Introduction

Top Abstract Introduction Materials and Methods Results Discussion References

Cytochrome P450 (P450) monooxygenases comprise a superfamily (termed CYP) of membrane-bound hemoproteins that catalyze the metabolism of a wide variety of endogenous and exogenous compounds (Porter and Coon, 1991; Capdevila et al., 1992; Gonzalez, 1992). Sometimes these reactions lead to more reactive and consequentially more cytotoxic metabolites. P450s have also been implicated in the metabolism of several clinically important chemotherapeutic agents to more reactive intermediates (Kivisto et al., 1995). In the case of cyclophosphamide, several P450 isoforms that catalyze the conversion of this prodrug to a cytotoxic drug (Ren et al., 1997) have been identified. Using human liver microsomes and P450 isoform-specific inhibitors, P450s have also been shown to be involved in the metabolic activation of doxorubicin and of the more cytotoxic morpholino-doxorubicin (Lewis et al., 1992; Goeptar et al., 1994). Doxorubicin is reduced by P450s as well as by NADPH-P450 reductase (hOR) to radicals that covalently modify DNA. P450s have been shown to metabolize other important chemotherapeutic agents such as epipodophyllotoxins, taxol, and vinca alkaloids (Zhou et al., 1993a,b; Relling et al., 1994; Royer et al., 1996). However, the relevance of this metabolism for the toxicity and the chemotherapeutic value of these anticancer drugs remains unfortunately unknown.

Vinca alkaloids such as vincristine and vinblastine are clinically important chemotherapeutic drugs, and are used for the treatment of a variety of tumors (Rowinsky and Donehower, 1996). Vincristine and vinblastine differ in their therapeutic value for different neoplasms. In addition, they show a different spectrum of organ toxicity (Rowinsky and Donehower, 1996). Using human liver microsomes and P450 isoform-specific inhibitors, it was shown that vinca alkaloids are metabolized by CYP3A enzymes (Zhou et al., 1993a,b), and inducers of that P450 have been shown to increase the elimination of vincristine (Villikka et al., 1999). The microsomal P450-mediated metabolism of these compounds showed a broad interindividual variability that correlated with the CYP3A levels in human liver microsomes. This may explain the large interindividual variation in the clinical pharmacokinetics of vinca alkaloids. However, in the absence of data on the biological activity of the resulting primary or secondary metabolites, the relevance of this observation for the desired and adverse effects of vinca alkaloids remains unfortunately unknown.

Tumor cells may become refractory to treatment with vinca alkaloids by a variety of mechanisms. These include alterations in the structure of tubulin proteins (Houghton et al., 1985) or increased expression of the multidrug resistance protein MDR1 (P-glycoprotein), which is known to mediate the cellular export of several drugs (Ueda et al., 1987). In vitro, prolonged treatment of cells with compounds that are substrates for MDR1 often selects for cells that overexpress this protein. It is intriguing that most substrates of the major hepatic P450 isoform CYP3A4 are either transported by MDR1 or are modulating the activity of this transporter. For example, vinca alkaloids, colchicine, and cyclosporine are substrates for both CYP3A4 and MDR1, whereas nifedipine and verapamil are substrates for the P450 but inhibitors for MDR1 (Kivisto et al., 1995; Schuetz et al., 1996). It could be speculated that both proteins coevolved to protect cells from toxic noxes from environmental toxins.

Some P450s, including CYP3A4, have been shown to be overexpressed in tumors (Murray et al., 1993a,b). However, mechanisms leading to the development of drug resistance by the overexpression of proteins have been mainly studied in cell culture. Because it is known that the expression of several P450 isoforms is usually strongly decreased on cultivation of cells (LeCluyse et al., 1996; Maurel, 1996), P450-mediated drug resistance may have escaped detection. In this study, we have determined the role of P450-mediated metabolism in the toxicity and in the development of drug resistance to vinca alkaloids.

	Materials and Methods

Top Abstract Introduction Materials and Methods Results Discussion References

Cell Culture and Establishment of Cell Lines. Cell lines described in this study were derived from a dihydrofolate reductase negative (DHFR) Chinese hamster ovary (CHO) cell line, designated DUKXB11 (Kaufman, 1990). The isolation of the cell lines that stably expressed CYP3A4 either alone or together with hOR has been described recently (Ding et al., 1997). A similar procedure was used to establish the CHO cell lines that expressed the cell line that coexpressed CYP2D6 and hOR. For mass propagation, the cells that coexpressed P450s and hOR were maintained in the presence of methotrexate and G-418 in -minimal essential medium (Life Technologies, Paisley, UK) containing dialyzed 10% fetal bovine serum (FBS). These drugs were removed at least 3 days before the cytotoxicity experiments. Cell lines were characterized for P450-mediated metabolism of testosterone (Ding et al., 1997) and bufuralol (Pritchard et al., 1998). P450 levels were determined spectrophotometrically in total cellular protein isolated from these cells (Ding et al., 1997). To generate a cell line that expressed the -galactosidase, the expression vector pSV--gal (containing the -galactosidase cDNA under the control of the SV40 promoter) was cotransfected with the vector pBSpacp [containing the puromycin selection marker (de la Luna et al., 1988)] into DUKXB11 cells. Transfectants were selected with puromycin (5 µg/ml) in Dulbecco's modified Eagle's medium (DMEM) containing 10% FBS to generate the cell line CHO/GAL After mass propagation in the presence of puromycin, the antibiotic was removed.

Coexpression of P450s and hOR in Escherichia coli The coexpression of hOR together with either CYP2D6 or CYP3A4 in E. coli has been described recently (Blake et al., 1996; Pritchard et al., 1998). In brief, CYP3A4 was modified within its N terminus for expression in E. coli as described by others (Gillam et al., 1993). CYP1A2, CYP2C8, and CYP2D6 were modified for expression by an N-terminal fusion to a bacterial ompA leader sequence and expressed from the plasmid pCW (Pritchard et al., 1997, 1998). hOR was coexpressed from a separate plasmid (pACYC184) under the control of the (tac)₂ promoter (Pritchard et al., 1998). The condition for the expression of P450s and the isolation of bacterial membranes was as described recently (Gillam et al., 1993; Blake et al., 1996; Pritchard et al., 1998).

Metabolism of Vinca Alkaloids. [³H]Vinblastine (14 Ci/mM) was purchased from Amersham-Pharmacia (Little Chalfont, UK). The assay conditions and the analysis of the metabolites by HPLC have been described by Zhou et al. (1993a), except that [³H]vinblastine (0.1 Ci/mM) was used at a concentration of 10 µM. Approximately 0.3 mg of cell lysate protein (2 mg/ml final concentration) were incubated with vinblastine at 37°C for 60 min in the presence and absence of NADPH.

Cytotoxicity Assay. Two thousand cells were seeded in a 96-well plate and cultured in DMEM medium containing 10% FBS, but no G-418 or methotrexate, for 24 h. Subsequently, the cells were exposed to varying concentrations of vinca alkaloids, usually for 3 days unless otherwise noted. At the end of the experiment, the number of viable cells was determined by the ATP bioluminescent assay (Dorr et al., 1988) according to the manufacturer (Sigma, Poole, UK) instructions.

Effects of Vinblastine on Immunodetectable CYP3A4. CHO cells that expressed CYP3A4 alone or together with hOR were seeded into DMEM containing FBS and grown for 24 h. Subsequently, cells were refed with the same medium without or with vinblastine (100 nM) and grown for another 3 days. The cells were harvested and total cellular protein was prepared by sonication and analyzed by immunoblotting using anti-CYP3A4 antibody or the anti-MDR1 JSB-1 (Chemicon, London, UK) antibody and horseradish peroxidase-coupled secondary antibody (SAPU, Carluke, Scotland). The blot was developed using an enhanced chemiluminescence kit (ECL; Amersham, Buckinghamshire, UK).

Determination of Selective Survival in Recombinant Cells. To determine whether CYP3A4 expression confers selective growth advantage in the presence of vinblastine, the following experiments were performed with recombinant CHO cells. The CHO/GAL cell line and the cell line that expressed CYP3A4 and hOR were cultured either separately or in coculture in the absence or presence of vinblastine (900 nM) for 24 h. -Galactosidase was visualized by incubating the cells with X-gal according to the method provided by the supplier of the vector pSV--gal (Promega, Southampton, UK). After this, the cells were permeabilized with 1% Triton X-100, fixed with 3% paraformaldehyde in PBS, subsequently reacted with anti-CYP3A4 antibody (raised in rabbit), washed, and incubated with horseradish peroxidase-coupled anti-rabbit IgG. The slides were developed with the Dab fast kit (Sigma), containing diaminobenzidine.

Determination of CYP3A4-Mediated Resistance in Tumor Cells. To investigate whether CYP3A4 expression in human tumor cells contributes to drug resistance to vinblastine, the following experiment was performed: LS174 cells were seeded and grown for 24 h in RPMI medium containing 10% fetal calf serum to a confluence of 20%. Subsequently, either rifampicin (final concentration 10 µM) or solvent [dimethyl sulfoxide (DMSO), final concentration less than 0.1%] was added. After 3 days, vinblastine (final concentration 80 nM) or DMSO (final concentration less than 0.2%) was added. After another 3 days, cells were fixed and permeabilized with formaldehyde and Triton X-100 as described above. Cells were reacted with anti-CYP3A4 antibody. For immunofluorescence, cells were incubated with fluorecein isothiocyanate-labeled secondary antibody. Nuclei were counterstained with 4,6-diamidino-2-phenylindole (Zhang et al., 1999).

Preincubation of Vinblastine with Recombinant P450s. Vinblastine was preincubated either with CHO cells coexpressing CYP3A4 and hOR or with membranes isolated from E. coli that coexpressed recombinant P450 isoforms and hOR. Subsequently, the cytotoxicity of the preincubation mix was assayed using parental CHO cells. For preincubation with mammalian cells, 1 × 10⁶ cells were seeded and grown for 24 h. Fresh medium containing 1 µM vinblastine was added and incubated for 3 days. This preincubation mixture was used for the cytotoxicity assay as indicated above after dilution with medium (DMEM) to achieve the final concentrations given in Fig. 3. For preincubation with membranes isolated from E. coli, vinblastine (30 µM) was incubated in 0.1 M phosphate buffer, pH 7.4, containing 30 mM Mg²⁺ in the presence and absence of 0.5 mM NADPH. Each incubation contained 500 pmol of recombinant P450. Preincubation was carried out at 37°C for 4 h. Subsequently, this preincubation mixture was sterile filtered through 0.2-µm filters. Cytotoxicity was assayed as described above using parental CHO cells after dilution of the mixture with medium to the concentrations indicated in Fig. 4.

	Results

Top Abstract Introduction Materials and Methods Results Discussion References

P450-Mediated Metabolism of Vinca Alkaloids. In this investigation, we used CHO cell lines that coexpressed hOR together with CYP2D6 or CYP3A4. CHO cells, which coexpressed CYP3A4 together with P450 reductase (hOR), have been described recently (Ding et al., 1997). In addition, we used cells that expressed CYP3A4 but no hOR. The levels of spectrally detected P450 and associated enzyme activities toward prototypical substrates for these cell lines are given in Table 1. The parental cell line did not contain spectrally detected P450 or display P450 enzyme activity toward the substrates given in Table 1. In the recombinant cell lines, the expression levels of the P450 isoforms and their activities were either similar or higher than those found in human liver microsomes. The exception was the cell line that expressed CYP3A4 alone, which had a 10-fold lower testosterone 6-hydroxylase activity than cells that contained CYP3A4 and hOR.

View this table: [in this window] [in a new window]   TABLE 1 Characterization of the recombinant cell lines used in this work The level of spectrally active P450 was determined from the Fe2+ versus Fe2+-CO difference spectrum of total cellular protein isolated from the various cell lines. The following enzyme activities were assayed in cultured cells: bufuralol-hydroxylase (CYP2D6) and testosterone 6-hydroxylase (CYP3A4). The level of hOR in the various recombinant cell lines was between 190 and 225 nmol of cytochrome c reduced/min/mg of total cellular protein, except the cell line that expressed CYP3A4 alone and the parental cell line for which this value was between 7 and 12 nmol of cytochrome c reduced/min/mg of total cellular protein. Testosterone 6-hydroxylase and bufuralol 1'-hydroxylase activities were reported to be 5000 and 70 nmol/min/mg of protein, respectively, in human liver microsomes (reviewed in Friedberg et al., 1999. Data for total cellular protein isolated from hepatocytes were not available.

1

Role of P450s in the Toxicity of Vinca Alkaloids. The cytotoxicity of vincristine and vinblastine was determined by incubating the recombinant cell lines with increasing concentrations of the vinca alkaloids for 3 days. Cytotoxicity curves for vincristine were rather similar for all cell lines tested up to a concentration of 3.6 nM (Fig. 1). At this concentration up to 80% of the cells survived. The IC₅₀ for all cell lines was approximately 11 nM. However, beyond this concentration, which is much lower than the systemic concentration (50-100 nM) reached during a bolus injection (Rowinsky and Donehower, 1996), an additional increase in the concentration of vincristine did not increase its cytotoxicity for the cell line that coexpressed CYP3A4 and hOR. In contrast, more than 90% of the cells that expressed CYP2D6 together with hOR or the hOR alone were sacrificed at a concentration of 300 nM. Note that coexpression of CYP3A4 and hOR did not shift the entire cytotoxicity curve to the right.

View larger version (15K): [in this window] [in a new window]   Fig. 1.   Cytotoxicity assay for vincristine. The parental DUKXB11 cells and the various recombinant cell lines were exposed to increasing concentrations of vincristine for 3 days. Subsequently, the number of viable cells was determined using the ATP bioluminescent assay as described under Materials and Methods. For each cell line, the number of cells obtained in the absence of vinblastine was set as 100%. The parental DUKXB11 cells () and the cells expressing the following proteins were assayed: hOR (); CYP3A4 (); CYP2D6 and hOR (); and CYP3A4 and hOR (). The data represent the means of two independent experiments with three determinations per concentration. The S.D. was less than 5% of the means. **, significantly different (P  .01) from control (DUKXB11 cells).

View larger version (14K): [in this window] [in a new window]   Fig. 2.   Cytotoxicity assay for vinblastine. The experiment was performed as described in the legend to Fig. 1 but using vinblastine. The symbols for the different cell lines are the same as in Fig. 1. The data represent the means of two independent experiments with three determinations per concentration. The S.D. was less than 5% of the means. **, significantly different (P  .01) from control (DUKXB11 cells).

View larger version (24K): [in this window] [in a new window]   Fig. 3.   Cytotoxicity of vinblastine after preincubation with mammalian cells. Vinblastine (1 µM) was preincubated with parental CHO (DUKXB11) cells () or with cells that coexpressed CYP3A4 and hOR () for 3 days. Subsequently, the preincubation mix (as defined under Materials and Methods) was diluted to the concentrations indicated in the figure and used in a cytotoxicity assay as described in the legend to Fig. 2, using parental CHO cells as the target. The S.D. was less than 10% of the means. **, significantly different (P  .01) from control (DUKXB11 cells).

View larger version (28K): [in this window] [in a new window]   Fig. 4.   Cytotoxicity of vinblastine after preincubation with E. coli membranes containing recombinant P450s. Vinblastine (30 µM) was preincubated in the absence (open symbols) or presence (closed symbols) of NADPH with membranes isolated from E. coli expressing the hOR together with the following P450 isoforms: CYP3A4 (A); CYP1A2 (B); CYP2C8 (C); and CYP2D6 (D). After incubation for 4 h at 37°C, the mixture (composition see Materials and Methods) was sterile filtered, diluted to the concentration indicated in the figure, and subsequently tested for cytotoxicity using parental CHO cells. The S.D. was less than 10% of the means. *, significantly different (P  .05) from control (DUKXB11 cells).

P450s and Development of Drug Resistance. The role of CYP3A4 for the development of drug resistance in a heterogeneous cell population during treatment with chemotherapeutics was evaluated in a coculture of a cell line that coexpressed CYP3A4 and hOR together with a cell line that expressed -galactosidase. The selective cytotoxicity of vinblastine was determined either histochemically by staining for CYP3A4 and for -galactosidase (see Materials and Methods). Histochemical analysis demonstrated that cells that reacted with the CYP3A4 antibody, as revealed by brown staining, survived treatment with vinblastine much better than the cells that expressed -galactosidase as revealed by blue staining (Fig. 5).

View larger version (112K): [in this window] [in a new window]   Fig. 5.   Effects of vinblastine on a coculture of cells that expresses either CYP3A4 or -galactosidase. Cells that expressed -galactosidase (Gal) or CYP3A4 together with hOR (3A4) and a coculture of these two cell lines (Gal/3A4; in a ratio of 2:1) were exposed to vinblastine (+V) at a concentration of 900 nM for 3 days. Subsequently, the cells were stained for -galactosidase (blue) and for CYP3A4 (brown) as described under Materials and Methods. Also shown is the analysis of the cocultured cell lines that had been grown in the absence of vinblastine (V).

View larger version (29K): [in this window] [in a new window]   Fig. 6.   Influence of vinblastine on the level of immunodetectable CYP3A4 in recombinant CHO cells and in intestinal cancer cells. A, CHO cells that expressed recombinant CYP3A4 alone or CYP3A4 and hOR were grown with (+) and without () vinblastine (V, 100 nM) for 3 days and, subsequently, the level of CYP3A4 was determined by immunoblotting as described under Materials and Methods. Human liver microsomes (HLM) were included for comparison. In each lane, 20 µg of protein was analyzed. B, human intestinal cancer cell line LS174T was grown in the presence (+) or absence () of rifampicin (R) as described under Materials and Methods. Subsequently, the medium was exchanged against medium containing, in addition, vinblastine (V) at the indicated concentrations. After 3 days the levels of CYP3A4 were determined by immunoblotting. Forty micrograms of total cellular lysate from LS174T cells was analyzed. For comparison, cell lysate from CYP3A4 and hOR CHO cells (2.5 µg) was included. C, expression of mdr1 in CHO, CHO (CYP3A4 and hOR), and LS174T cells. Total cellular protein (20 µg) isolated from either CHO or LS174T cells was analyzed by immunoblotting using anti-MDR1 (JSB-1) primary antibody, which, according to the supplier (Chemicon, London, UK), will recognize Chinese hamster mdr1 and human MDR1. As positive control, total cellular lysate (5 µg) isolated from the MDR1-overexpressing and vinblastine-resistant human ovarian cell line A1847 (Louie et al., 1986) was used. CHO (CYP3A4 and hOR) cells were also analyzed after exposure to vinblastine (100 nM) for 3 days. LS174T cells were either untreated or treated with rifampicin (10 µM) or vinblastine (80 nM) or rifampicin (3 days exposure) followed by vinblastine (subsequent exposure for 3 days).

View larger version (76K): [in this window] [in a new window]   Fig. 7.   Immunofluorimetric detection of CYP3A4 in untreated and treated LS174T cells. Cells were seeded and grown on a coverslip for 24 h. Growth was continued for 3 days in the presence or absence of rifampicin (10 µM). Subsequently, vinblastine (80 nM final) or DMSO (less than 0.2% final) was added. After 3 days, cells were fixed and permeabilized as described under Materials and Methods. For detection of CYP3A, cells were reacted with anti-CYP3A4 primary antibody and subsequently with fluorescein isothiocyanate-labeled secondary antibody. Nuclear DNA was detected using 4,6-diamidino-2-phenylindole. Note that the central region of the cell in the lower part of the third panel did not react with the anti-CYP3A4 antibody even though its nucleus is not located in this area. This is most likely due to incomplete permeabilization.

	Discussion

Top Abstract Introduction Materials and Methods Results Discussion References

Using human liver microsomes, previous studies have demonstrated that vinca alkaloids were metabolized by CYP3A proteins, and no evidence for the involvement of other P450 isoforms was found (Zhou et al., 1993a,b). We confirmed these results by showing that cells that coexpressed CYP3A4 and hOR were able to metabolize vinblastine. In our experiments, metabolism of vinblastine could not be detected in membranes isolated from E. coli that expressed either CYP1A2 or CYP2C8 or CYP2D6 (data not shown). These results strongly suggest that the interindividual variability in the pharmacokinetics of vinblastine are governed by the variation in CYP3A4 levels as proposed previously. Until now, however, the role of P450-catalyzed metabolism for the systemic toxicity of vinca alkaloids and the toxic potential of the resulting metabolites was not known.

Our data demonstrate, using several experimental approaches, that CYP3A4 detoxifies vinca alkaloids. One may argue that the increased resistance of the CYP3A4-expressing cell lines to vinca alkaloids may be unrelated to P450 expression but could have been due to increased expression of mdr1 in the recombinant cell lines. Expression of mdr1 was not detectable in the various recombinant cell lines used in this investigation (Fig. 6C), confirming that CHO cells contain only low levels of mdr1 (Turner and Curtin, 1996; Petriz et al., 1997), which are not increased to a detectable level during establishment of the cell lines used in this study. MDR1 expression also remained undetectable after treatment of the cell lines with vinblastine. However, it should be mentioned that attempts to study the effects of inhibitors of CYP3A4 (troleandomycin and ketoconazole) on the cytotoxicity of vinblastine were not conclusive, because these compounds already increased the cytotoxicity in the parental cell line, possibly due to an inhibition of drug transport proteins for vinca alkaloids. These transporters may include mdr1 (inhibitable by ketoconazole but not detectable in our cell lines) or alternatively mrp1, which has been shown to transport vinblastine and which is also inhibited by ketoconazole or its derivatives (Siegsmund et al., 1994; Hollo et al., 1996). These results are in agreement with earlier observations that drug efflux can be reduced in CHO cells by inhibitors of mdr1 (or related proteins), despite the extremely low level of this protein in this cell line (Turner and Curtin, 1996; Petriz et al., 1997). Because CYP3A4 inhibitors could not be used to confirm the role of CYP3A4 in the detoxication of vinca alkaloids, an alternative approach was used (Figs. 3 and 4). The results from these experiments clearly show that preincubation of vinblastine with membranes containing recombinant CYP3A4 reduced the cytotoxicity of the preincubation mixture compared with preincubations with membranes that were devoid of this P450 isoform.

Surprisingly, in experiments using intact cells, expression of CYP3A4 and hOR did not shift the entire cytotoxicity curves for vinca alkaloids to the right (Figs. 1 and 2). Importantly, however, this shift was observed in experiments that assayed the cytotoxicity of vinblastine after preincubation with CHO cells or with E. coli membranes containing CYP3A4 (Figs. 3 and 4A). It is likely that the initial drop in the cytotoxicity curves (Figs. 1 and 2) for cells that have been transfected with CYP3A4 and hOR was caused by a fraction of cells that inadvertently expressed little CYP3A4. The remaining high expressing cells were then resistant to concentrations of vinca alkaloids up to 1 µM. This increased resistance on expression of CYP3A4 can be explained if the import of vinca alkaloids into the cells was becoming saturated compared with their metabolism by CYP3A4, resulting altogether in a decreased intracellular concentration of the cytotoxic parental compounds. The mean uptake rate for vinblastine into hepatocytes has been reported to be 0.568 pmol/min/10⁶ cells (Zhou et al., 1994), which is more than one order of magnitude less than the CYP3A4-mediated metabolism of this compound in the cell line that expressed CYP3A4 together with hOR. The heterogeneity of CYP3A4 expression levels in the cell line that had been transfected with CYP3A4 and hOR was also revealed by immunostaining (Fig. 5). We were not able to eliminate this heterogeneity despite repeated subcloning. Interestingly, exposure of cells that expressed CYP3A4 to vinblastine led to an elevated level of CYP3A4 as revealed by immunoblotting (Fig. 6A) most likely by eliminating cells that expressed low levels of this P450.

The latter result also suggests that CYP3A4-mediated detoxication of vinca alkaloids is not only relevant to the systemic toxicity of these anticancer drugs, but also implied that CYP3A4 could be involved in the development of drug resistance of tumors during chemotherapy. This was also confirmed by treating a coculture of cells that expressed -galactosidase together with cells that expressed CYP3A4 with vinblastine. In this experiment, the latter cell line survived preferentially. This experiment also establishes that depletion of vinblastine in the medium by CYP3A4-mediated metabolism is not responsible for the decreased cytotoxicity of vinca alkaloids. It is more likely that this metabolism reduces the intracellular concentration of these compounds, which have been shown to diffuse slowly through cellular membranes (see above; Zhou et al., 1994).

While the above experiments relied on recombinant cell lines, the data on the effect of vinblastine on the CYP3A level in LS174 tumor cells (Figs. 6B and 7) strongly indicate that CYP3A plays a role in drug resistance in human tumor cells. Reverse transcriptase-polymerase chain reaction using primers that will only amplify CYP3A4 but not CYP3A5 and CYP3A7 verified that CYP3A4 was expressed in LS174 tumor cells (data not shown) even though it cannot be excluded that these cells also express CYP3A7 or CYP3A5. However, the former P450 was not detectable in the related intestinal tumor cell line LS180 and the latter P450 was not inducible by rifampicin in LS180 cells (Schuetz et al., 1996). In our experiments, it was observed that treatment of LS174T cells with rifampicin led to a strong induction of a CYP3A protein, which most likely was CYP3A4 in only some cells with the remaining population showing no effect. However, after treatment with vinblastine for another 3 days, expression was seen in most surviving cells, which often formed clones. Our interpretation of this experiment is that preferentially cells that expressed CYP3A4 survived treatment. An alternative interpretation is that vinblastine as a ligand of CYP3A4 stabilized this protein and acted synergistically together with vinblastine to result in a superinduction of P450 in each cell. Even though we cannot exclude this mechanism, one would expect that stabilization of CYP3A4 can only occur in the few cells in which this P450 was previously induced by rifampicin, thus maintaining a heterogeneous expression pattern. This was not observed. Furthermore, it is unlikely that LS174T cells, which displayed high levels of CYP3A4 after treatment with rifampicin, were resistant against vinblastine due to simultaneous induction of MDR1, as this protein was undetectable in untreated cells, and even after treatment with rifampicin or vinblastine (Fig. 6C). In this respect LS174T cells, at least under our culture conditions, appear to behave differently from LS180 cells, which have been shown to respond to treatment with rifampicin with a strong induction of MDR1 (Schuetz et al., 1996).

The relevance of our in vitro study for the in vivo situation remains to be established. It is important to note that P450s of the CYP3A family have been shown to be overexpressed in a variety of human tumors, such as those of sarcomas (Murray et al., 1993a), breast (Murray et al., 1993b), and prostate (Murray et al., 1995). With respect to drug resistance, it is of particular interest that only the invasive part of breast carcinomas has been found to express CYP3A proteins (Murray et al., 1993b). However, these histochemical studies did not distinguish between CYP3A4 and CYP3A5.

In conclusion, we have demonstrated that CYP3A4 metabolizes vinblastine, and that these reactions lead to less toxic compounds. The role of CYP3A4 in the detoxication of vinblastine and vincristine could explain the clinically important idiosyncratic toxicity of these compounds. The detoxication of chemotherapeutic agents by P450-catalyzed metabolism is very unusual because, in most cases, P450s are known to activate a variety of anticancer drugs (Goeptar et al., 1994). The only known exception to this appears to be 1,3-bis(2-chloroethyl)-1-nitrosourea (BCNU) which, as shown only in a cell-free system, is detoxified independently by P450 and by glutathione S-transferases (Weber and Waxman, 1993). In addition, our results strongly suggest that CYP3A4 could be involved in the development of drug resistance during chemotherapy. To establish the significance of these processes in vivo, it will be necessary to examine the P450 expression level in tumor samples derived from patients who have undergone chemotherapy with vinca alkaloids and other chemotherapeutics. The observation that CYP3A4 acts as a selectable marker in the presence of some chemotherapeutic drugs may also provide a mechanism for the induction of CYP3A4 in hepatocytes exposed to taxol (Kostrubsky et al., 1998), which has been shown to be metabolized by this important P450 isoform and by CYP2C8.