Abstract
The aim of this study was to explore the mechanisms governing the intra- to extracellular unbound concentration ratio (Kpu,u) for the HIV protease inhibitor atazanavir (ATV) in rat hepatocytes. We had previously proposed a new method to determine Kpu,u by using the unbound Km values from metabolism studies with suspended rat hepatocytes and rat liver microsomes. Following that method, we determined that the value of ATV Kpu,u was 0.32, indicating that ATV hepatocellular clearance is uptake rate–limited. This hypothesis was supported by the linear correlation between Kpu,u and active uptake clearance (P = 0.04; R2=0.82) in the presence of increasing concentrations of the uptake transport inhibitor losartan. Moreover, in contrast to an expected increase of Kpu,u upon inhibition of ATV metabolism, a decrease of Kpu,u was observed, suggesting an increased impact of sinusoidal efflux. In summary, involvement of active uptake transport does not guarantee high intracellular accumulation; however, it has a key role in regulating intracellular drug concentrations and drug metabolism. These findings will help improve future in vitro–to–in vivo extrapolations and likewise physiologically based pharmacokinetic models.
Introduction
Liver microsomes have been an established in vitro model to determine P450-mediated drug elimination for over 50 years (Rane et al., 1977). Currently, this relatively simple in vitro technique still stands as an invaluable preclinical high-throughput tool in drug discovery settings. However, absence of a plasma membrane with drug-transporting proteins [e.g., solute carrier (SLC)-transporters, ATP-binding cassette (ABC)-transporters] can cause liver microsome-based clearance predictions to deviate from the in vivo situation (Lam and Benet, 2004; Lu et al., 2006). Apart from drug transporters (uptake and efflux), intracellular metabolism and intracellular binding can also cause intracellular concentrations to differ greatly from medium concentrations (Parker and Houston, 2008). To overcome this problem, suspended hepatocytes that possess a cell membrane with drug transporting proteins, as well as both phase I and phase II metabolizing enzymes, are becoming the preferred in vitro drug-metabolism tool (Di et al., 2012). In suspended hepatocytes, the exposure of intracellular enzymes to unbound drug concentrations will more closely resemble in vivo conditions. However, when drug elimination data from these in vitro tools are used, unbound medium concentrations are often considered to equal intracellular unbound concentrations, which is rarely the case. The ratio between unbound intra- to extracellular concentrations (Kpu,u) can either be higher than 1 (metabolism/efflux rate-limited), equal to 1 (active/passive uptake approximates metabolism/efflux), or lower than 1 (uptake rate–limited) depending on the impact of each eliminating pathway (Pfeifer et al., 2013a). Not only clearance predictions, but also predictions regarding drug toxicity, drug efficacy, and drug-drug interactions could benefit from this information (Chu et al., 2013). Therefore, hepatic drug disposition models have been proposed to calculate intracellular drug exposure by combining uptake/efflux and metabolism data (Iwatsubo et al., 1999; Shitara et al., 2005; Webborn et al., 2007; Umehara and Camenisch, 2012). Still, the development of methods to accurately measure intracellular protein binding or intracellular drug distribution remains a great challenge. Several experimental techniques involving cell-homogenization have been proposed, but they disregard the dynamic interplay between drug transporters and drug metabolizing enzymes (Mateus et al., 2013; Pfeifer et al., 2013a). In contrast, indirect methods relying on differences in kinetic parameters such as the Ki-method presented by Brown et al. (2010) and the Km-method proposed by our group will take this dynamic interplay into consideration when calculating Kpu,u (Brown et al., 2010; Nicolaï et al., 2015). The Ki-method uses the ratio between the unbound Ki values of a drug metabolism inhibitor obtained in microsomes and hepatocytes to get an idea of the intracellular inhibitor accumulation. However, these calculations depend on the distribution of both substrate and inhibitor into the hepatocytes and are therefore potentially more challenging. On the other hand, the Km-method relies on the ratio between a compound’s unbound metabolic Km in liver microsomes (intracellular concentration) and its apparent Km in suspended rat hepatocytes (extracellular concentration), resulting in a direct value for Kpu,u. Inherent to this method, Kpu,u will reflect the impact of the processes controlling intracellular drug exposure. The current study aims to extend application of the Km-method—from calculating the Kpu,u during our previous experiments with verapamil (passive diffusion; Cyp3a1/2)—to atazanavir (active uptake/efflux; Cyp3a1/2) (Nicolaï et al., 2015). Thus, the HIV protease inhibitor atazanavir (ATV) was selected as a model compound since its elimination involves P450-mediated drug metabolism (hCYP3A4/5; rCyp3a1/2) and drug transport by sinusoidal ATP-binding cassette transporters (ABCC1; MRP1) as well as SLC-transporters (SLCO1B1/3) (Swainston Harrison and Scott, 2005; Kis et al., 2010; Wempe and Anderson, 2011; De Bruyn et al., 2015b). This will enable exploration of the interplay between active uptake/efflux transport and intracellular metabolism, which determine the intracellular unbound drug exposure.
Materials and Methods
Materials
Atazanavir sulfate (ATV) was obtained through NIH AIDS Reagent Program. Indinavir sulfate (IDV) was obtained from Hetero Drugs Limited (Hyderabad, India). 1-Aminobenzotriazole (ABT), benzbromarone, glucose 6-phosphate (G6P), collagenase type IV (C. hystolyticum, ≥125 CDU), Leibovitz-15 (L-15) powder (with glutamine), mineral oil, silicon oil, and sodium taurocholate hydrate were all purchased from Sigma-Aldrich (Schnelldorf, Germany). Trypan blue stain (0.4%), l-glutamine (200 mM), 10× phosphate-buffered saline (PBS), and fetal bovine serum (FBS) were obtained from Lonza SPRL (Verviers, Belgium). Dimethylsulfoxide (DMSO) and methanol (MeOH) were purchased from Acros Organics (Geel, Belgium). Acetonitrile (ACN), Na-acetate, and acetic acid were purchased from Analar-Normapur (VWR, Leicestershire, England). Na4-NADPH was obtained from MilliporeSigma (Billerica, MA). Ammonium acetate was obtained from Fisher Chemical (Landsmeer, Netherlands). HEPES (4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid) was purchased from MP Biochemical (Illkirch, France). 3-[[3-[(E)-2-(7-Chloroquinolin-2-yl)ethenyl]phenyl]-[3-(dimethylamino)-3-oxopropyl]sulfanylmethyl]sulfanylpropanoate sodium salt (MK571) was obtained from Merck Sharp & Dohme research laboratory (Rahway, NJ).
Animals
Male Wistar rats were purchased from Janvier (Le Genest-Saint-Isle, France) and housed according to Belgian and European laws, guidelines, and policies for animal experiments, animal housing, and animal care in the Central Animal Facility of the KU Leuven. All experiments involving laboratory animals were approved by The Institutional Ethical Committee for Animal Experimentation (license number: LA 1210261).
Determination of Unbound Fractions
Fumic, fuhep, fuIPRL, and fuplasma (unbound fraction in microsomes, hepatocytes, IPRL perfusate and plasma, respectively) were all determined by equilibrium dialysis using a HTDialysis apparatus (Gales Ferry, CT) fitted with membranes with a molecular mass cutoff of 12–14 kDa. The HTDialysis apparatus was subjected to circular agitation (175 rpm, 37°C) and samples were taken from donor and acceptor compartments in each well at 4 and 6 hours. Dialysis experiments were conducted in triplicate at an atazanavir concentration of 1 μM (0.2% v/v DMSO). Fuhep was determined with freshly-isolated and cryopreserved hepatocytes, metabolically inactivated by heat (50°C, 15 minutes) or by an incubation with ABT (1 mM) and compared with wells without cells. Integrity and viability of heat-inactivated cells were determined using the Trypan blue (0.04%) exclusion method. Fuhep and Fumic were determined in their respective incubation media. Binding of ATV to protein (20% blood) in the isolated perfused rat liver (IPRL) perfusate was determined by equilibrium dialysis with Krebs-Henseleit buffer (KHB) in the reference compartment (118 mM NaCl, 5.17 mM KCl, 1.2 mM CaCl2, 1.2 mM MgCl2, 23.8 mM NaHCO3, 12.5 mM HEPES, 5 mM d-glucose, and 5 mM Na-pyruvate, pH 7.4). Fuplasma was determined in plasma and compared with diffusion into a PBS-containing compartment.
Metabolism Studies in Rat Liver Microsomes
Rat liver microsomes (RLM) were prepared from male Wistar rats (177–244 g after 24 hours fasting) using sequential (ultra)centrifugation as described previously and stored at –80°C (Nicolaï et al., 2015). ATV and ABT solutions were prepared using microsomal incubation buffer (MIB) (3 mM MgCl2, 100 mM sodium phosphate buffer, pH 7.4) to acquire 4-fold concentrated solutions with a maximum DMSO content of 0.2%. RLM were gently thawed and kept on ice. Subsequently, they were diluted with MIB to obtain a 4- or 2-fold concentrated solution of RLM (i.e., 1 or 0.5 mg/ml). Incubations were performed on a shaking incubator (350 rpm, 37°C) in 48-well plates (Greiner-Bio-One, Wemmel, Belgium). ATV (100 μl) was preincubated with RLM (200 μl, 0.25 mg/ml final concentration) for 10 minutes before adding 100 μl of 4-fold concentrated prewarmed (37°C) NADPH solution (1 mM final concentration) containing glucose 6-phosphate (3 mM final concentration). During inhibition experiments, ABT (100 μl) was preincubated with RLM (100 μl, 0.25 mg/ml final concentration) and prewarmed (37°C) NADPH solution (1 mM final concentration) containing glucose 6-phosphate (3 mM final concentration) for 30 minutes, after which ATV (100 μl) was added (5–0.1 μM). Control wells were incubated in the absence of ABT or NADPH. Samples (75 μl) were taken at different time points and added to an equal volume of cold acetonitrile (ACN) containing internal standard (0.1 μM IDV). Samples were stored at –20°C for at least 1 hour prior to analysis. Before analysis, samples were thawed, vortexed, and centrifuged (10,500g) for 10 minutes at 21°C, the supernatant was transferred into microinserts for liquid chromatography–tandem mass spectrometry (LC-MS/MS) analysis. Linearity studies were performed with respect to ATV concentration, microsomal protein concentration, and incubation time. Subsequently, optimal conditions for determining the in vitro half-life were selected. All incubations were terminated at predetermined time points to ensure linear disappearance of ATV on an Ln(C)/time plot.
Metabolism Studies in Suspended Rat Hepatocytes
Suspended rat hepatocytes (SRH) were isolated from male Wistar rats (180–257 g; average weight, 215 g) according to the two-step collagenase perfusion method, cryopreserved, and thawed as described previously (Nicolaï et al., 2015). They were resuspended in L-15* (90% L-15, 3.6 mM l-glutamine, 9.9 mM d-glucose, 9 mM HEPES, 3.6 mM NaHCO3, pH 7.4), after which they were evaluated using the Trypan blue (0.04%) exclusion method to determine viability and cell density. Viability of freshly-isolated SRH and cryopreserved SRH ranged from 85 to 92% and from 75 to 85%, respectively. ATV and ABT solutions were prepared using L-15* to acquire 2- or 4-fold concentrated solutions with a maximum DMSO content of 0.2%. Cells were 2-fold concentrated and preincubated on a shaking incubator (37°C, 350 rpm) for 10 minutes in Advanced TC 24-well plates (Greiner-Bio-One, Wemmel, Belgium). Incubations were initiated by adding an equal volume of 2-fold concentrated ATV solution (L-15*, 0.05% DMSO). When inhibition experiments were performed, cells were concentrated 4-fold and preincubated (37°C, 350 rpm) in the presence of ABT for 30 minutes. Incubations were initiated by adding 2-fold concentrated ATV solution (L-15*, 0.05% DMSO). Linearity studies were performed with respect to ATV concentration, hepatocyte density, and incubation time. Subsequently, optimal conditions for determining the in vitro half-life were selected (i.e., different hepatocyte densities were selected for different incubation concentrations of ATV). All reactions were stopped by adding one volume of sample to two volumes of ice-cold MeOH containing internal standard (0.1 μM IDV). Samples were stored at –20°C for the duration of at least 1 hour prior to analysis. Finally, samples were thawed and centrifuged (20,816g) for 10 minutes at 21°C and the supernatant was transferred into microinserts for LC-MS/MS analysis.
Uptake Studies in SRH
Freshly isolated hepatocytes from male Wistar rats (180–220 g) were resuspended in Krebs-Henseleit buffer that had been sparged with carbogen (95%:5% O2/CO2), after which they were evaluated using the Trypan blue (0.04%) exclusion method to determine viability and cell density. Cells were diluted to a 4-fold concentrated cell density (1 million cells/ml final density). Two-fold concentrated ATV solutions and 4-fold concentrated uptake inhibitor solutions were prepared in KHB (0.5% final DMSO content). Uptake studies were performed using the oil-spin method. The hepatocytes (175 μl) were preincubated for 10 minutes (37°C) in the presence or absence of an uptake transporter inhibitor, i.e., benzbromarone or losartan (175 μl). Following preincubation, ATV (350 μl) was added and uptake was assessed. After 30 seconds, triplicate 200-μl aliquots were rapidly pipetted on top of an oil layer (82:18 silicon oil/mineral oil, 1.051 g/ml) above a NaCl solution (8% w/v) in 1.5-ml test tubes. The test tubes were centrifuged immediately (16,162g) for 3 minutes. Without delay, the tubes were snap-frozen in cooled ethanol (–80°C). After all tubes were frozen, the tube-pellets were collected in glass test tubes after cutting the bottom part (∼80 μl) with a tube cutter. Pellets were lysed with 300 μl 70:30 MeOH/H2O containing internal standard (0.01 μM IDV) and shaken for at least 30 minutes (300 rpm, room temperature). Following lysis, the mixture was collected in test tubes and stored at –20°C until analysis. Finally, samples were thawed and centrifuged (20,816g) for 10 minutes at 21°C and the supernatant was transferred into microinserts for LC-MS/MS analysis. To determine active uptake transporter kinetics, passive uptake of ATV was subtracted from total uptake. The passive uptake of ATV was assessed at different concentrations of ATV in the presence of benzbromarone (75 μM), which resulted in complete inhibition of active uptake (data not shown).
Monolayer Cultured Rat Hepatocytes
On the day before seeding, Advanced TC 24-well plates (Greiner Bio-One, Wemmel, Belgium) were coated with 50 μl of ice-cold (4°C) neutralized rat-tail collagen (∼1.5 mg/ml, pH 7.4). Plates were stored in a humidified incubator (37°C, 5% CO2) for 2 hours before hydrating them with 500 μl/well of warm (37°C) sterile PBS and storing them overnight in the humidified incubator (37°C, 5% CO2). In-house cryopreserved rat hepatocytes from male Wistar rats (180–206 g) were thawed as reported previously and suspended in day-0 cell culture medium (Williams’ E medium, 1 g/l glucose, 10% FBS, 2 mM glutamine, 100 IU/ml penicillin, 100 μg/ml streptomycin, 1 μM dexamethasone, 4 μg/ml human insulin). Just before seeding, viability and cell density were determined using the Trypan blue (0.04%) exclusion method (viability 80–90%). Following aspiration of PBS from the collagen-coated plates, hepatocytes were seeded at a density of 400,000 cells/well. Cells were left to attach to the collagen for 2 hours inside the humidified incubator (37°C, 5% CO2) before plates were shaken vigorously to remove unattached cells and the medium replaced with 500 μl of warm (37°C) day-0 medium. Cultures were kept inside the humidified incubator for 2 hours until the start of the experiment.
Efflux Studies in Monolayer Cultured Rat Hepatocytes
Four hours after seeding, monolayer cultured rat hepatocytes were washed two times with warm (37°C) Hanks’ balanced salt solution (HBSS) (10 mM HEPES, pH 7.4) before incubating for 30 minutes in the presence of 250 μl of ABT (500 μM) in KHB (10% FBS) to avoid interfering with ATV metabolism. Subsequently, 250 μl of KHB (10% FBS) containing 20 μM ATV was added to each well and cells were loaded for 20 minutes. Following loading, cells were washed three times with ice-cold (4°C) HBSS (10% FBS), after which ATV efflux was commenced by adding 500 μl of prewarmed (37°C) KHB (10% FBS) containing either 0.01% DMSO or 100 μM MK571. Medium samples (100 μl) were taken at 3, 5, 10, and 15 minutes and added to 210 μl of MeOH containing internal standard (0.1 μM IDV). Collagen-coated wells without cells were incubated to correct for diffusion of ATV from the collagen. Additional wells were lysed immediately after loading [70:30 MeOH/KHB (FBS 10%), containing 0.07 μM IDV] to evaluate loading efficiency. Samples were centrifuged (20,816g) for 10 minutes at 21°C and the supernatant was transferred into microinserts for LC-MS/MS analysis. Bioanalysis was performed on the day of experiments.
Isolated Perfused Rat Liver
Male Wistar rats (290–330 g) were anesthetized and the portal vein was cannulated. The thoracic vena cava inferior was severed without cannulation and the abdominal vena cava inferior was closed with a surgical clip. The bile duct was cannulated with 10–15 cm of PE-10 tubing (0.28 mm × 0.61 mm id × od). To maintain bile flow, sodium taurocholate hydrate was infused continuously into the circulating system (30 μmol/h) (Paumgartner et al., 1974). During excision, the liver was perfused (30 ml/min) with sparged (95%:5% O2/CO2) KHB (37°C, pH 7.4). Following hepatectomy, the liver was placed on a collecting platform in a humidified and temperature controlled (37°C) chamber. Perfusion with KHB was continued until 30 minutes after cannulation. After organ acclimatization, the perfusion system was switched to KHB + 20% heparinized rat blood (flow rate = 22.6 ± 2.1 ml/min), containing 0.4 μmol of ATV to reach an initial perfusate concentration of 5 μM, which was oxygenated while passing through silastic Q7-4750 semipermeable tubing (Dow Corning Europe SA, Belgium) inside an artificial lung ventilated with carbogen (95%:5% O2/CO2). The perfusate was sampled (100 μl) just before and directly after the liver at 2-minute intervals until 10 minutes and subsequently every 5 minutes until 25 minutes after dosing. Bile samples were collected every 5 minutes. Liver functionality was monitored by measuring bile flow, portal vein pressure (water column ∼10–15 cm water), and visual appearance. Perfusate samples were processed immediately (see Bioanalysis) and stored at –20°C until analysis. Adsorption studies were performed to assess adsorption of ATV to the materials of the perfusion setup. Tubing materials consisted of glass, THV220/221GZ (Polyfluor Plastics, Breda, The Netherlands), Tygon chemical, and Teflon.
Bioanalysis
IPRL samples (30 μl) were diluted with water (120 μl) and protein was precipitated with ACN (300 μl) containing internal standard (0.1 μM IDV). Following a vortex step (2 × 30 seconds), samples were centrifuged (20,816g) for 10 minutes at 21°C. The supernatant was transferred to microinserts and sample vials before being injected directly into the LC-MS/MS system. The LC-MS/MS system (Thermo Scientific, San Jose, CA) consisted of an Accela autosampler, an Accela pump, and a TSQ Quantum Access triple quadrupole mass spectrometer equipped with an electrospray ionization (ESI) source. A Kinetex XB–C18 column (50 mm × 2.1 mm, 2.6 μm), protected by a SecurityGuard ULTRA precolumn (Phenomenex, The Netherlands) was used for chromatographic separation. The mobile phase for ATV analysis consisted of a 0.5 mM ammonium acetate buffer (pH 3.5) (A) and MeOH (B). Gradient elution at a constant flow (400 μl/min) was performed as follows: 95% A decreased linearly to 5% in 2 minutes; was kept constant for 1 minute; followed by a linear increase back to 95% A in 10 seconds and re-equilibration for 1.0 minute with 95% A, before the next injection. The total run time amounted to 4 minutes. The column oven and autosampler tray temperature were set at 30°C and 15°C, respectively. The MS was operated in positive ionization mode. Spray voltage was 3500 V and argon was used as collision gas at a pressure of 1.5 mTorr. The MS was operated in a three-channel selected reaction monitoring mode with a scan time of 75 milliseconds. Appearance and disappearance of ATV mother compound, together with the internal standard (IDV), were measured during analysis. Transitions monitored were 705.4 → 168.1 m/z and 614.5 → 421.3 m/z with retention times of 3.16 and 2.94 minutes for ATV and IDV, respectively. Other ionization parameter settings were: capillary temperature (170°C), vaporizing temperature (300°C), sheet gas pressure (50 arbitrary units), auxiliary gas pressure (0 arbitrary units), ion sweep gas pressure (40 arbitrary units), and collision energy (43 arbitrary units). Intra- and intermediate precision of quality control samples (0.01 μM and 0.1 μM) was below 10%.
Data Analysis.
Uptake in SRH.
Net active uptake values of ATV in SRH were calculated by subtracting the uptake in the presence of 75 μM benzbromarone, representing passive uptake, from total uptake at 37°C. Using GraphPad Prism version 5.00 for Windows (GraphPad Software, Inc., San Diego, CA), the Michaelis-Menten equation (eq. 1) was fitted to the data, and Vmax and Km (± S.E.) were determined. Passive uptake clearance was calculated from the slope of the passive uptake rate over concentration curve. Active uptake clearance was calculated by dividing Vmax by Km or by dividing the active uptake rate (pmol/min per million cells) by the concentration of ATV at which active uptake was measured.(1)The inhibitory effect–Emax model (eq. 2) was used to describe the concentration-dependent inhibition of losartan on the uptake of ATV in SRH. Fits were obtained using nonlinear regression in GraphPad Prism 5.00 software.
(2)E represents the uptake of ATV in SRH; Emax, the uptake of ATV in the absence of inhibitor; E0, the uptake of ATV at the maximal inhibitory effect; (Emax – E0), the maximal inhibitory effect; n, the Hill factor; and IC50, the concentration at which the inhibitor exerts its half maximal inhibitory effect. Subsequently, the Cheng-Prusoff equation (eq. 3) was applied to determine Ki from IC50, where S is the ATV concentration at which IC50 was determined and Km the Michaelis-Menten constant for active uptake of ATV in SRH.

Metabolism in SRH and RLM.
The slope of ATV (Ln(C)) disappearance over time was applied using the in vitro half-life method to determine the rate of metabolism (pmol/min per million cells or milligrams of protein) for ATV metabolism in SRH and RLM as shown in eq. (4) (Obach, 1999):(4)The rate of metabolism was fitted as a function of the ATV concentration to acquire Michaelis-Menten parameters Vmax and Km (± S.E.). The unbound intrinsic metabolic clearance was calculated according to eq. 5.
(5)Following the methodology described in our previous study, the Kpu,u was determined by dividing the unbound Km value of ATV metabolism in RLM (Kmmic,u) by the unbound Km value of ATV metabolism in SRH (Kmhep,u) (eq. 6) (Nicolaï et al., 2015):
(6)By applying Kpu,u in eq. (7), the unbound intracellular intrinsic metabolic clearance in SRH was calculated (Clint,hep,u,intracellular).
(7)According to Webborn et al. (2007), the metabolic Clint,hep,u can be calculated by combining Clint,mic,u with passive and active uptake clearance of ATV (eq. 8). Subsequently, by combining eq. (8) and eq. (9), Kpu,u can be calculated from the unbound metabolic clearance in RLM and the active/passive uptake and efflux clearance values in SRH (eq. 10); the latter allows for the calculation of Kpu,u in the presence transporter or metabolism inhibitors (Supplemental Figures 1 and 2).
(8)
(9)
(10)Clint,up,act is the active uptake clearance, Clint,up,pass is the passive uptake clearance, Clint,mic,u is the microsomal unbound clearance, and Clint,eff is the total efflux clearance (passive and active). Clint,eff was calculated assuming that at steady state the sum of the influx rates (concentration × Clint) equals the sum of the efflux and metabolic elimination rates (Supplemental Equation 1). Influx rates were measured for the extracellular Kmhep,u of ATV metabolism in SRH. The microsomal rate of metabolism was calculated for the intracellular concentration attained at the extracellular concentration of Kmhep,u, which is Kmmic,u (Supplemental Table 1).
IPRL.
For the IPRL data, Cp0, kel, Vd, and Clint,IPRL were calculated as reported previously by using the outflow concentrations, reflecting the equilibrium between liver tissue and perfusate concentration. Clint,IPRL is subsequently scaled to in vivo with scaling factors (SF) for gram of liver per kilogram of body weight (determined for each individual IPRL experiment) (Nicolaï et al., 2015). Finally, Clint,IPRL was divided by fuIPRL to correct for binding of ATV in the perfusate (Clint,IPRL,u). The previously published blood concentration-time profile of ATV, obtained after intravenous administration to rats (n = 3), was fitted to a two-compartmental model (De Bruyn et al., 2015a). Intrinsic clearance values of ATV in RLM, SRH, and IPRL were used to simulate the in vivo data as reported previously (Supplemental Fig. 3) (Nicolaï et al., 2015). All data were scaled with conventional SF for MPPMC (0.37 mg of microsomal protein per million cells), MPPGL (61 mg of microsomal protein/g of liver), and HPGL (163 million cells/g of liver) (Smith et al., 2008) Activity-based SF for MPPMC, MPPGL, and HPGL were calculated as reported previously (eqs. 11–13) (Nicolaï et al., 2015).



Statistics.
Analysis of variance (α level of 0.05; Dunnett post-hoc test) was applied to compare control conditions to experiments including inhibitors, using GraphPad Prism version 5.00 for Windows (GraphPad Software).
Results
Determination of Unbound Fractions.
The unbound fractions of total ATV concentrations in the presence of hepatocytes, microsomes, and IPRL-buffer, were determined using equilibrium dialysis. Values for fuhep (fresh and cryopreserved), fumic, and fuIPRL were 0.96 ± 0.1, 0.77 ± 0.2, 0.76 ± 0.1, and 0.27 ± 0.02, respectively. Fu values were used to correct total (bound + unbound) clearance values, rendering the unbound clearance values. Fuplasma and fublood were determined previously by our group and valued 0.075 and 0.85, respectively (De Bruyn et al., 2015a)
Metabolism Studies in RLM and SRH.
The Michaelis-Menten equation was fitted to observed values of ATV metabolism in RLM (Fig. 1) and SRH (Fig. 2). All metabolic parameters are summarized in Table 1. There was no statistically significant difference (P = 0.68) between the unbound Kmhep values of ATV in freshly-isolated (0.83 ± 0.14) and cryopreserved (0.94 ± 0.24) SRH. Kpu,u values were 0.32 ± 0.07 and 0.28 ± 0.1 for ATV in freshly isolated and cryopreserved SRH, respectively. Using Kpu,u, Clint,hep,u was calculated to Clint,hep,u,intracellular, resulting in values of 514 ± 149 and 459 ± 185 μl/min per million cells for freshly isolated and cryopreserved SRH, respectively. Clint,hep,u,intracellular was comparable to the scaled Clint,mic,u. Clint,mic,u (1910 ± 323 μl/min per milligram of protein) was scaled to the cellular level using the activity-based scaling factor (0.30 mg of microsomal protein per million cells) for MPPMC, which was determined previously by our group, to attain a value of 554 ± 105 μl/min per million cells (Nicolaï et al., 2015).
Michaelis-Menten plot of atazanavir metabolism in RLM (A) and the corresponding atazanavir intrinsic clearance in RLM (B) as a function of atazanavir concentrations. Points represent mean values ± S.D. of four incubations performed in triplicate with two different batches of RLM.
Michaelis-Menten plots of atazanavir metabolism in freshly-isolated (●) and cryopreserved (○) SRH. (A) Corresponding atazanavir intrinsic metabolic clearance in SRH as a function of atazanavir concentrations. (B) Points represent mean values ± S.D. of triplicate incubations with two batches of SRH.
Michaelis-Menten parameters (Km, Km × fu, Vmax) describing atazanavir metabolism in RLM and SRH, together with corresponding total and unbound intrinsic clearance values (Clint and Clint/fu)
Km and Vmax values were obtained by fitting the Michaelis-Menten equation to the observed rate of metabolic atazanavir disappearance as a function of the atazanavir concentration (Fig. 2). Values are means ± S.D. of triplicate experiments with two batches of SRH and two batches of RLM.
Uptake Studies in Freshly-Isolated SRH.
The uptake of ATV in SRH in the presence and absence of the Oatp inhibitor benzbromarone is shown in Fig. 3A. The slope of the uptake rate of ATV in the presence of benzbromarone, representing the passive uptake clearance (Clint,up,pass), amounted to 134 ± 4 μl/min per million cells. The Michaelis-Menten equation was fitted to the observed values, resulting in a Km of 4.0 ± 0.5 μM and Vmax of 399 ± 22 pmol/min per million cells (Fig. 3B, Table 2).
Total (▪), passive (●), and active (○) uptake rates of atazanavir in SRH as a function of the atazanavir concentration. (A) Passive uptake rates were measured in the presence of the uptake transport inhibitor benzbromarone (75 μM), and active uptake rates were calculated from the difference between total and passive uptake rates. Close-up of the Michaelis-Menten plot of atazanavir active uptake rates in SRH as a function of the atazanavir concentration. (B) Points represent mean values ± S.D. of triplicate incubations with two batches of freshly isolated SRH.
Michaelis-Menten parameters for atazanavir uptake in SRH together with corresponding passive and active uptake clearance values
Km and Vmax were obtained by fitting the Michaelis-Menten equation to the observed rate of active atazanavir uptake in SRH as a function of atazanavir concentration (Fig. 3). Values are means ± S.D. of triplicate incubations in two batches of freshly-isolated SRH. Cmax,u (1.048 μM) equals the unbound Cmax in human plasma.
Effect of Uptake Inhibition on ATV Kpu,u.
To measure the effect of uptake transport inhibition on Kpu,u and likewise Clint,hep,u, the uptake transport inhibitor should not interfere with ATV metabolism. The effect of losartan (LOS) on the metabolism of ATV in RLM is shown in Fig. 4A. Only at concentrations higher than 10 μM LOS a statistically significant difference with the control condition was observed (P < 0.05). On the contrary, when the inhibitory effect of LOS on ATV uptake in SRH was determined, the maximal inhibitory effect had already been attained at 10 μM (Ki = 0.63 μM) (Fig. 4B). Therefore, concentrations of LOS lower than or equal to 10 μM (1, 5, and 10 μM) were selected to determine the Kpu,u of ATV in SRH in the presence of LOS. A linear correlation was observed (P = 0.04; R2 = 0.82) between the decrease in uptake clearance and the decrease in Kpu,u in the presence of different concentrations of LOS (Fig. 5).
Metabolism (% of control) of atazanavir in RLM in the presence of different concentrations of losartan. (A) Concentration-dependent inhibition of atazanavir uptake by losartan in SRH. (B) Points represent means ± S.D. of triplicate incubations with a single batch of RLM and two batches of SRH.
Linear correlation (P = 0.04; R2 = 0.82) between the Kpu,u and atazanavir uptake clearance, in the absence and presence of 1, 5, and 10 μM losartan. Values represent means ± S.E.M. of Kpu,u values determined with triplicate incubations in four different batches of freshly isolated SRH, compared with uptake clearance (means ± S.D.) of atazanavir in SRH in the presence of losartan (0 μM, 1 μM, 5 μM, 10 μM) in two batches of SRH.
Effect of Metabolism Inhibition on ATV Kpu,u.
Figure 6 shows the effect of 1-aminobenzotriazole (ABT) on ATV metabolism in both RLM and SRH. A proportional decrease in both Vmax and Km of ATV metabolism was observed when different concentrations of ABT were coincubated with RLM (Fig. 6A) [all fits are shown in (Supplemental Figs. 4–7]. However, when ATV was coincubated with ABT in SRH, the Km decreased relatively more slowly than the Vmax as a function of the ABT concentration (Fig. 6B). Hence, on the basis of eq. (6), ATV Kpu,u decreased with increasing ABT concentrations. In other words, the unbound intracellular ATV exposure decreases as a function of the ABT concentration (Fig. 7A).
(A) Michaelis-Menten parameters (Vmax, Km) for atazanavir metabolism in SRH in the presence of different concentrations of ABT. (B) In contrast to incubations with RLM, decrease of Km in SRH is less pronounced compared with Vmax. Bars represent means ± S.E. of triplicate incubations in single batches of RLM and SRH. Repeated experiments confirm this disconnect between RLM and SRH (Supplemental Figures 4–7).
Observed (▪) and calculated ([ATV] = Cmax,plasma,u; ●) change in atazanavir Kpu,u as a function of the ABT concentration. (A) The calculated change in Kpu,u as a function of ABT the concentration was obtained with eq. (10) (Webborn et al., 2007). Efflux of atazanavir in monolayer cultured rat hepatocytes in the absence (full line) and presence (dotted line) of the nonspecific Mrp inhibitor MK571 (100 μM). (B) Points represent means ± S.D. of efflux experiments performed in triplicate in three separate batches of rat hepatocytes.
ATV Efflux in Monolayer Cultured Rat Hepatocytes.
To evaluate whether ATV was a substrate for efflux transporters present in short term cultured rat hepatocytes, efflux of ATV was assessed in day-0 monolayer cultured rat hepatocytes in the absence and presence of 100 μM MK571. Figure 7B shows that 100 μM MK571 was able to decrease ATV efflux by >80%, confirming that ATV efflux is almost entirely reliant on an inhibitable efflux process.
Activity-Based Scaling Factors.
As reported in our previous work, activity-based SF for MPPMC, MPPGL, and HPGL can be calculated on the basis of data from different model systems (Nicolaï et al., 2015). To evaluate whether such SF are compound-dependent, they were calculated for ATV and compared with the SF obtained previously with verapamil. As described in eqs. 11 and 12, unbound intrinsic Cl of ATV in IPRL is needed to calculate MPPGL and HPGL. Clint,IPRL,u amounted to 826 ± 149 ml/min per kilogram of body weight (Fig. 8, Table 3). Activity-based SF are shown in Table 4. To evaluate the predictive value of preclinical data obtained during the present study, the in vivo pharmacokinetic profile of ATV in rats was simulated (Supplemental Fig. 3). All preclinical models performed well in predicting the in vivo elimination of ATV with RSS values of 0.92, 0.95, 0.96, and 0.98 for simulations with cryopreserved SRH, IPRL, freshly isolated SRH, and RLM, respectively.
Perfusate concentration (μM)-time profiles of atazanavir during IPRL experiments. Cin and Cout represent perfusate concentrations immediately before and after the liver, respectively. Points represent means ± S.D. of three separate liver perfusions.
Pharmacokinetic parameters describing atazanavir disposition in the IPRL system t1/2 and Vdss were determined as reported under Data Analysis
Values are means ± S.D. of IPRL experiments with three different livers.
Activity-based rat scaling factors for MPPMC (mg of protein per million cells), MPPGL (mg of protein/g of liver) and HPGL (million cells/g of liver)
Scaling factors were calculated for atazanavir following the described methodology and compared with scaling factors derived from verapamil data and “conventional” rat scaling factors as reported in literature. Values are means ± S.D.
Discussion
Transport-metabolism interplay of hepatic atazanavir (ATV) clearance was investigated to elucidate the mechanisms governing intracellular unbound ATV exposure. As a measure for intracellular drug exposure, the hepatocellular Kpu,u (0.32–0.28; Table 1) was calculated with unbound metabolic Km values retrieved from (in vitro) metabolism experiments with suspended rat hepatocytes (SRH) and rat liver microsomes (RLM). This approach for Kpu,u determination was recently introduced by our group with verapamil as a model compound (Nicolaï et al., 2015). Interestingly, the low (<1) value of Kpu,u for ATV was an indication for uptake rate-limited metabolic clearance (Fig. 9A). On the other hand, total liver-to-plasma concentration ratio of 3 has been reported (Fukushima et al., 2009). Uptake experiments with SRH revealed that intracellular ATV concentrations were controlled, at least to some extent, by a saturable uptake process (Fig. 3; Table 2). This is in line with previous work, which showed the role of OATP/Oatp for hepatocellular uptake and the rate-limiting effect of transporters on hepatic clearance of several HIV protease inhibitors (Brown et al., 2010; De Bruyn et al., 2015a). In contrast to findings from the present study, uptake transporter involvement is most often associated with high cell-to-medium concentration ratios (Parker and Houston, 2008; Brown et al., 2010). The hypothesis concerning uptake rate-limited metabolism was challenged further by investigation of the impact of uptake transport inhibition on Kpu,u. Given the nature of our approach to calculate Kpu,u (on the basis of metabolic Km values), an uptake transport inhibitor not interacting with ATV metabolism was required. Under these conditions, a shift of the metabolic Km of ATV in SRH was anticipated through inhibition of active uptake transport. LOS was evaluated as an inhibitor for ATV uptake in SRH (Ki = 0.62 μM; Fig. 4B). Whereas active uptake was completely inhibited at 10 μM, inhibition of metabolism at this concentration of LOS in RLM was excluded (P < 0.05) (Fig. 4A). Coincubation of ATV with LOS during metabolism experiments in freshly-isolated SRH revealed a statistically significant linear correlation (P = 0.04; R2 = 0.82; Fig. 5) between ATV uptake clearance and Kpu,u, confirming importance of ATV uptake for hepatocellular ATV metabolism (Fig. 9B). Overall, the current approach can be classified under “estimation of intracellular drug concentrations by modeling and simulation,” one of the main indirect methodology types referred to in a recent review on intracellular drug concentration determination (Chu et al., 2013). However, rather than applying a modeling approach, this article aims to quantitatively link hepatocellular drug disposition mechanisms to the hepatocellular Kpu,u.
Schematic overview illustrating the impact of atazanavir (ATV) disposition pathways on the hepatocellular Kpu,u in the absence of an inhibitor (A), presence of losartan (LOS) (B), or presence of ABT (C). Size of arrows and ATV indicate the extent of atazanavir flux and local atazanavir concentrations, respectively.
Intuitively, and with eq. (10) as a basis (Fig. 7A), inhibition of intracellular metabolism was expected to increase Kpu,u (Chang et al., 2014). Ideally, a noncompetitive metabolic inhibitor (no effect on Km) should be applied to investigate the effect of metabolism inhibition on Kpu,u. Thus, the inhibition profile of the general P450 inhibitor ABT for ATV metabolism was determined. Surprisingly, both Km and Vmax of ATV metabolism in RLM decreased proportionally when ABT concentrations were increased (Fig. 6A). This identified ABT as an uncompetitive inhibitor for ATV metabolism (Ki = 2.34 ± 0.02). To our knowledge, this is the first report of ABT being identified as an uncompetitive inhibitor. Notwithstanding the direct effect of ABT on the Km of ATV metabolism, it was still a suitable inhibitor to determine the effect of P450 inhibition on Kpu,u. When ATV was coincubated with ABT in SRH, the Km of ATV metabolism decreased at a relatively slower rate with respect to ABT concentrations compared with Vmax (Fig. 6B). However, inhibition of metabolism was expected to increase the intracellular unbound ATV accumulation, which would be reflected by a relatively faster decrease of Km compared with Vmax. The latter would be the result of an increase in Kpu,u with increasing concentrations of ABT, i.e., lower metabolic capacity (Fig. 7A). Since interaction of ABT with ATV uptake was ruled out (Supplemental Fig. 8) and ABT has never been shown to interact with efflux transporters, we deemed the direct influence of ABT on ATV transport unlikely (Kimoto et al., 2012). Additionally, no deviation from linearity was observed for the metabolism of ATV as a function of hepatocyte density curves (Supplemental Fig. 9), ruling out the hypothesis that ATV metabolites interact with ATV disposition in the SRH setup. Finally, in line with previous reports on transport-metabolism interplay, inhibition of intracellular metabolism could increase the susceptibility of ATV for efflux transporters present in SRH (Shi and Li, 2014).
Indeed, the current data already show a relatively high total efflux clearance (Clint,eff = 420 μl/min per million cells; Table 2; (Supplemental Equation 1; (Supplemental Table 1) compared with the passive uptake clearance (Clint,up,pass = 134 μl/min per million cells) of ATV. Additionally, the nonspecific Mrp inhibitor MK571 decreased ATV efflux in rat monolayer cultures by more than 80% (Fig. 7B). The intracellular gradient of unbound ATV, driven by intracellular metabolic ATV depletion, may become less steep upon inhibition of P450 and decrease overall flux of ATV toward the endoplasmic reticulum. As a consequence, deep intracellular drug exposure at the site of metabolism is decreased, thus increasing the likelihood of ATV binding to membrane-bound efflux transporters (Fig. 9C). This concept of inhomogeneous intracellular unbound drug distribution is not improbable, since several small molecules have been shown to have specific affinities for subcellular organelles, altering local intracellular drug concentrations (Matijašić et al., 2012; Pfeifer et al., 2013a; Fu et al., 2014). Likewise, the Km-method, which reflects intracellular unbound ATV concentrations at the level of P450 enzymes, depends on ATV concentrations in the vicinity of the endoplasmic reticulum. This inhomogeneous drug distribution could imply that different substrate concentrations should be considered at the intracellular locations of P450-enzymes and ABC-transporters, respectively.
In the present study, uptake and metabolism parameters were determined in SRH and RLM and combined to determine Kpu,u and Clint,eff for SRH (eqs. 6 and 10). If Clint,eff would be estimated from monolayer- or sandwich-cultured hepatocyte experiments, it should be corrected for altered expression levels of efflux transporters in these systems (De Bruyn et al., 2013). Alteration of expression levels as a function of culture time or internalization of canalicular efflux transporters upon isolation will affect intracellular ATV concentrations and likewise Kpu,u. Therefore, Kpu,u determined for SRH should not be transferred to other systems with different expression patterns of drug transporters and/or metabolic enzymes.
Consistent with our previous study, in which activity-based scaling factors were calculated using verapamil (Nicolaï et al., 2015), SF were calculated using ATV metabolic clearance in RLM, SRH, and IPRL. The calculated SF valued 0.27–0.24 mg of microsomal protein per million cells (MPPMC), 40–46 mg of protein per g liver (MPPGL), and 150–191 million cells/g of liver for freshly-isolated and cryopreserved SRH, respectively (Table 4). These values were similar to the activity-based SF calculated with verapamil and the P450-content–based SF reported in literature. However, as discussed in our previous article, the activity-based SF for MPPGL and HPGL in freshly-isolated SRH as determined with verapamil (80 mg of protein/g of liver; 269 million cells/g of liver), deviated from all other SF (Nicolaï et al., 2015). Confidence in current calculations improved, since the activity-based SF values calculated during this study coincided reasonably well with previously determined values, even though they were obtained with different RLM, SRH, IPRL preparations, and even another compound. Additionally, Clint,mic,u equaled the intracellular unbound clearance of ATV in both freshly isolated and cryopreserved SRH. This confirmed the previously reported preserved intracellular metabolic capacity of P450 enzymes following cryopreservation of SRH (Nicolaï et al., 2015).
In summary, during the present study, intracellular unbound ATV concentrations were correlated with the rate of active uptake transport, illustrating that total uptake is the rate-limiting process in ATV hepatic clearance in the rat. Consistently, despite active uptake transport but in line with sinusoidal efflux transport and significant intracellular metabolism, a low Kpu,u value was obtained (0.32). Inhibition of ATV metabolism with ABT unexpectedly decreased rather than increased the Kpu,u, pointing toward a possible mechanistic interplay between P450-mediated ATV metabolism and hepatocellular efflux transporters. Simultaneously, ABT was identified as an uncompetitive inhibitor of ATV metabolism. The current findings will help improve our understanding of the link between mechanisms governing intracellular hepatic drug disposition, aiming to ameliorate future PBPK modeling algorithms.
Acknowledgments
Janna Mertens, for providing understanding patience and loving care.
Authorship Contributions
Participated in research design: Nicolaï, Annaert.
Conducted experiments: Nicolaï, Thevelin.
Contributed new reagents or analytic tools: Nicolaï, De Bruyn.
Performed data analysis: Nicolaï, Annaert.
Wrote or contributed to the writing of the manuscript: Nicolaï, Annaert, De Bruyn, Augustijns.
Footnotes
- Received October 27, 2015.
- Accepted December 23, 2015.
This study was partially supported by a PhD scholarship awarded to Johan Nicolaï by the Agency for Innovation by Science and Technology [Agentschap voor innovatie door wetenschap en technologie (IWT), Flanders, Belgium, project number 111193], by the Scientific Research Network of the Research Foundation [FWO, Flanders, Belgium, grant number G.0662.09N], and by internal funds of the Laboratory for Drug Delivery and Disposition, KU Leuven Department of Pharmaceutical and Pharmacological Sciences.
↵
This article has supplemental material available at dmd.aspetjournals.org.
Abbreviations
- ABC
- ATP-binding cassette
- ABT
- 1-aminobenzotriazole
- ACN
- acetonitrile
- ATV
- atazanavir
- DMSO
- dimethylsulfoxide
- FBS
- fetal bovine serum
- Fumic
- unbound fraction in microsomes
- fuhep
- unbound fraction in microsomes
- fuIPRL
- unbound fraction in IPRL perfusate
- fuplasma
- unbound fraction in plasma
- HBSS
- Hanks’ balanced salt solution
- HPGL
- hepatocytes per gram liver
- IDV
- indinavir
- IPRL
- isolated perfused rat liver
- Kpu,u
- ratio of the intracellular to extracellular unbound concentration
- KHB
- Krebs-Henseleit buffer
- LC-MS/MS
- chromatography–tandem mass spectrometry
- LOS
- losartan
- MK-571
- 3-[[3-[(E)-2-(7-Chloroquinolin-2-yl)ethenyl]phenyl]-[3-(dimethylamino)-3-oxopropyl]sulfanylmethyl]sulfanylpropanoate sodium salt
- MeOH
- methanol
- MPPGL
- microsomal protein per gram liver
- MPPMC
- microsomal protein per million cells
- PBS
- phosphate-buffered saline
- RLM
- rat liver microsomes
- SF
- scaling factors
- SLC
- solute carrier
- SRH
- suspended rat hepatocytes
- Copyright © 2016 by The American Society for Pharmacology and Experimental Therapeutics