Experimental Procedures

Chemicals.

Ibuprofen and 7-ethoxy coumarin were purchased from Aldrich (Milwaukee, WI). Tolbutamide and (S)-warfarin were obtained from Salford Ultrafine Chemical and Research Ltd. (Manchester, England). Alprenolol, quinidine, propranolol and trypsin inhibitor (Type II-S: Soybean) were obtained from Sigma Chemical Co. (St. Louis, MO). Lidocaine, theophylline and hexobarbital were purchased from Tokyo Kasei Kogyo Co. (Tokyo, Japan). Antipyrine, caffeine, chlorpromazine, diazepam, phenytoin, and collagenase were obtained from Wako Pure Chemical Industries (Osaka, Japan). Compounds X and Y were synthesized in our laboratories. William's E medium was from Life Technologies (Grand Island, NY). All other chemicals were obtained from either Wako Pure Chemical Industries (Osaka, Japan) or Junsei Chemical Co. (Tokyo, Japan).

Serum Preparation.

Male Sprague-Dawley rats, 7 to 10 weeks of age, purchased from Charles River Japan, Inc. (Kanagawa, Japan), were used in all of the experiments. Rat blood was collected via the abdominal aorta under ether anesthesia. After stabilization for 3 h at room temperature to coagulate the blood, samples were centrifuged (10 min, 1800g). Serum was collected as the clear supernatant and stored at −80°C until use.

Hepatocyte Preparation.

Hepatocytes were freshly isolated from rats by a procedure similar to that described by Baur et al. (1975). After isolation, the hepatocytes were suspended in William's E medium (WE), pH 7.4, and kept on ice until use. Cell viability was routinely checked by the 0.4% Trypan Blue exclusion test, and only hepatocytes with viability greater than 90% were used in this study.

Effect of Serum on Antipyrine Metabolism.

Incubation media containing 0, 25, 50, and 100% serum were prepared by mixing rat serum and William's E medium (v/v). Hepatocytes were resuspended at a density of 2 × 10⁶cells/ml in each incubation medium at ice-cold temperature. An aliquot of DMSO solution of 12.5 mM antipyrine was pipetted at a volume of 1 μl per well into a 48-well plate. Each hepatocyte suspension was added at a volume of 250 μl per well at ice-cold temperature (the final concentration of antipyrine was 50 μM, DMSO was 0.4%). The final antipyrine concentration was set lower than its Michaelis-Menten constant, K_m = 2.2 mM (Buters and Reichen, 1990). The samples were incubated at 37°C with shaking at 150 rpm under an atmosphere of 95% O₂/5% CO₂. After 0.5-, 1-, and 2-h incubation times, the reaction was terminated by adding 500 μl of ice-cold CH₃CN/MeOH solution (2:1, v/v). The samples were centrifuged (10,000g × 10 min), and the amount of antipyrine remaining in the supernatant was measured by HPLC-UV (254 nm) (SPD10A; Shimadzu, Tokyo, Japan).

Serum Incubation Method.

The metabolism studies were typically performed as follows. Hepatocytes were resuspended at a density of 1 × 10⁶cells/ml in rat serum at ice-cold temperature. An aliquot of DMSO solution of a compound was pipetted at a volume of 1 μl per well into a 48-well plate with the exception of (S)-warfarin, which was added as an aliquot of water solution. The hepatocyte suspension was added in a volume of 250 μl per well at ice-cold temperature (final concentration of DMSO: 0.4%). The samples were incubated at 37°C with shaking at 150 rpm under an atmosphere of 95% O₂/5% CO₂. After a 1-h incubation, the reaction was terminated by adding 500 μl of ice-cold CH₃CN/MeOH solution (2:1, v/v). The samples were centrifuged (10,000g × 10 min), and the amount of the compound remaining in the supernatant was measured by HPLC-UV or liquid chromatography-tandem mass spectrometry. For low-clearance compounds such as antipyrine, caffeine, ibuprofen, theophylline, tolbutamide, and (S)-warfarin, cell density and incubation time were modified to 3 × 10⁶ cells/ml and 2 h, respectively. All substrate concentrations were set lower than their Michaelis-Menten constants (K_mvalues). Compounds with unknown K_m values, such as chlorpromazine, propranolol, verapamil, compound X, compound Y, and phenobarbital, were incubated at 1 μM, while theophylline was incubated at 5 μM.

Conventional Hepatocyte Incubation Method.

Hepatocytes were resuspended at a density of 0.2 × 10⁶ cells/ml in William's E medium, pH 7.4, at ice-cold temperature. Incubation and subsequent steps were performed as described under Serum Incubation Method. For low-clearance compounds such as antipyrine, caffeine, ibuprofen, theophylline, tolbutamide, and (S)-warfarin, cell density and incubation time were changed to 1 × 10⁶ cells/ml and 2 h, respectively. Compounds with unknown K_mvalues, such as chlorpromazine, verapamil, and phenobarbital, were incubated at 1 μM, while propranolol was incubated at 0.2 μM, compounds X and Y at 0.1 μM, and theophylline at 5 μM.

Theory and Calculation

The in vitro clearance of a drug is commonly expressed as follows:CL=rate of metabolism/CE Equation 1where C_E = the unbound drug concentration available at the enzyme site.

When the C_E of a drug is much lower than itsK_m value, the CL is calculated by the following eq. 2, using cell density (D), incubation time (T), and the ratio of unchanged compound remaining after incubation (R).CL=(−loge R)/(D×T) Equation 2Derivation of eq. 2: when the cell density (D) is used for an in vitro metabolism study, the rate of metabolism (dC/dt) at the concentration C(t) is expressed from eq. 1 as follows:dC/dt=−CL×D×C(t) By solving this differential equation, the amount of drug remaining after the incubation time (T) is expressed as follows:C(T)=C(0)×e(−CL×D×T) As C(T)/C(0) = R, the equation is converted to R = e^{(−CL×D×T)};then applying log_e to both sides, eq. 2 is produced.

When the compound is metabolized by hepatocytes suspended in William's E medium (the conventional incubation method), the intrinsic clearance of unbound fraction in William's E medium (CL_{u int, WE}) is calculated by eq. 2. Also, when an incubation is performed with hepatocytes suspended in serum (the serum incubation method), the intrinsic clearance in serum (CL_{int, serum}) is also calculated by eq. 2. To scale these in vitro clearance values up to in vivo liver values, the hepatocyte number per kilogram of body weight calculated from a report by Houston (1994) is used as a scaling factor (SF). The calculation is as follows:

SF = liver weight × hepatocyte number per gram of liver = 45 (g/kg of body weight) × 1.35 × 10⁸ (cells/g of liver) = 6075 × 10⁶ (cells/kg).

Then, CL_{H, int} is calculated as follows:

In the case of the conventional method:CLH,u int,WE=CLu int,WE×SF Equation 3By multiplying the unbound fraction in serum (f_ub, measured value),CLH,int,WE=fub×CLH,u int,WE Equation 4In the case of the serum incubation method, without using the f_ub value:CLH,int,serum=CLint,serum×SF Equation 5CL_H is predicted from CL_{H, int, WE} or CL_{H, int, serum} by using the dispersion model:CLH=QH×RB×(1−4a/((1+a)2exp[(a−1)/(2×DN)] Equation 6−(1−a)2exp[−(a+1)/(2×DN)])) RN=(CLH,int,(WE or serum))/(QH×RB)a=(1+4×RN×DN)1/2 FH=1−EH=1−CLH/(QH×RB) Equation 7where Q_H is liver blood flow rate (70 ml/min/kg) (an average value from the literature: Lin et al., 1982;Houston, 1994), D_N is dispersion number (0.17) (Roberts and Rowland, 1986), and R_B is blood to plasma concentration ratio (in compounds with unknown R_B values, a unity is used).

Results

Effects of Serum on Antipyrine Metabolic Activity in Isolated Rat Hepatocytes.

The effects of serum on the intrinsic metabolic potency of isolated rat hepatocytes were investigated using antipyrine as a test compound because antipyrine was reported not to bind to serum proteins (f_ub = 1; Singh et al., 1991). Aliquots of serum were added to the incubation medium (William's E medium) containing hepatocytes to give 0, 25, 50, and 100% serum. The metabolism of antipyrine in these incubation media was studied as described underExperimental Procedures.

No marked difference was observed in the metabolic rate of antipyrine in the different incubation media (Table1). Moreover, a linear metabolic reaction was observed over the 2 h following the start of the reaction. These observations demonstrate that serum has no effect on the intrinsic metabolic potency of antipyrine and can be used as an incubation medium for suspended isolated rat hepatocytes, at least during short-term metabolism studies.

Prediction of Metabolic Clearance and Availability in Liver by the Serum Incubation Method.

Eighteen compounds eliminated mainly by liver metabolism, including 16 commercial and 2 novel compounds, were evaluated by the serum incubation method in terms of prediction of CL_H and F_H in the liver. The following parameters of the compounds are listed in Table2: Michaelis-Menten constant (K_m), f_ub, in vivo plasma clearance (CL_p), and oral bioavailability (F_po). In vivo CL_{H, int} was calculated by the dispersion model with the assumption that CL_p was the same as the metabolic clearance. Incubation was performed as described under Experimental Procedures. The following parameters were calculated by the equations described under Theory and Calculation: CL_{int, serum}, CL_{H, int, serum}, CL_H, and F_H. As shown in Fig.1, a good correlation (r² = 0.94) was observed between the predicted CL_H and in vivo CL_p. Low-clearance drugs (CL_p < 15 ml/min/kg), such as antipyrine, caffeine, ibuprofen, phenytoin, tolbutamide, theophylline, and (S)-warfarin, were predicted to be low-clearance compounds. Moderately cleared drugs (15 < CL_p < 40 ml/min/kg), such as compounds X and Y, were predicted to be moderate-clearance compounds. Moreover, drugs showing high clearance (CL_p > 40 ml/min/kg), such as alprenolol, chlorpromazine, diazepam, 7-ethoxy coumarin, hexobarbital, lidocaine, propranolol, quinidine, and verapamil, were predicted to be high-clearance compounds. These results indicate that the serum incubation method is able to predict in vivo hepatic clearance with appropriate rank order.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

Prediction of metabolic clearance in the liver by the serum incubation method.

CL_H was determined by the serum incubation method as described under Experimental Procedures. Each value represents the mean of three incubations. In vivo CL_p in rats was quoted from references. r² = 0.94 [1, alprenolol; 2, chlorpromazine; 3, diazepam; 4, 7-ethoxy coumarin; 5, hexobarbital; 6, lidocaine; 7, phenytoin; 8, propranolol; 9, quinidine; 10, verapamil; 11, compound X; 12, compound Y; 13, antipyrine; 14, caffeine; 15, ibuprofen; 16, tolbutamide; 17, theophylline; 18, (S)-warfarin].

Figure 2 shows the correlation between the predicted F_H and in vivo F_po values. Compounds such as alprenolol, hexobarbital, lidocaine, propranolol, and verapamil, defined as having low oral bioavailability in vivo (F_po < 0.2), were predicted to have poor hepatic availability because of extensive hepatic metabolism (predicted F_H < 0.2). This finding indicates that compounds extensively metabolized by first-pass metabolism in the liver can be predicted by the serum incubation method. Also, compounds reported to have good oral bioavailability (F_po > 0.7), such as antipyrine, caffeine, ibuprofen, tolbutamide, and theophylline, were predicted to have good hepatic availability (predicted F_H > 0.7). However, phenytoin was also predicted to have F_H> 0.7, while the in vivo F_po is reported as 0.26 (Scriba et al., 1995). Phenytoin was reported to show poor absorption in the intestine because of low solubility (Scriba et al., 1995). For such compounds with insufficient availability in the intestine, F_po cannot be predicted only by F_H because F_po is calculated as the product of F_H and intestinal availability. Therefore, for compounds with excellent availability in the intestine, the serum incubation method predicts in vivo F_po values in rank order.

• Download figure
• Open in new tab
• Download powerpoint

Figure 2

Comparison of F_H and in vivo oral bioavailability F_po.

F_H was determined by the serum incubation method as described under Experimental Procedures. Each value represents the mean of three incubations. In vivo F_po in rats was quoted from references. [1, alprenolol; 3, diazepam; 5, hexobarbital; 6, lidocaine; 7, phenytoin; 8, propranolol; 10, verapamil; 11, compound X; 12, compound Y; 13, antipyrine; 14, caffeine; 15, ibuprofen; 16, tolbutamide; 17, theophylline].

In conclusion, these results indicate that the serum incubation method can be used to predict in vivo CL_H and maximum F_po values without the f_ubvalue.

Comparison with the Conventional Method.

Conventional hepatocyte incubation was performed as described underExperimental Procedures. The following parameters were calculated by the equations described under Theory and Calculation: CL_{int, WE}, CL_{H, u int, WE}, and CL_{H, int, WE}. The correlation coefficient value in this section is calculated based on low- to moderate-clearance compounds with in vivo CL_p values under 40 ml/min/kg [phenytoin, compounds X and Y, antipyrine, caffeine, ibuprofen, tolbutamide, theophylline, and (S)-warfarin] to focus on the relationship of intrinsic clearance and free fraction in serum.

Figure 3A shows the correlation between in vivo CL_{H, int} and in vitro CL_{H, u int, WE} (r² = 0.55). Because f_ub was not reflected, the values of CL_{H, u int, WE} were greater than those of in vivo CL_{H, int}, especially in the case of compounds with high serum protein binding (f_ub < 0.02), such as ibuprofen, tolbutamide, and (S)-warfarin, while the values of compounds with low serum protein binding, such as antipyrine, caffeine, theophylline, and hexobarbital, were in good agreement. On the other hand, when CL_{H, u int, WE} was converted to CL_{H, int, WE} by multiplying the f_ub, the values were well correlated with in vivo CL_{H, int} (Fig. 3B,r² = 0.86). These observations suggest that the f_ub value is necessary for precise prediction of in vivo hepatic clearance if the conventional incubation method is adopted.

• Download figure
• Open in new tab
• Download powerpoint

Figure 3

Comparison of in vitro and in vivo intrinsic clearance values obtained by the conventional and the serum incubation methods.

A, CL_{H, u int, WE} versus in vivo CL_{H, int}. B, CL_{H, int, WE} versus in vivo CL_{H, int}. C, CL_{H, int, serum} versus in vivo CL_{H, int}. In vivo CL_{H, int} values were calculated by the dispersion model assuming that hepatic clearance was the same as in vivo CL_p. [1, alprenolol; 2, chlorpromazine; 3, diazepam; 4, 7-ethoxy coumarin; 5, hexobarbital; 6, lidocaine; 7, phenytoin; 9, quinidine; 10, verapamil; 11, compound X; 12, compound Y; 13, antipyrine; 14, caffeine; 15, ibuprofen; 16, tolbutamide; 17, theophylline; 18, (S)-warfarin]. Phenytoin, compounds X and Y, antipyrine, caffeine, ibuprofen, tolbutamide, theophylline, and (S)-warfarin are classified as low- to moderate-clearance compounds. A, CL_{H, u int, WE} was compared with in vivo CL_{H, int}. In vitro CL_{H, u int, WE} was determined using the conventional hepatocyte incubation method under serum-free conditions as described underExperimental Procedures.r² = 0.55 in low- to moderate-clearance compounds; r² = 0.89 in all compounds. B, CL_{H, int, WE} was compared with in vivo CL_{H, int}. In vitro CL_{H, int, WE} was calculated by multiplying CL_{H, u int, WE} and f_ub as quoted from references. r² = 0.86 in low- to moderate-clearance compounds; r² = 0.90 in all compounds. C, CL_{H, int, serum} was compared with in vivo CL_{H, int}. In vitro CL_{H, int, serum} was determined by the serum incubation method as described underExperimental Procedures.r² = 0.98 in low- to moderate-clearance compounds; r² = 0.85 in all compounds.

Figure 3C shows the correlation between in vivo CL_{H, int} and in vitro CL_{H, int, serum}, which was directly predicted by the serum incubation method. As in Fig.3B, a good correlation was observed between the in vitro and in vivo values (r² = 0.98). This finding shows that the serum incubation method gives comparable prediction results without considering the f_ub.

Discussion

Many methods for predicting the in vivo metabolic hepatic clearance of drugs from in vitro values have been reported. These methods usually require two independent in vitro parameters, f_ub and CL_{H, u int}(Houston, 1994; Houston and Carlile, 1997; Iwatsubo et al., 1997). Because we used isolated rat hepatocytes to reflect CL_{H, u int} and rat serum to reflect f_ub, parameters of in vitro CL_{H, u int} obtained from incubation with isolated hepatocytes and f_ub are essential to predict in vivo metabolism.

However, the conventional methodology is inadequate as a screening technique in the early stage of drug discovery because independent measurements of these two parameters are time-consuming. Moreover, it is difficult to measure f_ub and/or CL_{H, u int} accurately. Although f_ub is usually obtained by ultrafiltration or equilibrium dialysis methods, some compounds adsorb to the membrane and/or apparatus, which causes an inaccurate estimation of the results (Bertilsson et al., 1979; Desoye, 1988). The same problem would also affect the calculation of CL_{H, u int} because some drugs may not only adsorb to the apparatus but may also bind to microsomes or hepatocytes in the incubates (Obach, 1997, 1999). Therefore, for accurate calculation of CL_{H, u int}, the concentration of a drug in the incubate should be corrected as the free fraction of the drug (Obach, 1997, 1999). Thus, additional studies would be required for this purpose.

However, with the serum incubation method, direct measurement of CL_{H, int} at the expected concentration can be performed without considering the unbound fraction in serum as demonstrated in Fig. 3C, and quantitative prediction of in vivo CL_H and F_H can be performed as shown in Figs. 1 and 2. Because the number of samples in the analysis step is thereby reduced at least by half compared with the conventional method, we can obtain the prediction result much faster. In addition, serum can reduce the adsorption of compounds to the apparatus or to hepatocytes because serum proteins such as albumin are sometimes used to prevent adsorption of compounds to the apparatus. There are several reports of in vitro metabolism studies with hepatocytes or microsomes performed with media containing purified albumin or other serum proteins (Gariepy et al., 1992; Obach, 1997). But as in vivo serum includes many drug-binding proteins, it would be difficult to reflect in vivo f_ub correctly using only purified serum proteins. Lavé et al. (1997) ranked the compounds according to the ratio of hepatic extraction into three groups (low, middle, and high) after calculation of CL_{H, u int} based on human hepatocyte suspensions, and their observations correspond fundamentally to those shown in Fig. 3A. Thus, the serum incubation method seems to be ideal for predicting in vivo metabolic CL_H and F_H, and the simplicity of the method may save considerable time and labor in the drug discovery stage. As far as we know, this is the first report to describe a method that can quantitatively predict in vivo metabolic CL_H and F_H using isolated hepatocytes directly suspended in serum.

It is true that serum protein binding improves F_Hand CL_H, but the unbound drug concentration in serum also has significant effects on many aspects of pharmacodynamics and should also receive attention at the drug screening stage. Therefore, the f_ub should be confirmed by an appropriate method such as ultrafiltration or equilibrium dialysis after screening by the serum incubation method. The f_ub can be roughly estimated according to the difference in metabolic rate of CL_{H, int, serum} versus CL_{H, u int, WE} by a procedure similar to that reported by Gariepy et al. (1992). However, the relevance of this approach remains unclear because the activation of metabolism by serum protein was reported (Ludden et al., 1997), and adsorption of a drug to the apparatus and/or hepatocytes, especially in serum-free condition, should be considered. A high unbound fraction is desirable for accessibility to the target, but a low unbound fraction may not always be unfavorable if the drug has good pharmacokinetic properties and good pharmacodynamic effects, for the in vivo drug effect is also dependent on its intrinsic potency to the target.

When orally available compounds are screened in the drug discovery stage, the prediction of in vivo CL_H and/or F_H by the serum incubation method can facilitate the exclusion of undesirable candidates that have high clearance and low oral bioavailability due to extensive hepatic metabolism, without the performance of labor-intensive and time-consuming in vivo pharmacokinetic studies. In addition, when an in vivo pharmacokinetic result has already been obtained, predicted CL_Hand F_H values obtained from in vitro experiments may help to estimate the contribution of hepatic metabolism to in vivo oral bioavailability and systemic clearance.

Conclusion

The serum incubation method appears to be more convenient than conventional methodology for predicting in vivo metabolic clearance and availability in the liver because these values are achieved without assessing f_ub. This method may be a useful tool for screening orally available compounds in the early stage of drug discovery.