In vitro approaches for the prediction of drug pharmacokinetic parameters in vivo (in vitro-in vivo extrapolation) have attracted widespread interest in recent years, particularly for the assessment of metabolic stability in preclinical drug development (Ward, 2005). In this regard, an intrinsic clearance (CL_int) for a metabolic pathway determined using human liver microsomes (HLM), fresh or cryopreserved human hepatocytes, or recombinant cytochrome P450 enzymes, may be extrapolated to a whole liver value using appropriate scaling factors (Houston, 1994; Griffin and Houston, 2004; Miners et al., 2004; Proctor et al., 2004). Hepatic clearance (CL_H) and extraction ratio are then estimated using a mathematical expression, typically the “well stirred model”, that relates this parameter to CL_int, liver blood flow (Q_H), and the fraction of drug unbound in blood (f_u,b) (Houston, 1994; Iwatsubo et al., 1997; Ito et al., 1998). It is similarly possible to predict the magnitude of an inhibitory drug-drug interaction in vivo based on measurement of an inhibitor constant (K_i) (Ito et al., 1998; von Moltke et al., 1998; Ito and Houston, 2004; Wienkers and Heath, 2005).

Despite the widespread use of in vitro-in vivo extrapolation, the use of HLM as the enzyme source generally results in underestimation of the in vivo CL_H (Ito and Houston, 2005). For drugs eliminated by glucuronidation, in vivo CL_H is typically underpredicted by an order of magnitude using physiologically based scaling factors (Mistry and Houston, 1987; Boase and Miners, 2002; Soars et al., 2002; Engtrakul et al., 2005; Riley et al., 2005). Prediction accuracy is improved with the use of fresh or cryopreserved human hepatocytes (Soars et al., 2002; McGinnity et al., 2005), but the bias for underprediction remains (Ito and Houston, 2005; Miners et al., 2006). Similar observations have been reported in studies investigating inhibition of drug glucuronidation with HLM as the enzyme source; experimental K_i values resulted in underprediction of the magnitude of inhibitory interactions involving glucuronidated drugs in vivo (Rowland et al., 2006; Uchaipichat et al., 2006).

It has been demonstrated previously that addition of bovine serum albumin (BSA) to human liver microsomal incubations decreases the K_m and increases the in vitro CL_int for drugs metabolized by CYP2C9 (Ludden et al., 1997; Carlile et al., 1999; Tang et al., 2002; Wang et al., 2002; Zhou et al., 2004). Improved predictivity of in vivo CL_int values occurred for experiments conducted in the presence of BSA. Subsequent studies conducted in this laboratory demonstrated that the addition of 2% BSA to incubations of HLM and recombinant UGT2B7 increased the in vitro CL_int for lamotrigine and zidovudine (AZT) glucuronidation 7- to 10-fold (Rowland et al., 2006; Uchaipichat et al., 2006). Moreover, the addition of BSA reduced the K_i values for valproic acid inhibition of lamotrigine glucuronidation and fluconazole inhibition of AZT glucuronidation by almost an order of magnitude, such that the magnitude of these interactions in vivo was predicted correctly from in vitro inhibition constants.

Despite the greatly improved predictivity of in vitro-in vivo clearance extrapolation for microsomal incubations conducted in the presence of BSA, the mechanism of the “albumin” effect remains unknown. The “mopping up” of endogenous inhibitors present in microsomal incubations or altered protein (CYP or UGT) conformation has been proposed (Tang et al., 2002; Rowland et al., 2006; Uchaipichat et al., 2006), but neither have been investigated in a systematic manner. Here, we compared the effects of BSA and human serum albumin (HSA) in their “crude”, fatty acid-free (FAF), globulin-free (GF), and fatty acid- and globulin-free (FAFGF) forms on the glucuronidation of AZT by HLM and recombinant UGT2B7 [expressed in human embryonic kidney (HEK)293 cells] and on 4-methylumbelliferone (4MU) glucuronidation by UGT2B7. The glucuronidation of AZT, a selective substrate for UGT2B7, follows Michaelis-Menten kinetics (Boase and Miners, 2002; Court et al., 2003), whereas 4MU glucuronidation by recombinant UGT2B7 exhibits sigmoidal kinetics characteristic of autoactivation (Uchaipichat et al., 2004). Because 4MU is a nonselective UGT substrate, 4MU glucuronidation was not investigated with HLM as the enzyme source. Results obtained with the albumin preparations and subsequent inhibition experiments demonstrated that unsaturated long-chain fatty acids present in the microsomal and HEK293 cell lysate preparations act as potent competitive inhibitors of AZT and 4MU glucuronidation and that sequestration of fatty acids is responsible for the higher binding affinities observed in the presence of certain albumin preparations.

Materials and Methods

Materials. Alamethicin (from Trichoderma viride), bovine serum albumin (crude BSA, product no. A7906), essentially fatty acid-free BSA (BSA-FAF; product no. A6003), essentially globulin-free BSA (BSA-GF; product no. A3059), essentially fatty acid- and globulin-free BSA (BSA-FAFGF; product no. A0281), human serum albumin (crude HSA; product no. A9511), essentially fatty acid-free HSA (HSA-FAF; product no. A1887), essentially globulin-free HSA (HSA-GF; product no. A8753), essentially fatty acid- and globulin-free HSA (HSA-FAFGF; product no. A3782), 4MU, 4-methylumbelliferone-β-d-glucuronide (4MUG), fatty acids (as the free acid), UDP-glucuronic acid (UDPGA, trisodium salt), [¹⁴C]UDPGA, and AZT were purchased from Sigma-Aldrich (Sydney, Australia). (Albumin product numbers have been provided to allow details of each preparation, including method of isolation, to be accessed from the supplier's Web site.) Human serum albumin TrueClone cDNA (reference sequence NM_000477.3, length 2110 base pairs) was purchased from Origene (Rockville, MD). Solvents and other reagents were of analytical reagent grade.

Human Liver Microsomes and Expression of UGT2B7. Pooled human liver microsomes were prepared by mixing equal protein amounts from five human livers (H7, 44yo female; H10, 67yo female; H12, 66yo male; H29 45yo male; and H40, 54yo female), obtained from the human liver “bank” of the Department of Clinical Pharmacology (Flinders University, Adelaide, Australia). Approval for the use of human liver tissue in xenobiotic metabolism studies was obtained from both the Clinical Investigation Committee of Flinders Medical Centre and from the donors' next of kin. HLM were prepared by differential centrifugation as described by Bowalgaha et al. (2005).

A cDNA encoding UGT2B7 was stably expressed in the HEK293 cell line as described previously (Stone et al., 2003). Cell lines were transfected with the UGT2B7 cDNA cloned into the expression vector pEF-IRES-puro6. Transfected cells were incubated in Dulbecco's modified Eagle's medium, which contained puromycin and 10% fetal calf serum in a humidified incubator at 37°C with an atmosphere of 5% CO₂. Following growth to at least 80% confluence, cells were harvested and washed with 0.1 M phosphate-buffered saline, pH 7.4. Cells were subsequently lysed by sonication using a sonicator (Heat Systems Ultrasonics, Plainsview, NY) set at microtip limit of 4, with four 1-s “bursts”, separated by 3 min with cooling on ice. Lysed samples were centrifuged at 12,000g for 1 min at 4°C, and the supernatant fraction was removed and stored at –80°C until use. Expression of UGT2B7 protein was demonstrated by immunoblotting with a nonselective UGT antibody (raised against purified mouse Ugt) according to Uchaipichat et al. (2004) and activity measurement (see below).

UGT2B7 and HSA cDNAs, each cloned into the expression vector pEF-IRES-puro6, were cotransfected in HEK293 cells following the procedure described above for UGT2B7. Transfected cells were treated as detailed for stably expressed UGT2B7. Expression of the HSA protein was quantified by immunoblotting (see below).

Immunoblotting of HSA. HEK293 cell lysate protein (20 μg) was subjected to 10% SDS-polyacrylamide gel electrophoresis. Proteins were rectilinearly transferred onto nitrocellulose and probed with goat anti-HSA primary antiserum (1:1000 dilution; Sigma Genosys, Sydney, Australia) and rabbit anti-goat IgG (1:1000 dilution; H+L-horseradish peroxidase) as the secondary antibody (Southern Biotechnology Associates, Birmingham, AL). Membrane-bound peptides conjugated with horseradish peroxidase were detected by chemiluminescence (Roche Diagnostics, Mannheim, Germany) and subsequently exposed to Omat autoradiographic film (Eastman Kodak, Rochester, NY). Autoradiographs were processed manually with AGFA developer, fixer, and replenisher reagents. Quantification of HSA expression was achieved by comparison of band intensity to those of a standard curve constructed using known concentrations of HSA.

AZT Glucuronidation Assay. Incubations, in a total volume of 200 μl, contained 0.1 M phosphate buffer, pH 7.4, 4 mM MgCl₂, 0.1 mg of HLM, 0 to 4% albumin, and 25 to 4000 μM AZT. HLM were fully activated by the addition of the pore-forming polypeptide alamethicin (50 μg/mg protein) with incubation on ice for 30 min (Boase and Miners, 2002). Following a 5-min preincubation, reactions were initiated by the addition of 5 mM UDPGA. Incubations were performed at 37°C in a shaking water bath for 60 min. Reactions were terminated by the addition of 6 μl of perchloric acid (70%, v/v). Samples were subsequently centrifuged at 4000g for 10 min, and a 30-μl aliquot of the supernatant fraction was injected directly into the HPLC column. For reactions performed using recombinant UGT2B7, incubation mixtures contained 0.15 mg of HEK293 cell lysate in place of HLM protein, and the incubation time was increased to 90 min. Under the reaction conditions used, AZT glucuronidation was linear with respect to incubation time to 120 min and protein concentration to 1.75 mg/ml for both HLM and HEK293 cell lysate as the enzyme source. AZT glucuronidation by lysate from untransfected HEK293 cells was not detectable.

Quantification of GAZT Formation. HPLC was performed using an Agilent 1100 series instrument (Agilent Technologies, Sydney, Australia) fitted with a NovaPak C18 analytical column (150 × 3.9 mm, 4 μm; Waters, Sydney, Australia). Zidovudine glucuronide (GAZT) was separated with a mobile phase containing 10 mM triethylamine (pH adjusted to 2.5 with perchloric acid) and 10% acetonitrile, at a flow rate of 1 ml/min. Column eluant was monitored by UV absorbance at 267 nm. Retention times for GAZT and AZT were 4.0 and 6.1 min, respectively. GAZT formation was quantified by comparison of peak areas to those of an AZT standard curve prepared over the concentration range 1 to 25 μM. Overall within-day assay reproducibility was assessed by measuring GAZT formation in eight separate incubations of the same batch of pooled HLM. Coefficients of variation were 3.8 and 3.1% for the AZT concentrations of 25 and 4000 μM, respectively.

4MU Glucuronidation Assay. Incubations, in a total volume of 200 μl, contained 0.1 M phosphate buffer, pH 7.4, 4 mM MgCl₂, 0 to 2% albumin, HEK293 cell lysate expressing 0.2 mg/ml UGT2B7, and 25 to 1500 μM 4MU. Following a 5-min preincubation, reactions were initiated by the addition of 5 mM UDPGA. Incubations were performed at 37°C in a shaking water bath for 90 min. Reactions were terminated by the addition of 6 μl of perchloric acid (70%, v/v). Samples were subsequently centrifuged at 4000g for 10 min, and a 25-μl aliquot of the supernatant fraction was injected directly into the HPLC column. 4MU β-d-glucuronidation by lysate from untransfected HEK293 cells was not detectable.

Quantification of 4MUG Formation. The HPLC instrument and column were as described for the measurement of GAZT. 4MUG was separated by gradient elution at a flow rate of 1 ml/min. Initial conditions were 96% 10 mM triethylamine buffer containing perchloric acid, pH 2.5, and 10% acetonitrile (mobile phase A) and 4% acetonitrile (mobile phase B). These conditions were held for 3 min, and then the proportion of mobile phase B was increased to 30% over 0.1 min and held for 1 min. Column eluant was monitored by UV absorbance at 316 nm. Retention times for 4MUG and 4MU were 3.6 and 5.8 min, respectively. 4MUG formation was quantified by comparison of peak areas to those of an authentic 4MUG standard curve prepared over the concentration range 1 to 10 μM. Overall withinday assay reproducibility was assessed by measuring 4MUG formation in eight separate incubations of the same batch of expressed UGT2B7. Coefficients of variation were 4.3 and 2.6% for 4MU concentrations of 50 and 1500 μM, respectively.

Fatty Acid Glucuronidation Assay. The glucuronidation of the fatty acids listed in legend to Fig. 3 was demonstrated using a radiometric thin layer chromatography method. Incubations (100 μl) contained 0.1 M phosphate buffer, pH 6.8, 4 mM MgCl₂, HEK293 cell lysate expressing 1.7 mg/ml UGT2B7, UDPGA [total concentration 50 μM, consisting of [¹⁴C]UDPGA (0.1 μCi; 5 μl) and nonlabeled UDPGA (47 μM)], and 100 μM fatty acid. Fatty acids were added as solutions in ethanol such that the final concentration of solvent in incubations was 1% (v/v). This concentration of ethanol has been demonstrated previously to cause <20% inhibition of recombinant UGT2B7 activity (Uchaipichat et al., 2004). Weakly acidic (pH 6.8) conditions were used to minimize hydrolysis of the fatty acid acyl glucuronides formed during the course of an incubation. Incubations were performed at 37°C in a shaking water bath for 120 min. For experiments demonstrating the glucuronidation of fatty acids present in preparations of HLM and recombinant UGT2B7, incubation mixtures contained 0 to 2% albumin, HLM, or cell lysate expressing 2 mg/ml UGT2B7 and 5 mM UDPGA (containing [¹⁴C]UD-PGA; 0.2 μCi, 10 μl). Arachidonic and linoleic acids were included as positive controls.

Separation of fatty acid glucuronides was achieved by development of TLC plates in butan-1-ol/acetic acid/35% ammonia/water (35:25:9:0.7:30.3). After development, plates were air-dried overnight and exposed in Storage PhosphorScreen Cassettes (GE Healthcare, Little Chalfont, Buckinghamshire, UK) for up to 10 days. Quantitative densitometry of glucuronides was undertaken using a scanning laser PhosphorImager and associated ImageQuant software (GE Healthcare) with reference to a standard curve constructed using known concentrations of [¹⁴C]UDPGA.

Quantification of the Fatty Acid Content of HLM, BSA, HSA, and HSA-FAF. The fatty acid contents of 1 mg/ml HLM, 2% BSA, 2% HSA, and 2% HSA-FAF were determined by gas-liquid chromatography using a modification of the method of Folch et al. (1956). In brief, triheptadecanoin (C17:0; internal standard) was added to a known volume of sample, and lipids were extracted with 5 ml of chloroform/methanol (2:l). Samples were transesterified with methanol (containing 1% H₂SO₄) at 70°C for 3 h. Fatty acid methyl esters (FAMEs) were extracted with water and n-heptane for analysis by gas-liquid chromatography. FAMEs were separated and quantified using a Hewlett Packard 6890 gas chromatograph equipped with a 50-m × 0.32-mm capillary column coated with BPX-70 (0.25-μm film thickness; SGE Pty Ltd., Victoria, Australia). Injector and detector (flame ionization) temperatures were set at 250 and 300°C, respectively, whereas the initial oven temperature of 140°C was programmed to rise to 220°C at a rate of 5°C/min. Helium was used as the carrier gas at a velocity of 35 cm/s. FAMEs were identified based on the retention times of authentic lipid standards (GLC-463; Nuchek Prep Inc., Elysian, MN) and quantified by comparison to the internal standard using ChemStation software (Agilent Technologies, Palo Alto, CA).

Binding of AZT and 4MU to Albumin, HLM, and HEK293 Cell Lysate. Binding of AZT and 4MU to HLM, HEK293 cell lysate, albumin (0.1, 0.5, 1, and 2%), and mixtures of albumin with each enzyme source was measured by an equilibrium dialysis method according to McLure et al. (2000). Binding measurements were performed using a Dianorm equilibrium dialysis apparatus that comprised Teflon dialysis cells (capacity of 1200 μl/side) separated into two compartments with Sigma-Aldrich dialysis membrane (molecular mass cut-off 12 kDa). One side of the dialysis cell was loaded with 1 ml of a solution of AZT (100–2000 μM) or 4MU (50–1000 μM) in 0.1 M phosphate buffer, pH 7.4. The other compartment was loaded with 1 ml of either a suspension of HLM in 0.1 M phosphate buffer, pH 7.4; HEK293 cell lysate in 0.1 M phosphate buffer, pH 7.4; 0.1, 0.25, 0.5, 1, and 2% albumin in 0.1 M phosphate buffer, pH 7.4; or a combination of albumin with each enzyme source in 0.1 M phosphate buffer, pH 7.4. The dialysis cell assembly was immersed in a water bath maintained at 37°C and rotated at 12 rpm for 4 h. Control experiments were also performed with phosphate buffer or albumin on both sides of the dialysis cells at low and high concentrations of all substrates to ensure that equilibrium was attained. A 200-μl aliquot was collected from each compartment, treated with ice-cold methanol containing 200 μl of 8% glacial acetic acid, and cooled on ice. Samples were subsequently centrifuged at 4000g for 10 min at 10°C, and an aliquot of the supernatant fraction (5 μl) was analyzed by HPLC.

Quantification of AZT and 4MU Binding. The HPLC instrument and column used was as described for the measurement of AZT and 4MU glucuronide formation. Separation of AZT was achieved using a 65:35 mixture of mobile phases A and B, as used for the GAZT assay. The mobile phase flow rate was 1.0 ml/min. Column eluant was monitored at 267 nm. The retention time for AZT under these conditions was 2.6 min. 4MU separation was achieved using a 70:30 mixture of mobile phases A and B (as described previously) at a flow rate of 1.0 ml/min. Column eluant was monitored at 316 nm. The retention time for 4MU was 2.9 min. AZT and 4MU concentrations in dialysis samples were determined by comparison of peak areas to those of a standard curve, in the respective concentration ranges 100 to 2000 μM and 10 to 1000 μM. Within-day assay variability was assessed by measuring 100 and 2000 μM AZT or 50 and 1000 μM 4MU (n = 5 for each concentration) in samples containing 0.1 M phosphate buffer, pH 7.4, or BSA in 0.1 M phosphate buffer, pH 7.4. Coefficients of variation were less than 5% in all cases.

Data Analysis. Kinetic data are presented as mean values derived from experiments with pooled HLM (four replicates) and duplicate experiments with recombinant UGT2B7. Kinetic constants (K_m and V_max) for AZT glucuronidation by HLM and recombinant UGT2B7 in the presence and absence of albumin were generated by fitting experimental data to the Michaelis-Menten equation (Uchaipichat et al., 2006). Kinetic constants for 4MU β-d-glucuronidation by recombinant UGT2B7, a reaction that exhibits homotropic positive cooperativity (or “autoactivation”), in the presence and absence of albumin were obtained by fitting data to the Hill equation to obtain estimates of S₅₀, V_max, and n (the Hill coefficient) (Uchaipichat et al., 2004). In all cases, fitting was based on unbound substrate concentrations in incubations and performed with EnzFitter (Biosoft, Cambridge, UK). CL_int for AZT glucuronidation by HLM and UGT2B7 were determined as V_max/K_m. Maximal clearance (CL_max) for 4MU glucuronidation by UGT2B7, which exhibits autoactivation, was calculated according to Houston and Kenworthy (2000), where n is the Hill coefficient (see above):

Where appropriate, statistical analysis (univariate General Linear Model, with Tukey post hoc analysis) was performed using SPSS for Windows, release 12.0.1, 2003 (SPSS Inc., Chicago, IL). Values of P less than 0.05 were considered significant.

Results

Binding of AZT and 4MU to Albumin. The binding of AZT and 4MU to incubation components, including various forms of albumin, was calculated as the concentration of drug in the buffer compartment divided by the concentration of drug in the protein compartment and expressed as the fraction unbound in incubations (f_u,inc). Binding of AZT to pooled HLM, HEK293 cell lysate, and combinations of each enzyme source with the various types of albumin (0.1–2%) was negligible (<5%) across the AZT concentration range investigated (10–2000 μM). Likewise, 4MU binding to HEK293 cell lysate was negligible (<5%) across the concentration range investigated. However, 4MU bound significantly to all forms of albumin. 4MU binding was independent of concentration over the range investigated (50–1500 μM), but it varied with albumin concentration and form. The mean f_u,inc value for 4MU binding to BSA ranged from 0.89 (0.1% BSA) to 0.27 (2% BSA). Significant binding of 4MU to HSA was also observed. Mean f_u,inc values for 4MU binding to mixtures of HSA and HEK293 cell lysate ranged from 0.79 (0.1% HSA) to 0.14 (2% HSA). There was a slight increase in the binding of 4MU to HSA-FAF; f_u,inc values ranged from 0.76 (0.1% HSA-FAF) to 0.09 (2% HSA-FAF). Where observed, the concentration of 4MU present in incubation mixtures was corrected for binding in the calculation of kinetic parameters.

Effect of Albumin on AZT Glucuronidation by HLM and Recombinant UGT2B7. Kinetic data for GAZT formation by human liver microsomal and recombinant UGT2B7, in the presence and absence of albumin, was well modeled by the expression for the Michaelis-Menten equation. Kinetic parameters for AZT glucuronidation in the presence of the various forms of BSA and HSA are shown in Tables 1 and 2, respectively. The mean K_m and V_max values for AZT glucuronidation by HLM and recombinant UGT2B7 in the absence of albumin were 1113 μM and 738 pmol/min · mg, and 439 μM and 50 pmol/min · mg, respectively. These values are similar to those reported in previous publications from this and other laboratories (Boase and Miners, 2002; Court et al., 2003; Uchaipichat et al., 2006).

All forms of BSA (at concentrations ranging from 0.1 to 4%) increased the rate of AZT glucuronidation by both human liver microsomal and recombinant UGT2B7 (Fig. 1). The activation of AZT glucuronidation plateaued at concentrations above 0.5% for all forms of BSA. Kinetic analysis at 0.5 and 2% BSA revealed that the enhanced rate of AZT glucuronidation occurred due to a decrease (P < 0.05) in the K_m for this pathway, without a significant effect on V_max (Table 1; Fig. 2). Crude (unmodified) HSA displayed a concentration-dependent and biphasic (activation and inhibition) interaction with human liver microsomal and recombinant UGT2B7 (Fig. 1). The K_m for AZT glucuronidation was initially reduced for incubations performed in the presence of 0.1% HSA, whereas the addition of 2% HSA caused an increase in the K_m for AZT glucuronidation (Table 2; Fig. 2). In contrast, HSA-FAF increased (4- to 9-fold) the rate of AZT glucuronidation by HLM and recombinant UGT2B7 in a concentration dependent manner up to 0.5%, after which activity plateaued (Fig. 1). Like the effect of BSA preparations, this was due to a decrease in K_m (Table 2; Fig. 2). HSA-GF and HSA-FAFGF also caused activation of human liver microsomal and recombinant UGT2B7 (Fig. 1; Table 2). However, the magnitude of the effect decreased with increasing concentrations of these albumin preparations.

Effect of Albumin on 4MU Glucuronidation by Recombinant UGT2B7. Based on the data for AZT, the effects of BSA, HSA, and HSA-FAF on 4MU glucuronidation by UGT2B7 were also investigated. The kinetics of 4MUG formation by recombinant UGT2B7, in the presence and absence of albumin, was best described by the Hill equation, with a value of n > 1, indicating positive cooperativity. Kinetic parameters for 4MU β-d-glucuronidation in the presence of albumin (BSA, HSA, and HSA-FAF) are shown in Table 3. The S₅₀, n, and V_max values for 4MU β-d-glucuronidation by recombinant UGT2B7 in the absence of albumin were 462 μM, 1.6 pmol/min/mg, and 1065 pmol/min/mg, respectively. These values are in agreement with previous reports for this reaction (Uchaipichat et al., 2004). BSA and HSA-FAF increased the rate of 4MU β-d-glucuronidation by recombinant UGT2B7 in a concentration-dependent manner. The S₅₀ for this pathway was reduced by 50 and 75% for BSA and HSA-FAF, respectively, without a significant effect on either n or V_max (Table 3). Crude HSA did not have a significant effect on any of the kinetic parameters for the glucuronidation of 4MU by recombinant UGT2B7 at any albumin concentration (up to 2%; Table 3).

Fatty Acid Content of HLM and Albumin Preparations.Table 4 details the concentrations of C16 to C20 series fatty acids present in HLM, HEK293 cell lysate, BSA, HSA, and HSA-FAF preparations. The total concentration of fatty acid in 1 mg/ml HLM, 1 mg/ml HEK293 cell lysate, 2% HSA, 2% BSA, and 2% HSA-FAF was 334, 28, 202, 28, and 35 μM, respectively, of which unsaturated fatty acids made up 70, 58, 56, 36, and 58%, respectively. Of the fatty acids known to inhibit UGT2B7 (Tsoutsikos et al., 2004), oleic acid (C18:1n-9), linoleic acid (C18:2n-6), and arachidonic acid (C20:4n-6) were observed in the highest concentrations. The fatty acid content of HLM was comparable with values published previously (Waskell et al., 1982; Kapitulnik et al., 1987), and it was approximately 10-fold higher than the fatty acid content of HEK293 cell lysate.

Fatty Acid Glucuronidation by UGT2B7. Selected saturated and unsaturated fatty acids identified in HLM and human albumin preparations were screened to determine whether they were glucuronidated by UGT2B7. Recombinant UGT2B7 glucuronidated all C14, C16, C18, and C20 saturated and unsaturated fatty acids screened (at an added concentration of 100 μM; Fig. 3). Rates of glucuronidation of the C18 series fatty acids increased with the degree of unsaturation: 21, 30, 36, 59 pmol/min · mg for C18:0, C18:1, C18:2, and C18:3, respectively. In contrast, with the exception of eicosenoic acid (C20:1), rates of UGT2B7 catalyzed glucuronidation of the C20 fatty acids decreased as the degree of unsaturation increased: 19, 27, 99, 82, 75, and 43 pmol/min · mg for C20:0, C20:1, C20:2, C20:3, C20:4, and C20:5, respectively. Like the C18 and C20 series, rates of glucuronidation of the saturated C14 (12 pmol/min · mg) and C16 (17 pmol/min · mg) fatty acids were lower than their respective monounsaturated derivatives (54 and 43 pmol/min · mg).

Formation of Fatty Acid Glucuronides by Preparations of HLM and HEK293 Cell Lysate. The formation of glucuronides of compounds present in (or released by) incubation mixtures of HLM and HEK293 cell lysate expressing UGT2B7, in the presence and absence of BSA and crude and fatty acid-free forms of HSA, was determined in the absence of an exogenous substrate. For comparison, incubations were conducted with added linoleic acid (10 μM), arachidonic acid (10 μM), and 4MU (5 μM; as a representative xenobiotic substrate.). Bands corresponding to the R_f (retention factor) values of the arachidonic acid and linoleic acid glucuronides were detected by TLC for incubations conducted in the presence of HLM and ¹⁴C-labeled UDPGA without albumin (Fig. 4). Based on band intensity, the concentration of the putative fatty acid glucuronides present in the incubation of HLM was 0.4 μM. The intensity of bands was substantially lower using HEK293 cell lysate expressing UGT2B7; the lower band intensity precluded meaningful product quantification. Increasing concentrations of HSA (0.1–2%) in incubations with each enzyme source increased the intensity of these bands in a concentration-dependent manner. For example, the concentration of fatty acid glucuronides present in the incubation of HLM with crude 2% HSA was 1.5 μM. In contrast, addition of BSA and HSA-FAF at 0.1 and 2% to incubations resulted in a concentration-dependent reduction in the intensity of the bands, with essentially no product visible for incubations conducted in the presence of either form of albumin at an added concentration of 2% (Fig. 4).

Inhibition of AZT and 4MU Glucuronidation by Fatty Acids in the Presence and Absence of Albumin. Inhibition of AZT glucuronidation at an added concentration of 400 μM (the approximate K_m) by oleic acid (C18:1), linoleic acid (C18:2), arachidonic acid (C20:4), and a mixture made up of 40% C18:1, 40% C18:2, and 20% C20:4 was measured in the presence and absence of albumin preparations using recombinant UGT2B7 as the enzyme source. Each fatty acid caused between 25 and 60% inhibition when present at 1/20th of the concentration observed in HLM (see above) (Table 5). When added as a mixture, the fatty acids did not increase in the magnitude of inhibition above that observed for the most potent inhibitor (arachidonic acid). The presence of 2% BSA or 2% HSA-FAF in incubations reversed the inhibitory effects of the fatty acids (Table 5).

The effect a combination of fatty acids (totaling 7.5 μM, made up of 3 μM C18:1, 3 μM C18:2, and 1.5 μM C20:4) on the kinetics of AZT and 4MU glucuronidation by recombinant UGT2B7 was characterized in the presence and absence of 2% BSA. In the absence of 2% BSA, the combination of fatty acids caused a 3-fold increase in the K_m for AZT glucuronidation by recombinant UGT2B7, from 439 to 1342 μM without a significant effect on V_max (56 versus 61 pmol/min · mg). In the presence of 2% BSA, the combination of fatty acids had no effect on the kinetics of AZT glucuronidation by recombinant UGT2B7; respective K_m and V_max values in the presence of BSA were 53 and 50 μM and 60 and 61 pmol/min · mg, for experiments performed in the absence and presence of fatty acids. In the absence of 2% BSA, the combination of fatty acids caused a 50% increase in the S₅₀ for 4MU glucuronidation by recombinant UGT2B7, from 462 to 612 μM without a significant effect on n or V_max (987 versus 972 pmol/min · mg). In the presence of 2% BSA, the combination of fatty acids had no effect on the kinetics of 4MU glucuronidation by recombinant UGT2B7; respective S₅₀, n, and V_max values in the presence of BSA were 227 and 238 μM, 1.7 and 1.7, and 933 and 984 pmol/min · mg, for experiments performed in the absence and presence of fatty acids.

AZT and 4MU Glucuronidation by Recombinant UGT2B7 Coexpressed with HSA. The content of HSA, determined by immunoblotting, in HEK293 cell lysates coexpressing HSA and UGT2B7 was approximately 9 nmol/mg (Fig. 5); this corresponds to an HSA concentration of 0.6 mg/ml (i.e., 0.06%) in a 0.2-ml incubation containing 1 mg/ml cell lysate. The kinetics of GAZT formation by recombinant UGT2B7 coexpressed with recombinant HSA was well modeled using the Michaelis-Menten equation, with K_m and V_max values of 212 μM and 47 pmol/min · mg, respectively. The K_m value is approximately half that observed for this pathway in the absence of coexpressed recombinant HSA (439 μM). In contrast, the V_max value (47 pmol/minzmdmg) for AZT glucuronidation in HEK293 cell lysates expressing both UGT2B7 and HSA was essentially identical to the value (50 pmol/min · mg) for cells expressing UGT2B7 alone. The kinetics of 4MUG formation by recombinant UGT2B7 coexpressed with recombinant HSA was well modeled using the Hill equation. S₅₀, n, and V_max values were 370 μM, 1.8 and 1152 pmol/min · mg, respectively. The S₅₀ is approximately 25% lower than that observed for this pathway in the absence of coexpressed recombinant HSA (462 μM). In contrast, the V_max and n values (1152 pmol/min · mg and 1.8) for AZT glucuronidation in HEK293 cell lysates expressing both UGT2B7 and HSA were essentially identical to the values (1065 pmol/min · mg and 1.6) for cells expressing UGT2B7 alone.

Discussion

All forms of BSA increased the rate of AZT glucuronidation by both HLM and UGT2B7 in a concentration-dependent manner, due to a decrease in K_m. A similar effect was observed with HSA-FAF, but crude HSA exhibited a biphasic effect (decreased K_m at 0.1%; increased K_m at 2%) on AZT glucuronidation. There was also a trend to increasing K_m at higher concentrations of HSA-GF and HSA-FAFGF. Consistent with the effects on AZT glucuronidation, BSA and HSA-FAF, but not HSA, enhanced the rate of 4MU glucuronidation due to a decrease in S₅₀. Similar data have been reported for the effects of different albumin preparations on CYP2C9 activity (Zhou et al., 2004). These results led us to hypothesize that fatty acids present in, or released during, incubations acted as competitive inhibitors of UGT2B7 catalyzed AZT and 4MU glucuronidation. We have demonstrated previously that unsaturated long-chain fatty acids (oleic, linoleic, and arachidonic) are potent inhibitors of UGT2B7 and UGT1A9 (Tsoutsikos et al., 2004). K_i values for linoleic and arachidonic acid inhibition of UGT2B7 were 6.3 and 0.15 μM, respectively. By contrast, saturated and short- and medium-chain fatty acids minimally affect UGT activity (Tsoutsikos et al., 2004).

Analysis of the fatty acid content of HLM, HEK293 cell lysate, BSA, HSA, and HSA-FAF was thus undertaken. Consistent with previous reports (Waskell et al., 1982; Kapitulnik et al., 1987), the unsaturated long-chain fatty acid content of HLM was high, with oleic (C18:1), linoleic (C18:2), and arachidonic (C20:4) acids the most prevalent. The fatty acid content of HEK293 cell lysate was significantly (10-fold) lower. Oleic, linoleic, and arachidonic acids were the most abundant unsaturated long-chain fatty acids present in crude HSA, whereas the content of these fatty acids in HSA-FAF and BSA was an order of magnitude lower. Oleic, linoleic, and arachidonic acids, and other unsaturated and saturated long-chain fatty acids [myristic (14:0), myristoleic (14:1), palmitic (16:0), palmitoleic (16: 1), stearic (18:0), linolenic (18:3), arachidic (20:0), eicosenoic (20:1), dihomo-linolenic (20:3), and eicosapentaenoic acids (20:5)] were all shown to be glucuronidated by UGT2B7. It has been demonstrated previously that linoleic and arachidonic acids are substrates of UGT2B7 (Jude et al., 2001; Turgeon et al., 2003; Little et al., 2004). Importantly, bands corresponding to the R_f values of unsaturated long-chain fatty acid glucuronides were observed in incubations of HLM with ¹⁴C-labeled UDPGA. The intensity of the bands was an order of magnitude lower for HEK293 cells expressing UGT2B7. Addition of BSA and HSA-FAF resulted in a concentration-dependent reduction in the band intensity. In contrast, the intensity of the putative FA glucuronide bands produced by incubations of HLM with ¹⁴C-labeled UDPGA increased in the presence of crude HSA.

Consistent with the postulated inhibitory effect of fatty acids on UGT2B7 activity, addition of oleic, linoleic, and arachidonic acids, individually and as a mixture, to incubations of HEK293 cell lysate expressing UGT2B7 decreased both AZT and 4MU glucuronidation activity, due to an increase in K_m (AZT) or S₅₀ (4MU) without a change in V_max, indicative of a competitive mechanism. Addition of BSA and HSA-FAF reversed the inhibition. The combined effect of oleic, linoleic, and arachidonic acids was similar to that of arachidonic acid alone. As noted previously, arachidonic acid is the most potent fatty acid inhibitor of UGT2B7 identified to date (Tsoutsikos et al., 2004). Similar to the reversal of fatty inhibition observed with HSA-FAF, the K_m/S₅₀ values for AZT and 4MU glucuronidation were lower when UGT2B7 was coexpressed with HSA, despite low HSA expression.

Whereas 2% BSA, 2% BSA-FAF, and 2% HSA-FAF caused an approximate 10-fold increase in the CL_int values for AZT glucuronidation by HLM and UGT2B7, the increase in CL_max for 4MU glucuronidation was somewhat lower (2- to 4-fold increase). It should be noted that although CL_int and CL_max both represent in vitro “clearances”, the two parameters are not equivalent and are thus not directly comparable (Houston and Kenworthy, 2000). Nevertheless, the data demonstrate that unsaturated long-chain fatty acids present in incubations influence the glucuronidation of UGT2B7 substrates that exhibit both “classic” Michaelis-Menten kinetics and positive homotropic cooperativity. Derived K_m values in previous studies have frequently been lower with recombinant UGT2B7 as the enzyme source compared with HLM. For example, K_m values for AZT, codeine, morphine (3- and 6-) and naproxen glucuronidation have variably been reported as 2- to 6-fold lower with recombinant UGT2B7, irrespective of whether the enzyme was expressed in a mammalian cell line (HEK293) or as “baculosomes” (baculovirus-mediated expression) (Soars et al., 2002; Court et al., 2003; Uchaipichat et al., 2006). These observations are consistent with a lower content of inhibitory unsaturated long-chain fatty acids in the expression systems.

Extrapolated “whole liver” CL_int values, from scaling of microsomal data, typically underpredict the known hepatic clearance of UGT2B7 substrates (Boase and Miners, 2002; Soars et al., 2002; Engtrakul et al., 2005; Riley et al., 2005). Despite the 10-fold higher CL_int values observed here for human liver microsomal AZT glucuronidation in the presence of BSA and HSA-FAF, predicted hepatic clearances (based on the approach of Uchaipichat et al., 2006) ranged from approximately 25 to 30 l/h (cf. actual value of 87 l/h). Predicted AZT hepatic glucuronidation clearances of 41 and 63 l/h have been reported based on kinetic constants obtained with human hepatocytes (Naritomi et al., 2003; Engtrakul et al., 2005). Other studies have generally demonstrated improved predictivity of in vivo drug glucuronidation clearance using human hepatocytes (Soars et al., 2002; Riley et al., 2005), although a trend to underprediction remains (Ito and Houston, 2004; Miners et al., 2006). Notably, mean K_m values observed here for AZT glucuronidation by HLM (87–89 μM) and recombinant UGT2B7 (40–52 μM) in the presence of BSA and HSA-FAF were essentially identical to the K_m (87 μM) for AZT glucuronidation by human hepatocytes reported by Engtrakul et al. (2005), which correctly predicted AZT as a high hepatic clearance drug. Because there is no evidence of extrahepatic AZT metabolic clearance, the poorer predictivity of the human liver microsomal CL_int (in the presence of albumin) must arise from a lower V_max. The reason for the lower turnover by the microsomal enzyme is unclear, but it may result from loss of active protein during HLM preparation, the presence of a noncompetitive inhibitor in microsomal incubations, or limited substrate and/or cofactor access to the lumenally orientated enzyme (despite addition of the pore forming agent alamethicin). The V_max values observed here for AZT glucuronidation by HLM are similar to those reported previously by this laboratory (Boase and Miners, 2002; Uchaipichat et al., 2006) and other laboratories (Court et al., 2003; Engtrakul et al., 2005).

The formation of fatty acid glucuronides during incubations of HLM with UDPGA indicates that fatty acids present in the microsomal membrane (as phospholipids) are hydrolyzed, presumably by the action of phospholipases, and released during the course of an incubation. It should be noted that dialysis of HLM against BSA in phosphate buffer did not change AZT glucuronidation kinetics (data not shown). Both BSA and HSA-FAF have the capacity to sequester inhibitory fatty acids, whereas fatty acid binding sites are presumably saturated in crude HSA. It is also likely that crude HSA contributes fatty acids to the incubation mixture, because fatty acids can desorb from binding sites on albumin despite high binding affinities (Hamilton, 2002). Fatty acid glucuronide formation was enhanced for incubations conducted in the presence of HSA (Fig. 4).

On the basis of improved predictivity of in vivo hepatic clearances from human hepatocyte kinetic data (i.e., CL_int), it has been suggested that HLM should not be used for predicting glucuronidation clearance or any kinetic parameters for metabolism or inhibition involving UGT enzymes (Engtrakul et al., 2005). Indeed, based on the under-prediction generally observed using microsomal kinetic data, there seems to be a general preference for using human hepatocytes to assess drug metabolic stability (Naritomi et al., 2003; McGinnity et al., 2005; Riley et al., 2005). Experimentally, however, HLM provide advantages over hepatocytes, particularly ease of use, decreased requirement for fresh tissue, and considerably lower cost. Previous results from this laboratory demonstrated that K_i values generated using HLM in the presence of BSA accurately predict the magnitude of inhibitory interactions involving glucuronidated drugs (Rowland et al., 2006; Uchaipichat et al., 2006). The present work similarly demonstrates that K_m values obtained with incubations of HLM and BSA/HSA-FAF are comparable with those observed with hepatocytes. Studies are in progress to further improve the predictivity of human liver microsomal kinetic data and characterize the universality of the albumin effect with enzymes other than UGT2B7.

Based on reported K_m (and K_i) values for UGT substrates, there is a general perception that glucuronidation is a “low-affinity” metabolic pathway (e.g., Williams et al., 2004). Data presented here suggest that previously reported K_m values for UGT2B7 substrates, and possibly those of other UGT enzymes, are overestimated by up to an order of magnitude due to an experimental artifact. Reconsideration of current notions regarding the binding affinities of UGT substrates is clearly warranted.