Abstract
Aldehyde oxidase (AO) metabolism could lead to significant underestimation of clearance in prediction of human pharmacokinetics as well as unanticipated exposure to AO-generated metabolites, if not accounted for early in drug research. We report a method using cryopreserved human hepatocytes and the time-dependent AO inhibitor hydralazine (KI = 83 ± 27 μM, kinact = 0.063 ± 0.007 min−1), which estimates the contribution of AO metabolism relative to total hepatic clearance. Using zaleplon as a probe substrate and simultaneously monitoring the AO-catalyzed formation of oxozaleplon and the CYP3A-catalyzed formation of desethyzaleplon in the presence of a range of hydralazine concentrations, it was determined that >90% inhibition of the AO activity with minimal effect on the CYP3A activity could be achieved with 25 to 50 μM hydralazine. This method was used to estimate the fraction metabolized due to AO [fm(AO)] for six compounds with clearance attributed to AO along with four other drugs not metabolized by AO. The fm(AO) values for the AO substrates ranged between 0.49 and 0.83. Differences in estimated fm(AO) between two batches of pooled human hepatocytes suggest that sensitivity to hydralazine varies slightly with hepatocyte preparations. Substrates with a CYP2D6 contribution to clearance were affected by hydralazine to a minor extent, because of weak inhibition of this enzyme. Overall, these findings demonstrate that hydralazine, at a concentration of 25 to 50 μM, can be used in human hepatocyte incubations to estimate the contribution of AO to the hepatic clearance of drugs and other compounds.
Introduction
Aldehyde oxidase (AO) is a soluble molybdenum cofactor-containing enzyme that is capable of oxidizing aldehydes, imines, and aromatic azaheterocyclic compounds (Beedham, 2002; Pryde et al., 2010; Garattini and Terao, 2012). On aromatic azaheterocyclic compounds, it catalyzes the oxidation of relatively electrophilic carbons adjacent to the nitrogen to generate lactam metabolites, with the molybdopterin cofactor participating in a nucleophilic attack on the electrophilic carbon. Although oxygen is the ultimate electron acceptor, the oxygen inserted into the lactam product derives from water; the reducing equivalents from the substrate are passed along to oxygen via FAD and FeS cofactors (Pryde et al., 2010; Garattini and Terao, 2012). A specific endogenous substrate has not been definitively identified, but AO potentially participates in the metabolism of neurotransmitters, oxidation of products involved in various metabolic pathways, and also degradation of vitamins (Garattini et al., 2003). Notable substrates used in in vitro work include N-methylnicotinamide and phthalazine. Drugs known to have an important contribution of aldehyde oxidase in human include zaleplon (Lake et al., 2002; Renwick et al., 2002) and famciclovir in which AO is involved in metabolism of a prodrug to the active antiviral agent penciclovir (Clarke et al., 1995; Rashidi et al., 1997).
Although there has been a very high focus on the cytochrome P450 family of drug-metabolizing enzymes in the research and development of new drugs, there has been considerably less attention on AO. However, an increase in the prevalence of the use of aromatic azaheterocyclics as substituents in drug design has caused an increase in the importance of AO in drug metabolism (Pryde et al., 2010). When left unexamined in drug design, an impact of AO on the clearance (CL) of a new chemical entity can result in an unexpected low exposure in humans. Examples of instances in which human pharmacokinetics were unacceptable because it was not known that AO contributed to a large extent in metabolic clearance before administration to humans include carbazeran (Kaye et al., 1985), zoniporide (Dalvie et al., 2010), N8-(3-chloro-4-fluorophenyl)-N2-(1-methyl-4-piperidinyl)-pyrimido[5,4-d]pyrimidine-2,8-diamine (BIBX 1382) (Dittrich et al., 2002), and a ketolide antibiotic (Magee et al., 2009). It is also possible that AO-generated metabolites could be responsible for toxicity (Diamond et al., 2010). One of the challenges in drug discovery regarding AO is that enzyme expression in commonly used laboratory animal species (mouse, rat, and dog) differs from that in human. In particular, the dog does not express the AOX1 gene that is important in human (Terao et al., 2006).
Prediction of human in vivo clearance of new drug candidates is an important activity in drug discovery so that the pharmacokinetics in humans will be consistent with a reasonable dosing regimen (i.e., low hepatic first-pass metabolism that can result in good oral bioavailability; clearance that will yield a half-life that permits an appropriate dosing frequency). Methods to predict human clearance from in vitro metabolism data have been well established for the cytochrome P450 enzymes (Emoto et al., 2010; Obach, 2011), as well as the glucuronyl transferase enzymes (Kilford et al., 2009). However, quantitative prediction of human clearance for AO-metabolized agents has not been accomplished. This may be in part due to the distribution of AO in extrahepatic tissues including lung, gastrointestinal tract, and kidney (Pryde et al., 2010). Species-dependent tissue distribution also is confounding development of predictive tools for AO (Garattini and Terao, 2011). In a recent article, an in vitro-in vivo correlation approach has been described wherein 11 compounds with varying rates of AO-mediated clearance in humans were studied (Zientek et al., 2010). The investigators proposed that new compounds that are subject to AO-catalyzed metabolism could be placed into this correlation to gain an estimate of whether clearance would be unacceptably high, moderate, or low.
However, in the attempt to predict in vivo clearance for a new compound from in vitro data, it is important to measure not only the rate of metabolism but also the relative contribution that the enzyme (or enzyme family) makes to overall metabolism. Such information is important for predictions of clearance, drug interactions, and interpatient pharmacokinetic variability from in vitro data. In this report, we describe the development of a method whereby the relative contribution that AO makes to overall hepatic metabolic clearance in humans is quantified. We show that the AO time-dependent irreversible inhibitor, hydralazine (Johnson et al., 1985), can be used to selectively and completely inhibit AO in human hepatocytes without inhibiting P450 enzymes.
Materials and Methods
Materials.
Hydralazine, zaleplon, 3-descladinosyl-11,12-dideoxy-6-O-methyl-3-oxo-12,11-(oxycarbonyl-(1-(1R-(1,8-naphthyridin-4-yl)-ethyl)-azetidin-3-yl)-imino)-erythromycin A (PF-0945863), carbazeran, zoniporide, oxozoniporide, propranolol, midazolam, DACA, naloxone, (E)-3-(4-((2S,3S,4S,5R)-5-1-(3-chloro-2,6-difluorobenzyloxyimino)ethyl)-3,4-dihydroxytetrahydrofuran-2-yloxy)-3-hydroxyphenyl)-2-methyl-N(3aS,4R,5R,6S,7R,7aR)-4,6,7-trihydroxyhexahydrobenzo[d][1,3]dioxol-5-yl) acrylamide (PF-05218881) and dextromethorphan, were obtained from the Pfizer sample bank (Pfizer, Groton, CT). 1-Aminobenzotriazole (ABT) and O6-benzylguanine was purchased from Sigma-Aldrich (St. Louis, MO). Desethylzaleplon was purchased from Santa Cruz Biotechnology Inc. (Santa Cruz, CA). Cryopreserved human hepatocytes from five individual donors, male and female, were purchased from In Vitro Technologies (Baltimore, MD) (batch 1: lots AGR, FKM, EHI, TDH, and ZFB) as well as a 10-donor mixed gender prepooled lot (batch 2: lot RTH). Both were stored in liquid nitrogen until use. Williams' E medium (WEM) (GIBCO custom formula) was supplemented with 26 mM NaHCO3 and filtered through a 0.22-μm sterile filtration flask. Pooled human liver cytosol was purchased from Celsis IVT (Chicago, IL). Pooled human liver microsomes were purchased from BD Biosciences (San Jose, CA). All other reagents and chemicals were of the highest purity available.
Biosynthesis of Oxozaleplon.
Zaleplon (20 μM) was incubated with human liver cytosol (10 mg/ml) in a total volume of 10 ml of potassium phosphate buffer (0.1 M, pH 7.4). The incubation was performed at 37°C open to air for 3 h. The reaction was terminated by addition of 8 ml of CH3CN containing 0.32 ml of formic acid. The mixture was spun in a centrifuge at 1700g for 5 min. To the supernatant was added 0.1% formic acid to a final volume of 100 ml, followed by centrifugation at 40,000g for 30 min. The supernatant was applied through a Jasco HPLC pump at 0.5 ml/min onto a Varian Polaris C18 column equilibrated in 0.1% formic acid containing 10% CH3CN. After the entire supernatant was loaded, the column was moved to an HPLC-mass spectrometry system (Thermo Finnigan LTQ mass spectrometer with Surveyor HPLC system; Thermo Fisher Scientific, Waltham, MA), and the oxozaleplon product was eluted using a mobile phase gradient that commenced with 0.1% formic acid containing 10% CH3CN and was held for 5 min, followed by a linear gradient to 70% CH3CN at 50 min. The eluent was collected into 20-s fractions; fractions containing oxozaleplon (which eluted at ∼27 min) were pooled and evaporated by vacuum centrifugation, and the residue was reconstituted in 0.075 ml of [2H6]DMSO for analysis by quantitative proton NMR (Walker et al., 2011). The resulting stock solution was 1.86 mM and was diluted as appropriate to make standard curves for bioanalysis.
Substrate and Inhibitor Preparations.
All substrate stock solutions were prepared at 3 mM in DMSO. Further dilutions were made with WEM for a final substrate concentration of 1 μM (propranolol had a final concentration of 0.1 μM). For studies in which the zaleplon metabolites were quantified, the zaleplon final concentration was increased to 20 μM. The P450 inhibitor, ABT, was prepared at 400 mM in DMSO for a final concentration of 1 mM. Hydralazine was prepared in water before each study at various concentrations.
Hepatocyte Preparations.
Immediately before each experiment, the individual donor hepatocytes were thawed by gently shaking in a 37°C water bath for 90 s and then were pooled and diluted 25× the hepatocyte volume into prewarmed and O2/CO2 (95:5) bubbled WEM. The pooled mixture was centrifuged at 100g for 5 min at room temperature. After centrifugation, the supernatant was discarded, and the hepatocyte pellet was resuspended in WEM to either 0.75 × 106 or 2.25 × 106 cells/ml. The hepatocyte number and viability were determined using trypan blue exclusion staining in a hemocytometer. Cell preparations with viability greater than 80% were diluted with WEM, and by using a Thermo Labsystems Multidrop DW instrument (Thermo Fisher Scientific) 30 μl of cell suspension was added to individual wells of 96-well tissue culture treated polystyrene plates (final cell density was 0.5 × 106 or 1.5 × 106 cells/ml). For the donor variability study, individual donors were kept separate and prepared by the same method as the pooled hepatocytes.
Hepatocyte Incubations.
Cells were placed in a 37°C incubator under an atmosphere of O2/CO2 (95:5) with 95% relative humidity for 30 min. After the 30-min incubation, 15 μl of 3 μM substrate or substrate/inhibitor mix was added to individual wells using an Apricot Designs Personal Pipettor 550. Incubations were performed in triplicate and were initiated by the addition of substrate or substrate/inhibitor solution to the hepatocytes. Reactions were terminated at 0, 5, 15, 30, 60, 120, and 240 min by adding 135 μl of ice-cold CH3CN containing internal standard (100 ng/ml PF-05218881). After the termination of the reaction, plates were centrifuged at 3000g at 4°C for 5 min. The supernatants were transferred to 96-deep well plates for LC-MS/MS analysis.
LC-MS/MS Analysis.
Samples were analyzed by LC-MS/MS using a Shimadzu (Nakagyo-ku, Kyoto, Japan) quaternary HPLC pump with an Agilent 1100 series membrane degasser (Agilent Technologies, Palo Alto, CA) and Leap autosampler (CTC Analytics; LEAP Technologies Inc., Carrboro, NC) coupled to a PE Sciex API 4000 QTRAP mass spectrometer (Applied Biosystems/MDS Sciex, Foster City, CA). Electrospray ionization in positive mode with multiple reaction monitoring was used. Mass spectrometer parameters were individually optimized for each compound and internal standard (Table 1). Chromatographic separation was achieved using a Synergi Polar-RP column (4 μm 50 × 2.0 mm; Phenomenex, Torrance, CA). The mobile phases, both containing 0.1% formic acid, were water-CH3CN (95:5) (solvent A) and water-CH3CN (5:95) (solvent B). A linear gradient of solvent B from 5 to 95% was applied over 3.5 min on the column at a flow rate of 0.5 ml/min. The column was then re-equilibrated to initial conditions. The total sample analysis time was approximately 4 min. All analytes eluted between 1.5 and 2.5 min. A standard curve was used to quantify zaleplon, oxozaleplon, and desethylzaleplon. Linearity was observed between 0.5 and 25, 0.001 and 5, and 0.0025 and 25 μM, respectively. Acceptable assay performance was based on linearity throughout the dynamic range of the standard curve. In addition, standards were included only if within 20% of the nominal value. AB Sciex Analyst 1.4.2 software was used to analyze all data.
Compound mass spectrometer parameters
Calculations for Clint, app and fm(AO).
The area under the concentration-time curve [AUC(0–∞)] was calculated from the substrate depletion time course using the linear trapezoidal approximation and extrapolation from the last quantifiable time to infinity from the estimated half-life (t1/2). All CLint, app values were calculated as follows:
The t1/2 was estimated as ln2/slope, where the slope is that of the plot of the terminal elimination phase on a logarithmic scale. Fraction metabolized by AO [fm(AO)] was calculated by
in which CLint, app, hyd is the apparent intrinsic clearance in the presence of hydralazine, calculated as above.
Cytochrome P450 Inhibition Study.
The P450 inhibition assay used a cocktail of six probe substrates metabolized by major P450 isoforms and human liver microsomes to assess the inhibition potential of a test compound for each P450 isoform (Zientek et al., 2008).
Time-Dependent Inhibition of Aldehyde Oxidase by Hydralazine.
A progress curve approach was used to determine the time-dependent inhibition of human AO activity by hydralazine. Incubation mixtures consisted of pooled human liver cytosol (5 mg/ml), zoniporide (20 μM), and hydralazine (0–500 μM) in 0.1 M potassium phosphate, pH 7.4. Reactions were commenced with the addition of cytosol and incubated at 37°C. At times of 0, 2.5, 5, 10, 15, 20, 25, 30, 40, 50, and 60 min, an aliquot (0.075 ml) of the incubation mixture was removed and added to 0.025 ml of CH3CN containing 5% formic acid and 0.02 mM metoprolol as an internal standard. The mixtures were centrifuged (Eppendorf; 14,000 rpm, 5 min), and supernatants were analyzed by HPLC-mass spectrometry. The injection volume was 10 μl. The HPLC system consisted of an Inertsil C8 column (100 × 4.6 mm; 3 μm) equilibrated in 0.1% formic acid at a flow rate of 0.8 ml/min. This mobile phase composition was maintained for 1.5 min followed by a linear increase in CH3CN composition to 80% at 6 min, held at this condition for 1 min, and followed by a 3-min reequilibration period at initial conditions. The eluent was introduced into the source of a Thermo Finnigan LTQ mass spectrometer operated in the positive ion mode. The mass transitions of m/z 337 → 278 and 268 → 191 were monitored for oxozoniporide and internal standard, respectively. Oxozoniporide was quantitated from a standard curve ranging from 0.05 to 10 μM. Data analysis was done for progress curve analysis as described in Morrison and Walsh (1988).
Metabolite Profiles of Drugs in Human Hepatocytes.
O6-Benzylguanine, PF-0945863, zaleplon, zoniporide, DACA, carbazeran, and propranolol were incubated at 10 μM with pooled human hepatocytes (∼750,000 cells/ml) in 2 ml. Incubations were performed at 37°C under an atmosphere of O2/CO2 (95:5). Aliquots were removed at time 0, 30 min, 1 h, or 3 h (depending on the expected turnover of the individual drug) and terminated with 5 volumes of CH3CN. The mixture was centrifuged at 1700g for 5 min, and the supernatant was removed under nitrogen. The residue was reconstituted in 0.2 ml of 1% formic acid and injected onto a Thermo Finnigan Surveyor high-performance liquid chromatograph in line with a diode array detector (200–400 nm) and ion trap mass spectrometer (LTQ). The HPLC system consisted of a Varian Polaris C18 column (4.6 × 250 mm; 5 μm) equilibrated in 0.1% formic acid containing 5% CH3CN at a flow rate of 0.8 ml/min. This mobile phase condition was held for 5 min followed by a linear gradient to 80% CH3CN at 30 min, which was held for 5 min more before returning to initial conditions to re-equilibrate the column. The LTQ was operated in the positive ion mode with data-dependent scanning; tune file parameters and collision energies were optimized for each compound based on the response for the protonated molecular ion and fragment ions, respectively.
Results
Metabolism of Zaleplon in Human Hepatocytes.
The metabolism of zaleplon in cryopreserved pooled human hepatocytes was examined to determine the enzyme kinetic parameters for the formation of the AO metabolite oxozaleplon and the P450 metabolite desethylzaleplon. These are the two major metabolic pathways reported for this drug (Kawashima et al., 1999). Preliminary experiments had determined 30 min to be an optimal incubation time for formation of both metabolites. Clinical concentrations for zaleplon are in the low micromolar range (Greenblatt et al., 1998); however, it was determined from this study that a 20 μM zaleplon incubation concentration was necessary to produce both metabolites in a readily measurable amount. Zaleplon concentrations greater than 50 μM did not result in a corresponding increase in either metabolite. Because 20 μM zaleplon produced both metabolites in sufficient quantity for quantification throughout the incubation time course, this concentration was selected for subsequent experiments. Although zaleplon kinetics have been reported (Lake et al., 2002) in human liver cytosol and liver slices, to date, the enzyme kinetics of zaleplon metabolism in cryopreserved human hepatocytes have not been reported. However, the data shown in Fig. 1 preclude making reliable estimates of KM and Vmax because of the apparent complexity of the v versus [S] relationship.
Enzyme kinetics of zaleplon metabolism in a five-donor pool of cryopreserved human hepatocytes (0.5 million cells/ml) monitored for the formation of oxozaleplon (AO-mediated) and desethylzaleplon (P450-mediated).
Effect of Hydralazine and ABT on Zaleplon Metabolism in Human Hepatocytes.
Zaleplon (20 μM) was incubated with hydralazine between 0.1 and 200 μM in a five-donor pool of human hepatocytes for 30 min to determine the inhibitory effect on the formation of the AO-mediated oxozaleplon metabolite and the P450-mediated desethylzaleplon metabolite (Fig. 2). Oxozaleplon decreased as a percentage of control with increasing hydralazine concentration. Greater than 90% inhibition of AO-mediated metabolite formation resulted from 25 μM hydralazine, whereas the desethylzaleplon P450-mediated metabolite formation was not inhibited. At concentrations greater than 100 μM, readily measurable inhibition of desethylzaleplon formation was observed. At 50 μM hydralazine, there was a slight effect on the CYP3A-catalyzed deethylation reaction; thus, it is concluded that concentrations should not exceed this value to selectively inhibit aldehyde oxidase.
Inhibition of zaleplon (20 μM) metabolism by hydralazine in a five-donor pool of cryopreserved human hepatocytes (0.5 million cells/ml); error bars represent the S.D. of n = 3 data points.
The pan-P450 inhibitor ABT was coincubated with zaleplon to better characterize the reliability of assessing a specific AO inhibitor in hepatocytes (Fig. 3). Zaleplon (20 μM) was coincubated with between 0 and 1.5 mM ABT. Desethylzaleplon decreased with increasing ABT concentration. P450 metabolite formation was inhibited >90% in the presence of 1 mM ABT. The AO-derived oxozaleplon metabolite formation was unaffected by ABT at the concentrations tested. This result confirms that oxozaleplon is generated by AO and desethylzaleplon is generated by P450 and that these pathways can be useful for probing the selectivity of inhibitors of these two enzymes.
Inhibition of zaleplon (20 μM) metabolism by ABT in a five-donor pool of cryopreserved human hepatocytes (0.5 million cells/ml); error bars represent the S.D. of n = 3 data points.
Effect of Hydralazine on Individual Human Cytochrome P450 Enzymes.
Whereas the experiment described above shows that hydralazine does not affect CYP3A-catalyzed zaleplon N-deethylation, it is important to determine the potential potency for inhibiting other P450 enzymes. Across the major drug-metabolizing P450 enzymes, hydralazine at 25 μM showed little to no inhibition (Table 2). When hydralazine was tested at 50 μM, the percentage of control for CYP2D6 and CYP3A was reduced to 77 and 76%, respectively.
Cytochrome P450 inhibition by hydralazine in human liver microsomes
The following probe substrates were used: 10 μM phenacetin for CYP1A2, 5 μM paclitaxel for CYP2C8, 5 μM diclofenac for CYP2C9, 40 μM S-mephenytoin for CYP2C19, 5 μM dextromethorphan for CYP2D6, and 2 μM midazolam for CYP3A.
Inactivation of Human Aldehyde Oxidase by Hydralazine.
In a previous report, it was proposed that hydralazine was a time-dependent inhibitor of aldehyde oxidase using guinea pig enzyme (Critchley et al., 1994). However, this was not known for human aldehyde oxidase; thus, measurement of the time dependence and determination of inactivation kinetic parameters was undertaken for the human enzyme. Using the oxidation of zoniporide to oxozoniporide as a probe reaction (Dalvie et al., 2010) and pooled human cytosol as the source of enzyme, the inactivation kinetics of aldehyde oxidase by hydralazine were determined. The maximum inactivation rate constant (kinact) was 0.063 ± 0.007 min−1, and the concentration yielding 50% of the maximum inactivation rate (KI) was 83 ± 27 μM. This was determined using a progress curve approach, in which substrate and inactivator are simultaneously incubated (Fig. 4).
Inactivation of AO in human cytosol by hydralazine. Top, time course of formation of oxozoniporide; bottom, relationship between inactivation rate constants and hydralazine concentration to determine KI and kinact that was derived from the data in the top panel according to the method described by Morrison and Walsh (1988).
Determination of Fraction Metabolized by Aldehyde Oxidase in Pooled Human Hepatocytes.
The use of hydralazine to determine fraction metabolized for compounds that are metabolized by AO was tested using 10 compounds with diverse enzymatic pathways (Table 3). O6-Benzylguanine, PF-0945863, zaleplon, zoniporide, DACA, and carbazeran were selected because these drugs have been shown to possess an AO contribution to their total clearance (Zientek et al., 2010). The results showed that hydralazine can have a substantial effect on drugs possessing an aldehyde oxidase component to their metabolic clearance (Table 3). Two different batches of pooled hepatocyte lots as well as four individual lots of hepatocytes were examined to assess interlot variability (Fig. 5). Some minor differences were observed in sensitivity to hydralazine, with batch 1 demonstrating an apparent greater effect of 25 μM hydralazine, whereas batch 2 required the use of 50 μM hydralazine.
Metabolic pathway, apparent intrinsic clearance, and fm(AO) identified for 10 selected compounds and compared in two human hepatocyte batches
Metabolic pathways were confirmed using biotransformation. Human hepatocytes were suspended at 1.5 million cells/ml. Termination time points = 0, 5, 15, 30, 60, 120, and 240 min; n = 3/time point. Clint, app values were calculated from averaged AUC0–∞ extrapolated data. Batch 1: pooled lots AGR, FKM, EHI, TDH, and ZFB; batch 2: lot RTH.
Effect of hydralazine (25 μM) on individual lot and pooled hepatocytes (1.5 million cells/ml) relative to the control (0 μM hydralazine); error bars represent the S.D. of n = 3 data points.
Naloxone, propranolol, midazolam, and dextromethorphan were selected because they have no AO-mediated clearance. These compounds were considered negative controls that could expose an effect by hydralazine on other metabolic enzymes. Dextromethorphan is primarily metabolized by CYP2D6 (Gorski et al., 1994), which is one of the enzymes that can be inhibited by hydralazine (Table 2), and propranolol has a component of its metabolism catalyzed by CYP2D6 (Yoshimoto et al., 1995). Hydralazine has an effect on intrinsic clearance of propranolol and dextromethorphan, whereas a minimal effect was notable for midazolam or naloxone.
To confirm that these AO substrates have other metabolic pathways besides the aldehyde oxidase-mediated reactions, the profile of metabolites was qualitatively determined in human hepatocytes (Table 3). DACA, zaleplon, and PF-0945863 all demonstrated other types of oxidative pathways commonly associated with P450 enzymes, zoniporide demonstrated a hydrolysis reaction (as described previously) (Dalvie et al., 2010), and carbazeran demonstrated a considerable extent of direct glucuronidation (presumably on the phthalazine nitrogen). Only O6-benzylguanine appeared to demonstrate a single metabolite that is presumably generated by aldehyde oxidase, but this oxidation could also possibly be catalyzed by other enzymes (e.g., xanthine oxidase and P450s). The large effect of hydralazine on O6-benzylguanine intrinsic clearance supports the fact that AO is the dominant enzyme involved in its clearance.
Discussion
Although aldehyde oxidase has been an enzyme known to be involved in the metabolism of some drugs for several years, it has been gaining importance in drug metabolism over recent years (Pryde et al., 2010). This has been posing new challenges in drug design, because methods for predicting various human pharmacokinetic attributes (e.g., clearance, drug-drug interactions, and interpatient variability), which have been reasonably well established for compounds metabolized by cytochrome P450 enzymes (Houston, 1994; McGinnity et al., 2004), are not well known for drugs metabolized by aldehyde oxidase. In a recent study, Zientek et al. (2010) proposed a correlative method for categorizing new compounds shown to be metabolized by aldehyde oxidase as potentially high-, moderate-, or low-clearance drugs. That method used human liver cytosol or S9 fraction as a source of enzyme, measurement of in vitro CLint, and comparison with a set of 11 drugs known to be metabolized by aldehyde oxidase and for which human pharmacokinetic data were available. By comparison of the CLint value for a new compound with that for the 11 known drugs, the in vivo CL can be predicted, albeit with low precision. More recently, Hutzler et al. (2012) extended this type of approach to human hepatocytes as an in vitro system and showed quantitative prediction of clearance by aldehyde oxidase for substrates of high clearance.
In addition to the prediction of human CL for aldehyde oxidase substrates, it is also important to understand the relative contribution of this enzyme to overall clearance. This is essential for understanding of the potential for interindividual variability in pharmacokinetics that can arise by interindividual differences in enzyme expression or drug-drug interactions. Furthermore, it has been reported recently that human aldehyde oxidase is subject to genetic polymorphisms that can have an impact on activity (Hartmann et al., 2012), thus potentially serving as a source of interindividual variability. The greater percentage to which a specific enzyme contributes to clearance, the greater the potential impact that variability in the activity of that enzyme will have on interindividual variability in clearance. For example, it has been previously estimated that zaleplon is metabolically cleared by both aldehyde oxidase and cytochrome P450 3A4 by approximately a two-thirds/one-third ratio (Renwick et al., 2002). Thus, even if CYP3A4 were completely inhibited in vivo, the exposure to zaleplon would only increase by approximately 30%, which is what is observed with coadministration of erythromycin, a potent CYP3A4 inhibitor (Sonata Product Label, http://www.accessdata.fda.gov/drugsatfda_docs/label/2007/020859s011lbl.pdf). Thus, the aldehyde oxidase route of clearance serves to blunt the effect of a potent CYP3A4 inhibitor. It is therefore important to be able to predict the relative contribution of aldehyde oxidase to the overall clearance of individual drugs. The development of an in vitro method to make this prediction was the objective of the studies described in this article. To develop an in vitro method to predict the impact of aldehyde oxidase to overall clearance, two elements are important: 1) an in vitro system that possesses aldehyde oxidase activity within as complete a complement of drug-metabolizing enzymes as is possible and 2) a selective tool that will knock out aldehyde oxidase activity with acceptable selectivity. Pooled cryopreserved human hepatocytes used in suspension were selected for the in vitro system, on the basis of an assumption that aldehyde oxidase and other drug-metabolizing enzyme activities are representative of what is present in the human liver in vivo. The potential for extrahepatic clearance and the possibility that relative enzyme activities significantly change throughout the tissue acquisition, cell preparation, storage, and in vitro incubation processes must be accepted as possible limitations (as they are for any drug metabolism study conducted in hepatocytes) (Akabane et al., 2012). As for a tool compound that will selectively inhibit aldehyde oxidase as completely as possible, several were considered. Menadione has been used extensively as a selective inhibitor of aldehyde oxidase relative to the related enzyme xanthine oxidase. However, menadione is subject to rapid metabolism and in preliminary experiments it also showed substantial inhibition of several cytochrome P450 activities and was thus not pursued further (data not shown). Raloxifene is a very potent uncompetitive inhibitor of human aldehyde oxidase in cytosol preparations (Obach, 2004); however, raloxifene has also been shown to be an inactivator of CYP3A4 (Chen et al., 2002) and was therefore deemed not selective enough to be used for this purpose. Hydralazine had been previously shown to be an inactivator of guinea pig aldehyde oxidase (Critchley et al., 1994) and was selected for further exploration as a selective inhibitor of human aldehyde oxidase that could be used in hepatocytes. It should be noted that during our investigations, another group showed that hydralazine could inhibit aldehyde oxidase in human hepatocytes (Hutzler et al., 2012), albeit it in that report it was not used for the estimation of fm(AO).
Our investigations showed that hydralazine possessed suitable properties as a selective inhibitor of aldehyde oxidase in human hepatocytes. It possessed little activity on the major drug-metabolizing P450 enzymes (Table 2). Using zaleplon oxidase and deethylase activities as simultaneous probes for aldehyde oxidase and cytochrome P450 activities, respectively, hydralazine at 25 μM demonstrated the necessary selectivity for the former enzyme (Figs. 2 and 3). Higher concentrations (i.e., ≥100 μM) started showing some effect on P450 activity. Hydralazine was shown to be a time-dependent inhibitor of human aldehyde oxidase (Fig. 4), as it had been previously shown to be for the guinea pig enzyme (Critchley et al., 1994). Overall, hydralazine demonstrated acceptable properties as an aldehyde oxidase-selective probe inhibitor in human hepatocytes.
Once the experimental conditions were established (i.e., pooled cryopreserved human hepatocytes as the in vitro system, hydralazine as the selective inhibitor at 25 μM, and monitoring the decline in test compounds at 1 μM over 4 h as the endpoint measurement), we tested these conditions with a wide array of drugs known to be metabolized, at least in part, by aldehyde oxidase. Several drugs were shown to have half or more of their metabolic clearance catalyzed by aldehyde oxidase including O6-benzylguanine, PF-0945863, zaleplon, zoniporide, DACA, and carbazeran [i.e., fm(AO) ≥0.50] (Table 3). Four negative controls were also tested. Naloxone and midazolam, which are primarily metabolized by UDP-glucuronosyltransferase and P450 enzymes, respectively, were minimally affected by hydralazine. However, a more significant effect was observed on the consumption of propranolol and dextromethorphan (Table 3). Both of these compounds are metabolized by CYP2D6, and it was noted that among the P450 enzymes tested, hydralazine had the greatest effect on CYP2D6 (Table 2). Thus, it will be important that when this method is used to estimate fm(AO), a known CYP2D6 probe substrate also be included as a control and that if an effect of hydralazine is also observed on that drug, it is possible that the new compound(s) being tested may be a substrate of CYP2D6 and not aldehyde oxidase. This could be easily assessed by using a CYP2D6 inhibitor in a parallel incubation. A second potential limitation of the approach is that there must be measurable turnover of the test compound to determine the impact of hydralazine. To this end, we used a concentration of 1.5 × 106 cells/ml to reduce the observed half-life of substrates. This led to a measurable difference between the CLint for a substrate exposed to hydralazine and not exposed, thus enabling estimation of fm(AO). Compounds with very low CLint will not be readily addressed using a substrate depletion approach; an alternate experimental design will need to be used to estimate fm(AO), such as quantitative monitoring of the formation of metabolites in the presence and absence of hydralazine. Nevertheless, these findings support the use of hydralazine (at 25–50 μM) in human hepatocytes as an acceptable probe inhibitor of aldehyde oxidase.
Examination in a second batch of human hepatocytes showed that a greater concentration of hydralazine was needed (50 μM) and demonstrates the potential for interlot variability in the sensitivity to hydralazine and/or differences in the content of AO and various P450 isoforms in the hepatocytes. On the basis of our observations of the subtle differences in the effect of hydralazine among two hepatocyte batches, it is recommended that investigators using this method establish a concentration of the inhibitor between 25 and 50 μM that is optimal for their own hepatocyte preparations. This can be done using one or more of the AO substrates described in this work, along with a CYP2D6 substrate to ensure that too high a concentration that would sacrifice selectivity is not used.
The best way to determine whether the fm(AO) values estimated using this in vitro method match the contribution of AO to drug clearance in humans is to use data from a human metabolism and excretion study with radiolabeled substrate and to sum up the excretory metabolites that can be attributed to AO catalysis. However, among the six AO substrates that we tested, zoniporide is the only one that also has such clinical metabolism data reported (Dalvie et al., 2010). In that study, fm(AO) can be estimated to be between 0.52 and 0.69. This range correlates well with our in vitro estimates of 0.64 and 0.55. Despite this agreement, there is insufficient clinical metabolism and excretion data for drugs with known aldehyde oxidase-mediated clearance to assess the quantitative correlation of fm(AO) between in vitro and in vivo measurements. The reported fm(AO) are relative estimations based on the in vitro studies described.
In conclusion, a method whereby fm(AO) can be estimated for the metabolism of drugs in humans using hepatocytes with hydralazine as a selective inhibitor has been demonstrated. This method should prove useful in the design of new drugs when the prediction of human pharmacokinetic attributes such as clearance and potential for drug-drug interactions is important. It should also prove useful in the design of a drug-drug interaction study strategy, in that observation of a substantial contribution to total CL by aldehyde oxidase will have a bearing on the types of drug-drug interaction clinical studies that should be considered.
Authorship Contributions
Participated in research design: Strelevitz, Orozco, and Obach.
Conducted experiments: Strelevitz, Orozco, and Obach.
Performed data analysis: Strelevitz, Orozco, and Obach.
Wrote or contributed to the writing of the manuscript: Strelevitz, Orozco, and Obach.
Acknowledgments
We thank Mike West for support with the drug-drug interaction assay and Greg Walker for NMR quantification of oxozaleplon. We also thank Larry Tremaine and Louis Leung for their review of this work and helpful suggestions.
Footnotes
Article, publication date, and citation information can be found at http://dmd.aspetjournals.org.
ABBREVIATIONS:
- AO
- aldehyde oxidase
- CL
- clearance
- BIBX 1382
- N8-(3-chloro-4-fluorophenyl)-N2-(1-methyl-4-piperidinyl)-pyrimido[5,4-d]pyrimidine-2,8-diamine
- DACA
- N-[(2′-dimethylamino)ethyl]acridine-4-carboxamide
- PF-0945863
- 3-descladinosyl-11,12-dideoxy-6-O-methyl-3-oxo-12,11-(oxycarbonyl-(1-(1R-(1,8-naphthyridin-4-yl)-ethyl)-azetidin-3-yl)-imino)-erythromycin A
- PF-05218881
- (E)-3-(4-((2S,3S,4S,5R)-5-1-(3-chloro-2,6-difluorobenzyloxyimino)ethyl)-3,4-dihydroxytetrahydrofuran-2-yloxy)-3-hydroxyphenyl)-2-methyl-N(3aS,4R,5R,6S,7R,7aR)-4,6,7-trihydroxy hexahydrobenzo[d][1,3]dioxol-5-yl) acrylamide
- ABT
- 1-aminobenzotriazole
- WEM
- Williams' E medium
- HPLC
- high-performance liquid chromatography
- DMSO
- dimethyl sulfoxide
- P450
- cytochrome P450
- LC
- liquid chromatography
- MS/MS
- tandem mass spectrometry.
- Received February 22, 2012.
- Accepted April 20, 2012.
- Copyright © 2012 by The American Society for Pharmacology and Experimental Therapeutics
