INTRODUCTION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

GLYCOGEN SYNTHESIS is one of the major metabolic events triggered by exposure of cells to insulin. Activation of glycogen synthase, the rate-limiting enzyme in glycogen storage, is mediated by a complex series of phosphorylation and dephosphorylation events. Glycogen synthase activity exists in two interconvertible forms: synthase D, which is highly phosphorylated and has an absolute dependence on its allosteric activator glucose-6-phosphate, and synthase I, which is less phosphorylated and considered to be the active form in vivo. Diabetes-related abnormalities in glycogen metabolism are attributable to a lower percentage of active glycogen synthase. Whereas it has been well established that glycogen synthase is dephosphorylated (activated) by protein phosphatase 1 (PP1) (7) and phosphorylated (inactivated) by glycogen synthase kinase 3 (GSK3) (36), the precise molecular mechanism by which insulin regulates this complex process is not well understood.

The binding of insulin to its plasma membrane receptor stimulates an intrinsic receptor tyrosine kinase activity, which in turn results in multisite phosphorylation of the cytosolic insulin receptor substrate 1 (IRS1). Tyrosine-phosphorylated IRS1 acts as a docking site for several signaling molecules, including the enzyme phosphatidylinositol 3-kinase (PI3K). PI3K is a heterodimer composed of an 85-kDa regulatory subunit (p85), which binds to other molecules via its src-homology 2 domains, and a 110-kDa catalytic subunit (p110), which phosphorylates the cellular lipids phosphatidylinositol (PI) and its 4' and 4',5' derivatives in the D-3 position of the inositol ring (4, 41). The fungal inhibitor wortmannin has been shown to be a potent and selective inhibitor of PI3K activity in a variety of tissues (18, 23, 25). Wortmannin inhibits the insulin-induced activation of glycogen synthase (29, 33, 35, 44) and the inactivation of GSK3 (6, 40) in a variety of cell types, suggesting that glycogen synthesis is regulated via a PI3K-mediated pathway.

Previous studies have examined the relationship between PI3K and glycogen synthase activation using cell lines or tissues in which glycogen synthesis is not of primary physiological importance, including adipocytes (33, 35, 42), Chinese hamster ovary (CHO) cells (29), and PC12 cells (44). Insulin is a potent regulator of glycogen synthesis in the heart and liver, however, and previous work from our laboratory has demonstrated that treatment of primary cultures of rat hepatocytes and cardiomyocytes with insulin leads to a reproducible increase in glycogen synthase activity (14, 19, 20). Cardiomyocytes isolated from normal rats and maintained in chemically defined medium show insulin-stimulated synthase activation that parallels that seen in the working heart. In diabetes, however, insulin activation of glycogen synthase is severely impaired. This loss of acute insulin activation has been shown to correlate with altered synthase phosphorylation and loss of PP1 activity in both liver and primary culture hepatocytes isolated from alloxan-diabetic rats. Tissues from perfused working diabetic hearts, as well as cardiomyocytes isolated from diabetic animals, display this same defect in synthase activation.

Results of the present study demonstrate that, in primary culture cardiomyocytes, the activation by insulin of glycogen synthase and PP1 is abolished by the PI3K inhibitor wortmannin. This lack of PP1 activation mimics that seen in cardiomyocytes isolated from diabetic animals. Furthermore, diabetic cardiomyocytes exhibit a significant increase in insulin-stimulated PI3K activity without an alteration in PI3K expression. These novel findings suggest that PI3K plays a key role in the acute regulation of glycogen metabolism in cardiomyocytes and implicate defects in signaling between PI3K and PP1 in the altered glycogen metabolism associated with insulin-dependent diabetes.

	MATERIALS AND METHODS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Animals. Cells were isolated from fed 8-wk-old (200-250 g) male Sprague-Dawley rats (Taconic Farms, Germantown, NY). Diabetes was induced in overnight-fasted rats by tail vein injection of alloxan (47 kg/mg) and was diagnosed 3-5 days after injection by positive urinary glucose (>5,000 mg/dl) and urinary ketones using Ames Keto-Diastix.

Cardiomyocyte isolation and culture. Cardiomyocytes were prepared and placed in primary culture as previously described (43). Briefly, hearts were perfused through the aorta in retrograde fashion with Krebs-Henseleit buffer containing 5 mM glucose and 2.5 mM calcium. Hearts were then perfused briefly with calcium-free Krebs-Henseleit buffer to stop contraction, followed by a final perfusion in buffer containing 50 µM calcium, 0.1% bovine serine albumin (BSA), 312 U/ml hyaluronidase, and 0.1% collagenase. Hearts were removed from the perfusion apparatus, trimmed of aortic and atrial tissue, bisected, and further digested by shaking to dissociate tissue. The resulting cell solution was filtered through nylon mesh, allowed to pellet by gravity, and rinsed by settling through 1% BSA. Cells were placed in minimal essential medium (MEM; GIBCO, Grand Island, NY) containing Earl's salts, 0.125% BSA, 5 mM creatinine, and 10 mM HEPES before they were seeded onto laminin-coated dishes. Cells were maintained at 37°C in a humidified incubator (95% air-5% CO₂). All cells were harvested on the day of plating after a 2-h equilibration period. For each experiment, cells from four hearts were pooled to provide a sufficient number of cells for all experiments.

Treatment of cells with insulin and PI3K inhibitors. Cells were rinsed and refed with insulin-free medium for 1 h. After a 10 min preincubation period in the presence or absence of wortmannin (1 µM), cells were stimulated with 10⁷ M insulin (Lilly, Indianapolis, IN) for 10 min at 37°C. In some experiments, cells were pretreated with the PI3K inhibitor LY-294002 [2-(morpholinyl)-8-phenyl-4H-1-benzopyran-4-one; 100 µM]. Because the final concentration of DMSO in dishes treated with wortmannin or LY-294002 was 0.1%, the same concentration of DMSO was added to all dishes. After treatment, cells were rinsed with ice-cold phosphate-buffered saline, frozen by floating on liquid nitrogen, and stored at 80°C until use. In experiments in which cells were used for PI3K measurements, the wash buffer included 200 µM Na₃VO₄.

Glycogen synthase assays. Cells were harvested on ice in 5 mM Tris, pH 7.5, 5 mM EDTA, and 20 mM NaF and homogenized using a Brinkman Polytron. Glycogen synthase activity was assayed in the presence or absence of glucose-6-phosphate (G6P) according to the method of Thomas et al. (38). Briefly, enzyme activity was measured as the incorporation of glucose from [¹⁴C]uridine diphosphoglucose (UDPG) into glycogen by using rabbit liver glycogen as substrate. The reaction was run for 60 min at 30°C and stopped by aliquoting the incubation mixture onto filter paper. [¹⁴C]glycogen bound to the filter paper was quantitated by scintillation counting. Synthase activity is expressed as the ratio of G6P-independent activity to total activity measured in the presence of G6P (%I). Insulin-stimulated synthase activation is expressed as %I measured in the presence of insulin. Each experiment represents a single data point (n = 1).

Glycogen synthase phosphatase assays. Cells were harvested in the same manner as described for synthase assays. Synthase phosphatase activity was measured as the conversion of synthase D to the I form. Aliquots of cell homogenates were incubated at 30°C for 10 min with exogenous synthase D from purified rabbit skeletal muscle. The phosphatase assay was terminated by transferring an aliquot of the homogenate into a glycogen synthase mixture, and synthase activity was measured after a 60-min incubation at 30°C. Because the synthase assay buffer contains potassium fluoride (21), an inhibitor of synthase phosphatase, phosphatase activation does not continue during the synthase assay. Phosphatase activity is expressed as nanomoles of glucose incorporated into glycogen from UDPG per minute per milligram of protein. Each experiment represents a single data point (n = 1).

PI3K activity assays. PI3K activity was assayed in cell homogenates using thin-layer chromatography (TLC) as previously described (27) with some modifications. Cells were scraped into ice-cold homogenization buffer (20 mM Tris, pH 7.4, 145 mM NaCl, 5 mM EDTA, 1 mM NaF, 1 mM Na₃VO₄, 1 mM phenylmethylsulfonyl fluoride, 10 µg/ml aprotinin, 10 µg/ml leupeptin, and 1 µg/ml pepstatin) and sonicated on ice. PI3K activity in 10,000-g supernatants was assayed by measuring the in vitro phosphorylation of the substrate PI (Avanti Polar Lipids, Alabaster, AL). Each reaction mixture consisted of 40 µg cell protein, 20 µg PI, 10 mM MgCl₂, 50 µM ATP, and 30 µCi [-³²P]ATP in TNE buffer (10 mM Tris, pH 7.4, 150 mM NaCl, and 5 mM EDTA). The kinase reaction was run for 15 min at room temperature and was stopped by the addition of 40 µl of 8 N HCl and 160 µl of CHCl₃/CH₂OH (1:1). After the organic and aqueous phases were separated by centrifugation, the organic phase was evaporated to dryness, dissolved in 40 µl of CHCl₃, and spotted onto silica gel plates (Aldrich, Milwaukee, WI) pretreated with oxalate. Phosphoinositides were resolved in 1-propanol:2 M acetic acid (65:35) for 5 h, and reaction products were visualized by autoradiography. The radioactivity in spots that comigrated with a PI 4-monophosphate standard was quantitated by scintillation counting of plate scrapings.

Quantification of PI 3-monophosphate formation. The TLC procedure does not permit the distinction between reaction products of PI3K and PI 4-kinase (PI4K). To overcome this problem, PI3K activity was assayed in cells that had undergone the following treatments: 1) control (no insulin or wortmannin added to dishes), 2) control + wortmannin, 3) insulin, and 4) insulin + wortmannin. The incorporation of [-³²P]ATP into PI-monophosphate in samples treated with wortmannin should represent PI4K activity only, because wortmannin is a selective inhibitor of PI3K. Therefore, the formation of PI 3-monophosphate [PI(3)P] was determined by subtracting ³²P counts per minute (cpm) in control + wortmannin samples (PI4K activity) from ³²P cpm in control samples (total PI kinase activity); likewise, insulin-stimulated PI(3)P formation was determined by subtracting ³²P cpm in insulin + wortmannin samples (PI4K activity) from ³²P cpm in insulin samples (total PI kinase activity). Each condition represents pooled homogenates from two dishes of cells.

High-performance liquid chromatography of phospholipids. To confirm that TLC gave a valid estimation of PI3K activity, high-performance liquid chromatography (HPLC) was used to analyze phospholipids from cells labeled in vivo with ³²P. Normal and diabetic cardiomyocytes plated in 10-cm dishes were rinsed with phosphate-free MEM (Sigma) and incubated for 20 min at 37°C. This medium was then removed and replaced with 4 ml of fresh phosphate-free MEM containing 0.125% BSA and 2 mCi ³²P_i (HCl free). After cells were labeled for 2 h at 37°C, they were preincubated with or without wortmannin (1 µM) for 10 min and stimulated with insulin (10⁷ M) for 10 min at 37°C. Cells were rinsed with Tris-buffered saline (TBS, pH 7.4) containing 200 µM Na₃VO₄, and cellular lipids were extracted as previously described (24). Phospholipids were concentrated in vacuo, resuspended in 60 µl of CHCl₃, and separated by TLC. After phosphoinositides were extracted from TLC plates, they were chemically deacylated with methylamine and subjected to anion-exchange HPLC using a Partisphere SAX column (Whatman, Clifton, NJ) according to the method of Serunian et al. (32). A preparation of PI3K purified from bovine brain was used to produce PI(3)P in vitro for verification of monophosphate elution time.

Electrophoresis and immunoblotting. Insulin signaling proteins were detected in whole cell extracts by immunoblotting with commercially available antibodies. Cultured cardiomyocytes were harvested in homogenization buffer, and whole heart homogenates were prepared as previously described (10) with modifications. Proteins were resolved by SDS-PAGE using 10% gels and were transferred to Immobilon membranes (Millipore, Bedford, MA). After blots were blocked in 5% nonfat dry milk, they were incubated with anti-p85, washed in TBS-Tween 20, pH 7.4, and incubated with sheep anti-rabbit IgG-horseradish peroxidase. After extensive washing, immunoreactive bands were visualized using enhanced chemiluminescence. An AlphaImager 2000 system was used for quantification of bands (Alpha Innotech, San Leandro, CA).

Statistical analysis. Statistical significance was determined using either paired t-tests for comparison of basal versus stimulated conditions or two-way ANOVA for comparisons of cell types across treatments (Statistica software). Tukey's post hoc test was used when main effects were significant.

Chemicals and antibodies. Anti-PI3K (p85 subunit) antibody was purchased from Transduction Laboratories (Lexington, KY). Sheep anti-rabbit IgG conjugated to horseradish peroxidase was obtained from Boehringer Mannheim (Indianapolis, IN). Chemiluminescence reagents used in immunoblot procedures were from KPL (Gaithersburg, MD). The purified PI3K preparation used for in vitro PI3K assays was a generous gift from Dr. Chris Vlahos (Lilly, Indianapolis, IN). Wortmannin (Sigma, St. Louis, MO) and LY-294002 (Biomol, Plymouth Meeting, PA) were dissolved in DMSO, stored at 20°C in the dark, and diluted with distilled water just before use. [-³²P]ATP and ³²P_i were purchased from NEN (Boston, MA). The Bio-Rad assay (Hercules, CA) was used to determine total cell protein in all experiments.

	RESULTS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Effect of wortmannin on glycogen synthase and synthase phosphatase activity in normal cardiomyocytes. Treatment of normal cardiomyocytes with insulin resulted in a significant increase in glycogen synthase activity (P < 0.001) to a level nearly threefold higher than that of unstimulated controls (Fig. 1). However, this insulin-induced activation of glycogen synthase was almost completely abolished by pretreatment of cells with wortmannin. Glycogen synthase activity in unstimulated cells pretreated with wortmannin, LY-294002, or DMSO, the vehicle for both inhibitors, was not significantly different from that of unstimulated controls (data not shown). Synthase phosphatase activity was also significantly increased in cardiomyocytes stimulated with insulin (P < 0.005), whereas wortmannin or LY-294002 pretreatment blocked this effect (Fig. 2). Basal phosphatase activity was not affected by PI3K inhibitors (Fig. 2) or DMSO (data not shown).

View larger version (33K): [in this window] [in a new window]   Fig. 1.   Wortmannin inhibits insulin-induced activation of glycogen synthase in normal cardiomyocytes. Cardiomyocytes were insulin-starved for 1 h and incubated for 10 min in presence or absence of inhibitor. Cells were then stimulated for 10 min with insulin, as noted. Homogenates were assayed for glycogen synthase activity as described in MATERIALS AND METHODS. Each bar represents mean ± SE of data from 5 cardiomyocyte preparations. %I, ratio of glucose-6-phosphate (G6P)-independent activity to total activity measured in presence of G6P; N0, normal cardiomyocytes, no inhibitor, no insulin; NI, normal cardiomyocytes, no inhibitor, 107 M insulin; NW, normal cardiomyocytes, 1 µM wortmannin, 107 M insulin. * Significantly different from N0 and NW.

View larger version (34K): [in this window] [in a new window]   Fig. 2.   Insulin does not activate phosphatase in diabetic cardiomyocytes or normal cardiomyocytes pretreated with phosphatidylinositol 3-kinase (PI3K) inhibitors. Cardiomyocytes isolated from normal or diabetic rats were insulin-starved for 1 h and incubated for 10 min in presence or absence of inhibitor. Cells were then incubated with 107 M insulin, as noted. Homogenates were assayed for phosphatase activity as described in MATERIALS AND METHODS. Each bar represents mean ± SE of data from 42 normal and 12 diabetic cardiomyocyte preparations. N0, normal cardiomyocytes, no inhibitor; NW, normal cardiomyocytes, 1 µM wortmannin; NL, normal cardiomyocytes, 100 µM LY-294002; D0, diabetic cardiomyocytes, no inhibitor. * Significantly different from basal N0. ** Significantly different from N0 stimulated with insulin.

Effect of diabetes on PP1 activity in cardiomyocytes. Insulin-stimulated phosphatase activity was effectively absent in diabetic cardiomyocytes. As shown in Fig. 2, phosphatase activity in stimulated diabetic cells was not significantly different from that observed in unstimulated diabetic cells, unstimulated normal cells, or cells pretreated with inhibitors of PI3K.

Effect of diabetes on PI3K activity in cardiomyocytes. PI3K activity was measured as the incorporation of ³²P into PI(3)P. As shown in Fig. 3A, insulin-stimulated PI3K activity was significantly elevated in diabetic cardiomyocytes compared with that in normal cells (P < 0.02). Whereas PI3K activity in normal cells was increased by insulin treatment to levels 77% above that of unstimulated controls, this activity was increased to 300% of levels in unstimulated controls in diabetic cells. However, basal PI3K activity in diabetic cells was not significantly different from that observed in normal cells (Fig. 3A). Insulin stimulation had no effect on PI4K activity, and, in fact, PI4K activity was similar in normal and diabetic cells under all conditions (Fig. 3B). In both normal and diabetic cells, basal PI3K activity (wortmannin sensitive) was found to be ~17% of total PI kinase activity (measured in the absence of wortmannin), whereas insulin-stimulated PI3K activity was ~25 and 35% of total PI kinase activity in normal and diabetic cells, respectively. These values were very consistent from experiment to experiment, reinforcing the validity of our method for quantifying PI3K. Two additional pieces of evidence are also important in this regard: 1) wortmannin (1 µM) added to cell homogenates in vitro was effective in reducing insulin-stimulated PI3K activity to near-control levels, as measured by TLC (data not shown); and 2) when cellular lipids were analyzed by HPLC, wortmannin (1 µM) reduced PI3K activity to negligible levels, while having no effect on PI4K activity (Fig. 4). It should also be noted that insulin-stimulated PI3K activity as measured by HPLC was higher in diabetic cardiomyocytes compared with that in normal cells (data not shown).

View larger version (52K): [in this window] [in a new window]   Fig. 3.   Insulin-stimulated PI3K activity is increased in diabetic cardiomyocytes. Cardiomyocytes isolated from normal and diabetic rats were insulin-starved for 1 h and incubated for 10 min with 1 µM wortmannin. Cells were stimulated for 10 min with 107 M insulin, and homogenates were assayed for PI3K activity by TLC as described in MATERIALS AND METHODS. A: PI3K activity was measured as wortmannin-sensitive incorporation of [-32P]ATP into PI 3-monophosphate. B: PI 4-kinase activity was determined by correcting for wortmannin-sensitive activity. Each bar represents mean ± SE (in counts/min) of data from 6 normal and 8 diabetic cardiomyocyte preparations. * Significantly different from basal cells. ** Significantly different from stimulated normal cells.

View larger version (28K): [in this window] [in a new window]   Fig. 4.   Determination of PI3K activity by anion-exchange HPLC. Cardiomyocytes isolated from normal and diabetic rats were insulin-starved for 1 h and labeled for 2 h with 2 mCi of 32Pi. Cells were incubated for 10 min in presence or absence of 1 µM wortmannin and stimulated for 10 min with 107 M insulin. Cellular lipids were deacylated, applied to a Partisphere SAX column, and eluted with NH4H2PO4 as described in MATERIALS AND METHODS. Representative HPLC chromatograms of 32P-labeled phosphoinositides from diabetic cells treated in presence (A) or absence (B) of wortmannin are shown. PI3P, PI 3-monophosphate; PI4P, PI 4-monophosphate; cpm, counts/min.

Effect of diabetes on expression of insulin-signaling proteins. The expression of PI3K (p85 subunit) in normal and diabetic cardiomyocytes was determined by SDS-PAGE and immunoblotting. Despite the diabetes-induced increase in PI3K activity, there were no differences in levels of p85 expression between normal and diabetic cardiomyocytes, or normal and diabetic heart tissue, as judged by densitometry (Fig. 5).

View larger version (6K): [in this window] [in a new window]   Fig. 5.   Immunologic detection of PI3K in normal and diabetic cardiomyocytes. Cardiomyocytes or heart tissue samples isolated from normal and diabetic rats were homogenized and analyzed by Western blot using an anti-PI3K (p85 subunit) antibody. Immunoreactive bands were detected by autoradiography. Blot is representative of 4 separate experiments. NC, normal cultured cardiomyocytes; DC, diabetic cultured cardiomyocytes; NT, normal heart tissue; DT, diabetic heart tissue.

	DISCUSSION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

The inability of insulin to stimulate glycogen synthesis is a hallmark of diabetes. The enzyme glycogen synthase is activated in a hierarchical manner by multisite dephosphorylation, but the phosphate released in response to insulin is mainly removed from a tryptic peptide containing three serine residues (sites 3a, 3b, and 3c), which are phosphorylated by the enzyme GSK3 and dephosphorylated by the glycogen-bound form of protein phosphatase 1 (PP1G) (7, 36). Therefore, the activation of glycogen synthase by insulin is mediated primarily by the activation of PP1 and the deactivation of GSK3. Because phosphatase activity in this investigation was measured as the activation of an exogenous glycogen synthase substrate, the activity in question is clearly that of PP1G. Although numerous studies on glycogen synthase regulation have been conducted using whole animals or perfused tissues, the advantages of studying this complex process under the tightly controlled conditions afforded by a primary culture system are clear.

Alloxan- or streptozotocin-induced diabetes in rats induces resistance to the effects of insulin in multiple tissues. Diabetes has been shown to result in a loss of activity in PP1G and a subsequent impairment of glycogen synthase activation (21). This investigation demonstrates an absence of insulin-stimulated PP1 activity in diabetic cardiomyocytes as well as an inhibition of insulin-stimulated glycogen synthase activation and PP1 activation in normal cardiomyocytes treated with wortmannin and LY-294002. This effect of LY-294002, a PI3K inhibitor that is structurally distinct from wortmannin (39), further supports the hypothesis that PI3K is involved in the regulation of glycogen synthesis and shows a direct effect of PI3K on phosphatase activation. These findings are in agreement with recent reports in which wortmannin was found to abolish synthase activation in PC12 cells (44), CHO cells (29), rat adipocytes (33, 35, 42), and mouse soleus muscle (18). It should also be noted that previous studies have shown a lack of inactivation of GSK3 by insulin in the presence of wortmannin (6, 40). This observation alone cannot explain the loss of synthase activation by insulin, however, because the primary role of GSK3 is to inactivate synthase.

Results presented here now demonstrate an involvement of PI3K in PP1 regulation. It has been reported that wortmannin has no effect on PP1G activation in isolated adipocytes (1); however, it is likely that PI3K regulates PP1G through multiple pathways that are tissue specific. In fact, an apparent liver-specific regulatory subunit of glycogen-associated PP1 has recently been identified (22). This subunit differs significantly from that of skeletal muscle (9), suggesting that the regulation of glycogen-associated PP1 may be remarkably different in muscle compared with liver. Interestingly, Saltiel and colleagues (26) have recently cloned a novel glycogen-targeting subunit of PP1, named PTG for protein targeting to glycogen, which is expressed predominantly in insulin-sensitive tissues. Current speculation is that PTG acts as a molecular scaffold, assembling glycogen synthase, PP1, and other metabolic enzymes onto the glycogen molecule. PI3K may mediate this assembly through subcellular protein trafficking and/or covalent modification of proteins by phosphorylation. Furthermore, it is conceivable that the D-3 phosphorylated lipid products of PI3K create a localized negative charge within the cell, thus facilitating the assembly and activation of critical regulatory enzymes.

Although the precise physiological function of PI3K is unknown, the D-3 phosphorylated lipids are thought to act as second messengers distinct from those in the classic phosphatidylinositol pathway. There is substantial evidence to suggest that PI3K mediates intracellular events such as protein sorting, vesicular trafficking, and cytoskeletal reorganization (11, 16). Therefore, the role of PI3K in PP1 activation may be directly related to its involvement in vesicular trafficking, because PP1 has been shown to be associated with microsomal membranes as well as with glycogen (12, 30). Interestingly, the yeast homolog (Vps34) of PI3K regulates vesicle budding and sorting processes (31). This Vps34 PI3K and a serine/threonine protein kinase (Vps15) function together as components of a membrane-associated signal transduction complex in yeast (34). Along these lines, it is conceivable that, in mammalian cells, PI3K regulates PP1 activity through the coordination of a protein and lipid phosphorylation activity. For example, PI3K may be responsible for the targeting of PP1 to glycogen, whereas a PI3K-associated serine/threonine kinase mediates the phosphorylation state (and thus activation) of PP1. PI3K has recently been shown to activate the downstream protein kinase B (Akt-1), a serine/threonine kinase with strong sequence homology to the catalytic domains of cAMP-dependent protein kinase and protein kinase C. Protein kinase B appears to play a role in promoting insulin-stimulated glucose transport in adipocytes (37).

The need to strictly coordinate glucose uptake and glycogen synthesis gives rise to the hypothesis that, in tissues in which glucose uptake is insulin sensitive, the role of PI3K in glycogen metabolism is directly related to its role in glucose uptake. PI3K has been shown to act as an intermediary molecule facilitating insulin-stimulated glucose transport in a variety of cell types. This enhanced glucose transport in target tissues is due to a wortmannin-sensitive translocation of GLUT-4 transporters from an intracellular vesicle pool to the plasma membrane (5, 13, 18). In fact, PI3K has been localized to GLUT-4-containing vesicles. Because GLUT-4 protein levels and transporter translocation are known to be downregulated in response to streptozotocin-induced diabetes (15), the 300% increase in PI3K activity we observed in diabetic cardiomyocytes may represent a compensatory mechanism to permit glucose uptake in the presence of insulin, despite a reduced pool of transporters. In other words, this hyperstimulation of PI3K may serve to prime the glucose transport machinery in the event that insulin levels are normalized. Our primary culture cardiomyocytes have been shown to express the GLUT-4 protein, and studies are currently underway in our laboratory to examine the relationship between PI3K activity and glucose transport in normal and diabetic cells.

The hyperstimulation of PI3K bears an interesting resemblance to the hyperstimulation of glycogen phosphorylase that is also observed in diabetic cardiomyocytes. In cardiomyocytes isolated from alloxan-diabetic rats, phosphorylase exhibits an acute hypersensitivity to epinephrine stimulation, although this response is normalized after 24 h in culture (3). It is intriguing to consider the possibility that the hyperstimulation of PI3K by insulin and the hyperstimulation of glycogen phosphorylase by epinephrine share a similar mechanism, perhaps a diabetes-related unmasking of effectors that are held in check in normal cells.

The importance of PI3K in glycogen synthase regulation has recently been questioned (29) due to the observation that insulin-stimulated glycogen synthase activity is unaltered in a CHO cell line overexpressing a dominant negative mutant of PI3K. The use of two structurally unique inhibitors of PI3K in the present study strongly suggest that PI3K is involved in synthase activation in the heart and apparently mediates the activation of PP1 by insulin. Clearly, additional studies are needed to elucidate the molecular mechanism by which PI3K regulates glycogen metabolism and to identify diabetes-induced derangements in insulin signaling pathways.

Surprisingly few studies have examined the effect of diabetes on PI3K activation. Whereas it has been reported (2, 28) that insulin-stimulated PI3K activity and PI3K levels in the liver and muscle of streptozotocin-diabetic rats are similar to normal controls, another study (10) has shown that PI3K levels are higher in the livers of diabetic rats than in controls. In this investigation, PI3K protein levels (p85 subunit) were similar in normal and diabetic cardiomyocytes, despite a diabetes-related elevation in PI3K activity after insulin stimulation. However, preliminary studies in our laboratory have shown that insulin receptor substrate 2 (IRS2) protein and mRNA levels are higher in the diabetic versus the normal heart, suggesting that PI3K hyperstimulation may be due to a preferential interaction of the enzyme with the IRS2 docking protein. PI3K is also known to possess an intrinsic serine kinase activity that phosphorylates the p85 subunit and subsequently downregulates the lipid kinase activity of the p110 catalytic subunit (8, 17). It will be interesting to determine whether the hyperactivation of PI3K in diabetic cardiomyocytes is associated with a decrease in this autophosphorylation of p85. A diabetes-induced alteration in PI3K activity may indeed control the location, substrate specificity, and activity of PP1.