Comparative effects of GTPgammaS and insulin on the activation of Rho, phosphatidylinositol 3-kinase, and protein kinase N in rat adipocytes. Relationship to glucose transport.

Electroporation of rat adipocytes with guanosine 5'-3-O-(thio)triphosphate (GTPgammaS) elicited sizable insulin-like increases in glucose transport and GLUT4 translocation. Like insulin, GTPgammaS activated membrane phosphatidylinositol (PI) 3-kinase in rat adipocytes, but, unlike insulin, this activation was blocked by Clostridium botulinum C3 transferase, suggesting a requirement for the small G-protein, RhoA. Also suggesting that Rho may operate upstream of PI 3-kinase during GTPgammaS action, the stable overexpression of Rho in 3T3/L1 adipocytes provoked increases in membrane PI 3-kinase activity. As with insulin treatment, GTPgammaS stimulation of glucose transport in rat adipocytes was blocked by C3 transferase, wortmannin, LY294002, and RO 31-8220; accordingly, the activation of glucose transport by GTPgammaS, as well as insulin, appeared to require Rho, PI 3-kinase, and another downstream kinase, e.g. protein kinase C-zeta (PKC-zeta) and/or protein kinase N (PKN). Whereas insulin activated both PKN and PKC-zeta, GTPgammaS activated PKN but not PKC-zeta. In transfection studies in 3T3/L1 cells, stable expression of wild-type Rho and PKN activated glucose transport, and dominant-negative forms of Rho and PKN inhibited insulin-stimulated glucose transport. In transfection studies in rat adipocytes, transient expression of wild-type and constitutive Rho and wild-type PKN provoked increases in the translocation of hemagglutinin (HA)-tagged GLUT4 to the plasma membrane; in contrast, transient expression of dominant-negative forms of Rho and PKN inhibited the effects of both insulin and GTPgammaS on HA-GLUT4 translocation. Our findings suggest that (a) GTPgammaS and insulin activate Rho, PI 3-kinase, and PKN, albeit by different mechanisms; (b) each of these signaling substances appears to be required for, and may contribute to, increases in glucose transport; and (c) PKC-zeta may contribute to increases in glucose transport during insulin, but not GTPgammaS, action.

GTP␥S, 1 like insulin, has been found to activate GLUT4 translocation and/or glucose transport in rat adipocytes (1,2) and 3T3/L1 adipocytes (3). The mechanisms whereby GTP␥S and insulin activate GLUT4 translocation and glucose transport, however, are unclear. In 3T3/L1 cells, unlike insulin, GTP␥S was not found to activate cytosolic phosphatidylinositol (PI) 3-kinase (3), and this suggested that (a) PI 3-kinase was not essential for the activation of glucose transport and (b) GTP␥S may operate through different or more distal processes. In keeping with the latter possibility, small G-proteins in the Rab group are present in GLUT4 vesicles, appear to translocate or mobilize in response to insulin stimulation (4), and could act as direct mediators for GTP␥S stimulation of glucose transport; accordingly, GTP␥S-stimulated glucose transport is only partly inhibited by the PI 3-kinase inhibitor, wortmannin, in 3T3/L1 adipocytes (5), and GTP␥S therefore appears to act, at least partially, independently of PI 3-kinase in 3T3/L1 cells. On the other hand, we have recently found that the small G-protein, RhoA, is activated by insulin in rat adipocytes, and, based upon Clostridium botulinum C3 transferase sensitivity, Rho appears to be required for insulin-stimulated glucose transport in these cells (6). Further, it seems clear that Rho is directly activated by GTP␥S in rat adipocytes, since it was found that the addition of GTP␥S to rat adipocyte homogenates stimulates the translocation of Rho to plasma membranes and Rho-dependent activation of phospholipase D (6). Of further note, it has been reported that GTP-Rho directly activates PI 3-kinase in some (7), but not all (8,9), cell-free systems. Presently, we examined the possibility that GTP␥S activates PI 3-kinase via Rho in rat adipocytes. We also examined the role of Rho, PI 3-kinase, and protein kinases that are known to be downstream of Rho and PI 3-kinase (e.g. PKN and protein kinase C-(PKC-)) in the activation of glucose transport during treatment of rat adipocytes with GTP␥S or insulin. Our findings suggested that (a) both GTP␥S and insulin, albeit by different mechanisms, activate Rho, PI 3-kinase, and PKN; and (b) each of these factors may be required for, and may contribute to, the activation of GLUT4 translocation and glucose transport in rat adipocytes.

EXPERIMENTAL PROCEDURES
Rat Adipocytes (Preparation, Incubation, and Electroporation)-Rat adipocytes were prepared by collagenase digestion of epididymal fat pads obtained from male Sprague-Dawley rats weighing approximately 200 -250 g, as described previously (6). The adipocytes were suspended in glucose-free Krebs Ringer phosphate (KRP) buffer containing 1% bovine serum albumin for acute incubations, or in Dulbecco's modified Eagle's medium (DMEM; Life Technologies, Inc.) containing 1% bovine serum albumin for overnight incubations. GTP␥S (Sigma or ICN) and C. botulinum C3 transferase (List) were introduced into adipocytes (in 25 or 50% suspensions, cell volume/total volume) by electroporation (Bio-Rad Gene Pulsar; 350 V and 960 microfarads with a time constant of 12 ms) in either an intracellular buffer (118 mM KCl, 5 mM NaCl, 0.38 mM CaCl 2 , 1 mM EGTA, 1.2 mM Mg 2 SO 4 , 1.2 mM KH 2 PO 4 , 3 mM sodium pyruyate, 25 mM HEPES, and 20 mg/ml bovine serum albumin) or in DMEM, respectively, as described in the text.
Glucose Transport Studies in Rat Adipocytes-Glucose transport was measured in adipocytes suspended (6%, cell volume/total volume) in glucose-free KRP buffer as described by Karnam et al. (6). Where indicated, inhibitor (wortmannin, Sigma; LY294002, Biomol; RO 31-8220, Alexis) was added to the incubation 15 min prior to agonist addition. The cells were then treated with vehicle alone (control), insulin (Elanco), or the indicated concentrations of GTP␥S. In the case of GTP␥S treatment, immediately following electroporation in intracellular buffer (see above), the cells were diluted with glucose-free KRP buffer containing indicated inhibitor concentrations. After 30 min of treatment with vehicle (controls), 10 nM insulin, or 0 -500 M GTP␥S, the uptake of 2-[ 3 H]deoxyglucose (2-DOG; 0.1 mM; NEN Life Science Products) was measured over a 1-min period as described (6). In these assays, it should be noted that cytochalasin B blank values, which reflect trapping of medium or nonspecific uptake, were relatively small (approximately 10% of stimulated values) and were not influenced significantly by electroporation or overnight incubation. In cells that were assayed directly without electroporation, insulin-induced increases in the uptake of 2-DOG generally ranged from 4-to 10-fold. In cells that were electroporated and then immediately assayed, there usually was a small, but variable, increase in basal 2-DOG uptake, with little or no change in maximal insulin-stimulated values; consequently, the relative effect of insulin on 2-DOG uptake generally tended to be slightly less, i.e. approximately 2-3-fold, in electroporated adipocytes (this is illustrated in Fig. 8). Overnight incubation of adipocytes also increased basal transport activity, and the relative insulin effect was similarly diminished to approximately 2-3-fold in these cells. However, the effect of combined electroporation and overnight incubation on 2-DOG uptake was not significantly different from that of either electroporation or overnight incubation alone, since insulin effects on 2-DOG uptake were also approximately 2-3-fold in these cells.
GLUT4 Translocation Studies in Transiently Transfected Rat Adipocytes-Effects of transiently transfected Rho and PKN on GLUT4 translocation were measured in rat adipocytes co-transfected with hemagglutinin (HA)-tagged GLUT4, as described by Quon et al. (10,11). In brief, adipocytes (as a 50% suspension in DMEM) were electroporated in the presence of eukaryotic expression vector pCIS2 containing cDNA encoding HA-tagged GLUT4 (kindly supplied by Drs. Michael Quon and Simeon Taylor) and either (a) pCDNA3 (Invitrogen) eukaryotic expression vector alone or pCDNA3 containing cDNA encoding (i) wild-type Rho (kindly supplied by Dr. David Lambeth), (ii) V14 mutant, constitutive Rho (kindly supplied by Dr. Alan Hall), or (iii) dominant negative Rho (kindly supplied by Dr. Gary Bokoch); or (b) pTB701 eukaryotic expression vector alone or pTB701 containing cDNA encoding (i) wildtype or (ii) dominant negative, kinase-inactive mutant forms of PKN (prepared by Dr. Ono). After overnight incubation to allow time for expression of cDNA inserts (see Refs. 10 and 11 and verified by expression of HA-tagged forms of GLUT4, Rho, and PKN), the cells were equilibrated in glucose-free KRP buffer and treated with or without 10 nM insulin for 30 min, prior to the addition of 2 mM KCN and measurement of cell count and cell-surface, HA-tagged GLUT4 as described (10,11); for the latter purpose, the primary anti-HA mouse monoclonal antibody was obtained from Babco, and 125 I-labeled second antibody (sheep anti-mouse IGG) was obtained from Amersham Pharmacia Biotech. Blank values (nonspecific binding), obtained by incubating cells transfected with vectors alone (i.e. without the HA-GLUT4 and other inserts), were subtracted from values observed in cells in which HA-GLUT4 was expressed. In each experiment, a single batch of adipocytes was used, so that the level of HA-tagged GLUT4 was identical in each experimental group, and treatment groups could be directly compared with each other. Absolute values of 125 I bound per 10 6 cells varied somewhat from experiment to experiment, depending upon the batches of antibodies used and the level of 125 I in the second antibody; nevertheless, relative changes induced by GTP␥S, insulin, and other treatments were similar in all experiments.
Stable Transfection Studies in 3T3/L1 Cells-Effects of stably transfected, wild-type Rho or PKN on glucose transport (2-DOG uptake) and membrane-associated PI 3-kinase activity were evaluated in 3T3/L1 fibroblasts and/or adipocytes, using transfection methods described previously (12). Effects of dominant negative forms of Rho and PKN were also studied in 3T3/L1 fibroblasts, but these forms inhibited adipogenesis, thus precluding experiments with plasmids containing these cDNA inserts in 3T3/L1 adipocytes. In these stable transfection experiments, all cells (untransfected controls or cells transfected with pCDNA3 vector alone or pCDNA3 containing cDNAs encoding Rho (supplied by Dr. David Lambeth) or PKN (supplied by Dr. Peter Parker)) were grown, differentiated, and assayed simultaneously. As in previous transfection studies (12), all reported clones (selected by G418 resistance) were shown to contain immunoreactive GLUT4 and GLUT1 levels that were indistinguishable from those observed in untransfected control cells; consequently, alterations in glucose transport in Rho or PKN-transfected cells could not be ascribed to changes in levels of GLUT4 or GLUT1. In some experiments, a tetracycline-inducible system (CLONTECH, Tet-On kit) was used to turn on the expression of transfected, wild-type Rho in 3T3/L1 fibroblasts 72 h prior to measurement of 2-DOG uptake (described more fully in text, see Fig. 7).
Enzyme Assays for PI 3-Kinase, PKC-, and PKN-PI 3-kinase enzyme activity was measured either in total membranes or in IRS-1 immunoprecipitates of rat adipocytes or 3T3/L1 adipocytes, as described previously (13). Immunoprecipitable PKC-enzyme activity was measured as described (12,14). To measure immunoprecipitable PKN enzyme activity, cells were lysed in buffer containing 150 mM NaCl, 250 mM sucrose, 20 mM Tris/HCl (pH, 7.5), 1.2 mM EGTA; 1 mM EDTA, 5 mM MgCl 2 , 20 mM ␤-mercaptoethanol, 25 mM NaF, 3 mM Na 4 P 2 O 7 , 10 mM Na 3 VO 4 , 20 g/ml aprotinin, 20 g/ml leupeptin, 1 mM PMSF, 1% Triton X-100, and 2 M Microcystin-LR (Calbiochem). Rabbit anti-PKN polyclonal antibody (raised in Dr. Ono's laboratories) was added in sufficient amounts to quantitatively precipitate PKN from 300 g of lysate protein. After overnight incubation at 0 -4°C, the precipitate was collected on Sepharose A/G beads (Santa Cruz Laboratories), washed three times with lysis buffer and twice with assay buffer (50 mM Tris/HCl (pH, 7.5), 10 mM MgCl 2 , 1 mM EGTA, 1 mM NaF, 1 mM Na 3 VO 4 , 1 mM Na 4 P 2 O 7 , 10 mM ␤-glycerophosphate, 100 M phenylmethylsulfonyl fluoride), and then incubated at 30°C for 6 min in the presence of 40 M serine-25 PKC-␣ pseudosubstrate (Life Technologies) and 50 M ATP containing 2 Ci of [␥-32 P]ATP (NEN Life Science Products). Aliquots of the reaction mixture were spotted on p81 filter paper, washed in 30% acetic acid, and counted for 32 P. It may be noted that phosphatidylserine was not present in PKN assays, and, moreover, PKN activity was not increased by the addition of phosphatidylserine, Ca 2ϩ , or diolein: it may therefore be surmised that PKN immunoprecipitates were not contaminated with significant amounts of conventional, novel, or atypical PKCs. Fig. 1, electroporation of rat adipocytes in the presence of increasing amounts of GTP␥S led to dose-related increases in 2-DOG uptake. At a concentration of 500 M, GTP␥S-stimulated increases in 2-DOG uptake were in most cases slightly less than those of insulin, as measured in cells electroporated in parallel. We did not attempt to use higher concentrations of GTP␥S, but it may be noted that we probably did not reach maximal transport rates with 500 M GTP␥S, electroporation opens membrane pores only fleetingly, and intracellular GTP␥S concentrations were most likely less than those present in the electroporation buffer. In addition to increasing glucose transport, 500 M GTP␥S provoked increases in the translocation of HA-tagged GLUT4 to the plasma membrane (cell surface 125 I-labeled anti-HA antibody (reflecting the level of exofacial HA-tagged GLUT4) was 1772 Ϯ 238 (n ϭ 4) versus 3748 Ϯ 321 (n ϭ 6) cpm/10 6 cells (p Ͻ 0.005, t test), control versus GTP␥S, respectively); these increases in HA-GLUT4 translocation were in some experimental groups similar to those provoked by insulin (viz. approximately 2-fold) or, in some groups (e.g. see below), slightly less.

Studies of Glucose Transport in Rat Adipocytes-As shown in
Since PI 3-kinase is required for insulin effects on glucose transport, it was of interest to see if inhibitors of PI-3-kinase

Effects of GTP␥S and Insulin on Rho, PI 3-Kinase, and PKN
altered the effects of GTP␥S on 2-DOG uptake. As shown in Fig. 1, A and B, both 100 nM wortmannin and 100 M LY294002, (concentrations that fully inhibit PI 3-kinase) fully inhibited GTP␥S-stimulated, as well as insulin-stimulated, 2-DOG uptake. These findings suggested that GTP␥S, like insulin, requires PI 3-kinase for the activation of glucose transport in rat adipocytes.
We have previously reported that the effects of insulin on glucose transport are blocked by C. botulinum C3 transferase, which specifically ADP-ribosylates, inhibits, and, after overnight incubation of rat adipocytes, leads to a complete loss of immunoreactive RhoA in these cells (see Ref. 6). As shown in Fig. 1C, overnight C3 transferase treatment blocked subsequent effects of both GTP␥S and insulin on 2-DOG uptake in rat adipocytes. These results (as well as results from transfection studies; see below) suggested that Rho is required for the effects of GTP␥S, as well as insulin, on glucose transport in rat adipocytes.
Studies of PI 3-Kinase Activation-Since wortmannin and LY294002 fully blocked the effects of both GTP␥S and insulin on 2-DOG uptake in the rat adipocyte, we questioned whether GTP␥S activates PI 3-kinase. As shown in Fig. 2, both GTP␥S and insulin provoked increases in membrane-associated PI 3-kinase activity in rat adipocytes. Interestingly, C3 transferase blocked the activating effect of GTP␥S, but not insulin, on membrane-associated PI 3-kinase (Fig. 2C). Also, insulin-induced increases in PI 3-kinase immunoreactivity and enzyme activity that are specifically associated with IRS-1, were not blocked by C3 transferase treatment (Fig. 3). These findings suggested that activating effects of insulin on PI 3-kinase, both in total membranes and as specifically activated through IRS-1 in the rat adipocyte, were not dependent upon Rho. On the other hand, activating effects of GTP␥S on membrane-associated PI 3-kinase in the rat adipocyte appeared to be fully dependent upon Rho. In keeping with a role for Rho in PI 3-kinase activation, as described below, stable overexpression of Rho in 3T3/L1 adipocytes led to increases in membraneassociated PI 3-kinase activity.
Studies on Effects of RO 31-8220 on PKN Activity and Glucose Transport-Considerable evidence suggests that one or more protein kinases, apparently distal to PI 3-kinase, is (are) required for insulin stimulation of glucose transport. For example, we have previously shown that the bisindolemaleimidetype PKC inhibitor, RO 31-8220, inhibits insulin-stimulated glucose transport, without inhibiting the activation of PI 3-kinase by insulin (15). Recently, we have found (14)  . Presently, we found that RO 31-8220 inhibited immunoprecipitated PKN with an IC 50 of approximately 30 nM (Fig. 4). Although the exact identity of the RO 31-8220-sensitive kinase that is required for insulin-stimulated glucose transport is not certain, it was of interest to find that RO 31-8220 inhibited the effects of GTP␥S, as well as insulin, on 2-DOG uptake in intact rat adipocytes (Fig. 5). However, the concentrations of RO 31-8220 that were required to inhibit GTP␥S effects on 2-DOG uptake were considerably less (IC 50 Ͻ 1 M) than those required for inhibition of insulin effects (IC 50 ϭ 5 M) on 2-DOG uptake in the rat adipocyte (Fig. 5). It therefore seems likely that different RO 31-8220-sensitive protein kinases are required for glucose transport effects of GTP␥S and insulin in the rat adipocyte.
Studies on the Activation PKC-and PKN-Inasmuch as RO 31-8220 does not inhibit the activity or activation of PI 3-kinase

FIG. 1. Effects of wortmannin (WORT) (A), LY294002 (B), and C3 transferase (C3) on GTP␥S-and insulin-stimulated 2-DOG uptake in rat adipocytes.
A and B, cells were treated with 100 nM wortmannin or 100 M LY294002 for 15 min, electroporated in intracellular buffer with or without the indicated concentrations of GTP␥S, and then diluted with glucose-free KRP medium and treated with or without 10 nM insulin as indicated. After 30 min of treatment with GTP␥S or insulin, the uptake of 2-DOG uptake was measured over 1 min. C, cells were electroporated in DMEM with or without C3 transferase (0.5 g/ml), and incubated overnight to deplete immunoreactive Rho in C3 transferase-treated cells (see Ref. 6). The cells were then electroporated a second time in intracellular buffer with or without 500 M GTP␥S and then diluted with glucose-free KRP medium and treated with or without 10 nM insulin, as indicated. After 30 min of treatment with GTP␥S or insulin, 2-DOG uptake was measured over 1 min. Values are mean Ϯ S.E. of four determinations.

Effects of GTP␥S and Insulin on Rho, PI 3-Kinase, and PKN
in the rat adipocyte (15), as alluded to above, it may be surmised that one or more protein kinases, distinct from PI 3-kinase, is required during the activation of glucose transport. Accordingly, we questioned whether GTP␥S or insulin activates PKC-or PKN, i.e. kinases that appear to be downstream of PI 3-kinase and/or Rho. Whereas insulin (see Ref. 14 for more detailed studies) activated PKC-, GTP␥S, if anything, diminished the activity of immunoprecipitable PKC-in intact adipocytes (Table I). In addition, GTP␥S, unlike insulin, failed to activate PKB (data not shown). On the other hand, GTP␥S activated immunoprecipitable PKN mildly (23%), but significantly, in intact adipocytes (Table II) and more dramatically in adipocyte homogenates (Table III). (Note that the in vitro in-cubation of adipocyte homogenate in low Mg 2ϩ conditions prior to immunoprecipitation markedly lowered basal immunoprecipitable PKN activity, and this probably facilitated the observance of greater relative effects of GTP␥S in the cell-free system.) Similarly, insulin provoked approximately 60% increases in immunoprecipitable PKN activity throughout a 1-10-min treatment period in intact rat adipocytes, and these increases were blocked by C3 transferase, but not by wortmannin (Fig.  6). Thus, in keeping with our previous report that PI 3-kinase is not required for insulin stimulation of GTP-loading of Rho (6), PI 3-kinase did not appear to be required for insulininduced activation of PKN. Also, in keeping with the possibility that PKN is downstream of Rho, Rho appeared to be required for insulin-induced activation of PKN.
Transfection Studies-Since our findings with C3 transferase suggested that Rho is required for (a) GTP␥S-and insulininduced increases in glucose transport and (b) GTP␥S-induced increases in PI 3-kinase, we questioned whether Rho itself could provoke increases in PI 3-kinase activity and glucose transport and/or GLUT4 translocation. To pursue these questions and to further examine the requirement for Rho in glucose transport, we used several transfection approaches. First, in both 3T3/L1 fibroblasts and adipocytes, the stable overexpression of wild-type Rho led to increases in both immunoreactive Rho and basal and insulin-stimulated glucose transport (Fig. 7); in addition, stably transfected Rho provoked a 73 Ϯ 10% increase (mean Ϯ S.E.; n ϭ 7; p Ͻ 0.001, paired t test, Rho transfectants versus untransfected and vector-transfected con-trols) in membrane-associated PI 3-kinase enzyme activity in 3T3/L1 adipocytes. Second, in rat adipocytes, transient transfection of wild-type and, even more so, constitutive, Rho led to increases in the translocation of transiently co-transfected HAtagged GLUT4 to the plasma membrane (Fig. 8); in contrast, dominant-negative Rho largely inhibited the effects of insulin (Fig. 8) and fully inhibited the effects of GTP␥S (Fig. 8) on HA-GLUT4 translocation. Third, we stably transfected 3T3/L1 fibroblasts with eukaryotic expression vectors (pTRE and pTet-On; CLONTECH Tet-On kit) that (a) placed the expression of wild-type Rho under the control of a promoter (PminCMV), which, in turn, is controlled by a tetracycline response element, and (b) provided the tetracycline-controlled transcription activator, i.e. a Tet repressor (mutated to reverse its response to tetracycline and thereby activate transcription) fused to the VP16 activation domain of a herpes simplex virus. Upon the addition of doxycycline to these cells (i.e. after they had become confluent), there were increases in (a) Rho expression and (b) control and insulin-stimulated 2-DOG uptake (Fig. 7); this approach avoided having Rho overexpressed until 72 h prior to

TABLE I Effects of GTP␥S and insulin on immunoprecipitable PKCactivity in rat adipocytes
All adipocytes were electroporated in intracellular buffer with or without 500 M GTP␥S and then treated with or without 10 nM insulin, as indicated. After a 10-min incubation, cells were homogenized in ice-cold 250 mM sucrose, 20 mM Tris-HCl (pH 7.5), 1.2 mM EGTA, 20 mM ␤-mercaptoethanol, 1 mM phenylmethylsulfonyl fluoride, 20 g/ml leupeptin, 20 g/ml aprotinin, 1 mM Na 3 VO 4 , 1 mM Na 4 P 2 O 7 , and 1 mM NaF. Homogenates were centrifuged at 500 ϫ g for 10 min to remove nuclei, debris, and the fat cake. After adding 1% Triton X-100, 0.5% Nonidet, and 150 mM NaCl, PKC-was immunoprecipitated (polyclonal antibodies from Santa Cruz Biotechnology) and assayed as described previously (12,14). Values are the mean Ϯ S.E. of n determinations (in parentheses). p was determined by t test.   rat adipocyte homogenates Adipocytes were homogenized in buffer containing 250 mM sucrose, 20 mM Tris-HCl (pH 7.5), 1.2 mM EGTA, 1 mM EDTA, 20 g/ml aprotinin, 20 g/ml leupeptin, 1 mM phenylmethylsulfonyl fluoride, and 20 mM ␤-mercaptoethanol, 1 mM Na 3 VO 4 , 1 mM Na 4 P 2 O 7 , and 1 mM NaF. Homogenates were centrifuged at 500 ϫ g for 10 min to remove nuclei, debris, and the fat cake, incubated first for 20 min at 37°C to facilitate the release of endogenous GDP and GTP, and then incubated for 5 min with 10 mM MgCl 2 , with or without 20 M GTP␥S, as indicated. After the addition of 1% Triton X-100, 0.5% Nonidet, 150 mM NaCl, and other substances (see "Experimental Procedures"), PKN was immunoprecipitated and assayed (see "Experimental Procedures"). Values are the mean Ϯ S.E. of n determinations (in parentheses).  6. Effects of insulin, C3 transferase, and wortmannin on immunoprecipitable PKN activity in rat adipocytes. Where indicated, adipocytes were either incubated directly for 15 min with or without 100 nM wortmannin (first six groups), or electroporated with or without 1 g/ml C3 transferase and then incubated for 20 h (last two groups) as in Figs. 1C and 3. After these initial treatments, the cells were equilibrated in glucose-free KRP medium and treated with or without 10 nM insulin for the indicated times (1, 5, or 10 min), following which PKN was immunoprecipitated and assayed for enzyme activity as described under "Experimental Procedures." Shown here are mean Ϯ S.E. values of n determinations (shown in parentheses). Although not shown separately in the figure, insulin effects on PKN activity were comparable in adipocytes that were used directly or placed into primary culture for 20 h. p was determined by t test. NS, not significant. using cells for 2-DOG assay. Fourth, stable expression of dominant-negative Rho partially (25-35%) inhibited insulin-stimulated 2-DOG uptake in 3T3/L1 fibroblasts (Fig. 9). From these transfections studies, it appears that (a) Rho itself can activate PI 3-kinase, GLUT4 translocation, and glucose transport; and (b) in keeping with studies using C3 transferase, Rho is required for effects of both insulin and GTP␥S on GLUT4 translocation and glucose transport in rat adipocytes.
In keeping with the possibility that PKN may operate downstream of Rho during glucose transport activation, we found that transient expression of wild-type PKN in rat adipocytes resulted in increases in the translocation of co-transfected HAtagged GLUT4 to the plasma membrane (Fig. 9); in contrast, dominant-negative PKN partially inhibited the effects of insulin (Fig. 9) and fully inhibited the effects of GTP␥S (Fig. 8) on HA-GLUT4 translocation. Similarly, we found that stable over-

FIG. 7. Effects of stable (A) and tetracycline-inducible (B) overexpression of Rho on 2-DOG uptake in 3T3/L1 fibroblasts and adipocytes.
A, fibroblasts were stably transfected with pCDNA3 alone (vector, V) or pCDNA3 containing cDNA encoding wild-type Rho (R). Clones were selected by G418 resistance, and grown, differentiated, and assayed in parallel with untransfected (0) cells. Cells were incubated in glucose-free KRP medium and, after treatment for 30 min, with indicated concentrations (0, 5, 100 nM) of insulin, 5-min uptake of 2-DOG was measured. Values are the mean Ϯ S.E. of (n) clones, each assayed in triplicate at each insulin concentration. Insets show increases in immunoreactivity in cells transfected (TX) with Rho (R). B, cells were stably transfected with (a) plasmid (pTet-On) that contains cDNA encoding a mutated tetracycline repressor fused to the VP16 activation domain of a herpes simplex virus controlled by a constitutive P cmv promoter and (b) a plasmid (pTRE) containing cDNA encoding wildtype Rho whose expression is dependent upon a tetracycline response element that controls the activation of the P minCMV promoter and subsequent transcription of the Rho cDNA (prepared according to instructions in the CLONTECH Tet-On kit). Colonies were selected both by resistance to G418 and hygromycin, grown to confluence in 24-well plates, induced by doxycycline for 72 h as indicated, and then (after changing to a glucose-free KRP medium) treated with the indicated concentrations of insulin for 30 min, prior to measurement of 2-DOG uptake over 5 min. Values are mean Ϯ S.E. of n (shown in parentheses) clones, each assayed in triplicate at each insulin concentration. Results in clones treated with empty vectors alone were indistinguishable from results in untransfected cells, and these results were pooled (controls in panel A, left). The insets show levels of immunoreactive Rho in noninduced controls (Ϫ) and tetracycline-induced (ϩ) cells; note that increases were observed only in tetracycline-treated cells that were transfected with plasmid containing cDNA encoding Rho. P values (t test) were determined by comparison of results in tetracycline-treated cells that contained and expressed the Rho insert, relative to results in control cells that contained vectors lacking the Rho insert. Note that 2-DOG uptake was not influenced by tetracycline treatment in the control group. expression of PKN enhanced, and dominant negative PKN partially (55%) inhibited, insulin effects on glucose transport in 3T3/L1 fibroblasts (Fig. 9). DISCUSSION Our findings suggested that GTP␥S activates PI 3-kinase through a Rho-dependent mechanism in intact rat adipocytes. GTP-Rho has also been found to activate PI 3-kinase in platelet homogenates (7) but, for uncertain reasons, not in other cellfree systems (8,9). Our findings with wortmannin and LY294002 also suggested that PI 3-kinase is required for the activation of glucose transport during GTP␥S stimulation of rat adipocytes. Thus, GTP␥S, as presently used, did not appear to activate glucose transport in rat adipocytes simply by activating small G-proteins such as Rab that are thought to function distal to PI 3-kinase in regulating Glut4 translocation (4).
Whereas GTP␥S appeared to activate PI 3-kinase through Rho, insulin effects on PI 3-kinase were largely independent of Rho. Thus, although insulin activates Rho (6), Rho was not a major contributor to insulin-stimulated PI 3-kinase activation, which probably occurs largely through tyrosine phosphorylation of IRS-1 and/or other proteins (16). Along these lines, it is pertinent to note that PI 3-kinase is required for the translocation, but not GTP loading, of Rho during insulin action (6); thus, PI 3-kinase can operate upstream (e.g. during insulin action), as well as downstream (e.g. during GTP␥S activation), of Rho. During insulin action, Rho may translocate to specific sites of PI 3-kinase-induced increases in polyphosphoinositides (accordingly, we have found that Rho avidly binds to artificial phosphatidylcholine vesicles containing 5% PI-3, 4-(PO 4 ) 2 , PI-3,4,5-(PO 4 ) 3 , or PI-4,5-(PO 4 ) 2 ), 2 and this may coordinate certain actions of PI 3-kinase and Rho. During GTP␥S action, GTP␥S stimulates the translocation of Rho to plasma and microsomal membranes (6), and this may explain how membrane-activated PI 3-kinase is activated by GTP-Rho.
In addition to requirements for Rho and PI 3-kinase, our findings with RO 31-8220 suggested a requirement for one or more protein kinases in the activation of glucose transport by GTP␥S as well as by insulin. In the case of insulin, the required protein kinase(s) appears to operate distally to, or in parallel with, PI 3-kinase, since RO 31-8220 does not inhibit insulininduced activation of either PI 3-kinase (15) or PI 3-kinase-dependent PKB activation 2 ; presumably, the same situation pertains during GTP␥S action, i.e. the RO 31-8220-sensitive protein kinase(s) required for glucose transport is distal or parallel to PI 3-kinase. Although the identity of the protein kinase is uncertain, note that both PKC-and PKN are activated by PI 3-kinase lipid products (i.e. polyphosphoinositides) (17)(18)(19), and PKN is directly activated by GTP-Rho (20,21). Also, as reported for other bisindolemaleimides (see Refs. 22 and 23), we have found (14) that RO 31-8220 inhibits recombinant conventional (␣, ␤, and ␥) and novel (␦, ⑀, and ) PKCs at relatively low concentrations (IC 50 values of 15-100 nM) and the atypical PKC, PKC-, at relatively high concentrations (IC 50 of 1-4 M). Presently, we found that RO 31-8220, at relatively low concentrations (IC 50 ϭ 30 nM), inhibited immunoprecipitated PKN. Presumably, inhibitory effects of RO 31-8220 on PKC and PKN reflect homology in the catalytic domains of PKN and most PKCs (24).
With respect to PKC-and PKN as RO 31-8220-inhibitable protein kinases that may be required for glucose transport during the actions of GTP␥S and insulin, the following are germane. First, PKC-was activated by insulin, but not by GTP␥S; thus, PKC-seems unlikely to be involved in the action of GTP␥S but may play a role during insulin action. Second, as in other systems in which GTP-Rho directly activates PKN (20,21), we found that GTP␥S activated both Rho and PKN, and insulin activated PKN by a Rho-dependent mechanism; accordingly, PKN is probably activated via Rho during the actions of both GTP␥S and insulin in rat adipocytes. On the other hand, our studies suggested that different RO 31-8220-sensitive protein kinases were required for glucose transport effects of GTP␥S and insulin in rat adipocytes; thus, insulin required a protein kinase sensitive to higher (IC 50 of 4 -5 M) concentrations of RO 31-8220, e.g. PKC-, whereas GTP␥S required a protein kinase sensitive to lower (IC 50 Ͻ 1 M) concentrations of RO 31-8220, e.g. PKN. Although these findings with RO 31-8220 might suggest that PKN is required for glucose transport effects of GTP␥S, but not insulin, note that expression of dominant-negative PKN partially inhibited (a) the effects of insulin as well as GTP␥S on the translocation of HA-GLUT4 to 2  the plasma membrane in rat adipocytes and (b) insulin effects on glucose transport in 3T3/L1 fibroblasts. It is presently uncertain if these seemingly divergent findings reflect shortcomings in our experimental approaches (e.g. effective local concentrations of inhibitors such as RO 31-8220 at specific enzyme sites in situ are uncertain, and transfections of dominantnegative proteins may cause untoward effects).
Our observation of activation of membrane-associated PI 3-kinase by GTP␥S in rat adipocytes appears to differ from that of a previous report in which GTP␥S failed to activate cytosolic PI 3-kinase in 3T3/L1 adipocytes (3); this may reflect differences in cell types or the fact that we measured membrane, rather than cytosolic, PI 3-kinase activity. Along these lines, note that (a) we found that insulin and GTP␥S activated membrane, but not cytosolic, PI 3-kinase in rat adipocytes; and (b) our failure to observe increases in cytosolic PI 3-kinase may reflect the large pool of insulin-independent PI 3-kinase that is activated indiscriminately during our assays of crude rat adipocyte cytosol. Although we did not examine the effects of GTP␥S on membrane PI 3-kinase activity in 3T3/L1 adipocytes, we did find that overexpression of Rho activated membrane PI 3-kinase in these cells. Accordingly, it may be surmised that, as in rat adipocytes, GTP␥S, by activating Rho, may activate PI 3-kinase in 3T3/L1 adipocytes; this could explain why GTP␥S, at least partly (approximately 50%, as per wortmannin studies in Ref. 5) requires PI 3-kinase for the activation of glucose transport in 3T3/L1 adipocytes; on the other hand, glucose transport effects of GTP␥S that are independent of PI 3-kinase (also approximately 50%; see Ref. 5) may be explained by direct activating effects of GTP␥S on Rab (4) or other G-proteins that act distally to PI 3-kinase.
Finally, it was of interest to find that, in addition to inhibitory effects of C3 transferase and dominant-negative forms of Rho and PKN on GTP␥S-and insulin-stimulated glucose transport and/or GLUT4 translocation, transfected Rho (particularly if constitutively activated) and its downstream kinase, PKN, provoked increases in GLUT4 translocation and/or glucose transport in rat adipocytes and 3T3/L1 cells. It therefore may be conjectured that Rho is not only required for, but may actively participate in, the activation of GLUT4 translocation and glucose transport in the actions of insulin, GTP␥S, and other agonists.
In summary, like insulin, GTP␥S provoked increases in 2-DOG uptake and HA-GLUT4 translocation in rat adipocytes. Also, like insulin, (a) GTP␥S provoked increases in membraneassociated PI 3-kinase, and PI 3-kinase appeared to be required for GTP␥S-induced activation of glucose transport; and (b) both Rho and an RO 31-8220-sensitive protein kinase appeared to be required for GTP␥S-induced activation of glucose transport. In studies of RO 31-8220-sensitive protein kinases, both GTP␥S and insulin activated PKN, and PKN appeared to be required for activation of GLUT4 translocation by GTP␥S and insulin. Unlike insulin, however, GTP␥S appeared to activate PI 3-kinase primarily through Rho, rather than through IRS-1; PKCwas activated by insulin but not by GTP␥S; and effects of GTP␥S on glucose transport were inhibited by lower concentrations of RO 31-8220 than were effects of insulin. It may therefore be surmised that, although there are similarities in the signaling factors (i.e. Rho, PKN, and PI 3-kinase) that are used by GTP␥S and insulin to activate glucose transport, these agents activate Rho and PI 3-kinase by different mechanisms and appear to use different distal protein kinases to activate glucose transport.