Abstract
In vitro experiments were conducted to compare kinact, KI and inactivation efficiency (kinact/KI) of cytochrome P450 (P450) 2C9 by tienilic acid and (±)-suprofen using (S)-flurbiprofen, diclofenac, and (S)-warfarin as reporter substrates. Although the inactivation of P450 2C9 by tienilic acid when (S)-flurbiprofen and diclofenac were used as substrates was similar (efficiency of ∼9 ml/min/μmol), the inactivation kinetics were characterized by a sigmoidal profile. (±)-Suprofen inactivation of (S)-flurbiprofen and diclofenac hydroxylation was also described by a sigmoidal profile, although inactivation was markedly less efficient (∼1 ml/min/μmol). In contrast, inactivation of P450 2C9-mediated (S)-warfarin 7-hydroxylation by tienilic acid and (±)-suprofen was best fit to a hyperbolic equation, where inactivation efficiency was moderately higher (10 ml/min/μmol) and ∼3-fold higher (3 ml/min/μmol), respectively, relative to that of the other probe substrates, which argues for careful consideration of reporter substrate when mechanism-based inactivation of P450 2C9 is assessed in vitro. Further investigations into the increased inactivation seen with tienilic acid relative to that with (±)-suprofen revealed that tienilic acid is a higher affinity substrate with a spectral binding affinity constant (Ks) of 2 μM and an in vitro half-life of 5 min compared with a Ks of 21 μM and a 50 min in vitro half-life for (±)-suprofen. Lastly, a close analog of tienilic acid with the carboxylate functionality replaced by an oxirane ring was devoid of inactivation properties, which suggests that an ionic binding interaction with a positively charged residue in the P450 2C9 active site is critical for recognition and mechanism-based inactivation by these close structural analogs.
Cytochrome P450 (P450) enzymes are a superfamily of oxidative enzymes involved in the metabolism and clearance of a variety of structurally diverse therapeutic drugs. Within this family, P450 2C9 represents approximately 20% of the total hepatic P450 (Guengerich, 2002) and has recently been found to also represent approximately 15% of the total intestinal P450 (Paine et al., 2006). P450 2C9 is known to be involved in the metabolism of numerous acidic drugs, including nonsteroidal anti-inflammatory drugs such as flurbiprofen, naproxen, and diclofenac (Tracy et al., 1996, 1997; Klose et al., 1998), as well as the low therapeutic index drugs warfarin and phenytoin (Rettie et al., 1992; Giancarlo et al., 2001). The significant role that P450 2C9 plays in the metabolism of drugs (roughly 20% of drugs on the market), in addition to its polymorphic expression, makes this enzyme of profound interest from a drug-drug interaction perspective.
Cytochrome P450 enzymes are often implicated in drug-drug interactions (DDIs) because of the inhibition of metabolic clearance, ultimately leading to increased exposure and potentially serious adverse events in the clinic. The prediction of in vivo DDIs with new chemical entities has become an increasingly important component of drug metabolism research within drug discovery, and continuous improvements have been made in our ability to predict this risk in the clinic (Brown et al., 2006; Obach et al., 2006; Lu et al., 2007). However, in vitro-in vivo extrapolations of DDIs have been complicated by multiple binding regions within the enzyme active site that produce substrate-dependent inhibition, a characteristic initially observed with P450 3A4 (Stresser et al., 2000; Wang et al., 2000). Information from crystal structures of P450 2C9 (Williams et al., 2003; Wester et al., 2004) suggests that the active site is quite voluminous, which has been hypothesized to at least partially explain the numerous reports of complex kinetics that this enzyme displays (Korzekwa et al., 1998; Hutzler and Tracy, 2002; Hutzler et al., 2003). Kumar et al. (2006) have recently provided direct evidence of the potential complexities of P450 2C9 by testing a large set of competitive inhibitors against P450 2C9 using various probe substrates, indicating that inhibition potency was probe substrate-dependent. As a result, it was suggested that to fully evaluate inhibition of P450 2C9 for predictive purposes, multiple probe substrates should be used, a practice that has become common when inhibition of P450 3A4 is evaluated.
Structures of the P450 2C9 inactivators tienilic acid and (±)-suprofen and the P450 2C9 probe substrates (S)-flurbiprofen, diclofenac, and (S)-warfarin that were tested. *, chiral carbon in (±)-suprofen; arrows, major site of metabolism for the probe substrates.
Of equal importance to the identification of competitive inhibitors is the characterization of mechanism-based inactivation of cytochrome P450. Not evaluating the time dependence of P450 inhibition may lead to underprediction of DDI risk. This concern is indicated by the numerous methodologies reported for predicting clinical DDIs using in vitro inactivation data (Obach et al., 2007; Riley et al., 2007; Venkatakrishnan et al., 2007). One of the first examples of mechanism-based inactivation of P450 2C9 involved the thiophene derivative tienilic acid (Lopez-Garcia et al., 1994), which was withdrawn from the market because of hepatotoxicity, the apparent result of an immune response to alkylation by the P450 2C9 protein (Lecoeur et al., 1994). More recently, the structural analog (±)-suprofen was shown to also be a mechanism-based inactivator of P450 2C9 (O'Donnell et al., 2003). Tienilic acid and (±)-suprofen are reported to cause mechanism-based inactivation by a similar mechanism, leading to covalent modification of the P450 2C9 apoprotein within the active site (Koenigs et al., 1999; O'Donnell et al., 2003), making these inactivators ideal for evaluating the substrate dependence of inactivation kinetics. In the studies described herein, tienilic acid and (±)-suprofen (Fig. 1) were used to investigate the following: the dependence of in vitro inactivation kinetics of P450 2C9 by tienilic acid and (±)-suprofen on the reporter substrate used for measuring residual activity and the mechanism behind the superior P450 2C9 inactivation kinetics observed with tienilic acid relative to (±)-suprofen. It is anticipated that this work may improve our rationale regarding probe substrate selection for measuring P450 2C9 inactivation rates in vitro and concurrently provide additional insight into the structure-activity relationship for inactivation of P450 2C9 by carboxylate-containing molecules.
Materials and Methods
Chemicals. Potassium phosphate buffer, NADPH, NADP+, isocitric acid, isocitrate dehydrogenase, magnesium chloride, tolbutamide, diclofenac, and 4-hydroxydiclofenac were purchased from Sigma-Aldrich (St. Louis, MO). Tienilic acid, analog 1 [(2,3-dichloro-4-(oxiran-2-ylmethoxy)phenyl)(thiophen-2-yl)methanone], (±)-suprofen, (S)-flurbiprofen, 4′-hydroxyflurbiprofen, (S)-warfarin, and 7-hydroxywarfarin were obtained from the Pfizer internal compound library (St. Louis, MO). Cytochrome P450 2C9 Baculosomes were purchased from Invitrogen (Carlsbad, CA). All other chemicals were obtained from commercial sources and were of the highest purity available.
P450 2C9 Inactivation Assays. Human recombinant P450 2C9 was assayed for residual activity with probe substrate [(S)-flurbiprofen, diclofenac, or (S)-warfarin] (Fig. 1) after a primary incubation with various concentrations of tienilic acid, analog 1, or (±)-suprofen (0.5% total organic v/v) in triplicate in a 96-well plate format at 37°C. Incubations consisted of P450 2C9 Baculosomes (0.2 μM primary incubation and 0.01 μM secondary incubation) in 100 mM potassium phosphate buffer (pH 7.4) with a NADPH-regenerating system (3 mM MgCl2, 1 mM, NADP+, 5 mM isocitric acid, and 1 unit/ml isocitrate dehydrogenase). At selected time intervals [0–5 min for tienilic acid and analog 1 and 0–30 min for (±)-suprofen], 15-μl aliquots were taken out of the primary incubation mixture and transferred into 285 μl of a secondary incubation mixture containing (S)-flurbiprofen, diclofenac, or (S)-warfarin at 25 μM, a concentration >3 times the apparent Km for each reaction, to measure residual activity. The 20-fold dilution and saturating concentration of probe substrate in the secondary incubation served to minimize competitive inhibition and any further inactivation. Secondary incubations [4 min for diclofenac, 20 min for (S)-flurbiprofen, and 40 min for (S)-warfarin] were quenched with 300 μl of cold acetonitrile containing an internal standard (200 μM tolbutamide). Plates were then centrifuged at 3000g for 10 min and metabolite formation was analyzed by liquid chromatography/tandem mass spectrometry (LC-MS/MS) as described below. Observed rates of inactivation (kobs) were calculated for each inactivator concentration by measuring loss in enzyme activity, determined by linear regressions of the natural log (ln) percentage remaining activity as a function of primary incubation time using GraphPad Prism (GraphPad Software Inc., San Diego, CA), and subsequently replotted as kobs as a function of inactivator concentration to estimate inactivation rate constants (KI and kinact) for each inactivator-substrate combination by nonlinear regression using Sigma Plot 9.0 Systat software (Systat Software Inc., San Jose, CA) using either a hyperbolic (eq. 1) or Hill equation (eq. 2): where kobs represents the estimated rate constant for inactivation, kobs[I] = 0 is the apparent inactivation rate measured in the absence of inactivator, kinact is the theoretical maximal rate of inactivation, assessed in vitro, KI is the concentration of inactivator that yields an inactivation rate that equals half kinact, [I] is the concentration of inactivator in the primary incubation [tienilic acid or (±)-suprofen], and n is the Hill coefficient. Comparisons of goodness of the fits were determined by examination of the residuals, R2 value, residual sum of squares, and inspection of Eadie-Hofstee plots.
Spectrophotometric Studies. Spectral binding titration studies were carried out with recombinant P450 2C9 Baculosomes (0.2 μM) in 100 mM potassium phosphate buffer (pH 7.4) containing 20% glycerol, 0.3% CHAPS, 0.1 mM EDTA, and 0.2 mM dithiothreitol. The mixture was evenly divided into two 10-mm black-walled cuvettes, and experiments were performed at room temperature by titrating in aliquots of tienilic acid (0–70 μM) or (±)-suprofen (0–150 μM) into a sample cuvette, with vehicle controls added to the reference cuvette (dimethyl sulfoxide <1% v/v). After each addition, cuvette contents were mixed manually and allowed to sit for 1 min to allow the samples to reach equilibrium. Difference spectra were recorded between 350 and 500 nm using a U3300 dual-beam spectrophotometer Hitachi (Danbury, CT) with UV Solutions 1.2 software. The spectral binding constants (Ks) were determined by monitoring the type I spectral changes (increase in high spin state at 390 nm with a concomitant decrease in low spin state at 420 nm) in the difference spectra as the enzyme was titrated with each compound. The difference in absorbance between the peak and trough was then plotted versus the titrant concentration and fit to a hyperbolic curve function (eq. 3) for estimation of the Ks value using GraphPad Prism software: where Bmax is the maximal binding as measured by the maximal difference in absorbance between 420 and 390 nm and Ks is the concentration of ligand that yields half Bmax.
Separation of (±)-Suprofen Enantiomers. Separation of (±)-suprofen enantiomers was performed by preparing a 10 mg/ml stock solution of (±)-suprofen in ethanol and delivering 1 ml/injection onto a CHIRALPAK AD-H 30 × 250 mm semiprep column (Chiral Technologies, West Chester, PA) with 20% ethanol solvent flowing isocratically at 70 ml/min using an supercritical fluid chromatography preparation system (Berger SFC MultiGram II; Mettler-Toledo Autochem, Newark, DE). Flow was introduced directly into a UV diode array detector set at 254 nm, followed by an 1100 single-quadrupole mass spectrometer (Agilent Technologies, Santa Clara, CA) fitted with an APCI source operating in positive ion mode (corona voltage 4 V and temperature 325°C). Characterization of (R)- and (S)-isomers after separation was performed simply by comparison of retention time. The (R)-isomer was characterized by a retention time of 4.45 min [agreeing with that of the available (R)-suprofen standard], as well as by a protonated ion of [M + H]+ at m/z 261. The (S)-isomer of suprofen eluted at 4.94 min and was also characterized by UV absorption and mass spectrometry as indicated above. Fractions were collected and dried down in vacuo, yielding 35 mg of (R)-suprofen and 33 mg of (S)-suprofen. Absorption and mass spectral data were processed using ChemStation software (version B.03.01).
Metabolic Stability. The metabolism of tienilic acid, (±)-suprofen, (R)-suprofen, and (S)-suprofen (1 μM) by recombinant P450 2C9 (0.04 μM) was monitored by substrate depletion in 100 mM potassium phosphate buffer (pH 7.4) at 37°C. Reactions were initiated by the addition of NADPH (2 mM) and at selected time intervals (0–30 min) 50-μl aliquots were taken out of the incubation mixture and quenched by addition to 150 μl of cold acetonitrile containing an internal standard (200 μM tolbutamide). Plates were then centrifuged at 3000g for 5 min, and substrate depletion was analyzed by LC-MS/MS as described below. The in vitro half-life (t1/2) for each substrate was calculated by linear regression of the ln percentage remaining of substrate as a function of incubation time using GraphPad Prism software according to equation t1/2 = ln 2/slope.
Partition Ratio. The partition ratio (the molar ratio of inactivator metabolized per P450 2C9 inactivated) was estimated using the titration method, similar to previous work by Lin et al. (2002) and Jushchyshyn et al. (2003). Similar to methods described above, P450 2C9 Baculosomes (0.2 μM primary incubation) were incubated in the presence of tienilic acid (0–60 μM) or (±)-suprofen (0–160 μM) for 30 min to ensure completion of the inactivation process. Aliquots (15 μl) were then removed from the primary incubation and transferred into 285 μl of the secondary incubation mixture containing 25 μM (S)-flurbiprofen to measure residual activity. Metabolite formation was analyzed by LC-MS/MS as described above. The turnover number was estimated by plotting the percentage of remaining activity as a function of the molar ratio of inactivator to P450 2C9, followed by regression of the linear portions of both the low and the high [inactivator]/[P450 2C9] ratios and extrapolation of their intersection to the x-axis. The turnover number includes the 1 molecule of inactivator [tienilic acid or (±)-suprofen] that inactivates the P450 enzyme and is thus equal to the partition ratio + 1.
LC-MS/MS Analysis. Metabolite formation was analyzed on a PE Sciex API-3000 triple quadrupole instrument. The mass spectrometer was equipped with an electrospray ionization interface connected in line with a Shimadzu LC20AD pump and a CTC PAL (Leap Technologies, Carrboro, NC) autosampler. Analytes were separated using a Zorbax 3.5-μm Eclipse Plus C18 2.1 × 50 mm column with a gradient elution profile. Mobile phase was flowing at 0.4 ml/min, and the gradient was initiated and held for the first 0.3 min at 95% A-5% B (A, 0.1% formic acid in H2O; B, 0.1% formic acid in acetonitrile) and was then ramped linearly to 5% A-95% B over the next 2.5 min and held for 1.2 min. The profile was then immediately returned to initial conditions and allowed to re-equilibrate for 2 min. The source temperature was set to 400°C, and mass spectral analyses were performed using multiple reaction monitoring, with transitions for 4′-hydroxyflurbiprofen (m/z 259.1 → 215.2), 4-hydroxydiclofenac (m/z 309.9 → 265.8), 7-hydroxywarfarin (m/z 323.1 → 176.9), and tienilic acid (m/z 328.9 → 270.8) and tolbutamide (m/z 269.1 → 170.0) with a turbo ion spray source in negative ionization mode (–3.5 kV spray voltage). Multiple reaction monitoring for suprofen (m/z 261.06 → 111.0) was performed with a turbo ion spray source in positive ionization mode (5 kV spray voltage). All data were analyzed using PE Sciex Analyst 1.4.1 software.
Results
Inactivation of P450 2C9. In effort to compare kinetic rates of inactivation with different reporter substrates, both tienilic acid and (±)-suprofen were preincubated with P450 2C9 Baculosomes, and residual enzymatic activity was measured using (S)-flurbiprofen, diclofenac, or (S)-warfarin as probe substrates. As shown in Table 1, inactivation by tienilic acid and (±)-suprofen was described by comparable kinact and KI values against P450 2C9-mediated (S)-flurbiprofen and diclofenac hydroxylation. In addition, the kinetics of inactivation were described by a sigmoidal kinetic profile (Fig. 2, A–D). However, inactivation of (S)-warfarin 7-hydroxylation by both tienilic acid and (±)-suprofen was best fit to a standard hyperbolic equation (Fig. 2, E and F), which yielded noticeably lower KI estimates [12.5 and 6.7 μM with tienilic acid and (±)-suprofen, respectively], resulting in moderately higher inactivation efficiencies, in particular for (±)-suprofen (Table 1). Tienilic acid was a markedly better inactivator of P450 2C9 compared with (±)-suprofen, independent of probe substrate, with a 4- to 8-fold higher inactivation efficiency (Table 1). It is interesting to note that a close structural analog of tienilic acid with the carboxylate functionality replaced by an oxirane ring (analog 1) did not demonstrate inactivation properties against P450 2C9 (Fig. 3).
Kinetic parameters describing the inactivation of P450 2C9 by tienilic acid and (±)-suprofen when different probe substrates are used to measure residual activity
Data were fit to both a hyperbolic equation (eq. 1) and the Hill equation (eq. 2), as described under Materials and Methods. Efficiency of inactivation was calculated as the ratio of kinact to KI.
Spectrophotometric Analysis. The titration of both tienilic acid and (±)-suprofen with P450 2C9 Baculosomes resulted in a type I difference spectrum, characterized by an absorbance peak at ∼390 nm, and a trough at ∼420 nm (Fig. 4, A and B, insets). When data [ΔAbs 420–390 nm versus tienilic acid or (±)-suprofen] were fitted to eq. 3, a spectral binding affinity constant (Ks) was estimated to be 2 μM for tienilic acid (Fig. 4A) and 21 μM for (±)-suprofen (Fig. 4B).
Metabolism by P450 2C9 Baculosomes. When tienilic acid and (±)-suprofen were incubated with P450 2C9 Baculosomes, there was a substantial difference observed in their metabolic rates, as measured by substrate depletion (Fig. 5). The half-life of tienilic acid was approximately 5 min, whereas for (±)-suprofen, a 50-min half-life was observed. To examine potential enantiomeric differences between (R)- and (S)-suprofen metabolism (and potentially, inactivation rates), the enantiomers were separated, and their substrate depletion rates were compared. There was little evidence for enantioselective metabolism of suprofen, as the half-lives of both enantiomers were comparable (35–38 min), although slightly shorter than that for (±)-suprofen (50 min).
Partition Ratio. To compare the partition ratios (molar ratio of inactivator metabolized per P450 2C9 inactivated) between tienilic acid and (±)-suprofen in the same P450 2C9 enzymatic system, a titration method was used. Tienilic acid (0–60 μM) and (±)-suprofen (0–160 μM) were preincubated for 30 min to ensure complete inactivation of (S)-flurbiprofen hydroxylation (see Materials and Methods). With use of the titration methodology demonstrated in Fig. 6, the turnover number (the total number of metabolic events that results in inactivation; partition ratio + 1) for tienilic acid inactivation was ∼35 (partition ratio of 34), whereas during inactivation by (±)-suprofen, the turnover number was estimated to be 99 (partition ratio of 98).
Kinetic plots demonstrating observed rates of inactivation (Kobs) of P450 2C9 by tienilic acid and (±)-suprofen with diclofenac (A and B), (S)-flurbiprofen (C and D), and (S)-warfarin (E and F) as probe substrates. Data were fit to either eq. 1 (hyperbolic) or eq. 2 (Hill equation).
Discussion
Tienilic acid and (±)-suprofen are thiophene-containing mechanism-based inactivators of P450 2C9, the former being withdrawn from the market because of hepatotoxicity, proposed to be related to an immune response resulting from covalent modification of the P450 2C9 protein (Lecoeur et al., 1994). The bioactivation mechanism for both tienilic acid and (±)-suprofen involves oxidation of the thiophene ring system (Lopez-Garcia et al., 1994; Koenigs et al., 1999; O'Donnell et al., 2003). It is interesting to note that although these molecules are structurally related and seem to be bioactivated via similar mechanisms, tienilic acid is noticeably more efficient at inactivating P450 2C9 than (±)-suprofen. In separate but related work, the inhibition potency of numerous competitive inhibitors of P450 2C9 was shown to be dependent on the marker substrate used to measure enzymatic activity (Kumar et al., 2006). In addition, Cheng et al. (2007a,b) observed that the mechanism of inhibition of P450 2B4 by the mechanism-based inhibitor 2-ethynylnaphthalene depended on the probe substrate. Thus, the focus of our research was the use of tienilic acid and (±)-suprofen as in vitro tools to investigate whether using different probe substrates to measure residual activity in vitro would have an impact on mechanism-based inactivation kinetic rates/efficiencies and also to investigate mechanistically why the inactivation kinetics of tienilic acid and (±)-suprofen are so disparate, despite their close structural similarities.
To investigate substrate-dependent inactivation kinetics with P450 2C9, three reporter substrates were selected: (S)-flurbiprofen, diclofenac, and (S)-warfarin. These substrates were selected strategically as representatives of the three subgroups designated by a cluster analysis of potency of competitive inhibitors of P450 2C9 (Kumar et al., 2006). Whereas inactivation of both (S)-flurbiprofen and diclofenac 4-hydroxylation was best described by a sigmoidal model (Hill equation) and similar inactivation kinetics, inactivation of (S)-warfarin 7-hydroxylation was best described by a classic hyperbolic equation (Table 1), which yielded KI estimates that were considerably lower than those estimated for inactivation of both (S)-flurbiprofen and diclofenac 4-hydroxylation (Table 1). Although the lower estimates of KI with (S)-warfarin as a probe substrate for these inactivation studies are consistent with that reported by Kumar et al. (2006), in which as much as 20-fold lower inhibition constants (KI) were observed against (S)-warfarin for several competitive inhibitors, the magnitude of the observed difference here was not as substantial. Unfortunately, the question remains whether the observed kinetic differences are significant enough to affect predictions of clinical drug-drug interactions using established methods (Obach et al., 2007). However, this type of analysis is difficult with the selected probe substrates from these studies, as (S)-warfarin is the only probe that is primarily cleared by P450 2C9 in humans, with an estimated fraction metabolized through P450 2C9 (fm2C9) of 0.91 (Obach et al., 2007). In addition to P450 2C9-mediated hydroxylation, both (S)-flurbiprofen and diclofenac are cleared roughly 50% by direct glucuronidation of the carboxylate functional groups (Kumar et al., 2002; Mano et al., 2007). In the absence of a clear comparison to understand the clinical implications of these data, it may be prudent to consider including (S)-warfarin as a reporter substrate in vitro when inactivation of P450 2C9 is evaluated, because this seems to be the most sensitive probe substrate and is likely to be one of the clinical interactions of most concern because of the known low therapeutic index of this drug.
Inactivation of P450 2C9 by tienilic acid and a close structural analog (analog 1) with the carboxylic acid replaced by an oxirane ring, demonstrating the importance of the carboxylic acid to the inactivation properties of tienilic acid.
In addition to the kinetic differences observed for reporter substrates, it was clear that tienilic acid was substantially more effective for inactivating P450 2C9 than (±)-suprofen (Table 1). This was not a surprise, as inactivation kinetics for these compounds have been reported previously, although from separate studies using different enzyme preparations (Lopez-Garcia et al., 1994; O'Donnell et al., 2003; McGinnity et al., 2006). In this direct comparison, we report that the enhanced efficiency of P450 2C9 inactivation by tienilic acid relative to (±)-suprofen is independent of which reporter substrate is used. In consideration of their structural similarities, these differences are intriguing from a structure-activity relationship perspective, and, to our knowledge, the underlying causes of these rate differences have yet to be investigated. First, to investigate and compare binding affinities of tienilic acid and (±)-suprofen, difference spectra were collected, which indicated that both inactivators bind in a type I fashion, with characteristic peak and trough absorbances at 390 and 420 nm, respectively. This binding is consistent with both tienilic acid and (±)-suprofen being substrates for P450 2C9. However, it is clear upon visual inspection of the spectral data in Fig. 4 that (±)-suprofen binds with substantially less affinity, supported by a higher spectral binding affinity constant (Ks) of 21 μM, compared with 2 μM for tienilic acid. It is worth noting that the binding kinetics of tienilic acid and (±)-suprofen were not described by a sigmoidal profile, thus indicating that the sigmoidicity observed for the inactivation of (S)-flurbiprofen and diclofenac 4-hydroxylation was probably not due to the initial binding properties of tienilic acid and (±)-suprofen alone. Last, spectral binding data were consistent with data from incubations in P450 2C9 Baculosomes showing that tienilic acid is markedly less metabolically stable than (±)-suprofen, with an in vitro disappearance half-life of 5 min, compared with a 50-min half-life for (±)-suprofen (Fig. 5). A higher metabolic rate for tienilic acid suggests that it is probably bioactivated more rapidly than is (±)-suprofen, perhaps explaining the disparate inactivation rates displayed by these two inactivators.
Other factors potentially contributing to the inactivation kinetic differences of tienilic acid and (±)-suprofen with P450 2C9 were also investigated. The enantiomers of (±)-suprofen were separated and tested to determine whether there may be enantioselective metabolism by P450 2C9, which has been observed for other racemic substrates. For example, (S)-warfarin is a substrate for P450 2C9 (Rettie et al., 1992), whereas (R)-warfarin is a substrate for P450 1A2 and P450 3A4 (Yamazaki and Shimada, 1997). As shown in Fig. 5, there does not seem to be enantio-selective metabolism of (±)-suprofen. Although we did not test both suprofen isomers for inactivation of P450 2C9, one can infer on the basis of a similar metabolic half-life (35–38 min) that one isomer probably does not substantially inhibit or protect the P450 2C9 enzyme from inactivation by the other when the racemic mixture was used in our inactivation studies. The possibility of some minor effects cannot be completely discounted, however, because the substrate depletion half-life is slightly longer for (±)-suprofen (50 min). Partition ratios of tienilic acid and (±)-suprofen were also compared using (S)-flurbiprofen as the probe substrate. In theory, a higher partition ratio (more metabolite formed per enzyme inactivated) would lead to less efficient inactivation. As indicated in Fig. 6, the partition ratio of (±)-suprofen is estimated to be ∼98, whereas that for tienilic acid is ∼34. It seems then that a higher partition ratio for (±)-suprofen may play a role in less efficient inactivation, although it seems more likely that slower metabolism and thus bioactivation is the primary reason for the decreased inactivation rates relative to tienilic acid.
In an effort to more completely understand the observed differences in inactivation rates with tienilic acid and (±)-suprofen, as well as the probe substrate-dependent kinetic profiles of inactivation, computational studies may be necessary using the crystal structure of P450 2C9 (Wester et al., 2004). Similar approaches would also have to be taken to understand whether existing information from the crystal structure of P450 2C9 could explain the fact that both tienilic acid and (±)-suprofen seem to inactivate (S)-warfarin 7-hydroxylation with a distinct hyperbolic kinetic profile, yielding up to a 3-fold difference in inactivation efficiency. Interestingly, Melet et al. (2003) have shown evidence that the serine residue at position 365 (Ser-365) seems to play a role in inactivation of P450 2C9 by tienilic acid, a conclusion based on the observation that replacement of the nucleophilic serine by either alanine (S365A) or glycine (S365G) resulted in attenuation of mechanism-based inactivation of P450 2C9. In addition, cocrystallization of (S)-warfarin bound to P450 2C9 protein suggests that it lies predominantly in a hydrophobic pocket lined by several residues, including Ser-365 (Williams et al., 2003). Although one could argue that this reported binding orientation of (S)-warfarin may represent only one potential conformation, especially because this likely represents a nonproductive binding orientation (i.e., 10 Å from heme), if inactivation by tienilic acid results in covalent modification of Ser-365, then perhaps this may offer a reasonable explanation for the kinetic mechanism for inactivation of (S)-warfarin 7-hydroxylation relative to that of (S)-flurbiprofen and diclofenac hydroxylation. Although somewhat speculative at this time, it is possible that covalent modification of P450 2C9 at Ser-365 may result in two enzyme pools: adducted P450 2C9 that is inactive against (S)-warfarin but still able to catalyze (S)-flurbiprofen and diclofenac metabolism; and unadducted P450 2C9, which is active against all three substrates used in this study. If this hypothesis is true, then it may explain the sigmoidal inactivation kinetics of (S)-flurbiprofen and diclofenac 4-hydroxylation (i.e., both P450 2C9 enzyme pools active, but distinct kinetically), whereas the inactivation kinetics against (S)-warfarin are hyperbolic (i.e., only the unadducted P450 2C9 enzyme remains active). This is a potentially novel hypothesis as it pertains to inactivation mechanisms for when the apoprotein is covalently modified and certainly needs to be tested. Wester et al. (2004) have also provided evidence that Arg-108 probably plays a role in the high-affinity binding of (S)-flurbiprofen and propose that other lipophilic anions such as diclofenac and naproxen probably bind in this mode as well. The fact that analog 1 of tienilic acid (minus the carboxylate) did not demonstrate inactivation of P450 2C9 is consistent with these findings and further suggests that this residue most likely plays a role in the inactivation of tienilic acid and (±)-suprofen. Based on this information, we also hypothesize that the distance between the carboxylate and the bioactivated thiophene, which appears longer for tienilic acid, would allow the thiophene of tienilic acid to position itself closer to the heme prosthetic, relative to the thiophene of (±)-suprofen, and may be key in the observed inactivation rate differences. This hypothesis regarding the importance of distance will be the focus of future investigations.
Spectral binding titration of tienilic acid (A) and (±)-suprofen (B) with P450 2C9. A type I spectral shift was observed for both inactivators, and a spectral binding affinity constant of 2 μM for tienilic acid and 21 μM for (±)-suprofen was estimated. ΔAbs, change in absorbance.
Comparison of metabolic rates of tienilic acid (▪) and (±)-suprofen (○) and those of the separated enantiomers, (R)-suprofen (▴) and (S)-suprofen (▿), determined by the substrate depletion method. The calculated in vitro half-life of tienilic acid when incubated with P450 2C9 Baculosomes was 5 min, whereas those for (±)-suprofen, (R)-suprofen, and (S)-suprofen were 50, 38, and 35 min, respectively.
Titration study showing loss of P450 2C9 enzymatic activity as a function of the ratio of (A) tienilic acid to P450 2C9 (tienilic acid/P450 2C9) and (B) (±)-suprofen to P450 2C9 [(±)-suprofen/P450 2C9]. Partition ratios of 34 and 98 were estimated for tienilic acid and (±)-suprofen, respectively.
In conclusion, these studies demonstrate that inactivation efficiencies and to a greater extent inactivation kinetic profiles of P450 2C9 by both tienilic acid and (±)-suprofen seem to be substrate-dependent, with (S)-warfarin being the most sensitive reporter substrate. Although the differences in inactivation kinetic profile (sigmoidal versus hyperbolic) are not completely understood, this profile may have implications for prediction of clinical drug-drug interactions using in vitro data and suggests that (S)-warfarin should be considered for use as a reporter substrate when in vitro mechanism-based inactivation studies with P450 2C9 are performed.
Acknowledgments
We thank Roger Melton for assistance with development of bioanalytical assays, Sascha Freiwald for bioanalytical assistance, Joe Collins and Robert Hughes for assistance with separation of the enantiomers of (±)-suprofen, and R. Scott Obach for thoughtful comments.
Footnotes
-
This work was supported by the National Institutes of Health [Grant GM069753].
-
Article, publication date, and citation information can be found at http://dmd.aspetjournals.org.
-
doi:10.1124/dmd.108.023358.
-
ABBREVIATIONS: P450, cytochrome P450; DDI, drug-drug interaction; LC-MS/MS, liquid chromatography/tandem mass spectrometry; CHAPS, 3-[(3-cholamidopropyl) dimethylammonio]-1-propane sulfonate.
- Received July 12, 2008.
- Accepted October 3, 2008.
- The American Society for Pharmacology and Experimental Therapeutics