Protein kinase A phosphorylates hepatocyte nuclear factor-6 and stimulates glucose-6-phosphatase catalytic subunit gene transcription.

Glucose-6-phosphatase is a multicomponent system that catalyzes the terminal step in gluconeogenesis. To examine the effect of the cAMP signal transduction pathway on expression of the gene encoding the mouse glucose-6-phosphatase catalytic subunit (G6Pase), the liver-derived HepG2 cell line was transiently co-transfected with a series of G6Pase-chloramphenicol acetyltransferase fusion genes and an expression vector encoding the catalytic subunit of cAMP-dependent protein kinase A (PKA). PKA markedly stimulated G6Pase-chloramphenicol acetyltransferase fusion gene expression, and mutational analysis of the G6Pase promoter revealed that multiple cis-acting elements were required for this response. One of these elements was mapped to the G6Pase promoter region between -114 and -99, and this sequence was shown to bind hepatocyte nuclear factor (HNF)-6. This HNF-6 binding site was able to confer a stimulatory effect of PKA on the expression of a heterologous fusion gene; a mutation that abolished HNF-6 binding also abolished the stimulatory effect of PKA. Further investigation revealed that PKA phosphorylated HNF-6 in vitro. Site-directed mutation of three consensus PKA phosphorylation sites in the HNF-6 carboxyl terminus markedly reduced this phosphorylation. These results suggest that the stimulatory effect of PKA on G6Pase fusion gene transcription in HepG2 cells may be mediated in part by the phosphorylation of HNF-6.

contained within the lumen of the endoplasmic reticulum (1,2). In this model the other components of the glucose-6-phosphatase system act as transport proteins to shuttle both substrate and product across the endoplasmic reticulum membrane (1,2). These include a glucose-6-phosphate transporter and putative transporters for inorganic phosphate and glucose (1,2). Inactivating mutations in the G6Pase and glucose-6-phosphate transporter genes gives rise to glycogen storage disease types 1a and 1b, respectively (4). Type 1 glycogen storage disease is characterized by severe hypoglycemia in the postabsorbtive state, hyperlipidemia, hyperuricemia, and lactic acidemia (4 -6). In addition, patients are prone to growth retardation, hepatic steatosis and cirrhosis, hepatic adenoma, and renal failure (4 -7). Because of the wider tissue distribution of the glucose-6-phosphate transporter, patients afflicted with glycogen storage disease type 1b also suffer from infectious complications as a result of functional deficiencies in neutrophils and monocytes, indicating an important role for this transporter in the immune response (4).
In contrast to glycogen storage disease type 1a, which is caused by decreased G6Pase activity, increased G6Pase activity contributes to the pathophysiology of diabetes. In both type 2 and poorly controlled type 1 diabetics, the ability of insulin to stimulate peripheral glucose utilization and to repress hepatic glucose production is reduced as a consequence of insulin resistance. Although the causes of insulin resistance are unclear (8), it is apparent that the elevation in hepatic glucose production is caused by an increased rate of gluconeogenesis rather than glycogenolysis in both type 1 (9) and type 2 diabetics (10). Several lines of evidence suggest that increased expression of key gluconeogenic enzymes including G6Pase contribute to this increase in hepatic glucose production. Thus, hepatic G6Pase expression is markedly elevated in various diabetic animal models (11)(12)(13)(14)(15), and overexpression of G6Pase in hepatocytes using recombinant adenovirus was associated with enhanced rates of gluconeogenesis as well as defects in glycogen metabolism (16). Furthermore, a modest overexpression of G6Pase in rats, again using recombinant adenovirus, resulted in approximately a 1.6 -3-fold increase in hepatic G6Pase enzymatic activity that was associated with glucose intolerance, hyperinsulinemia, decreased hepatic glycogen content, and increased peripheral triglyceride stores, changes similar to those found in early stage type 2 diabetic patients (17). These observations (16,17) suggest that G6Pase is a major control point in the glucose-6-phosphatase system and represents a prime therapeutic target for the treatment of diabetes.
In the liver cAMP, glucocorticoids, glucose, fatty acids, leptin, and ␤ 3 adrenergic receptor agonists all stimulate G6Pase gene expression (11, 14, 18 -24), whereas insulin both inhibits basal G6Pase gene expression and overrides the stimulatory effects of cAMP, glucocorticoids, glucose, and fatty acids (11, 18 -20, 22, 24). Multiple cis-acting elements in the G6Pase promoter are required for the full stimulatory effect of the cAMP signal transduction pathway on G6Pase gene expression (25)(26)(27). This paper shows that one of these elements is a binding site for hepatocyte nuclear factor (HNF)-6 and demonstrates that HNF-6 is a substrate for the catalytic subunit of protein kinase A (PKA).
Plasmid Construction-The construction of a series of 5Ј truncated G6Pase-CAT fusion genes has been described previously (29,30). A site-directed mutant of the G6Pase HNF-6 motif was generated within the context of the Ϫ231 to ϩ66 G6Pase promoter fragment using a previously described three-step PCR strategy (29,31). The resulting construct was used as the template in a second PCR to create a sitedirected mutant of the G6Pase HNF-6 motif within the context of the Ϫ129 to ϩ66 G6Pase promoter fragment (Fig. 1). The heterologous XMB vector contains a minimal Xenopus 68-kDa albumin promoter ligated to the CAT reporter gene (32). Double-stranded complementary oligonucleotides representing the wild-type or mutated HNF-6 motif ( Fig. 1B) were synthesized with HindIII-compatible ends and ligated into HindIII-cleaved XMB in multiple (5-6) copies. The number of inserts was determined by restriction enzyme analysis and confirmed by DNA sequencing.
Expression vectors encoding the ␣ and ␤ forms of PKA were a generous gift from Dr. Richard Maurer (33). An empty vector control was generated by digesting the PKA␤ plasmid with XhoI and HindIII to remove the open reading frame, filling in the noncompatible ends using the Klenow fragment of E. coli DNA polymerase I, and then religating. A mammalian cell expression vector encoding the full-length form of HNF-6 was constructed by cloning the open reading frame of mouse HNF-6, isolated as an EcoRI-EcoRI fragment from the plasmid HNF-6 pGEM1 (a generous gift from Dr. Robert Costa) (34), into EcoRI-digested pcDNA3 (Invitrogen). A bacterial cell expression vector for HNF-6 was constructed by re-isolating the coding region of HNF-6 from the pcDNA3 plasmid, minus nine amino acids at the N terminus, as a PvuI-XhoI fragment and ligating into XhoI-digested pET15b (Novagen). The noncompatible ends were then filled in using the Klenow fragment of E. coli DNA polymerase I prior to blunt-end ligation. In the resulting plasmid designated full-length HNF-6 pET15b, the HNF-6 coding sequence is in frame with that of a 6x histidine tag. A carboxyl-terminal truncation of HNF-6 was constructed by digesting this HNF-6 pET15b plasmid with SacI and BamHI, filling in the noncompatible ends using the Klenow fragment of E. coli DNA polymerase I, and then religating.
Convenient restriction enzyme sites flanking the mutated codons were then utilized to isolate smaller fragments that contained these mutated codons. Thus, the plasmid encoding the mutated serine 309 codon was digested with SacI and PstI, and the plasmids encoding the mutated serine 411 and 440 codons were digested with PstI and PvuII and PvuII and SacI, respectively. The fragments generated were then ligated together into SacI-digested pGEM7, and the resulting plasmid was designated HNF-6 site 1-3 SDM pGEM7. This plasmid was then digested with SacI and EcoRI, and the fragment generated that contained the mutated HNF-6 sequence was exchanged for the equivalent fragment in the wild-type HNF-6 pcDNA3 plasmid described above to create a plasmid designated HNF-6 SDM pcDNA3. In addition, this same SacI-EcoRI fragment was exchanged for the equivalent fragment of HNF-6 in the wild-type HNF-6 pET15b bacterial expression vector, which corresponded to a SacI-BamHI fragment. The noncompatible ends were filled in using the Klenow fragment of E. coli DNA polymerase I and then religated. The resulting plasmid was designated HNF-6 SDM pET15b. All fragments generated by PCR were completely sequenced using the USB Sequenase™ kit to verify the absence of polymerase errors. Plasmid constructs were purified by centrifugation through cesium chloride gradients (35).
Cell Culture, Transient Transfection, CAT, and ␤-galactosidase Assays-Human HepG2 hepatoma cells were grown and transiently transfected in suspension using the calcium phosphate DNA co-precipitation method as described previously (29,30). CAT and ␤-galactosidase assays were also performed exactly as described previously (29,30). The CAT activity directed by the various fusion gene constructs was corrected for the ␤-galactosidase activity in the same samples, and each construct was analyzed in duplicate in multiple transfections as specified in the figure legends.
Gel Retardation Assays-IPTG was used to induce the expression of HNF-6 in the BL21(DE3) pLysS E. coli strain (Stratagene) transformed with the full-length HNF-6 pET15b plasmid described above. Bacterial extracts were prepared by sonication in 20 mM HEPES, pH 7.5, 50 mM KCl, 1 mM MgCl 2 , 0.5 mM EGTA, and 1 mM phenylmethylsulfonyl fluoride at 4°C. The soluble fraction was separated from the particulate fraction via centrifugation. Complementary oligonucleotides representing the mouse G6Pase HNF-6 motif ( Fig. 1B) were synthesized with HindIII-compatible ends, gel purified, annealed, and then labeled with [␣-32 P]dATP by using the Klenow fragment of E. coli DNA polymerase I to a specific activity of ϳ2.5 Ci/pmol. The labeled HNF-6 oligonucleotide (ϳ7.5 fmol, ϳ30,000 cpm) was incubated with bacterial extract in a final reaction volume of 20 l containing 20 mM HEPES, pH 7.5, 50 mM KCl, 1 mM MgCl 2 , 0.5 mM EGTA, 1 mM dithiothreitol, 10% glycerol (v/v), and 2 g of poly(dI-dC)⅐poly(dI-dC). After incubation for 10 min at room temperature, the reactants were loaded onto a 6% polyacrylamide gel and electrophoresed at room temperature for 120 min at 150 V in 0.25ϫ TBE buffer (1ϫ TBE is 89 mM Tris, 89 mM boric acid, and 2 mM EDTA). After electrophoresis, gels were dried and exposed to Kodak XAR5 film, and binding was analyzed by autoradiography. For competition experiments, a 100-fold molar excess of the unlabeled doublestranded wild-type or mutated HNF-6 oligonucleotides ( Fig. 1B) was incubated with the labeled oligomer prior to the addition of bacterial extract. Binding was then analyzed by polyacrylamide gel electrophoresis as described above.
HNF-6 Purification-The full-length, wild-type and mutated, and carboxyl-terminal truncated forms of HNF-6 were expressed in the BL21-CodonPlus™(DE3)-RIL E. coli strain (Stratagene) transformed with the HNF-6 pET15b plasmids described above. Once bacterial cultures (1 liter) had reached an A 600 of ϳ0.6, protein expression was induced with IPTG (1 mM) by incubation for 4 h at 37°C. After centrifugation bacterial pellets were stored at Ϫ20°C overnight. The histidine-tagged HNF-6 protein was then partially purified using metal affinity chromatography. Briefly, bacterial pellets were thawed on ice for 15 min and then resuspended in 20 ml of lysis buffer, pH 8.0, containing 50 mM NaH 2 PO 4 , 500 mM NaCl, 20 mM imidazole, 10 mM ␤-mercaptoethanol, 1 mM phenylmethylsulfonyl fluoride, and 1 mg/ml lysozyme. After incubation on ice for 30 min, bacteria were sonicated at 4°C 10 times for 30 s each with a 1-min incubation on ice between each round of sonication. Lysates were then diluted with an additional 20 ml of lysis buffer, and particulate matter was removed by centrifugation. The supernatant was incubated with 5 ml of nickel-nitrilotriacetic acid agarose (Qiagen) for 1 h at 4°C before the application of the slurry to a column. After washing twice with 50 ml of a buffer, pH 8.0, comprising 50 mM NaH 2 PO 4 , 500 mM NaCl, 60 mM imidazole, 10 mM ␤-mercaptoethanol, 0.5 mg/ml bovine serum albumin, and 1 mM phenylmethylsulfonyl fluoride, HNF-6 was eluted in a buffer, pH 7.2, containing 50 mM NaH 2 PO 4 , 500 mM NaCl, 250 mM imidazole, 10 mM ␤-mercaptoethanol, 0.5 mg/ml bovine serum albumin, and 1 mM phenylmethylsulfonyl fluoride. The purification of HNF-6 was analyzed via SDS-PAGE, and protein was visualized by Coomassie Brilliant Blue staining (35). The yield of HNF-6 was estimated by comparison to the staining of a bovine serum albumin standard.
Phosphorylation Experiments-Partially purified histidine-tagged HNF-6 was phosphorylated at room temperature by PKA (0.12 M) in a final reaction volume of 40 l, pH 8.0, containing 5 mM NaH 2 PO 4 , 2.5 mM HEPES, 25 mM imidazole, 50 mM NaCl, 200 M [␥-32 P]ATP (1.25 Ci/nmol), 0.6 mg/ml bovine serum albumin, and 0.5% (v/v) ethylene glycol. The concentrations of HNF-6 and magnesium acetate as well as the time course of the reaction were as indicated in the figure legends. All reactions were terminated by the addition of 20 l of 3ϫ SDS sample buffer and boiling for 5 min. Samples were analyzed by SDS-PAGE, and protein was visualized via Coomassie Brilliant Blue staining (35). Gels were subsequently dried and exposed to Kodak XAR5 film. Bands corresponding to phosphorylated HNF-6 were then cut out of the dried gels, and the incorporation of phosphate into HNF-6 was quantified by scintillation counting.

An Element Located between Ϫ129 and Ϫ85 in the G6Pase Promoter Contributes to the Full Stimulatory Effect of the cAMP Signal Transduction Pathway on G6Pase-CAT Fusion
Gene Expression-To examine the molecular mechanisms by which the cAMP signal transduction pathway stimulates mouse G6Pase gene expression, the liver-derived HepG2 cell line was transiently co-transfected with a series of G6Pase-CAT fusion genes and an expression vector encoding PKA. PKA markedly stimulated G6Pase-CAT fusion gene expression, and mutational analysis of the G6Pase promoter revealed that multiple regions were required for this response ( Fig. 1A; Ref. 27). This strategy led to the identification of a cAMP response element (CRE) in the mouse G6Pase promoter located between Ϫ162 and Ϫ155 (27). However, even when this element was deleted, PKA still induced G6Pase-CAT fusion gene expression ϳ4 -5-fold ( Fig. 1A; Ref. 27). Further deletion of the G6Pase promoter sequence between Ϫ129 and Ϫ85 resulted in an additional reduction in the stimulatory effect of PKA on G6Pase-CAT fusion gene expression (Fig. 1A). These results indicate that the cAMP signal transduction pathway stimulates the expression of G6Pase through a complex cAMP response unit (CRU; Ref. 36) and that an element located between Ϫ129 and Ϫ85 in the G6Pase promoter contributes to the full stimulatory effect of PKA on G6Pase-CAT fusion gene expression.
An HNF-6 Binding Site Located between Ϫ110 and Ϫ101 in the G6Pase Promoter Contributes to the Full Stimulatory Effect of PKA on G6Pase-CAT Fusion Gene Expression-Examination of the mouse G6Pase promoter sequence between Ϫ129 and Ϫ85 for known transcription factor binding sites revealed a putative binding site for HNF-6 located between nucleotides Ϫ110 and Ϫ101 ( Fig. 1B; Ref. 37). The sequence of this HNF-6 motif is highly conserved among the mouse, rat, and human G6Pase genes (Fig. 1B); this seems to be the only HNF-6 motif in the mouse G6Pase promoter. Samadani and Costa (38) have previously shown that HNF-6 is expressed in HepG2 cells. To determine whether this putative HNF-6 binding site was required for the stimulatory effect of PKA on G6Pase-CAT fusion gene expression, this motif was mutated by site-directed mutagenesis in the context of the Ϫ129 to ϩ66 G6Pase promoter fragment. The fusion gene containing this mutation, designated Ϫ129 HNF-6 SDM, was transiently co-transfected into HepG2 cells in combination with the expression vector encoding PKA. Mutation of the G6Pase HNF-6 motif resulted in a decreased stimulation of G6Pase-CAT fusion gene expression  5 g), and either an expression vector (5 g) encoding PKA or the same vector (5 g) with the PKA open reading frame deleted. The G6Pase-CAT fusion genes contained either distinct lengths of the wild-type promoter sequence as indicated by the 5Ј deletion end points or a sitedirected mutation of the HNF-6 site designated Ϫ129 HNF-6 SDM. The HNF-6 motif was mutated as shown in B. After transfection, cells were incubated for 18 -20 h in serum-free medium. The cells were then harvested, and both CAT and ␤-galactosidase activity were assayed as described previously (29,30). Results are presented as the ratio of CAT activity, corrected for ␤-galactosidase activity in the cell lysate, in PKA-transfected versus empty vector-transfected cells (expressed as -fold induction) and represent the mean Ϯ S.E. of 5-18 (A) or 3 (C) experiments in which each construct was assayed in duplicate. SDM, site-directed mutant. B, comparison of the mouse G6Pase promoter sequence between Ϫ114 and Ϫ99 with the equivalent sequence from the rat and human G6Pase promoters. The putative HNF-6 binding motif is boxed. The consensus HNF-6 sequence is taken from Ref. 37. by PKA compared with the wild-type Ϫ129 G6Pase-CAT fusion gene construct (Fig. 1C) that was equivalent to that seen when the entire HNF-6 site was deleted, as is the case with the Ϫ85 G6Pase-CAT fusion gene (Fig. 1C). This result suggests that HNF-6 could be the factor that is mediating the stimulatory effect of PKA on G6Pase-CAT fusion gene expression through this region.
HNF-6 Binds to the G6Pase HNF-6 Motif-To provide evidence that HNF-6 is indeed the factor that is mediating the effect of PKA through the Ϫ129 to Ϫ85 promoter sequence, HNF-6 binding to this region was analyzed using the gel retardation assay. IPTG was used to induce the expression of a histidine-tagged form of mouse HNF-6 in bacteria, and then a soluble extract from these cells was incubated with a labeled double-stranded oligonucleotide representing the G6Pase HNF-6 motif ( Fig. 2A). No protein binding was detected in the gel retardation assay using lysate from non-IPTG-treated bacteria (data not shown); however, a single protein-DNA complex was detected using lysate from IPTG-treated cells ( Fig. 2A). Competition experiments, in which a 100-fold molar excess of unlabeled DNA was included with the labeled probe, were used to correlate protein binding with the PKA response. An oligonucleotide representing the wild-type G6Pase HNF-6 binding site competed effectively with the labeled probe for protein binding ( Fig. 2A). By contrast, an oligonucleotide that contains a mutation identical to that in the Ϫ129 HNF-6 SDM construct (Fig. 1) failed to compete ( Fig. 2A). Thus, the binding of HNF-6 to the G6Pase promoter ( Fig. 2A) correlates with the stimulatory effect of PKA on G6Pase-CAT fusion gene expression (Fig. 1).

The G6Pase HNF-6 Motif Can Confer a Direct Stimulatory Effect of PKA on the Expression of a Heterologous Fusion
Gene-To determine whether the G6Pase HNF-6 motif was sufficient to mediate a direct stimulatory effect of PKA on gene transcription, six copies of a double-stranded oligonucleotide representing the wild-type G6Pase promoter sequence between Ϫ114 and Ϫ99 (Fig. 1B) were ligated into the heterologous XMB expression vector (32). Transient co-transfection of the resulting construct, designated HNF-6 WT XMB, into HepG2 cells with the expression vector encoding PKA resulted in an approximately 6-fold stimulation of reporter gene expression (Fig. 2B). Similarly, three copies of a larger double-stranded oligonucleotide representing the wild-type G6Pase promoter sequence between Ϫ114 and Ϫ77 conferred a 4.12 Ϯ 0.19-fold induction (n ϭ 4) of reporter gene expression by PKA when ligated into the heterologous XMB expression vector (data not shown). To verify that the stimulatory effect of PKA was mediated through the HNF-6 binding site, five copies of a doublestranded oligonucleotide containing the same mutation (Fig.  1B) that abolished HNF-6 binding in gel retardation assays ( Fig. 2A) were ligated into the XMB vector. No basal reporter gene expression was detected when the resulting construct, designated HNF-6 MUT XMB, was transiently transfected into HepG2 cells. Furthermore, co-transfection with the expression vector encoding PKA failed to induce reporter gene expression (Fig. 2B). These results support a model in which HNF-6 is a target of PKA signaling.
HNF-6 Is Phosphorylated by PKA in Vitro-Rousseau and co-workers (39) have previously noted that rat HNF-6 contains five potential PKA phosphorylation sites, but only three of these strongly match the consensus PKA phosphorylation sequence (40). All five sites are perfectly conserved among human, rat, and mouse HNF-6. Fig. 3 shows that PKA can phosphorylate a histidine-tagged form of HNF-6 in a time-and concentration-dependent manner in vitro. The kinetics of the phosphorylation were markedly affected by the concentration of magnesium in the reaction (Fig. 3, compare A and B). In the presence of low magnesium (0.1 mM), PKA phosphorylates HNF-6 with an apparent K m of ϳ0.25 M (Fig. 3A). However, in the presence of high magnesium (5 mM), we were not able to calculate an apparent K m , because we were unable to add sufficient HNF-6 to the phosphorylation reaction to reach V max (Fig. 3B). In the presence of either low (0.1 mM) or high (5 mM) concentrations of magnesium, the time course of HNF-6 phosphorylation was similar with maximal phosphorylation by ϳ120 min (Fig. 3, C and D). However, the maximal incorporation of phosphate into HNF-6 was much greater when phosphorylation reactions contained 5 mM magnesium. A maximum stoichiometry of ϳ1.75 mol of phosphate/mol of HNF-6 was calculated in the presence of 5 mM magnesium. In contrast, in FIG. 2. HNF-6 binds to the G6Pase HNF-6 motif and confers a direct stimulatory effect of PKA on the expression of a heterologous fusion gene. A, a labeled double-stranded oligonucleotide representing the wild-type mouse G6Pase HNF-6 binding site (Fig. 1B) was incubated in the absence (Ϫ) or presence of a 100-fold molar excess of the unlabeled oligonucleotides shown, representing either the wild-type (WT) or mutated (MUT) G6Pase HNF-6 motif (Fig. 1B). Extract from IPTG-treated E. coli transformed with an expression vector encoding a 6x histidine-tagged HNF-6 fusion protein was then added, and protein binding was analyzed using the gel retardation assay as described under "Experimental Procedures." A representative autoradiograph is shown. B, HepG2 cells were transiently co-transfected as described under "Experimental Procedures" with heterologous XMB fusion genes (15 g), an expression vector encoding ␤-galactosidase (2.5 g), and either an expression vector (5 g) encoding PKA or the same vector (5 g) with the PKA open reading frame deleted. The heterologous XMB fusion genes were generated by ligating oligonucleotides representing either the wild-type or mutated G6Pase HNF-6 motif (Fig. 1B) into the HindIII site of the XMB vector in multiple (5-6) copies. After transfection, cells were incubated for 18 -20 h in serum-free medium. The cells were then harvested, and both CAT and ␤-galactosidase activity were assayed as described previously (29,30). Results are presented as the ratio of CAT activity, corrected for ␤-galactosidase activity in the cell lysate, in PKA-transfected versus empty vector-transfected cells (expressed as -fold induction). Results are the mean Ϯ S.E. of 3-4 experiments in which each construct was assayed in duplicate. No reporter gene expression was detected when the basic XMB vector was transiently transfected into HepG2 cells even in the presence of the PKA expression vector (data not shown). the presence of 0.1 mM magnesium, a maximum stoichiometry of ϳ0.3 mol of phosphate/mol of HNF-6 was obtained. Such magnesium-dependent variations in the kinetics and stoichiometry of phosphorylation by PKA have been reported for other PKA substrates (41)(42)(43).
The three consensus PKA phosphorylation sites in HNF-6 are located in the carboxyl-terminal region of the protein (39). Therefore, an expression vector was constructed that encoded a histidine-tagged carboxyl-terminal truncated form of HNF-6 in which these three putative PKA phosphorylation sites were deleted. The phosphorylation of this truncated form of HNF-6 by PKA was markedly reduced compared with the nontruncated form of the protein (Fig. 4). This result suggests that HNF-6 is phosphorylated in vitro by PKA predominantly on one or more of the carboxyl-terminal sites that match the consensus PKA phosphorylation sequence. Further support for this conclusion was obtained by constructing an expression vector that encoded a histidine-tagged full-length form of HNF-6 in which these three putative PKA serine phosphorylation sites were changed to alanine residues by site-directed mutagenesis (Fig. 4). The phosphorylation of this mutated form of HNF-6 by PKA was markedly reduced compared with the wild-type form of the protein (Fig. 4).   FIG. 3. HNF-6 is phosphorylated by PKA in vitro. The ability of PKA to phosphorylate a partially purified 6x histidine-tagged HNF-6 fusion protein was assessed as described under "Experimental Procedures." Phosphorylation reactions were analyzed by SDS-PAGE, and phosphate incorporation was quantitated by scintillation counting. A and B show the relationship between HNF-6 concentration and 32 P incorporation in the presence of 0.1 mM or 5 mM magnesium acetate, respectively. Phosphorylation reactions were performed for 5 (A) or 1 min (B) at room temperature. Under these conditions the rate of 32 P incorporation into HNF-6 was linear at all HNF-6 concentrations tested. C and D show the relationship between time and 32 P incorporation in the presence of 0.1 mM or 5 mM magnesium acetate, respectively. Phosphorylation reactions were performed using 1 M HNF-6. Each panel shows the mean data Ϯ S.E. from three experiments with a representative autoradiograph shown as an inset. Plots without error bars indicate that the S.E. values were smaller than the plot symbol.
To explore the functional consequence of mutating these three serine residues on PKA-stimulated G6Pase-CAT fusion gene expression, expression vectors encoding full-length wildtype and mutated HNF-6 were constructed. Fig. 5 shows that the co-transfection of the wild-type Ϫ129 G6Pase-CAT fusion gene construct with either of these expression vectors stimulated reporter gene expression to the same extent as that achieved by co-transfection with the expression vector encoding the catalytic subunit of PKA. Because phosphorylation by PKA has little effect on HNF-6 binding to DNA (data not shown), we hypothesize that it increases the transactivation potential of HNF-6 but that this effect is only apparent under conditions in which the concentration of HNF-6 is limiting. DISCUSSION Multiple promoter elements are required for the full stimulatory effect of the cAMP signal transduction pathway on G6Pase gene transcription in hepatoma cells that together comprise a CRU (27,36). The large induction of G6Pase-CAT fusion gene expression obtained by using the PKA co-transfection technique was critical for the delineation of such a multiple component CRU ( Fig. 1A; Ref. 27). Thus, in contrast, using cAMP analogs both Chou and co-workers (25) and Burchell and co-workers (26) reported the involvement of single elements in the cAMP response. Lin et al. (25) found that a region of the human G6Pase promoter encompassing the sequence between Ϫ136 and Ϫ134 was required for the stimulatory effect of cAMP on G6Pase fusion gene expression in HepG2 cells, whereas Schmoll et al. (26) found that the sequence located between Ϫ161 and Ϫ152 was critical for the combined stimulatory effects of cAMP and glucocorticoids in H4IIE hepatoma cells. The reason for these disparate results with the human promoter is unclear, but our data indicate that both of the equivalent regions in the mouse G6Pase promoter contribute to the induction of G6Pase-CAT fusion gene expression by PKA ( Fig. 1A; Ref. 27). However, even with both of these regions deleted, mouse G6Pase-CAT fusion gene expression was still induced by ϳ5-fold in response to PKA (Fig. 1A). Further truncation of the G6Pase promoter sequence between Ϫ129 and Ϫ85 resulted in a reduction in this stimulatory effect of PKA on G6Pase-CAT fusion gene expression (Fig. 1A). The data presented in this paper suggest that the stimulatory effect of PKA through the Ϫ129 to Ϫ85 region of the G6Pase promoter is mediated by the phosphorylation of HNF-6. Mutation of this HNF-6 motif in the context of an otherwise intact CRU has little effect on the induction of G6Pase-CAT fusion gene expression by PKA (data not shown). In contrast, mutation of the CRE, located between Ϫ162 and Ϫ155, in the context of an otherwise intact CRU almost abolishes the induction of G6Pase-CAT fusion gene expression by PKA in HepG2 cells (data not shown) and LLC-PK cells (27). These results are consistent with the 5Ј deletion analysis (Fig. 1A) that shows a much greater contribution of this CRE than the HNF-6 motif to the PKA response. Whether the relative contribution of the HNF-6 motif to the induction of G6Pase gene transcription by PKA increases under certain metabolic conditions and whether HNF-6 is important for the induction of other hepatic genes by PKA remains to be determined.
HNF-6 is a member of the ONECUT family of transcription factors that is characterized by a bipartite DNA binding domain consisting of a single cut domain and an atypical homeodomain (39). Classical homeodomains are 60 amino acids long and contain a conserved tryptophan and histidine at positions 48 and 50 of the homeodomain as opposed to the homeodomain in the ONECUT transcription factor family, in which the amino acid residues located at positions 48 and 50 are phenylalanine and methionine, respectively (37). The cut domain has been shown to be required for HNF-6 binding to DNA in all target FIG. 5. Overexpression of wild-type or mutated HNF-6 stimulates basal G6Pase-CAT fusion gene expression. HepG2 cells were transiently co-transfected as described under "Experimental Procedures" with a G6Pase-CAT fusion gene (15 g) containing the promoter sequence between Ϫ129 and ϩ66, an expression vector encoding ␤-galactosidase (2.5 g), and either an expression vector (5 g) encoding PKA (P) or the same vector (5 g) with the PKA open reading frame deleted (C). Cells were also co-transfected as indicated with 1 g of the empty pcDNA3 vector, the wild-type HNF-6 pcDNA3, the HNF-6 SDM pcDNA3 expression vectors described under "Experimental Procedures," or no further addition (None). After transfection, cells were incubated for 18 -20 h in serum-free medium. The cells were then harvested, and both CAT and ␤-galactosidase activity were assayed as described previously (29,30). Results are presented as a ratio relative to the CAT activity, corrected for ␤-galactosidase activity in the cell lysate, in empty PKA vector, no pcDNA3-transfected cells (expressed as -fold induction) and represent the mean Ϯ S.E. of 3-5 experiments assayed in duplicate. genes examined, whereas for a subset of HNF-6 target genes, the homeodomain seems to be dispensable for DNA binding (37). Both the cut domain and the homeodomain of HNF-6 are also involved in transcriptional activation by HNF-6 (37,44). Activation of HNF-6 target gene transcription on promoters that do not require the HNF-6 homeodomain for DNA binding involves the recruitment of the CREB-binding protein (44). The interaction of CREB-binding protein with rat HNF-6 requires an LXXLL motif (where L is a leucine residue and X is any amino acid) in the cut domain and the amino acid residues located at positions 48 and 50 (phenylalanine and methionine, respectively) of the homeodomain (44). The LXXLL motif has previously been shown to be important for the interaction of other transcription factors with CREB-binding protein (45). In contrast, activation of gene transcription by HNF-6 on target genes that require the homeodomain for DNA binding involves the recruitment of the coactivator p300/CREB-binding proteinassociated factor through an unidentified domain (44).
Of the three consensus PKA phosphorylation sites in HNF-6, one is located in the vicinity of the LXXLL motif in the cut domain (serine residue 309), and the other two are located in the homeodomain (serine residues 411 and 440). As described above, both the cut domain and the homeodomain are involved in DNA binding and in recruitment of coactivators, and thus the phosphorylation of HNF-6 by PKA could potentially have affected either or both parameters. Because phosphorylation by PKA has little effect on HNF-6 binding to DNA (data not shown), we hypothesize that it increases the transactivation potential of HNF-6. However, this putative effect of PKA-dependent phosphorylation on HNF-6 transactivation potential is only apparent under conditions in which the concentration of HNF-6 is limiting (Fig. 5). Thus, overexpression of either wildtype or mutated HNF-6 stimulates basal G6Pase-CAT fusion gene expression to the same extent as that achieved by cotransfection with the expression vector encoding PKA (Fig. 5). It may be possible to prove that PKA-dependent phosphorylation increases the transactivation potential of HNF-6 by analyzing the effect of mutating the three PKA phosphorylation sites in the context of an HNF-6 molecule in which the basal activation domains (37,44) have been mutated but only if these same domains are not also required for the PKA response. These observations are somewhat related to those recently reported by Quinn and co-workers (46,47), who investigated the relative contributions of different domains in CREB to transcription initiation. Of particular note is the observation that the constitutive activation domain in CREB mediates recruitment of the polymerase complex, whereas the kinase-inducible domain mediates later PKA-stimulated steps in transcription initiation. In the case of HNF-6 we hypothesize that when the HNF-6 binding site in the G6Pase promoter is fully occupied, the basal activation domains in HNF-6 are sufficient to mediate a maximal rate of transcription initiation. Interestingly, protein kinase C and casein kinase II phosphorylate the Cut/CCAAT displacement protein, a member of the superclass of cut homeodomain proteins, on residues located in the cut domain and alter the binding of the Cut/CCAAT displacement protein to DNA (48,49).
Because overexpression of HNF-6 stimulates G6Pase fusion gene expression (Fig. 5), this raises the possibility that hormones/metabolites could regulate G6Pase gene expression indirectly through an action on HNF-6 gene expression. There is circumstantial evidence that HNF-6 expression may be regulated by cAMP/PKA. Thus a CRE is present in the HNF-6 promoter (50) and growth hormone, which activates PKA in liver (51), stimulates HNF-6 expression (52). However, whether the CRE in the HNF-6 promoter contributes to this stimulation is unknown (50). The available data show that the effect of growth hormone on HNF-6 expression is mediated at least in part through increases in signal transducer and activator of transcription-5 and HNF-4 binding (53) and a decrease in CCAAT/enhancer-binding protein-␣ binding (54) to the HNF-6 promoter. The relative roles of growth hormone-stimulated PKA and Janus Kinase/signal transducer and activator of transcription activation in mediating these changes remain to be determined.
The phosphorylation of HNF-6 by PKA varied with changes in the magnesium ion concentration (Fig. 3). Other substrates are also differentially phosphorylated by PKA when the magnesium ion concentration is altered (41)(42)(43). Thus, Singh et al. (41,43) demonstrated that PKA phosphorylates a single site in the ␣ subunit of phosphorylase kinase at low magnesium ion concentrations; however, at high magnesium ion concentrations, PKA phosphorylates the ␣ subunit of phosphorylase kinase on three additional sites. Furthermore, Berglund et al. (42) showed that the optimal phosphorylation of L-type pyruvate kinase by PKA was found within a very narrow range of magnesium ion concentration. However, when the magnesium ion concentration was increased above 10 mM or decreased below 4 mM, PKA was less active (42).
In summary, the data presented in this manuscript demonstrate that an HNF-6 site located between Ϫ110 and Ϫ101 in the G6Pase promoter may contribute to the full stimulatory effect of PKA on G6Pase-CAT fusion gene expression. In addition, HNF-6 was shown to be phosphorylated by PKA in vitro. Taken together, these results suggest that the stimulatory effect of PKA on G6Pase gene transcription may be mediated in part by the phosphorylation of HNF-6.