Characterization of the Human Blood Coagulation Factor X Promoter*

Blood coagulation Factor X is a serine protease re- quired for both the intrinsic and extrinsic pathways of coagulation. The gene for Factor X spans 27 kilobases and is located on chromosome 13, in close proximity to the gene encoding Factor VII. Expression of Factor X is restricted to the liver. We have characterized the human Factor X promoter by mapping the start sites of transcription and carrying out a functional analysis of the promoter. The first 279 base pairs (bp) of 5’- flanking sequence upstream from the first AUG are sufficient to confer maximal promoter activity in HepG2 cells. Protein-binding sites within the 279-bp fragment are defined using gel mobility shift assays. Mutagenesis of two specific sequences within the 279- bp fragment (CCAAT at -120 to -116, and ACTTTG at -56 to -61), results in loss of ability to bind proteins from a HepG2 nuclear extract, and profound reduction in promoter activity of the 279-bp fragment. We con- clude that these two protein-binding sites are critical for the activity of the Factor X promoter. The

human Factor X promoter, including the sequence of 745 bp' of flanking DNA, the start site of transcription, activity of the promoter using a reporter gene assay, and delineation of specific protein-binding domains within the promoter. One of these protein-binding regions, conserved in the promoters of Factors IX and VII, may be involved in coordinate regulation of the genes encoding the vitamin K-dependent clotting factors. Mutagenesis of the protein-binding domains described here is associated with loss of the protein-binding property and loss of promoter activity as well.

Anchored PCR
Anchored PCR was carried out using the procedure of Loh et al. (13)  Hill) as starting material, first strand cDNA was synthesized using a Factor X-specific primer (5'-CCAGAATTCATTCGTCTTGTCGCT GTCCTC-3', located 244 bp downstream from the 1st AUG) and 200 units of Moloney murine leukemia virus H-reverse transcriptase (Bethesda Research Laboratories), which lacks RNase H activity. 20 units of RNasin was included in this reaction. Excess primer and unincorporated deoxynucleosides were removed by two passages through Sephadex G-50, and first strand cDNA was then tailed with poly(G) using terminal deoxynucleotidyl transferase (Boehringer Mannheim). The tailed FX cDNA was amplified using a second Factor X-specific primer (5'-TGGCCTGCTCCCTGCGGATGAAC AGACTTT-3', located 95 bp downstream from the 1st AUG) and a GGATCCCCCCCCCCCCCC-3') and AN (5"GCACAAGCTTGAA mixture of nonspecific primers ANC (5"GCACAAGCTTGAATTC T T C G G A T C C -3 ' ) in a ratio of 1:lO for the poly(G) end of the cDNA. After amplification, the PCR product was gel-purified and subcloned into an M13 vector by blunt-end ligation for sequence analysis (14).
SI Nuclease Analysis S1 nuclease analysis was carried out using the procedure of Sharp et al. (15). A 357-base single-stranded DNA probe, beginning 279 bases upstream from the first AUG and continuing through exon 1, was 5' end-labeled and hybridized to 12.5 pg of total human liver RNA. After treatment with 50 units of S1 nuclease (Bethesda Research Laboratories) at 30 "C, the DNA-RNA heteroduplexes were electrophoresed on a 6% denaturing acrylamide gel. Nucleotide sequence was run on the same gel as a size marker. Hybridization of the probe to yeast tRNA (Boehringer Mannheim) served as a negative control.

Isolation and Characterization of 5"Flanking Region of Human Factor X
Previously published data on the 5'-flanking region of human Factor X included only 281 bp upstream from the first AUG (11).
This sequence was amplified from human genomic DNA using primers based on the published sequence. To obtain additional 5'flanking sequences, 2.8 X lo5 cosmid clones from a human genomic The abbreviations used are: bp, base pair(s); kb, kilobase pair(s); PCR, polymerase chain reaction; hGH, human growth hormone; SEAP, secreted alkaline phosphatase plasmid; HEPES, 442-hydrox-yethy1)-1-piperazineethanesulfonic acid.

15440
This is an Open Access article under the CC BY license. placenta library (Stratagene, La Jolla, CA) were screened using the available 5"flanking sequence (-279 bp + +12) as a probe. One positive cosmid clone was obtained. Based on restriction analysis and PCR amplification of downstream exons within the clone (exon 1 was present, exon 2 was not), the clone appears to contain at least 10 kb of 5"flanking sequence. A 3.5-kb BamHI fragment, spanning the region from 2.8 kb upstream from the first AUG to the first intron, was subcloned into pBR322, sequenced, and used as starting material for preparing transfection constructs. The sequencing strategy consisted of synthesizing a sequencing primer based on the known 5'flanking sequence, and preparing subsequent primers based on data from each run to "walk" further upstream.

Tissue Culture
HepG2 cells (16) (originally from Wistar Institute) and HeLa cells (17) were obtained from the Tissue Culture Facility in the Lineberger Cancer Research Center at the University of North Carolina-Chapel Hill. The following materials were also obtained through the Tissue Culture Facility in the Lineberger Cancer Research Center at UNC-CH: fetal bovine serum (HyClone Laboratories, Inc.), Eagle's minimal essential medium (containing Eagle's salt and L-Glutamine), phosphate-buffered saline (pH 7.5), 0.4% trypan blue stain (Bethesda Research Laboratories), 0.05% trypsin, 0.53 mM EDTA.4Na (Bethesda Research Laboratories), sterile H20 (endotoxin-free), dimethyl sulfoxide (Sigma), and CaCI2.2Hz0 (cell culture grade, Sigma). HepG2 cells were grown in Eagle's minimal essential medium supplemented with 10% fetal calf serum, 0.01% penicillin, and 0.01% streptomycin. HeLa cells were grown in Dulbecco's minimal essential medium-H with 10% fetal calf serum, and 0.025% kanamycin.

Reporter Gene Assays
Preparation of Constructs inpOGH Vector-The pOGH vector (18) employs secretion of hGH into the medium as an assay for eukaryotic promoter strength. The vector contains hGH structural sequences cloned into the BamHI-EcoRI sites of pUC12. For the assays for Factor X promoter activity, varying lengths of Factor X 5"flanking sequence were cloned into the HindIII-Xbal site of pOGH by directional cloning (see Fig. 1). The flanking sequence fragments were generated by PCR and deleted at the 5' end using convenient restriction sites. The 3' amplifying oligonucleotide contains an engineered XbaI site within the translated sequence, used for subcloning into pOGH (see Fig. 2). All inserts contain the entire 5'-untranslated sequence. The PCR-generated fragments were sequenced to verify that no errors had been introduced in the amplification step.
hGH Assay-Levels of hGH in the medium were measured with a solid-phase two-site radioimmunoassay kit under the conditions recommended by the manufacturer (Nichols Insitute Diagnostics, CA). The lower limit of detection of the assay is 0.5 ng of hGH/ml; the assay is linear in the range of 0.5-50 ng/ml. Transfection Conditions-Optimal transfection conditions were determined for each cell line using a secreted alkaline phosphatase plasmid (SEAP) (19). For assays of promoter activity, cells were cotransfected with pOGH constructs and the SEAP plasmid; levels of SEAP activity were used to normalize transfection efficiency. For HepG2 cells, transfection efficiency varied over a %fold range and for HeLa cells over as much as 5-fold. For HepG2 cells, lo6 cells in 60-mm tissue culture dishes were transfected with 2 pg of SEAP and 2 pg of pOGH FX5' by the Capo4 precipitation method (20). After 5 h of exposure to DNA precipitates, the cells were washed with phosphate-buffered saline and fresh medium was added. Forty-eight hours after transfection, the medium was harvested to perform SEAP and hGH assays. For HeLa cells, optimal transfection conditions are slightly different. 5 X lo5 cells in 60-mm tissue culture dishes were transfected with 6 pg of pOGH FX5' and 2 pg of SEAP using the CaPO, method. Sixteen hours later the cells were washed with phosphate-buffered saline and fresh medium supplemented with 5 mM sodium butyrate was added. Seventy-two hours after transfection, the medium is assayed for hGH and SEAP.

Preparation of Nuclear Extracts
Approximately 1 X lo9 HepG2 cells were used to prepare the nuclear extract. The cells were grown to 50% confluence and procedures were performed as previously described (21-23). All steps were carried out at 0-4 "C. Following gentle extraction of nuclear proteins in the presence of protease inhibitors, extracts are dialyzed, divided into 100-pl aliquots, frozen in liquid N,, and stored at -70 "C. Varying lengths of Factor X promoter were subcloned into the HindIII-XbaI site of pOGH by directional cloning. The Hind111 site in the vector was blunt-ended with Klenow fragment. The Factor X promoter fragments were generated using PCR the 3'-oligonucleotide contained an XbaI site located within exon 1. The creation of the XbaI site does not alter any sequences within the Factor X promoter. The 5' ends were generated by restriction enzymes and filled in with Klenow fragment if necessary. The constructs were sequenced through the regions of the inserts to insure that no errors had been introduced through cloning or PCR.

In Vitro Synthesis of C/EBP Protein
The cDNA of rat C/EBP (24) was the gift of Professor S. McKnight (Carnegie Institution of Washington). The cDNA was released from pMSV and subcloned into pSP6/T3 (BRL) in the same orientation as the T3 promoter. Following transcription with T3 RNA polymerase, in vitro translation was performed using a rabbit reticulocyte lysate system as described by the manufacturer (Promega). Purified C/EBP fragment containing the 88 COOH-terminal amino acid residues was the gift of Dr. Z. Cao

Gel Mobility Shift Assays
The basic procedure is that of Chodosh et al. (25). Oligonucleotide probes were prepared on an oligonucleotide synthesizer (Millipore), annealed, and 5' end-labeled using T4 polynucleotide kinase. Probes included the following: Factor X promoter sequence from -133 to -103 (CCAAT); same probe except CCAAT sequence changed to AGCTA (mCCAAT); promoter sequence from -68 to -39 (ACTTTG); same probe except that ACTTTG has been changed to GACAAT (mACTTTG); and a probe spanning the same region of the promoter and containing a single nucleotide change from wildtype, T-A at -52 (MT-A). These were used for both the binding and competition assays. Binding conditions were optimized by varying the ionic strength of the binding buffer and the amount of nonspecific DNA or poly(dI.dC) added. For binding reactions with HepG2 nuclear extract, the DNA probe was incubated with 10 pg of the extract in 24 pl of 10 mM HEPES (pH 7.61, 40 mM KC], 3 mM MgCl,, 4% Ficoll, 0.5 mM dithiothreitol, and 2 pg of dI.dC for 15 min at room temperature. For binding reactions with purified protein (either native protein or translation product), the DNA probe was incubated with either 5 ng of purified C/EBP fragment or 1/25 (2 pl) of the translation product from 1 pg of C/EBP transcript in 20 pl of 10 mM HEPES (pH 7.5), 50 mM KCl, 1 mM EDTA, 1 mM dithiothreitol, 12% glycerol, 1 mg/ml bovine serum albumin, and 0.5-1.0 pg of dI. dC at room temperature for 15 min. Varying amounts of dI.dC (0.1-2.0 pg) or salmon sperm DNA (0-2.0 pg) were tested to determine optimal binding conditions. In all cases, the whole reaction was loaded onto a 5% nondenaturing acrylamide gel (bis:acrylamide = 1:30) containing 0.5 X TBE and 5% glycerol and run in 0.4 X TBE at 300 volts at 4 "C. When retarded bands were observed, 15-150-fold concentration of unlabeled oligonucleotide probes were included in the reaction for the competition. A double stranded DNA fragment containing the C/EBP-binding site, spanning the sequence from -178 to +33 of the factor IX promoter, was included as a positive control for C/EBP binding (10).

Mutagenesis of Factor X Promoter at Potential Protein-binding Sites
Mutagenic oligonucleotides were synthesized in an oligonucleotide synthesizer by including the three non-wild-type precursors at each step of the synthesis, so that the sequence of the protein-binding region is replaced by randomly incorporated non-wild-type nucleotides. The oligonucleotide is designed such that both the 3' and the 5' ends contain 12 nucleotides of wild-type DNA sequence. Uracilcontaining DNA templates were prepared by the method of Kunkel (26) and annealed to the primers described above. In vitro primer extension was carried out using 10 units of T4 DNA polymerase (New England Biolabs) and 800 units (2 pl) of T4 DNA ligase (New England Biolabs) and incubating at room temperature for 4 h. 1/50 to 1/20 of the reactions (10-25 ng of DNA) were used to transform Escherichia coli wild-type strain JM 103. Mutations were identified by sequence analysis.
The mutated M13 templates were subjected to PCR, and the resultant DNA fragments were subcloned into pOGH plasmid vectors. Each individual clone was sequenced to insure that no mistakes were introduced into the promoter fragments during PCR and to verify the DNA sequences in the mutated regions.

Start Site of Transcription-The start site of transcription
for Factor X has not previously been determined. We have used both anchored PCR and S1 nuclease analysis to map the start sites of transcription. Briefly, anchored PCR involves reverse transcription of RNA, priming with a Factor Xspecific oligonucleotide. The reverse transcripts are tailed with poly(G) using terminal deoxynucleotidyl transferase, and the tailed reverse transcripts are enzymatically amplified using another Factor X-specific primer and poly(C). The resulting PCR products are subcloned into M13 and sequenced to disclose transcription start sites. Human liver poly(A+) RNA was used as starting material. The total number of Factor X clones sequenced was 21. The 21 clones mapped to 6 different start sites, indicated by arrows on Fig. 2. Most of the start sites clustered within a 20-bp region, at sites ranging from 13 to 33 bp upstream from the first AUG. Three of the 21 clones mapped to a longer transcript, with a start site 57 bp upstream from the first AUG. S1 nuclease analysis confirmed these results (data not shown), and suggested one additional transcription start site at -130 bp upstream from the first AUG. Based on band intensity, this appeared to be a minor start site and was not seen in the anchored PCR clones. The finding of multiple start sites of transcription, by both anchored PCR and S1 nuclease analysis, is consistent with results previously reported for other TATA-less promoters ( Ref. 27 and references therein). This is the first report of the start site of transcription of the Factor X gene.
Since anchored PCR has not been used extensively to map start sites of transcription, we used the technique to map the start site of human Factor IX, as a check on the accuracy of the method. Using total RNA as starting material, the experiment was performed as described above. Five clones were sequenced; three of these mapped to -26 and two others to -10 upstream from the first AUG. The former result (-26 start site) is in close agreement with the start site previously determined by S1 nuclease analysis by Anson et al. (28). Characterization of Additional 5'-Flanking Sequence-Results are shown in Fig. 2. At 5 residues in the 3' 281 bp, this sequence differed from that previously published (11). The remaining sequence data have not been previously published. All data presented here are the results of at least two independent sequencing runs. The sequence data reported here have been entered into the GenBank of Los Alamos National Laboratory.
Factor X Promoter Activity in HepG2 Cells-Following identification of the transcription start sites in Factor X, promoter activity was examined using a reporter gene assay in a transient expression system. Varying lengths of 5'-flanking DNA from Factor X were cloned into the human growth hormone-containing plasmid pOGH (18) (Fig. 2) and introduced into HepG2 cells by calcium phosphate transfection. Deletion constructs were made using appropriate restriction sites or PCR fragments; pOGH, the growth hormone plasmid without any 5"flanking sequence, was used as a negative control. Optimal transfection conditions for HepG2 cells were first determined, and involved addition of 4 pg of DNA X 5 h without glycerol shock. Transfection efficiency was monitored using pBC12/CMV/SEAP (19) Table I.) The highest levels of promoter activity were present in the fragment containing 279 bp of DNA upstream from the first AUG. The addition to the construct of the next 109 bp upstream has little effect on promoter activity. Inclusion of additional 5"flanking sequences gradually diminishes activity, so that the 2.8-kb construct has only 55% of the maximal promoter activity observed with the 279-bp construct. For constructs shorter than the 279-bp fragment, the 209-bp fragment again has close to maximal activity. Further deletions result in substantial loss of activity, so that the 121-bp fragment has only 23% of maximal activity, and deletion of the next 13 bp (the 108-bp fragment) containing the CCAAT sequence (see below) results in a 4-fold further reduction in activity. Thus, the major elements required for Factor X promoter activity FIG. 3. Promoter activity of deletion constructs. A, in HepG2 cells promoter activity is measured by quantitating growth hormone activity in' the medium. Results reported are the average of 4-10 transfection experiments. The activity of the maximally active construct (279-bp fragment) is arbitrarily set at 10076, and other results are reported as a percent of the maximum. B, in HeLa cells maximal activity is seen with the 209-bp fragment, and other results are expressed as a percentage. All results are the average of at least two transfection experiments. are contained in the 279 bp immediately upstream from the first AUG.

Promoter Activity Is Diminished in a Nonhepatocyte
Line-In order to determine whether the promoter element under study is adequate to confer tissue specificity, the Factor X promoter-containing hGH constructs were transfected into HeLa (17) cells, a malignant cell line which does not normally express Factor X. Direct comparison of results with HepG2 is difficult, since optimal transfection conditions differ in the two cell lines. However, maximal activity using optimal transfection conditions in HeLa cells was only 20% of the maximum observed with HepG2 cells. Comparison of relative activities in the two cell lines is also informative (see Fig. 3). I n HepG2 cells, the fragment between 209 and 388 bp upstream from the first AUG confers maximal promoter activity; in HeLa cells, however, maximal promoter activity is seen with the 209-bp fragment, and inclusion of the next 179 bp of the promoter decreases activity to 28% of maximum. This reduced expression suggests the presence of a HeLa-specific negative acting element in this region, and is in marked contrast to the results with the same fragment in HepG2 cells.

Identification of Protein-binding Sites within the Factor X
Promoter-Within the X promoter, two regions were selected for further study, one characterized by the presence of the putative protein-binding site CCAAT (29) and the other by the presence of the sequence ACTTTG, which is conserved in the promoters of Factors VII, IX, and X (11).
As a first step, gel mobility shift assays using normal and mutant oligonucleotides from these regions were carried out. Results with the CCAAT-containing probe are shown in Fig.  4a. The location within the promoter of the 30-base oligonucleotide probe (designated CCAAT) is shown on Fig. 2. Lane 1 contains oligonucleotide probe alone; in lane 2, the probe has been incubated with 10 pg of HepG2 nuclear extract.
Three slower moving bands, representing protein-DNA complexes, are seen. Addition of a 15-fold excess of unlabeled competitor (unlabeled oligonucleotide, lane 3) markedly reduces the intensity of these bands and a 60-fold excess (lane 4) eliminates them, suggesting that the binding is specific for the fragment used.
In lanes 5-7, the same 30-mer has been used as probe, but a mutated 30-mer (mCCAAT), in which the sequence CCAAT has been changed to AGCTA by random mutagenesis, is used as unlabeled competitor. Note that even at 150-fold excess, the mutated sequence does not compete well with the wildtype sequence for binding of proteins in the HepG2 nuclear extract. There is perhaps a very modest decrease in the intensity of the slowest moving band at 150-fold excess. When the mutated sequence is used as probe (lanes 8-10), the slowest moving band is faintly detected (lane 9), but the other bands are not detected at all. This band is no longer detected after the addition of a 150-fold excess of the mutant probe. Thus the slowest moving band may result from binding at a site other than CCAAT within the 30-base sequence used as probe.
Similar studies with the ACTTTG-containing probe (see Fig. 2 for location within the promoter) are shown in Fig. 4b. Lane 1 contains probe alone and lane 2 probe and HepG2 nuclear extract. Two bands are apparent in close proximity, with the slower moving band being more intense. Addition of 15-fold excess of unlabeled competitor abolishes the faster moving band and greatly reduces the intensity of the slower moving band; addition of a 60-fold excess of competitor abolishes this band as well. When mutant 30-mers, in which the sequence ACTTTG has been altered by random mutagenesis to GACAAT, are used as unlabeled competitors (lanes 5-7, designated mACTTTG) no reduction in either band is seen, even at 150-fold excess of competitor. When the mutant oligonucleotides are used as probe (lanes 8-10), no retarded bands are seen, confirming that alteration of the ACTTTG site results in loss of the characteristic protein-binding property of the DNA sequence. Fig. 4c shows that even a single base change within the ACTTTG sequence can result in the loss of its proteinbinding property. In this experiment, binding with the wildtype sequence is compared to binding in the presence of a 30mer which is identical except that ACTTTG has been changed to ACTTAG. Again, incubation of HepG2 nuclear extract with labeled wild-type probe results in the generation of two bands, which are abolished by excess unlabeled competitor.
When the mutant sequence (designated MT-~) is used as 1- competitor (lanes 5 and 6), the diminution in intensity of the two bands is much less, even with a 60-fold excess of the mutant oligonucleotide. Thus, the single-base  FIG. 4. Gel mobility shift assays using normal and mutant sequences as probes. a, the element CCAAT in the Factor X promoter binds HepG2 nuclear proteins. Protein-DNA complexes are analyzed on a 4% nondenaturing acrylamide gel. The end-labeled wild-type oligonucleotide 5'-GCAGGTCTCGGCTCCAATCAGGA-GGCCTAA-3' is used as probe in lanes 1-7 (designated CCAAT). The oligonucleotide in mCCAAT is identical except that the sequence AGCTA has been substituted for CCAAT by random mutagenesis. Lane 1 contains labeled oligonucleotide alone; in lanes 2-7, the oligonucleotide is incubated with 10 pg of HepG2 nuclear extract. Unlabeled competitor is added in lanes 3-7 as indicated. Binding to the wild-type probe is competed specifically by unlabeled wild-type oligonucleotide (lanes 3 and 4 ) , whereas binding to the probe is unaffected by the addition of excess mutant oligonucleotide (lanes 5-7), except for a modest decrease in the intensity of the slowest moving band. Lane 8 contains mutant oligonucleotide alone, and lane 9, mutant oligonucleotide and 10 pg of HepG2 nuclear extract. Only the slowest moving band is present, suggesting that the protein-binding site within the oligonucleotide is a t a site other than the CCAAT sequence. Addition of unlabeled mCCAAT (lane 10) eliminates the band. b, the element ACTTTG in the Factor X promoter binds hepatic nuclear proteins. The labeled oligonucleotide 5"GCTGGGGC-GTGGACTTTGCTCCACAGCCT-3' (designated ACTTTG) is used change results in a marked reduction in affinity for the bound protein(s). In lanes 7-10, the mutant oligonucleotide MT+A is used as labeled probe. A faint band corresponding to the slower moving band is seen with the HepG2 nuclear extract (lane 8), but the intensity is markedly diminished compared to binding to the wild-type sequence (compare lane 2). The faint band is abolished by addition of unlabeled (mutant) competitor, so the binding, although greatly reduced, is specific for the fragment.
To determine whether these putative protein-binding site(s) within the fragments tested have any functional role in the promoter, the sites were altered by mutagenesis and tested using the reporter gene assays previously described. In all of these assays, the wild-type and mutated fragments were 279 bp in length, and differed from each other only at the sites designated. Mutations were installed in the 279-bp wildtype fragment using site-directed mutagenesis and then subcloned back into pOGH. Results, which represent the average of 2-6 transfection experiments, are shown in Fig. 5. (Standard deviations for these experiments are shown in Table 11.) Note that substitution of AGCTA for CCAAT within the 279bp promoter reduces promoter activity to 11.8% of wild-type, and similarly that the change from ACTTTG to GACAAT reduces promoter activity to 17.2% of wild-type. Even the single-base change from ACTTTG to ACTTAG reduces promoter activity to 19.2% of wild-type. In contrast, when the potential Spl-binding site (30) at -66 to -61 was changed from GGGGCG to TCTATC, no change in promoter activity was seen. Thus, the two elements, CCAAT and ACTTTG, in addition to binding proteins within a HepGZ nuclear extract, appear to be critical for Factor X promoter activity, since mutations in these sequences result in a 5-10-fold drop in promoter activity.
Transcription Factor C/EBP Does Not Bind to the CCAAT Sequence within the Factor X Promoter-The transcription factor C/EBP, present in liver and adipose tissue (4, 31), binds at a number of different DNA elements, some characterized by the presence of the sequence CCAAT. Crossley and Brownlee (lo), using DNase footprinting studies, have shown that C/EBP binds to the Factor IX promoter. We tested the ability of C/EBP to bind to the CCAAT-containing fragment of the Factor X promoter in a gel mobility shift assay. The labeled probe was the 30-mer designated CCAAT protein in the binding assay consisted of either C/EBP translation product (1/25 of translation product from 1 pg of C/EBP transcript), or 5 ng of purified C/EBP fragment (88 COOHterminal amino acids capable of binding DNA). Under several different sets of binding conditions, neither the translation product nor the purified protein bound to the CCAAT oligonucleotide, whereas both bound to a positive control sequence (residues -178 to +33 from the Factor IX promoter), confirmas probe in lanes 1-7. mACTTTG is the same sequence except that ACTTTG has been changed to GACAAT by random mutagenesis. Again, lane 1 contains labeled probe alone, and lanes 2-7, probe and 10 pg of HepG2 nuclear extract. In lanes 3-7 unlabeled competitor is added as indicated, and lanes 8-10 contain mACTTTG as labeled probe. Two prominent bands are generated by addition of HepG2 nuclear extract (lane 2) which are competed specifically by unlabeled wild-type oligonucleotide (lanes 3 and 4 ) but not by the mutant oligonucleotide (lanes 5-7). The mutant oligonucleotide is unable to bind protein from the nuclear extract (lane 9). c, gel mobility shift base change

5'-GCTGGGGCGTGGACTTAGCTCCACAGCCT-3'
assay. Oligonucleotide ACTTTG is as above. MT-A contains a single within the highly conserved ACTTTG sequence. Note that MT-A competes poorly with the wild-type oligonucleotide (ACTTTG) for binding nuclear proteins (compare lane 2 with lanes 5 and 6), and indeed appears to have very low affinity for nuclear proteins (lane 8 ) , compared to the wild-type sequence (lane 2). ing by gel shift Crossley and Brownlee's earlier DNase footprinting result. Thus C/EBP does not appear to be the transcription factor binding to the CCAAT-containing element within the Factor X promoter.

DISCUSSION
Start Site of Transcription-A necessary prerequisite for the study of the Factor X promoter is definition of the start site of transcription. We used anchored PCR to determine the start site, and confirmed these data using S1 nuclease analysis. In both cases the starting material was RNA from normal human liver, so that potential problems related to the use of RNA from transformed cell lines (e.g. aberrant start sites) are avoided.
Anchored PCR is a technique which allows one to amplify a specific segment of DNA or RNA when sequence information is available for only one end of the segment. The method has general applicability but offers special advantages for mapping start sites in genes belonging to superfamilies, since the start site is provided as sequence data, which allows the identification and exclusion of extraneous sequence. This is in contrast to the potential pitfalls of S1 nuclease mapping or primer extension, where there is always the concern that signals on the autoradiograph may have arisen from hybridization with a closely related transcript rather than the transcript under study.
Anchored PCR has not been extensively used previously to map start sites of transcription (13). Thus we confirmed the results using S1 nuclease analysis. As an additional check on the method, we used anchored PCR to determine the start site of transcription for human Factor IX. Using total human liver RNA as starting material, we identified two start sites, one at 26 bp upstream and the other at 10 bp upstream from the first AUG. These data are in close agreement with those of Anson et al. (28), who identified a start site 29 bp upstream, using S1 nuclease analysis.
The results for the Factor X transcript show the presence of multiple start sites of transcription, most clustered in a segment 13 to 33 bp upstream from the first AUG. The finding of multiple start sites is consistent with results reported for other TATA-less promoters (27). The cluster of start sites that we report here predicts a relatively short 5"untranslated segment for human Factor X. This is similar to the findings for human Factor IX ( Ref. 28 and this report), where the 5'untranslated region appears to be short, <30 bp.
Functional Characterization of the Human Factor X Promoter-In HepG2 cells, maximal promoter activity is attained with a 279-bp fragment of 5'-flanking DNA (Fig. 3A). Inclusion of an additional 109 bp (388 bp fragment) results in virtually no change in promoter activity, while inclusion of additional 5"flanking elements (beyond 388 bp) results in a gradual decrease in activity. These results are similar to those previously reported for the Factor IX promoter (8) where maximal activity is attained using a 303-bp promoter element (numbering from the first AUG), and inclusion of an additional 5'-flanking sequence, up to 445 bp upstream from the first AUG, results in little change in promoter activity. In HeLa cells, which do not express Factor X, the pattern of expression is different, with maximal activity seen with the 209-bp fragment, and marked loss of promoter activity with inclusion of additional 5"flanking sequence. The 388-bp construct has only 28.4% of the activity seen with the 209-bp construct. Although direct comparisons between cell lines are difficult, since optimal transfection conditions are different, it should be noted that maximal GH activity for the HepG2 cells was &fold higher than that observed for HeLa cells. The relative promoter activities, and the differing patterns of promoter activity with addition of the fragment from -209 to -388 (sustained maximal levels of expression in HepG2 uersus reduced expression in HeLa cells) both point toward a possible tissue-specific regulatory element within this fragment.
Analysis of promoter activity in HepG2 cells reveals that deletion of the 88 bp between the 209-and 121-bp constructs results in a 4-fold decrease in promoter activity, and deletion of the next 13 bp has an additional 4-fold loss of activity. Thus critical promoter elements reside in the fragment from -209 to -108. Whether any of these elements are tissuespecific is unclear, but the difference in overall promoter activity in the two cell lines is consistent with the notion that these elements may also possess some degree of tissue specificity for hepatocytes.
We chose for further study two specific elements within the 5"flanking region: the sequence from -133 to -103, containing the element CCAAT (at -120 to -116), and the sequence ACTTTG located -56 to -51 upstream from the first ATG. The former was chosen because of the marked decrease in promoter activity resulting from deletion of this element (compare -121 to -108 in Fig. 3A) and the latter because of its conservation in the promoter regions of FVII, IX, and X. Based on gel mobility shift assays (Fig. 4), both elements bind proteins contained in a HepG2 nuclear extract. In both cases specificity of binding is demonstrated by disappearance of binding complexes following addition of unlabeled competitor. For the more 5'-element, alteration of the CCAAT sequence by random mutagenesis (to AGCTA) results in a sequence that cannot compete for binding to HepG2 nuclear proteins (lanes 5-7). When the altered sequence is used as a probe (lane 9) the two faster migrating complexes are entirely absent, indicating that they both require an intact CCAAT sequence for binding, whereas the slowest moving complex is still seen, suggesting that binding for this protein does not require the CCAAT sequence. Further evidence of the critical nature of the CCAAT sequence is seen in the promoter activity assays displayed in Fig. 5. Alteration of the sequence to AGCTA within the 279-bp fragment of 5"flanking results in a 10-fold drop in promoter activity. Thus a change in the CCAAT element abolishes protein binding and profoundly reduces promoter activity of the Factor X promoter. Since the element CCAAT can serve as a recognition sequence for the transcription factor C/EBP, which has been shown to bind to a site within the F.IX promoter (lo), we wished to determine whether C/EBP bound to the 30-mer spanning -133 to -103. Results using both the purified protein and the translation product were negative, suggesting that the protein(s) binding to this element within the X promoter are distinct from this previously characterized liver transcription factor.
Results for analysis of the ACTTTG element were similar to those described for the CCAAT element. An oligonucleotide probe spanning the sequence from -68 to -39 and containing ACTTTG binds proteins from a HepG2 nuclear extract; the complexes are abolished by the addition of unlabeled competitor. Mutant sequences in which ACTTTG has been changed to GACAAT cannot compete for binding (lanes 5-7) nor can they bind protein from the nuclear extract (lane 9). The substitution of even a single nucleotide within this sequence (ACTTTG + ACTTAG) results in the complete loss of the faster moving band (lane 8). In the reporter gene assay (Fig.  5) both of these mutations result in a %-fold loss of promoter activity compared to wild-type. Again loss of ability to bind a nuclear protein (or proteins) is associated with a dramatic decrease in promoter activity for the sequence under study.
The point mutation in the ACTTTG sequence is identical to a naturally occurring mutation described in the F.IX promoter (T-A at -20, within the conserved ACTTTG sequence) (32). This variant displays the Leyden phenotype, with absence of F.IX protein (antigen and activity) in affected children, but gradually increasing levels following puberty. The pathophysiology of the Leyden phenotype remains unexplained, but further characterization of the protein which binds at this site may shed light on this interesting problem.