Abstract
Current challenges with the in vitro–in vivo extrapolation (IVIVE) of hepatic uptake clearance involving organic anion–transporting polypeptide (OATP) 1B1/1B3 hinder drug design strategies. Here we evaluated the effect of 100% human plasma on the uptake clearance using transfected human embryonic kidney (HEK) 293 cells and primary human hepatocytes and assessed IVIVE. Apparent unbound uptake clearance (PSinf,u) increased significantly (P < 0.05) in the presence of plasma (vs. buffer incubations) for about 50% of compounds in both OATP1B1-transfected and wild-type HEK cells. Thus, plasma showed a minimal effect on the uptake ratios. With cultured human hepatocytes, plasma significantly (P < 0.05) increased PSinf,u for 11 of 19 OATP1B substrates (median change of 2.1-fold). Cell accumulation in HEK cells and hepatocytes was also increased for tolbutamide, which is not an OATP substrate. Plasma-to-buffer ratio of PSinf,u obtained in hepatocytes showed a good correlation with unbound fraction in plasma, and the relationship was best described by a “facilitated-dissociation” model. IVIVE was evaluated for the 19 OATP1B substrates using hepatocyte data in the presence of buffer and plasma. PSinf,u from buffer incubations markedly underpredicted hepatic intrinsic clearance (calculated via well stirred and parallel tube models) with an estimated bias of 0.10–0.13. Predictions improved when using PSinf,u from plasma incubations; however, considerable systemic underprediction was still apparent (0.19–0.26 bias). Plasma data with a global scaling factor of 3.8–5.3 showed good prediction accuracy (95% predictions within 3-fold; average fold error = 1.7, bias = 1). In summary, this study offers insight into the effect of plasma on the uptake clearance and its scope in improving IVIVE.
SIGNIFICANCE STATEMENT Our study using diverse anionic compounds shows that human plasma facilitates organic anion–transporting polypeptide 1B–mediated as well as passive uptake clearance, particularly for the highly bound compounds. Leveraging data from transfected human embryonic kidney 293 cells and primary human hepatocytes, we further evaluated mechanisms involved in the observed plasma-facilitated uptake transport. Enhanced hepatic uptake rate in the presence of plasma could be of relevance, as such mechanisms likely prevail in vivo and emphasize the need to maintain physiologically relevant assay conditions to achieve improved translation of transport data.
Introduction
Organic anion–transporting polypeptide (OATP) 1B1 and OATP1B3 are often involved in the hepatic uptake of high-molecular-weight acids and zwitterions irrespective of their passive permeability [i.e., extended clearance classification system (ECCS) 1B/3B compounds] (Varma et al., 2015, 2017a). However, once cleared from the blood compartment to liver, highly permeable OATP substrates (class 1B) are metabolized and excreted from the body as phase I and/or phase II metabolites, whereas low-permeability (class 3B) OATP substrates are predominantly eliminated unchanged in the bile (Kimoto et al., 2017; Varma et al., 2017c; Alluri et al., 2020). Although considerable progress has been made in the mechanistic characterization of the various disposition pathways, accurate prediction of hepatic clearance via transporters from in vitro tools is still challenging (Jones et al., 2012; Ménochet et al., 2012; Zamek-Gliszczynski et al., 2013; Li et al., 2014a). The noted disconnect in the in vitro–in vivo extrapolation (IVIVE) was speculated to be at least partly due to the expression/activity differences in the in vitro reagents, such as primary human hepatocytes and/or suboptimal assay conditions. It is noteworthy that quantitative proteomic studies suggest that expression levels of OATPs in the in vitro systems (e.g., primary human hepatocytes) are close to those in liver samples (Kimoto et al., 2012; Prasad et al., 2014). More recently, Kumar et al. (2019) studied total abundance and plasma membrane abundance of drug transporters and suggested lack of differences in the expression of uptake transporters (including OATPs) across hepatocyte models and human liver tissue, confirming transporter abundance cannot explain the IVIVE disconnect per se. Empirical scaling factors are suggested to offset underprediction of clearance and enable subsequent pharmacokinetic modeling and simulations (Jones et al., 2012; Ménochet et al., 2012; Varma et al., 2012; Li et al., 2014b).
Recently, a few studies have shown that human hepatocytes can better predict in vivo clearance of OATP substrates in the presence of bovine or human serum albumin and human plasma (Poulin and Haddad, 2015; Fukuchi et al., 2017; Kim et al., 2019; Riccardi et al., 2019). The enhancement in the uptake clearance demonstrated in some of these studies was suggested to be due to a phenomenon described as “albumin-mediated transport,” wherein the protein-bound drug presents more efficient uptake rate than the unbound drug (Poulin and Haddad, 2015; Poulin et al., 2016; Bowman and Benet, 2018; Bteich et al., 2019). Very early studies postulated mechanisms primarily involving enhanced dissociation of the drug from the albumin-drug complex at the cell surface after interaction of the complex with the hepatocyte membrane. This led to proposed variations of “facilitated-dissociation” models to describe the experimental data (Forker and Luxon, 1981, 1983; Tsao et al., 1988). Recent studies by Miyauchi et al. (2018) and Kim et al. (2019) further substantiated the facilitated-dissociation hypothesis. To account for the albumin-facilitated uptake in the IVIVE, Poulin and colleagues (Poulin et al., 2012, 2016; Poulin and Haddad, 2018) proposed an “fraction unbound in plasma (fu,p)-adjusted” model, wherein unbound fraction at the surface of hepatocytes is estimated based on the plasma-to-liver concentration ratio of albumin in vivo. Other potential mechanisms involved in the albumin-mediated uptake are extensively reviewed recently (Bowman and Benet, 2018; Bteich et al., 2019). Of note is a hypothesis described as “the transporter-induced protein-binding shift,” which relies on the idea that transporter proteins can directly strip the drug from the plasma protein if the binding affinity is relatively higher toward the transporter versus plasma proteins (Baik and Huang, 2015; Bowman et al., 2019). Despite various efforts described above, the quantitative effect of albumin or plasma and the underlying mechanisms are not completely understood. Use of 100% human plasma may provide physiologically relevant conditions for in vitro experiments and enable improved IVIVE and, arguably, reduce the uncertainty associated with unbound fraction measurements for highly bound compounds.
The main objective of this investigation was to evaluate the effect of plasma on the in vitro uptake clearance of OATP1B1/1B3 substrate drugs and further characterize the mechanism involved therein. For a set of about 21 compounds (ECCS class 1B/3B/1A/2), which included 19 OATP substrates, in vitro uptake clearance was measured in the buffer and 100% plasma incubations using OATP1B1-transfected and wild-type HEK293 cells as well as primary human hepatocytes. Moreover, IVIVE was evaluated to quantitatively predict human hepatic clearance using hepatocyte uptake data obtained on the plasma incubations.
Materials and Methods
Chemicals and Reagents.
Atorvastatin, bosentan, cerivastatin, fluvastatin, pitavastatin, pravastatin, and rosuvastatin were purchased from Sequoia Research Products Ltd. (Oxford, UK). Glyburide and rifamycin SV were purchased from Sigma-Aldrich (St. Louis, MO). All other compounds were obtained from the Pfizer Chemical inventory system. Pooled human plasma [lot HMN167544, prepared using whole-blood donations of 13 donors (seven males and six females)] was purchased from Celsis IVT (Baltimore, MD). InVitroGro-HT, CP, and HI hepatocyte media were purchased from Celsis IVT. Collagen I–coated 24-well plates were obtained from BD Biosciences (Franklin Lakes, NJ). Cryopreserved human hepatocytes Hu8246 (female, Caucasian, 37 years old) were purchased from Thermo Fisher Scientific (Carlsbad, CA). Cryopreserved human hepatocyte lots HH1025 (female, Caucasian, 59 years old) and HH1062 (female, Caucasian, 33 years old) were purchased from In Vitro ADMET Laboratories (Malden, MA). BCA protein assay kit was purchased from PIERCE (Rockford, IL). NP-40 protein lysis buffer was purchased from Thermo-Fisher (Franklin, MA). Human embryonic kidney (HEK) 293 cells stably transfected with human OATP1B1 were generated at Pfizer Inc. (Sandwich, UK). Dulbecco’s modified Eagle’s medium, FBS, nonessential amino acids, GlutaMAX-1, sodium pyruvate, penicillin, and streptomycin solution were obtained from Invitrogen.
In Vitro Transport Studies Using Transporter-Transfected Cells.
HEK293 cells, both wild-type (WT) and stably transfected with OATP1B1, were seeded at a density of 0.3 × 106 cells/well on BioCoat 24-well poly-D-lysine–coated plates (Corning Inc., Corning NY) and grown in Dulbecco’s modified Eagle’s medium containing 10% FBS and 1% sodium pyruvate for 48 hours at 37°C, 90% relative humidity, and 5% CO2. OATP1B1-HEK293 cells were supplemented with nonessential amino acids and GlutaMAX.
Pooled plasma was stored at −20°C and thawed at room temperature before the in vitro studies. For uptake studies, the test compounds were equilibrated with 100% pooled human plasma for 30 minutes prior to incubation. OATP1B1-HEK293 and HEK293-WT cells were rinsed two times with 0.5 ml warm uptake buffer (HBSS, pH 7.4), and this was followed by preincubation for 10 minutes without (i.e., buffer condition) or with 100% human plasma (referred as plasma condition). Uptake study was initiated by adding test compounds in triplicates without or with 100% plasma (0.5 ml/well). Substrate concentrations for without and with plasma were 0.5 and 3 µM, respectively. Cellular uptake was terminated at designed time points (0.5, 2, 5, and 10 minutes) by washing the cells three times with ice-cold HBSS buffer. The samples were lysed with 0.5 ml of methanol containing internal standard and quantified by liquid chromatography tandem mass spectrometry (LC-MS/MS) system. The total cellular protein content was determined by using a Pierce BCA protein assay kit according to the manufacturer’s specifications. Uptake rates were estimated from the initial time course (typically, 0.5–2 minutes) by linear regression (Bi et al., 2017, 2019; Bergman et al., 2019) assuming negligible metabolic clearance and efflux. For each condition, mean and %CV are presented from three independent experiments, with each run in triplicate.
Uptake Studies in Cultured Primary Human Hepatocytes.
The hepatic uptake assay was performed using short-term culture primary human hepatocytes as described previously with some modification (Bi et al., 2017, 2019). Briefly, cryopreserved hepatocytes were thawed at 37°C in InVitroGro-HT media and seeded into 24-well collagen I–coated plates with 0.35 × 106 cells/well in a volume of 0.5 ml/well in InVitroGro-CP media. The cells were cultured in the InVitro-CP media overnight (∼18 hours). The cell culture of either was preincubated for 10 minutes at 37°C with protein-free HBSS or 100% pooled human plasma. In case of inhibition studies, rifamycin SV was added at 1 mM in buffer (plasma-free) incubations and 3 mM in plasma incubations. The preincubation buffer or plasma was aspirated, and the uptake and inhibition reaction were initiated by addition of prewarmed buffer or plasma (0.5 ml/well) containing test compounds with or without rifamycin SV. The reactions were terminated at designated time points (0.5, 1, 2, 5 minutes) by adding ice-cold HBSS immediately after removal of the incubation buffer. The cells were washed three times with ice-cold HBSS and lysed with 100% methanol containing internal standard, and the samples were analyzed by LC-MS/MS. Unless mentioned otherwise, mean and %CV are presented from two to five independent experiments, with each run in triplicate. Cellular accumulation was presented as cell-to–unbound media ratio (C/M ratio) (microliter per milligram-protein, eq. 1). Apparent unbound uptake clearance (PSinf,u) (microliter per minute per milligram-protein, eq. 2) was estimated from the initial time course (typically, 0.5–2 minutes) by linear regression (Bi et al., 2017, 2018, 2019) assuming negligible metabolic clearance and efflux.
(1)
(2)wherein fu is fraction unbound in the incubation media. fu was assumed to be unity for buffer incubations and fu,p for plasma incubations.
LC-MS/MS Method.
LC-MS/MS analyses were performed on an SCIEX Triple Quad 6500+ mass spectrometer (SCIEX, ON, Canada) equipped with TurboIonSpray interface. The high-pressure liquid chromatography systems consisted of an Agilent 1290 Infinity binary pump and ADDA autosampler (Agilent Technologies, Santa Clara, CA). All instruments were controlled and synchronized by SCIEX Analyst software (version 1.6.2) working in tandem with the ADDA software. Mobile phases were 0.1% formic acid in water (mobile phase A) and 0.1% formic acid in acetonitrile (mobile phase B). The gradient was maintained at 5% B for 0.2 minutes, followed by a linear increase to 95% B in 0.7 minutes, and kept at 95% B for 1.0 minute before a linear decrease to 5% in 1.02 minutes. The column was equilibrated at 5% B for 0.5 minutes. The total run time for each injection was 1.5 minutes. The chromatographic separation was carried out on a Phenomenex Kinetex C18 100 Å 30 × 2.1–mm column with a flow rate of 0.8 ml/min. The injection volume was 10 µl. A Phenomenex Kinetex C18 (2.6 µm, 2.1 × 30 mm) column with a C18 guard column was used. The following gradient was used to elute samples: 5% solvent B for 0.2 minutes increased to 95% B for 0.5 minutes, held at 95% B for 0.3 minutes, reduced to 5% B over 0.02 minutes, and held at 5% solvent B for 0.48 minutes (total run time 1.5 minutes). For mass spectrometer, the TurboIonSpray interface was operated in the positive/negative switching ion mode at 5000/−4500 V and 600°C. Quadrupoles Q1 and Q3 were set on unit resolution. Multiple-reaction-monitoring mode using specific precursor/product ion transitions was used for quantification. The assay sensitivity is ≥0.0001 µM, and the assay accuracy is at least within +/−20% at the lowest concentration assessed for all the compounds. Linear regression was fitted to data of standard solutions using 1/Y2 weighting. Data processing was performed using MultiQuant software (version 3.0.2; Sciex).
In Vitro–In Vivo Extrapolation of Hepatic Uptake Clearance.
Assuming hepatic uptake as the rate-determining step in the overall hepatic clearance of the OATP1B substrate drugs evaluated, IVIVE was evaluated using the in vitro PSinf,u obtained after buffer and plasma incubations. Standard physiologic scalars were used in this process: 1 × 106 hepatocytes/mg–measured protein, 120 × 106 hepatocytes/g-liver, and 21 g-liver/kg-body weight (Davies and Morris, 1993; Hosea et al., 2009). Hepatic plasma clearance (CLh) was calculated by using both well stirred liver model (eq. 3) and parallel tube model (eq. 4).
(3)
(4)Rbp is blood-to-plasma ratio, fu,p is plasma protein binding, and Qh is hepatic blood flow (20 ml/min per kilogram). Similarly, in vivo (observed) intrinsic hepatic clearance (CLint,h) was calculated using well stirred clearance model (eq. 5) and parallel tube model (eq. 6).
(5)
(6)Prediction accuracy and precision were assessed with absolute average fold error (AAFE) (eq. 7) and bias (eq. 8).
(7)
(8)N is the number of observations.
“Facilitated-Dissociation” Model and Plasma Effect on Uptake Clearance.
Change in unbound uptake clearance in the plasma versus buffer incubations was assessed using a “facilitated-dissociation” model, which is based on hypothesis that ligand-albumin complex interacts with the cell membrane resulting in enhanced dissociation of the ligand in the vicinity of cell surface (Bteich et al., 2019). Model previously proposed (Forker and Luxon, 1983; Tsao et al., 1988; Miyauchi et al., 2018) was adopted to describe the ratio (R) of apparent unbound uptake clearance in the presence (PSinf,u,plasma) and absence (PSinf,u,buffer) of human plasma.
(9)in which Bmax represents the capacity of albumin binding sites on the surface of hepatocytes, and Kd,m is the dissociation constant of the bound albumin from the cell surface. The parameter “r” represents a ratio of uptake clearance of the unbound ligand dissociated from the ligand-albumin complex at cell surface to the uptake clearance of the unbound ligand dissociated in the plasma away from cell surface. For data fitting, “r · Bmax” was considered as a hybrid constant because they were individually unidentifiable. Albumin concentration ([Alb]) in the plasma was assumed to be about 680 µM (Peters, 1996). Kd,m was initially fixed to 45.2 µM, a value previously reported by Kim et al. (2019). We refer to the term “r.Bmax/(Kd,m+[Alb])” in eq. 9 as “albumin-mediated uptake factor” (AMUF).
“fu,p-Adjusted” Model and Plasma Effect on Uptake Clearance.
Change in unbound uptake clearance in the plasma versus buffer incubations was also assessed using a “fu,p-adjusted” model proposed by Poulin et al. (2012); Poulin and Haddad (2018). This model suggests need for adjustment in the fu,p to account for pH gradient in the medium (extracellular, pH 7.4) and hepatocytes (intracellular, pH 7.1) as well as the albumin-mediated hepatic uptake occurring in the presence of plasma. Here, R value is estimated based on the following expressions:
(10)
(11)in which PLR refers to the plasma-to-liver concentration ratio of albumin in human. A value of 13.3, as suggested by Poulin et al. (2012) and Poulin and Haddad (2018), was used for this parameter. The parameters funionized,pH7.4 and funionized,pH7.1 are the corresponding fraction of unionized drugs in the plasma (pH 7.4) and hepatocyte intracellular water (pH 7.1). Because initial intrinsic uptake clearance was determined assuming negligible intracellular passive permeability clearance back to the extracellular, the pH gradient funionized ratio was not applicable, thus simplifying to eq. 12:
(12)
Results
Effect of Human Plasma on the In Vitro Uptake in OATP1B1-Transfected and Wild-Type HEK293 Cells.
Time course of cell accumulation was measured in OATP1B1-transfected and wild-type HEK293 cells for 10 OATP1B substrates (Fig. 1). Additionally, midazolam and tolbutamide were also included as controls with no OATP1B1 substrate activity. C/M ratio, a ratio of amount in cell to unbound drug in media, increased considerably in the presence of 100% human plasma compared with incubations with buffer for several OATP1B1 substrates, including atorvastatin, cerivastatin, entacapone, fluvastatin, pitavastatin, repaglinide, and telmisartan. In comparison, the effect of plasma on C/M ratio was minimal for bosentan, pravastatin, and rosuvastatin. For further analysis, PSinf,u values were estimated based on the initial rate, and uptake ratios (i.e., ratio of uptake PSinf,u in OATP1B1 cells to wild-type cells) were derived (Fig. 1; Table 1). Compared with the no-plasma (buffer) condition, the presence of plasma increased PSinf,u by more than 2-fold for atorvastatin, entacapone, repaglinide, telmisartan, and pitavastatin in the OATP1B1-HEK293 cells. Interestingly, a similar effect was also seen for these compounds in wild-type cells. Consequently, uptake ratios (OATP1B1/wild-type) were not affected by plasma for OATP1B1 substrates. Interestingly, the effect of plasma on tolbutamide (not a substrate to OATP1B1) was noted to be high in both OATP1B1-transfected and wild-type cells (about 5-fold). However, no clear trend was seen with C/M ratio for midazolam (Fig. 1). Our previous studies with buffer incubations implied that the uptake clearance of probe substrates in the OATP1B1-transfected HEK293 cells in the presence of 100 µM rifampicin is similar to the uptake in HEK293-WT cells, suggesting that the passive uptake clearance is similar between these cells (Bi et al., 2019).
Effect of plasma on the uptake in HEK293 wild-type and OATP1B1-transfected cells of 12 drugs. Time course (representative) of cell-to-media ratio is depicted on the left-side plots, and the estimated intrinsic clearance (CLint) and uptake ratio (OATP1B1/WT cells) are presented in the right-side plots for each compound. N = 3, for each data point. Data variability for intrinsic clearance is captured in Table 1.
Effect of plasma on the uptake clearance in wild-type and OATP1B1-transfected HEK293 cells
Statistical significance of plasma-induced increase in uptake clearance was assessed using unpaired t test (*P < 0.05; **P < 0.01; ***P < 0.001).
Effect of Human Plasma on the In Vitro Uptake by Primary Human Hepatocytes.
Uptake of 21 compounds, which included 19 OATP1B substrates, midazolam, and tolbutamide, was measured in cultured human hepatocytes without and with rifamycin SV in buffer and plasma incubations. Consistent with HEK293 cell data (Fig. 1), the C/M ratio of several drugs, including atorvastatin, entacapone, fluvastatin, pitavastatin, telmisartan, and tolbutamide, was higher with plasma incubations compared with uptake studies in buffer (Fig. 2). Presence of plasma increased PSinf,u by more than 2-fold for 11 of 19 (∼58%) OATP1B substrates, although only 4 (∼21%) substrates (montelukast, PF-05089771, PF-05241328, and pitavastatin) showed a more than 3-fold enhancement. Montelukast showed highest effect (∼16-fold). Rifamycin SV, an OATP1B inhibitor, significantly reduced the PSinf,u of most of the OATP1B substrates in both buffer and plasma incubations, with the exception of GSK269984A, montelukast, repaglinide, and telmisartan (Fig. 2; Table 2). Consistent with the observations in HEK293 cells, plasma markedly increased C/M ratio of tolbutamide, which is not a substrate to OATP1B1/1B3 but is transported by OAT2. However, a reliable estimate for PSinf,u could not be made for tolbutamide and midazolam because of flat C/M ratio–time profiles in the presence of plasma (Fig. 2).
Effect of human plasma on the uptake in primary human hepatocytes (lot Hu8246) of 12 representative drugs. Time course (representative) of cell-to-media ratio is depicted on the left-side plots, and the estimated intrinsic clearance (CLint) and percent active uptake are presented in the right-side plots for each compound. N = 3–6, for each data point. Data variability for intrinsic clearance is captured in Table 2. Rif SV, rifamycin SV.
Summary of in vitro uptake clearance data in plated human hepatocytes (lot Hu8246) and other parameters of 19 OATP1B1 substrates and midazolam and tolbutamide
fu,p and blood-to-plasma ratio (Rbp) were measured as previously described (Riccardi et al., 2019). Statistical significance of plasma-induced increase in uptake clearance was assessed using unpaired t test (*P < 0.05; **P < 0.01; ***P < 0.001).
In general, there was a good correlation for uptake ratios measured in HEK293 cells (OATP1B1/wild-type ratio) between buffer and plasma incubations (Fig. 3A). Apart from entacapone and repaglinide, uptake ratios noted in plasma incubations were within 50% of uptake ratios in buffer (Table 1), suggesting that plasma affected both OATP1B1-mediated transport and passive transport to a similar extent. For the 12 compounds evaluated in both HEK293 cells and human hepatocytes, the effect size of plasma on the total uptake clearance was well within 2-fold between the two systems (Fig. 3B). Moreover, there is a good correlation between buffer and plasma conditions for percent active uptake in human hepatocytes—this was estimated assuming rifamycin SV completely inhibited OATP1B (Fig. 3C). Here the values were generally within ±20% error with the exception of cerivastatin, PF-05241328, and repaglinide, which showed higher increase in percent active uptake in plasma incubations, whereas PF-05089771 showed opposite trend.
Correlations between uptake in transfected cells and plated human hepatocytes (lot Hu8246) without and with plasma. (A) Uptake ratio in transfected cells (ratio of uptake in OATP1B1 and wild-type cells) in buffer vs. plasma incubations. (B) Plasma effect on total uptake clearance in OATP1B1-transfected cells vs. plated human hepatocytes. (C) Percent active uptake in plated human hepatocytes in buffer vs. plasma.
For six representative OATP1B1 substrates (fluvastatin, glyburide, montelukast, pitavastatin, PF-05241328, and rosuvastatin), which showed plasma/buffer ratio of PSinf,u in human hepatocytes (R value) in the range of unity to >15 using the hepatocyte lot Hu8246 (Table 2), we measured PSinf,u in the buffer and plasma incubations in two additional cryopreserved human hepatocyte lots (HH1025 and HH1062). Data summarized in Table 3 generally suggested similarity in the uptake clearance, and the plasma-induced enhancement ratio (R value) across the three lots is similar.
Comparison of in vitro uptake clearance in plated human hepatocytes across three different hepatocyte lots for a representative six OATP1B1 substrates
Statistical significance of plasma-induced increase in uptake clearance was assessed using unpaired t test (*P < 0.05; **P < 0.01; ***P < 0.001).
Finally, the plasma/buffer ratio (R value) was noted to decrease with increase in fu,p (Fig. 4). This inverse correlation for the 19 substrates was reasonably well described by the “facilitated-dissociation” model (eq. 9) using a previously reported Km,d of 45.2 µM (Kim et al., 2019) (root-mean-square error ∼0.45). The hybrid constant, r·Bmax, was estimated as 5.2 ± 1.1. Sensitivity analysis of Km,d (±10×) suggested minimal impact of this parameter on the overall goodness of fit (no change), and r·Bmax ranged from 5.0–8.3. Within this range of Km,d (4.5–452 µM), the AMUF in the facilitated-dissociation model (eq. 9) was found to be a constant (AMUF ∼0.0073). On the contrary, the R value estimated based on the “fu,p-adjusted” model (eq. 10) was considerably different from the observations (Fig. 4).
Correlation between fu,p and the plasma/buffer ratio of total uptake clearance in human hepatocytes. Data were fitted to facilitated-dissociation (FD) model (eq. 9); solid curve and shaded area represent facilitated-dissociation model fitting and 95% confidence interval. Triangles represent estimated ratio using fu,p-adjusted model (eq. 10). Data for all compounds were obtained using hepatocyte lot Hu8246 (circles); for six representative OATP1B1 substrates, data were also obtained in two additional hepatocyte lots (HH1025, squares; HH1062, diamonds).
In Vitro–In Vivo Extrapolation of Uptake Clearance to Predict Human Hepatic Clearance.
IVIVE was assessed for 19 OATP1B substrates, assuming that the overall hepatic clearance is primarily determined by the total uptake clearance (active and passive) (Fig. 5; Table 4). In vitro PSinf,u obtained from buffer incubations markedly underpredicted in vivo CLint,h, which was estimated using the well stirred model (0.10 bias). The PSinf,u obtained after incubation with plasma improved the predictions to a certain extent; however, considerable underpredictions (∼0.20 bias) are still apparent with only 26% within 3-fold of observed values. Nonetheless, the presence of plasma seems to reduce the skew in the IVIVE correlation, particularly for ECCS class 1B compounds. Consequently, an empirical scaling factor (ESF) was applied to scale up the in vitro PSinf,u from plasma incubations to the in vivo hepatic CLint,h. The resulting global ESF of 5.3 well recovered CLint,h for 15 of 18 (79%) drugs within 3-fold (Fig. 5C). In this case, total CLh was predicted within 2-fold for 84% of compounds (Table 3).
In vitro–in vivo extrapolation of CLint,h of 19 OATP1B1 substrate drugs, assuming hepatic uptake as rate-determining step. Well stirred (A–D) and parallel tube (E–H) models were employed in the IVIVE. Predictions are based on the total uptake measured using plated human hepatocyte incubations in buffer (A and C), buffer with a global scaling factor (B and F), human plasma (C and G), and human plasma with a global scaling factor (D and H). In vitro uptake data obtained using human hepatocyte lot Hu8246. Diagonal dashed, dotted, solid, and dotted-dash lines represent unity line, 3-fold error, bias, and skew, respectively. Circles, ECCS class 1B; squares, class 3B; triangle, class 1A compounds.
Statistical summary of IVIVE performance in predicting CLint,h and total CLh using in vitro total PSinf,u measured in cultured human hepatocytes (Hu8246) with buffer and plasma incubations
A somewhat smaller ESF (3.8) was required when employing the parallel tube model in the IVIVE exercise (Fig. 5, D–F). Compared with the well stirred model, the parallel tube model tended to show improved predictability for CLint,h, although no clear difference was noted with CLh predictions (Table 4).
Discussion
The present study evaluating the effect of plasma on uptake transport of 21 compounds using HEK293 cells (OATP1B1-transfected and wild-type cells) and cultured primary human hepatocytes suggested that: 1) the presence of plasma in the in vitro incubations may increase the PSinf,u of OATP1B substrates; 2) changes in the uptake ratio in HEK293 cells and % active uptake in hepatocytes are, however, minimal; 3) effect of plasma on PSinf,u is comparable across wild-type and OATP1B1-HEK293 cells and the primary human hepatocytes; 4) plasma also enhances uptake of non-OATP substrates like tolbutamide in both cell types; and 5) magnitude of plasma-induced enhancement in cultured human hepatocytes varies across the OATP1B substrates. However, a good correlation with fu,p is evident. These findings collectively suggest that plasma enhances both OATP1B1-mediated and passive transport simultaneously and generally support the hypothesis postulating “facilitated dissociation” of anionic drugs from the plasma proteins, probably at the vicinity of the cell membrane, and thus provide higher unbound ligand at the cell surface for uptake (Fig. 6). The observed plasma-induced effect could be of relevance since such mechanisms may prevail in vivo; moreover, it emphasizes the need to maintain physiologically relevant assay conditions to achieve improved translation of in vitro transport data.
Schematic of potential mechanism involved in the enhanced active and passive uptake rate in the presence of plasma. According to the data-supported “facilitated-dissociation” hypothesis, bound ligand to plasma proteins is in equilibrium with unbound ligand in the bulk plasma, which is similar to the experimental fu,p. However, plasma proteins (likely albumin) may interact with cell membrane (or membrane receptors/proteins), possibly leading to conformational changes in plasma protein and shift in unbound fraction in the vicinity of the cell membrane (fu,surface). It is postulated that the intrinsic (active and passive) transport activity across the cell membrane does not change in the presence of plasma/albumin; however, the unbound concentration of ligand at the cell surface is higher than the unbound concentration at the cell surface for highly bound ligands, which acts as a driving force for apparent plasma-induced enhancement. Consequently, both active and passive uptake rates are driven by the unbound ligand, which is determined by a combination of fu,p and fu,surface instead of fu,p alone. Drug-independent factors, such as affinity of albumin to cell surface (Kd,m) and the capacity of albumin binding sites on the surface of hepatocytes, also contribute to the observed plasma-induced enhancement of active and passive uptake.
The PSinf,u of a diverse set of 19 OATP1B substrates—spanning ECCS classes 1B (n = 14), 3B (n = 4), and 1A (n = 1)—were determined in the cultured human hepatocytes with buffer and plasma incubations (Table 2). The presence of plasma enhanced PSinf,u by more than 2-fold for about 60% of the compounds, although the enhancement ratio was >3 for only four compounds (∼20%). Notably, the four compounds (montelukast, PF-05089771, PF-05241328, and pitavastatin) are highly bound with measured fu,p < 0.003, whereas compounds with limited plasma-induced effect (R value < 2) are relatively less bound (fu,p generally above 0.03) (Fig. 4). The relationship between plasma enhancement ratio and fu,p was best described by a “facilitated-dissociation” model (eq. 9) [previously proposed by Forker and Luxon (1983) and Tsao et al. (1988); Miyauchi et al. (2018)], which assumes enhanced dissociation of ligand from the albumin-ligand complex when the complex interacts with the surface (membrane or receptor/protein) of the hepatocytes, resulting in a higher unbound drug for uptake. In theory, there are certain differences between the models proposed by Forker and Luxon (1983) and Tsao et al. (1988); however, given that our studies were run at a single concentration of albumin (i.e., ∼680 µM albumin in plasma), the present data set cannot distinguish between the two models. It is noteworthy that these two models are statistically indistinguishable even when ligand uptake was studied at multiple albumin concentrations (Miyauchi et al., 2018). Recently, Miyauchi et al. (2018) studied the concentration-dependent effect of albumin on hepatocyte uptake of pitavastatin and estimated Kd,m (i.e., dissociation constant of the bound albumin from hepatocyte surface) to be about 200 µM. Subsequent studies by Kim et al. (2019) estimated the parameter to be ∼45 µM based on simultaneous fitting of albumin concentration–dependent data of 10 OATP1B substrates. We adopted 45 µM as an initial value for Kd,m in fitting fu,p versus the plasma enhancement ratio (Fig. 4). However, the model fitting to the current data set was not sensitive to this parameter, implying that at physiologically relevant albumin concentration (i.e., in human plasma) association/dissociation of complex from cell surface is not rate-limiting to overall uptake (Fig. 6). The plasma-induced enhancement in apparent uptake rate was well described while assuming a constant value for r·Bmax, although in theory this parameter is considered substrate-dependent (Miyauchi et al., 2018). We further analyzed the hepatocyte uptake data using a “fu,p-adjusted” model (eq. 10), which was suggested to account for the albumin-mediated uptake assuming a similar facilitated-dissociation hypothesis (Poulin et al., 2012; Poulin and Haddad, 2018). Clearly the fu,p-adjusted model did not recover the observed R value (Fig. 4) and in almost all cases overestimated the plasma effect in vitro for the anionic compounds assessed here. Finally, the relationship established here with a diverse set of ECCS class 1B/3B compounds can be used to prospectively predict the potential for plasma-induced uptake enhancement based on fu,p values. With this data set, AMUF was found to be a constant (∼0.0073) irrespective of the Kd,m input. Therefore, plasma enhancement ratio (R value) for a given compound can be reasonably predicted using experimental fu,p alone, which can subsequently inform the need to conduct uptake studies in the presence of plasma for IVIVE purposes.
Recent reports have suggested “transporter-induced protein-binding shift” as a possible mechanism for enhanced uptake on the basis that compounds with high affinity for OATP1B1 show higher plasma- or albumin-mediated uptake (Baik and Huang, 2015; Bowman and Benet, 2018; Bowman et al., 2019). Our results with HEK cells and hepatocytes do not substantiate this hypothesis. Notably, plasma enhancement ratios were similar between OATP1B1-transfected and wild-type HEK cells, implying plasma enhanced both active and passive transport (Fig. 3). Similar inference could be drawn with no major difference in % active uptake in hepatocytes in plasma versus buffer incubations. Plasma-induced increase in passive uptake (about 4–10 fold for pitavastatin and atorvastatin) was also noted by Bowman et al. (2019) and, as implied in their report, cannot be rationalized based on the transporter-induced protein-binding shift mechanism when the unbound passive uptake clearance is also influenced by the presence of plasma protein. Moreover, we also studied the effect of plasma on the substrate concentration–dependent uptake in cultured human hepatocytes to evaluate a potential shift in unbound OATP1B affinity (Km,u) (Supplemental Fig. 1; Supplemental Table 1). Of the eight OATP1B substrates studied, only pravastatin showed a minimal change in unbound OATP1B affinity (Km,u) and Vmax, whereas reliable kinetic parameters could not be obtained for six substrates since the data fitting produced large errors in parameter estimation. In these cases, the measured total uptake was statistically not separable from the model-estimated passive uptake in the unbound concentration range (mostly below 2 µM) studied in the plasma incubations. These data are somewhat similar to those previously reported for the highly bound drugs atorvastatin and pitavastatin (Bowman et al., 2019). We believe that plasma incubations for such compounds need to be studied at much higher unbound concentrations to achieve a dynamic range for estimating reliable active and passive transport kinetics. Ideally, the higher end of unbound concentration in the plasma incubations should be above the affinity constant (Km) obtained in plasma-free buffer to make a reliable assessment of plasma-induced affinity shift. However, obtaining transport kinetics at higher unbound concentrations was hindered by high plasma binding. Moreover, plasma protein binding could be saturated at higher total drug concentrations, which may further alter the kinetics of albumin-mediated uptake.
Similar effect of plasma across three cell types (OATP1B1-HEK293 cells, wild-type HEK293 cells, and primary human hepatocytes) suggests the involvement of a plasma-induced uptake mechanism for anionic drugs across basolateral membranes of not only liver but also other disposition organs (like kidney, muscle, etc.). We also studied midazolam (a nonacid drug), which showed no clear differences in the plasma versus buffer incubations. Further studies with basic and neutral drugs are currently underway in our laboratory to investigate whether such mechanisms contribute to the hepatic uptake of nonacids.
IVIVE to predict hepatic clearance was assessed for the 19 OATP substrates considering the well stirred and parallel tube models (Fig. 5, Table 4). In line with the previous reports, hepatic clearance was considerably underpredicted using PSinf,u obtained from buffer incubations (Watanabe et al., 2009; Jones et al., 2012; Ménochet et al., 2012; Varma et al., 2014; Kimoto et al., 2017). Indeed, the bias of about 0.1 in the IVIVE indicates about 10-fold underprediction, which is similar to our earlier observation for 10 OATP substrates and using three different hepatocyte lots (Varma et al., 2014). The PSinf,u values from plasma incubations improved the IVIVE, although the bias is well below unity (∼0.2–0.25), consequently requiring a global ESF of about 3.8–5.3 to recover in vivo clearance. A limitation of our study is the use of single hepatocyte batch for IVIVE exercise, which may potentially deviate from population mean. However, our liquid chromatography/mass spectrometry–based quantitative proteomic studies suggest that expression of OATP1B1 in the current hepatocyte lot is within 30% of the geometric mean of expression levels measured across 52 individual healthy liver samples (Supplemental Table 2). Moreover, for six compounds with plasma-induced enhancement ratio (R value) ranging from unity (rosuvastatin) to >15 (montelukast), we found that PSinf,u in buffer and plasma incubations were comparable across the three different hepatocyte lots (Table 3). Additionally, recent studies by Kim et al. (2019) suggested the need for similar ESF to best describe in vivo clearance based on the PSinf,u measured in the presence of 5% human serum albumin. Overall, the IVIVE disconnect cannot be readily explained. Nonetheless, in vitro studies using plasma incubations offer a better approach over the existing methodologies for predicting uptake clearance. Notably, plasma reduced the skew in the predictions, particularly for ECCS class 1B compounds, and thus can enable prospective predictions based on a global ESF without the need for previously contended “compound-specific” scalars (Jones et al., 2012; Ménochet et al., 2012; Li et al., 2014a).
Conclusions
This study shows plasma-induced enhancement in the transporter-mediated and passive uptake clearance for several OATP1B substrates and generally supports the hypothesis of “facilitated dissociation” of drug from the albumin-drug complex at the cell surface as a possible mechanism for the observed effect. On the basis of this mechanism, the plasma enhancement effect observed in cultured human hepatocytes correlated well with the fu,p. Collectively, in vivo hepatic uptake can exceed uptake projected based on conventional free-drug hypothesis for highly bound anionic drugs. Leveraging the current data set, we propose that the plasma-induced uptake effect size can be predicted based on readily available fu,p and a derived constant (i.e., AMUF). A considerable improvement in the IVIVE of hepatic clearance was observed with the in vitro transporter data obtained with plasma incubations, although systemic underprediction was still apparent. Plasma data along with a relatively smaller global ESF therefore offer a pragmatic approach for prospective clearance prediction of high-molecular-weight acids and zwitterions (ECCS class 1B/3B).
Authorship Contributions
Participated in research design: Bi, Tess, Rodrigues, Varma.
Conducted experiments: Bi, Ryu.
Performed data analysis: Bi, Ryu, Tess, Varma.
Wrote or contributed to the writing of the manuscript: Bi, Ryu, Tess, Rodrigues, Varma.
Footnotes
- Received June 1, 2020.
- Accepted October 19, 2020.
No funding was received for the work reported here.
All authors are full-time employees of Pfizer Inc. The authors have no conflicts of interest that are directly relevant to this study.
↵
This article has supplemental material available at dmd.aspetjournals.org.
Abbreviations
- AAFE
- absolute average fold error
- AMUF
- albumin-mediated uptake factor
- CLh
- hepatic plasma clearance
- CLint,h
- intrinsic hepatic clearance
- C/M ratio
- cell-to–unbound media ratio
- ECCS
- extended clearance classification system
- ESF
- empirical scaling factor
- fu,p
- fraction unbound in plasma
- HBSS
- Hanks’ balanced salt solution
- HEK
- human embryonic kidney
- IVIVE
- in vitro–in vivo extrapolation
- Kd,m
- dissociation constant of the bound albumin from the cell surface
- LC-MS/MS
- liquid chromatography tandem mass spectrometry
- OATP
- organic anion–transporting polypeptide
- PSinf,u
- apparent unbound uptake clearance
- WT
- wild type
- Copyright © 2020 by The American Society for Pharmacology and Experimental Therapeutics
