Functional characterization of the human factor VII 5'-flanking region.

Factor VII is a vitamin K-dependent coagulation protein essential for proper hemostasis. The human Factor VII gene spans 13 kilobase pairs and is located on chromosome 13 just 2.8 kilobase pairs 5′ to the Factor X gene. In this report, we show that Factor VII transcripts are restricted to the liver and that steady state levels of mRNA are much lower than those of Factor X. The major transcription start site is mapped at −51 by RNase protection assay and primer extension experiments. The first 185 base pairs 5′ of the translation start site are sufficient to confer maximal promoter activity in HepG2 cells. Protein binding sites are identified at nucleotides −51 to −32, −63 to −58, −108 to −84, and −233 to −215 by DNase I footprint analysis and gel mobility shift assays. A liver-enriched transcription factor, hepatocyte nuclear factor-4 (HNF-4), and a ubiquitous transcription factor, Sp1, are shown to bind within the first 108 base pairs of the promoter region at nucleotide sequences ACTTTG and CCCCTCCCCC, respectively. The importance of these binding sites in promoter activity is demonstrated through independent functional mutagenesis experiments, which show dramatically reduced promoter activity. Transactivation studies with an HNF-4 expression plasmid in HeLa cells also demonstrate the importance of HNF-4 in promoting transcription in nonhepatocyte derived cells. Additionally, the sequence of a naturally occurring allele containing a previously described decanucleotide insert polymorphism at −323 is shown to reduce promoter activity by 33% compared with the more common allelic sequence.

Factor VII, a vitamin K-dependent clotting protein, plays a critical role in coagulation. Vessel wall injury exposes the membrane-bound glycoprotein tissue factor to circulating Factor VII, allowing the two proteins to form a complex. This facilitates cleavage of the Factor VII zymogen to its active form, Factor VIIa, which in concert with tissue factor further activates the coagulation cascade (1).
Six proteins of the coagulation system undergo a unique posttranslational modification, ␥-carboxylation of glutamic acid residues in the amino terminus, which requires the presence of vitamin K (2, 3). These vitamin K-dependent clotting proteins (prothrombin; protein C; protein S; and Factors VII, IX, and X) share many structural and functional similarities. More specifically, Factors VII, IX, and X and protein C have 8 exons, each of which corresponds to conserved functional domains of the proteins (4). Unlike the promoter regions of most eukaryotic promoters, the promoter regions of the vitamin Kdependent clotting proteins lack TATA boxes (5)(6)(7)(8). Additionally, the sequence element ACTTTG is found in similar locations in the proximal promoters of Factors VII, IX, and X. In Factor IX, this element binds to the liver-enriched transcription factor hepatocyte nuclear factor-4 (HNF-4) 1 (9).
Among the coagulation proteins, Factor VII has the shortest half-life, approximately 2-3 h (10). Factor VII also circulates at extremely low concentrations (0.5 g/ml) (11), less than onetenth of the concentrations of the other vitamin K-dependent coagulation proteins (12). However, the molecular mechanisms controlling these low Factor VII plasma concentrations have not been defined.
Clinical studies have shown correlations between Factor VII levels and a large number of variables, most notably cardiovascular disease. In 1986, the Northwick Park Heart Study, a prospective study of adverse cardiovascular events in 1511 middle-aged males, showed that higher concentrations of Factor VII coagulant activity were associated with an increased risk of ischemic heart disease (13). Factor VII levels have also been correlated with age (14,15), cholesterol, and triglyceride levels (16), dietary saturated fat intake (17), menopause (15,18), oral contraceptive pills (19), body mass index (20), testosterone concentrations in men (21), and a polymorphism within exon 8 encoding amino acid residue 353 (22). The presence of a common polymorphic decanucleotide insert (allele frequency 0.2) in the proximal human Factor VII promoter region (23) was recently shown to correlate with both a lower Factor VII antigen and coagulant activity (24).
In the present study, we characterize important cis-and trans-acting functional elements in the human Factor VII promoter, including sequence analysis of 3958 bp of 5Ј-flanking DNA, delineation of transcription start sites, analysis of promoter activity using reporter gene assays, and determination of protein binding regions by DNase I footprint analysis and gel mobility shift assays. We also compare promoter activity between the two alleles with and without the decanucleotide insert 323 bp 5Ј of the translation start site (23). We show by Northern blot analysis that synthesis of Factor VII mRNA is restricted to the liver. Using mutagenesis of protein-binding sites along with reporter gene assays, we demonstrate the critical role played by cis elements at Ϫ63 3 Ϫ58 and Ϫ108 3 Ϫ84, and their cognate binding proteins, HNF-4 and Sp1.

EXPERIMENTAL PROCEDURES
Genomic Library Screening, Subcloning, and Sequencing-A PCR fragment generated from the 279 bp immediately upstream of the Factor X translation start site was used as a probe to screen a human cosmid library as described previously (7). This cosmid clone contains 3.5 kb of the Factor X gene, the 2.8-kb Factor VII-Factor X intergenic region (8), the entire 12.8-kb Factor VII gene and approximately 10 -15 kb of the Factor VII 5Ј-flanking sequence.
DNA Sequence Analysis-The 5Ј-flanking DNA generated by EcoRI cleavage between base pairs Ϫ309 and Ϫ3958 of the human Factor VII gene was subcloned into a pUC19 vector for DNA sequence analysis. Oligonucleotide sequencing primers (Integrated DNA Technologies) were synthesized at approximately 300-bp intervals, and sequencing was performed using 7-deaza-dGTP termination mixes (U. S. Biochemical Corp.) to minimize compression bands using the dideoxy Sanger sequencing method (25). All sequences were confirmed on both strands. The sequence was scanned for putative transcription factor binding sites using version 2.3 of the TRANSFAC data base. 2 Northern Blot Analysis-A 1.3-kb EcoRI-EcoRI fragment of Factor X cDNA and a 1.5-kb HindIII-EcoRI fragment of Factor VII cDNA were randomly labeled with [␣-32 P]dCTP using a Random Primed DNA labeling kit (Boehringer Mannheim) according to the specifications of the manufacturer, yielding probes with specific activities of 1.2 ϫ 10 9 cpm/g and 8 ϫ 10 8 cpm/g, respectively. On consecutive days, the Factor X and Factor VII cDNA probes were hybridized for 1 h at 65°C in ExpressHyb hybridization solution (Clontech) with multiple tissue Northern blots containing approximately 2 g/lane of human poly(A) ϩ RNA from 16 different tissues (Clontech). Densitometric measurements were performed on autoradiograms exposed to the membrane for 12 h using a Molecular Dynamics densitometer and ImageQuant software.
Primer Extension Assay-Primer extension was performed using an end-labeled 22-bp oligonucleotide (ϩ21 to ϩ42). Total human liver RNA was isolated from discarded segments of a donor transplant specimen using an RNA extraction kit (Pharmacia Biotech Inc.). One million cpm of primer was added to 50 g of total human liver RNA and incubated for 90 min at 65°C. The RNA was reverse transcribed for 1 h at 42°C using 8 units of avian myeloblastosis virus reverse transcriptase (Promega) in a buffer containing 20 mM Tris-HCl (pH 8.3), 10 mM MgCl 2 , 0.5 mM dithiothreitol, 0.15 mg/ml actinomycin D, 14 mM 4-dNTP mix, and 50 mM NaCl. After reverse transcription, RNase A (2.1 g) and 10.5 g of salmon sperm DNA were added to the reaction in a total volume of 150 l. The product was precipitated, phenol/chloroform extracted, resuspended in an 80% formamide loading solution, and analyzed on a 6% polyacrylamide gel alongside a sequencing reaction using [ 35 S]dATP and the ϩ21 to ϩ42 bp probe as a sequencing primer. The primer extension reaction was also performed with Superscript II® (Life Technologies, Inc.) and poly(A) ϩ RNA using the ϩ21 to ϩ42 bp primer. The procedure was as described previously with the following minor changes. 1 g of poly(A) ϩ human liver RNA was used in place of the total liver RNA, and the reverse transcriptase buffer contained the following: 50 mM Tris-HCl (pH 8.3), 75 mM KCl, 3 mM MgCl 2 , 10 mM dithiothreitol, and 5 mM 4-dNTP mix. Poly(A) ϩ liver RNA was prepared using a PolyATtract® system 1000 kit (Promega).
RNase Protection Assay-The transcriptional start site of the human Factor VII gene was analyzed using an RNase protection assay essentially as described previously (27). A 550-bp PCR-generated fragment spanning from Ϫ508 to ϩ42 was subcloned into the SmaI site of pBlue-script® II SK (ϩ) (Stratagene), and the correct sequence was verified to ensure that no polymerase-induced errors were introduced. The plasmid was linearized and transcribed with [␣-32 P]CTP using 10 units of T7 RNA Polymerase (Promega). Approximately 500 ng of the Factor VII riboprobe, with a specific activity of 1 ϫ 10 6 cpm/l, was dissolved in hybridization buffer (80% formamide, 40 mM PIPES (pH 6.4), 0.4 M NaCl, 1 mM EDTA), combined with 10 g of total liver RNA and denatured for 5 min at 85°C before hybridization for 16 h at 45°C. The hybridized samples were then incubated for 30 min at 30°C with either RNase T1 alone (2 g/ml) or RNase T1 (2 g/ml) ϩ RNase A (10 g/ml). The positive control, pTRI-␤-actin (Ambion), a 249-bp ␤-actin fragment cloned into the pTRIPLEscript vector, was hybridized with 10 g of human liver RNA under the same conditions described above for the Factor VII probe. The negative control was the Factor VII probe hybridized with 10 g of tRNA (Life Technologies, Inc.) under identical conditions. The RNA Century Marker template set (Ambion) was used to generate 100-bp RNA markers using the protocol described by the manufacturer. The distance migrated by the Factor VII product on a 6% polyacrylamide gel was used to calculate the nucleotide length of the protected fragment based on an equation generated by plotting the known RNA marker size versus distance traveled using semilog analysis (28).
Construction of Growth Hormone Reporter Gene Plasmids-The promoterless plasmid pOGH (Nichols Diagnostics Institute), a 4.817-kb modified pUC12 plasmid containing the structural human growth hormone gene, was used for reporter gene plasmid constructs (29). All inserts were cloned into the multiple cloning site between the BamHI and HindIII sites, spaced 17 bp apart. The 3Ј-end of all inserts ended at position ϩ1, the first nucleotide of the translation start site (6). The 75-, 186-, 291-, 404-, and 501-base pair inserts were synthesized using PCR primers with HindIII and BamHI sites artificially installed at the 5Јand 3Ј-ends, respectively. The PCR fragments were subcloned into the pOGH vector by directional cloning. The 109-bp constructs (wild-type and mutant) and the 162-bp construct were generated using a 3Ј-end primer with an artificially installed BamHI site and 5Ј-end primers containing 24 bp of the desired insert sequence. The PCR products were digested with BamHI and subcloned into a pOGH plasmid. A 186-bp construct (p186⌬ACTTTG) with a mutation installed between Ϫ63 and Ϫ58 (ACTTTG to GACAAT) was generated by overlapping PCR (30,31). The 5Ј and 3Ј fragments were first synthesized using primers within the pOGH vector, and a mutated oligonucleotide sequence between bases Ϫ76 and Ϫ47 using the wild-type 186-bp pOGH construct as template. The two fragments were subsequently combined in a second PCR step to generate a mutated fragment, which was digested with BamHI and HindIII and subcloned by directional cloning into the pOGH vector. Inserts for the constructs containing 713, 1127, 1426, 1746, 3959 bp were produced by digestion of the pUC19-Factor VII 5Ј-flanking DNA plasmid with SfiI on the 3Ј-end and digestion and blunt ending with SmaI, StuI, EcoRV, XmnI, and EcoRI, respectively, on the 5Ј-ends. These fragments were subcloned into the 501-bp Factor VII promoter pOGH construct with an SfiI-digested 3Ј-end and a Klenow-blunt ended HindIII site at the 5Ј-end. The following portions of sequences were verified by sequence analysis to ensure that no errors were introduced: all PCR-generated inserts, the flanking regions of the pOGH plasmid, and 200 bp of sequence surrounding the SfiI-ligated inserts. Two constructs were synthesized using different DNA samples as template, one from an individual whose DNA contained a decanucleotide sequence insert (723 bp), and the other from an individual who lacked the insert (713 bp); duplicate preparations were made of each of these constructs. All constructs were purified using double CsCl gradient centrifugation preparation and run on a 1% agarose gel to verify the integrity of the in-tact closed circular plasmids.
Transient Transfection Assay-Calcium phosphate transfection was used for all transient transfection assays of HepG2 (32) and HeLa cells. HepG2 and HeLa cells were cultured to confluency in 75-cm flasks in Dulbecco's modified Eagle's medium containing 10% heat-inactivated fetal bovine serum, 2 mM L-glutamine, 100 units/ml penicillin, and 100 g/ml streptomycin sulfate (Life Technologies, Inc.). 24 h prior to transfection, cells were plated on 60-mm tissue culture dishes in 5 ml of medium at a density of 1 ϫ 10 6 and 5 ϫ 10 5 cells/plate for HepG2 and HeLa cells, respectively. Calcium phosphate transfection was carried out as described previously (33). For transfection in HepG2 cells, 1 pmol (approximately 3 g) of the growth hormone reporter gene construct and 2.5 g of an internal control plasmid pLuc109 (34) were transfected. 5 pmol of reporter gene constructs were used for transfection of HeLa cells. Medium and cell lysate were sampled 48 h after replacement with fresh medium. Transactivation experiments in HeLa cells were performed with 15 g of the 185-bp reporter gene construct; 10 g of pLEN4S (35) or an equimolar amount of pLEN was added. The HNF-4 expression plasmid, pLEN4S (gift from Dr. James Darnell, Rockefeller University), contains the entire 3-kb HNF-4 cDNA cloned into pLEN, a 5-kb expression plasmid containing the SV40 enhancer, the human metallothionein promoter, and the human growth hormone 3Ј-untranslated region. The difference in DNA amount between pLEN4S and pLEN was made up by addition of pUC19 plasmid DNA to transfections with pLEN. The Student's t test was used for statistical comparisons.
Cell Extract Preparation and Luciferase Activity Measurement-Cell lysate preparation and luciferase activity measurement were performed as described by Ramachandran et al. (36). Luciferase activity was measured in duplicate using a Berthold Lumat LB 9501 luminometer.
Human Growth Hormone Assay-100 l of medium was used for radioimmunoassays to determine concentrations of human growth hormone using a commercially available kit (Nichols Institute Diagnostics).
DNase I Footprint Analysis-The DNase I footprint analysis was carried out as described by Lichtsteiner et al. (37). Approximately 10 g of pOGH containing 309 bp of Factor VII (Ϫ309 to Ϫ1) was digested with BamHI, dephosphorylated by calf intestine alkaline phosphatase (Life Technologies, Inc.), end-labeled with [␥-32 P]ATP, and subsequently digested with HindIII. Approximately 10 ng of probe (6 ϫ 10 4 cpm) was incubated for 30 min on ice with 50 g of nuclear extract in 40 l of buffer (25 mM Hepes (pH 7.6), 60 mM KCl, 7.5% glycerol, 0.1 mM EDTA, 0.75 mM dithiothreitol, 5 mM MgCl 2 , and 1 g of poly(dI⅐dC). At room temperature, digestion with the addition of 2.5 units of DNase I (Boehringer Mannheim) was allowed to proceed for exactly 1 min. To the control without nuclear extracts, only 0.1 unit of DNase I was added. The corresponding Factor VII sequence generated by the Maxam and Gilbert method (38) was run alongside the reactions.
Preparation of Nuclear Extracts-Nuclear extracts were prepared according to the method of Dignam et al. (39). Human liver nuclear extracts were prepared as described by Gorski et al. (40). The protein concentration was determined by the Bradford method (41). Extracts were dialyzed and frozen in 50-l aliquots at Ϫ80°C.
Gel Mobility Shift Assays-The procedure for gel mobility shift assays was similar to that of Chodosh et al. (42). Complementary singlestranded oligonucleotides (24 -36 nucleotides in length) were annealed, labeled, and purified using Sephadex G-50 quick spin columns (Boehringer Mannheim). Specific activities of approximately 10 6 cpm/ng DNA were routinely obtained. 10 -20 g of nuclear extracts were preincubated for 10 min on ice with 0.5-3 g bovine serum albumin, 2.5 g of poly(dI⅐dC) (Pharmacia), and 1 g of salmon sperm DNA, in 20 l of binding buffer (25 mM Hepes, (pH 7.6), 14 mM KCl, 10% glycerol, 0.1 mM EDTA, 0.75 mM dithiothreitol, 5 mM MgCl 2 ) before 10 5 cpm of DNA probe was added. Samples were incubated on ice for 30 min. In competition studies, unlabeled oligonucleotide was added during the preincubation step. The Sp1 consensus oligonucleotide (Promega) contains the following sequence: 5Ј-ATTCGATCGGGGCGGGGCGAGC-3Ј. In supershift experiments, antibody was added after the 10-min incubation period of the probe with the nuclear extract mixture. The Sp1 antibody (Santa Cruz Biotechnology) does not cross-react with Sp2, Sp3, or Sp4. The HNF-4 antiserum (rabbit) was raised against a peptide corresponding to amino acids 445-455 of the rat HNF-4 (gift from Dr. James Darnell and Dr. Frances Sladek, Rockefeller University). In studies with purified Sp1, one footprint unit of Sp1 (Promega) was used in place of the nuclear extract. Gel mobility shift assays were performed on 5% polyacrylamide gels (acrylamide/bisacrylamide, 29:1) in 0.5 ϫ Tris borate/EDTA for 2 h at 4°C after prerunning the gel for 2 h at 200 V.

Northern Analysis of FVII Reveals Liver-specific Expres-
sion-Northern analysis of human poly(A) ϩ RNA revealed a single 2.7 kb band present only in liver (Fig. 1, arrow at left) using blots containing 16 different human tissue sources. Steady state Factor VII mRNA levels were 6% of those of Factor X mRNA levels, 3 as determined by densitometric measurement of autoradiograms of the blot incubated sequentially with Factor VII and Factor X cDNA probes of comparable specific activity. These levels of steady state mRNA correlate well with molar plasma levels (12) of Factor VII (0.009 M) and Factor X (0.18 M).
DNA Sequence of the Human Factor VII Gene 5Ј-Flanking Region-The 5Ј-flanking sequence of the proximal promoter of the human Factor VII gene is shown in Fig. 2. Minor differences were observed between the first 521-bp sequence reported by O'Hara et al. (6) (isolated from HepG2 cells) and this sequence. In our sequence an extra C was seen at Ϫ149, and an extra G was seen at Ϫ459.
The TRANSFAC data base 2 was used to search for putative transcription factor binding sites. Potential binding sites were The proximal 455 bp of sequence are derived from an allele, which does not contain the previously described decanucleotide insert at Ϫ323 bp. The sequence from Ϫ3958 to Ϫ456 was obtained from an allele that contains the decanucleotide insert. The major transcription start site is marked with an asterisk at Ϫ51. The 5Ј-flanking region is shown in lower case. The first 42 bp in exon 1 are shown in capital letters with the corresponding single-letter amino acid code underneath. Protein-binding sites identified by DNase I footprint analysis and gel mobility shift assays are shaded. Putative binding sites identified within the first 400 bp are as follows: AP-1, GATA factors, Sp1, AP-4, TFEB, XPF-1 (exocrine pancreas factor-1), HRE (hormone response element), Sp1, GAGA factor, H4TF-1 (Histone H4 transcription factor), GR (glucocorticoid receptor), and HNF-4. seen for the liver-enriched transcription factors HNF-4 at Ϫ67 to Ϫ56 and LF-A1 (Liver Factor-A1) (43) at Ϫ2178 to Ϫ2162. Putative binding sites were also seen for the following transcription factors at the indicated positions (using the numbering system of O'Hara et al. (6) where Ϫ1 represents the first nucleotide 5Ј to the translation start site): Sp1 at Ϫ286 to Ϫ276, Ϫ204 to Ϫ195, and Ϫ101 to Ϫ92; XPF-1 (exocrine pancreas factor-1) at Ϫ221 to Ϫ207; AP-1 at Ϫ391 to Ϫ 382; AP-2 at Ϫ1684 to Ϫ1675; AP-4 at Ϫ3487 to Ϫ3478 and Ϫ234 to Ϫ225; GR (glucocorticoid receptor) at Ϫ91 to Ϫ82; H4TF-1 (histone H4 transcription factor-1) at Ϫ100 to Ϫ92; PTF-1 (pancreatic acinar cell-specific factor) at Ϫ1403 to Ϫ1392; GATA transcription factors at Ϫ363 to Ϫ353, TFEB at Ϫ224 to Ϫ213; a hormone response element at Ϫ226 to Ϫ212; GAGA factor at Ϫ143 to Ϫ130, and ISGF-1 (interferon-stimulated gene factor-1) at Ϫ3552 to Ϫ3543. An Alu repeat is present at Ϫ956 to Ϫ634. Additionally, two nucleotide differences were noted between an individual containing a decanucleotide sequence insert (5Ј-CCTATATCCT-3Ј) at Ϫ323 and the sequence of an individual lacking the insert; these changes were at Ϫ122 (T to C) and Ϫ401 (G to T). The sequence data beyond Ϫ456 is based on the sequence present in the cosmid clone, which was constructed from an individual in whom the previously de-scribed decanucleotide insert was present.
Transcription Start Sites of Human Factor VII-The start sites of transcription of the TATA-less human FVII gene were determined by primer extension and by RNase protection assays. Primer extension was performed using avian myeloblastosis virus reverse transcriptase and a primer located at ϩ21 to ϩ42 in exon 1 to reverse transcribe total human liver RNA (Fig. 3A). A cluster of start sites was seen after a 2-week exposure of the autoradiogram; the most prominent bands mapped to Ϫ71, Ϫ69, and Ϫ68 (upper triple arrow). Bands at positions Ϫ51, Ϫ50, and Ϫ48 (lower triple arrow) and Ϫ32 (single arrow) were also visible. Less distinct bands were seen around Ϫ105, Ϫ117, Ϫ130 (within the bracket at right) with the reverse transcribed product generated by avian myeloblastosis virus reverse transcriptase. The results were confirmed using liver poly(A) ϩ RNA and Superscript II primed with the ϩ21 to ϩ42-bp oligonucleotide (data not shown). In addition to bands seen at Ϫ71, Ϫ69, Ϫ68, Ϫ51, and Ϫ50, this second primer extension also showed bands at Ϫ38, Ϫ45, Ϫ79, and Ϫ80.
RNase protection was performed using total human liver RNA hybridized to a riboprobe (Fig. 3B). Results were obtained using digestion with RNase T1, which cleaves 3Ј to guanine residues, and with RNase T1 ϩ RNase A, which together cleave 3Ј to guanine and pyrimidine residues. The product incubated with RNase T1 alone revealed a prominent band of 95 bp, which maps to a transcription start point at Ϫ53 (lane 4, upper arrow on right). This was two to three base pairs longer than the most prominent RNase T1 ϩ RNase A digested products, which yielded two adjacent bands corresponding approximately to positions Ϫ51 and Ϫ50 (lane 5, lower arrows on right). The positive control, a 249-bp ␤-actin riboprobe, was seen at the appropriate size (arrow on left) with the RNase T1-digested product (lane 2) but was again several base pairs longer than the RNase T1 ϩ RNase A product (lane 1). No bands were seen in the negative control lane (lane 6) using tRNA as a source for hybridization with the Factor VII probe.
Taken together, the primer extension and RNase protection data suggest a major start site at approximately Ϫ51. The primer extension data also suggest the presence of start sites between Ϫ80 and Ϫ30, although these were not confirmed by RNase protection.
Factor VII Promoter Activity in HepG2 and HeLa Cells-To determine the fragments of the Factor VII 5Ј-flanking region required for maximal promoter activity, segments were cloned into a reporter gene plasmid containing the human growth hormone structural gene and transfected into either HepG2 cells or HeLa cells. HepG2 cells were chosen to study the Factor VII promoter in a hepatocyte-like environment. HeLa cells, derived from a human cervical carcinoma, were chosen as a nonhepatocyte cell line.
Optimization experiments were first performed to determine the quantity of DNA and incubation times to use in transient CaPO 4 transfections. For HepG2 cells, 3 g of test plasmid DNA (1 pmol) in 12-h incubations were used; samples were taken 48 h later. Identical conditions were used for transfections of HeLa cells; however, transfections were performed with 15 g of test plasmid DNA (5 pmol). To correct for variations in transfection efficiency in both HepG2 and HeLa cell experiments, a luciferase internal control plasmid (2.5 g) was cotransfected along with the growth hormone reporter gene constructs. All results were normalized in relation to luciferase expression (Fig. 4A). Results in HepG2 cells showed that the first 75 bp alone were insufficient to increase reporter gene activity above base line. However, the addition of the adjacent 5Ј 34 bp in a 109-bp construct dramatically increased promoter activity to approximately 20-fold above base line. Promoter activity increased slightly upon including the next 53 bp (162-bp construct) and again with adding the next 24 bp (186-bp construct). The 186-bp construct, Ϫ185 to ϩ1, conferred maximal activity. A statistically significant decrease (p Ͻ 0.01) to 59% of the 186-bp construct was seen after including the adjacent 105 bp. Levels increased again at 501 bp and rose to a maximum at 1.746 kb. However, the increase above the promoter activity seen with the 186-bp construct was not significant. A decrease was seen between 1.7 and 3.9 kb.
HeLa cells were used to compare promoter activity to that seen in HepG2 cells in order to study differences between a cell with hepatocyte-like characteristics and a nonhepatocyte derived cell. The 109-bp construct had maximal promoter activity but showed levels only 3 times the base-line value. For the remaining constructs, including 162 bp or more, activity was at or below base line. These results indicate that the Factor VII promoter is more active in HepG2 cells than in HeLa cells. The promoter activity of two different polymorphic alleles was compared in HepG2 cells; one construct contained the FVII sequence from an individual carrying a decanucleotide insert at Ϫ323 (bases Ϫ324 to Ϫ333); the other did not contain this insert (23). The construct containing the decanucleotide insert had a promoter activity that was only 67% that of the more common allele lacking the insert when assayed in four independent transfection experiments with duplicate preparations of each of the two constructs. This difference reached a statistical significance of p Ͻ 0.005 using the Student's t test to compare the results (67% Ϯ 21% for the construct containing the decanucleotide insert and 100% Ϯ 23% for the construct lacking the insert (n ϭ 28)).

Identification of Protein Binding Sites in the Factor VII Promoter by DNase I Footprint Analysis and Gel Mobility Shift
Assays-DNase I footprint analysis of the proximal 309 bp of 5Ј-flanking DNA of the FVII gene showed three protein-protected sites (Fig. 5): region A (Ϫ51 to Ϫ32), region B (Ϫ108 to Ϫ84), and region C (Ϫ233 to Ϫ215). These regions were protected using both HepG2 and HeLa nuclear extracts as shown in lanes 4 and 5. The protected DNA sequence in region B includes a G-rich sequence (lower strand) at Ϫ101 to Ϫ94, which contains a previously described Sp1 binding site (CCCCTCCCCC) (44).
Gel mobility shift assays confirmed the presence of proteins, which bound specifically to regions A, B, and C. Fig. 6A shows the gel mobility shift assay using an oligonucleotide spanning the sequence between Ϫ51 and Ϫ29 bp. In HepG2 and HeLa nuclear extracts, two faint bands are seen (lanes 1 and 3) in addition to a faster migrating complex. A slower migrating band is also seen (top arrow) more prominently in HeLa than in HepG2 nuclear extracts. The ability of cold competitor to eliminate the binding of the radiolabeled oligonucleotide indicates the specificity of this sequence for the nuclear proteins. Fig. 6, B and C, shows gel mobility shift assays using HepG2 nuclear extracts (lanes 1-4), HeLa nuclear extracts (lanes 5-8), and human liver nuclear extracts (lanes 9 -12) with probes designed to study regions B and C as defined by DNase I footprint analysis (Fig. 5). Two slowly migrating complexes are seen using an oligonucleotide spanning the sequence Ϫ108 to Ϫ83 (Fig. 6B). Both complexes are seen more prominently using the HeLa nuclear extracts (lanes 5-8) than using either HepG2 (lanes 1-4) or liver nuclear extracts (lanes 9 -12), suggesting a higher abundance of proteins in these complexes in HeLa cells. Both also disappear in the presence of cold competitor, suggesting that binding is specific for this DNA sequence. A fast migrating complex is also present; this complex disappeared only in the presence of the highest concentrations of competitor (for HeLa and human liver nuclear extracts). In Fig.  6C, using a probe that extended from Ϫ236 to Ϫ201, two closely spaced complexes are seen that are reliably competed away by excess self competitor (lanes 2-4, 6 -8, and 10 -12). A third faster migrating band was variably present and not thought to represent specific binding to this sequence.
An Sp1 Binding Site Is Identified in the Ϫ108 to Ϫ83 bp Region-Data from gel mobility shift assays using an oligonucleotide from Ϫ108 to Ϫ83 suggest that Sp1 is the protein present in the slower migrating complex (Fig. 7A). An Sp1 consensus oligonucleotide is able to compete away only the slower migrating band (lane 5), whereas cold self competitor is able to compete away both of the closely spaced complexes (lane 2). An Sp1 antibody, which does not cross-react with known Sp1-like transcription factors, supershifts the slower migrating complex (lane 3). Additionally, purified Sp1 binds to the oligonucleotide and migrates to the same position as the slower migrating band (lane 4). All of these data strongly support the presence of an Sp1 binding site within the Ϫ108 to Ϫ83 bp sequence.
Two mutations were installed in the G-rich (lower strand) region between Ϫ103 and Ϫ92 base pairs at residues Ϫ100 and Ϫ94 by changing cytosine residues to adenosines. An oligonucleotide probe containing this Ϫ94/Ϫ100 mutation (MT94/100) fails to bind either the slower migrating protein or the protein in complex B (Fig. 7A, lane 6) and no longer binds to purified Sp1 protein (Fig. 7, lane 8). Independent mutations of the Ϫ94 and Ϫ100 bp sites revealed that the C to A change at Ϫ100 noticeably reduced binding on gel shift assay (Fig. 7B, lane 3) whereas mutation of the Ϫ94 bp site alone has a more profound effect, completely eliminating binding of both Sp1 and complex B on gel shift assays (Fig. 7B, lane 2). This result, the loss of two distinct complexes on gel shift assay with the mutation of a single nucleotide, suggests the existence of overlapping binding sites for Sp1 and another nuclear protein.
HNF-4 Antiserum Supershifts Binding of Liver Nuclear Protein to a Factor VII Proximal Promoter Sequence-A 30-bp oligonucleotide sequence from the Factor VII promoter (Ϫ76 to Ϫ47), which contains an ACTTTG element at Ϫ63 to Ϫ58, was analyzed using gel mobility shift assays. The oligonucleotide was shown to bind a nuclear protein present in HepG2 cells and human liver but not in HeLa cells (Fig. 8A, lanes 1, 4, and 7). The binding could be competed away by incubating with excess unlabeled self competitor oligonucleotide (Fig. 8A, lanes 2 and  3 and lanes 5 and 6). Polyclonal antiserum directed against the carboxyl terminus of HNF-4 supershifted the complex formation by liver nuclear extracts (Fig. 8B). Furthermore, mutation of the ACTTTG element to GACAAT eliminated binding to the liver nuclear extract (Fig. 8A, lane 9).
Mutagenesis of the Sp1 and HNF-4 Binding Sites Reveals Critical Elements in the Factor VII Promoter-Mutagenesis was done to analyze the functional consequences of mutations in the binding sites for Sp1 and HNF-4 (Fig. 4B). Mutations installed at Ϫ94 and Ϫ100 (MT94/100) in the Sp1 binding site and at Ϫ94 alone (MT94) reduced promoter activity to 2% of the wild-type 109 bp construct in HepG2 cells. In contrast, a mutation at Ϫ100 alone, which moderately affects binding on gel mobility shift assay, reduced promoter activity to 35% of wild-type.
To study the functional importance of this ACTTTG element, a 6-bp mutation (⌬ ACTTTG: ACTTTG 3 GACAAT) was installed in the 186-bp pOGH construct and used in transfection experiments with HepG2 cells and in transactivation studies with HeLa cells. This 6-bp mutation (⌬ ACTTTG) was previously shown to dramatically reduce activity in the Factor X promoter (7). In HepG2 cells, mutation of the putative HNF-4 binding site reduced promoter activity to 2% of the 186-bp wild-type (Fig. 4B). Transactivation experiments were performed to assess whether the presence of HNF-4 in trans could increase expression of the human growth hormone reporter gene above base-line levels. A 4-fold increase in promoter activity of the 186-bp wild-type construct was seen when cotransfected with 10 g of pLEN4S, an expression plasmid containing the HNF-4 cDNA. This was not seen with cotransfection of an equimolar amount of the pLEN expression plasmid, lacking the HNF-4 cDNA insert (wild-type ϩ pLEN4S: 392% Ϯ 33.9% versus wild-type ϩ pLEN: 100% Ϯ 9.6%). When this experiment was repeated using a 186-bp reporter gene construct and C, ؊233 to ؊215) of the human Factor VII promoter as determined by DNase I footprinting. Areas protected in the DNase I footprint analysis (Fig. 5) were analyzed for their ability to bind specifically to nuclear extracts in three different cell types: HepG2 cells, HeLa cells, and human liver cells. Increasing concentrations of unlabeled self competitors are used to differentiate specific from nonspecific binding. A, an oligonucleotide probe containing the sequence between Ϫ51 and Ϫ29 shows formation of four complexes (indicated by arrows) with nuclear extracts from HeLa cells (lanes 1 and 2) and HepG2 cells (lanes 3-6). Lane 2 contains 100-fold of cold self competitor. Lanes 4 -6 contains self competitors at 10-fold, 100-fold, and 500-fold excess, respectively. B, an oligonucleotide spanning the sequence from Ϫ108 to Ϫ83 was incubated with nuclear extracts from HepG2 (lanes 1-4), HeLa (lanes [5][6][7][8], and human liver (lanes 9 -12). Specific DNA-protein complexes are indicated by arrows. Indicated amounts of molar excess of cold self competitors were added. C, an oligonucleotide spanning the sequence from Ϫ236 to Ϫ201 was incubated with nuclear extracts from HepG2 cells (lanes 1-4), HeLa cells (lanes [5][6][7][8], and human liver (lanes 9 -12). Two specific complexes are indicated by thick arrows. A faster migrating band as indicated by a thin arrow is not consistently seen and is therefore considered nonspecific. Molar excess of cold self competitors were added as indicated. Nucleotide sequence of the probe is shown at the bottom. containing a mutation at the putative HNF-4 binding site (p186⌬ACTTTG), no transactivation occurred (p186⌬ACTTTG ϩ pLEN4S: 137% Ϯ 15.9% versus p186⌬ACTTTG ϩ pLEN: 136% Ϯ 20.9%). The levels of promoter activity were comparable with levels of the wild-type construct cotransfected with pLEN. These results suggest that the binding of HNF-4 to the ACTTTG element in Factor VII is critical for promoter activity. DISCUSSION A longstanding assumption in the field of coagulation has been that the vitamin K-dependent coagulation proteins are synthesized exclusively in the liver. Recent data indicate, however, that this assumption is not true, at least at the mRNA level, where several groups have documented the presence of transcripts in non-liver tissues of protein S (45), protein C (46), prothrombin (47), and Factor X. 4 For human coagulation Factor VII in contrast, within the sensitivity of the assay, Northern analysis using poly(A) ϩ RNA from a wide variety of tissues demonstrates that expression is confined to the liver (Fig. 1). Thus, at least for Factor VII, the older notion in fact appears true.
Factor VII, along with Factor X, is located on the long arm of chromosome 13 (13q34) (48) only 2.8 kb 5Ј of the Factor X gene (8). Densitometric analysis of autoradiograms demonstrated that steady state levels of Factor VII mRNA were only 6% those of Factor X mRNA levels, correlating well with the 10 -20-fold difference in plasma levels. Thus, the difference in plasma concentrations of the two factors is due to a difference in rates of transcription or in RNA stability (or some combination of the two).
A qualitative comparison of our Factor VII promoter data with those previously published for Factor X (7,8) shows similarities in that both promoters are TATA-less with the essential regulatory elements located within the first several hundred bp 5Ј to the translation start site. An HNF-4 binding site, including the same ACTTTG motif seen in the promoters of Factors IX (9) and X (7,8), is shown to be critical in the Factor VII promoter. Although a CAAT box, a conserved element present in approximately 30% of eukaryotic promoters (49), is critical in the promoters of both factor IX (9) and Factor X (7), no such element is apparent in the Factor VII promoter.
Identification of the transcription start site is an essential step in the characterization of a promoter. Like the promoters of many other clotting factor genes, the Factor VII promoter lacks a TATA box, a sequence present in approximately 80% of RNA polymerase II eukaryotic promoters (49), which is usually located about 30 bp upstream from the transcription start site. Many TATA-less promoters are thought to depend on the presence of the transcription factor Sp1 upstream from the start site (50,51). TATA-less promoters are also frequently characterized by the presence of an initiator element encompassing the transcription start site (52). In this report, a major transcription start site is identified at Ϫ51 by both primer extension and RNase protection. This start site occurs in the context TCA ϩ1 GTCCC, a site that is in 100% agreement with both a mathematically calculated consensus sequence CANYYY (49) based on the cap signal of 502 eukaryotic promoters, and an experimentally defined consensus PyPyA ϩ1 NT/APyPy determined using synthetically prepared promoter sequences (53). An Sp1 binding site, which we show to be crucial in promoter activity of Factor VII, is located approximately 41 base pairs upstream of the identified start site. This position corresponds well with the observed optimal distance between a GC-rich Sp1 binding element and the transcription start site (52,54). It should also be noted that, as is the case for most other TATAless promoters, multiple other start sites are identified in the region surrounding the major transcription start site.
Functional characterization of the human Factor VII promoter was performed in both HeLa cells, a non-hepatocytederived cell line, and HepG2 cells, a hepatoma-derived cell line previously shown to secrete Factor VII (55). Analysis of promoter activity using varying lengths of 5Ј-flanking region adjacent to the translation start site revealed a dramatic increase in promoter activity from base line to maximal activity upon the addition of a G-rich (lower strand) region, which includes an Sp1 binding site. The importance of this binding site was demonstrated by a dramatically reduced promoter activity (2% of wild-type) in a construct containing a single base change (C to A), which eliminated binding of Sp1 by gel mobility shift assay. However, the possibility cannot be excluded that a second protein, with an overlapping binding site to that of Sp1, plays an equally important role.
In reporter gene assays, a reproducible and statistically significant decrease of 41% was seen between constructs containing 186 and 291 base pairs of Factor VII promoter sequence. This region includes a site at Ϫ233 to Ϫ215, which was protected on DNase I footprint analysis and bound nuclear proteins on gel mobility shift assays. Interestingly, part of the protected sequence (5Ј-PuGGTCANNTGACCPy-3Ј) bears great resemblance to the sequence of several hormone response elements (56). Experimental studies testing the influence of spacing, orientation, and the exact sequence of the motif and flanking sequence have documented the importance of all these factors in determining receptor binding to a particular combination of half-sites (26). The sequence, which contains two PuGGTCA half-sites in an inverted orientation separated by 2 bp, does not exactly match any documented hormone responsive elements. 2 It should be noted that putative binding sites for several other factors were also seen in this region.
Experiments revealed that an intact binding site for HNF-4, a known member of the steroid hormone receptor superfamily (35), is critical for Factor VII promoter activity in vitro. First, a protein present in both human liver and HepG2 nuclear extracts but not in HeLa cells could be supershifted with the addition of HNF-4 antiserum. Mutation of the sequence ACTTTG (Ϫ63 to Ϫ58), which is well documented to bind HNF-4 in the Factor IX promoter (9), eliminated binding in gel mobility shift assays. Furthermore, an HNF-4 expression plasmid was able to increase promoter activity 4-fold when cotransfected with a 186-bp Factor VII promoter construct in HeLa cells. Mutation of the ACTTTG not only dramatically reduced promoter activity in HepG2 cells (2% of wild-type) but also resulted in the loss of ability of the HNF-4 expression plasmid to transactivate the Factor VII construct in HeLa cells. These experiments demonstrate that the HNF-4 binding site is critically required for Factor VII promoter activity. Given these findings, it is perhaps surprising that protection of the HNF-4 binding site is not seen on footprint analysis, but in this case this appears to result from the resistance of this region of DNA to digestion by DNase I, since no ladder is seen in the "no FIG. 8. A, gel mobility shift assays reveal binding of a protein from liver and HepG2 nuclear extracts but not HeLa nuclear extracts to an oligonucleotide sequence containing the ACTTTG motif (bases Ϫ63 to Ϫ58) in the proximal Factor VII promoter. A probe spanning the sequence from Ϫ76 to Ϫ47 forms a complex (arrow at left) with a protein present in HepG2 nuclear extracts (lanes 1-3) and human liver nuclear extracts (lanes 4 -6). Addition of excess unlabeled self competitor results in a gradual disappearance of the complex at 10ϫ (lanes 2 and 5) and 100ϫ (lanes 3 and 6) concentrations. HeLa nuclear extracts do not show binding of a protein in this region (lanes 7 and 8). Mutation of the ACTTTG to GACAAT between Ϫ63 and Ϫ58, within the Ϫ76 to Ϫ47 sequence, eliminates binding of human liver nuclear extracts to the mutated radiolabeled probe (lane 9). B, anti-HNF-4 antibodies supershift the binding of human liver nuclear extracts to a probe containing the sequence between Ϫ76 to Ϫ47 bp. Lane 1 shows the formation of a complex (left arrow) by the Ϫ76 to Ϫ47 radiolabeled probe with human liver nuclear extract. Lane 2 shows the binding of the same components with the addition of 1 l of antiserum raised against the carboxyl terminus of HNF-4. The position of a slower moving immune complex is indicated (right arrow).
A recent report documented a reduced level of Factor VII antigen in individuals heterozygous for a decanucleotide insert at position Ϫ323 in the Factor VII promoter (81% antigen level in individuals heterozygous for the insert, 112% antigen level in individuals without the insert, p Ͻ 0.005) (24). We investigated whether the presence of this decanucleotide in reporter gene constructs could influence promoter activity in HepG2 cells. Our findings showed that a 33% reduction in activity was associated with the naturally occurring allelic sequence containing the insert. This provides additional evidence that the presence of a decanucleotide insert may be directly related to the lower Factor VII levels seen in the population tested.
A schematic model of the promoter region of Factor VII based on data presented in this report is shown in Fig. 9. Four cis elements were shown to bind nuclear proteins. These elements are located at Ϫ51 to Ϫ32, Ϫ63 to Ϫ58, Ϫ108 to Ϫ84, and Ϫ233 to Ϫ215. Binding sites for the liver-enriched transcription factor, HNF-4, and the ubiquitous factor, Sp1 were identified at Ϫ63 3 Ϫ58, and Ϫ101 to Ϫ94, respectively. A major transcription start site was identified at approximately Ϫ51, located between two protein binding regions. This start site was favorably located approximately 40 bp downstream of a functionally important Sp1 binding site. A sequence with homology to a hormone responsive element was seen at Ϫ227 to Ϫ213 within a region protected by DNase I footprint analysis. Additionally, transfection studies of the FVII promoter comparing polymorphic alleles, one of which included a decanucleotide insert at Ϫ323, suggested that this region may also influence promoter activity.