Metabolomic analysis and identification of a role for the orphan human cytochrome P450 2W1 in selective oxidation of lysophospholipids.

Human cytochrome P450 (P450) 2W1 is still considered an “orphan” because its physiological function is not characterized. To identify its substrate specificity, the purified recombinant enzyme was incubated with colorectal cancer extracts for untargeted substrate searches using an LC/MS-based metabolomic and isotopic labeling approach. In addition to previously reported fatty acids, oleyl (18:1) lysophosphatidylcholine (LPC, lysolecithin) was identified as a substrate for P450 2W1. Other human P450 enzymes tested showed little activity with 18:1 LPC. In addition to the LPCs, P450 2W1 acted on a series of other lysophospholipids, including lysophosphatidylinositol, lysophosphatidylserine, lysophosphatidylglycerol, lysophosphatidylethanolamine, and lysophosphatidic acid but not diacylphospholipids. P450 2W1 utilized sn-1 18:1 LPC as a substrate much more efficiently than the sn-2 isomer; we conclude that the sn-1 isomers of lysophospholipids are preferred substrates. Chiral analysis was performed on the 18:1 epoxidation products and showed enantio-selectivity for formation of (9R,10S) over (9S,10R). The kinetics and position specificities of P450 2W1-catalyzed oxygenation of lysophospholipids (16:0 LPC and 18:1 LPC) and fatty acids (C16:0 and C18:1) were also determined. Epoxidation and hydroxylation of 18:1 LPC are considerably more efficient than for the C18:1 free fatty acid.

task in modern biochemistry is the elucidation of protein functions, including the establishment of the catalytic activities of novel enzymes with unknown substrates ( 1,2 ). P450 enzymes play important roles in the metabolism of a large number of compounds, including sterols, fatty acids, eicosanoids, vitamins, and xenobiotics ( 3 ). It has been estimated that P450 reactions are involved in ‫ف‬ 75% of the enzymatic transformations of small molecule drugs ( 4,5 ). There are 57 human P450 genes identifi ed in the human genome, and about one fourth of them can be termed "orphans" because of their unknown physiological or other functions ( 5,6 ).
Human P450 2W1 is considered one of the orphan P450 enzymes. It is preferentially expressed in colorectal cancer tissue ( 7,8 ), and it is regulated by gene methylation and reverse membrane orientation ( 8,9 ). Moreover, expression of the P450 2W1 variant allele G541A (Ala181Thr) in tumors has been reported to be associated with lower survival rates ( 10 ). Interestingly, P450 2W1 expression is seen in colon, ileum, and testes in mice ( 11 ), but more sensitive searches in corresponding human tissues have not been reported. Some P450 2W1-catalyzed reactions have been identifi ed, including N -demethylation of benzphetamine ( 12 ), reduction of The enzyme reactions were initiated by the addition of an NADPHgenerating system including 100 l of 100 mM glucose 6-phosphate, 50 l of 10 mM NADP + , and 2 l of a 1 mg ml Ϫ 1 solution of yeast glucose 6-phosphate dehydrogenase ( 33 ). After incubation at 37°C for 60 min, the contents of the two Thunberg tubes were combined (equal volumes) and quenched with CH 2 Cl 2 . After centrifugation at 2 × 10 3 g for 10 min, the organic phase (lower) was carefully separated, taken to dryness under a N 2 stream, and redissolved in CH 3 CN for LC/MS analysis. The oxidation products were identifi ed as M and M+2 doublets ( 19 ) with a newly developed approach, which was based on the software MZmine2 ( 34 ) and an in-house made Matlab program, as described in the supplementary data.
LC separation was performed with a Waters Acquity UPLC system (Waters, Milford, MA) with an Acquity BEH octadecylsilane (C 18 ) UPLC column (1.7 m, 1.0 mm × 100 mm) at 50°C. Samples (10 l) were injected onto the UPLC column, and components were eluted with a linear gradient increasing from 95% (v/v) mobile phase A (10 mM NH 4  . Data was collected with a ThermoFinnigan LTQ ion trap mass spectrometer (ThermoFisher, Watham, MA) equipped with an ESI source or APCI source scanning from m/z 80 to 800 in the profi le mode, using the same instrument parameters as previously described ( 19 ). A ThermoFinnigan Orbitrap mass spectrometer was used for the collection of HRMS data.

Characterization of oxidation products
Characterization of oxidation products was performed by GC/MS after preparing the corresponding TMS ethers. Oxidation products of FFAs were obtained by incubating each FFA (100 M) in 1.0 ml of reaction mixture containing phosphate buffer, P450 2W1, NADPH-P450 reductase, L-␣ -1,2-dilauroylsn -glycero-3-phosphocholine, and an NADPH-generating system (see above). The products were extracted with 2.0 ml of CH 2 Cl 2 and dried under a N 2 stream. Epoxides were converted to dihydrodiols after incubation with H 2 O (adjusted to pH 2) at 23°C for 10 min and extracted again with CH 2 Cl 2 . Oxidation products of LPCs were obtained by incubating each LPC (100 M) with 1.0 ml of the reaction mixture (see above). The reactions were quenched with 2.0 ml of CH 3 OH containing butylated hydroxytoluene (0.005%, w/v) and 1.0 ml of aqueous KOH (15%, w/v). The mixtures were then mixed with a vortex device, purged with Ar, and incubated at 37°C for 30 min to hydrolyze the oxidized LPCs and release the oxidized fatty acids ( 35 ). The mixtures were acidifi ed to pH 2 with HCl, and the oxidized fatty acids were extracted into CH 2 Cl 2 . TMS derivatization was performed with 20 l of silylation reagent (BSTFA/TMCS/TMSI/pyridine, 3:2:3:10, v/v/v/v) at 60°C for 30 min. The resulting TMS derivatives were analyzed by GC/MS in the electron impact mode as previously described ( 19 ).

Kinetic analysis of P450 reactions
Substrate concentrations ranging from 0 to 200 M were used for steady-state kinetic studies. Reactions were run in duplicate at 37°C for 15 min. Oxidation products of FFAs and LPCs were extracted as described above. The products were derivatized with 20 l of 10% (v/v) N , N -diisopropylethylamine in CH 3 CN and 40 l of 10% (v/v) pFBB in CH 3 CN at 37°C for 20 min ( 35 ). pFBB-derivatized samples were dried under a N 2 stream and then derivatized with 20 l of BSTFA and 7 l of dry dimethylformamide at 37°C for 20 min ( 35 ) and analyzed by GC/MS in the chemical ionization mode. FFA C17:0 and C19:0 standards were used to LC/MS is one of the most widely used analytical methods for metabolomic analysis and has proved to be a powerful approach in substrate searches (15)(16)(17)(18). Recently we developed a general strategy for the identifi cation of endogenous substrates of human P450s in tissue extracts using LC/MS assays and the program DoGEX ( 19,20 ). The approach is based on the fact that the majority of P450-mediated reactions involve the incorporation of an oxygen atom into the substrate, i.e., the product is 16 amu heavier than the substrate. Incubation of a 1:1 mixture of 18 O-and 16 O-labeled oxygen gas with tissue extracts generates products as M/M+2 doublets in the MS spectra, which can be identifi ed by the program DoGEX ( 19,20 ). This strategy has been validated ( 19 ) and used to identify endogenous substrates for P450 4F11 (i.e., FFAs) ( 21 ).
The aim of the present work was to identify endogenous substrates for human P450 2W1. The purifi ed enzyme was used to conduct untargeted substrate searches in human co lorectal cancer samples (i.e., site of P450 2W1 expression) using the LC/MS metabolomic and isotopic labeling approach ( 19,21 ). A series of lysophospholipids and FFAs were identifi ed as novel substrates for P450 2W1, and the isomerand enentiomer-selectivity of P450 2W1-catalyzed lysophospholipid oxidations have been characterized. The identities of the oxidation products were defi ned, and steady-state kinetics of the P450 reactions were determined.  ( 29 ), and rat NADPH-P450 reductase ( 30 ) were expressed in E. coli and purifi ed as previously described. P450 2C19 was expressed ( 31 ) and purifi ed using the same protocol as that for P450 2C9 ( 32 ). All phospholipids were purchased from Avanti Polar Lipids (Alabaster, AL). The BSTFA:TMCS:TMSI:pyridine mixture (3:2:3:10, v/v/v/v) was purchased from Regis Technologies (Morton Grove, IL). Preparative TLC was conducted on precoated 2,000 m silica gel GF-254 plates (Analtech Inc., Newark, DE). All other reagents and solvents were obtained from general commercial suppliers.

Colorectal cancer extracts
Malignant human colorectal cancer samples were obtained from the Translational Pathology Shared Resource, Vanderbilt University School of Medicine. Human liver samples (from organ donors) were obtained from Tennessee Donor Services. Extracts were prepared from pooled samples from fi ve individuals using Folch reagent (CHCl 3 :CH 3 OH, 2:1, v/v) as previously described ( 19 ). O 2 isotopic labeling experiments, reactions were performed using the method described previously ( 19 ), except that 100% 16 O 2 and 97% 18 O 2 gas were used in two individual Thunberg tubes. endogenous substrates, with the general concept described previously ( 20 ). To profi le as many metabolites as possible, samples were analyzed with both ESI and APCI sources in both the positive and negative ionization modes. A new approach, based on the software MZmine2 and an inhouse made Matlab program, was used for doublet searches due to its improved performance regarding both precision and recall (supplementary Table I).

LC/MS metabolomics and data analysis
The doublets m/z 538/540 ( Fig. 1A ) in the ESI positive ion mode and m/z 269/271, 271/273, 295/297, 297/299, and 319/321 in the ESI negative ion mode were identifi ed (supplementary Table II). Product candidates were found only in the samples incubated with P450 2W1, NA-DPH-P450 reductase, and NADPH but not in the samples absent any of these. All doublets were further confi rmed to be oxidation products by comparison with the incubations done only with 16 O 2 gas, in which the m/z M+2 peaks were absent.
The m/z values of the respective substrates can be deduced from the m/z values of the products by subtracting 16 amu (oxygen). Therefore, the molecular masses of the putative substrates were calculated to be 521, 254, 256, 280, 282, and 304. MS fragmentation of the m/z 538 ion in the ESI positive ion mode produced a daughter ion of m/z 184 ( Fig. 1B ), indicative of a phosphocholine group ( 39 ). The LIPIDMAPS database (http://www.lipidmaps.org) suggested that 18:1 LPC ( m/z 521) was a likely substrate. MS fragmentation analysis of the products detected in the ESI negative ion mode and the search of LIPIDMAPS database suggested that FFAs C16:0, C16:1, C18:1, C18:2, and C20:4 were likely substrates. To confi rm the identities of the substrates, 18:1 LPC and fi ve FFAs were incubated with P450 2W1 and NADPH, and the extracted products were analyzed by LC/MS/MS. All of the product peaks formed in the incubations with the authentic compounds yielded exactly the same peaks identifi ed by the new software. prepare calibration curves for kinetic analysis of FFA C16:0 and C18:1 oxidation; 17:0 LPC and C19:0 LPC were used to prepare calibration curve for kinetic analysis of 16:0 LPC oxidation. Epoxy-18:1 LPC and epoxy-16:1 PC, purifi ed and quantifi ed by a phosphorus assay ( 36 ), were used to prepare calibration curves for kinetic analysis of 18:1 LPC oxidation. Epoxy-18:1 LPC was chemically synthesized by incubating 18:1 LPC with an excess amount of m -chloroperoxybenzoic acid. The reaction mixture was streaked on a preparative fl uorescent TLC plate, developed with CH 3 OH:CHCl 3 (1:1 v/v), and visualized by UV light. The lower band was eluted by the same solvent, taken to dryness using a rotary evaporator, and dissolved in C 2 H 5 OH containing 1% diisopropylethylamine (v/v). LC/MS analysis confi rmed that all 18:1 LPC was converted into epoxy-18:1 LPC. Epoxy-16:1 PC was synthesized and quantifi ed with the same method (see above). For rate comparisons of different lysophospholipids, P450 2W1 was incubated with 100 µM 18:1 LPC, 18:1 LPI, 18:1 LPS, 18:1 LPG, 18:1 LPE, or 18:1 LPA in triplicate, and the rates were determined as described above.

Purifi cation of sn -1 and sn -2 LPC
HPLC was used to separate the two isomers. sn -1 and sn -2 LPCs were monitored at 196 nm and baseline separation was achieved with a Phenomenex prodigy ODS ( 3 )

Chiral analysis
Optically pure (9 S ,10 R )-and (9 R ,10 S )-epoxystearic acids were produced by hydrogenating pure (9 S ,10 R )-epoxy-12 Z -octadecenoic acid and (9 R ,10 S )-epoxy-12Z-octadecenoic acid ( 37 ) with Pd powder under a H 2 stream for 3 min ( 19 ). The enantiomers of 9,10-epoxystearic acid were separated by normal phase HPLC with a Waters Alliance 2695 HPLC pump (Waters, Milford, MA) and a Chiralpak AD column (5 m, 4.6 mm × 25 cm ). An isocratic solvent of a 100:2:0.05 (v/v/v) hexanes/CH 3 OH/CH 3 CO 2 H mixture was used to resolve the enantiomers at a fl ow rate of 1 ml min Ϫ 1 at room temperature. The retention times of (9 S ,10 R )-and (9 R ,10 S )-epoxystearic acids were determined to be 16.9 min and 18.7 min, respectively. Epoxide generated from FFA C18:1 was extracted with CH 2 Cl 2 after enzymatic reaction. Epoxide generated from 18:1 LPC was subjected to hydrolysis as described above, and fi ve volumes of 1 M potassium phosphate buffer (pH 7.4) was added to neutralize the pH before extraction with CH 2 Cl 2 . After dried under a N 2 stream, the epoxide was analyzed as described above and detected with the APCI negative ion mode.

Searches for P450 2W1 substrates in malignant human colorectal cancer extracts
Purifi ed P450 2W1 was incubated with malignant human colorectal cancer extracts, NADPH, and 16   acids prior to silylation. Multiple doublets in the LC/MS data shared the same m/z , suggesting that each substrate may have multiple products (supplementary Table II). As summarized in Table 1 , P450 2W1 catalyzed both hydroxylation and epoxidation at the middle of fatty acid chains.

Steady-state kinetic analysis of P450 2W1 reactions
Kinetic studies were performed for 16:0 LPC and 18:1 LPC, as well as for the FFAs C16:0 and C18:1. The kinetic parameters k cat and K m were estimated based on Michaelis-Menten plots and nonlinear regression analysis ( Table 2 ). The catalytic effi ciencies ( k cat / K m ) of P450 2W1-catalyzed fatty acid oxidations were comparable with those catalyzed by other human P450 enzymes ( 19,21 ). The catalytic efficiency of 18:1 LPC oxidation was ‫ف‬ 6-fold greater than that of the FFA C18:1.

Characterization of oxidation products
GC/MS assays of TMS derivatives were used to characterize the oxidation products of P450 2W1 reactions. For FFAs, oxidation products were extracted with CH 2 Cl 2 and their identities were established after silylation. For LPCs, base hydrolysis was performed to release the oxidized fatty  reaction mixture and analyzed using the same conditions as for LPCs; no oxidation products were detected. Other classes of lysophospholipids, including 18:1 LPI, 18:1 LPS, 18:1 LPG, 18:1 LPE, and 18:1 LPA, were also confi rmed to be substrates for P450 2W1, with similar rates of oxidation ( Fig. 4 ).

DISCUSSION
To elucidate the substrate specifi city and possible physiological function of P450 2W1, untargeted substrate searches were performed using an LC/MS-based metabolomic and isotopic-labeling approach ( 19 ). In addition to FFAs, 18:1 LPC was identifi ed as a substrate for P450 2W1 using a pH 5 solvent in HPLC (supplementary Fig. I). The identities of recovered isomers were confi rmed by MS/MS fragmentation patterns, due to the different ratios of fragmentation ions at m / z 184 and 104 ( 39 ): the sn -1 isomer produces more m/z 104 daughter ion ( Fig. 3D, E ) than the sn -2 isomer ( Fig. 1B ). A mock incubation of pure sn -1 and sn -2 isomers in a P450 reaction mixture was performed for 10 min, and little interconversion between the two isomers was observed. To determine the isomer specifi city of lysophospholipid oxidation, equivalent amounts of sn -1 and sn -2 isomers were incubated with P450 2W1 and NADPH for 10 min. The reactions were quenched with three volumes of CH 3 CN, and the pH was decreased to 5 with one-half volume of 1 M NH 4 CO 2 H buffer (pH 5). Oxidation of 18:1 LPC was observed only in the incubation with sn -1 isomer ( Fig. 3A ). The sn -1 isomeric nature of the oxidation product, confi rmed by MS/MS fragmentation ( Fig. 3D, E ), also indicated that sn -1 18:1 LPC is the preferred substrate. We conclude that the sn -1 isomers of lysophospholipids are preferred substrates.
the fi rst report that the FFAs C16:0, C16:1, C18:1, and C18:2 are substrates for P450 2W1. P450 2W1 catalyzes both hydroxylation and epoxidation at the middle of fatty acid chains ( Table 1 ). Although some of oxidation products of FFAs have been shown to have interesting physiological functions in vivo ( 44 ), the fact that many P450 enzymes catalyze FFA oxidations at very slow rates raises doubts about the physiological importance of many of the products (e.g., -1, -2). To our knowledge, this is the fi rst report that lysophospholipids are substrates for any P450 enzyme. The most abundant LPCs in plasma are 16:0 LPC, 18:0 LPC, 18:1 LPC, 18:2 LPC, and 20:4 LPC ( 45 ). All of these, except 18:2 LPC and 20:4 LPC, were confi rmed to be substrates for P450 2W1. 18:2 LPC and 20:4 LPC are also likely to be substrates for P450 2W1, but these were not commercially available and, therefore, were not tested. The results of our kinetic studies suggest that unsaturated acyl LPCs are more effi ciently oxidized by P450 2W1 than are saturated acyl LPCs.
Although commercial LPCs are composed of ‫ف‬ 90% sn -1 and ‫ف‬ 10% sn -2 isomers, the percentage of each isomer in plasma has been reported to be ‫ف‬ 50% ( 46 ). In plasma, 90% of the unsaturated acyl LPCs were sn -1 isomers ( 46 ), which can be more effi ciently oxidized by P450 2W1 compared with saturated acyl LPC.
Lysophospholipids are lipid mediators involved in a vast variety of biological functions ( 47 ). In particular, LPCs are endogenous proinfl ammatory lipids that stimulate chemotaxis of T lymphocytes ( 48 ) and macrophages ( 49 ). Decreased concentrations of LPCs have been identifi ed in the plasma of colorectal cancer and lung cancer patients ( 39,50 ). Thus, decreased LPC levels may be an important contributing factor for tumor development. LPAs are also potent lipid mediators that lead to a plethora of biological actions, including cell proliferation, survival, motility, and invasion, which are critically required for tumor initiation and progression ( 51,52 ). 18:1 LPA, one of the substrates of P450 2W1, has been reported to enhance the metastatic potential of human colon cancer cells and to protect them from apoptosis (53)(54)(55). One aspect of LPA action is its role as a ligand for several cell surface G-protein coupled receptors [e.g., LPA1, LPA2, LPA3, LPA4/GPG23, and LPA5/GPR92 ( 56 )]; it is not known whether these receptors are isomer-selective or enantio-selective. LPA enhances cell proliferation by activation of the transcription factor Krüppel-like fractor 5 (KLF5) ( 54,57 ). Considering the presence of hydroxyl-and epoxy-lysophospholipids/ phospholipids in vivo ( 58,59 ), it is possible that P450 2W1catalyzed lysophospholipid oxidations are involved in infl ammation and tumor development by producing ligands to these receptors and modulate downstream signaling pathways, although further information is not available.
FFAs are common substrates for a number of human and other P450 enzymes ( 12,21,(41)(42)(43). Among the fi ve fatty acids identifi ed in our study, only C20:4 has been reported as a substrate for P450 2W1 previously ( 12 ). This is lysophospholipase-based pathways. It is possible that the upregulation of P450 2W1 expression in colorectal cancer tissues disturbs the balance of oxidized lysophospholipids and leads to pathological consequences, but further conjecture about the role of P450 2W1 in cancer is speculative.
In conclusion, we have identifi ed P450 oxidation reactions that selectively occur on lysophospholipid fatty acid chains (but not on diacylphospholipids). They were found to be selectively catalyzed by an orphan human P450, P450 2W1, that had not been clearly shown to have defi nitive catalytic activities with physiological substrates previously. The physiological functions of these oxidized lysophospholipids, if any, remain to be established. We have also introduced new software for the analysis of isotopic compositions of compounds in MS, which can be used in other metabolomic studies.