The molecular physiology of hepatic nuclear factor 3 in the regulation of gluconeogenesis.

Glucocorticoids stimulate gluconeogenesis by increasing the rate of transcription of genes that encode gluconeogenic enzymes such as phosphoenolpyruvate carboxykinase (PEPCK) and glucose-6-phosphatase. Previous studies have shown that hepatic nuclear factor 3 (HNF3) is required as an accessory factor for several glucocorticoid-stimulated genes, including PEPCK. Here, we show that adenovirus-mediated expression of an HNF3beta protein with a deleted C-terminal transactivation domain (HNF3betaDeltaC) reduces the glucocorticoid-induced expression of the PEPCK and glucose-6-phosphatase genes in H4IIE hepatoma cells. Furthermore, expression of this truncated HNF3 protein results in a proportionate reduction of glucocorticoid-stimulated glucose production from lactate and pyruvate in these cells. The expression of HNF3betaDeltaN, in which the N-terminal transactivation domain is deleted, does not exhibit any of these effects. These results provide direct evidence that members of the HNF3 family are required for proper regulation of hepatic gluconeogenesis. Modulation of the function of the HNF3 family of proteins might be used to reduce the excessive hepatic production of glucose that is an important pathophysiologic feature of diabetes mellitus.

Glucocorticoids stimulate gluconeogenesis by increasing the rate of transcription of genes that encode gluconeogenic enzymes such as phosphoenolpyruvate carboxykinase (PEPCK) and glucose-6-phosphatase. Previous studies have shown that hepatic nuclear factor 3 (HNF3) is required as an accessory factor for several glucocorticoid-stimulated genes, including PEPCK. Here, we show that adenovirus-mediated expression of an HNF3␤ protein with a deleted C-terminal transactivation domain (HNF3␤⌬C) reduces the glucocorticoidinduced expression of the PEPCK and glucose-6-phosphatase genes in H4IIE hepatoma cells. Furthermore, expression of this truncated HNF3 protein results in a proportionate reduction of glucocorticoid-stimulated glucose production from lactate and pyruvate in these cells. The expression of HNF3␤⌬N, in which the N-terminal transactivation domain is deleted, does not exhibit any of these effects. These results provide direct evidence that members of the HNF3 family are required for proper regulation of hepatic gluconeogenesis. Modulation of the function of the HNF3 family of proteins might be used to reduce the excessive hepatic production of glucose that is an important pathophysiologic feature of diabetes mellitus.
The stimulation of gluconeogenesis is one of the important functions of glucocorticoids. This effect is due in part to the ability of glucocorticoids to induce the transcription of genes that encode gluconeogenic enzymes such as phosphoenolpyruvate carboxykinase (PEPCK) 1 and glucose-6-phosphatase (G-6-Pase) (1,2). The mechanism by which glucocorticoids regulate transcription of the PEPCK gene has been studied in detail. A glucocorticoid response unit (GRU) that consists of two glucocorticoid receptor (GR)-binding sites (GR1 and GR2) and four accessory elements (gAF1, 2 gAF2, gAF3, and cyclic AMP response element) in the PEPCK gene promoter is required for a complete glucocorticoid response (3)(4)(5)(6)(7). A mutation of any of these accessory elements results in a 50% reduction of glucocorticoid-stimulated PEPCK gene transcription, and a combination of any two mutations of the accessory elements abolishes the glucocorticoid response (3,5,8). It follows from these results that GR1 and GR2 cannot mediate the glucocorticoid response by themselves in the context of the PEPCK gene promoter. Indeed, GR1 and GR2, either alone or in combination, cannot mediate a glucocorticoid response when placed in the context of a heterologous promoter (9). Thus, the glucocorticoid response of the PEPCK gene is completely dependent on accessory factor elements and the proteins that bind to these sites. In recent studies we have shown that chicken ovalbumin upstream promoter transcription factor and HNF4 bind to gAF1 (4), HNF3 binds to gAF2 (7), chicken ovalbumin upstream promoter transcription factor binds to gAF3 (5), and CAAT enhancer-binding protein binds to the cyclic AMP response element (10).
Members of the hepatic nuclear factor 3 (HNF3) family, HNF3␣, ␤ and ␥, play a central role in the regulation of a number of genes that are involved in various aspects of cellular metabolism (7,(11)(12)(13)(14). Much of this regulation is accomplished through N-and C-terminal transactivation domains (15). In addition, the C-terminal domain is required for accessory factor activity in the PEPCK gene GRU, whereas the N-terminal transactivation domain does not support this function (16). Interestingly, HNF3 also functions as an accessory factor in many other GRUs, such as those located in the tyrosine aminotransferase, insulin-like growth factor-binding protein-1, and 6-phosphofructo-2-kinase/fructose-2,6-bisphosphatase gene promoters (12)(13)(14). Thus, the functional interaction between HNF3 and GR appears to be a common mechanism for glucocorticoid-regulated gene transcription. Although a considerable amount is known about the organization of the GRUs of various promoters, little is known about how the different components actually influence hepatic glucose production from gluconeogenesis.
In this study recombinant adenoviruses that express structural variants of HNF3␤ were constructed to study the role this family of proteins plays in the regulation of gluconeogenesis in H4IIE hepatoma cells. We show that adenovirus-mediated ex-pression of a form of HNF3␤ that lacks the C-terminal transactivation domain results in a decrease of glucocorticoid-stimulated PEPCK and G-6-Pase gene transcription. Exposure of H4IIE hepatoma cells to the same adenovirus reduces glucose production in proportion to the decreased rate of expression of these genes. By contrast, an HNF3␤ variant protein that lacks the N-terminal transactivation domain does not support the glucocorticoid response on gene expression, nor does it influence gluconeogenesis. These results confirm the importance of HNF3␤ in the regulation of glucocorticoid-stimulated gene expression and provide direct evidence for the critical role of the protein in the regulation of gluconeognesis.

MATERIALS AND METHODS
Cell Culture, Treatment with Recombinant Adenovirus, and CAT Assays-The maintenance of H4IIE and HL1C cells and the measurement of CAT activity have been previously described (3,17). Cells were treated with an appropriate volume of cell lysates containing recombinant adenovirus in fresh Dulbecco's modified essential medium supplemented with 2.5% newborn calf serum and 2.5% fetal bovine serum for 16 -24 h (see below). After washing with phosphate-buffered saline, cells were incubated with serum free Dulbecco's modified essential medium with or without 500 nM dexamethasone (DEX) for another 16 -24 h.
Preparation of Recombinant Adenovirus-A cDNA that encodes amino acids 1-366 of HNF3␤ (HNF3␤⌬C) and another that encodes amino acids 52-458 of HNF3␤ (HNF3␤⌬N) were subcloned into the pACCMV plasmid (18). These plasmids were then cotransfected with the pJM17 plasmid into 293 human embryonal kidney cells to construct recombinant adenoviruses by homologous recombination as described previously (18). The adenoviruses were named Ad-HNF3␤⌬C and Ad-HNF3␤⌬N, respectively. The construction of a recombinant adenovirus containing the cDNA that encodes the Escherichia coli ␤-galactosidase gene (Ad-␤-Gal) was described previously (19). The amount of Ad-␤-Gal virus used was determined by first finding the volume of clarified 293 cell lysate that stained more than 80% of a monolayer of HL1C cells after fixation and treatment with 5-bromo-4-chloro-3-indolyl ␤-D-galactopyranoside (X-gal) (20). The amount of the Ad-HNF3␤⌬C and Ad-HNF3␤⌬N viruses used was determined from a dose-response experiment in which the maximum amount of virus that did not demonstrate toxicity to the cells was established. For the Ad-HNF3␤⌬C virus, this volume of clarified lysate inhibited dexamethasone-stimulated PEPCK gene promoter activity from semiconfluent HL1C cells in a 100-cm 2 dish by 50%, as measured by CAT activity. Overexpression of protein from both viruses was confirmed by gel shift analysis. Expression levels of HNF3␤⌬C and HNF3␤⌬N from the selected amount of virus were approximately equal as determined by this method, and both proteins were expressed in excess of endogenous HNF3␤.
Gel Mobility Shift Assay-The preparation of whole cell lysates and the method of performing the gel mobility shift assays have been previously described (16). The double-stranded oligonucleotide used in the gel mobility shift assays has the sequence 5Ј-CACTAGCAAAA-CAAACTTATTTTGAACAC-3Ј (13). The polyclonal HNF3␤ antibody was obtained from James Darnell of Rockefeller University.
Ribonuclease Protection Assay-Total cellular RNA was isolated from H4IIE cells using TRI REAGENT (Molecular Research Center, Cincinnati, OH) according to the manufacturer's instructions. RNase protection assays were performed according to the instructions provided with the Ambion (Austin, TX) RPAII kit. The rat PEPCK and G-6-Pase RNA probes were generated from polymerase chain reactions in which the downstream primer contained the T7 promoter. A 5-l aliquot of the polymerase chain reaction was added directly to the components of the Ambion Maxiscript kit, with [␣-32 P]UTP, to produce radiolabeled RNA. The rat ␤-actin control probe purchased from Ambion generated a 126-base pair fragment after digestion with RNase A and T1. The sequences of the oligonucleotides used to generate the probes for rat PEPCK (150 base pairs after digestion) and rat G-6-Pase (175 base pairs after digestion) are 5Ј-GGGTGGAAAGTTGAATGTG (upstream) and 5Ј-GGATCCTAATACGACTCACTATAGGGAGGGGA-TGGTCTTAATGGCGTTC (downstream) for PEPCK and 5Ј-CCA-GTCGACTCGCTACCTC (upstream) and 5Ј-GGATCCTAATACGACT-CACTATAGGGAGGAACCAGTCTCCAACCACTG (downstream) for G-6-Pase.
Glucose Production Assay-H4IIE cells were incubated in 100-cm 2 plates 3 h at 37°C in 15% atmospheric air/5% CO 2 in 5 ml of glucose production buffer (glucose-free Dulbecco's modified essential medium, pH 7.4, containing 20 mM sodium lactate and 2 mM sodium pyruvate without phenol red). At the end of this incubation, 0.5 ml of medium was taken to measure the glucose concentration in the culture medium using a glucose assay kit (Sigma 510-A). A 2-fold concentration of the kit reagents was used to increase the sensitivity. Cells were collected and lysed, and the total protein concentration was measured (Bio-Rad) to correct for cell count. Statistical evaluations were obtained using the Student's t test method.

RESULTS AND DISCUSSION
Previous experiments have shown that HNF3␤, by binding to the gAF2 element in the GRU, serves as an accessory factor for glucocorticoid-mediated activation of PEPCK gene expression. This effect requires the C-terminal transactivation domain of HNF3␤ but is not dependent on the N-terminal transactivation domain (16). We reasoned that a protein lacking the C-terminal transactivation domain would bind DNA and compete with the endogenous protein with regard to glucocorticoid accessory factor activity. This protein, if abundantly expressed in H4IIE cells, should blunt the response of the PEPCK gene to glucocorticoids in a manner equivalent to a mutation or deletion of gAF2 (3,16,21). If this gene is centrally involved in gluconeogenesis, as is suggested by metabolic control strength studies, glucocorticoid-mediated glucose production from noncarbohydrate precursors should also be decreased (22). Accordingly, a recombinant adenovirus (Ad-HNF3␤⌬C) that lacks this C-terminal domain was constructed. A second recombinant adenovirus was prepared in which the N-terminal transactivation domain of HNF3␤ was deleted (Ad-HNF3␤⌬N). The latter construct served as a control because it retains accessory factor activity (16). A gel mobility shift assay was used to confirm that these two HNF3␤ mutants/proteins are expressed in H4IIE hepatoma cells after the cells are exposed to the adenoviruses. The C-terminal truncated protein is smaller than the N-terminal truncated protein, thus the protein-DNA complex formed from extracts of HNF3␤⌬C-treated cells migrates faster than its counterpart from HNF3␤⌬N-treated cells (Fig. 1, compare  lanes 7 and 10). The formation of both protein-DNA complexes was blocked by adding an antiserum directed against HNF3␤ but not by an antiserum directed against LexA (Fig. 1, lanes  7-9 and 10 -12). These experiments confirm that the truncated HNF3␤ proteins are expressed in H4IIE cells. Furthermore, the expression level of these two HNF3␤ proteins was approximately equal, and both were expressed in excess of the endog- enous HNF3␤ (Fig. 1, lanes 1-6). The treatment of H4IIE cells with a recombinant adenovirus that contains a cDNA that expresses ␤-glactosidase (Ad-␤-Gal) did not influence the binding pattern of HNF3␤ proteins in the gel mobility shift assay (Fig. 1, compare lanes 1-3 with lanes 4 -6).
The HL1C cell line is a stable transfectant of H4IIE cells that contains the PEPCK gene promoter from Ϫ2100 to ϩ69, relative to the transcription start site, ligated to the CAT reporter gene. The regulation of CAT gene expression and activity recapitulates the hormonal regulation of the endogenous PEPCK gene in these cells so they are very useful in studying PEPCK promoter activity (17,23). HL1C cells exposed to 500 nM DEX overnight exhibit a 20-fold induction of CAT activity (Fig. 2). As previously demonstrated, treatment of HL1C cells with the Ad-␤-Gal adenovirus does not influence the DEX response of these cells (Fig. 2 and Ref. 20). The DEX response was, however, reduced by about 50% when HL1C cells were treated with Ad-HNF3␤⌬C. By contrast, the DEX response of HL1C cells treated with HNF3␤⌬N was unaffected (Fig. 2). These results are consistent with our previous findings showing that a deletion or a mutation of the HNF3-binding site (gAF2) of the PEPCK gene promoter resulted in a 50 -70% reduction of the response of this gene to glucocorticoids (3,7,8). These results also confirm the observation that the C-terminal transactivation domain is required for the accessory activity of the PEPCK glucocorticoid response and that the N-terminal domain is not necessary (16).
H4IIE cells were treated with Ad-␤-Gal, Ad-HNF3␤⌬N, or Ad-HNF3␤⌬C to determine the role HNF3␤ plays in the regulation of endogenous PEPCK gene expression by glucocorticoids. Ribonuclease protection assays were used to measure PEPCK mRNA (Fig. 3). Overnight exposure of H4IIE cells to DEX resulted in a 3-fold induction of PEPCK mRNA (Fig. 3,  compare lanes 1 and 2). The prior treatment of the cells with either Ad-␤-Gal (lanes 3 and 4) or Ad-HNF3␤⌬N (lanes 7 and 8) had no significant effect on this induction, which is due to an increased rate of transcription of the gene (24). By contrast, treatment of cells with Ad-HNF3␤⌬C decreased the glucocorticoid-stimulated PEPCK mRNA response by about 50% and the basal expression of the PEPCK gene by about 20% (lanes 5 and 6). ␤-Actin mRNA was not affected by DEX or by the prior treatment with any of the adenovirus constructs. These results demonstrate the importance of HNF3 in the regulation of the endogenous PEPCK gene by glucocorticoids and are consistent with the conclusions derived from previous studies (7,16). In addition, expression of the PEPCK gene is markedly reduced when the HNF3 DNA-binding domain, which does not contain either of the transactivation domains, is stably transfected into hepatoma cells (25). Furthermore, expression of the PEPCK gene is reduced in mice carrying a null mutation of the gene encoding HNF3␥ (26).
The rate of transcription of the G-6-Pase gene, which encodes another rate-controlling gluconeogenic enzyme, is also stimulated by glucocorticoids (27). There are several HNF3-binding sites in the G-6-Pase gene promoter (28), so we tested for effects of the HNF3␤ mutants on the expression of the endogenous G-6-Pase gene (Fig. 3). Adenovirus-mediated expression of HNF3␤⌬C, but not HNF3␤⌬N, repressed glucocorticoid-stimulated G-6-Pase mRNA levels (Fig. 3, compare lanes 5 and 6 with  lanes 7 and 8). Although HNF3 is not known to be an accessory factor for the glucocorticoid response of the G-6-Pase gene, the results here suggest that it may serve this function. This would not be surprising because HNF3 is an accessory factor for the glucocorticoid response of the tyrosine aminotransferase, insulin-like growth factor-binding protein-1, and 6-phosphofructo-2-kinase/fructose-2,6-bisphosphatase genes (12)(13)(14).
Because HNF3␤⌬C reduced the glucocorticoid-mediated induction of PEPCK and G-6-Pase gene expression, we predicted that this treatment would also decrease glucocorticoid-stimulated gluconeogenesis. We therefore measured glucose production in H4IIE cells after exposure to the recombinant adenovi-

FIG. 3. Adenovirus-mediated expression of HNF3␤⌬C represses glucocorticoid-induced PEPCK mRNA in H4IIE cells.
Control H4IIE cells or cells infected with Ad-␤-Gal, Ad-HNF3␤⌬N, or Ad-HNF3␤⌬C overnight were treated with or without 500 nM dexamethasone for an additional 18 h. The amount of virus used was determined as described under "Materials and Methods." Total cellular RNA was prepared, and the abundance of PEPCK, G-6-Pase, and ␤-actin RNA was determined using the ribonuclease protection assay described under "Materials and Methods." A typical autoradiogram is shown. ruses. The low basal glucose production from gluconeogenesis in H4IIE cells is increased 5-fold by an overnight exposure of the cells to DEX (Fig. 4). Adenovirus-mediated expression of HNF3␤⌬C reduced glucocorticoid-induced glucose production in a dose-dependent manner (Fig. 4). By contrast, the expression of HNF3␤⌬N had a small stimulatory effect on glucocorticoid-stimulated gluconeogenesis in these cells, and the expression of ␤-galactosidase had no effect (Fig. 4). Thus, the HNF3␤⌬C-mediated effect on gluconeogenesis corresponds to its ability to decrease the expression of two genes, PEPCK and G-6-Pase, that encode gluconeogenic enzymes. The important role of HNF3 in the regulation of glucose homeostasis has been proposed previously, but its direct role in gluconeogenesis has not been previously explored. HNF3 could be a master regulator of metabolism because it can regulate the expression of HNF4 and HNF1 genes, which in turn regulate the expression of several genes that encode enzymes involved in glucose metabolism (29 -31). In fact, mutations in either the HNF4 or the HNF1 gene are associated with specific forms of maturity onset diabetes of the young, in which defective insulin secretion is the primary pathophysiologic feature (32,33). Our results provide direct evidence for a critical role of HNF3 in hepatic gluconeogenesis.
The different isoforms of HNF3 may have specific effects. The hypoglycemia of HNF3␣ knock-out mice has been linked to the reduction of glucagon gene expression in pancreatic ␣ cells, but the expression of hepatic gluconeogenic enzyme genes was not altered in these animals (34). By contrast, targeted disruption of the HNF3␥ gene in mice results in decreased expression of the PEPCK and tyrosine aminotransferase genes, whereas the expression of the G-6-Pase, albumin, and transthyretin genes was not affected (26). The lack of HNF3␥ results in a mild phenotype that could be explained by a complementary increase of HNF3␣ and ␤ gene expression (26). Our strategy of using the expression of HNF3␤ mutants that can bind to all HNF3-binding sites cannot be used to distinguish the specific role of each member of HNF3 family, but it does avoid the complications that arise from the compensation provided by other HNF3 proteins when one member of the family is eliminated. We were able to use this strategy to study the function of the HNF3 family in the regulation of gene expression and glucose production.
In summary, this report provides direct evidence for the role of the HNF3 family of proteins in the regulation of gluconeogenesis. The importance of HNF3 in glucocorticoid-stimulated expression of the PEPCK and G-6-Pase genes suggests that modulation of the function of HNF3 could influence the excessive rate of hepatic gluconeogenesis that is part of the pathophysiology of type 2 diabetes mellitus. As an extension of this view, the intricate structure of the PEPCK GRU, which requires many transcription factors other than GR for a complete glucocorticoid response, may provide an opportunity for the pharmaceutical manipulation of gluconeogenesis.
FIG. 4. Adenovirus-mediated expression of HNF3␤⌬C represses glucocorticoid-induced glucose production in H4IIE cells. Control H4IIE cells or cells infected with Ad-␤-Gal, Ad-HNF3␤⌬N, or Ad-HNF3␤⌬C overnight were treated with or without 500 nM dexamethasone for another 18 h. The amount of virus used was determined as described under "Materials and Methods." The cells were washed twice with phosphate-buffered saline and then were incubated in glucose-free Dulbecco's modified Eagle's medium, pH 7.4, supplemented with 20 mM sodium lactate and 2 mM sodium pyruvate for 3 h. The glucose concentration was measured in the extracellular medium as described under "Materials and Methods." Results are presented as percentages relative to the glucose produced by DEX-treated H4IIE cells (100%). Data represent the means (Ϯ S.E.) glucose, corrected for total cell protein, of at least three experiments. The DEX response of cells transfected with Ad-HNF3␤⌬C was significantly different (p Ͻ 0.0001) from that of cells transfected with Ad-␤-Gal (double asterisks). The DEX response of cells transfected with Ad-HNF3␤⌬N was different (p Ͻ 0.05) from cells transfected with Ad-␤-Gal.