The Regulation of Glycogen Synthase by Protein Phosphatase 1 in 3T3-L1 Adipocytes

The stimulation of glycogen-targeted protein phosphatase 1 (PP1), glycogen synthase, and glycogen synthesis by insulin was examined during the differentiation of 3T3-L1 fibroblasts into adipocytes. Insulin treatment barely changed the low levels of glycogen synthesis measured in fibroblasts. Following differentiation into adipocytes, insulin increased glycogen synthesis up to 40-fold. After further culturing of the adipocytes for a week, insulin stimulated glycogen accumulation 700-fold. Differentiation of 3T3-L1 cells also resulted in the increased expression of glycogen synthase and in increases in both total glycogen synthase activity and -fold stimulation by insulin. While the levels of PP1 protein were unchanged by differentiation, PP1 specific activity decreased over 60%, although sensitivity to insulin treatment was augmented. Concurrently, levels of the PP1 inhibitor protein DARPP-32 were dramatically induced upon 3T3-L1 adipogenesis. DARPP-32 in both 3T3-L1 and primary rat adipocytes was exclusively localized to the particulate fractions, including the glycogen-enriched pellet. PP1 activity from 3T3-L1 adipocytes exhibited a kinetic lag in vitro, which was not present in fibroblast extracts. Insulin pretreatment of the adipocyte cells overcame thein vitro lag in PP1 activity, resulting in up to 5-fold stimulation of PP1 activity being measured at early assay time points. These results suggest that in 3T3-L1 adipocytes, DARPP-32 may maintain glycogen-targeted PP1 activity in a low basal state, priming the phosphatase for stimulation by insulin.

Protein phosphorylation plays a critical role in the regulation of lipid and glucose metabolism (1). As the major anabolic hormone regulating glucose utilization and storage, insulin exerts many of its effects by promoting the net dephosphorylation of enzymes such as glycogen synthase, glycogen phosphorylase, and phosphorylase kinase, resulting in the stimulation of glycogen synthesis. Several lines of evidence indicate that these insulin-stimulated dephosphorylations are catalyzed by protein phosphatase 1 (PP1). 1 This phosphatase is found in nearly all cellular compartments and is thought to be targeted intracellularly by specific proteins (2,3). Such targeting proteins have been identified that localize PP1 to the nucleus (4), the sarcoplasmic reticulum (5), glycogen, and myofibrils (reviewed in Ref. 3). The best characterized of these are the mammalian glycogen-targeting proteins. Three related members have been described: G M , isolated from skeletal muscle (6,7), the hepatic G L protein (8,9), and the recently cloned PTG protein (10,11). Unlike the extremely restricted expression of G M and G L , PTG is highly expressed in the major insulinresponsive tissues, including fat, skeletal muscle, heart, and liver. 2 PP1 activity is controlled by targeting proteins in several ways. First, by localizing PP1 to subcellular compartments, such as glycogen, targeting proteins serve to increase PP1 specific activity against co-localized substrates. The two glycogen-targeting subunits, G M and G L , have also been reported to differentially regulate PP1 activity toward the same glycogenbound substrates (9,12). Phosphorylation of G M has been proposed to directly regulate PP1 activity in response to both anabolic and catabolic signals (12), although this model has been questioned (13)(14)(15)(16)(17). To add to the uncertainty, the two putative regulatory phosphorylation sites of G M are not conserved in PTG or G L , and both proteins are poorly phosphorylated in vitro (9,18). Since it is unlikely that PTG is regulated by phosphorylation in vivo (18), other models for the stimulation of glycogen-targeted PP1 need to be explored. PP1 activity is also regulated by inhibitor (Inh) proteins (reviewed in Ref. 19). There are three inhibitors of PP1 activity, Inh-1, its neuronal homologue DARPP-32, and Inh-2. Phosphorylation of a conserved threonine residue on Inh-1 and DARPP-32, by cAMP-dependent protein kinase, converts these molecules into highly potent, reversible, and very specific inhibitors of PP1 activity. Dephosphorylation of these proteins results in their inactivation and disassociation from PP1. Inh-2 is also regulated by phosphorylation, but it acts more as a regulatory subunit. The direct regulation of glycogen-targeted PP1 by these inhibitors has not been extensively investigated.
We have investigated the regulation of glycogen synthesis by insulin in the 3T3-L1 cell line. Differentiation of these cells from fibroblasts into adipocytes results in a large increase in both total metabolic activity and insulin responsiveness. To evaluate the regulation of the components directly involved in glycogen synthesis, we examined the effects of insulin treatment on glycogen synthase and glycogen-targeted PP1 activity during differentiation of 3T3-L1 cells. The results suggest that in 3T3-L1 adipocytes, the inhibitor peptide DARPP-32 may regulate glycogen-associated PP1 activity and may be a target of insulin signaling to increase glycogen synthesis.

EXPERIMENTAL PROCEDURES
Materials-Cell culture reagents, phosphorylase, and phosphorylase kinase were from Life Technologies, Inc. Okadaic acid was obtained from Calbiochem. [␥-32 P]ATP (3000 Ci/mmol) and [U-14 C]glucose (251 mCi/mmol) were purchased from NEN Life Science Products, while UDP-[U-14 C]glucose (261 mCi/mmol) was from ICN. Chicken affinitypurified anti-PP1 and anti-glycogen synthase antibodies were generous gifts of Dr. J. Lawrence (University of Virginia). For ECL detection, horseradish peroxidase-conjugated rabbit anti-chicken and goat antimouse IgG were purchased from Accurate Chemical Corp. (Westbury, NY) and Bio-Rad, respectively.
Cell Culture and Fractionation-3T3-L1 fibroblast cultures were maintained and differentiated as described previously (15). Except where indicated, adipocytes were used 7-12 days postdifferentiation. Cells were routinely examined using a microscope to estimate the percentage of differentiated cells and to monitor lipid accumulation. Prior to treatments, adipocyte cultures were serum-starved for 3-24 h in Krebs-Ringer buffer with 30 mM Hepes (KRBH, pH 7.4) plus 0.5% bovine serum albumin and 2.5 mM glucose. Following stimulation, the plates were washed three times with ice-cold phosphate-buffered saline. Cells were collected in the appropriate buffer and lysed by sonication. Cellular fractions were then prepared as described previously (18). For crude particulate preparation, the PNS fraction was directly subjected to the ultracentrifugation step. Primary rat adipocytes were isolated from epididymal fat pads as described previously (20), and frozen lysates were provided by Dr. Cynthia Mastick (Parke-Davis).
Assay of PP1 Activity-For PP1 assays, cells were scraped in PP1 homogenization buffer (50 mM Hepes (pH 7.2), 2 mM EDTA, 0.2% 2-mercaptoethanol, and 2 mg/ml glycogen) plus 10 g/ml aprotinin, 1 mM benzamidine, and 0.1 mM phenylmethylsulfonyl fluoride added just before use. Cells were lysed by sonication and fractionated as described above. PP1 activity was measured by diluting 0.5-2 g of lysate into 20 l of PP1 homogenization buffer plus 4.5 nM okadaic acid. Samples were incubated for 2-3 min at the indicated temperature, and the reaction was initiated by the addition of 10 l (15 g) of 32 P-labeled phosphorylase a (3 nM okadaic acid and 5 mM caffeine, final concentration). Phosphate release was measured as described previously (15), or the reactions were terminated by the addition of 90 l of ice-cold 20% trichloroacetic acid and 5 l of 2% bovine serum albumin (modified from Ref. 21). Following a 10-min incubation on ice, samples were centrifuged for 2 min at 15,000 ϫ g, and phosphate released into the supernatant was measured by liquid scintillation counting. 32 P-Phosphorylase a was prepared as described by Lazar et al. (15).
Assay of Glycogen Synthase Activity and Glycogen Synthesis-For glycogen synthase assays, cells were collected into 50 mM Hepes (pH 7.8), 10 mM EDTA, and 100 mM NaF plus protease inhibitors. Following sonication, extracts were subjected to centrifugation (10 min; 2500 ϫ g). Glycogen synthase activity was then measured in the PNS fraction as described (15). Glycogen synthesis was assayed as described (15) with minor modifications. Briefly, six-well dishes were serum-deprived in KRBH containing 2.5 mM glucose and 0.5% bovine serum albumin. Triplicate wells were treated either with water or 100 nM insulin for 15 min at 37°C. Subsequently, cells were incubated with 2.5 mM D-[U-14 C]glucose (1 Ci/well) for 45 min. Glucose incorporation into glycogen was then determined.
Other Procedures-Immunoblotting was performed as described previously (15), and immunoreactivity was detected by horseradish peroxidase-conjugated secondary antibodies and ECL. Protein measurements were by Bradford.

RESULTS
The Responsiveness of Glycogen Synthesis to Insulin Dramatically Increases during Differentiation of 3T3-L1 Cells-3T3-L1 fibroblasts were differentiated into adipocytes by a standard 4-day protocol, and the adipocytes were then maintained in FBS-containing medium for an additional 1 or 8 days. The culturing of the adipocytes in the FBS medium for 7 more days did not change the percentage of cells containing lipid droplets (Ͼ90%), but it did result in increased lipid stores per cell (data not shown). Basal glycogen synthesis was barely detectable in fibroblast cells and was increased only 3-fold by 100 nM insulin treatment (Table I, Fibroblast). After differentiation of the fibroblasts into lipid-containing adipocytes, glycogen synthesis was stimulated 30 -40-fold by insulin, with no change in the basal rate (Table I, Adipo/1). In day 8 adipocytes, basal glycogen synthesis was still comparable with that seen in 3T3-L1 fibroblasts, but insulin caused a nearly 700-fold increase in glycogen accumulation (Table I, Adipo/8).
Regulation of Glycogen Synthase and PP1 Expression and Activities during 3T3-L1 Differentiation-Because of the dramatic changes in insulin-stimulated glycogen synthesis during adipogenesis, we sought to compare the relative expression of different enzymes that might be critical to the process. Indeed, several proteins have been reported to be expressed upon 3T3-L1 adipocyte differentiation, including the insulin receptor (22) and Glut-4 (23), which contribute to the regulation of glycogen synthesis by insulin. The expression of glycogen synthase and PP1 on the glycogen particle was measured during differentiation of 3T3-L1 fibroblasts into adipocytes. Across the 3T3-L1 differentiation protocol, replicate plates were stimulated in the absence and presence of 100 nM insulin. Glycogenenriched pellets were prepared, and glycogen synthase and PP1 protein levels were examined by immunoblotting. As seen in Fig. 1A, glycogen synthase levels in this fraction increased during the differentiation protocol (days 2-6), and continued to increase as the adipocytes were cultured in FBS medium (days 6 -17). Insulin treatment had no effect on the amount of glycogen synthase present in the glycogen-containing fraction at any time. In contrast, there was no change in the amount of glycogen-associated PP1 during fibroblast differentiation or upon continued culturing of the adipocytes (Fig. 1B). At no time did insulin cause any measurable change in glycogen-targeted PP1 levels, indicating that the stimulation of PP1 activity by insulin occurs independently of translocation of the phosphatase to glycogen. It should be noted that since the total protein content of the glycogen-enriched pellet increased by at least 8-fold during differentiation, increases in the total amounts of glycogen synthase and PP1 in the adipocyte samples are significantly understated.
In parallel to the protein measurements, both glycogen synthase and PP1 activities were directly measured on the indicated days. In agreement with Fig. 1A, differentiation of the 3T3-L1 fibroblasts into adipocytes resulted in a dramatic increase in total glucose-6-phosphate (G6P)-dependent glycogen synthase activity ( Fig. 2A, ϩG6P). Insulin activates glycogen synthase through dephosphorylation of the enzyme, resulting in the stimulation of G6P-independent synthase activity (ϪG6P). Very low levels of active synthase were detected in the fibroblasts, which increased slightly in response to insulin ( Fig.  2A, Fibroblast, ϪG6P). In day 1 adipocytes, insulin now caused a greater -fold stimulation and an absolute increase in G6Pindependent synthase activity ( Fig. 2A, Adipo/1, ϪG6P). This trend continued as the adipocytes were cultured for another week, with active synthase being increased 7-fold in response to insulin in the day 8 adipocytes (Fig. 2A). These measurements were adjusted for protein, so the amounts of active and total glycogen synthase activities in the adipocytes are underrepresented approximately 8-fold relative to the fibroblasts.
Since glycogen synthase activation by insulin results from dephosphorylation, PP1 activity during differentiation was also examined. In contrast to observations regarding glycogen synthase, 3T3-L1 fibroblasts contained the highest PP1 basal activity, which was modestly affected by exposure to insulin (Fig. 2B). Differentiation and continued culturing of the 3T3-L1 adipocytes resulted both in a large decrease in PP1 basal activity and a corresponding increase in the stimulation of the phosphatase by insulin (Fig. 2B). In the day 8 adipocytes, insulin now caused a 2-fold increase in PP1 activity. Numerous experiments in 3T3-L1 adipocytes revealed an inverse correlation between basal PP1 activity and its stimulation by insulin (data not shown), suggesting that reducing PP1 basal activity may prime the phosphatase for activation by insulin.
Glycogen-targeted PP1 Is Regulated by DARPP-32 in 3T3-L1 Adipocytes-The significant decrease in basal PP1 specific activity without a corresponding change in protein levels during adipogenesis suggested that a PP1 inhibitor may play an important role in the regulation of phosphatase activity. Since PP1 is known to be inhibited by Inh-1 and DARPP-32, the presence of these proteins in 3T3-L1 adipocytes was examined by immunoblotting. No Inh-1 protein could be detected in 3T3-L1 lysates, although purified Inh-1 protein was recognized by the affinity-purified antibody used (data not shown). However, DARPP-32 was readily detected in 3T3-L1 adipocyte lysates using an affinity-purified mouse monoclonal antibody (Fig. 3A, lane 1). Subsequent fractionation of the PNS fraction revealed that DARPP-32 is largely absent from the cytosol and present in both the plasma membrane and glycogen-enriched fractions (Fig. 3A, lanes 2-4). Similar results were obtained when primary rat adipocytes were fractionated and subjected to anti-DARPP-32 immunoblotting (Fig. 3A, lanes 5-8), indicating that DARPP-32 expression is not a cell line artifact. In bovine parathyroid gland, DARPP-32 has recently been shown to be isoprenylated, resulting in an exclusively particulate localization (24). To examine this possibility, crude particulate fractions were prepared from 3T3-L1 adipocytes by ultracentrifugation. This fraction was resuspended in PP1 homogenization buffer containing salt or detergent and was then recentrifuged. The resulting supernatants and particulate fractions were then examined by anti-DARPP-32 immunoblotting. As seen in Fig. 3B, resuspension of the original pellet in homoge- During differentiation of L1 fibroblasts into adipocytes, replicate plates were incubated in the absence and presence of 100 nM insulin (Ins) for 15 min. 20 g of the glycogen-enriched pellets were separated by SDSpolyacrylamide gel electrophoresis, transferred to nitrocellulose, and probed with an anti-glycogen synthase antibody. Antibody binding was detected with a horseradish peroxidase-conjugated rabbit anti-chicken secondary antibody and ECL. Differentiation of 3T3-L1 fibroblasts started on day 2 (D) and was completed on day 6 (A). Adipocytes were maintained in FBS-containing medium for up to 11 days (day 17). Two replicate plates were not differentiated and were fed calf serum-containing medium in parallel with the differentiating cells (CS). The figure is representative of two independent experiments. B, PP1 immunoblot. Differentiation of 3T3-L1 fibroblasts was initiated on day 0 (D) and finished on day 4 (A). Adipocytes were cultured in FBS-containing medium for another 10 days (days 4 -14). Samples were treated and prepared as in A. Blots were probed with an anti-PP1 antibody, which recognizes ␣, ␤, ␥, and ␦ PP1 isoforms, and detected by ECL. The immunoblot is representative of three independent experiments.

FIG. 2. Glycogen synthase and PP1 activities during 3T3-L1 adipocyte differentiation.
A, glycogen synthase activity. Replicate 10-cm plates of 3T3-L1 fibroblasts and adipocytes cultured for either 1 (Adipo/1) or 8 (Adipo/8) days postdifferentiation were serum-deprived for 3 h. Cells were treated with either buffer (B) or 100 nM insulin (Ins) for 20 min and then were scraped in glycogen synthase homogenization buffer. Glycogen synthase activity in the the PNS fraction was then measured in triplicate, in the absence and presence of 10 mM glucose 6-phosphate (Ϫ/ϩG6P), as described under "Experimental Procedures." Shown are means Ϯ S.D., and results are representative of three independent experiments. B, PP1 activity. Following a 3-h serum deprivation, replicate plates of 3T3-L1 fibroblasts and day 1 or day 8 adipocytes were exposed to either buffer or 100 nM insulin for 20 min. PNS fractions were prepared, and PP1 activity in triplicate was measured at 37°C for 3 min as described under "Experimental Procedures." Results are means Ϯ S.D. and are representative of two separate experiments. nization buffer containing 0 or 500 mM NaCl did not extract any of the DARPP-32 from the particulate fraction (lanes 1-2 and 4 -5). However, the inclusion of 0.1% Triton in the resuspension buffer led to the complete extraction of DARPP-32 from the pellet into the resulting supernatant (lanes 3 and 6). These results indicate that in 3T3-L1 adipocytes, DARPP-32 is most likely integrally associated with membranes, perhaps due to its isoprenylation as described previously (24). 500 mM NaCl completely removed PP1 from the particulate fraction (data not shown), supporting the notion that the association of DARPP-32 with the glycogen pellet and plasma membrane is independent of its binding to PP1.
Levels of DARPP-32 protein in the glycogen-enriched pellet were next measured during the differentiation of 3T3-L1 cells. As seen in Fig. 4, no DARPP-32 was detected in the 3T3-L1 fibroblasts. However, the inhibitor was expressed upon adipocyte differentiation and continued to increase during culturing of the adipocytes in FBS medium. At no time did insulin treatment result in any detectable change in the amount of DARPP-32 present in the glycogen-enriched fraction (Fig. 4). The expression of DARPP-32 correlated with the reduction in PP1 basal specific activity seen during 3T3-L1 differentiation and suggests a possible role for DARPP-32 in the regulation of glycogen-targeted PP1 activity.
The Stimulation of PP1 Activity by Insulin Results in Increased Initial Enzymatic Rates Measured in Vitro-Elucidation of the mechanism by which PP1 is activated by insulin has been hampered by the failure to detect in vitro marked elevation of the phosphatase by the addition of hormone to cells in culture. One possible explanation for the small observed effect might lie in an artificially elevated basal activity, which may arise from the disassociation of inhibitor peptides upon cell lysis. The kinetics of PP1 activity were evaluated in full differentiated 3T3-L1 adipocyte lysates. Extracts from basal adipocytes were rapidly prepared, and PP1 activity was measured in the PNS fraction for indicated times at 37°C (Fig. 5A, basal). During the first 4 min of the assay, PP1 activity was low. After 4 min, phosphatase activity significantly increased, reaching constant activity kinetics. This activity state persisted until substrate consumption exceeded 30 -40% (data not shown).  1 and 4), buffer plus 0.5 M NaCl (lanes 2 and 5), or buffer plus 0.1% Triton X-100 (lanes 3 and 6). Samples were then subjected to ultracentrifugation (100,000 ϫ g, 30 min), and subsequent pellets (lanes 1-3) and supernatants (lanes 4 -6) were analyzed by DARPP-32 immunoblotting. FIG. 4. DARPP-32 is expressed in the glycogen pellet during 3T3-L1 adipocyte differentiation. A DARPP-32 immunoblot is shown. During differentiation of 3T3-L1 fibroblasts into adipocytes, two replicate plates were serum-starved for 3 h and treated in the absence (Ϫ) or presence (ϩ) of 100 nM insulin for 15 min. Glycogen-enriched pellets were prepared as described under "Experimental Procedures." 25 g of protein were resolved by SDS-polyacrylamide gel electrophoresis, transferred to nitrocellulose, and probed with an anti-DARPP-32 monoclonal affinity-purified antibody; immunoreactivity was detected by ECL. Day 0, confluent fibroblasts. Differentiation (D) was initiated on day 2 and completed on day 6 (A). Adipocytes were maintained in FBS-containing medium for up to 11 more days (days 6 -17). Two confluent plates of fibroblast cells were maintained in parallel for 17 days in calf serum-containing medium (CS). The immunoblot is representative of four independent experiments.
FIG. 5. Kinetics of PP1 activity in vitro: insulin stimulation overcomes lag. A, PP1 activity measured at 37°C. Postnuclear supernatants from basal and insulin-stimulated (20 min, 100 nM) 3T3-L1 adipocytes (days 7-12) were prepared. Phosphatase assays were performed at 37°C and terminated at the indicated times. Phosphate released from the substrate was measured as described under "Experimental Procedures," and PP1 activity was calculated. B, PP1 activity measured at 30°C. The phosphatase assay was performed at 30°C in the presence of 3 nM okadaic acid, and total phosphate released was measured at the indicated times. Results are representative of four experiments.
Treatment of the cells with insulin largely overcame the lag in PP1 activity seen in vitro (Fig. 5A, Ins). Thus, the highest stimulation of PP1 activity by insulin was observed during the initial phase of the assay (Fig. 5A). During the linear phase of PP1 activity, however, only 20 -40% stimulation was measured in extracts from insulin-treated cells, consistent with previous reports (25,26). Performing the assays at 30°C resulted in an extension of the basal PP1 activity lag, which was not detected in extracts prepared from insulin-treated cells (Fig. 5B). In contrast, PP1 activity from 3T3-L1 fibroblast lysates exhibited linear kinetics at all time points (data not shown). These results suggest that insulin can increase the initial rate of PP1 activity measured in vitro, by inducing the disassociation of inhibitor peptide, allowing for the full expression of catalytic activity.

DISCUSSION
Insulin action is characterized by the paradoxical regulation of serine/threonine phosphorylation (1). The stimulation of protein kinase cascades play a crucial role in the mitogenic effects of this hormone. However, regulation of glucose-and lipidmetabolizing enzymes by insulin occurs largely through dephosphorylation of these proteins. Although inactivation of upstream kinases may be involved (27), the key enzymes that regulate glycogen synthesis, namely glycogen synthase, glycogen phosphorylase, and phosphorylase kinase, are all excellent substrates for PP1. Further, PP1 has been shown to be specifically activated by insulin, suggesting an important role for this phosphatase in insulin action. Although PP1 is present in nearly all cellular compartments, only a specific, limited number of phosphoproteins are dephosphorylated in response to insulin (1). This paradox raises two important questions: how is PP1 targeted to the relevant substrates, and how are specific pools of PP1 preferentially activated by insulin?
We have utilized the 3T3-L1 cell line to investigate the regulation of glycogen metabolism by insulin. 3T3-L1 fibroblasts are poorly responsive to insulin; however, differentiation of these cells into lipid-containing adipocytes results in a dramatic increase in insulin-mediated glucose utilization and storage. The levels of many proteins are increased during adipogenesis, including key mediators of insulin signaling, such as the insulin receptor (22) and . In this study, we examined the hormonal regulation of glycogen synthesis during differentiation of 3T3-L1 cells. In fibroblasts, insulin only caused a 3-fold increase in glucose accumulation as glycogen; in the fully differentiated day 8 adipocytes, insulin treatment resulted in a 700-fold increase in glycogen synthesis. Correspondingly, the protein and activity levels of glycogen synthase were dramatically increased upon adipogenesis, as was the stimulation of synthase activity in response to insulin. In contrast, the specific activity of PP1 decreased by over 60% during differentiation, with no corresponding change in protein levels. However, this decrease in PP1 specific activity correlated with a significant increase in the -fold activation of PP1 by insulin. These findings suggest that mechanisms may exist that maintain PP1 in a low activity state, allowing the phosphatase to be maximally stimulated by insulin.
The precise events leading to the activation of glycogenlocalized PP1 activity remain unclear. Three mammalian proteins have been identified that target PP1 to glycogen: G M , G L , and PTG. Dent et al. (12) reported that the phosphorylation of site 1 on G M by pp90 rsk in vitro increased the activity of the bound PP1, while phosphorylation of sites 1 and 2 on G M by cAMP-dependent protein kinase resulted in the disassociation of PP1 from G M . This attractive model would explain how insulin, through the stimulation of the MAP kinase/pp90 rsk cascade, could increase PP1 activity, while also explaining how catabolic agents, through activation of adenylate cyclase and cAMP-dependent protein kinase, could reduce PP1 activity at the glycogen particle. However, several groups have since convincingly shown that the MAP kinase/pp90 rsk pathway is neither sufficient nor necessary for the increases in glucose storage or PP1 activity caused by insulin (13)(14)(15)(16)(17). Further, since neither site 1 or 2 of G M is conserved in either PTG or G L , it is impossible to apply the G M phosphorylation model to these proteins. Not surprisingly, PTG and G L are not phosphorylated in vitro by cAMP-dependent protein kinase (9,18). Additionally, PTG, is not phosphorylated in vivo in response to anabolic agents such as insulin or catabolic agents such as isoproterenol (18). Moreover, neither treatment had any effect on the levels of glycogen-associated PP1 in 3T3-L1 or primary rat adipocytes ( Fig. 2B; data not shown; Ref. 18). Therefore, the activation of PP1 targeted to glycogen by PTG occurs independently of PP1 translocation and PTG phosphorylation.
Several lines of evidence have been presented here implicating DARPP-32 as a key regulator of glycogen-targeted PP1 activity in adipocytes. DARPP-32 was not detectable in 3T3-L1 adipocytes, but it was dramatically up-regulated during adipocyte differentiation. This expression of DARPP-32 correlated well with both the decrease in PP1 specific activity measured during differentiation and the increased stimulation of PP1 by insulin. DARPP-32 in 3T3-L1 and primary rat adipocytes is exclusively localized to particulate fractions, including the glycogen pellet. Finally, an assay of PP1 activity in 3T3-L1 adipocyte lysates revealed a lag in PP1 activity that was largely overcome by insulin treatment of the cells. The largest stimulation of PP1 activity was thus detected at the earliest assay time points. PP1 activity from fibroblast cells did not exhibit this in vitro lag, nor was phosphatase activity significantly affected by insulin. One explanation of these results is that PP1 in fully differentiated adipocytes is regulated by DARPP-32 binding, and stimulation of PP1 by insulin occurs through disinhibition of the phosphatase. Although DARPP-32 is considered to be the neuronal homologue of Inh-1, both pig brown fat and bovine adipose tissue have previously been reported to contain DARPP-32 (28,29). DARPP-32 immunoreactivity was also detected in primary rat adipocytes (Fig. 3A), so DARPP-32 is likely to be the 32-kDa inhibitor of phosphorylase phosphatase activity in rat adipose tissue reported by Nemenoff et al. (30). These results suggest that in adipocytes DARPP-32 may play an important role in insulin signaling.
The mechanisms of regulation of glycogen-targeted PP1 by both DARPP-32 and insulin are unclear. The organic PP1 inhibitor microcystin has been used as an affinity matrix to isolate PP1 and a large number of associated proteins from tissues (8,31). Many PP1-binding proteins, therefore, do not appear to obstruct the catalytic site, allowing targeted PP1 to interact not only with substrates but also with physiological inhibitors such as DARPP-32. Thiophosphorylated DARPP-32 could inhibit PP1 bound to recombinant glutathione S-transferase-PTG by 70% without any detectable disassociation of the phosphatase from PTG (18), indicating that a trimeric PTG⅐PP1⅐DARPP-32 complex may be formed. The possible disruption of the PP1/DARPP-32 interaction by insulin could occur in several ways. First, insulin may stimulate the dephosphorylation or prevent the phosphorylation of DARPP-32, most likely through reduction of cAMP levels and/or cAMP-dependent protein kinase activity. Alternatively, insulin stimulation of adipocytes may result in the generation of a soluble second messenger, which could disrupt DARPP-32 binding to PP1 independently of inhibitor dephosphorylation. We have previously shown that PP1 bound to PTG is less sensitive to inhibition by DARPP-32 (18). Conversely, the PP1⅐PTG complex may be more easily disinhibited by insulin treatment, possibly explaining the ability of insulin to specifically activate glycogentargeted PP1, leading to the dephosphorylation of a limited number of substrates. Because of the high off rate and the need to solubilize with detergent, it has thus far not been possible to capture a complex between PP1 and DARPP-32 in vivo. Efforts are under way to explore the effect of insulin on PP1/DARPP-32 interaction.