Structural Determination of the Substrate Specificities and Regioselectivities of the Rat and Human Fatty Acid ω-Hydroxylases

Archives of Biochemistry and Biophysics Vol. 373, No. 1, January 1, pp. 63–71, 2000 Article ID abbi.1999.1504, available online at http://www.idealibrary.com on

Structural Determination of the Substrate Specificities and Regioselectivities of the Rat and Human Fatty Acid v-Hydroxylases 1 Ute Hoch,* Zhoupeng Zhang,* Deanna L. Kroetz,† and Paul R. Ortiz de Montellano* ,2 *Department of Pharmaceutical Chemistry and †Department of Biopharmaceutical Sciences, University of California, San Francisco, California 94143-0446

Received July 14, 1999, and in revised form September 16, 1999

The substrate and regiospecificities of the known CYP4A enzymes from rat (CYP4A1, -4A2, -4A3, and -4A8) and human (CYP4A11) have been determined using lauric (C12), myristic (C14), palmitic (C16), oleic (C18:1), and arachidonic (C20:4) acids. The CYP4A2 and CYP4A8 cDNAs required to complete the enzyme set were cloned from a rat kidney library. All five proteins were expressed in Escherichia coli and were purified with the help of a six-histidine tag at the carboxyl terminus. Two complementary CYP4A2CYP4A3 chimeras fused at residue 119 (CYP4A2) and 122 (CYP4A3) were constructed to explore the roles of the 18 amino acid differences between the parent proteins in determining their catalytic profiles. The chimera in which the first 119 amino acids are from CYP4A2 indicates that the first 120 amino acids control the substrate specificity. The chimera in which the first 122 amino acids are from CYP4A3 is inactive due to a defect in electron transfer to the heme group. The highest activity for lauric acid was obtained with CYP4A1 and CYP4A8, but for all the proteins the activity decreased with increasing fatty acid chain length. The fact that none of the rat and human CYP4A enzymes exhibits a high activity with arachidonic acid appears to limit their role as catalysts for the physiologically important conversion of arachidonic acid to 20-hydroxyeicosatetraenoic acid (20-HETE). © 2000 Academic Press

1 This work was supported by National Institutes of Health Grants GM25515 (PROM) and HL53994 (DLK). 2 To whom correspondence and reprint requests should be addressed at the Department of Pharmaceutical Chemistry, University of California, San Francisco CA 94143-0446. Fax: (415) 502-4728. E-mail: [email protected].

0003-9861/00 $35.00 Copyright © 2000 by Academic Press All rights of reproduction in any form reserved.

The CYP4A family of P450 3 enzymes catalyzes the metabolism of fatty acids and exhibits a unique selectivity for hydroxylation of the thermodynamically disfavored terminal methyl group (1, 2). Four isoforms have been identified in rat: CYP4A1, -4A2, -4A3, and -4A8 (1, 2). The first three enzymes are present in rat liver and are induced by clofibrate (2). All four enzymes are constitutively expressed in the kidney, albeit at varying levels (2– 4). CYP4A11 is the only characterized CYP4A human fatty acid v-hydroxylase (5–7). The four rat enzymes and human CYP4A share 72–96% amino acid similarity (1, 2, 8). This high sequence identity makes these enzymes well suited for structure-function studies, particularly as there is no crystal structure available for a eukaryotic, membranebound P450 enzyme. However, the relative substrate specificities of the different isoforms have not been definitively established. CYP4A1 is by far the best-characterized isoform with respect to substrate specificity. In addition to the wildtype enzyme, a fusion protein with cytochrome P450reductase has been heterologously expressed and characterized using different fatty acids and fatty acid analogs as substrates (9 –12). The only other rat isoform so far cloned, expressed, and purified is CYP4A2 (13– 15). Little information is available about the activity of CYP4A3 (9, 15) and nothing is known about CYP4A8. Even though the studies with CYP4A1 indicate a 12carbon chain length substrate preference for CYP4A1, the few studies performed with purified (16) or selectively inhibited CYP4A2 (17) have focused on arachidonic acid hydroxylation. We describe here our characterization of CYP4A1, -4A2, -4A3, -4A8, and -4A11 with regard to substrate 3

Abbreviation used: P450, cytochrome P450. 63

64

HOCH ET AL.

specificity and hydroxylation regioselectivity using lauric, myristic, palmitic, arachidonic, and oleic acids as test substrates. Moreover, we describe the generation and characterization of complementary CYP4A3_2 and CYP4A2_3 chimeras. This study takes advantage of the high sequence similarity between the parent CYP4A2 and CYP4A3 enzymes (96%) to shed some light on the structural basis of their exceptional specificity for the v-position. METHODS AND MATERIALS Materials. IPTG was purchased from Promega. Ampicillin, d-aminolevulinic acid, glycerol, lysozyme, dilauroylphosphatidylcholine, glutathione, catalase, NADPH, lauric acid, and 12-hydroxylauric acid were obtained from Sigma. Tris, EDTA, sodium cholate, NaCl, MgCl 2, H 2SO 4 and MeOH were purchased from Fisher. Emulgen 913 was a gift from Kao Chemicals. Tryptone and yeast extract were from Difco Laboratories. [ 14C]Lauric acid (55 mCi/mmol) and [ 14C]arachidonic acid (55 mCi/mmol) were purchased from American Radiolabeled Chemicals, Inc. [ 14C]Myristic acid (21.6 mCi/mmol), [ 14C]palmitic acid (40.2 mCi/mmol), and [ 14C]oleic acid (35 mCi/ mmol) were purchased from Sigma. The CYP4A1 (18) and CYP4A11 (19) proteins were obtained as described previously. Frank J. Gonzalez (National Cancer Institute, Bethesda, MD) provided the CYP4A3 cDNA. The human NADPH-cytochrome P450 reductase and cytochrome b 5 cDNAs were provided by Stephen M. Black (University of California, San Francisco), and the proteins were expressed and purified according to the protocol outlined previously (18). The pCWori vector was a gift from Robert Fletterick (University of California, San Francisco) (20). Isolation of CYP4A2 and CYP4A8 cDNAs. A full-length CYP4A2 cDNA (1673 bp) was isolated from Wistar Kyoto kidney RNA by a reverse transcription-polymerase chain reaction (RT-PCR) technique. Total kidney RNA was reverse transcribed using a poly(dT) primer and M-MLV reverse transcriptase. The reaction product was subsequently amplified by 30 cycles of PCR using the following primers based on the published genomic sequence of CYP4A2 (23): forward-59-ggggtaccccagaccctagtgatccaga-39 and reverse-59-ccatcgatggcagaaggatgggaatcaaag-39. The underlined nucleotides indicate restriction sites for KpnI and ClaI, respectively, which were present in the primer regions and used for ligating the PCR product into pGEM7Zf(1). A full-length 1712-bp CYP4A8 cDNA was also isolated by RT-PCR from Wistar-Kyoto kidney using the following primers based on its published sequence (24): forward, 59-ccatcgatggcatgagtggctct-39 and reverse-59-gctctagaaaagactgacagacaagg-39 with ClaI and XbaI restriction sites, respectively. The full-length CYP4A8 cDNA was ligated directly into pT7Blue(R) (Novagen, Madison, WI). The identity of CYP4A2 and CYP4A8 clones isolated by RT-PCR was confirmed by DNA sequencing of the entire inserts using dideoxymediated chain termination and Sequenase 2.0 (United States Biochemical, Cleveland, OH). The sequence of the CYP4A2 cDNA was in complete agreement with the genomic sequence (23). Distinct CYP4A2 cDNAs have been isolated by RT-PCR from Lewis–Wistar outer medulla RNA (19) and Sprague–Dawley kidney RNA (17), which contain a 121- or 123-bp deletion in the 39-untranslated region of the transcript. The primers used for isolation of the CYP4A2 cDNA in the present study were located in this deletion region and the resultant clone contains the first 81 bp of this sequence. Our studies found no evidence for a second CYP4A2 cDNA but the possibility of splice variants is currently being investigated. Helvig et al. also reported a C for A substitution at nucleotide 87 that results in the replacement of a Gln for a Lys at residue 20 and five additional substitutions in the 39-noncoding region (17). These substitutions were not present in our CYP4A2 cDNA.

Mutations in CYP4A2, CYP4A3, and CYP4A8. A sequence encoding a six-histidine tail at the carboxy terminus was added to all CYP4A genes via the polymerase chain reaction. The primers introduced an NdeI site, a SalI site, and the His 6-tail. Due to the high sequence similarity between CYP4A2 and CYP4A3 the same set of primers could be used: Primer 1 (amino terminus) (59-39), ggaattccatatggctctgttattagcagtttttctgcttttgctcagtctatttctg, primer 2 (carboxyl terminus) (59-39) acgcgtcgacttaatgatgatgatgatgatgtcctcttagcttcttgagacg. The CYP4A8 DNA was mutated by using the following primers: Primer 1 (amino terminus) (59-39), ggaattccatatggctctgttattagcagtttttctgcttgtgctcactgtgctcctg, primer 2 (carboxyl terminus) (59-39) acgcgtcgacttaatgatgatgatgatgatgtccttggagctttttgagacg. The sequences coding for the first 23 amino acids of the CYP4A enzymes were replaced by an oligonucleotide coding for the first 23 amino acids of the bovine sterol 17a-hydroxylase with the codon modifications introduced by the Waterman group (25). Introduction of an NcoI restriction site. In order to construct the CYP4A3_2 and CYP4A2_3 chimeras a new restriction site (NcoI) had to be introduced at amino acid position 119 (CYP4A2) and 122 (CYP4A3). The NcoI restriction site was introduced using overlapping oligos and the polymerase chain reaction (26). The overlapping oligos a and b for CYP4A2 and oligos c and d for CYP4A3 (oligo a, nucleotides 2130 –367; oligo b, nucleotides 354 –1524; oligo c, nucleotides 2130 –382; oligo d, nucleotides 363–1532) were obtained by using the following set of primers: primer a1 and c1 (amino terminus) (59-39): caggaaacaggatcagc;, primer a2 (carboxyl terminus (5939): cataaccaatccatggagcaagggattg; primer b1 and d1 (amino terminus) (59-39): tgctccatggattggttatg; primer b2 and d2 (carboxyl terminus (59-39): aaggcctgtcgacttatcttagcttctt; primer c2 (carboxyl terminus (59-39): tccatggagcaaggaattg. Oligos a– d were purified and used in a secondary PCR with the primers a1 and b2. The final PCR product was digested with NdeI and SalI and ligated into the pCWori vector. Construction of the CYP4A3_2 and CYP4A2_3 chimeras. In order to construct the CYP4A3_2 and CYP4A2_3 chimeras the pCWori vectors containing the CYP4A2 and CYP4A3 genes with the NcoI restriction site were digested with NdeI and NcoI resulting in 357 and 6132 bp fragments in the case of CYP4A2 and 366- and 6127-bp fragments in the case of CYP4A3. For construction of the CYP4A3_2, the small 357 bp fragment from the CYP4A2 digestion was ligated into the 6127-bp fragment of the CYP4A3 digestion. The CYP4A2_3 chimera was constructed by ligating the 366-bp fragment of the CYP4A3 digestion into the 6132-bp fragment of the CYP4A2 digestion. After confirming the sequence integrity, the chimeric enzymes were expressed in Escherichia coli as described below. Heterologous expression and purification of recombinant CYP4A2, CYP4A3, CYP4A8, CYP4A2_3, and CYP4A3_2. A slightly modified protocol based on that previously described for expression and purification of CYP4A1 (18) was used for CYP4A2, -4A3, -4A8, and -4A3_2. Briefly, the pCWori plasmid containing the appropriate CYP4A was transformed into E. coli XL-1 Blue (Stratagene), and a single colony was grown overnight in 23 YT with 100 mg/ml ampicillin. One liter of terrific broth with 200 mg/liter ampicillin was inoculated with 2 ml of the saturated overnight culture and grown at 37°C to an OD600 of 0.5 to 0.6. At this point, 80 mg/liter d-aminolevulinic acid and 1 mM isopropyl-b-D-thiogalactopyranoside were added. The flasks were cooled to 28°C and incubated for an additional 18 h (36 h for CYP4A8). To purify the proteins, the cells were harvested by centrifugation at 4000g for 20 min. The pellet was resuspended in 100 mM Tris (pH 7.8) containing 0.1 mM EDTA and 20% glycerol, and the suspension was stirred with lysozyme for 1 h at 4°C. The suspension on ice was sonicated for 4 min (1 min on, 1 min off, 50% power). The cell debris and unbroken cells were removed by centrifugation at 12,000g for 10 min. The supernatant was centrifuged at 100,000g for 1 h to pellet the cell membranes. The membranes were resuspended in 20 mM Tris (pH 7.4) containing 20% glycerol, and the CYP4As solubilized by the addition of 1%

FATTY ACID SUBSTRATE SPECIFICITY OF CYP4A ENZYMES Emulgen 913. The solution was stirred gently for 1 h at 4°C. The nonsolubilized protein was removed by centrifugation at 100,000g for 1 h. The solubilized CYP4As were purified on a Ni 21-NTA-agarose column (Novagen) as described previously for CYP4A1 and were desalted on a PD-10 column prior to use. Spectroscopic methods. Reduced CO-difference and substrate perturbation spectroscopic analyses were done on a Varian Cary 1E UV/visible dual-beam spectrophotometer. Absolute spectra were recorded on a Hewlett-Packard 8452 diode array spectrophotometer. Both instruments were equipped with a temperature control accessory. The P450 content was determined using the method of Omura and Sato (27). To obtain substrate binding difference spectra, the fatty acids were dissolved in dimethylsulfoxide and titrated into the different CY4A (1.5 mM) solutions using a 10-ml Hamilton syringe, resulting in a final sample volume change of less than 1%. All spectra were obtained at 25°C. The spectral binding constant K s was determined from the hyperbolic plot of the respective differences in the 420- to 390-nm peak to trough absorbance versus the ligand concentration. Measurement of CYP4A hydroxylation activity. CYP4A v-hydroxylation was measured by a modification of the method of Ortiz de Montellano and Reich (28). The enzyme reaction was set up in an Eppendorf vial (1.5 ml): 10 mg of dilauroylphosphatidylcholine, 0.1 mg of sodium cholate, 50 pmol of CYP4A, 500 pmol of cytochrome P450 reductase, 50 pmol of cytochrome b 5, 1.5 mmol of glutathione, and 5 mg catalase were gently mixed together and incubated for 10 min at room temperature. Next 15 mmol MgCl 2 was added and 50 mM Tris (pH 7.4) containing 0.02% sodium cholate, 250 mM NaCl and 20% glycerol was added to a final volume of 490 ml. Finally, 10 ml of 100 mM NADPH was added. Aliquots of 50 ml of the mixture were transferred to 0.6-ml Eppendorf vials. The reaction was started by adding 1 ml of a 5:1 unlabeled to 14C-labeled fatty acid mixture that produced a final fatty acid concentration of 100 mM. The reaction was carried out for 10 min for lauric acid and 30 min for the other fatty acids at 37°C and was then quenched by adding 30 ml of ice-cold 94:6 acetonitrile/acetic acid. The precipitated proteins were pelleted by centrifugation in a microcentrifuge for 2–5 min, and the samples were stored at 220°C until analyzed by HPLC. HPLC analysis. Reverse-phase HPLC was performed on a C 18 Alltech Econosil column, 3.2 3 100 mm, coupled to a Packard Radiomatic Flow-One model A500 radioisotope detector. Linear gradients ranging from 40:60:0.1 acetonitrile:water:acetic acid to 95:5:0.1 acetonitrile:water:acetic acid at a flow rate of 0.6 ml/min were used for the separation of the products derived after enzymatic catalysis according to the methods of Okita et al. (29) and Adas et al. (30). Metabolic rates were calculated using the percentage of metabolite area to the total product area, and were expressed as nmol/min/nmol of protein.

RESULTS

Expression and spectroscopic characterization of CYP4A2, CYP4A3, and CYP4A8. The genes for CYP4A2 and CYP4A8 were cloned from Wistar Kyoto rat kidney RNA by reverse transcription-polymerase chain reaction methods. The sequence of the CYP4A2 cDNA was in complete agreement with the known genomic sequence (23). The first 23 amino acids of these fatty acid hydroxylases were replaced by the corresponding 23 amino acids of the bovine sterol 17ahydroxylase with the modifications introduced by Barnes et al. to improve expression of mammalian P450 enzymes in E. coli (25). CYP4A2, CYP4A3, and CYP4A8 were then heterologously expressed in E. coli

65

FIG. 1. Spectroscopic properties of purified, recombinant CYP4A2 (—), -4A3 (- - -), and -4A8 (. . .). The absolute absorbance spectra are shown in (A) and the reduced CO-difference spectra in (B). Spectra were recorded at 25°C with a 0.5 mM solution of the protein in 25 mM Tris–HCl buffer (pH 7.5) containing 20% glycerol, 0.02% cholate, and 250 mM NaCl.

with a six-histidine tag on the carboxy terminus to simplify protein purification. All three enzymes were expressed at a level of 80 –120 nmol L 21, as judged by quantitation of the native P450 chromophore. The absolute spectra of CYP4A2, -4A3, and -4A8 (Fig. 1) show that all three proteins have Soret, a, and b bands at 418, 571, and 537 nm, respectively. The presence in CYP4A2 of a shoulder at around 390 nm indicates that this recombinant protein was isolated in a mixed spin state. After reduction with sodium dithionite and addition of CO, CYP4A2, and CYP4A3 developed a strong absorption band with a maximum at 452 nm, as expected for intact P450 enzymes (Fig. 1B) (27). A minor cytochrome P420 shoulder can be seen for CYP4A2. The CYP4A8 protein was obtained, after purification, as a 3:1 mixture of the inactive cytochrome P420 and active P450 forms. The CYP4A8 protein has not previously been obtained in purified, active form, although its expression in Spodoptera frugiperda cells was reported as this manuscript was submitted for publication (31). The spectroscopic binding constants for lauric, myristic, arachidonic, and oleic acids are summarized in Table I. All the difference spectra were Type I (32),

66

HOCH ET AL. TABLE I

Spectroscopic Binding Constants (K s) at Which Half of the Maximum Spectroscopic Change Is Observed upon Titration of the Enzymes with the Indicated Substrates a

Fatty acid Lauric Myristic Arachidonic Oleic

CYP4A1 K s (mM) 30 46 ND b ND

CYP4A2 K s (mM)

CYP4A3 K s (mM)

CYP4A11 K s (mM)

8 4 3 18

3 13 19 64

22 7 1 136

a Spectroscopic binding constants were measured as reported under Materials and Methods using a 1.5-mM enzyme solution and a 0.5–500 mM concentration of the fatty acid. The values are the averages of two independent measurements that differed by no more than 10%. b Not detectable (ND), no spin state change was observed upon titration with the indicated fatty acid.

indicating a net shift of the heme from the low to the high spin state. The spectral perturbation was characterized by a maximum at 390 nm, a trough with an absorbance minimum at 420 nm, and an apparent isosbestic point at 407 nm. No perturbation was observed when arachidonic and oleic acids were titrated into a CYP4A1 solution, suggesting that these fatty acids were bound very poorly or, when bound, did not induce a spin state change of the iron atom. The binding constants with palmitic acid could not be measured due to the low solubility of this fatty acid in the assay solution, and the CYP4A8 values could not be determined because of the high proportion of cytochrome P420 to P450 chromophore and the relatively low purified enzyme concentration (250 mM) that was available. In general, arachidonic acid appears to be the highest affinity ligand for ferric CYP4A2 and CYP4A11 but not CYP4A1 or CYP4A3 (Table I). Construction and expression of the CYP4A2_3 and CYP4A3_2 chimeras. The design of the chimeras was based on a sequence alignment of the rat and human CYP4A isoforms. The alignment was done with the GCG, Clustal V, and PIMA alignment programs (21, 33, 34). A total of 18 amino acid differences are distributed along the entire length of the CYP4A2 and CYP4A3 proteins. However, the region of highest variability is found between amino acid residues 113 and 119 (4A3 numbering), as CYP4A3 possesses an additional Ser-Gly-Ile sequence in this region (Fig. 2). In addition to this insert, two substitutions are found in this region. The amino acid differences between CYP4A2 and CYP4A3 are not identical if one uses our CYP4A2 sequence rather than that recently published by Helvig et al. (17). Our sequence does not have the Lys to Gln substitution at position 20, or the Ser for Pro substitution of residue 366; instead we find an additional difference in amino acid 163 with a Ser residue

in CYP4A2 and an Asn residue in CYP4A3. As our alignment matches the Pro-113 residue in CYP4A2 with the Ala-113 residue in CYP4A3, an additional Ser-Gly-Ile rather than Ala-Ser-Gly insert is inferred (Fig. 2). Of all the differences found between CYP4A2 and CYP4A3, the above mentioned region was thought the most likely to engender differences in catalytic activity. We therefore engineered a new restriction site into the nucleotides coding for Pro-119 (4A2) and Pro122 (4A3) (see Fig. 2). We then fused the oligonucleotide sequence coding for the first 119 amino acids of CYP4A2 to the segment of the CYP4A3 cDNA encoding the polypeptide that extends from residue 122 to the protein carboxy terminal. This fusion removes the SerGly-Ile insert of CYP4A3 from the CYP4A3_2 chimera. However, the CYP4A2_3 chimera, in which the first 122 amino acids of CYP4A3 are fused to the CYP4A2 region from amino acid residue 120 to the carboxy

FIG. 2. Alignment of the sequences of CYP4A2 and CYP4A3 obtained in this study. The concensus sequences is show at the bottom of each row and the amino acid differences between the two proteins are indicated in the gray boxes. The residue numbering scheme is given for CYP4A3. Amino acid deletions are indicated by dashes.

67

FATTY ACID SUBSTRATE SPECIFICITY OF CYP4A ENZYMES

FIG. 3. Concentration-dependent activity of CYP4A1 (h) and CYP4A2 (E) with lauric acid. The y-axis on the left side reflects the 4A1 activity and that on the right the 4A2 activity. Activities were measured as described under Materials and Methods.

terminal, retains this three amino acid insert. The chimeric enzymes were expressed exactly as described under Materials and Methods for the wild-type proteins. The expression levels for CYP4A3_2 were similar to the wild-type proteins, whereas the CYP4A2_3 chimera was expressed at the lower level of 40 nmol/L 21. After affinity chromatography on the Ni 21-NTA column, the ferric and ferrous CO-spectra of both chimeric proteins were identical to those of CYP4A2 (Fig. 1). Substrate specificity and regioselectivity of the wildtype enzymes. The published data on the substrate specificity of the five CYP4A enzymes investigated here was obtained by different laboratories using either purified or microsomal preparations of recombinant proteins expressed in E. coli, insect cells, or mammalian cells. Due to the diversity of the experimental systems employed, it is difficult to compare the activities from the various studies. To correct this shortcoming, we

have expressed and purified all four rat CYP4A isoforms as well as human CYP4A11 and have investigated their substrate specificities and regioselectivities with a common set of fatty acids under identical reaction conditions. The set of substrates investigated consisted of the fatty acids of 12, 14, and 16 carbons in chain length (C12–C16) as well as the physiologically active arachidonic (C20:4) and oleic (C18:1) acids. The plots obtained for the concentration-dependent activity of CYP4A1 and CYP4A2 with lauric acid as the substrate are shown in Fig. 3. Increasing activities were measured with increasing lauric acid concentrations up to a fatty acid concentration of 100 mM. Further increases in the concentration of lauric acid (100 – 400 mM) resulted in a gradual (CYP4A2) or drastic (CYP4A1) drop in activity. A similar effect was observed for all the enzymes and substrates investigated. This decrease in activity may be caused by substrate or product inhibition, by micellar aggregation of the fatty acids, or by physical effects of the fatty acids on the reconstituted enzyme. Long chain fatty acids can act as detergents and thereby may disturb the arrangement of the electron transfer partners in the reconstituted system. However, increasing the amount of phosphatidylcholine up to 50 mg did not prevent the drop in activity at high fatty acid concentrations. Table II summarizes the apparent k cat values and the substrate concentrations at which they were determined for the different CYP4A isoforms in incubations carried out at 37°C with 50 pmol of purified enzyme and a 1:10:1 ratio of CYP4A/cytochrome P450 reductase/cytochrome b 5. Analysis of the lauric acid reaction indicates that the conditions yield initial velocity values. However, the different solubilities of the various fatty acids required that the rates be measured with a

TABLE II

Apparent k cat Values and the Fatty Acid Concentration at Which They Were Determined for the Five CYP4A Isoforms a Fatty acid (mM) Lauric Myristic Palmitic Arachidonic Oleic

CYP4A1 min 21

CYP4A2 min 21

CYP4A3 min 21

CYP4A8 min 21

CYP4A11 min 21

649 (100) 230 (100) 60 (100) 6 (80) 1.4 (80)

35 (100) 40 (100) 6 (100) 2 (40) 0.5 (40)

73 (100) 70 (100) ND b

230 (100) 6 (100) 1.5 (100) 1 (40) 1.4 (40)

42 (100) 50 (100) 10 (100) 0.4 (20) 0.4 (20)

2 (80) 0.4 (80)

a Turnover numbers for the combined v- and v-1 hydroxylations were measured as reported under Materials and Methods using 50 pmol of enzyme. The values are the average of two independent measurements that differed by no more than 10%. The optimal concentration of the fatty acid (mM) used for the measurement is shown in parenthesis. The number in parenthesis is the substrate concentration at which maximum turnover was observed. The substrate concentrations ranged from 10 to 400 mM. b ND, not detectable.

68

HOCH ET AL. TABLE III

Comparison of CYP4A v: v-1 Regioselectivity for Different Fatty Acids a CYP4A1

CYP4A2

CYP4A8

CYP4A11

2.5:1 1.6:1 1.6:1 ND b

17:1 4:1 2.2:1 ND

v: v-1

Fatty acid Lauric Myristic Palmitic Arachidonic

CYP4A3

40:1 3:1 1:1 6:1

6:1 3:1 — 2.4:1

3:1 1.2:1 — 2:1

a Hydroxylated fatty acids have been separated by reversed-phase HPLC as described under Methods and Materials. The metabolites have been identified by injecting reference samples or by comparison to the relative elution times published in the literature. b ND, Not determined.

limited number of substrate concentrations and are therefore apparent k cat values. CYP4A1 and CYP4A8 show high activities of 649 and 230 min 21, respectively, with lauric acid as the substrate. Both hydroxylation rates are higher than the previously reported values of 150 min 21 (18) for CYP4A1 and 5.5 min 21 (9) for CYP4A3. The difference in the CYP4A3 activity is probably due to the fact that microsomal membranes of HepG2 cells were used for the earlier assay. In the case of CYP4A1 the difference most likely reflects differences in the reconstitution conditions, a finding that emphasizes the usefulness of a common set of data. CYP4A2, with a turnover number of 35 min 21 is the least active toward lauric acid. This value is slightly higher than the published value of 10 min 21 (15). The CYP4A11 rate of 42 min 21 agrees with a recently published value (19). Increasing the fatty acid chain length by two carbon atoms, as in myristic acid, decreased the hydroxylation rate for CYP4A1 and CYP4A8. The hydroxylation rate remained approximately the same for CYP4A2, CYP4A3, and CYP4A11 in going from lauric to myristic acid (Table II). As found for lauric acid, the highest overall activity with myristic acid was seen with CYP4A1 and CYP4A8, which oxidized myristic acid at rates of 230 and 100 min 21, respectively. A further two-carbon increase in the fatty acid chain length to palmitic acid again decreased the hydroxylase activity of CYP4A1 (18 min 21), CYP4A8 (1.5 min 21) and CYP4A11 (3 min 21). Low activity was observed with CYP4A2 and none with CYP4A3. Although the unsaturated fatty acids oleic and arachidonic acids are longer than palmitic acid, all five CYP4A enzymes catalyzed their hydroxylation. Turnover numbers between 0.4 and 6 min 21 for the sum of v- plus v-1 hydroxylation of arachidonic acid and 0.4 to 1.4 min 21 for oleic acid were obtained (Table II). The numbers obtained for arachidonic acid are higher than the previously reported values of 0.6 (15) and 0.89 min 21 (14). Under our conditions we did not observe

formation of the epoxyeicosatrienoic acid reported by the Schwartzman group (14, 31). Regioselectivity of the CYP4A isoforms. In addition to the substrate specificity of the CYP4A isoforms, we investigated the regioselectivity of the hydroxylation reactions (Table III). All the products were resolved by reverse phase HPLC using different acetonitrile/water/ acetic acid gradients. In all cases a baseline separation was obtained between the v- and (v-1)-hydroxylated products. The identification of the products was achieved by coinjection with authentic reference compounds and/or by comparison of the retention times with those published in the literature (6, 29, 35). The exception was oleic acid, the products of which were not separated by our chromatographic system. For all the enzymes, the highest regioselectivity was obtained with lauric acid. CYP4A1 showed the highest regioselectivity (40:1) for the intrinsically least reactive v-position (Table III), followed by CYP4A11 (17:1) and CYP4A2 (6:1). CYP4A3 and CYP4A8 were the least specific, with only a two- to threefold excess of the v-hydroxylated product. For all other fatty acids investigated the preference for the v-position decreased to the point that there was only a 1.6- to 6-fold excess of the v-hydroxylated metabolite. The least selective isoforms were CYP4A3 and CYP4A8, for which the v to v-1 ratio did not exceed 3 or 2.5, respectively. Regioselectivities of the CYP4A3_2 and CYP4A2_3 chimeras. CYP4A2 and CYP4A3 exhibited different activities and regioselectivities for the hydroxylation of lauric acid (35 vs 73 min 21, and 6:1 vs 3:1, respectively) and myristic acid (40 vs 70 min 21, and 3:1 vs 1.2:1, respectively) (Tables II and III). These pronounced differences allowed us to address the question of which amino acid(s) control the specificity and selectivity of these enzymes. The results show that the CYP4A3_2 chimera retained the regiospecificity of the CYP4A2 enzyme from which it derived the first 119 amino acids (Table IV). The CYP4A2_3 chimera was not catalytically active. This lack of activity appears to be due to a

69

FATTY ACID SUBSTRATE SPECIFICITY OF CYP4A ENZYMES TABLE IV

Substrate Specificity and Regioselectivity of the CYP4A3_2 Chimera

Enzyme

Rate of lauric acid hydroxylation (min 21)

Lauric acid v- to v-1 hydroxylation

Rate of myristic acid hydroxylation (min 21)

Myristic acid v- to v-1 hydroxylation

CYP4A2 CYP4A3 CYP4A3_2 CYP4A2_3

35 73 95 ND a

6:1 3:1 7:1 ND

40 70 52 ND

3:1 1.2:1 3:1 ND

a

ND, not detected.

perturbation of the electron transfer system, as a reduced CO-difference spectrum with a maximum at ;450 nm did not emerge when NADPH and lauric acid were added to a CO-saturated solution of the reconstituted enzyme (not shown). A ferrous-CO difference spectrum was obtained when the parent proteins were similarly treated. Helvig et al. (15) recently reported the generation of a CYP4A3/2 chimera, in which the first 182 amino acids of CYP4A2 were replaced by the corresponding amino acids from CYP4A3. This chimeric protein derived its regioselectivity for lauric acid from the CYP4A3 protein. Our results, in conjunction with theirs, suggest that the selectivity of the CYP4A enzymes is governed by the first 120 or so amino acids. DISCUSSION

Our data clearly shows that the four rat CYP4A enzymes and the human CYP4A11 have similar substrate specificities even though their absolute activities, particularly for the hydroxylation of lauric acid, vary widely. In particular, CYP4A1 and CYP4A8 have a much higher activity for lauric acid than the other enzymes (Table II). However, for all the enzymes the highest catalytic activity was obtained with lauric acid and elongation of the fatty acid chain decreased the activity. Most notable is the fact that none of the five enzymes exhibited a preference for longer chain fatty acids. In an earlier study in which the activities of CYP4A1 and CYP4A3 were measured after transfection into HepG2 cells, CYP4A1 was found to oxidize palmitic acid with a higher turnover number than lauric acid (15). This result conflicts with the data on the purified enzymes and is probably an artifact of the unoptimized membrane preparations used in the earlier study. Our results with CYP4A11 are consistent with the available data (21, 36). In a study published just before this paper was submitted it was reported that Spodoptera frugiperda cell membranes expressing CYP4A8 did not oxidize arachidonic acid (31), but our results show that the activity of the purified enzyme is low but detectable. The results with the rat and human enzymes contrast with the activities found for the rabbit CYP4A4, -4A5, -4A6, and -4A7 enzymes transfected

into COS-1 cells (37). All the rabbit enzymes except CYP4A5 were found to oxidize arachidonic acid, and the rate of arachidonic acid oxidation by CYP4A6 and CYP4A7 was found to be as high as 50% of that for lauric acid. The low rate of CYP4A5 for the oxidation of arachidonic acid was recently confirmed with the purified enzyme (38), but palmitic acid was found to be oxidized at a rate not much lower than that for oxidation of lauric acid (38, 39). A strong association between CYP4A levels and renal arachidonic acid v-hydroxylase with blood pressure suggests that the CYP4A isoforms are important in the formation of 20hydroxyeicosatetraenoic acid formation and the associated regulation of blood pressure (40, 41). In contrast, the finding that the four rat CYP4A enzymes and human CYP4A11 have a marginal activity for long chain fatty acid v-hydroxylation suggests that the physiologically important v-hydroxylation of arachidonic acid to 20-HETE (42– 44), and the oxidation of very long chain fatty acids in the epidermis and other tissues (45, 46), may be mediated, at least in part, by other enzymes. An alternative catalyst for arachidonic acid v-hydroxylation in human tissue is CYP4F2 (47). In the absence of a crystal structure for any mammalian P450 enzyme, including any CYP4A enzyme, inferences concerning the relationship between structure and activity must rest on homology models based on sequence alignment of the proteins of interest with the bacterial enzymes for which structures are available. The most reliable homology model for a CYP4A enzyme is that of CYP4A11 constructed by Chang and Loew (40). As there is high sequence identity between CYP4A2 and CYP4A3 with CYP4A11, we have used the CYP4A11 model to determine the approximate location of the 18 amino acid differences between CYP4A2 and CYP4A3 and to visualize the structures of the two chimeras investigated in this study. The first 119 amino acids of CYP4A2 are shown in green and the remainder of the amino acids from CYP4A3 in blue in the CYP4A11-based ribbon model of the CYP4A3_2 chimera (Fig. 4). In the CYP4A2_3 chimera, the green fragment is from CYP4A3 and the blue fragment from CYP4A2. The residues that differ between the two

70

HOCH ET AL.

FIG. 4. Computer model of the CYP4A3_2 chimera structure based on the homology model for CYP4A11 (40). The ribbon representing the backbone of the protein derived from CYP4A2 (residues 1–119) is in green. The blue strand in the model represents the backbone of the protein derived from CYP4A3. The side chains of the residues shown in red indicate the residues that differ between CYP4A2 and CYP4A3. The three yellow residues in the upper center are the three amino acids (Ser-Gly-Ile) that are present in CYP4A3 and CYP4A2_3 but not in CYP4A2 or CYP4A3_2. The heme group and Arg-96, the proposed binding site for the carboxylic acid group of the substrate, are shown in black.

proteins are shown in red as individual amino acids on the ribbon background. The amino acids at some distance from the heme in the blue fragment of the structure appear to have little or no effect on substrate specificity, in accord with the finding that substrate specificity is primarily governed by the amino acids in the green fragment. The most striking difference between CYP4A2 and CYP4A3 is the presence in CYP4A3 of a Ser-Gly-Ile insert that is shown in the model in yellow at the interface between the green and blue domains (Fig. 4). The Ser-Gly-Ile insert is not present in the CYP4A3_2 chimera, but is present in the CYP4A2_3 enzyme. As shown in Fig. 4, CYP4A11 has the three amino acid insert although its sequence is His-Gly-Ser. Site-specific mutagenesis studies will be required to determine the specific roles of these three amino acids, or the vicinal residues that differ, in determining the specificity of CYP4A2 and CYP4A3. In addition to the Ser-Gly-Ile insert in CYP4A3, six amino

acid residue differences are found in this region: Ser for Thr at position 7, Asn for Asp at position 63, Thr for Lys at position 91, Ala for Thr at position 92, Pro for Ala at position 113, and Ser for Phe at position 116 in CYP4A2. It is surprising that the CYP4A2_3 chimera is inactive, given that it folds and binds heme, gives a normal 450-nm spectrum in the reduced-CO state, and is closely related to the active CYP4A3_2 chimera. Furthermore, addition of lauric acid to the chimera gives a normal difference spectrum, indicating that the lauric acid binds, displaces the distal water ligand from the iron, and induces a normal low to high spin shift in the iron state. The inactivity of the protein is therefore not due to an inability to bind the substrate in a normal productive manner. The finding that CO binds and gives a normal 450-nm spectrum when the iron is reduced with dithionite clearly show that reduction of the iron and ligand binding can occur normally. However, efforts to convert the protein to the ferrous–CO complex by incubating the ferric CYP4A2_3 chimera with cytochrome P450 reductase and NADPH under an atmosphere of CO failed even though the CO complex could be formed in similar experiments with the parent proteins and the CYP4A3_2 chimera. These results suggest that the CYP4A2_3 chimera has a defect in its ability to interact with P450 reductase or in the ability of the reductase to transfer an electron to the iron. This difference must involve an interaction of amino acids between the CYP4A2 and CYP4A3 fragments in the chimera, as reduction occurs normally with both the parent proteins and the CYP4A3_2 chimera. ACKNOWLEDGMENTS We thank Angela Wilks and Elizabeth Dirks for help in making the constructs for expression of the CYP4A isoforms.

REFERENCES 1. Simpson, A. E. C. M. (1997) Gen. Pharmacol. 28, 351–359. 2. Kupfer, D. (1980) Pharmacol. Ther. 11, 469 – 496. 3. Kimura, S., Hanioka, N., Matsunaga, E., and Gonzalez, F. J. (1989) DNA Cell Biol. 8, 503–516. 4. Kimura, S., Hardwick, J. P., Kozak, C. A., and Gonzalez, F. J. (1989) DNA Cell Biol. 8, 517–525. 5. Stromstedt, M., Hayashi, S., Zaphiropoulos, P. G., and Gustafsson, J. A. (1990) DNA Cell Biol. 9, 569 –577. 6. Kroetz, D. L., Huse, L. M., Thresson, A., and Grillo, M. P. (1997) Mol. Pharmacol. 52, 362–372. 7. Palmer, C. N. A., Richardson, T. H., Griffin, K. J., Hsu, M.-H., Muerhoff, A. S., Clark, J. E., and Johnson, E. F. (1993) Biochim. Biophys. Acta 1172, 161–166. 8. Nhamburo, P. T., Gonzalez, F. J., McBride, O. W., Gelboin, H. V., and Kimura, S. (1989) Biochemistry 28, 8060 – 8066. 9. Kikuta, Y., Kusunose, E., Kondo, T., Yamamoto, S., Kinosha, H., and Kusunose, M. (1994) FEBS Lett. 348, 70 –74.

FATTY ACID SUBSTRATE SPECIFICITY OF CYP4A ENZYMES 10. Imaoka, S., Ogawa, H., Kimura, S., and Gonzalez, F. J. (1993) DNA Cell Biol. 12, 893– 899. 11. Ayoma, T., Hardwick, J. P., Imaoka, S., Funae, Y., Gelboin, H. V., and Gonzalez, F. J. (1990) J. Lipid Res. 31, 1477–1482. 12. Alterman, M. A., Chaurasia, C. S., Lu, P., Hardwick, J. P., and Hanzlik, R. P. (1995) Arch. Biochem. Biophys. 320, 289 –296. 13. Bambal, R. B., and Hanzlik, R. P. (1996) Biochim. Biophys. Res. Commun. 219, 445– 449. 14. Lu, P., Alterman, M. A., Chaurasia, C. S., Bambal, R. B., and Hanzlik, R. P. (1997) Arch. Biochem. Biophys. 337, 1–7. 15. Aoyama, T., Hardwick, J. P., Imaoka, S., Funae, Y., Gelboin, H. V., and Gonzalez, F. J. (1990) J. Lip. Res. 31, 1477–1482. 16. Wang, M.-H., Stec, D. E., Balazy, M., Mastyugin, V., Yang, C. S., Roman, R. J., and Schwartzman, M. L. (1996) Arch. Biochem. Biophys. 336, 240 –250. 17. Helvig, C., Dishman, E., and Capdevila, J. H. (1998) Biochemistry 37, 12546 –12558. 18. Imaoka, S., Tanaka, S., and Funae, Y. (1989) Biochem. Int. 18, 731–740. 19. Wang, M.-H., Guan, H., Nguyen, X., Zand, B. A., Nasjletti, A., and Laniado-Schwartzman, M. (1999) Am. J. Physiol. 276, F246 –F253. 20. Dierks, E. A., Davis, C. S., and Ortiz de Montellano, P. R. (1998) Biochemistry 37, 1839 –1847. 21. Dierks, E. A., Zhang, Z., Johnson, E. F., and Ortiz de Montellano, P. R. (1998) J. Biol. Chem. 273, 23055–23061. 22. Muchmore, D. C., McIntosh, L. P., Anderson, D. E., and Dalqhuist, F. W. (1989) Methods Enzymol. 177, 44 –73. 23. Kimura, S., Hanioka, N., Matsunaga, E., and Gonzalez, F. J. (1989) DNA 8, 503–516. 24. Stromstedt, M., Hayashi, S., Zaphiropoulos, P. G., and Gustafsson, J. A. (1990) DNA Cell Biol. 9, 569 –577. 25. Barnes, H. J., Arlotto, M. P., and Waterman, M. R. (1991) Proc. Natl. Acad. Sci. USA 88, 5597–5601. 26. Chen, G.-Q., Choi, I., Ramachandran, B., and Gouaux, J. E. (1994) J. Am. Chem. Soc. 116, 8799 – 8800. 27. Omura, T., and Sato, R. (1964) J. Biol. Chem. 239, 2370 –2378. 28. Ortiz de Montellano, P. R., and Reich, N. O. (1984) J. Biol. Chem. 259, 4136 – 4141. 29. Okita, R. T., Clark, J. E., Okita, J. R., and Masters, B. S. S. (1991) Methods Enzymol. 206, 432– 441.

71

30. Adas, F., Berthou, F., Picart, D., Lozac’h, P., Beauge, F., and Amet, Y. (1998) J. Lipid Res. 39, 1210 –1219. 31. Nguyen, X., Wang, M.-H., Reddy, K. M., Falck, J. R., and Schwartzman, M. L. (1999) Am. J. Physiol. 276, R1691– R1700. 32. Estabrook, R. W., Peterson, J., Baron, J., and Hildebrandt, A. (1972) In Methods in Pharmacology (Chignell, C. F., Ed.), Vol. 2, pp. 303–350, Appleton-Century Crofts, NY. 33. Higgins, D. G., Bleasby, A. J., and Fuchs, R. (1992) Comput. Appl. Biosci. 8, 189 –191. 34. Smith, R. F., and Smith, T. F. (1992) Protein Eng. 5, 35– 41. 35. Kroetz, D. L., Huse, L. M., Thuresson, A., and Grillo, M. P. (1997) Mol. Pharmacol. 52, 362–372. 36. Palmer, C. N. A., Richardson, T. H., Griffin, K. J., Hsu, M.-H., Muerhoff, A. S., Clark, J. E., and Johnson, E. F. (1993) Biochim. Biophys. Acta 1172, 161–166. 37. Roman, L. J., Palmer, C. N. A., Clark, J. E., Muerhoff, A. S., Griffin, K. J., Johnson, E. F., and Masters, B. S. S. (1993) Arch. Biochem. Biophys. 307, 57– 65. 38. Hosny, G., Roman, L. J., Mostafa, H. M., and Masters, B. S. S. (1999) Arch. Biochem. Biophys. 366, 199 –206. 39. Sawamura, A., Kusunose, E., Satouchi, K., and Kusunose, M. (1993) Biochim. Biophys. Acta 1168, 30 –36. 40. Chang, Y.-T., and Loew, G. H. (1999) Proteins 34, 403– 415. 41. Su, P., Kaushal, K., Kroetz, D. (1998) Am. J. Physiol. 275, R426 –R438. 42. Capdevila, J. H., Zeldin, D., Makita, K., Karara, A., and Falck, J. R. (1995) In Cytochrome P450: Structure, Mechanism, and Biochemistry (Ortiz de Montellano, P. R., Ed.) pp. 443– 471, 2nd ed., Plenum Press, NY. 43. Harder, D. R., Lange, A. R., Gebremedhin, D., Birks, E. K., and Roman, R. J. (1996) J. Vasc. Res. 34, 237–243. 44. McGiff, J. C., Quilley, C. P., and Carroll, M. A. (1993) Steroids 58, 573–579. 45. Wertz, P. W., and Downing, D. T. (1987) Biochim. Biophys. Acta 917, 108 –111. 46. Alexander, J. J., Snyder, A., and Tonsgard, J. H. (1998) Neurochem. Res. 23, 227–233. 47. Powell, P. K., Wolf, I., Jin, R., and Lasker, J. M. (1998) J. Pharmacol. Exp. Therap. 285, 1327–1336.