Introduction

Top Abstract Introduction Materials and Methods Results Discussion References

Acetaminophen (paracetamol; APAP) is a widely used nonprescription drug with mild analgesic and antipyretic properties. When used at therapeutic dosages, the principal route of excretion of APAP is glucuronidation (52-57% total urinary metabolites) with a lesser contribution by sulfation (30-44%) and oxidation (<5%) (Prescott, 1983; Patel et al., 1992). Although normally a very safe drug, APAP may cause hepatotoxicity in some patients after accidental or intentional overdose. Under these circumstances, sulfation becomes saturated such that glucuronidation is the predominant pathway (66-75% total urinary metabolites), whereas an increased amount of drug (7-15%) is oxidized to the highly reactive metabolite, N-acetyl-p-benzoquinone imine (Prescott, 1983; Nelson, 1990; Vermeulen et al., 1992).

Interindividual variability in APAP glucuronidation is likely to influence the risk for developing hepatotoxicity after APAP overdose. In support of this, fasting, which depletes essential cofactors necessary for efficient conjugation including the glucuronidation cosubstrate UDP-glucuronic acid (UDPGA), has been identified as an important risk factor for APAP hepatotoxicity (Whitcomb and Block, 1994). Furthermore, patients with Gilbert's syndrome, who have an inherited defect of UDP-glucuronosyltransferase 1A1 (UGT1A1), show lower serum APAP glucuronide concentrations and higher urinary excretion of N-acetyl-p-benzoquinone imine-derived metabolites after administration of APAP compared with normal subjects (de Morais et al., 1992). Finally Gunn rats, which are deficient in all of the UGT1A subfamily isoforms but are normal in other metabolic pathways, are from 110- to 230-fold more susceptible to APAP toxicity compared with normal rats (de Morais and Wells, 1988).

Although APAP has been used extensively as a probe substrate for the study of glucuronidation in humans and human tissues, at present it is unclear as to which UGT isoform is primarily responsible for this activity. Currently only UGT1A6 and UGT1A9 have been shown to glucuronidate APAP (Bock et al., 1993). Enzyme kinetic studies demonstrated that UGT1A6 is a relatively high-affinity isoform (K_m = 2 mM), whereas UGT1A9 is a low-affinity isoform (K_m = 50 mM). In addition, arylhydrocarbon receptor agonists, which induce UGT1A6 in human primary hepatocytes and Caco-2 cells (Munzel et al., 1996, 1998), also enhance APAP-UGT activity in humans (Bock et al., 1987). Based on these findings, APAP-UGT activity has been proposed as an indicator of UGT1A6-mediated glucuronidation in human tissues (de Wildt et al., 1999; Fisher et al., 2000), despite evidence for involvement of other UGT isoforms.

Deficient APAP glucuronidation in Gilbert's patients suggests that UGT1A1, the relevant bilirubin-UGT (Bosma et al., 1994), also contributes to this activity (Douglas et al., 1978; de Morais et al., 1992). This is further supported by the finding that APAP-UGT activity is induced by phenobarbital-type compounds (Tomlinson et al., 1996), which are reported to induce UGT1A1 in primary human hepatocytes (Ritter et al., 1999). In addition, UGT1A1 also appears to be induced strongly by arylhydrocarbon receptor agonists in this hepatocyte model (Ritter et al., 1999). On the other hand, some studies have failed to show reduced APAP glucuronidation in Gilbert's patients (Ullrich et al., 1987). Furthermore, APAP was reported not to be a substrate for UGT1A1 (King et al., 1996). However, in that study, expressed UGTs known to mediate APAP-UGT activity were not included as positive controls, and incubations contained saccharolactone, which can inhibit APAP glucuronidation (Alkharfy and Frye, 2001).

Consequently, the aim of the present study was to initially characterize interindividual variability in APAP glucuronidation using a bank of HLMs. UGT isoforms relevant to this variability were then identified by screening a panel of recombinant UGTs for activity, comparison of enzyme kinetics in recombinant UGTs with liver microsomes, and correlation of APAP-UGT activity with other UGT activities and immunoquantified UGT1A6 content. Finally, these data were used to model the effect of APAP concentration on the relative contribution of the identified UGT isoforms to hepatic microsomal APAP-UGT activity.

	Materials and Methods

Top Abstract Introduction Materials and Methods Results Discussion References

Chemicals and Other Reagents. Unless otherwise indicated, most reagents including Brij 58 (polyoxyethylene 20-cetyl ether), alamethicin, UDPGA (sodium salt), APAP, and APAP glucuronide were purchased from Sigma (St. Louis, MO), whereas 2-acetamidophenol was from Aldrich Chemical (Milwaukee, WI). Recombinant UGTs 1A1, 1A3, 1A4, 1A6, 1A9, 2B7, 2B15, and the WB-UGT1A6 antibody were obtained from Gentest (Woburn, MA). Recombinant UGTs 1A7 and 1A10 were from Panvera (Madison, WI). All recombinant UGTs were expressed using recombinant baculovirus-infected Sf-9 insect cells. The glucuronidation activity of each of the expressed UGTs had been substantiated by the manufacturers using the following substrates: bilirubin and estradiol (UGT1A1); 7-hydroxy-4-trifluoromethylcoumarin (UGTs 1A3, 1A8, 1A10, and 2B15); trifluoperazine (UGT1A4); 1-napthol (UGTs 1A6 and 1A10); propofol (UGT1A9); morphine (UGT2B7); and octyl-gallate (UGT1A7). In addition, immunoblotting with an antibody specific for the conserved C-terminal region of all UGT1A isoforms (WB-UGT1A; Gentest) showed that the content of immunodetectable UGT protein was similar for all of the expressed UGT1A isoforms used.

Liver Microsomes. Livers were selected randomly from frozen banks maintained at the Division of Clinical Pharmacology, Department of Pharmacology and Experimental Therapeutics, Tufts University School of Medicine (n = 36) and the Department of Clinical Pharmacology, Flinders University School of Medicine (n = 20). Donors were primarily Caucasian, but included seven African-Americans and three Hispanics. Other available donor information included gender (24 females, 32 males) and age (average, 36 years; range, 2-80 years). Microsomes were prepared from frozen liver by differential centrifugation as previously described (Court and Greenblatt, 1997b). The resultant pellet was reconstituted in 20% glycerol/phosphate buffer, aliquoted, and stored at 80°C. Frozen microsomes were thawed once only immediately before use. The quality of the liver samples used was ascertained by reference to other UGT and CYP enzyme activities measured in this laboratory using the same set of liver samples. Livers that consistently showed low-activity values (>2-fold lower for all measured activities) relative to the median-activity value for the entire liver set were excluded from the study.

APAP Glucuronidation Assay. Microsomal APAP glucuronidation activities were measured by high-performance liquid chromatography with ultraviolet detection using the method described previously (Court and Greenblatt, 1997a, 1997b). Unless otherwise indicated, incubations were performed in 50 mM phosphate buffer, pH 7.5, at 37°C with 5 mM MgCl₂. Initial rate conditions with respect to time and protein concentration for the formation of APAP glucuronide were established in preliminary studies. For determination of liver microsomal activities at substrate concentrations of 0.5 mM APAP and 20 mM UDPGA, incubation time was 120 min, and protein concentration was 1 mg/ml. Recombinant UGTs and vector control were also assayed for APAP-UGT activity at 50 and 0.5 mM APAP using 20 mM UDPGA, 0.5 mg/ml protein concentration, and 120 min of incubation time.

Enzyme Kinetic Analysis. Each set of substrate concentration (S) and velocity (V) data were fitted to the appropriate enzyme kinetic model by nonlinear least-squares regression analysis (Sigmaplot; SPSS, Chicago, IL) to derive the enzyme kinetic parameters V_max (maximal velocity) and K_m (substrate concentration at half-maximal velocity) as described previously (Court et al., 2001). Both the Michaelis-Menten model (eq. 1) and the substrate activation model (eq. 2), which incorporates the exponent (n), were used:

(1)

(2)

In addition to these two single enzyme models, three dual enzyme models were used consisting of two Michaelis-Menten enzymes, two substrate activated enzymes, or one Michaelis-Menten enzyme with one substrate activated enzyme. Choice of the appropriate model was determined by a number of criteria, including visual inspection of data plots (Michaelis-Menten and Eadie-Hofstee), distribution of residuals, size of the sum of squared residuals, and the standard error of the estimates. In a number of livers, apparent inhibition was observed at high APAP concentrations in addition to activation at low APAP concentrations. Consequently, kinetic parameters were derived by truncation of activity values obtained at APAP concentrations above 30 mM and fitting of the data to the substrate activation model.

Other Activities. Four other UGT activities (bilirubin-UGT, imipramine-UGT, propofol-UGT, and androsterone-UGT) and a CYP activity (phenacetin o-deethylation) were also measured using the 20 livers from the Australian donors. Linearity of product formation with respect to protein concentration and incubation time were established in preliminary experiments. Phenacetin o-deethylation was measured by high-performance liquid chromatography with UV absorbance detection as previously described (Venkatakrishnan et al., 1998). Substrate concentration was 50 µM, protein concentration was 0.5 mg/ml, and incubation time was 20 min. For the UGT assays, product formation rates were determined by measuring the incorporation of [¹⁴C]UDPGA as previously described (Bansal and Gessner, 1980). Product was separated from substrate by thin-layer chromatography and quantified by PhosphorImager (Molecular Dynamics, Sunnyvale, CA) using a standard curve constructed from [¹⁴C]UDPGA. Final UDPGA concentration was 2 mM except for androsterone-UGT activity, which used 2.5 mM. Protein concentration was 0.5 mg/ml, except for propofol-UGT activity, which used 0.1 mg/ml, and androsterone-UGT activity, which used 1 mg/ml. MgCl₂ concentration was 5 mM, except for androsterone-UGT activity, which used 4 mM. Aglycone substrate concentrations and incubation times were 20 µM and 60 min for bilirubin-UGT activity, 500 µM and 90 min for imipramine-UGT activity, 50 µM and 20 min for propofol-UGT activity, and 30 µM and 120 min for androsterone-UGT activity.

UGT1A6 Protein Content. The relative content of UGT1A6 protein in hepatic microsomes from the 20 Australian donors was determined by adapting a previously described immunoblotting method (Court et al., 2001). Briefly, 25 µg of microsomal protein was separated by sodium dodecyl sulfate-polyacrylamide gel electrophoresis using a 7% separating gel with a 4% stacking gel. Proteins were then transferred electrophoretically to polyvinyl difluoride membrane (Immobilon-P; Millipore, Bedford, MA). Blots were blocked in 10% powdered nonfat milk in Tris-buffered saline-Tween (0.15 M NaCl, 0.04 M Tris, pH 7.7, and 0.1% Tween 20) and then incubated in Tris-buffered saline-Tween/1% milk containing a 1:1000 dilution of a polyclonal antipeptide UGT1A6 antibody (WB-UGT1A6). After washing, the blots were incubated in a 1:10,000 dilution of horseradish peroxidase-conjugated secondary antibody (Sigma) and washed, and chemiluminescence detection was performed (Super Signal; Pierce, Rockford, IL) with exposure to radiographic film (Biomax MR; Kodak, Rochester, NY). The film was then scanned, and the area and density of identified bands was quantified using a densitometer (Bio-Rad, Hercules, CA). The linearity of the measurements under the conditions used was established by loading serial dilutions of microsomes from the liver containing the highest amount of UGT1A6 (HL45). Values are expressed relative to the liver containing the lowest amount of UGT1A6 (HL48).

Statistical Analyses and Modeling. Statistical analyses were performed using the Sigmastat software package (SPSS, Chicago, IL). All activities were determined in duplicate and averaged. A P value of less than 0.05 was considered statistically significant. Comparisons between gender groups were performed using the Mann-Whitney rank-sum test. APAP-UGT activities measured at 0.5 mM APAP concentration were correlated to each of the other measured activities and relative UGT1A6 protein content using the Spearman rank order correlation analysis. This nonparametric method was used because of the skewed distribution of the APAP-UGT activity data. Multiple regression analysis of rank transformed data was then performed using bilirubin-UGT activity (as an isoform-specific marker for UGT1A1), relative UGT1A6 content (for UGT1A6), and propofol-UGT activity (for UGT1A9) as independent variables and APAP-UGT activity as the dependent variable as follows:
where b₀ is the intercept and b₁, b₂, and b₃ are the standardized regression coefficients. The relative contribution of each independent variable to the dependent variable was then calculated by dividing each standardized regression coefficient by the sum of the standardized regression coefficients and expressing the result as a percentage.

	Results

Top Abstract Introduction Materials and Methods Results Discussion References

Interindividual Variability in APAP Glucuronidation Activity. APAP glucuronidation measured in hepatic microsomes at a clinically relevant APAP concentration (0.5 mM) ranged from 33 to 508 pmol/min/mg representing more than a 15-fold range in activities. The distribution of these data was not normal (Kolmogorov-Smirnov normality test; P = 0.026) with significant skewing toward lower activities (Fig. 1). Mean and median activity values were 174 and 145 pmol/min/mg, respectively.

View larger version (35K): [in this window] [in a new window]   Fig. 1.   Frequency distribution of APAP-UGT activities measured in vitro at 0.5 mM APAP concentration using HLMs derived from 56 donors including 24 females and 32 males.

APAP Glucuronidation by Expressed UGTs. APAP glucuronidation activities were measured at a clinically relevant (0.5 mM) and saturating (50 mM) APAP concentration using all commercially available recombinant UGTs. At 0.5 mM APAP concentration (Fig. 2A), four isoforms were shown to be active, with UGT1A9 showing the highest activity followed by UGT1A1, UGT1A6, and UGT1A10. At 50 mM APAP concentration (Fig. 2B), UGT1A9 clearly showed the highest activity, more than 3-fold higher than UGT1A1, the next highest. Most of the remaining UGTs showed very low although measurable activity. No activity could be detected with UGT1A4 or the vector control preparation.

View larger version (22K): [in this window] [in a new window]   Fig. 2.   APAP-UGT activities measured at 0.5 mM (A) and 50 mM (B) APAP concentration using microsomes from baculovirus-infected insect cells expressing individual UGT isoforms or baculovirus vector control.

Hepatic Microsome and Expressed UGT Kinetics. Enzyme kinetic analyses were performed using representative high (HLM1, HLM2, and HLM3), intermediate (HLM4, HLM5, and HLM6), and low (HLM7, HLM8, and HLM9) activity liver microsomes, and recombinant UGTs normally expressed in liver that showed activity at 0.5 mM APAP concentration (UGT1A1, UGT1A6, and UGT1A9). Analyses were conducted initially using a fixed high APAP concentration and varied UDPGA concentration (UDPGA kinetics) followed by a fixed high UDPGA concentration and varied APAP concentration (APAP kinetics). Fitted kinetic parameters are given in Table 1.

View larger version (39K): [in this window] [in a new window]   Fig. 3.   Enzyme kinetics of acetaminophen glucuronidation by HLMs. Shown are representative Eadie-Hofstee plots of data generated using high-activity (A and D), intermediate-activity (B and E), and low-activity (C and F) livers. Kinetics were measured using either 50 mM APAP and UDPGA concentrations between 0.25 and 50 mM (A-C) or 20 mM UDPGA and APAP concentrations between 0.1 and 50 mM (D-F). Also shown for each plot are curves representing best-fit estimates of these data determined by nonlinear curve fit to either eq. 1 (A, B, and F) or eq. 2 (C, D, and E).

View larger version (38K): [in this window] [in a new window]   Fig. 4.   Enzyme kinetics of acetaminophen glucuronidation by microsomes from insect cells expressing UGT1A1 (A and D), UGT1A6 (B and E), and UGT1A9 (C and F). Shown are Eadie-Hofstee plots of data generated using either 50 mM APAP and UDPGA concentrations between 0.25 and 50 mM (A-C) or 20 mM UDPGA and APAP concentrations between 0.1 and 50 mM (D-F). Also shown for each plot are curves representing best-fit estimates of these data determined by nonlinear curve fit to either eq. 1 (A, B, C, and E) or eq. 2 (D and F).

Correlational Analyses. Correlational analyses (Fig. 5, A-C) with APAP-UGT activity as the dependent variable indicated a strong relationship with propofol-UGT activity (Spearman rank order correlation coefficient: r = 0.85; P < 0.001) and bilirubin-UGT activity (r = 0.66; P = 0.0016). A weaker correlation was also found with imipramine-UGT activity (r = 0.47; P = 0.037), but not with any other regressor (Table 2). Exclusion of data from livers with the two highest APAP-UGT activities in this analysis had minimal or no influence on correlation values. Multiple regression analysis (Table 2) was then performed using variability markers for hepatic UGT isoforms shown in this study to have significant APAP-UGT activity, including bilirubin-UGT activity (UGT1A1), relative UGT1A6 protein content (UGT1A6), and propofol-UGT activity (UGT1A9). The r value for the correlation increased to 0.92 (Fig. 5D; Table 2) with propofol-UGT as a major coregressor (P < 0.001), bilirubin-UGT as a minor coregressor (P = 0.045), and relative UGT1A6 protein content (P = 0.46) having insignificant influence.

View larger version (44K): [in this window] [in a new window]   Fig. 5.   Comparison of APAP-UGT activities to activities mediated by UGT1A1 (bilirubin-UGT; A) and UGT1A9 (propofol-UGT; C), and immunoquantified UGT1A6 (B) measured in the same set of HLMs (n = 20). Shown for each comparison are the correlation coefficients of rank transformed data (r). Also shown is a comparison of the rank-transformed APAP-UGT activity data with the rank values predicted from a multiple linear regression analysis using rank-transformed bilirubin-UGT activities, propofol-UGT activities, and UGT1A6 protein content as independent variables and rank-transformed APAP-UGT activity as the dependent variable (D).

Relative Contribution of UGT Isoforms. Based on the standardized regression coefficients from the multiple regression analysis, it was estimated that 23 ± 11% (mean ± S.E.) of total APAP-UGT activity (at 0.5 mM APAP concentration) was contributed by UGT1A1, 15 ± 10% was contributed by UGT1A6, and 63 ± 10% was contributed by UGT1A9. The influence of clinically relevant APAP concentrations (50 µM-5 mM) on this predicted relative contribution to total APAP-UGT activity was then modeled using the APAP kinetic parameters previously determined for expressed UGT1A1, UGT1A6, and UGT1A9. V_max values were adjusted such that the relative contribution of each isoform at 0.5 mM APAP concentration approximated values determined by the multiple regression analysis. Based on this model (Fig. 6), UGT1A9 was found to be the predominant isoform (61% of total activity) at 50 µM APAP concentration with lesser contributions from UGT1A6 (29%) and UGT1A1 (10%). With APAP concentrations as high as 5 mM, the relative contribution of UGT1A1 increased almost 4-fold to 39%, approaching that of UGT1A9, which had decreased slightly to 58%; whereas UGT1A6 had decreased to only 3% of total activity.

View larger version (23K): [in this window] [in a new window]   Fig. 6.   Effect of APAP concentration on the relative contribution of UGT1A1, UGT1A6, and UGT1A9 to total APAP-UGT activity in HLMs. Rates of APAP glucuronidation were calculated using APAP kinetic parameters experimentally determined for expressed UGT1A1, UGT1A6, and UGT1A9 at 10-µM intervals from 50 µM to 5 mM and expressed as a percentage of the sum of activities for all three isoforms. Vmax values for each UGT isoform were adjusted such that relative activity values calculated for each isoform at 0.5 mM APAP equaled relative activity values determined for each isoform in the multiple regression analysis.

Gender Differences in APAP-UGT Activity. In initial studies using the entire set of liver microsomes, APAP-UGT activities measured at 0.5 mM APAP concentration were found to be almost 50% higher (P = 0.047) in livers from 32 male donors (median, 174 pmol/min/mg; 25-75% percentiles, 109-245 pmol/min/mg) compared with 24 female donors (median, 117 pmol/min/mg; 25-75% percentiles, 72-172 pmol/min/mg) (Figs. 1 and 7A). The microsomal activities and UGT1A6 protein content determined in the subset of samples for the correlational analyses were used additionally to isolate a possible cause for this gender dependent difference in APAP-UGT activity (Table 3). Of these, only relative UGT1A6 protein content approached a statistically significant difference with approximately 50% higher UGT1A6 protein in male livers compared with female livers (P = 0.076; Fig. 7B).

View larger version (18K): [in this window] [in a new window]   Fig. 7.   Gender differences in human liver microsomal APAP-UGT activity (A; 32 males and 24 females) and immunoquantified UGT1A6 protein content (B; 10 males and 10 females). Data points represent the average of duplicate determinations for each HLM sample. Also shown are the median values for each group (dashed lines) and the P value for the comparison using the Mann-Whitney rank sum test.

	Discussion

Top Abstract Introduction Materials and Methods Results Discussion References

Hepatic microsomal APAP-UGT activities showed a substantial degree of interindividual variability (more than 15-fold). This is somewhat higher than the 7-fold variability recently reported (Fisher et al., 2000), although fewer livers were used (20 versus 56 in the present study) decreasing the likelihood of finding slow and fast metabolizers. In an earlier in vivo study investigating interindividual differences in urinary excretion of APAP metabolites by 158 human subjects, only about 4-fold variation was observed in the fractional urinary excretion of APAP glucuronide (Patel et al., 1992). This difference could be explained by factors other than hepatic glucuronidation, which can affect the fractional urinary excretion of a metabolite, including extrahepatic glucuronidation, biliary excretion of glucuronide, and urinary flow rate (Miners and Birkett, 1993).

At least three different UGT isoforms were found to mediate APAP glucuronidation in HLMs. Confirming results from a previous study (Bock et al., 1993), UGT1A6 was found to be a high-affinity, low-capacity enzyme, whereas UGT1A9 was a low-affinity, high-capacity enzyme. To our knowledge this is the first study to show that UGT1A1 is also a proficient APAP-UGT with intermediate affinity and capacity compared with UGT1A6 and UGT1A9. Most other UGTs (except UGT1A4) showed some activity at high substrate concentrations, whereas UGT1A10 showed activities similar to UGT1A6. Although UGT1A10 is not expressed in liver, it may be important in the extrahepatic metabolism of APAP particularly in the gastrointestinal tract (Tukey and Strassburg, 2001). Recombinant preparations of at least two other UGT isoforms expressed in the liver, (UGT2B4 and UGT2B17) were not available for this study, but should be evaluated for activity in future studies. In addition, the contribution of other, yet to be identified UGT isoforms cannot be excluded.

Because the relative abundance of UGT isoforms in liver is unknown, several different approaches were used in an attempt to identify the relevant APAP-UGT. Initially, enzyme kinetic parameters were determined using representative liver microsomes and the three most active expressed UGTs with the rationale that K_m values of liver microsomes should most closely approximate the K_m value of the responsible UGT. APAP K_m values suggest that UGT1A1 is primarily responsible for activity in high- and intermediate-activity livers, whereas UGT1A9 and perhaps other low-affinity UGTs contribute to the activity in two of the three low-activity livers. UDPGA kinetics although primarily performed to establish an appropriate UDPGA concentration for the APAP kinetic study were not useful in identifying a UGT isoform. UDPGA K_m values differed minimally between livers and tended to be more than 2-fold higher than values determined for the expressed UGTs. This may be because at such a high acetaminophen concentration (50 mM) the liver microsomal activity primarily reflects low-affinity UGT isoforms. Alternatively, there may be enzyme inhibition at this acetaminophen concentration as was found for six of the nine livers in the APAP kinetic study. Enzyme inhibition at high acetaminophen concentrations was also observed with expressed UGT1A1 and 1A6, but not with UGT1A9. Interestingly, two of the three livers that did not show inhibition kinetics (HLM7 and HLM8) also had relatively high K_m values, suggesting involvement of UGT1A9.

Consistent with previous reports (Court and Greenblatt, 1997b; Fisher et al., 2000), most livers in this study showed substrate activation kinetics. Atypical enzyme kinetics are explained commonly by evoking a two-site model in which the enzyme can bind simultaneously two substrate molecules, either within the same active site or at two distinct binding sites, one of which is the active site and the second is a modulatory site (Korzekwa et al., 1998). Because UGTs can form dimers (Gschaidmeier and Bock, 1994; Ikushiro et al., 1997; Meech and Mackenzie, 1997), these sites could also be present on separate molecules within the same oligomeric complex. Interestingly, UGT1A6 was the only UGT studied that did not show activation kinetics. Compared with other UGTs, substrates glucuronidated by UGT1A6 tend to be smaller, planar compounds suggesting relatively restricted access to the enzyme active site (Ebner and Burchell, 1993). Differential activation kinetics could therefore be explained by a modulatory binding site that is located within the catalytic sites of UGT1A9 and UGT1A1, but is not present within the more restricted active site of UGT1A6.

Correlational analysis confirmed involvement of UGT1A1 and UGT1A9 in APAP glucuronidation by HLMs. UGT activities used in the analysis were chosen based on evidence in the literature for isoform selectivity. Of these activities, bilirubin-UGT activity and to a lesser extent propofol-UGT activity are most likely selective for UGT1A1 and UGT1A9 (respectively) in liver microsomes. UGT1A6 protein was immunoquantified because all of the characterized UGT1A6 substrates also appear to be good substrates for UGT1A9 and other isoforms. The weak correlation to imipramine-UGT activity indicates that imipramine is probably not a selective UGT1A4 substrate because expressed UGT1A4 did not glucuronidate APAP. Lack of correlation to the CYP1A2 index activity (phenacetin-o-deethylation) suggests that interindividual variability in APAP glucuronidation is not related to arylhydrocarbon receptor-mediated enzyme induction.

The use of rank-transformed data for the multiple linear regression enabled estimation of the relative contribution of UGT isoforms to microsomal APAP-UGT activity at 0.5 mM APAP concentration. Based on this analysis, UGT1A9 (63% of activity) appears to be the predominant isoform, with lesser contributions from UGT1A1 (23% of activity) and UGT1A6 (15% of activity). Enzyme kinetic studies with these isoforms indicated that APAP concentration could impact this relationship. Consequently a kinetic model was constructed by combining multiple regression data and enzyme kinetic parameters. This model predicts that whereas UGT1A9 consistently provides more than 55% of activity over the clinically relevant concentration range (50 µM-5 mM), the relative contribution of UGT1A6 and UGT1A1 is highly concentration-dependent. UGT1A6 appears to be most active (>29% contribution) at relatively low concentrations (<50 µM), whereas UGT1A1 is most active (>28% contribution) at toxic concentrations (>1 mM).

In vivo evidence from patients with Gilbert's syndrome, which results in decreased UGT1A1 activity, appears to support a role for UGT1A1 at relatively high APAP concentrations. In the previously cited study (de Morais et al., 1992), Gilbert's patients showed 31% lower APAP clearance by glucuronidation compared with control subjects, with the majority of the difference restricted to the first 2 h after APAP administration. During this time, average plasma APAP concentrations decreased from approximately 400 to 80 µM, corresponding to a concentration range predicted by the kinetic model to show a marked decrease in the contribution of UGT1A1 to APAP-UGT activity (from 21 to 12%). Such a concentration effect may also explain why differences in APAP glucuronidation are not consistently seen in studies of Gilbert's patient's in that dosage and administration route of APAP differs substantially between studies (de Morais et al., 1992; Patel et al., 1992). On the other hand, heterogeneity of APAP metabolism has also been described within a group of Gilbert's patients in that some patients have relatively low APAP glucuronidation relative to control subjects, whereas some are not different despite receiving the same dose of APAP (Esteban and Perez-Mateo, 1999). This could be explained by relatively higher APAP-UGT activity mediated by UGT1A9 (and perhaps UGT1A6) in the apparently normal Gilbert's patients, which compensates for reduced UGT1A1 expression.

In agreement with previous reports of higher in vivo clearance of APAP by glucuronidation in male compared with female subjects (Abernethy et al., 1982; Divoll et al., 1982; Miners et al., 1983; Critchley et al., 1986; Bock et al., 1994), APAP-UGT activity was approximately 50% higher in livers from male compared with female donors. Further investigation of this phenomenon indicated that it does not result from differences in UGT1A9- or UGT1A1-mediated activity, but may result from greater UGT1A6 protein content of male compared with female liver microsomes, although study of more livers is needed to confirm this difference. A similar gender difference in p-nitrophenol glucuronidation and immunoquantified phenol-UGT protein (presumably UGT1A6) has also been reported for rat liver microsomes (Catania et al., 1995).

In conclusion, the results of this study indicate that there is substantial interindividual variability in APAP glucuronidation by HLMs. At least three UGTs were found to contribute to this variability, with the relative contribution of each UGT isoform dependent on APAP concentration. The majority of interindividual variability appears to reflect differences in activity of UGT1A9 over a clinically relevant range of APAP concentrations. UGT1A1 was also predicted to contribute substantially at toxic APAP concentrations, whereas UGT1A6 was most active at relatively low APAP concentrations. APAP-UGT activity was not found to be a useful predictor of hepatic microsomal UGT1A6 protein content under the assay conditions used.