Activation and Translocation of Rho (and ADP Ribosylation Factor) by Insulin in Rat Adipocytes APPARENT INVOLVEMENT OF PHOSPHATIDYLINOSITOL 3-KINASE

Insulin reportedly (Standaert, M. L., Avignon, A., Yamada, K., Bandyopadhyay, G., and Farese, R. V. (1996) Biochem. J. 313, 1039-1046) activates phospholipase D (PLD)-dependent hydrolysis of phosphatidylcholine (PC) in plasma membranes of rat adipocytes by a mechanism that may involve wortmannin-sensitive phosphatidylinositol (PI) 3-kinase. Because Rho and ADP ribosylation factor (ARF) activate PC-PLD, we questioned whether these small G-proteins are regulated by insulin and PI 3-kinase. We found that insulin provoked a rapid translocation of both Rho and ARF to the plasma membrane and increased GTP loading of Rho. Wortmannin and LY294002 inhibited Rho translocation in intact adipocytes, and the polyphosphoinositide, PI 4,5-(PO4)2, stimulated Rho translocation in adipocyte homogenates. On the other hand, wortmannin did not block GTP loading of Rho. Guanosine 5′-3-O-(thio)triphosphate stimulated both Rho and ARF translocation and activated PC-PLD in homogenates. C3 transferase, which inhibits and depletes Rho, inhibited PC-PLD activation by insulin in intact adipocytes. C3 transferase also inhibited insulin stimulation of [3H]2-deoxyglucose uptake. Our findings suggest that: (a) insulin translocates Rho by a PI 3-kinase-dependent mechanism, but another factor is responsible for GTP loading of Rho; (b) both Rho and ARF may contribute to PC-PLD activation during insulin action; and (c) Rho may be required for insulin stimulation of glucose transport.

Phospholipase D (PLD)-mediated 1 hydrolysis of phosphati-dylcholine (PC) is a major signaling system for agonists that activate tyrosine kinases. Insulin activates PC-PLD in rat adipocytes (1) and other cells (2)(3)(4), and this may be important for activation of signaling and targeting processes in the plasma membrane (1,4,5). In adipocytes, insulin-induced activation of PC-PLD is inhibited by wortmannin (1), an inhibitor of phosphatidylinositol (PI) 3-kinase, which is activated through its SH 2 domains by specific phosphotyrosine motifs in proteins, such as insulin receptor substrate-1 (IRS-1) (6); thus, PI 3-kinase may be required for PC-PLD activation. Accordingly, polyphosphoinositides, which are increased by insulin (7)(8)(9) through PI 3-kinase action (9) and which may be required for PC-PLD activation (10 -13), may contribute directly to the stimulation of PLD by insulin. However, PC-PLD is also activated by Rho and ARF (10 -15), and we presently questioned whether these small G-proteins are regulated by insulin and PI 3-kinase. In addition, because Rho and PI 3-kinase are thought to be involved in vesicle trafficking and because PI 3-kinase appears to play an important role in insulin-stimulated glucose transport (6), we questioned whether Rho may also be required for the latter process.
To study Rho/ARF translocation in intact cells, after incubation, the cells were chilled and sonicated in Buffer I, which contained 250 mM sucrose, 20 mM Tris-HCl (pH 7.5), 1.2 mM EGTA, 20 g/ml aprotinin, 20 g/ml leupeptin, 1 mM phenylmethylsulfonyl fluoride, and 20 mM ␤-mercaptoethanol. Homogenates were centrifuged first at 500 ϫ g for 10 min to remove nuclei, debris, and the fat cake and then for 30 min at 100,000 ϫ g to obtain membrane and cytosol fractions. Membranes were suspended in Buffer I supplemented with 5 mM EGTA, 2 mM EDTA, and 1% Triton X-100, and insoluble (cytoskeleton) substances were removed by centrifugation. Plasma membranes, microsomes, and mixed nuclear/mitochondrial fractions were obtained as described (1). Routinely, 60 g of cytosolic protein and 80 g of membrane protein were used for immunoblotting (note that cytosolic protein is three or four times more abundant than membrane protein).
To study Rho/ARF translocation in vitro, post-nuclear homogenates were prepared in Buffer I containing 1 mM EDTA and incubated first for 20 min at room temperature to release GDP and then for 20 min at 37°C after adding 10 mM MgCl 2 with or without GTP␥S (Sigma) or PI 4,5-(PO 4 ) 2 (PIP 2 ; Fluka). Subcellular fractions were then obtained as described above. Subcellular fractions were resolved by SDS-polyacrylamide gel electrophoresis, transferred to nitrocellulose membranes, and immunoblotted, using chemiluminescence (ECL, Amersham Corp.) as described (1). Immunoblots were quantified with a Bio-Rad Molecular Analyst Chemiluminescence/ 32 P Imaging System. All results were expressed relative to controls (set at 100%) that were developed simultaneously on the same blots.
Mouse monoclonal antibodies raised against synthetic peptide corresponding to amino acids 120 -150 in human RhoA were purchased from Santa Cruz Biotechnology, Inc.; these antibodies do not cross-react with RhoC, RhoG, Rac1, Rac2, or Cdc42Hs. Mouse monoclonal antibodies raised against human ARF1 proteins (ID9) were kindly provided by Dr. Richard Kahn. Antibodies for Cdc42HS, Rac1, and Rac2 were obtained from Santa Cruz. Antibodies for PI 3-kinase were from Upstate Biotechnology, Inc.
In some experiments, 30 -40 ml of adipocytes were incubated in batches for 2 h in 2-3 volumes of low phosphate (0.12 mM NaH 2 PO 4 ) Krebs Ringer buffer supplemented with 10 mM HEPES, 2.5 mM glucose, 1% BSA, and 10 mCi of 32 PO 4 (DuPont NEN). Aliquots were then * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked treated with insulin, and Rho was quantitatively immunoprecipitated from total cell lysates in buffer containing 20 mM Tris-HCl (pH 7.4), 250 mM sucrose, 150 mM NaCl, 2 mM EGTA, 10 mM MgCl 2 , 1 mM Na 4 P 2 O 7 , 1 mM NaF, 1 mM Na 3 VO 4 , 1% Triton X-100, 0.5% Nonidet, 1 mM phenylmethylsulfonyl fluoride, 20 g/ml leupeptin, and 20 g/ml aprotinin. Precipitates were collected on protein AG-Sepharose beads, washed with lysis buffer and phosphate-buffered saline, and then heated for 20 min at 68°C in buffer containing 5 mM EDTA, 2 mM dithiothreitol, 0.2% SDS, 0.5 mM GTP, and 0.5 mM GDP. GDP and GTP were separated on polyethyleneimine-cellulose plates (Merck) developed with 1 M KH 2 PO 4 (pH 3.4). Plates were scanned and quantified in the Bio-Rad 32 P Imaging System. PLD activation was measured both in situ in intact adipocytes and in vitro in post-nuclear homogenates of adipocytes after overnight labeling with[ 3 H]oleic acid as described (1). For in situ assays, cells were electroporated as described (16) in Dulbecco's modified Eagle's medium buffer with or without 1 g/ml Clostridium botulinum C3 transferase (List), a selective inhibitor of Rho (17). After overnight labeling (with or without C3 transferase), cells were equilibrated in KRP buffer containing 1% BSA and 2% ethanol and treated for 0 -20 min with 10 nM insulin (see Ref. 1). For in vitro assays, labeled cells were sonicated in Buffer I containing 25 mM HEPES (pH 7.4), 1 mM EDTA, 100 mM KCl, and 3 mM NaCl. After equilibration of homogenates for 20 min at room temperature to release GDP, 1 M CaCl 2 , 5 mM MgCl 2 , 2.5% ethanol, and GTP␥S were added, and incubation at 37°C was continued for 30 min. [ 32 H]Phosphatidylethanol was isolated by TLC as described (1).
Glucose transport was assessed by measurement of [ 3 H]2-deoxyglucose (2-DOG) uptake during a 1-min period, following treatment of adipocytes with vehicle (controls) or 10 nM insulin for 30 min as described (1). Where indicated, cells were electroporated in the absence or the presence of C3 transferase and incubated for designated times prior to assessment of basal and insulin-stimulated 2-DOG uptake.

Rho Translocation in Intact Cells-26-kDa
Rho was largely localized (Ͼ80%) in the cytosolic fraction in control cells. Upon the addition of 300 nM insulin (required to fully activate PI 3-kinase; see Ref. 18), there was a rapid 2-3-fold increase in membrane-associated Rho and a 25% decrease in cytosolic Rho ( Fig. 1). At an insulin concentration of 10 nM, which only partially activates PI 3-kinase (18) but fully activates glucose transport, membrane Rho levels increased submaximally by approximately 50% at 0.5, 1, 5, 10, and 20 min of insulin treatment (not shown). These findings suggested that insulin-induced activation of both PI 3-kinase and Rho may occur at similar levels of post-receptor signaling.
Rho translocated specifically to the plasma membrane in response to insulin treatment (Fig. 1). Only small amounts of Rho were detectable in microsomes, and insulin had no effect on microsomal Rho levels ( Fig. 1). We did not detect Rho in nuclear/mitochondrial or cytoskeletal preparations (not shown).
ARF Translocation in Intact Cells-21-kDa ARF was localized almost exclusively (Ͼ95%) in the cytosolic fraction in control adipocytes. Like Rho, ARF rapidly translocated to the plasma membrane, but not to microsomes, in response to 300 nM insulin treatment (Fig. 1); cytosolic ARF levels did not decrease significantly, probably reflecting a smaller fraction undergoing translocation (Fig. 1). Increases in membrane ARF levels were also observed at 1-20 min of treatment with 10 nM insulin, but as in the case of Rho, these increases were approximately 50% less than those seen with 300 nM insulin (not shown).
Effects of Wortmannin and LY294002 on Rho Translocation-Because wortmannin inhibits insulin effects on PLD (1), we questioned whether Rho translocation was also sensitive to PI 3-kinase inhibitors. As shown in Fig. 2, both wortmannin and LY294002 inhibited insulin-induced increases in Rho translocation but did not alter the subcellular distribution of Rho in control cells.
[ 32 P]GTP Loading of Rho-Upon the addition of 300 nM insulin to 32 P-labeled adipocytes, [ 32 P]GTP loading of immunoprecipitable Rho increased 3-fold at 1 and 10 min post-treatment (Fig. 3). Surprisingly, in contrast to inhibitory effects on Rho translocation, wortmannin did not inhibit GTP loading of Rho (Fig. 3).
Translocation of Rho and ARF in Vitro-We examined the possibility that GTP loading and/or polyphosphoinositides, e.g. as resulting from PI 3-kinase activation, may be important for Rho/ARF translocation. Accordingly, both Rho and ARF (but not Cdc42Hs or PI 3-kinase) translocated from the cytosol to total, plasma and microsomal membrane fractions when GTP␥S was added to post-nuclear homogenates; PIP 2 , on the other hand, stimulated Rho, but not ARF, translocation to plasma and microsomal membranes ( Fig. 4 and Table I). The translocation of Rho and/or ARF to microsomal as well as plasma membranes in response to GTP␥S and/or PIP 2 suggested that insulin treatment in intact cells must also activate a process that targets Rho and ARF to the plasma membrane.
Effects of C3 Transferase on 2-DOG Uptake-Because of the apparent requirement for PI 3-kinase in the activation of Rho translocation and glucose transport by insulin, we questioned whether Rho may be required for insulin stimulation of glu- cose. As shown in Fig. 5 (A and C), when cells were electroporated in the presence of 0.5-5 g/ml C3 transferase and then incubated for 20 h, there was a large decrease in insulinstimulated 2-DOG uptake. Uptake of 2-DOG in control cells also decreased mildly in response to C3 transferase treatment, and this may reflect the fact that electroporation and simple overnight incubation themselves cause small increases in 2-DOG uptake (relative to cells assayed directly; see Ref. 16), and control cells in these conditions may not be in a fully basal state. It may be noted that: (a) C3 transferase in concentrations up to 5 g/ml was ineffective in the absence of electroporation (Fig. 5C); (b) it was necessary to incubate C3 transferasetreated cells more than 3 h (i.e. overnight) to observe inhibitory effects on 2-DOG uptake (Fig. 5B); and (c) electroporation/ overnight treatment with C3 resulted in a marked diminution in immunoreactive Rho (Fig. 5A, inset), similar to that reported by others (19). It may also be noted that C3 transferase-treated adipocytes were morphologically (as per light microscopy) and metabolically intact, as evidenced by incorporation of labeled glucose and fatty acids into all major classes of lipids and normal activation of IRS-1-dependent PI 3-kinase by insulin (not shown).
PLD Activation-In conjunction with Rho and ARF translocation, GTP␥S stimulated PLD activity in post-nuclear homogenates (Fig. 4). In intact adipocytes, C3 transferase pretreatment largely inhibited insulin-induced activation of PLD (Fig. 4). DISCUSSION The present findings suggested that Rho and ARF may participate in the activation of PC-PLD by insulin in rat adipocytes. Both Rho and ARF translocated to plasma membranes sufficiently rapidly to contribute to the rapid activation of plasma membrane PC-PLD by insulin (see Ref. 1), and both G-proteins have been reported to activate PC-PLD (10 -15). In addition, GTP␥S stimulated Rho and ARF translocation to the plasma membrane, as well as PLD activity, in adipocyte homogenates, and insulin increased GTP loading of Rho in intact cells. Moreover, insulin effects on PLD were inhibited by C3 transferase, which markedly depleted Rho.
The present findings also suggested that PI 3-kinase activation, perhaps via polyphosphoinositides, may contribute to Rho translocation. Accordingly, both wortmannin and LY294002 inhibited the translocation of Rho in intact adipocytes, and PIP 2 stimulated Rho translocation in vitro. In addition, polyphosphoinositides may be required for PC-PLD activation (10 -13), and wortmannin inhibits PC-PLD activation in both insulin-treated rat adipocytes (1) and f-Met-Leu-Phe-stimulated human neutrophils (20). It is therefore possible that insulin-induced activation of PI 3-kinase leads to production of polyphosphoinositides in the plasma membrane, followed by translocation of Rho, GTP loading, and PLD activation. Along these lines, PIP 2 increases GDP dissociation from ARF (11) and Cdc42Hs and Rho (see Ref. 22); however, this dissociation does not necessarily result in GTP loading (22), and, as shown presently, GTP loading of Rho appeared to be due to a factor distinct from PI 3-kinase.
Recently, there has been keen interest in mechanisms whereby small G-proteins activate PC-PLD on the one hand and, on the other hand, regulate vesicle formation and traffick-

FIG. 4. Effects of C3 transferase on insulin-stimulated PLD activation in intact adipocytes (D) and effects of GTP␥S and/or PIP 2 on translocation of Rho and ARF from cytosol to various membrane fractions (A and B) and PLD activity (C) in adipocyte homogenates.
In A and C, designated amounts of GTP␥S, and in B, 20 M GTP␥S or 1 M PIP 2 were added directly to post-nuclear homogenates in vitro, and assays for translocation (of Rho, ARF, Cdc42Hs, and PI 3-kinase) and PLD were conducted as described under "Experimental Procedures" (there were approximately 30,000 cpm in total cellular phospholipids in each PLD assay). In D, adipocytes were electroporated in the presence or the absence of 1 g/ml C3 transferase, cultured overnight in the presence of [ 3 H] oleic acid (there were approximately 60,000 cpm in total cellular phospholipids/assay in both control and C3-treated cells), and then treated with or without 10 nM insulin for indicated times, as described under "Experimental Procedures." In all cases, similar results were observed in repeat experiments. C, control; G, GTP␥S; P, PIP 2 ; Cyto, cytosol; Mem, total membranes; PM, plasma membrane; MS, microsomes; IB, immunoblot; ␣, antibodies or antiserum.
ing and in a variety of cytoskeletal events. Of Rho family members (Rho, Rac, and Cdc42Hs subtypes), Rho A is important in PC-PLD activation (10 -14) and formation of actin stress fibers and focal adhesions (23). Rho has also been suggested to operate upstream of PI 3-kinase in human platelets (24,25), although in insulin-stimulated adipocytes, PI 3-kinase appears to be primarily activated by IRS-1 or other phosphotyrosine-containing proteins (6). Rac1 is important in membrane ruffling (23) and seems to operate downstream of PI 3-kinase in insulin-induced membrane ruffling (26). Rac1 and Cdc42Hs (but not Rho A) activate a 62-65-kDa protein kinase that regulates stress-activated protein kinase/c-Jun NH 2 -terminal kinase (27)(28)(29)(30). GTP␥S-containing forms of Cdc42Hs and Rac1 also bind to and activate PI 3-kinase, and although this suggested that PI 3-kinase may be a downstream effector, it was surmised from other evidence that PI 3-kinase operates upstream of Cdc42Hs and Rac1 (31). Of further interest, PI 3-kinase and PC-PLD may control vesicle formation through the generation of membrane curvature-perturbing, acidic phospholipids, i.e. PI 3,4,5-(PO 4 ) 3 , PI 3,4-(PO 4 ) 2 , and PI 3-PO 4 , via PI 3-kinase action and phosphatidic acid via PLD action (32). Also, conventional (␣, ␤, and ␥) and novel (␦, ⑀, , and ) PKCs may be activated by diacylglycerol derived through PC-PLD action, and these PKCs, along with atypical PKCs ( and ) and protein kinase C-related kinase (PRK1 or PKN; see below), may be activated by D3-PO 4 polyphosphoinositides (33, 34) derived from PI 3-kinase action. Although not a member of the Rho family, ARF, like Rho, activates PC-PLD (10 -14), and, as suggested by the finding that polyphosphoinositides stimulate GTP/GDP exchange in ARF (11), PI 3-kinase may function upstream of ARF, as well as Rho. Obviously, both Rho and ARF may operate through or in conjunction with a variety of lipid and protein kinases and other signaling factors in regulating vesicle trafficking and cytoskeletal events.
Similar to our present observation of Rho and ARF translocation, f-Met-Leu-Phe stimulates ARF and Rho translocation in HL-60 cells (35) and guanosine nucleotide exchange in Rho in lymphocytes (36). Thus, ARF and Rho may function in parallel or in tandem and co-ordinately activate PC-PLD and other processes in response to agonist treatment in various cell types.
In addition to PLD activation, our observation that insulininduced Rho translocation is sensitive to polyphosphoinositides and PI 3-kinase inhibitors is of interest, because: (a) GTP-Rho binds to an activation site in the NH 2 -terminal regulatory domain of a 120-kDa protein kinase, variously called PKN (21,37) or PRK1 (34), and (b) PRK1 (34), like various PKCs (33,34), is activated by polyphosphoinositides. Thus, insulin-induced increases in polyphosphoinositides (7)(8)(9), which occur through the activation of PI 3-kinase (9), may activate PKN (PRK1), both directly via polyphosphoinositides and indirectly through Rho. The co-activation of Rho and PKN (PRK1) by polyphosphoinositides may facilitate their co-localization and may also co-ordinate the activation of PKN (PRK1) with other PKCs that are activated by PI 3-kinase through its lipid products or PC-PLD. With respect to PKC, it should be noted that we have not observed significant activation of PC-PLD or translocation of Rho during phorbol ester treatment in rat adipocytes.
Finally, it was of interest to find that insulin-stimulated glucose transport was inhibited in cells depleted of Rho by C3 transferase treatment. This apparent requirement for Rho, coupled with the fact that Rho is translocated and activated by insulin, suggests that Rho may have a role in insulin stimulation of glucose transport. Clearly, more studies are needed to test this possibility and further define the role of Rho.
In summary, insulin provoked rapid increases in Rho and ARF translocation to the plasma membrane and GTP loading of Rho in rat adipocytes. In addition, wortmannin and LY294002 inhibited insulin effects on Rho translocation in intact adipocytes, but wortmannin did not inhibit GTP loading of Rho. Of further note, PIP 2 and GTP␥S stimulated Rho translocation in adipocyte homogenates, and C3 transferase inhibited PLD activation in intact adipocytes. Collectively, these findings suggest that insulin translocates Rho by a PI 3-kinasedependent mechanism but stimulates GTP loading of Rho by a PI 3-kinase-independent mechanism, and both Rho and ARF may participate in the activation of PLD. Further studies will be required to define: (a) the precise mechanisms for activation and translocation of Rho and ARF and (b) the interrelated roles of PI 3-kinase and these small G-proteins in the activation of PC-PLD, various lipid-regulated protein kinases, vesicle trafficking, cytoskeletal events, and other cellular processes.  5. Effects of C3 transferase on insulin stimulation of 2-DOG uptake in rat adipocytes. In A, all cells were electroporated in the absence or the presence of the indicated concentration of C3 transferase, then incubated for 20 h (overnight) in Dulbecco's modified Eagle's medium, and finally suspended in glucose-free KRP medium to measure 2-DOG uptake in the absence (CON) or the presence of 10 nM insulin (INS). As shown in the inset, this C3 transferase treatment resulted in complete loss of immunoreactive Rho. In B, all cells were electroporated in the presence or the absence of 5 g/ml C3 transferase and then incubated for 3 h prior to measurement of 2-DOG uptake in the presence or the absence of 10 nM insulin. In C, C3 transferase was used with or without electroprotein (Ep), and cells were incubated for 20 h prior to measurement of 2-DOG uptake in the presence or the absence of 10 M insulin.