Introduction

Top Abstract Introduction Materials and Methods Results Discussion References

Dextromethorphan (DXM¹), a synthetic analog of codeine, is a widely used over the counter antitussive agent. DXM is metabolized to dextrorphan (DXO) through O-demethylation, to 3-methoxymorphinan (MEM) through N-demethylation, and to 3-hydroxymorphinan (HYM) through N,O-didemethylation in humans (Fig. 1; Beinhart, 1980; Pfaff et al., 1983). The formation of DXO is mediated by the polymorphic CYP2D6 and cosegregates with the debrisoquine-sparteine-type oxidative polymorphism (Schmid et al., 1985; Küpfer et al., 1986; Meyer et al., 1990; Kerry et al., 1994; Sachse et al., 1997). Therefore, the cough-suppressing DXM, available universally without prescription, is established and widely used as a probe drug for polymorphic CYP2D6 activity both in vivo (Baumann et al., 1992; Sachse et al., 1997) and in vitro (Jacqz-Aigrain et al., 1993; Wu et al., 1993; Kerry et al., 1994).

View larger version (24K): [in this window] [in a new window]   Fig. 1.   Dextromethorphan demethylation pathways catalyzed by recombinant human CYP2D6 and CYP3A4. The apparent Km for the major contributor to each reaction estimated in the present study is shown.

However, although DXM N-demethylation data from human liver microsomes can be fit to a single-enzyme kinetic model (Jacqz-Aigrain et al., 1993; Kerry et al., 1994; Schmider et al., 1997), investigators have not always agreed on a single, responsible enzyme (Jacqz-Aigrain et al., 1993; Gorski et al., 1994; Ducharme et al., 1996; Jones et al., 1996a,b; Schmider et al., 1997; Von Moltke et al., 1998; Kashuba et al., 1999; Krecic-Shepard et al., 1999; Wang and Unadkat, 1999; McGinnity et al., 2000). In a separate study from our laboratory, we confirmed that highly purified, recombinant human CYP2D6 itself produces significant amounts of N-demethylated MEM and secondary metabolite HYM in addition to the major DXO product. Therefore, it seems possible that the observation of single-enzyme kinetics for MEM formation in human liver microsomes may be due to contributions by a group of closely related enzymes rather than a single entity. Inhibition studies have been performed by others using selective inhibitors, and the results largely exclude the participation of CYP1A2, CYP2A6, CYP2C8, CYP2C9, CYP2C19, CYP2D6, and CYP2E1 in MEM formation. The significant correlations between DXM N-demethylase activity and erythromycin N-demethylase, testosterone 6-hydroxylase activities, and immuno-quantified levels of CYP3A4 provide strong evidence that CYP3A4 is the principal enzyme involved in the DXM N-demethylation pathway.

Thus, DXM has been suggested as a marker to monitor CYP3A activity simultaneously with the activity of CYP2D6 (Gorski et al., 1994). This hypothesis is evaluated in vivo (Ducharme et al., 1996; Jones et al., 1996a,b; Krecic-Shepard et al., 1999) and shows confirmative correlations using a well characterized inducer (rifampin) and inhibitor (erythromycin) of CYP3A4. Nevertheless, DXM N-demethylation as an index reaction for CYP3A4 activity raises questions because of the incomplete inhibition by selective CYP3A4 chemical and antibody inhibitors (Jacqz-Aigrain et al., 1993; Gorski et al., 1994; Schmider et al., 1997). In one study using P450 expressed from cDNA in lymphoblastoid cell culture, CYP3A4 was identified as the main enzyme mediating DXM N-demethylation (50%), whereas CYP2C9 and CYP2C19 are estimated to also be main contributors (22 and 26%, respectively; Von Moltke et al., 1998). More recently, CYP2C9 was estimated to be responsible for up to 43% of MEM formation, CYP3A4 42%, CYP2C19 8%, and CYP2D6 7% using P450 isoforms and human NADPH-P450 reductase coexpressed in Escherichia coli membranes (McGinnity et al., 2000). The CYP2B6 isoform, overlooked for decades, has also recently been identified to mediate MEM formation, in fact with the lowest apparent K_m and a relatively high V_max. The residual MEM activity (referring to the portion of MEM formation activity resistant to inhibition by indinavir in human liver microsomes) is highly correlated with CYP2B6 activity, suggesting the use of residual MEM activity as a measure of CYP2B6 activity (Wang and Unadkat, 1999). In addition, DXM N-demethylation as an index reaction for phenotypic estimation of CYP3A activity has been questioned by in vivo results (Kashuba et al., 1999).

In a previous study from our laboratory, the catalytic activity of highly purified recombinant human CYP2D6 was characterized by HPLC and confirmed using liquid chromatography-mass spectrometry analysis of both protein and metabolites (Yu et al., 2001). With DXM as a probe, we found that CYP2D6 produced significant amounts of MEM and HYM, as well as DXO. In the present study, we investigated and compared the contributions to DXM metabolism among CYP2B6, CYP3A4, CYP2C8, CYP2C9, CYP2C18, CYP2C19, and CYP2D6 using purified enzymes or cDNA-expressed cell microsomes. The results were then analyzed after correction for the average content and activity of each P450 estimated for human liver and compared with results obtained from human liver microsomes.

	Materials and Methods

Top Abstract Introduction Materials and Methods Results Discussion References

Reagents. Dextromethorphan HBr, dextrorphan D-tartrate, 3-methoxymorphinan HCl, and 3-hydroxymorphinan HBr were purchased from Research Biochemicals International (Natick, MA). Ketoconazole, reduced NADPH, L--dilaurylphosphatidylcholine (DLPC) were purchased from Sigma (St. Louis, MO). HPLC solvents and other chemicals were of the highest grade commercially available.

P450 Isoforms and Human Liver Microsomes. Human CYP2C8, CYP2C9, CYP2C18, and CYP2D6 proteins were expressed in Trichoplusia ni suspension cultures using a baculovirus expression system, purified, and characterized essentially as described previously (Haining et al., 1996; Yu and Haining, 2001). Human CYP2C19 was also expressed with this system, whereas cell microsomes were used for this study rather than purified P450. Microsomes containing human CYP3A4 or CYP2B6 along with P450 reductase and cytochrome b₅, expressed in BTI-TN-5B1-4 insect cells using a baculovirus expression system, were purchased from GENTEST (Woburn, MA) along with control insect cells infected with wild-type virus. Human liver microsomes (HG3, HK23, HG42, HG43, HG56, HG112, and H161) were also purchased from GENTEST.

Incubation Conditions. Incubation reactions for the determination of enzyme kinetic parameters (K_m and V_max) were carried out in 100 mM potassium phosphate, pH 7.4, containing 0.1 µM P450, 0.2 µM P450 reductase, 10 µg of DLPC, 1 mM NADPH, and substrate in a final volume of 200 µl. P450 and reductase were added together first and left to incubate at room temperature for 15 min. DLPC was then added for a further 15-min incubation period before the addition of buffer and substrate. The reactions were incubated at 37°C for 5 min and initiated by the addition of NADPH. The reactions were terminated after a 10 min-incubation (except CYP2D6 for which a 5-min incubation period was applied) by the addition of 10 µl of 60% perchloric acid. DXM concentrations ranged from 0 to 3000 µM for CYP3A4 and CYP2C18, from 0 to 2000 µM for CYP2C9 and CYP2B6, from 0 to 1000 µM for CYP2C19, and from 0 to 4000 µM for CYP2D6. MEM concentrations ranged from 0 to 50 µM for CYP2D6 and from 0 to 1000 µM for CYP3A4; DXO concentrations ranged from 0 to 4000 µM for CYP2D6 and from 0 to 1000 µM for CYP3A4 and CYP2B6.

HPLC Conditions. HPLC analyses were carried out on a Waters Alliance System consisting of 2690 separation module, 2487 dual absorbance detector, and 474 scanning fluorescence detector controlled with Millennium³² software (Waters, Milford, MA). A 250-mm × 4.6-mm i.d. Hi-chrom phenyl column (Regis Technologies, Inc., Morton Grove, IL) was used to separate the metabolites (Ducharme et al., 1996). The composition of the mobile phase was 50% acetonitrile and methanol mixture (250/200, v/v) and 50% 10 mM potassium phosphate buffer adjusted to pH 3.5 with orthophosphoric acid. The flow rate through the column at ambient temperature was 1 ml/min. The excitation and emission wavelengths, the gain, and the attenuation of the fluorescence detector were set at 280 and 310 nm, 10, and 32, respectively.

Data Analysis. Michaelis-Menten Enzyme parameters (K_m and V_max) were estimated by nonlinear regression (GraphPad Prism v3.02, GraphPad Software, San Diego, CA). Initial estimates for nonlinear regression were generated graphically using Eadie-Hofstee plots (V₀ versus V₀/[S]). Inhibitory constants were determined using Dixon plots (1/V₀ versus [I]). Linear regression analyses were conducted using Microsoft Excel 2000.

	Results

Top Abstract Introduction Materials and Methods Results Discussion References

HPLC Determination of DXM and Its Metabolites. The reaction pathways in question are illustrated in Fig. 1. DXM and its three demethylated metabolites were baseline-separated, with retention times of 7.1 (HYM), 9.1 (DXO), 11.1 (MEM), and 15.4 min (DXM) (Fig. 2) under the HPLC conditions described. The addition of quinidine or ketoconazole did not interfere with the separation analyses. Linear response relationships were observed between peak area and the amount of compound. The calibration curves used in this study ranged from 5 to 1000 pmol for MEM and HYM, and 10 to 2000 pmol for DXO and DXM, although the limits of detection were less than 5 pmol for all compounds. In most cases, control reactions lacking NADPH contained small amounts of DXO, MEM, and HYM, which were subtracted out.

View larger version (26K): [in this window] [in a new window]   Fig. 2.   HPLC separation of dextromethorphan and its metabolites formed by CYP2B6, CYP2C19, CYP2D6, and CYP3A4 at 20 µM initial substrate concentration during a 5- (2D6) or 10-min (2B6, 2C19, and 3A4) reaction. Each compound was baseline-separated, with retention times of 7.1 (HYM), 9.1 (DXO), 11.1 (MEM), and 15.4 min (DXM).

DXO Formation. DXO was produced by cytochromes P450 with calculated intrinsic clearance (V_max/K_m) for DXM O-demethylation decreasing in the order CYP2D6 > CYP3A4 > CYP2C9 > CYP2C19 > CYP2C18 > CYP2B6 (Fig. 3A). CYP2C8 did not mediate DXO formation under these conditions. The Michaelis-Menten kinetic parameters determined for each are shown in Table 1. As expected, the P450 isoform with the highest affinity (lowest K_m) of those involved in DXM O-demethylation was CYP2D6 in which the K_m value was 3.7 ± 1.2 µM, consistent with the K_m value calculated with a narrow DXM concentration range (0-50 µM) in a 5-min assay (Yu and Haining, 2001). DXM was also metabolized to DXO by CYP2D6 with the highest V_max (11.9 pmol/pmol of P450/min) and consequently with the highest intrinsic clearance (2.98 µl/pmol of P450/min). It was found that CYP2B6, CYP2C18, and CYP3A4 were involved in DXO formation but each proceeded with a much higher K_m and lower intrinsic clearance compared with CYP2D6.

View larger version (21K): [in this window] [in a new window]   Fig. 3.   Rates of DXO (A) and MEM (B) formation catalyzed by CYP2D6 (), CYP3A4 (), CYP2B6 (), CYP2C9 (×), CYP2C19 (*), and CYP2C18 () isoforms over a large (1-4000 µM) DXM concentration range.

View larger version (18K): [in this window] [in a new window]   Fig. 4.   Quinidine inhibition of CYP2D6-catalyzed dextrorphan formation (A) (Ki = 0.034 µM) and ketoconazole inhibition of CYP3A4-catalyzed 3-methoxymorphinan formation (B) (Ki = 0.13 µM) using recombinant enzymes in vitro. A, , 1 µM DXM; , 3 µM DXM; , 10 µM DXM. B, , 50 µM DXM; , 100 µM DXM; , 800 µM DXM.

View larger version (18K): [in this window] [in a new window]   Fig. 5.   Estimation of isoform contribution to dextromethorphan O- and N-demethylations catalyzed by P450 at 2 µM initial substrate concentration to mimic plasma dextromethorphan concentrations in human extensive metabolizers. Values shown are not adjusted for the APC or RAF for each P450 in human liver.

MEM Formation. Unexpectedly, CYP3A4 did not appear to have the highest turnover for the N-demethylation reaction (Table 1). Rates of MEM formation by P450s decreased in the order CYP2B6 > CYP3A4 > CYP2C18 > CYP2D6 > CYP2C19 > CYP2C9. CYP2C8 was again found not to be involved in DXM metabolism (Fig. 3B). CYP2B6 proceeded with the lowest calculated K_m (105 ± 7 µM) and a relatively high V_max (29.9 ± 0.6 pmol/pmol of P450/min). Also surprisingly, CYP2D6 catalyzed DXM N-demethylation with the highest apparent V_max (38.5 ± 7.1 pmol/pmol of P450/min) and the lowest affinity (K_m, 1290 ± 575 µM). The highest intrinsic clearance was attributable to CYP2B6 (0.283 µl/pmol of P450/min), followed by CYP3A4 (0.092 µl/pmol of P450/min). However, the contribution of CYP2B6 to MEM formation appears insignificant since its APC in human liver is only 0.2% (Shimada et al., 1994). Thus, CYP3A4 was primarily responsible for MEM formation (more than 92%) after correction. CYP2C18 was found to be involved in DXM N-demethylation with comparable K_m (354 ± 81 µM) and intrinsic clearance (0.049 µl/pmol of P450/min) values compared with CYP3A4 and CYP2B6 (Table 1). However, its contribution to MEM formation also appears insignificant due to its content of less than 2.5 pmol/mg of protein in human liver microsomes (Richardson et al., 1997).

View larger version (32K): [in this window] [in a new window]   Fig. 6.   Inhibition of CYP3A4-catalyzed dextromethorphan (200 µM initial) N-demethylation in human liver microsomes by 10 µM ketoconazole. Liver samples were chosen for varying CYP3A4 and CYP2B6 content as quantified by immunoblotting (www.gentest.com) as follows: sample H161, pooled microsomes from 11 different human livers; sample HG42, 310 pmol/mg of 3A4, 53 pmol/mg of 2B6 (ratio 5.8:1); sample HK23, 85 pmol/mg of 3A4, 7 pmol/mg of 2B6 (ratio 12.1:1); sample HG43, 94 pmol/mg of 3A4, 4 pmol/mg of 2B6 (ratio 23.5:1).

View larger version (21K): [in this window] [in a new window]   Fig. 7.   Correlation between dextromethorphan N-demethylase activity and testosterone 6 -hydroxylase activity (A) (n = 7) and immuno-quantified CYP3A contents (B) (n = 6) in human liver microsomes.

HYM Formation. No detectable HYM was formed from DXM by CYP2C8, CYP2C9, CYP2C18, or CYP2C19 individually. HYM was produced in the reactions containing CYP3A4, CYP2B6, and CYP2D6, but the reaction rate appears to be independent of DXM concentration. No kinetic parameters of HYM formation from DXM were calculated in this study. Nevertheless, the contribution to HYM formation by P450 isoforms could be estimated from the model mimicking human plasma DXM concentration. At 2 µM initial DXM, CYP2D6 was identified as the only isoform producing detectable HYM (data not shown).

	Discussion

Top Abstract Introduction Materials and Methods Results Discussion References

The use of dextromethorphan as a dual probe to simultaneously phenotype CYP3A4 activity along with CYP2D6 has been debated due to results from chemical and antibody inhibition studies in human liver microsomes, which show that CYP3A cannot fully account for the observed MEM formation from DXM (Fig. 6; Jacqz-Aigrain et al., 1993; Gorski et al., 1994; Schmider et al., 1997; Wang and Unadkat, 1999). In some cases, the incomplete inhibition observed in studies with chemical or antibody inhibitors may be due to the artificially high substrate concentration employed or a relatively low inhibitor concentration. From the Dixon equation, it is clear that the inhibition of enzyme activity is affected not only by the concentration of inhibitor but also by the concentration of the substrate. The unwanted biotransformation or binding of selective inhibitors in human liver microsomes is also possible, further complicating interpretations. In an attempt to clarify this disparity and to better understand the data derived from in vitro models, enzymes expressed individually in insect cells using a baculovirus-mediated system were used to examine the contribution to DXM metabolism by each of seven P450 isoforms under standardized conditions.

In practice, the apparent inhibitory potency of specific inhibitors may be affected by lab-to-lab variations, including enzyme source, buffers, incubation conditions, sample variability, and inhibitor, substrate, and enzyme concentrations. Therefore, chemical inhibition experiments were carried out for comparison with literature values. The K_i of quinidine against DXM O-demethylation reported in one study (Schmider et al., 1997) is approximately 0.1 µM on average, about 3 to 7 times higher than the K_i values (0.015-0.043 µM) obtained with human liver microsomes or recombinant protein by other investigators (Broly et al., 1989; Ching et al., 1995; Abdel-Rahman et al., 1999) and the current study (K_i, 0.034 µM; Fig. 4A). Published K_i values for the inhibition of human liver microsomal CYP3A4 by ketoconazole also vary considerably, from 4 nM to 8 µM (Wrighton and Ring, 1994; Bourrie et al., 1996; Von Moltke et al., 1996b; Gibbs et al., 1999). In the current study, the mean K_i for ketoconazole inhibition of DXM N-demethylation was found to be 0.13 µM (Fig. 4B), whereas that reported by Schmider et al. (1997) is 0.37 µM and that reported by Yamano et al. (1999) is 0.16 µM. The mean ketoconazole K_i values for in vitro hydroxylation of alprazolam (Von Moltke et al., 1994), triazolam (Von Moltke et al., 1996a), and midazolam (Von Moltke et al., 1996b) consistently fall below 0.1 µM.

As is clear from the present study, at high enough substrate concentrations, not only CYP2D6 but also CYP2C9, CYP2C18, CYP2C19, CYP2B6, and CYP3A4 all appear relevant in DXM O-demethylation (Fig. 3A). This may explain why biphasic Eadie-Hofstee plots are at times observed when using human liver microsomes because many experiments are carried out over broad DXM concentrations, from 1 to 1000 µM or higher (Jacqz-Aigrain et al., 1993; Gorski et al., 1994; Kerry et al., 1994; Schmider et al., 1997). Conversely, monophasic plots are obtained when a narrow DXM concentration range (0-40 µM) is used (Dayer et al., 1989). Indeed, above 50 µM, CYP3A4 appears to be the main contributor to DXO formation following normalization by the APC or RAF criteria (Fig. 3A). These in vitro results, therefore, cannot be used to predict in vivo drug biotransformation since the actual drug concentration in human blood is much lower. Using a concentration nearer the reported peak plasma DXM concentration in human extensive metabolizers (Capon et al., 1996), CYP2D6 is rightfully identified as the prime isoform catalyzing DXO formation, consistent with the results obtained from clinical and experimental data (Table 1; Fig. 5). Therefore, DXM O-demethylation (or DXM/DXO ratio) can safely be used to phenotype CYP2D6 activity.

CYP2B6, which was not included in many other studies, was surprisingly identified as the isoform with the lowest K_m and the highest intrinsic clearance for MEM formation not CYP3A4. However, since the average content of CYP2B6 in most livers is relatively much lower (0.2%; Shimada et al., 1994), more than 89% of DXM N-demethylation is attributable to CYP3A4 following correction by APC or RAF (Table 1). Nonetheless, in HLMs the concentrations of individual P450 isoforms can vary independently 100-fold or more. Recent studies indicate that CYP2B6 may be expressed at much higher levels (mean 44.6 pmol/mg) in human livers than previously known (Venkatakrishnan et al., 2000). Likewise, CYP3A4 levels may vary considerably between samples. In an attempt to address this variability, microsomal preparations of human liver were chosen for the present study based on their CYP3A4 to CYP2B6 ratio, ranging from 5.8:1 to 23.5:1 (Fig. 6). At a "mid-range" substrate concentration (200 µM DXM), 10 µM ketoconazole inhibited only some (50%) of the observed N-demethylation activity in each of these samples and in a pooled (11 livers) microsomal sample. DXM N-demethylation activity was further evaluated using a correlation analysis. Surprisingly, MEM formation correlated not only with CYP3A4 and 3A4-activity (Fig. 7) but also reasonably with CYP2B6 content (r² = 0.89) and (S)-mephenytoin N-demethylation (r² = 0.61) as well (data not shown). Wang and Unadkat (1999) likewise found that such correlation changes only minimally upon knocking out CYP3A activity with indinavir. Therefore, it appears that when present, CYP2B6 may be an important but still minor overall contributor to MEM formation.

CYP2B6 produced only marginal amounts of HYM from DXM, DXO, or MEM. Thus, didemethylation to HYM can occur by two separate pathways (Fig. 1): from DXM first to DXO catalyzed by CYP2D6, then to HYM mediated by CYP3A4; and from DXM to MEM catalyzed by CYP3A4 or CYP2B6, then to HYM mediated by CYP2D6. In the present study, CYP2D6 was found to catalyze O-demethylation with a much lower K_m and higher intrinsic clearance compared with CYP3A4 in both pathways. Therefore, HYM formation is attributable to CYP2D6 enzyme activity, providing the first in vitro evidence that the DXM/HYM ratio could be used equally to phenotype CYP2D6 activity and the DXM/DXO ratio.

In conclusion, all approaches used in this study provide consistent results indicating that CYP2D6 is the prime isoform catalyzing DXM O-demethylation, and CYP3A is the prime isoform mediating DXM N-demethylation at in vivo concentrations. The use of DXM as a dual probe to phenotype both CYP2D6 and CYP3A4 activity simultaneously, therefore, is practical despite possible contributions to DXO and MEM formation from other P450 isoforms. However, only a small sample set was included in the present study, and caution is still advised when working with samples or individuals with unusually high or unknown CYP2B6 content.