The Repression of Hormone-activated PEPCK Gene Expression by Glucose Is Insulin-independent but Requires Glucose Metabolism*

Phosphoenolpyruvate carboxykinase (PEPCK) is a rate-controlling enzyme in hepatic gluconeogenesis, and it therefore plays a central role in glucose homeostasis. The rate of transcription of the PEPCK gene is increased by glucagon (via cAMP) and glucocorticoids and is inhibited by insulin. Under certain circumstances glucose also decreases PEPCK gene expression, but the mechanism of this effect is poorly understood. The glucose-mediated stimulation of a number of glycolytic and lipogenic genes requires the expression of glucokinase (GK) and increased glucose metabolism. HL1C rat hepatoma cells are a stably transfected line of H4IIE rat hepatoma cells that express a PEPCK promoter-chloramphenicol acetyltransferase fusion gene that is regulated in the same manner as the endogenous PEPCK gene. These cells do not express GK and do not normally exhibit a response of either the endogenous PEPCK gene, or of thetrans-gene, to glucose. A recombinant adenovirus that directs the expression of glucokinase (AdCMV-GK) was used to increase glucose metabolism in HL1C cells to test whether increased glucose flux is also required for the repression of PEPCK gene expression. In AdCMV-GK-treated cells glucose strongly inhibits hormone-activated transcription of the endogenous PEPCK gene and of the expressed fusion gene. The glucose effect on PEPCK gene promoter activity is blocked by 5 mm mannoheptulose, a specific inhibitor of GK activity. The glucose analog, 2-deoxyglucose mimics the glucose response, but this effect does not require GK expression. 3-O-methylglucose is ineffective. Glucose exerts its effect on the PEPCK gene within 4 h, at physiologic concentrations, and with an EC50 of 6.5 mm, which approximates theK m of glucokinase. The effects of glucose and insulin on PEPCK gene expression are additive, but only at suboptimal concentrations of both agents. The results of these studies demonstrate that, by inhibiting PEPCK gene transcription, glucose participates in a feedback control loop that governs its production from gluconeogenesis.

Phosphoenolpyruvate carboxykinase (PEPCK 1 ; EC 4.1.1.32) is a rate-controlling enzyme of gluconeogenesis and is regulated by hormones that are involved in the maintenance of glucose homeostasis. PEPCK is not allosterically regulated, nor is its activity altered by phosphorylation or dephosphorylation. Rather, the activity of PEPCK is modulated by alterations in the abundance of the protein, achieved principally by variations of the rate of the transcription of the gene (for reviews, see Refs. [1][2][3][4]. In animal models of diabetes, gluconeogenic enzymes are inappropriately overexpressed, while glycolytic enzymes are underexpressed despite the presence of hyperglycemia (5). Indeed, overexpression of PEPCK in transgenic animals leads to a non-insulin-dependent diabetes mellituslike condition, underscoring the importance of the proper regulation of the PEPCK gene in the maintenance of normal plasma glucose levels (6,7). In healthy animals, a decrease in plasma glucose results in the release of glucagon and glucocorticoids, hormones that increase the rate of PEPCK gene transcription. High plasma glucose stimulates insulin secretion and insulin inhibits the actions of glucagon and glucocorticoids on PEPCK gene transcription in a dominant fashion (8 -11). Glucose, under certain circumstances, also inhibits PEPCK gene expression (9,12). While a great deal is known about the how the glucostatic hormones regulate PEPCK gene expression, it is also of interest to know how glucose directly regulates the expression of the gene.
The synthesis of a number of genes that encode glycolytic or lipogenic enzymes, including liver pyruvate kinase, 6-phosphofructo-2-kinase/fructose-2,6-bisphosphatase, spot 14 (S14), and fatty acid synthase, is induced by feeding with a diet that is high in carbohydrate and low in fat and protein (13)(14)(15)(16)(17)(18). Initially it was assumed that insulin, released as a result of carbohydrate feeding, increased the rate of transcription of these genes rather than elevated levels of blood glucose per se (14). However, studies using primary hepatocytes demonstrated that insulin and glucose are both required for the induction of the transcription of these genes, since neither glucose nor insulin can activate these genes independently. In the presence of a fixed amount of insulin, the effect of glucose on these genes is concentration-dependent (19). These observations led to the hypothesis that insulin facilitates the glucose response by providing adequate amounts of liver glucokinase (GK, hexokinase IV), a high K m hexokinase that mediates the first committed step in glucose metabolism (2,20). By this view, an increase in GK activity leads to an increase in glucose flux, and this generates a metabolite that modulates gene expression (2). This hypothesis was at least partially confirmed when Kahn and colleagues demonstrated that glucose activates a reporter gene driven by the pyruvate kinase gene promoter in the absence of insulin when isolated hepatocytes are cotransfected with a vector that expresses GK (21). Likewise, the fatty acid synthase gene can be induced by glucose in hepatocytes that have been pre-treated with hormones (including insulin) that stimulate GK gene expression (22). The mechanism of glucose action on these genes is not fully understood. However, the metabolism of glucose is required since metabolizable analogs of glucose mimic the effect, whereas nonmetabolizable analogs do not (15). Thus, it is thought that a metabolite of glucose confers a signal that results in the altered expression of a number of genes involved in the regulation of glucose and lipid metabolism (13)(14)(15)(16).
Glucose decreases hormone-activated PEPCK gene expression in primary hepatocytes and Fao hepatoma cells, but fails to regulate PEPCK in H4IIE cells (9,12). This difference in the ability of glucose to regulate PEPCK gene expression may reflect the intrinsic glucose phosphorylating capacity in the different cells. In the present study, we test the hypothesis that an increase in glucose metabolism allows glucose to repress hormone-activated PEPCK gene transcription in cells that do not normally respond to glucose. HL1C rat hepatoma cells were used, a cell line that is a derivative of H4IIE cells that affords the direct measurement of PEPCK promoter activity and that does not express GK. We find that the inhibition of glucocorticoid-and cAMP-stimulated PEPCK gene transcription by glucose in these cells is dependent on GK expression, controlled in this study by exposure of the cells to a recombinant adenovirus that contains the GK cDNA (23). This inhibition by glucose correlates closely with GK activity and glucose metabolism. In addition, physiologic concentrations of glucose inhibit PEPCK gene promoter activity within 4 h. This effect is independent of insulin and is additive with this hormone at suboptimal concentrations of both agents.

EXPERIMENTAL PROCEDURES
Cell Culture, Treatment with Recombinant Adenovirus, and CAT Assays-The isolation of the H4IIE rat hepatoma-derived stable transfectant, HL1C, was described previously (24). This cell line contains the PEPCK promoter segment from Ϫ2100 to ϩ69, relative to the transcription start site, ligated to the CAT reporter gene. The maintenance of H4IIE and HL1C cells and the measurement of CAT activity have been described previously (25)(26)(27). Cells were treated with an appropriate volume (see below) of recombinant adenovirus in fresh Dulbecco's modified essential medium supplemented with 2.5% newborn calf serum and 2.5% fetal bovine serum for 30 -36 h prior to treatment with hormones and/or glucose or mannitol. Hormone and sugar or sugar alcohol treatments were carried out in serum and glucose free Dulbecco's modified essential medium supplemented with 1 mM sodium pyruvate. In addition, the medium was adjusted, as needed, with various concentrations of mannitol so that it always contained at least 5.0 mM glucose or mannitol to avoid hypo-osmotic conditions.
RNA Isolation and Primer Extension Analysis-The synthesis of the PC28 and ACT25 oligonucleotides was accomplished as described previously (30). These oligonucleotides are complimentary to the mRNAs of the rat PEPCK and rat ␤-actin genes at positions 102-129 and 42-67, respectively (31,32). Total RNA was isolated with Tri-Reagent (Molecular Research Center, Inc., Cincinnati, OH) using the instructions provided by the manufacturer. A primer extension assay was used to measure PEPCK and ␤-actin mRNA amounts, as described previously (24). The products of the primer extension assay were quantified using a Bio-Rad GS-20 Molecular Imager in conjunction with the Bio-Rad Molecular Analyst software.
Enzyme and Metabolic Assays-Glucose phosphorylating activities were measured by a radioisotopic assay using [U-14 C]glucose in whole cell extracts in the presence of 10 mM glucose 6-phosphate (Glu-6-P) to distinguish between low K m , Glu-6-P-sensitive hexokinase activities and GK activity (33). Glucose usage was measured in intact cells using [2-3 H]glucose and the subsequent production of 3 H 2 O as described previously (34,35). Lactate accumulation was determined using a colorometric assay kit (Sigma).

AdCMV-GK Increases Glucose Metabolism in HL1C Cells-
HL1C cells are a stable transfectant of H4IIE rat hepatoma cells that express the CAT reporter gene driven by the PEPCK promoter from Ϫ2100 to ϩ69 relative to the transcription start site. CAT gene expression in this system reflects the hormonal regulation of the endogenous PEPCK gene in that the transcription of the fusion gene is strongly induced (20 -50-fold) by the combination of dexamethasone/cAMP, and this response is inhibited by insulin (24,30). The ability of glucose to alter the expression of a number of glucose-responsive genes requires GK activity in order to increase glucose metabolism (13)(14)(15)(16). Intermediary metabolites of either glycolysis or the pentose phosphate shunt, as well as Glu-6-P, have been implicated as potential mediators of the signal generated by glucose in the liver (22,36). Although derived from hepatocytes, HL1C cells have no intrinsic GK activity and do not respond to glucose (data not shown and see Fig. 1). Therefore, HL1C cells were treated with a recombinant adenovirus that expresses rat liver GK (AdCMV-GK) to test whether this treatment increases glucose metabolism and, if so, whether increased glucose metabolism has an affect on hormone-activated PEPCK gene promoter activity. HL1C cells treated with AdCMV-GK and incubated in 3 mM glucose displayed a 2.5-fold increase in glucose utilization over control cells, as measured by 3 The effect of GK expression on the response of the PEPCK gene to glucose. Untreated HL1C cells, or cells treated with AdCMV-␤-Gal or AdCMV-GK, were incubated for 12 h in serum-free Dulbecco's modified essential medium (which contains 5.5 mM glucose) in the presence of dexamethasone (Dex)/cAMP, or dexamethasone/ cAMP supplemented with mannitol or additional glucose. Cells that were incubated in Dulbecco's modified essential medium plus dexamethasone/cAMP (solid bars) were compared with those that were incubated in the same medium plus dexamethasone/cAMP and 20 mM mannitol (hatched bars), or an additional 20 mM glucose (shaded bars). CAT activity was determined in the cell lysates. The results are shown as the mean of the maximal dexamethasone/cAMP response (set at 100%) from each group of cells (ϮS.E., n ϭ 3). generated from [2-3 H]glucose. This effect increased to 6-fold when the cells were incubated in 20 mM glucose. Similarly, lactate accumulation was increased about 2-fold when the cells treated with AdCMV-GK were incubated in 3 mM glucose and was markedly increased when 20 mM glucose was used in the medium (Table I). In contrast, cells treated with a recombinant adenovirus that expresses the E. coli ␤-galactosidase gene (Ad-CMV-␤-Gal) exhibited levels of glucose utilization and lactate accumulation about equal to untreated control cells in the presence of either 3 or 20 mM glucose (Table I). These data demonstrate that infection of HL1C cells with AdCMV-GK leads to the expression of GK and a subsequent increase in glucose metabolism. Furthermore, the increase in glucose metabolism is not due to adenoviral infection alone, since infection with AdCMV-␤-Gal had no affect on glucose metabolism.
Glucose Represses PEPCK Gene Transcription in a GK-dependent Manner-We tested whether elevated concentrations of glucose could inhibit hormone-activated, PEPCK promoterdriven, CAT expression in AdCMV-GK-treated HL1C cells and, if so, whether this effect was dependent on the expression of GK. Untreated cells, or cells treated with either AdCMV-␤-Gal or AdCMV-GK, were treated with dexamethasone/cAMP and challenged with either high concentrations of glucose or mannitol. A 12-h treatment with 20 mM glucose decreased the dexamethasone/cAMP activation of the PEPCK-CAT fusion gene by 80% in cells that express GK (Fig. 1). In contrast, CAT activity was not affected by the elevated glucose concentration in control cells or in cells treated with AdCMV-␤-Gal. Together, these observations indicate that the glucose effect on PEPCK gene promoter activity is dependent on the expression of GK. Additionally, treatment with adenovirus per se had no effect on PEPCK gene promoter activity, since cells treated with Ad-CMV-␤-Gal did not respond to elevated concentrations of glucose. Furthermore, 20 mM mannitol had no effect on the dexamethasone/cAMP-activated CAT activity, which demonstrates that, in this system, increased osmolarity does not alter the activity of the PEPCK gene promoter. Finally, glucose had no effect on basal PEPCK promoter activity (data not shown).
HL1C cells were next treated with various amounts of a 293 cell lysate that contained AdCMV-GK to determine whether there is a correlation between GK activity and the inhibitory effect of glucose on dexamethasone/cAMP-stimulated PEPCK gene transcription. As shown in Fig. 2, the volume of the lysate used to infect HL1C cells correlates well with the glucokinase activity measured in HL1C lysates. However, there was no change in the ability of 20 mM glucose to repress the dexamethasone/cAMP-mediated induction of the PEPCK-CAT fusion gene until the cells were treated with 100 l of the lysate, even though there was a linear increase in the activity of GK from HL1C cell extracts starting with the 25-l treatment of the AdCMV-GK-containing lysate. This observation suggests that a threshold amount of GK activity may be required before a signal sufficient to elicit an effect on PEPCK gene transcription is generated. An alternative explanation is that the two assays used in this experiment have different sensitivities. By this view, the number of cells treated with virus that is required to observe an effect on the dexamethasone/cAMP response is greater than that needed to measure an increase in GK activity. Experiments using an inducible promoter to drive the expression of GK will be needed to distinguish between these two possibilities.
As an additional control, mannoheptulose, an inhibitor of GK enzyme activity (37), was employed to confirm that the glucose effect is the result of GK enzyme activity rather than an artifact of AdCMV-GK-mediated GK protein expression. In the experiment depicted in Fig. 3, 15 mM glucose inhibited hormone-activated PEPCK-CAT gene promoter activity by 70%. Five mM mannoheptulose suppressed the dexamethasone/ cAMP response slightly (about 20%) and the same concentration of mannoheptulose blocked the inhibition by glucose. One nM insulin decreased the dexamethasone/cAMP response by 75%, and did so in the absence of glucose. This confirms previous observations which demonstrated that the insulin effect on PEPCK gene promoter activity is independent of glucose (8,9) (see also Fig. 7 below). Mannoheptulose did not relieve the insulin-mediated inhibition of hormone-stimulated CAT expression in these cells, which demonstrates the specificity of its effect. Together, these data demonstrate that glucose inhibits hormone-activated PEPCK promoter activity in HL1C cells in a GK-dependent manner.
Glucose Represses the Expression of the Hormone-activated Endogenous PEPCK Gene in H4IIE Cells Treated with AdCMV-GK-H4IIE rat hepatoma cells, the parental cell line of HL1C cells, were treated with AdCMV-GK or AdCMV-␤-Gal to test whether the hormone-activated endogenous PEPCK gene could also be repressed by glucose. Primer extension experiments were performed to measure mRNA PEPCK (Fig. 4). An overnight treatment with dexamethasone/cAMP increased mRNA PEPCK by about 9-fold, in agreement with previous results (30). Mannitol had no affect on the induction of mRNA PEPCK by dexamethasone/cAMP. However, the presence of 20 mM glucose in the culture media blunted the dexamethasone/cAMP effect in untreated cells, and in cells treated with AdCMV-␤-Gal, by 40 and 35%, respectively. This is in contrast with the observation  that glucose had no effect on the expression of the PEPCK-CAT fusion gene in untreated HL1C cells, or in HL1C cells treated with AdCMV-␤-Gal (Fig. 1). This may be because the endogenous gene in the parental cell line is more sensitive to glucose metabolism than is the PEPCK-CAT fusion gene. Alternatively, the increased glucose may be affecting the stability of mRNA PEPCK , an effect that has been previously reported (12). Nonetheless, glucose was much more effective in decreasing the dexamethasone/cAMP induction in cells that were treated with AdCMV-GK; in such cells mRNA PEPCK was reduced by 71%. Importantly, glucose had no effect on the amount of mRNA ␤-actin , an observation which demonstrates that glucose does not have a general affect on mRNA abundance. Indeed, in other experiments we found that glucose had no effect on unstimulated mRNA PEPCK levels (data not shown). Thus, expression of GK increases the ability of glucose to repress hormoneactivated PEPCK gene expression in these cells.
Time Course and Dose Response of the Glucose Effect-Adenovirus-treated HL1C cells were incubated in the presence or absence of 20 mM glucose for various periods to determine the time course of the glucose effect (Fig. 5). Untreated control cells had only background levels of CAT expression throughout the course of these experiments. However, the dexamethasone/ cAMP treatment stimulated PEPCK promoter-driven CAT activity by 4 h, and the increase continued for an additional 4 h and then it plateaued between 8 and 12 h. In contrast, the addition of 20 mM glucose to the medium resulted in a significant repression of the dexamethasone/cAMP-mediated rise in CAT activity by 4 h, an effect that was maintained throughout the course of the experiment.
Glucose inhibited, in a concentration-dependent manner, the dexamethasone/cAMP-stimulated PEPCK gene promoter activity in HL1C cells treated with AdCMV-GK (Fig. 6). Incubation in 5 mM glucose resulted in a significant decrease of dexamethasone/cAMP-stimulated CAT activity, and 15 mM glucose inhibited the dexamethasone/cAMP response by 80%. The calculated EC 50 of the glucose effect on PEPCK-CAT fusion gene promoter activity was 6.5 mM, which corresponds to the K m of GK and is well within the physiologic range of plasma glucose (20).
Suboptimal Concentrations of Insulin and Glucose Are Additive-HL1C cells were treated with AdCMV-GK, and with dexamethasone/cAMP and various concentrations of insulin in the presence or absence of 5 mM glucose, to test the relationship between glucose and insulin signaling to the PEPCK promoter (Fig. 7). Insulin at 0.01 nM repressed the dexamethasone/cAMP induction of the PEPCK gene significantly, and the maximal effect was reached at 10 nM, a result that is consistent with previous observations (24,30). The combination of 5 mM glucose and insulin at very low concentrations (0.001 and 0.01 nM) was more effective at repressing the dexamethasone/cAMP induction of the PEPCK-CAT fusion gene than was insulin alone. At higher concentrations of insulin (0.1-10 nM), the presence of glucose made no difference, since the PEPCK-CAT gene was essentially completely repressed. When both sets of data are expressed relative to the control data point within each group (no insulin), the two curves are virtually superimposable (Fig.  7, inset). Thus, the repressive effects of insulin and glucose on hormone-stimulated PEPCK gene promoter activity are additive, but only at suboptimal concentrations of both agents.
The Effects of Glucose Analogs on PEPCK Promoter Activity-HL1C cells were treated with AdCMV-GK, and with dexamethasone/cAMP in the presence or absence of glucose or the glucose analogs, 3-O-methylglucose, and 2-deoxyglucose, to test whether these agents could mimic the glucose response. 3-O-Methylglucose enters the cell but cannot be further metabolized and, as shown in Fig. 8, it cannot repress hormoneactivated PEPCK promoter activity. This result confirms that glucose metabolism to at least glucose-6-P is necessary for the glucose response. In contrast, treatment with 20 mM 2-deoxyglucose, which is readily phosphorylated to 2-deoxyglucose-6-P, resulted in a repression of hormone-activated PEPCK promoter activity in excess of that achieved by the same concentration of glucose. HL1C cells that had not been treated with AdCMV-GK were incubated with dexamethasone/cAMP in the presence of 20 mM glucose or the glucose analogs. In this experiment 2-deoxyglucose repressed hormone-activated PEPCK promoter activity, but glucose and 3-O-methylglucose were ineffective. Thus, the repression of PEPCK promoter activity by 2-deoxyglucose does not require GK activity (Fig. 8). DISCUSSION This study demonstrates that the glucose-mediated repression of hormone-activated PEPCK gene promoter activity is dependent on the metabolism of glucose. In the liver, glucose enters the cell through the facilitative glucose transporter, GLUT2, and is phosphorylated by GK to generate Glu-6-P, which can then enter a number of metabolic pathways, including glycogen synthesis, glycolysis, and the pentose phosphate pathway (for reviews, see Refs. 2, 15, 16, 20, 38, and 39). In hepatoma cells that do not express GK, such as H4IIE or HL1C cells, the phosphorylation of glucose must be performed by another hexokinase (predominantly hexokinase I). Hexokinase I has a low K m for glucose (10 M) and is maximally active well below the glucose concentration in the culture medium. Also, in contrast to GK, hexokinase I is inhibited by physiologic concentrations of Glu-6-P, so that the capacity for glucose phosphorylation is restricted in cells that do not express GK (20,40). Thus, our results suggest that in cells with a sufficiently high glucose flux, such as those that express GK, glucose generates a signal that regulates the expression of the PEPCK gene.
The importance of GK in liver glucose metabolism is underscored by the fact that overexpression of GK in hepatoma cells or isolated hepatocytes leads to an increase in glycogen deposition and lactate production (35,41,42). Conversely, GK expression is undetectable in streptozotocin-induced diabetes (streptozotocin destroys the pancreatic ␤-cells and so deprives the animal of insulin), a condition that is associated with low glycogen stores, markedly decreased glucose uptake and increased net hepatic glucose output from increased and unrestrained gluconeogenesis (5,40). Data collected from transgenic animals verifies the important role GK plays in glucose homeostasis. Overexpression of GK in the livers of transgenic mice greatly reduces the hyperglycemia induced by streptozotocin. This is accompanied by a decrease of the mRNAs that encode the gluconeogenic enzymes PEPCK and tyrosine aminotransferase and an increase of the mRNA that encodes pyruvate kinase, a glycolytic enzyme (5). These findings, and the results of the present study, suggest that GK-mediated glucose metabolism generates a signal (or signals) that coordinately regulates hepatic genes involved in glucose homeostasis, both positively and negatively, and can do so in the absence of insulin. In this manner glucose participates in the regulation of its own production and utilization.
Glucose analogs have often been used to explore the mechanism(s) by which glucose metabolism affects gene transcription (15,16,39). Although useful, these experiments must be interpreted with caution, as these compounds can have several effects. For example, 2-deoxyglucose, which is phosphorylated to 2-deoxyglucose-6-phosphate, mimics the ability of glucose to induce fatty acid synthase and acetyl-CoA carboxylase mRNA in adipocytes and pacreatic beta cells. These experiments, along with many others that show a correlation between Glu-6-P levels and glucose action, form the basis of the hypothesis that Glu-6-P is the metabolite that mediates the glucose response (16). In the present study, we found that 2-deoxyglucose mimics the ability of glucose to repress hormone-activated PEPCK promoter activity, but GK is not needed for this effect (Fig. 8). Cells exposed to 2-deoxyglucose are depleted of ATP (43). Sutherland et al. (44) demonstrated that sodium metaarsenite, which inhibits oxidative phosphorylation and therefore results in intracellular ATP depletion, decreases hormoneactivated PEPCK promoter activity through a stress-activated pathway (45). Thus it is possible that 2-deoxyglucose represses the PEPCK gene by a mechanism unrelated to its role as a glucose analog. On the other hand, we cannot rule out the possibility that 2-deoxyglucose-6-phosphate, generated by hexokinase I, could accumulate to a significant level owing to the fact that it is not readily metabolized. It could then act as an analog of Glu-6-P and affect the activity of the PEPCK gene promoter.
The PEPCK gene promoter consists of a series of hormone response units, each of which is comprised of two or more DNA elements (10,11,46,47). The glucocorticoid response unit is composed of three accessory factor elements, AF1, AF2, and AF3, two glucocorticoid receptor binding sites, and a cAMP response element (10,48,49). A mutation in any one accessory factor element (or the CRE) results in a blunted glucocorticoid response, whereas a mutation of any combination of two elements abolishes the response of the PEPCK gene to glucocorticoids (10,48,49). Thus, the multielement structure of the glucocorticoid response unit (and the other hormone response units in the PEPCK gene promoter) provides functional redundancy to the PEPCK gene so that a single mutation in a promoter element or accessory factor does not deprive the organism of the ability to make glucose. Given the importance of gluconeogenesis to the survival of the organism, it is not surprising that there is considerable redundancy built into the system. Indeed, another layer of redundancy is provided by the multiple hormones (glucagon, glucocorticoids, adrenaline, and retinoic acid) that stimulate PEPCK gene transcription (1)(2)(3)(4). This reasoning can be extended to the repression of the hormone-activated PEPCK gene by insulin and glucose. The secretion of insulin serves as the primary response of the organism to hyperglycemia (38). While insulin effectively inhibits gluconeogenesis by dominantly suppressing the transcription of the PEPCK gene, also through a complex insulin response unit (47), glucose may act as a backup for insulin, ensuring that the liver perceives and responds appropriately to hyperglycemia by decreasing glucose production from gluconeogenesis.
Glucose response elements have been described in some detail for the pyruvate kinase and S14 genes, and have recently been characterized in the 6-phosphofructo-2-kinase/fructose-2,6-bisphosphatase gene (15,16,18). In the case of the pyruvate kinase and S14 genes, the elements, termed carbohydrate response elements (ChoREs), are composed of two E-box-like motifs (CANNTG) separated by 5 base pairs (50). In addition, both the pyruvate kinase and the S14 ChoREs require accessory factors. Thus, the carbohydrate responses of these genes require multiple elements and can be thought of as glucose response units. The pyruvate kinase accessory factor is hepatic nuclear factor 4 and the S14 accessory factor has not been identified (50,51). The identity of the glucose-responsive protein that binds to the pyruvate kinase ChoRE remains somewhat controversial. The basic helix-loop-helix/leucine zipper transcription factor protein upstream stimulatory factor binds to both ChoREs very well in vitro (51,52). Kahn and colleagues have suggested that upstream stimulatory factor is responsible for the glucose effect, while the Towle laboratory has recently questioned this idea (53)(54)(55). Experiments are in progress to define the glucose response unit of the PEPCK promoter.