Introduction

Top Abstract Introduction Materials and Methods Results Discussion References

The glucuronidation of tertiary aliphatic or aromatic amines leading to the formation of a polar quaternary ammonium-linked glucuronide metabolite (N⁺-glucuronide) (Tukey and Strassburg, 2000) plays a significant role in the metabolism of various xenobiotics. Examples of these substrates include various H₁ antihistamine and tricyclic antidepressant drugs (Hawes, 1998; Dalgaard and Larsen, 1999; Mey et al., 1999) in the case of aliphatic amines and anastrozole (McCann et al., 1997), lamotrigine (Sinz and Remmel, 1991), nicotine (Caldwell et al., 1992), and tioconazole (MacRae et al., 1990) in the case of aromatic amines. Basic knowledge is lacking regarding this metabolic route, especially in the case of glucuronidation at an aromatic tertiary amine. These deficiencies include the identification of the UDP-glucuronosyltransferase (UGT³) enzymes involved in catalysis at an aromatic tertiary amine and substrate specificities.

In humans, members of the UGT1A and UGT2B subfamilies are considered to be mainly involved in the catalysis of the glucuronidation of xenobiotics. However, primarily UGT1A enzymes are known to catalyze N-glucuronidation, where metabolite formation is far more selective for tertiary than primary or secondary amine substrates. In fact, only UGT1A3 and UGT1A4 have been demonstrated to catalyze the glucuronidation of tertiary amine substrates (Green et al., 1995, 1998; Green and Tephly, 1996, 1998; Breyer-Pfaff et al., 2000). However, whereas such has been demonstrated for many substrates with respect to aliphatic tertiary amines, only in the case of UGT1A4 catalysis of lamotrigine glucuronidation had aromatic tertiary amines been investigated (Green et al., 1995). Previous structure-metabolism relationship studies of glucuronidation that relate physicochemical features to metabolic kinetics are limited and involve O-glucuronidation.

To investigate both the UGT enzyme(s) involved in the catalysis of glucuronidation at an aromatic tertiary amine and the structure-metabolism relationships thereof, a series of 1-substituted imidazoles (1a-1h) (Fig. 1), which appropriately differ in their electronic (^para,  value for the substituent in the para position of the phenyl ring), lipophilic (log P; or ^para,  value for the substituent in the para position of the phenyl ring), and pK_a properties, was selected for study (Kubinyi, 1995). The reasons why this series was selected for study include the following.

View larger version (12K): [in this window] [in a new window]   Fig. 1.   Chemical structures of the 1-substituted imidazoles and the site of N-glucuronidation.

1) Many drugs contain five-membered heterocyclic ring systems as part of their structure, and examples are known where a major N⁺-glucuronide metabolite is formed in humans (MacRae et al., 1990; Rush et al., 1990, 1992; McCann et al., 1997). 1-Substituted imidazoles serve as a model for such ring systems, having only one aliphatic-like tertiary amine and one aromatic-type tertiary amine at the 1 and 3 positions, respectively. Also, there is generally greater commercial availability or synthetic accessibility of imidazoles than polyaza ring systems such, as triazoles or tetrazoles.

2) 1-Substituted imidazoles, including 1-phenyl substituted imidazoles, are appropriate substrates to investigate correlations between physicochemical properties (electronic, lipophilic, and pK_a) of the substrates and their kinetic parameters (K_m, V_max, and V_max/K_m).

3) A previous study from this laboratory of the prototype molecule 1-phenylimidazole (1a) demonstrated the definitive identification of the N⁺-glucuronide metabolite in human liver microsomes and that only human UGT1A3 and UGT1A4 of the nine UGT enzymes examined catalyzed N-glucuronidation (Vashishtha et al., 2000).

The specific objectives of this study were to identify the UGTs enzyme(s) involved in the catalysis of the glucuronidation of the series 1 compounds and to determine, for the identified UGT(s), the kinetic parameters for each 1-substituted imidazole and the relative importance of various physicochemical parameters in the N-glucuronidation of the selected series 1 compounds.

	Materials and Methods

Top Abstract Introduction Materials and Methods Results Discussion References

Chemicals. 1-Phenylethylimidazole (1c) was synthesized by a literature procedure (Salerno et al., 1999). All other substrates [1-phenylimidazole (1a), 1-(4-methoxyphenyl)imidazole (1d), 1-(4-chlorophenyl)imidazole (1e), 1-(4-bromophenyl)imidazole (1f), 1-(4-trifluoromethylphenyl)imidazole (1g), 1-(4-nitrophenyl)imidazole (1h) (Transworld Chemicals, Rockville, MD), and 1-benzylimidazole (1b) (Lancaster, Windham, NH)] were obtained from commercial sources. The metabolite 1-phenylimidazole N⁺-glucuronide was synthesized by a literature procedure (Vashishtha et al., 2000). Perchloric acid (BDH Chemicals, Toronto, Ontario, Canada), methimazole, Brij-58, CHAPS, Lubrol PX, Triton X-100, Tris base, alamethicin, ethyl 2-pyridylacetate, UDP-glucuronic acid (UDPGA) (ammonium salt), magnesium chloride, and D-saccharic acid 1,4-lactone (Sigma, St. Louis, MO) were also obtained commercially. Emulgen 911 was a gift from the Kao Corporation (Tokyo, Japan). Although all the chemicals were of reagent grade, the organic solvents (EM Science, Gibbstown, NJ) were of HPLC grade. Double-distilled water, deionized and purified by a Milli-Q water system (18 ± 0.05 ohm/cm) (Millipore Corporation, Bedford, MA), was used. HPLC mobile phase solvents were filtered through Millipore 0.45-µm filters before use.

HPLC Analysis. Separation and quantification of the glucuronide metabolites produced in recombinant cell microsomes were achieved by gradient reversed-phase HPLC on a chromatographic system consisting of two HPLC pumps (Waters model 501, Milford, MA), an automated gradient controller (Waters model 680), and a variable wavelength absorbance detector (Waters model 486). Samples were injected via an autosampler SCL-10A (Shimadzu Corporation, Kyoto, Japan), and the separation was achieved on an Ultracarb C₁₈ analytical column (ODS 30; 4.6 × 250 mm; Phenomenex, Torrance, CA) packed with 5-µm diameter particles. The analytical column, maintained at ambient temperature, was protected using security guard C₁₈ cartridges (4 × 3.0 mm; Phenomenex). Data acquisition and integration were performed using a Waters Maxima 820 chromatography workstation. The total flow rate of the mobile phase was 1.0 ml/min. The chromatographic conditions regarding the solvent compositions (solvent A/solvent B) of the mobile phase consisting of solvent A (10 mM perchloric acid, pH 2.5) and solvent B (acetonitrile), the wavelength of maximum absorption, and the external standards (in parenthesis) for the quantitative analysis of the glucuronide of each substrate were as follows: 1a, 88:12 to 99:1 for 9 min, 88:12 for 10 to 15 min, 236 nm (2-ethylpyridyl acetate); 1b, 88:12 for 5 min, 84:16 for 6 to 20 min, 220 nm (methimazole); 1c, 88:12 for 10 min, 80:20 for 11 to 28 min, 220 nm (methimazole); 1d, 88:12 for 5 min, 85:15 for 6 to 15 min, 240 nm (methimazole); 1e, 90:10 for 10 min, 80:20 for 11 to 25 min, 240 nm (methimazole); 1f, 88:12 for 10 min, 84:16 for 11 to 25 min, 240 nm (methimazole); 1g, 78:22 for 5 min, 75: 25 for 6 to 25 min, 243 nm (1e).

Identification of Human UGT Enzyme(s) Catalyzing the N-Glucuronidation of 1-Substituted Imidazoles. Microsomal recombinant human UGT enzymes, expressed in human lymphoblastoid cells [UGT1A1, UGT1A4, UGT1A6, UGT1A9, UGT2B15, and control microsomes (without vector)] were purchased from GENTEST (Woburn, MA). Microsomes from Spodoptera frugiperda-9 insect cells infected with a baculovirus containing cDNA for human UGT1A3, UGT1A7, UGT1A10, and UGT2B7, and S. frugiperda-9-WT control baculosomes were purchased from Pan Vera (Madison, WI). The suppliers gave the protein content and glucuronidation activity data of a reference substrate of each expressed enzyme. Also, in the case of UGT1A3, 4-nitrophenol served as an in-house positive control, where under the following general UGT-screening conditions, the glucuronidation activity was determined to be 92.7 ± 2.4 pmol/min/mg of protein. The control products were used as negative controls for their respective UGT preparations.

Enzyme Kinetic Studies of N-Glucuronidation of 1-Substituted Imidazoles using Human UGT1A4 Expressed in Baculovirus-Infected Insect Cells. Supersomes (GENTEST) prepared from baculovirus-infected insect cells (BTI-TN-5B1-4) expressing human UGT1A4 cDNA were used to determine the kinetic parameters of the substrates. For this preparation, the manufacturer determined that the rate of formation is 1.14 nmol/min/mg of protein for glucuronidation at an aliphatic tertiary amine group of trifluoperazine. Using 1-phenylimidazole as a prototypic substrate, the incubation conditions were optimized with respect to pH, latency-disrupting agents, time of incubation, and the protein concentration required to give the linear rate of formation of the glucuronide. The effect of pH on the rate of glucuronidation was examined in the range of 6.0 to 9.0 (6.0, 6.5, 7.0, 7.4, 8.0, 8.5, and 9.0). Various latency-disrupting agents, a pore-forming peptide (alamethicin) and five detergents (Brij-58, CHAPS, Emulgen 911, Lubrol PX, and Triton X-100) were examined initially at a concentration of 100 µg/mg of protein. Subsequently, alamethicin (0, 2, 5, 10, 20, 100, and 400 µg/mg of protein) and Emulgen 911 (20, 100, and 400 µg/mg of protein) were examined over a wide concentration range. Also, the effectiveness of sonication was investigated that involved one, two, or three 4- to 5-s bursts with at least 1 min on ice between bursts. For all substrates, the time of incubation and the protein concentration were varied from 15 to 75 min (15, 30, 45, 60, and 75 min), and 125 to 1000 µg/ml (125, 250, 500, 750, and 1000 µg/ml), respectively. Each experiment was carried out in triplicate.

Calculations. Kinetic parameters were calculated according to the Michaelis-Menten equation for one or two enzymes by nonlinear least-squares regression analysis (GraphPad Prism, GraphPad Software, San Diego, CA). V_max/K_m ratios were determined as a rough calculation of metabolic clearance. Data are given as mean ± S.D. Linear correlation analyses between kinetic and physicochemical parameters were examined where Hammet  and  values were from a standard text (Kubinyi, 1995), and partition coefficient (log P, n-octanol/water) and pK_a were calculated with the ACD program (Advanced Chemistry Development Inc., Toronto, ON, Canada).

	Results

Top Abstract Introduction Materials and Methods Results Discussion References

The separation of the glucuronide metabolites was achieved using a gradient reversed-phase HPLC method. Since this method used acetonitrile and 10 mM perchloric acid (pH 2.5) as solvent systems and also 1% perchloric acid was used in sample preparation for analysis, a stability study was conducted with the glucuronide of 1a (5 mM) over the pH range encountered in analysis. After 48 h of incubation at 37°C in 1 N HCl, 1% perchloric acid, and Tris buffer, pH 7.4, the percentage of the glucuronide remaining was determined to be 98.3 ± 3.2, 98.8 ± 2.0, and 98.2 ± 4.9%, respectively. The identity of each N-glucuronide metabolite produced by expressed enzymes was indicated by comparison of the HPLC retention times under various chromatographic conditions with the metabolite produced by incubation of the same substrate with hepatic microsomes. In the latter case, each metabolite was identified by mass spectrometry, reaction dependence on the addition of UDPGA, and susceptibility to -glucuronidase treatment (unpublished work).

1-Substituted imidazoles 1b to 1h were screened against a panel of nine recombinant expressed human UGT enzymes. Where N-glucuronidation occurred, the data determined by the analysis of the microsomal mixture by HPLC-UV are given in Table 1. For six of the seven substrates examined, only UGT1A4 (human lymphoblast-expressed) was detected to give the N⁺-glucuronide metabolite. This is in contrast to the 1-phenyl compound 1a (Table 1), where in previous work (Vashishtha et al., 2000) and confirmed in the present work, both UGT1A3 and UGT1A4 were identified to catalyze N-glucuronidation. In the case of the 1-(4-nitrophenyl) compound 1h, none of the UGTs examined were detected to produce a metabolite. The observed UGT1A4-catalyzed activities for the series of 1-substituted substrates were in the range 3 to 82 pmol/min/mg of protein.

N-Glucuronidation kinetics of the series 1 compounds were determined for UGT1A4 expressed in baculovirus-infected insect cells. The N-glucuronidation activities (1a, 1b, 1d-1g) observed with this preparation were greater than for the preparation involving UGT1A4 expressed in human lymphoblastoid cells, although the difference in activity was not accurately determined. The incubation conditions for the kinetic measurements were optimized. The optimum pH for the N-glucuronidation was observed either at pH 7.4 or to plateau in the pH 7.4 to 9 range, as shown for the 1-phenyl compound 1a (Fig. 2). In further work all substrates were investigated at pH 7.4. Regarding latency disrupting agents, preliminary investigation of alamethicin and five detergents at 100 µg/mg of protein concentration gave a change in catalytic activity of the N-glucuronidation of 1a that varied from a slight increase to a marked decrease (alamethicin, 14.2%; Brij-58, 21.8%; CHAPS, 92.5%; Emulgen 911, +11.2%; Lubrol PX, 3.5%; and Triton X-100, 9.5%). Subsequent study of alamethicin and Emulgen 911 over wide concentration ranges, as well as various sonication techniques, showed that in each case the affected increase in catalytic activity did not exceed 25% (Fig. 3). Alamethicin at a concentration of 10 µg/mg of protein gave observed optimal activity for both substrates examined (1a and 1d), and hence this latency-disrupting agent condition was used for kinetic studies. Depending upon the substrate, the rate of N-glucuronidation was linear up to 45 (1f), 50 (1a), and 60 min (1b, 1d, 1e, and 1g). Regarding protein concentration, the rate of glucuronidation was linear up to 1.0 mg of protein for all the substrates examined.

View larger version (11K): [in this window] [in a new window]   Fig. 2.   Effect of pH on the rate of N-glucuronidation of 1-phenylimidazole by human UGT1A4 expressed in baculovirus-infected insect cells.

View larger version (25K): [in this window] [in a new window]   Fig. 3.   Effect of latency-disrupting agents and sonication on the rate of N-glucuronidation of 1-phenylimidazole by human UGT1A4 expressed in baculovirus-infected insect cells.

The kinetic parameters obtained for the series 1 compounds by nonlinear regression analysis of the data plotted between the rate of formation under optimized conditions of the metabolite and substrate concentration are shown in Table 2. For 1a, 1b, and 1d to 1g there were 17.5-, 2.2-, and 8.7-fold variation in the apparent K_m (range, 0.18-3.15 mM) and V_max (range, 0.16-0.35 nmol/min/mg of protein) values, and V_max/K_m ratios (range, 0.11-0.95 µl/min/mg of protein), respectively. Linear correlation analyses between these three apparent kinetic parameters and either pK_a, electronic properties (^para), or lipophilicity (^para or log P) revealed the following correlations:


where n is the number of substrates with available data, r² is the correlation coefficient, and p is the statistical probability.

	Discussion

Top Abstract Introduction Materials and Methods Results Discussion References

Six of the seven 1-substituted imidazoles examined were glucuronidated. The site of glucuronidation was probably at the imidazole 3N position, as proven in the case of the previously studied 1-phenyl-substituted compound (Vashishtha et al., 2000; unpublished work). This is consistent with the general pattern observed for published reports involving the site of N-glucuronidation of five-membered heterocyclic ring systems with more than one ring nitrogen atom. In general, where a ring nitrogen atom is substituted, as in the present study, a N⁺-glucuronide metabolite is formed at a ring aromatic-type tertiary amine (Midgley et al., 1981; Takeuchi et al., 1989; MacRae et al., 1990; Rush et al., 1990, 1992; McCann et al., 1997), whereas where ring substitution is only via carbon atoms, a tertiary amine glucuronide is formed at a ring secondary amine group (Stearns et al., 1991; Huskey et al., 1993, 1994a,b; Colletti and Krieter, 1994; Perrier et al., 1994; Kuo et al., 1999; Kaivosaari et al., 2000; Stevens et al., 2001).

The UGT isoforms involved in the glucuronidation of the series 1 compounds were identified. Of the nine UGT isoforms investigated, only UGT1A4 was found to affect catalysis of the six newly investigated 1-substituted imidazoles 1b to 1g where metabolism was observed. It is surprising that UGT1A3 was not detected to catalyze the glucuronidation of 1b to 1g, since this UGT isoform, as well as UGT1A4, glucuronidated the 1-phenyl-substituted compound 1a (Vashishtha et al., 2000). To the best of our knowledge, there are no other published investigations of UGT1A3 catalysis at an aromatic tertiary amine, although various substrates with an aliphatic tertiary amine have been demonstrated to be glucuronidated by this UGT isoform (Green et al., 1998; Breyer-Pfaff et al., 2000). However, in virtually all these documented cases, UGT1A3 catalyzed glucuronidation at lower rates than UGT1A4. In fact, a few substrates, such as trifluoperazine, were observed to be glucuronidated by UGT1A4 but not UGT1A3. Thus, there is great potential to develop UGT1A3- and UGT1A4-specific substrates by investigation of compounds where glucuronidation occurs at either an aliphatic or aromatic tertiary amine group.

The kinetic parameters for the glucuronidation of all series 1 substrates by UGT1A4 were determined at the same incubation pH of 7.4, since the glucuronidation activity either peaked at this pH value or plateaued in the pH 7.4 to 9.0 range. A similar observation has been made regarding the incubation pH for the glucuronidation of the same compounds by human liver microsomes (Vashishtha et al., 2000). The optimization of the incubation conditions for the glucuronidation of series 1 compounds by UGT1A4 with respect to the use of membrane-disrupting agents or sonication was investigated. In general, minimal effect was observed in removing the so-called latency for glucuronidation. It was previously noted (Green and Tephly, 1996) in studies with a different UGT1A4 preparation that certain detergents inhibited glucuronidation at a tertiary amine group. Notably, the use of detergents or alamethicin enhanced glucuronidation of series 1 compounds 2- to 3-fold in human liver microsomes (unpublished work), whereas in the present work, use of the same agents or sonication enhanced glucuronidation by UGT1A4 1.25-fold at most. This far lower activating effect of certain detergents or pore-forming peptides with expressed enzymes rather than microsomes may be a general phenomenon. Alamethicin was used in the kinetic studies at the optimal concentration of 10 µg/mg of protein. It was noted that as the concentration of alamethicin increased, the removal of latency effect was eventually replaced by an inhibitory effect (Fig. 3). Therefore, like detergents, the appropriate concentration of this pore-forming peptide should be determined for each substrate in the incubation mixture employed.

The enzyme activities and kinetic parameters for glucuronidation at an aliphatic tertiary amine group have been reported for various substrates with both human liver microsomes (Le Bigot et al., 1983; Dahl-Puustinen and Bertilsson, 1987; Coughtrie and Sharp, 1991; Styczynski et al., 1992; Breyer-Pfaff et al., 1997, 2000; Mey et al., 1999) and expressed UGT isoforms (Green et al., 1995, 1998; Green and Tephly, 1996, 1998; Breyer-Pfaff et al., 2000). However, such data are lacking regarding glucuronidation at an aromatic tertiary amine and are limited to the antiepileptic drug lamotrigine (Magdalou et al., 1992; Green et al., 1995; Furlan et al., 1999). Comparison of the present data for expressed UGT1A4 can be made with reported kinetic data for the same UGT isoform involving glucuronidation at an aliphatic tertiary amine group and available kinetic data for lamotrigine with human liver microsomes (Furlan et al., 1999; K_m for healthy livers, 5.5 ± 5.2 mM). The apparent K_m is considered an appropriate kinetic parameter to make a comparison between substrates of an enzyme, whereas the V_max shows wide variability between different expressed enzyme preparations of the same UGT, demonstrated in this work by the difference in catalytic activities obtained for the same substrate by the two UGT1A4 preparations investigated. The K_m data obtained for the glucuronidation of series 1 compounds by expressed UGT1A4 (range, 0.2-3.2 mM) are either in the same range or within one order of magnitude higher than the data reported (Green and Tephly, 1996; Green et al., 1998; Breyer-Pfaff et al., 2000) for the glucuronidation of seven substrates at an aliphatic tertiary amine (range, 0.03-0.3 mM).

To undertake kinetics/physicochemical parameters correlation analysis, an adequate number of compounds is necessary, especially since more than one physicochemical parameter is probably important with respect to enzyme-substrate interaction. Eight compounds were included in the present study, but glucuronidation could not be detected for one of the substrates, and a key substrate for investigation of correlations involving lipophilicity, the phenylethyl compound 1c, could not be accurately quantified. Regarding the latter, since there were data available for only two compounds of a homologous series (1a and 1b), the study of lipophilicity correlations was hampered. Nevertheless, despite the relatively low number of substrates with available kinetic data, there were statistically significant correlations between the following values: V_max and log P or pK_a and log 1/K_m and pK_a. The inverse correlation found between log P and V_max values can be contrasted with the parabolic relationship found for similar correlation analysis in a previous study involving only reaction rates (Kim, 1991). These previous analyses of various data sets for the O-glucuronidation of phenols by rat liver microsomes indicated that the reaction velocity was optimum at an approximate log P value of 2. In the present study, investigation of an optimum log P value is precluded due to the low number of substrates with a limited range of log P values. The correlations involving pK_a values indicate that for the series 1 compounds, a high pK_a is indicative of both a high K_m and a high V_max. Notably, a relationship of an increase in reaction velocity with an increase in pK_a was reported for one of the series of compounds in structure-metabolism relationship studies of the O-glucuronidation of phenols by rat liver microsomes (Kim, 1991). However, that glucuronidation, an S_N2 reaction, only occurs within a range of pK_a values could be anticipated. In fact, no glucuronidation could be detected for the 4-nitro compound (1h), the series 1 compound with a para-phenyl substituent with the largest electron withdrawing effect, and hence the lowest pK_a value. A further indication that the availability of the substrate lone pair of electrons at the site of glucuronidation has relevance to the kinetics of the metabolic reaction was the relatively wide variability observed in the optimum pH for glucuronidation at an aliphatic or aromatic tertiary amine group. For example, the optimum incubation pH value of 7.4 presently found for glucuronidation at an aromatic tertiary amine of some substrates can be contrasted with the optimum value of pH 8.4 found for glucuronidation of the more basic aliphatic tertiary amine group using the present or another UGT1A4 preparation (Green et al., 1995; unpublished work). Finally, in view of the low number of substrates examined and the limited range of values for each parameter, especially V_max values, further study with more substrates of greater structural diversity is warranted to substantiate the present correlation analyses.

In conclusion, a series of eight 1-substituted imidazole compounds served as a prototype model for glucuronidation at an aromatic tertiary amine group of a polyaza five-membered heterocyclic ring system. Where N-glucuronidation occurred, UGT1A4 catalyzed the metabolism. Small changes in substrate structure resulted in large changes in UGT1A4 catalysis that included loss of activity and marked change in kinetic parameters, especially K_m. Despite the low number of data points for correlation analyses between substrate physicochemical parameters and enzyme kinetic constants, correlations were found between V_max and log P or pK_a and between K_m and pK_a. These correlations indicate that the lipophilicity and the ease of availability of the tertiary amine lone pair of electrons of the substrate are important with respect to enzyme catalysis.