Abstract
The effect of human serum albumin (Hsa) and human liver cytosol (Hlc) on the in vitro enzyme kinetics of the formation of hydroxytolbutamide (CYP2C9 marker reaction) and the inhibitory effect of gemfibrozil on tolbutamide hydroxylation were examined using human liver microsomes. The addition of Hsa greatly decreased the unbound concentrations of tolbutamide and gemfibrozil in the incubation medium, whereas Hlc only slightly decreased them. The unboundKm value for tolbutamide hydroxylation was 123 μM without Hsa and Hlc, and 73, 88, and 64 μM in the presence of Hsa (5 mg/ml), Hlc (0.5 mg/ml), and Hsa plus Hlc, respectively. The predicted in vivo hepatic clearance (CLh) of tolbutamide based on enzyme kinetics without Hsa and Hlc (0.06 ml/min/kg) was 40% of its in vivo clearance (0.15 ml/min/kg) based on published data. Addition of 5 mg/ml Hsa and 0.5 mg/ml Hlc to the incubation medium distinctly improved the prediction, with the coaddition of Hsa and Hlc yielding the most accurate value (0.14 ml/min/kg). TheKi (6 μM) of gemfibrozil for CYP2C9, calculated using total drug concentrations, was increased by Hlc (8 μM), Hsa (40 μM), or both (72 μM). However, when the unbound substrate and inhibitor concentrations were considered, theKi (6 μM without Hsa and Hlc) was not markedly altered by Hsa (4 μM), Hlc (8 μM), or both Hsa and Hlc (9 μM). The present findings suggest that the addition of Hsa and Hlc to microsomal incubation media may yield enzyme kinetic estimates more comparable with in vivo results.
Numerous attempts have been made to predict in vivo pharmacokinetics and drug-drug interactions on the basis of in vitro hepatic microsomal data (for reviews, see Bertz and Granneman, 1997; von Moltke et al., 1998). However, the utilization of human microsomal data for predicting the in vivo situation has remained uncertain and controversial. Factors such as the in vitro incubation conditions, extrahepatic drug metabolism, and transporters related to drug absorption and disposition can confound the in vitro-in vivo extrapolation.
The addition of bovine serum albumin to the microsomal incubation medium has decreased the Km estimates and increased the intrinsic clearance (CLint) of phenytoin p-hydroxylation, a reaction mainly catalyzed by cytochrome P-450 CYP2C9, yielding predicted clearance values more comparable with the in vivo values (Ludden et al., 1997; Carlile et al., 1999). However, because the albumin concentrations used in microsomal incubations have been unphysiologically high (comparable with serum levels), a recent consensus on the conduct of studies of metabolic and transport interactions warranted further studies to resolve whether albumin should be added to microsomal incubation mixtures (Tucker et al., 2001). In another study, phenytoin oxidation in human liver microsomes was substantially promoted by the addition of rat liver cytosol (Komatsu et al., 2000b). Both albumin and cytosolic components are probably present at the metabolic enzyme site in vivo (Shroyer and Nakane, 1987; Komatsu et al., 2000b). However, it appears that there are no published studies that have assessed the usefulness of adding both human serum albumin (Hsa) and human liver cytosol (Hlc) to microsomal incubation media for predicting the metabolic clearance of substrates of cytochrome P-450 isoforms. Furthermore, it is unclear how adding Hsa or Hlc affects inhibition of cytochrome P-450-mediated reactions.
In this study, the effect of addition of a low concentration of albumin and cytosol to microsomal incubation medium on the enzyme kinetic estimates of the formation of hydroxytolbutamide (a marker reaction for CYP2C9) and its utility in predicting the in vivo partial metabolic clearance (CLmet) of tolbutamide were investigated. In addition, the effects of cytosol and albumin on the inhibitory effect of gemfibrozil on tolbutamide hydroxylase activity (Wen et al., 2001a) were examined using human liver microsomes.
Experimental Procedures
Materials.
Gemfibrozil was obtained from Orion Pharma (Espoo, Finland). Hsa (analytical grade, fatty acid-free), tolbutamide, and NADPH were purchased from Sigma-Aldrich (St. Louis, MO). Hydroxytolbutamide was purchased from Ultrafine Chemicals (Manchester, UK). Pooled human liver microsomes (prepared from five male and five female human liver microsomal samples) containing a representative activity of CYP2C9 were obtained from Gentest Corp. (Woburn, MA). Pooled Hlc, prepared from five male and five female human liver samples, was obtained from Tebu International Co. (Paris, France). Other chemicals and reagents were obtained from Merck (Darmstadt, Germany).
Binding Studies with Tolbutamide and Gemfibrozil.
Unbound fractions of tolbutamide and gemfibrozil in incubation buffers containing microsomal protein (0.1 mg/ml), Hsa (0.5–5 mg/ml), and Hlc (0.5–5 mg/ml) were separated by ultrafiltration (Sallustio et al., 1997; Carlile et al., 1999). Various concentrations of tolbutamide (12–14 concentrations, 5–1000 μM; dissolved in ethanol, final concentration 1%) and gemfibrozil (4 concentrations, 10–100 μM; dissolved in acetonitrile; residue reconstituted after evaporation to dryness, using incubation buffers) were prepared in each of the media used in the microsomal incubations, except that without NADPH. After incubation for 30 to 120 min (the time courses used were validated by pilot tests to attain equilibrium), the sample was centrifuged for 30 min at 37°C in a micropartition filter (Centrifree, part number 4104; Millipore Corporation, Bedford, MA). After the centrifugation, an aliquot of the filtrate was subjected to analysis of the unbound drug concentration by high-performance liquid chromatography (HPLC) as described below.
Assays for Tolbutamide Hydroxylation in Vitro.
The assay method for tolbutamide hydroxylase activity was similar to that of a previous study (Wen et al., 2001a) except that tolbutamide (5–1000 μM) was preincubated with the incubation medium containing 0.1 M sodium phosphate buffer (pH 7.4) and 5 mM MgCL2at 37°C for 30 min in the absence of Hsa and Hlc, in the presence of Hlc (0.5–5 mg/ml), and for 120 min in the presence of Hsa (0.5–5 mg/ml) and Hlc plus Hsa. The reaction was initiated by the addition of 1.0 mM NADPH followed by 0.1 mg/ml microsomes. The time of incubation and the concentration of microsomal protein used in the assay were determined to be in the linear range for the rate of hydroxytolbutamide formation. After incubation for 60 min, the reactions were terminated by adding 20 μl of phosphoric acid (85%) to precipitate the proteins. After centrifugation at 10,000g for 5 min, an aliquot of the supernatant was subjected to analysis of hydroxytolbutamide by HPLC, as described below. All our results represent the average of duplicate determinations, and if there was more than a 10% difference between the duplicates, the samples were discarded and the experiments were repeated.
Inhibition Studies.
The incubation conditions used to study the effect of gemfibrozil on CYP2C9 activity were adapted from a previous report (Wen et al., 2001b) with some modifications. In brief, gemfibrozil was dissolved in acetonitrile. After evaporation of acetonitrile to dryness, gemfibrozil was reconstituted in an incubation medium (final concentrations 0–100 μM) containing 0.1 M sodium phosphate buffer (pH 7.4), 5 mM MgCL2, tolbutamide (5–250 μM), and 5 mg/ml Hsa and/or 0.5 mg/ml Hlc. Otherwise, the incubations were carried out as described above. All incubations were performed in duplicate.
HPLC Analysis.
Assays for tolbutamide, hydroxytolbutamide, and gemfibrozil were carried out using HPLC, as described previously (Hsu et al., 1991; Ko et al., 1997). The intraday and interday coefficients of variation were <7% at relevant concentrations (n = 6).
Analysis of Data.
The apparent total (unbound) kinetic parameters for the formation of hydroxytolbutamide [Km,Vmax(Km,u,Vmax,u)], and the apparent inhibitory constant (Ki) values of gemfibrozil were determined by nonlinear regression analysis using Systat for Windows 6.0.1 (SPSS Inc., Chicago, IL). An assessment of goodness of fit of the models was made using the size of the residual sum of squares, the random distribution of the residuals, the standard error, and the 95% confidence interval of the parameter estimates. When necessary, an F test was performed to determine whether there was a significant difference in the size of the residual sum of squares between models (Motulsky and Ransnäs, 1987). A one-enzyme Michaelis-Menten model was found to be the best fit enzyme model relating hydroxytolbutamide formation rates to the concentrations of either total or unbound tolbutamide:
Statistical Analysis.
A t test was used to compare the observed in vivo metabolic clearance of tolbutamide with that predicted from literature microsomal data. The unbound fractions of tolbutamide and gemfibrozil in incubation media in the presence or absence of 5 mg/ml Hsa and 0.5 mg/ml Hlc were compared using the Wilcoxon test.
Scaling Microsomal Parameters to in Vivo Hepatic Clearance.
The total (unbound) intrinsic clearance [CLint(CLint,u)] values were calculated usingVmax/Km(Vmax,u/Km,u). The CLint (CLint,u) values were scaled to in vivo using the standard scaling factors of 45 mg of microsomal protein per gram of liver and 20 g of liver per kilogram of body weight for humans (Obach, 2000). The CLh values were calculated using the well stirred model.
Results
Comparison of Observed Values of in Vivo Metabolic Clearance of Tolbutamide with Those Predicted from Previous Microsomal Studies.
On the basis of a MEDLINE search, six separate in vitro studies including 20 individual human liver microsomal samples on the in vitro enzyme kinetic estimates of the formation of hydroxytolbutamide were summarized and listed in Table 1. The inclusion of these studies was based on the following criteria: the incubation conditions (the buffer, ionic strength, microsomal protein concentration used, and incubation time) were comparable with the standard incubation conditions and were identical to each other; a larger than 10-fold substrate concentration range was used in each study; and the kinetic parameters were estimated by a nonlinear regression analysis using a single Michaelis-Menten equation. The predicted CLh values of tolbutamide based on these in vitro data were significantly (P < 0.0001) lower than the values of actual metabolic clearance (CLmet) of tolbutamide, adopted from a clinical pharmacokinetic study including 10 healthy subjects (Table 1).
Binding Studies.
At a spiked concentration of 50 μM, tolbutamide and gemfibrozil showed essentially no binding with the 0.1 mg/ml microsomal protein in incubation medium containing microsomal protein (Table 2). The addition of 0.5 to 5 mg/ml Hsa and Hlc to the incubation medium concentration dependently decreased the unbound fractions of tolbutamide and gemfibrozil; the effects of Hlc were much smaller than those of Hsa (Table 2). At a fixed concentration of 0.5 mg/ml, Hlc had no statistically significant effect on the unbound fractions of tolbutamide (5–250 μM) and gemfibrozil (10–100 μM) (P > 0.05; Figs.1 and 2). However, at a fixed concentration of 5 mg/ml Hsa, the unbound fractions of tolbutamide (5–250 μM) and gemfibrozil (10–100 μM) increased with increasing drug concentrations (Figs. 1 and 2). The coaddition of Hsa and Hlc yielded no apparent difference (P > 0.05) in the unbound fractions of tolbutamide and gemfibrozil compared with Hsa alone (Figs. 1 and 2).
Effects of Hsa and Hlc on the Enzyme Kinetics of Tolbutamide Hydroxylation.
The Michaelis-Menten and Eadie-Hofstee plots were consistent with a single enzyme playing a predominant role in the formation of hydroxytolbutamide in all incubation mixtures (Fig.3). The apparentKm,u andVmax,u values in standard incubation medium were 123 μM and 282 pmol/mg/min, respectively (Table3). The predicted CLh,u value for tolbutamide using these estimates (0.06 ml/min/kg) was 40% of the actual in vivo data (0.15 ml/min/kg) (Table 3), agreeing well with the previous in vitro data (Table 1). The addition of 5 mg/ml Hsa or 0.5 mg/ml Hlc to the microsomal incubation medium substantially changed the unbound apparent kinetic estimates of the formation of hydroxytolbutamide (Table 3), making the predicted CLh,u values more accurate (0.11 or 0.09 ml/min/kg, respectively). The coaddition of Hsa and Hlc yielded the predicted CLh,u value (0.14 ml/min/kg) closest to the in vivo data (0.15 ml/min/kg).
Effects of Hsa and Hlc on Inhibition of CYP2C9 by Gemfibrozil.
At concentrations of 5 to 100 μM, gemfibrozil was an effective inhibitor of tolbutamide hydroxylase, with an IC50 value of 13 μM under standard incubation conditions. The IC50 value based on total drug concentrations was substantially increased by the addition of 5 mg/ml Hsa or 0.5 mg/ml Hlc to the incubation medium (67 or 24 μM, respectively) and more than 8-fold (>100 μM) by the coaddition of Hsa and Hlc (Fig. 4). A similar change was seen in the Ki of gemfibrozil when total drug concentrations were used (Fig.5; Table4). However, when the unbound substrate and inhibitor concentrations were considered, theKi (6 μM under standard conditions) was not significantly altered by the addition of Hsa (4 μM), Hlc (8 μM), or both Hsa and Hlc (9 μM) (Table 4).
Discussion
In the current study, a single enzyme Michaelis-Menten model described well the formation of hydroxytolbutamide in all experiments, consistent with previous findings about the kinetics of tolbutamide metabolism in human liver microsomes (Miners et al., 1988; Doecke et al., 1991; Hickman et al., 1998; Hemeryck et al., 1999; Shin et al., 1999; Komatsu et al., 2000a). The observed apparentKm (123 μM) andVmax (282 pmol/mg/min) values in pooled human liver microsomes were comparable with the mean values of published reports (Km = 135 μM andVmax = 257 pmol/mg/min) (Table 1). Both CYP2C9 and CYP2C19 are involved in the formation of hydroxytolbutamide in human liver microsomes, with CYP2C9 being the predominant isoform and CYP2C19 being the minor isoform (∼14–22%) (Wester et al., 2000). However, CYP2C19 possesses aKm value for the formation of hydroxytolbutamide similar to that of CYP2C9 (Lasker et al., 1998), and thereby, the kinetics of the formation of hydroxytolbutamide obviate biphasic behavior. Accordingly, the observed kinetic estimates of tolbutamide metabolism may, to some extent, still reflect a minor contribution of CYP2C19.
The comparison of the observed in vivo metabolic clearance of tolbutamide with the predicted values based on published in vitro studies using standard incubation conditions showed that the predicted mean CLh value of 0.059 ml/min/kg was significantly lower than the observed value of 0.15 ml/min/kg, in vivo (P < 0.0001). A similar underprediction phenomenon has been observed also with other CYP2C9 substrates. In two recent in vitro studies, the clearance estimates of phenytoin oxidation, based on in vitro data determined using standard microsomal incubation conditions, were substantially lower than actual in vivo values (Ludden et al., 1997; Carlile et al., 1999).
Substrate-nonspecific binding to incubation matrices has been suggested as the major cause of the underprediction of pharmacokinetic estimates (Venkatakrishnan et al., 2000). However, our present findings, and those of Ludden et al. (1997) and Carlile et al. (1999), suggest that nonspecific binding may not be the major cause for the underprediction in the case of the CYP2C9 substrates phenytoin and tolbutamide. At the 0.1 mg/ml microsomal protein concentration used in the current study, the microsomal binding of tolbutamide was negligible, i.e., its consideration in the calculation could not improve the prediction. Similarly, Carlile et al. (1999) found that, even when using the unbound phenytoin concentration in the microsomal incubation medium, the estimated mean Km and CLint estimates of phenytoin were considerably different from those of the actual in vivo values (20 μM versus 2–3 μM, and 0.031 liter/min/70 kg versus 0.28 liter/min/70 kg, respectively).
Addition of 5 mg/ml (0.5%) Hsa and 0.5 mg/ml Hlc to the microsomal incubation medium substantially increased the unbound CLint of tolbutamide. Consequently, the predicted CLh of tolbutamide was distinctly improved. In line with these findings, Ludden et al. (1997) reported that the addition of 4% bovine serum albumin promoted the microsomal metabolism of phenytoin, resulting in a mean unboundKm value of 2.2 μM, comparable with that of the in vivo value (2–3 μM). Similarly, a distinct improvement in the prediction of the CLint of phenytoin was observed after the addition of 2% bovine serum albumin (Carlile et al., 1999). In the same study, Carlile et al. (1999) also observed a considerable promotion of microsomal tolbutamide hydroxylation by the addition of 2% bovine serum albumin, but the predicted CLint value of tolbutamide (0.59 l/min/kg) was about 300% higher than the actual in vivo data (0.14 l/min/kg). A higher albumin concentration (2%) was used in the study by Carlile et al. (1999) than in ours (0.5%).
Albumin is the most abundant protein synthesized by hepatocytes and is localized in the rough endoplasmic reticulum and Golgi complexes in hepatocytes (Shroyer and Nakane, 1987). The concentration of albumin in cultured rat hepatocyte S9 fractions (containing the cytosolic and microsomal fractions) has been about 65 mg/g protein, i.e., 6.5% of total protein weight (Milosevic et al., 1999). Although the real in vivo concentration of albumin around the metabolic enzyme site is not known, it is, probably, clearly lower than the concentration of albumin in serum (about 4%). Thus, the addition of 0.5% Hsa to in vitro microsomal incubations (as in our study) may more accurately reflect the physiological conditions in vivo than the higher bovine serum albumin concentrations used in previous studies.
The mechanism underlying the enhancing effects of albumin and cytosol on microsomal drug metabolism (Mori et al., 1984; Ludden et al., 1997;Carlile et al., 1999; Komatsu et al., 2000b) is unclear. In the current study, Hsa and Hlc mainly decreased theKm of tolbutamide hydroxylation without much effect on the Vmax, suggesting that the binding affinity between the substrate and the enzyme may be increased by albumin and cytosol. A possible explanation is that albumin and cytosol components may act by binding to the enzyme, or by sequestering some endogenous compounds that inhibit CYP2C9. In any case, the mechanism of stimulation of CYP2C9 by cytosol seems not to be related to binding of the substrate by cytosol, because the 0.5 mg/ml cytosol concentration did not change the unbound fraction of tolbutamide.
When total concentrations of tolbutamide and gemfibrozil were used in the calculations, Hsa substantially, and Hlc to a smaller extent, increased the IC50 and Kivalues of gemfibrozil for the CYP2C9-mediated reaction. However, when the unbound substrate and inhibitor concentrations were considered in the estimation, the differences among theKi values were only minimal. These results indicate that the unbound fraction of gemfibrozil contributes primarily to the gemfibrozil-mediated CYP2C9 inhibition in vitro. Interestingly, although the addition of Hsa may increase the binding affinity of tolbutamide to the enzyme (as suggested by theKm values), it did not change the unbound Ki value of gemfibrozil, suggesting that Hsa did not affect or only slightly increased the binding affinity of gemfibrozil for CYP2C9. The good correlation of the in vitro-in vivo enzyme kinetic results suggests that the unbound Ki values in the presence of both Hsa and Hlc (9 μM) may best reflect the real in vivo inhibitory effect of gemfibrozil.
To conclude, addition of Hsa and Hlc to microsomal incubations substantially changed the enzyme kinetic estimates of tolbutamide. The predicted value of the in vivo CLh of tolbutamide, based on the coaddition of Hsa and Hlc to the incubations, showed a good agreement with the actual in vivo clearance values. However, addition of Hsa and Hlc to the microsomal incubations only slightly altered the in vitro inhibitory effect of gemfibrozil on tolbutamide hydroxylase activity when unbound drug concentrations were used. The present findings represent interesting phenomena and warrant further mechanistic explanation. Testing for the prediction of in vivo intrinsic clearance requires further exploration with other substrates and enzymes.
Acknowledgments
We thank Jouko Laitila for skillful technical assistance.
Footnotes
-
↵1 These authors contributed equally to this study.
-
This study was supported by grants from the Helsinki University Central Hospital Research Fund and the National Technology Agency of Finland (TEKES).
- Abbreviations:
- CLint
- intrinsic clearance
- CLh
- hepatic clearance
- CLmet
- partial metabolic clearance
- Hlc
- human liver cytosol
- HPLC
- high-performance liquid chromatography
- Hsa
- human serum albumin
- Received December 17, 2001.
- Accepted March 11, 2002.
- The American Society for Pharmacology and Experimental Therapeutics