Bioactivation of Phencyclidine in Rat and Human Liver Microsomes

1488

Chem. Res. Toxicol. 2007, 20, 1488–1497

Bioactivation of Phencyclidine in Rat and Human Liver Microsomes and Recombinant P450 2B Enzymes: Evidence for the Formation of a Novel Quinone Methide Intermediate James P. Driscoll, Katherine Kornecki, Joanna P. Wolkowski, Lou Chupak, Amit S. Kalgutkar, and John P. O’Donnell* Department of Pharmacokinetics, Dynamics and Metabolism and DiscoVery Chemistry, Pfizer Global Research and DeVelopment, Groton, Connecticut 06340 ReceiVed May 1, 2007

The hypothesis that the psychological side effects associated with the anesthetic phencyclidine (PCP) may be caused by irreversible binding of PCP or its reactive metabolite(s) to critical macromolecules in the brain has resulted in numerous in vitro studies aimed at characterizing pathways of PCP bioactivation. The studies described herein extend the current knowledge of PCP metabolism and provide details on a previously unknown metabolic activation pathway of PCP. Following incubations with NADPH- and GSH-supplemented human and rat liver microsomes and recombinant P450 2B enzymes, two sulfhydryl conjugates with MH+ ions at 547 and 482 Da, respectively, were detected by LC/MS/MS. Shebley et al. [(2006) Drug Metab. Dispos. 34, 375–383] have also observed the GSH conjugate 1 with MH+ at 547 Da in PCP incubations with rat P450 2B1 and rabbit P450 2B4 isoforms fortified with NADPH and GSH. The molecular weight of 1 is consistent with a bioactivation pathway involving Michael addition of the sulfhydryl nucleophile to the putative 2,3-dihydropyridinium metabolite of PCP obtained via a four-electron oxidation of the piperidine ring in the parent compound. The mass spectrum of the novel GSH adduct 2 with an MH+ ion at 482 Da was suggestive of a unique PCP bioactivation pathway involving initial ortho- or para-hydroxylation of the phenyl ring in PCP followed by spontaneous decomposition to piperidine and an electrophilic quinone methide intermediate, which upon reaction with GSH yielded adduct 2. The LC retention times and mass spectral properties of enzymatically generated 2 were identical to those of a reference standard obtained via reaction of GSH with synthetic p-hydroxyPCP in phosphate buffer (pH 7.4, 37 °C). 1H NMR and 13C-distortionless enhancement by polarization transfer (DEPT) NMR spectral studies on synthetically generated 2 suggested that the structural integrity of the p-hydroxyphenyl and cyclohexyl rings likely was preserved and that the site of GSH addition was the benzylic carbon joining the two scaffolds. The formation of 2 in human microsomes was reduced upon addition of the dual P450 2C19/P450 2B6 inhibitor (+)-N-3-benzylnirvanol. Consistent with this finding, both recombinant P450 2B6 and P450 2C19 catalyzed PCP bioactivation to 2. In the absence of GSH, synthetic p-hydroxyPCP underwent rapid decomposition (t1/2 ∼ 5.2 min) to afford p-hydroxyphenylcyclohexanol and p-hydroxyphenylcyclohexene, presumably via the quinone methide intermediate. Overall, our findings on the facile degradation of synthetic p-hydroxyPCP to yield an electrophilic quinone methide intermediate capable of reacting with nucleophiles, including GSH and water, suggest an inherent instability of the putative phenolic PCP metabolite. Thus, if formed enzymatically in vivo, p-hydroxyPCP may not require further metabolism to liberate the quinone methide, which can then react with macromolecules. To our knowledge, this is the first report of a quinone methide reactive intermediate obtained in humanliver microsomal metabolism of PCP. Introduction 1

Phencyclidine (PCP) (Scheme 1) was originally developed in the mid-1950s for use as an anesthetic agent, but its use in human medicine was discontinued soon after because it produced serious adverse drug reactions (ADRs) that included a dose-dependent psychosis resembling schizophrenia (1, 2). In addition, a long-lasting and idiosyncratic PCP-induced * To whom correspondence should be addressed: Pharmacokinetics, Dynamics and Metabolism Department, Pfizer Global Research and Development, Groton, CT 06340. Phone: (860) 441-6002. Fax: (860) 7154695. E-mail: [email protected]. 1 Abbreviations: PCP, phencyclidine; ADRs, adverse drug reactions; PCP-Im+, iminium ion of PCP; LC/MS/MS, liquid chromatography tandem mass spectrometry; CID, collision-induced dissociation; DEPT, distortionless enhancement by polarization transfer.

psychosis was also observed in susceptible patient populations (1, 2). The etiology of these long-term and idiosyncratic side effects has not yet been determined; however, it has been proposed that these side effects could result from the irreversible binding of PCP or its reactive metabolites to critical proteins in the central nervous system (CNS) (3–8). More than two decades ago, the metabolism of PCP by cytochrome P450 enzymes was shown to result in the formation of covalent adducts with microsomal protein and of reactive metabolites that lead to P450 inactivation (3–8). On the basis of these studies, it was thought that a PCP iminium ion (PCP-Im+) metabolite obtained from the P450-mediated R-carbon oxidation of PCP was responsible for both the covalent-binding and the enzyme-inactivation events (4, 5). This proposal was supported by the isolation of a PCP–cyano adduct (Scheme 1) and the

10.1021/tx700145k CCC: $37.00  2007 American Chemical Society Published on Web 09/25/2007

Phencyclidine BioactiVation to an Electrophilic Quinone Methide

Chem. Res. Toxicol., Vol. 20, No. 10, 2007 1489

Scheme 1. Known Bioactivation Pathways of Phencyclidine in Liver Microsomes from Preclinical Species and Humans

observation that inclusion of cyanide in incubations of liver microsomes protected against covalent binding of PCP to microsomal proteins (4, 5). However, subsequent studies revealed that P450 inactivation by PCP-Im+ required the presence of NADPH (6–8), suggesting that PCP-Im+ formation is an intermediary step in PCP bioactivation and that additional reactive intermediate(s) capable of inactivating P450 and/or binding to macromolecules may be formed from this iminium species. Recent studies (9–13) of the PCP-mediated inactivation of rat and human P450 2B enzymes have also resulted in the characterization of additional reactive intermediates derived from PCP bioactivation in these systems. For example, co-incubation of PCP and GSH in NADPH-supplemented rat P450 2B1 and rabbit P450 2B4 enzymes led to the characterization of a GSH conjugate formed by the addition of the sulfhydryl nucleophile to a putative 2,3-dihydropyridinium intermediate presumably generated by a two-electron oxidation of PCP-Im+ (Scheme 1) (12). Under similar experimental conditions, use of human P450 2B6 led to the detection of a GSH conjugate of PCP that had also undergone dihydroxylation on the piperidine ring system (Scheme 1) (12). The possibility that electrophilic intermediates of PCP that are reactive toward GSH also covalently interact with microsomal proteins cannot be ruled out, especially since earlier studies showed that addition of GSH to liver microsomes incubated with PCP resulted in an ∼50% reduction in metabolismdependent covalent binding of PCP to microsomal proteins (14). In the present set of studies, we have assessed the bioactivation of PCP in NADPH- and GSH-supplemented human and rat liver microsomes and recombinant rat P450 2B1 and human P450 2B6 enzymes. In addition to the GSH conjugate 1 (structure shown in Figure 3) previously reported by Shebley et al. (12) in recombinant rat P450 2B1, we have detected a novel GSH conjugate 2 (structure shown in Figure 4) obtained via reaction of the thiol with an electrophilic quinone methide metabolite of PCP in both the human and rat liver microsomes and the recombinant enzyme systems. Insights into the structure of this de novo GSH conjugate and the PCP bioactivation pathway leading to formation of the quinone methide intermediate have been obtained via additional mechanistic studies on a

synthetic standard of a PCP derivative bearing a p-hydroxy group on its aromatic ring.

Experimental Procedures Materials. PCP was purchased from U.S. Pharmacopeia (Rockville, MD). p-HydroxyPCP was synthesized as the corresponding hydrobromide salt according to the method of Kamenka et al. (15). Mp: 143–145 °C. 1H NMR (CD3OD): δ 7.4 (d, 2H), 6.9 (d, 2H), 2.58 (broad s, 1H), 2.10 (m, 2H), 1.75 (m, 1H), 1.30–1.70 (m, 5H), 1.25 (m, 1H). Electrospray ionization analysis in the positive-ion mode revealed a single peak with a molecular ion at m/z 260 and fragment ions at m/z 175 and 86. 4,4′-Cyclohexylidinebisphenol was purchased from Sigma-Aldrich (Milwaukee, WI) and then converted to p-hydroxyphenylcyclohexene via thermal decomposition under inert conditions using the methodology of Hiramine et al. (16). Briefly, 4,4′-cyclohexylidinebisphenol (1g, 3.73 mmol) was treated with 50% aqueous NaOH (100 mL), and this mixture was heated at 250 °C for 6 h under an argon atmosphere. The reaction mixture was then cooled and treated with aqueous HCl (pH ∼ 1.0), and the aqueous solution was extracted with ethyl acetate (2 × 50 mL). The combined organic solution was washed with brine (100 mL) and water (100 mL), dried using MgSO4, and filtered. The solvent was evaporated under reduced pressure, and the oily residue was purified by column chromatography (1:99 ethyl acetate: hexanes) to afford the target compound as a colorless oil in 60% yield. 1H NMR (DMSO-d6): δ 9.30 (s, 1H, phenolic OH), 7.20 (d, J ) 8.7 Hz, 2H, Ar H), 6.65 (d, J ) 8.7 Hz, 2H, Ar H), 5.97 (t, 1H, olefin H), 2.35 (m, 2H, CH2), 2.18 (m, 2H, CH2), 1.7 (m, 2H, CH2), 1.58 (m, 2H, CH2). LC/MS: m/z 175 (M + H+). MgCl2, NADPH, GSH, furafylline, quinidine, sulfaphenazole, and ketoconazole were obtained from Sigma-Aldrich. (+)-N-3-Benzylnirvanol was synthesized at Pfizer. Unless otherwise specified, all other chemicals and solvents were obtained from Sigma-Aldrich and were reagent grade or better. Rat liver microsomes, human liver microsomal fractions pooled from 53 individual donors, and recombinant rat P450 2B1 and human P450 2B6 membrane preparations were purchased from BD Biosciences Discovery Labware (Woburn, MA). 1H and 13C-distortionless enhancement by polarization transfer (DEPT) NMR spectra in CD3OD were recorded on a Varian Unity M-400 MHz spectrometer (Varian, Inc., Palo Alto, CA); chemical shifts are expressed in parts per million calibrated to the deuterium lock signal for the deuterated solvent.

1490 Chem. Res. Toxicol., Vol. 20, No. 10, 2007 Synthesis of the GSH Adduct [2-Amino-5-(1-(carboxymethylamino)-3-(1-(4-hydroxyphenyl)cyclohexylthio)-1-oxopropan-2ylamino)-5-oxopentanoic Acid, 2]. p-HydroxyPCP (34 mg, 0.1 mmol) was added to a solution of GSH (60 mg, 0.2 mmol) in pH 7.4 sodium phosphate buffer (5 mL). The reaction mixture was stirred for 72 h at 25 °C. The solvent was removed in vacuo, and the residue was purified by reversed-phase HPLC to afford the target compound 2 (12 mg, 25% yield). 1H NMR (CD3OD): δ 1.3–1.6 (m, 4H), 1.7–1.8 (m, 2H), 1.95–2.20 (m, 6H), 2.4 (m, 3H, CH2S and C(O)CH2CH2), 2.55 (dd, J ) 4.98 and 12.87 Hz, 1H, CH2S), 3.60 (t, J ) 6.23 Hz, 1H, NH2CH), 3.78 (s, 2H, alanine CH2), 4.16 (dd, J ) 5.4 and 8.7 Hz, 1H, CHCH2S), 6.75 (d, J ) 8.7 Hz, 2H, Ar H), 7.32 (d, J ) 8.7 Hz, 2H, Ar H). 13C NMR (CD3OD): δ 22.7, 25.8, 26.5, 29.9, 31.8, 37.1, 37.2, 41.3, 51.9, 53.6, 54.1, 114.9, 128.1, 155.7, 164.9, 171.8, 172.8, 173.6. LC/MS: m/z 482.2 (M + H+). Liver Microsome Incubations. General incubation conditions for rat and human liver microsomes and recombinant rat/human P450 2B microsomal systems were essentially identical. Microsomes containing 0.25–0.5 µM P450 were thawed and suspended in 100 mM potassium phosphate (pH 7.4) containing 3.3 mM MgCl2. A PCP concentration of 10 µM [comparable to the estimated total liver concentrations of 0.9–5.6 µM (17) and in the linear range of metabolism] was used for all metabolite characterization studies. Reactions were initiated with the addition of PCP, and incubations were conducted at 37 °C in the presence or absence of NADPH cofactor (1.2 mM). For trapping experiments, GSH was added at a concentration of 5 mM. For metabolite identification experiments, reaction volumes were typically 2 mL, with 1 mL drawn at t ) 0 and the remainder quenched at t ) 30 min. All reactions were quenched with a 2-fold dilution in ice-cold acidified acetonitrile. The samples were then centrifuged (3000g, 5 min); the supernatant was dried under a steady stream of nitrogen, and the residue was reconstituted in 200 µL of 0.1% aqueous formic acid prior to analysis by liquid chromatography tandem mass spectrometry (LC/ MS/MS). Stability of p-HydroxyPCP at pH 3.0, 7.4, and 10. pHydroxyPCP (10 µM) was incubated at 25 °C in 100 mM potassium phosphate buffer at pH 3.0, 7.4, or 10.0. The buffer was acidified or basicified using concentrated phosphoric acid or potassium hydroxide, respectively. Periodically (every 3 min for 30 min), aliquots (5 µL) were injected directly out of the reaction vessel into the LC/MS/MS system and assessed for remaining starting material. The in vitro t1/2 was determined using the analyte response (expressed as a percentage of the response at t ) 0) for each time point. The slope (-k) of the linear regression of the natural logarithm of the percentage of p-hydroxyPCP remaining versus time was used to calculate t1/2 via the equation t1/2 ) -0.693/k. Values of t1/2 are reported as the mean of two independent incubations. Stability of p-HydroxyPCP under Physiological Conditions: Characterization of the Degradation Products in the Absence or Presence of GSH. p-HydroxyPCP (10 µM) was incubated at 37 °C in 100 mM potassium phosphate buffer at pH 7.4 in the presence or absence of GSH (5 mM). Aliquots (100 µL) of the reaction mixtures were removed at 0, 2, 5, 10, 15, and 30 min and quenched with 100 µL of 5 N phosphoric acid. A proprietary internal standard (100 µL of a 5 µg/mL stock solution in acidified acetonitrile) was added prior to analysis. Samples (5 µL) were injected into the LC/ MS/MS system to assess the amounts of GSH adduct 2 and p-hydroxyphenylcyclohexene formed and the remaining concentrations of p-hydroxyPCP. The slope (-k) of the linear regression of the natural logarithm of the percentage of p-hydroxyPCP remaining versus time was used to calculate t1/2 via the equation t1/2 ) -0.693/k. Incubations were conducted in duplicate. Determination of the Kinetic Constants for PCP Bioactivation to GSH Adduct 2 in Human and Rat Liver Microsomes and Recombinant P450 2B Enzymes. PCP (10–500 µM) was incubated in human and rat liver microsomes ([P450] ) 0.5 µM) and recombinant P450 2B6 and 2B1 enzymes (0.025 µM) in 100 mM potassium phosphate buffer (pH 7.4). The reaction mixtures (1 mL), containing MgCl2 (3.3 mM), NADPH (1.2 mM), and GSH

Driscoll et al. (5 mM), were prewarmed at 37 °C for 2 min prior to adding substrate. PCP was incubated for 10 min, which preliminary experiments had established as the duration of linear product formation over the concentration range (10–500 µM). Aliquots (100 µL) were taken at 0, 1, 2, 5, and 10 min and quenched with 200 µL of ice-cold acidified acetonitrile containing a proprietary internal standard (5 µg/mL). The samples were then centrifuged (3000g, 5 min), and 100 µL of supernatant was diluted with 100 µL of 0.1% aqueous formic acid prior to analysis for 2 by LC/MS/MS. Incubations for each concentration were performed in duplicate. Michaelis–Menten constants were derived from nonlinear regression of the reaction velocity on the substrate concentration using SigmaPlot (version 8.0, SPSS Science, Chicago, IL). Identification of the Human P450 Isozymes Responsible for the Formation of GSH Conjugate 2 from PCP. 1. Chemical Inhibition Studies. For P450 1A2 studies, human liver microsomes ([P450] ) 0.25 µM) were preincubated with MgCl2 (3.3 mM), NADPH (1.2 mM), and GSH (5 mM) at 37 °C in the presence of the P450 1A2 inactivator furafylline at a final concentration of 10 µM for 10 min. PCP (102 µM, the apparent Km value measured for human liver microsomes) was then added, and the reaction mixture was further incubated for 10 min at 37 °C. For competitive P450 inhibition studies, human liver microsomes ([P450] ) 0.25 µM) were incubated for 10 min at 37 °C with PCP (102 µM), MgCl2 (3.3 mM), NADPH (1.2 mM), GSH (5 mM), and a P450 inhibitor at the given final concentration(s) for the specified enzyme(s): quinidine (5 µM, P450 2D6), sulfaphenazole (5 µM, P450 2C9), (+)-N-3-benzylnirvanol (3 µM, P450 2C19; 300 µM, P450 2B6), and ketoconazole (1 µM, P450 3A4/5). Incubations were conducted in duplicate, and the amount of 2 formed relative to that in a control containing no inhibitor was determined as described above. 2. Metabolism by Heterologously Expressed P450 Isozymes. P450 isozyme activity was further verified using heterologously expressed P450 2B6 and 2C19, both of which were identified by inhibition studies as participants in the formation of 2 following incubation of PCP in human liver microsomes. Incubations were conducted using 0.025 µM isozyme in 100 mM potassium phosphate (pH 7.4). The reaction mixtures (1 mL), containing MgCl2 (3.3 mM), NADPH (1.2 mM), and GSH (5 mM), were prewarmed at 37 °C for 2 min prior to adding substrate (500 µM PCP) and then incubated for 30 min. The reactions were quenched after 30 min with 2 mL of acidified acetonitrile, and the reaction mixtures were processed for LC/MS/MS analysis as detailed above. Preparative HPLC Separation of p-HydroxyPCP Degradation Products. Following incubation of p-hydroxyPCP (5 mg) in 100 mM potassium phosphate buffer (pH 7.4) at 37 °C for 4 h, degradation products exhibiting an m/z of 175 were separated by preparative HPLC using a Symmetry 19 × 50 mm column and UV detection (λ ) 254 nm). The flow rate was 25 mL/min, and a 10 min binary gradient consisting of 0.1% trifluoroacetic acid in water (solvent A) and 0.1% trifluoroacetic acid in acetonitrile (solvent B) was used to elute major m/z 175 peaks. A linear gradient started at 10% solvent B and ramped to 90% solvent B in 10 min separated two major peaks. Isolated fractions were evaporated to dryness at 40 °C using a GeneVac prior to analysis by 1H NMR and LC/MS/MS. LC/MS/MS Methods. Qualitative metabolite formation and characterization was assessed on a Sciex API 3000 LC/MS/MS triple-quadrupole mass spectrometer in line with a Shimadzu LC20AD HPLC system (Applied Biosystems/MDS Sciex, Foster City, CA). An autosampler was programmed to inject 20 µL of the sample on a Synergi Polar RP 4 µm, 4.6 × 150 mm column (Phenomenex Inc., Torrance, CA) using a binary gradient consisting of a mixture of 95% water, 5% acetonitrile, and 0.1% formic acid (solvent A) and a mixture of 95% acetonitrile, 5% water, and 0.1% formic acid (solvent B) at a flow rate of 0.8 mL/min. The LC gradient was programmed as follows: the solvent A:solvent B ratio (v/v) was held at 90:10 for 0.2 min, then adjusted from 90:10 to 10:90 over 16.5 min, held at 10:90 for 30 s, and finally adjusted from 10:90 back to the initial ratio of 90:10 over 3 min (a total run

Phencyclidine BioactiVation to an Electrophilic Quinone Methide time of 20.2 min). The column was re-equilibrated for 3 min prior to the next analytical run. Postcolumn flow was split such that the mobile phase was introduced into the mass spectrometer via an ion-spray interface at a rate of 300 µL/min. Collision-induced dissociation (CID) was conducted in the positive-ion mode at an ion-spray interface temperature of 400 °C using nitrogen as the collision gas at a collision energy of 10 eV. The ion-spray voltage was 4.5 kV, and the orifice voltage was optimized at 40 eV. Initial Q1 scans were performed between m/z 75 and 800. Metabolites were identified by comparing t ) 0 samples to t ) 30 min samples, and structural information was generated from CID spectra of the corresponding protonated molecular ions and/or comparison to LC retention times (tR) and product-ion spectra of synthetic standards. Quantitative analysis of key PCP metabolites and degradation products was completed using a 3 min analytical method utilizing the instrumentation described above. A proprietary internal standard (MW ) 392, 5 µg/mL in acidified acetonitrile) was added prior to analysis. An autosampler was programmed to inject 5 µL on a Synergi Polar RP 2 µm, 30 × 2.0 mm column using a mobile phase consisting of 95% water and 5% acetonitrile containing 0.1% formic acid (solvent A) and 95% acetonitrile and 5% water (solvent B). The LC conditions were programmed as follows: 10% acetonitrile for 0.5 min, increasing to 90% over the next 1.8 min and then decreasing back to 10% (i.e., the original condition) over the final 3 min, at a flow rate of 0.5 mL/min. Ionization was conducted in the positive-ion mode at an ion-spray interface temperature of 400 °C, using nitrogen as the nebulizing and heating gas. The ion-spray voltage was 5.0 kV, and the orifice voltage was optimized at 30 eV. Synthetic standards of GSH adduct 2, p-hydroxyPCP, and p-hydroxyphenylcyclohexene were analyzed in the multiple-reaction monitoring (MRM) mode using the transitions m/z 482 f 175, m/z 260 f 175, and m/z 175 f 107, respectively. Calibration curves were prepared by plotting the appropriate peak-area ratios (analyte/ internal standard) against the analyte concentrations using 1/x weighting. The concentration of the analytes was determined by interpolation from the standard curve. The typical dynamic range for each assay varied from 2.5 to 10 000 ng/mL.


Figure 1. Extracted ion chromatograms of GSH conjugate 1 (MH+ ) 547, tR ) 7.02 min) following incubation of PCP (10 µM) in GSHsupplemented human liver microsomes: (A) microsomes also supplemented with NADPH; (B) microsomes not supplemented with NADPH.

Results Trapping Reactive Metabolites of PCP with GSH. Reduced GSH (5 mM) was used as an exogenous nucleophile to trap reactive metabolites of PCP in incubations of NADPHsupplemented rat and human liver microsomes and recombinant P450 enzymes. Qualitative LC/MS/MS analysis of NADPHsupplemented rat and human liver and P450 2B1/2B6 microsome incubations containing PCP (10 µM) and GSH led to the detection of the two GSH conjugates 1 (MH+ ) 547, tR ) 7.02 min) and 2 (MH+ ) 482, tR ) 8.04 min). The formation of 1 and 2 in these incubations was NADPH-dependent, as illustrated by representative chromatograms for incubations of human liver microsomes (Figures 1 and 2). The molecular weight of GSH adduct 1 was consistent with the addition of one molecule of GSH to one molecule of dehydrogenated PCP. The product-ion spectrum of GSH adduct 1 (Figure 3) was similar to the one published previously by Shebley et al. (12) for PCP incubations with rat P450 2B1. Thus, the diagnostic fragment ions at m/z 308, 240, and 82 were consistent with the presence of a GSH molecule and the tetrahydropyridine ring of PCP, suggesting that GSH had added to the 2,3-dihydropyridinium intermediate of PCP in a 1,2-Michael fashion, as previously described (12). Since details on GSH adduct 1 have been described earlier (12), no further characterization of this product was attempted. In the case of GSH adduct 2, the product-ion spectrum obtained by CID of the MH+ ion at m/z 482 (Figure 4) included fragment ions at m/z 308, 233, 179, 175, 162, and 107. While the fragment ions at m/z 308, 233, 179, and 162 were consistent with the presence of a GSH

Figure 2. Extracted ion chromatogram of GSH conjugate 2 (MH+ ) 482, tR ) 8.04 min) following incubation of PCP (10 µM) in GSHsupplemented human liver microsomes: (A) microsomes also supplemented with NADPH; (B) microsomes not supplemented with NADPH.

molecule, those at m/z 175 and 107 were consistent with the presence of p-hydroxyphenylcyclohexane and of a hydroxylated tropylium species, respectively. The fragment ions at m/z 175 and 107 were also observed in the mass spectrum of the synthetic standard of p-hydroxyPCP. A proposed structure for GSH adduct 2 that is consistent with the product-ion spectrum is shown in Figure 4. Elucidation of the Structure of the Major GSH Adduct 2. Since the yield of GSH adduct 2 was insufficient for LC/ NMR work, we decided, for the purposes of LC/MS/MS and NMR characterization, to prepare a synthetic standard of the proposed adduct 2 via the reaction of GSH with synthetic p-hydroxyPCP (15) in potassium phosphate buffer (pH 7.4, 25 °C for 72 h). The resulting GSH adduct possessed a tR and molecular mass (MH+ ) 482) identical with those of the enzymatically generated GSH adduct 2. The 1H NMR spectrum (Figure 5A) of the synthetic GSH adduct 2 revealed the appearance of characteristic signals between 2.0 and 4.5 ppm for the glutathionyl moiety as well as two aromatic signals, each integrating for two protons and appearing as a doublet (J ) 8.7 Hz). This characteristic pattern clearly indicated that the p-hydroxyphenyl ring in the synthetic standard of 2 was

1492 Chem. Res. Toxicol., Vol. 20, No. 10, 2007

Figure 3. Product-ion spectrum of GSH adduct 1 obtained by CID of the MH+ ion (m/z 547). The origins of the characteristic ions are as indicated. The mass spectrum is consistent with that previously reported by Shebley et al. (12).

Figure 4. Product-ion spectrum of GSH adduct 2 obtained by CID of the MH+ ion (m/z 482). The origins of the characteristic ions are as indicated.

unaltered and the site of addition of GSH was on the cyclohexyl ring. Regiochemistry of GSH attachment on the cyclohexyl ring was examined by 13C-DEPT NMR spectroscopy, which sorts 13 C NMR signals by the number of protons attached to each carbon. In this experiment, four CH signals and seven signals from fully substituted carbon atoms were observed. The CH peaks at 128.1 and 115.9 ppm were consistent with a symmetric para-substituted phenol. The CH peaks at 54.1 and 53.6 ppm were consistent with the methine protons present in GSH. No new methine proton signals were discernible in the DEPT experiment. Finally, the 13C NMR spectrum of GSH adduct 2 revealed the presence of a new aliphatic quaternary carbon signal at 51.97 ppm (the quaternary carbon signal in p-hydroxyPCP appeared at ∼74 ppm), consistent with benzylic substitution (carbon e in Figure 5B). Identification of the Human P450 Isozymes Responsible for the Formation of GSH Conjugate 2. To identify the human P450 isozymes involved in the formation of GSH adduct 2, PCP (102 µM, the apparent Km for formation of 2 from PCP in human liver microsomes) was incubated with human liver microsomes containing NADPH and GSH in the absence or presence of isozyme-selective P450 inhibitors. Pretreatment of human liver microsomes with quinidine (a P450 2D6 inhibitor), ketoconazole (a P450 3A4 inhibitor), sulfaphenazole (a P450 2C9 inhibitor), and furafylline (a P450 1A2 inactivator) did not attenuate GSH adduct formation (Figure 6). However, the formation of GSH adduct 2 was markedly inhibited in human

Driscoll et al.

liver microsomes pretreated with (+)-N-3-benzylnirvanol at a concentration of 300 µM (Figure 6). At a concentration of 3 µM (+)-N-3-benzylnirvanol, moderate inhibition was observed as well, suggesting a role for P450 2C19. While the selectivity of (+)-N-3-benzylnirvanol as an inhibitor of P450 2C19 has been established (18), Walsky and Obach (19) have recently shown that at higher concentrations (+)-N-3-benzylnirvanol also functions as a P450 2B6 inhibitor. Thus, further inhibition of the formation of 2 at higher (+)-N-3-benzylnirvanol concentrations suggests that P450 2B6 is likely to catalyze the formation of this GSH conjugate. An incubation of PCP (500 µM) in heterologously expressed P450 2B6 and 2C19 for 30 min in the presence of MgCl2 (3.3 mM), NADPH (1.2 mM), and GSH (5 mM) at 37 °C demonstrated the formation of GSH adduct 2 (data not shown). Enzyme Kinetics for the Reaction Sequence PCP f GSH Adduct 2 in Human and Rat Liver Microsomes and in Recombinant P450 2B Enzymes. Table 1 lists the apparent Km and Vmax estimates for the formation of the GSH adduct 2 from PCP in microsomes and recombinant P450 2B enzyme systems. Consistent with the results obtained from the chemical inhibition studies in human microsomes, both P450 2C19 and 2B6 catalyzed the formation of GSH adduct 2. Values of kinetic parameters could not be fully established for P450 2C19 because of insufficient sensitivity to the formation of 2 at low substrate concentrations for this enzyme; however, a formation rate of 0.05 nmol (nmol of P450)–1 min–1 was observed at 500 µM. Vmax estimates of 0.17 and 0.80 nmol (nmol of P450)–1 min–1 were determined for recombinant P450 2B6 and 2B1, respectively. The Km value of 102 µM for the formation of 2 using human liver microsomes is more than an order of magnitude higher than the estimated total liver concentration of PCP (0.9–5.6 µM) (17), suggesting that the process of conversion of PCP to GSH adduct 2 is not expected to be saturated in vivo. pH Stability of p-HydroxyPCP at Room Temperature. Stability of synthetic p-hydroxyPCP (10 µM) at room temperature (25 °C) was assessed at pH 3.0, 7.4, and 10 (Figure 7). First-order degradation rate constants of 180, 43, and 0.4 min at pH 3.0, 7.4, and 10, respectively. Stability of p-HydroxyPCP under Physiological Conditions. Under physiological conditions (pH ∼ 7.4, 37 °C), the observed degradation half-life of synthetic p-hydroxyPCP was 5.2 min and did not change when GSH (5 mM) was present in the incubation mixture (Figure 8). Nearly 3.5 nmol of GSH adduct 2 was formed in the presence of GSH, with the remainder of the starting material forming a mixture of p-hydroxyphenylcyclohexanol [facile dehydration of the compound in the mass spectrometer produced a molecular ion (MH+) observed at m/z 175] and p-hydroxyphenylcyclohexene, as suggested by chromatographically separated peaks bearing molecular ions (MH+) at m/z 175. As shown in Figure 9, two peaks from an extracted ion chromatogram of m/z 175 (MH+) corresponded to a mixture of p-hydroxyphenylcyclohexanol (tR ) 14.15 min) and phydroxyphenylcyclohexene (tR ) 18.67 min). Apart from these products, piperidine (MH+ ) 86, tR ) 0.44 min) that was liberated in the decomposition process was also detected in this analysis (Figure 9, inset). In connection with these results, it is worth noting that in the absence of GSH, incubation of PCP with NADPH-supplemented human liver microsomes and recombinant P450 2B systems also led to the formation of the


Figure 5. 1H and

13


C NMR spectra (CD3OD) of the synthetic GSH adduct 2 generated via the reaction of p-hydroxyPCP with GSH.

Table 1. Km and Vmax Values for the Reaction Sequence PCP f GSH Adduct 2 in Human and Rat Liver Microsomes and in Recombinant P450 2B Enzymes

Figure 6. Effect of isozyme-selective P450 inhibitors on the formation of GSH adduct 2 in incubations of PCP with human liver microsomes. Data are reported as the mean of two separate determinations.

p-hydroxyphenylcyclohexanol/p-hydroxyphenylcyclohexene mixture, via the P450-catalyzed metabolism of PCP to p-hydroxyPCP. The structural identities of p-hydroxyphenylcyclohexene and p-hydroxyphenylcyclohexanol in these incubation mixtures were confirmed by 1H NMR and LC/MS/MS following preparative HPLC isolation of the two degradation products. The LC/MS/MS total ion chromatogram of each compound

P450 source

Km (µM)

Vmax [nmol (nmol of P450)–1 min–1]

human liver microsomes rat liver microsomes P450 2B1 P450 2B6

102 143 76 39

0.003 0.010 0.80 0.17

revealed a molecular ion (MH+) at m/z 175; these ions had identical product-ion spectra (data not shown). 1H NMR spectroscopy confirmed the p-hydroxyphenyl substructure with a characteristic AA′BB′ coupling in the aromatic region (Figure 10). The later-eluting (less polar) p-hydroxyphenylcyclohexene contained characteristic exchangeable phenolic and olefinic protons at 9.3 and 5.9 ppm, respectively (Figure 10, bottom panel). The earlier-eluting (more polar) peak demonstrated additional exchangeable protons at 4.5 and 9.1 ppm, consistent with the tertiary-alcohol and phenolic OH groups, respectively, of p-hydroxyphenylcyclohexanol. 13CDEPT NMR spectroscopy also confirmed the lack of protons


Driscoll et al.

derivative under physiological conditions also led to the formation of GSH adduct 2 (data not shown).

Discussion

Figure 7. First-order degradation of p-hydroxyPCP at pH 3.0, 7.4, and 10.0 at 25 °C.

Figure 8. Degradation of p-hydroxyPCP and formation of GSH adduct 2 in a GSH-supplemented pH 7.4 phosphate buffer at 37 °C.

Figure 9. Extracted ion chromatogram of m/z 175 (MH+) following incubation of p-hydroxyPCP in a pH 7.4 buffer at 37 °C. Inset: Extracted ion chromatogram of piperidine (MH+ ) 86) obtained as a byproduct during the decomposition of p-hydroxyPCP in pH 7.4 buffer at 37 °C. The tR value and the CID spectrum were identical to those of a synthetic standard.

on the benzylic carbon, providing further evidence for the proposed tertiary alcohol structure. Interestingly, this spectrum retained shifts at 5.9 and 9.3 ppm, corresponding to the olefinic and phenolic protons, respectively, of p-hydroxyphenylcyclohexene. Closer inspection of the NMR spectra demonstrated that the polar peak was a mixture of phydroxyphenylcyclohexanol and p-hydroxyphenylcyclohexene. In regard to this, it is worth noting that addition of excess GSH (5 mM) to the isolated p-hydroxyphenylcyclohexene

Following the initial hypothesis that the ADRs associated with PCP could be attributed to covalent interactions of a reactive metabolite(s) of PCP with critical biomacromolecules in the CNS, there have been several elegant in vitro mechanistic studies dedicated toward understanding PCP metabolism and bioactivation (Scheme 1). A key finding in most of these studies was that R-carbon oxidation of PCP, which leads to PCP-Im+, is a major pathway for oxidative metabolism of the drug by P450 enzyme(s). Furthermore, on the basis of trapping studies with cyanide, it was originally thought that PCP-Im+ is the reactive PCP intermediate responsible for covalent interactions with liver protein as well as for P450 inactivation. However, the finding that the mechanism-based inactivation of P450 by PCP-Im+ is NADPH-dependent suggested that PCP-Im+ undergoes further biotransformation by P450 to yield the ultimate reactive species. Consequently, attempts aimed at characterizing secondary reactive species derived from the metabolism of PCP-Im+ in liver microsomes have been carried out. Hoag et al. (7, 20) have characterized a conjugated amino-enone metabolite of PCP-Im+, 1-(1-phenylcyclohexyl)-2,3-dihydro-4-pyridone, in liver microsomes (Scheme 1), but this product was also shown not to be directly responsible for P450 inactivation. More recently, using PCP as a tool to probe structure–function relationships for recombinant rat and human P450 2B isoforms, Shebley et al. (12) have characterized additional electrophilic intermediates, including the 2,3-dihydropyridinium metabolite of PCP and the GSH adducts of mono- and dihydroxylated metabolites derived from oxidation of the 2,3-dihydropyridinium intermediate (Scheme 1). Incidentally, the 2,3-dihydropyridinium intermediate is thought to be a precursor leading to the amino-enone metabolite observed by Hoag et al. (20). While most of the efforts to elucidate PCP bioactivation pathways have focused on the metabolism of the piperidine ring, metabolic activation mechanisms involving the phenyl ring have received very little attention. The original report by Law (3) on metabolism-dependent covalent binding of radiolabeled PCP to microsomal protein speculated that an electrophilic aryl epoxide metabolite of PCP may be responsible for covalent binding. However, to date no evidence has been presented to support the existence of such a reactive intermediate, and on the basis of in vitro metabolite identification studies in liver microsomes from preclinical species (21, 22) and from humans (23), it appears that the piperidine and cyclohexyl rings in PCP are the primary sites for P450-catalyzed oxidation. However, there are some instances where P450-mediated aromatic ring hydroxylation in PCP has been inferred on the basis of characterization of downstream metabolites (24–26) and/ or comparison to synthetic standards. Indirect evidence for the formation of phenolic PCP metabolites was first provided by Wong and Beimann (24) in their characterization of hydroxyphenylcyclohexene as a PCP metabolite in rat urine. In that report, the investigators proposed that hydroxyphenylcyclohexene was produced by thermal degradation of a hydroxyPCP metabolite as an artifact of the GC/MS analysis. Since PCP is known to undergo pyrolysis in cigarette smoke to yield free piperidine and phenylcyclohexene (27), thermal degradation within the injection port is a plausible explanation for the observed urinary metabolite. However, in the course of our studies of the stability of synthetic p-hydroxyPCP, we observed the formation of both p-hydroxyphenylcyclohexanol and p-



Figure 10. 1H NMR spectra (DMSO-d6) of a mixture of p-hydroxyphenylcyclohexanol and p-hydroxyphenylcyclohexene (top) and of p-hydroxyphenylcyclohexene only (bottom).

Scheme 2. P450-Mediated PCP Bioactivation Involving Formation of an Electrophilic Quinone Methide Intermediate

hydroxyphenylcyclohexene (detected at m/z 175) following incubation in a pH 7.4 phosphate buffer. These metabolites were chromatographically separated from each other and therefore were not thermal degradation products. Rather, the formation of p-hydroxyphenylcyclohexanol and p-hydroxphenylcyclohexene from p-hydroxyPCP appears to be derived from solvolysis of the released quinone methide with water and subsequent dehydration during sample workup. In fact, after isolation and concentration of a single chromatographic peak for p-hydroxyphenylcyclohexanol, more p-hydroxyphenylcyclohexene was

generated, suggesting that the tertiary alcohol is unstable. p-Hydroxyphenylcyclohexene also appears to exist as its quinone methide tautomer, as suggested by the formation of the GSH adduct 2 following incubation with the isolated p-hydroxyphenylcyclohexene metabolite (data not shown). These data are consistent with the observed behavior of p-hydroxyPCP alone in a pH 7.4 buffer, where it rearranges to the quinone methide by releasing the piperidine ring and then reacts with either solvent (H2O) or GSH at the benzylic carbon. The competition of solvent and GSH (or other endogenous protein


nucleophiles) for addition is dependent upon the reactivity of the quinone methide (28). Our studies reveal that the instability of p-hydroxyPCP is dependent on pH and is driven predominantly by deprotonation of the phenol, which is consistent with its accelerated instability at higher pH. Addition of GSH and/ or water to the quinone methide occurs readily at pH 7.4 and 37 °C. Direct evidence for formation of phenolic metabolites of PCP was first presented by Ohta et al. (26), who reported that incubation of PCP in rat and mouse liver microsomes led to the formation of a phenol derivative whose tR and mass spectral properties corresponded to those of a reference standard of m-hydroxyPCP. Given our present results showing rapid degradation of the synthetic p-hydroxyPCP derivative to the quinone methide under physiological conditions, it is not surprising that those investigators did not detect p-hydroxyPCP as a metabolite. Similarly, we could not confirm concentrations of p-hydroxyPCP following incubation of PCP with rat and human microsomes even when conducting the incubation at 25 °C (room temperature), which could theoretically afford greater chemical stability of this metabolite after its formation (data not shown). The amounts of major monohydroxylated metabolites formed under these nonphysiological conditions, however, were quite small. Since most of the previous PCP bioactivation studies were conducted using liver microsomes from preclinical species (e.g., rabbit, rat, and mouse) and this information has been extended to the human situation, we decided to re-assess PCP bioactivation in incubations of rat and human liver and recombinant rat P450 2B1 and human P450 2B6 microsomes. To our surprise, we discovered a new pathway of PCP bioactivation in rat and human hepatic tissue involving a rate-limiting P450mediated aromatic ring hydroxylation of PCP to p-hydroxyPCP (Scheme 2). Once formed, p-hydroxyPCP possesses the propensity to rapidly decompose to the quinone methide, without the need for additional metabolism to decrease electron density around the piperidine nitrogen (e.g., R-carbon oxidation of the p-hydroxyPCP metabolite to the corresponding iminium ion). The quinone methide can react with GSH or water, resulting in the formation of the sulfhydryl conjugate 2 or p-hydroxyphenylcyclohexanol, respectively. While comparison of the Km value for the formation of 2 in vitro with the estimated liver concentration of 2 suggests that the pathway is not expected to be saturated in vivo, extrapolation of the kinetic data to an in vivo situation needs to be evaluated in greater detail because the estimated Km represents a sum of three steps: enzymatic conversion of PCP to p-hydroxyPCP, nonenzymatic elimination of piperidine to form the corresponding quinone methide, and nonenzymatic conjugation of the quinone methide with GSH. In conclusion, the present study demonstrates the formation of a novel, electrophilic quinone methide following parahydroxylation of the phenyl ring of PCP by human liver microsomes and recombinant P450 2B enzymes. Using synthetic p-hydroxyPCP, we have shown that the extent of reactive quinone methide formation increases with increasing pH. The finding that PCP and synthetic p-hydroxyPCP form a stable GSH conjugate after rearrangement to the quinone methide in NADPH-fortified liver microsomes and buffer, respectively, suggests a potential role for the p-hydroxyPCP metabolite, alongside other reactive intermediates such as PCP-Im+ and its bioactivation products, in the formation of covalent protein adducts. The relative contributions of these various pathways toward covalent adduct formation in vivo and their role in

Driscoll et al.

neurological toxicities associated with PCP remain to be investigated. Acknowledgment. We thank Ms. Beth Cooper for her assistance in isolating p-hydroxyphenylcyclohexanol and phydroxyphenylcyclohexene.

References (1) Greifenstein, F. E., Yoshitake, J., DeVault, M., and Gajewski, J. E. (1958) A study of 1-aryl cyclohexylamine for anesthesia. Anesth. Analg. 37, 283–294. (2) Fauman, B., Aldinger, G., Fauman, M., and Rosen, P. (1976) Psychiatric sequelae of phencyclidine abuse. Clin. Toxicol. 9, 529– 538. (3) Law, F. C. P. (1981) Metabolic disposition and irreversible binding of phencyclidine in rats. Toxicol. Appl. Pharmacol. 57, 263–272. (4) Ward, D., Kalir, A., Trevor, A., Adams, J., and Baillie, T., Jr. (1982) Metabolic formation of iminium species: Metabolism of phencyclidine. J. Med. Chem. 25, 491–492. (5) Ward, D. P., Trevor, A. J., Kalir, A., Adams, J. D., and Baillie, T. A., Jr. (1982) Metabolism of phencyclidine: The role of iminium ion formation in covalent binding to rabbit microsomal protein. Drug Metab. Dispos. 10, 690–695. (6) Hoag, M. K. P., Trevor, A. J., Asscher, Y., and Weissman, J., Jr. (1984) Metabolism-dependent inactivation of liver microsomal enzymes by phencyclidine. Drug Metab. Dispos. 12, 371–375. (7) Hoag, M. K. P., Trevor, A. J., and Kalir, A., Jr. (1987) Phencyclidine iminium ion: NADPH-dependent metabolism, covalent binding to macromolecules, and inactivation of cytochrome(s) P-450. Drug Metab. Dispos. 15, 485–490. (8) Osawa, Y., and Coon, M. J. (1989) Selective mechanism-based inactivation of the major phenobarbital-inducible P-450 cytochrome from rabbit liver by phencyclidine and its oxidation product, the iminium compound. Drug Metab. Dispos. 17, 7–13. (9) Crowley, J. R., and Hollenberg, P. F. (1995) Mechanism-based inactivation of rat liver cytochrome P4502B1 by phencyclidine and its oxidative product, the iminium ion. Drug Metab. Dispos. 23, 786– 793. (10) Sharma, U., Roberts, E. S., Kent, U. M., Owens, S. M., and Hollenberg, P. F. (1997) Metabolic activation of cytochrome P4502B1 by phencyclidine: Immunochemical and radiochemical analyses of the protective effects of glutathione. Drug Metab. Dispos. 25, 243–250. (11) Jushchyshyn, M. I., Kent, U. M., and Hollenberg, P. F. (2003) The mechanism-based inactivation of human cytochrome P450 2B6 by phencyclidine. Drug Metab. Dispos. 31, 46–52. (12) Shebley, M., Jushchyshyn, M. I., and Hollenberg, P. F. (2006) Selective pathways for the metabolism of phencyclidine by cytochrome P450 2B enzymes: Identification of electrophilic metabolites, glutathione, and N-acetyl cysteine adducts. Drug Metab. Dispos. 34, 375–383. (13) Jushchyshyn, M. I., Wahlstrom, J. L., Hollenberg, P. F., and Wienkers, L. C. (2006) Mechanism of inactivation of human cytochrome P450 2B6 by phencyclidine. Drug Metab. Dispos. 34, 1523–1529. (14) Kalir, A., Trevor, A. J., Ward, D. P., Adams, J. D., Baillie, T. A., and Castagnoli, N., Jr. (1983) Reactive metabolites of phencyclidine and covalent binding to microsomal proteins. In Phencyclidine and Related Arylcyclohexylamines: Present and Future Applications (Kamenca, J. M., Domino, E. F., and Geneste, P., Eds.) pp 267–278, NPP Books, Ann Arbor, MI. (15) Kamenka, J. M., Chiche, B., Goudal, R., Geneste, P., Vignon, J., Vincent, J. P., and Lazdunski, M. (1982) Chemical synthesis and molecular pharmacology of hydroxylated 1-(1-phenylcyclohexyl)piperidine derivatives. J. Med. Chem. 25, 431–435. (16) Hiramine, T., Ono, H., Isoda, Y., Yao, K., and Kojima, H. (2002) Preparation of cyclohexenylphenols. Jpn. Kokai Tokkyo Koho JP 2002212123. (17) Walberg, C. B., McCarron, M. M., and Schulze, B. W. (1983) Quantitation of phencyclidine in serum by enzyme immunoassay: results in 405 patients. J. Anal. Toxicol. 7, 106–110. (18) Suzuki, H., Kneller, M. B., Haining, R. L., Trager, W. F., and Rettie, A. E. (2002) (+)-N-3-Benzyl-nirvanol and (-)-N-3-benzyl-phenobarbital: new potent and selective in vitro inhibitors of CYP2C19. Drug Metab. Dispos. 30, 235–239. (19) Walsky, R. L., and Obach, R. S. (2003) Verification of the selectivity of (+)-N-3-benzylnirvanol as a CYP2C19 inhibitor. Drug Metab. Dispos. 31, 343. (20) Hoag, M. L. P., Schmidt-Peetz, M., Lampen, P., Trevor, A. J., and Castagnoli, N., Jr. (1988) Metabolism studies on phencyclidine: characterization of a phencyclidine iminium ion metabolite. Chem. Res. Toxicol. 1, 128–131.



(21) Gole, D. J., Pirat, J.-L., Kamenka, J.-M., and Domino, E. F. (1988) Hydroxy metabolites of phencyclidine: Identification and quantitation of two novel metabolites. Drug Metab. Dispos. 16, 386–391. (22) Holsztynska, E. J., and Domino, E. F. (1986) Biotransformation of phencyclidine. Drug Metab. ReV. 16, 285–320. (23) Laurenzana, E. M., and Owens, S. M. (1997) Metabolism of phencyclidine by human liver microsomes. Drug Metab. Dispos. 25, 557–563. (24) Wong, L. K., and Biemann, K. (1976) Metabolites of phencyclidine. Clin. Toxicol. 9, 583–591. (25) Misra, A. L., Pontani, R. B., and Bartolomeo, J. G. (1980) Disposition of [3H]-phencyclidine in the rat after single and multiple doses. Life Sci. 27, 2501–2508.

(26) Ohta, S., Masumoto, H., Takeuchi, K., and Hirobe, M. (1987) A phenolic metabolite of phencyclidine. The formation of a pharmacologically active metabolite by rat liver microsomes. Drug Metab. Dispos. 15, 583–584. (27) Martin, B. R., Bailey, B. B., Awaya, H., May, E. L., and Narasimhachari, N. (1982) In vitro metabolism of 1-phenyl-1-cyclohexene, a pyrolysis product of phencyclidine. Drug Metab. Dispos. 10, 685–689. (28) Thompson, D. C., Perera, K., Krol, E. S., and Bolton, J. L. (1995) o-Methoxy-4-alkylphenols that form quinone methides of intermediate reactivity are the most toxic in rat liver slices. Chem. Res. Toxicol. 8, 323–327.

TX700145K

Bioactivation of Phencyclidine in Rat and Human Liver Microsomes

Recommend Documents