Abstract
The human UDP-glucuronosyltransferase (UGT) enzymes UGT1A9 and UGT2B7 play important roles in the hepatic glucuronidation of many drugs. The presence of bovine serum albumin (BSA) during in vitro assays was recently reported to lower the Km values of both these UGTs for their aglycone substrates without affecting the corresponding Vmax values. Nonetheless, using the specific substrates entacapone and zidovudine (AZT) for UGT1A9 and UGT2B7, respectively, and using an improved ultrafiltration method for measuring drug binding to BSA and to biological membranes, we found that the presence of BSA during the glucuronidation reaction leads to a large increase in the Vmax value of UGT1A9, in addition to lowering its Km value. On the other hand, in the case of UGT2B7, our results agree with the previously described effect of BSA, namely lowering the Km value without a large effect on the enzyme's Vmax value. The unexpected BSA effect on UGT1A9 was independent of the expression system because it was found in a recombinant enzyme that was expressed in baculovirus-infected insect cells as well as in the native enzyme in human liver microsomes. Moreover, the effect of BSA on the kinetics of 4-methylumbelliferone glucuronidation by recombinant UGT1A9 was similar to its effect on entacapone glucuronidation. Contrary to the aglycone substrates, the effect of BSA on the apparent Km of UGT1A9 for the cosubstrate UDP-α-d-glucuronic acid was nonsignificant. Our findings call for further investigations of the BSA effects on different UGTs and the inhibitors that it may remove.
Introduction
Human UDP-glucuronosyltransferases (UGTs) play major roles in the metabolic elimination of numerous endo- and xenobiotics. They are membrane enzymes of the endoplasmic reticulum that catalyze glucuronic acid transfer from the cosubstrate, UDP-α-d-glucuronic acid (UDPGA), to nucleophilic groups of chemically diverse substrates. There are 19 functional human UGTs and they are divided into three subfamilies: 1A, 2A, and 2B (Mackenzie et al., 2005). Closely related enzymes that use somewhat different nucleotide cosubstrates have recently been discovered and assigned to subfamily 3A (MacKenzie et al., 2011), but they will not be further considered in this work. Individual UGT isoforms have distinctive substrate and inhibitor selectivity (Miners et al., 2010) and are differentially expressed in various tissues, most notably liver, intestine, and kidney (Ohno and Nakajin, 2009).
Because of partial overlaps in the substrate specificity of individual UGTs and the expression of multiple isoforms in each tissue that expresses these enzymes, in vitro studies on the UGTs and drug glucuronidation are often performed using recombinant human UGTs that are either expressed in insect cells (mainly Spodoptera frugiperda Sf9 cells), or in human embryonic kidney (HEK) 293 cells (Radominska-Pandya et al., 2005). Enzyme kinetic constants (e.g., Km and Vmax) and inhibition (IC50 or Ki) parameters of drug glucuronidation, determined from in vitro assays, are commonly used to estimate the extent of glucuronidation in vivo (Miners et al., 2010). Because there is currently no good method to extract the UGTs from the membrane and purify them as fully active enzymes, in vitro glucuronidation assays are performed with different cell fractions rather than with highly purified enzymes. In such systems, nonspecific substrate binding to the membrane and different proteins within it, as well as the presence of inhibitors within the membrane, can lead to erroneous estimation of UGT activity. The acquired errors may lead to erroneous estimation of in vivo glucuronidation activity, namely poor in vivo-in vitro extrapolation, significantly weakening the ability to predict the pharmacokinetics properties of a drug under development, a problem that was already faced in the case for many therapeutic drugs that are eliminated by glucuronidation (Boase and Miners, 2002; Kilford et al., 2009; Raungrut et al., 2010). Hence, there is a clear need for better understanding of factors that affect the outcome of UGT assays in vitro.
In a series of studies, it was found that adding purified bovine serum albumin (BSA) to assays with several recombinant human UGTs that were expressed in HEK293 cells, or to the native enzymes in human liver microsomes (HLM), can significantly decrease the Km for drugs that are glucuronidated by UGT1A9 and UGT2B7 without affecting the reaction Vmax (Uchaipichat et al., 2006; Rowland et al., 2007, 2008). These authors suggested that long-chain fatty acids (e.g., oleic, linoleic, and arachidonic acid) competitively inhibit the UGTs, and that BSA addition reversed that inhibition by binding the inhibitory fatty acids (Rowland et al., 2007). Because similar inhibition was not observed in cultured hepatocytes (Engtrakul et al., 2005), the authors speculate that the inhibitory fatty acids are released during microsome preparation from either human liver or the cells that were used for recombinant UGT expression.
Raungrut et al. (2010) have recently studied the effect of BSA on codeine glucuronidation by the recombinant UGT2B4 and UGT2B7 that were expressed in insect cells. However, thus far, this is the only study that examined the BSA effect on UGTs that were expressed in insect cells, even if most of the research on different aspects of the UGTs is currently conducted using such recombinant enzymes because the commercial UGTs are expressed in insect cells. There are differences in lipid composition between the HEK293 cells, insect cells, and HLM (Marheineke et al., 1998), but it is unclear if these differences are “translated” into differences in the BSA effect on individual UGTs. Our original goal was to investigate the effect of BSA on the activities of recombinant UGTs that were expressed in insect cells. As reported below, the obtained results led us also to reexamine the earlier reports about the native UGT enzymes in HLM, particularly the BSA effect on UGT2B7 and UGT1A9. Although the BSA effect does not appear to be dependent on the expression system, the new findings should raise general awareness about factors that can influence UGT assays in vitro and the complexity of the BSA effect. They may also be instrumental for better understanding of the glucuronidation reaction mechanism and how it may be inhibited as well as for better predictability of the in vitro assays.
Materials and Methods
Compounds and Reagents.
4-Methylumbelliferone (>99%, CAS 90-33-5), UDPGA (triammonium salt, 98–100%, CAS 63700-19-6), alamethicin (>90%, CAS 27061-78-5), 4-methylumbelliferone-β-d-glucuronide (4-MU; ≥98%, CAS 6160-80-1), zidovudine (3′-azido-3′-deoxy-thymidine, ≥98%, CAS 30516-87-1), sodium phosphate monobasic dihydrate (≥99%, CAS 13472-35-0), and BSA (≥96%, CAS 9048-46-8, essentially fatty acid free, <0.004%) were purchased from Sigma-Aldrich (St. Louis, MO). Entacapone (batch 1044842) was a generous gift from Orion Corporation (Espoo, Finland). Entacapone-β-d-glucuronide was synthesized in our laboratory (Luukkanen et al., 1999). Tween 20 (CAS 9005-64-5) and Tween 80 (CAS 9005-65-6) were purchased from Acros Organics (Fairlawn, NJ). Magnesium chloride hexahydrate and perchloric acid were obtained from Merck (Darmstadt, Germany). Formic acid (98–100%) was from Riedel-deHaën (Seelze, Germany). Disodium hydrogen phosphate dihydrate was purchased from Fluka (Buchs, Switzerland). Radiolabeled [14C]UDPGA was acquired from PerkinElmer Life and Analytical Sciences (Waltham, MA). High-performance liquid chromatography (HPLC)-grade solvents were used throughout the study.
Enzyme Sources.
Recombinant human UGT2B7 and UGT1A9 were expressed as His-tagged proteins in baculovirus-infected Sf9 insect cells as described previously (Kurkela et al., 2007). The collected cells were osmotically lysed and the suspension was centrifuged at 41,000g for 2 h. The resulting pellets were homogenized, suspended in 25 mM Tris-HCl buffer (pH 7.5) and 0.5 mM EDTA and stored in aliquots at −70°C until use (Kurkela et al., 2003).
Control samples, insect cell membranes without any human UGT, were prepared by infecting insect cells with baculovirus that does not encode any human UGT and then treating the cells as described above. Pooled HLM (lot 18888) and recombinant UGT1A9 “supersomes” (lot 81661; expressed in Sf9 insect cells) were purchased from BD Gentest (Woburn, MA). Protein concentrations were determined by the BCA protein assay (Thermo Fisher Scientific, Waltham, MA).
Drug Binding Assays.
We developed an ultrafiltration method to measure the binding of AZT, entacapone, and 4-MU to BSA, control insect cell membranes, and HLM. The technical component of the assay and basic calculations to determine the drug fraction that is bound to the device are described under Nonspecific Binding to the Filter Device and Filter Pretreatment, whereas the results for the different drugs and the determination of the unbound drug fraction under different conditions are described under Results. The filter devices for the drug binding assays were Amicon Ultra filters with 10-kDa Ultracel regenerated cellulose membrane, 500 μl volume, and they were purchased from Millipore Corporation (Billerica, MA).
Nonspecific Binding to the Filter Device and Filter Pretreatment.
A sample of the compound in phosphate buffer (500 μl solution, 50 mM, pH 7.4) was transferred to a filter device and centrifuged twice at 2500g, 1 min each time, to collect two separate 50-μl filtrate fractions. In total, a maximal volume of 100 μl was allowed to pass through the filter, namely ≤20% of the total loaded volume. The first filtrate sample was removed, and a 30-μl aliquot from the second 50-μl filtrate fraction, as well as a similar sized sample from the prefiltered solution, were collected, each mixed with 60 μl of 4 M perchloric acid/methanol (MeOH) (1:5 mix) and submitted to ultra-performance liquid chromatography (UPLC) analysis. The experiments were performed in triplicate, and the nonspecific binding to the filter device was calculated using the following equation:
in which [S] is the substrate concentration in the respective solutions.
The nonspecific binding to the filter device (NSBf) was significantly lowered by the following pretreatment: filter device wash twice with 400 μl of 1% Tween 20 and removal of the remaining detergent solution by 5 min of centrifugation at 5000g and a subsequent wash with 500 μl of phosphate buffer (50 mM, pH 7.4).
The integrity of the filter device membrane was tested after the pretreatment by filling the device with 450 μl of either 2% BSA solution or 1 mg/ml of control insect cell membranes, centrifuging for 10 min at 5000g, transfer of a 200-μl aliquot of the resulting filtrate to a 1.5 ml-centrifuge tube, acidification of the sample with 20 μl of 4 M perchloric acid, transferring to ice for 20 min, centrifuging for 10 min at 16,000g, and visually inspecting the centrifuge tube for protein precipitates.
Determination of Substrate Binding to BSA, HLM, and Control Insect Cell Membranes.
The substrate of interest was first incubated with BSA, HLM, or insect cell membranes in phosphate buffer (50 mM, pH 7.4) in a total volume of 500 μl for 60 min at 37°C. The solution was then transferred to the filter device and centrifuged twice for 1 min at 2500g. In total, a maximal volume of 100 μl was allowed to pass through the filter (≤20% of total volume). The 30-μl aliquots from the second filtrate fraction and from the prefiltered solution were each mixed with 60 μl of 4 M perchloric acid/MeOH (1:5 mix), transferred to ice for 20 min, centrifuged for 10 min at 16,000g, and then submitted to substrate concentration determination by a UPLC analysis. The experiments were performed in triplicates.
Drug Glucuronidation Assays.
Stock solutions of AZT, entacapone, and 4-MU were prepared in methanol and diluted with methanol to the desired concentrations immediately before use. Appropriate amounts of these dilutions were transferred into 1.5-ml centrifuge tubes and the solvent was evaporated in vacuo at ambient temperature. The solid residues were dissolved in the reaction mixture containing phosphate buffer (50 mM, pH 7.4), MgCl2 (10 mM), BSA (0–2%), and an enzyme source (0.02–0.2 mg/ml total protein in the membrane, depending on the enzyme source) to a final volume of 100 μl.
The reaction mixtures for the HLM assays also contained alamethicin at a final concentration of 5% of the microsomal protein concentrations, and they were placed on ice for 30 min (Fisher et al., 2000) before continuing as with the recombinant UGT-containing samples. Alamethicin was not added to the incubations with recombinant UGTs because it has no significant effect on the glucuronidation activity of such samples (Zhang et al., 2011).
The reaction mixtures (in the case of HLM, after the preincubation with alamethicin) were incubated first for 30 min at room temperature, followed by 5 min at 37°C, initiated by the addition of UDPGA to a final concentration of 5 mM, and performed at 37°C (15–60 min) protected from light. Negative controls, including without UDPGA, without substrate, or with control (“empty”) insect cell membranes, were performed for each set of assays. The 4-MU glucuronidation reactions were terminated by the addition of 10 μl of ice-cold perchloric acid (4 M). In the cases of AZT and entacapone, the reactions were terminated by the addition of 60 μl of ice-cold 4 M perchloric acid/MeOH (1:5 mix). After reaction termination, the tubes were transferred to ice for 30 min and then centrifuged at 16,000g for 10 min. Aliquots of the resulting supernatants were transferred to dark glass vials and subjected to HPLC or UPLC analyses.
Analytical Methods.
The HPLC system consisted of an Agilent 1100 series degasser, binary pump, 100-vial autosampler, thermostated column compartment, multiple wavelengths UV detector, and fluorescence detector (Agilent Technologies, Santa Clara, CA). The resulting chromatograms were analyzed with Agilent ChemStation software (revision B.01.01) on Windows XP Professional software (Microsoft, Redmond, WA). For separation and detection of 4-MU-β-d-glucuronide, we used a Chromolith SpeedROD RP-18e (50 × 4.6 mm, 3 μm; Merck) column (at a column temperature of 40°C and injection volume of 20 μl). The mobile phase consisted of 80% 50 mM phosphate buffer, pH 3 (A) and 20% methanol (B) at a constant flow rate of 2 ml/min. A fluorescence detector with an excitation wavelength of 316 nm and emission wavelength of 82 nm was used for detection. The retention time of 4-MU-β-d-glucuronide under these conditions was 1.45 min. The quantification was based on a standard curve prepared using an authentic glucuronide standard.
The UPLC system consisted of a Waters Acquity UPLC (Waters, Milford, MA) system equipped with an Acquity UPLC BEH C18 column (2.1 × 100 mm, 1.7 μm; Waters) and a precolumn (column temperature of 40°C), column manager, sample manager, binary solvent pump, and photodiode array UV detector. The UV detector was equipped with high-sensitivity 2.4-μl flow cell. The resulting chromatograms were analyzed with Empower 2 software (Build 2154; Waters) on a Windows XP Professional operating system. We developed UPLC methods to separate and detect zidovudine-β-d-glucuronide and entacapone-β-d-glucuronide on the basis of their UV absorbance as well as zidovudine, entacapone, and 4-MU substrates for analyses of the binding assay results. The injection volume was 10 μl for all samples.
For the separation of AZT-β-d-glucuronide, the mobile phase consisted of 0.1% formic acid (A) and acetonitrile (B) and the flow rate was 0.6 ml/min. UV absorbance at 267 nm was used for detection. The gradient in this method was as follows: 0 to 1 min of 5% B, 1 to 4 min of 5 to 30% B, 4.5 to 5 min of 30 to 80% B, and 5 to 6 min of 5% B. The AZT-β-d-glucuronide retention time was 2.40 min. The quantification of AZT-β-d-glucuronide was based on a standard curve constructed using the UV absorption of zidovudine.
For the separation of entacapone-β-d-glucuronide, the mobile phase consisted of 50 mM phosphate buffer, pH 3 (A), and acetonitrile (B), and the flow rate was 0.5 ml/min. UV absorbance at 309 nm was used for detection, and the quantification was done using a standard curve made with an authentic glucuronide standard. The gradient in this method was as follows: 0 to 3 min of 20 to 30% B, 3 to 3.2 min of 30 to 80% B, 3.2 to 4 min of 80% B, 4 to 4.1 min of 80 to 20% B, and 4.1 to 6 min of 20% B. The entacapone-β-d-glucuronide retention time was 2.18 min.
A single UPLC method was developed for separation and analysis of AZT, entacapone, and 4-MU in drug binding assays. The mobile phase consisted of 50 mM phosphate buffer, pH 3 (A), and acetonitrile (B), and the flow rate was 0.6 ml/min throughout. UV absorbance at 267, 309, and 321 nm was used for detection of AZT, entacapone, and 4-MU, respectively. The run was isocratic with 35% B for 1.5 min. The retention times of AZT, entacapone, and 4-MU and were 0.72, 0.87, and 1.32 min, respectively. The quantification was based on standard curves prepared with the respective compounds.
Glucuronidation activities are reported as the average and S.E. of at least three replicate determinations. Please note that because of the lack of suitable isoform-specific anti-UGT antibodies, the glucuronidation rates and Vmax values of the different recombinant UGTs and HLM cannot be compared directly.
Enzyme Kinetic Analyses.
The protein concentrations and incubation times for the kinetic analyses reactions were selected based on preliminary assays to ensure that product formation was within the linear range with respect to protein concentration and time, and that the substrate consumption during the reaction was less than 10%. The substrate concentration ranges for AZT, entacapone, and 4-MU enzyme kinetic experiments were 50 to 2000, 5 to 750, and 5 to 500 μM, respectively. The UDPGA enzyme kinetic assays were performed with either 75 μM entacapone or 50 μM 4-MU as the aglycone substrate. The incubation times varied from 15 to 60 min.
The enzyme kinetic parameters were obtained by fitting kinetic models to the experimental data using GraphPad Prism version 5.01 for Windows (GraphPad Software Inc., San Diego, CA). The best model was selected based on the corrected Akaike's information criterion, the calculated r2 values, residuals graph, parameter S.E. estimates, 95% confidence intervals, and visual inspection of the Eadie-Hofstee plots. In assays containing BSA, the free substrate concentrations (fu, or fraction unbound), were corrected according to the estimated drug binding to BSA under the specific conditions of each glucuronidation assay. Data were fitted with the following models:
Michaelis-Menten equation
where v is the initial velocity of the enzyme reaction, Vmax is the maximal velocity, [S] is the substrate concentration, and Km is the Michaelis-Menten constant (concentration of substrate at 0.5 of Vmax).
Substrate inhibition equation
where Ki is the constant describing the substrate inhibition interaction.
Allosteric sigmoidal model (Hill equation)
where S50 is the concentration of substrate at 0.5 of Vmax (analogous to Km in the Michaelis-Menten model) and h is the Hill coefficient.
Two-site biphasic model equation (Korzekwa et al., 1998)
where Vmax1 and Km1 are estimated from the curved portion of the plot at lower substrate concentrations. The CLint represents the ratio of Vmax2/Km2 and describes the linear portion of the plot exhibited at higher substrate concentrations.
Two-sites model equation (Houston and Kenworthy, 2000)
where Ks is a substrate dissociation constant, α describes the change in substrate binding affinity for the second enzyme site, and β describes the change in rate of product formation from the substrate-enzyme-substrate (S · E · S) complex compared with the enzyme-substrate (E · S) complex.
Results
Drug Binding to the Filter Device, BSA and Insect Cell Membranes, and HLM.
A prerequisite for correct interpretation of the BSA effect on enzyme kinetics is being able to determine the concentration of free substrate (the drug) in the presence of BSA, the so-called fraction unbound (fu). Rowland et al. (2007) have used an equilibrium dialysis device for this, and in the current study, in the absence of such an instrument, we developed an ultrafiltration method. To obtain values of fu that are as accurate as possible, we took into consideration the amount of drug that binds to the filter device even in the absence of BSA, the so-called nonspecific binding (NSBf). The method is detailed in the Materials and Methods, and examples of the obtained results with an untreated filter device for 20 μM AZT, entacapone, and 4-MU solutions are NSBf values of 20, 99, and 40%, respectively. In an effort to decrease the NSBf, we tested several ways to decrease that nonspecific binding by a suitable filter pretreatment. The best results were achieved by a double wash with a solution of the mild detergent Tween 20, 1% final concentration, followed by phosphate buffer rinse, as described under Materials and Methods. This pretreatment reduced the NSBf of 20 μM AZT, entacapone, and 4-MU solutions to the acceptable values of 1, 27, and 7%, respectively.
We further examined the effect of drug concentration on its NSBf and found that for entacapone (the most lipophilic compound among the tested drugs in this study) it is saturable in nature and exponentially decreases with increasing entacapone concentration (Fig. 1A, inset). To obtain a good estimation of the NSBf of entacapone at any given entacapone concentration, the determined binding values were fitted to the following exponential decay (empirical) equation:
where NSBf-max is the maximal measured NSBf, NSBf-min is the minimal measured NSBf at the plateau region, k is the exponential decay rate constant, and [S] is the concentration of entacapone.
Binding of entacapone to 0.1% BSA (A) and binding of 4-MU to 0.1 and 1% BSA (B). The NSBf of entacapone and 4-MU is presented as figure insets. The results are presented as fu and represent mean ± S.E. (n = 3). The entacapone binding data were fitted to a empirical hyperbolical equation; the entacapone NSBf data were fitted with an empirical exponential decay equation; 4-MU binding data to 0.1 and 1% BSA were fitted with empirical exponential association equation and linear equation, respectively; and the 4-MU NSBf data were fitted with linear equation (see Materials and Methods for all details).
Unlike entacapone, the NSBf of 4-MU was essentially concentration independent. Because of this, for calculating its fu, we took the mean value of the obtained data points (fu = 7%). The NSBf of AZT to pretreated filter devices was negligible.
After the clarification of the NSBf and its dependence on drug concentration, we turned to the determination of drug binding to BSA. To avoid impractically high fu values, particularly for entacapone, we tested the effects on enzyme kinetics of lower BSA concentrations than 2%, the value used in the previous studies (Rowland et al., 2007, 2008). It turned out that the presence of as low as 0.1% BSA in the reaction mixture was sufficient to yield the stimulatory effect (see below) and, therefore, for entacapone we mainly used 0.1% BSA. In the case of 4-MU, the kinetic experiments were performed in the presence of either 0.1 or 1% BSA, whereas 1% BSA was used for the AZT studies to make them more comparable to the previously published studies.
The binding of AZT, entacapone, and 4-MU to BSA and the different biological membranes (enzyme sources) was studied in the concentration ranges of 5 to 2000, 5 to 750, and 5 to 500 μM, respectively. The fu was calculated by the following equation:
where [S]filtrate is the drug concentration in filtrate, [S]prefilter is its concentration in the prefiltered solution, and NSBf is the nonspecific binding to the filter in the absence of BSA and/or an enzyme source.
The fu of entacapone in the presence of 0.1% BSA increased hyperbolically with increasing entacapone concentration and, therefore, the obtained values were fitted to the following empirical equation that appears to provide a good description of the results:
where fu-max is the maximal fraction unbound achieved in the presence of an unlimited amount of entacapone and Ks is the concentration of entacapone at half fu-max. The binding of entacapone to 1% BSA was also tested, but because it was excessive and the estimated fu was well below 1% at low entacapone concentrations (data not shown), all subsequent experiments on the effect of BSA on entacapone glucuronidation were performed in the presence of 0.1% BSA.
The fu of 4-MU in the presence of 0.1% BSA increased curvilinearly with increasing 4-MU concentration, and the results were well described by the following empirical equation of exponential association:
In this equation, fu-min is the minimal measured fraction unbound, fu-max is the maximal fu, k is an association rate constant, and [S] is the concentration of 4-MU. The fu of 4-MU at 1% BSA increased linearly with increasing 4-MU concentration.
The binding of AZT to either 0.1 or 1% BSA was negligible (≤1%), a result that is in good agreement with the previously reported binding properties of this drug (Rowland et al., 2007).
The dependence of entacapone and 4-MU binding to BSA may also be presented in the form of a binding plot (supplemental text and Supplemental Fig. 1). These analyses revealed that the binding of entacapone and, to a lesser extent, of 4-MU is biphasic, suggesting that two or more binding sites on albumin may be involved. Because of this finding, we further tested the binding of entacapone to 0.1% BSA in the presence of 4-MU. However, the results indicated that the fu of entacapone is only modestly increased in the presence of 4-MU, suggesting minor binding competition between the two substrates (Supplemental Fig. 2).
The binding of AZT, entacapone, and 4-MU to the biological membranes that carry the tested UGTs was tested using HLM and control insect cell membranes. The results indicated that up to 0.2 mg/ml (total protein), the highest protein concentration used in this work, the binding of the three tested drugs is very low. However, it was noted that binding of entacapone to higher than 0.5 mg/ml enzyme source was considerable and should be taken into account if such high concentrations of an enzyme source are used (data not shown).
We also studied drug binding to BSA in the presence of either insect cell membranes or HLM. It was surprising to note that the results indicated that the presence of high amounts of an enzyme source lowers entacapone binding to 0.1% BSA, whereas 4-MU binding to either 0.1 or 1% BSA was not significantly affected by the presence of insect cell membranes. Although the effect of membrane presence on entacapone binding to BSA was not large, it might be significant when more membranes (enzyme source) are added to the reaction mixture. We thus interpolated an empirical three-dimensional (3D) function over the experimental data points (Fig. 2B) that enables estimation of entacapone binding to 0.1% BSA at any given concentration of entacapone and enzyme source used during an in vitro assay. Scatchard plots for the binding of entacapone for either BSA or insect cell membranes, as well as to different combinations of the two, were also drawn from the data presented in Fig. 2. They show that although the character of entacapone binding to BSA remains biphasic, the addition of enzyme source decreases the apparent affinity of entacapone for albumin in a concentration-dependent manner (Supplemental Fig. 3).
Binding of entacapone to 0.1% BSA without Sf9 control membrane (●), and in the presence of 0.032 (▴), 0.080 (▾), and 0.16 (♦) mg/ml of Sf9 control membrane (A). The results are presented as fu and represent mean of 3 determinations. The S.E. was very small and for the sake of clarity, in this condensed figure the S.E. bars were left out. The data were fitted to an empirical hyperbolical equation. The correlation between measured fu and Sf9 control membrane concentration at different entacapone concentrations is presented as the inset in panel A. An identical set of data are presented in the form of a 3D scatter (B). An empirical 3D function was fitted to data points (see supplemental material for all details).
We also tested if other reaction components, such as MgCl2 and UDPGA, affect drug binding. The results indicated that neither MgCl2 nor UDPGA significantly changes the binding of all of the tested compounds to BSA or the enzyme sources.
Enzyme Kinetics of AZT Glucuronidation with HLM and UGT2B7.
After all of the needed experiments and analyses that are described above, we turned to the UGTs and the effect of the presence of BSA on their kinetics. The studies on the effect of 1% BSA on the AZT glucuronidation kinetics by HLM and recombinant UGT2B7, expressed in insect cells, are shown in Fig. 3, and the derived kinetic parameters are presented in Table 1. AZT glucuronidation by HLM and UGT2B7 is best described by the Michaelis-Menten model, and the addition of 1% BSA significantly decreased the Km values for both enzyme sources, without affecting the respective Vmax values (the latter values differ from each other because a different enzyme source was used, and we do not have a good way to determine the UGT2B7 concentration within each membrane sample). A further increase of the BSA concentration to 2% did not significantly affect the kinetic parameters. The obtained kinetic parameter values and the effect of BSA addition on them are in good agreement with previous results with locally made HLMs and recombinant UGT2B7 that was expressed in a different system than the one used in this study, HEK293 cells (Uchaipichat et al., 2006; Rowland et al., 2007).
Enzyme kinetics of AZT glucuronidation by HLM (A) and UGT2B7 (B) without BSA and in the presence of 1% BSA. The points represent an average of three samples ± S.E. Glucuronidation rates are presented as actual (measured) rates in nmol · min−1 · mg−1 recombinant protein. The derived kinetic constants are presented in Table 1. The data were fitted to the Michaelis-Menten equation. The Eadie-Hofstee transforms of the data are presented as insets.
AZT, entacapone, and 4-MU glucuronidation kinetic type and parameters
The values represent a best-fit result ± S.E. The reaction velocity is given in actual rates for all of the enzymes. See Materials and Methods for additional details.
Enzyme Kinetics of Entacapone with HLM and UGT1A9.
Entacapone is a nearly selective UGT1A9 substrate in the liver (Lautala et al., 2000), and we used it as a model compound for studying the effect of BSA on either recombinant UGT1A9 or the native enzyme in HLMs. Because the preliminary results were significantly different from the previously published finding for this enzyme (Rowland et al., 2008), we performed these studies using two different preparations of recombinant UGT1A9, the one from our laboratory that carry a short C-terminal fusion peptide (Kurkela et al., 2003) and the commercial UGT1A9 Supersomes that lack such a fusion peptide. It may be added here that the concentration of active UGT1A9 in the Supersomes sample appears to be significantly higher than in our recombinant UGT1A9 sample, but, as shown below, the differences between the two preparations do not much affect their kinetics.
The clear and unexpected result from the experiments with all three different UGT1A9 samples was that in addition to decreasing the reaction Km value, the presence of 0.1% BSA in the reaction mixture led to a large increase in the three respective Vmax values (Fig. 4; Table 1). Mild substrate inhibition was observed at higher entacapone concentrations, and the obtained Ki values were 75- to 1500-fold higher than the respective Km values (Table 1). Because of the large differences between the Km values and the respective Ki values, the obtained Vmax values were not significantly influenced by the Km/Ki ratio and were not largely underestimated as they would have been if the Km/Ki ratio was closer to 1. The presence of BSA also somewhat changes the apparent enzyme kinetic model of entacapone glucuronidation from mild substrate inhibition to partial substrate inhibition. The latter means that on the basis of the expectation from the substrate inhibition model, the inclusion of BSA led to a less-than-expected decrease in the glucuronidation rate in the presence of high entacapone concentrations. Therefore, in addition to the empirical substrate inhibition equation, we fitted the experimental data to a mechanistic two-site model equation (Houston and Kenworthy, 2000). This model assumes the existence of two identical substrate binding sites and can be used for sigmoidal and substrate inhibition kinetics. But because no autoactivation was observed, we constrained parameter α to 1 and used parameter β to describe the changes in the rate of product formation from the S · E · S complex in comparison to its formation from an E · S complex (Table 1).
Enzyme kinetics of entacapone glucuronidation by HLM (A), in-house-produced UGT1A9 (B), and commercial UGT1A9 (C) without BSA and in the presence of 0.1% BSA. The points represent an average of three samples ± S.E. The concentrations of entacapone were corrected for nonspecific binding. Glucuronidation rates are presented as actual (measured) rates in nmol · min−1 · mg−1 recombinant protein. The derived kinetic constants are presented in Table 1. The data were fitted to the two-site equation (see Materials and Methods for details). The Eadie-Hofstee transforms of the data are presented as insets.
Enzyme Kinetics of 4-MU Glucuronidation by UGT1A9.
To further explore the effect of BSA on the Vmax of UGT1A9 in entacapone glucuronidation (Fig. 4; Table 1) and find out if this is only a peculiarity of this substrate, we examined the effect of BSA addition on the glucuronidation of 4-MU by UGT1A9. Because 4-MU is not a UGT1A9-specific substrate, the 4-MU glucuronidation assays with UGT1A9 were limited to the recombinant enzymes. One reason for the selection of 4-MU as the second test substrate for UGT1A9 is that its use enables a direct comparison of our results with the previous study on the effect of BSA on UGT1A9 (Rowland et al., 2008). The examination of the two different samples of recombinant UGT1A9 that we tested (the locally made and the commercial sample) most clearly showed that the addition of BSA to either 0.1 or 1% resulted in a significant Km decrease and a concomitant increase in the Vmax values, with a slightly larger effect on the Vmax by 1% BSA than by 0.1% (Fig. 5; Table 1). A detailed kinetic analysis revealed that 4-MU glucuronidation by UGT1A9 also follows the substrate inhibition equation in the absence and the presence of BSA. It is important to note that, as in the case of entacapone, the Ki values of UGT1A9 for 4-MU were significantly higher than the corresponding Km values, allowing for correct determination of Vmax from the experimental data (Table 1).
Enzyme kinetics of 4-MU glucuronidation by in-house-produced UGT1A9 (A) and commercial UGT1A9 (B) without BSA and in the presence of 0.1 and 1% BSA. The concentrations of 4-MU were corrected for nonspecific binding. The points represent an average of three samples ± S.E. Glucuronidation rates are presented as actual (measured) rates in nmol · min−1 · mg−1 recombinant protein. The derived kinetic constants are presented in Table 1. The data were fitted to the substrate inhibition equation. The Eadie-Hofstee transforms of the data are presented as insets.
Enzyme Kinetics for the Cosubstrate UDPGA with UGT1A9.
We also explored the possibility that the presence of BSA affects the kinetics of UGT1A9 with the cosubstrate UDPGA. In these assays, we used either entacapone or 4-MU as the aglycone substrate, and the results show that, regardless of the aglycone used, the reaction was best described by the Michaelis-Menten model (Fig. 6; Table 1). The inclusion of 0.1% BSA when entacapone was the aglycone substrate resulted in a Vmax increase without a Km change (Fig. 6A). However, when 4-MU was the aglycone substrate, the addition of 0.1% BSA led to a Km decrease and a Vmax increase (Fig. 6B; Table 1). The reason for these differences is currently unclear and should be examined in the future.
Enzyme kinetics of UDPGA using entacapone (A) and 4-MU (B) as aglycone substrates without BSA and in the presence of 0.1% BSA. The points represent an average of three samples ± S.E. Glucuronidation rates are presented as actual (measured) rates in nmol · min−1 · mg−1 recombinant protein. The derived kinetic constants are presented in Table 1. The data were fitted to the Michaelis-Menten equation. The Eadie-Hofstee transforms of the data are presented as insets.
Discussion
Drug glucuronidation rates and kinetics that are determined using in vitro assays tend to underestimate the in vivo rates (Boase and Miners, 2002; Miners et al., 2006). Rowland et al. (2007, 2008) have found that the addition of BSA significantly enhances the activity of UGT2B7 and UGT1A9 by decreasing their Km for the aglycone substrate without affecting the Vmax values. They offered an interesting explanation for this “BSA effect”—removal of a long-chain fatty acid that competitively inhibits the UGTs. At the start of the study presented here, we wondered if the proposed inhibitory fatty acids are also present in the recombinant UGTs that we often use for glucuronidation studies—in-house-produced recombinant UGTs that are expressed in insect cells and carry a C-terminal fusion peptide (Kurkela et al., 2003, 2007). However, the results took us to different directions, mainly to validate the initial observation that the BSA effect in UGT1A9 is different and more complex than in UGT2B7. However, before doing this, we developed the needed methods for measuring drug binding to BSA and different biological membranes.
Many drugs bind nonspecifically to macromolecules; therefore, determining the fraction of the added drug that is free under the experimental conditions (the fu) during in vitro assays is important because only the unbound fraction interacts with the target enzyme (Grime and Riley, 2006; Varshney et al., 2010). We measured drug binding to BSA and two enzyme sources, insect cell membranes and HLM, by a newly developed ultrafiltration assay. Because ultrafiltration systems often “suffer” from high NSBf values (Lee et al., 2003; Taylor and Harker, 2006), we took care to minimize, determine, and take into account the nonspecific binding to the filter device, regardless of whether the binding was to the filter itself or to the walls of the tube. Our results concerning AZT binding to BSA, negligible binding, are the same as in a previous study that used a dialysis system (Rowland et al., 2007). On the other hand, our finding that 4-MU binding to BSA is concentration dependent to some extent, particularly at 4-MU concentrations less than 100 μM 4-MU and in the presence of 0.1% BSA (Fig. 1B), is not in full agreement with the published studies/results (Rowland et al., 2008). A possible reason for this is that the substrate concentration dependence of 4-MU is mainly visible at substrate concentrations less than 50 μM in the presence of 0.1% BSA (Fig. 1B), whereas the lowest 4-MU concentration tested in the previous study (in the presence of 0.1, 1, and 2% BSA) was 50 μM (Rowland et al., 2007).
Entacapone binds strongly to the filter device and to BSA (Figs. 1A and 2). Our analysis clearly demonstrates that, at least for some drugs, the binding is concentration dependent, and therefore the correct fu value for each point on the kinetic curve should be taken into account. This outcome differs from the simpler view that emerged from the earlier reports of Rowland et al. (2007, 2008), but our findings are very clear and indicate that the substrate dependence of binding to BSA, and perhaps to plasma also, should be evaluated for each drug under study. A further small complication in the case of entacapone binding, and perhaps other highly lipophilic compounds, is possible displacement from BSA by the enzyme source such as insect cell membranes (Fig. 2). This displacement was more significant at substrate concentrations less than 200 μM and in the presence of higher enzyme source concentrations (Fig. 2), and although it was not significant in our case, it might turn out to be meaningful with other drugs that exhibit similar features but that serve as poor substrates to the tested UGT, perhaps leading to the addition of excessive amounts of recombinant enzyme. In such cases, a 3D binding curve, similar to the one we present in Fig. 2B, may be essential to estimate the fu at different points of the experiment.
The current results on the effect of BSA addition on AZT glucuronidation by UGT2B7, either in its native state within HLM or as a recombinant enzyme that was expressed in insect cell membranes and carries a short C-terminal fusion peptide (Fig. 3), are very similar to the earlier results for HLM and recombinant UGT2B7 that was expressed in HEK293 cells (Rowland et al., 2007). This finding strongly suggests that, at least with respect to the putative competitive inhibitor that is removed by the addition of BSA to the reaction mixture, there is no significant difference between the recombinant UGTs, regardless of whether they were expressed in HEK293 or insect cells, to the native UGT in HLM. Likewise, our results on the BSA effect on UGT1A9 do not give any reason to suspect that significant differences in enzyme kinetics and the interactions of UGT1A9 with its aglycone substrates exist between the native and the recombinant enzyme that was expressed in insect cells, with or without a C-terminal fusion peptide (Figs. 4 and 5). On the other hand, there is a major difference between our results with UGT1A9 and those from a previous study (Rowland et al., 2008).
We found that the addition of BSA to the entacapone glucuronidation reaction, which in HLM is primarily catalyzed by UGT1A9 (Lautala et al., 2000), not only lowers the Km value, as previously found for propofol glucuronidation by UGT1A9 (Rowland et al., 2008), but also leads to a large increase in the Vmax value (Fig. 4; Table 1). Moreover, our results for 4-MU glucuronidation by recombinant UGT1A9 demonstrate that the differences between the current and previous results are unlikely to be simply due to the use of different aglycone substrates (something that would have been important by itself). The general importance of our findings stems from the fact that they do not fit into the previously suggested scheme according to which the BSA effect on UGTs is solely the removal of a competitive inhibitor, possibly a long-chain fatty acid, from the aglycone substrate binding site of the enzyme (Rowland et al., 2007, 2008).
It is currently difficult to explain the differences between the two sets of results, particularly that there is such a good agreement between them on the effect of AZT glucuronidation by UGT2B7. Nevertheless, the combined results allow us to construct a theoretical framework in which the observed BSA effect can be rationalized and further studied (Fig. 7). The suggestion is based on the assumption that UGT-catalyzed glucuronidation reactions follow a compulsory ordered bi-bi mechanism in which UDPGA is the first binding substrate (Luukkanen et al., 2005). Because a Km decrease and a Vmax increase take place in UGT1A9 catalyzing entacapone glucuronidation in the presence of BSA, and because the enzyme affinity to UDPGA was relatively unaffected, it is suggested that the inhibitor(s) that BSA removes binds to either enzyme-UDPGA (E · AX) or enzyme-UDPGA-substrate (E · AX · B) complexes rather than to the free enzyme, “E” (Fig. 7).
The proposed mechanism for inhibitor interactions with a compulsory ordered bi-bi mechanism of UGT catalysis (see Discussion for details). E, enzyme (UGTs); AX, UDP-α-d-glucuronic acid; A, UDP; B, substrate (aglycone); I, inhibitor.
The exact nature of the tentative inhibitor(s) and the reasons for the differential effect of BSA toward UGT1A9 and UGT2B7 remain unknown at this stage. One possibility is that UGT1A9 is inhibited by a mixture of different inhibitors, perhaps including competitive, noncompetitive, and/or mixed-type inhibitors, all of which are probably released from the cells/membranes during the preparation of the samples for in vitro glucuronidation assays. Such a mixture will probably include inhibitors with different affinities to different individual UGTs and thereby might also explain the small change of the kinetic model for UGT1A9 that was caused by BSA addition (Fig. 4; Table 1).
In summary, our results show that addition of BSA enhances the in vitro activities of UGT2B7 and UGT1A9, regardless of whether a native enzyme in HLM was used, or a recombinant UGT with or without a C-terminal fusion peptide. Drug binding to BSA differs in the extent and dependence on drug concentration, indicating that it should be determined carefully in each case. The BSA effect is (even) more complex than described previously, but its full clarification may lead us to a better and deeper understanding of the full kinetic mechanism of drug glucuronidation by the UGTs.
Authorship Contributions
Participated in research design: Manevski and Finel.
Conducted experiments: Manevski and Morelo.
Performed data analysis: Manevski, Morelo, and Finel.
Wrote or contributed to the writing of the manuscript: Manevski, Yli-Kauhaluoma, and Finel.
Acknowledgments
We thank Johanna Mosorin for skillful technical assistance.
Footnotes
This study was supported by the Graduate School in Pharmaceutical Research, Academy of Finland (Project Number 120975); the Sigrid Juselius Foundation; and a Helsinki University Research Foundation grant for young researchers.
Article, publication date, and citation information can be found at http://dmd.aspetjournals.org.
doi:10.1124/dmd.111.041418.
↵
The online version of this article (available at http://dmd.aspetjournals.org) contains supplemental material.
-
ABBREVIATIONS:
- UGT
- UDP-glucuronosyltransferase
- AZT
- zidovudine (3′-azido-3′-deoxythymidine)
- BSA
- bovine serum albumin
- 3D
- three-dimensional
- HEK
- human embryonic kidney
- HLM
- human liver microsomes
- HPLC
- high-performance liquid chromatography
- 4-MU
- 4-methylumbelliferone
- NSBf
- nonspecific binding to the filter device
- UDPGA
- UDP-α-d-glucuronic acid
- UPLC
- ultraperformance liquid chromatography
- MeOH
- methanol
- fu
- fraction unbound.
- Received June 28, 2011.
- Accepted August 18, 2011.
- Copyright © 2011 by The American Society for Pharmacology and Experimental Therapeutics