Some compounds used for phenotyping of cytochrome P450s are substrates of P-glycoprotein (pgp). It is likely that in these cases, the level of pgp modulates the metabolism of in vivo probes. To address this important issue, we have analyzed the effects of pgp on CYP3A4-mediated reactions in two newly established cell lines (3A4/HR/MDR−and 3A4/HR/MDR+), which express CYP3A4 in the absence and presence of pgp, respectively. In cultured cells, the presence of pgp increased the apparent K m for the 6β-hydroxylase activity of CYP3A4 toward testosterone and cortisol by a factor of 1.7 and 4, respectively. These steroids are poor and good substrates of pgp, respectively, and cortisol 6β-hydroxylase has been frequently used as an in vivo probe for CYP3A4. Interestingly, we also found that pgp modulated the inhibition of CYP3A4-mediated metabolism by several compounds in intact cells. Although quinidine inhibited testosterone 6β-hydroxylase activity in membranes or in intact cells that expressed recombinant CYP3A4 in the absence of pgp, low concentrations of this compound increased CYP3A4 activity in intact cells that expressed pgp. These results imply that pharmacokinetic drug-drug interactions involving CYP3A4 can be influenced by pgp.
Drug disposition may be conceptualized as consisting of absorption, distribution, biotransformation, and excretion (Parkinson, 1996). The latter three processes also control the disposition of several endogenous compounds such as steroids. Phase I of drug metabolism is mainly catalyzed by cytochrome P450s (P450). The cellular uptake and export of xenobiotics or their metabolites is mediated by drug transporters.
P450-mediated metabolism displays interindividual variability either due to genetic polymorphisms, which affect the catalytic properties of the various P450 isoforms, or due to differences in the level of P450 enzymes, which are determined by endogenous and exogenous factors (Daly et al., 1993; Kroemer and Eichelbaum, 1995; Smith et al., 1998). Because interindividual differences of P450-catalyzed drug metabolism have important implications for drug therapy and also determine disease susceptibility (Wormhoudt et al., 1999), geno- and phenotyping of these enzymes have become important issues. Ideally, the metabolism of the in vivo metabolic probes should be only dependent on the level and the catalytic properties of the P450 that metabolizes them. Phase II metabolism of the resulting metabolites would interfere with accurate phenotyping. Similarly, drug uptake or export is likely to affect the probe reaching its target P450, thereby influencing these assays. This can occur with compounds that are simultaneously substrates of drug transporters and of P450s.
Most studies have investigated the interaction of CYP3A4 or orthologous CYP3A enzymes, and an important drug transporter, namely, P-glycoprotein (pgp, human MDR1), which share many substrates and/or inhibitors (Kivisto et al., 1995). For example, cortisol, quinidine, and erythromycin, which are substrates of in vivo phenotyping assays for CYP3A4, are also substrates of pgp (Ged et al., 1989; Barnes et al., 1996; Kim et al., 1999). Several reports have addressed the important question of how the combined action of CYP3A4 and pgp influences the pharmacokinetics and the cytotoxicity of therapeutic drugs and how drug-drug interactions impinge on this process (Gan et al., 1995; Wacher et al., 1995; Lown et al., 1997; Watkins, 1997; Hall et al., 1999). However, it has been extremely difficult to evaluate the contribution of each component to these parameters, mostly because selective inhibitors that discern between both processes have been until recently unavailable. One attempt to circumvent this problem has been to correlate the levels of CYP3A and pgp in tissues with pharmacokinetic data (Lown et al., 1997). Another approach used transgenic mice in which the mdr1a and the mdr1bgenes have been inactivated by homologous recombination, resulting in changes in bioavailability and the toxicity of several compounds (Schinkel et al., 1995). However, to our knowledge, no data are available on the effect of pgp on CYP3A-mediated metabolism in this model. This approach may also be problematic because the catalytic properties of rodent P450s differ from those of human P450s (Gonzalez, 1992). It has been also demonstrated that human and mouse pgp have different substrate specificities (Tang-Wai et al., 1995). Consequently, the interplay between P450 and pgp in rodents is likely to be different from that in humans.
The development of cell lines that coexpress pgp and CYP3A4 was hampered by the rapid repression of CYP3A4 in vitro. One study achieved culture conditions that lead to CYP3A4 expression in the human intestinal cell line Caco-2 (Schmiedlin-Ren et al., 1997), which expresses low levels of pgp. Another report described the heterologous expression of CYP3A4 in Caco-2 cells. These cell lines could be used to study the influence of pgp on CYP3A4-mediated metabolism, by comparing metabolism in intact cells with that in membrane fractions. However, no data are available on this issue. This approach has also the disadvantage that other factors such as membrane permeability, cellular milieu, and cofactor availability may contribute to differences in intact cells compared with lysates.
Some pharmacokinetic drug-drug interactions are often a composite, resulting from the interplay of pgp and CYP3A4, which unfortunately complicates interpretation of the observed effects. In one study, it was shown that the antifungal ketoconazole or the immunosuppressant cyclosporin A increased the area under the curve for vinca alkaloids, most likely by inhibiting pgp and CYP3A. However, the contribution of both pathways to the increased bioavailability of vinca alkaloids is not clear (Chan, 1998). Similarly simultaneous inhibition of pgp and CYP3A has been postulated to be responsible for the effects of ketoconazole on the bioavailability of digoxin or the cysteine protease inhibitor KO2 (Salphati and Benet, 1998).
Because the impact of pgp on CYP3A4-mediated reactions and pharmacokinetic drug-drug interactions is difficult to dissect in vivo, we developed an in vitro model to address this important issue.
Bovine serum albumin, verapamil, cortisol, testosterone, α-naphthoflavone, ketoconazole, and quinidine were obtained from Sigma (Poole, UK). 6β-Hydroxytestosterone was kindly provided by Steraloids Inc. (New Port, RI) and 6β-hydroxycortisol was obtained from Research Plus (Bayonne, NJ). The multidrug resistance modulator LY335979 was provided by Lilly (Windlesham, UK). Cell culture media and the Lipofectin reagent were purchased from GIBCO (Paisley, Scotland). Calcein acetoxymethyl ester (calcein AM) was purchased from Molecular Probes (Eugene, Oregon). [3H]Vinblastine (12.7 Ci/mmol) was obtained from Amersham (Cardiff, UK). Other reagents were of highest grade commercially available.
Cell Culture, DNA Transfection, and Isolation of Cell Lines.
1847 human ovarian cancer cells (Louie et al., 1986) were grown under standard cell culture conditions in 1640 RPMI medium supplemented with 10% fetal bovine serum, penicillin (50 IU/ml), and streptomycin (50 μg/ml).
This cell line was made resistant to paclitaxel (Taxol, Bristol-Myers Squibb, Stamford, CT) by initially treating the cells with the mutagen ethylmethanesulfonate (EMS) for 24 h, allowing the cells to recover in the absence of EMS for 48 h before treating with 1 nM paclitaxel. The cells were allowed to adapt to this concentration of paclitaxel for three passages before the paclitaxel concentration was increased to 2.5 nM. This incremental exposure to drug was continued until cells became resistant to 10 μM paclitaxel. Expression of pgp in different clones was analyzed by immunoblotting.
1847 cells overexpressing human pgp (termed MDR+cells) as well as parental cells (termed MDR−cells) were cotransfected with the plasmids pDHFR/3A4 carrying the CYP3A4 cDNA under the control of the cytomegalovirus promoter and pCDNA containing the G-418 resistance gene (Ding et al., 1997) using the Lipofectin. In brief, 2 × 105 cells were incubated in 60-mm tissue culture plates in 4 ml of standard growth medium for 24 h. Plasmid DNA (2 μg) and 20 μl of Lipofectin reagent were each diluted in 100 μl of Opti-MEM I medium. Both solutions were allowed to stand at room temperature for 45 min, mixed, and incubated at room temperature for further 15 min. Opti-MEM I (1.8 ml) was added and the 1847 cells were overlaid with this solution after being washed with Opti-MEM1 once. After 24 h of incubation at 37°C, the DNA-containing medium was replaced with 4 ml of normal growth medium and the incubation continued for another 48 h. Finally, cells were split and selected by G-418 (400 μg/ml). Cells derived from MDR+ clones were grown in the presence of additional paclitaxel (1 μM). Expression of CYP3A4 was detected by immunoblotting.
Lipofection was also used for the cotransfection of cells with the vectors pcDNAHR and pcDNA/Zeo+ carrying the human P450 reductase cDNA under the control of the cytomegalovirus promoter (Ding et al., 1997) and, respectively, the marker conferring resistance to zeocin. Cell clones were selected using zeocin (100 μg/ml). After isolation of zeocin-resistant colonies, the concentration of zeocin was changed to 50 μg/ml and 400 μg/ml G-418. Cells derived from MDR+ clones were grown in the presence of additional paclitaxel (1 μM).
Immunochemical Detection of pgp, CYP3A4, and P450 Reductase.
For isolation of total cellular protein, cells were harvested by trypsin treatment and resuspended in a buffer containing 10 mM sodium phosphate (pH 8.0), 1 mM EDTA, and 2 mM dithiothreitol. Cells were lysed by two 8-s bursts at an amplitude of 12 μm using an MSE Soniprep 150 sonicator, leaving the samples frequently on ice. Protein concentration was determined spectrophotometrically using the Bradford reagent (Bio-Rad, Hemel Hempstead, UK) and bovine serum albumin as standard.
The expression of pgp, CYP3A4, and P450 reductase was analyzed by immunoblotting (Towbin et al., 1979) using as primary antibodies the monoclonal anti-pgp C219 (Signet Lab, Glasgow, UK) and polyclonal antibodies directed against CYP3A4 or human P450 reductase (Li et al., 1999). Secondary antibodies were coupled to horseradish peroxidase and were detected by fluorography using the enhanced chemiluminescence Western blotting kit (Amersham, Cardiff, UK), according to the manufacturer's instructions. For detection of pgp, samples were denatured by suspension in loading buffer containing 200 mM dithiothreitol and subsequent incubation at room temperature for 5 min. For the detection of CYP3A4 and P450 reductase, samples were boiled in loading buffer. Generally, 60 μg of protein was loaded per lane.
Determination of the CYP3A4 Activity in Cultured Cells and Cell Lysates.
For the detection of the testosterone 6β-hydroxylase activity in cultured cells, 5 × 105 cells were seeded per 60-mm culture dish and incubated in standard growth medium at 37°C for 24 h (Ding et al., 1997). Hemin was added to a final concentration of 5 μg/ml and cells cultured for a further 24 h. Medium was replaced with 4 ml of fresh medium and incubated for 1 h at 37°C. Testosterone (dissolved in methanol) was added directly into the culture medium to a final concentration of 100 μM if not stated otherwise. After 3 h of incubation, 3.3 ml of medium was transferred to a polypropylene tube containing 1.1 ml of ice-cold methanol and left on ice for about 10 min. Subsequently, the remaining medium was removed and cells were harvested and lysed by sonication for determination of the total protein concentration. Before HPLC, metabolites were concentrated using Isolute C18 columns (IST Ltd, Midglamorgan, UK) and after washing with 4 ml of 25% methanol, the samples were eluted with 1 ml of methanol. The solvent was evaporated and the residue was resuspended in 200 μl of 35% methanol, and then centrifuged to pellet-insoluble material. Metabolites were separated by HPLC.
For the detection of testosterone 6β-hydroxylase activity in cell lysate, 5 × 106 cells were cultured for 3 days as described above and harvested by trypsinization, resuspended, and sonicated (see above). Cell lysate (5 mg) was incubated for 1 h at 37°C in an assay buffer containing 0.05 M Hepes (pH 7.4), 30 mM MgC12, 200 μM testosterone, and 1 mM NADPH. The HPLC assay was performed as described above. For determination of CYP3A4 activity in membranes isolated from recombinantEscherichia coli (Blake et al., 1996), the cell lysate in the assay was replaced with membranes containing 100 pmol CYP3A4 and P450 reductase (approximately 1000 nmol of cytochrome creduced per minute per milligram of protein). Incubation was for 20 min at 37°C. The assay conditions were found to be linear with time and protein.
The determination of the cortisol 6β-hydroxylase was conducted as follows: 1.5 × 106 cells were cultured for 3 days as described above. The medium was replaced by 7 ml of fresh medium and incubated for 1 h at 37°C. Cortisol (dissolved in DMSO) was added directly into the culture medium to a final concentration of 200 μM unless stated otherwise. After 14 h of incubation, 6.6 ml of medium was transferred to a polypropylene tube containing 2.2 ml of ice-cold methanol and left on ice for about 10 min. Samples were prepared as described above using Isolute C18 columns for concentration of the metabolites and the total protein concentration was determined. Separation of the metabolites by HPLC was performed on an Inertsil ODS-2 column (5 μm, 4.6 × 250 mm; Phase Separations, Ltd, Watford, UK) at a flow rate of 1 ml/min using isocratic elution with 8% ethanol. The effluent was monitored at 240 nm. The assay conditions were found to be linear with time and protein.
Determination of pgp Activity.
Calcein AM ester (Molecular Probes) was used to study the efflux activity of pgp (Homolya et al., 1993). Cells were seeded at 5 × 105cells/ml for 5 days in 10 ml of cell culture medium in an 80-cm2 flask. Briefly, cells were trypsinized and washed once with HPMI buffer [120 mM NaCl, 5 mM KCl, 0.4 mM MgCl2, 10 mM HEPES-Na (pH 7.4), 10 mM NaHCO3, 10 mM glucose, and 5 mM NaHPO4]. They were then resuspended in the HPMI buffer and incubated with 0.25 μM calcein AM at 37°C for 10 min (concentration of solvent in medium, 0.5% DMSO). Incubation was terminated by centrifugation, followed by washing of the cell pellet with HPMI buffer. The final pellet was resuspended in 1 ml of lysis solution [0.25 M sucrose, 10 mM Hepes (pH 7.6), and 1 mM EDTA]. Fluorescence was measured at 493/515 nm. A high level of fluorescence indicates a high intracellular retention of calcein, reflecting a low activity of pgp.
Effects of pgp on the accumulation of vinblastine, testosterone, and cortisol in these cell lines were determined by a modification of the published procedure using radiolabeled substrate (Stein et al., 1994). Cells were seeded onto six-well plates at a density of 105 cells/well and grown for 3 to 4 days in 5 ml of medium. For the accumulation of the steroids, cells were washed three times with 5 ml of phosphate-buffered saline, leaving them for 10 min at room temperature between each wash. Subsequently, the experiment was carried out using serum-free medium. To measure vinblastine accumulation, cells were washed only once and accumulation was studied using serum-containing medium. To initiate the experiment, cells were incubated with 1 ml of fresh medium. After 30 min at 37°C in 5% CO2, another 1 ml of medium containing 1 μM tritiated substrate was added to a final concentration of 0.5 μM (concentration of solvent in medium, 0.5% DMSO). After 30 min (for vinblastine) or 1 h (for steroids), reactions were terminated by aspirating the medium and rinsing the cells three times with 5 ml of cold phosphate-buffered saline. Cells were lysed by adding 0.5 ml of 0.4 M NaOH and neutralized with 0.5 ml of 0.5 M NH4OAc, pH 6.6. Lysate (400 μl) was counted in a scintillant. Separate wells were seeded to determine the number of cells per well. Results were normalized against cell number.
Establishment and Characterization of Cell Lines.
The cell lines that were used for the coexpression of CYP3A4 and P450 reductase were the human ovarian cancer cell line 1847 (Louie et al., 1986) and a derivative cell line that had been previously selected to overexpress endogenous pgp. The latter cell line was isolated by treating the parental pgp-negative 1847 cells (termed MDR−cells) with the mutagen EMS followed by selection on increasing concentrations of paclitaxel (up to 10 μM). One of the resistant clones (termed MDR+ cells) and MDR− cells were subsequently transfected with an expression vector containing the CYP3A4 cDNA and selected on G-418. In the case of MDR+-derived cells, the medium also contained paclitaxel to prevent loss of pgp. The resulting clones were screened by immunoblotting for the expression of CYP3A4. One of the positive clones derived either from MDR− or MDR+ cells was transfected with an expression vector containing the human P450 reductase. Subsequent selection was on zeocin and G-418. Transfectants were screened for expression of the recombinant proteins and endogenous pgp.
Figure 1 shows the immunoblot analysis of the cell lines that were used in the present work. Pgp was undetectable in cells that had been grown in the absence of paclitaxel (Fig. 1, lanes 1 and 3). The expression of pgp in the 3A4/HR/MDR+ cells was weaker than in MDR+ cells (Fig. 1, cf. lanes 4 and 2). The antibody against CYP3A4 recognized two proteins in total cellular lysates, however the protein with the slower mobility was only detected in cells that had been transfected with the CYP3A4 cDNA. The level of this protein was rather similar in 3A4/HR/MDR−and 3A4/HR/MDR+ cells (Fig. 1, cf. lanes 3 and 4). These two cell lines also displayed similar expression levels of P450 reductase as determined by immunoblotting. This was confirmed by assaying the cytochrome c reductase activity (Table1), which was more than 10 times higher than the activity in cells that did not express recombinant P450 reductase. Importantly, cell lysates isolated from 3A4/HR/MDR− and 3A4/HR/MDR+ cells displayed rather similar CYP3A4-mediated testosterone 6β-hydroxylase activities (Table 1). This activity was not detected in cells that did not express the recombinant proteins, namely, MDR+ and MDR− cells.
The activity of pgp in 3A4/HR/MDR− and 3A4/HR/MDR+ cells was determined by measuring the accumulation of calcein AM ester, vinblastine, testosterone, and cortisol (Fig. 2). Compared with 3A4/HR/MDR− cells, accumulation of calcein, vinblastine, and cortisol was strongly reduced in 3A4/HR/MDR+ cells. This effect was not as strong for testosterone. These experiments were also performed at a concentration of testosterone and cortisol of 25 instead of 0.5 μM, yielding similar results (data not shown).
Effect of pgp on CYP3A4-Dependent Metabolism.
Testosterone has been shown to be a poor substrate of pgp, whereas cortisol has been reported to be efficiently transported by pgp (Barnes et al., 1996). The effects of pgp on CYP3A4-mediated metabolism were investigated by determining the kinetic parameters of the testosterone 6β-hydroxylase and the cortisol 6β-hydroxylase in cultured 3A4/HR/MDR+ and 3A4/HR/MDR− cells without prior trypsinization (Fig. 3; Table2), under conditions that were linear with time and cell number (data not shown). The two reactions followed linear Lineweaver-Burk kinetics irrespective of whether pgp was expressed (Fig. 3). The apparent K m of the testosterone 6β-hydroxylase was almost 2-fold higher in 3A4/HR/MDR+ cells compared with their counterparts that did not express pgp (Table 2). Differences in the apparent K m were even more pronounced for the cortisol 6β-hydroxylase activity, with 3A4/HR/MDR+ cells displaying a 4-fold higherK m compared with 3A4/HR/MDR− cells. TheV max of testosterone 6β-hydroxylase activity in the cell line that expressed pgp was 3-fold lower than in cells that did not express this transporter (cf. 3A4/HR/MDR− versus 3A4/HR/MDR+ cells). Interestingly, theV max of the cortisol 6β-hydroxylase was similar in cells that expressed the transporter compared with those that were deficient in pgp.
Effects of CYP3A4 Inhibitors on P450 Activity Are Modulated by pgp.
The following experiments were performed to investigate the impact of pgp on CYP3A4-mediated metabolism in the presence of weak and strong inhibitors of this P450.
The effects of quinidine, LY335979, which had been originally developed as a reversal agent, verapamil, ketoconazole, and α-naphthoflavone on the testosterone 6β-hydroxylase activity of CYP3A4 were analyzed in membrane fractions containing CYP3A4 as well as in intact 3A4/HR/MDR− cells and 3A4/HR/MDR+ cells at a concentration of 40 μM testosterone. In these experiments, membranes isolated from E. coli, which coexpressed CYP3A4 and human P450 reductase, were used (Fig. 4). LY335979 and ketoconazole were the most potent inhibitors of CYP3A4, with the metabolic IC50 of both compounds being less than 1 μM. The IC50 of verapamil and α-naphthoflavone was higher than 5 μM. Quinidine was a very weak inhibitor of testosterone 6β-hydroxylase activity, with the IC50 being higher than 100 μM. Compared with the results obtained with membrane fractions, qualitatively similar effects of the inhibitors on the activity of CYP3A4 were noted in cultured cells that expressed CYP3A4 in the absence of pgp (3A4/HR/MDR− cells; Fig.5, striped columns). As expected from the experiments using bacterially expressed CYP3A4, increasing concentrations of the compounds lead to increased inhibition of CYP3A4 activity. Again, ketoconazole and LY335979 were the most potent inhibitors of testosterone 6β-hydroxylase activity. Interestingly, α-naphthoflavone affected the testosterone 6β-hydroxylase activity of intact cells only slightly. Importantly, in the presence of pgp (3A4/HR/MDR+ cells; Fig. 5, striped columns) the effects of quinidine and LY335979 on testosterone 6β-hydroxylase activity were different than those seen in membranes or in intact 3A4/HR/MDR− cells. Quinidine at a concentration of 5 and 20 μM stimulated CYP3A4 activity and LY335979 failed to inhibit this enzyme.
Cell lines were developed that allow investigations into some pharmacokinetic consequences resulting from P450-catalyzed metabolism in the presence of pgp. Pgp is absent in 3A4/HR/MDR− cells as revealed by immunoblotting (Fig. 1) and by PCR (data not shown) but is clearly present in 3A4/HR/MDR+ cells. This is also reflected by the decreased accumulation of calcein, vinblastine, testosterone, and cortisol in 3A4/HR/MDR+ compared with 3A4/HR/MDR− cells (Fig. 2). As expected from published data on the activity of pgp toward cortisol and testosterone (Barnes et al., 1996), the presence of this transporter in the former cell line reduced the accumulation of cortisol more strongly than that of testosterone. Even though anti-CYP3A4 antibodies reacted with a protein in cells that had not been transfected with the CYP3A4 cDNA (Fig. 1) this protein is unlikely to represent a CYP3A isoform since untransfected cells did not display any testosterone 6β-hydroxylase activity.
Steroids were used for the characterization of CYP3A4-mediated metabolism in pgp-positive and -negative cell lines, because the activities of both proteins toward these entities were previously well characterized. The kinetics of testosterone 6β- and cortisol 6β-hydroxylase activity in both cell lines did not deviate from linearity when analyzed according to Lineweaver-Burk (Fig. 3). Also analyzing the data according to Eadie-Hofstee failed to detect any cooperativity in these reactions. A small cooperative effect (Hill coefficient of 1.3) had been previously noted for the testosterone 6β-hydroxylase activity of CYP3A4 in reconstituted membrane vesicles (Ueng et al., 1997). The K m of CYP3A4 toward cortisol and testosterone (as determined by the 6β-hydroxylase activity of CYP3A4) had been found by others to be 10 and 56 μM, respectively (Ged et al., 1989; Ding et al., 1997). These values are similar to those found by us in intact 3A4/HR/MDR− cells. However, in the presence of pgp, the apparent K m for the 6β-hydroxylase activity of CYP3A4 toward testosterone and cortisol was increased by a factor of 1.7 and 4, respectively (Table 2, 3A4/HR/MDR+ cells). This difference can be predicted from the higher activity of pgp toward cortisol compared with testosterone (Barnes et al., 1996; our data), resulting in a stronger intracellular depletion of the former compared with the latter steroid. It is unlikely that pgp affected the apparentK m of the P450 by somehow altering the intrinsic affinity of CYP3A4 for the substrate. It is furthermore unlikely that differences in the apparentK m resulted from the slightly lower level of CYP3A4 in 3A4/HR/MDR+ compared with 3A4/HR/MDR− cells (Table 1).
In accordance with literature (Ged et al., 1989; Ding et al., 1997), the V max of CYP3A4 toward cortisol was much lower than that toward testosterone and was not significantly modulated by the expression of pgp (Table 2). However, the expression of pgp lead to a 3-fold reduction of theV max for the testosterone 6β-hydroxylase activity. This strong reduction cannot be explained by the slightly lower level of CYP3A4 in cells that overexpress pgp (Table1). One may speculate that the net accumulation of substrate, which is determined by uptake and pgp-mediated efflux, became rate limiting for the rapid metabolism of testosterone but not for the slow metabolism of cortisol.
The ratio of cortisol to 6β-hydroxycortisol has been used as an in vivo marker activity for CYP3A4 (Ged et al., 1989; Nakamura et al., 1997). Our finding that pgp strongly affected the apparentK m of CYP3A4 toward cortisol could have relevance for the use of this marker, provided that the intracellular level of cortisol is below theK m of the cortisol 6β-hydroxylase and that the level of hepatic pgp influences the concentration of this steroid in hepatocytes. The plasma level of free cortisol, which unlike protein-bound cortisol is responsible for the biological effects of this steroid, has been reported to be maximally 0.5 μM and its entry into cells is mainly by diffusion (Milgrom, 1990). Modulation of cortisol 6β-hydroxylase activity by pgp in the liver could explain the observation that the correlation between this enzyme activity and CYP3A4 protein level was only moderate when applied separately to either patients that had been untreated or to individuals that had been treated with rifampicin (Ged et al., 1989). This correlation became only strong when applied to both groups together, thus showing, as discussed previously (Ged et al., 1989), that cortisol 6β-hydroxylase activity can be used as a marker of CYP3A induction but not to assess interindividual variations in CYP3A protein levels. In this respect, it is important to note that levels of pgp in the liver showed marked interindividual differences (Schuetz et al., 1995) and that rifampicin is a potent inducer of pgp (Schuetz et al., 1996). It remains to be seen whether intraindividual differences in the level of pgp may also be partially responsible for the lack of correlation between marker activities of CYP3A4 such as erythromycin N-demethylase, dapsone N-hydroxylase, cortisol 6β-hydroxylase, and midazolam 1′-hydroxylase (Kinirons et al., 1999), with the added complication that some of these metabolic probes will be more sensitive to intestinal pgp, whereas others may be more influenced by hepatic pgp.
Metabolic in vivo probes have been also used to predict pharmacokinetic drug-drug interactions relating to CYP3A4. One study investigated the effects of human immunodeficiency virus protease inhibitors on dapsoneN-hydroxylation and cortisol 6β-hydroxylation in vivo (Gass et al., 1998). Interestingly these compounds, some of which have been shown to interact with pgp, failed to affect both activities of CYP3A4 in vivo as would have been expected from in vitro experiments. Another group investigated the effects of stiripentol on CYP3A4 activities in vivo and found that this compound inhibited CYP3A4-catalyzed carbamazepine metabolism but that the cortisol/6β hydroxycortisol ratio did not prove to be a reliable indicator of this inhibition (Tran et al., 1997). Here, we wanted to explore the possibility that pgp may modulate drug-drug interactions acting via CYP3A4. We have investigated the effects of some pgp inhibitors on CYP3A4 mediated reactions in intact 3A4/HR/MDR− and 3A4/HR/MDR+ cells and in membrane fractions containing recombinant CYP3A4 (Figs. 4 and 5). Verapamil, LY335979, and quinidine have been previously tested as reversal agents. LY335979 was found to be more than 1000-fold as potent as verapamil in the inhibition of pgp-dependent accumulation of daunomycin (Dantzig et al., 1996). Moreover, this study showed that LY335979 at a concentration of 100 nM lowered the cytotoxicCIC50 of vinblastine in pgp-overexpressing cells 400-fold, whereas verapamil at a concentration of 5 μM lowered the cytotoxicCIC50 only 50-fold. However, no data are available on the interaction of LY335979 with CYP3A4. Recently, the selectivity of several reversal agents toward CYP3A4 and pgp was published (Wandel et al., 1999). This study showed that quinidine was more selective than verapamil to inhibit preferentially pgp over CYP3A4, which is in agreement with our data that quinidine, but not verapamil, stimulated the testosterone 6β-hydroxylase activity in intact 3A4/HR/MDR+cells due to preferential inhibition of pgp compared with CYP3A4 (Fig.5), thereby making more substrate available to the P450. A similar but weaker effect was also noted with LY335979. This demonstrates that compounds that are stronger inhibitors of pgp than of CYP3A4 can stimulate the activity of this P450 in intact cells. Stimulation of testosterone 6β-hydroxylase activity by quinidine (5 μM) was 2-fold in cells that expressed pgp (Fig. 5), whereas the presence of pgp decreased the concentration only by 1.5-fold. This difference is not entirely unexpected because Michaelis-Menten kinetics predicts thatV a/V b= (V max ×S a/K m+S a)/(V max×S b/K m+ S b), withS a andS b being the concentrations of testosterone in the presence and absence, respectively, of pgp andV a andV b being the velocities of the testosterone 6β-hydroxylation under these two conditions. Note that the concentration of testosterone in the experiments displayed in Fig.5 was 40 μM and the K m for the testosterone hydroxylase is 60 μM.
The inhibitory effects of verapamil on CYP3A4 activity seem to plateau in 3A4/HR/MDR+ but not in 3A4/HR/MDR− cells. This most likely results from inhibition of both CYP3A4 and pgp with both inhibitions having an opposing impact on CYP3A4 activity, albeit only in pgp-positive cells. However, the resulting net effect on CYP3A4 activity is difficult to predict accurately.
The data presented demonstrate that the expression of pgp, as predicted by theoretical considerations, impinges on CYP3A4-mediated metabolism by influencing the apparent K m values of the enzyme in intact cells. In cases such as quinidine where a compound is oxidized by CYP3A4 at two different positions with different K m values (Sanwald et al., 1996; Nielsen et al., 1999), product ratios can be expected to be influenced by the activity of pgp. Even though our observations are mainly pertinent to reactions catalyzed by CYP3A4, it should be noted that debrisoquine, which has been used as a metabolic in vivo probe for CYP2D6, is a substrate of pgp (Kim et al., 1999). In addition, we show that the influence of pharmacokinetic drug-drug interactions on CYP3A4 is strongly modulated by the activity of pgp.
- Received July 3, 2000.
- Accepted September 27, 2000.
Send reprint requests to: Dr. T. Friedberg, Biomedical Research Center, University of Dundee, Ninewells Hospital and Medical School, Dundee, DD1 9SY, UK. E-mail:
This work was supported by a grant of the German Research Council (to J.M.B).
- cytochrome P450
- human P-glycoprotein
- calcein AM
- calcein acetoxymethylester
- high performance liquid chromatography
- dimethyl sulfoxide
- concentration of inhibitor that reduces metabolism of substrate by 50%
- concentration of cytotoxic compound that reduces cell survival by 50%
- The American Society for Pharmacology and Experimental Therapeutics