CCAAT/Enhancer-binding Protein (cid:1) (C/EBP (cid:1) ) Activates Transcription of the Human Microsomal Epoxide Hydrolase Gene ( EPHX1 ) through the Interaction with DNA-bound NF-Y*

Microsomal epoxide hydrolase (mEH) plays a central role in xenobiotic metabolism as well as mediating the sodium-dependent uptake of bile acids into the liver, where these compounds regulate numerous biological processes such as cholesterol metabolism and hepatocyte signaling pathways. Little is known, however, about the factors that control the constitutive and inducible expression of the mEH gene ( EPHX1 ) that is altered during development and in response to numerous xenobiotics. In previous studies we have established that GATA-4 binding to the EPHX1 core promoter is critical for EPHX1 expression. The (cid:2) 80/ (cid:3) 25 bp core promoter also contained a reversed CCAAT box ( (cid:2) 5/ (cid:2) 1 bp), integrity of which was required for maximal basal EPHX1 transcription in HepG2 cells. Transient transfection of CCAAT/enhancer-binding protein (cid:1) (C/EBP (cid:1) ) substantially stimulated EPHX1 promoter activity. Electrophoretic mobility shift assays, however, revealed that nuclear factor Y (NF-Y), but not C/EBP (cid:1) , directly bound to this site although increased expression of NF-Y had no effect on EPHX1 promoter activity. These results suggested that C/EBP (cid:1) activated EPHX1 expression through its interaction with NF-Y

Microsomal epoxide hydrolase (mEH) plays a central role in xenobiotic metabolism as well as mediating the sodium-dependent uptake of bile acids into the liver, where these compounds regulate numerous biological processes such as cholesterol metabolism and hepatocyte signaling pathways. Little is known, however, about the factors that control the constitutive and inducible expression of the mEH gene (EPHX1) that is altered during development and in response to numerous xenobiotics. In previous studies we have established that GATA-4 binding to the EPHX1 core promoter is critical for EPHX1 expression. The ؊80/؉25 bp core promoter also contained a reversed CCAAT box (؊5/؊1 bp), integrity of which was required for maximal basal EPHX1 transcription in HepG2 cells. Transient transfection of CCAAT/enhancer-binding protein ␣ (C/EBP␣) substantially stimulated EPHX1 promoter activity. Electrophoretic mobility shift assays, however, revealed that nuclear factor Y (NF-Y), but not C/EBP␣, directly bound to this site although increased expression of NF-Y had no effect on EPHX1 promoter activity. These results suggested that C/EBP␣ activated EPHX1 expression through its interaction with NF-Y bound to the CCAAT box. The existence of a C/EBP␣[NF-Y] complex was supported by electrophoretic mobility shift assays using antibodies against NF-Y and C/EBP␣ as well as by the ability of a dominant-negative NF-Y expression vector to inhibit promoter activity. The interaction between these transcription factors was established by co-immunoprecipitation analysis and glutathione S-transferase pull-down assays, whereas the association of the two factors and the interaction of NF-Y with the CCAAT box in vivo was confirmed by chromatin immunoprecipitation assays. C/EBP␣-dependent EPHX1 activation was also supported by reconstitution studies in HeLa cells that lack this protein. These results establish that EPHX1 expression is regulated by C/EBP␣ interacting with DNA-bound NF-Y.
Microsomal epoxide hydrolase (mEH) 1 is a bifunctional protein that is expressed with two distinct topological orientations on the hepatic endoplasmic reticulum (1,2), where it plays a central role in the metabolism of many carcinogenic xenobiotics (3). One of these topological forms is targeted to the plasma membrane, where it mediates the uptake of bile acids (4 -10). Bile acids play a critical role in the digestion of dietary lipids, excretion of cholesterol and metabolized xenobiotics, cholesterol metabolism, and the modulation of hepatocyte signaling pathways. The greatly reduced expression of mEH resulting from naturally occurring mutations at a 5Ј HNF-3␤ (FOXA2) element and an intron 1 putative USF site in the human mEH gene (EPHX1) has been shown to result in defects in bile acid uptake (11), leading to highly elevated serum bile acid levels (hypercholanemia). In addition, polymorphisms in exon 3 and 4 (12) that affect mEH stability (13), when combined with the intron 1 mutation, can affect the metabolism of numerous compounds such as polyaromatic hydrocarbon carcinogens, resulting in alterations in the susceptibility to various forms of cancer such as colorectal adenomas (14). Despite the important role that mEH plays in xenobiotic detoxification and bile acid transport, little is known about the factors that regulate EPHX1 expression, which is altered during development (7,15) and in response to various chemicals such as carcinogens and bile acids (16,17). In recent studies we have demonstrated that GATA-4 binding to two cis elements in the core promoter plays a critical role in mediating the constitutive expression of EPHX1 (18).
Sequence analysis of the EPHX1 core promoter region (Ϫ80 to ϩ25 bp) also revealed a reversed CCAAT box, which is a frequently occurring cis-acting element in eukaryotic promoters that has been shown to occur close to the transcription initiation site in TATA-less promoters such as found in EPHX1 (19). Several transcription factors have been characterized that can potentially bind to CCAAT or similar sequences such as the CCAAT/enhancer-binding proteins (C/EBPs), the CCAAT transcription factor (CTF), and the CCAAT box binding factor/ nuclear factor Y (NF-Y) (19,20). The latter factor has been shown to be the major CCAAT-binding protein in diverse promoters in vivo (21). NF-Y is composed of three subunits: NF-YA, NF-YB, and NF-YC, which are all required for DNA binding (20,22). In the absence of DNA, NF-YB and NF-YC, which contain histone-like domains, form a stable heterodimer prior to the association with NF-YA and binding to the CCAAT motif (20). NF-Y is an important regulator of numerous genes and has been shown to physically interact with many other adjacent transcription factors through protein-protein interactions, which subsequently affect the interaction of these factors with their DNA binding site as observed for USF-1/2 (23), SP1 (24), SF1 (25), SREBP (26), ATF6 (27), and HNF-4 (28). In addition, NF-Y also interacts with components of the transcriptional machinery such as TFIID (29), as well as other coactivators such as PCAF (30), GCN5 (31), and p300/CBP (32) that play an important role in transcriptional regulation by altering chromatin structure.
Our initial observations demonstrated that EPHX1 core promoter activity was stimulated by C/EBP␣ despite the absence of a functional DNA binding site for this factor. C/EBPs have been shown to play critical roles in regulating the expression of numerous hepatocyte-specific genes (33). In this study we have demonstrated that, although NF-Y interacts with the EPHX1 Ϫ5/Ϫ1 bp CCAAT box in vitro and in vivo, it was unable to stimulate EPHX1 promoter activity. C/EBP␣ activation of EPHX1 promoter activity was, however, dependent on CCAAT box integrity. Together, these results demonstrate that transcription of the EPHX1 promoter is mediated by the interaction of C/EBP␣ with NF-Y bound to the CCAAT box element.

EXPERIMENTAL PROCEDURES
Plasmids-The Ϫ80/ϩ25 EPHX1 core promoter fragment in the promoterless expression vector pOGH, which contains a human growth hormone (hGH) reporter gene (Nichols Institute Diagnostics), was prepared as previously described (18). The construct containing a mutated CCAAT box was prepared by polymerase chain reaction using the Ϫ80/ϩ25 EPHX1 construct as a template and the following oligonucleotides as primers: 5Ј primer, 5Ј-AGCAAGCTTTGTATCTCTCAGGTCAGC-3Ј, containing a HindIII site (underlined); 3Ј primer, AGCGGATCCCAATTGCA-CAGTCCTGCCAAGTCAGCTGATCTCATGTGTGCACAG-3Ј, containing a BamHI site (underlined). The mutated nucleotides are in boldface type. The PCR products were digested with HindIII and BamHI and inserted into pOGH. The expression vectors for the three subunits of NF-Y (pCMV-NF-YA, pCMV-NF-YB, pCMV-NF-YC) were provided by Dr. Hiroyoshi Ariga (Hokkaido University, Hokkaido, Japan). The expression vectors for GST-fused NF-YA, NF-YB, and NF-YC in pGEX-4T were provided by Dr. Keiko Funa (Goteborg University, Goteborg, Sweden). The expression vector for full-length mammalian C/EBP␣ (pSEW-CO1) was provided by Dr. Gary S. Hayward (Johns Hopkins University School of Medicine, Baltimore, MD). The expression vectors for the GST-fused C/EBP␣ fulllength (1-360), N-terminal (1-226), and C-terminal (281-360) portions in pGEX-4T-1 were obtained from Dr. Gretchen Darlington (Baylor College of Medicine, Houston, TX). The cDNA encoding the C/EBP␣ N-terminal portion (1-226) was isolated from the pGEX-4T-1 vector by BamHI/XhoI digestion and inserted in pCDNA1.1/Amp (Invitrogen) for expression in HepG2 cells. The C-terminal portion of the C/EBP␣ cDNA was also cloned into pCDNA1.1 so that the ATG codon at 219 would be used to translate the 219 -360 sequence in HepG2 cells. ⌻he expression vector C/EBP␤ (pCMV-C/EBP␤) was provided by Dr. Magnus Nord (Karolinska Institute, Huddinge University Hospital, Huddinge, Sweden). The expression vector for CTF was provided by Dr. Richard Gronostajski (State University of New York, Buffalo, NY). The dominant negative NF-YA expression vector (NF-YAm29) was provided by Dr. Timothy Osborne (University of California, Irvine, CA).
Promoter Reporter Assay-HepG2 and HeLa cells were obtained from the American Type Culture Collection (ATCC) and grown at 37°C in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum. The EPHX1 Ϫ80/ϩ25 promoter constructs and transcription factor expression vectors or the NF-YA dominant negative vector were transiently transfected into HepG2 or HeLa cells using DEAEdextran transfection methods as previously described (34). A pSV-␤galactosidase control vector (Promega, Madison, WI) was included as an internal control of transfection efficiency. The hGH produced was excreted into the medium and measured by radioimmunoassay (Nichols Institute Diagnostics) 48 h after transfection. Construct pXGH5, which contained the mouse metallothionein promoter, was also used as a positive control to monitor transfection efficiency in HeLa cells. The experiments were carried out in triplicate in 100-mm dishes and repeated at least three times with freshly made plasmid preparations to ensure that the differences in promoter activity between different constructs was attributable to their sequence differences. The amounts of hGH produced were normalized to the activity of ␤-galactosidase after subtracting background values of the promoterless pOGH plasmid.
Electrophoresis Mobility Shift Assay (EMSA)-The oligonucleotide probes for the unmodified and mutated (M1-M7) EPHX1 Ϫ23/ϩ15 fragment were synthesized, annealed, and radiolabeled with [ 32 P]dNTP in fill-in reactions with Klenow large fragment of DNA polymerase. The probes containing the consensus binding sites (Santa Cruz Biotechnology) for NF-Y, C/EBP, and CTF were also prepared with the following sense strands: 5Ј-AGACCGTACGTGATTGGTTAATCTCTT-3Ј (NF-Y), 5Ј-TGCAGATTGCGCAATCTGCACT-3Ј (C/EBP), and 5Ј-TTTTGGATT-GAAGCCAATATGATAACT (CTF). Expression plasmids for NF-Y subunits A, B, and C (1 g) were linearized by XhoI and in vitro transcription/translation was carried out by using the T7 quick coupled transcription/translation system (Promega) (50 l) according to the instructions from the manufacturer. To eliminate the nonspecific effects of the TNT system and the plasmids on C/EBP␣ DNA binding, the total amount of these two materials were identical in all experiments. Nuclear proteins were extracted from cultured HepG2 cells as previously described (35). A protease inhibitor mixture for mammalian tissues containing 4-(2-aminoethyl)-benzenesulfonyl) fluoride, aprotinin, leupeptin, bestatin, pepstatin A, and E-64 (Sigma) was also included in Buffers A and C during isolation steps at concentrations recommended by the manufacturer. HeLa nuclear extracts were purchased from Promega (Madison, WI). Poly(dI-dC)⅐poly(dI-dC) (1.25 g) was preincubated with 3-18 g of nuclear proteins for 10 min at room temperature in a mix of 19 l containing 20 mM Tris-HCl, pH 7.8, 1 mM MgCl 2 , 50 mM NaCl, 0.5 mM EDTA, 0.5 mM dithiothreitol, and 10% glycerol. 0.5-1 l of radiolabeled oligonucleotides (approximately 5-10 fmol/5000 -10,000 cpm) were added, and the incubation was continued for another 20 min at room temperature. Where indicated, unlabeled competitors (50ϫ labeled oligo), antibodies (1.0 g) (Santa Cruz Biotechnology) and synthesized NF-YA, -B, and -C subunits (1 l) were also included in the binding mixture prior to the addition of the radiolabeled probes. After mixing with 1 l of loading buffer (250 mM Tris-HCl, pH 7.8, 0.2% bromphenol blue, 40% glycerol), the mixture was loaded on a 6% nondenaturing acrylamide (acrylamide:bisacrylamide ϭ 29:1) in 0.5ϫ TBE buffer prerun at 50 V for 1 h. Gel electrophoresis was carried out at 150 V (approximately 10 V/cm) for 1.5 h. The bands were visualized by autoradiography at Ϫ75°C with an intensifying screen.
Co-immunoprecipitation Assay-A HepG2 nuclear extract (35) (60 g) was incubated with 40 l of protein A-agarose beads and 8 g of anti-NF-YA antibody for 2 h in 150 mM NaCl, 50 mM Tris-HCl, pH 7.4, 0.3% Nonidet P-40, 1 mM EDTA, 1 mM dithiothreitol, and 0.5 mM phenylmethylsulfonyl fluoride. As negative controls, the HepG2 nuclear proteins were also incubated with an anti-GATA-3 antibody and with protein A-Sepharose alone. The beads were extensively washed with the same buffer and the eluted proteins separated by SDS-PAGE. The proteins were then transferred to a nitrocellulose filter and Westernblotted with anti-C/EBP␣ antibody as well as anti-NF-YA antibody. The bands were visualized with alkaline phosphatase-conjugated secondary antibody. The Western analysis was performed with reagents from the WesternBreeze chromogenic Western blot immunodetection kit (Invitrogen) according to the recommendations from the manufacturer.
GST Pull-down Assays-The expression vectors containing the cDNAs encoding full-length (1-360), N-terminal (1-226), and C-terminal (281-360) parts of C/EBP␣ or NF-YA, NF-YB, or NF-YC in pGEX-4T (Invitrogen) were transformed into BL21-Gold (DE3) cells (Stratagene), grown to A 600 ϭ 0.6, followed by the addition of isopropyl-␤-D-1-thiogalactopyranoside to 1 mM and incubation for 2 h at 37°C. The cells were collected, suspended in PBS, 0.5% Nonidet P-40, and protease inhibitor mixture (Sigma), lysed by sonication, and the cell debris removed by centrifugation at 14,000 rpm for 10 min. Glutathione-Sepharose 4B beads (50 l) (Amersham Biosciences) was added to the supernatant and the tubes inverted for 1 h at room temperature, washed three times with PBS, and the associated GST fusion proteins quantitated by SDS-PAGE. The GST pull-down assay was performed with 1 g of each fusion protein in PBS, 0.5% Nonidet P-40, 1 mM dithiothreitol, protease inhibitor mixture, and 200 g of HepG2 nuclear protein extract in a total of 500 l by inverting the tubes for 2 h at 4°C. The beads were washed three times with PBS and the bead-associated proteins separated by SDS-PAGE and analyzed by Western blotting using anti-NF-YA and anti-NF-YB antibodies for the C/EBP␣ pulldown assay and anti-C/EBP␣ antibody for the NF-Y pull-down assay.

EPHX1 Promoter Activity Is Dependent on the Ϫ5/Ϫ1 bp CCAAT Box and Is Induced by C/EBP␣ Expression-Previous
studies concerning the regulation of EPHX1 expression by cis elements in the core promoter region have established that the Ϫ80/ϩ25 bp fragment affords maximal basal promoter activity in HepG2 cells (18). In addition to two GATA sites that mediated the activation of EPHX1 transcription by GATA-4 (18), sequence analysis also revealed the presence of a reversed CCAAT box at Ϫ5/Ϫ1 bp (Fig. 1A). To establish the functional role of this cis element in the transcriptional regulation of EPHX1, the CCAAT box was mutated to CTGAT. Transient transfection of the WT and M reporter constructs into HepG2 cells revealed a 50% loss of promoter activity, demonstrating that the CCAAT box element contributed to the core promoter activity (Fig. 1A). To establish which transcription factor interacts with this element, the Ϫ80/ ϩ25 bp pOGH EPHX1 core promoter construct was cotransfected into HepG2 cells with the expression vectors for several putative CCAAT box-binding proteins: NF-Y, C/EBP␣, C/EBP␤, and CTF. As shown in Fig. 1B, C/EBP␣ overexpression resulted in approximately a 64-fold increase in promoter activity, whereas expression of the other factors resulted in little or no effect on EPHX1 promoter activity.
NF-Y Binds to the EPHX1 CCAAT Box and Interacts with C⁄EBP␣-To characterize the transcription factor that binds to the Ϫ5/Ϫ1 CCAAT box, EMSA was performed with nuclear extracts from HepG2 cells and with radiolabeled consensus probes for NF-Y, C/EBP␣, and CTF. As shown in Fig. 2B, all three probes formed DNA-protein complexes (lanes 1, 7, and 13), which could be inhibited by a 50-fold excess of unlabeled probe (WT). To establish which of these factors binds to the CCAAT box or adjacent regions in the Ϫ80/ϩ25 bp EPHX1 promoter fragment, four overlapping DNA fragments (a-d) ( Fig. 2A) were prepared and utilized as competitors in the EMSA analysis. Fragments a-d did not affect the C/EBP␣ complex (lanes 9 -12), whereas fragment c did not affect the CTF complex (lane 15), establishing that these factors did not directly bind to the CCAAT box and, in addition, that C/EBP␣ did not bind to adjacent domains in the Ϫ80/ϩ25 bp core promoter fragment. In contrast, fragment c (50-fold) was an effective competitor of the NF-Y-DNA complex, indicating that, as opposed to C/EBP␣ and CTF, NF-Y bound directly to the CCAAT box. However, increased expression of NF-Y had no stimulatory effect on EPHX1 promoter activity as shown in Fig.  1. These results suggest that activation of EPHX1 by C/EBP␣ may be mediated by its interaction with NF-Y bound to the CCAAT box. To further characterize this mechanism, EMSA was carried out with a radiolabeled EPHX1 Ϫ23/ϩ15 bp probe, which contained the CCAAT box. As shown in Fig. 2C, incubation of the probe with a HepG2 nuclear extract produced a DNA-protein complex (lane 1) that was inhibited in the presence of the unlabeled Ϫ23/ϩ15 fragment (WT) (lane 2), showing that the binding was specific. Competition with an NF-Y consensus oligonucleotide also blocked the formation of the complex (lane 3), further supporting the notion that NF-Y binds directly to the CCAAT element. Competition with the C/EBP␣ oligonucleotide, however, had little effect on the complex (lane 4), a result consistent with the conclusion that this factor did not directly bind to the EPHX1 core promoter. To further identify the components of this complex EMSA was carried out in the presence of antibodies against NF-Y, C/EBP␣, C/EBP␤, and CTF. As shown in Fig. 2C, an anti-NF-Y antibody supershifted the whole complex (lane 5), confirming that NF-Y is a component of the complex. In contrast, an anti-C/EBP␣ antibody abolished the complex (lane 6), whereas antibodies against C/EBP␤ and CTF had no effect (lanes 8 and 9), indicating that C/EBP␣ is also a component of the protein-DNA complex. The inhibitory effect of the anti-C/EBP␣ antibody may result from the antibody binding close to the NF-Y DNA binding domain or from an induced conformation change resulting in a decreased affinity for the DNA site. A mixture (1:1) of these antibodies resulted in a supershifted band of diminished intensity (lane 7), suggesting that the supershifted band also contained both NF-Y and C/EBP␣. These results further support the idea that NF-Y and C/EBP␣ are components of a complex that activates EPHX1 transcription mediated by the interaction of NF-Y with the CCAAT element, where it serves as an anchor for C/EBP␣.
To investigate whether the formation of the C/EBP␣[NF-Y] complex inhibits C/EBP␣ DNA binding, EMSA was carried out as shown in Fig. 2B with in vitro translated NF-Y subunits. As shown in Fig. 2D (lane 2), competition with NF-Y(AϩBϩC) inhibited binding to the C/EBP␣ oligonucleotide by 79%, indicating that excess NF-Y would compete for C/EBP␣ binding to DNA. This inhibitory effect was less in the presence of NF-YA (45%) or NF-YB (30%) alone and was not observed with NF-YC, suggesting that subunits A and B are directly involved in the formation of this complex.
C⁄EBP␣ Induction of EPHX1 Activity Is Dependent on NF-Y Binding to the Ϫ5/Ϫ1 CCAAT Box and on Both the DNA Binding and Activation Domains of C/EBP␣-To further establish the functional role of NF-Y in mediating the induction of EPHX1 promoter activity by C/EBP␣, HepG2 cells were transiently transfected with the EPHX1 Ϫ80/ϩ25 bp pOGH core promoter construct together with a dominant-negative DNA binding-defective NF-Y mutant (NF-YAm29) whose product prevents NF-Y from binding to the CCAAT box. As shown in Fig. 3 (lane 2), the presence of the mutant NF-YA subunit significantly reduced the basal EPHX1 promoter activity, thereby demonstrating the functional participation of NF-Y in this complex. HepG2 cells were also transiently transfected with increasing concentrations of C/EBP␣ expression plasmid in addition to the EPHX1 core promoter, resulting in a dosedependent stimulation of promoter activity as shown in Fig. 3  (lanes 3-5). The C/EBP␣-induced stimulation was also dependent on NF-Y binding to the CCAAT box, as evidenced by the inhibitory effect of the dominant-negative NF-YA mutant (lane 6). The integrity of the CCAAT box was also critical in mediating the C/EBP␣ induction, as established by utilizing the FIG. 2. Identification of the HepG2 nuclear protein factors that are associated with the EPHX1 ؊5/؊1 CCAAT box. A, map of the EPHX1 Ϫ80/ϩ25 promoter region containing the Ϫ5/Ϫ1 CCAAT box and the four overlapping fragments (a-d) used as competitors in the EMSA analysis. B, EMSA analysis was performed with a HepG2 nuclear protein extract and radiolabeled probes containing the consensus binding site for NF-Y (lanes 1-6), C/EBP␣ (lanes 7-12), and CTF (lanes [13][14][15] in the absence of competitors (Ϫ) (lanes 1, 7, and 13) and in the presence of a 50-fold excess of unlabeled oligonucleotide competitor, WT (lanes 2, 8, and 14), oligo a (lanes 3 and 9), oligo b (lanes 4 and 10), oligo c (lanes 5, 11, and 15), and oligo d (lanes 6 and 12). C, EMSA was performed with a radiolabeled EPHX1 Ϫ23/ϩ15 oligonucleotide probe containing the CCAAT box, analyzed on 6% polyacrylamide gels, and visualized by autoradiography. Prior to the addition of the probe, the reaction mixture was incubated, when indicated, with mutated EPHX1 Ϫ80/ϩ25 promoter construct, where the -fold induction was now greatly reduced (lane 7). The residual activity may result from the induction of other genes, whose products may, in turn, affect EPHX1 activity. To investigate which domain of C/EBP␣ is involved in the induction of EPHX1 promoter activity, the N-terminal part (1-226), which contains the transactivation domains, and the C-terminal part (219 -360), which contains the DNA binding domain were coexpressed with the EPHX1 Ϫ80/ϩ25 core promoter in HepG2 cells. As shown in Fig. 3 (lanes 8 and 9), stimulation of EPHX1 was greatly reduced when compared with the full-length C/EBP␣-(1-360) probe (lane 5), suggesting that both domains of C/EBP␣ are required for the induction of EPHX1 activity.

C⁄EBP␣ Forms a Complex with CCAAT-bound NF-Y and Does Not Bind to Other Contiguous Regions of the Ϫ23/ϩ15
bp Probe-Although the above experiments have indicated that C/EBP␣ interacts directly with CCAAT-bound NF-Y, the possibility cannot be excluded that it might also bind weakly to an adjacent DNA sequence similar to the CCAAT element, where this binding could be stabilized by NF-Y in a fashion similar to that observed for the weak binding of NF-Y to an imperfect site, which is stabilized by Sp1 binding to an adjacent site (36). To examine this possibility, the Ϫ5/Ϫ1 bp CCAAT box and the flanking sequences in the Ϫ23/ϩ15 bp probe were systematically mutated (Fig. 4A) and the effects of these sequence substitutions on NF-Y and C/EBP␣ binding assessed by EMSA analysis as shown in Fig. 4 (B and C). Utilizing the WT probe (lanes 8 and 9) resulted in the formation of a complex that contained both NF-Y (Fig. 4C) and C/EBP␣ (Fig. 4B), as also shown in Fig. 2C. The M4 probe in which the reverse CCAAT box had been mutated was the only probe that resulted in the loss of the DNA-protein complex (Fig. 4, B and C, lane 7), confirming that NF-Y bound to this site. All the other probes still formed the protein-DNA complex, which was sensitive to the anti-C/EBP␣ and anti-NF-YA antibodies, demonstrating that the complex still contained NF-Y and C/EBP␣. The increased intensity of the bands obtained with M3 (Fig. 4, B and C, lane 5) most likely results from the substitutions that result in a superior NF-Y consensus sequence (20), whereas the lighter intensity observed with the M5 probe, which has no change in the NF-Y consensus sequence, may be attributable to the lower specific activity of the probe. If C/EBP␣ interacted with DNA sequences adjacent to the CCAAT box, one of the introduced mutations would have affected the formation of the complex. Thus, these results establish that C/EBP␣ only binds to NF-Y and not to adjacent DNA sites and eliminates the possibility that an additional factor binding to an adjacent sequence is necessary for the formation of the NF-Y[C/EBP␣] complex.

NF-Y Associates with C⁄EBP␣ in Vivo -
The binding of C/EBP␣ to NF-Y in vivo was evaluated by immunoprecipitation of HepG2 nuclear extracts with an anti-NF-YA antibody. The precipitated proteins were separated by SDS-PAGE and immunoblotted with anti-C/EBP␣ and anti-NF-YA antibodies. As shown in Fig. 5, both transcription factors were detected in the anti-NF-YA antibody-immunoprecipitated proteins (lanes 1  and 6). These proteins were also detected in total nuclear extracts (lanes 4 and 5). The ability of anti-NF-YA antibody to immunoprecipitate both NF-YA (lane 6) and C/EBP␣ (lane 1) demonstrates the physical association of these two factors. C/EBP␣ was not detected when protein A-Sepharose alone (lane 2) or an irrelevant antibody (anti-GATA-3) (lane 3) was used.
Interaction of NF-Y and C⁄EBP␣ with the EPHX1 Ϫ5/Ϫ1 bp CCAAT Box in Vivo-ChIP procedures were utilized to establish that NF-Y interacted with the EPHX1 core promoter CCAAT box and formed a complex with C/EBP␣ on the endogenous EPHX1 promoter. Cross-linked chromatin from HepG2 cells was immunoprecipitated with an anti-NF-YA antibody. The presence of the EPHX1 promoter fragment was established by PCR using primers from Ϫ104 to ϩ395 bp. As negative controls, immunoprecipitations were run without antibody or with pre-immune IgG. PCR amplification was also performed with primers (ϩ21,998 to 22,324 bp) that bind to the EPHX1 exon 9 region that does not have a CCAAT box. As shown in Fig. 6, anti-NF-YA antibody immunoprecipitated chromatin that contained the EPHX1 core promoter (Fig. 6A, lane 4). The specificity of this interaction was verified by demonstrating the absence of amplification in the EPHX1 exon 9 region (Fig. 6B, lane 4) or when no antibody (lane 2) or pre-immune IgG (lane 3) were used. Immunoprecipitation of cross-linked chromatin with an anti-C/EBP␣ antibody also resulted in the formation of an EPHX1 promoter fragment (Fig. 6A, lane 5) that was not observed in the EPHX1 exon 9 region (Fig.  6B, lane 5). These results establish that NF-Y binds the EPHX1 core promoter CCAAT in vivo and forms a complex with C/EBP␣ associated with the endogenous promoter.
Characterization of the C⁄EBP␣ Domains and NF-Y Subunits Involved in the EPHX1 CCAAT Box Complex-To further characterize the NF-Y subunit(s) that interact with C/EBP␣, GSTfused subunits of NF-Y were utilized in a GST pull-down assay. As shown in Fig. 7A, only the NF-YB fusion protein (lane 3) was  able to interact with C/EBP␣, a result that was supported by the EMSA competition study (Fig. 2D, lane 4). The NF-YA fusion protein, however, was not able to interact with C/EBP␣ (lane 2), in contrast to the co-immunoprecipitation (Fig. 5) and EMSA competition (Fig. 2D, lane 3) studies, suggesting that the portion of NF-YA involved in the interaction with C/EBP␣ may be too close to the GST moiety and the interaction is inhibited by steric hindrance. To characterize the C/EBP␣ domain that interacts with NF-Y, a GST-fused full-length (1-360), a N-terminal portion containing the activation domains (1-226), and a C-terminal portion containing the DNA binding domain (281-360) were utilized in a GST pull-down assay. As shown in Fig. 7B, full-length GST-C/EBP␣ (lane 2) as well as the C-terminal region (lane 4) interacted with NF-YA in a HepG2 nuclear extract as detected by anti-NF-YA antibody. In contrast, no interaction was observed with the N-terminal region (lane 3) or GST alone (lane 1), thereby establishing that the DNA-binding region of C/EBP␣ interacted with NF-Y in this complex. Analysis of the pull-down assays with anti-NF-YB antibody, however, failed to detect any proteins interacting with the C/EBP␣ probes, suggesting that the portion of C/EBP␣ involved in the interaction with NF-YB may be too close to the GST moiety as described above for GST-NF-YA interaction with anti-C/EBP␣ antibody. These results demonstrate the importance of using complementary methods in the study of protein-protein interactions. Ϫ23/ϩ15 oligonucleotides shown in A and analyzed as in Fig. 2. B, binding of the Ϫ23/ϩ15 oligonucleotide probes to HepG2 nuclear protein in the absence (Ϫ) and presence (ϩ) of anti-C/EBP␣ antibody (lanes 1-15). C, binding of the Ϫ23/ϩ15 oligonucleotide probes in the absence and presence of anti-NF-YA antibody (lanes 1-15).

Reconstitution of EPHX1 Promoter Activity in HeLa Cells
and Induction by C⁄EBP␣-The roll of C/EBP␣ in stimulating EPHX1 promoter activity was further investigated using HeLa cells. EMSA analysis using a consensus NF-Y oligonucleotide demonstrated abundant NF-Y binding, similar to that observed in HepG2 cells (Fig. 8A, lanes 1 and 2); however, binding to a consensus C/EBP␣ probe was negligible compared with the levels in HepG2 cells (lanes 3 and 4). Commensurate with this observation was the observed absence of promoter activity when the EPHX1 Ϫ80/ϩ25 bp pOGH core promoter construct was transiently transfected into HeLa cells (Fig. 8B). The positive control plasmid pXGH5 transfected into HepG2 and HeLa cells indicated similar amounts of activity for the mouse metallothionein promoter (data not shown), indicating similar transfection efficiencies for the two cell lines, and thus establishing that the negligible EPHX1 promoter activity in HeLa cells resulted from differences in the repertoire of transcription factors in each cell. These results further establish that NF-Y binding to the CCAAT box alone cannot activate EPHX1 expression. Transfection of HeLa cells with the expression vector for C/EBP␣, in addition to the EPHX1 promoter construct, however, now resulted in the concentration-dependent expression of EPHX1 promoter activity (Fig. 8B), where a 55-fold increase in promoter activity was observed (0.1 g of C/EBP␣ expression vector) compared with a 7-fold increase observed in HepG2 cells (Fig. 3, lanes 1 and 5) under the same conditions. The different extent of C/EBP␣-induced stimulation observed in HepG2 and HeLa cells is likely the result of the high basal level of EPHX1 expression in HepG2 cells because of differences in amounts of other factors such as GATA-4, which is also absent in HeLa cells (18). These results further establish the requirement for C/EBP␣ for EPHX1 promoter activity mediated by NF-Y. DISCUSSION In this report we have demonstrated that EPHX1 expression in HepG2 cells is regulated, in part, by a complex composed of C/EBP␣ and NF-Y bound to a CCAAT box in the EPHX1 core promoter. Results obtained with EMSA in vitro (Figs. 2 and 4), with chromatin immunoprecipitation assays in vivo (Fig. 6), and with transient transfection assays utilizing a NF-YA dominant negative vector (NF-YAm29) (Fig. 3) established that NF-Y bound directly to the CCAAT box but was not sufficient to stimulate EPHX1 promoter activity (Figs. 1B and 8). In contrast, increased expression of C/EBP␣, which did not bind directly to the Ϫ23/ϩ15 bp oligonucleotide containing the CCAAT box (Fig. 2, B (lane 11) and C (lane 4)), was shown to  1 and 2) and C/EBP␣ (lanes 3 and 4) in nuclear extracts from HepG2 and HeLa cells using radiolabeled oligonucleotides containing the consensus binding site for the respective transcription factors. B, HepG2 and HeLa cells were transiently transfected with the EPHX1 Ϫ80/ϩ25 WT constructs in pOGH and basal activity measured by hGH release as in Fig. 1. HeLa cells were also cotransfected with the expression vector for C/EBP␣ (0.1 and 0.2 g).
stimulate EPHX1 promoter activity in two different cell lines (Figs. 1B and 8B) and to physically interact with NF-Y by co-immunoprecipitation (Fig. 5), ChiP assay (Fig. 6), and GST pull-down assays (Fig. 7). Furthermore, EMSA analyses also supported the notion of a complex formed between these two factors (Figs. 2C and 4). The co-immunoprecipitation of C/EBP␣ by an anti-NF-YA antibody (Fig. 5) and GST pull-down studies (Fig. 7) demonstrate that C/EBP␣ can interact with NF-YA and NF-YB subunits. The inability of the activation domains of NF-Y to induce EPHX1 activity when bound to the Ϫ5/Ϫ1 bp CCAAT box may result from an unfavorable juxtaposition to the transcription initiation complex. The binding of C/EBP␣ to NF-Y bound to the CCAAT box site, however, is able to activate EPHX1 expression, possibly mediated by the activation domains of C/EBP␣, a conclusion based on the requirement of both the activation and DNA binding domains in a functional assay (Fig. 3, lanes 8 and 9) and the observed interaction of the C/EBP␣ DNA binding domain with NF-Y (Fig. 7B).
The undetectable level of EPHX1 activity and C/EBP␣ in HeLa cells, which express NF-Y at levels similar to those found in HepG2 cells, also supports the conclusion that C/EBP␣ is required for EPHX1 promoter activity (Fig. 7). Although expression of C/EBP␣ in HeLa cells resulted in the stimulation of EPHX1 promoter activity, it was at greatly reduced level from that observed in HepG2 cells, suggesting that additional transcription factors such as GATA-4, which is also not expressed in HeLa cells (18), are required for a more robust level of expression. Two GATA-4 binding sites in the core promoter are involved in EPHX1 expression (18); however, there is no evidence for an interaction with the CCAAT box described in this report (data not shown). In addition, the EPHX1 core promoter also contains a putative binding site for Nrf2 on the antioxidant response element (ARE) that, based on preliminary evidence, may be involved in the inducible expression of EPHX1 by xenobiotics such as oltipraz.
Although NF-Y has been shown to physically and functionally interact with several transcription factors that are also bound to DNA (23)(24)(25)(26)(27)(28), this study appears to be the first report of C/EBP␣ stimulating promoter activity, not through its DNA binding capacity, but through its interaction with NF-Y bound to the EPHX1 CCAAT box. Although NF-Y and C/EBP␣ appear to be the components in this complex, we cannot exclude the possibility that other protein(s) may be associated with the C/EBP␣[NF-Y] complex to promote transcription. Several other genes have been shown to be regulated by transcription factors or coactivators that interact with NF-Y bound to the CCAAT box. The heat shock proteins HSP70 and HSP40 are induced by the coactivator HSP-CBF, which binds to the HSP promoters by forming a complex with CCAAT-bound NF-Y (37). This type of functional complex is similar to those observed for other coactivators that interact with NF-Y such as P/CAF (29) and GCN5 (31). These factors possess histone acetyltransferase activity that affects local chromatin structure (31). The transcriptional regulation of the MDR1 gene by histone acetyltransferase and deacetylate is also mediated by NF-Y (38). The cell cycle-dependent expression of the HSP70 gene has also been shown to be dependent on CCAAT-bound NF-Y where the level of HSP70 expression is regulated by the intracellular concentration of the transcription factor, c-Myc that interacts with NF-Y but not with DNA (39). c-Myc has also been shown to play a critical role in the cell cycle-dependent expression of platelet-derived growth factor ␤-receptor mRNA (40) by also forming a complex with CCAAT-bound NF-Y. A similar model was proposed for the action of p73 that also represses plateletderived growth factor ␤-receptor (41) and the p53 tumor sup-pressor protein that decreases the transcription of the cdc2 gene (42) by interacting with CCAAT-bound NF-Y. The surface of CCAAT-bound NF-Y thus permits the subsequent binding of a variety of factors that, in concert with NF-Y, regulate the promoter activity of numerous genes. The observed binding of C/EBP␣ to NF-Y in this study may inhibit C/EBP␣ DNA binding activity as suggested in Fig. 2D that could result in the negative regulation of C/EBP␣ target genes. C/EBP␣ also forms complexes with DNA-bound factors in addition to NF-Y, as described in liver, where an age-specific C/EBP␣-Rb-Brm-E2F4 complex binds to an E2F-dependent (c-Myc) promoter resulting in gene repression and the loss of a proliferative response (43). C/EBP␣ also autoregulates the human C/EBP␣ gene by its interaction with DNA-bound USF (44).
C/EBP␣, which is an important regulator of hepatocyte proliferation and a variety of liver functions (33), is first expressed at approximately day 16 during rat fetal development when many of these processes are initiated (45) and increases in concentration during the first 2.5 months (46). In previous studies we have also established that rat mEH is first expressed in late fetal development and reaches adult levels in ϳ2 months (8). The inductive effect of C/EBP␣ on EPHX1 expression suggests that this factor may play a significant role in regulating the expression of EPHX1 during development. This association is further supported by the observation that C/EBP␣ mRNA expression is transiently decreased by more that 80% within 24 h after partial hepatectomy (47,48). A similar response is also observed for the expression of mEH activity during liver regeneration (49).
In conclusion, these studies have demonstrated that C/EBP␣ plays a critical role in establishing the constitutive expression of hepatic EPHX1 whose gene product (mEH) is involved in carcinogen metabolism and in sodium-dependent transport of bile acids. Furthermore, the results support a model where C/EBP␣ regulates EPHX1 transcription not by binding to DNA, but through its interaction with NF-Y bound to the Ϫ5/Ϫ1 bp CCAAT box in the EPHX1 core promoter.