Carbon Flux via the Pentose Phosphate Pathway Regulates the Hepatic Expression of the Glucose-6-phosphatase and Phosphoenolpyruvate Carboxykinase Genes in Conscious Rats*

Hepatic gene expression of P-enolpyruvate carboxykinase (PEPCK) and glucose-6-phosphatase (Glc-6-Pase) is regulated in response to changes in the availability of substrates, in particular glucose (Glc; Massillon, D., Barzilai, N., Chen, W., Hu, M., and Rossetti, L. (1996) J. Biol. Chem. 271, 9871–9874). We investigated the mechanism(s) in conscious rats. Hyperglycemia per se caused a rapid and marked increase in Glc-6-Pase mRNA abundance and protein levels. By contrast, hyperglycemia decreased the abundance of PEPCK mRNA. Importantly, inhibition of glucokinase activity by glucosamine infusion blunted both the stimulation of Glc-6-Pase and the inhibition of PEPCK gene expression by Glc, suggesting that an intrahepatic signal (metabolite) generated by the metabolism of glucose at or beyond Glc-6-P was responsible for the regulatory effect of Glc. The effect of Glc on the l-type pyruvate kinase gene is mediated by xylulose-5-P (Doiron, B., Cuif, M., Chen, R., and Kahn, A. (1996) J. Biol. Chem. 271, 5321–5324). Thus, we next investigated whether an isolated increase in the hepatic concentration of this metabolite can also reproduce the effects of Glc on Glc-6-Pase and PEPCK gene expression in vivo. Xylitol, which is directly converted to xylulose-5-P in the liver, was infused to raise the hepatic concentration of xylulose-5-P by ∼3-fold. Xylitol infusion did not alter the levels of Glc-6-P and of fructose-2,6-biphosphate. However, it replicated the effects of hyperglycemia on Glc-6-Pase and PEPCK gene expression and resulted in a 75% increase in the in vivo flux through Glc-6-Pase (total glucose output).

We have recently demonstrated that the marked increase in Glc-6-Pase mRNA and protein levels in the liver of diabetic rats was restored to normal levels following normalization of the plasma glucose concentrations by either insulin or the glycosuric agent phlorizin (19). Since phlorizin administration failed to normalize Glc-6-Pase gene expression when hyperglycemia was maintained by glucose infusion (19), we suggested that Glc-6-Pase gene expression in the diabetic liver is regulated by glucose independent of insulin. This preliminary observation has received support by a recent comprehensive study in primary cultured hepatocytes which demonstrated stimulation of Glc-6-Pase gene expression by glucose and fructose 2,6-biphosphate (22). Thus, Glc-6-Pase belongs to a family of genes whose expression is regulated by carbohydrates. In particular, this family includes genes coding for key enzymes within the gluconeogenic, glycolytic, and lipogenic pathways, such as L-type pyruvate kinase (23,24), PEPCK (25,26), and fatty acid synthase (27,28).
While changes in the plasma glucose concentration are likely to regulate the gene expression of liver enzymes in the intact organism as well (19), the impact of these changes on in vivo glucose fluxes has not been delineated. It has long been recognized that acute hyperglycemia decreases HGP in perfused liver (29) and in non-diabetic animals (10,30,31) and humans (32). This decrease is associated with increased flux through glucokinase (increased glucose cycling), with no measurable changes in the in vivo fluxes through gluconeogenesis and Glc-6-Pase (10). Thus, in the presence of hyperglycemia, it may be difficult to isolate the impact of the stimulation of hepatic Glc-6-Pase gene expression on hepatic glucose fluxes.
The aim of the present work was to test the hypothesis that the flux of glucose through glucokinase modulates Glc-6-Pase gene expression in conscious rats. We also wished to determine whether glucose itself or an intracellular metabolite generated by glucose metabolism is responsible for this stimulatory effect on the expression of the Glc-6-Pase gene. Our results indicate that, similar to what has been recently observed for the hepatic pyruvate kinase gene in cultured hepatocytes (33), an intrahepatic signal generated by the metabolism of glucose in the nonoxidative branch of the pentose phosphate pathway (Fig.  1A) can reproduce the stimulatory effect of glucose on the expression the Glc-6-Pase gene in vivo. Most important, this stimulation resulted in a marked increase in the rate of glucose production. We propose that the nonoxidative branch of the pentose phosphate shunt represents a glucose sensing regulatory pathway which activates a feedback control system limiting the size of the Glc-6-P pool in response to sustained increases in the rate of hepatic glucose phosphorylation.

MATERIALS AND METHODS
Animals-Male Sprague-Dawley rats (Charles River Breeding Laboratories, Inc., Wilmington, MA) were used in all studies. Rats were housed in individual cages and subjected to a standard light (6 a. m. to 6 p. m.) and dark (6 p. m. to 6 a. m.) cycle. Five to seven days before the in vivo study, rats were anesthetized with an intraperitoneal injection of pentobarbital (50 mg/kg body weight) and indwelling catheters were inserted into the right internal jugular vein and in the left carotid artery. The venous catheter was extended to the level of the right atrium and the arterial catheter was advanced to the level of the aortic arch (34,35).
Insulin/Somatostatin/Glucose/Xylitol Infusions-Studies were performed in awake, unstressed, chronically catheterized rats using a combination of the pancreatic and hyperglycemic clamp techniques (10,36,37) (Fig. 1B). All rats were fasted for 6 h before the in vivo studies. At the beginning of the basal period and 120 min before starting the glucose/insulin/xylitol infusions, a prime continuous infusion of high performance liquid chromatography-purified [3-3 H]glucose (NEN Life Science Products, Boston, MA; 20 Ci of bolus, 0.2 Ci/min) was initiated and maintained throughout the remainder of the study. Briefly, in protocol 1, a primed continuous infusion of somatostatin (1.5 g/kg⅐min) and regular insulin (ϳ0.7 milliunits/kg⅐min) were administered, and a variable infusion of a 25% glucose solution was started at time 0 and periodically adjusted to clamp the plasma glucose concentration at ϳ7 mM for 5 h (euglycemic studies) or at ϳ7 mM for 2 h followed by 1, 3, or 5 h at ϳ17 mM (hyperglycemic studies). When indicated, glucosamine (GlcN) was infused at a rate of 30 mol/kg⅐min, designed to elevate and maintain the plasma GlcN concentrations at ϳ2 mM. During the first 2 h of the studies (euglycemic period) the rate of insulin infusion was adjusted as required to maintain normoglycemia during the somatostatin infusion. The average insulin infusion required to maintain normoglycemia in all groups was 0.7 Ϯ 0.1 milliunits/kg⅐min, without the need for glucose infusion. In one group of rats the rate of insulin infusion was increased to 18 milliunits/kg⅐min for 3 h while the plasma glucose concentration was raised and maintained at ϳ17 mM by a variable infusion of glucose. In protocol 2, a primed continuous infusion of somatostatin (1.5 g/kg⅐min) and regular insulin (1.0 milliunits/kg⅐min) were administered, and a variable infusion of a 25% glucose solution was started at time 0 and periodically adjusted to clamp the plasma glucose concentration at ϳ7 mM for 5 h. During the last 3 h of the studies, vehicle or xylitol (30 mol/kg⅐min) were also infused. Blood samples (ϳ25 l) were taken every 10 min to monitor the plasma glucose concentrations and adjust the rates of glucose infusion. Plasma samples for determination of plasma insulin, glucagon, and FFA concentrations were obtained every 30 min during the study. At the end of the insulin infusion, rats were anesthetized (pentobarbital 60 mg/kg body weight, intravenously), the abdomen was quickly opened, portal vein blood obtained and liver was freeze-clamped in situ with aluminum tongs precooled in liquid nitrogen. The time from the injection of the anesthetic until freeze-clamping of the liver was less than 45 s. Tissue samples were stored at Ϫ80°C for analysis. The protocol was approved by the Institutional Animal Care and Use Committees of Albert Einstein College of Medicine.
Immunoblotting Analysis-Microsomes were prepared according to van de Werve et al. (11,19,36). Briefly, liver tissue (100 mg) was homogenized in 10 volumes of a Tris sucrose/phenylmethylsulfonyl fluoride buffer (50 mM Tris buffer, pH 7.3, 250 mM sucrose, 1 mM phenylmethylsulfonyl fluoride, and 1 mM EGTA. This homogenate was centrifuged for 10 min at 10,000 ϫ g; the cytosol was then centrifuged for 1 h at 100,000 ϫ g, and the pellet resuspended in 1 ml of Tris sucrose/phenylmethylsulfonyl fluoride buffer. The resuspended pellet was incubated at 4°C for 30 min in the presence of Triton X-100 at a final concentration of 0.1%. Protein content was measured by the Bio-Rad assay (Bio-Rad) using bovine serum albumin as standard. Equal amounts of proteins (20 g) were subjected to a 10% SDS-polyacrylamide gel electrophoreis and electrophoretically transferred to nitrocellulose membranes. After blocking, the membranes were incubated with a 1:2500 dilution of polyclonal anti-Glc-6-Pase antibody (16, 19, 36) followed by a 1:2500 dilution of goat anti-rabbit horseradish peroxidase-conjugated secondary antibodies kit (Amersham, ECL, Arlington Heights, IL).
Northern Blot Analysis-Total RNA was isolated from freezeclamped liver tissues according to the Trizol method (Life Technologies, Gaithersburg, MD). The isolated RNA was assessed for purity by the 260/280 absorbance ratio. 20 g of total RNA were electrophoresed on a 1.2% formaldehyde-denatured agarose gel in 1 ϫ MOPS running buffer. The RNA was visualized with ethidium bromide and transferred to a Hybond-N ϩ membrane (Amersham). We used a 1. 25 1. A, simplified scheme of the sites of entry of glucose and xylitol in the pentose phosphate pathway. Both glucose and xylitol can generate xylulose-5-P in hepatocytes. Glucose is rapidly phosphorylated to glucose 6-phosphate, thus becoming available for glycogen synthesis, glycolysis, or for entry into the pentose phosphate pathway. Under most circumstances, the pentose phosphate pathway receives only a small fraction (ϳ1-2%) of the incoming glucose via the conversion of glucose-6-P to 6-phosphogluconolactone. Ribulose 5-phosphate is the end product of the oxidative branch of this pathway. The nonoxidative branch of the pentose phosphate pathway begins with the conversion of ribulose-5-P to xylulose-5-P or ribose-5-P. Triose phosphates are also in equilibrium with xylulose-5-P via a reversible reaction with glyceraldehyde-3-P. Alternatively, infused xylitol is directly phosphorylated to xylulose-5-P. In the fed state, the hepatic xylulose-5-P levels are highly sensitive to changes in the extracellular glucose concentration. B, schematic representation of the experimental protocols. To control the plasma insulin and glucagon concentrations during the in vivo studies, somatostatin and insulin were infused throughout. Protocol 1 was designed to assess the role of hyperglycemia per se on hepatic glucose fluxes and PEPCK and Glc-6-Pase gene expression. Studies in which euglycemia was maintained for 5 h (control studies) were compared with studies in which 2 h of euglycemia (glucose, ϳ7 mM) was followed by 1, 3, or 5 h of hyperglycemia (glucose, ϳ17 mM). Insulin was infused at the rate of ϳ0.7 milliunits/kg⅐min to generate plasma hormone concentrations similar to those measured during the basal period. In an additional group of rats, the rate of insulin infusion was raised to 18 milliunits/ kg⅐min during the last 3 h of the clamp studies, to generate marked hyperinsulinemia in the presence of hyperglycemia. Protocol 2 was designed to assess whether the infusion of the pentose xylitol could reproduce some of the effects of hyperglycemia per se on hepatic glucose fluxes and PEPCK and Glc-6-Pase gene expression. Somatostatin and insulin were infused for 2 h (basal) or for 5 h and xylitol (xylitolϩ) or vehicle (xylitolϪ) were infused during the last 3 h (3 h).
formed for 4 h at 42°C in 5 ϫ SSC, 50% formamide (v/v), 5 ϫ Denhardt's, 100 g/ml salmon sperm DNA, 1% SDS, 100 mM phosphate buffer, pH 6.5, 10 mM EDTA. Hybridization was carried out for 16 h in the same buffer with the 32 P-labeled probe. The filters were washed 3 times for 10 min in 2 ϫ SSC, 0.1% SDS at room temperature and 2 times in 0.1 ϫ SSC, 0.1% SDS for 30 min at 55°C, then exposed to Fuji x-ray films for 12-48 h at Ϫ80°C with intensifying screens. Quantification was done by scanning densitometry.
Analytical Procedures-Plasma glucose was measured by the glucose oxidase method (Glucose Analyzer II, Beckman Instruments, Inc., Palo Alto, CA). Plasma insulin was measured by radioimmunoassay, using rat and porcine insulin standards. The plasma concentration of FFA was determined by an enzymatic method with an automated kit according to the manufacturer's specifications (Waco Pure Chemical Industries, Osaka, Japan). Plasma [ 3 H]glucose radioactivity was measured in duplicates in the supernatants of Ba(OH) 2 and ZnSO 4 precipitates (Somogyi procedure) of plasma samples (20 l) after evaporation to dryness to eliminate tritiated water. Fructose 2,6-biphosphate was extracted at 80°C in 0.1 M NaOH and measured using the 6-phosphofructo-1-kinase assay (38). Xylulose-5-P concentration was assayed according to the method of Casazza and Veech (39,40). Uridine diphosphoglucose and uridine diphosphogalactose concentrations and specific activities in the liver were obtained through two sequential chromatographic separations, as previously reported (10). Plasma glucosamine concentrations were determined by high performance liquid chromatography following quantitative derivatization with phenyl isothiocyanate as described previously (41). Differences between groups were determined by ANOVA analysis of variance. All values are presented as the mean Ϯ S.E.
Terminology (10)-The term total glucose output (TGO) is intended as total in vivo flux through Glc-6-Pase. The term hepatic glucose production (HGP) is intended as the net rates of Glc-6-P dephosphorylation to glucose. Finally, glucose cycling is defined as the input of extracellular glucose into the Glc-6-P pool followed by exit of plasmaderived Glc-6-P back into the extracellular pool.
Calculations-Under steady-state conditions for plasma glucose concentrations, the rate of glucose disappearance equals the rate of glucose appearance. The latter was calculated as the ratio of the rate of infusion of [3-3 H]glucose (dpm/min) and the steady-state plasma [ 3 H]glucose specific activity (dpm/mg). When exogenous glucose was given, the rate of endogenous glucose production was calculated as the difference between rate appearance and the infusion rate of glucose. The percent of the hepatic glucose-6-phosphate pool directly derived from plasma glucose was calculated as the ratio of [ 3 H]uridine diphosphoglucose and plasma [ 3 H]glucose specific activities (10).

RESULTS
Biochemical Parameters during the in Vivo Studies-There was no significant difference in age and body weight among groups. Table I displays biochemical parameters during the euglycemic and hyperglycemic clamp studies. The plasma glucose levels were kept at approximately 7 mM in the euglycemic studies and 17 mM in the hyperglycemic studies. Insulin, glucagon, and FFA levels were not significantly different between groups. During the hyperglycemic studies in which glucosamine was infused the plasma glucosamine concentration averaged 2.1 Ϯ 0.3 mM. Table II displays biochemical parameters  during the pancreatic clamp studies with vehicle or xylitol infusions. There were no significant differences in the plasma glucose, insulin, FFA, and glucagon concentrations between the two groups. Fig. 2A depicts the rate of hepatic glucose production (HGP) during the euglycemic and hyperglycemic clamp studies. In the presence of similar plasma insulin and glucagon levels, 2-3 h hyperglycemia per se inhibited HGP by 40% compared with either basal studies (saline infusion) or time control studies at euglycemia. However, this suppression of HGP could not be detected following 4 -5 h of hyperglycemia. In fact, HGP was similar between 4 and 5 h of hyperglycemia and euglycemia. Fig. 2, B and C, display the rates of in vivo Glc-6-Pase flux (TGO) and glucose cycling during pancreatic clamp studies at euglycemia and hyperglycemia. While TGO was not significantly affected by hyperglycemia at 3 h, it was significantly increased (by 58%) following 5 h at hyperglycemia versus euglycemia. As previously reported (10), hyperglycemia markedly stimulated the rate of glucose cycling. However, this rate was further increased between 3 and 5 h of hyperglycemia, probably reflecting the marked stimulation of the flux through Glc-6-Pase.

Effect of Glucose per se on Hepatic Glucose Fluxes in Conscious Rat-
Effect of Glucose per se on Hepatic Glc-6-Pase mRNA Levels in Conscious Rat-To examine in vivo the effect of hyperglycemia per se on the gene expression of key hepatic enzymes, we infused somatostatin to inhibit the endogeneous secretion of insulin and glucagon, replaced insulin either at basal or high levels, and infused glucose at a variable rate to achieve and maintain either euglycemia or hyperglycemia for 1, 3, or 5 h (Fig. 1B). The relative abundance of Glc-6-Pase mRNA was examined by Northern blot analysis using a 1.25-kilobase cDNA that recognizes the catalytic portion of this enzyme. Fig.  3A shows that hyperglycemia markedly increased the Glc-6-Pase mRNA in the liver of nondiabetic rats. Analysis of the time course of this effect demonstrates a rapid increase in the mRNA abundance within 1 h reaching a maximal value (more than 5-fold increase) at 3 and 5 h.
Role of Glucose Phosphorylation-The stimulatory effect of glucose on Glc-6-Pase gene expression could be mediated by glucose directly or by a metabolite derived from glucose metabolism. The first step in hepatic glucose metabolism is its phosphorylation by glucokinase. Glucosamine is a potent competitive inhibitor of glucokinase and it has been previously used for this purpose in vitro (42) and in vivo (37,43). Specifically, we  have recently shown that glucosamine, at plasma concentrations similar to those achieved in the present studies (ϳ2 mM), decreases hepatic glucokinase activity in vitro (by 60%) and the flux through glucokinase in vivo (37). Thus, to delineate whether glucose needs to be phosphorylated to regulate the hepatic gene expression of Glc-6-Pase, we performed additional 3-h hyperglycemic clamp studies while the hepatic activity of glucokinase was markedly inhibited by maintaining throughout the plasma glucosamine concentration at ϳ2 mM. This duration of hyperglycemia (3 h) was selected since substantial accumulation of Glc-6-Pase mRNA was consistently observed at this time. The stimulatory effect of glucose on Glc-6-Pase gene expression (Fig. 3A) was prevented by the glucosamineinduced inhibition of glucokinase activity, indicating that glucose needs to be phosphorylated to exert its effect on the Glc-6-Pase gene.

Effect of Glucose per se on Hepatic PEPCK and Glucokinase (GK) mRNAs in Conscious
Rats-To assess the specific effect of glucose on Glc-6-Pase enzyme expression, we next examined the effect of hyperglycemia per se on PEPCK and GK mRNA levels. Indeed, it has been previously shown that glucose independent of insulin decreases PEPCK gene transcription and accelerates the degradation of the PEPCK mRNA in streptozotocin-diabetic rats (26). Thus, liver PEPCK mRNA was used as negative marker for Glc-6-Pase enzyme expression since its expression is expected to be regulated in an opposite fashion by glucose per se. In fact, PEPCK mRNA displayed opposite regulation to Glc-6-Pase mRNA. Hyperglycemia markedly, rapidly, and progressively decreased PEPCK mRNA starting within 1 h of hyperglycemia (Fig. 3B). Although the inhibitory effect of hyperglycemia was less pronounced on GK mRNA than on PEPCK mRNA, Fig. 4 shows that high glucose levels were also associated with a moderate decrease in the abundance of glucokinase mRNA both at the end of 3-and 5-h hyperglycemic clamp studies. Most important, glucosamine infusion completely blocked the effect of 3-h hyperglycemia on PEPCK and GK mRNA levels suggesting that, as for Glc-6-Pase mRNA, this effect required glucose phosphorylation (Figs. 3B and 4).

Effect of Glucose per se on Hepatic Glc-6-Pase Protein in Conscious
Rats-To study if the increase in the mRNA coding for the catalytic portion of Glc-6-Pase was paralleled by a concomitant increase in the protein levels in liver microsomes, we performed Western blot analysis using a polyclonal antibody against the catalytic portion of Glc-6-Pase (16,19,36). As can be seen in Fig. 5, a marked (ϳ3-fold) increase in the catalytic portion of Glc-6-Pase protein was detected within 3 h of hyperglycemia and sustained at 5 h.
Effect of Combined Hyperglycemia and Hyperinsulinemia on Glc-6-Pase mRNA-Under most physiologic conditions, increases in the plasma glucose concentration are tightly coupled with elevations in the circulating insulin levels. Thus, we next wished to examine whether hyperinsulinemia would effectively antagonize the induction of Glc-6-Pase gene expression by hyperglycemia. As can be seen in Fig. 6, hyperinsulinemia completely prevented the glucose-induced increase in liver Glc-6-Pase mRNA. In fact, in the presence of combined hyperglycemia and hyperinsulinemia, Glc-6-Pase mRNA abundance was similar to that observed in 6-h fasted euglycemic animals at basal insulin. Since the transcriptional regulation of the L-type pyruvate kinase gene by glucose has been recently shown to be mediated by xylulose-5-P (33), we hypothesized that an isolated increase in the hepatic concentration of this metabolite may reproduce the complex effects of Glc on Glc-6-Pase and PEPCK gene expression.
Effect of Xylitol Infusion on Hepatic Glucose Fluxes in Conscious Rat-The infusion of xylitol for 3 h resulted in a ϳ3-fold increase in the hepatic concentration of xylulose-5-P (from 19 Ϯ 5 to 59 Ϯ 11 nmol/g; p Ͻ 0.01). Conversely, the hepatic levels of Glc-6-P (289 Ϯ 28 and 279 Ϯ 36 nmol/g in xylitolϩ and xylitolϪ, respectively) and fructose-2,6-biphosphate (21.0 Ϯ 1.8 and 18.9 Ϯ 2.0 nmol/g in xylitolϩ and xylitolϪ, respectively) were not significantly different at the end of the xylitol and vehicle infusions. As shown in Fig. 7, the infusion of xylitol resulted in marked increases in the rates of HGP, TGO, and glucose cycling. These results suggested that a moderate increase in the hepatic xylulose-5-P levels results in a marked increase in the in vivo flux through Glc-6-Pase. Fig. 8 shows that increasing the hepatic concentration of xylulose-5-P by ϳ3-fold markedly increased the abundance of Glc-6-Pase mRNA in the liver of nondiabetic rats. This effect was quantitatively similar to that observed in the presence of a 2.5-fold increase in the plasma glucose concentration. Xylitol infusion also led to a marked decrease in the hepatic abundance of PEPCK mRNA. A significant increase (2-3-fold) in the hepatic concentration of xylulose-5-P (to 43 Ϯ 12 nmol/g; p Ͻ 0.01) was also observed at the end of hyperglycemic clamp studies similar to those described in protocol 1. The similarities of the effects of xylitol and glucose on these two genes and the physiologic relevance of the observed increases in the hepatic concentration of xylulose-5-P (39,40,44,45) provide strong evidence in support of a common mechanism mediating the actions of these substrates on Glc-6-Pase and PEPCK gene expression in the liver. DISCUSSION We provide direct evidence for a marked and rapid in vivo regulation of the hepatic levels of Glc-6-Pase mRNA and protein by glucose in conscious rats. Using a combination of the hyperglycemic and pancreatic clamp techniques, a glycemic target was achieved while carefully controlling other major hormonal and metabolic variables. The induction of hyperglycemia in conscious rats increased the level of Glc-6-Pase gene expression by Ͼ5-fold. The increase in Glc-6-Pase mRNA abundance was paralleled by an increase in Glc-6-Pase protein levels in vivo. Thus, we were able to demonstrate that the regulation of the gene expression of Glc-6-Pase by glucose occurs in conscious nondiabetic animals. In vivo this effect was observed within 1 h after induction of hyperglycemia and it achieved a plateau between 3 and 5 h. The plasma glucose concentration which elicited the induction of the hepatic Glc-6-Pase gene in the present study was ϳ17 mM, which is physiologically relevant in the context of poorly controlled diabetes or following a meal, when the portal glucose concentration is in this range. This stimulation was specific for Glc-6-Pase since under the same experimental conditions, PEPCK mRNA abundance was markedly diminished by hyperglycemia, consistent with the notion that this gene is under negative control by the hexose (26). A moderate decrease (ϳ40%) in GK mRNA abundance was also observed following 3 and 5 h of hyperglycemia.

Xylitol Infusion Reproduces the Effect of Glucose on Glc-6-Pase and PEPCK Gene Expression in Conscious Rats-
Glucose must be metabolized to modulate the gene expression of these enzymes and the mechanism by which it stimu- This figure shows the immunoblot of the liver microsomal fraction of euglycemic control rats (CON), and rats maintained at hyperglycemia for 3 h (Glc 3 h) and 5 h (Glc 5 h). 20 g of microsomal protein were subjected to a 10% SDS-polyacrylamide gel electrophoresis transfer to nitrocellulose membranes and blotted with antibodies against the catalytic unit of Glc-6-Pase. Samples are from rats which were sacrificed and liver freeze-clamped in situ following 3 and 5 h at the desired plasma glucose levels (as displayed in Table I). Analysis was performed multiple times for all rats. lates liver Glc-6-Pase expression appeared to involve an intracellular metabolite resulting from hepatic glucose metabolism. We used an in vivo experimental approach to generate support for this hypothesis. Glucosamine is a potent inhibitor of the activity of GK (37,42,43), the enzyme which catalyzes the rate-limiting step for the hepatic utilization of glucose. This amino sugar was infused during hyperglycemic clamp studies to achieve plasma concentrations capable of markedly inhibiting the hepatic activity of GK. This maneuver almost completely abolished the effects of hyperglycemia on Glc-6-Pase and PEPCK gene expression in conscious rats. Taken together, these data indicated that glucose phosphorylation is a prerequisite for the effect of glucose on hepatic Glc-6-Pase expression and that the signal is generated by glucose metabolism downstream of Glc-6-P.
Several glycolytic, gluconeogenic, and lipogenic enzymes are controlled by cabohydrates at the transcriptional level (23-26, 28, 46). Carbohydrate responsive elements have been identified in some of these genes (27,47). As we observed for the Glc-6-Pase gene in the present study, this regulation generally requires the phosphorylation of glucose (22,28,46). While the metabolic pathway of hepatic glucose utilization which regulates the transcription of most of these genes remains to be delineated, Doiron et al. (33) recently reported that the transcriptional regulation of the L-type pyruvate kinase gene by carbohydrates is mediated by a signal generated via the pentose phosphate pathway.
Since increasing the extracellular glucose concentrations in isolated cell systems or feeding a high carbohydrate diet in vivo increases the hepatic levels of xylulose-5-P by severalfold (39,40,44,45,48), we next examined whether a similar increase in the tissue levels of this metabolite, induced by means of an infusion of xylitol, would be sufficient to replicate the effects of hyperglycemia on hepatic Glc-6-Pase and PEPCK gene expression. In fact, xylulose-5-P is an intermediate in the nonoxidative branch of the pentose phosphate shunt which has been proposed to function as a sensor for hepatic glucose fluxes and to regulate the activity of fructose-6-phosphate 2-kinase:fructose-2,6-biphosphatase (44,45) and the transcription of the L-type pyruvate kinase (33) in the liver. During euglycemicpancreatic clamp studies the infusion of xylitol resulted in a marked increase in the hepatic abundance of Glc-6-Pase mRNA and in the stimulation of the in vivo flux through Glc-6-Pase. Most important, this occurred under carefully matched hormonal and metabolic conditions and with moderate increases in the hepatic levels of xylulose-5-P well within the range observed with high glucose or sucrose feeding (39,40,44). Furthermore, these tissue levels were not sufficient to increase the hepatic concentrations of other key metabolites, such as Glc-6-P, UDP-glucose, and fructose-2,6-biphosphate. Since changes in fructose-2,6-biphosphate induced by adenovirus vector overexpression altered Glc-6-Pase gene expression in FAO cells (22), it was particularly important to ascertain that the in- crease in xylulose-5-P concentration did not induce sizeable changes in the levels of fructose-2,6-biphosphate. The use of postabsorptive rather than starved rats in the present studies may account for the lack of increase in fructose-2,6-biphosphate levels in the liver (39,40,48). While the dramatic effect of glucose and xylitol on Glc-6-Pase and PEPCK mRNA levels and on hepatic glucose fluxes strongly suggest a common mechanism for their action, our findings do not exclude the possibility that increased concentration of fructose-2,6-biphosphate could also play a role in the regulation of the Glc-6-Pase gene, particularly in the starved liver. A lag time was observed between the increase in Glc-6-Pase gene expression and changes in the in vivo glucose fluxes. This may reflect the time required for the synthesis and intracellular targeting of an active catalytic unit of the enzyme.
It may appear to be paradoxical that hyperglycemia increases Glc-6-Pase gene expression since the latter effect would favor further release of glucose by the liver. However, it should be noted that insulin, which increases concomitantly with glucose during meal absorption, has a potent inhibitory effect on the transcription of the Glc-6-Pase gene (8,16,18,19,21). The ability of hyperglycemia and other nutrients (36) to balance the effect of insulin on the regulation of this gene may serve to avoid complete depletion of the enzyme during meal absorption. This may, in turn, help prevent excessive hepatic storage of glucose and prepare the transition to the post-absorptive and fasting periods when increased glucose output is needed. Changes in the level of Glc-6-Pase expression during feeding may be reflected in changes in the enzyme activity occurring later during the postabsorptive state.
Finally, the ability of liver cells to sense glucose and to coordinately modify the gene expression of several key enzymes may represent an important feedback control system in response to sustained changes in the rate of glucose flux and/or metabolism. In fact, it may be argued that the overall substrate/energy regulation of glycolytic and gluconeogenic enzymes tends to favor the utilization of hexose phosphates and to limit further phosphorylation of glucose. For example, the expression of glycolytic and lipogenic enzymes is increased while that of a pivotal gluconeogenic enzyme, i.e. PEPCK, is decreased by glucose. Conversely, the expression of GLUT2 and Glc-6-Pase is stimulated while that of GK is inhibited. Interestingly, while glucose and insulin tend to regulate the gene expression of most hepatic enzymes in the same direction (21,28,46), they affect the gene expression of GK, GLUT2 (27,49), and Glc-6-Pase (16 -19, 21, 22) in an opposite fashion.
In conclusion, we provide evidence for the in vivo regulation of Glc-6-Pase and PEPCK mRNA levels by glucose and by a pentose entering the nonoxidative branch of the pentose phosphate pathway. We propose that the latter pathway acts as a hepatic sensor of hexose and triose phosphate availability capable of coordinately regulating the gene expression of key hepatic enzymes involved in their formation and utilization.