Abstract
Drug depletion-time profiles in isolated hepatocytes, as well as microsomes, have become a standard method of assessing hepatic metabolic clearance in vitro. There is a previously described adaptation of the depletion approach to allow determination of hepatic uptake by transporters in addition to metabolism (Drug Metab Dispos 35:859–865, 2007). Dual incubations are performed where one set of incubations undergo conventional methodology, whereas for the second set, cells and media are separated for determination of drug loss from the media. The utility of this dual incubation approach has been assessed using eight drugs (atorvastatin, clarithromycin, erythromycin, fexofenadine, pitavastatin, repaglinide, rosuvastatin, and saquinavir) with a range of active uptake, passive permeability, cell binding, and metabolic characteristics. Four of these compounds (fexofenadine, rosuvastatin, pitavastatin, and atorvastatin) show a biphasic time profile when assessing drug loss from media indicative of hepatic uptake before elimination within the hepatocyte, which is distinct from the time profile in a conventional incubation, and show higher clearances. The four other compounds (clarithromycin, saquinavir, erythromycin, and repaglinide) show identical depletion-time profiles (and clearances) in both sets of incubations. Whether or not the biphasic nature (and higher clearance) is evident, indicating transporter activity for a particular drug, appears to be dependent on its passive permeability. Using the parameter Kpu to reflect the relative importance of hepatic transporters versus passive diffusion, a value of 10 was identified as a cutoff for whether the biphasic nature was evident; those compounds in excess of 10 show this characteristic clearly. There appears to be no relationship between the presence of the biphasic nature and any other parameter, including cellular binding, extent of metabolism, or the magnitude of active uptake.
Introduction
The use of isolated hepatocytes, either in suspension or as a monolayer, has allowed many aspects of qualitative and quantitative drug metabolism to be assessed (Hewitt et al., 2007; Soars et al., 2007a). Investigations performed as early as the mid 1970s (Inaba et al., 1975; Yih and van Rossum, 1977) demonstrated the use of the substrate depletion-time profile within an incubation of isolated hepatocytes to estimate clearance and to characterize the nonlinear nature of this process, in a way analogous to an in vivo pharmacokinetic study. The popularity of in vitro systems to make quantitative predictions of in vivo clearance has led to similar investigations using hepatic microsomes (Obach 2001; Riley et al., 2005). This substrate depletion approach is widely used because formal kinetic characterization and metabolite quantification are not required, allowing the rapid screening of compounds.
The basis of predicting in vivo clearance values from in vitro kinetic parameters is the use of Michaelis-Menten kinetics and experiential data obtained under initial rate conditions (Houston, 1994). For a number of drugs with simple metabolic pathways, it has been demonstrated that drug depletion and metabolite formation can provide equivalent clearance terms (Jones and Houston, 2004; Sjögren et al., 2009; Stringer et al., 2009). However, Jones and Houston (2004) have also indicated the importance of enzyme stability for longer incubation times because instability may generate biphasic depletion-time profiles. The drug depletion method assumes that the concentrations of drug in media and cells are the same, which may not be true because of transporter protein activity. The most common approach to measure hepatic uptake is analogous to the Michaelis-Menten approach, involving the measurement of cellular uptake under initial rate conditions after an “oil-spin” separation of cells from medium (Petzinger and Fückel, 1992; Ishigami et al., 1995). A range of substrate concentrations are usually investigated, and the kinetic parameters Vmax, Km, CLactive (Vmax/Km), and Pdiff are obtained (Yabe et al., 2011).
Soars et al. (2007b) have described an adaptation of the depletion approach to allow determination of hepatic uptake, in addition to hepatic metabolism. Dual incubations are performed where one set of incubations undergo conventional methodology (drug depletion assessed from samples of the incubation matrix of cells and buffer) and in the second set of incubations, the cells and media from samples of the incubation matrix are separated by centrifugation to determine drug loss from the media alone (media loss methodology). The second set of incubations yields a clearance reflecting total loss from the media (that is, both uptake and subsequent metabolism), whereas the conventional incubations provide a measure of metabolic (irreversible) clearance (see Fig. 1). By adopting this dual approach, Soars et al. (2007b) were able to demonstrate for 36 proprietary compounds that much closer (higher) estimates of in vivo clearance could be obtained than were achieved using the conventional approach. A similar improvement was also reported by Gardiner and Paine (2011) for seven marketed drugs. This experimental design is similar to that used previously by others for various investigations including intracellular binding of lipophilic amines and the impact of saturable events within the lysosomes (Hallifax and Houston, 2006). The methodology offers a more pragmatic, if arguably less robust, approach to identifying and characterizing hepatic uptake by transporters than the oil-spin method. As with other drug depletion approaches, clearance is derived from only one substrate concentration, and there are the usual assumptions regarding linearity (with time and cell number) to allow scaling of the CLobs parameter.
To test the general utility of the dual incubation approach to substrate depletion, eight drugs (atorvastatin, clarithromycin, erythromycin, fexofenadine, pitavastatin, repaglinide, rosuvastatin, and saquinavir) known to be organic anion transporter polypeptide (OATP) substrates (Shitara et al., 2006; Kalliokoski and Niemi 2009; Giacomini et al., 2010) have been selected. They show a range of active uptake, passive permeability, cellular binding, and metabolic characteristics. All eight compounds are actively transported into hepatocytes (range of CLact 100-fold); however, because of differences in passive permeability (range of Pdiff 40-fold), they show differing degrees of importance for active uptake (Yabe et al., 2011). These characteristics can be seen in Table 1, together with the intracellular binding metric fucell, which covers a range of 50-fold. These drugs have been investigated using a protocol suggested by Soars et al. (2007b) to obtain clearance values from the slopes of the depletion-time profiles. Three initial substrate concentrations are investigated to establish whether saturation can be identified. The data have also been analyzed by a simple pharmacokinetic model (see Fig. 2) to obtain an alternative set of parameters to describe hepatocellular events including uptake and metabolism.
Materials and Methods
Chemicals.
Atorvastatin calcium, pitavastatin calcium, rosuvastatin calcium, and saquinavir were obtained from Sequoia Research Products (Pangbourne, UK); erythromycin, clarithromycin, repaglinide, and fexofenadine were obtained from Sigma-Aldrich (Buchs, Switzerland).
Animal Source, Housing, and Diet.
Male Sprague-Dawley rats (240–260 g) were obtained from the Biological Sciences Unit, Medical School, University of Manchester (Manchester, UK). They were housed in groups of two to four, in opaque boxes on a bed of sawdust in rooms maintained at a temperature of 20 ± 3°C, with a relative humidity of 40 to 70% and a 12-h light/dark cycle. The animals were allowed free access to Chow Rat Mouse diet (Special Diet Services, Essex, UK) and fresh drinking water. All animal protocols were approved by the University of Manchester review committee.
Hepatocyte Studies.
Anesthetized rats were sacrificed by cervical dislocation, and hepatocytes were prepared using an adaptation of the collagenase perfusion method as described previously (Hayes et al., 1995). Hepatocyte viability was determined using the trypan blue exclusion test, and only those hepatocyte preparations with viabilities greater than 85% were used. All hepatocyte studies were performed using three independent hepatocyte preparations.
Depletion Studies in Rat Hepatocytes.
Conventional assay.
Drug diluted in Williams' E medium (125 μl, final concentration range, 0.1–10 μM) was preincubated for 5 min in an Eppendorf Thermomixer (Eppendorf UK Ltd., Stevenage, UK) (37°C, 900 rpm). To initiate the reaction, 125 μl of prewarmed (37°C) hepatocyte suspension (final concentration range, 0.1–1 × 106 cells/ml) was added, giving a final incubation volume of 250 μl. Experiments were performed in duplicate, and the organic solvent concentration (methanol) in the incubation was 0.5%. At 10 specified time points (up to 90 min), reactions were terminated by snap-freezing in liquid nitrogen. Samples were then thawed, and ice-cold acetonitrile containing an appropriate internal standard for liquid chromatography-tandem mass spectrometry (LC-MS/MS) analysis was added.
Media-loss assay.
The media-loss assay (based on the methods by Soars et al., 2007b) was performed in parallel to the conventional assay in each case (n = 3 for each drug). The only difference between the protocols was that at the specific time points (up to 90 min) in the media-loss assay, the incubation was centrifuged at 7000g for 30 s, and 80 μl of supernatant was snap-frozen in liquid nitrogen to terminate the reaction.
LC-MS/MS Analysis.
The LC-MS/MS system used consisted of a Waters 2790 with a Micromass Quattro Ultima triple quadruple mass spectrometer (Waters, Milford, MA). Hepatocyte samples were vortexed and centrifuged for 10 min at 11,600g (Eppendorf centrifuge 5413), and an aliquot (10 μl) of the supernatant was analyzed by LC-MS/MS. Conditions used have been described in detail by Yabe et al. (2011).
Data Analysis.
All depletion data were fitted to either a monoexponential decay model or a biexponential decay model as shown in eqs. 1 and 2: where C0 is the substrate concentration in the incubation media at time 0 and kdep is initial depletion rate constant. where A and B represent the back-extrapolated substrate concentration in the incubation media for the first and the second phase, respectively, and kdep and k2 are the initial depletion rate constants for the first phase and the second phase, respectively.
kdep values obtained for the three drug concentrations were used to determine the theoretical depletion constant at low substrate concentration (kdep ([S] → 0)) using eq. 3 (Obach and Reed-Hagen, 2002).
The observed in vitro clearance (CL) was determined using eq. 4, where V was the incubation volume and normalized for number of cells.
Data were also modeled by a two-compartment model (Fig. 2) to obtain specific clearance terms: CLuptake, CLmet, and Pdiff [reverse transport out of the cell was assumed to occur only by passive permeability because efflux transporters are known to be internalized after isolation (Bow et al., 2008)]. The relationships between the modeling parameters and the experimentally determined parameters are shown in Fig. 1.
For comparative purposes, in vitro estimates of the same parameters (CLuptake and Pdiff) obtained from the oil-spin method (Yabe et al., 2011) were used. In addition, the hepatocyte-to-medium partition coefficient for unbound drug (Kpu) and the fraction unbound in the hepatocyte (fucell) were calculated from eqs. 5 and 6, respectively (Yabe et al., 2011). where Kptotal is the ratio of the total cellular concentration relative to the external medium.
Kpu provides a measure of the cytosolic cellular concentration relative to the external medium and hence reflects the importance of active uptake. The parameter Kptotal reflects intracellular binding in addition to active uptake processes. These two partitioning parameters are related by intracellular binding (Parker and Houston, 2008), and hence fucell can be calculated. The experimental derived parameters used in eqs. 5 and 6 were taken from Yabe et al. (2011) and were based on the oil-spin method.
Results
Both monophasic and biphasic depletion-time profiles were observed in the conventional and media-loss experiments. Whereas six of the eight drugs studied showed biphasic profiles, only four of the eight cases (atorvastatin, fexofenadine, pitavastatin, and rosuvastatin) were unique to the media-loss approach (type A behavior). Clarithromycin and saquinavir were biphasic in both types of experiments. Monophasic depletion-time profiles were observed for erythromycin and repaglinide in both the media-loss and conventional experiments. This latter type of behavior (no difference between approaches) was designated type B.
A characteristic type A profile obtained with fexofenadine is shown in Fig. 3A, where the fraction remaining over time is shown for 0.1 μM fexofenadine in the two systems. With fexofenadine, the clearance is clearly biphasic and faster using the media-loss assay, with a parallel decline for both systems seen at the later time points. Fexofenadine CLint is 17 μl · min−1 · 106 cells−1 using the conventional assay and 117 μl · min−1 · 106 cells−1 using the media-loss assay (Table 2). This higher in vitro CLint in the media-loss assay, compared with the conventional assay, may be explained by the characterization of the uptake, which is faster than metabolism; hence, fexofenadine accumulates into the cells with subsequent metabolism/excretion. Similar profiles were obtained with rosuvastatin, pitavastatin, and atorvastatin.
A second characteristic (type B) profile was exemplified by clarithromycin (see Fig. 3B). The profiles obtained with the two methods are similar (Fig. 1B), with CLint values of 46 and 68 μl · min−1 · 106 cells−1 using the conventional assay and the media-loss assay, respectively (Table 2). In this case, no conclusion can be made concerning the importance of active uptake from the present data. Although there is good evidence for active uptake of clarithromycin in rat hepatocytes, uptake is probably not the rate-limiting step for its hepatic clearance; once the drug is actively transported into cells, it may or may not undergo intracellular accumulation before metabolism (depending on whether metabolism is the rate-limiting step). Similar characteristics were obtained with saquinavir, erythromycin, and repaglinide.
The apparent clearance values seen in the conventional approach ranged from 3.3 to 1675 μl · min−1 · 106 cells−1, whereas there is a smaller range in the media-loss experiments (35 to 1657 μl · min−1 · 106 cells−1). Extreme values were obtained for rosuvastatin and saquinavir and erythromycin and rosuvastatin, respectfully. As seen in Table 2, the rank order across the eight drugs was different for the two methods. Also shown in Table 2 is the ratio of the two methods, which varied from 1 to almost 500.
Concentration dependence was seen over the range of 0.1 to 10 μM for five of the eight drugs either in both systems (atorvastatin, erythromycin, and saquinavir) or in one system [conventional (fexofenadine) or media-loss (pitavastatin) approaches]. In all cases, Km values obtained for the two systems were not statistically different and were in the low micromolar region (see Table 2). For clarithromycin, repaglinide, and rosuvastatin, there was no evidence of concentration dependence over the initial concentration range studied. Figure 4 shows examples of the concentration dependence for atorvastatin and fexofenadine. In the latter case, the Km is similar to that reported by us using the oil-spin method; however, for atorvastatin, this value was an order of magnitude lower (Yabe et al., 2011).
The data for the four drugs (atorvastatin, fexofenadine, pitavastatin, and rosuvastatin) that showed a clear difference between the two methods were modeled according to the two-compartment approach described (Fig. 2), and the parameter values obtained are shown in Table 3. Five parameters were obtained to describe clearances for uptake and metabolism (CLuptake and CLmet) and passive transport (Pdiff), together with the media and the cell volume terms Vmedia and Vcell, respectively (Table 3). The clearance caused by metabolism showed good agreement with the apparent clearance observed in the conventional method for all four compounds. However, greater variability was evident in these parameters (Table 3). This is larger than reported previously for the oil-spin method (Yabe et al., 2011) and for comparable data analysis with experiments using monolayers of hepatocytes (Ménochet et al., 2012).
In the case of the uptake clearance, good agreement was seen between the model parameter and the media-loss slope only for fexofenadine. For atorvastatin (where there is extensive metabolism), the model value was reasonably similar, but in the case of pitavastatin and rosuvastatin, a lower value was obtained using the model. Pdiff values were also obtained from the model and were lowest for fexofenadine (13 μl · min−1 · 106 cells−1) but similar for the other three compounds (162–234 μl · min−1 · 106 cells−1). The Vmedia values were similar for all four compounds and close to the actual volume of the incubation (250 μl). The model parameters showed no marked difference when this volume term was restrained to 250 μl. Vcell showed a 10-fold range over the four compounds, with the rank order being rosuvastatin < pitavastatin < fexofenadine < atorvastatin, which is consistent with the fucell values determined previously (see Table 1).
Compared with published values using the oil-spin method (Yabe et al., 2011), Pdiff values were considerably greater, and CLuptake values were lower for atorvastatin and fexofenadine, whereas they were higher for rosuvastatin and pitavastatin. In addition, the rank order did not coincide; however, the spread of the parameter values was quite minimal for these four drugs.
Figure 5 illustrates the relationship between CLuptake and Pdiff in defining the ratio of cellular to media concentrations (Kpu) on the basis of eq. 5. Each of the eight drugs studied are identified on this three-dimensional surface, and it is clear that the four drugs corresponding to type A fall within the top half, with Kpu values in excess of 10. In contrast, type B compounds have combinations of Pdiff and CLuptake that gave Kpu values < 10.
Discussion
The underprediction of in vivo clearance from both hepatic microsomal and isolated hepatocyte studies has been evident in several analyses of published reports (Soars et al., 2007a; Hallifax et al., 2010). Although it has been acknowledged that in vitro methods underperform kinetically relative to in vivo methods, there is also a need to ensure that all clearance processes are considered so the scaling of in vitro parameters is not compromised (Houston, 1994). Attention has been drawn to the impact of hepatic transporters as a mechanism for underprediction and hence the need to incorporate this phenomenon into in vitro/in vivo extrapolation (Soars et al., 2007b). Examples of successful scale up of OATP-transported drugs from in vitro kinetic studies using the oil-spin method have been reported (Watanabe et al., 2009, 2010).
There is a need for a methodology for routine assessment of hepatic uptake that has faster throughput than the established oil-spin approach. The use of a monolayer of hepatocytes that allows successive washing before cellular drug determination offers one alternative approach (Ménochet et al., 2012). Another methodology has been proposed by Soars et al. (2007b), who have adapted the standard depletion approach, which is widely used for metabolic clearance determination, to allow assessment of hepatic uptake. The notion of dual incubations to determine hepatic uptake in addition to hepatic metabolism is attractive. However, certain criteria need to be established regarding the rate-limiting processes operating in this system for particular drugs, namely uptake or metabolism/efflux. If uptake clearance exceeds metabolic clearance, a difference in the two methods would be anticipated, i.e., when the rate-limiting step is not the initial uptake process.
In the current investigations with eight drugs known to be OATP substrates, comparison of the conventional assay and the media-loss assay has led to the identification of two groups. The first group (type A, which includes atorvastatin, fexofenadine, pitavastatin, and rosuvastatin) is characterized by a higher predicted CLint using the media-loss assay than the conventional assay, and this can be explained by a markedly faster uptake clearance than metabolic clearance. In the second group (type B, which includes clarithromycin, erythromycin, repaglinide, and saquinavir), the same profiles were obtained with both methods, and this may be explained by several situations including either the lack of active transport, an uptake rate limitation for metabolic clearance, or relatively similar clearance values; further experimentation would be needed to discriminate between these options.
Pdiff in addition to CLactive appeared to play a strong role in defining whether type A or type B behavior was observed in this particular series of eight compounds, where Pdiff ranged 40-fold. For type A compounds, Pdiff was minor (<10% of CLactive), whereas for type B compounds, Pdiff was at least 20% of CLactive; in two cases (erythromycin and saquinavir), the two parameters were comparable. Kpu is a valuable metric reflecting the relative values of these two terms. The combined role of passive and active processes in controlling drug uptake has been reviewed previously (Sugano et al., 2010).
Binding within the cell (fucell) appears to be independent of the type of behavior observed. For example, atorvastatin and rosuvastatin have widely different fucell values but yet are both type A, and saquinavir and erythromycin, both type B, also have quite different fucell values (Table 1). There is good agreement between the fucell values reported previously and the Vcell values obtained from modeling the current data. Although cellular binding might be expected to result in biphasic depletion-time profiles, these would not be unique to the media-loss assay, as exemplified by clarithromycin and saquinavir in this study.
The CLmet values for the eight compounds differ significantly. In the case of saquinavir, metabolic clearance is known to be uptake rate-limited (Parker and Houston, 2008), whereas clarithromycin (Brown et al., 2010) and repaglinide (Ménochet et al., 2012) are known to be metabolism rate-limited. It would be expected that erythromycin would show similar behavior to clarithromycin; atorvastatin also appears to show similar characteristics (Paine et al., 2008). Thus, whether a compound falls into the type B category or the type A category appears not to be related to its rate-limiting step in hepatic clearance.
Three of the four type A drugs are known to show minimal metabolism, and the slow metabolic clearance from the conventional method is to be expected. Thus, the biphasic behavior using the media-loss methodology indicating rapid uptake before metabolism/excretion is consistent. Although efflux transporters are internalized during the isolation of fresh hepatocytes (Bow et al., 2008), the clearance term designated as metabolism may include some cellular efflux activity. Atorvastatin also showed type A behavior despite its extensive and rapid metabolism. It would appear that the uptake process for atorvastatin is faster than its metabolism. For type A compounds, there will be accumulation within the cell because metabolism/efflux is the rate-limiting step governing the intracellular unbound concentration for these drugs.
There is some ambiguity in the transporter literature regarding the terms rate-limiting and rate-determining in describing the role of transport in hepatocellular uptake (Parker and Houston, 2008; Brown et al., 2010; Watanabe et al., 2010). The rate-limiting step in kinetic terms reflects the slowest step in a catenary sequence; thus, hepatic disposition will be uptake rate-limited only when the uptake by transporters into the cell has a slower clearance than the metabolism. However, use of the term rate-determining is more descriptive and can be applied to the overall loss of drug from the plasma (external media) and should be viewed differently. Thus, the action of transporters can influence hepatocellular uptake without being the rate-limiting step. There are examples where biliary excretion clearance is less than hepatic uptake clearance, for example, napsagatran (Poirier et al., 2009), as well as examples where metabolic clearance is slower than uptake clearance, as in the case of saquinavir and nelfinavir (Parker and Houston, 2008). For drugs with metabolism/efflux as the rate-determining step for hepatic disposition, extensive intracellular accumulation of unbound drug would be expected. However, this is not always the case because passive permeability also contributes to the parameter Kpu (see eq. 5); this parameter appears to be a particularly valuable guide to the hepatocellular kinetic behavior.
The use of a range of initial drug concentrations allowed saturation of uptake as well as metabolism to be identified. This was evident for atorvastatin and pitavastatin, but no indication of nonlinearity was seen for clarithromycin, repaglinide, or rosuvastatin. For erythromycin, saquinavir, and fexofenadine, there are inconsistencies between the two methods, suggesting that a more robust methodology (oil-spin or monolayer methods where a substantial amount of kinetic data are generated) is required for more detailed analysis. The high imprecision evident in the modeling of type A drugs would support this notion because the use of biexponential equations (eq. 2) requires a reasonable number of data points in both phases, which is difficult to achieve, as illustrated in Fig. 3A.
These investigations have demonstrated that the dual incubation approach cannot be relied upon to identify the importance of hepatic transporters for a given drug. In the present study, this approach proved to be successful in 50% of the eight cases where a biphasic decline was evident with the media loss but not the conventional methodology (type A behavior). However, although transporter activity may not always be identified in the dual incubation approach, an accurate estimate of hepatic clearance from the media is still obtained, and this may reflect uptake or metabolism. Hence, if the outcome is type B behavior, the same clearance value is obtained in both methods. The media loss rather than the conventional approach may prove to be the method of choice for drugs where transporters are anticipated, and this safer option may provide more reliable predictions. However, detailed studies involving more robust methods such as the oil-spin or monolayer methods will be required subsequently to improve precision of parameter estimation and to provide mechanistic information reflected in the parameters Vmax, Km, Pdiff, and Kpu.
Authorship Contributions
Participated in research design: Jigorel and Houston.
Conducted experiments: Jigorel.
Performed data analysis: Jigorel.
Wrote or contributed to the writing of the manuscript: Jigorel and Houston.
Acknowledgments
We thank Sue Murby and David Hallifax for valuable assistance with LC-MS/MS and Aleksandra Galetin for useful discussions.
Footnotes
This work was supported by a consortium of pharmaceutical companies (GlaxoSmithKline, Lilly, Pfizer, and Servier) within the Centre of Applied Pharmacokinetic Research at the University of Manchester.
This work was previously published in part in the following publication: Jigorel E and Houston JB (2009) Involvement of drug uptake in hepatic clearance. Drug Metab Rev 41 (Suppl 1): 66.
Article, publication date, and citation information can be found at http://dmd.aspetjournals.org.
ABBREVIATIONS:
- CLactive
- clearance by active uptake
- CLmet
- clearance by metabolism
- CLobs
- observed in vitro clearance (either by uptake or by metabolism)
- CLuptake
- total uptake clearance by active and passive processes
- fucell
- fraction of unbound drug in the hepatocyte
- kdep
- initial depletion rate constant
- Kptotal
- tissue-to-medium total drug concentration ratio
- Kpu
- hepatocyte-to-medium unbound drug concentration ratio
- OATP
- organic anion transporter polypeptide
- Pdiff
- passive uptake clearance
- LC-MS/MS
- liquid chromatography-tandem mass spectrometry.
- Received March 16, 2012.
- Accepted May 16, 2012.
- Copyright © 2012 by The American Society for Pharmacology and Experimental Therapeutics