The Role of Glucose Metabolites in the Activation and Translocation of Glycogen Synthase by Insulin in 3T3-L1 Adipocytes*

The role of increased glucose transport in the hormonal regulation of glycogen synthase by insulin was investigated in 3T3-L1 adipocytes. Insulin treatment stimulated glycogen synthase activity 4–5-fold in these cells. Cytosolic glycogen synthase levels decreased by 75% in response to insulin, whereas, conversely, the glycogenolytic agent isoproterenol increased cytosolic enzyme levels by 200%. Removal of extracellular glucose reduced glycogen synthase activation by 40% and completely blocked enzymatic translocation. Addition of 5 mm 2-deoxyglucose did not restore glycogen synthase translocation but did augment dephosphorylation of the protein by insulin. The translocation event could be reconstitutedin vitro only by the addition of UDP-glucose to basal cell lysates. Amylase pretreatment of the extracts suppressed glycogen synthase translocation, indicating that the enzyme was binding to glycogen. Incubation of 3T3-L1 adipocytes with 10 mmglucosamine induced a state of insulin resistance, blocked the translocation of glycogen synthase, and inhibited insulin-stimulated glycogen synthesis by 50%. Surprisingly, glycogen synthase activation by insulin was enhanced 4-fold, in part due to allosteric activation by a glucosamine metabolite. In vitro, glucosamine 6-phosphate and glucose 6-phosphate stimulated glycogen synthase activity with similar concentration curves. These results indicate that glucose metabolites have an impact on the regulation of glycogen synthase activation and localization by insulin.

The role of increased glucose transport in the hormonal regulation of glycogen synthase by insulin was investigated in 3T3-L1 adipocytes. Insulin treatment stimulated glycogen synthase activity 4 -5-fold in these cells. Cytosolic glycogen synthase levels decreased by 75% in response to insulin, whereas, conversely, the glycogenolytic agent isoproterenol increased cytosolic enzyme levels by 200%. Removal of extracellular glucose reduced glycogen synthase activation by 40% and completely blocked enzymatic translocation. Addition of 5 mM 2-deoxyglucose did not restore glycogen synthase translocation but did augment dephosphorylation of the protein by insulin. The translocation event could be reconstituted in vitro only by the addition of UDP-glucose to basal cell lysates. Amylase pretreatment of the extracts suppressed glycogen synthase translocation, indicating that the enzyme was binding to glycogen. Incubation of 3T3-L1 adipocytes with 10 mM glucosamine induced a state of insulin resistance, blocked the translocation of glycogen synthase, and inhibited insulinstimulated glycogen synthesis by 50%. Surprisingly, glycogen synthase activation by insulin was enhanced 4-fold, in part due to allosteric activation by a glucosamine metabolite. In vitro, glucosamine 6-phosphate and glucose 6-phosphate stimulated glycogen synthase activity with similar concentration curves. These results indicate that glucose metabolites have an impact on the regulation of glycogen synthase activation and localization by insulin.
Insulin is responsible for the maintenance of blood glucose levels in a narrow physiological range. Under hyperglycemic conditions, insulin increases glucose uptake and stimulates the rate-limiting enzymes that regulate glucose oxidation and storage. In healthy subjects, 80 -90% of glucose disposal occurs in skeletal muscle, where it is primarily stored as glycogen. Insulin acutely stimulates glycogen synthesis through activation of glycogen synthase (GS) 1 and inactivation of glycogen phosphorylase. The coordinate increase in glucose uptake and regulation of glycogen metabolizing enzymes accounts for the marked stimulation of glycogen synthesis by insulin.
GS, the rate-limiting enzyme in glycogen synthesis, is regulated allosterically and by covalent modification (1,2). The protein is phosphorylated on six residues by a variety of kinases, which cumulatively inhibit its activity. Insulin-stimulated activation of glycogen targeted protein phosphatase-1 (PP1) results in the dephosphorylation of GS (3). The hormone can also produce the inactivation of glycogen synthase kinase-3 (GSK-3), resulting in the disinhibition of GS by preventing its phosphorylation (4 -6). Additionally, binding of glucose 6-phosphate (G6P) allosterically activates the enzyme and increases its susceptibility to dephosphorylation (7)(8)(9). The increase in glucose transport and phosphorylation caused by insulin results in elevated levels of intracellular G6P, producing further activation of GS.
Although the signaling pathways involved the metabolic actions of insulin are unclear, recent work has implicated an important role for phosphatidylinositol-3Ј kinase (PI3Ј-K). Stimulation of PI3Ј-K by insulin results in the sequential activation of phosphoinositide-dependent protein kinase and Akt kinase (reviewed in Ref. 10). Overexpression of Akt has been reported to increase glucose transporter 4 vesicle translocation (11)(12)(13), whereas blockade of Akt activation by PI3Ј-K inhibitors prevented insulin induced increases in glucose transport (14 -16) and GS activation (17)(18)(19)(20). Additionally, Akt phosphorylates and inactivates GSK-3 (21), which may increase the amount of dephosphorylated, active GS in the cell (4,10). However, endogenous Akt activation is not sufficient for glucose transporter 4 vesicle translocation (22) or GS activation (5,23,24), and overexpression of a dominant negative Akt construct did not block insulin-stimulated glucose uptake (25). Furthermore, the importance of GSK-3 inactivation in regulating GS activity has not been conclusively demonstrated, and does not appear to contribute to the robust stimulation of GS by insulin in the 3T3-L1 adipocytes (5,24).
Previous work has demonstrated that glucose metabolism plays an important role in the activation of GS by insulin. In primary adipocytes, insulin or elevated glucose alone had little effect on GS activity, but together, they produced a synergistic dephosphorylation and activation of the enzyme (8,26). Exposure of hepatocytes to glucose resulted in a dose-dependent increase in GS activity, consistent with changes in intracellular G6P levels (27,28). Additionally, hyperglycemia also resulted in the translocation of hepatic GS from the cytosol to a particulate fraction in normal and diabetic rats (29,30). These results suggest that glucose uptake and metabolism correlate well with the activation of GS in the liver (reviewed in Ref. 31).
The specific activation of discrete pools of intracellular enzymes underlies the unique metabolic effects of insulin (32). Signaling enzymes, such as PP1 and PI3Ј-K, are differentially activated by insulin as compared with other growth factors (24,33,34). We therefore investigated changes in GS activation and localization in response to insulin in 3T3-L1 cells. We demonstrate here that increased glucose uptake and metabolism play an important role in the intracellular localization of GS and contributes to the regulation of the enzyme by insulin.

EXPERIMENTAL PROCEDURES
Materials-Cell culture reagents were obtained from Life Technologies, Inc, with the exception of sera, which were supplied by Summit Biotechnology (Ft. Collins, CO). Insulin, glucose metabolites, and differentiation agents were from Sigma. UDP-[U-14 C]glucose (286 mCi/ mmol) was from ICN, and [U-14 C]glucose (251 mCi/mmol) was obtained from NEN Life Science Products. ECL reagent was purchased from Amersham Pharmacia Biotech, whereas GF/A filters were supplied by Whatman. Affinity purified chicken anti-PP1␣, anti-pan-PP1, and antiglycogen synthase antibodies were the generous gift of Dr. J. Lawrence (University of Virginia). Horseradish peroxidase-conjugated rabbit anti-chicken IgG was from Accurate Chemical Corp. (Westbury, NY).
Cell Culture and Extract Preparation-3T3-L1 fibroblasts were maintained in Dulbecco's modified Eagle's medium (DMEM) (high glucose) plus 10% calf serum. Following 2 days at confluence, differentiation was initiated by the addition of DMEM containing 10% fetal bovine serum, 167 nM insulin, 0.25 mol/liter dexamethasone, and 0.5 mM isobutylmethylxanthine. Three days later, the medium was replaced with DMEM plus 10% fetal bovine serum and 167 nM insulin. After 2 more days, the medium was switched to DMEM plus 10% fetal bovine serum. Adipocytes were used 5-14 days after completion of the differentiation protocol, when Ͼ95% of the cells expressed the adipocyte phenotype. Prior to experiments, cells were rinsed two times with low serum medium (DMEM containing 5 mM glucose, 0.5% fetal bovine serum, 25 mM Hepes (pH 7.4), 100 units/ml penicillin, 100 units/ml streptomycin, and 0.29 mg/ml glutamine), and then incubated in the same medium for 2.5-3 h. When external glucose conditions were varied, cells were rinsed twice with low serum medium without glucose and once with low serum medium containing the indicated addition and then were preincubated in the same medium. Following treatment, cells were rapidly washed three times with ice-cold phosphate-buffered saline and were harvested in glycogen synthase buffer (50 mM Hepes (pH 7.8), 100 mM NaF, 10 mM EDTA, 2 mg/ml glycogen, with 0.1 mM phenylmethylsulfonyl flouride, 1 mM benzamidine. and 10 mg/liter aprotinen added just before use). 0.1 mM sodium orthovanadate was also added when tyrosine phosphorylation was measured. Adipocytes were lysed by sonication (10 s, 20% output), and centrifuged at 2500 ϫ g for 5 min at 4°C to pellet nuclei. The resulting postnuclear supernatant (PNS) fractions were centrifuged at 10,000 ϫ g for 15 min to pellet plasma membranes and then at 100,000 ϫ g for 30 -60 min to separate the cytosolic fraction from the glycogen-enriched pellet. Particulate fractions were resuspended in fresh glycogen synthase buffer using a 23-gauge needle.
Glycogen Synthase Assay-Glycogen synthase activity in the various cellular fractions was measured as described previously (35). Briefly, 25-50 l of lysate was assayed in a final volume of 100 l of glycogen synthase buffer containing 5 mM UDP-glucose and 1 Ci/ml [U-14 C]UDP-glucose, in the absence and presence of 10 mM G6P. Samples were incubated at 37°C and then placed on ice for 15 min. 90 l of the reaction was spotted on GF/A filters, dried for 3 s, and then placed in 70% ethanol on ice. Filters were washed for 20 min at 4°C and then washed two more times in 70% ethanol at room temperature. Filters were air dried, and [U-14 C]UDP-glucose incorporation into glycogen was measured by liquid scintillation counting.
Glycogen Synthesis Assay-3T3-L1 adipocytes grown in six-well dishes were serum-starved for 3 h in the presence of either 10 mM glucose or 10 mM glucosamine (pH 7.4). The cells were then switched to low serum medium containing either 10 mM glucose or 5 mM glucose and glucosamine. Following a 15-min stimulation with 100 nM insulin, 1 or 2 Ci of [ 14 C]glucose was added to each well; a final specific activity of approximately 230 cpm/nmol glucose was attained for both conditions. After a 45-min incubation at 37°C, cells were washed three times with phosphate-buffered saline at 4°C. [ 14 C]Glucose incorporation into glycogen was then determined as described previously (35).
In Vitro Translocation of GS-3T3-L1 adipocytes were serumstarved, harvested, and lysed by sonication. Nuclei and plasma membranes were pelleted by centrifugation, and the resulting supernatants were transferred to fresh tubes. Stocks of glucose metabolites were freshly prepared and diluted into the samples, which were nutated at 4°C for 15 min, and then subjected to ultracentrifugation (100,000 ϫ g for 30 -60 min). The resulting cytosolic fractions were assayed for glycogen synthase activity; the identical samples were also analyzed by anti-glycogen synthase/anti-PP1␣ immunoblotting. For experiments involving UDP-glucose addition, the final UDP-glucose concentration in the assay was accounted for in the specific activity determination. In experiments involving amylase, extracts were preincubated for 15 min at 30°C in the absence and presence of 40 g/ml of amylase. Samples were then chilled on ice and processed as indicated above.
Other Procedures-Immunoblotting was performed as described previously (35). Protein determination was by the Bradford method.

Glycogen Synthase Localization in 3T3-L1 Adipocyte Subcellular Fractions Is Subject to Hormonal
Regulation-The effects of insulin and isoproterenol on GS activity in different intracellular fractions were measured. Cells were serum-starved for 3 h, followed by treatment with 100 nM insulin for 15 min. Adipocytes were lysed by sonication, and a PNS fraction was prepared by centrifugation at 2000 ϫ g. The PNS lysate was further fractionated by differential centrifugation, to prepare the plasma membrane, cytosolic, and glycogen-enriched fractions. Samples were analyzed by immunoblotting with anti-GS and anti-PP1 antibodies. Neither treatment had any effect on the total amount of GS or PP1 within the cell (Fig. 1, lanes 1-3). However, insulin treatment caused a 75% reduction in cytosolic GS levels (lanes 4 and 5), and a reciprocal increase in the amount of GS found in the plasma membrane fraction (lanes 7 and 8). Conversely, the glycogenolytic agent isoproterenol doubled the amount of cytosolic GS and reduced plasma membrane enzymatic levels (Fig. 1, lanes 4 and 6 and lanes 7 and 9). Identical results were obtained in GS activity assays (data not shown). Neither agent had any effect on the localization of PP1 (Fig. 1A), GSK-3␤, and the catalytic subunit of phosphorylase kinase (data not shown). However, the hormone-induced changes in cytosolic GS were greater than the corresponding shifts in plasma membrane levels, indicating that another cellular compartment was involved. Subsequent data indicated that GS also translocated to the glycogen-enriched fraction in response to insulin. These results suggest that the intracellular localization of GS is subject to reciprocal regulation by opposing hormonal agents that control glycogen metabolism. The time course of the effect of insulin on cytosolic GS was measured. 3T3-L1 adipocytes were treated for various times with 100 nM insulin, cytosolic fractions were prepared and GS levels were measured by activity assay and immunoblotting. Insulin induced a rapid decrease in cytosolic GS activity, which was maximal by 15-30 min ( Fig. 2A). Anti-GS immunoblots revealed corresponding changes in GS protein levels, whereas cytosolic PP1 levels were unchanged by insulin (Fig. 2B). The time course of GS activation by insulin in the glycogen fraction slightly preceded the ability of insulin to reduce cytosolic GS levels (Fig. 2C). The translocation of GS by insulin described in Fig. 2A was stable. Following a 15 min stimulation with insulin, cells were extensively washed and then placed in medium to recover. Under these conditions, the tyrosine phosphorylation of the insulin receptor and insulin receptor substrate-1 returned to the basal state within 1.5 h (data not shown). However, cytosolic GS levels did not recover until 6 -24 h later (Fig. 2D). Interestingly, isoproterenol increased cytosolic GS levels in cells that had been pretreated with insulin for 30 min, indicating that GS localization was responsive to sequential, opposing stimuli (data not shown).
Role of Glucose in the Activation and Translocation of GS by Insulin-Ferná ndez-Novell et al. (27,29) have previously implicated a role for glucose uptake and phosphorylation in the regulation of hepatic GS localization. To determine whether the effects of insulin described above could be attributed to increased glucose metabolism, the effects of extracellular glucose on the insulin-stimulated activation and translocation of GS in 3T3-L1 adipocytes were investigated. Following serum starvation, cells were stimulated with 100 nM insulin for 15 min, and GS activity and localization were examined. In the presence of 5 mM extracellular glucose, insulin caused a 4-fold increase in the G6P-independent GS activity in the PNS fraction (Fig. 3A). Removal of extracellular glucose caused a 40% reduction in GS activation by insulin. Inclusion of 5 mM 2-deoxyglucose in the media led to a dramatic increase in both basal and insulin-stimulated GS activity, probably due to the intracellular accumulation of 2-deoxyglucose 6-phosphate (26). In FIG. 2. Time course of insulin-induced GS translocation and activation. 3T3-L1 adipocytes were treated with 100 nM insulin for the indicated times, and cytosolic fractions were prepared and analyzed by GS activity assay (A) or anti-GS immunoblotting (B). Assays in A were performed in the presence of 10 mM G6P, to measure total GS activity present. C, time course of GS translocation and activation. Cells were treated for the indicated times with 100 nM insulin, cytosolic and glycogen fractions were prepared, and GS activity was measured. Values for maximal GS activity ratio in the glycogen fraction and maximal translocation of cytosolic GS activity were set at 100%. D, reversal of cytosolic GS translocation. Replicate plates of 3T3-L1 adipocytes were treated for 15 min in the absence and presence of 100 nM insulin and then washed four times with phosphate-buffered saline. Cells were either harvested immediately (time 0) or allowed to recover in DMEM/0.5% fetal bovine serum for the indicated times. Cytosolic fractions were prepared, and GS/PP1␣ immunoreactivity was measured. All panels are representative of at least three independent experiments. parallel, cells were further fractionated, and cytosolic GS levels were measured. In the presence of 5 mM extracellular glucose, insulin caused a 75% reduction in cytosolic GS total activity (Fig. 3B), with a commensurate change in GS protein levels (Fig. 3B, inset). In the absence of glucose, the translocation of GS was completely blocked (Fig. 3B). In the presence of 5 mM 2-deoxyglucose, insulin produced a dramatic increase in the gel mobility of GS (Fig. 3B, inset), indicative of a decrease in the net phosphorylation state of the enzyme (8). However, under these conditions, the translocation of cytosolic GS induced by insulin was not restored (Fig. 3B). These results are in contrast to previous work in hepatocytes, in which insulin reportedly caused a similar translocation of GS in the presence of either glucose or 2-deoxyglucose (27,29). Variations in endogenous glucose production or different GS isoforms in the two cell types may explain this discrepancy.
These data indicate that the insulin-stimulated translocation of cytosolic GS requires increased glucose uptake and further metabolism after phosphorylation by hexokinase. To identify the glucose metabolite responsible for this activity, an in vitro assay for the translocation of GS was developed. Lysates were prepared from basal 3T3-L1 adipocytes, and a combined cytosolic/glycogen fraction was prepared. 5 mM G6P, G1P, or UDP-glucose was added, mixed at 4°C for 15 min, and then subjected to ultracentrifugation to pellet the glycogen particles. GS was assayed in the resulting cytosolic supernatant. Addition of both G6P and G1P had no effect on GS total activity (Fig. 4A) or protein levels (Fig. 4B), further confirming that G6P does not mediate the effect of insulin on GS localization in 3T3-L1 adipocytes. However, addition of UDP-glucose reduced cytosolic GS total activity by over 50% (Fig. 4A). Anti-GS immunoblots revealed a similar decrease in GS levels (Fig. 4B). Pretreatment of the lysate with amylase prior to UDP-glucose addition completely blocked the loss of cytosolic GS (Fig. 4B, lanes 5 and 6), indicating that the enzyme was translocating directly to glycogen. The effect was specific for UDP-glucose, as the addition of UDP-galactose and UDP-mannose had no effect (Fig. 4C). UDP-glucose induced the translocation of cytosolic GS from basal extracts with an EC 50 of approximately 0.7 mM (Fig. 5). Pretreatment of the cells with insulin and 5 mM 2-deoxyglucose, to maximize GS dephosphorylation, resulted in a 5-fold leftward shift of the UDP-glucose concentration curve (Fig. 5), indicating that activated GS may be more sensitive to changes in intracellular UDP-glucose levels.
Differential Inhibition of Insulin Signaling by Glucosamine in 3T3-L1 Adipocytes-Glucosamine has been proposed to mediate insulin resistance resulting from chronic hyperglycemia (36 -38). Because this sugar may undergo a metabolic fate similar to glucose, its effects on GS activation by insulin were examined. 3T3-L1 adipocytes were serum-starved for 3 h in the presence of either 10 mM glucose or glucosamine prior to insulin stimulation. Under this protocol, insulin signaling was reduced by approximately 50% as measured by either the tyrosine phosphorylation of the insulin receptor or insulin receptor substrate-1 (Fig. 6A), or the phosphorylation of Akt or mitogenactivated protein kinase ( Fig. 6B and C). These results are consistent with another report in 3T3-L1 adipocytes (39), despite differences in the glucosamine preincubation protocols.
Substitution of glucose with glucosamine in the medium completely blocked the insulin-stimulated translocation of GS from the cytosolic fraction, similar to incubation of the cells in the absence of glucose (Fig. 7A). However, the dephosphorylation of GS by insulin was still readily apparent in the presence of glucosamine as determined by gel mobility shift (Fig. 7A). Surprisingly, glucosamine preincubation resulted in an apparent 4-fold increase in insulin-stimulated GS activity in the PNS fraction (Fig. 7B). However, when glycogen-enriched pellets from the same lysates were prepared by ultracentrifugation and resuspended in fresh buffer prior to assay, the effect of glucosamine on insulin stimulated-GS activity was markedly reduced (Fig. 7B). These data suggested that glucosamine treatment resulted in the intracellular accumulation of an allosteric activator of GS, which was carried over into the in vitro GS assays.
Because glucosamine is structurally related to 2-deoxyglucose, the accumulation of glucosamine 6-phosphate following insulin treatment may result in direct activation of GS. Therefore, the effects of glucosamine 6-phosphate on GS activity were compared with those of G6P. PNS fractions were prepared from basal 3T3-L1 adipocytes, and GS activity was measured in vitro in the presence of the increasing concentrations of either G6P or glucosamine 6-phosphate. The two metabolites activated GS with identical concentration curves, although ab- solute GS activity was about 20% lower at all glucosamine 6-phosphate concentrations (Fig. 7C). Addition of either 10 mM glucose 1-phosphate or 10 mM glucosamine 1-phosphate to the assay had no effect on GS activity (data not shown).
The effects of glucosamine on insulin-stimulated glycogen synthesis in 3T3-L1 adipocytes were next examined. 3T3-L1 adipocytes were serum-starved for 3 h in the presence of either 10 mM glucose or glucosamine. Immediately prior to insulin stimulation, the medium was switched to one containing either 10 mM glucose or 5 mM glucose plus 5 mM glucosamine.

. Reconstitution of GS translocation in vitro.
A, extracts were prepared from basal 3T3-L1 adipocytes and subjected to centrifugation at 10,000 ϫ g. 10 mM concentrations of the indicated glucose metabolites were added to the resulting supernatant. Samples were nutated for 15 min at 4°C and then centrifuged at 100,000 ϫ g to separate the cytosol and glycogen-enriched pellet. Cytosolic GS activity was assayed in the presence of 10 mM G6P. G1P, glucose 1-phosphate; UDP-Glu, UDP-glucose. B, basal extracts were treated as in A, and samples were analyzed by immunoblotting (lanes 1-4). Additionally, the cytosol/glycogen fraction was pretreated with 40 g/liter amylase for 15 min at 30°C, prior to addition of UDP-glucose and ultracentrifugation (lanes 5 and 6). the cells pretreated with glucose (Fig. 7D). However, despite the dramatic enhancement of insulin-stimulated GS activity in the presence of glucosamine, glycogen synthesis was reduced by 60% in the glucosamine-pretreated cells (Fig. 7D). This deficit most likely results from diminished insulin receptor signaling and ATP levels (Fig. 6A) (37), as well as direct inhibition of glucose uptake and phosphorylation by glucosamine. The reported reduction in UDP-glucose levels during glucosamine treatment (40, 41) may arise from inhibition of glucose uptake coupled with the allosteric activation of GS by glucosamine 6-phosphate. DISCUSSION The molecular basis for the specificity in signal transduction for insulin remains uncertain. One factor that is likely to play an important role is spatial compartmentalization. Indeed, the specific activation of discrete pools of proteins may explain the unique spectrum of effects of insulin. Moreover, many of the FIG. 7. Regulation of GS activity and localization by glucosamine. 3T3-L1 adipocytes were pretreated for 3 h in low serum medium containing the indicated addition. Following a 15-min treatment with 100 nM insulin, cells were harvested and fractions were prepared. The cytosolic fractions were analyzed by anti-GS/PP1␣ immunoblotting (A), whereas the PNS and glycogen fractions were analyzed by GS activity assay (B). C, allosteric activation of GS in vitro. PNS fractions were prepared from basal cells, and GS activity was measured in the presence of the indicated amounts of G6P or GlcN6P. GS activity measured at 10 mM G6P was set at 100%. D, inhibition of insulin-stimulated glycogen synthesis by glucosamine. Replicate wells were preincubated 2.5 h with low serum medium containing either 10 mM glucose or glucosamine. Immediately prior to insulin stimulation, cells were switched to medium containing either 10 mM glucose or 5 mM glucose and GlcN. After a 15-min incubation with 100 nM insulin, 1-2 Ci of [ 14 C]glucose was added; specific activity was approximately 230 cpm/nmol in all wells. After 45 min, cells were washed three times, and glycogen was isolated. A and B are representative of three experiments, and C and D are the average Ϯ S.D. of three independent experiments. cellular actions of insulin result from changes in cellular localization of signaling molecules. We report here that the activation of GS by insulin involves changes in its cellular compartmentalization, in addition to previously described covalent modifications. Insulin primarily stimulated GS dephosphorylation in the glycogen-containing fraction. Additionally, the hormone caused the dramatic reduction in the cytosolic levels of GS protein, as measured both by enzyme activity and immunoblotting. Agents that promoted glycogen breakdown, such as isoproterenol (Fig. 1) and norepinephrine (data not shown), result in a 2-3-fold increase in cytosolic GS levels. The effects of insulin and isoproterenol on cytosolic GS were mirrored by reciprocal changes in the crude plasma membrane fraction, which also contains glycogen. However, the amount of cytosolic GS translocating in response to hormonal treatment was greater than the corresponding change in plasma membrane enzyme levels, indicating that another cellular compartment was also involved. In vitro experiments (Fig. 4B) suggest that GS translocates to glycogen in response to insulin, secondary to an elevation in intracellular UDP-glucose levels. The high basal enzyme levels most likely masked any changes in GS in the glycogen-enriched pellet. In contrast, there was no detectable change in the localization of other enzymes that regulate glycogen metabolism, namely PP1, GSK-3␤, and the catalytic subunit of phosphorylase kinase. Together, these data suggest a specific, reversible translocation of GS between the cytosolic and glycogen-containing fractions, paralleling changes in glycogen accumulation and breakdown.
Previous work in skeletal muscle had suggested that changes in the cellular localization of PP1 (42,43) are primarily responsible for the regulation of glycogen synthesis. Differential phosphorylation of the PP1 glycogen targeting subunit, G M , by insulin and agents that elevated intracellular levels of cAMP was reported to underlie the regulation of phosphatase activity present at the glycogen particle (44). Thus, during conditions of increased glycogen synthesis, more PP1 activity was stimulated, whereas under glycogenolytic conditions, PP1 translocated away from glycogen into the cytosol. However, in 3T3-L1 adipocytes, we were unable to detect any translocation of PP1 following stimulation with either insulin or adenylate cyclase activators (Fig. 1), in agreement with previous results from primary adipocytes (45). The principal PP1 glycogen targeting subunit in adipocytes, PTG, does not contain the two putative regulatory phosphorylation sites found in G M (46), nor is PTG phosphorylated in response to insulin or forskolin (45). Thus, translocation of PP1 does not appear to be required for the hormonal regulation of GS activity in adipocytes.
Hepatic GS has been reported to translocate from the cytosol to a particulate fraction in response to elevations of extracellular glucose (27,29). However, the translocation pattern was unaffected by amylase treatment of the lysates (29), indicating that hepatic GS was bound to a site other than glycogen. Interestingly, GS translocated to an actin-rich area near the plasma membrane, where most of the de novo glycogen synthesis occurred (47). The degree of GS translocation in hepatocytes correlated tightly with changes in intracellular G6P levels (27,29), but presumably, UDP-glucose levels also increased in parallel. However, GS translocation was unaffected by substitution of extracellular glucose with 2-deoxyglucose, indicating that UDP-glucose does not mediate hepatic GS translocation (27,29). These data suggested that G6P regulates GS activity by several mechanisms: inducing translocation of the enzyme, enhancing its dephosphorylation by PP1, and allosterically activating the enzyme.
In 3T3-L1 adipocytes, it appears that GS translocation is regulated in a different fashion. Removal of extracellular glu-cose completely blocked the translocation of cytosolic GS (Fig.  3A) and diminished the insulin-stimulated enzymatic activation, although not to the extent observed in primary adipocytes (8). Substitution of extracellular glucose with 2-deoxyglucose dramatically enhanced the insulin-induced dephosphorylation of GS (Fig. 3B). However, the translocation of cytosolic GS in response to insulin was completely blocked by 2-deoxyglucose (Fig. 3B) and 3-O-methyl glucose (data not shown), indicating that a glucose metabolite formed downstream of G6P was involved. Cytosolic GS translocation was specifically replicated in vitro by the addition of UDP-glucose to basal 3T3-L1 adipocyte lysates, whereas G6P and several other sugar metabolites were without effect. In contrast to results in hepatocytes (29), pretreatment of the lysates with amylase completely blocked the effect of UDP-glucose on cytosolic GS levels, indicating that GS translocated directly to glycogen. Thus, in 3T3-L1 adipocytes, elevation of UDP-glucose increases the affinity of GS for its other substrate, glycogen. The reason for the discrepancy between the results in hepatocytes and 3T3-L1 adipocytes is unclear, but it may represent tissue-specific differences in the regulation of GS.
In the absence of extracellular glucose, insulin stimulation of GS activity was diminished, but it occurred without any change in cytosolic GS levels. This result suggests that enzymatic activation and translocation are unrelated. However, the regulation of GS localization on the glycogen particle has not been extensively examined. Presumably, to mediate glycogen synthesis, GS must be both activated and bound to a portion of the glycogen polymer that can be elongated. It is tempting to speculate that UDP-glucose may specifically induce GS to bind to the ends of the glycogen chains, where the enzyme could efficiently catalyze the incorporation of UDP-glucose into glycogen. If so, GS translocation may represent a disproportionate component of enzyme mediating the stimulation of glycogen synthesis by insulin. Conversely, the removal of activated GS from glycogen to the cytosol by isoproterenol is an additional mechanism for the enhancement of glycogenolysis.
The stimulation of glucose transport and metabolism by insulin thus appears to regulate GS activity by a several mechanisms. Increased intracellular levels of G6P can allosterically activate GS, as well as enhance the dephosphorylation and activation of GS by PP1. Furthermore, elevation of UDP-glucose levels appears to induce the translocation of GS to glycogen, where it is in proximity to both its substrate, glycogen, and its insulin-responsive activator, glycogen-targeted PP1. Thus, a system for the coordinate translocation of glucose transporter and GS may potentiate the insulin-meditated increase in glycogen storage. In contrast, insulin and isoproterenol had no effect on the subcellular distribution of PP1. Therefore, in 3T3-L1 adipocytes, the translocation of GS, rather than PP1, appears to partially underlie the hormonal regulation of glycogen metabolism. Experiments are currently under way to assess the relative importance of subcellular localization of these two enzymes in skeletal muscle, in response to hormonal stimulation and exercise.