Glucose 6-Phosphate Produced by Glucokinase, but Not Hexokinase I, Promotes the Activation of Hepatic Glycogen Synthase*

In a previous study (O’Doherty, R. M., Lehman, D. L., Seoane, J., Gómez-Foix, A. M., Guinovart, J. J., and Newgard, C.B. (1996) J. Biol. Chem. 271, 20524–20530), we demonstrated that adenovirus-mediated overexpression of glucokinase but not hexokinase I has a potent enhancing effect on glycogen synthesis in primary hepatocytes. In an effort to understand the underlying mechanism of this differential effect of the two hexokinase isoforms, we have investigated changes in key intracellular metabolites and the activation state of glycogen synthase in cells treated with recombinant adenoviruses expressing the liver isoform of glucokinase (AdCMV-GKL) or hexokinase I (AdCMV-HKI). Glucose 6-phosphate (Glu-6-P) levels are elevated from approximately 1.5 nmol/mg protein to 8–10 nmol/mg protein in both AdCMV-GKLand AdCMV-HKI-treated hepatocytes as glucose is raised from 1 to 5 mM, levels four times higher than those in untreated cells. In AdCMVGKL-treated cells, Glu-6-P continues to accumulate at glucose levels greater than 5 mM, reaching a maximum of 120 nmol/mg protein in cells incubated at 25 mM glucose, a value 10 and 50 times greater than the maximal levels achieved in AdCMV-HKI-treated and untreated cells, respectively. In parallel with the changes observed in Glu-6-P levels, increases in UDP-Glc in AdCMV-HKIand AdCMV-GKL-treated cells were most pronounced at low (1–5 mM) and high (25 mM) glucose levels, respectively. Despite the significant increases in Glu-6-P and UDP-Glc achieved in AdCMV-HKI-treated cells, only AdCMV-GKL-treated cells exhibited increases in glycogen synthase activity ratio and translocation of the enzyme from a soluble to a particulate form relative to untreated control cells. We conclude that Glu-6-P produced by overexpressed glucokinase is glycogenic because it effectively promotes activation of glycogen synthase. Glu-6-P produced by overexpressed hexokinase, in contrast, appears to be unable to exert the same regulatory effects, probably due to the different subcellular distribution of the two glucose-phosphorylating enzymes.

In mammals in the postabsorptive state, glycogen synthesis is potently activated in the liver in response to increased circulating glucose levels. The mechanism by which this occurs is not a settled matter. It was originally suggested that unmetabolized glucose can indirectly activate glycogen synthase (GS), 1 the key enzyme in the control of glycogen synthesis, by relieving the inhibitory effect of phosphorylase a on the protein phosphatases (phosphatases 1 and 2A) responsible for dephosphorylation of, and hence activation of, GS (1). Several lines of evidence, however, argue that free glucose cannot be solely responsible for the activation of glycogen synthesis. First, activation of glycogen synthase and accumulation of phosphorylase a can occur simultaneously in liver cells (2,3). Second, glucose phosphorylation is required to activate glycogen synthase but not to inactivate phosphorylase a (4 -7). Finally, increases in the glycogen synthase activation state are proportional to the intracellular glucose 6-phosphate (Glu-6-P) level (7,8). These observations suggest that glucose phosphorylation is a key step in the activation of glycogen synthesis.
There are several mechanisms by which Glu-6-P can regulate activation of glycogen formation. Increases in intracellular Glu-6-P lead to the allosteric activation of GS, this effect being reversible when Glu-6-P returns to basal levels (9). Additionally, Glu-6-P promotes the covalent activation of glycogen synthase, possibly by inducing a conformational change that favors the dephosphorylation of the enzyme by phosphatases 1 and 2A (10). Finally, increases in Glu-6-P levels trigger translocation of GS between a supernatant fraction and a pellet fraction, a process that appears to play an important role in the regulation of glycogen synthesis (11,12).
Glucose is phosphorylated to Glu-6-P in mammalian cells by members of the hexokinase gene family. Hepatocytes contain primarily glucokinase (GK or HKIV), and small amounts of hexokinase I (HKI) (13). GK differs from HKI, HKII, and HKIII in that it is approximately half as large, is not allosterically inhibited by Glu-6-P, and has a much higher S 0.5 for glucose (8 mM versus 50 -100 M) (14). In a recent study, we showed that overexpression of GK in hepatocytes by adenovirus-mediated gene transfer causes profound enhancement of glycogen synthesis, with the major effect occurring at glucose concentrations in excess of 5 mM, whereas overexpression of HKI has no effect at any concentration of the sugar (15). The current study was undertaken to elucidate the mechanism of this surprising differential effect of the two glucose-phosphorylating enzymes, with particular emphasis on their impact on Glu-6-P levels and the activation state of glycogen synthase.

MATERIALS AND METHODS
Preparation of Recombinant Adenoviruses-Recombinant adenoviruses containing the cDNAs encoding rat hexokinase I (AdCMV-HKI) or * This work was supported by Grant DGICYT91/0276 from the Spanish government (to J. J. G.), Grant 1P50H2598801 from the National Institutes of Health (to C. B. N.), an FPI fellowship from the Spanish government (to J. S.), and an American Physiological Society/Genentech Inc. fellowship (to R. M. O.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
rat liver glucokinase (AdCMV-GKL) were prepared as described previously (16,17). Hepatocyte Isolation, Culture, and Treatment with Recombinant Adenovirus-Hepatocytes were isolated from 24-h fasted male Wistar rats (180 -225 g) by collagenase perfusion as described (18). Cells were suspended in Dulbecco's modified Eagle's medium (Whittaker) supplemented with 10 mM glucose, 10% fetal bovine serum (Whittaker), 100 nM insulin (Boehringer Mannheim), and 100 nM dexamethasone (Sigma) and seeded onto plastic plates of 60-mm diameter treated with 0.1% gelatin (Sigma) at a final density of 6 ϫ 10 4 cells/cm 2 . After cell attachment (5 h at 37°C), hepatocytes were treated with stocks of AdCMV-GKL or AdCMV-HKI at a multiplicity of infection of 10 for 2 h at 37°C. After treatment with the virus, the medium was replaced with Dulbecco's modified Eagle's medium supplemented with 1 mM glucose, 10 nM dexamethasone, and 10 nM insulin. The medium was changed after 16 h, and cells were incubated for a further period of 24 -26 h in the same medium. Following this incubation period (a total of 40 -42 h after viral treatment to allow full expression of the adenovirus-transferred genes) experimental manipulations were performed as detailed in the text and figure Legends. At the end of each experimental manipulation, cell monolayers were washed in phosphate-buffered saline (Whittaker) and frozen in liquid N 2 until analysis.
Enzyme Activity Assays-To measure enzyme activities, 100 l of homogenization buffer consisting of 10 mM Tris-HCl (pH 7.0), 150 mM KF, 15 mM EDTA, 15 mM 2-mercaptoethanol, 10 g/ml leupeptin, 1 mM benzamidine, and 1 mM phenylmethylsulfonyl fluoride was added to frozen plates containing the cell monolayers, and cells were collected with a plastic scraper. Cell bursting was caused by thawing and checked under light microscopy for completeness. Homogenates were collected in Eppendorf tubes and centrifuged at 10,000 ϫ g for 15 min at 4°C, and supernatants and/or pellets were used for determinations. Protein concentration was measured as described by Bradford (19) using the Bio-Rad assay reagent. Glucose-phosphorylating activity was measured spectrophotometrically in 10,000 ϫ g supernatant fractions of hepatocyte extracts at 37°C in the presence of 100 mM glucose as described (20). Glucose phosphorylation was also assayed in whole cell extracts by a previously described radiometric assay (20). Glycogen synthase activity was measured in the presence or absence of 6.6 mM Glu-6-P as described (21) in both supernatant and pellet fractions. The activity measured in the absence of Glu-6-P represents the active form of the enzyme (I or a form), whereas the activity tested in the presence of 6.6 mM Glu-6-P is a measure of total activity. The ratio of these two activities is known as the glycogen synthase activity ratio and is an estimate of the degree of activation of the enzyme.
Metabolite Determinations-Glycogen content was measured by scraping cells into 30% KOH, boiling the extract for 15 min, and centrifuging at 5,000 ϫ g for 15 min. Glycogen in the cleared supernatants was measured as described (22). The intracellular concentrations of Glu-6-P and UDP-Glc were measured by previously described spectrophotometric assays (23,24). L-Lactate levels in the incubation medium were measured as described (25).

GK and HKI Overexpression in Rat
Hepatocytes-Demonstration of the efficacy of the AdCMV-HKI and AdCMV-GKL recombinant adenoviruses for overexpression of glucose-phosphorylating enzymes has been documented previously (15). In the current series of studies, glucose-phosphorylating capacity in untreated hepatocytes was 13.5 Ϯ 0.8 mol/min/g protein when assayed by the spectrophotometric method at 100 mM glucose and 2.3 Ϯ 0.2 mol/min/g protein with 20 mM glucose when assayed by the radiometric assay. The reasons for the difference in sensitivity of the two assays has been detailed in a previous article (15). When assayed by the radiometric method, the glucose phosphorylation capacity was found to be increased by 9-and 8-fold at 20 mM glucose and by 10-and 23-fold in 3 mM glucose in extracts from AdCMV-GKL-and AdCMV-HKI-treated hepatocytes, respectively, relative to extracts from untreated cells assayed at the same glucose concentrations. These values are comparable to those previously reported (15).
Glu-6-P and UDP-Glc Accumulation in AdCMV-GKL-and AdCMV-HKI-treated Hepatocytes-Our previous article demonstrated that overexpression of GK but not HKI increases glycogen synthesis in hepatocytes (15). To address potential mechanisms underlying this observation, Glu-6-P and UDP-Glc levels were measured in untreated, AdCMV-GKL-treated, and AdCMV-HKI-treated hepatocytes. Untreated and transduced cells preincubated for 42 h at 1 mM glucose and then transferred to media containing variable glucose concentrations in the range of 1-25 mM for 2 h had Glu-6-P levels that increased in a glucose concentration-dependent manner (Fig.  1). In both AdCMV-HKI-and AdCMV-GKL-treated cells Glu-6-P was significantly increased over untreated control cells at all glucose levels (Fig. 1A). In AdCMV-HKI-treated cells, however, the greatest increases in Glu-6-P occurred between 1 and 5 mM glucose, with smaller increases at higher glucose levels (Fig. 1B). In AdCMV-GKL-treated cells, there was a sigmoidal increase in Glu-6-P accumulation with an inflection at glucose levels greater than 5 mM, reaching a value of 120 nmol/mg protein (approximately equivalent to a 30 mM intracellular concentration) in cells incubated in 25 mM glucose (Fig. 1A). The level of Glu-6-P in AdCMV-GKL-treated cells incubated in 25 mM glucose was 10 and 50 times greater than the corresponding levels in AdCMV-HKI and untreated cells, respectively. In parallel with the changes observed for Glu-6-P, the increases in UDP-Glc in AdCMV-HKI-and AdCMV-GKL-

FIG. 1. Effects of HKI and GK overexpression on Glu-6-P levels.
Rat hepatocytes were treated with AdCMV-GKL or AdCMV-HKI adenoviruses or left untreated and incubated for 42 h in the presence of 1 mM glucose to allow full expression of the adenovirus-transferred genes. Cells were then washed, exposed to the concentrations of glucose indicated for 2 h, and collected for measurement of intracellular Glu-6-P levels. The same data are shown in A and B, but B uses a smaller scale to allow a clearer view of the changes caused by AdCMV-HKI treatment. Data represent the mean Ϯ S.E. for five independent experiments. The symbols indicate statistical significance at the following levels: ‫,ءء‬ p Ͻ 0.005; and ‫,ءءء‬ p Ͻ 0.001, for comparisons with untreated cells; and a, p Ͻ 0.01 for comparison with AdCMV-HKI-treated cells. treated cells were most pronounced at low (1-5 mM) and high (25 mM) glucose levels, respectively (Fig. 2). At 1 mM glucose, UDP-Glc levels in AdCMV-HKI-treated cells were more than double those in AdCMV-GKL-treated or untreated cells, whereas at 25 mM glucose, AdCMV-GKL-treated cells had UDP-Glc levels that were 3-fold greater than in AdCMV-HKItreated cells and 7-fold greater than in untreated control cells.
Glycogen Synthase Activity in AdCMV-GKL-and AdCMV-HKI-treated Hepatocytes-A possible mechanism that can explain the ability of GK but not HK overexpression to activate glycogen synthesis (15) is that the Glu-6-P derived from GK but not HKI increases the proportion of glycogen synthase that is in the active state. To investigate this hypothesis, we measured changes in the glycogen synthase activation state in response to changes in media glucose and intracellular Glu-6-P levels. Total glycogen synthase activity (measured in the presence of 6.6 mM Glu-6-P) was not statistically different in untreated cells, AdCMV-HKI-treated cells, and AdCMV-GKL-treated cells (2.06 Ϯ 0.05, 2.30 Ϯ 0.09, and 2.12 Ϯ 0.11 milliunits/10 6 cells, respectively; mean Ϯ S.E. for five determinations). A dramatic difference was noted, however, in the glycogen synthase activity ratio (activity measured in the absence of Glu-6-P divided by activity measured in the presence of 6.6 mM Glu-6-P), which increased sharply as a function of glucose concentration in the supernatant fraction from AdCMV-GKLtreated cells to a ratio of 0.56 Ϯ 0.04 at 25 mM glucose, a value three times greater than achieved in extracts from untreated or AdCMV-HKI-treated cells (Fig. 3A). Similar results were obtained in the 10,000 ϫ g pellet fraction from these cells (Fig.  3B). Fig. 4, A and B, demonstrates that the increase in the GS activation state in both the supernatant and pellet fractions from AdCMV-GKL-treated hepatocytes is correlated with Glu-6-P levels. No increase in the GS activity ratio was observed in AdCMV-HKI-treated cells relative to untreated cells (Fig. 3), despite the fact that Glu-6-P levels are increased to a similar extent in AdCMV-GKL and AdCMV-HKI-treated hepatocytes (Fig. 1B) at glucose concentrations (5 mM, for example) that induce GS activation in AdCMV-GKL-treated cells. These data support the hypothesis that Glu-6-P produced from GK has a different regulatory impact on glycogen synthase activation than Glu-6-P produced by hexokinase I.
In addition to its role in promoting the activation of glycogen synthase, Glu-6-P can also induce the translocation of the enzyme from a cytosolic pool to a 10,000 ϫ g pellet pool (12). Total glycogen synthase activity (measured in the presence of 6.6 mM Glu-6-P) decreased in the supernatant fraction from 1.5 Ϯ 0.1 milliunits/10 6 cells in untreated cells to 0.73 Ϯ 0.07 milliunits/10 6 cells in AdCMV-GKL-treated cells incubated at 25 mM glucose. Concomitantly, total GS activity increased in the pellets from 0.53 Ϯ 0.08 to 1.44 Ϯ 0.08 milliunits/10 6 cells, respectively. The combination of translocation and activation led to an increase of active GS in the pelletable fraction in AdCMV-GKL-treated but not AdCMV-HKI-treated cells. As shown in Fig. 5, incubation of AdCMV-GKL-treated but not AdCMV-HKI-treated hepatocytes with increasing concentrations of glucose caused a large increase in active glycogen synthase (measured in the absence of Glu-6-P) in the 10,000 ϫ g pellet fraction. In cells incubated at 5 mM glucose, active glycogen synthase in the pellet from glucokinase-overexpressing cells was twice that in untreated or hexokinase-overexpressing cells, becoming five times greater in cells incubated at 25 mM glucose. The corresponding increases in the supernatant fraction were less marked, being only 33% and 2-fold at 5 and 25 mM glucose, respectively. These data provide further support for a differential regulatory impact of Glu-6-P produced by  Fig. 1 and collected for assay of glycogen synthase activity. A, glycogen synthase activity ratio (glycogen synthase activity measured in the absence of Glu-6-P divided by activity measured in the presence of 6.6 mM Glu-6-P) in the 10,000 ϫ g supernatant fraction of hepatocyte homogenates. B, glycogen synthase activity ratio in the 10,000 ϫ g pellet fraction of hepatocyte homogenates. Data represent the mean Ϯ S.E. of five independent experiments. The symbols indicate statistical significance at the following levels: ‫,ء‬ p Ͻ 0.01; ‫,ءء‬ p Ͻ 0.005; and ‫,ءءء‬ p Ͻ 0.001, for comparisons with untreated cells; and a, p Ͻ 0.01; b, p Ͻ 0.005; and c, p Ͻ 0.001, for comparisons with AdCMV-HKI-treated cells.
hexokinase versus glucokinase in liver cells.
Relationship between Glu-6-P and Lactate Accumulation-The foregoing results indicate that Glu-6-P produced from glucokinase has a greater glycogenic potential than Glu-6-P produced from hexokinase. To investigate the relationship between the source of Glu-6-P and the glycolytic rate, intracellular Glu-6-P levels and lactate levels were measured in untreated and AdCMV-GKL-and AdCMV-HKI-treated cells cul-tured at glucose concentrations in the range of 1-5 mM (Fig.  1B). As shown in Fig. 6, a tight linear relationship exists between the intracellular Glu-6-P level and the media lactate concentration for all three groups of cells, indicating that Glu-6-P produced by glucokinase has no advantage relative to Glu-6-P produced by hexokinase for entry into the glycolytic pathway.

DISCUSSION
The purpose of this study was to further evaluate the role of glucose phosphorylation in the control of glycogen synthase and to elucidate the mechanism underlying the differential metabolic impact of glucokinase and hexokinase I overexpression in hepatocytes. Previous studies have shown that glucose must be phosphorylated to promote the activation of hepatic glycogen synthase (4,6,7). The results of the current study clearly support this hypothesis but also introduce the new concept that the potency of Glu-6-P for activation of glycogen synthase and glycogenesis is determined by the hexokinase isoform that is responsible for its production.
We have used the recombinant adenovirus system to overexpress glucokinase or hexokinase I, thereby raising Glu-6-P to levels much higher than found in untreated hepatocytes expressing only endogenous hexokinases. The accumulation of Glu-6-P in AdCMV-HKI-and AdCMV-GKL-treated cells reflects the known kinetic properties of hexokinase I and glucokinase. The increase in Glu-6-P levels in cells overexpressing glucokinase is described by a sigmoidal curve, with an inflection point at 5 mM glucose, in accord with the enzyme's high S 0.5 for glucose and its cooperative behavior (13,14). In cells overexpressing hexokinase I, Glu-6-P accumulation is enhanced to the same degree as in AdMCV-GKL-treated cells up to glucose concentrations of 5 mM, but at higher concentrations of the sugar, much more Glu-6-P accumulates in the glucokinase-overexpressing cells. Since glucose-phosphorylating activity in extracts of AdCMV-GKL-and AdCMV-HKI-treated cells was equivalent when measured at 20 mM glucose, the modest further accumulation of Glu-6-P at high glucose concentrations in hexokinase-overexpressing cells is likely explained by the inhibition of this enzyme by its reaction product (14).
At low glucose concentrations, Glu-6-P does accumulate in FIG. 4. Correlation between glycogen synthase activity ratio and Glu-6-P levels in glucokinase-overexpressing hepatocytes. Glu-6-P levels measured in AdCMV-GKL-treated hepatocytes at different glucose levels (Fig. 1A) were plotted against the glycogen synthase activity ratio achieved at the same glucose concentration (Fig. 3). A, plot using the glycogen synthase activity ratio values measured in the 10,000 ϫ g supernatant fraction. B, plot using the glycogen synthase activity ratio values measured in the 10,000 ϫ g pellet fraction. The regression coefficient (r) is shown for each plot.

FIG. 5. Effects of HKI and GK overexpression on glycogen synthase translocation.
Hepatocytes were treated at different glucose concentrations as described in the legend to Fig. 1 and collected for measurement of the active form of glycogen synthase (activity measured in the absence of Glu-6-P) in the 10,000 ϫ g pellet fraction of hepatocyte homogenates. Data represent the mean Ϯ S.E. for five independent experiments. The symbols indicate statistical significance at the following levels: ‫,ء‬ p Ͻ 0.01; and ‫‪p‬ءءء‬ Ͻ 0.001, for comparisons with untreated cells.
FIG. 6. Correlation between lactate production and Glu-6-P levels. Hepatocytes were treated at different glucose concentrations as described in the legend to Fig. 1. After the 2-h incubation at different glucose concentrations, media were collected for lactate determination, cells were collected for measurement of intracellular Glu-6-P levels, and the values were plotted against each other. Data represent the means of lactate and Glu-6-P measurements in five independent groups of cells. The regression coefficient (r) is shown for the plot.
AdCMV-HKI-treated cells to levels that are 5-9 times higher than in untreated control cells. Importantly, this Glu-6-P does not have the same regulatory impact as the identical amount of Glu-6-P produced by overexpression of glucokinase. This is best illustrated by comparison of events in glucokinase-versus hexokinase-overexpressing cells incubated at 5 mM glucose. Glu-6-P levels are elevated from approximately 1.5 nmol/mg protein (approximately 0.3 mM intracellular concentration) to 8 -10 nmol/mg protein (2.0 -2.5 mM) in both AdCMV-GKL-and Ad-CMV-HKI-treated cells (Fig. 1B), but only the glucokinaseoverexpressing cells exhibit increases in the glycogen synthase activity ratio (Fig. 3), translocation of glycogen synthase (Fig.  5), and, as shown in our previous study (15), activation of glycogen synthesis at this concentration of the sugar. Glu-6-P is thought to bind to glycogen synthase and to cause a conformational change that activates the enzyme and renders it a better substrate for protein phosphatases (10). The increase in the activation state of glycogen synthase in AdCMV-GKL-treated cells suggests that accumulation of glucokinase-derived Glu-6-P enhances dephosphorylation of the enzyme more effectively than Glu-6-P derived from overexpressed hexokinase I.
One potential explanation for the lack of metabolic impact of Glu-6-P on glycogen synthesis in hexokinase I-overexpressing cells could have been a failure to divert this pool of the intermediate to UDP-Glc. Our data show, however, that this is not the case, since UDP-Glc levels are actually more elevated in hexokinase-overexpressing cells than in glucokinase-overexpressing cells incubated at 1 or 5 mM glucose (Fig. 2). Thus, UDP-Glc formation is clearly not limiting at 5 mM glucose, a concentration sufficient to activate glycogen accumulation in glucokinase-overexpressing cells but not hexokinase-overexpressing cells. These data show that the increase in UDP-Glc achieved at 5 mM glucose is not sufficient to activate glycogen synthesis through a "push" mechanism, as has been proposed (26), and that other regulatory events, such as Glu-6-P-mediated activation of glycogen synthase, must accompany accumulation of the proximate precursor.
Recent studies have demonstrated that hepatic glucokinase translocates from a bound to a free state in response to high glucose concentrations or to micromolar concentrations of fructose or sorbitol, which are precursors of fructose 1-phosphate (27). Translocation of glucokinase by these substrates correlates with stimulation of glycogen synthesis (28). In contrast to glucokinase, the low K m hexokinases do not translocate in response to either elevated glucose concentrations or the presence of fructose or sorbitol (29). The failure to activate glycogen synthase with increasing Glu-6-P in hexokinase-overexpressing cells may be explained by compartmentation of this pool of Glu-6-P at a site that is not accessible to glycogen synthase. Translocation of glucokinase to a "glycogenic site" may also help explain the results of Cahill et al. (30), who showed nearly 40 years ago in rat liver slices that glycolysis becomes saturated at approximately 20 mM glucose, whereas rates of glycogen synthesis continue to increase at much higher levels of the hexose (30).
In sum, we have demonstrated that the distinct effects of glucokinase and hexokinase I overexpression on glycogen synthesis in hepatocytes can be explained by the differential impact of Glu-6-P produced by the two enzymes on activation and translocation of glycogen synthase. These studies confirm the importance of Glu-6-P in regulation of glycogenesis and provide important new evidence for a second level of control conferred by the hexokinase isoform that is responsible for glucose phosphorylation. Whether the differential potency of glucokinase and hexokinase in regulation of glycogen synthase is related to differences in their subcellular localization will be an important topic for future study.