Abstract
The intrinsic metabolic clearance of saquinavir, nelfinavir, and ritonavir was determined over a range of concentrations (0.02–20 μM) in both rat liver microsomes and fresh isolated rat hepatocytes in suspension. Clearance values were found to be concentration dependent for both systems, and at low concentrations, microsomal clearance was much greater (7–14-fold) than in hepatocytes. Kinetic parameters showed substantially lower microsomal Km values (5–42 nM) compared with suspended rat hepatocytes (34–270 nM) but similar scaled Vmax values (2–26 nmol/min/g liver). In the absence of metabolism (achieved by pretreating hepatocytes with a mechanism-based inhibitor of cytochrome P450), saquinavir, nelfinavir, and ritonavir were actively and rapidly taken up into hepatocytes (cell/medium concentration ratios of 306-3352), and intracellular unbound drug concentrations between 5- and 12-fold higher than extracellular unbound concentrations were achieved. Comparison of the rate of uptake into hepatocytes with the rate of metabolism in hepatocytes and microsomes indicates that the former is the rate-limiting step at low concentrations. The rate of metabolism saturates at lower concentrations (100–400-fold) than the rate of uptake; hence, at the high concentrations metabolic rate-limited clearance occurs. In conclusion, the clearance of saquinavir, nelfinavir, and ritonavir is extremely rapid, and it is proposed that in the case of hepatocytes and by inference in vivo, the rate of uptake limits the metabolic clearance of these three drugs.
In vitro systems are well established as valuable tools for studying various aspects of drug metabolism; in particular, kinetic data obtained from in vitro systems can be scaled and used in the prediction of in vivo clearance (Houston and Carlile, 1997; Obach, 2001). Hepatic microsomes are at present the dominant in vitro system but have obvious disadvantages, such as the need for exogenously supplied cofactors and the lack of nonmicrosomal enzymes, in particular the phase II enzymes, and membrane transporter systems. Because more drugs are being developed with noncytochrome P450-dependent clearance, studies in hepatocytes are increasingly more important. Hepatocytes have traditionally been perceived as a complicated and time-consuming metabolizing system due to, in part, the need for fresh tissue for the cell isolation procedure. With recent advances in cryopreservation technologies, the use of freshly isolated cells is no longer a necessity, and hepatocytes can be stored at -80°C after isolation for a period of months, retain activity, and provide useful human clearance prediction (Lau et al., 2002; McGinnity et al., 2004; Brown et al., 2007).
An obvious major advantage of hepatocytes is their intact structural integrity, which may result in intracellular drug concentrations that differ from those in the surrounding medium and are more representative of the in vivo situation. Several processes may contribute to this situation including intracellular binding (Hallifax and Houston, 2007) and membrane transporter activity (Lau et al., 2006). The study of transporter proteins expressed in hepatocytes has resulted in a substantial amount of literature including many comprehensive reviews (Mizuno et al., 2003; Shitara et al., 2006). It is becoming increasingly apparent that the interplay between transporters and metabolism may have an impact on the absorption and clearance of drugs (Benet et al., 2004; Zamek-Gliszczynski et al., 2006). For instance, an increase in the plasma levels of cerivastatin has been observed when coadministered with the immunosuppressant drug cyclosporin A (Mück et al., 1999). Shitara et al. (2003) confirmed that cerivastatin is actively taken up into human hepatocytes and that cyclosporin A inhibited this uptake (Ki values of 0.3–0.7 μM). Inhibition of cerivastatin metabolism by cyclosporin A gave only a small reduction in turnover, hence suggesting that the interaction occurs via the inhibition of transporter-mediated uptake and not via metabolic inhibition. A similar scenario may occur with repaglinide and gemfibrozil (Hinton et al., 2007). The inability to recognize these hepatic uptake and metabolic interplay phenomena may explain why for certain types of drugs in vivo clearance is not well predicted from simple hepatic microsomal studies (Lam and Benet, 2004). Human immunodeficiency virus protease inhibitors (PIs) are known to be substrates for hepatocyte uptake and efflux transporters (Kim et al., 1998; McRae et al., 2006) and are extensively and efficiently metabolized; thus, they provide a good example for study of the interplay between transporter and metabolism.
The aims of this study were first to delineate the kinetics and metabolism of saquinavir, nelfinavir, and ritonavir in both isolated hepatocytes and hepatic microsomes. The source of these in vitro tissues was the rat to allow direct comparison of parameter values without the complication of intraindividual donor variability, which occurs with human preparations (Hallifax et al., 2005; Rawden et al., 2005). The second aim was to characterize the uptake of these three PIs in freshly isolated hepatocytes. Comparison of these data provides direct evidence that the clearance of saquinavir, nelfinavir, and ritonavir in intact hepatocytes is extremely rapid in vitro and likely to be rate limited by their hepatic uptake in vivo.
Materials and Methods
Chemicals. Saquinavir, nelfinavir, [14C] saquinavir and [3H] nelfinavir were all gifts from Roche Products Ltd (Welwyn, UK and Basel, Switzerland), and ritonavir was a gift from Abbott Laboratories (Abbott Park, IL). [3H]Ritonavir was purchased from Moravek Biochemicals (Brea, CA). All radiochemicals were >98% purity. Collagenase A and H were purchased from Roche Molecular Biochemicals, and silicone oil (AR20 and AR200) were purchased from Wacker Chemie GmbH (Munich, Germany. All other chemicals and reagents were purchased from Sigma Chemical (Poole, Dorset, UK), BDH (Poole, Dorset, UK), or Fisher Scientific Co. (Pittsburgh, PA) and were of the highest grade available. Uptake buffer contained sodium chloride, potassium chloride, potassium dihydrogen phosphate, magnesium chloride, sodium bicarbonate, d-glucose, HEPES, and calcium chloride (all from Sigma).
Preparation of Tissue. Hepatocytes were prepared using the collagenase perfusion method from Sprague-Dawley male rats (200–250 g) sacrificed by cervical dislocation according to the procedure described by Jones et al. (2005). Preparation of hepatocytes for uptake experiments was based on the same method using male Wistar rats anesthetized with phenobarbital prior to hepatocyte isolation. Only cells with viability greater than 85% were used. Hepatic microsomes were prepared using the standard differential centrifugation method described by Jones et al. (2005) and were prepared from male Sprague-Dawley rats.
Incubation Methods. Incubations were performed in Eppendorf tubes in a Thermomixer (Eppendorf AG, Hamburg, Germany) set at 37°C and 900 rpm and in a volume of 400 μl. The depletion of saquinavir, nelfinavir, and ritonavir was measured over a range of concentrations (0.1–10, 0.1–10, and 0.01–2 μM for saquinavir, nelfinavir, and ritonavir, respectively) with freshly isolated suspended rat hepatocytes (0.2 × 106 cells/ml) or liver microsomes (0.2 mg protein/ml). Liver microsomes were diluted in phosphate buffer, and a methanol solution of drug was added [final solvent concentration of 1% (v/v)]. After 5 min of preincubation at 37°C, 200 μl of prewarmed NADPH regenerating system was added to initiate reactions. Reactions were terminated by addition of 600 μl of ice-cold acetonitrile containing an internal standard at the appropriate time point. Rat hepatocyte incubations were carried out in Williams' medium E prewarmed to 37°C, and reactions were initiated by the addition of 200 μl of drug solution (in Williams' medium E and 2% methanol) to 200 μl of cell suspension [final solvent concentration of 1% (v/v)]. Termination of reactions was by snap freezing in liquid nitrogen at the appropriate time point, and 600 μl of acetonitrile containing internal standard was added on thawing of the samples. All incubations were performed in duplicate.
Drug uptake into isolated hepatocytes was determined using the centrifugal filtration technique through a silicone oil layer (Hallifax and Houston, 2007). Following isolation, hepatocytes were resuspended in uptake buffer (pH 7.4 and 37°C) containing 2 mM aminobenzotriazole (to prevent P450-mediated metabolism) and 2% BSA at 4 × 106 cells/ml. Hepatocytes were dispensed into Eppendorf tubes and preincubated at 4 or 37°C. The addition of saquinavir (0.1–200 μM), nelfinavir or ritonavir (1–200 μM) in uptake buffer containing a fixed amount of the appropriate radiolabeled substrate and 2% BSA (final methanol concentration of 1%) initiated the uptake process. Aliquots were removed from the incubation at 6, 16, 26, and 36 s and dispensed into microtubes containing 10 μl of 3 M potassium hydroxide (lower layer) and 150 μl of silicone oil (upper layer) and were immediately centrifuged for 30 s at high speed. After freezing in liquid nitrogen, the KOH layer was removed using clippers and dropped into scintillation vials, which were then shaken overnight prior to analysis. Uptake rates at 4 and 37°C were determined by linear regression and expressed as picomoles per second. The rate at 4°C (passive uptake) was subtracted from 37°C rate (passive plus active) to determine temperature-dependent (active) uptake rate. Resultant active rates were plotted against drug concentration and corrected for binding to BSA in uptake buffer, and kinetic parameters were estimated.
The uptake of saquinavir, nelfinavir, and ritonavir at 5 μM was also monitored over a longer time period to determine cellular drug concentration at equilibrium to determine the cell/medium concentration ratio (Kp). Time points were taken at 6, 16, 26, 36, 70, and 120 s, and a first order rate exponential input equation was applied to the 37 minus the 4°C data.
Measurement of Protein and Nonspecific Binding. Protein binding was determined using Dianorm dialysis equipment (Diachema AG, Zurich, Switzerland) and a dialysis membrane with a 5000-kDa molecular mass cut-off. Drug solutions (1 μM) were prepared in each of the three incubation media containing protein (0.2 mg protein/ml liver microsomes, 0.02% BSA, or 2% BSA), and 1 ml of dialyzed against protein-free media in a water bath at 37°C for 4 h. Samples were taken from each side of the chamber, and fraction unbound (fu) was calculated by dividing the drug concentration in the dialysate by the drug concentration in the sample chamber. Nonspecific binding was measured in hepatocyte and microsomal incubations with 1 μM saquinavir, nelfinavir, or ritonavir, where phosphate buffer replaced the NADPH-regenerating system in microsomal incubations, and 2 mM aminobenzotriazole was added to hepatocyte incubations to prevent metabolism. Aliquots were taken after 5 min and compared with known standards to calculate loss of compound due to nonspecific binding. Previous experience has shown that the use of BSA in hepatocyte incubations is valuable in both minimizing nonspecific binding and stabilizing the function of hepatocytes.
The extent of nonspecific binding of saquinavir, nelfinavir, and ritonavir to plastic and glass was also measured after the initial experiments indicated that back-calculated initial concentration was consistently lower than the nominal initial concentration. This was particularly noticeable in the hepatocyte metabolism incubations but was also detected in microsomal incubations. The degree of nonspecific binding to plastic and glass varied between the drugs studied, and the cumulative effect for these incubations is shown in Table 1. The greater effect seen in hepatocyte incubations is partly a consequence of the experimental design in which there is no protein present in the incubation until the addition of cells at zero time. The use of the observed initial drug concentration as opposed to the nominal initial concentration in all appropriate calculations corrected for this loss. This value was then corrected for fu in the incubation medium.
Sample Preparation and Analysis. Liver microsomes and hepatocytes samples were vortexed and centrifuged at 10,000g for 10 min. The supernatant was transferred to high-performance liquid chromatography vials for analysis via LC-tandem mass spectrometry.
Saquinavir together with verapamil (internal standard) were separated on a Luna C18 (2) 50- × 4.6-mm 3-μm column (Phenomenex, Torrance, CA) at 40°C using a binary gradient maintained at 1 ml/min by a Waters Alliance 2795 HT LC system (Waters, Milford, MA). An initial mobile phase of 90% 0.001 M ammonium acetate/10% acetonitrile was ramped linearly to 82% acetonitrile/18% 0.001 M ammonium acetate (from 1–4 min), after which the initial ratio was immediately re-established and equilibrated from 4 to 5 min. The retention times were approximately 3.4 (verapamil) and 3.5 (saquinavir) min. Nelfinavir and its metabolite (M1) (Zhang et al., 2001) and ritonavir and its metabolites (M1, M2, M9, and M11) (Denissen et al., 1997) together with verapamil (internal standard) were separated using the same system as for saquinavir except that the initial mobile phase of 90% 0.001 M ammonium acetate/10% acetonitrile was ramped linearly to 90% acetonitrile/10% 0.001 M ammonium acetate from 1 to 4 min. The retention times were approximately 3.2 (verapamil), 3.3 (nelfinavir-M1), 3.4 (nelfinavir), 3.9 (ritonavir-M11), 4.1 (ritonavir-M2 and ritonavir-M9), and 4.6 (ritonavir and ritonavir-M1) min.
Saquinavir, nelfinavir, and ritonavir and associated metabolites, and internal standards were detected and quantified by atmospheric pressure electrospray ionization tandem mass spectrometry using a Micromass Quattro Ultima triple quadruple mass spectrometer. The LC column eluate was split, and one quarter was delivered into the mass spectrometer, where the desolvation gas (nitrogen) flow rate was 600 l/h, the cone gas (nitrogen) flow rate was 100 l/h, and the source temperature was 125°C. Using positive ion mode, protonated molecular ions were formed using a capillary energy of 3.5 kV and cone energies of 38 (saquinavir), 50 (nelfinavir, nelfinavir-M1), and 60 (verapamil, ritonavir, ritonavir-M1, ritonavir-M2, ritonavir-M9, ritonavir-M11) V. Product ions formed in argon at a pressure of 2 × 10-3 millibar and at varying collision energies were monitored as ion chromatograms that were subsequently integrated and quantified by quadratic regression of standard curves using Micromass QuanLynx 3.5 software.
Data Processing. For drug depletion studies, the log of drug concentration was plotted against time, and the elimination rate constant was calculated by fitting a single exponential decay. Initial drug concentration was corrected for protein binding prior to calculation of clearance. Metabolism data were also expressed as rates, representing a measure of substrate turnover per unit of time per unit of enzyme (as either microsomal protein or cell number) for each substrate concentration. This was achieved by multiplying the initial unbound concentration by clearance. These rate data were analyzed by the Michaelis-Menten equation to determine by nonlinear regression the kinetic parameters Km and Vmax.CLint and Perm were subsequently calculated by dividing Vmax by Km for the metabolic and transport data, respectively.
For the calculation of Kp (total drug at equilibrium) at an initial concentration of 5 μM, two assumptions were made; first, that drug concentration in adherent water was equivalent to medium concentration (correction for drug in adherent water was found to make minimal difference to calculated Kp); and second, that the residual amount of drug (i.e., not detected in cells) represented drug in the medium. The unbound concentration of drug in the medium at equilibrium (Cmu) was calculated using eq. 1, where C0 is initial concentration of drug in media, and Dm is fraction of drug in media at equilibrium.
The concentration of drug in cells at equilibrium (Cc) was calculated using eq. 2, where Ae was the amount of drug in cells at equilibrium, Cm the concentration of total drug in media at equilibrium, Va was the volume of the adherent water layer, and Vc was the intracellular volume.
Kp (total drug at equilibrium) was expressed as the ratio of Cc to Cmu. Values for Va and Vc were taken from Hallifax and Houston (2007). Kpu (unbound drug at equilibrium) was calculated by dividing the initial uptake rate at 37°C (rate due to both transporters and passive permeability) by the initial uptake rate at 4°C (rate due to only passive permeability). The unbound intracellular fraction of drug was calculated by dividing Kpu by Kp.
Data modeling was carried out using GraFit (version 4), Erithacus Software (Horley, Surrey, UK), and data are shown as mean values ± S.D. Metabolism and uptake parameters from microsomal and hepatocyte incubations were calculated based on PI concentration corrected for both protein and nonspecific binding, scaled, and expressed as “per grams of liver” using the scaling factors 60 mg protein/g liver and 109 × 106 cells/g liver (Houston and Carlile, 1997).
Results
Nonspecific and Protein Binding Characteristics. The extent of binding at the drug concentrations used within the hepatocyte metabolism (0.02% BSA), microsomal (0.2 mg/ml), and hepatocyte uptake (2% BSA) incubation matrices generally increased with the protein content. However, the rank order in terms of fu for the three drugs differed among the three matrices: for example, ritonavir > saquinavir > nelfinavir in microsomes compared with saquinavir > ritonavir > nelfinavir in hepatocyte uptake buffer (Table 1). Substantial binding (90% or more) was evident within the hepatocyte uptake and microsomal incubates for nelfinavir. Saquinavir and ritonavir fu were similar in both matrices (fu were approximately 0.3 and 0.5, respectively), but nelfinavir demonstrated a much greater affinity for microsomal protein (fu = 0.1 in 0.2 mg protein/ml for liver microsomes compared with 0.5 in 0.02% BSA). All parameters refer to unbound drug concentrations because they are corrected for both protein and nonspecific (plastic and glass, see Materials and Methods) binding.
Drug Depletion Studies. The depletion of saquinavir, nelfinavir, and ritonavir was measured over a range of concentrations (0.1–10, 0.1–10, and 0.01–2 μM for saquinavir, nelfinavir, and ritonavir, respectively) in freshly isolated suspended rat hepatocytes (0.2 × 106 cells/ml) and microsomes (0.2 mg protein/ml). All depletion profiles appeared log-linear over the time course studied (up to a maximum of 20 min) and involved a minimum of 15% parent drug loss.
Expression of data in terms of the metabolic rate at each concentration (corrected for fu in incubation) for saquinavir is shown in Fig. 1, A and B. Saquinavir metabolism followed Michaelis-Menten kinetics in microsomes and hepatocytes and the kinetic parameters Vmax, Km, and Clint are listed in Tables 2 and 3 for these two in vitro systems, respectively. The data were also expressed as clearance plots, as shown for saquinavir in Fig. 1, C and D. Clearance in suspended hepatocytes approached concentration independence (representing Clint) at concentrations below 0.02 μM, illustrated by the plateau at low saquinavir concentration. Above 0.02 μM, clearance became saturated as illustrated by the progressive decline in saquinavir clearance with respect to increasing saquinavir concentration. In microsomes due to the apparently low Km, Clint could only be estimated by back extrapolation because very rapid turnover and analytical sensitivity prevented determination below initial unbound drug concentration of 0.03 μM.
Clearance plots for nelfinavir and ritonavir are shown in Figs. 2 and 3, respectively, for both the hepatocytes and microsomal incubations. As in the case of saquinavir, the turnover of these two PIs was substantially faster in microsomes than in hepatocytes. Michaelis-Menten kinetic behavior was observed in hepatocytes for both drugs, and the Km and Vmax values are listed in Table 2. In microsomes, again there is the issue of lack of experimental data at very low substrate concentrations, below 0.008 and 0.003 μM for nelfinavir and ritonavir, respectively. Also, there was some suggestion of substrate inhibition for nelfinavir, demonstrated by a slight decrease in metabolic rate at high nelfinavir concentrations (>0.5 μM). Thus, in addition to the Michaelis-Menten equation, a substrate inhibition equation was also applied to the nelfinavir microsomal data; however, only minor changes were evident (see Table 3). The kinetic parameters obtained in both in vitro systems (Tables 2 and 3) represent the sum of more than one pathway because saquinavir, ritonavir, and nelfinavir form multiple metabolites.
Metabolite Formation Studies. In addition to measuring the time course of depletion of parent drug, metabolite appearance was also investigated for ritonavir and nelfinavir. Four ritonavir metabolites were identified in hepatocyte incubations and were quantified in arbitrary units because metabolite standards were not available. The metabolites M1 (loss of the thiazolyl carbamate moiety), M2 (oxidation at the terminal isopropyl group), M9 (oxidation on the methylthiozolyl moiety), and M11 (loss of the isopropylthiazolylmethyl moiety) were detected (previously identified by Denissen et al., 1997). The Michaelis-Menten kinetic equation was applied to the formation of all four metabolites (Fig. 4A), and the Km values for the individual metabolites were 0.034 ± 0.015, 0.13 ± 0.055, 0.029 ± 0.014, and 0.024 ± 0.014 μM for M1, M2, M9, and M11, respectively. Assuming approximately equal mass spectrometry sensitivities, a rank order of Clint can be made by dividing the Vmax (in arbitrary units) by Km. The highest Clint was observed with M1, and the Clint of M2, M9, and M11 were all approximately half that of M1.
Four metabolites, M1, M2, M9, and M11, were also detected in microsomal incubations containing ritonavir, and again, Michaelis-Menten kinetics were observed in each case (Fig. 4B). The Km values for the individual metabolites were 0.028 ± 0.005, 0.091 ± 0.021, 0.015 ± 0.007, and 0.029 ± 0.019 μM for M1, M2, M9, and M11, respectively. The rank order of Clint for the four metabolites, calculated using the Vmax values (in arbitrary units) and assuming equal mass spectrometry sensitivity, was M1 > M2 = M9 > M11.
Two metabolites of nelfinavir were detected after incubating with rat hepatocytes, previously identified as M1 by Zhang et al. (2001). Only M1 was quantifiable, and its rate of formation was consistent with Michaelis-Menten kinetics, with a Km of 2.3 ± 1.2 μM.
M1 and M8 were also identified in microsomes, but only M1 was quantifiable. An increase in initial rate of formation with increasing nelfinavir concentration was observed up to 5 μM, but the rate at 10 μM was slower than that at 5 μM, indicating possible substrate inhibition. Both substrate inhibition and Michaelis-Menten kinetic fits were applied, and Km values for the formation of M1 were 0.027 (Michaelis-Menten fit) and 0.08 (substrate inhibition fit) μM.
There was excellent agreement between the Km values for ritonavir M1 formation and parent drug depletion in hepatocytes. However, Km values were consistently higher than parent drug depletion for the specific metabolites of ritonavir in microsomes and nelfinavir in both systems. This observation highlights the hybrid nature of the kinetic parameters obtained by drug depletion. However, the much faster turnover of both ritonavir and nelfinavir, evident in the drug depletion studies, was confirmed from the metabolite formation studies.
Hepatocyte Uptake Studies. Initial uptake rates of saquinavir, nelfinavir, and ritonavir into rat hepatocytes were assessed between 0.1 and 200 μM at both 37 and 4°C (Fig. 5, A–C). Significant temperature-dependent uptake of saquinavir, nelfinavir, and ritonavir was observed, indicative of cell uptake via an active process. For saquinavir, saturation of uptake was not observed at unbound drug concentrations below 28 μM; therefore, kinetic parameters were limited to the calculation of the linear permeability term. For nelfinavir and ritonavir, saturation of uptake was observed at high concentrations (see Fig. 5, B and C), and the Michaelis-Menten equation was applied. The resultant kinetic parameters and Perm (estimated by dividing Vmax by Km) values for saquinavir, nelfinavir, and ritonavir are shown in Table 2. It can be seen that all transporter parameters were substantially higher than the corresponding metabolic parameters.
Figure 5D shows uptake time profiles for active uptake into hepatocytes for saquinavir, nelfinavir, and ritonavir, obtained by subtracting the 4°C rate from the rate at 37°C. The rate constants for the first order uptake obtained from the exponential time profile were 0.022, 0.047, and 0.0504 s-1 for saquinavir, nelfinavir, and ritonavir, respectively, and the rank order of these measures of permeability were consistent with those obtained in the initial rate experiments (Table 2). Comparison of the plateau values of these curves indicates that nelfinavir showed the highest extent of intracellular accumulation, and saquinavir and ritonavir accumulation were substantially lower. From the cell and medium drug concentrations at plateau, Kp values for saquinavir, nelfinavir, and ritonavir were calculated and showed a 10-fold range (Table 4).
To resolve the Kp values into transporter and intracellular binding components, Kpu values were calculated to obtain a pure measure of the former component. Similar Kpu values were found for the three drugs (within a 2-fold range) and constituted a minor component of the overall Kp (<2%). The rank order of intracellular binding (as reflected in the unbound values in Table 4) was saquinavir = ritonavir >> nelfinavir. The intracellular unbound fractions for ritonavir and saquinavir are similar because both the Kp and the Kpu show a 2-fold difference.
Comparison of Hepatocellular Uptake and Metabolism with Microsomal Metabolism. To allow direct comparison between the microsomally and hepatocyte-derived kinetic parameters, rates and clearances were expressed per gram of intact liver by the use of scaling factors. Figure 6 illustrates the scaled data for uptake and metabolism. At low substrate concentrations, microsomal values markedly exceed hepatocyte metabolism, whereas the latter are in close agreement with the hepatocyte uptake values. In Table 5, the corresponding linear parameters for clearance and permeability in hepatocytes and microsomes are expressed in terms of units of liver to allow direct comparison of the in vitro systems.
Discussion
PIs are a class of drugs known to be extensively and rapidly metabolized in vivo and to be substrates for several transporter proteins (McNicholl, 2004; Kashuba, 2005). We have investigated the metabolism of saquinavir, nelfinavir, and ritonavir in liver microsomes and in suspended hepatocytes to elucidate hepatic clearance mechanisms and to allow rational prediction of in vivo clearance. Both substrate depletion and metabolite formation approaches were used. PIs generally are highly plasma protein bound due to their high lipophilicity; therefore, binding to protein present in in vitro incubation matrices was measured. Saquinavir, nelfinavir, and ritonavir all showed significant nonspecific binding, which was corrected for throughout this study to provide true kinetic parameters. Nelfinavir has previously been shown to exhibit a higher degree of intracellular accumulation compared with other PIs (Jones et al., 2001; Khoo et al., 2002).
Significantly higher metabolic rates (corresponding to lower Km values) were observed when the kinetics of saquinavir, nelfinavir, and ritonavir in rat liver microsomes were compared with suspended rat hepatocytes, ranging from 6-fold lower for ritonavir and saquinavir to 30-fold lower for nelfinavir. When nelfinavir metabolite data are compared, a similar picture is observed, where the Km for nelfinavir metabolite M1 in rat microsomes was 80-fold lower than the Km in suspended rat hepatocytes. The differences between ritonavir metabolite Km values in rat microsomes and hepatocytes were less. Vmax values for rat microsomes and suspended rat hepatocytes, when scaled to common units of liver weight, were relatively similar, ranging from 3.2 and 2.4 for ritonavir to 26 and 13 nmol/min/g liver for saquinavir for rat microsomal and rat hepatocytes, respectively. It could be argued that the nelfinavir Vmax in microsomes (6 nmol/min/g liver) was underestimated due to either nonlog-linear depletion or substrate inhibition, whereas in suspended hepatocytes (25 nmol/min/g liver), there was no evidence of substrate inhibition, possibly due to the presence of phase II enzymes, preventing a build up of competing metabolite(s) (Jones et al., 2005).
To date, there is minimal literature information on in vitro metabolism of saquinavir, nelfinavir, and ritonavir in rat liver microsomes, and information on rat hepatocytes would appear to be nonexistent. Yamaji et al. (1999) studied kinetics in Wistar rat liver microsomes of a number of PIs and observed Km values of 8.3 and 5.9 μM and Vmax values of 1.4 and 0.7 nmol/min/mg protein for saquinavir and nelfinavir, respectively. Shibata et al. (2002b) also studied saquinavir metabolism in Wistar rat liver microsomes and determined Km and Vmax values of 37 μM and 4700 nmol/min/mg protein. These parameters differ from those determined in this work. They also differ from each other, and this affords very different estimates of saquinavir Clint: 0.17, 126, and 11 μl/min/mg protein for Yamaji et al. (1999), Shibata et al. (2002b), and this work, respectively. Discrepancies in parameter estimates may result from differences in methodology, although all three studies estimated kinetics by determining depletion of parent drug, together with the lack of correction for protein and nonspecific binding, leading to significantly higher Km estimates. No literature was identified on in vitro metabolism of ritonavir in rat hepatocytes, but in vivo and microsomal data published by Denissen et al. (1997) and Kumar et al. (1996) determined that metabolites M1, M2, M9, and M11 were the predominant primary metabolites, which agrees with observations in our work. Studies have been carried out in human liver microsomes at comparable substrate concentrations with the present study, and low Km estimates were reported: 0.61 and, 0.92 μM for saquinavir (Eagling et al., 2002) and ritonavir (Koudriakova et al., 1998).
Comparison of saquinavir-scaled Clint (Vmax/Km) (Table 5) mainly reflects the disparity in Km values, and the Clint of saquinavir rat liver microsomes is over 12 times higher than Clint in suspended rat hepatocytes. A comparison of nelfinavir-scaled Clint shows that the Clint of rat liver microsomes is between 5 and 8 times higher than Clint in suspended rat hepatocytes, depending on whether a Michaelis-Menten or substrate inhibition fit was applied. Ritonavir-scaled Clint value in rat liver microsomes is 9 times higher than Clint in suspended rat hepatocytes, again mainly due to the difference in Km. Consideration of Figs. 1 to 3 not only highlights the very rapid turnover in microsomes with respect to hepatocytes but also the issue of the lack of experimental data at the lower substrate concentrations; thus, the Km value and intrinsic clearance values are obtained by back extrapolating the higher concentration data as part of the nonlinear regression process. However, the differences between the hepatocyte and microsomal kinetic parameters are entirely consistent with the metabolic rate data shown in these figures.
The higher Km values observed in rat hepatocytes suggest that the unbound intracellular concentrations of saquinavir, nelfinavir, and ritonavir differ from the unbound extracellular concentration. It would seem reasonable to assume that the microsomal Clint represents the “true” Clint, with no nonmetabolic limitations and unbound drug concentration in the microsomal incubation being equivalent to the unbound drug concentration available to the drug-metabolizing enzymes. However, for hepatocyte Clint, the intracellular free drug concentration available to the enzyme is unknown. Various factors will affect the free intracellular drug concentration, including influx of drug into the cell (via active and passive means), efflux out of the cell, and intracellular protein binding. These results may result in a different Km value in hepatocytes.
Saquinavir, nelfinavir, and ritonavir all showed temperature-dependent uptake into suspended rat hepatocytes, and kinetic parameters were calculated for nelfinavir and ritonavir (saquinavir uptake showed no saturation over the concentration range studied). The Km values for nelfinavir and ritonavir were similar, whereas Vmax differed by 4-fold. Permeability values estimated from the ratio of Vmax to Km (nelfinavir and ritonavir) or from the slope of the rate plot (saquinavir) gave a rank order of permeability of nelfinavir > ritonavir > saquinavir (Table 2), comparable with the rank order of permeability determined by measuring uptake rate to equilibrium (Fig. 5D). Saquinavir, nelfinavir, and ritonavir all demonstrate high Kp values (310–3350), with the nelfinavir Kp significantly higher than the others. In contrast, Kpu was found to be similar for the three drugs (6–11, see Table 4), indicating a similar dependence on transporters. Intracellular fraction unbound values for all three drugs were low, particularly for nelfinavir, where 99.8% of intracellular nelfinavir was estimated to be bound. In contrast, high Kp values (100–1300) have also been measured for lipophilic amines such as propranolol and fluoxetine; however, once Kp is corrected for intracellular binding, there is essentially no evidence for active uptake (Hallifax and Houston, 2007).
Comparison of Perm and Clint values measured in rat hepatocytes (Table 2) reveals the same rank order across the three drugs and a similar order of magnitude between the parameters for each drug. Interestingly, hepatocyte metabolic parameters are consistently lower than microsomal Clint for all three compounds, indicating that uptake into cells, rather than enzyme factors, governs hepatocellular clearance. At low saquinavir concentrations, the metabolic rate in rat microsomes is greater than uptake rate in rat hepatocytes, reflecting an uptake limitation of metabolism in rat hepatocytes. In contrast, at higher concentrations once metabolic clearance becomes saturated, the metabolic rate is no longer limited by uptake. This is illustrated in Fig. 6 (saquinavir, ritonavir, and nelfinavir). Although the transport and metabolic studies were carried out using different rat strains, we believe this does not compromise the interpretation of these data. However, because rat strain differences in transporters have yet to be investigated, this remains a moot point.
In vivo, hepatocytes express both efflux and uptake transporters, and it is important to consider that the apparent uptake rate may be affected by efflux rate, especially because PIs are well documented to be substrates for efflux transporters (Srinivas et al., 1998). An increase in uptake rate in cell lines expressing both P-gp and uptake transporters has been observed using P-gp inhibitors (Jones et al., 2001; Su et al., 2004). In freshly isolated rat hepatocytes, expression of efflux transporters may be significantly lower than in vivo due to down-regulation and internalization of the apical membrane, and although P-gp has been detected on the hepatocyte apical membrane after isolation using immunohistochemical techniques, it is at a lower abundance than in vivo (Bow et al., 2008). Lam et al. (2006) have observed changes in uptake and clearance of erythromycin in rat hepatocytes following treatment with rifampin and GF120918, consistent with decreased and increased transporter activity (OATPs and P-gp), respectively.
Although there is little published work on uptake of PIs in the rat, some studies have been carried out in various human systems. Saquinavir, nelfinavir, and ritonavir have all been shown to inhibit uptake of the OATP substrates estradiol 17β-glucuronide (Tirona et al., 2003) and fexofenadine (Dresser et al., 2002), suggesting that they may possibly be substrates. Saquinavir has also been confirmed to be a substrate for OATP-A (OATP 1A2/SLC21A3) in HepG2 cells and Xenopus oocytes, and Km values determined were 94.6 and 36.4 μM, respectively (Su et al., 2004), which is consistent with this work, where a Km > 28 μM was estimated.
In the rat, all three PIs studied are high-clearance drugs (Lin, 1997; Gao et al., 2002; Shibata et al., 2002b). Using the well stirred liver model, hepatic extraction ratios were predicted from microsomal and hepatocellular Clint and Perm values, and as shown in Table 5, high extraction ratios were obtained in all cases. Thus, it is difficult at this level of comparison to conclude which in vitro system provides the most accurate prediction. However, it is clear from the above discussion that microsomal Clint values are misleading in terms of the magnitude of the turnover of these PIs in vivo. Thus, although cellular uptake rate limits the metabolism of saquinavir, nelfinavir, and ritonavir within intact hepatocytes, this process is efficient and strongly indicates that the hepatic clearances for these PIs are all blood flow limited in the in vivo situation.
Acknowledgments
We thank Nadine Warner (Roche Products Ltd, Welwyn Garden City, UK) for initiating the BBRSC Collaborative Award in Science and Engineering (CASE) project and Christoph Funk (Hoffman-La Roche Ltd, Basel, Switzerland) for valuable assistance and laboratory facilities for the hepatocyte uptake studies.
Footnotes
-
A.J.P. was financially supported by a Biotechnology and Biological Sciences Research Council Collaborative Award in Science and Engineering (CASE) studentship with Roche Products Limited.
-
Article, publication date, and citation information can be found at http://dmd.aspetjournals.org.
-
doi:10.1124/dmd.108.020917.
-
ABBREVIATIONS: PI, protease inhibitor; BSA, bovine serum albumin; Kp, tissue-to-medium total drug concentration ratio; fu, fraction(s) of unbound drug; LC, liquid chromatography; M1, 3′methoxy-4′hydroxynelfinavir); CLint, intrinsic clearance; Perm, linear permeability parameter; Kpu, tissue-to-medium unbound drug concentration ratio; M8, nelfinavir hydroxy-t-butamide; GF120918, N-(4-[2-(1,2,3,4-tetrahydro-6,7-dimethoxy-2-isoquinolinyl)ethyl]-phenyl)-9,10-dihydro-5-methoxy-9-oxo-4-acridine carboxamide; OATP, organic anion transporter polypeptide; P-gp, P-glycoprotein.
- Received February 7, 2008.
- Accepted April 17, 2008.
- The American Society for Pharmacology and Experimental Therapeutics