Vincristine is a commonly used chemotherapeutic agent in the treatment of multiple types of malignancies including hematological cancers. For pediatric acute lymphoblastic leukemia (ALL), a cancer routinely treated with vincristine, the clinical outcomes for patients are often unpredictable and differ by race (McCune and Lindley, 1997; Pollock et al., 2000; Bhatia et al., 2002; Lange et al., 2002). For example, in an ALL pediatric study, the mortality rates of African-American children with vincristine combination chemotherapy were 42% greater than those of Caucasian children after correcting for known prognostic factors such as compliance, clinical presentation, and tumor characteristics (Pollock et al., 2000). In addition, in a retrospective analysis with ALL patients, Caucasian children compared with African-American children were more likely to experience vincristine-associated neurotoxicity (Renbarger et al., 2008). Thus, interracial variability in vincristine exposure due to differences in metabolism may in part explain the differences in clinical outcomes between African-American and Caucasian children. This hypothesis is the basis for the current study of vincristine metabolism and the role of CYP3A5, an enzyme more commonly expressed in African-Americans.

Vincristine is a substrate of CYP3A enzymes including CYP3A4 and the genetically polymorphic CYP3A5 (Dennison et al., 2006). Whereas only low-frequency genetic mutations of the CYP3A4 gene are known to affect CYP3A4 activity (He et al., 2005), CYP3A5 genetic variants are common (CYP3A5^*3, CYP3A5^*6, and CYP3A5^*7). Individuals without at least one CYP3A5^*1 allele are effectively devoid of hepatic CYP3A5 because only a small amount of active CYP3A5 is produced (“CYP3A5 low expressers”). For individuals with at least one active CYP3A5^*1 allele, significant quantities of CYP3A5 are expressed (“CYP3A5 high expressers”); approximately 75% of African-Americans, 47% of East Asians, and 19% of Caucasians express high levels of CYP3A5 (Xie et al., 2004). The relative amounts of CYP3A4 and CYP3A5 protein were summarized in a recent meta-analysis of human liver microsomal banks (n = 45 for CYP3A5 high expressers, Caucasian livers) (Perrett et al., 2007). For CYP3A5 high expressers, the mean contribution of CYP3A5 was approximately 40% of the total CYP3A protein. Thus, for CYP3A5 high expressers, substantial amounts of hepatic CYP3A5 protein are available to potentially metabolize a CYP3A substrate.

Previous studies from this laboratory investigated the potential of CYP3A5 genetic polymorphisms to influence vincristine metabolism using various noncellular enzyme systems (Dennison et al., 2006, 2007). Of the cDNA-expressed hepatic cytochrome P450 (P450) enzymes, only CYP3A4 and CYP3A5 readily metabolized vincristine to one primary metabolite, M1 (Fig. 1) (Dennison et al., 2006). We were surprised to find that unlike other known CYP3A substrates (Williams et al., 2002; Patki et al., 2003; Huang et al., 2004; McConn et al., 2004), vincristine was selectively metabolized by CYP3A5; the intrinsic clearance of vincristine with CYP3A5 was approximately 10-fold higher than that with CYP3A4 (Dennison et al., 2006). In addition, using human liver microsomes, vincristine was more efficiently metabolized by CYP3A5 high expressers compared with low expressers (5-fold higher hepatic clearance) (Dennison et al., 2007). Specifically for CYP3A5 high expressers, the contribution of CYP3A5 to vincristine metabolism was approximately 80% (Dennison et al., 2007). Taken together, the kinetic data using cDNA-expressed enzymes and human liver microsomes suggest that CYP3A5 expression significantly influences the P450-mediated metabolism of vincristine.

Although vincristine may be a substrate of CYP3A5, the in vivo clearance of vincristine may also be influenced by phase II metabolism or drug transport. To study these potential effects, human hepatocytes in suspension or other in vitro cell systems can be used. In addition, even for drugs that do not undergo phase II metabolism, metabolism with human hepatocytes may be more representative of in vivo metabolism than noncellular in vitro models because the isolated hepatocytes contain physiological concentrations of enzymes, coenzymes, and cofactors. In recent reports of in vitro/in vivo scaling with CYP3A substrates, the predicted clearances using cryopreserved human hepatocytes were on average 5-fold lower than the in vivo clearances, but they were more predictive than the clearances estimated using human liver microsomes (Brown et al., 2007). Thus, the current study evaluates the role of CYP3A5-mediated metabolism of vincristine using cryopreserved human hepatocytes. The objectives of the investigation were to identify any metabolites, evaluate the role of CYP3A5 genotype on metabolism, and evaluate the impact of transporter and P450 inhibitors on metabolite formation.

Materials and Methods

Chemicals. Vincristine sulfate (VCR), vinorelbine ditartrate (VRL), testosterone (TST), 6β-hydroxytestosterone, N-desmethyl diazepam, ketoconazole (KTZ), and NADPH were purchased from Sigma-Aldrich (St. Louis, MO). Vincristine sulfate [³H(G)] was purchased from American Radiolabeled Chemicals (St. Louis, MO). LSN335984, the dichlorinated derivative of the difluorinated compound LY335979 (Dantzig et al., 1996), was provided by Eli Lilly & Co. (Indianapolis, IN). The hepatocyte-thawing media were purchased from Celsis/In Vitro Technologies (Baltimore, MD). The maintenance media were purchased from Lonza Walkersville, Inc. (Walkersville, MD). All other reagents were of high-performance liquid chromatography (HPLC) grade and were purchased from Thermo Fisher Scientific (Waltham, MA).

CYP3A5 Genotyping. DNA samples isolated from the Indiana University human liver bank samples and cryopreserved hepatocytes were genotyped for the CYP3A5^*3 and CYP3A5^*6 allelic variants using allele-specific polymerase chain reaction methods and primers described previously (Hiratsuka et al., 2002) with SYBR Green detection (Le Corre et al., 2004). A TaqMan allelic discrimination assay (Applied Biosystems, Foster City, CA) was used to determine the CYP3A5^*7 allelic variant (Eap et al., 2004).

Cell Preparation. Cryopreserved human hepatocytes (lots AIT, CHD, EHI, FKM, MRS, REL, RML, SCA, ZIJ, and ZYZ; Celsis/In Vitro Technologies) (lot Hu418; CellzDirect, Pittsboro, NC) (lot 652; XenoTech, LLC, Lenexa, KS) were rapidly thawed in a water bath at 37°C for 75 to 90 s. The vials were emptied into prewarmed hepatocyte-thawing media and suspended by gentle inversion. To isolate the cells, the suspension was centrifuged at 50g for 5 min at room temperature. After the supernatant was removed, the cells were suspended in 2 ml of hepatocyte maintenance media (37°C, perfused with oxygen/CO₂, 95:5). The viable cell concentration was quantified using an automated viability analyzer (Vi-Cell; Beckman-Coulter, Fullerton, CA). The cell suspension was diluted with hepatocyte maintenance media to target a concentration 2-fold higher than the final incubation concentration (i.e., 3.0 × 10⁶ viable cells/ml for a final concentration of 1.5 × 10⁶ viable cells/ml).

Metabolite Identification. VCR stock solution in water was added to hepatocyte maintenance media (8–40 μM VCR) and aliquoted to 96-well (50 μl) or 24-well plates (250 μl each). The plate with VCR was warmed at 37°C in a humidified CO₂ incubator. The cells were prepared in hepatocyte maintenance media as described in the previous section to yield a final incubation concentration of 0.5 × 10⁶ (lots Hu418 and 652) or 1.5 × 10⁶ (lots MRS and EHI). The plates were incubated at 37°C in a humidified CO₂ incubator with orbital shaking (100 rpm) for up to 4 h. The contents were transferred to wells with equal volumes of acetonitrile with VRL internal standard and mixed by pipetting. After the plate was centrifuged, the supernatants were removed and frozen at -80°C until analysis by HPLC as described previously (Dennison et al., 2006). In brief, the mobile phases consisted of 0.2% formic acid (mobile phase A) and methanol (mobile phase B). The analytes were separated on a C18 column (Inertsil ODS-3, 2.1 × 150 mm, 5-μm particle size; MetaChem Technologies, Inc., Torrence, CA) at a flow rate of 0.2 ml/min using the following gradient conditions: 0 to 7 min 20% B; 7 to 42-min linear increase to 56% B; and 42 to 52 min 80% B. For liquid chromatography/mass spectrometry (LC/MS) analysis, full-scan mass spectra were obtained between 150 to 1050 kDa for representative lots EHI, MRS, Hu418, and 652. To confirm the identity of VCR-related compounds, positive ion tandem mass spectrometry was conducted for the ions of interest using argon as the collision gas at 1.5 mTorr and the collision energy of -40 V. The formation of known metabolites was also assessed by UV detection at 254 nm.

To ensure that all VCR-related compounds were identified, [³H]VCR was incubated with cryopreserved hepatocytes (lot EHI). The HPLC eluent fractions of purified [³H]VCR (10 μCi) were divided between microcentrifuge tubes and evaporated to dryness at room temperature. [³H]VCR (8 μM, approximately 1 × 10⁷ dpm/tube, 50 μl final volume) was dissolved in media, transferred to individual wells in a 96-well plate, and incubated with intact cells or intact cells with ketoconazole (10 μM) for 4 h. The [³H]VCR at the same concentration was also incubated with a cell lysate preparation (described under Vincristine and Testosterone Incubations with Intact Cells and Cell Lysates) in microsomal incubation buffer with and without NADPH for 30 min. For the intact cell incubation without ketoconazole, the media and cells were separated by centrifugation, and radiochromatograms were generated from both fractions. All incubations were quenched with an equal volume of acetonitrile and stored at -80°C.

To generate radiochromatograms, the samples were thawed and centrifuged at 1800g for 10 min. The supernatant was diluted with an equal volume of 0.2% formic acid and directly injected on a C18 column (Inertsil ODS-3, 3 × 150 mm, 5-μm particle size; MetaChem Technologies, Inc.). Separation of analytes was achieved at an eluent flow rate of 0.4 ml/min using gradient conditions described earlier for LC/MS full-scan analysis. The samples were fractionated into 96-well ScintiPlate-96 plates (PerkinElmer Life and Analytical Sciences, Waltham, MA) at 15 s/well using a Gilson 215 Liquid Handler (Gilson, Inc., Middleton, WI). The plates were dried using a GeneVac Personal Evaporator (Genevac Inc., Valley Cottage, NY) and counted using a Perkin Elmer 1450 LCS and Luminescence Counter (Micro-Beta TriLux; PerkinElmer Life and Analytical Sciences, Waltham, MA).

Vincristine and Testosterone Incubations with Intact Cells and Cell Lysates. For intact cell incubations with VCR, 50 μl of the cell suspension (150,000 cells/well) was added to an individual well of a 96-well flat-bottom plate with 50 μl of VCR (4 μM final concentration) in hepatocyte maintenance media. VCR was incubated with and without chemical inhibitors [ketoconazole, 10 μM; cyclosporine A (CsA), 20 μM; and LSN335984, 50 μM]. The plates were incubated in a humidified CO₂ incubator with shaking for 0, 1, 2, and 4 h at 37°C. All wells were quenched with the addition of 100 μl of 20 μM VRL (internal standard) in acetonitrile. In some cases, just before the quench, the cells were separated from the media by centrifugation at 50g for 5 min. In this case, an aliquot of the media after centrifugation was added to an equal volume of quench solution. After the media was removed, the cell pellet was dissolved in 100 μl of a 1:1 mixture of quench solution/fresh media. Two additional negative control experiments were performed at each time point using the same 96-well flat-bottom plates; VCR was incubated in media without cells, and the cell suspension was sonicated before mixing with VCR. All solutions were stored at -80°C until analysis by LC/MS (below).

Additional experiments were performed with cryopreserved hepatocyte lysates supplemented with NADPH. After the hepatocytes were counted and diluted to the target concentration as described in the previous section, the cells were centrifuged at 50g for 5 min. The medium was removed and replaced with microsomal incubation buffer (100 mM Na₂HPO₄ with 5 mM MgCl₂, pH 7.4). The cells in the hepatocyte suspension were lysed by 5 × 1-s pulses using a Branson Sonifier 450 probe sonicator (Branson Ultrasonics Corporation, Danbury, CT). The resulting “cell lysate” (50 μl) was added to 50 μl of a VCR (4 μM final concentration) and NADPH (0.5 mM final concentration) solution in incubation buffer. The lysate preparation with VCR was incubated for 0, 10, 20, and 40 min at 37°C in a shaking water bath and then quenched with 100 μl of 20 μM VRL (internal standard) in acetonitrile. The samples were stored at -80°C before analysis by LC/MS.

To quantify the CYP3A4 activity, testosterone (final concentration 200 μM, 0.75% methanol) was added in methanol to microsomal incubation buffer (100 mM Na₂HPO₄ with 5 mM MgCl₂, pH 7.4). An appropriate volume of cell lysate was added to the testosterone solution (200 μl) with NADPH (1 mM) for a presonicated cell density of 0.3 × 10⁶ cells/ml. The suspension was incubated in a shaking water bath for 10 min, quenched with 200 μl of ethyl acetate, and stored at -80°C before extraction and analysis by HPLC as described previously with slight modifications (Dennison et al., 2007). For an incubation volume of 200 μl, the chromatographic separation of the metabolites was achieved on a C18 column (Inertsil ODS-3, 2.0 × 150 mm, 5-μm particle size; MetaChem Technologies, Inc.) at an eluent flow rate of 0.2 ml/min. For lots AIT, EHI, RML, ZIJ, and ZYZ, the testosterone hydroxylase activities of the lysates were estimated in triplicate or duplicate with S.D.s or percentage differences of less than 10%. For the remaining lots, one-point estimates were reported. No estimates of error were provided by the vendor for testosterone hydroxylase activities of the intact cells.

Michaelis-Menten Kinetics of M1 Formation. VCR at approximately 100, 50, 25, 12.5, 6.3, and 3.1 μM was incubated with intact cells or cell lysates as described earlier (lot EHI, 1.5 × 10⁶ cells/ml). The incubations were quenched with 20 μM VRL in acetonitrile after 90 min (intact cells) or 13 min (cell lysate). The no-cell controls or NADPH-negative incubations were also performed at each concentration of VCR and quenched after 90 or 13 min, respectively. The samples were stored at -80°C before HPLC-UV analysis as described earlier.

Quantification of M1 and Vincristine by LC/MS. For the experiments described under Vincristine and Testosterone Incubations with Intact Cells and Cell Lysates, the quenched hepatocyte incubations containing VCR were thawed, vortexed, and centrifuged at 1800g for 10 min. After the supernatant was diluted with an equal volume of 0.2% formic acid, the M1 and VCR concentrations were quantified by LC/MS analysis (ThermoQuest; Thermo Fisher Scientific) in positive electrospray ionization mode using selected ion monitoring of VCR (m/z 413), M1 (m/z 397), and VRL (m/z 390). Xcalibur software (version 1.0; Thermo Fisher Scientific) was used for data acquisition and processing. The VCR/M1 standard solution was prepared and used to generate appropriate standard curves as described previously (Dennison et al., 2006). Separation of analytes was achieved using a C18 column (Inertsil ODS-3, 2.1 × 150 mm, 5-μm particle size; MetaChem Technologies Inc.) at a flow rate of 0.2 ml/min. The mobile phase compositions were 0.2% formic acid in water/methanol (80:20, v/v) (mobile phase A) and 0.2% formic acid in water/methanol (20:80, v/v) (mobile phase B). Starting at 0% B, the gradient conditions were as follows: 0 to 0.5-min linear increase to 10% B; 0.5 to 6.5-min linear increase to 18% B; 6.5 to 10.5-min linear increase to 100% B; 10.5 to 14.5 min 100% B; and 14.5 to 20 min 0% B.

Data Analysis. The apparent Michaelis-Menten constants, K_m and V_max, were estimated by fitting the data to a one-enzyme model using nonlinear least-square regression (GraphPad Prism version 5.00 for Windows; GraphPad Software Inc., San Diego, CA). With the same software, the rates of vincristine depletion and M1 formation, k, were estimated by nonlinear regression using at least four time points. For vincristine depletion, an exponential one-phase decay equation was used with a plateau constraint of zero (total depletion at infinite time). For M1 formation, M1 was assumed to be the primary metabolite of vincristine metabolism. Thus, an exponential one-phase association equation was used with the initial vincristine concentration as a plateau constraint. The S.D.s of the fitted parameters (also termed S.E.s of the estimates) are reported. For lots where M1 was degraded and/or metabolized at late time points (FKM and SCA), the rates of M1 formation were estimated using the earliest time points (10 min) before significant M1 loss. For lots FKM and SCA, two rate constants were also estimated: k₁ (the first-order constant for VCR depletion and M1 formation) and k₂ (first-order constant for M1 depletion) (WinNonlin 4.0; Pharsight, Mountain View, CA). The reported p values were determined using the Student's t test. The p values of less than 0.05 were considered statistically significant.

Results

Classification of Hepatocytes by CYP3A5 Genotype. By CYP3A5 genotyping, seven hepatocyte preparations were predicted to be CYP3A5 high expressers, and two were predicted to be CYP3A5 low expressers (Table 1). The CYP3A5 expression could not be predicted for three lots (CHD, MRS, and REL) by CYP3A5 genotype alone because the donors had multiple CYP3A5 single nucleotide polymorphisms, which may have prevented CYP3A5 expression from one or both chromosomes. These livers were later categorized as high and low expressers by phenotype using VCR activity (Table 1). Consistent with previous reports (Xie et al., 2004), 78% of the African-American livers (7 of 9) were ultimately classified by genotype and phenotype (when necessary) as CYP3A5 high expressers.

Metabolite Identification, Intact Cells. By LC/MS analysis, incubation with VCR and 14 lots of cryopreserved hepatocytes produced two phenotypically distinct groups. For the first group designated as CYP3A5 low expressers (n = 4) (Table 1), VCR metabolites were not detected. For these donors, for example lot MRS, the VCR concentrations for the intact cells and the control incubations (VCR without cells, VCR + ketoconazole with intact cells, and VCR + sonicated cells in media) did not detectibly change with incubation time (Fig. 2a). Low levels of M1 were detected in the cell incubations and the controls as a contaminant at time 0. However, the M1 concentrations of the intact cell incubations did not increase with incubation time; in fact, the concentrations of M1 decreased with time for the intact cell and control incubations (Fig. 2b). The concentrations also did not change with time for other known VCR degradation products or impurities including the VCR-epoxides (m/z 823) and VCR-N-oxide (m/z 841) (data not shown). By LC/MS analysis, no previously unknown VCR-related metabolites were detected. Similar results were obtained for lots 652, CHD, and Hu418 (Table 1).

For the second group of hepatocytes designated as CYP3A5 high expressers (n = 10, Table 1), a time-dependent increase in M1 concentration was observed with a corresponding reduction in VCR concentration (lot EHI) (Fig. 1, c and d). M1 was the dominant metabolite with no evidence of other potential metabolites by UV detection (Fig. 3a) or by LC/MS in scan mode. Using [³H]VCR and lot EHI, four major areas of activity were evident from the radiochromatogram (Fig. 3c): epoxide 1 region (14–15 min), M1/epoxide 2 region (15–17 min), the VCR region (17–19 min), and the VCR degradation and impurities region (19–20 min). The distribution of the areas did not differ between the cell pellet and the media (data not shown); only 4% of the radioactivity was recovered in the cell pellet. The M1/epoxide region area was reduced with coincubation of KTZ, a CYP3A inhibitor (9.9 versus 3.2% of total count) (Fig. 3c). The VCR degradation and impurities region seemed to be higher than the nonradiolabeled parent incubation (Fig. 3a), and they also reduced with coincubation of KTZ (Fig. 3c). However, without baseline separation from VCR, the quantification of the VCR degradation and impurities region was not possible by radiochromatography. To evaluate the compounds in this region, the radiolabeled incubation products were also analyzed by LC/MS. As predicted using the LC/MS data from the nonradiolabeled incubations, the concentrations of the epoxides and the VCR-related compounds that eluted later than VCR were not different between the intact cells and the KTZ control; only the M1 concentration was decreased in the KTZ control (data not shown).

Metabolite Identification, Cell Lysates. Cell lysates were prepared from cryopreserved hepatocytes and incubated with VCR to evaluate the role of the drug translocation and non-NADPH-dependent processes on VCR metabolism. For all lots of hepatocytes in which the cell lysates were prepared, regardless of CYP3A5 genotype, VCR was metabolized by the cell lysates to one major metabolite, M1. For lot MRS, a lot unable to detectably metabolize VCR as intact cells, the cell lysate preparation efficiently metabolized VCR to M1, albeit at a lower rate than that of CYP3A5 high expressers (Table 1). For lot EHI with high CYP3A5 expression, VCR was almost exclusively metabolized to M1 as determined by HPLC and LC/MS for the intact cell and cell lysate preparations (Fig. 3b). Compared with the NADPH-negative control, the cell lysate incubation produced one major VCR metabolite, M1, as determined by radiochromatography (Fig. 3d). The M1 formation by retention time (15–16 min) accounted for approximately 50% of the vincristine loss due to metabolism. An additional smaller region of activity (21 min, less than 40% of the M1 area) was also observed on the radiochromatogram (Fig. 3d). At a retention time of 21 min, a vincristine-related product was detected by LC/tandem mass spectrometry (m/z = 753.3) with two key secondary ions (m/z = 471 and 614). The first mass of 471 was the intact N-formyl vindoline moiety of VCR (Dennison et al., 2007). Based on the second mass (M–139), the compound was putatively identified as a degradation product or metabolite of M1 with a loss of a methyl ester group on the catharanthine moiety.

Testosterone 6β-Hydroxylase Activity. For lots purchased from Celsis/In Vitro Technologies, the vendor reported testosterone 6β-hydroxylase (CYP3A) activity of the intact cells (Table 1). To provide additional information on the CYP3A activity after thawing the cells, we quantified the testosterone 6β-hydroxylase activity for the cell lysates supplemented with NADPH. The activities of the cell lysates were highly variable between hepatocyte lots with a 100-fold range in values and not correlated to the vendor-reported intact cell activities (r² < 0.01) (Table 1).

Vincristine Depletion and M1 Formation for CYP3A5 High Expressers. To better understand the unexpectedly low rate of VCR metabolism compared with previous human liver microsomal kinetic data (Dennison et al., 2007), the rates of VCR depletion and M1 formation were quantified from the same donor using both intact hepatocytes and cell lysates (Table 1). To estimate the rates of metabolism, M1 and VCR concentrations were quantified by LC/MS at multiple time points. The rate constants for VCR depletion (k) were estimated by nonlinear regression, assuming a first-order reaction (for example, lot EHI) (Fig. 3, a and b). Likewise, the rate constants for M1 formation were also estimated assuming that M1 was the terminal product and the only metabolite of VCR. For these two methods, the rate constant estimated from VCR depletion was typically higher than the rate constant estimated from M1 formation (Table 1). However, in all cases, the M1 formation accounted for at least 42% of the VCR depletion (range = 42–131%) as shown by the differences in intrinsic clearances calculated using the rates of M1 formation and VCR depletion (Table 1).

When the M1 formation rate constants were estimated for the lysates, the assumption of first-order kinetics was not valid for two lots of hepatocytes (FKM and SCA). Although the mechanism is currently unknown, for these two lots, M1 was rapidly formed from VCR and then degraded and/or was metabolized at later time points. With hepatocyte lot FKM, the initial VCR concentration was reduced 80% by an apparent first-order reaction over 40 min (Fig. 4c). The M1 concentration correspondingly increased early in the reaction but later stabilized, even as the VCR concentration continued to decline (Fig. 4c). As a result, for lots FKM and SCA, the rate constants for M1 formation and subsequent intrinsic clearances (Table 1) were estimated using the 10-min time point only. For lots FKM and SCA, the concentrations of VCR and M1 were effectively modeled by nonlinear regression using a kinetic model with two rate constants: k₁ and k₂ (Fig. 4c).

Prediction of Intrinsic Clearance for CYP3A5 High Expressers. Using the estimated rate constants for CYP3A5 high expressers, the intrinsic clearances of VCR were predicted with the intact cells and the cell lysates (Table 1). The concentration of VCR used to calculate the rate constants was 4 μM, a value well below 30 μM, the K_m for intact cells (Fig. 5b). For most lots, the estimated intrinsic clearances were dramatically different for the two preparations; the cell lysates predicted values at least 10-fold higher than those predicted with intact cells. In addition, the values from the cell lysate were more variable than those from intact cells. Compared with the intrinsic clearances predicted with human liver microsomes (Dennison et al., 2007), the cell lysate values were similar for CYP3A5 high expressers. For the cell lysate preparations, the activities for M1 formation and VCR depletion were correlated to the rates of TST 6β-hydroxylation (r² = 0.67 and 0.68, respectively; p = 0.01 for both). However, the cell lysate activities for TST 6β-hydroxylation did not predict the rates of M1 formation with intact cells (r² = 0.03). Although the values were similar to each other (Table 1), the highest activity hepatocytes for M1 formation with intact cells were lots with the CYP3A5^*1/^*1 genotype compared with the CYP3A5^*1/^*0 livers (mean = 106 for CYP3A5^*1/^*1 and 61 ml/min for CYP3A5^*1/^*0; p = 0.02).

Effects of Inhibitors on M1 Formation. To determine whether M1 formation was mediated by CYP3A enzymes, ketoconazole was used as a selective CYP3A inhibitor (Fig. 2, c and d). For each lot of hepatocytes, the cells were incubated with VCR for 4 h with and without ketoconazole. For CYP3A5 high expressers, the M1 concentrations of the incubations with ketoconazole were 8 to 38% of the positive control incubations without inhibitor (Table 2). The M1 levels in the incubations with ketoconazole were reduced to essentially background levels and did not increase with time of incubation (Fig. 2d). For CYP3A5 low expressers, the low levels of M1 in the incubations were similar to the controls and were not substantially affected by ketoconazole addition (Fig. 2c).

Two other chemical inhibitors were also coincubated with VCR: CsA (a CYP3A4 inhibitor) and LSN335984 [a selective P-glycoprotein (P-gp) inhibitor]. CsA at 25 μM was chosen because the drug at this concentration was previously used to selectively inhibit the formation of M1 by CYP3A4 in human liver microsomes (Dennison et al., 2007). For all hepatocyte lots incubated with CsA, the rates of M1 formation were unaffected (Table 2). The P-gp inhibitor LSN335984 was initially chosen to reduce the rate of VCR efflux from the hepatocytes and thus potentially increase the intracellular, unbound concentration of VCR. In screening experiments with hepatocyte lot EHI, 5 μM LSN335984 was metabolized faster than VCR and did not affect the rate of M1 formation (data not shown). To ensure that significant levels of LSN335984 remained during the incubations, the concentration of LSN335984 was increased to 50 μM, although the nonselective inhibition on P450 enzymes at this concentration was unknown. With 50 μM LSN335984, the rates of M1 formation with hepatocytes were partially inhibited (Table 2).

Michaelis-Menten Kinetics of M1 Formation. To better understand the differences in rates of M1 formation between intact cells and cell lysates, the apparent Michaelis-Menten kinetic parameters were estimated for lot EHI using both cell preparations. The K_m of the cell lysate was 3-fold lower than the K_m of the intact cells; the V_max of the cell lysate was 3-fold higher than that of intact cells (Fig. 5). Altogether, the intrinsic clearance of the cell lysate (V_max/K_m) was 9-fold higher than that of the intact cells.

Prediction of Intrinsic Clearance for CYP3A5 Low Expressers. VCR depletion and M1 formation were not observed for specific lots of intact hepatocytes: 652, CHD, Hu418, and MRS (Table 1). For these lots, categorized as CYP3A5 low expressers, the M1 concentration, initially present in low concentrations as a contaminant, unexpectedly decreased with time (Fig. 2b). As a result, the intrinsic clearance of VCR for these lots was estimated as zero.

Discussion

Vincristine is primarily metabolized by CYP3A4 and CYP3A5 to one metabolite, M1, in human P450-reconstituted systems (Dennison et al., 2006, 2007). In the present study, human cryopreserved hepatocytes were used to determine whether CYP3A metabolism remains the major route of elimination in an integrated cellular system. We specifically investigated the role of CYP3A5, a genetically polymorphic enzyme, using multiple hepatocyte donors with apparent high and low CYP3A5 expression (n = 12). Commercially available lots of African-American hepatocytes (n = 9) were used to increase the probability of obtaining CYP3A5 high expressers. The hepatocyte preparations were classified as high (at least one CYP3A5^*1 allele) and low CYP3A5 expressers by the CYP3A5 genotype when possible. However, for three African-American donors with multiple allelic variants, CYP3A5 expression could not be predicted by genotype alone (Table 1). Although additional cells from these lots were unavailable for quantification of CYP3A5 protein or RNA, two of the three preparations with multiple allelic variants did not detectably metabolize vincristine and thus were classified as CYP3A5 low expressers by phenotype (Table 1). For all hepatocyte preparations, the CYP3A5 genotypes were consistent with the CYP3A5 expression classification (Table 1).

Excluding the three hepatocyte preparations with ambiguous CYP3A5 genotype, the metabolism of vincristine was only detectable in intact cells for preparations with apparent high CYP3A5 expression (n = 7) (Table 1). The metabolic profile for the CYP3A5 high expressers (Fig. 3a) was characteristic of the P450-mediated metabolism previously described using cDNA-expressed enzymes and human liver microsomes (Dennison et al., 2006, 2007). For these enzyme systems, M1 was detected as the major vincristine-related metabolite. No other secondary metabolites or phase II metabolites were detected by LC/MS. Based on this metabolic profile, P450 enzymes were exclusively responsible for the metabolism of vincristine in hepatocytes.

As noted previously for human liver microsomes, the metabolism of vincristine by hepatocytes was probably mediated by CYP3A enzymes, specifically CYP3A5 (Dennison et al., 2007). With CYP3A5 high expresser hepatocytes, the formation of M1 was almost completely inhibited by ketoconazole, a selective CYP3A4 and CYP3A5 inhibitor (Fig. 2d; Table 2). The role of CYP3A4 in the metabolism of vincristine was substantially less than that predicted using reconstituted human enzymes. We previously reported that vincristine was selectively metabolized by CYP3A5 compared with CYP3A4 (7-fold higher specific activity). However, in contrast with human liver microsomes, M1 formation by hepatocytes was not inhibited by cyclosporin A, a selective CYP3A4 inhibitor (Table 2). Although the intracellular concentration of this inhibitor is unknown, the lack of chemical inhibition by cyclosporin A is consistent with a limited role in the formation of M1 by CYP3A4. With intact hepatocytes, those classified as CYP3A5 low expressers were unable to detectably metabolize vincristine (Table 1). In fact, the concentration of M1 declined with time in an identical pattern to the controls (Fig. 2b).

Based on our previous work with human reconstituted enzymes, the lack of metabolism by CYP3A5 low-expressing hepatocytes was an unexpected result. For these hepatocytes, control experiments were performed to ensure that the activity of CYP3A4 and the viability of the cells were normal. The CYP3A5 low-expressing hepatocytes used during the metabolite-profiling experiments effectively metabolized a probe CYP3A substrate, verapamil (lots 652 and Hu418; data not shown). In addition, the CYP3A4 activities of the CYP3A5 low expressers were typical of the CYP3A5 high expressers as determined by the testosterone 6β-hydroxylase activities of the intact cells from the vendor (lots CHD and MRS) and of the cell lysate (lot MRS) (Table 1). Intraday variability does not seem to be a factor; MRS was incubated with vincristine on two separate occasions, once at the same time as lot SCA, a CYP3A5 high expresser (Table 1). In direct contrast to the intact cell results, the cell lysate of MRS also effectively metabolized VCR to M1 (Table 1). Using the cell lysate rate of M1 formation for MRS, the calculated intrinsic clearance of 293 ml/min is just higher than the median intrinsic clearance for a CYP3A5 low expresser as reported previously using human liver microsomes (Dennison et al., 2007). Thus, for the intact cells, CYP3A4 seems to be active in CYP3A5 low expressers but unable to detectably metabolize vincristine.

Even though intact cells from CYP3A5 high expressers were able to metabolize vincristine, the rates of metabolism were much slower than the clearances predicted using human reconstituted enzymes and the adult in vivo clearances. Based on the rates of M1 formation for the most active hepatocyte lot SCA, the predicted hepatic extraction ratio was only 0.04 compared with in vivo extraction ratios of 0.03 to 0.2 in Caucasian adults (Sethi et al., 1981; Nelson, 1982). Other investigators have reported a 5-fold systematic underprediction of in vivo intrinsic clearance, using cryopreserved human hepatocytes (Brown et al., 2007). Although underprediction of in vivo clearances is typical using human hepatocytes, the hepatocyte kinetic data were expected to more closely match the previously reported human liver microsome data (Dennison et al., 2007). Compared with the human liver microsome kinetic data, the human hepatocyte data in this study predicted at least 10-fold lower in vivo clearances for vincristine.

The published studies that have compared the predicted clearances of hepatocytes and microsomes are conflicting and confounded by the variability of the preparations. For example, one study reported that the midazolam intrinsic clearance estimated with human liver microsomes was 22-fold higher compared with the cryopreserved hepatocyte estimated value (Lu et al., 2006). However, in common with all such studies, the lots of cryopreserved hepatocytes and the microsomes were not from the same liver, and the numbers of different lots tested were small (n = 3–7). In a different study, the cryopreserved hepatocytes predicted a 3-fold higher midazolam intrinsic clearance compared with human liver microsomes (Hallifax et al., 2005). Again, the variability between lots may in part explain the discrepancy between the reports. We designed our experiments to remove this variable by preparing cell lysates from the same lots of hepatocytes. Even so, for the various lots of cryopreserved hepatocytes, the intrinsic clearances of CYP3A5 high expressers were 4 to 69-fold lower than those predicted using paired cell lysate preparations (Table 2). This finding is consistent with an intracellular concentration of vincristine that is significantly lower than the extracellular concentration and a corresponding underestimate of intrinsic clearance. Thus, the predicted hepatic clearance with cryopreserved hepatocytes may be a function of metabolic intrinsic clearance and vincristine translocation across the plasma membrane.

The uptake of vincristine by hepatocytes is hypothesized to be primarily diffusion-rate limited, whereas the cellular efflux is probably mediated by active transporters at the canalicular membrane, specifically P-gp and multidrug resistance-associated protein 2 (Jackson Jr. et al., 1978; Zhou et al., 1994; Ho and Kim, 2005). To understand the potential role of active transport in our hepatocyte study, vincristine was coincubated with transporter inhibitors: cyclosporin A (P-gp and multidrug resistance-associated protein 2) and LSN335984 (P-gp). However, the rates of M1 formation were unaffected or slightly reduced by the inhibitors, perhaps as a result of a combination of efflux transporter and CYP3A inhibition (Table 2). To quantify the impact of putative extra- to intracellular concentration gradients of vincristine on the rates of M1 formation, the apparent Michaelis-Menten kinetic parameters of M1 formation were estimated for both intact cells and cell lysate preparations with NADPH (Fig. 5). For lot EHI, the K_m value for the intact cells was 3-fold higher than that of the cell lysate. In isolation, this observation suggests that membrane transport for this lot of hepatocytes reduced the intracellular concentration and subsequently the apparent intrinsic clearance. However, only 50% of the observed differences in intrinsic clearance were accounted for by the K_m; the V_max values were concommittally 3-fold higher for the cell lysate compared with the intact cell preparation. The catalytic properties of P450 enzymes in the lysed cells seem to be fundamentally different from those in the intact cells. Thus, the K_m values may have differed between the preparations without the influence of vincristine transport or diffusion. Additional studies will be required to understand the functional differences between enzymes in the lysed and intact hepatocytes.

In conclusion, vincristine was metabolized to a single oxidative metabolite, M1, by human cryopreserved hepatocytes, but only those with apparent high CYP3A5 expression. In contrast to the prediction based on human liver microsomes, CYP3A4 did not seem to catalyze M1 formation in intact hepatocytes. Genetic polymorphisms that determine CYP3A5 expression are expected to alter vincristine exposure and clinical outcomes and may be used to recommend dose adjustments for individualized therapy.