Regulation of both glycogen synthase and PHAS-I by insulin in rat skeletal muscle involves mitogen-activated protein kinase-independent and rapamycin-sensitive pathways.

Incubating rat diaphragm muscles with insulin increased the glycogen synthase activity ratio (minus glucose 6-phosphate/plus glucose 6-phosphate) by approximately 2-fold. Insulin increased the activities of mitogen-activated protein (MAP) kinase and the Mr = 90,000 isoform of ribosomal protein S6 kinase (Rsk) by approximately 1.5-2.0-fold. Epidermal growth factor (EGF) was more effective than insulin in increasing MAP kinase and Rsk activity, but in contrast to insulin, EGF did not affect glycogen synthase activity. The activation of both MAP kinase and Rsk by insulin was abolished by incubating muscles with the MAP kinase kinase (MEK) inhibitor, PD 098059; however, the MEK inhibitor did not significantly reduce the effect of insulin on activating glycogen synthase. Incubating muscles with concentrations of rapamycin that inhibited activation of p70S6K abolished the activation of glycogen synthase. Insulin also increased the phosphorylation of PHAS-I (phosphorylated heat- and acid-stable protein) and promoted the dissociation of the PHAS-I*eIF-4E complex. Increasing MAP kinase activity with EGF did not mimic the effect of insulin on PHAS-I phosphorylation, and the effect of insulin on increasing MAP kinase could be abolished with the MEK inhibitor without decreasing the effect of insulin on PHAS-I. The effects of insulin on PHAS-I were attenuated by rapamycin. Thus, activation of the MAP kinase/Rsk signaling pathway appears to be neither necessary nor sufficient for insulin action on glycogen synthase and PHAS-I in rat skeletal muscle. The results indicate that the effects of insulin on increasing the synthesis of glycogen and protein in skeletal muscle, two of the most important actions of the hormone, involve a rapamycin-sensitive mechanism that may include elements of the p70S6K signaling pathway.

Glycogen synthesis in skeletal muscle has a key role in the control of blood glucose levels by insulin. The large majority of postprandial glucose uptake occurs in skeletal muscle (1,2), and most of the glucose that enters muscle fibers in response to insulin is converted to glycogen (3). This hormonal effect in-volves activation of glycogen synthase, the enzyme that catalyzes the rate-limiting step in the conversion of intracellular glucose to glycogen (4,5). Insulin activates synthase by promoting dephosphorylation of sites in the COOH-and NH 2terminal regions of the enzyme (4,5). The pattern of dephosphorylation is consistent with the hypothesis that insulin activates PP1 G , the glycogen-bound form of type I protein phosphatase, as this phosphatase is able to dephosphorylate multiple sites in the synthase subunit (6). PP1 G is controlled by phosphorylation of sites in its regulatory subunit (6,7), designated R GL (8). Phosphorylation of site 1 increases phosphatase activity toward glycogen synthase. This site is readily phosphorylated in vitro by Rsk-2 (9 -11), a kinase that is phosphorylated and activated by MAP 1 kinase when cells or tissues are incubated with insulin (12,13). Injecting insulin into rabbits has been reported to increase phosphorylation of site 1 in R GL (9), and it is widely believed that the activation of glycogen synthase by insulin in skeletal muscle involves the sequential activation of MAP kinase, Rsk-2, and PP1 G .
Insulin also stimulates protein synthesis in many cells, but again skeletal muscle is of particular importance as this tissue is the largest reservoir of body protein (14). Muscle wasting is a hallmark of untreated diabetes mellitus in humans, and inducing diabetes in rats decreases by half the rate of protein synthesis in skeletal muscle (15). Experimental diabetes also causes dispersion of polysomes and accumulation of free ribosomal subunits (16,17). Within 2 h of treating diabetic rats with insulin, the polysome profile returns to the prediabetic state. These effects of insulin and diabetes are indicative of regulation of translation initiation, which is generally the ratelimiting phase of protein synthesis (18 -20). Initiation involves recognition of capped mRNA, melting of secondary structure in the 5Ј-nontranslated region of the mRNA, and binding of the 40 S ribosomal subunit. Initiation is mediated by several factors, the least abundant of which is the mRNA cap-binding protein, eIF-4E. Several lines of evidence indicate that eIF-4E activity is limiting for initiation (18 -20).
Insulin increases eIF-4E activity in adipocytes by stimulating the phosphorylation of the translational regulator, PHAS-I (21)(22)(23)(24). PHAS-I cDNA was originally cloned from a rat skeletal muscle library, and PHAS-I mRNA was found in highest levels in skeletal muscle and fat (25). Nonphosphorylated PHAS-I binds tightly to eIF-4E and inhibits translation (24), probably by preventing the association of eIF-4E with eIF-4␥ (26). When PHAS-I is phosphorylated in response to insulin, the PHAS-I⅐eIF-4E complex dissociates (21)(22)(23)(24), allowing eIF-4E to participate in translation initiation. PHAS-I is an excellent substrate for MAP kinase in vitro, and the major site (Ser 64 ) phosphorylated by MAP kinase in vitro is phosphorylated in response to insulin in adipocytes (27). Moreover, essentially all of the insulin-stimulated PHAS-I kinase activity in adipocyte extracts is accounted for by the ERK-1 and ERK-2 isoforms of MAP kinase (23). Based on these results, MAP kinase was proposed to mediate the phosphorylation of PHAS-I by insulin in adipocytes. Thus far, all studies of the regulation of PHAS-I by insulin have been confined to adipocytes.
Recent findings indicate that activation of MAP kinase is neither necessary nor sufficient for the effects of insulin on glycogen synthase in fat cells (28 -32). EGF and other agents that are as effective as insulin in activating MAP kinase and Rsk did not activate glycogen synthase in either primary (29,31) or 3T3-L1 adipocytes (28,32). MAP kinase activation by insulin in 3T3-L1 adipocytes was blocked by a novel inhibitor of MEK, PD 098059, without inhibiting the effect of insulin on glycogen synthase (30). PHAS-I phosphorylation is increased by agonists that activate MAP kinase; however, the phosphorylation of PHAS-I by insulin was not attenuated by inhibition of MAP kinase activation with PD 098059 (22). In contrast, the effect of insulin on increasing PHAS-I phosphorylation and promoting the dissociation of the PHAS-I⅐eIF-4E complex was markedly inhibited by rapamycin (21,22). Rapamycin acts to inhibit transduction through a pathway that is distinct from the Ras-MAP kinase pathway and that leads to the activation of p70 S6K (33)(34)(35)(36). Rapamycin was without effect on the activation of glycogen synthase by insulin in rat adipocytes (29,31) but markedly inhibited the activation of synthase in 3T3-L1 adipocytes (37). Multiple pathways had been previously shown to exist for the activation of glycogen synthase by insulin in rat adipocytes (38). Different cell types might utilize different transduction pathways to activate synthase, and it has been suggested that the mechanism of activation of synthase in skeletal muscle differs from that in adipocytes.
In view of the importance of skeletal muscle in the regulation of glycogen and protein metabolism by insulin, it is important to determine which signaling pathways are utilized by insulin in this tissue. In the present experiments, we have investigated the roles of the MAP kinase and rapamycin-sensitive pathways in mediating the effects of insulin on glycogen synthase and PHAS-I in rat diaphragm.

EXPERIMENTAL PROCEDURES
Incubation of Muscle in Vitro-Male rats (60 -80 g, Sprague-Dawley, Sasco) were fed ad libitum before diaphragms with the surrounding ribs were removed as described by Goldberg et al. (39). Incubations were conducted essentially as described previously (40). Briefly, to remove endogenous hormones, muscles were incubated at 37°C in Dulbecco's modified Eagle's medium for 45 min. For treatments, muscles were transferred to Krebs-Henseleit buffer (118 mM NaCl, 4.7 mM KCl, 2.5 mM CaCl 2 , 1.2 mM potassium phosphate, 1.2 mM MgSO 4 , and 25 mM NaHCO 3 , pH 7.4) plus 5 mM glucose and incubated at 37°C with insulin (20 nM) or EGF (100 nM) for the desired time. When present, rapamycin or PD 098059 was added 30 min before insulin or EGF. All media were gassed continuously by bubbling with a 19:1 mixture of O 2 :CO 2 . To terminate the incubations, the muscles were frozen in liquid nitrogen. Diaphragms were chipped out of the rib cages, ground into powder in a porcelain mortar chilled with liquid nitrogen, and stored at Ϫ75°C.
For measurements of glycogen synthase, powdered muscle (100 mg) was homogenized using a tissue grinder (Teflon-glass) at 4°C in a solution (1 ml) composed of 100 mM potassium fluoride, 10 mM EDTA, 2 mM EGTA, 5 mM sodium potassium phosphate, 2 mM potassium phosphate, 50 mM Tris-HCl, pH 7.8, and protease inhibitors benzamidine (1 mM), leupeptin (10 g/ml), aprotinin (10 g/ml), and phenylmethylsulfonyl fluoride (0.1 mM). Homogenates were centrifuged at 10,000 ϫ g for 30 min, and supernatants were retained. When protein kinase activities were to be measured, powders were homogenized in buffer containing 10 mM MgCl 2 , 1 mM sodium orthovanadate, 1 mM dithiothreitol, 0.1 M microcystin, 1 mM EDTA, 5 mM EGTA, 10 mM potassium phosphate, 50 mM sodium ␤-glycerophosphate, pH 7.4, and the same protease inhibitors as used in the synthase homogenization buffer. After preparation of extracts by centrifugation (10,000 ϫ g for 30 min), the protein content of each sample was measured (41) and adjusted to a concentration of 1 mg/ml by adding homogenization buffer.
Measurements of Glycogen Synthase Activities-Synthase activities were measured by monitoring the incorporation of [ 14 C]glucose from UDP-[U-14 C]glucose into glycogen (42). Activity ratios represent the activity measured without glucose 6-phosphate divided by the activity measured in the presence of 10 mM glucose 6-phosphate. Total activity represents that measured in the presence of 10 mM glucose 6-phosphate and is expressed relative to extract protein (43).
Antibodies-ERK-1 antisera were generated by immunizing rabbits with a peptide (CQETARFQPGAPEAP) based on a sequence in the COOH-terminal region of ERK-1, and antibodies were affinity-purified before use as described previously (44). The p70 S6K and Rsk antisera were from the laboratory of Dr. Edwin Krebs (University of Washington) and were generated by immunizing rabbits with the peptides corresponding to sequences found in mouse Rsk (ESSILAQRRVRKLP-STTL) (45) or in p70 S6K (CQAFPMISKRPEHLRMNL). PHAS-I antiserum was generated by immunizing rabbits with purified [His 6 ]PHAS-I (27). Glycogen synthase antibodies were made by immunizing chickens with rabbit skeletal muscle glycogen synthase. Purified synthase protein (46) was administered subcutaneously (0.2 mg/injection) at monthly intervals as emulsions (complete adjuvant for the initial injection and incomplete adjuvant for boosts). After the second boost, eggs were collected, and IgY was purified from the yolks by using the dextran sulfate method described by Jensenius et al. (47). Glycogen synthase was coupled to cyanogen bromide-activated Sepharose 4B (2 mg of synthase/ml) as directed by the supplier (Pharmacia Biotech Inc.), and the synthase resin was used to affinity purify the antibodies as described previously (44).
MAP Kinase Activities-The activities of MAP kinase in extracts were measured by using [␥-32 P]ATP and MBP as substrates (48). The composition of the reaction mixture and the incubation procedures used were exactly as described previously (29). An immune complex assay was used to assess the activity of the ERK-1 isoform of MAP kinase. Except for the use of diaphragm extracts (20 g of protein), this assay was performed exactly as described previously for measuring ERK-1 activity in adipocyte extracts (44). Activation of ERK-1 and ERK-2 was also assessed by using the "in-gel" assay described by Wang and Erikson (49). In this assay, extract samples (25 g) were subjected to SDS-PAGE in gels containing MBP fixed within the matrix of the gel. Kinase activities were measured after renaturation as described previously (49). In some experiments, single muscle fibers were manually dissected from freeze-dried specimens of control and EGF-treated diaphragms and dissolved in SDS sample buffer (50). The "in-gel" assay was used to assess ERK-1 and ERK-2 activities in samples (0.2 g, dry weight) of individual fibers.
Ribosomal Protein S6 Kinase Activities-Protein A-agarose beads (Bio-Rad, 0.1 ml of serum per ml of packed beads) were incubated at 23°C for 60 min with nonimmune serum or antisera to either Rsk or p70 S6K . The beads were then washed five times with phosphate-buffered saline (PBS) (145 mM NaCl, 4 mM KCl, and 10 mM sodium P i , pH 7.4) and once with homogenization buffer. Samples of extract (100 l) were incubated with beads (10 l) for 60 min at 4°C with constant mixing and then washed twice (0.5 ml of homogenization buffer/wash) and suspended in 100 l of homogenization buffer.
The S6 peptide, KEAKEKRQEQIAKRRRLSSLRASTSKSGGSQK, is based on a phosphorylated region of ribosomal protein S6 and was used by Dent et al. (9) to assay the activity of the R GL kinase, ISPK, later identified as Rsk-2 (10,11). With S6 peptide as substrate, the reactions were terminated by spotting samples (15 l) onto phosphocellulose papers (1 ϫ 2 cm, Whatman P81) and immediately immersing the papers in 175 mM H 3 P0 4 . The amounts of 32 P incorporated into the peptide were determined after washing the papers as described previously (29). When [His 6 ]⌬R GL was used as substrate for Rsk, the reactions were terminated by adding SDS sample buffer. Samples were subjected to SDS-PAGE (52), and the relative amounts of 32 P incorporated into [His 6 ]⌬R GL were determined by using a phosphorimager (Molecular Dynamics).
To measure p70 S6K activities, immune complexes were incubated as described for measuring Rsk activities, except that 40 S ribosomes (2 mg/ml final concentration) were used as substrate. The ribosomes were purified from rat liver as described by Krieg et al. (53). The protein kinase reaction was terminated by adding SDS sample buffer, and samples were subjected to SDS-PAGE (52). The relative amounts of 32 P incorporated into ribosomal protein S6 were determined by phosphorimaging.
Preparation of [His 6 ]⌬R GL -The 1329-base pair NcoI-NcoI fragment was excised from G⅐pET-8c plasmid (8), blunt-ended, and then inserted into blunt-ended XhoI and Bpu1102 sites in pET-15b (Novagen). This construct encodes the first 443 amino acid residues of R GL plus 6 NH 2 -terminal His residues and 4 additional amino acids (Val-Ser-Asn-Asn) at the COOH terminus. The orientation and reading frame of the insert were determined to be correct by restriction mapping and nucleotide sequencing. The resulting plasmid, ⌬C-G⅐pET, was transformed into Escherichia coli [BL21(DE3)], and after induction with isopropyl-1-thio-␤-D-galactopyranoside, [His 6 ]⌬R GL was extracted and purified by chromatography using Ni 2ϩ -NTA-agarose (Qiagen). After SDS-PAGE and staining with Coomassie Blue, the recombinant protein appeared as a species of apparent M r ϭ 65,000. This is somewhat higher than the M r ϭ 52,300 expected of the protein encoded by the cDNA, but intact R GL also exhibits lower electrophoretic mobility than would be expected of its actual M r (8). The purified protein was identified as [His 6 ]⌬R GL by amino acid sequencing of the NH 2 -terminal residues and by immunoblotting with R GL antibodies (8).
Immunoblotting-Samples were subjected to SDS-PAGE (52) before proteins were electrophoretically transferred to nylon membranes (Immobilon, Millipore), which were then incubated in a solution of PBS containing 5% milk (Carnation) and 1% Triton X-100. Membranes were then incubated for 2 h with 1000-fold dilutions of Rsk, p70 S6K , or PHAS-I antisera or 0.1 g/ml affinity-purified glycogen synthase antibodies. After washing for six times with 1% Triton X-100 in PBS, the membranes were incubated with either protein A conjugated to alkaline phosphatase (5000-fold dilution) or with rabbit anti-chicken antibodies conjugated to horseradish peroxidase (3000-fold dilution). Membranes were then washed six times with 1% Triton X-100 in PBS before antibody binding was detected by using enhanced chemiluminesence kits containing reagents for the alkaline phosphatase reaction (Tropix) or horseradish peroxidase reaction (Amersham).
Purification of PHAS-I⅐eIF-4E Complexes-Muscle extracts (100 g of protein), prepared as described above for the extraction of glycogen synthase, were mixed with 10 l of m 7 GTP-Sepharose (Pharmacia) and incubated for 30 min at 21°C. The resin was washed three times (0.5 ml/wash) with kinase homogenization buffer. Proteins were eluted with SDS sample buffer and subjected to SDS-PAGE (52). PHAS-I that copurified with eIF-4E was determined by immunoblotting.
Other Materials-UDP-[U-14 C]glucose was prepared as described by Thomas et al. (42) by using [U-14 C]glucose from ICN. [␥-32 P]ATP was obtained from DuPont NEN. Porcine insulin (27 units/mg) and EGF were from Eli Lilly and United Biotechnology International, respectively. PD 098059 was provided by Parke-Davis. Microcystin and okadaic acid were from LC Labs, and rapamycin was purchased from Calbiochem.

Effects of Insulin and EGF on Glycogen Synthase and MAP
Kinase Activities-Incubating rat diaphragm muscles for increasing times with insulin resulted in a rapid activation of glycogen synthase (Fig. 1). The activity ratio in control muscles prior to incubation with the hormone was 0.25, and the ratio did not significantly change during the subsequent incubation period. After 30 min of incubation with insulin, the activity increased to approximately 0.44. Phosphorylation of certain sites in purified glycogen synthase has been shown to decrease the electrophoretic mobility of the protein in SDS-PAGE (54). The synthase subunit was readily detected by immunoblotting after extract samples were subjected to SDS-PAGE (Fig. 1,  inset). Neither insulin nor EGF changed the amount of synthase detected, consistent with the observation that total glycogen synthase activity was changed by neither agent (Table I).
Insulin promoted a shift in the electrophoretic mobility of the synthase to forms of higher electrophoretic mobility (Fig. 1,  inset), consistent with its well established action to promote dephosphorylation of the protein (4,5). In contrast to insulin, EGF was without effect on the synthase activity ratio or the electrophoretic mobility of the synthase subunit.
EGF was much more effective than insulin in activating MAP kinase. In Fig. 2, the ERK-1 and ERK-2 isoforms of MAP kinase in extracts of insulin-and EGF-treated muscles were resolved by SDS-PAGE before kinase activities were measured after renaturation by using an in-gel assay (49). Bands of activity corresponding to species of apparent M r ϭ 44,000 and 42,000 were detected (Fig. 3A, for example). These species had exactly the same electrophoretic mobilities as the ERK-1 and ERK-2 isoforms from adipocytes. 2 There was no indication of sequential activation (ERK-1 before ERK-2) of the isoforms as was reported to occur in hindlimb muscles when rats were injected with insulin (55). Insulin increased the activities of ERK-1 ( Fig. 2A) and ERK-2 (  comparison, EGF increased the activities of the ERK-1 and ERK-2 by approximately 3-and 5-fold, respectively. The finding that both isoforms of MAP kinase were markedly increased by EGF under conditions in which glycogen synthase activity was not changed indicates that MAP kinase activation is not sufficient for the activation of glycogen synthase. Because of the important implications of the findings, control experiments were performed to verify that EGF was more effective than insulin in activating MAP kinase. EGF produced a larger increase than insulin when total MAP kinase activity was measured in nonfractionated extracts 2 or when the activity of the ERK-1 isoform was measured in immune complex assays (Fig. 3D). Thus, the difference between the effectiveness of EGF and insulin was not dependent on the method used to assay MAP kinase. MAP kinase activity was found to be increased in single muscle fibers that had been manually dissected from control and EGF-treated muscles, 2 indicating that the effect of EGF was not due to effects on other cell types present in skeletal muscle.
Failure of the MEK Inhibitor, PD 098059, to Prevent Activation of Glycogen Synthase by Insulin-PD 098059 acts noncompetitively with respect to ATP to inhibit activation of MEK1 and MEK 2 in vitro (56). The MEK inhibitor appears to block transduction through the MAP kinase pathway without inhibiting other protein or lipid kinases (22,30,56,57). The inhibitor markedly decreased activation of MAP kinase by EGF in diaphragms, and it abolished the effects of insulin on MAP kinase when ERK-1 or ERK-2 activities were measured by using the in-gel assay (Fig. 3, A-C) or when ERK-1 activity was assessed after immunoprecipitation (Fig. 3D). In the same muscles, the MEK inhibitor was without significant effect on glycogen synthase activity in either the absence or presence of EGF or insulin (Table I).
Inhibiting MAP kinase activation with the MEK inhibitor was associated with marked inhibition of Rsk (Fig. 4). As with MAP kinase activity, Rsk activity measured in an immune complex assay using a large S6 peptide as substrate was in-creased much more by EGF than by insulin. Treating muscles with PD 098059 abolished activation of Rsk by insulin and decreased the activity observed in the presence of EGF by approximately 80%. Rsk immunoprecipitated from muscle extracts readily phosphorylated [His 6 ]⌬R GL , a truncated form of R GL consisting of 443 residues that include site 1 and the other NH 2 -terminal sites (Fig. 4B). Both insulin and EGF increased the activity of Rsk with respect to the recombinant protein.

Incubating muscles with the MEK inhibitor abolished the effects of both insulin and EGF.
MAP Kinase-independent Control of PHAS-I by Insulin-PHAS-I from most tissues may be resolved into three species, denoted ␣, ␤, and ␥, by subjecting extracts to one-dimensional SDS-PAGE (23). In extracts of rat diaphragm, a relatively smaller amount of a fourth more slowly migrating species, denoted ␥Ј, was also detected (Fig. 5). Previous experiments have established that phosphorylation of the appropriate sites retards the electrophoretic mobility of PHAS-I in SDS-PAGE (23,25,27). Thus, the different bands represent protein phosphorylated to differing extents, and a shift to species of higher apparent M r is indicative of increased phosphorylation. In muscles incubated without insulin or EGF, approximately 25% of the PHAS-I was present as the nonphosphorylated (␣) form, which binds tightly to eIF-4E. As previously observed (21-24), PHAS-I ␣ and ␤ were the predominant forms found bound to eIF-4E when PHAS-I⅐eIF-4E complexes were isolated from ex-

FIG. 2. Effects of insulin and EGF on ERK-1 and ERK-2 activ-
ities. Diaphragms were incubated as described in Fig. 1. Samples (25 g of protein) of muscle extracts were subjected to SDS-PAGE in gels containing MBP, and MAP kinase activities were measured after renaturation by using the "in-gel" assay described by Wang and Erikson (49). Relative activities of ERK-1 and -2 were determined by phosphorimaging. Results are expressed as percent of controls, which were activities detected in extracts of cells incubated without insulin or EGF. tracts of control cells by affinity purification with m 7 GTP-Sepharose (Fig. 5).
Incubating muscles with insulin for 30 min dramatically changed the proportions of the different electrophoretic forms of PHAS-I, indicative of increased phosphorylation of the PHAS-I protein (Fig. 5). The ␥ ϩ ␥Ј forms were increased by approximately 10-fold, and PHAS-I ␣ and ␤ were decreased by approximately 90 and 75%, respectively. Insulin also markedly decreased the amount of PHAS-I that copurified with eIF-4E (Fig. 5), indicating that insulin-stimulated phosphorylation of PHAS-I promotes the dissociation of the PHAS-I⅐eIF-4E complex in skeletal muscle. EGF increased the amount of PHAS-I ␥Ј, but the growth factor did not decrease the amount of PHAS-I bound to eIF-4E. Inhibiting the activation of MAP kinase with PD 098059 had little if any effect on the changes in electrophoretic mobility of PHAS-I produced by insulin. 2 The results with EGF and PD 098059 indicate that the regulation of PHAS-I phosphorylation by insulin in skeletal muscle is medi-ated by a MAP kinase-independent pathway.
Rapamycin-sensitive Regulation of p70 S6K , Glycogen Synthase, and PHAS-I-When immunoprecipitated from extracts of control muscles and detected by immunoblotting, most of the p70 S6K appeared as a band of apparent M r ϭ 70,000 (Fig. 6). Activation of p70 S6K is caused by multisite phosphorylation that results in phosphorylated forms that exhibit decreased electrophoretic mobility when subjected to SDS-PAGE (58). Insulin treatment generated several forms of p70 S6K having reduced electrophoretic mobility, but EGF was without effect on the mobility of the kinase (Fig. 6). In the absence of rapamycin, insulin increased p70 S6K activity by approximately 20fold (Fig. 7A). Basal p70 S6K activity was unaffected by rapamycin; however, insulin-stimulated activity was inhibited approximately 60 and 80% by incubation with rapamycin at 25 and 100 nM, respectively (Fig. 7A).
In the absence of insulin, rapamycin did not significantly affect glycogen synthase (Fig. 7B). However, incubating muscles with 25 nM rapamycin attenuated the effect of insulin, and 100 nM rapamycin abolished the effect of the hormone on activating glycogen synthase (Fig. 7B). Incubating diaphragms with rapamycin in the absence of insulin also had little effect on the phosphorylation state of PHAS-I, as the electrophoretic pattern of the protein was only slightly changed (Fig. 7C). Rapamycin attenuated, but did not abolish, the effect of insulin on increasing PHAS-I phosphorylation (Fig. 7C). In muscles incubated with insulin, rapamycin increased the ␣ and ␤ forms by approximately 4-and 2-fold, respectively. DISCUSSION Activation of MAP kinase by insulin in rat skeletal muscle is neither necessary nor sufficient for the activation of glycogen synthase or the phosphorylation of PHAS-I. Supporting this conclusion are observations that MAP kinase activation by insulin could be abolished by incubating diaphragms with an inhibitor of MEK, which did not significantly attenuate the effects of insulin on activating glycogen synthase (Table I) or on decreasing PHAS-I binding to eIF-4E (Fig. 5). Moreover, MAP kinase was markedly increased by EGF without activating glycogen synthase (Table I), and EGF did not mimic the effects of insulin on PHAS-I (Fig. 5). Consequently, it is unlikely that a small amount of insulin-stimulated MAP kinase activity, as might have been undetected in experiments with the MEK inhibitor, could account for the effects of insulin on PHAS-I and glycogen synthase.
The activation of Rsk by insulin was also abolished by the MEK inhibitor (Fig. 4). This was expected as Rsk is phosphorylated and activated by MAP kinase (12,59). However, the findings that EGF activated Rsk without activating glycogen synthase and that inhibiting Rsk activation with the MEK inhibitor did not abolish synthase activation by insulin (Fig. 4, Table I)   FIG. 6. Effects of insulin and EGF on p70 S6K . Muscles were incubated as described in the legend to Fig. 5. Immunoprecipitations were preformed using antiserum to p70 S6K (lanes 1, 3, and 5) or nonimmune serum (lanes 2, 4, and 6), and samples were subjected to SDS-PAGE. p70 S6K was detected by immunoblotting. The large band below p70 S6K is IgG heavy chain (HC), which was derived from the antibodies used in the immunoprecipitation. (9) in which synthase activation is mediated by ISPK, the rabbit equivalent of Rsk mo2 (10,11,60). Rsk immunoprecipitated from diaphragm extracts phosphorylated [His 6 ]⌬R GL , indicating that the antibodies recognize forms of the kinase that are capable of phosphorylating PP1 G (Fig. 4). Our interpretation of the results is that Rsk can phosphorylate R GL in vitro, but Rsk does not mediate the activation of glycogen synthase by insulin in intact skeletal muscle. It should not be inferred that synthase activation does not involve phosphorylation of PP1 G , although experiments are needed to confirm that site 1 is phosphorylated in response to insulin in skeletal muscle of rats and other species in which the hormone activates glycogen synthase.
The finding that the effects of insulin on activating glycogen synthase and p70 S6K were inhibited by similar concentrations of rapamycin is suggestive of a role of the p70 S6K pathway in the control of synthase (Fig. 7, A and B). Rapamycin is believed to act by binding FKBP-12, an M r ϭ 12,000 protein that is also the receptor for the immunosuppressant, FK-506 (61,62). FKBP-12 possesses peptidyl-prolyl isomerase activity that is inhibited by both rapamycin and FK506. This action does not appear to explain the effect of rapamycin to attenuate insulin signaling, as FK506 blocked the activation of neither p70 S6K (21,36) nor glycogen synthase 2 by insulin. Insensitivity to FK506 would also seem to exclude the calcium-sensitive phosphatase, calcineurin, in the activation of synthase, as the FKBP-12⅐FK506 complex binds to and inhibits calcineurin (61,62). The FKBP-12⅐rapamycin complex does not inhibit calcineurin but instead binds to TOR (target of rapamycin), a protein originally identified in yeast and more recently cloned from mammalian cells (63)(64)(65). The intervening steps between TOR and p70 S6K have not been identified.
Somewhat higher concentrations of rapamycin were needed to inhibit synthase or p70 S6K in the diaphragm than to inhibit p70 S6K in suspended or cultured cells (21,22,33,34,36), where most previous studies of rapamycin sensitivity have been performed. The possibility that rapamycin acts nonspecifically to inhibit activation of p70 S6K and glycogen synthase cannot be excluded. However, to achieve effective concentrations in fibers within the intact muscle, rapamycin must cross diffusion barriers that are not present in cultured cells. This may be why higher concentrations of rapamycin were needed to inhibit p70 S6K in muscle. Alternatively, TOR or other elements in the signal transduction pathways leading to the activation of p70 S6K in skeletal muscle may be less sensitive to rapamycin than those in other cells.
Under conditions in which MAP kinase activity was increased severalfold by EGF, neither the electrophoretic mobility (Fig. 6) nor the activity 2 of p70 S6K was significantly changed. Thus, the failure of EGF to activate synthase is not inconsistent with a role of the p70 S6K pathway in regulating synthase. One potential mechanism by which activation of p70 S6K could lead to synthase activation involves GSK-3, the only protein kinase known to phosphorylate site 3(aϩbϩc) in synthase (4,66). GSK-3 can be phosphorylated and inactivated by p70 S6K in vitro (67), and insulin has been shown to inhibit GSK-3 activity in adipocytes (68), L6 muscle cells (69), and Chinese hamster ovary cells overexpressing the human insulin receptor (CHO.T cells) (70). Inhibition of GSK-3 alone would be insufficient to account for dephosphorylation of synthase in rat skeletal muscle, where insulin promotes dephosphorylation not only of site 3(aϩbϩc) but also of site 2(aϩb) (5, 71), which is not phosphorylated by GSK-3 (4,66). Moreover, there is reason to believe that GSK-3 inhibition is not involved in the rapamycinsensitive regulation of glycogen synthase, as rapamycin did not block the effect of insulin on inhibiting GSK-3 activity by insulin in CHO.T cells (70) or in L6 cells (69).
The effect of rapamycin on glycogen synthase activity in skeletal muscle is similar to that observed by Shepherd et al. (37) in 3T3-L1 cells but differs from findings in rat adipocytes (29,31) and CHO cells (72), where rapamycin did not block the activation of synthase by insulin. Thus, different cells may utilize different pathways for activation of synthase. There is other evidence to support this view. In rat adipocytes glucose potentiates activation of synthase by insulin (38), whereas in skeletal muscle (5) or 3T3-L1 adipocytes (28) glucose has little if any effect on activating the enzyme. In Swiss mouse 3T3 FIG. 7. Rapamycin attenuates the effects of insulin on p70 S6K , glycogen synthase, and PHAS-I. Diaphragms were incubated at 37°C for 30 min with increasing concentrations of rapamycin. Incubations were then continued for 30 min without (circles) or with (squares) 20 nM insulin. A, p70 S6K was immunoprecipitated from extracts, and kinase activity was measured by using 40 S ribosomes and [␥-32 P]ATP as substrates. Activities are expressed as percentages of maximum kinase activity, which in all experiments was observed in extracts of muscles that had been incubated with insulin in the absence of rapamycin. The inset is an immunoblot showing p70 S6K after immunoprecipitation from extracts of muscles that had been incubated without additions (CON), with 25 nM rapamycin (RAP), with 20 nM insulin (INS), or with insulin plus rapamycin (INS ϩ RAP). B, glycogen activity ratios were measured in muscle extracts, and mean values Ϯ S.E. are presented. Total synthase activities (nmol/min/mg extract protein) were as follows: control, 33 Ϯ 7; 25 nM rapamycin, 39 Ϯ 4; 100 nM rapamycin, 35 Ϯ 4; insulin, 30 Ϯ 3; 25 nM rapamycin plus insulin, 34 Ϯ 4; and 100 nM rapamycin plus insulin, 27 Ϯ 6. C, Extracts were subjected to SDS-PAGE, and the relative amounts of the ␣ and ␤ forms of PHAS-I were determined after immunoblotting (see inset). The results are expressed as percentages of the total PHAS-I and are mean values Ϯ S.E. of three experiments. Error bars not shown fall within the symbol. cells, EGF and platelet-derived growth factor activated glycogen synthase (73,74), but EGF did not increase synthase activity in adipocytes (28,29,32) or skeletal muscle (Fig. 1). As skeletal muscle is the most important site of insulin-stimulated glycogen deposition (3), it will be most important to determine the mechanism of synthase activation by insulin in this tissue.
The present study demonstrates that insulin increases the phosphorylation of PHAS-I and dissociation of the PHAS-I⅐eIF-4E complex in skeletal muscle. As PHAS-I inhibits eIF-4E (24), dissociation would be expected to increase rates of initiation, providing a potential explanation of the well established and important action of the hormone on mRNA translation in skeletal muscle (14). The phosphorylation of PHAS-I in response to insulin in skeletal muscle involves a rapamycinsensitive pathway (Fig. 7); however, as observed in adipocytes, rapamycin did not completely inhibit insulin-stimulated phosphorylation of PHAS-I (21,22). Consequently, the control of phosphorylation of PHAS-I by insulin in skeletal muscle also involves a rapamycin-insensitive pathway.
The kinase responsible for the rapamycin-sensitive phosphorylation of PHAS-I has not been identified. It is unlikely that the PHAS-I kinase is p70 S6K , as purified p70 S6K did not directly phosphorylate recombinant PHAS-I (27). This does not exclude the p70 S6K pathway, as kinases either upstream or downstream of p70 S6K might mediate PHAS-I phosphorylation. Similarly, other kinases in the p70 S6K pathway should be considered as candidates for mediating the activation of glycogen synthase, perhaps by phosphorylating site 1 in PP1 G . The recent discovery by Hartley et al. (75) that TOR is homologous to the DNA-dependent protein kinase catalytic subunit raises the intriguing possibility that TOR itself may signal as a protein kinase. Although additional investigation is needed to identify signaling intermediates, our findings indicate that two of the most important actions of insulin, the stimulation of both protein and glycogen synthesis in skeletal muscle, involve a rapamycin-sensitive pathway that is distinct from the MAPkinase signaling pathway.