Expression of the CYP4F3 Gene

Cytochrome P450 4F3 (CYP4F3) catalyzes the inactivation of leukotriene B4 by ω-oxidation in human neutrophils. To understand the regulation of CYP4F3 expression, we analyzed the CYP4F3 gene and cloned a novel isoform (CYP4F3B) that is expressed in fetal and adult liver, but not in neutrophils. The CYP4F3 gene contains 14 exons and 13 introns. The cDNAs for CYP4F3A (the neutrophil isoform) and CYP4F3B have identical coding regions, except that they contain exons 4 and 3, respectively. Both exons code for amino acids 66–114 but share only 27% identity. When expressed in COS-7 cells, the K m of CYP4F3B was determined to be 26-fold higher than the K m of CYP4F3A using leukotriene B4 as a substrate. 5′-Rapid amplification of cDNA end studies reveal that the CYP4F3A and CYP4F3B transcripts have 5′-termini derived from different parts of the gene and are initiated from distinct transcription start sites located 519 and 71 base pairs (bp), respectively, from the ATG initiation codon. A consensus TATA box is located 27 bp upstream of the CYP4F3B transcription start site, and a TATA box-like sequence is located 23 bp upstream of the CYP4F3A transcription start site. The data indicate that the tissue-specific expression of functionally distinct CYP4F3 isoforms is regulated by alternative promoter usage and mutually exclusive exon splicing.

Cytochrome P450 4F3 (CYP4F3) catalyzes the inactivation of leukotriene B 4 by -oxidation in human neutrophils. To understand the regulation of CYP4F3 expression, we analyzed the CYP4F3 gene and cloned a novel isoform (CYP4F3B) that is expressed in fetal and adult liver, but not in neutrophils. The CYP4F3 gene contains 14 exons and 13 introns. The cDNAs for CYP4F3A (the neutrophil isoform) and CYP4F3B have identical coding regions, except that they contain exons 4 and 3, respectively. Both exons code for amino acids 66 -114 but share only 27% identity. When expressed in COS-7 cells, the K m of CYP4F3B was determined to be 26-fold higher than the K m of CYP4F3A using leukotriene B 4 as a substrate. 5-Rapid amplification of cDNA end studies reveal that the CYP4F3A and CYP4F3B transcripts have 5-termini derived from different parts of the gene and are initiated from distinct transcription start sites located 519 and 71 base pairs (bp), respectively, from the ATG initiation codon. A consensus TATA box is located 27 bp upstream of the CYP4F3B transcription start site, and a TATA box-like sequence is located 23 bp upstream of the CYP4F3A transcription start site. The data indicate that the tissue-specific expression of functionally distinct CYP4F3 isoforms is regulated by alternative promoter usage and mutually exclusive exon splicing.
Leukotriene B 4 (LTB 4 ) 1 is synthesized from arachidonic acid by the sequential action of 5-lipoxygenase and leukotriene A 4 hydrolase enzymes in neutrophils, monocytes and macrophages (1). It is a potent chemoattractant for human neutrophils and also has chemotactic activity for monocytes (2)(3)(4)(5)(6). LTB 4 exerts this activity via a G protein-coupled receptor in target cells (7) and induces a cascade of cellular events that amplify the inflammatory response (8). The generation of LTB 4 is implicated in the pathogenesis of inflammatory disorders that involve prominent neutrophil infiltration of tissues (8,9).
Recently, it has been suggested that LTB 4 can also play an intracellular role in regulating transcription by functioning as a ligand for peroxisome proliferator-activated receptor ␣ (10). The bioactivity of LTB 4 is determined by the regulation and kinetic properties of the enzymes that control its synthesis and catabolism. LTB 4 is inactivated by -oxidation of its terminal carbon to yield 20-OH LTB 4 and 20-COOH LTB 4 (10,11). The initial -hydroxylation is catalyzed by CYP4F3 (12)(13)(14)(15)(16), which by biochemical analysis has a restricted localization to neutrophils, and to a lesser extent monocytes (12,17). The CYP4 family of enzymes catalyze the -hydroxylation of a large number of fatty acids and arachidonic acid derivatives (18). The CYP4F subfamily utilizes LTB 4 as a substrate. Two members of this subfamily have been identified in humans and are designated CYP4F3 (16,19) and CYP4F2 (20). Studies of neutrophil microsomes or recombinant CYP4F3 have determined the specificity of CYP4F3 for LTB 4 (15,21). CYP4F1, -4, -5, and -6 are designations given to rat enzymes (22,23).
The two human 4F subfamily members differ in their tissue localization and kinetic properties (16,20). CYP4F3 is expressed in neutrophils and monocytes, whereas CYP4F2 is expressed in liver (20). CYP4F2 has a K m for the -hydroxylation of LTB 4 that is ϳ60-fold higher than CYP4F3 (20). Both enzymes have 520 amino acids and share 93% amino acid sequence identity throughout most of the molecule. However, the sequence diverges between amino acids 66 and 114, where there is only 27% identity. It is not known whether the sequence variation within this 48-amino acid segment accounts for the different K m values of the two enzymes.
We have cloned both a novel isoform of CYP4F3 expressed in fetal and adult liver and the CYP4F3 gene. The liver and neutrophil transcripts of CYP4F3 are generated by tissue-specific splicing of exons 3 and 4, respectively, and are initiated from different transcription start sites. Exons 3 and 4 are identical in size and each code for amino acids 66 -114 of CYP4F3. Mutually exclusive selection of the exons is correlated with the generation of high K m and low K m forms of the enzyme; liver CYP4F3 containing exon 3 has a K m for LTB 4 which is ϳ26-fold higher than neutrophil CYP4F3 containing exon 4. Analysis of the gene indicates that putative TATA boxes are located upstream of both the liver and neutrophil transcription start sites. These findings suggest that the generation of CYP4F3 isoforms is regulated by a combination of alternative promoter usage and tissue-specific splicing. In addition, we propose a general model for CYP4F family gene organization and expression.

EXPERIMENTAL PROCEDURES
Isolation and Analysis of Genomic Clones-A human genomic P1 library (Genome Systems, St. Louis, MO) was screened by PCR using primers N1 and N3 (Table I). The PCR conditions were 94°C for 1 min, 55°C for 1 min, and 72°C for 1.5 min; 35 cycles were followed by 1 cycle with a 10 min extension time. Clones that gave a PCR product of 78 bp on a 1% agarose gel were isolated. Two P1 clones, each with genomic inserts spanning more than 75 kb, were isolated to purity after 3 rounds of screening. A BamHI restriction digest of one clone (P1-A) was shotgun subcloned into pZErO-1 vector (Invitrogen). Positive subclones were identified by Southern blotting using the 32 P-labeled coding region of CYP4F3 cDNA as a probe. Five different BamHI fragments were identified, which collectively span the CYP4F3 gene when arranged in a linear order. Plasmid DNA was purified using Qiagen plasmid kits and sequenced at the Massachusetts General Hospital core facility using a combination of forward and reverse primers. Exons and intronexon junctions were sequenced using primers designed from the cDNA of CYP4F3. Introns and the 3Ј-and 5Ј-flanking regions of the gene were sequenced by primer walking. The distances of the large introns were determined by PCR using the Expand Long Template System (Roche Molecular Biochemicals) and primer pairs from flanking exons.
RNA Isolation, RT-PCR, and 5Ј-RACE-Human peripheral blood neutrophils were separated from whole blood by fractionation on Mono-Poly resolving medium (ICN) and total RNA was prepared from the cells by the RNA STAT-60 procedure (Tel-Test Inc.). Total RNA from human liver and fetal liver (15-24 weeks) was purchased from CLON-TECH. Total RNA was isolated from COS-7 cells using the RNeasy Mini Kit (Qiagen). First-strand cDNA synthesis was performed using the cDNA Cycle Kit (Invitrogen) with avian myeloblastosis virus reverse transcriptase and random primers. The cDNA was purified by phenolchloroform extraction and ethanol precipitation. Alternative splice forms of CYP4F3 and CYP4F2 were assayed by PCR using forward primers N5 and L5 and reverse primers NL6e3 and NL6e4. The PCR conditions were 94°C for 1 min, 55°C for 1 min, and 72°C for 1 min; 30 cycles were followed by 1 cycle with a 10-min extension time. PCR products were analyzed on a 2% agarose gel.
A novel isoform of CYP4F3 containing exon 3 (CYP4F3B) was cloned from fetal liver total RNA following first strand cDNA synthesis with a specific primer for the 3Ј-UTR of CYP4F genes (NL10). The product was amplified by a single round of PCR using primers N1 and NL9 and cloned into pCR2.1-TOPO (Invitrogen). The PCR conditions were 94°C for 1 min, 50°C for 1 min, and 72°C for 1.5 min; 30 cycles were followed by 1 cycle with a 10-min extension time. Clones containing exon 3 were identified by PCR screening with primers N5 and NL6e3 and were sequenced.
The start sites of transcription were determined by 5Ј-RACE (Life Technologies, Inc.). First-strand cDNA synthesis was performed with a primer (NL10) that recognized the 3Ј-UTR of CYP4F3 and CYP4F2. The cDNA was purified and tailed with dCTP by terminal transferase. First-round PCR was with the anchor primer and either primer N7 (CYP4F3) or L7 (CYP4F2). Second-round PCR was with the universal amplification primer and either primer NL6e3 (exon 3 of CYP4F2 and CYP4F3) or NL6e4 (exon 4 of CYP4F2 and CYP4F3). PCR conditions were as before. The final PCR product was ligated into pCR2.1-TOPO vector (Invitrogen) and transformed into TOP10 cells. Plasmid DNA was purified with SNAP kits (Invitrogen) and sequenced using primer N2 (CYP4F3) or L4 (CYP4F2).
Preparation and Affinity Purification of a Polyclonal Antiserum to CYP4F3-(410 -520)-A partial CYP4F3 cDNA coding for amino acids 410 -520 was amplified by PCR using primers N8 and NL10 and ligated into pCRII TA vector (Invitrogen). The partial cDNA was then sub-cloned in frame into the EcoRI site of pGEX-2TK (Amersham Pharmacia Biotech), which generates an N-terminal GST fusion protein, and also into the EcoRI site of pMAL-p2 (New England Biolabs), which generates an N-terminal MBP fusion protein. To induce the expression of fusion proteins, DH5␣ cells were transformed with the pGEX-2TK or pMAL-p2 constructs and incubated with 0.1 mM isopropyl-1-thio-␤-Dgalactopyranoside for 2 h at 37°C. The cells were collected by centrifugation at 10,000 ϫ g for 10 min and were resuspended in PBS containing 0.1% Triton X-100. The cells were disrupted by sonication for 3 ϫ 1 min on ice using a Vibracell probe sonicator (50% output, setting 6), and the sonicate was centrifuged at 10,000 ϫ g for 10 min. The GST fusion protein was found in the pellet of sonicated DH5␣ cells. The pellet was resuspended in SDS gel loading buffer and the proteins separated on a 10% SDS gel and detected with Coomassie Blue staining. The portion of the gel containing the GST fusion protein (38 kDa) was excised, ground in a minimal volume of PBS, and mixed with complete Freund's adjuvant for immunization of rabbits. Blood samples (15 ml) were obtained biweekly from the ear veins, and antiserum was prepared and stored at Ϫ20°C.
The MBP-CYP4F3-(410 -520) fusion protein was found in the supernatant of sonicated DH5␣ cells and was purified by amylose affinity chromatography according to the manufacturer's instructions. The purity of the fusion protein was confirmed by SDS-polyacrylamide gel electrophoresis with Coomassie Blue staining. It was then coupled to cyanogen bromide-activated Sepharose 4B. Antibodies to CYP4F3 were precipitated from serum at a concentration of 55% saturated ammonium sulfate, resuspended in 10 mM potassium phosphate buffer, pH 7.2, and collected in the flow-through of a DEAE-Sepharose column equilibrated in the same buffer. The antibodies were then bound to the CYP4F3-(410 -520) affinity column, washed with 10 column volumes of PBS, and eluted with glycine chloride buffer (0.05 M glycine, 0.15 M NaCl, pH 2.3) into 0.5 M sodium phosphate buffer, pH 7.7. The final composition of each fraction was 0.5 ml of eluate and 0.25 ml of sodium phosphate buffer. The fractions were screened at a dilution of 1:1000 against Western blots containing 1 g of MBP-CYP4F3-(410 -520) fusion protein. Peak fractions (2, 3, and 4) were pooled, dialyzed against PBS, and stored in PBS containing 0.05% azide.
Cell Transfections and Fractionation-COS-7 cells were maintained in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum. The cells were grown to 50% confluence and transfected with 10 g of plasmid DNA/100-mm dish using SuperFect Reagent (Qiagen). After 48 h the cells were washed three times in PBS and harvested in disruption buffer (50 mM Tris-HCl, pH 7.4, 0.25 M sucrose, 1 mM EDTA). They were disrupted with a Vibracell probe sonicator (3 ϫ 30 s, 50% output, 5 ϫ 10 7 cells/ml) and centrifuged at 12,000 ϫ g for 10 min. The supernatant was centrifuged at 100,000 ϫ g for 60 min. The microsomal membrane pellet was resuspended in disruption buffer by brief sonication. All fractionation procedures were at 4°C. Total protein content was assayed by the Bio-Rad method, and the membranes were stored at Ϫ70°C.
Western Blot Analysis-Proteins in cell extracts and microsome preparations were fractionated by SDS-polyacrylamide gel electrophoresis and electrophoretically transferred to nitrocellulose Trans-Blot membranes (Bio-Rad). The membranes were blocked for 1 h in PBS containing 3% nonfat dried milk, pH 7.4, and then incubated for 1 h with affinity purified anti-CYP4F3-(410 -520) diluted 1:100 in blocking buffer. The membranes were washed three times in PBS and incubated TABLE I Summary of primers Sequences of primers referred to in text are shown. Primers are numbered sequentially in order of their cDNA location and are labeled to indicate whether the sequence is specific for CYP4F3 (N) or CYP4F2 (L), or is identical in both CYP4F3 and CYP4F2 (NL). Numbering of cDNA positions is relative to the ATG initiation codon (A ϭ ϩ1). ACTGCTTCATTCCTTTTTATGG for 1 h with peroxidase-conjugated protein A (Roche Molecular Biochemicals) diluted 1:5000 in blocking buffer. Immunoreactive protein bands were visualized using the enhanced chemiluminescence method (Amersham Pharmacia Biotech). LTB 4 -Hydroxylase Assay-The conversion of LTB 4 to 20-OH LTB 4 was determined as described previously (14). Reaction mixtures containing 20 g of microsomal protein, [ 3 H]LTB 4 (40,000 cpm/nmol), 100 M NADPH, and 50 mM Tris-HCl, pH 8.0, were incubated in a final volume of 0.1 ml for 20 min at 37°C. The reaction was terminated by the addition of 4 drops of 2 M citric acid, and the substrate and product were extracted twice with 0.1 ml of ethyl acetate, resolved by thin layer chromatography, and quantitated by ␤-scintillation counting.
Chromosomal Localization-The genomic clone P1-A was used as a probe to map the chromosomal localization of the CYP4F3 gene by fluorescence in situ hybridization at Genome Systems. The clone was labeled with digoxigenin dUTP by nick translation and then hybridized to normal metaphase chromosomes derived from Phaseolus vulgaris agglutinin-stimulated lymphocytes from a male donor. The hybridization buffer was 50% formamide, 10% dextran sulfate, and 2 ϫ SSC. Specific hybridization signals were detected by incubating hybridized chromosome slides with fluoresceinated anti-digoxigenin, followed by counterstaining with 4,6-diamidino-2-phenylindole. An anonymous genomic clone known to map to chromosome 19q was used as an internal control in cohybridization experiments with the P1-A probe to confirm the identity of chromosome 19.

Identification of a New Exon in the CYP4F3
Gene-A map of the CYP4F3 gene comprising 14 exons and 13 introns was constructed by DNA sequencing and PCR analysis of genomic subclones as described under "Experimental Procedures." The first exon is non-coding, and the ATG translation start codon is located at the beginning of exon 2. All introns conform to the gt-ag rule (24), and a summary of the splice junctions is shown in Table II. These data are in agreement with a recent description of the gene (19), but include a previously unrecognized exon, designated exon 3, which is not incorporated into the characterized cDNA of neutrophil CYP4F3 (16). Examination of the sequence of exons 3 and 4 and the genomic flanking regions (Fig. 1A) revealed several features that suggested exon 3 might have the capacity to participate in alternative splicing reactions with exon 4. 1) The sequences bordering exon 3 conform to recognizable splice junctions. 2) Exon 3 is identical in size to exon 4 (145 bp), and its predicted translation product has 13 amino acids in common with exon 4. 3) Exon 3 has a homolog, which is expressed in the CYP4F2 gene. Analysis of the genomic sequence of CYP4F2 deposited by the Human Genome Project (GenBank accession AC005336) reveals that it contains 14 exons that are directly analogous to those of CYP4F3. A comparison of the similarities in gene sequences in the regions of exons 3 and 4 is shown in Fig. 1A. Examination of the characterized CYP4F2 cDNA from liver (20) indicates that exon 3 from the gene is incorporated in this case whereas exon 4 is excluded. 4) The intron separating exons 3 and 4 (intron 3) of CYP4F3 contains several pyrimidine-rich tracts, which resemble PTB binding sites that regulate alternative splicing in certain pre-mRNAs (25)(26)(27)(28)(29)(30)(31)(32). These polypyrimidine tracts are characterized by the inclusion of one or more specific sequence motifs (31). They are functional as splicing repressor signals in pre-mRNA, but their existence can be inferred from the gene sequence. No comparable polypyrimidine tracts were observed in other introns.
To confirm the possibility that exon 3 provides an alternative substrate for splicing we searched for CYP4F3 transcripts which incorporate this exon. An important consideration is that the assay used must be capable of distinguishing related CYP4F2 transcripts that are encoded by a separate gene. The sequences of exons 3 and 4 shown in Fig. 1A were used to develop an RT-PCR assay that distinguishes between all possible isoforms that might be generated by alternative splicing of CYP4F3 and CYP4F2. Forward primers specific for exon 2 of either CYP4F3 (N5) or CYP4F2 (L5) were paired with reverse primers, which had an identical sequence in the genes for CYP4F2 and CYP4F3 but which were specific for either exon 3 (NL6e3) or exon 4 (NL6e4). The four different combinations of primers each generate an isoform-specific PCR product of 158 bp and collectively give a profile of the alternatively spliced transcripts expressed in a given tissue (Fig. 1B).
The RT-PCR assay demonstrated that peripheral blood neutrophils derived from an adult donor exclusively express CYP4F3 containing exon 4 (Fig. 1B, lanes 1-4). This isoform corresponds to the previously characterized cDNA from neutrophils (16). It could not be detected in adult liver (lane 8), but is expressed in fetal liver (lane 12). This observation may derive from hematopoiesis in fetal liver (33). Adult liver expresses CYP4F2 containing exon 3 (lane 5), and this corresponds to the previously characterized cDNA from liver (20). This CYP4F2 isoform could also be detected in fetal liver (lane  There is no evidence for expression of exon 4 in CYP4F2, but differences in predicted amino acid sequence are included for comparison. Primers N5, L5, NL6e3, and NL6e4 were used to detect alternative splice products in the RT-PCR assay. Intron sequences are shown in lowercase letters. Polypyrimidine tracts (underlined) are located within intron 3 and contain splicing repressor motifs (boxed). Sequences that can be classified as DNA repetitive elements (retrotransposons) are abbreviated as MLT2d (mammalian LTR-transposon 2d) and L2 (LINE 2). Arrows indicate sense orientation of each repetitive element. B, PCR analysis of exon 3 and 4 expression. Total RNA from adult blood neutrophils (lanes 1-4), adult liver (lanes 5-8), or fetal liver (lanes 9 -12) was used as a substrate for reverse transcription using random primers. The cDNA was analyzed by PCR using the following primer pairs: L5 and NL6e3 to detect CYP4F2 containing exon 3 (lanes 1, 5, and 9), L5 and NL6e4 to detect CYP4F2 containing exon 4 (lanes 2, 6, and 10), N5 and NL6e3 to detect CYP4F3 containing exon 3 (lanes 3, 7, and 11), and N5 and NL6e4 to detect CYP4F3 containing exon 4 (lanes 4, 8, and 12). Each primer pair is predicted to generate a product of 158 bp. cal to the previously characterized cDNA from neutrophils (16) except for the following: 1) residues 199 -343 (numbering from ATG initiation codon) are coded for by exon 3 in the CYP4F3 gene, not exon 4; 2) the 5Ј-UTR is derived from a different part of the CYP4F3 gene (see below). The results confirm that the cDNA is an alternative splice product of the CYP4F3 gene, which shows a distinct tissue-specific pattern of expression.
The CYP4F3 cDNAs derived from neutrophils (containing exon 4) and fetal liver (containing exon 3) were designated CYP4F3A and CYP4F3B, respectively. They were cloned into pcDNA3 (Invitrogen) and expressed in COS-7 cells. CYP4F3A has been purified and characterized previously (19,21) but was included as a control in expression and kinetic studies to provide a direct experimental comparison with CYP4F3B. Total RNA was isolated from the COS-7 cells and analyzed by RT-PCR using the same four primer combinations as before ( Fig.  2A). Cells transfected with CYP4F3A gave a PCR product of 158 bp with the CYP4F3-specific primer (N5) and exon 4-specific primer (NL6e4) as shown in Fig. 2A (lane 4). All other primer combinations were negative (lanes 1-3). Cells transfected with CYP4F3B gave a PCR product only with the CYP4F3-specific primer (N5) and exon 3-specific primer (NL6e3) as shown in Fig. 2A (lane 7). No PCR products were obtained from non-transfected cells (lanes 9 -12).
The expression of CYP4F3 protein was confirmed using af-finity-purified anti-CYP4F3 antibody. Microsomal fractions of transfected COS-7 cells were analyzed by Western blotting (Fig. 2B). An immunoreactive band with an apparent molecular mass of 61 kDa was observed in microsomes from cells transfected with CYP4F3A. An immunoreactive band with an identical size was observed in microsomes from cells transfected with CYP4F3B, but this band was not observed in microsomes from non-transfected cells. This confirms that the novel isoform CYP4F3B is expressed in cells as a full-length protein in its predicted subcellular location. The microsomal fractions of COS-7 cells transfected with CYP4F3A or CYP4F3B were used to determine the K m of the two isoforms using LTB 4 as a substrate. In a representative experiment (Fig. 2C), the K m was determined to be 3.9 M for CYP4F3A and 90 M for CYP4F3B. The V max was 5.9 nmol/ min/mg for CYP4F3A, and 7.  1-4), pcDNA3/CYP4F3B (lanes 5-8), or were not transfected (lanes 9 -12). Total RNA was extracted from the cells 48 h after transfection and used as a substrate for reverse transcription using random primers. The cDNA was analyzed by PCR using primer pairs L5 and NL6e3 (lanes 1, 5, and 9), L5 and NL6e4 (lanes 2, 6, and 10), N5 and NL6e3 (lanes 3, 7, and 11), or N5 and NL6e4 (lanes 4, 8, and 12) differences between these exons must therefore account for the different kinetic properties. A summary of the alternative splicing pathways of CYP4F3 is shown in Fig. 3. A revised map of the genomic organization, originally suggested to comprise 13 exons and 12 introns (19), is included.
5Ј-RACE Analysis of CYP4F Transcripts-The sequence for the CYP4F3 gene shares 80% identity with the reported sequence of the CYP4F2 gene over a region that extends into the 5Ј-flanking region (Fig. 4). The close sequence similarities between CYP4F2 and the two isoforms of CYP4F3 must be taken into account when distinguishing their respective transcription start sites in tissues with overlapping expression. We used a 5Ј-RACE system, which exploits a series of nested PCR reactions with isoform-specific primers to generate 5Ј-cDNA ends of determined identity (Fig. 4A).
We first tested the 5Ј-RACE procedure using neutrophil RNA, which contains a single type of CYP4F transcript (CYP4F3A). The sequence of 10 5Ј-RACE products were identical, and indicated that transcription of CYP4F3A begins 49 bp upstream from the 3Ј-end of exon 1 (Fig. 4B, start site A). We numbered the gene sequence relative to this start site (ϩ1). The sequence matched the previously reported cDNA sequence of neutrophil CYP4F3 with an additional 17 bp at the 5Ј-end. These data are in agreement with conclusions derived from S1 nuclease protection analysis of neutrophil RNA (19). The presence of a TATA box-like sequence and a consensus site for the myeloid-specific factor MZF-1 upstream of this initiation site has been noted previously (19).
We then applied the 5Ј-RACE system to fetal liver RNA to selectively identify the 5Ј-end of the novel CYP4F3B transcript cloned from this source. Ten CYP4F3B RACE products were cloned and sequenced. Eight clones identified a major start site of transcription located within intron 1 at position ϩ449, 70 bp upstream of the 5Ј-end of exon 2 where translation is initiated (Fig. 4B, start site B1). Two clones identified a minor start site of transcription located at ϩ489, 30 bp upstream of the 5Ј-end of exon 2 (Fig. 4B, start site B2). Sequence analysis confirmed that exon 1 is excluded from CYP4F3B cDNA and exon 2 is extended by an additional 70 bp at its 5Ј-end (or 30 bp if initiated from the minor start site). The 5Ј-UTRs of CYP4F3A and CYP4F3B are therefore derived from different parts of the gene. A TATA box is located 27 bp upstream of the major start site B1. This suggests that the CYP4F3 gene contains two promoter regions: one upstream of start site A, which may direct expression in myeloid cells, and a second upstream of start site B, which may direct expression in liver.
5Ј-RACE analysis of ten CYP4F2 cDNAs from liver and fetal liver identified a single start site of transcription. Its location exactly matches the initiation site of CYP4F3A (start site A) when the two gene sequences are aligned. There is a TATA box located 23 bp upstream of the CYP4F2 start site (GCTATAT-CAA), which is directly aligned with the TATA-like box upstream of start site A in the CYP4F3 gene (CCTACATCAG). The MZF-1 site in CYP4F3 (TGTGGGGA) is not present in CYP4F2 because of two nucleotide differences (TGCAGGGA). The TATA box upstream of start site B in the CYP4F3 gene (TCTATCTCCA) is not observed in CYP4F2 because of two nucleotide differences (TCTACCTCCG). These observations would be consistent with the CYP4F2 gene containing a single promoter region that is not myeloid-specific. A model for CYP4F family gene organization and expression is shown in Fig. 5.
Chromosomal Localization of the CYP4F3 Gene-The chromosomal location of the P1-A genomic clone containing the CYP4F3 gene was mapped by fluorescence in situ hybridization (Fig. 6). A total of 80 metaphase cells were analyzed, with 71 exhibiting specific labeling on chromosome 19p. Measurements of 10 specifically hybridized chromosomes demonstrated that the CYP4F3 gene is localized at a position that is 28% of the distance from the centromere to the telomere in an area that corresponds to band 19p13.1. Cohybridization experiments with the P1-A probe and a control probe for 19q gave twin-spot signals on both arms of chromosome 19 (Fig. 6A), whereas hybridization with the P1-A probe alone gave a twin-spot signal only on the short arm (data not shown). These results make an adjustment to the previously estimated location of the CYP4F3 gene (19), and support the possibility that it is part of a gene cluster with the CYP4F2 gene, which has been localized to 19p13.1 by the Human Genome Project.
FIG. 3. CYP4F3 gene organization and splicing pathways. Four major BamHI fragments that span the CYP4F3 gene, two of ϳ4 kb and two of ϳ10 kb, were isolated as genomic subclones in pZErO-1 and arranged in a linear order as shown. A subclone containing a fragment that extends 6.5 kb further into the 5Ј-flanking region was also isolated. A map of 14 exons and 13 introns was constructed; solid boxes represent exons. Exon 3 is excluded from the previously characterized cDNA of peripheral blood neutrophils, which is designated as CYP4F3A. Exon 3 replaces exon 4 in a novel cDNA identified in fetal and adult liver (CYP4F3B) and a different transcription start site excludes exon 1 and extends exon 2 (hatched box). The tissue-specific splicing of exon 4 or exon 3 is correlated with the generation of low K m and high K m forms of the enzyme, respectively.

DISCUSSION
Human cytochrome P450 4F3 (CYP4F3) catalyzes the -hydroxylation of LTB 4 and has the potential to act as an inhibitory control point in inflammation. The gene spans approximately 20 kb and comprises 14 exons and 13 introns (Table II), including the previously unrecognized exon 3. Exons 3 and 4 are identical in size but only have 27% amino acid sequence identity (Fig. 1A). We developed an RT-PCR assay, which selectively determined the expression of exons 3 and 4 (Fig. 1B), and demonstrated that the two exons are alternatively spliced in a mutually exclusive fashion to generate distinct CYP4F3 isoforms.

FIG. 4. Transcription initiation sites of the CYP4F3 gene.
A, single-strand cDNAs were generated from tissue RNA by reverse transcription with primer NL10 and tailed with dCTP using terminal transferase. Each of the three types of CYP4F transcript were uniquely selected for subsequent 5Ј-RACE analysis by the sequential use of two isoform-specific primers in nested PCR reactions. DNA sequences are in the 5Ј to 3Ј direction with respect to the gene orientation. The sequences shown extend from the 5Ј-end to the ATG translation start codon (underlined). B, the relationship of the 5Ј-RACE sequences to the gene sequence is shown. Transcription initiation of CYP4F3A in neutrophils from start site A generates a transcript (gray shading) with a 5Ј-UTR consisting of exon 1, which is spliced to exon 2. Initiation of CYP4F3B in fetal liver from start sites B1 (major) or B2 (minor) generates a transcript (underlined), which excludes exon 1 and extends exon 2 to make a distinct 5Ј-UTR. The region of exon 2 common to both CYP4F3A and CYP4F3B transcripts begins 1 bp prior to the ATG translation initiation codon (clustered arrowheads). Predicted TATA boxes are labeled A or B to distinguish whether they are located in the region upstream of start site A (putative neutrophil promoter) or start site B (putative liver promoter), respectively. A single start site of CYP4F2 transcription is aligned with start site A in CYP4F3, and nucleotides are numbered relative to this location (ϩ1). Exon sequences are shown in uppercase letters. Intron sequences and 5Ј-flanking sequences are shown in lowercase letters. Differences in the CYP4F2 gene are shown below the CYP4F3 sequence. 5Ј-RACE products were sequenced with primer N2.
Analysis of neutrophil RNA by RT-PCR indicated that myeloid CYP4F3 consists exclusively of the isoform containing exon 4 (CYP4F3A). We identified a novel CYP4F3 cDNA containing exon 3 (CYP4F3B) in adult liver but could not detect CYP4F3A in this tissue. Alternative splicing of CYP4F3 therefore exhibits a distinct tissue specificity. Fetal liver contains both alternative splice forms of CYP4F3 but it is possible that a cell-specific pattern of splicing is maintained. The liver is a site of hematopoiesis in the fetus (33), and although the production of neutrophils may be limited (34,35), neutrophil-specific proteins can be detected by RT-PCR (35).
The novel CYP4F3 isoform (CYP4F3B) was cloned and expressed in COS-7 cells to enable a comparison of its properties with the previously characterized isoform (CYP4F3A). The K m values for LTB 4 (Fig. 2C) were directly correlated with inclusion of exon 4 (low K m ) or exon 3 (high K m ). The differences in amino acid sequence between the two exons must therefore account for the different kinetic properties of these two isoforms of CYP4F3. The selection of exon 4 in myeloid cells may have importance in maximizing the efficiency of LTB 4 inactivation in situations that require fine control of inflammatory reactions.
Exon switching may depend on positive signals to promote selection of exon 3 in liver or negative signals to repress exon 4 selection. Recent work on cell-specific splicing of pre-mRNAs for tropomyosins, c-Src, and GABA A receptor ␥2 suggest that a complex balance of positive and negative regulatory elements is likely to be required (25)(26)(27)(28)(29)(30)(31)(32). The negative signals in these pre-mRNAs generally consisted of polypyrimidine tracts with specific sequence motifs, which act as binding sites for PTB. We identified typical polypyrimidine tracts within intron 3 of the CYP4F3 gene. Their distribution is most reminiscent of the rat ␤-tropomyosin gene, where a cluster of PTB binding sites within one intron suppresses splicing of a skeletal musclespecific exon immediately downstream. By analogy, PTB may regulate CYP4F3 splicing by repressing exon 4 selection in liver, and some form of derepression may occur in neutrophils. However, polypyrimidine tracts can be located upstream or downstream of a regulated exon, and PTB is a positive regulator of exon splicing in calcitonin pre-mRNA (36). Further analysis is therefore required to determine the functional status of these elements in CYP4F3. There is a duplication of a 5Ј-splice donor sequence in CYP4F3, which comprises 6 bp at the 3Ј-end of exon 4 and the following 19 bp of intron 4 (Fig. 1A). It is not clear how this might affect assembly of the spliceosome, but cooperative binding of factors to duplicate binding sites could provide a competitive advantage in the splicing reaction, or could inhibit splicing by hyperstabilizing the spliceosome complex.
The human CYP4F3 and CYP4F2 genes are highly related and have a similar organization. Alternative splicing of CYP4F2 has not been demonstrated, but we suggest a similar designation of 14 exons and 13 introns to the gene. The characterized form of CYP4F2 (20) has a number of properties in common with the novel CYP4F3B isoform described in this study. Both contain exon 3 from their respective genes; they show a similar tissue distribution in liver and fetal liver, but not in neutrophils; and they have a high K m relative to neutrophil CYP4F3A. This is consistent with the interpretation that the K m of CYP4F enzymes is determined by the inclusion of exon 3 or 4 into the mature mRNA transcript, and that splicing of these exons occurs in a tissue-specific pattern. Polypyrimidine tracts containing splicing repressor motifs are observed within intron 3 of the CYP4F2 gene, suggesting that splicing might be regulated by a similar mechanism. An alternative splice form of CYP4F2 containing exon 4 has not yet been observed. This exon has 90% homology with exon 4 in the CYP4F3 gene, and its splice junctions are almost identical in sequence (Fig. 1A). However, the splice junction at the 3Ј-end of exon 4 is not duplicated in CYP4F2.
Neutrophil CYP4F3A and fetal liver CYP4F3B have different 5Ј-UTRs initiated from distinct transcription start sites (referred to as start sites A and B, respectively, in Fig. 4B). A sequence resembling a TATA box is located 23 bp upstream of start site A, but a stronger consensus TATA box sequence is located 27 bp upstream of start site B. The CYP4F3 gene therefore contains two putative promoter regions. Promoter A (upstream of transcription start site A) may direct myeloidspecific expression. Promoter B (upstream of start site B) may direct expression in liver. There is no evidence for two promoters in the CYP4F2 gene, and its single transcription start site can be aligned exactly with the CYP4F3A start site in neutrophils (start site A). The gene sequences of CYP4F2 and CYP4F3 are 80% identical over a region that extends 168 bp upstream of this start site. However, CYP4F2 is not expressed in neutrophils. A small number of nucleotide changes may therefore have been sufficient to convert the promoter region upstream of start site A to a myeloid-specific form subsequent to the divergence of the two genes. These could include the gain of sites for myeloid factors such as MZF-1, and perhaps the loss of TATA box activity. The consensus TATA box in the CYP4F2 gene is replaced by a weaker TATA-like sequence upstream of start site A in the CYP4F3 gene. Whatever the evolutionary sequence of events, selection has led to greater specificity for LTB 4 (CYP4F family), lower K m (exon 4), and myeloid expression (CYP4F3A).
A hypothetical model for the regulation of CYP4F genes is shown in Fig. 5. The model makes some simple assumptions which are easily testable and provides a basis for future studies. Promoters A and B in CYP4F3 are depicted as independent control elements, but it remains a possibility that they share some regulatory sites. Tissue-specific splicing pathways are shown to be determined by either neutrophil-or liver-specific factors. However, it is possible that regulated splicing occurs in only one of the two tissues while the other exhibits a default pathway determined by the basal splicing factors. Some of the features described in Fig. 5 have similarities in other genes. Mutually exclusive exon splicing is regulated by PTB in the rat ␣and ␤-tropomyosin genes (25)(26)(27)(28). The chicken ␤-tropomyosin gene uses two promoters and mutually exclusive exon splicing to generate three tissue-specific isoforms (37). Alternative promoter usage accounts for tissue-specific patterns of expression of CYP19 (38) and CYP27 (39) but generates an identical enzyme in all cells.
CYP4F3A plays an important role in the control of inflammation by catalyzing the -hydroxylation of LTB 4 in inflammatory cells with a unique low K m for the reaction. Bioactivity of CYP4F3A probably depends on the interplay of cell-specific transcription and splicing factors. Characterization of these factors and their target elements in the CYP4F3 gene may help to elucidate the regulatory pathways that determine the extent of an inflammatory response.