Abstract
The present study represents a comparison of three approaches to transform recombinant cytochrome P-450 (rCYP) enzyme kinetic data to human liver activity using mirtazapine (MIR) biotransformation as a model. MIR metabolite rCYP formation rates were corrected using I) relative activity factors (RAFs) determined on site, II) RAFs based on activity data provided by the rCYP manufacturer, and III) immunologically determined human liver abundance of CYP isoforms reported in the literature. For 2.5, 25, and 250 μM MIR, predictions of 1) the relative contribution of CYP isoforms to a particular reaction, 2) absolute metabolite formation rates, 3) the relative contribution of each pathway to net MIR biotransformation, and 4) the relative contribution of CYP isoforms to net MIR biotransformation were generated, and the results were compared with data obtained with human liver microsomes (HLM). We found that RAFs determined on site most accurately predict the results observed in HLM. Estimations based on liver abundance systematically underestimated CYP1A2 and overestimated CYP3A and CYP2C9 contributions to MIR metabolism and, therefore, seem less suitable to predict CYP isoform involvement in a particular reaction. Normalized RAFs calculated from the manufacturer activity data fell within the range of RAFs determined on site and lead to similar results for CYP isoform contribution to metabolic reactions and to net MIR biotransformation. Considering the time and resource-intensive step of RAF determination, manufacturer RAFs are an alternative to RAFs determined on site for the transformation of rCYP enzyme kinetic data; both of them provide more accurate estimations than immunologically determined human liver CYP isoform content.
Over the past years, cDNA-expressed human recombinant cytochrome P-450 (rCYP) (Crespi, 1995) has become increasingly important for in vitro drug metabolism studies. The availability of a single rCYP isoform allows rapid screening for metabolic activity. However, a positive result does not directly translate into a clinically important contribution of this CYP to the metabolism of the compound in question, because rCYP formation rates need to be corrected for the relative activity or abundance of the respective isoforms in human liver. Normalization of metabolic rates based on immunologically determined human liver CYP isoform content is widely used for this purpose, and most authors refer to a study by Shimada et al. (1994). Unfortunately, this approach does not take into account genetic polymorphisms that alter enzyme activity without affecting protein expression (e.g., CYP2C9) (Bhasker et al., 1997) and assumes that rCYP activity is independent of the expression system used. However, there is evidence that rCYP activity of the same isoform varies considerably between different expression systems (Crespi and Penman, 1997; Venkatakrishnan et al., 1998). Other factors, such as the level of cytochrome P-450 NADPH oxidoreductase and coexpression of cytochromeb5, also influence reaction velocity (Imai, 1981; Watkins et al., 1990; Crespi and Penman, 1997; Rendic and Di Carlo, 1997). Relative activity factors (RAFs), calculated as the ratio of human liver microsome (HLM) activity divided by rCYP activity for an isoform-specific index reaction, represent an approach that integrates these variables (Crespi, 1995). The approach is based on the assumption that any variable that affects the rate of metabolism for one substrate (the index drug) applies equally to other substrates (the study drug) (Crespi, 1995; Crespi and Penman, 1997; Venkatakrishnan et al., 1998). Although there is evidence that in some cases RAFs for a CYP isoform may depend on the index reaction used (Kenworthy et al., 1999; Roy et al., 1999; K. Venkatakrishnan, unpublished data), the RAF approach has been successfully used to estimate CYP isoform contributions to drug metabolism (Kobayashi et al., 1997; von Moltke et al., 1998, 1999; Greenblatt et al., 1999; Nakajima et al., 1999). Sufficiently selective index reactions are available for most CYP isoforms (Ono et al., 1996; Rendic and Di Carlo, 1997; Hickman et al., 1998), but determination of RAFs for each rCYP and HLM preparation requires considerable time and resources. The use of literature data is limited, because the nature of the rCYP cannot always be identified in detail (cell line, transfection method, reductase content,b5 coexpression, etc.). However, when rCYP are obtained from commercial sources, activity data is usually provided by the manufacturer and can be used for RAF calculation. Since activities for rCYP and HLM for a particular index reaction need to be determined under the same experimental conditions to obtain valid RAFs, we computed RAFs from the rCYP activity and the activity of pooled HLM from the same manufacturer. We assumed that this pool of HLM, although not actually used in the present study, represents an estimate of average CYP isoform activity in HLM.
Identification of a single CYP isoform accounting for a large fraction of the biotransformation of a given drug (>50%) is usually straightforward, while more complex metabolic patterns may yield different results depending on the method used for transforming the rCYP data. The antidepressant mirtazapine (MIR) is extensively metabolized to 8-hydroxy-mirtazapine (OHM),N-desmethyl-mirtazapine (DMM), and mirtazapine-N-oxide (MNO) (Dahl et al., 1997; Delbressine et al., 1998), and we have previously identified five CYP isoforms (CYP1A2, CYP2C8, CYP2C9, CYP2D6, and CYP3A4) involved in MIR metabolism in vitro (Störmer et al., 2000). This complex biotransformation was used to assess the accuracy of different strategies for the transformation of rCYP enzyme kinetic data.
Applied to MIR biotransformation, the present study represents a comparison of three different approaches to correct rCYP enzyme kinetic data for human liver activity using:
I. RAFs determined on site for the 10 HLM preparations and the lot of rCYP used for MIR metabolism experiments.
II. RAFs computed from activity data provided by the manufacturer of the rCYP.
III. Human liver abundance of CYP isoforms reported in the literature (Shimada et al., 1994; Lasker et al., 1998).
Each method was used to generate predictions of 1) the relative contribution of CYP isoforms to a particular reaction, 2) absolute formation rates of the metabolites, 3) the relative contribution of each metabolic pathway to net MIR biotransformation, and 4) the relative contribution of CYP isoforms to net MIR biotransformation. Results were compared with data obtained with HLM. The study is based on MIR enzyme kinetic parameters and chemical inhibition data determined previously (Störmer et al., 2000).
Three MIR concentrations were chosen to reflect 1) anticipated in vivo liver concentrations: 2.5 μM (Anderson et al., 1999; Moore et al., 1999), 2) possible in vivo concentrations after intentional (suicidal) or accidental ingestion of MIR overdose: 25 μM (Gerritsen, 1997;Holzbach et al., 1998; Retz et al., 1998), and 3) the approximateKm for MIR biotransformation in HLM: 250 μM MIR (Störmer et al., 2000).
Materials and Methods
Chemicals.
MIR, DMM, OHM, and MNO were kindly provided by N.V. Organon (Oss, The Netherlands). Other drugs and chemicals were purchased from commercial sources, or were kindly provided by their pharmaceutical manufacturers.
Human Liver Samples and cDNA-Expressed Enzymes.
Healthy liver tissue was obtained from the International Institute for the Advancement of Medicine (Exton, PA) or the Liver Tissue Procurement and Distribution System (University of Minnesota, Minneapolis, MI). The tissue was kept at −80°C until the time of microsome preparation. Microsomes were prepared and stored as described previously (von Moltke et al., 1993). Microsomal protein content was determined using bicinchoninic acid protein assay (Pierce, Rockford, IL) and bovine serum albumin as a standard. None of the liver samples were phenotypically poor metabolizers for CYP2D6 (mean activity at 10 μM dextrorphan: 177.5 nmol of dextrorphan/mg of protein/min, S.D. 67.1) or CYP2C9 (mean activity at 600 μM tolbutamide: 188.4 nmol of hydroxy-tolbutamide/mg of protein/min, S.D. 71.6).
Microsomes from cDNA-transfected human lymphoblastoid cells expressing CYP1A2, CYP2A6, CYP2B6, CYP2C8, CYP2C9, CYP2C19, CYP2D6, CYP2E1, and CYP3A4 were obtained from Gentest Corp. (Woburn, MA).
Incubations.
Methanolic solutions of substrates and inhibitors were evaporated to dryness before addition of buffer and cofactors. Incubation mixtures (250 μl) contained 0.5 mM NADP, 3.75 mM dl-isocitric acid, 1 U/ml isocitrate dehydrogenase, and 5 mM Mg2+ in 0.05 M KH2PO4 (pH 7.4). Following the manufacturer's instructions, phosphate buffer was replaced by Tris (0.05 M, pH 7.4) for rCYP2C9. Mirtazapine: in vitro studies on MIR metabolism have been described in detail elsewhere (Störmer et al., 2000). Briefly, MIR (2.5–2000 μM) was incubated with 250 μg of protein/ml for HLM (500 μg/ml for rCYP) for 30 min, except for rCYP2D6 (5 min). Index reactions: incubation conditions for determination of RAFs are summarized as follows: CYP1A2: phenacetin-O-deethylation at 100 μM, 250 μg of protein/ml, incubation time: 20 min for HLM, 30 min for rCYP. CYP2C8: paclitaxel-6α-hydroxylation at 10 μM, 500 μg/ml, 30 min. CYP2C9: tolbutamide-methylhydroxylation at 600 μM, 250 μg/ml, 20 min for HLM, 30 min for rCYP. CYP2D6: dextromethorphan-O-demethylation at 10 μM, 500 μg/ml, 30 min for HLM, 5 min for rCYP. CYP3A4: triazolam-4-hydroxylation at 750 μM, 250 μg/ml, 20 min for HLM, 30 min for rCYP. Incubations were done in triplicate (determination of RAFs), duplicate (MIR with HLM), or single (MIR with rCYP). Solubility of substrates and inhibitors was validated for the concentration ranges used. Incubation times and microsomal protein concentrations were within the linear range of metabolite formation.
HPLC.
Separation of MIR and its metabolites has been described elsewhere (Störmer et al., 2000). HPLC conditions for index reactions were derived from previously described methods (Rahman et al., 1994; Schmider et al., 1996; von Moltke et al., 1996, 1998;Miners and Birkett, 1998). Identity of metabolites was verified by comparing the retention times with those of authentic standards.
Data Analysis.
Kinetic parameters were determined by nonlinear least square regression (SigmaPlot 4.01, SPSS Inc., Chicago, IL).
Correction of rCYP Data for Human Liver Activity.
Three different approaches to correct rCYP data for human liver activity were compared. Formation rates were multiplied with:
I. RAFs determined in our laboratory for the 10 HLM preparations and the lot of rCYP used for MIR metabolism experiments (RAFs determined on site).
II. RAFs computed from activity data provided by Gentest Corp. for the lot of rCYP used in MIR metabolism experiments and pooled HLM from the same manufacturer (not used in MIR studies) (Manufacturer RAFs).
III. Average human liver CYP isoform content reported by Shimada et al. (1994). Since CYP2C8 content was not assessed in this study, a reference by Lasker et al. (1998) was used to identify the mean ratio of CYP2C8:CYP2C9:CYP2C19 in human liver, which was then applied to the total CYP2C content reported by Shimada et al. (1994)(Liver abundance).
Relative activity factors (RAFs) were calculated using eq. 1 (Crespi, 1995; Venkatakrishnan et al., 1998):
Results
Determination of Correction Factors.
RAFs determined on site were obtained for recombinant CYP1A2, CYP2C8, CYP2C9, CYP2D6, and CYP3A4 for 10 HLM preparations, manufacturer RAFs were computed from activity data provided by the manufacturer of the rCYP, and human liver abundance of CYP isoforms was identified in the literature (Table 1).
Liver abundance values were generally lower and manufacturer RAFs tended to be higher than the RAFs determined in our laboratory (Table1). For most applications, however, the magnitude of an RAF relative to other isoforms rather than its absolute value is critical. Normalized values of manufacturer RAFs (Fig. 1) fell within the range of RAFs determined on site with CYP1A2 accounting for >50% of total RAF value, CYP3A4 representing 20% to 30%, and CYP2C8, CYP2C9, and CYP2D6 showing normalized RAFs of <10%. Liver abundances, however, showed a different pattern, with a normalized CYP3A liver content of >50% and CYP2C and CYP1A2 accounting for 20% to 30% each. The normalized factor associated with CYP2D6 was <10% for all three approaches (Fig. 1).
Contribution of CYP Isoforms to a Particular Reaction.
The relative contribution of CYP isoforms to a metabolic pathway was calculated at 2.5, 25, and 250 μM MIR (Figs.2a, 3a, and 4a). Results were compared with the degree of inhibition observed with chemical inhibitors in HLM (Figs. 2b, 3b, and 4b). To facilitate this comparison, inhibition data are expressed as percentage inhibition rather than percentage of control activity, because this value directly corresponds to the relative contribution of the isoform.
OHM.
CYP1A2, CYP2C9, CYP2D6, and CYP3A4 showed MIR-hydroxylase activity (Fig. 2).
CYP1A2 contribution was predicted by RAFs determined on site to increase from 20% (mean 20.1 ± 8.9% S.D.) at 2.5 μM MIR to 55% (53.4 ± 13.0%) at 250 μM MIR, which is in agreement with the observed increase in α-naphthoflavone inhibition from 20% (20.2 ± 14.1%) at 25 μM MIR to 50% (49.2 ±10.1%) at 250 μM MIR. Manufacturer RAFs predicted similar CYP1A2 contributions, but liver abundance underestimated CYP1A2 contribution with increasing MIR concentrations.
In contrast, CYP2C9 contribution to OHM formation was overestimated by liver abundance, predicting a 40% contribution of CYP2C9 at 250 μM MIR compared with <10% (8.9 ± 3.3%) predicted by RAFs determined on site. Sulfaphenazole inhibition confirmed the lower value, causing 10% (9.7 ± 2.9%) inhibition at 250 μM MIR and showing no effect at 25 μM MIR corresponding to the CYP2C9 contribution of <2% (1.3 ± 0.5%) predicted by RAFs determined on site at this concentration.
The values predicted for CYP2D6 are in good agreement among the three approaches, indicating a decrease in the CYP2D6 contribution from a range of 60 to 80% at 2.5 μM MIR, to a range of 40 to 60% at 25 μM, and to a range of 15 to 30% at 250 μM. The inhibition observed with quinidine in HLM also decreased from 40% (37.3 ± 17.7%) at 25 μM to 30% (26.6 ±8.6%) at 250 μM MIR.
Predictions for the contribution of CYP3A4 to OHM formation at 250 μM MIR ranged from <10% (9.3 ± 4.8%) for RAFs determined on site to >20% for liver abundance, while inhibition by ketoconazole was negligible (2.5 ± 4.9%).
DMM.
CYP1A2, CYP2C8, and CYP3A4 showed MIR-N-demethylase activity (Fig.3). RAFs determined on site predicted a decrease in CYP1A2 contribution to DMM formation from 60% (57.0 ± 13.8%) at 2.5 μM MIR to 30% (31.7 ± 11.3%) at 25 μM to 10% (10.9 ± 5.2%) at 250 μM, while CYP3A4 contribution increased correspondingly from 40% (40.7 ± 13.9%) to 75% (75.8 ± 8.3%). The corresponding inhibition by α-naphthoflavone and ketoconazole was approximately 10% and 40%, respectively, at both, 25 and 250 μM MIR. Manufacturer RAFs predicted an approximately 10% lower CYP1A2 contribution and a 10% greater CYP3A4 contribution at 2.5 and 25 μM MIR, while CYP3A4, based on liver abundance, accounted for >80% of DMM formation at any substrate concentration. The predicted CYP2C8 contribution was generally low (<15%).
MNO.
CYP1A2 and CYP3A4 showed MIR-N-oxidase activity (Fig. 4). CYP1A2 was identified as the high affinity enzyme with a predicted contribution of >80% at 2.5 μM MIR decreasing to 10 to 20% at 250 μM MIR (RAFs determined on site and manufacturer RAFs). The low affinity enzyme CYP3A4 contributed <20% (10.3 ± 6.5%) at 2.5 μM MIR but accounted for >80% (80.0 ± 8.9%) of MNO formation at 250 μM. With a pattern similar to DMM formation, liver abundance predicted a smaller contribution of CYP1A2 and in turn a more important role of CYP3A4. Inhibition studies in HLM were limited to 250 μM, because MNO formation was not detectable in HLM at MIR concentrations ≤25 μM. At 250 μM MIR, MNO formation was inhibited by α-naphthoflavone and ketoconazole by 10% (8.6 ± 11.4%) and 50% (46.5 ± 12.8%), respectively.
Absolute Formation Rates.
The predicted total OHM and DMM formation rates (equal to the sum of corrected formation rates for each isoform catalyzing the reaction) at 2.5, 25, and 250 μM MIR were plotted against the formation rates observed in HLM (Fig.5). As expected from the absolute factor values (Table 1), liver abundance generally underestimated the true formation rates, particularly those for OHM, while use of manufacturer RAFs tended to overestimate reaction velocity of DMM formation (Fig. 5). MNO formation was not detectable at MIR concentrations ≤25 μM.
Relative Contribution of Metabolic Pathways to Net MIR Biotransformation.
Figure 6 shows the relative contribution of MIR-8-hydroxylation, MIR-N-demethylation, and MIR-N-oxidation to net MIR metabolism over a range of substrate concentrations. The RAFs determined on site provided the most accurate estimate of the true (observed) biotransformation pattern in HLM. Manufacturer RAFs and liver abundance both overestimated the role of DMM and MNO formation while underestimating the role of MIR-hydroxylation by about 20% (Fig.6). All estimated values fell within two standard deviations of the mean contribution observed in HLM indicating that, at 2.5 μM MIR, OHM and DMM each accounted for >35% of net MIR biotransformation, respectively, while MNO contributed <10%.
Relative Contribution of CYP Isoforms to Net MIR Biotransformation.
At 2.5 μM MIR, both RAF approaches predicted a 40% (37.8 ± 12.2%) contribution of CYP1A2 to MIR metabolism, while liver abundance indicated a contribution of <20% for this enzyme (Fig. 7). Predicted CYP2D6 contribution ranged from 30% to 50%. RAFs determined on site indicated that CYP3A4 accounts for 15% (12.8 ± 6.4%) of net MIR biotransformation, while manufacturer RAFs and liver contents predicted 20% and 30%, respectively, for the same enzyme. CYP2C9 and CYP2C8 were consistently of minor importance for total MIR clearance (<10%).
Discussion
The RAF approach (Crespi, 1995; Crespi and Penman, 1997;Venkatakrishnan et al., 1998) and the use of immunologically determined CYP isoform liver content (Shimada et al., 1994; Rodrigues, 1999) have been the two main strategies to correct rCYP formation rates for native human liver enzyme activity. In contrast to immunoquantified CYP levels, an RAF does not only depend on the liver abundance of the CYP isoform but also reflects the specific activity of the rCYP preparation used. A recently introduced RSF (relative substrate-activity factor) method (Roy et al., 1999), is based on the same concept and yields identical results with the same limitations as the RAF approach.
To facilitate future in vitro studies using recombinant human CYP, we compared three different approaches to transform rCYP enzyme kinetic data to the situation in vivo. The biotransformation of the antidepressant MIR was used as an example involving five CYP isoforms in three metabolic pathways. Based on previously determined enzyme kinetic parameters for MIR-8-hydroxylation, MIR-N-demethylation, and MIR-N-oxidation by rCYP (Table 2) (Störmer et al., 2000), formation rates were corrected for human liver activity using I) RAFs determined on site, II) RAFs based on manufacturer activity data, or III) CYP isoform liver abundance (Table 1).
Immunoquantification factors were generally lower and manufacturer RAFs tended to be higher than the RAFs determined in our laboratory (Table1), but normalized manufacturer RAFs fell within the range of RAFs determined on site. Immunoquantified CYP levels, however, showed a different pattern, with values for CYP3A4 and CYP2C9 at least twice as high as the corresponding RAF, and a CYP1A2 factor of about 50% the RAF values (Fig. 1).
The accuracy of predictions of CYP isoform contribution to a particular reaction can be easily assessed by comparison with chemical inhibition data in HLM. At a given substrate concentration, a CYP isoform-specific inhibitor (Newton et al., 1995; Bourrié et al., 1996; Ono et al., 1996) should reduce the formation rate of the metabolite in HLM by approximately the same fraction that the particular CYP is estimated to account for. In general, inhibition studies at different substrate concentrations can verify predicted changes in contribution of high and low affinity enzymes in the same reaction. For MIR, the increased contribution of CYP1A2 to OHM formation with increasing substrate concentrations (Fig. 2) was confirmed by an increase in α-naphthoflavone inhibition between 25 and 250 μM MIR, while the CYP2D6 contribution as well as quinidine inhibition decreased correspondingly. Also, the small increments in CYP2C9 and CYP3A4 contributions with increasing MIR concentrations were reflected by the inhibitory effects of sulfaphenazole and ketoconazole, respectively, in HLM (Fig. 2).
Estimation of absolute formation rates of a particular metabolite seems to require the use of RAFs determined for the livers and rCYP actually tested, because manufacturer RAFs tended to overestimate metabolic rates, particularly for DMM, while OHM formation rates predicted by liver abundance were too low (Fig. 5). In practice, studies rarely focus on absolute formation rates, but rather evaluate the relative contribution of one metabolite to the overall biotransformation of the parent compound: this appears to be less sensitive to variations among different correction methods. While estimated formation rates can differ by >100% from the observed values (Fig. 5), the predicted relative contributions of OHM, DMM, and MNO formation deviate less than 20% from the observed data, regardless of the correction approach used (Fig. 6).
By combining the results of the different pathways of MIR metabolism, the contribution of a particular CYP isoform to net biotransformation can also be estimated from rCYP data. Predicted relative contributions of CYP isoforms varied up to 20% at 2.5 μM MIR. CYP1A2, CYP2D6, and CYP3A4 were identified as the major enzymes involved in MIR biotransformation independent of the method used (Fig.6), while CYP2C8 and CYP2C9 contributed less than 10% (Fig. 7).
Comparing the three strategies to correct rCYP enzyme kinetic data for human liver activity, we found that RAFs determined on site most accurately predicted the results observed in HLM. Estimations based on liver abundance systematically underestimated CYP1A2 and overestimated CYP3A and CYP2C9 contributions (Figs. 2 and 3) to MIR metabolism, and therefore seem less suitable to predict CYP isoform involvement in a particular reaction. However, normalized RAFs calculated from manufacturer-provided activity data fell within the range of the RAFs determined on site (Fig. 1) and lead to similar results for CYP isoform contribution to metabolic reactions and to net MIR biotransformation (Figs. 2-4, and 7).
Considering the time- and resource-intensive step of RAF determination, manufacturer RAFs are an alternative to RAFs determined on site for the transformation of rCYP enzyme kinetic data, providing more accurate estimations than human liver CYP isoform contents. However, considering the possible substrate dependence of RAFs, the most appropriate approach for a given drug may partly depend on the index reactions used to determine the RAFs (Kenworthy et al., 1999). Consequently, studies involving different drugs or classes of drugs will be needed to further investigate the subject.
Footnotes
-
Send reprint requests to: David J. Greenblatt, M.D., Department of Pharmacology and Experimental Therapeutics, Tufts University School of Medicine, 136 Harrison Ave., Boston, MA 02111. E-mail: dj.greenblatt{at}tufts.edu
-
↵1 This work was supported by Grants MH-34223, MH-01237, and DA-05258 from the U.S. Department of Health and Human Services.
-
↵2 Recipient of an HSP III doctoral grant by the German Academic Exchange Service (DAAD).
- Abbreviations:
- rCYP
- recombinant cytochrome P-450
- CYP
- cytochrome P-450
- MIR
- mirtazapine
- OHM
- 8-hydroxymirtazapine
- DMM
- N-desmethylmirtazapine
- MNO
- mirtazapine-N-oxide
- HLM
- human liver microsomes
- RAF
- relative activity factor
- Received April 4, 2000.
- Accepted July 19, 2000.
- The American Society for Pharmacology and Experimental Therapeutics