C/EBPα and HNF6 protein complex formation stimulates HNF6-dependent transcription by CBP coactivator recruitment in HepG2 cells^†

Expression and Reporter Plasmids and Cotransfection Assays.

The cytomegalovirus (CMV) promoter expression vectors containing the HNF6 complementary DNA (cDNA) alone or tagged with green fluorescent protein (GFP) or V5 and the C/EBPα, C/EBPβ, or C/EBPδ expression vectors were previously described.17, 21, 22 The SV40 promoter pECE expression vectors containing mutant HNF6 ΔSTP box (Δ98-123) or HNF6 Δpolyhistidine cDNAs were a generous gift from Frédéric Lemaigre (Brussels, Belgium) and have been previously described.16 The CMV pCDNA3.1 expression vectors containing the hemagglutinin (HA)-tagged C/EBPα or HA-C/EBPα deletion mutants lacking either the transcriptional AD2 (amino acids Δ116-254) or AD1 sequences (119-360) and the C/EBPα K300E DNA binding mutant have been previously described.23, 24 The pGEX4-T-1 plasmid was used to express the glutathione-S-transferase (GST)-C/EBPα fusion proteins consisting of either AD1 (1-119), both AD1 and AD2 (1-226 or 1-286), the basic leucine zipper DNA binding domain (280-336) or the basic domain (313-336) and has been previously described.23 To generate the GST-C/EBPα AD2 (92-271) fusion protein, we blunted the BssHII fragment from the human C/EBPα cDNA into the SmaI site of the pGEX4T-1 expression plasmid. The pGEM1 HNF6 expression plasmids were used to synthesize radioactively labeled full-length or truncated HNF6 proteins and have been previously described.19 The HNF6-dependent reporter gene (6X HNF6 TATA luciferase plasmid) has been previously described.17, 19

Human hepatoma HepG2 cells, mouse hepatoma Hepa1-6 cells (ATCC #CRL-1830), or human osteosarcoma U20S cells were maintained in monolayer cultures and grown as previously described18, 25 and cotransfected using Fugene 6 reagent (Roche, Mannheim, Germany) according to the manufacturer's protocol. HepG2 or U2OS cells were transfected with 1.6 μg of the 6X HNF6 TATA luciferase plasmid and 200 ng of CMV C/EBPα (wild-type or mutant) or CMV C/EBPβ expression plasmid with or without 200 ng of the CMV HNF6 expression vector. We also performed cotransfection assays with 1.6 μg of the 6X HNF6 TATA luciferase reporter plasmid and 200 ng of CMV HNF6 and CMV C/EBPα with or without 200 ng of the CMV adenovirus E1A expression vector.

In Vitro GST-C/EBPα Pull-Down Assay, Western and Co-immunoprecipitation With Transfected HepG2 Nuclear Extracts.

Production and isolation of GST fusion proteins from BL21 Escherichia coli was performed as previously described.26 Binding buffer and wash buffer for the in vitro GST pull-down assay and co-immunoprecipitation assays with nuclear extracts were performed as previously described.19, 22 The immunoprecipitates were eluted by the addition of SDS sample buffer containing 1% β-mercaptoethanol and boiled for 5 minutes. Eluted proteins were resolved via SDS-PAGE and transferred to nitrocellulose membranes for Western blot analysis with either monoclonal antibody specific to GFP antibody (Clontech, Palo Alto, CA, for GFP-HNF6), HA epitope tag (Invitrogen, Carlsbad, CA, for HA-C/EBPα), V5 epitope tag (Invitrogen, for V5-HNF6), or C/EBPβ (sc-150 [C-19]; Santa Cruz Biotechnology, Santa Cruz, CA) and detected as previously described.19, 21, 27 For co-immunoprecipitation assays, 750 μg of total protein extract from quiescent mouse liver (0h) or regenerating mouse liver was immunoprecipitated with HNF627 or rabbit serum (Vector Laboratories, Burlingame, CA) and then subjected to Western analysis with rabbit anti-C/EBPα antibody (sc-61; Santa Cruz Biotechnology), C/EBPβ antibody (sc-150 [C-19]; Santa Cruz Biotechnology) or rabbit anti–CBP C-terminal antibody (Upstate, Lake Placid, NY). Peroxidase-conjugated ImmunoPure Recombinant Protein A/G (Pierce, Rockford, IL) was used to identify antibody–antigen bands and was detected via enhanced chemiluminescence (ECL-+ Amersham Pharmacia Biotech, Piscataway, NJ) followed by autoradiography.

Construction of Adenovirus With Inducible C/EBPα Expression and Adenovirus Purification.

The adenovirus (Ad)-tetracycline responsive element–HA-C/EBPα adenoviral construct was produced using the Adeno-X Tet-Off expression system by inserting the HA-tagged human C/EBPα cDNA (1-360) into the tetracycline responsive element shuttle vector following the manufacturer's instructions (BD Biosciences, Palo Alto, CA). The adenovirus vector used 7 copies of the tetracycline operator sequence linked to the minimal CMV promoter (CMV-TetO) to drive expression of the HA-C/EBPα protein. Induced expression was accomplished by coinfection with a second adenovirus containing the CMV promoter driving expression of the tetracycline transcriptional activator (AdCMV-TA), which is able to transcriptionally activate in the absence of doxycycline (Adeno-X Tet-Off system). Expression of HA-C/EBPα from this inducible adenovirus delivery system was detected via Western blot analysis with either the C/EBPα antibody or HA-tag antibody (Upstate). Recombinant adenoviruses were used to infect HEK 293A cells, and cell lysates were harvested at 72 hours postinfection. Generation and infection with the adenovirus containing the CMV-HNF6 cDNA expression cassette (AdHnf6) has been previously described,28 and adenovirus CMV GFP was purchased from BD Biosciences. The large-scale production of recombinant adenovirus particles was performed via infection of QBO-293 cells and subsequent purification following the protocol of Quantum Biotechnologies (Montreal, Canada). The number of infectious particle units was determined using the Adeno-X Rapid Titer Kit (BD Biosciences) according to the manufacturer's instructions.

Infection of Hepatoma Hepa1-6 Cells With Recombinant Adenoviruses and Chromatin Immunoprecipitation Assays.

Mouse hepatoma Hepa1-6 cells (1 × 107 cells per 150-mm dish) were infected at a multiplicity of infection of 10 infectious particle units per cell with AdHNF6 or AdC/EBPα (AdCMV-TetO-HA-C/EBPα and AdCMV-TA) either separately or together or with AdGFP control virus alone, as previously described.19, 29 We also used the Nucleofector II apparatus and buffers recommended by the manufacturer (Amaxa, Gaithersburg, MD) to transfect Hepa1-6 cells with CMV expression vectors containing either mouse HNF6 cDNA or rat C/EBPβ cDNA separately or together or with the CMV-GFP expression plasmid alone. The Nucleofector II system transfected approximately 60% of the Hepa1-6 cells as determined by electroporation of CMV GFP expression vector and counting the number of cells positive for GFP fluorescence (data not shown). At 24 hours after infection or electroporation, the cells were processed for chromatin immunoprecipitation (ChIP) assays using published methods.30, 31 The following antibodies were used for ChIP assays as previously described30: Hnf6 rabbit antibody,27 CBP rabbit antibody (sc-369 [A-22]; Santa Cruz Biotechnology), C/EBPα or C/EBPβ rabbit antibody (Santa Cruz Biotechnology; see above), or control rabbit serum (Vector Laboratories). Crosslinks were reversed on all chromatin samples via RNase A and proteinase K digestion, and DNA was then purified using polymerase chain reaction (PCR) purification columns according to the manufacturer's instructions (Qiagen, Valencia, CA) as previously described.30 The total input sample was diluted 1:10, and 2.5 μL was used for PCR (10% total input). To amplify the mouse Foxa2 promoter in ChIP assays, we used the −166 bp sense primer (ctcctgaagtcatcccacaagg) and −67 bp anti-sense primer (ggtgcccaaagcatttcgtaac) with 54°C annealing temperature. The following reaction mixture was used for all PCR samples: 1X of IQ SybrGreen Supermix (Biorad, Carlsbad, CA), 100 nmol/L of each primer, and 2.5 μL of each purified ChIP extract in a 25 μL total volume. Reactions were amplified and analyzed in triplicate using the MyiQ single-color real-time PCR detection system (Biorad). Normalization was performed using the ΔΔCT method as previously described.30, 31 The levels of either AdGFP-infected or CMV-GFP–transfected control samples were set at one. The experimental ChIP binding levels were expressed as a fold induction with respect to these control samples (relative promoter binding) ± SD.

ChIP Assays With Wild-Type Mouse Liver.

Liver tissue was isolated from 3 separate wild-type (WT) mice and washed in phosphate-buffered saline; the weight of each mouse liver was then recorded. The liver tissue was diced with a razor blade, homogenized in phosphate-buffered saline for 2 seconds with a PowerGen 125 Polytron (Fisher Scientific, Pittsburgh, PA), and fixed in 10 mL 1% formaldehyde (diluted in phosphate-buffered saline) for 15 minutes at room temperature with rotation. Glycine was added to 0.125 mol/L and suspension was incubated with rotation for an additional 5 minutes at room temperature to neutralize the formaldehyde. Fixed liver tissue was then pelleted 1,500-2,500 rpm at 4°C for 10 minutes. The cell pellet was homogenized for 1 minute in a Medimachine (Becton-Dickinson, Franklin Lakes, NJ) by resuspending a tissue pellet in 1 mL of cold phosphate-buffered saline containing protease inhibitor cocktail (Roche), then transferring it to a 50-μm medicon using an 18-gauge needle and syringe. Nuclei were recovered through the port in the medicon using a syringe and needle, and the homogenate was reapplied to the medicon 2 more times to purify liver nuclei. The liver nuclei were then pelleted and frozen at −80°C overnight. At this point, the chromatin was sonicated, and the ChIP assay was performed and analyzed as described above for cultured cells.

HNF6 Transcriptional Synergy Requires the C/EBPα AD1/AD2 Sequences and Is Abrogated by E1A-Mediated Inhibition of CBP Activity.

In this study, we examined whether HNF6 associates with another liver transcription factor to stimulate HNF6 activity. We cotransfected human hepatoma HepG2 cells with the HNF6-dependent reporter plasmid (6× HNF6 TATA luciferase) and the CMV HNF6 expression plasmid with or without expression vectors containing either HNF4α, fetoprotein transcription factor, C/EBPα, C/EBPβ, or C/EBPδ cDNAs. These cotransfection experiments revealed that only the C/EBPα expression vector stimulated HNF6 transcriptional activity (data not shown). Cotransfection assays with C/EBPα and HNF6 expression vectors and 6× HNF6 TATA luciferase plasmid demonstrated that C/EBPα stimulated HNF6-dependent transcription in either human hepatoma HepG2 or osteosarcoma U2OS cells (Fig. 1A -B). Furthermore, a DNA binding–deficient C/EBPα (K300E) mutant was able to stimulate HNF6-dependent transcription in cotransfection assays (Fig. 1A), suggesting that this transcriptional activation was independent of C/EBPα DNA binding activity. This HNF6 transcriptional activation was specific to the C/EBPα isoform because neither C/EBPβ nor C/EBPδ expression vectors provided significant stimulation of HNF6 transcriptional activity in HepG2 or U2OS cotransfection assays (Fig. 1A-B and data not shown).

Cotransfection studies with expression vectors containing C/EBPα deletion mutants demonstrated that both AD1 and AD2 sequences were essential for C/EBPα-HNF6 transcriptional synergy (Fig. 1C; C/EBPα ΔAD1, or ΔAD2). Because the AD1 sequences were shown to mediate transcriptional activation by the p300/CBP coactivators,11 we performed cotransfection assays with the CMV vector expressing the adenovirus E1A protein, which inhibits histone acetyltransferase activity of the p300/CBP coactivator proteins.32 Cotransfection assays showed that the E1A protein suppressed the ability of C/EBPα to stimulate HNF6 transcriptional activity (Fig. 1C), a finding that supports the hypothesis that C/EBPα stimulates HNF6 transcriptional activity by recruiting the CBP coactivator proteins. In contrast, cotransfection assays with the E1A and HNF6 expression vectors and the 6X HNF6 TATA Luciferase plasmid was unable to inhibit the basal transcriptional activity of HNF6, a finding consistent with previous studies.19 Taken together, these structure–function studies indicated that both the C/EBPα AD1 and AD2 sequences are essential for C/EBPα-HNF6 transcriptional synergy.

The HNF6 Polyhistidine and STP Box Sequences Are Essential for C/EBPα Transcriptional Synergy in Cotransfection Assays.

We examined whether deletion of either the N-terminal HNF6 transcriptional activation STP box domain (HNF6ΔSTP; Δ98-123) or the polyhistidine (HNF6ΔPH; Δ122-139) sequences were necessary for transcriptional synergy with the C/EBPα protein (Fig. 2A ). We performed HepG2 cotransfection studies with the 6X HNF6 TATA luciferase plasmid and expression vectors containing WT HNF6, HNF6ΔSTP, or HNF6ΔPH mutant proteins with or without the CMV C/EBPα expression construct. Consistent with published studies without C/EBPα,16 the HNF6ΔPH mutant displayed higher transcriptional activity than the WT HNF6 protein, while the HNF6ΔSTP mutant exhibited a 30% reduction in transcriptional activity (Fig. 2B). However, both the HNF6ΔSTP and HNF6ΔPH mutant proteins showed a reduction in HNF6-dependent transcriptional activity when combined with the CMV C/EBPα expression vector (Fig. 2B). These results demonstrated that retention of both the HNF6 N-terminal STP box and PH sequences was required for HNF6-C/EBPα transcriptional synergy.

HNF6 Associates With the C/EBPα Protein, and HNF6 Cut Domain and C/EBPα AD 1/2 Sequences Are Required for the Interaction.

To define sequences required for association between HNF6 and C/EBPα proteins, we performed pull-down experiments with GST C/EBPα fusion proteins and radioactively labeled in vitro transcription and translation HNF6 proteins as shown in Fig. 3A -B. The HNF6 expression plasmids was used to synthesize ³⁵S-methionine–labeled HNF6 proteins via in vitro transcription and translation (Fig. 3B, right panel), which were then used for binding reactions with various GST-C/EBPα fusion proteins (Fig. 3A) immobilized on glutathione-sepharose beads as previously described.19 These GST-C/EBPα pull-down assays demonstrated that labeled HNF6 proteins that retained the cut domain sequences (Fig. 3C, WT HNF6, Cut-Homeo and Cut) bound to the GST-C/EBPα AD1/AD2 fusion proteins (Fig. 3C, GST-C/EBPα proteins 2 and 3), and that deletion of the HNF6 cut–homeodomain sequences abrogated this binding (Fig. 3C). None of the labeled HNF6 proteins interacted with GST-C/EBPα fusion proteins containing only the AD1, the basic leucine zipper, or basic domain sequences (Fig. 3A,C, proteins 1, 4, and 5). Furthermore, labeled WT HNF6 protein bound weakly to the GST-C/EBPα AD2 fusion protein (Fig. 3D, protein 6), suggesting that both the C/EBPα AD1 and AD2 sequences were required for efficient HNF6 binding. Taken together, these findings demonstrate that formation of the C/EBPα-HNF6 complex requires the HNF6 cut domain and the AD1/AD2 C/EBPα sequences.

The C/EBPα Protein Co-immunoprecipitates With HNF6 in Transfected HepG2 Nuclear Extracts.

The C/EBPα protein is a potent inhibitor of cell proliferation23, 33; therefore, hepatoma cell lines do not express C/EBPα at high levels. To perform HNF6-C/EBPα co-immunoprecipitation (Co-IP) assays, we prepared nuclear extracts from HepG2 cells cotransfected with plasmids expressing either the GFP-tagged HNF6 protein19 or the HA-tagged C/EBPα protein.23 These transfected HepG2 nuclear extracts were subjected to immunoprecipitation with an HA-tag antibody followed by Western blot analysis using a monoclonal GFP antibody (Fig. 4A ). These Co-IP studies demonstrated that the GFP-HNF6 protein was able to associate with the HA-tagged C/EBPα protein (Fig. 4A, right panel). Co-IP assays with nuclear extracts prepared from HepG2 cells transfected with V5-tagged HNF621 and C/EBPβ22 expression vectors demonstrated that HNF6 protein failed to associate with the C/EBPβ protein (Fig. 4B, right panel). Western blot analysis of these transfected HepG2 nuclear extracts with the GFP, HA, V5, or C/EBPβ antibody showed protein expression from the transfected vectors (Fig. 4A-B, left panel, 10% input). These Co-IP results demonstrate that the HNF6 transcription factor associates with only the C/EBPα protein.

Functional Association Between the HNF6 and C/EBPα Transcription Factors and the CBP Coactivator.

To verify that HNF6 and C/EBPα transcription factors formed a complex with the CBP coactivator protein in vivo, we performed Co-IP assays with protein extracts from quiescent mouse liver or regenerating mouse liver isolated at 40 hours following partial hepatectomy, the latter of which represents the peak in hepatocyte DNA replication.34 Quiescent or regenerating mouse liver extracts were Co-IP with either the HNF6 or control rabbit serum; immunoprecipitated proteins were then subjected to Western blot analysis with either the C/EBPα or C/EBPβ antibody. This Co-IP experiment demonstrated that HNF6 and C/EBPα proteins formed a stable complex in extracts from both quiescent and regenerating liver, but HNF6 failed to interact with the related C/EBPβ protein (Fig. 5A ). Furthermore, Western blot analysis demonstrated that the CBP coactivator protein was Co-IP with C/EBPα in liver extracts using the HNF6 antibody (Fig. 5A). None of these transcription factors was immunoprecipitated by the control immunoblobulin G antibody (Fig. 5A). Taken together, these Co-IP studies with liver protein extracts demonstrated an association between the HNF6 and C/EBPα transcription factors and the CBP coactivator protein in vivo.

Increased Expression of HNF6 and C/EBPα Stimulates Their Binding and CBP Recruitment to the Endogenous Mouse Foxa2 Promoter.

To increase expression of C/EBPα in hepatoma cells, we generated an adenovirus vector that conditionally expressed the HA-tagged human C/EBPα protein under the control of CMV-TetO. Expression of HA-C/EBPα protein is induced by coinfection with a separate adenovirus containing the CMV promoter driving expression of AdCMV-TA, which is active in the absence of doxycycline (Tet-Off system). Coinfection of hepatoma cells with both AdCMV-TetO-HA-C/EBPα and AdCMV-TA (AdC/EBPα) is sufficient to induce expression of the HA tagged C/EBPα protein, as evidenced by Western blot analysis with antibody specific to either the HA tag or the C/EBPα protein (Fig. 5B). The proximal mouse Foxa2 promoter region contains a high affinity binding site for the HNF6 (−138 and −126 bp) and C/EBP (−84 to −74 bp) proteins as previously described.18, 22 To determine whether increased expression of both C/EBPα and HNF6 protein enhances association of these transcription factors and the CBP coactivator protein with the endogenous mouse Foxa2 promoter region, we used ChIP assays as previously described.30, 31

We prepared cross-linked chromatin from mouse hepatoma Hepa1-6 cells 24 hours after infection with AdHNF6 or AdC/EBPα, either alone or combined, and then sonicated chromosomal DNA into fragments between 500 to 1,000 nucleotides in length. We also prepared cross-linked and sonicated chromatin from Hepa1-6 cells that were efficiently transfected with CMV-HNF6 or CMV-C/EBPβ expression vector alone or combined by electroporation as described in Materials and Methods. This chromatin was then immunoprecipitated with antibodies specific to HNF6, CBP, C/EBPα, C/EBPβ or rabbit serum (control) and the immunoprecipitated genomic DNA was analyzed for the amount of mouse Foxa2 promoter DNA using quantative real-time PCR with primers specific to the mouse Foxa2 promoter region. We also performed ChIP assays with Hepa1-6 cells either infected with AdGFP or electroporated with CMV-GFP expression plasmid as normalization controls.

These hepatoma cell ChIP assays demonstrated that increased expression of both HNF6 and C/EBPα proteins stimulated association of either HNF6 (Fig. 6A) or C/EBPα (Fig. 6B) protein to the endogenous Foxa2 promoter region compared with hepatoma cells infected separately with AdHNF6 or AdC/EBPα (Fig. 6A-B). Interestingly, ChIP assays revealed that increased levels of both HNF6 and C/EBPα stimulated recruitment of the CBP coactivator protein to the endogenous Foxa2 promoter region compared with control hepatoma cells infected singly with AdHNF6 or AdC/EBPα (Fig. 6C). Although we were unable to demonstrate that HNF6 and C/EBPβ proteins interact using Co-IP assays with transfected cells or liver protein extract, cotransfection of both HNF6 and C/EBPβ expression vectors stimulated HNF6 binding to the endogenous Foxa2 promoter region (Fig. 6D), suggesting that these factors may interact in the context of the chromatin associated Foxa2 promoter region. However, cotransfection of both HNF6 and C/EBPβ expression vectors was unable to efficiently recruit either C/EBPβ (Fig. 6E) or the CBP coactivator protein (Fig. 6F) to the endogenous Foxa2 promoter region. ChIP assays with cross-linked and sonicated mouse liver tissue showed that HNF6, C/EBPα, and C/EBPβ bound at similar levels to the endogenous mouse Foxa2 promoter region (Fig. 6G). Interestingly, we observed reduced levels of CBP coactivator recruitment to the mouse liver Foxa2 promoter compared with that observed in Hepa1-6 cells with increased levels of HNF6 and C/EBPα proteins (compare Fig. 6C and 6G). These results are consistent with the inability of the HNF6-C/EBPβ complex to efficiently recruit the CBP coactivator protein to the endogenous Foxa2 promoter region. These studies support our hypothesis that formation of the C/EBPα-HNF6 complex stimulated binding of these transcription factors and recruitment of the CBP coactivator protein to the chromatin-associated Foxa2 promoter region.