independent experiments. When unindicated, the statistical significance was assessed in comparison with the data from insulin-treated cells expressing normal IRS-1 without IRS-N proteins. *, p<0.05 versus control; **, p<0.01 versus control; ***, p<0.001 versus control.

Discussion

In the present study, we analyzed the functional roles in which the N-terminal regions of IRS proteins play by the overexpression of the PH and PTB domain fragments of IRS proteins or the N-terminal PH-interacting region of PHIP. Although the analysis of the PH domain deletion verified that this domain is essential for insulin-induced tyrosine phosphorylation of IRS proteins (Myers et al., 1995; Voliovitch et al., 1995; Xu et al., 1999; Yenush et al., 1996), we observed that overexpression of the isolated PH domains did not affect tyrosine phosphorylation of IRS proteins (data not shown). In contrast, as another group has previously reported (Matsumoto et al., 2002), we also demonstrated the dominant-negative effects of the N-terminal fragment containing the PH and PTB domains of IRS proteins on insulin-induced tyrosine phosphorylation of IRS-1, IRS-2 or IRS-3 (Fig. 1), which may suggest that the cooperation of the PH domain binding to its ligands in the plasma membrane and the PTB domain interacting with the insulin receptor may also be required for the dominant-negative effects of the IRS N-termini as similarly required for efficient tyrosine phosphorylation of IRS proteins (Kaburagi et al., 2001).

A previous study showed that both PI(3,4,5)P₃ and PI(4,5)P₂ are in vitro ligands for the PH domain of IRS-1 (Dhe-Paganon et al., 1999). If so, IRS-1 should not be translocated to the plasma membrane in response to insulin in a PI 3-kinase inhibitor-sensitive manner as we reported (Fig. 4) (Kaburagi et al., 2001) but rather targeted to the plasma membrane irrespective of the insulin treatment, because the abundance of PI(4,5)P₂ is much greater than PI(3,4,5)P₃ even in stimulated cells (Czech, 2000). Another group reported using in vitro binding assay that the PH domain of IRS-1 preferentially binds to PI(3,4,5)P₃, whereas the PH domain of IRS-2 and that of IRS-3 bind to PI(3,4)P₂ and PI(3)P, respectively (Razzini et al., 2000). Although the proposed ligands for the PH domains of IRS-1 and IRS-2 are consistent with the PI 3-kinase-dependent translocation of these IRS molecules in response to insulin (Kaburagi et al., 2001; Razzini et al., 2000), the insulin-insensitive distribution of IRS-3 in the plasma membrane may not be consistent with the affinity of its PH domain to PI(3)P because the FYVE domain, which preferentially binds to PI(3)P, was shown to be highly enriched in endosomes but not in the plasma membrane (Gillooly et al., 2000). Recently, it was reported that insulin induced PI(3)P formation through the activation of a small G protein TC10 at the membrane raft in the cell surface (Maffucci et al., 2003a), which may take part in the targeting of IRS-3 to the plasma membrane. However, as insulin-independent localization of IRS-3 to the cell surface, as well as distribution in a diffuse pattern but not in a spotted distribution indicative of targeting to the raft in the plasma membrane (Fig. 4) (Kaburagi et al., 2001; Maffucci et al., 2003b), indicates that the major ligand for the PH domain of IRS-3 may not be PI(3)P, further studies are required to identify the physiological targets of the PH domain of IRS proteins.

Regardless of these different affinities to phosphoinositides, we also analyzed IRS-1 with the mutations of the PH and PTB domains disrupting their affinity for phosphoinositides (R28C/3KQ) (Jacobs et al., 2001). The R28C/3KQ mutations of IRS-1 significantly reduced but did not completely abrogate the tyrosine phosphorylation of IRS-1, which was further suppressed by the expression of all the IRS-N proteins examined (Fig. 2A & 2B). In addition, partially impaired translocation of IRS-1 R28C/3KQ to the plasma membrane induced by insulin was further suppressed by the coexpression of IRS1-N (Fig. 4). Razzini et al. previously showed that the PH domain fragment of IRS-1 with the R28C mutation neither localized to the plasma membrane upon insulin stimulation nor bound to phosphoinosides in an in vitro assay. They also showed that the distribution of the mutant PH domain was not affected by the treatment with the PI 3-kinase inhibitors or the expression of the dominant negative mutant of the PI 3-kinase regulatory subunit (Razzini et al., 2000). In contrast, Jacobs et al. demonstrated that both the R28C and 3KQ mutations do not completely abolish translocation of the full-length and PH/PTB fragment of IRS-1 to the plasma membrane in response to insulin whereas the R28C mutation disrupts plasma membrane targeting of the IRS-1 PH domain (Jacobs et al., 2001). Therefore, the dominant negative effects of IRS-N proteins on translocation of IRS-1 with the R28C/3KQ mutation may suggest the existence of novel non-phosphoinositide molecules interacting with IRS-1 via regions other than the PH domain in the plasma membrane.

In the present study, we demonstrated that the IRS-N proteins non-specifically suppressed insulin-induced tyrosine phosphorylation of IRS-1, IRS-2 or IRS-3 (Fig. 1). The lack of specificity in IRS-N proteins may be attributed in part to the PTB domains contained in these dominant-negative proteins, which interact with the juxtamembrane region of the insulin receptors, thereby interfering with the access of intact IRS molecules irrespective of the binding specificities of the PH domains. However, the immunostaining experiments demonstrated that IRS-N proteins impaired in fact the targeting of IRS-1 to the plasma membrane in a similar fashion to the suppression of IRS-1 tyrosine phosphorylation, indicating that there exists an overlap in the cell surface ligands for the N-terminal regions of IRS proteins other than phosphoinositides, which may be a major cause of the dominant-negative effects of IRS-N molecules (Fig. 4). We also observed that the dominant-negative effect of IRS3-N on every IRS protein was most potent in all the IRS-N molecules analyzed in this study (Fig. 1). Previous studies analyzing the subcellular distribution of IRS proteins showed that IRS-3 is constitutively localized to the plasma membrane via its PH domain whereas targeting of IRS-1 and IRS-2 to the cell surface is sensitive to insulin (Kaburagi et al., 2001; Razzini et al., 2000; Maffucci et al., 2003b). As the analysis using confocal microscopy revealed that targeting of IRS-1 to the plasma membrane in response to insulin was impaired in parallel with suppressed IRS-1 tyrosine phosphorylation by the expression of the IRS-N proteins (Fig. 1 & 4), constitutive localization of IRS3-N to the plasma membrane may be the reason why the dominant negative effect of IRS3-N on IRS-mediated signaling is most effective.

As the non-phospholipid ligand for the PH domains of IRS proteins, we tested the role of PHIP in signal transduction and subcellular distribution of IRS proteins, because the dominant-negative mutant protein of PHIP reportedly suppressed insulin-induced mitogenesis and glucose transport as well as tyrosine phosphorylation of IRS-1 or IRS-2 (Farhang-Fallah et al., 2000; Farhang-Fallah et al., 2002). Contradictory to these data, we demonstrated that the N-terminal region of PHIP interacting with the PH domains of IRS proteins does not affect tyrosine phosphorylation or insulin-stimulated localization to the plasma membrane of IRS proteins (Fig. 3 & 4). One of the causes of these inconsistencies may be the difference in the expression level of the PHIP-N protein, because we detected a statistically insignificant decrease in tyrosine phosphorylation of IRS-2 and IRS-3 but not IRS-1 by the maximal overexpression of PHIP-N (Fig. 3). We speculate that the PHIP-N expression level analyzed by Farhang-Fallah et al. was much higher than that examined in this study, although the dose response effects of PHIP-N were not analyze on tyrosine phosphorylation of IRS proteins were not evaluated in their study (Farhang-Fallah et al. 2000; Farhang-Fallah et al. 2002). We believe that PHIP plays a minor role in signal transduction and subcelluar distribution of IRS proteins, because the PHIP-N expression comparable to all the IRS-N proteins did not affect tyrosine phosphorylation or plasma membrane targeting of IRS proteins in our experimental conditions. Collectively, we conclude from our data that PHIP does not play a major role in tyrosine phosphorylation of IRS through the targeting of IRS to the plasma membrane via its PH domain, although there may exist a possibility that the C-terminus of PHIP may play some role in transducing signals downstream of IRS proteins.

Taken together, the analyses of dominant-negative IRS proteins and PHIP showed that: 1) the N-terminal fragments of each IRS non-selectively suppressed tyrosine phosphorylation of all IRS proteins; 2) the N-terminal fragments of each IRS also suppressed insulin-induced localization of IRS-1 to the plasma membrane; 3) the mutations disrupting phosphoinositide binding sites of IRS-1 did not block insulin-induced tyrosine phosphorylation or cell surface targeting, both of which were further impaired by IRS-N expressions; 4) PHIP does not affect tyrosine phosphorylation of IRS proteins or insulin-induced cell surface targeting of IRS-1. These findings suggest that insulin-induced tyrosine phosphorylation and the cell surface targeting of IRS proteins are regulated in the same manner and that there exist molecules other than phosphoinositides or PHIP causing the translocation of IRS proteins to the plasma membrane thereby transducing signals via the insulin receptor tyrosine kinase.

Acknowledgements

We would like to thank Dr. Naoto Kubota (The Department of Metabolic Disease, Graduate School of Medicine, University of Tokyo) for generously providing us the mouse genomic DNA containing the IRS-2 gene. We also thank Ms. Asuka Otomo (Research Institute, International Medical Center of Japan), Ms. Kumiko Kimura (Institute for Adult Diseases, Asahi Life Foundation) and Dr. Takashi Kadowaki (The Department of Metabolic Disease, Graduate School of Medicine, University of Tokyo) for their helpful suggestions and support. This work was supported in part by a research grant from the Ministry of Health, Labor, and Welfare of Japan (to Y. K.) and a grant from the Ministry of Education, Science, Sports and Culture, Japan (to Y. K.).