Identification of CYP4F8 in Human Seminal Vesicles as a Prominent 19-Hydroxylase of Prostaglandin Endoperoxides*

A novel cytochrome P450, CYP4F8, was recently cloned from human seminal vesicles. CYP4F8 was expressed in yeast. Recombinant CYP4F8 oxygenated arachidonic acid to (18R)-hydroxyarachidonate, whereas prostaglandin (PG) D2, PGE1, PGE2, PGF2α, and leukotriene B4 appeared to be poor substrates. Three stable PGH2 analogues, 9,11-epoxymethano-PGH2 (U-44069), 11,9-epoxymethano-PGH2 (U-46619), and 9,11-diazo-15-deoxy-PGH2 (U-51605) were rapidly metabolized by ω2- and ω3-hydroxylation. U-44069 was oxygenated with aV max of ∼260 pmol min− 1 pmol P450− 1 and a K m of ∼7 μm. PGH2decomposes mainly to PGE2 in buffer and to PGF2α by reduction with SnCl2. CYP4F8 metabolized PGH2 to 19-hydroxy-PGH2, which decomposed to 19-hydroxy-PGE2 in buffer and could be reduced to 19-hydroxy-PGF2α with SnCl2. 18-Hydroxy metabolites were also formed (∼17%). PGH1 was metabolized to 19- and 18-hydroxy-PGH1 in the same way. Microsomes of human seminal vesicles oxygenated arachidonate, U-44069, U-46619, U-51605, and PGH2, similar to CYP4F8. (19R)-Hydroxy-PGE1 and (19R)-hydroxy-PGE2 are the main prostaglandins of human seminal fluid. We propose that they are formed by CYP4F8-catalyzed ω2-hydroxylation of PGH1 and PGH2 in the seminal vesicles and isomerization to (19R)-hydroxy-PGE by PGE synthase. CYP4F8 is the first described hydroxylase with specificity and catalytic competence for prostaglandin endoperoxides.

Expression of CYP4F8 -The coding region of CYP4F8 was amplified with a sense primer (5Ј-TTGGGATCCAAAATGTCGCTGCTGAGC-CTGTCTTG-3Ј) that contained a BamHI linker sequence, three A residues to increase the expression efficiency (21), and 23 base pairs of the translation start site of CYP4F8 cDNA. The antisense primer (5Ј-CT-GGAATTCTCAGCCCAGGGGTTCTACTCGC-3Ј) contained 19 base pairs of the end of the coding sequence, the stop codon, and an EcoRI linker sequence. Polymerase chain reaction was performed with Pfu polymerase using a full-length cDNA clone of CYP4F8 (17) as template with an annealing temperature of 60°C. The polymerase chain reaction product was subcloned into the V60 yeast expression vector at the BamHI and EcoRI sites. The expression of CYP4F8 was carried out with the S. cerevisiae strain W(R), which has been genetically modified to also overexpress the yeast reductase (20). A galactose-inducible promoter in the plasmid and in the yeast genome, respectively, was used to control the expression. After transformation of the plasmid into the W(R) yeast strain by a lithium acetate method, selection of clones was achieved by growing the yeast in adenine-and uracil-deficient medium. To achieve higher expression levels, the yeast cells were first grown to high density with glucose as the main energy source; thereafter, galactose was added to induce expression. In brief, the procedure was as follows. Transformed yeast cells were grown to a density of ϳ30 ϫ 10 6 cells/ml in SGI medium (containing per liter: casamino acids 1 g, yeast nitrogen base 7 g, glucose 20 g, tryptophan 20 mg). The cells were then diluted to ϳ2.5 ϫ 10 6 cells/ml and grown for 22-24 h in YPGE medium (containing per liter: yeast extract 10 g, bactopeptone 10 g, glucose 5 g, ethanol 16 g). Galactose (2%) was added, and the cells were harvested 16 h later by centrifugation (22). The cell walls were disrupted with glass beads, and the microsomal fraction was obtained by differential centrifugations at ϩ4°C (20,000 ϫ g for 10 min and 100,000 ϫ g for 60 min). The pelleted microsomes were homogenized in 0.05 M Tris-HCl, 20% glycerol, and 1 mM EDTA (pH 7.4) and stored at Ϫ80°C. Cytochrome P450 and cytochrome P450 reductase were measured as described (22). Control microsomes were prepared from yeast transfected with the V60 plasmid without an insert. Protein was determined as described (22,23).
Experimentation-Yeast microsomes (50 g, ϳ2 pmol of CYP4F8) or microsomes of human seminal vesicles (1-2 mg/ml) were incubated with 1 mM NADPH and 5-200 M substrate in a total volume of 100 l of 0.1 M KHPO 4 and 2 mM EDTA (pH 7.4) for 45 s to 30 min at 37°C. For GC-MS analysis, 25 pmol of CYP4F8 were used. Substrates were added in Ͻ1 l of solvent. The reactions were terminated with 4 volumes of ethanol. Incubations with PGH 2 were also terminated by addition of buffered SnCl 2 (18). The metabolites were extracted on a Sep-Pak C 18 column with ϳ90% recovery (24).
Kinetics-CYP4F8 (3 pmol in 0.3 ml) was preincubated with 5, 10, 50, and 200 M U-44069 in triplicate for 2 min at 37°C. NADPH (1 mM) was added, and the reaction was stopped after 45 s. The biosynthesis of hydroxy metabolites of U-44069 was quantified by LC-MS analysis using selective ion monitoring of the carboxylate anion (m/z 365). 19-Hydroxy-PGB 2 was added as an internal standard in some experiments. Standard curves were constructed from known amounts of U-44069 (selective ion monitoring of the carboxylate anion at m/z 349). We assumed that U-44069 and its hydroxy metabolites yielded the same response, as the increase in the hydroxy metabolites corresponded to the decrease in U-44069. V max and K m were calculated according to the Lineweaver-Burk plot. The rate of biosynthesis of 19-and 18hydroxy metabolites of endoperoxide analogues, PGD 2 , PGE 1 , PGE 2 , PGF 2␣ , and PGH 2 by CYP4F8 (2 pmol; 1-5 min at 37°C) was estimated at a fixed substrate concentration (10 M). Metabolite formation was estimated by LC-MS using standard curves of parent compounds or by percent conversion of substrate.
LC-MS Analysis-Equipment for LC-MS analysis was as described (24). The column contained octadecasilane silica (5 m, 250 ϫ 2 mm; Chromasil 5 C 18 100 A, Phenomenex Inc., Torrance, CA) and was eluted at 0.2 ml/min. The mobile phase was CH 3    effluent first passed a UV detector and was then subjected to negative ion electrospray ionization in an ion trap mass spectrometer (LCQ, ThermoQuest, San Jose, CA). The source voltage was 2.4 kV; the capillary temperature was 230°C; and the collision energy was set to 30%. PGF 1␣ was used for tuning. Products were assayed by selective ion monitoring of the carboxylate anions and identified by MS/MS analysis. These mass parameters were adjusted to the mass of the carboxylate anions of the various compounds as they eluted from the HPLC column.
GC-MS Analysis-Methyl esters and trimethylsilyl ethers were prepared as described (10). PGE compounds were dehydrated to PGB compounds with 0.25 M KOH for 20 min (5). A gas chromatograph (Varian 3100) with a nonpolar capillary column (30 m; DB-5, J & W Scientific; film, 0.25 m; diameter, 0.25 mm) was connected to an ion trap mass spectrometer (ITS40, Finnigan MAT). The gas chromatograph was programmed from 120 to 270°C in 5 min and from 270 to 294°C in 8 min and then kept at 294°C. C-values (number of apparent carbons) were determined from the retention times of fatty acid methyl esters (10,25). Steric analysis of 18-HETE methyl esters was performed after hydrogenation by analysis of (2S)-phenylpropionic acid derivatives of methyl 18-hydroxyeicosanoates (10).

Expression of CYP4F8 in Yeast-
The yield of CYP4F8 was 26 -62 pmol/mg of microsomal proteins, and the yield of cytochrome P450 reductase was Ͼ10-fold higher (430 -850 pmol/ mg). The reduced CYP4F8⅐CO complex had an absorption max-imum at 449.6 nm. Control microsomes did not contain appreciable amounts of cytochrome P450 and did not metabolize any of the substrates discussed below.
Hydroxylation of Fatty Acids-Recombinant CYP4F8 and NADPH metabolized arachidonic acid to one major metabolite (Fig. 1A). This metabolite was identified as 18-HETE by LC-MS analysis. The MS/MS analysis of the carboxylate anion (m/z 319 3 full scan) yielded a mass spectrum that was identical to that of authentic 18-HETE. Characteristic signals were noted at m/z 261 (319-58, loss of CH 3 CH 2 CHO) and 217 (261-44, loss of CO 2 ) (24). Steric analysis by GC-MS showed that the metabolite consisted of Ͼ95% of the 18R-stereoisomer (Fig. 1B). The mass spectra, which were obtained at C-values of ϳ29.0 and ϳ29.5 for the S-and R-stereoisomers, respectively, were identical to previous data (10). CYP4F8 was also incubated with linoleic acid. LC-MS analysis (MS/MS 295 3 full scan) showed that the major metabolite had the same mass spectrum as that of 16-hydroxylinoleic acid (24). Arachidonic and linoleic acids were thus oxygenated to the very same metabolites formed by monkey seminal vesicle microsomes (10, 26). A, CYP4F8 was incubated with PGH 2 . PGH 2 and its metabolites were then allowed to decompose in buffer, and formation of hydroxy-PGs and PGs was analyzed by LC-MS. MS/MS analysis suggested that peak I contained 19-hydroxy-PGF 2␣ , that peak II contained 19-hydroxy-PGE 2 (with trace amounts of 19hydroxy-PGD 2 on the right shoulder), and that peak III contained 18-hydroxy-PGE 2 . PGF 2␣ , PGE 2 , and PGD 2 eluted as marked. Hydroxylation of PGs and Leukotriene B 4 -Recombinant CYP4F8 was incubated with PGD 2 , PGE 1 , PGE 2 , PGF 2␣ , and leukotriene B 4 , and the formation of hydroxylated metabolites was analyzed by LC-MS with MS/MS. These compounds were all poor substrates, and metabolites could not be identified in many experiments (Table I). However, using a high substrate concentration (100 M) and long incubation times, small amounts of 19-hydroxy metabolites of PGE 2 and PGF 2␣ could be identified by LC-MS analysis and by comparison with authentic standards. Cytochrome b 5 at the same concentration as CYP4F8 did not augment the oxygenation of PGE 1 and PGE 2 .
Hydroxylation of PGH 2 Analogues-CYP4F8 was found by LC-MS to metabolize U-44069 to two hydroxy metabolites at a ratio of ϳ6:1 ( Fig. 2A). LC-MS analysis was consistent with 19and 18-hydroxy metabolites. GC-MS analysis of the trimethylsilyl ether methyl ester derivative of the main metabolite showed a mass spectrum at a C-value of 25.1 with a strong signal at m/z 117 (H 3 C-CH-O ϩ -Si(CH 3 ) 3 ). The C-value of the trimethylsilyl ether methyl ester derivative of U-44069 was 22.8. The difference in the C-values of the metabolite and the parent compound and the mass spectrum of the metabolite suggested that it was 19-hydroxy-(9,11-epoxymethano)-PGH 2 (25). The minor metabolite showed a mass spectrum with a strong signal at m/z 131 (H 3 C-CH 2 -CH-O ϩ -Si(CH 3 ) 3 ) at a Cvalue of 24.9 and was identified as 18-hydroxy-(9,11-epoxymethano)-PGH 2 (25).
The hydroxylation of U-44069 at 37°C was linear with time for at least 6 min and linear with protein (0.1-1.6 mg/ml). Cytochrome b 5 was without effect. K m and V max values were estimated by a Lineweaver-Burk plot. The apparent K m was ϳ7 M, and the V max ϳ260 pmol min Ϫ1 pmol P450 Ϫ1 (Fig. 2B).
CYP4F8 also efficiently metabolized U-46619 and U-51605 (Table I). LC-MS analysis suggested that the main products of U-46619 were 19-and 18-hydroxy metabolites (Table II). The structures were confirmed by GC-MS analysis (trimethylsilyl ether methyl ester derivatives). The mass spectrum of the main metabolite showed a characteristic signal at m/z 117 and had a C-value of 25.6. The mass spectrum of the other metabolite showed a characteristic signal at m/z 131 and had a slightly smaller C-value of 25.4, whereas the C-value of U-46619 was 23.3. The two metabolites were thus identified as 19-hydroxy-(11,9-epoxymethano)-PGH 2 and 18-hydroxy-(11,9-epoxymethano)-PGH 2 (25). The latter was partially separated by HPLC into two equal components with identical MS/MS spectra, conceivably stereoisomers at C-18, but this was not further investigated. In U-51605, the endoperoxide oxygens are replaced by a -NϭN-group, and the hydroxy group at C-15 has been deleted. LC-MS with triple MS (m/z 347 3 m/z 319 3 full scan) suggested that CYP4F8 metabolized U-51605 to 19and 18-hydroxy metabolites (Table II). U-51605 and its 19-and 18-hydroxy metabolites were too unstable for GC-MS analysis.
Hydroxylation of PGH 1 and PGH 2 -Recombinant CYP4F8 was incubated with 10 M PGH 2 and 1 mM NADPH at 37°C for 10 min, and the products were analyzed by LC-MS (Fig. 3). Under these conditions, PGH 2 will decompose in buffer to PGD 2 , PGE 2 , and PGF 2␣ as shown in Fig. 3 and to (12S)hydroxyheptadecatrienoic acid (data not shown) (18). Hydroxy metabolites of PGH 2 will decompose to hydroxy metabolites of PGs in the same way. The first eluting product (peak I) was identified as 19-hydroxy-PGF 2␣ by its MS/MS spectrum (MS/MS 369 3 full scan). Peak II mainly contained 19-hydroxy-PGE 2 (MS/MS 367 3 full scan), but trace amounts of 19-hydroxy-PGD 2 eluted on the right shoulder of the peak as judged by MS/MS analysis (due to the relative intensity at m/z 251). The formation of 19-hydroxy-PGE 2 was confirmed by GC-MS analysis as described below. 18-Hydroxy-PGE 2 was identified in peak III by MS/MS analysis, which showed a characteristic signal at m/z 273 (loss of 2H 2 O and CH 3 CH 2 CHO). PGF 2␣ , PGE 2 , and PGD 2 eluted after 18, 24, and 29 min, respectively.
The structure of the main hydroxy metabolite of PGH 2 was confirmed by GC-MS analysis. PGE 2 compounds that were formed during incubation with PGH 2 and CYP4F8 were converted to PGB 2 compounds by alkali treatment. A polar PGB 2 metabolite was purified by HPLC and analyzed by GC-MS (trimethylsilyl ether methyl ester derivative). The mass spectrum and the C-value of 26.1 was identical to those reported for 19-hydroxy-PGB 2 (4,5).
An experiment with chemical reduction of products confirmed that PGH 2 was metabolized to 19-hydroxy-PGH 2 . PGH 2 was incubated with CYP4F8 for 2 min, and one-half of the incubation was terminated with buffered SnCl 2 , whereas the other half was terminated with ethanol. LC-MS analysis of the sample reduced with SnCl 2 showed two major peaks (Fig. 4A). 19-Hydroxy-PGF 2␣ was present in peak I, and 19-hydroxy-PGE 2 and 18-hydroxy-PGF 2␣ were present in peak II. In contrast, the second sample yielded 19-hydroxy-PGE 2 as the main product due to decomposition of 19-hydroxy-PGH 2 in aqueous ethanol (Fig. 4B). Small amounts of 18-hydroxy-PGE 2 were present in peak III in both chromatograms (Fig. 4B). PGH 1 was metabolized by CYP4F8 in the same way as PGH 2 . The main metabolites decomposed in buffer to 19-hy- FIG. 4. Identification of 18-and 19hydroxy-PGH 2 as CYP4F8 metabolites of PGH 2 by reduction to 18-and 19-hydroxy-PGF 2␣ . One-half of the incubation was terminated with buffered SnCl 2 , and the other half with ethanol. The metabolites were analyzed by selective ion monitoring of the interval m/z 367-369 with MS/MS analysis of hydroxy-PGE 2 (MS/MS 367 3 full scan) and hydroxy-PGF 2␣ (MS/MS 369 3 full scan). A, products formed after termination with SnCl 2 ; B, products formed after termination with ethanol. The LC-MS analysis suggested that peak I contained 19-hydroxy-PGF 2␣ , that peak II contained mainly 19-hydroxy-PGE 2 (and 18-hydroxy-PGF 2␣ in A), and that peak III contained 18-hydroxy-PGE 2 . The mobile phase was CH 3 CN/H 2 O/acetic acid (35:65:0.01). droxy-PGE 1 and 18-hydroxy-PGE 1 ( Table II).
As shown in Table I, U-44069 and U-51605 appeared to be the best substrates of CYP4F8. PGH 2 was metabolized at a lower rate. However, PGH 2 is unstable, and our in vitro conditions may not mimic in vivo biosynthesis. It is noteworthy that PGH 2 and the stable endoperoxide analogues were metabolized more efficiently than PGE 2 , PGF 2␣ , PGD 2 , and PGE 1 .
Human Seminal Fluid-18-Hydroxy-PGE compounds in pooled seminal fluid (n ϭ 12) were analyzed by LC-MS. 18-Hydroxy-PGE compounds were present in slightly lower rela-tive amounts in seminal fluid compared with those formed by hydroxylation of PGH 1 and PGH 2 by CYP4F8 in vitro (Table  II). 18-HETE could not be detected in these samples by LC-MS.
Human Seminal Vesicles-Microsomes of seminal vesicles (n ϭ 3) and NADPH metabolized U-44069 to 19-and 18-hydroxy metabolites at a ratio of ϳ6:1 as judged by LC-MS analysis (Fig. 5A). U-44069, U-46619, and U-51605 were metabolized to 19-and 18-hydroxy metabolites in the same relative amounts as by CYP4F8 (Table II). To determine whether PGH 2 was metabolized to 19-hydroxy-PGH 2 by microsomes of seminal vesicles, we repeated the experiments described above with SnCl 2 . Treatment with buffered SnCl 2 yielded significant formation of 19-hydroxy-PGF 2␣ , whereas 19-hydroxy-PGE 2 was the main product formed in aqueous ethanol (Fig. 5B). Small amounts of 18-hydroxy-PGE 2 were detected in some experiments. The yield of 19-hydroxy metabolites of PGH 2 was rather poor. Microsomal fractions of seminal vesicles may contain PGE synthase and other PGH 2 -metabolizing enzymes. For example, PGH 2 was also transformed to 6-keto-PGF 1␣ in some experiments. As previously reported, microsomes of human seminal vesicles only slowly convert PGE 1 and PGE 2 to their 19-hydroxy metabolites (4). Attempts to compare the biosynthesis of 19-hydroxy-PGE 2 from PGE 2 with the biosynthesis of 19-hydroxy-PGH 2 from PGH 2 were unsuccessful due to the low metabolism of PGE 2 .
Arachidonic acid was metabolized to 18-HETE by CYP4F8 as described above, and we confirmed that microsomes of human seminal vesicles and NADPH also formed this metabolite (Fig.  5C). Small amounts of 15-HETE (retention time of 23 min) and 11-HETE (retention time of 25 min) were also identified by LC-MS, but these HETEs can also be formed by PGH synthase of seminal vesicles (27). DISCUSSION mRNA of a new enzyme, CYP4F8, was recently discovered in human seminal vesicles (17). We have expressed CYP4F8 in yeast and report as our main finding that recombinant CYP4F8 metabolizes PGH 1 and PGH 2 to 19-hydroxy-PGH 1 and 19-hydroxy-PGH 2 , respectively. Three stable PGH 2 analogues were also substrates, whereas PGD 2 , PGE 1 , PGE 2 , and PGF 2␣ were metabolized poorly (Table I). Our results suggest that PGH 1 and PGH 2 are endogenous substrates for CYP4F8, which thus can be named PGH 19-hydroxylase. CYP4F8 is the first described hydroxylase with pronounced specificity for prostaglandin endoperoxides. 9,11-Epoxymethano-PGH 2 (U-44069) was the best substrate of recombinant CYP4F8. U-44069 was converted to 19-and 18-hydroxy metabolites at a ratio of ϳ6:1, with a K m of ϳ7 M and a V max of ϳ260 pmol min Ϫ1 pmol P450 Ϫ1 . This V max value is remarkably high for a mammalian cytochrome P450 (20,22). 9,11-Diazo-15-deoxy-PGH 2 (U-51605), PGH 1 , PGH 2 , and 11,9epoxymethano-PGH 2 (U-46619) were also good substrates, although there was a remarkable difference between U-44069 and U-46619 (Table I). PGH 1 and PGH 2 and their hydroxy metabolites decompose in buffer with a half-life of ϳ5 min at 37°C (18), which decreases the metabolism by CYP4F8 and complicates the analysis of the products.
To address the question of whether CYP4F8 is present in human seminal vesicles, we compared the catalytic properties of microsomes and recombinant CYP4F8. Both metabolized U-44069, U-46619, U-51605, and arachidonic acid to virtually the same profile of 19-and 18-hydroxy metabolites (Table II). Both preparations also oxygenated PGH 2 to 19-hydroxy metabolites and to small amounts of 18-hydroxy metabolites. These observations suggest that CYP4F8 is present in seminal vesicles. Human and primate semen contains Ͼ98% of the 19Rstereoisomers of 19-hydroxy-PGs (1)(2)(3)(4). It remains to be determined whether CYP4F8 oxidizes PGH 2 to the 19R-stereoisomer of 19-hydroxy-PGH 2 . However, we confirmed that CYP4F8 metabolizes arachidonate to Ͼ95% of the 18R-stereoisomer of 18-HETE, which is the stereoisomer formed by microsomes of primate seminal vesicles (10).
PGH synthase-1 is an integral membrane protein with its catalytic domains on the luminal side of the endoplasmic reticulum (28). The substrate channel of PGH synthase-1 faces the endoplasmic reticulum. Its entrance is surrounded by the membrane-binding surface, which does not extend beyond one leaflet of the lipid bilayer. Human cytochromes P450 are also integral membrane proteins, but their catalytic domains are located on the cytosolic side of the endoplasmic reticulum (29). PGH 2 must cross the lipid bilayer to gain access to CYP4F8. It is known that an important PGE synthase is also associated with the same membrane system (30). PGE synthase might efficiently metabolize PGH compounds, as non-enzymatic breakdown products, e.g. PGD compounds, have not been detected in human seminal fluid. The rapid biosynthesis of 19hydroxy-PGE 1 and 19-hydroxy-PGE 2 in vivo suggests that PGH synthase, PGH 19-hydroxylase, and PGE synthase are closely linked (1,2,4). A proposed mechanism for biosynthesis of (19R)-hydroxy-PGE 2 in human seminal vesicles is shown in Fig. 6.
There is a marked inter-individual variation in the amount of PGE and 19-hydroxy-PGE compounds in normal human seminal fluid (8,9). This may be due to variable expression of CYP4F8 or due to isoforms with reduced enzyme activity. CYP4F8 and CYP4F3 are located on the CYP4F cluster on chromosome 19p13.1 (17). CYP4F3 catalyzes -hydroxylation of leukotriene B 4 and was recently found to consist of distinct isoforms that were regulated by alternative promoter usage and exon splicing (31,32). Whether isoforms of CYP4F8 exist is now under investigation.
Two thiolate hemoproteins, thromboxane synthase (CYP5A) of platelets and prostacyclin synthase (CYP8A) of the vascular endothelium, can rearrange PGH 2 into thromboxane A 2 and prostacyclin, respectively (33,34). These enzymes do not require NADPH and differ from CYP4F8 in this respect. It is possible that PGH 2 might be a substrate of other cytochromes P450. Recombinant CYP4A11 and microsomes from human kidneys hydroxylate U-44069 at the -side chain. 2 The NADPHdependent metabolism of PGH 2 in the kidney and other tissues merits further investigation.
In summary, our results suggest that CYP4F8 of human seminal vesicles catalyzes 2-hydroxylation of PGH 1 and PGH 2 , which will lead to biosynthesis of the two main PGs of human seminal fluid, (19R)-hydroxy-PGE 1 and (19R)-hydroxy-PGE 2 . CYP4F8 is the first described enzyme with both specificity and kinetic competence for hydroxylation of PGH 1 and PGH 2 .