Action of Insulin ReceptorSubstrate-3 (IRS-3) and IRS-4 toStimulate Translocation of GLUT4in Rat Adipose Cells

Lixin Zhou, Hui Chen, Pin Xu, Li-Na Cong,Salvatore Sciacchitano, Yunhua Li, David Graham,Aviva R. Jacobs, Simeon I. Taylor, and Michael J. Quon

Diabetes Branch (L.Z., P.X., S.S., D.G., A.R.J., S.I.T.)National Institute of Diabetes and Digestive and Kidney Disease andHypertension-Endocrine Branch (H.C., L-N.C., Y.L., M.J.Q.)National Heart, Lung, and Blood InstituteNational Institutes of HealthBethesda, Maryland 20892-1754

The insulin receptor initiates insulin action byphosphorylating multiple intracellular substrates.Previously, we have demonstrated that insulin re-ceptor substrates (IRS)-1 and -2 can mediate insu-lin’s action to promote translocation of GLUT4 glu-cose transporters to the cell surface in rat adiposecells. Although IRS-1, -2, and -4 are similar in over-all structure, IRS-3 is '50% shorter and differswith respect to sites of tyrosine phosphorylation.Nevertheless, as demonstrated in this study, bothIRS-3 and IRS-4 can also stimulate translocation ofGLUT4. Rat adipose cells were cotransfected withexpression vectors for hemagglutinin (HA) epitope-tagged GLUT4 (GLUT4-HA) and human IRS-1, mu-rine IRS-3, or human IRS-4. Overexpression ofIRS-1 led to a 2-fold increase in cell surfaceGLUT4-HA in cells incubated in the absence ofinsulin; overexpression of either IRS-3 or IRS-4elicited a larger increase in cell surface GLUT4-HA.Indeed, the effect of IRS-3 in the absence of insulinwas '40% greater than the effect of a maximallystimulating concentration of insulin in cells notoverexpressing IRS proteins. Because phosphati-dylinositol (PI) 3-kinase is essential for insulin-stimulated translocation of GLUT4, we also studieda mutant IRS-3 molecule (IRS-3-F4) in which Phewas substituted for Tyr in all four YXXM motifs (thephosphorylation sites predicted to bind to and ac-tivate PI 3-kinase). Interestingly, overexpression ofIRS-3-F4 did not promote translocation of GLUT4-HA, but actually inhibited the ability of insulin tostimulate translocation of GLUT4-HA to the cellsurface. Our data suggest that IRS-3 and IRS-4 arecapable of mediating PI 3-kinase-dependent met-abolic actions of insulin in adipose cells, and that

IRS proteins play a physiological role in mediatingtranslocation of GLUT4. (Molecular Endocrinology13: 505–514, 1999)

INTRODUCTION

Tyrosine phosphorylation plays a necessary role inmediating the biological actions of insulin (1–5). Afterinsulin binds to its receptor, the activated receptortyrosine kinase phosphorylates multiple proteins in-cluding insulin receptor substrates (IRS)-1, -2, -3, and-4 (6–10), Grb2-associated binder-1 (Gab1) (11), Shc(12), and pp120/HA4 (13). Phosphotyrosine residues inthese substrates bind to SH2 domains in proteins suchas phosphatidylinositol (PI) 3-kinase, resulting in acti-vation of multiple intracellular signaling pathways.There is apparent functional redundancy in the systembecause multiple proteins share the ability to activatethe same signaling pathway (e.g. activation of PI 3-ki-nase). Nevertheless, several observations suggest thateach IRS may serve distinct functions in mediatinginsulin action. Each substrate has a characteristic pat-tern of expression in specific tissues. In addition, eachsubstrate possesses unique structural features thatdetermine which downstream signaling pathways willbe triggered when the protein becomes phosphory-lated. For example, IRS-3 has a structure that is quitedistinct from the three other members of the IRS fam-ily. In contrast to IRS-1, -2, and -4, IRS-3 is a smallermolecule and has fewer phosphorylation sites (8, 9).

Although the necessary role of the insulin receptortyrosine kinase in insulin action is well established(1–5), it remains controversial whether IRS proteins areabsolutely required to mediate the metabolic actionsof insulin (14, 15). We have previously demonstratedthat both IRS-1 and IRS-2 are capable of mediating

0888-8809/99/$3.00/0Molecular EndocrinologyCopyright © 1999 by The Endocrine Society

505

insulin’s action to stimulate translocation of GLUT4glucose transporters from intracellular vesicles to theplasma membrane in rat adipose cells (16, 17). Fur-thermore, expression of an antisense ribozyme di-rected against IRS-1 resulted in a 4-fold decrease inthe sensitivity of the dose-response curve for insulin-induced translocation of GLUT4 in adipose cells (16).In contrast, studies in 3T3-L1 cells have suggestedthat phosphorylation of IRS-1 may not be required tomediate insulin’s action to stimulate glucose transport(14, 15). However, in knockout mice lacking IRS-1,adipose cells are partially resistant to the metabolicactions of insulin, strongly suggesting that IRS-1 in-deed participates in this action of insulin in this phys-iological target cell (18–21). Since adipose cells frommice lacking IRS-1 retain some ability to respond toinsulin, albeit with a less sensitive dose-reponse curve,other substrates may also contribute to this action ofinsulin in adipose cells. When IRS-3 was identified asthe major insulin-stimulated phosphotyrosine-contain-ing protein in adipose cells from mice lacking IRS-1(18, 19), this raised the question of whether IRS-3 canmediate the metabolic actions of insulin.

In the present study, we have demonstrated thatoverexpression of either mouse (m)IRS-3 or human(h)IRS-4 (in the presence or absence of insulin) leadsto an increase in the number of GLUT4 moleculesrecruited to the surface of rat adipose cells. Further-more, this ability of mIRS-3 requires the presence ofone or more of the four YXXM motifs in the molecule;thus, it is likely that activation of PI 3-kinase is requiredfor the action of mIRS-3 to promote translocation ofGLUT4-containing vesicles. Finally, mutant mIRS-3lacking all four YXXM motifs inhibits this action ofinsulin. Since this mutant would be predicted to com-pete with endogenous IRS proteins for binding to theinsulin receptor, our data support the hypothesis thatphosphorylation of IRS proteins plays a physiologicalrole in the pathway that mediates the ability of insulinto promote translocation of GLUT4.

RESULTS

Phosphotyrosine-Containing Proteins in RatAdipose Cells

In freshly isolated rat adipose cells, insulin increasedthe phosphotyrosine content of three major bands(data not shown). As reported previously, these bandscorrespond to the insulin receptor (Mr, 95,000), IRS-1and -2 (Mr, 185,000), and IRS-3 (Mr, 60,000) (8, 17).Because subcellular localization of IRS proteins maycontribute importantly to signal specificity (17, 22–25),we investigated the distribution of these phosphopro-teins between particulate and cytosolic fractions. Asexpected, the insulin receptor (which is an integralmembrane protein) was located exclusively in the par-ticulate fraction of the cell. Similarly, the Mr 185,000band (corresponding to IRS-1 and -2) and the Mr

60,000 band (corresponding to IRS-3) were also lo-cated in the particulate fraction, without any detect-able quantities in the cytosolic fraction (data notshown). Therefore, in subsequent studies, we assayedfor IRS proteins only in the particulate fraction of thecell.

In our transfection studies, we incubated rat adiposecells overnight after electroporation to allow for ex-pression of the recombinant proteins. Therefore, weassessed the effect of these experimental interven-tions upon the expression of insulin-stimulated phos-photyrosine-containing proteins. After electroporationand overnight incubation, we observed a decrease inthe intensity of phosphotyrosine-containing bands (Mr

185,000 and Mr 130,000) in the particulate fraction ofcells incubated in the absence of insulin (Fig. 1, lanes1 and 3). Acute stimulation with insulin (60 nM, 2 min)increased the phosphotyrosine content of severalbands. In cells incubated overnight, insulin induced anincrease in phosphorylation of the IRS-1/2 band thatwas comparable to that seen in freshly isolated cells.In contrast, insulin-stimulated phosphorylation of theinsulin receptor appeared to be increased and phos-phorylation of IRS-3 appeared to be decreased in cellscultured overnight (Fig. 1, lanes 2 and 4). In addition,

Fig. 1. Tyrosine-Phosphorylated Proteins in Freshly Isolatedand Cultured Adipose Cells

Freshly isolated cells or adipose cells that had undergonetransfection with pCIS2 (empty expression vector) and over-night culture were treated without or with insulin (60 nM, 2min). Membrane fractions were then prepared and 40 mgprotein from each group were separated by SDS-PAGE on a7.5% gel followed by immunoblotting with the antiphospho-tyrosine antibody 4G10. A representative blot is shown froman experiment that was repeated independently three times.

MOL ENDO · 1999 Vol 13 No. 3506

there is a phosphotyrosine containing band at Mr

130,000, the identity of which is not certain. Overnightculture led to a decrease in the basal phosphorylationof this band, without affecting the insulin-stimulatedlevel of phosphorylation. Consequently, insulin ap-peared to increase phosphorylation of the Mr 130,000protein in cultured cells, but not in freshly isolatedcells. Furthermore, there is a marked increase in theintensity of a band corresponding to a Mr 75,000 phos-photyrosine-containing protein in cultured cells afterstimulation with insulin (Fig. 1, lane 4).

Characterization of Overexpressed RecombinantIRS Proteins in Rat Adipose Cells

We assessed the ability of recombinant IRS proteins tobecome phosphorylated in response to insulin stimu-lation in transfected adipose cells. In control cellstransfected with the empty expression vector pCIS2,insulin increased the phosphotyrosine content ofbands corresponding to IRS-1/2 (17), IRS-3 (8), andthe insulin receptor (Fig. 2, lanes 1 and 2). Overexpres-sion of each IRS protein in rat adipose cells led tomarked increases in the intensities of bands corre-sponding to IRS-1, -2, -3, and -4 after insulin stimu-lation (Fig. 2, lanes 4, 6, 8, and 12, respectively). Underour experimental condition, we obtain expression of

recombinant proteins in approximately 5% of the ad-ipose cells (3). Therefore, these observations are con-sistent with high-level overexpression of IRS proteinsin the small fraction of cells that are actually trans-fected. In addition, the phosphotyrosine content of theMr 60,000 band in cells overexpressing mIRS-3 butincubated in the absence of insulin (Fig. 2, lane 7) was'2-fold greater than that seen in control cells incu-bated in the presence of insulin (Fig. 2, lane 2). We alsooverexpressed a mutant mIRS-3 (denoted IRS-3-F4) inwhich Phe was substituted for the four Tyr residues inTyr-Xaa-Xaa-Met motifs (Tyr341, Tyr350, Tyr361, andTyr390). While IRS-3-F4 underwent insulin-stimulatedtyrosine phosphorylation, the intensity of the band wasmarkedly reduced in comparison to mIRS-3. Note thatthe presence of the myc epitope tag retarded theelectrophoretic mobilities of both recombinant mIRS-3and IRS-3-F4 in comparison with endogenous ratIRS-3. Thus, in this experiment it is possible to evalu-ate the phosphorylation of IRS-3 and IRS-3-F4 inde-pendent of endogenous IRS-3. Since IRS-4 is not ex-pressed endogenously in adipose cells, phosphorylationof recombinant IRS-4 is also easily appreciated.

Several lines of evidence demonstrate that activa-tion of PI 3-kinase mediates insulin’s effect to stimu-late translocation of GLUT4 to the plasma membrane(19, 22, 23, 26–30). As a consequence of insulin bind-

Fig. 2. Insulin-Stimulated Phosphorylation of Overexpressed IRS Proteins in Rat Adipose CellsTotal membrane fractions were isolated from cells transfected (6 mg DNA/cuvette) with expression vector pCIS2 (lanes 1 and

2), hIRS-1 (lanes 3 and 4), mIRS-2 (lanes 5 and 6), mIRS-3 (lanes 7 and 8), mIRS-3-F4 (lanes 9 and 10), or hIRS-4 (lanes 11 and12) after treatment without or with insulin (60 nM, 2 min). The samples (50 mg protein) were separated by SDS-PAGE on a 7.5%gel and immunoblotted with antiphosphotyrosine antibody 4G10. The intensity of the Mr 185,000 band corresponding tophosphorylated IRS-1, -2, and -4 was increased 2-fold in cells overexpressing IRS-1 (lane 4), IRS-2 (lane 6), and IRS-4 (lane 12)as compared with the control cells (lane 2). Relative to the intensity of the phosphorylated Mr 60,000 band in the control cells (lane2), transfection led to a 6-fold increase in the level of IRS-3 in cells overexpressing mIRS-3 (lane 8). In addition, the level ofphosphorylated mIRS-3 in the absence of insulin (lane 7) was approximately twice the level of phosphorylated endogenous IRS-3after insulin stimulation of control cells (lane 2). Note also that the addition of a myc epitope tag to both the mIRS-3 and IRS-3-F4constructs resulted in a slower migration on the gel so that these recombinant proteins can be distinguished from the endogenousIRS-3.

Role of IRS-3 and 4 in Translocation of GLUT4 507

ing, insulin receptors phosphorylate Tyr-Xaa-Xaa-Met(YXXM) motifs on IRS-1, -2, -3, and -4. These phos-phorylated YXXM motifs bind to SH2 domains in thep85 regulatory subunit of PI 3-kinase, thereby activat-ing the p110 catalytic subunit. Therefore, we evaluatedthe ability of mIRS-3 to bind to the p85 regulatorysubunit of PI 3-kinase. Recombinant myc-taggedmIRS-3 was expressed in rat adipose cells; the par-ticulate fraction of the cell was solubilized and sub-jected to immunoprecipitation with anti-myc antibodyfollowed by immunoblotting with anti-p85 antibody.Even in the absence of insulin, there was a detectableassociation between mIRS-3 and p85. Stimulation ofcells with insulin led to 4.5-fold increase in the quantityof p85 co-immunoprecipitated with mIRS-3 (Fig. 3,lane 4). As expected, mutation of the four Tyr-Xaa-Xaa-Met motifs abolished the association of p85 withIRS-3-F4 (Fig. 3, lanes 5 and 6). Comparable levels ofp85 were detected in the particulate fractions, irre-spective of whether cells were transfected withpCIS-2, mIRS-3, or IRS-3-F4 or whether or not thecells were stimulated with insulin (Fig. 3, lanes 7–12).Furthermore, as judged by immunoblotting with anti-body to the myc-epitope tag, mIRS-3 and mIRS-3–F4were expressed at comparable levels (Fig. 3, bottompanel, lanes 3–6).

Effect of Overexpression of mIRS-3 and hIRS-4on Translocation of GLUT4

When rat adipose cells were cotransfected with pCIS2and GLUT4-HA, insulin treatment (60 nM) caused a2.5-fold increase in cell surface GLUT4-HA (Fig. 4).Furthermore, as shown previously (16, 17), overex-pression of hIRS-1 in adipose cells stimulated trans-location of GLUT4-HA to the cell surface. In the ab-sence of insulin, recombinant hIRS-1 led to a 2-foldincrease in cell surface GLUT4-HA; in the presence ofa maximally effective concentration of insulin (60 nM),the level of cell surface GLUT4-HA in cells overex-pressing IRS-1 exceeded by 25% the maximal recruit-ment observed in insulin-stimulated control cells. In-terestingly, recombinant mIRS-3 exerted a largereffect upon translocation of GLUT4-HA to the cellsurface. In the absence of insulin, recombinantmIRS-3 led to a 3.5-fold increase in cell surfaceGLUT4-HA. Moreover, the level of cell surfaceGLUT4-HA was greater in cells overexpressingmIRS-3 (incubated in the absence of insulin) than wasobserved in control cells (i.e., not overexpressing re-combinant IRS proteins) that had been incubated inthe presence of insulin. Addition of insulin did not elicita further increase in the content of GLUT4-HA on the

Fig. 3. Insulin-Stimulated Association between Recombinant IRS-3 Proteins and PI 3-KinaseMembrane fractions were obtained from adipose cells transfected with pCIS2, mIRS-3, or IRS-3-F4 (6 mg DNA/cuvette) after

overnight incubation and treatment without or with insulin (60 nM, 2 min). Aliquots of each sample containing 130 mg protein weresubjected to immunoprecipitation (ippt) with an anti-myc antibody followed by immunoblotting with either an antibody against thep85 subunit of PI 3-kinase or an antibody against the myc epitope (lanes 1–6). Aliquots of the same membrane fractions (20 mgprotein) were also immunoblotted with the antibody against p85 without prior immunoprecipitation to demonstrate that allsamples contained comparable amounts of p85 (lanes 7–12). Insulin treatment caused a 4.5-fold increase in the amount of p85coimmunoprecipitated with the mIRS-3 construct (lanes 3 and 4). In contrast, there was no detectable association of p85 withIRS-3-F4 either in the absence or presence of insulin (lanes 5 and 6).

MOL ENDO · 1999 Vol 13 No. 3508

cell surface. Similarly, even in the absence of insulin,overexpression of hIRS-4 led to a supramaximal in-crease in recruitment of GLUT4-HA to the cell surface(;30% greater than the stimulation of control cells inresponse to insulin), which was not increased furtherupon addition of insulin (Fig. 5).

Effect of Overexpression of mIRS-3-F4 onTranslocation of GLUT4

To evaluate the role of PI 3-kinase in the action ofIRS-3, we investigated the effect of overexpressing theIRS-3-F4 mutant. mIRS-3-F4 inhibited insulin-stimu-lated translocation of GLUT4 in transfected rat adi-pose cells (Fig. 6). In the absence of insulin, the level ofcell surface GLUT4-HA in cells overexpressing mIRS-3-F4 was approximately 30% lower than that of thecontrol cells. Furthermore, at an insulin concentrationof 0.3 nM, there was no significant increase in cellsurface GLUT4-HA in cells overexpressing mIRS-3-F4, although this concentration of insulin elicited anear-maximal effect in cells expressing only pCIS2.Nevertheless, at a maximally effective concentration ofinsulin (60 nM), mutant mIRS-3-F4 decreased the con-tent of GLUT4-HA on the cell surface by only 30% incomparison to control cells. It is likely that mIRS-3-F4

exerts its inhibitory action by competing with the abil-ity of the insulin receptor to phosphorylate endoge-nous IRS-1, -2, and -3. As a result, mutant mIRS-3-F4would inhibit activation of PI 3-kinase, thereby inhib-iting insulin’s action to stimulate translocation ofGLUT4.

In separate experiments, we verified that overex-pression of mIRS-3-WT, mIRS-3-F4, and hIRS-1 didnot significantly affect the level of expression ofGLUT4-HA (Fig. 7). Thus, the effects of recombinantIRS proteins are due to their effects upon the subcel-lular localization of GLUT4-HA, rather than altering thelevel of expression of the GLUT4-HA reportermolecule.

DISCUSSION

Necessary Role of IRS Proteins in Recruitmentof GLUT4

Considerable evidence suggests that IRS proteinsplay a necessary role in mediating metabolic actions ofinsulin such as the recruitment of GLUT4. For exam-ple, absence of IRS-1 in knockout mice leads to insulinresistance in the whole animal and also in adiposecells incubated in vitro (18, 19). Furthermore, micelacking IRS-2 are also insulin resistant although stud-

Fig. 4. Effects of Overexpression of hIRS-1 or mIRS-3 onTranslocation of GLUT4 in Rat Adipose Cells

Cells were cotransfected with GLUT4-HA (3 mg/cuvette)and pCIS2 (control), hIRS-1, or mIRS-3 (6 mg/cuvette). Wemeasured the amount of GLUT4-HA at the cell surface in thebasal state or in response to insulin (0.07 or 60 nM). Data areexpressed as a percentage of cell surface GLUT4-HA in thepresence of a maximally effective insulin concentration forthe control group (pCIS2). Results are the means 6 SEM offour independent experiments. The actual value for the spe-cific cell-associated radioactivity for the control group at 60nM insulin was 1620 6 80 cpm. In the absence of insulin(basal state), cells overexpressing either hIRS-1 or mIRS-3had significantly higher levels of GLUT4-HA at the cell surfacethan the corresponding control (P , 0.001). At 60 nM insulin,the level of GLUT4-HA at the surface of cells overexpressingmIRS-3 was significantly greater than for cells overexpress-ing hIRS-1 (P , 0.007). These studies were performed usingan mIRS-3 construct without the myc epitope tag.

Fig. 5. Effects of Overexpression of hIRS-4 on Translocationof GLUT4 in Rat Adipose Cells

Cells were cotransfected with GLUT4-HA (3 mg/cuvette)and either pCIS2 (control), or hIRS-4 (6 mg/cuvette). We mea-sured the amount of GLUT4-HA at the cell surface in thebasal state or in response to insulin (0.024, 0.072, 0.3, or 60nM). Data are expressed as a percentage of cell surfaceGLUT4-HA in the presence of a maximally effective insulinconcentration for the control group (pCIS2). In the absence ofinsulin (basal state), cells overexpressing hIRS-4 had levels ofGLUT4-HA at the cell surface that were significantly higherthan those seen in the control cells stimulated with 60 nM

insulin (P , 0.04). For cells overexpressing hIRS-4, treatmentwith insulin did not result in a further increase in cell surfaceGLUT4-HA. Results are the means 6 SEM of four independentexperiments. By MANOVA, the two dose-response curvesare significantly different (P , 1029).

Role of IRS-3 and 4 in Translocation of GLUT4 509

ies of isolated fat cells have not yet been reported (31).In addition to these studies of the effect of chronicdeficiencies of IRS molecules, we have previously re-ported that expression of an antisense ribozyme di-rected against IRS-1 mRNA in rat adipose cells led toa rightward shift in the dose-response curve for insulinaction (16). Nevertheless, some reports suggest thatIRS proteins may not be necessary to mediate theeffect of insulin to recruit GLUT4 in 3T3-L1 culturedadipocytes (14, 15). In those reports, microinjection ofvarious reagents (including anti-IRS-1 antibodies, iso-lated phosphotyrosine binding domains, and SAIN do-mains of IRS-1) blocked the mitogenic actions of in-sulin but failed to inhibit metabolic actions.

Our experiments using overexpression of IRS-3 and-4 directly demonstate that these substrates are ca-pable of mediating metabolic actions of insulin such asrecruitment of GLUT4. Furthermore, in contrast tostudies in 3T3-L1 cells, our present studies with IRS-3-F4 demonstrated a striking inhibition of insulin’sability to promote translocation of GLUT4. The obser-vation that IRS-3-F4 inhibits the cells’ metabolic re-sponse to insulin suggests that this mutant moleculeinterferes with the endogenous signaling pathway andis consistent with the hypothesis that IRS moleculesplay a necessary role in mediating insulin’s action tostimulate translocation of GLUT4. It is likely that mIRS-

3-F4 binds to pTyr960 in the juxtamembrane domain ofthe insulin receptor and thereby inhibits the phosphor-ylation of other substrates containing phosphotyrosinebinding domains (e.g. IRS-1, -2, -3) (32–35). Thus,IRS-3-F4 is probably inhibiting signaling mediated byIRS-1 and -2 in addition to IRS-3. However, our ex-perimental design does not permit us to test this hy-pothesis directly. Because only 5% of cells becometransfected under our experimental conditions (3), it isnot possible to quantitate phosphorylation of endog-enous IRS proteins in the small number of transfectedcells because this is obscured by the presence of amuch larger number of nontransfected cells. Further-more, we have attempted to use various methods toseparate transfected from nontransfected cells. How-ever, because of the fragility of the adipose cells afterelectroporation followed by overnight tissue culture,these efforts have not been successful (M. J. Quon,unpublished observations). In any case, it is notewor-thy that high concentrations of insulin can partiallyovercome the inhibition due to mIRS-3-F4. This mightbe explained if mIRS-3-F4 is less than 100% effectiveat inhibiting phosphorylation of endogenous sub-strates such as IRS-1, -2, and -3. However, we cannotcompletely rule out the possibility that there may alsobe IRS-independent pathways that contribute to someextent to the ability of insulin to promote GLUT4translocation.

It is not clear why our studies in rat adipose cellsreached different conclusions from those reported instudies of 3T3-L1 adipocytes (14, 15). Both studiesemployed reagents that are thought to inhibit insulinaction by similar mechanisms (i.e. competitively inhib-iting phosphorylation of endogenous IRS moleculesand Shc). It is possible that the different effects oninsulin-stimulated recruitment of GLUT4 might be ex-plained by differences in cell type, differences in theinhibitory reagents, or other differences in experimen-tal methods employed. For example, the full-lengthmIRS-3-F4 might be more effective than isolatedphosphotyrosine binding or SAIN domains at compet-itively inhibiting activation of PI 3-kinase.

Relative Roles of IRS Proteins in Rat Adipocytes

Previously, we have presented evidence that bothIRS-1 and IRS-2 are capable of mediating insulin’saction to stimulate glucose transport in rat adiposecells (16, 17). Studies in mice lacking IRS-1 have dem-onstrated that IRS-3 is the major phosphotyrosine-containing protein that binds to PI 3-kinase in insulin-treated adipose cells from the knockout mice (18, 19).The present study provides direct evidence that bothIRS-3 and IRS-4 can also mediate the action of insulinto promote translocation of GLUT4 glucose transport-ers in adipose cells. Thus, phosphorylation of any ofthese four substrates by the insulin receptor tyrosinekinase (i.e. IRS-1, -2, -3, or -4) is sufficient to triggertranslocation of GLUT4 to the cell surface. This raisesthe question as to the relative importance of the four

Fig. 6. Inhibitory Effect of mIRS-3-F4 on Insulin-StimulatedTranslocation of GLUT4

Cells were cotransfected with GLUT4-HA (3 mg/cuvette)and either pCIS2 (control) or mIRS-3-F4 (6 mg/cuvette). Wemeasured the amount of GLUT4-HA at the cell surface in thebasal state or in response to insulin (0–60 nM). Data areexpressed as a percentage of cell surface GLUT4-HA in thepresence of a maximally effective insulin concentration forthe control group (pCIS2). Results are the means 6 SEM offour independent experiments. The actual value for the spe-cific cell-associated radioactivity for the control group at 60nM insulin was 2280 6 380 cpm. In the absence of insulin(basal state), cells overexpressing mIRS-3-F4 had signifi-cantly lower levels of GLUT4-HA at the cell surface than thecorresponding control (P , 0.03). At 60 nM insulin, the level ofGLUT4-HA at the surface of cells overexpressing mIRS-3-F4was also significantly lower than for control cells (P , 0.02).By MANOVA, the two dose-response curves are significantlydifferent (P , 1029).

MOL ENDO · 1999 Vol 13 No. 3510

IRS proteins in mediating insulin action in rat adipo-cytes. Unlike IRS-1, -2, and -3, which are all endog-enously expressed in rat adipocytes, IRS-4 is not ex-pressed at detectable levels in this cell type (36).Previously, we (17) and others (37) have estimated thatIRS-1 contributes approximately twice as much asIRS-2 to the pathway leading to the activation of PI3-kinase. In the present study, the level of expressionof IRS-3 appears roughly comparable to the sum totalof IRS-1 plus IRS-2, as judged by the phosphotyrosinecontent of these proteins after insulin stimulation infreshly isolated rat adipocytes (Fig. 1, lane 2). How-ever, several considerations dictate caution in inter-preting the results of immunoblotting studies with an-tiphosphotyrosine antibodies. For example, the totalphosphotyrosine content does not necessarily reflectthe phosphorylation of a specific tyrosine phosphory-lation site. In addition, it is possible that the amino acidsequence flanking a phosphotyrosine residue may al-ter that avidity for binding of antiphosphotyrosine an-tibodies. Furthermore, the literature contains conflict-ing reports with respect to the intrinsic activities ofvarious IRS proteins to activate PI 3-kinase. Previ-ously, Smith-Hall et al. (18) suggested that IRS-3 maybe more effective than IRS-1 at activating PI 3-kinasein rat adipose cells. Specifically, they reported that thetime course for the association of PI 3-kinase withIRS-3 was faster and the extent of the association wasgreater than for IRS-1. In contrast, Kaburagi et al. (19)reached the opposite conclusion that IRS-1 is themore potent activator in murine adipose cells. In com-paring the activities of recombinant IRS-1 with IRS-3,we noted that recombinant IRS-3 appeared to be moreeffective than recombinant IRS-1 in promoting trans-location of GLUT4 to the plasma membrane (Fig. 1). Infact, the phosphotyrosine content of mIRS-3 in trans-

fected cells incubated in the absence of insulin ap-peared less than the phosphotyrosine content ofhIRS-1 in transfected cells incubated in the presenceof insulin. Nevertheless, the relatively low levels ofphosphorylated mIRS-3 in basal cells (Fig. 2, lane 7)led to a greater increase in the number of GLUT4molecules on the cell surface than was induced by thehigher levels of phosphorylated hIRS-1 in insulin-stim-ulated cells (Fig. 2, lane 4). Interestingly, althoughIRS-4 is not endogenously expressed in adipose cells,it resembles IRS-3 in that overexpression of eithermolecule led to qualitatively similar results. In fact,although the phosphotyrosine content of hIRS-4 in theabsence of insulin (Fig. 2, lane 11) was lower than thatof mIRS-3 (Fig. 2, lane 7), overexpression of eitherIRS-3 or IRS-4 led to supramaximal translocation ofGLUT4. There are many possible explanations thatmight account for different potencies of the variousIRS proteins to mediate the metabolic actions of insu-lin, e.g. different subcellular localization, activation ofdifferent downstream effector molecules, differentialsensitivity to inactivation by phosphotyrosine phos-phatases, etc.

When isolated rat fat cells are cultured for prolongedperiods in vitro, they tend to lose differentiated func-tions. For example, there is a progressive decrease inthe level of expression of GLUT4 mRNA and protein(38), and a concomitant decrease of the insulin re-sponse. Thus, another observation consistent with thehypothesis that IRS-3 plays a role in mediating meta-bolic actions of insulin is the fact that the level ofendogenous IRS-3 decreased after 1 day in culture.Interestingly, the levels of endogenous IRS-1 and -2were constant over the same period of time.

On balance, our data are consistent with the hypoth-esis that IRS-3 may be an important IRS molecule

Fig. 7. Expression of Recombinant GLUT4-HACells were cotransfected with GLUT4-HA (3 mg/cuvette) and either pCIS2 (control) or various expression vectors for IRS

proteins (6 mg/cuvette). Triton X-100 extracts of total membrane fractions (100 mg protein) were subjected to SDS-PAGE, followedby immunoblotting with anti-HA antibody.

Role of IRS-3 and 4 in Translocation of GLUT4 511

mediating the metabolic actions of insulin in rat adi-pocytes. However, to draw definitive conclusionsabout the relative importance of various IRS mole-cules, it is necessary to compare the roles of endog-enous IRS-1, -2, and -3 in freshly prepared adipo-cytes. Unfortunately, high quality antibodies to ratIRS-3 are not yet available. If these antibodies wereavailable, it would have been possible to carry outimmunoprecipitation experiments to determine the rel-ative quantities of PI 3-kinase associated with thevarious IRS molecules. Nevertheless, uncertaintywould still remain whether activated PI 3-kinase isequally effective at stimulating translocation of GLUT4irrespective of whether PI 3-kinase is bound to IRS-1,-2, or -3.

In addition to the known IRS molecules, we alsodetected at least two other proteins (Mr 130,000 andMr 75,000) in cultured adipose cells that underwenttyrosine phosphorylation in response to insulin (Fig. 1).The Mr 75,000 phosphoprotein was not detected infreshly isolated rat adipocytes. In contrast, there was aconstitutively phosphorylated protein band in the Mr

130,000 region of the gel in fresh cells. In the culturedcells, the phosphotyrosine content of the Mr 130,000band was decreased, which facilitated detection ofthis band upon insulin stimulation. Although the iden-tities of these insulin-sensitive phosphoproteins arenot known, it is possible that they may play a role inmediating insulin action in adipose cells.

Downstream from IRS Phosphorylation

In contrast to the apparent redundancy at the level ofIRS, the next step downstream in the metabolic sig-naling pathway (i.e. activation of PI 3-kinase) is essen-tial. For example, when we expressed a mutant p85that inhibits the activation of PI 3-kinase, the ability ofinsulin to promote translocation of GLUT4 in rat adi-pose cells was essentially abolished (27). Similarly,agents that inhibit the catalytic activity of PI 3-kinasealso lead to complete inhibition of this action of insulin(28). Our present studies demonstrating an inhibitoryeffect of a mutant IRS-3-F4 that does not associatewith the p85 regulatory subunit of PI 3-kinase is alsoconsistent with this view. That is, since the regulatorysubunit is required for activation of PI 3-kinase, it isreasonable to conclude that absence of p85 wouldcorrelate with absence of associated PI 3-kinase ac-tivity. It is generally accepted that activation of PI3-kinase plays a necessary role in mediating insulinaction to stimulate translocation of GLUT4 to theplasma membrane. In addition, while there is consid-erable evidence that activation of PI 3-kinase is suffi-cient to promote translocation of GLUT4 (30, 39, 40),activated PI 3-kinase must be located in the correctsubcellular compartment to elicit this response (22, 23,29, 41). Nevertheless, the cell also possesses othermechanisms that can stimulate translocation ofGLUT4, even if these mechanisms do not appear tomediate insulin’s action upon GLUT4. For example,

although activation of Ras is capable of triggeringtranslocation of GLUT4, inhibitory mutants of Ras donot block this action of insulin in rat adipose cells (27).Furthermore, although wortmannin inhibits insulin-stimulated translocation of GLUT4, wortmannin doesnot inhibit the effect of activated L61-ras (27). The factthat mIRS-3 is capable of mediating translocation ofGLUT4 provides additional evidence against a role ofRas in this pathway. Because mIRS-3 does not con-tain an obvious binding site for the SH2 domain ofGrb2 (9), it is unlikely that mIRS-3 would activate theRas pathway. Taken together, these data support thehypothesis that activation of PI 3-kinase is both nec-essary and sufficient to mediate insulin’s ability tostimulate translocation of GLUT4.

General Conclusions

Signals emanating from multiple cellular tyrosine ki-nases converge upon an overlapping set of substratesfor phosphorylation. Furthermore, as we have shownin this and other studies (16, 17), the signals emanatingfrom IRS-1, -2, and -3 converge to activate the samemetabolic action of insulin to promote translocation ofGLUT4. It is likely that the various members of the IRSfamily of proteins mediate activation of PI 3-kinase.However, in addition, there are divergent signals thatemanate from the IRS family of proteins resulting inactivation of other signaling pathways, e.g. Ras,SHP-2, Fyn, etc. Interestingly, the various IRS proteinsdo not share all of the same phosphorylation sites sothat they are not predicted to activate precisely thesame set of diverging signaling pathways. Detailedstudies of the role of the individual IRS proteins havethe potential to provide insights into the importantcontributions of both redundancy and specificity in thecomplex network of signaling pathways.

MATERIALS AND METHODS

Antibodies

Murine monoclonal antibodies directed against phosphoty-rosine and polyclonal antibodies directed against the p85subunit of PI 3-kinase were obtained from Upstate Biotech-nology, Inc. (Lake Placid, NY). Monoclonal antibody againstinfluenza HA epitope was obtained from BAbCO (Berkeley,CA). Polyclonal antibodies against the myc epitope wereobtained from Santa Cruz Biotechnology (Santa Cruz, CA).

Expression Plasmids

In these studies, we used the expression vector pCIS2 (42) todirect high level expression of various recombinant proteinsin transfected rat adipose cells. Expression vectors forGLUT4-HA (3), hIRS-1 (16), and mIRS-2 (17) were con-structed as previously described. Murine IRS-3 cDNA (9) wasligated into the HpaI site of pCIS2 expression vector byblunt-end ligation. IRS-3-F4 is a mutant form of IRS-3 inwhich phenylalanines were substituted for the four tyrosineresidues (Tyr341, Tyr350, Tyr361, and Tyr390) predicted to bind

MOL ENDO · 1999 Vol 13 No. 3512

the p85 regulatory subunit of PI 3-kinase. Mutant IRS-3-F4cDNA was ligated into pCIS2 expression vector as describedabove. In addition, a myc epitope tag was added at the Cterminus of both the wild-type mIRS-3 and IRS-3-F4 con-structs. Finally, the sequence at the translation start site wasmodified to match the Kozak consensus sequence for tran-scription start sites (43).

IRS4 was cloned by RT-PCR. Briefly, total RNA was iso-lated from 293 cells with TRIzol reagent (GIBCO BRL, Gaith-ersburg, MD). First-strand cDNA was synthesized from 5 mgof total RNA using Superscript II/RNase H2/MMLV reverseTranscriptase (GIBCO BRL) and a gene-specific primer (59-ccctaacactgtagactgtagcgcatcg-39). Two different fragmentswere generated by PCR. Using pfu DNA polymerase (Strat-agene, La Jolla, CA), we amplified two fragments of cDNAwith the following two pairs of primers: 59-ggaaaccagtgctcta-gagatggcc-39 plus 59-tatgggcccgacctcttttgggagagtcgaac-39;and 59-atagaattcgccaccatggcgagttgctccttcac-39 plus 59-ccctggccatctctagagcactgg-39. The cDNA encoding epitope-tagged IRS-4 was obtained by joining the two fragments ofamplified cDNA and ligating them into the pCDNA3.1 myc-hisA vector. Full-length IRS-4 cDNA was excised frompCDNA3.1 myc-his A vector using EcoRI and PmeI, blunt-ended, and subcloned into the HpaI site of the pCIS2 vector.The nucleotide sequence of IRS-4 cDNA was determined tobe identical to the published sequence (10) with two excep-tions: substitution of T and A for C705 and C1203, respectively.

Experiments with Isolated Rat Adipose Cells

Rat adipose cells were isolated from epididymal fat pads ofmale rats according to the method of Honnor et al. (44).Isolated adipose cells were transfected by electroporation asdescribed by Quon et al. (3, 16, 27) with slight modifications(17). Monoclonal anti-HA antibody was employed to measurethe quantity of epitope-tagged GLUT4 expressed on the cellsurface (3). For some immunoblotting experiments, freshlyisolated adipose cells were treated without or with insulin (60nM, 2 min), and whole-cell lysates were prepared as de-scribed (45). Whole-cell lysates were also fractionated intoparticulate and cytosolic fractions by centrifugation at100,000 3 g for 30 min as described (17). For immunoblottingexperiments with transfected cells, the particulate fractionwas isolated from the cells as described (17). Immunoblottingand immunoprecipitation were performed according to pre-viously described methods (17, 45). Quantitation of immuno-blots was performed by scanning laser densitometry (Molec-ular Dynamics, Sunnyvale, CA).

Statistical Analysis

Paired Student’s t tests were used to compare individualpoints where appropriate. Values of P , 0.05 were consid-ered to indicate statistical significance. The dose-responsecurves were fit using a nonlinear least squares method andcompared by multiple analysis of variance (MANOVA).

Acknowledgments

Received July 31, 1998. Re-revision received November 6,1998. Accepted November 11, 1998.

Address requests for reprints to: Michael J. Quon, M.D.,Ph.D., National Institutes of Health, Building 10, Room 8C-103, 10 Center Drive MSC 1754, Bethesda, Maryland 20892-1754. E-mail: quonm@gwgate.nhlbi.nih.gov.

REFERENCES

1. Kasuga M, Karlsson FA, Kahn CR 1982 Insulin stimulatesthe phosphorylation of the 95,000-dalton subunit of its

own receptor. Science 215:185–1872. Odawara M, Kadowaki T, Yamamoto R, Shibasaki Y,

Tobe K, Accili D, Bevins C, Mikami Y, Matsuura N,Akanuma Y, Takaku F, Taylor SI, Kasuga M 1989 Humandiatetes associated with a mutation in the tyrosine kinasedomain of the insulin receptor. Science 245:66–68

3. Quon MJ, Guerre MM, Zarnowski MJ, Butte AJ, Em M,Cushman SW, Taylor SI 1994 Tyrosine kinase-deficientmutant human insulin receptors (Met1153–.Ile) overex-pressed in transfected rat adipose cells fail to mediatetranslocation of epitope-tagged GLUT4. Proc Natl AcadSci USA 91:5587–5591

4. Quon M, Butte A, Taylor S 1994 Insulin signal transduc-tion pathways. Trends Endocrinol Metab 5:369–376

5. Cheatham B, Kahn CR 1995 Insulin action and the insulinsignaling network. Endocr Rev 16:117–142

6. Sun XJ, Rothenberg PL, Kahn CR, Backer JM, Araki E,Wilden PA, Cahill DA, Goldstein BJ, White MF 1991 Thestructure of the insulin receptor substrate IRS-1 defines aunique signal transduction protein. Nature 352:73–77

7. Sun XJ, Wang LM, Zhang Y, Yenush L, Myers MGJ,Glasheen E, Lane WS, Pierce JH, White MF 1995 Role ofIRS-2 in insulin and cytokine signalling. Nature377:173–177

8. Lavan BE, Lane WS, Lienhard GE 1997 The 60-kDaphosphotyrosine protein in insulin-treated adipocytes isa new member of the insulin receptor substrate family.J Biol Chem 272:11439–11443

9. Sciacchitano S, Taylor SI 1997 Cloning, tissue expres-sion, and chromosomal localization of the mouse IRS-3gene. Endocrinology 138:4931–4940

10. Lavan BE, Fantin VR, Chang ET, Lane WS, Keller SR,Lienhard GE 1997 A novel 160-kDa phosphotyrosineprotein in insulin-treated embryonic kidney cells is a newmember of the insulin receptor substrate family. J BiolChem 272:21403–21407

11. Holgado-Madruga M, Emlet DR, Moscatello DK, GodwinAK, Wong AJ 1996 A Grb2-associated docking protein inEGF- and insulin-receptor signaling. Nature 379:560–564

12. Isakoff SJ, Yu Y-P, Su Y-C, Blaikie P, Yajnik V, Rose E,Weidner KM, Sachs M, Margolis B, Skolnik EY 1996Interaction between the phosphotyrosine binding do-main of Shc and the insulin receptor is required for Shcphosphorylation by insulin in vivo. J Biol Chem 271:3959–3962

13. Margolis R, Schell MJ, Taylor SI, Hubbard A 1990 He-patocyte plasma membrane ecto-ATPase (pp120/HA4)is a substrate for tyrosine kinase activity of the insulinreceptor. Biochem Biophys Res Commun 166:562–566

14. Morris AJ, Martin SS, Haruta T, Nelson JG, VollenweiderP, Gustafson TA, Mueckler M, Rose DW, Olefsky JM1996 Evidence for an insulin receptor substrate 1 inde-pendent insulin signaling pathway that mediates insulin-responsive glucose transporter (GLUT4) translocation.Proc Natl Acad Sci USA 93:8401–8406

15. Sharma PM, Egawa K, Gustafson TA, Martin JL, OlefskyJM 1997 Adenovirus-mediated overexpression of IRS-1interacting domains abolishes indulin-stimulated mito-genesis without affecting glucose transport in 3T3–L1adipocytes. Mol Cell Biol 17:7386–7397

16. Quon MJ, Butte AJ, Zarnowski MJ, Sesti G, CushmanSW, Taylor SI 1994 Insulin receptor substrate 1 mediatesthe stimulatory effect of insulin on GLUT4 translocation intransfected rat adipose cells. J Biol Chem 269:27920–27924

17. Zhou LX, Chen H, Lin CH, Cong L-N, McGibbon MA,Sciacchitano S, Lesniak MA, Quon MJ, Taylor SI 1997Insulin receptor substrate-2 (IRS-2) can mediate the ac-tion of insulin to stimulate translocation of GLUT4 to thecell surface in rat adipose cells. J Biol Chem 272:29829–29833

18. Smith-Hall J, Pons S, Patti ME, Burks DJ, Yenush L, SunXJ, Kahn CR, White MF 1997 The 60 kDa insulin receptor

Role of IRS-3 and 4 in Translocation of GLUT4 513

substrate functions like an IRS protein (pp60IRS-3) inadipose cells. Biochemistry 36:8304–8310

19. Kaburagi Y, Satoh S, Tamemoto H, Yamamoto-Honda R,Tobo K, Veki K, Yamauchi T, Kono-Sugita E, Sekihara H,Aizawa S, Cushman SW, Akanuma Y, Yazaki Y, Kad-owaki T 1997 Role of insulin receptor substrate-1 andpp60 in the regulation of insulin-induced glucose trans-port and GLUT4 translocation in primary adipocytes.J Biol Chem 272:25839–25844

20. Araki E, Lipes MA, Patti ME, Bruning JC, Haag BI, John-son RS, Kahn CR 1994 Alternative pathway of insulinsignaling in mice with targeted disruption of the IRS-1gene. Nature 372:186–191

21. Tamemoto H, Kadowaki T, Tobe K, Yagi T, Sakura H,Hayakawa T, Terauchi Y, Ueki K, Kaburagi Y, Satoh S, al.e 1994 Insulin resistance and growth retardation in micelacking insulin receptor substrate-1. Nature 372:182–186

22. Heller-Harrison RA, Morion M, Czech MP 1995 Insulinregulation of membrane-associated insulin receptor sub-strate 1. J Biol Chem 270:24442–24450

23. Heller-Harrison RA, Morion M, Guilherme A, Czech MP1996 Insulin-mediated targeting of phosphatidylinositol3-kinase to GLUT4-containing vesicles. J Biol Chem271:10200–10204

24. Clark S, Martin S, Carozzi A, Hill M, James D 1998Intracellular localization of phosphatidylinositide 3-ki-nase and insulin receptor substrate-1 in adipocytes: po-tential involvement of a membrane skeleton. J Cell Biol140:1211–1225

25. Inoue G, Cheatham B, Emkey R, Kahn C 1998 Dynam-ics of insulin signaling in 3T3–L1 adipocytes. Differen-tial compartmentalization and trafficking of insulinreceptor substrate (IRS)-1 and IRS-2. J Biol Chem 273:11548–11555

26. Kelly KL, Ruderman NB 1992 Phosphatidylinoditol-3-ki-nase in isolated rat adipocytes. Activation by insulin andsubcellular distribution. J Biol Cehm 267:3423–3428

27. Quon MJ, Chen H, Ing BL, Liu ML, Zarnowski MJ, Yon-ezawa K, Kasuga M, Cushman SW, Taylor SI 1995 Rolesof 1-phosphatidylinositol 3-kinase and ras in regulatingtranslocation of GLUT4 in transfected rat adipose cells.Mol Cell Biol 15:5403–5411

28. Cheatham B, Vlahos CJ, Cheatham L, Wang L, Blenis J,Kahn CR 1994 Phosphatidylinositol 3-kinase activation isrequired for insulin stimulation of pp70 S6 kinase, DNAsynthesis, and glucose transporter translocation. MolCell Biol 14:4902–4911

29. Frevert EU, Kahn BB 1997 Differential effects of consti-tutively active phosphatidylinositol 3-kinase on glucosetransport, glycogen synthase activity, and DNA synthesisin 3T3–L1 adipocytes. Mol Cell Biol 17:190–198

30. Martin SS, Haruta T, Morris AJ, Klippel A, Williams LT,Olefsky JM 1996 Activated phosphatidylinositol 3-kinaseis sufficent to mediate actin rearrangement and GLUT4translocation in 3T3–L1 adipocytes. J Biol Chem271:17605–17608

31. Withers D, Gutierrez J, Towery H, Burks D, Ren J, PrevisS, Zhang Y, Bernal D, Pons S, Shulman G, Bonner-WeirS, White M 1998 Disruption of IRS-2 causes type 2diabetes in mice. Nature 391:900–904

32. White MF, Livingston JN, Backer JM, Lauris V, Dull TJ,

Ullrich A, Kahn CR 1988 Mutation of the insulin receptorat tyrosine 960 inhibits signal transmission but does notaffect its tyrosine kinase activity. Cell 54:641–649

33. Margolis B 1996 The PI/PTB domain: a new protein in-teraction domain involved in growth factor receptor sig-naling. J Lab Clin Med 128:235–241

34. Gustafson TA, He W, Craparo A, Schaub CD, O’Neill TJ1995 Phosphotyrosine-dependent interaction of SHCand insulin receptor substrate-1 with the NPEY motif ofthe insulin receptor via a novel non-SH2 domain. Mol CellBiol 15:2500–2508

35. Kavanaugh MW, Turck CW, Williams LT 1995 PTBdomain binding to signaling proteins through a se-quence motif containing phosphotyrosine. Science268:1177–1179

36. Fantin V, Sparling J, Slot J, Keller S, Lienhard G, Lavan B1998 Characterization of insulin receptor substrate 4 inhuman embryonic kidney 293 cells. J Biol Chem273:10726–10732

37. Rondinone CM, Wang L-M, Lonnroth P, Wesslau C,Pierce JH, Smith U 1997 Insulin receptor substrate (IRS)1 is reduced and IRS-2 is the main docking protein forphosphatidylinositol 3-kinase in adipocytes from sub-jects with non-insulin-dependent diabetes mellitus. ProcNatl Acad Sci USA 94:4171–4175

38. Gerrits P, Olson A, Pessin J 1993 Regulation of theGLUT4/muscle-fat glucose transporter mRNA in adiposetissue of insulin-deficient diabetic rats. J Biol Chem268:640–644

39. Tanti JF, Gremeaux T, Grillo S, Calleja V, Klippel A,Williams LT, Van Obberghen E, Le Marchand-Brustel Y1996 Overexpression of a constitutively active form ofphosphatidylinositol 3-kinase is sufficient to promoteGlut 4 translocation in adipocytes. J Biol Chem271:25227–25232

40. Katagiri H, Asano T, Ishihara H, Inukai K, Shibasaki Y,Kikuchi M, Yazaki Y, Oka Y 1996 Overexpression ofcatalytic subunit p110alpha of phosphatidylinositol 3-ki-nase increases glucose transport activity with transloca-tion of glucose transporters in 3T3–L1 adipocytes. J BiolChem 271:16987–16990

41. Frevert EU, Bjorbaek C, Venable CL, Keller SR, Kahn BB1998 Targeting of constitutively active phosphoinositide3-kinase to GLUT4-containing vesicles in 3T3–L1 adipo-cytes. J Biol Chem 273:25480–25487

42. Choi T, Huang M, Gorman C, Jaenisch R 1991 A genericintron increases gene expression in transgenic mice. MolCell Biol 11:3070–3074

43. Kozak M 1987 At least six nucleotides preceding theAUG initiator codon enhance translation in mammaliancells. J Mol Biol 196:947–950

44. Honnor RC, Dhillon GS, Londos C 1985 cAMP-depen-dent protein kinase and lipolysis in rat adipocytes. I. Cellpreparation, manipulation, and predictability in behavior.J Biol Chem 260:15122–15129

45. Chen H, Wertheimer SJ, Lin CH, Katz SL, Amrein KE,Burn P, Quon MJ 1997 Protein-tyrosine phosphatasesPTP1B and syp are modulators of insulin-stimulatedtranslocation of GLUT4 in transfected rat adipose cells.J Biol Chem 272:8026–8031

MOL ENDO · 1999 Vol 13 No. 3514