Protein Kinase C-ζ and Phosphoinositide-dependent Protein Kinase-1 Are Required for Insulin-induced Activation of ERK in Rat Adipocytes*

The mechanisms used by insulin to activate the multifunctional intracellular effectors, extracellular signal-regulated kinases 1 and 2 (ERK1/2), are only partly understood and appear to vary in different cell types. Presently, in rat adipocytes, we found that insulin-induced activation of ERK was blocked (a) by chemical inhibitors of both phosphatidylinositol 3-kinase (PI3K) and protein kinase C (PKC)-ζ, and, moreover, (b) by transient expression of both dominant-negative Δp85 PI3K subunit and kinase-inactive PKC-ζ. Further, insulin effects on ERK were inhibited by kinase-inactive 3-phosphoinositide-dependent protein kinase-1 (PDK-1), and by mutation of Thr-410 in the activation loop of PKC-ζ, which is the target of PDK-1 and is essential for PI3K/PDK-1-dependent activation of PKC-ζ. In addition to requirements for PI3K, PDK-1, and PKC-ζ, we found that a tyrosine kinase (presumably the insulin receptor), the SH2 domain of GRB2, SOS, RAS, RAF, and MEK1 were required for insulin effects on ERK in the rat adipocyte. Our findings therefore suggested that PDK-1 and PKC-ζ serve as a downstream effectors of PI3K, and act in conjunction with GRB2, SOS, RAS, and RAF, to activate MEK and ERK during insulin action in rat adipocytes.

Assay of Immunoprecipitable ERK-As described in previous studies (5, 15) of total mitogen-activated protein kinase activation, after incubation, adipocytes were sonicated in buffer containing 40 mM ␤-glycerophosphate (pH, 7.3), 0.5 mM dithiothreitol, 0.75 mM EGTA, 0.15 mM Na 3 VO 4 , 5 g/ml leupeptin, 5 g/ml aprotinin, 0.1 mM phenylmethylsulfonyl fluoride, and 5 g/ml trypsin inhibitor. The resulting homogenates were centrifuged at 700 ϫ g for 10 min to remove fat, cell debris, and nuclei. Post-nuclear supernatants were then supplemented with 0.154 M NaCl, 1% Triton X-100, and 0.5% Nonidet, and equal amounts of lysate protein in each experiment (varying from 200 to 500 g between experiments) were subjected to overnight immunoprecipitation at 4°C with mouse monoclonal anti-ERK2 antibodies (Santa Cruz Biotechnologies, Inc., Santa Cruz, CA), which, as shown below, immunoprecipitated ERK1, as well as ERK2, despite the fact that these antibodies reacted only with ERK2 in Western analyses. Precipitates were collected on Protein-AG-agarose beads, washed and incubated for 10 min at 30°C in 50 l of buffer containing 25 mM ␤-glycerophosphate (pH, 7.3), 0.5 mM dithiothreitol, 1.25 mM EGTA, 0.5 mM Na 3 VO 4 , 10 mM MgCl 2 , 1 mg/ml bovine serum albumin, 1 M okadaic acid, 0.1 mM [␥-32 P]ATP (NEN Life Science Products; approximate specific activity, 1,500,000 dpm/nmol), and 50 g of myelin basic protein (Sigma). After incubation, an aliquot of the reaction mixture was spotted on p81 filter paper, which was washed and counted for 32 P-radioactivity (5,15). Blank values were obtained by substituting a nonimmune antibody preparation instead of the anti-ERK2 antibodies, or by omitting myelin * This work was supported by funds from the Dept. of Veterans Affairs Merit Review Program and National Institutes of Health Research Grant 2R01DK38079-9A1. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
ʈ basic protein substrate (results were similar). Except for greater relative effects of insulin, results obtained with this ERK immune complex assay were similar in most aspects to those obtained in assays of total mitogen-activated protein kinase activity observed in crude cell extracts (5,15). Differences in absolute 32 P-incorporation values between individual experiments reflect variations in amounts of cell extracts immunoprecipitated and specific activity of [␥-32 P]ATP used, but relative effects of insulin and other agonists were comparable. In most cases, the actual data from individual experiments are depicted, but in all cases similar findings were observed in repeat experiments. As depicted in representative blots in Fig. 1, and as quantified in multiple samples in Table I, treatments with insulin and PI3K and PKCinhibitors, wortmannin and the myristoylated PKC-pseudosubstrate, did not have significant effects on the levels of ERK1 and ERK2, or their ratios, in these ERK2 immunoprecipitates, as determined by blotting with a rabbit polyclonal antiserum that recognizes both ERK1 and ERK2 in Western analyses. It may therefore be surmised that these ERK2 assays actually reflected activities of both ERK1 and ERK2, and our present finding of insulin effects on both ERK1 and ERK2 in these immunoprecipitates is in keeping with our previous findings, which showed that insulin activates both p44 ERK1 and p42 ERK2 in rat adipocytes, as determined following their electrophoretic resolution and assay in myelin basic protein-containing gels (15).
Transfection Studies-Rat adipocytes were transiently transfected using an electroporation method described previously (14,17). In brief, 0.4 ml of adipocyte was suspended in an equal volume of sterile Dulbecco's modified Eagle's medium containing 5% bovine serum albumin and 3.3 g of pCMV5 encoding MYC-tagged ERK2 or 1 g of pCEP4 encoding hemagglutinin (HA)-tagged ERK2 (both kindly supplied by Dr. Melanie Cobb), along with, as indicated in individual experiments: (a) 6.7 g of pCDNA3 encoding dominant-negative ⌬p85 PI3K subunit mutant (kindly supplied by Dr. Masato Kasuga; see Ref. 18); (b) 9 g of pRSV encoding N17 dominant-negative or V12 constitutive RAS mutants (both kindly supplied by Dr. Jane Reusch); (c) 6.7 g of pCEP4 encoding dominant-negative mutant forms of c-RAF-1 (a truncated form containing the N-terminal RAS-binding domain of c-RAF-1 that consequently inhibits RAS-dependent activation of endogenous ERK2) or MEKK1 (kinase-inactive, D1369A mutant) (both kindly supplied by Dr. Melanie Cobb); (d) 9 g of pCDNA3 encoding wild-type (WT), constitutive, or kinase-inactive (KI) PKC-(see Refs. 14 and 17); (e) 9 g of pCMVS encoding mutant T410A PKC-(kindly supplied by Dr. Alex Toker, see Ref. 19); (f) 9 g of pCDNA3 encoding WT or KI (K110N mutant) PDK-1 (both kindly supplied by Dr. Alex Toker; see Ref. 19); (g) 6.7 g of pSR␣ encoding WT or a dominant-negative SOS that interferes with GTP/GDP exchange in RAS (kindly supplied by Dr. Masato Kasuga, see Ref. 20); or (h) the vector alone. The amount of plasmid DNA used for transfection was kept constant in all samples by varying the amount of insert-free vector. After electroporation, cells were incubated overnight to allow time for expression, and then washed and suspended in glucose-free KRP medium and incubated for 10 min with or without 10 nM insulin. After incubation, cells were sonicated and MYC-or HA-tagged ERK2 was immunoprecipitated with rabbit polyclonal anti-MYC antiserum (Upstate Biotechnologies Inc., Lake Placid, NY) or mouse monoclonal anti-HA antibodies (Covance, Richmond, CA), respectively, and assayed for ERK-dependent myelin basic protein phosphorylation, as described above. As shown in the immunoblots depicted in Fig. 1, and, as may be surmised from observing comparable levels of basal enzyme activity of epitope-tagged ERK2 in various cotransfection groups (see Figs. [3][4][5], the transfection of dominant-negative forms of SOS, RAS, RAF, MEKK1, ⌬p85 PI3K, and various forms of PKC-and PDK-1 had relatively little or no significant effect on the levels of immunoprecipitable epitope-tagged ERK2. Also note that only ERK2 was recovered in immunoprecipitates obtained with anti-HA and anti-MYC antibodies (Fig. 1).

RESULTS AND DISCUSSION
Initially, we used inhibitors to identify factors required for insulin-induced activation of immunoprecipitable ERK2 in rat adipocytes. As seen in Fig. 2, PI3K inhibitors, wortmannin (100 nM) and LY294002 (100 M) (i.e. in concentrations required to largely inhibit insulin-stimulated glucose transport in the rat adipocyte), and the MEK1 inhibitor, PD098059 (10 M, which did not inhibit insulin-stimulated glucose transport), inhibited insulin-stimulated increases in immunoprecipitable ERK. Of particular interest, the cell-permeable myristoylated PKCpseudosubstrate inhibited insulin-induced increases in immunoprecipitable ERK activity over a concentration range comparable with that which is effective in inhibiting PKC- (14). In this regard, diacylglycerol-dependent PKCs are not required for insulin-induced activation of ERK in rat adipocytes (see Ref. 15; presently, we also confirmed that phorbol ester-induced PKC down-regulation inhibited the acute effects of phorbol esters but did not inhibit insulin-induced activation of immunoprecipitable ERK; data not shown), and it is therefore clear that the PKC-pseudosubstrate did not exert its effects through inhibition of diacylglycerol-dependent PKCs. These findings with inhibitors suggested that PI3K and PKC-(or PKC-, which is 72% homologous to PKC-and has an identical pseudosubstrate sequence), as well as MEK1, are required for insulin-induced activation of ERK in rat adipocytes.
Further evidence implicating PI3K in insulin-induced activation of ERK was obtained in experiments in which rat adipocytes were transiently co-transfected with plasmids encoding MYC-tagged ERK2 and a dominant-negative mutant form of the p85 subunit of PI3K, ⌬p85 (18). As seen in Fig. 3, insulininduced activation of MYC-ERK2 was inhibited by dominantnegative ⌬p85, a mutant form of the p85 subunit of PI3K that interacts with phosphotyrosine (pY) residues on activated forms of IRS family members but is unable to transmit activating signals to the p110 catalytic subunit of PI3K (18). These findings therefore suggested that the p85 regulatory subunit, as well as the p110 catalytic subunit (which is directly inhibited by wortmannin and LY294002), of PI3K was required for ERK2 activation, presumably reflecting a need for activation of the SH2 domain of the p85 subunit by pYXXM-containing proteins such as IRS (this assumes that ⌬p85 neither binds nor inhibits the p110 subunit of PI3K). I Levels of immunoprecipitable ERK1 and ERK2 following treatment of rat adipocytes with insulin, wortmannin, and/or myristoylated PKCpseudosubstrate Adipocytes were treated first without inhibitors or with 100 nM wortmannin for 15 min, or with 50 M myristoylated (MYR) PKCpseudosubstrate (PS) for 60 min, and second with or without 10 nM insulin for 10 min. ERK was immunoprecipitated with anti-ERK2 mouse monoclonal antibody, and precipitates were resolved by SDS-polyacrylamide gel electrophoresis and blotted with a rabbit polyclonal anti-ERK antiserum that recognizes both ERK1 and ERK2. After chemiluminescence development, p44 ERK1 and p42 ERK2 bands were quantitated with Bio-Rad molecular analyst chemiluminescence/ 32 P imaging system. Values are mean Ϯ S.E. of four determinations. Note that the mean control value was set at a relative value of 1.00, and four blots were compared simultaneously on the same Bio-Rad molecular analyst chemiluminescence imaging screen, thus allowing the direct comparison of 4 sets of immunoprecipitations, with each set containing each treatment. See Fig. 1  In addition to PI3K, we found that SOS, RAS, and RAF were required for insulin-induced activation of ERK2 in rat adipocytes. As seen in Fig. 3, transient transfection of dominantnegative mutant forms of SOS, RAS, and c-RAF-1 inhibited the activation of co-transfected HA-or MYC-tagged ERK2 by insulin; in addition, constitutively active RAS markedly stimulated HA-ERK2 activity. In contrast to the c-RAF-1 mutant, transfection of a dominant-negative kinase-inactive mutant form of MEKK1, which like RAF and PI3K (21) can interact with RAS (22), had no effect on basal or insulin-stimulated MYC-ERK2 (Fig. 3). Also, in contrast to inhibitory effects of dominantnegative RAS on insulin-induced activation of HA-ERK2, this RAS mutant did not inhibit the acute activating effects of phorbol esters on HA-ERK2 (data not shown), which may in some cell types occur independently of RAS (23). Thus, the inhibitory effects of both dominant-negative RAS and c-RAF-1 on insulin-induced activation of HA-ERK2 appeared to be specific.
The above findings suggested that PI3K along with SOS, RAS, c-RAF-1, and MEK1 was required for insulin-induced activation of ERK2 in the rat adipocyte. Because PKC-(and/or ) is known to serve as an effector of PI3K during insulin action in rat adipocytes (14) and other cells (24 -27), and in view of the above-described inhibitor studies, we tested the possibility that PKC-may function downstream of PI3K during ERK activation by transiently co-transfecting rat adipocytes with plasmids encoding HA-ERK2 and various forms of PKC-. As seen in Fig.  4, whereas WT PKC-had no effect on HA-ERK2 activity, both a KI form of PKC-(K271N mutant) and an activation-resistant form of PKC-(T410A mutant that cannot be activated by PDK-1; see Refs. 19 and 28) markedly inhibited insulin-stimulated HA-ERK2 activity but had little effect on basal or phorbol ester-stimulated HA-ERK2 activity. Moreover, the inhibitory effect of KI-PKC-could be reversed (or prevented) by co-transfecting plasmid encoding WT-PKC-, which alone had no effect on basal or insulin-stimulated ERK2. These findings suggested that the point-mutation per se in KI-PKC-was responsible for its inhibitory effects on insulin-induced activation of ERK2 and further implied that the kinase activity of PKC-is specifically required, presumably to phosphorylate a presently undefined substrate that is required for subsequent ERK2 activation in rat adipocytes.
Whereas mutant forms of PKC-inhibited insulin-induced activation of HA-ERK2, constitutive PKC-provoked insulinlike increases in HA-ERK2 activity, even in the absence of In panels B and C, adipocytes were co-transfected with plasmids encoding WT, constitutive (Constit), dominant-negative (DN), KI, or other mutant (e.g. activation-resistant T410A PKC-) signaling proteins, along with plasmid encoding HA-or MYC-tagged ERK2. After incubation, immunoprecipitates were prepared as described under "Experimental Procedures," and all were blotted with a rabbit polyclonal antiserum that recognizes both ERK1 and ERK2. Note that both ERK1 and ERK2 were recovered in precipitates brought down by mouse monoclonal anti-ERK2 antibodies (A), whereas only ERK2 was recovered in precipitates brought down with rabbit polyclonal anti-MYC antiserum (B) and mouse monoclonal anti-HA antibody (C). Shown here are representative immunoblots, repeated at least 4 times with similar results. Levels of ERK1 and ERK2, as determined by quantitation of chemiluminescence in a Bio-Rad molecular analyst chemiluminescence/ 32 P-imaging system, are given in Table I. FIG . 2. Effects of wortmannin, LY294002, PD098059, and the cell-permeable myristoylated PKC-pseudosubstrate on insulin-induced activation of immunoprecipitable ERK. Adipocytes were equilibrated with 100 nM wortmannin, 100 M LY294002, or 10 M PD098059 for 15 min in panels A, B, and C and for 60 min with myristoylated PKC-pseudosubstrate (see Ref. 14) in panel D, in glucose-free KRP medium containing 1% bovine serum albumin, and then treated with or without 10 nM insulin for 10 min (this time was optimal for observing changes in ERK2). After incubation, adipocytes were sonicated, and cell lysates were assayed for immunoprecipitable ERK activity as described under "Experimental Procedures." Values are mean Ϯ S.E. of (n) determinations. insulin (Fig. 4). Further stimulatory effects of insulin on ERK2 activity in cells expressing constitutive PKC- (Fig. 4) were also observed, and this may reflect the activation of endogenous PKC-, or the fact that insulin can provoke further increases in activity of "constitutive" PKC-. 2 Because PDK-1, in conjunction with PI3K-dependent increases in PI-3,4,5-(PO 4 ) 3 , has been reported to transmit activating signals from PI3K to PKC-(19, 28) (note that we have confirmed that PDK-1 and its target in PKC-, threonine-410, are required for PKC-activation by insulin in rat adipocytes; see Ref. 29), we examined the role of PDK-1 in insulin-induced activation of ERK2 in transiently transfected rat adipocytes. As seen in Fig. 5, WT-PDK-1 enhanced basal HA-ERK2 activity, and KI-PDK-1 inhibited insulin-stimulated HA-ERK2 activity. Further, the inhibitory effect of KI-PDK-1 on insulin-stimulated ERK2 activation was reversed by co-transfection of WT-PDK-1 (Fig. 5), indicating that its kinase activity (presumably to phosphorylate Thr-410 in PKC-), like that of PKC-, is specifically required for insulin-induced activation of ERK2.
Our findings suggested that PI3K, PDK-1, and PKC-, along with SOS, RAS, c-RAF-1, and MEK1 were required for ERK2 activation during insulin stimulation of rat adipocytes. In this regard, RAS is known to bind to the p110 subunit of PI3K (10,11), and we presently found that both the p110 and p85 subunits of PI3K were recovered in RAS immunoprecipitates prepared from lysates of rat adipocytes (Fig. 6). However, we did not observe any effects of insulin on (a) p85 or p110 PI3K subunit levels in RAS immunoprecipitates or (b) PI3K enzyme activity recovered in RAS immunoprecipitates (Fig. 6). Moreover, as discussed above, the inhibitory effects of dominantnegative ⌬p85 on insulin-induced activation of ERK2 suggested that the activation of PI3K that is relevant to ERK activation requires input from a factor that is capable of activating the p85 subunit of PI3K, e.g. an IRS family member. Collectively, these findings suggested that RAS may serve in the localization but not in the activation of PI3K. Rat adipocytes were transiently co-transfected with plasmids encoding HA-or MYC-tagged ERK2 and indicated proteins, and after overnight incubation to allow time for expression, cells were incubated in glucose-free KRP medium for 10 min with or without 10 nM insulin as described under "Experimental Procedures." After incubation, epitope-tagged ERK2 was immunoprecipitated and assayed as described under "Experimental Procedures." Values are mean Ϯ S.E. of (n) determinations. As shown in Fig.  1, co-transfection of mutant signaling proteins had relatively small or no effect on the level of immunoprecipitable epitope-tagged ERK2 (also note similar levels of basal ERK2 enzyme activity in precipitates obtained from various co-transfected cells in each experimental group).

FIG. 4. Effects of WT, KI, constitutively active (Constit), and activation-resistant (T410A) forms of PKC-on insulin-induced activation of epitope-tagged ERK2.
Adipocytes were transiently co-transfected with plasmids encoding HA-ERK2 and indicated proteins, and after overnight incubation to allow time for expression, the cells were incubated in glucose-free KRP medium for 10 min with or without 10 nM insulin as described in Methods. After incubation, epitope-tagged ERK2 was immunoprecipitated and assayed as described under "Experimental Procedures." Values are mean Ϯ S.E. of (n) determinations. As shown in the inset (immunoblot developed with polyclonal anti-C-terminal PKC-/ antiserum obtained from Santa Cruz Biotechnologies, Inc., Santa Cruz, CA), the T410A mutant form of PKC-was expressed to about the same extent as KI-PKC-(see Ref. 4 for expression data for other forms of PKC-in rat adipocytes). As shown in Fig. 1, co-transfection of mutant signaling proteins had relatively small or no effect on the level of immunoprecipitable epitopetagged ERK2 (also note similar levels of basal ERK2 enzyme activity in precipitates obtained from various co-transfected cells in each experimental group).
Finally, we found that inhibitors of tyrosine kinase and the GRB2 SH2 domain, viz., genistein and a N␣-oxalyl-tripeptide-pY mimetic (see above and Ref. 16), respectively, also inhibited insulin-induced activation of immunoprecipitable ERK in the rat adipocyte (Fig. 7). These findings therefore suggested that a tyrosine kinase-dependent substrate, i.e. IRS and/or SHC, along with GRB2 and SOS, operate upstream of RAS in insulin-induced activation of ERK2 in rat adipocytes.
To summarize, our findings suggested that factors in two signaling pathways that are frequently considered to be functionally separate during insulin action, viz., the GRB2/SOS/ RAS/RAF pathway and the PI3K/PDK-1/PKC-pathway, are jointly required for insulin-induced activation of ERK in rat adipocytes. Although there are still gaps in our understanding of how these pathways are activated and interact with each other, it seems likely that one or more activated forms of IRS family members acts upon a specific pool of PI3K that operates in conjunction with GRB2/SOS and RAS. This PI3K apparently activates a subset of PDK-1 and PKC-, which in turn may contribute, along with RAS, to the activation of c-RAF-1. This postulation is in keeping with other findings indicating that RAS-dependent activation of RAF apparently requires the phosphorylation of RAF by serine/threonine kinases and subsequent recruitment of a 14⅐3⅐3 protein (30 -32). Alternatively, despite our negative finding, RAS may activate PI3K (see Refs. 10 and 11), or IRS-stimulated PI3K may activate RAS either directly (12) or indirectly via GRB2/SOS, as suggested to occur during G␤␥ signaling through PI3K and RAS (33); however, in the latter case, PI3K is postulated to activate a nonreceptor tyrosine kinase before activating GRB2/SOS (33), and in the case of insulin, this would imply an element of redundancy, as it would mean that tyrosine kinase activation is required both before and after PI3K activation. With any of these alternatives, the ability of RAS to bind both PI3K (10, 11) and RAF (21), coupled with localizing and activating effects of PI-3,4,5-(PO 4 ) 3 , may facilitate the assembly of a functional complex that contains or subsequently recruits RAS, PI3K, PDK-1, PKCand RAF. Further studies are needed to more precisely define (a) how PI3K and RAS operate with respect to each other and (b) the role of PKCin insulin-induced activation of ERK.
FIG. 6. PI3K association with RAS in rat adipocytes. Cells were incubated with or without 10 nM insulin for indicated times. After incubation, equal amounts (1 mg of protein) of cell lysates were subjected to immunoprecipitation with rat monoclonal anti-RAS antibodies or rabbit polyclonal anti-p110/PI3K antiserum (Santa Cruz Biotechnologies, Inc.) and precipitates were blotted with anti-RAS antibodies (same source), anti-p110/PI3K antiserum (same source), and rabbit polyclonal anti-p85/PI3K antiserum (Upstate Biotechnologies, Inc.), as indicated, or assayed for PI 3-kinase activity as described (26) .   FIG. 7. Effects of inhibitors of tyrosine kinase and the SH2 domain of GRB2 on insulin-induced activation of ERK in rat adipocytes. As in Fig. 2, cells were equilibrated for 15 min with indicated concentrations of genistein and for 180 min with indicated concentrations of a N␣-oxalyl-tripeptide-pY mimetic, i.e. compound L-20d, which contains a (phosphonomethyl)phenylalanine residue, and effectively inhibits the GRB2 SH2 domain both in vitro and in intact cells (16). Cells were then treated without or with 10 nM insulin, as indicated, for 10 min. After incubation, ERK was immunoprecipitated and assayed as in Fig. 2. Values are mean Ϯ S.E. of four determinations.