Abstract
Apparent intrinsic clearance (CLint,app) of 7-ethoxycoumarin, phenacetin, propranolol, and midazolam was measured using rat and human liver microsomes and freshly isolated and cryopreserved hepatocytes to determine factors responsible for differences in rates of metabolism in these systems. The cryopreserved and freshly isolated hepatocytes generally provided similar results, although there was greater variability using the latter system. The CLint,app values in hepatocytes are observed to be lower than that in microsomes, and this difference becomes greater for compounds with high CLint,app. This could partly be attributed to the differences in the free fraction (fu). The fu in hepatocyte incubations (fu,hep-inc) was influenced not only by the free fraction of compounds in the incubation buffer (fu,buffer) but also by the rate constants of uptake (kup) and metabolism (kmet). This report provides a new derivation for fu,hep-inc, which can be expressed as fu,hep-inc = [kup/(kmet + kup)]/[1 + (Chep/Cbuffer) × (Vhep/Vbuffer)], where the Chep, Cbuffer, Vhep, and Vbuffer represent the concentrations of a compound in hepatocytes and buffer and volumes of hepatocytes and buffer, respectively. For midazolam, the fu,hep-inc was calculated, and the maximum metabolism rate in hepatocytes was shown to be limited by the uptake rate.
The determination of in vitro intrinsic clearance (CLint) for drug candidates in the early discovery stage is a common practice in the pharmaceutical industry (Houston, 1994; Lave et al., 1997; Obach et al., 1997). The CLint values of drug candidates can help to confirm whether metabolism is the main clearance pathway when it is compared with the total body clearance in vivo. It is also helpful in rank-ordering drug candidates based on their metabolic stabilities, assessing species and gender differences in metabolic clearance, and projecting the metabolic clearance of drug candidates in humans. The in vitro CLint may be derived from enzyme kinetic data such as Vmax/Km (Lin et al., 1996; Tan and Pang, 2001; Griffin and Houston, 2004) or from the in vitro t1/2 values where subKm substrate concentrations are used (Lave et al., 1997, Obach, 1999; Lau et al., 2002; Jones and Houston, 2004). The CLint can be calculated from the experimental apparent intrinsic clearance, CLint,app, by correcting for free fraction of test compounds in the incubations. To further predict the in vivo hepatic clearance from the in vitro intrinsic clearance, a well stirred model is often used (Naritomi et al., 2001; Ito and Houston, 2004). A survey of literature revealed that in hepatocyte incubations, the free fraction of test compound has not been well defined. Simply assuming a steady state where the intracellular free concentration equals the extracellular free concentration may allow one to roughly estimate CLint for some compounds. However, clearance, after a dose in vitro or in vivo, is actually a dynamic system such that at any given time the amount of compound getting into a cell typically equals the amount of compound leaving the cell by diffusion and by metabolism (Fig. 1). Thus, the intracellular free concentration is always somewhat lower than the extracellular free concentration because metabolism constantly removes compound from hepatocytes, and an extracellular-intracellular free concentration gradient is needed to replenish the metabolized and outfluxed compound. For some rapidly metabolized compounds, the intracellular free concentration may be much lower than the extracellular free concentration because the removal of compounds by metabolism could be faster than the uptake. In the literature, a few attempts have been reported on measurement of free fraction using metabolically inactivated or dead hepatocytes by equilibrium dialysis (Austin et al., 2005) or using the ratio of free concentration in the buffer over the total concentration in hepatocytes, free and bound (Witherow and Houston, 1999). Both approaches assumed the buffer concentration equals the intracellular free concentrations of compound without considering metabolism. The current report provides a derivation of free fraction in the hepatocyte incubations using the more dynamic system with intact diffusional and metabolic processes.
Among the routinely used in vitro systems, such as microsomes and hepatocytes (Houston and Carlile, 1997), microsomes are usually used to determine P450-mediated metabolism (phase I). Hepatocytes, having intact cell membranes and physiological concentrations of enzymes and cofactors, are believed to be a model close to whole liver for drug clearance measurements (Bachmann et al., 2003; Ito and Houston, 2004; McGinnity et al., 2004). The aims of this work were 1) to compare intrinsic clearance determinations in hepatic microsomes and hepatocytes for a set of compounds, including marketed drugs that are primarily metabolized by phase I enzymes; 2) to explore the relationship of the free fraction with uptake and metabolism of the test compound in hepatocytes; and 3) to provide an explanation for the low apparent intrinsic clearances observed in hepatocyte incubations compared with that in microsomal incubations.
Hepatocyte suspension model, a dynamic system where the rate of uptake equals the rate of metabolism plus diffusion from inside to outside at any given time. The rate constant of diffusion across either sides are assumed to be the same. The active uptake and efflux transporters were ignored for simplicity.
Materials and Methods
Reagents. Pooled human and rat liver microsomes and pooled cryopreserved rat hepatocytes were purchased from XenoTech LLC (Kansas City, KS). Cryopreserved human hepatocytes (pool of four in the experiment) were purchased from In Vitro Technologies, Inc. (Baltimore, MD). Fresh human hepatocytes were purchased from BD Gentest (Woburn, MA) and In Vitro Technologies, Inc. Fresh rat hepatocytes were prepared in-house. All hepatocytes used in this study had viability of >80%. Midazolam, 7-ethoxycoumarin, phenacetin, and propranolol were purchased from Sigma-Aldrich (St. Louis, MO).
CLint,app Determination in Microsomes and Hepatocytes. Microsomes (0.5 mg/ml) were preincubated with 2 μM test compound for 5 min at 37°C in 0.1 M phosphate buffer, pH 7.4. The reactions were initiated by adding prewarmed cofactors (2 mM NADPH and 3 mM MgCl2). After 0-, 3-, 7-, 12-, 20-, and 30-min incubations at 37°C, the reactions were stopped by adding an equal volume of acetonitrile containing 1 μM carbutamide (internal standard). The samples were kept in a refrigerator for 30 min and then centrifuged at 3000g for 10 min. The supernatants were analyzed with LC/MS/MS for the amount of parent compound remaining.
Calculation of apparent intrinsic clearance: a: 20 and 45 g of liver/kg of body weight were used for human and rat, respectively (Lin et al., 1996).
Likewise, CLint,app determinations in hepatocytes were performed in Krebs-Henseleit buffer, pH 7.4, with 1.0 × 106 hepatocytes/ml (viability >80%) and 2 μM test compound. The incubations were carried out in a 37°C CO2 (5%) incubator as described previously (Li et al., 1999a). The reactions were stopped at 0, 30, 60, 120, 180, and 240 min with the addition of an equal volume of acetonitrile containing 1 μM carbutamide. A value of 135 × 106 hepatocyte/g of liver (Houston, 1994) was used in the CLint,app calculation (equation above).
Hepatocyte Uptake. Cryopreserved hepatocytes were thawed out and suspended in Krebs-Henseleit buffer at 2.0 × 106 hepatocytes/ml as described previously (Shitara et al., 2003). Centrifuge tubes were prepared by loading 150 μl of silicone/mineral oil on top of 50 μl of 5 N NaCl/0.2% saponin. In a 37°C water bath, 0.5-ml hepatocyte suspensions were preincubated in the presence or absence of 10 μM ketoconazole for 10 min. Uptake was initiated by adding 0.5 ml of 4 μM midazolam into the hepatocytes. At 0.5, 2, 4, 10, 20, and 30 min, 100 μl of the cell suspensions were transferred into the centrifuge tubes containing the silicone/mineral oil and the NaCl/0.2% saponin, and the uptake was terminated by separating the hepatocytes from the midazolam solution with a 10-s centrifugation at 10,000g. Midazolam concentrations in the hepatocytes, as well as the uptake solution, were analyzed using LC/MS/MS with appropriate standard curves.
Microsomal Protein Binding. Microsomal protein binding assay was adapted from a published procedure (Obach, 1997): microsomes (0.5 mg/ml) were mixed with 2 μM test compound in 0.1 M phosphate buffer, pH 7.4, containing 3 mM MgCl2 (the dialysis buffer). The mixture was subjected to an overnight dialysis at 37°C against the dialysis buffer using the Spectrum apparatus (Spectrum, Los Angeles, CA). The retrieved microsomes were then diluted in 2 volumes of the dialysis buffer, and dialysate from the receiving side was diluted in half-volume of the control microsomes. After the protein in the samples from the donor and the receiver was precipitated in an equal volume of acetonitrile containing 1 μM carbutamide, the supernatants were analyzed using LC/MS/MS for the amount of parent compound remaining. The free fractions were calculated as:
LC/MS/MS Analyses. Peak area ratios of test compounds and carbutamide (internal standard) were determined by a LC/MS/MS system which consisted of an Agilent 1100 HPLC (Agilent, Palo Alto, CA), a Leap CTC PAL autosampler (LEAP Technologies Inc., Carrboro, NC), and a SCIEX API 4000 detector (Applied Biosystems, Concord, ON, Canada). Separation was performed on a Waters YMC Basic, 3 μM, 50 × 2.0-mm column (Waters, Milford, MA), eluted at a flow rate of 0.5 ml/min. Mobile phase A was 0.1% (v/v) formic acid in water, and mobile phase B was 0.1% (v/v) formic acid in acetonitrile. The gradient consisted of 30% of mobile phase B for 0.5 min after injection and increased linearly to 90% B from 0.5 to 1.5 min. Mobile phase B was held at 90% from 1.5 to 2.3 min, and the column was re-equilibrated to 30% B from 2.3 to 3.5 min. All compounds were detected by positive ion spray in the multiple-reaction monitoring mode using predetermined parent/product mass transition ion pairs.
Results
The apparent intrinsic clearance results are presented in Table 1. Phenacetin, 7-ethoxycoumarin (7-EC), propranolol, and midazolam are primarily metabolized by phase I enzymes. The CLint,app values for all compounds in hepatocytes were much lower compared with those in microsomes, with midazolam showing the highest differential and, hence, has been studied in detail. All four compounds showed comparable CLint,app in freshly isolated (three experiments) and cryopreserved (five experiments) rat hepatocytes with a similar interexperimental variation (Table 2). Furthermore, the mean CLint,app values in freshly isolated and cryopreserved rat hepatocytes were comparable, suggesting that the enzyme activity in the cryopreserved rat hepatocytes was at levels comparable with those in the freshly isolated rat hepatocytes. In the cryopreserved human hepatocyte study, these four compounds also showed a similar interday variation in enzyme activity as that in cryopreserved rat hepatocytes. However, in freshly isolated human hepatocytes, the interexperimental variation was notably higher (Table 2). Unlike the cryopreserved hepatocyte study, which uses the same characterized and pooled hepatocytes, the fresh human hepatocyte studies were subject to interindividual variation in enzyme activities, limited characterization at the time of the experiment, and a possibility of damaged hepatocytes during shipment. On the other hand, cryopreservation is known to preserve most of the P450 activities of the original freshly isolated hepatocytes (Li et al., 1999b; Madan et al., 1999). Because the mean values of CLint,app in cryopreserved and freshly isolated human hepatocytes in our study seemed to be similar, cryopreserved human hepatocytes could be a valid model for CLint,app determinations.
Apparent intrinsic hepatic clearances (CLint,app) from different in vitro systems
Apparent clearance (l/h/kg) in freshly isolated and cryopreserved rat and human hepatocytes
In the hepatocyte uptake study, after separating hepatocytes from the incubation buffer at different time points, the concentrations of midazolam in both hepatocytes and buffer were measured. In the human hepatocyte incubations, midazolam was depleted from the incubation buffer at the same rate as in the rat hepatocyte incubations. In contrast to hepatocytes, the clearance of midazolam in the liver microsomal incubation from rats was faster than that from humans (Table 1). When the selective CYP3A4 inhibitor ketoconazole was coincubated with midazolam (a selective CYP3A4 substrate), midazolam was no longer cleared from the incubation buffer (Fig. 2). These data suggested that 1) the removal of midazolam from the incubation buffer is also due to metabolism in hepatocytes, and 2) the uptake is the rate-limiting step. Taken together, these observations suggest that the midazolam uptake rates in rats and humans are similar to each other and that both rates are lower compared with the corresponding metabolism rates; i.e., as soon as midazolam gets into the hepatocytes, it is cleared rapidly by metabolism. It is reasonable to believe that midazolam uptake rates are similar in human and rat hepatocytes because they reflect simple diffusion across the cell membranes. Midazolam is not reported to be a substrate of any active uptake or efflux transporters; therefore, its uptake is not affected by ketoconazole. Ketoconazole is an inhibitor of several uptake and efflux transporters and CYP3A (Azer et al., 1995; Salphati and Benet, 1998).
Table 3 summarizes the free fractions (fu) of midazolam in human and rat liver microsomal incubations. From the protein binding studies, midazolam had fu values of 0.83 and 0.84 in human and rat microsomes, respectively, showing that the binding was similar in rat and human systems. The fu values in hepatocyte incubations were calculated from the parameters obtained from the hepatocyte uptake study using eq. 12 (derivation discussion below). The uptake rate constants in rats and humans were determined to be 0.0419 and 0.0371/min/(million hepatocytes/ml), respectively; the volume ratios (Vhep/Vbuffer) were 1:100 for both rats and humans, and the concentration ratios (Chep/Cbuffer) were 35 and 41 for rats and humans, respectively. The CLint values of midazolam in microsomal and hepatocyte incubations were calculated from the CLint,app and the fu and are presented in Table 3 for comparison.
Midazolam clearances (l/h/kg) and unbound fraction in microsomes and hepatocytes
Midazolam uptake in human and rat hepatocytes. Measurement of midazolam disappearance in the incubation buffer after removing hepatocytes at each incubation time point.
Discussion
In our study, the CLint,app values of 7-EC, phenacetin, propranolol, and midazolam, all cleared by phase I metabolism as the first step, were significantly lower in hepatocyte incubations compared with microsomal incubations. Similar observation was recently reported by Hallifax and colleagues (2005a), where they compared CYP1A2, 2C9, 2D6, 2E1, and 3A4 activities in cryopreserved hepatocytes from approximately 200 donors with that in microsomes from approximately 100 donors. The P450 activities in hepatocytes were found to be 2.5- to 20-fold lower than in microsomes. For example, using testosterone as the substrate, CYP3A4 activity was 11-fold lower in hepatocytes. Many factors could attribute to this difference, such as nonspecific binding or quality of cryopreserved hepatocytes. The present work provides an explanation for this difference from the perspective of free fraction in hepatocyte incubations for compounds that involve diffusional movement across cell membrane. Because hepatocytes are intact cells, compounds have to get into the cells before they can reach the metabolizing enzymes. The rate of compound clearance is, therefore, dependent on the uptake rate as well as the metabolism rate. Two important scenarios arise. If the uptake rate is much faster than the metabolism rate, then the overall clearance is metabolic rate limited. Conversely, if the metabolism rate is much faster than the uptake rate, the overall clearance is uptake rate limited.
Transcellular (diffusional) uptake is the primary route of entry into hepatocytes for most of compounds (Brayden, 1997). It is usually fast because it does not require an energy source or a salt gradient as most active uptake processes do. This discussion focuses on the transcellular uptake to quantitatively explore the relationship between hepatocyte clearance and hepatocyte uptake/metabolism. In hepatocytes, when the intracellular free concentration (Chep,u) is « the Michaelis-Menten constant (Km), the apparent metabolic rate is: and the uptake rate in the hepatocytes is:
where kup is the uptake rate constant, kmet is the metabolism rate constant, and Cbuffer,u is the free concentration in buffer which is same as the extracellular free concentration (Fig. 1). Assuming that diffusion across membrane is identical in both directions, the removal rate from the hepatocytes is:
where the term kmet × Chep,u represents metabolic rate, and kup × Chep,u represents the diffusion of compound from inside the hepatocytes to the outside buffer. After the compound is added to the incubation mixture, a dynamic equilibrium is established where the rate of compound being taken up by the hepatocytes equals the rate of compound being removed from the hepatocytes (by diffusional and metabolism) at any given time (Fig. 1), thus the uptake rate can be expressed as:
Dividing eq. 1 by eq. 4:
By multiplying both the numerator and denominator of the right side of eq. 5 by Cbuffer,u and cancelling out the Chep,u, eq. 5 becomes
Equation 6 can then be rewritten as:
where vmet,max = kmet × Cbuffer is the maximum metabolic rate if Cbuffer,u = Chep,u. In other words, this would be the highest possible metabolic rate of the compound in hepatocytes as if the enzymes were exposed to the extracellular concentration. Equation 7 demonstrates the hyperbolic relationship between apparent metabolic rate and the maximum metabolic rate, where the uptake rate is the upper limit. For compounds that have a lower rate of metabolism, the effect of vup on the clearance rate is minimal. However, for a rapidly metabolized compound, the apparent metabolic rate could plateau at vup.
Intrinsic clearance can be calculated from the apparent intrinsic clearance by correcting with the free fraction of the compound to which the enzymes are exposed (Obach, 1999). The fu in hepatocyte incubation (fu,hep-inc) could be expressed as the ratio of intracellular free concentration (Chep,u) to total incubation concentration (Ctotal), which is the amount of compound in hepatocytes (Chep × Vhep) and buffer (Cbuffer × Vbuffer) divided by the incubation volume (Vbuffer + Vhep): where Chep is the total concentration of compound in hepatocytes, i.e., free + bound to intracellular and membrane proteins, and Cbuffer equals the Cbuffer,u because no protein is present in the incubation buffer. The fu,hep-inc could be simplified assuming that the volume of the hepatocytes compared with the volume of the buffer is minimal (Vhep « Vbuffer):
Equation 9 can be rearranged to:
Equation 4 can be rearranged as:
By placing eq. 11 into eq. 10, we get:
At a true steady state where there is no metabolism to remove the compound (kmet = 0 or « kup), the intracellular free drug concentration equilibriums with the extracellular free drug concentration. Equation 12 can be simplified to:
Thus, the intracellular free drug concentration approaches the extracellular free drug concentration. However, in eq. 12, when metabolism is very high compared with the uptake, the numerator becomes a small fraction, and the fu,hep-inc becomes less than fu,buffer. Therefore, the intracellular free concentration becomes much lower than the extracellular free concentration. This explains why the rapidly metabolized compound midazolam has a much lower CLint,app in hepatocytes compared with that in microsomal incubations. As illustrated in eq. 12, the fu in hepatocyte incubations can be calculated from the experimental data using Chep, Vhep, Cbuffer, Vbuffer, kmet, and kup.
The uptake rate in a hepatocyte incubation can be calculated from the amount of a compound appearing in hepatocytes at any given period of time. But this method cannot distinguish the fraction of compound nonspecifically bound to the outer membrane of the hepatocytes from the fraction uptaken into the hepatocytes. In an incubation where a compound's metabolism rate is greater than its uptake rate, monitoring the disappearance of the compound from the incubation buffer provides an easier way to determine the uptake rate. In our study, the metabolism rate constants of midazolam in rats and humans, converted from the microsomal data with correction of fu, were 0.370 and 0.167/min/(million hepatocytes/ml), respectively. Putting values of rate constants of metabolism and uptake, as well the concentrations and volumes of the hepatocytes and incubation buffer in eq. 12, the fu values of midazolam in rat and human hepatocyte incubations were calculated to be 0.075 and 0.129, respectively (Table 3). Noticeably, before being corrected for fu, the apparent intrinsic clearance of midazolam in rat microsomal incubation was approximately 40-fold higher than that in hepatocytes. However, after a correction for fu in both systems, the difference in CLint is reduced to approximately 3.8-fold. Considering that the microsomal and hepatocyte incubations are two widely different experimental systems involving two different scale-up factors to calculate the fu, this 3.8-fold difference may be considered acceptable. Likewise, a smaller difference (3.5-fold) was observed in intrinsic clearances between human microsomal and hepatocyte incubations after a correction of both values with fu. The intrinsic clearance of midazolam using human hepatocytes was significantly lower than that in rat hepatocytes. This was consistent with the reported lower plasma clearance of 0.40 l/h/kg in humans (Hardman et al., 2001) compared with 4.75 l/h/kg in rats (Kotegawa et al., 2002). A recent study suggested that the metabolism of midazolam in human hepatocytes would not reach saturation until the total incubation concentration reaches 100 μM (Zhao et al., 2005). This is consistent with our study. The fu,hep-inc of midazolam in human hepatocytes calculated from eq. 12 is 0.129 (Table 3). Therefore, a 100 μM concentration in hepatocyte incubation would have an intracellular free concentration around 12.9 μM. At such a concentration midazolam, which has Km around 4 μM determined in human liver microsomes (Pelkonen et al., 1998), would have its metabolism rate at saturation.
The CLint,app values for midazolam in hepatocytes, measured at subKm concentrations using the method of disappearance of parent compound, are consistent with those reported by Lave et al. (1997) and Lau et al. (2002). It is interesting to note that several reports (Obach, 1999; Kotegawa et al., 2002; Walsky and Obach, 2004; Hallifax et al., 2005) show different CLint,app values measured using the alternate method of Vmax/Km. This discrepancy may be attributable to the Vmax/Km approach for estimating CLint,app, where saturating concentrations of midazolam were used and enzymes in hepatocytes metabolized midazolam at full capacity. One has to be cautious when using the data from the Vmax/Km approach because midazolam, at high nonphysiological concentrations, is known to be a “substrate inhibitor” of CYP3A (Houston and Kenworthy, 2000; Galetin et al., 2003; Hallifax et al., 2005), thereby also showing reduced clearance in microsomes.
In conclusion, hepatocytes, being closer to the in vivo liver system, are ideally suited for the prediction of in vivo hepatic clearance. Microsomes are still valuable for clearance determination for compounds metabolized primarily by phase I enzymes. The determination of intrinsic clearance using hepatocytes is dependent on CLint,app, which in turn, is dependent on the metabolism and cellular uptake rates. A derivation of an equation to calculate fu,hep-inc is provided, and its use to account for the lower in vitro clearance in human and rat hepatocytes compared with microsomes for midazolam, a compound which shows high clearance using microsomal system, is demonstrated.
Footnotes
-
Article, publication date, and citation information can be found at http://dmd.aspetjournals.org.
-
doi:10.1124/dmd.106.010793.
-
ABBREVIATIONS: CLint, intrinsic clearance; CLint,app, apparent intrinsic clearance; P450, cytochrome P450; LC/MS/MS, liquid chromatography coupled to tandem mass spectrometry; 7-EC, 7-ethoxycoumarin; fu,hep-inc, free fraction in hepatocyte incubations.
- Received April 24, 2006.
- Accepted June 19, 2006.
- The American Society for Pharmacology and Experimental Therapeutics