Characterization of Transcriptional Regulatory Elements in the Promoter Region of the Murine Blood Coagulation Factor VII Gene*

To identify the 5′ sequences of the murine coagulation factor VII (fVII) gene that resulted in its efficient transcription, a variety of 5′-flanking sequences up to 7 kilobase pairs upstream of the translation ATG initiation codon were fused to the reporter gene, bacterial chloramphenicol acetyltransferase, and relative expression levels of this gene in mouse Hepa 1–6 cells were determined. It was found that the 5′ region extending approximately 85 base pairs (bp) upstream of the transcriptional initiation site served as the minimal DNA region that provided full relative promoter activity for chloramphenicol acetyltransferase expression. This region of the gene also contains consensus sequences for liver-enriched transcription factors, C/EBPβ and HNF4, as well as for the ubiquitous protein factors, AP1, H4TF1, NF1, and Sp1. In vitro DNase I footprinting of the 200-bp proximal region of the promoter with a murine Hepa 1–6 cell nuclear extract revealed a clear footprint of a region corresponding to −80 to −28 bp of the murine fVII gene, suggesting that liver factors interact with this region of the DNA. Competitive gel shift and supershift assays with different synthetic oligonucleotide probes demonstrate that proteins contained in the nuclear extract, identified as C/EBPβ, H4TF1, and HNF4, bind to a region of the murine fVII DNA from 85 to 32 bp upstream of the transcription start site. Purified Sp1 also interacts with this region of the DNA at a site that substantially overlaps, but is not identical to, the H4TF1 binding locus. Binding of Sp1 to the mouse DNA was not observed with the nuclear extract as the source of the transcription factors, suggesting that Sp1 is likely displaced from its binding site by H4TF1 in the crude extract. In vivo dimethyl sulfate footprint analysis confirmed the existence of these sites and additionally revealed two other binding regions slightly upstream of the CCAAT/enhancer-binding protein (C/EBP) binding locus that are homologous to NF1 binding sequences. The data demonstrate that appropriate transcription factor binding sites exist in the proximal promoter region of the murine fVII gene that are consistent with its strong liver-based expression in a highly regulated manner.

Factor VII (fVII) 1 is a vitamin K-dependent plasma protein that serves as the precursor for fVIIa, a serine protease that functions as a procoagulant in the extrinsic blood coagulation pathway. This latter activity is derived in part from the ability of fVIIa to activate the coagulation zymogens, fIX and fX to their respective enzymes, fIXa and fXa (1,2). The presence of a cofactor for fVII, TF, along with Ca 2ϩ , creates optimal conditions for the functioning of fVIIa. Tissue factor pathway inhibitor serves to down-regulate this coagulation pathway by inactivating both fXa and the fVIIa⅐TF complex (3).
Human fVII is synthesized in liver as a single-chain protein containing 406 amino acids. Activation of this zymogen occurs consequent to cleavage of the Arg 152 -Ile 153 peptide bond. This step is catalyzed by fXa (4), thrombin (4), TF⅐fVIIa (5), or hepsin (6), and leads to an enzyme composed of a 152-amino acid light chain, disulfide-linked to a 254-residue heavy chain. Factor VII is organized as a series of domain units (7) that are common to fIX (8), fX (9), and protein C (10). Specifically, the light chain of fVIIa consists of a Gla-containing domain, followed by a short helical spacer region and two epidermal growth factor-like motifs. The functions of these modules are to provide binding sites for regulators of fVII activity, such as TF (11,12). The heavy chain of fVIIa contains the serine protease catalytic machinery, as well as binding sites for TF (13)(14)(15).
The intron-exon organization of the human fVII gene (7) and the entire genomic sequence of murine fVII (16) have been established. Both of these genes contain approximately 12 kb, and include transcriptional and translation start sites, 5Јflanking transcriptional regulatory regions, and 3Ј-capping polyadenylation sites and polyadenylation enhancer elements. The positions of the introns in these genes are in identical locations, and splice the exonic regions that encode the protein domains. The human fVII DNA is organized in eight exons, with an optional exon in the leader peptide region. The murine fVII gene possesses seven introns and eight exons. The major transcriptional start site of murine fVII has been located nine nucleotides upstream of the ATG translation initiation site (16). This short transcriptional initiator site in murine fVII presents a similar situation to that of the human gene, where in this latter case the major transcriptional start site was located 51 nucleotides 5Ј of the translational initiation site (17).
Although genetic deficiencies of human fVII have been re-ported, few patients present total fVII deficiency states. However, very low levels of this zymogen (Ͻ2% of normal) are associated with severe coagulation disorders (18 -22). The clinical manifestations of less severe deficiencies are variable (18,19,(23)(24)(25). Such observations suggest that only low levels of fVII activity are required to maintain hemostasis in unchallenged patients. Other data suggest that fVII levels may be reciprocally involved in coronary heart disease, perhaps through a linkage with hypertriglyceridemia (26). The prospective Northwick Park Heart (27), PROCAM (28), and PROCAM follow-up studies (29) suggest that elevated plasma levels of fVII coagulant activity constitute an independent risk factor for fatal outcomes of coronary heart disease in middle-aged men, the chances of which worsen when other risk factors are present. In addition, severe arterial thrombosis in a fVII-deficient patient was observed after infusion of fVII (21), showing that even artificial elevation of fVII levels can result in thrombotic complications.
To investigate the role of fVII in embryonic development and survival, mice were generated with a complete fVII deficiency (30). Unlike genetic deficiency of TF, which results in embryonic lethality at approximately 10.5 days post coitum, fVIIdeficient mice develop normally. However, the fVII-deficient mice suffer severe perinatal lethality mainly from intra-abdominal bleeding. Those surviving birth expire within the next several weeks, primarily from intracranial bleeding (30). Since human clinical data, as well as the fVII-deficiency state produced in mice, indicates that fVII activity levels are important in hemostatic and cardiovascular pathology, we undertook an investigation aimed at defining the factors that regulate transcription of the fVII gene. The results of this study constitute the present contribution.

EXPERIMENTAL PROCEDURES
Proteins-Recombinant transcription factors C/EBP␣ and C/EBP␤ were gifts from Dr. Peter F. Johnson (NCI, National Institutes of Health, Frederick, MD). Sp1 was a product of Promega (Madison, WI).
The rabbit C/EBP␤ polyclonal antibody was produced against a peptide corresponding to amino acids 258 -276 of rat C/EBP␤ (Santa Cruz Biotechnology, Santa Cruz, CA). The rabbit Sp1 antibody (Santa Cruz Biotechnology) was raised against a peptide corresponding to amino acids 520 -538 of human Sp1, a region with an identical sequence to the mouse protein. The NF-B control polyclonal antibody (Santa Cruz Biotechnology) was generated against a peptide corresponding to amino acids 350 -363 of human p50. HNF4 rabbit antiserum, raised against a peptide corresponding to amino acids 445-455 of rat HNF4, was obtained from Dr. Frances Sladek (University of California, Riverside, CA).
Aprotinin and leupeptin were products of Sigma. Proteinase K was a product of Boehringer Mannheim. DNase I was obtained from Worthington.
Expression Vectors-The expression vectors, pSV-␤-gal control (C) vector and pCAT-control (C) vector, pCAT-enhancer (E) vector, and pCAT-basic vector were purchased from Promega. pSV-␤-gal-C vector is a 6.8-kb plasmid containing the lacZ gene, transcription of which is driven by the SV40 promoter/enhancer. pCAT-C vector is a 4.8-kb plasmid containing the CAT gene driven by the SV40 promoter/enhancer. pCAT-E vector is a 4.6-kb plasmid containing the SV40 enhancer, whereas pCAT-B vector is a 4.4-kb plasmid without the enhancer sequences.
Cell Lines and Nuclear Extracts-Mouse Hepa 1-6 cells (American Type Culture Collection, Rockville, MD) were maintained in DMEM containing 1.45 g of glucose/liter (Sigma), supplemented with 2 mM glutamine, 50 g/ml gentamycin sulfate, and 10% (v/v) heat-inactivated fetal calf serum (Life Technologies, Inc.). The cultures were incubated at 6.6% CO 2 , 37°C, in a humidified atmosphere in 150-cm 2 flasks. The cells were grown to confluence in DMEM with 4.5 g/liter glucose.
Nuclear extracts were prepared as described (31). All solutions con-tained 0.05 mM dithiothreitol, and the protease inhibitors aprotinin (1 g/ml), leupeptin (1 g/ml), and phenylmethylsulfonyl fluoride (0.5 mM). Protein concentrations were measured with the Bradford assay. Cell extracts were frozen in 100-l aliquots and stored at Ϫ70°C. Genomic Clones Containing Murine fVII Inserts-pND95 is a 16.0-kb HindIII-NotI fragment generated from murine fVII genomic clone E.80.11C (16), which was reinserted in the HindIII-NotI sites of pBSKIIϩ (Stratagene, La Jolla, CA). This plasmid contains 7.0 kb of 5Ј-flanking sequences of the murine fVII gene and proceeds downstream through exon 5 of this gene.
pND73 is a 2.0-kb NotI-BamHI fragment of clone E.50.4A1 (16) subcloned in the NotI-BamHI sites of pBSKIIϩ. The DNA insert of this plasmid contains 45 bp of the FIXII polylinker DNA, 1.1 kb of the 5Ј-flanking region of the murine fVII gene, and 0.7 kb of the coding region of the gene, terminating at a location between genomic exons 1 and 2.
Preparation of Cell Lysates for ␤-Gal Assays-Adherent cells were washed three times with a solution of 0.05 M sodium phosphate, 0.1 M NaCl, pH 7.5, and harvested by incubating the cells for 5 min at room temperature with a buffer containing 0.04 M Tris-HCl, 0.01 M EDTA, 1.5 M NaCl, pH 7.5. The cells were scraped from the bottom of the Petri dish with tissue culture cell lifter (Fisher) and transferred to a microcentrifuge tube. After centrifugation for 1.5 min (10,000 rpm at 4°C), the cells were resuspended in 0.25 mM Tris-HCl, pH 7.7, and lysed by three cycles of liquid N 2 freeze-thaw steps. The cell debris were pelleted at 4°C for 2.5 min at 10,000 rpm. One-third of the cell lysate was transferred to sterile Eppendorf tubes and stored at Ϫ70°C for ␤-galactosidase enzyme assays. The remaining cell lysate was heated at 60°C for 10 min. The denatured proteins were pelleted for 2.5 min at 10,000 rpm at 4°C. The resulting supernatant was stored at Ϫ70°C for CAT assays.
Reporter Gene Assays-␤-Gal assays using the substrate, p-nitrophenyl-␤-D-galactopyranoside, were performed with protein extracts of transfected hepatocytes to quantify transfection efficiency.
Assays for CAT activity were conducted with approximately 50 g of cell extract protein in a 150-l reaction mixture consisting of 35 l of 1 M Tris-HCl, pH 7.7, 5 l [ 14 C]chloramphenicol, 20 l of acetyl-CoA (3.5 g/l), and an amount of water required to bring the reaction mixture to 150 l. These components were incubated at 37°C for 4 h. The extent of acetylation of the substrate was analyzed on thin layer chromatography plates (J.T. Baker) using a phosphorimager screen (Eastman Kodak Co., Rochester, NY). Normalizations of the CAT assays for transfection efficiencies were accomplished using the cotransfected ␤-gal activity.
Electrophoretic Mobility Shift Assays-Oligonucleotides for mobility shift assays were end-labeled with [␣-32 P]dATP and [␣-32 P]dCTP (Ͼ3,000 Ci/mM; ICN, Costa Mesa, CA), using the Klenow fragment of DNA polymerase (Promega). The binding reactions were carried out in a total volume of 25 l containing 50,000 cpm of the labeled probe, 1 g of poly (dI-dC) (Pharmacia Biotech Inc.), 12 g of nuclear extract, 1 mM EDTA, 5 mM MgCl 2 , 10 M ZnCl 2 , 1 mM ␤-mercaptoethanol, 4% (v/v) glycerol, 100 mM KCl, in 10 mM Tris-HCl, pH 7.5. The reaction mixtures were incubated for 30 min at room temperature and loaded onto a 6% polyacrylamide gel (acrylamide:bisacrylamide ratio, 19:1, w/w) in running buffer (50 mM Tris-HCl, 0.38 M glycine, 2 mM EDTA, pH 8.0). The samples were subjected to electrophoresis at 8 V/cm. The resulting gels were dried and exposed to the x-ray film with an intensifying screen overnight at Ϫ80°C. In competition experiments, oligonucleotide competitors were added in the amounts indicated in the figure legends.
Reactions containing recombinant C/EBP␣ (30 g/ml) or C/EBP␤ (26 g/ml) contained 2.0 l of the protein and 1.0 l of normal rabbit serum/25 l of binding reaction. Reaction mixtures with Sp1 included 1.0 footprint unit (fpu) of purified Sp1 (1 fpu is defined as the amount of Sp1 needed to give full protection against DNase 1 digestion on the SV40 early promoter). Sp1-containing mobility shift assays were performed on 6% polyacrylamide gels in 0.5 ϫ Tris borate/EDTA (44.5 mM Tris-HCl, pH 8.0, 44.5 mM boric acid, 1 mM EDTA). In supershift experiments, the antibodies were added 15 min after the initiation of the 30-min incubation period of the probe with the nuclear extract mixture.
The sequences of double-stranded oligonucleotides used in the assays are as follows (the consensus sequences are indicated in bold lettering, the lowercase letters represent overhangs, the mutant sequences are indicated with an "m," and the mutated bases are The wild-type and mutant Sp1 site oligonucleotides from the human fVII promoter (17) were gifts from Dr. Eleanor Pollak (University of Pennsylvania, Philadelphia, PA).
DNase I Footprint Analysis-The DNase I footprint analysis was carried essentially as described (34). A quantity of 20 pmol of pCAT-200, containing 200 bp of the murine fVII promoter (Ϫ200 bp to Ϫ1 bp), was linearized with either restriction endonucleases HindIII or XbaI, endlabeled as described above with [␣-32 P]dATP and [␣-32 P]dCTP (Ͼ3,000 Ci/mmol; ICN), and digested with the second enzyme to release the oligonucleotide insert. The 32 P-labeled DNA was fractionated electrophoretically on a 5% polyacrylamide gel in TBE buffer (89 mM Tris-HCl, 89 mM boric acid, 2 mM EDTA, pH 8.0). The 32 P-labeled DNA insert was electroeluted from an excised gel slice, extracted with phenol/CHCl 3 (1:1, v/v) followed by CHCl 3 /isoamyl alcohol (24:1, v/v), and precipitated by addition of 1 ml of 100% ethanol. The pellet was washed with 1 ml of 80% ethanol, dried, and resuspended in 50 l of a solution of 10 mM Tris-HCl, 1 mM EDTA, pH 8.0.
An amount of 20,000 cpm of probe was incubated for 15 min on ice with 50 g of nuclear extract protein in 50 l of a solution containing 50 mM KCl, 10% (v/v) glycerol, 1.0 mM dithiothreitol, 2.5 mM MgCl 2 , 0.02% Nonidet P-40 (Sigma), 2.0% polyvinyl alcohol, 1 g of poly(dI-dC), in 10 mM sodium Hepes, pH 7.6. A volume of 50 l of 5 mM CaCl 2 , 10 mM MgCl 2 was included in the reaction mixture and incubated for 1 min at room temperature. This was followed by the addition of 2 l of a 0.016 mg/ml (or 0.0025 mg/ml for the control without nuclear extract) stock solution of DNase I (Worthington). The mixture was incubated for 1 min and followed by addition of 90 l of DNase I stop buffer (1%, w/v, sodium dodecyl sulfate, 0.2 M NaCl, 250 g/ml glycogen, 20 mM EDTA, pH 8.0). The samples were then digested with 10 l of 2.5 mg/ml proteinase K for 5 min at room temperature. The solution was extracted with phenol/ chloroform (1:1, v/v) and precipitated by addition of 1 ml of 100% ethanol. The samples were the resuspended in formamide loading buffer (80% formamide, 1 mM EDTA, 0.1% xylene cyanol, 0.1% bromphenol blue) at 1,500 cpm/l, and a total of 5,000 cpm was loaded on a 6% polyacrylamide sequencing gel. The corresponding murine fVII sequence generated by the Maxam and Gilbert method (35) was placed in a lane alongside the reactions.
Methylation Interference Assays-Methylation interference was conducted as described previously (34). A 30-bp fragment extending from bp Ϫ45 to bp Ϫ66 of the mouse fVII promoter was subcloned into the SmaI site of the pBSKIIϩ vector (Stratagene). The resulting plasmid was linearized with either EcoRI or BamHI. A quantity of 20 pmol of plasmid was end-labeled with [␣-32 P]dATP and [␣-32 P]dCTP (Ͼ3,000 Ci/mmol), as described above. The labeled plasmid was then digested with the second enzyme to release the oligonucleotide insert. Labeled DNA was purified by polyacrylamide gel electrophoresis. An amount of 1 ϫ 10 7 cpm of labeled probe was mixed with 0.5 l of DMS in 200 l of DMS reaction buffer (50 mM sodium cacodylate, 1 mM EDTA, pH 8.0) for 2.5 min at room temperature. This was followed immediately by the addition of 40 l of DMS stop buffer (1.5 M NaOAc, 1 M ␤-mercaptoethanol, pH 7.0), 1 l of 10 mg/ml yeast tRNA solution, and 600 l of 100% ethanol. This solution was mixed and placed for 10 min in a dry ice/ethanol bath and microcentrifuged for 15 min at 4°C. The supernatant was discarded. This step was repeated three times, and the final pellet obtained was dried and resuspended in a buffer containing 10 mM Tris-HCl, 1 mM EDTA, pH 8.0, at 100,000 cpm/l. Five standard gel shift reactions using either Hepa 1-6 nuclear extract or purified Sp1 were conducted as described above. Following electrophoresis, the gels were exposed to film overnight at room temperature. Gel slices containing the bound and free probe were removed. The probe was then electroeluted from the slices, extracted with phenol/CHCl 3 (1:1, v/v), and precipitated with 100% ethanol. The probe was then incubated with 1 M piperidine at 95°C for 30 min, followed by three rounds of lyophilization, and resuspended in formamide loading dye (see previous section) at 1,500 cpm/l. A total of 5,000 cpm was loaded onto a 12% sequencing gel for electrophoretic analysis at 1500 V for 1.5 h. The gel was exposed to x-ray film overnight at Ϫ80°C with an intensifying screen.
Primers 3 and 6 were then end-labeled with polynucleotide kinase (Amersham Life Science) in the following manner. A concentration of 20 pmol of primer was combined with 2 l of 10 ϫ kinase buffer, 120 Ci of [␥-32 P]ATP (ICN, Costa Mesa, CA; 4500 Ci/mmol), and water to a final volume of 20 l. A 1:10 dilution of polynucleotide kinase (30 units/ml) was prepared with kinase buffer, and 1 l of this solution was added to the reaction tube. The reaction was incubated at 37°C for 30 min, after which a second 1-l aliquot of enzyme was added and incubation allowed to proceed for another 30 min. Next, water was added to bring the total volume to 50 l, followed by 50 l of 5 M NH 4 OAc. The reaction was then extracted with one volume of phenol-CHCl 3 (1:1, v/v) followed by another extraction with an equal volume of CHCl 3 -isoamyl alcohol (24:1, v/v), and then precipitated twice with 100% ethanol. The pellet was washed in 80% ethanol, dried, and resuspended in 20 l of autoclaved distilled water. A total of 1 l was removed and counted in a liquid scintillation counter.
In Vivo DNA Methylation-Hepa 1-6 cells (5 ϫ 10 7 cells) were centrifuged at 1500 rpm, resuspended in 2 ml of complete medium, and treated with 0.1% (v/v) DMS for 10 min at room temperature. The reaction was quenched by 10-fold dilution with cold PBS (0.137 M NaCl, 2.6 mM KCl, 10 mM Na 2 HPO 4 , 1.8 mM KH 2 PO 4 , pH 7.4), and the cells pelleted by centrifugation at 1500 rpm for 10 min at 4°C. The cell pellet was washed two additional times with 20 ml of 1 ϫ PBS and lysed overnight in 1.5 ml of harvest buffer containing 20 mM Tris-HCl, 20 mM NaCl, 20 mM EDTA, 0.1% SDS, and 300 g/ml proteinase K, pH 8.0, at 37°C. The methylated DNA was diluted to 3 ml with a buffer containing 10 mM Tris, 1 mM EDTA, pH 8.0, and then extracted once with an equal volume of phenol-CHCl 3 (1:1, v/v) and once with one volume of CHCl 3 -isoamyl alcohol (24:1, v/v). The DNA was precipitated with 100% ethanol (2.5 volumes), and the precipitate then washed with 80% ethanol, dried, and resuspended in 200 l of 1 M piperidine. Cleavage was accomplished at 95°C for 30 min, after which the sample was lyophilized, resuspended in 10 mM Tris, 1 mM EDTA, pH 8.0, and precipitated with ethanol as above. The DNA was then resuspended in 500 l of water and lyophilized for three cycles. After lyophilization, the DNA was resuspended in 200 l of water and its concentration determined by fluorimetry.
In Vitro Methylation of Genomic DNA-Genomic DNA was isolated from 2.5 ϫ 10 7 cells by centrifugation at 1500 rpm for 10 min and washing the pellet twice with 10 ml of 1 ϫ PBS. The resulting pellet was resuspended in 2 ml of harvest buffer and incubated overnight at 37°C. The sample was extracted with an equal volume of phenol/CHCl 3 (1:1, v/v), followed by 1 volume of CHCl 3 : isoamylalcohol (24:1, v/v), and precipitated with 2.5 volumes of 100% ethanol. The resulting DNA was methylated by treating 200 g of genomic DNA with 5 l of 20% DMS, 80% ethanol (v/v), followed by incubation at room temperature for 30 s. The methylation reaction was terminated by the addition of 50 l of cold stop buffer (1.5 M NaOAc, 1.0 M 2-mercaptoethanol, 100 g/ml yeast t-RNA) and 750 l of 100% ethanol. The DNA was pelleted, washed with 80% ethanol, dried, and resuspended in 200 l of 1.0 M piperidine. The cleavage reaction and lyophilizations were as listed in the preceding section.
LMPCR Protocol-This procedure was conducted as described (34), except that during exponential amplification, the time of extensions were increased by increments of 3 s/cycle, and during primer extension two to four cycles of PCR were completed, depending on the extent of labeling of the primers.
Analytical Techniques-Oligonucleotides were synthesized using the phosphoramidite method on a Beckman Oligo 1000M DNA synthesizer. DNA samples were sequenced by the dideoxy method (36) using the Sequenase version 2.0 DNA sequencing kit (U. S. Biochemical Corp.). For the larger pCAT constructs, nucleotide sequences were determined using the Automated Laser Fluorescent (ALF) Express DNA sequencer (Pharmacia) employing Cyp-5Ј-dATP labeling with two primers constructed within the pCAT vector sequences flanking the inserts on both sides. Electroelution of DNA samples was conducted in a Electrophoretic Concentrator Series 1750 -100 (ISCO, Inc., Lincoln, NE), according to product protocols.

RESULTS
CAT expression plasmids containing as possible promoter elements a variety of 5Ј sequences immediately upstream of the transcription initiation site of the murine fVII gene (16) were constructed, and their comparative abilities to drive expression of the CAT gene were assessed. The 5Ј-flanking constructs ranged in size from 7000 bp to 4 bp. Since fVII is present at a very low concentration in plasma, it is not surprising that its gene possesses a weak promoter, and only low expression levels of CAT were observed when the CAT gene was linked to the fVII 5Ј regions using the pCAT-B vector. Thus, all of the investigations reported herein were accomplished with the SV40 enhancer element as part of the CAT constructs (CAT-E vector). CAT assays were performed on each of the samples and the resulting thin layer chromatography data quantitated by phosphorimaging. All CAT activities of individual constructs were normalized for individual transfection efficiencies by determination of ␤-gal activities resulting from cotransfection of the murine Hepa 1-6 cells with the lacZ gene. Two separate expression experiments were performed on each construct, with the variation not greater than 10%. The construct with the highest expression, pCAT-97, was assigned a value of 100% (which was 9-fold over promoterless pCAT-E), with all other values relative to this construct. The final data are illustrated in Fig. 1 and show that a region 200 bp upstream of the ATG site contains the minimal promoter for the fVII gene. In confirmation of this point, removal of this 200-bp segment from the 1100-bp 5Ј-flanking region of the fVII DNA (providing pCAT-900Ј) reduced the promoter activity to base-line levels. This 200-bp DNA segment also contains the major transcription initiation site, which is positioned 9 bp upstream of the initiator ATG codon (16).
Several of the high probability transcription factor binding sites that reside within the 200-bp promoter region, as predicted by the MatInspector data base search (37), are depicted in Fig. 2, along with the nucleotide sequence of this segment of the fVII 5Ј-flanking DNA region. Notably, the area beginning around bp Ϫ120 is particularly well endowed with consensus sites for the liver-enriched transcription factors C/EBP␤ and HNF4, and the ubiquitous factors Sp1, H4TF1, and NF1.
DNase I footprinting of the 200-bp proximal region of the promoter with a mouse Hepa 1-6 liver cell nuclear extract showed a clear protection of a DNA segment encompassing Ϫ80 to Ϫ28 bp (Fig. 3), suggesting that liver factors are binding to this region of the fVII DNA. This footprint was observed on both the sense and antisense strands (Fig. 3), thus confirming the observation. This is the same region predicted from Fig. 2 to contain an array of potential transcription factor binding sites.
A synthetic oligonucleotide (M7H4) corresponding to nucleotides Ϫ44 to Ϫ66 of the gene, which contains the potential H4TF1 site, has been synthesized and labeled with 32 P. A gel shift assay with Hepa 1-6 cells (Fig. 4) demonstrated that this oligonucleotide binds to the cell extract (lane 2) and is effectively competed by the same unlabeled oligonucleotide (lanes 3  and 4). However, a synthetic unlabeled oligonucleotide (mM7H4) containing mutations in the H4TF1 consensus region did not compete for this same interaction (lanes 5 and 6). The corresponding sequence of the human fVII promoter (HSp1) also competed with labeled M7H4 for binding to the Hepa 1-6 nuclear extract (lanes 7 and 8), but a synthetic oligonucleotide with a mutation in the G-rich area (mHSp1) did not compete effectively with [ 32 P]M7H4 (lanes 9 and 10). Similarly, a synthetic unlabeled oligonucleotide containing the H4TF1 consensus sequence (33) was competitive in this assay (lanes 11 and  12), whereas a similar oligonucleotide containing the Sp1 consensus sequence was not competitive (lanes 13 and 14). This suggests that Sp1 is not a functional transcription factor in Hepa 1-6 cell extracts for regulation of the proximal murine fVII promoter element. To strengthen the finding that H4TF1 binds to this region of the promoter, and to differentiate between H4TF1 and Sp1 activities in Hepa 1-6 nuclear extract, the oligonucleotide containing the Sp1 consensus sequence was radiolabeled with 32 P and used in gel shifts with Hepa 1-6 nuclear extract (Fig. 5). In competition experiments, the unlabeled probe effectively displaced the labeled oligonucleotide (lanes 3 and 4), while the probe containing the H4TF1 consensus sequence did not ( lanes  5 and 7). The M7H4 oligonucleotide from the mouse fVII promoter slightly displaced the labeled probe (lanes 7 and 8), but the synthetic oligonucleotide containing the sequence from Ϫ83 to Ϫ108 of the human fVII promoter, which was proposed to contain the Sp1 transcription factor site (17), did not compete at all (lanes 9 and 10). This experiment, in combination with the results in Fig. 4, clearly showed a distinction between Sp1 and H4TF1 activities in Hepa 1-6 nuclear extract. The Ϫ44 to Ϫ66 bp region of the mouse promoter weakly competed for the Sp1 complex, although not as efficiently as the Sp1 consensus oligonucleotide, suggesting that transcription factor Sp1 may have a weak affinity for this site on the murine fVII promoter.
To identify further which of the two factors, viz. Sp1 or H4TF1, interacted with this region of the promoter of fVII, the Ϫ44 to Ϫ66 bp synthetic oligonucleotide (M7H4) of the murine fVII gene was labeled and used in gel supershifts with human factor Sp1 and mouse Hepa 1-6 nuclear extract (Fig. 6). The presence of the anti-human Sp1 antibody (which cross-reacts with murine Sp1) does indeed supershift the M7H4-human Sp1 complex (lanes 1-4), but does not supershift the M7H4-Hepa 1-6 complex (lanes 5-7). The data strengthen the claim that the complex formed with mouse Hepa 1-6 nuclear extract and the murine fVII proximal promoter element does not contain a contribution from Sp1. However, it does appear that purified Sp1 can bind to this sequence. One possible resolution of this apparent conflict is that H4TF1 possesses a higher affinity than Sp1 for the mouse-derived oligonucleotide probe and is responsible for the observed shift when using nuclear extract, whereas pure Sp1 is capable of interacting with this site only in the absence of H4TF1.
An extension of this experiment was performed to determine whether differences existed between the guanosine contacts observed in the M7H4-Hepa 1-6 nuclear extract complex and the M7H4-purified Sp1 complex. To address this question, methylation interference analysis was performed. The gels of Fig. 7 clearly indicate that the Hepa 1-6 nuclear extract (lanes 1-3) and purified Sp1 (lanes 4 -6) do not generate identical methylation protection patterns. Although they are very similar, it is seen from Fig. 7 that Sp1 does not protect the 5Ј-  4. Gel mobility shift assay confirms the binding of H4TF1 or a highly related factor, but not Sp1, to the region ؊44 to ؊66 of the murine fVII promoter. A probe consisting of bp Ϫ44 to Ϫ66, a region protected in the DNase I footprint analysis (Fig. 2), was tested for its ability to bind specifically to Hepa 1-6 nuclear extract. Panel A, the sequences of oligonucleotides used for competition are indicated. Regions very similar or identical to the H4TF1 consensus, 5Ј-GGGG-GAGGGG-3Ј, were aligned and are indicated by the boxed region. Panel B, competition experiment using a labeled probe consisting of bp Ϫ44 to Ϫ66 and various unlabeled competitor oligonucleotides with Hepa 1-6 nuclear extract. Lane 1 contains probe with no nuclear extract, whereas lane 2 contains probe with 12 g of Hepa 1-6 nuclear extract. A specific complex was formed, which is indicated by an arrow and was only competed by unlabeled M7H4 (lanes 3 and 4), by bp Ϫ83 to Ϫ108 of the fVII promoter, HSP1 (lanes 7 and 8), and the H4TF1 consensus oligonucleotide, H4TF1 (lanes 11 and 12). Mutant probes of the respective mouse (mM7H4) and human (mHSp1) sites showed no competition (lanes 5, 6, 9, and 10). In addition, a high affinity Sp1 oligonucleotide (Sp1) probe was unable to compete for this complex (lanes 13 and 14).
terminal guanosine (lane 5), while the Hepa 1-6 nuclear extract does. As expected, methylation interference was not observed in the coding strand (data not shown). Overall, these data indicate that the guanosine contacts supplied by H4TF1 in Hepa 1-6 nuclear extract with M7H4, and those of purified Sp1 and M7H4, are very similar, but not identical.
A putative HNF4 site is present in the mouse fVII proximal promoter in the region of nucleotides Ϫ33 to Ϫ50 (Fig. 2). To examine whether this factor actually interacted with this promoter element in nuclear extracts, a 32 P-labeled synthetic oligonucleotide (M7HNF4) corresponding to residues Ϫ33 to Ϫ50 of the mfVII gene was prepared and subjected to competition experiments with Hepa 1-6 nuclear extract. The results are shown in Fig. 8. Unlabeled M7HNF4 effectively competed for the observed complex (lanes 2-4). Additionally, an oligonucleotide containing a consensus HNF4 binding site (32) effectively competed with the radiolabeled M7HNF4 for the complex, as observed in lanes 5 and 6. However, a synthetic mutant HNF4 consensus oligonucleotide (mHNF4, Fig. 8) was not competitive for this binding. These results suggest that a HNF4 transcription site is harbored in residues Ϫ33 to Ϫ50 of the murine fVII gene. These results were verified by the antibody supershift experiments illustrated in Fig. 9, using the same 32 P-labeled M7HNF4 probe and nuclear extract. Lanes 3 and 4 of Fig. 9 show that addition of HNF4 antibody resulted in a supershift of the complex of M7HNF4 with Hepa 1-6 extract. This confirms that HNF4 binds to this region of the fVII gene.
Although a highly probable C/EBP␤ transcription site was identified by the consensus search at residues Ϫ47 to Ϫ78 of the mouse fVII gene (Fig. 2), the use of a 32 P-labeled synthetic oligonucleotide (M7C/H) containing this region of the murine fVII promoter as a probe in gel shifts with nuclear extract did not reveal a specific complex (data not shown). However, the data of Fig. 10 show that when the Hepa 1-6 nuclear extract is supplemented with Escherichia coli-expressed recombinant C/EBP␤, an additional complex was formed with M7C/H (lane 4). On the other hand, addition of bacterially expressed recombinant C/EBP␣ did not result in a complex with M7C/H. When a probe (M7mC/H) is used in which the C/EBP␤ site has been mutated, no new complex is observed upon addition of recombinant C/EBP␤ (lane 8). That the complex observed in the presence of C/EBP␤ represents specific binding of this protein factor is confirmed from the antibody supershift data of Fig. 11. Here, the entire complex formed (lane 3) after addition of nuclear extract, C/EBP␤, and radiolabeled M7C/H, is supershifted by the addition of C/EBP␤ antibody (lane 4), whereas that is not the case when a control antibody is added (lane 5). These experiments show that C/EBP␤ can form a complex with the murine fVII promoter in a sequence-specific manner. The observation that a specific C/EBP␤ complex was not observed when using Hepa 1-6 extract alone may be the result of the fact that C/EBP␤ is present at low concentrations in mature liver cells. Thus, its concentration in Hepa 1-6 nuclear extracts FIG. 5. Gel shift assay using an Sp1 consensus oligonucleotide reveals the formation of a specific Sp1 complex with Hepa 1-6 nuclear extract (indicated by arrows). Lane 1 contains probe with no nuclear extract, and lane 2 contains the labeled Sp1 consensus oligonucleotide with 12 g of Hepa 1-6 nuclear extract. This complex is competed by the addition of excess unlabeled Sp1 consensus oligonucleotide (lanes 3 and 4), but not by the addition of a unlabeled H4TF1 consensus oligonucleotide, H4TF1 (lanes 5 and 6). Neither the addition of the unlabeled murine fVII H4TF1 site oligonucleotide, M7H4 ( lanes  7 and 8), nor the addition of the unlabeled human fVII SP1 site oligonucleotide, HSP1 (lanes 9 and 10), resulted in efficient competition of this complex. The exact sequences of these oligonucleotides are provided in panel A of Fig. 3.   FIG. 6. Anti-Sp1 antibodies supershift the binding of purified human Sp1 to bp ؊44 to ؊66 of the murine fVII promoter but do not supershift the complex formed from Hepa 1-6 nuclear extract. The murine fVII H4TF1 site (bp Ϫ44 to Ϫ66) was labeled and used in an antibody supershift experiment with purified human Sp1 and Hepa 1-6 nuclear extract. Lane 1 contains probe with no nuclear extract. Lane 2 contains probe and 1 fpu of purified human Sp1. Lanes 3 and 4 contain probe and 1 fpu of purified human Sp1 followed by the addition of 1 and 2 l of anti-Sp1-IgG, respectively. Lane 5 contains probe with 12 g of Hepa 1-6 nuclear extract. Lanes 6 and 7 contain probe with 12 g of Hepa 1-6 nuclear extract followed by the addition of 1 and 2 l of anti-Sp1-IgG, respectively. may be too low for such a complex to be observed. Such a precedent with this transcription factor has been established (32).
The importance of these transcription factors in the in vivo regulation of the fVII gene was assessed by in vivo DMS methylation and analysis of the resulting genomic footprints by LMPCR. Autoradiograms of these experiments are provided in Fig. 12 and show protection of several G residues in each of the sites assigned by the above in vitro data to C/EBP␤, H4TF1/ Sp1, and HNF4. In addition, clear footprints have been found upstream of the C/EBP␤ site. These protected G residues, at sequence locations Ϫ89, Ϫ90, Ϫ108, and Ϫ109, are contained within two DNA segments (residues Ϫ81 to Ϫ98 and Ϫ106 to Ϫ121) predicted in Fig. 2 to be capable of binding to NF1. These latter two regions were not revealed in the in vitro footprint of Fig. 3. Additionally, in vitro gel shift experiments with synthetic oligonucleotides encompassing this region of the gene, as well as with NF1 consensus oligonucleotides, did not confirm the presence of NF1 sites. These negative results may be due to the experimental conditions chosen for the in vitro binding assays, or NF1 binding to the murine fVII promoter may require that other regions of the gene are brought into proximity through folding. In similar experiments with the top strand of DNA, no footprints were observed. DISCUSSION Coagulation fVII is a member of a closely related group of proteins, which also include fIX, fX, and PC and which share considerable homology in their domain organizations. These proteins play different roles in blood coagulation and anticoagulation, suggesting that only minor overall evolutionary changes have occurred in diversifying their functions. The correspondence of the domain structures of the proteins and the exon placements within the genes demonstrate that evolution of this family of proteins probably occurred by exon shuffling among their genes, perhaps facilitated by intronic recombination mechanisms (38). Other proteins that contain some of these domains, such as those containing kringle and epidermal growth factor modules, may also have evolved, at least in part, in this manner.
It is relevant to determine whether expression of genes of these related proteins is regulated by similar factors, and whether tissue-specific factors are obvious regulatory elements. In this regard, fVII is an excellent paradigm for such an investigation because, unlike fX (39) and PC (40), and similar to fIX (41), fVII biosynthesis is much more highly confined to the liver. In addition, steady state mRNA levels of fVII appear to correlate well with the plasma concentration of this protein (17). The recent characterization of mice containing a targeted deletion of the fVII gene (30) provides further impetus for an understanding of the transcription factors that regulate fVII FIG. 7. Identification of the contact sites for Hepa 1-6 nuclear extract and purified human Sp1 on the ؊44 to ؊66 region of the murine fVII promoter by methylation interference footprinting. The partially methylated, radiolabeled, murine fVII H4TF1 site probe (bp Ϫ44 to Ϫ66) was incubated with Hepa 1-6 nuclear extract (lanes 1-3) or purified human Sp1 (lanes 4 -6). Free and bound probes were separated on a 6% nondenaturing polyacrylamide gel, isolated, and cleaved with piperidine. Cleavage products of both free (lanes 1, 3, 4, and 6) and bound probe (lanes 2 and 5) were analyzed on a 12% sequencing gel. The non-coding strand is shown.

FIG. 8. Gel mobility shift assay confirms the binding of HNF4
to the ؊33 to ؊50 bp region of the murine fVII promoter. A probe consisting of bp Ϫ33 to Ϫ50 of the mouse fVII promoter, containing a putative HNF4 site, was tested for its ability to bind specifically to Hepa 1-6 nuclear extract. Lane 1 contains probe with no protein. Lane 2 contains probe with 12 g of Hepa 1-6 nuclear extract. A specific complex is formed (indicated by the arrow), which is competed by the addition of unlabeled self, M7HNF4 (lanes 3 and 4), and by the addition of a unlabeled HNF4 consensus oligonucleotide probe, HNF4 (lanes 5 and 6). The complex is not competed, however, by the addition of an unlabeled mutant HNF4 consensus probe, mHNF4 (lanes 7 and 8). The sequences of the probes used in the competition experiments are shown above.
gene transcription in order that the full applicability of this murine system to that of humans can be more completely assessed. This investigation has been accomplished, and the totality of in vitro and in vivo data present in this manuscript show that H4TF1, HNF4, and C/EBP␤ potentially function as transcriptional regulators of the murine fVII gene. There also exists a strong possibility that two additional NF1 sites are present in the 5Ј-flanking DNA region. All of these factor binding sites exist within 130 bp upstream of the transcription initiation site, thus showing that defined regulatory elements for expression of murine fVII are present in a relatively small region of the 5Ј-flank of this gene. H4TF1 was first identified as a factor that binds the human histone H4 promoter (33). It was later partially purified and shown to be distinct from Sp1 (42), although its recognition sequence, 5Ј-GGGGGAGGGG, is very similar to that of Sp1, 5Ј-GGGGCGGGG. H4TF1 has been implicated as an important regulatory element in the promoters of human eosinophil peroxidase (43), and human pyruvate dehydrogenase-␤ (44). In addition, a non-Sp1 factor that recognizes the sequence 5Ј-GGGGGAGGGG has been described as regulating the human ␤-enolase gene (ENO-3) (45) and the immunoglobulin heavy chain enhancer, 3Ј-␣E(hsl,2) (46). A G-rich factor binding element, 5ЈGGGGGAGGGG, was identified in the human fVII promoter, which matches the consensus sequences for both H4TF1 and Sp1 (17). Gel shift analysis revealed two closely spaced complexes (A and B), both shown to be specific. The slower moving complex A was shown to contain Sp1, while complex B did not contain this factor. A single point mutation completely abolished the formation of both complexes. It was concluded that the loss of both complexes on a gel shift assay with the mutation of a single nucleotide was consistent with the existence of overlapping binding sites for Sp1 and another nuclear protein. A separate study identified the same G-rich sequence as being important in transcription factor binding to human fVII (47). This latter group found that when a Sp1 sequence from the human metallothionein IIA gene (MIIA probe) was labeled, unlabeled MIIA probe efficiently competed for the complex formed, but the fVII oligonucleotide probe of the putative Sp1 site did not. This is difficult to reconcile if this site is in fact a Sp1 binding locus, but not if it is a H4TF1 site. Our results clearly show the existence of a specific complex on gel shifts, the formation of which is inhibited by a H4TF1 consensus oligonucleotide, but not by one containing the Sp1 consensus sequence (Fig. 4). We also show that a monoclonal antibody to Sp1 did not react with this complex formed from Hepa 1-6 cell nuclear extract. This antibody, does, however, completely supershift recombinant human Sp1 bound to the fVII G-rich element (Fig. 6). Thus, while recombinant human Sp1 can bind to this sequence in the absence of Hepa 1-6 nuclear extract, it appears that H4TF1 in Hepa 1-6 nuclear extract has a much higher affinity for the G-rich element in the murine fVII promoter, thereby effectively precluding Sp1 binding. It was possible to effectively distinguish between the binding activities of Sp1 and H4TF1 in nuclear extract by using a labeled Sp1 consensus oligonucleotide in gel shifts. In this case, unlabeled Sp1 oligonucleotide was able to compete for the complex, whereas the H4TF1 consensus oligonucleotide did not. In addition, the mouse fVII G-rich element competed very poorly, A probe consisting of bp Ϫ47 to Ϫ78 of the murine fVII promoter, and containing a putative C/EBP␤ site, was tested for its ability to bind specifically to Hepa 1-6 nuclear extract alone or Hepa 1-6 nuclear extract plus 100 ng of recombinant C/EBP␣ or C/EBP␤. Lane 1 contains probe without protein. Lane 2 contains probe with 12 g of Hepa 1-6 nuclear extract. Lane 3 contains probe with 12 g of Hepa 1-6 nuclear extract, plus 100 ng of recombinant C/EBP␣. Lane 4 contains probe with 12 g of Hepa 1-6 nuclear extract, plus 100 ng of recombinant C/EBP␤. The C/EBP␤-DNA complex is indicated by an arrow. Lanes 5-8 are identical to lanes 1-4 but contain a probe consisting of bp Ϫ47 to Ϫ78 of the murine fVII promoter with a mutant C/EBP␤ site. The sequences of the oligonucleotide probes used are shown above the gels. and the human G-rich element did not compete at all in this set of experiments. Methylation interference assays (Fig. 7), using Hepa 1-6 extract and recombinant human Sp1, showed that although both H4TF1 in Hepa 1-6 nuclear extract and recombinant Sp1 recognize overlapping regions of the fVII promoter, these binding sites are not identical. Taken together, these data strongly implicate the binding of H4TF1, and not Sp1, to this sequence in nuclear extracts.
HNF4 is a novel member of the steroid hormone receptor superfamily and is expressed in liver, kidney, and intestine (32,48). It has been described as a factor crucial for the liverselective transcription of genes encoding ornithine transcarbamylase (49), transthyretin (50), and apolipoprotein CIII (51). This transcription factor binds to the promoter region of the human fX gene and has been shown to be required for its expression (52). Disruption of a binding site for HNF4 in the human fIX promoter results in one of the hemophilia B-Leiden variants (53). HNF4 also functionally binds to a similar region of the promoter of human fVII (17,47). We have clearly identified a HNF4 site within nucleotides Ϫ33 to Ϫ50 of the murine fVII promoter through gel mobility shift assays (Fig. 8) and antibody supershift assays (Fig. 9). The importance of this site resides in the liver-specific expression of this gene and shows that the necessary machinery for this critical property of the gene is indeed present.
C/EBP␤ is a liver-enriched member of the basic domainleucine zipper protein family (54). The originally identified member of this family, which is also liver-enriched, is C/EBP␣ (55). Other designations for C/EBP␤ are NF-IL6 (56), LAP (57), IL-6DBP (58), AGP/EBP (59), and CRP2 (60). Regulation of the expression of this transcription factor is involved in the expression of several liver-specific genes since C/EBP␤ is not highly expressed in adult hepatocytes but is expressed in the hepatoma cell lines, HepG2 and Hepa3B (61). In adult hepatocytes, however, it is strongly induced after lipopolysaccharide, IL-1, or IL-6 stimulation, suggesting that it may be involved in acute phase gene induction (56,59,62). In confirmation of this, C/EBP␤ has been shown to bind to promoters of certain human acute phase responsive proteins, such as albumin (61), ␣ 1antitrypsin (50), ␣ 1 -acid glycoprotein (63), C-reactive protein FIG. 11. Anti-C/EBP␤ antibodies supershift the binding of recombinant C/EBP␤ to bp ؊47 to ؊78 of the murine fVII promoter. A probe consisting or bp Ϫ47 to Ϫ78 of the murine fVII promoter and containing a putative C/EBP␤ site was labeled and used in an antibody supershift experiment with Hepa 1-6 nuclear extract and recombinant C/EBP␤. Lane 1 contains probe without protein. Lane 2 contains probe with 12 g of Hepa 1-6 nuclear extract. Lane 3 contains probe with 12 g of Hepa 1-6 nuclear extract plus 100 ng of recombinant C/EBP␤. Lane 4 is identical to lane 3, except that 2 l of anti C/EBP␤ polyclonal antibody was added to the binding reaction. Lane 5 is identical to lane 3, except that 2 l of anti NF-B polyclonal antibody was added to the binding reaction. The C/EBP␤-DNA complex and supershift are indicated by arrows.
FIG. 12. In vivo methylation pattern of the lower strand DNA using primers 1, 2, and 3, located 5 of the transcription factor binding sites. Along side each symbol or marker denoting protection is the corresponding nucleotide number of the G residue on the promoter sequence. Control DNA was derived from a murine B lymphocyte cell line (S194), which is identical in sequence to the HEP 1-6 cells in the areas studied. The location of each transcription factor binding site is denoted by solid bars, and the identity of each is listed. (62), haptoglobin (62), and hemopexin (64). Additionally, a switch from C/EBP␣ to C/EBP␤ as a transcriptional regulatory element accompanies the acute phase induction of ␣ 1 -acid glycoprotein (50). The induction of C/EBP␤ could also explain the higher concentration of this factor in HepG2 and Hepa3B cells, as a response to the transformation events required to establish these cell lines. We have assigned a C/EBP site within the Ϫ47 to Ϫ78 region of the gene of the murine fVII promoter by gel shift (Fig. 10) and antibody supershift (Fig. 11) assays. In addition, we have shown that recombinant C/EBP␤, but not recombinant C/EBP␣, binds to this site (Fig. 10). A specific C/EBP␤ complex with nuclear extract from the murine hepatoma cell line Hepa 1-6 was not observed, but unlike HepG2 and Hepa3B, Hepa 1-6 may more accurately reflect the physiological expression pattern of C/EBP␤ in that little or no C/EBP␤ is present in adult liver cells prior to induction with lipopolysaccharide or cytokines. The fact that C/EBP␤ binds the murine fVII promoter raises the possibility that murine fVII may be an inducible protein (65,66), and responsive to the state of differentiation of cells and to hormones (54).
In addition to these sites, which have been confirmed through in vitro experiments, in vivo DMS footprint analyses (Fig. 12) showed protection of G residues involved in each of the C/EBP␤, H4TF1, and HNF4 binding sites (Fig. 2). In addition, four other G residues upstream of the C/EBP␤ site were protected from methylation, two each of which are contained in separate regions of the promoter. These two 5Ј regions include TGG sequences, which are located near the protected G residues. These TGG sequences have been found to be important in recognition of the NF1 family of transcription factors (67).
Finally, an illustration of the transcription factors that have been identified to be potentially functional in liver as regulators of transcription of the genes of this gene family, including fVII, fIX, fX, and PC, is provided in Fig. 13. These sites have all been identified through in vitro studies. The only exception to this is the murine fVII gene in the current investigation, wherein in vivo experiments (Fig. 12) confirm the sites identified in vitro and locate additional factor sites. There are many similarities in the genes in the regulatory elements in this general location, which are within Ϯ200 bp of their major transcription start sites. Specifically, a HNF4 binding site is found within 60 bp of the major transcription initiation site in all of the genes, except that of human PC. However, in this case, both HNF1 and HNF3 sites are present in this region. Thus, there is an obvious basis for liver-enriched transcription of these genes.
Only two of the genes possess C/EBP binding sites, viz. mouse fVII (C/EBP␤) and human fIX (C/EBP␣). The importance of this transcription factor in liver expression of genes is underpinned by the finding that disruption of its binding site in the fIX gene results in a hemophilia B-Leiden variant (68). Additionally, because of the existence of the C/EBP␤ transcription factor binding site in the murine fVII gene, it is possible that fVII is an acute phase protein in the mouse. The proximal promoter region of human fVII does not interact with this transcription factor, perhaps indicative of a species difference in the acute phase response of fVII.
A site for the ubiquitous H4TF1 transcription factor overlaps with that of a Sp1 site in the murine fVII gene. A Sp1 site is present in a similar location in the human fVII and human fX genes, with another further downstream in the human PC gene. It is possible that this overlapping H4TF1 site is present in the Sp1 location in these other genes, especially that of human fVII, since the properties described in the relevant papers for the Sp1 site would be consistent with additional binding of H4TF1, or with H4TF1 as a replacement for Sp1. Additionally, the oligonucleotide corresponding to this region of the human fVII gene did not appear to react with purified Sp1 protein (Fig. 5).
Of the fVII-related genes discussed in this report, only human fIX, and possibly murine fVII, possess NF1 binding sites, and these exist in similar locations in the promoter element. Since the human fX gene does not have a defined major transcription start site, but multiple initiation sites, the numbering in this case was from the translation start site. The lengths of the ovals represent the nucleotides involved in binding of that particular factor. DBP, D-site-binding protein; NF1L, nuclear factor 1-liver specific; NFY, nuclear factor Y; Sp1, stimulating protein 1. The unfilled ovals represent factors that bind to the indicated regions of the genes that have not been identified. The following sources were used to prepare this illustration; human fVII (17,47,73), murine fVII (this study); human fIX (74 -76), human fX (52) (77,78), and human PC (79). This additional transcription factor binding site(s) in these genes may partly explain their strong liver expression capabilities, especially since at least one member of this family of proteins, murine NF1-B3 (identical to the rat and chicken isoforms, NF1-L and cNF1-A1.1, respectively), is a liver-enriched transcription factor. Also, it is known that NF1 cooperates with other liver factors, such as HNF1, HNF3, and HNF4, to enhance target gene transcription (69 -71), or transcription of genes for other regulatory factors, such as C/EBP␤ (72). Thus, NF1 may play these types of direct or indirect roles in human fIX and mouse fVII expression.
In summary, we have shown that HNF4, C/EBP␤, H4TF1, and possibly NF1 binding sites exist within the minimal promoter region of the murine fVII gene, as defined in mouse Hepa 1-6 liver cells. It is likely that other positive and negative regulatory factors that bind further upstream or downstream of the region investigated also influence transcription of this gene. However, those factors that have been identified herein are capable of eliciting the known expression pattern of the gene, and further suggest directions for investigation of other regulatory properties of fVII gene expression.