Stability of the Hepatocyte Nuclear Factor 6 Transcription Factor Requires Acetylation by the CREB-binding Protein Coactivator*

We previously demonstrated that the formation of complexes between the DNA binding domains of the hepatocyte nuclear factor 6 (HNF6) and Forkhead Box a2 (Foxa2) transcription factors resulted in synergistic transcriptional activation of a Foxa2 target promoter. This Foxa2·HNF6 transcriptional synergy was mediated by the recruitment of CREB-binding protein (CBP) coactivator through the HNF6 Cut-Homeodomain sequences. Although the HNF6 DNA binding domain sequences are sufficient to recruit CBP coactivator for HNF6·Foxa2 transcriptional synergy, paradoxically these HNF6 Cut-Homeodomain sequences were unable to stimulate the transcription of an HNF6-dependent reporter gene. Here, we investigated whether the CBP coactivator protein played a different role in regulating HNF6 transcriptional activity. We showed that acetylation of the HNF6 protein by CBP increased both HNF6 protein stability and its ability to stimulate transcription of the glucose transporter 2 promoter. Mutation of the HNF6 Cut domain lysine 339 residue to an arginine residue abrogated CBP acetylation, which is required for HNF6 protein stability. Furthermore, the HNF6 K339R mutant protein, which failed to accumulate detected protein levels, was transcriptionally inactive and could not be stabilized by inhibiting the ubiquitin proteasome pathway. Finally, increased HNF6 protein levels stabilized the Foxa2 protein, presumably through the formation of the Foxa2·HNF6 complex. These studies show for the first time that HNF6 protein stability is controlled by CBP acetylation and provides a novel mechanism by which the activity of the CBP coactivator may regulate steady levels of two distinct liver-enriched transcription factors.

We previously demonstrated that the formation of complexes between the DNA binding domains of the hepatocyte nuclear factor 6 (HNF6) and Forkhead Box a2 (Foxa2) transcription factors resulted in synergistic transcriptional activation of a Foxa2 target promoter. This Foxa2⅐HNF6 transcriptional synergy was mediated by the recruitment of CREB-binding protein (CBP) coactivator through the HNF6 Cut-Homeodomain sequences. Although the HNF6 DNA binding domain sequences are sufficient to recruit CBP coactivator for HNF6⅐Foxa2 transcriptional synergy, paradoxically these HNF6 Cut-Homeodomain sequences were unable to stimulate the transcription of an HNF6-dependent reporter gene. Here, we investigated whether the CBP coactivator protein played a different role in regulating HNF6 transcriptional activity. We showed that acetylation of the HNF6 protein by CBP increased both HNF6 protein stability and its ability to stimulate transcription of the glucose transporter 2 promoter. Mutation of the HNF6 Cut domain lysine 339 residue to an arginine residue abrogated CBP acetylation, which is required for HNF6 protein stability. Furthermore, the HNF6 K339R mutant protein, which failed to accumulate detected protein levels, was transcriptionally inactive and could not be stabilized by inhibiting the ubiquitin proteasome pathway. Finally, increased HNF6 protein levels stabilized the Foxa2 protein, presumably through the formation of the Foxa2⅐HNF6 complex. These studies show for the first time that HNF6 protein stability is controlled by CBP acetylation and provides a novel mechanism by which the activity of the CBP coactivator may regulate steady levels of two distinct liver-enriched transcription factors.
Many of these liver transcription factors recruit the p300/ CREB-binding protein (CBP) family of histone acetyltransferases to activate transcription of hepatocyte-specific genes. These include the HNF4␣ (15,16), HNF1␣ (17,18), HNF6 (19), and C/EBP transcription factors (20). Furthermore, stimulation of HNF4␣ transcriptional activity is mediated by CBP acetylation of a lysine residue within the nuclear localization sequence of the HNF4␣ protein, causing an increase in its nuclear retention (21). The p300/CBP coactivator proteins stimulate gene transcription by acetylating positively charged amino groups of lysine residues on histone proteins and by interacting with the basal transcriptional machinery (22)(23)(24)(25)(26). Acetylation of histone proteins causes their partial dissociation from the DNA regulatory region, thus providing increased accessibility of the DNA regulatory regions to bind other transcription factors.
The mammalian Fox family of transcription factors consists of Ͼ50 mammalian proteins (3) that share homology in the winged helix DNA binding domain, which mediates monomeric recognition of its DNA target sequences (27). The Foxa1, Foxa2, and Foxa3 proteins are coexpressed in hepatocytes, share greater than 90% homology in their winged helix binding sequence, and recognize similar DNA target sequences (1,5,28). The Foxa proteins also display sequence homology within both their N-and C-terminal transcriptional activation domains (29,30) and are known to position nucleosomes within the albumin enhancer sequences in vivo (31,32).
During liver organogenesis, the mouse HNF6 protein is expressed at high levels in hepatoblasts adjacent to the portal mesenchyme and in the epithelial cells of the extrahepatic bile ducts and gall bladder but HNF6 exhibits lower expression in the hepatoblasts of the liver parenchyma (33). Consistent with this expression pattern, Hnf6 Ϫ/Ϫ mouse embryos also fail to develop a gall bladder and exhibit severe abnormalities in both extrahepatic and intrahepatic bile ducts, demonstrating that HNF6 plays a critical role in biliary and gall bladder development (33,34). The HNF6 transcription factor binds to DNA as a monomer utilizing a single Cut domain and a divergent homeodomain motif located at its C terminus (7)(8)(9)35). Recent NMR studies of the HNF6 DNA binding domain demonstrated that the Cut domain folds into a topology homologous to the POU DNA binding domain, even though there is no sequence homology between the Cut and Pou domain sequences (36). Transcriptional activation of an HNF6-dependent reporter gene was shown to involve both the HNF6 N-terminal STP Box, which is rich in serine, threonine, and proline residues, and the C-terminal Cut-Homeodomain sequences, the latter of which are necessary for recruitment of the CBP coactivator protein (19).
Both an LSDLL CBP-docking sequence in the HNF6 Cut domain and residues in the HNF6 homeodomain are necessary for recruitment of the CBP coactivator protein (19,37). Recent studies from our laboratory demonstrated that the formation of complexes between the DNA binding domains of the HNF6 and Foxa2 transcription factors resulted in synergistic transcriptional activation of a Foxa2 target promoter through the recruitment of the CBP coactivator protein by the Cut-Homeodomain sequences (37). Furthermore, we showed that HNF6⅐Foxa2 transcriptional synergy was abrogated by E1Amediated inhibition of CBP acetyltransferase activity and that the HNF6 LSDLL mutant protein (L330A) was unable to support this HNF6⅐Foxa2 transcriptional synergy (37). Although the HNF6 DNA binding domain sequences are sufficient to recruit CBP coactivator for HNF6⅐Foxa2 transcriptional synergy of a Foxa2 target promoter, paradoxically these Cut-Homeodomain sequences were unable to stimulate the transcription of a 6ϫ HNF6 TATA luciferase reporter gene (37). These results suggest that when the HNF6 Cut-Homeodomain binds to its target sequence, the HNF6 protein is unable to recruit p300/CBP coactivators and HNF6 may utilize its N-terminal activation domain to stimulate transcription of HNF6 target genes (19).
Here, we investigated whether the CBP coactivator protein played a different role in regulating HNF6 transcriptional activity. We show that CBP acetylation of the HNF6 protein results in a significant increase in HNF6 protein levels and stability and that retention of the HNF6 acetylation site is required for its protein stability. Furthermore, increased HNF6 protein levels stabilized the Foxa2 protein, presumably through the formation of the Foxa2⅐HNF6 complex. These studies show for the first time that HNF6 protein stability is controlled by CBP acetylation, suggesting that limiting the availability of the CBP protein will cause diminished accumulation of both the HNF6 and Foxa2 transcription factors.

MATERIALS AND METHODS
Expression Plasmids and Reporter Plasmids-Expression vectors consisted of the CMV promoter driving expression of the mouse HNF6 cDNA or the rat Foxa1, Foxa2, and Foxa3 cDNAs as described previously (8,29). The pECE HNF6 L330A mammalian expression plasmid in which the SV40 promoter drives the expression of the mutant HNF6 L330A cDNA was a generous gift from Frédéric Lemaigre (Brussels, Belgium) and was described previously (19). For V5-tagged constructs, rat Foxa2 cDNA, mouse HNF6 cDNA, and mouse HNF6 L330A mutant cDNA were PCR-generated, digested with EcoRI and NotI, and then ligated into the pcDNA V5/His expression vector (Invitrogen). The following sense and antisense primers were used for PCR amplification of the designated HNF6 coding regions: HNF6 full length-(1-465), sense 5Ј-CGCGAATTCATGAACGCACAGCTGACC-3Ј and antisense 5Ј-CGCGGATCCTGCTTTGGTACAAGTGCT-3Ј, and Foxa2 full length-(1-458), sense 5Ј-CGCGAATTCATGCTGGGAGCCGTGAAG-3Ј and antisense 5Ј-CGCTCTAGAGGACGAGTTCATAATAGG-3Ј. The Ϫ188-bp glucose transporter 2 (Ϫ188 Glut2) promoter contained two tandem HNF6 binding sites between Ϫ188 and Ϫ134 nucleotides and was previously shown to be transcriptionally activated by the HNF6 protein (38). We generated HNF6 amino acid point mutations using the Altered Sites II mammalian mutagenesis system (Promega). The 1.6-kb mouse HNF6 cDNA fragment was cloned into the EcoRI and NotI sites of the pAlter-Max vector. The following 5Ј-phosphorylated antisense site-directed mutation primer was used to convert the Lys residue at 339 to an Arg residue in the HNF6 coding region, 5Ј-P-ACGCGACTTGAGTCT-GCTCCAGGGCTT-3Ј. Mutagenesis was performed as described by the manufacturer, and the HNF6 point mutations were confirmed by sequencing (University of Chicago DNA Sequencing Facility). The HNF6 K339R mutant cDNA was cloned in into the pcDNA V5/His expression vector (Invitrogen).
Cell Culture and Transient Transfection-Human hepatoma HepG2 cells were maintained in monolayer cultures and grown in Ham's F-12 medium supplemented with 7% fetal calf serum, 100 units/ml penicillin/ streptomycin, 0.5ϫ MEM amino acids, and 0.5 units/ml human recombinant insulin as previously described (9). For transient transfection assays, cells were plated in six-well plates, grown in Dulbecco's modified Eagle's medium, 10% fetal calf serum, and 100 units/ml penicillin/ streptomycin and transfected using FuGENE 6 reagent (Roche Applied Science) according to the manufacturer's protocol. Cells were transfected with 100 ng of CMV-HNF6 or HNF6 mutant expression vector with 1.6 g of the Ϫ188 Glut2 promoter luciferase plasmid, which was previously shown to be an HNF6 target gene (38). Transfections included the 30 ng of CMV-Renilla internal control and were performed in the presence or absence of 100 ng of either CMV-CBP or CMV E1A. Protein extracts were isolated 24 h after transfection and assayed for dual luciferase activity as described previously (39). Results are expressed as the fold induction of transcriptional activity with respect to CMV empty expression vector derived from triplicate transfections.
Western Blot Analysis-Nuclear protein extracts were prepared from cultured HepG2 cells that were transiently transfected with V5 epitopetagged HNF6 WT or mutant expression vector with or without CMV Foxa2, CMV E1A, or CMV-CBP expression plasmid as described previously (37,38,40). 100 g of these nuclear extracts were subjected to 10% SDS-polyacrylamide gel electrophoresis, transferred to nitrocellulose membranes, and immunoblotted as described previously (41). Membranes were incubated with primary monoclonal antibodies (1:100 dilution) specific to either V5 epitope tag (Invitrogen), green fluorescent protein (GFP, BD Biosciences), or acetylated lysine residue (Upstate Biotechnology) in 5% milk at 4°C. Following washing in Tris-buffered saline buffer, membranes were incubated for 2 h with horseradish peroxidase-conjugated anti-mouse or anti-rabbit-IgG secondary antibody (Vector Laboratories) and detection was achieved using ECL Plus reagent (Amersham Biosciences) per the manufacturer's instructions. Membranes were stripped (42) and incubated for 1 h with polyclonal antibody against cyclin-dependent kinase 2 (Cdk2) (Santa Cruz Biotechnology, 1:200 dilution) in 5% milk at 4°C. Secondary antibody and detection are as described above. 24 h after transfection, cycloheximide (CHx) (ICN Pharmaceuticals) was added to the cells at a final concentration of 20 g/ml. The cells were harvested at 0, 2, 4, and 6 h following CHx treatment, and protein extracts were prepared at the indicated times with lysis buffer (20 mM HEPES, pH 7.9, 400 mM NaCl, 1 mM EDTA, 0.1% Nonidet P-40, 1 mM phenylmethylsulfonyl fluoride, and 10% glycerol). The extracts then were subjected to Western blot analysis as described above. We also used affinity-purified HNF6 antibody to analyze HNF6 protein levels during the cycloheximide decay experiments.

Increased HNF6 Protein Levels Are Mediated by CBP Acetylation-
Published studies have demonstrated that the CBP protein acetylated a lysine residue within the nuclear localization sequence of the HNF4␣ protein and that acetylation of this lysine residue was required to target all of the HNF4␣ protein to the nucleus (21). Therefore, we examined whether increased CBP protein levels stimulated acetylation of the HNF6 protein.
HepG2 cells were transiently transfected with the CMV HNF6 expression plasmid, which contained the HNF6 cDNA sequence fused with a C-terminal V5 epitope tag (CMV V5-HNF6) in the presence or absence of the CMV-CBP expression vector. We also included cotransfections of the E1A expression vector as a control to inhibit CBP acetyltransferase activity (43). Nuclear extracts were prepared from cells 24 h after transfection and then used for Western blot analysis with monoclonal antibodies specific to either an acetylated lysine residue to examine HNF6 protein acetylation or to the V5 epitope tag to measure the levels of transfected V5-HNF6 protein (Fig. 1A). This Western blot analysis demonstrated that the elevating levels of the CBP coactivator caused a significant increase in acetylation of the transfected V5-HNF6 protein and that E1A-mediated inhibition of CBP acetyltransferase activity abrogated the acetylation of the V5-HNF6 protein (Fig. 1A). CBP acetylation of the HNF6 protein also caused a significant increase in steady-state levels of the transfected V5-HNF6 protein (Fig. 1A). This increase in HNF6 protein levels was not due to CBP-mediated increase in CMV promoter activity, because expression levels of the CMV-GFP were unchanged by cotransfection with either CMV-CBP alone or in combination with the CBP and E1A expression vectors (Fig. 1B).
We next determined whether CBP-mediated increase in HNF6 protein levels influenced steady-state levels of the Foxa2 protein through the formation of HNF6 and Foxa2 protein complexes. Indeed, cotransfection of CMV GFP-HNF6 with CMV V5-Foxa2 increased steady-state levels of transfected V5-Foxa2 protein (Fig. 1C). The steady-state levels of V5-Foxa2 protein were further increased when we cotransfected both the CMV GFP-HNF6 and CMV-CBP expression vectors with the CMV V5-Foxa2 plasmid (Fig. 1C). Western blot analysis demonstrated no detectable acetylation of the Foxa2 protein (data not shown), suggesting that Foxa2 was not a target for CBPmediated acetylation. These results suggest that CBP acetyla-tion of HNF6 protein resulted in stimulating levels of the HNF6 protein, which in turn caused elevated Foxa2 protein levels, presumably because of the formation of the HNF6⅐Foxa2 protein complex.
Because CBP acetylation influenced the nuclear retention of the HNF4␣ protein (21), we determined whether the increased levels of CBP or E1A protein altered the nuclear distribution of the transfected V5-HNF6 protein. Cotransfection of HepG2 cells with the CMV V5-HNF6 expression vector and either CMV-CBP or CMV E1A did not alter the nuclear localization of the V5-HNF6 protein as visualized by immunofluorescence with the V5 monoclonal antibody (Fig. 2, A-C). Furthermore, nuclear staining was observed with the mutant V5-HNF6 L330A protein (Fig. 2D), suggesting that retention of the CBP docking sequence was not necessary for HNF6 nuclear localization. These results suggested that CBP acetylation of the HNF6 protein increased its steady-state levels without influencing nuclear localization of the HNF6 protein, a finding that differs from published studies regarding the acetylation of the HNF4␣ transcription factor (21).
Retention of the CBP Acetylation Site in the HNF6 Cut Domain Is Required for HNF6 Protein Stability and Transcriptional Activity-We next scanned the HNF6 protein sequence with the p300/CBP lysine acetylation consensus sequence (G/ S)K (44) and found one potential p300/CBP acetylation site within the HNF6 Cut DNA binding domain at lysine 339, which is adjacent to the LSDLL-(327-331) CBP docking sequence (Fig. 3A). Published studies demonstrated that the mutation of a Lys residue to an Arg residue in the E2F transcription factors was sufficient to inhibit p300/CBP-mediated acetylation (45). To disrupt the CBP acetylation sequence in the HNF6 Cut domain, we used oligonucleotide-mediated sitedirected mutagenesis to change the Lys residue at amino acid 339 in the Cut domain to an Arg residue within the context of full-length V5-HNF6 protein. We transfected expression vectors containing either V5-HNF6 WT or V5-HNF6 K339R mutant proteins with or without CMV-CBP expression vector. We also included transfections with the V5-HNF6 L330A expres-FIG. 1. CBP acetylates HNF6 protein, mediating increased HNF6 protein levels. A, cotransfection studies demonstrate that CBP acetylates HNF6 protein, increasing HNF6 protein levels. HepG2 cells were transiently transfected with CMV V5 epitope-tagged HNF6 (CMV V5-HNF6) expression plasmid in the presence or absence of the CMV-CBP expression vector. We also included cotransfected CBP and the E1A expression vectors as a control, because E1A protein inhibits CBP acetyl transferase activity. Nuclear extracts were prepared from cells 24 h after infection and then used for Western blot analysis with monoclonal antibodies specific to either acetylated lysine residue to examine protein acetylation or V5 epitope tag to measure the levels of transfected HNF6 protein. Cdk2 protein levels serve as a protein loading control. B, CBP protein did not stimulate CMV-driven expression of the GFP protein. HepG2 cells were transfected with the CMV GFP expression vector either alone or together with CMV-CBP to show that the CBP protein did not stimulate CMV-driven expression of the GFP protein as detected by Western blot analysis. C, cotransfection studies demonstrate that increased levels of GFP-HNF6 and CBP protein stimulate Foxa2 protein levels. HepG2 cells were transiently transfected with CMV V5-Foxa2 expression plasmid in the presence or absence of either CMV GFP-HNF6 either alone or together with CMV-CBP expression vector. mAb, monoclonal antibody; Ab, antibody.

FIG. 2. HNF6 nuclear localization is not influenced by cotransfection with either CBP or E1A expression vectors.
A-C , nuclear staining of V5-HNF6 protein is unchanged following cotransfection with either CBP or E1A expression vectors. D, nuclear staining of V5-HNF6 L330A mutant protein that fails to recruit the CBP coactivator protein. The V5 monoclonal antibody was used for immunofluorescent staining of transfected V5-HNF6 protein.
sion vector to determine whether this CBP-docking sequence was required for efficient acetylation of the HNF6 protein in vivo.
Nuclear extracts were prepared from these transfected cells and then used for Western blot analysis with monoclonal antibodies specific to either an acetylated lysine residue or V5 epitope tag to measure the levels of transfected HNF6 protein (Fig. 3B). This Western blot analysis demonstrated that increased CBP levels caused a significant increase in both the steady-state levels and acetylation of either transfected V5-HNF6 WT or V5-HNF6 L330A proteins (Fig. 3B). However, in the absence of exogenous CBP coactivator, the V5-HNF6 L330A mutant protein is not detected by the acetylated lysine antibody, suggesting that the HNF6 CBP LSDLL-docking sequence may be required for efficient acetylation when CBP levels are limiting (Fig. 3B). In contrast, the retention of the HNF6 K339 residue was required for CBP acetylation and this HNF6 K339R mutant protein failed to accumulate detectable protein levels (Fig. 3B). Furthermore, the V5-HNF6 K339R mutant protein was undetectable by immunofluorescence in transfected cells (data not shown), indicating that this acetylation site did not influence nuclear levels of the HNF6 protein.
These results demonstrate that HNF6 protein stability requires acetylation by the CBP coactivator.
To examine whether CBP acetylation of HNF6 protein is required for transcriptional activation, we performed cotransfection assays with the Ϫ188 Glut2 promoter, which was previously shown to be transcriptionally activated by HNF6 expression vector in cotransfection assays (38). We transfected HepG2 cells with the Ϫ188 Glut2 promoter luciferase plasmid and CMV expression vectors containing V5-HNF6 WT, V5-HNF6 L330A mutant, or HNF6 K339R mutant proteins. We also performed cotransfection assays in the presence or absence of the CMV-CBP or CMV E1A (an inhibitor of CBP acetyltransferase activity) expression vectors. Protein extracts were isolated 24 h after transfection and assayed for dual luciferase activity. These transfection studies demonstrated that HNF6 provided a 16-fold increase in Glut2 promoter activity and, including the CBP expression vector, stimulated the ability of HNF6 to activate transcription of the Glut2 promoter (Fig. 3C). Cotransfection of the E1A expression vector, which inhibits CBP-mediated stabilization of HNF6 protein, diminished the ability of HNF6 to activate transcription of the Glut2 promoter (Fig. 3C). Likewise, the HNF6 K339R acetylation site mutant protein, which is an inherently unstable protein, was unable to activate the transcription of the Glut2 promoter (Fig. 3C). Consistent with published studies (19), we observe a significant reduction in HNF6-transcriptional activity with the HNF6 L330A mutant protein, suggesting that the LSDLL sequences contribute to HNF6-transcriptional activity (Fig. 3C).
To examine whether the instability of the V5-HNF6 K339R mutant protein was mediated by the proteasome degradation pathway, we transfected expression vectors containing either V5-HNF6 WT or mutant V5-HNF6 L330A or V5-HNF6 K339R proteins with or without treatment with the proteasome inhibitor MG132. Nuclear extracts were prepared from cells 24 h after transfection, and then the levels of transfected HNF6 protein were measured by Western blot analysis with antibody specific to the V5 epitope tag (Fig. 4). This analysis revealed that steady-state levels of the V5-HNF6 K339R protein remained undetectable following treatment with the proteasome inhibitor MG132 (Fig. 4). Likewise, steady-state levels of either V5-HNF6 WT or mutant V5-HNF6 L330A proteins were not increased by MG132 treatment (Fig. 4). These results suggested that the HNF6 K339R mutant protein could not be

FIG. 3. Retention of the CBP acetylation site in the HNF6 Cut domain is required for HNF6 protein stability and transcriptional activity. A, amino acid sequence of the HNF6 Cut domain
showing the location of the CBP binding site and a potential CBP acetylation sequence. B, the HNF6 K339R acetylation site mutant protein fails to accumulate detectable protein levels. HepG2 cells were transiently transfected with CMV expression vectors containing V5-HNF6 WT, V5-HNF6 L330A mutant, or HNF6 K339R mutant proteins in the presence or absence of the CMV-CBP expression vector. Nuclear extracts were prepared from cells 24 h after infection and then used for Western blot analysis with monoclonal antibodies specific to either acetylated lysine residue to examine protein acetylation or V5 epitope tag to measure the levels of transfected HNF6 protein. Cdk2 protein levels serve as a protein loading control. C, cotransfection of CBP stimulates HNF6 transcriptional activity, whereas HNF6 K339R mutant is transcriptionally inactive. HepG2 cells were transiently transfected with the Ϫ188 Glut2 promoter luciferase plasmid and CMV expression vectors containing V5-HNF6 WT, V5-HNF6 L330A mutant, or HNF6 K339R mutant proteins. We also performed cotransfection assays in the presence or absence of the CMV-CBP or CMV E1A (inhibitor of CBP acetyltransferase activity) expression vectors. Protein extracts were isolated 24 h after transfection and assayed for dual luciferase assay. Results are expressed as the fold induction of transcriptional activity with respect to CMV empty expression vector derived from triplicate transfections. mAb, monoclonal antibody; Ab, antibody. stabilized through the inhibition of the proteasome degradation pathway and that acetylation of the HNF6 protein was not protecting the Lys residue from ubiquitination and subsequent proteasomal-dependent degradation.

CBP Acetylation of the HNF6 Protein Increases the Half-life of the HNF6 and Foxa2
Proteins-We next wanted to examine whether increased levels of CBP caused an increase in the half-life of transfected V5-HNF6 protein and whether stability of the transfected V5-Foxa2 protein was influenced through stabilization by the HNF6 protein. To determine whether CBP acetylation increased the stability of either the V5-HNF6 or V5-Foxa2 proteins, protein synthesis was inhibited in transfected cells by CHx, and protein decay rates were measured at various time points thereafter. For these protein stability studies, HepG2 cells were transfected with CMV V5-HNF6 either in the absence or presence of the CMV-CBP expression vector (Fig. 5, A and B) or were transfected with CMV V5-Foxa2 in the absence or presence of the CMV-HNF6 expression vector (Fig.  5, C and D). HepG2 cells were also transfected with CMV V5-Foxa2 together with CMV HNF6 and CMV-CBP expression vectors (Fig. 5E). 24 h following transfection, protein synthesis was inhibited with CHx treatment and nuclear extracts were prepared at 0, 2, 4, and 6 h following CHx addition. Western blot analysis was performed with the V5 epitope antibody to measure protein decay rates of either V5-HNF6 or V5-Foxa2 after inhibition of protein synthesis under the various combinations described above (Fig. 5).
We found that, in the absence of exogenous CBP coactivator, an 80% decline in V5-HNF6 protein levels occurred within 2 h following inhibition of protein synthesis by CHx (Fig. 5A). In contrast, cotransfection of CMV-CBP and CMV V5-HNF6 expression vectors caused a significant increase in the stability of the transfected V5-HNF6 protein (Fig. 5B). In the presence of exogenous CBP protein, we found no significant decreases in HNF6 protein levels at 2 h following CHx addition and a 60% reduction in HNF6 protein levels at the 4-h time point (Fig.  5B). Similar to the HNF6 protein, a 65% decrease in V5-Foxa2 protein levels was found at 2 h following CHx treatment (Fig.  5C). Furthermore, cotransfection of V5-Foxa2 and HNF6 expression vectors caused a significant increase in the stability of the V5-Foxa2 protein, presumably because of the formation of the HNF6⅐Foxa2 protein complex (Fig. 5D). Likewise, Foxa2⅐HNF6 complex formation caused increased stability of the HNF6 protein compared with the V5-HNF6 protein alone (Fig. 5, compare D with A). Moreover, cotransfection of V5-Foxa2, HNF6, and CBP expression vectors further stimulated the steady-state levels and the stability of the transfected V5-Foxa2 protein (Fig. 5E). These results suggest that HNF6 protein stability was increased by CBP acetylation and that formation of the HNF6⅐Foxa2 protein complex is probably involved in stimulating Foxa2 protein stability. DISCUSSION We previously demonstrated that formation of complexes between the DNA binding domains of the HNF6 and Foxa2 transcription factors resulted in synergistic transcriptional activation of a Foxa2 target promoter through the recruitment of CBP by the HNF6 Cut-Homeodomain sequences (37). In this study, we determined that CBP acetylation of the HNF6 protein resulted in a significant increase in stability and steadystate levels of the HNF6 protein. Furthermore, the HNF6 K339A mutant protein was not acetylated by CBP, failed to accumulate detectable protein levels, and was transcriptionally inactive on the Glut2 promoter (Fig. 3), a known target gene for the HNF6 transcription factor (38). Likewise, the E1A inhibition of CBP acetyltransferase activity resulted in diminished HNF6 protein levels and diminished HNF6-dependent tran-scriptional activation of the Glut2 promoter. Furthermore, cotransfection of CBP and HNF6 expression vectors stimulated the ability of HNF6 to activate the transcription of the Glut2 promoter. Interestingly, the HNF6 acetylation sequence in the Cut domain was conserved between the related OC-2 and OC-3 and the HNF6 (OC-1) proteins (35,46), implying that stabilization of these other ONECUT family proteins may involve CBP acetylation. These studies show for the first time that HNF6 protein stability is controlled by CBP acetylation and provide a novel mechanism by which the CBP coactivator activity may modulate steady-state levels of a tissue-specific transcription factor (Fig. 6).
We also found that the elevated expression of HNF6 protein caused increased stability and steady-state levels of the Foxa2 protein, which is probably occurring through the formation of To determine whether CBP acetylation increased the stability of either the V5-HNF6 or V5-Foxa2 proteins, protein synthesis was inhibited in transfected cells by CHx and protein decay rates were measured at various time points thereafter. For these studies, HepG2 cells were transfected with CMV V5-HNF6 either in the absence (A) or presence (B) of CMV-CBP. We also transfected HepG2 cells with CMV V5-Foxa2 without (C) or with CMV-HNF6 (D) or together with CMV-CBP (E) expression vector. 24 h following transfection, protein synthesis was inhibited with CHx and nuclear extracts were prepared at 0, 2, 4, and 6 h following CHx addition. Western blot analysis was performed with the V5 epitope antibody to measure protein decay rates of V5-HNF6 or V5-Foxa2 in the above conditions after inhibition of protein synthesis. In cycloheximide decay experiments with HepG2 cells transfected with CMV V5-Foxa2 and CMV-HNF6 (D), we also performed Western blot analysis with HNF6 antibody to determine whether the formation of the Foxa2⅐HNF6 complex increased steady-state levels of HNF6. Western blot signals were normalized to Cdk2 internal control, and relative protein levels were compared with the 0-h time point. the Foxa2⅐HNF6 complex. Cotransfection of both HNF6 and CBP expression vectors further increased Foxa2 protein levels through enhanced protein stability. Furthermore, no detectable acetylation of the Foxa2 protein was found by Western blot analysis with the monoclonal antibody specific to acetylated lysine (data not shown), suggesting that Foxa2 was not a target for CBP-mediated acetylation. These data support the hypothesis that increased activity of CBP will not only increase stability of HNF6 but may also indirectly cause increased levels of Foxa2 protein, presumably through formation of the HNF6⅐Foxa2 complex (Fig. 6). In addition to the Foxa2⅐HNF6 transcriptional synergy of Foxa target genes, these results imply that HNF6 protein levels regulate Foxa2 protein stability and thus HNF6 indirectly controls transcription of Foxa2 target genes.
CBP-mediated acetylation of other transcription factors has been shown to increase their transcriptional activity by stimulating either nuclear localization or protein stability. For example, CBP acetylation of a Lys residue within the nuclear localization sequence of the HNF4␣ protein increased its transcriptional activity by stimulating nuclear retention of the HNF4␣ protein (21). Inhibition of HNF4␣ acetylation by either mutating the HNF4␣ acetylation site or cotransfecting a dominant negative CBP protein caused cytoplasmic localization of ϳ50% of the transfected HNF4␣ protein, and this was mediated by the CRM1 nuclear export pathway (21). However, inhibiting acetylation of the HNF4␣ protein did not influence its protein stability but rather caused diminished nuclear levels of HNF4␣ protein (21). Although nuclear localization sequences are found in both the HNF6 Cut and Homeodomain sequences (37), we found that CBP acetylation of the HNF6 Cut domain sequences did not influence its nuclear localization (Fig. 2), a finding that is distinct from the published data regarding the HNF4␣ protein (21). Instead, CBP-mediated acetylation of the HNF6 protein is essential for HNF6 protein stability and HNF6 protein acetylation significantly increased its protein half-life and transcriptional activity (Figs. 3, 5, and 6). Likewise, stabilization of the sterol regulatory element-binding protein was mediated by CBP acetylation. In contrast to the HNF6 protein, mutation of the sterol regulatory element-binding protein acetylation sites caused enhanced protein stability by preventing its ubiquitination and subsequent degradation through the proteasomal-dependent pathway (47). Furthermore, we showed that MG132-mediated inhibition of the ubiquitin proteasome pathway was unable to stabilize the unacetylated HNF6 K339R mutant protein and did not increase levels of the WT HNF6 protein. Taken together, our results suggest that regulation of the HNF6 protein by CBP acetylation involves a mechanism that is distinct from that found with the sterol regulatory element-binding protein and HNF4␣ proteins and that degradation of the HNF6 K339R mutant protein does not involve the ubiquitin proteasome pathway.
Our data suggest that stimulating the activity or availability of the CBP coactivator will increase levels of the HNF6 protein.
Phosphorylation of CBP by Cdk, p42/44 mitogen-activated protein kinase/ERK1, or S6 kinase p90Rsk is known to stimulate CBP histone acetyltransferase activity in proliferating cells (48 -52). These results suggest the hypothesis that increased CBP activity in proliferating cells may cause elevated HNF6 protein levels and that HNF6 may also regulate proteins involved in proliferation. Consistent with this concept, published chromatin immunoprecipitation studies with cross-linked chromatin from hepatoma cells and an HNF6-specific antibody have demonstrated that the HNF6 protein binds to several endogenous promoters involved in proliferation such as Cdc25A and Cdk2 (53). Furthermore, many of the other liver transcription factors, including HNF1␣, HNF4␣, and C/EBP proteins, are known to recruit the CBP coactivator to their activation domains to stimulate the transcription of hepatocyte-specific genes (15)(16)(17)(18)20). This finding suggests the possibility that CBP recruitment by these other liver transcription factors may limit the availability of the CBP coactivator and may lead to diminished acetylation and protein levels of the HNF6 protein.
Previous studies had determined that an LSDLL sequence in the HNF6 Cut domain was essential for mediating recruitment of the CBP coactivator (19). Recent studies from our laboratory demonstrated that HNF6⅐Foxa2 transcriptional synergy was abrogated by E1A-mediated inhibition of CBP acetyltransferase activity (37). Furthermore, the HNF6 LSDLL mutant protein (L330A) was unable to recruit the CBP protein and failed to synergize with Foxa2 to activate transcription (37). Although the HNF6 DNA binding domain sequences are sufficient to recruit CBP coactivator for HNF6⅐Foxa2 transcriptional synergy of a Foxa2 target promoter, the HNF6 Cut-Homeodomain sequences were unable to stimulate transcription of a 6ϫHNF6 TATA luciferase reporter gene (37). These results suggested the hypothesis that HNF6 Cut domain LSDLL sequence is unable to recruit the CBP protein when HNF6 Cut-Homeodomain is involved in binding to its DNA target sequence. This is supported by NMR structural studies of the HNF6 Cut-Homeodomain showing that the LSDLL sequence resides in helix 3 of the HNF6 Cut domain, which is essential for the hydrophobic packing of the Cut domain structure (36). These structural studies suggest that LSDLL sequence may be inaccessible to recruit the CBP coactivator protein when HNF6 protein is bound to DNA, because the Cut domain helix 3 containing the LSDLL motif is predicted to be partially in the major groove of the DNA. Our studies are consistent with previous studies (19), demonstrating that HNF6 L330A mutant protein is unable to activate HNF6-dependent transcription, and suggest the hypothesis that disruption of the LSDLL motif may influence the ability of HNF6 Cut domain to efficiently recognize its DNA target sequence. This implies that the HNF6 Cut-Homeodomain sequences are only capable of recruiting the CBP coactivator when the HNF6 protein is tethered to the Foxa2 DNA binding sequence and not when HNF6 protein binds to its own target sequence. Our current studies are consistent with this hypothesis, demon-FIG. 6. Model summarizing CBP acetylation of HNF6 mediates increase stability of HNF-6 and Foxa2 proteins. Our model is that increased stability of the HNF6 protein is mediated by CBP-mediated acetylation. Formation of the HNF6⅐Foxa2 complex is probably involved in the ability of HNF6 to increase steady-state levels and stability of the Foxa2 protein. Our studies provide a novel mechanism by which activity of the CBP coactivator may regulate steady levels of two distinct liver-enriched transcription factors. strating that CBP plays a different role in regulating transcriptional activation of HNF6-dependent target genes. We show that the CBP acetylation of the HNF6 protein is required for its transcriptional activity and protein stability.