Materials and Methods

Chemicals.

Uridine 5′-diphosphate-glucuronic acid, sodium salt, and UDP-[U-¹⁴C]glucuronic acid (350 mCi/mmol) were purchased from Boehringer (Mannheim, Germany) and Isotopchim (Ganagobie-Peyruis, France), respectively. Adrenaline, apomorphine, 2,3-dihydroxybenzoic acid, 2-hydroxyestrone, 4-hydroxyestradiol, dobutamine, catechin, epicatechin, epicatechin gallate, epigallocatechin, epigallocatechin gallate, 4-methylcatechol, 4-tert-butyl-5-metoxycatechol, 4-nitrocatechol, 3,4-dihydroxymandelic acid, 5-hydroxydopamine, 6-hydroxydopa, carbidopa, 2,3-dihydroxybenzaldehyde, noradrenaline, and gallic acid were obtained from Sigma (Saint Quentin Fallavier, France). 4-tert-Butylcatechol, 3-fluorocatechol, tetrachlorocatechol, and methylgallate were purchased from Aldrich (Saint Quentin Fallavier, France).

Catechol, protocatechuic acid (3,4-dihydroxybenzoic acid), pyrogallol,l-dopa, dopamine, 6-hydroxydopamine, 3,4-dihydroxyphenyl acetic acid, and octylgallate were purchased from Fluka (Saint Quentin Fallavier, France). Tyrphostine A23, ethyl-3,4-dihydroxybenzylidenecyanoacetate, dihydrexidine [(±)-trans-10,11-dihydroxy-5,6,6α,7,8,12β-hexahydroxybenz[a]phenanthridine], 2-hydroxyestradiol, and 2-(1-thienyl)-ethyl-3,4-dihydroxybenzylidenecyanoacetate were obtained from ICN (Orsay, France). The 3-nitrocatechol inhibitors of COMT, entacapone, tolcapone, and nitecapone were kindly provided by Orion Pharma (Espoo, Finland). 3,5-Dinitrocatechol was obtained from Tocris Cookson (Bristol, UK). The chemical structure of the catechols is shown in Table 1.

Enzyme Sources.

Male Wistar rats (Domaine des Oncins, St. Germain l'Arbresle, France) were housed in an environmentally controlled room (a 12-h light cycle, 22–24°C) and fed rodent chow (UAR Alimentation, Villemoisson, France). For induction studies, 3-MC, dissolved in corn oil, was injected once intraperitoneally (100 mg/kg body weight; Sigma). Phenobarbital, dissolved in 0.9% (w/v) NaCl (Fluka) was administered once intraperitoneally (100 mg/kg body weight) and then was given with drinking water (1 g/l) for 4 days. Animals were killed on the fifth day. Clofibrate (0.3%, w/w; Sigma) was added to the rodent chow and given to rats for 7 days. As previously reported by Haumont et al. (1990), the doses of 3-MC, phenobarbital, and clofibrate were chosen for selectively inducing the UGT isoforms that glucuronidate planar phenols, bulky phenols, and bilirubin, respectively. In these conditions, glucuronidation of 3′-azido-3′-deoxythymidine was enhanced 4 times by phenobarbital, glucuronidation of 1-naphthol was enhanced 6-fold by 3-MC, and glucuronidation of bilirubin was enhanced 2-fold by clofibrate. Aroclor 1254 (Foxboro, North Haven, CT), 2,3-dihydroxynaphthalene, tolcapone, and entacapone were dissolved in olive oil and administered at the doses of 550, 300, 200, and 200 mg/kg body weight, respectively. Aroclor 1254 was injected once intraperitoneally. 4-Methylcatechol and 4-nitrocatechol were dissolved in a sucrose solution and administered at two doses, 100 and 150 mg/kg body weight and 100 and 200 mg/kg body weight, respectively. All catechols were given for 3 days by oral gavage. Liver microsomes from three individual rats treated by the inducers and from control rats were prepared by differential centrifugations, according to Hogeboom (1955), and stored at −80°C in 5 mM Hepes buffer, pH 7.4, containing 0.25 M sucrose.

The metabolic competent cell lines expressing the rat liver UGT1A6 and UGT2B1 were established as previously described (Fournel-Gigleux et al., 1991; Pritchard et al., 1994). The genetically engineered V79 cells expressing UGT2B1 and 1A6 were grown on Dulbecco's modified Eagle's medium (Invitrogen, Carlsbad, CA) with 5% and 8% (v/v) Nu-serum (Tebu, Le Perray en Yvelines, France), respectively, and containing 0.1 mg/ml streptomycin and 100 U/ml penicillin. One day after replating, the cells were treated with 2 mM butyrate. Sodium butyrate has been shown to increase the gene expression of recombinant proteins in heterologous systems (Palermo et al., 1991). According to our experiments, addition of sodium 2 mM butyrate in V79 cell cultures enhanced the expression of recombinant UGTs 2.5 times, as previously discussed (Fournel-Gigleux et al., 1991). Then, they were washed with phosphate-buffered saline, scraped from the plates, and broken in 5 mM Hepes buffer, pH 7.4, and 0.25 M sucrose by a 3 × 5-s sonication. Microsomes were prepared by differential centrifugations. The homogenate was centrifuged for 10 min at 10,000g. The corresponding supernatant was centrifuged for 60 min at 100,000g on a RCM 120 GX micro-ultracentrifuge (Sorvall, Les Ulis, France). The activity of the recombinant enzymes was tested with marker substrates, 1-naphthol and 4-methylumbelliferone for UGT1A6 (Bock and White, 1974; Lilienblum et al., 1982) and morphine and ketoprofen (Sigma) for UGT2B1 (Terrier et al., 1999), before their use toward catechols. The protein concentration of liver microsomes and cellular membranes was evaluated by the method of Bradford (1976), with the Bio-Rad protein assay kit (Hercules, CA).

Screening of Glucuronidation Activity.

The thin-layer chromatography method of Bansal and Gessner (1980) was used to measure the in vitro glucuronidation of catechols. The catechols (0.5 mM) dissolved in dimethyl sulfoxide (2 μl) were added to an incubation mixture containing 40 μg of microsomal proteins, UDP-glucuronic acid (2.5 mM, 0.1 μCi), and 100 mM Tris-HCl buffer, pH 7.4, and 5 mM MgCl₂ (final volume, 40 μl). A control was simultaneously run without catechol. When rat liver microsomes were used, they were previously treated by Triton X-100 (Sigma) at the final detergent/protein concentration of 0.4 (w/w) for 30 min on ice to obtain maximal enzyme activation. Incubations were performed for 30 min at 37°C, and the reactions were stopped by addition of 40 μl of ice-cold ethanol. The precipitated proteins were discarded by centrifugation with a tabletop centrifuge (5000g, 10 min), and the supernatant was placed onto silica gel Si250 F-PA thin-layer chromatography plates (Whatman, Clifton, NJ). The products were separated with a mobile phase composed ofn-butanol/acetone/acetic acid/30% aqueous ammoniac/water (70:50:18:1.5:60, v/v). After migration, the plates were dried and exposed to X-Omat or BioMax films (Eastman Kodak, Rochester, NY; Sigma) for 12 to 48 h. The films were developed, and the areas from the spots corresponding to the radioactive glucuronides were scraped off with the aid of a drop of water and put into 5 ml of Ultima Gold scintillant cocktail (Packard Instrument Co., Meriden, CT). A similar amount of silicagel from the control experiment was also scraped. The amount of radioactivity associated was measured on a LKB spectrometer (1214 Rack Beta; PerkinElmer Wallac, Turku, Finland). Activity values were expressed as nanomoles of glucuronides formed per minute and milligram of protein after substraction of the value found in controls.

Glucuronide formation was verified by the susceptibility of the products to hydrolysis by β-glucuronidase (bovine liver; Sigma), as previously reported (Pless et al., 1999). Data are the means ± S.D. (n = 3 rats; three determinations per rat and catechol). Results were considered significant when p< 0.05 (Student's t test).

Kinetics of Rat Recombinant UGT.

Apparent kinetic constants K_m andV_max toward catechols were determined at the optimal pH, protein concentration, and reaction time yielding linear product formation. Briefly, all assays mixtures (100 μl of total volume) contained 100 mM Tris-maleate buffer, pH 7.4, 5 mM MgCl₂, 2 mM UDP-glucuronic acid with 0.1 μCi radiolabeled UDP-glucuronic acid, 1 mM d-saccharic acid 1,4-lactone, 100 μg of proteins, and increasing concentrations of catechol (50 to 2500 μM) dissolved in a same volume (5 μl) of dimethyl sulfoxide. After a 40-min incubation, the reaction was terminated by the addition of 100 μl of prechilled methanol. The precipitated proteins were removed by centrifugation at 5000g for 10 min. The resulting supernatant was directly injected onto the high-performance liquid chromatography (HPLC) column according to the method of Ethell et al. (1998). Briefly, the HPLC system consisted of a binary acetonitrile gradient, 0 to 100% in 0.05 mM ammonium acetate, developed over 15 min on a 25-cm × 4.6-mm i.d. Spherisorb 5ODS2 column (HPLC Technology, Macclesfield, UK) at a flow rate of 1 ml/min. The radioactive UDP-glucuronic acid and glucuronide were detected with a model 9701 radioactivity monitor (Reeve Analytical, Glasgow, UK). Values were expressed as nanomoles of glucuronide formed per minute and milligram of protein. The apparent kinetic constants were calculated from nonlinear least-squares regression analysis of the double-reciprocal plots of initial activity versus substrate concentration.

PLS Modeling.

PLS analysis was made using the soft independent modeling by class analogy Simca-P 8.1 program (Umetrics AB, Umeå, Sweden). Two glucuronidation activities measured with liver microsomes of nontreated and 3-MC-treated rats were used as y-variables after transformation to logarithmic scale. The detection limit 0.01 was used instead of a zero activity value. Several physicochemical parameters and substructure descriptors were used as x-variables. The octanol-water partition coefficient, log P, was calculated using KOWWIN v.1.62 program (Syracuse Research Institute, North Syracuse, NY). The experimental log P value was used instead of the calculated one, when reported by the program. The molar volume (V_m) and pK_a values for the most acidic catecholic hydroxyl group were calculated using the ACD/log D 4.0 program (Advanced Chemistry Development, Inc., Richmond, Toronto, Canada). For the COMT inhibitors (entacapone, tolcapone, and nitecapone), experimental pK_a values reported by Wikberg et al. (1993) were used.

Substructure descriptors were defined based on the atom type, the distance of the substituent atom from the catechol ring, and the position of the substituent in the catechol ring. All nonhydrogen atoms of the substituents in the catechol ring were classified in one of five atom types: S (saturated C, ether O), U (double or triple bond or aromatic C, O, N), N⁺ (protonated amino nitrogen), O⁻ (carboxylate group), HD (hydrogen bond donor OH or NH other than N⁺). Depending on the distance from the catechol ring, substituent atoms were classified as proximal (p) or distal (d). In flexible side chains, the atoms one or two bonds away from the catechol ring were considered proximal. In addition, the substituent in the ortho position to one of the catechol hydroxyls was specified in a separate class (o). In case of polycyclic structures, all the atoms of the ring fused to the catechol ring were considered proximal. Altogether the following 20 substructure descriptors were considered: Spo, Upo, N⁺po, O⁻po, HDpo, Sdo, Udo, N⁺do, O⁻do, HDdo, Sp, Up, N⁺p, O⁻p, HDp, Sd, Ud, N⁺d, O⁻d, HDd. The descriptor O⁻po showed nonzero value only for one compound, and therefore it was combined with O⁻p. The descriptors N⁺po, N⁺do, O⁻do and HDdo showed zero value for all compounds and were therefore removed. Consequently, 14 substructure descriptors were used in the PLS analysis.

Results

Glucuronidation of Catechols in Rat Liver Microsomes.

The glucuronidation activity measured upon incubation of rat liver microsomes with the series of 42 catechols has been determined. In each case, the catechol glucuronides formed were identified on the thin-layer chromatography plate from 1) the associated radioactivity when catechols were incubated with microsomes and UDP-[U-¹⁴C]glucuronic acid, 2) their absence in control reactions in which UDP-glucuronic acid was omitted, and 3) their susceptibility to hydrolysis by β-glucuronidase. No attempt was made to determine the position of the glucuronide on the 1,2-dihydroxybenzene ring.

The glucuronidation rate of catechols by rat liver microsomes is reported in Table 2. The small catechols (compounds 1–4, 6, 7, 10, 11, 14, and 16) in which the structure derives from the reference compound 1,2-dihydroxybenzene (pyrocatechol) were actively glucuronidated by liver microsomes. Moreover, the activity depended on the nature and on the position of the substituent on the benzene ring. 4-Nitrocatechol, 4-methylcatechol, or 2,3-dihydroxybenzaldehyde was glucuronidated at the highest rate, whereas 3,5-dinitrocatechol or 4-tert-butylcatechol was the least glucuronidated. The carboxyl catechols were very poor substrates (compounds 35–38 and 40; Table 2).

On the other hand, the polyhydroxy catechols (the flavonoid catechins and related substances) that are found in green tea and are well known for their antioxidant properties were glucuronidated at a low rate, despite the fact that the molecules exhibit several potential glucuronidation sites (compounds 21, 24, and 28–30). The aliphatic amines that are positively ionized at physiological pH (dopamine, noradrenaline, adrenaline, 6- and 5-hydroxydopamine, carbidopa, dihydrexidine, and apomorphine) and the zwitterionic catecholsl-dopa and 6-hydroxydopa were not very efficiently glucuronidated. From this series, only the bulky catechol dobutamine was glucuronidated to an appreciable extent (Table 2). The glucuronidation rate observed for the acidic drugs tolcapone, nitecapone, and entacapone (compounds 18, 20, 27) was low, ranging from 0.20 to 0.87 nmol/min/mg of protein. By contrast, the bulky neutral hydrophobic catechol estrogens, except 4-hydroxyestradiol (compounds 9, 15, 26), and the cyanoacetate derivatives (compounds 5, 8, 13) were markedly glucuronidated (2.5 to 7.9 nmol/min/mg protein) (Table 2).

Effect of Treatment with Inducers on the Glucuronidation of Catechols.

Rats were administered with the inducing agents 3-MC, phenobarbital, or clofibrate, which are known to stimulate the expression of distinct classes of UGTs. The results reported in Figs.1 and 2 and Table 2 clearly show that 3-MC only could markedly enhance the glucuronidation of catechols of different chemical structure (small catechols, catechins, cyanoacetate derivatives, and catechol estrogens). The increase in activity was particularly high (10×) for 3,5-dinitrocatechol, epicatechin, and epicatechin gallate. On the other hand, administration of the peroxisome proliferator clofibrate exhibited a moderate inducing effect (2×) on a limited number of catechols (octylgallate, catechins, and apomorphine). By contrast, phenobarbital failed to stimulate the glucuronidation of catechols, whatever the type of catechol (Fig. 1 and 2). Aroclor 1254 treatment caused an increase in the glucuronidation rate of catechols that was similar to that measured with 3-MC (results not shown). Moreover, no significant induction was observed when rats were treated by catechols, whatever the dose used (4-methylcatechol, 4-nitrocatechol, 2,3-dihydroxynaphthalene, entacapone, and tolcapone) (results not shown). This experiment indicates that the glucuronidation of catechols is catalyzed by UGT isoforms that are mainly sensitive to the inducing effect of polycyclic arylhydrocarbons (3-MC and Aroclor 1254). No autoinduction by catechols was observed.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

Effect of inducers on the glucuronidation of small catechols by rat liver microsomes.

Rats were treated with 3-MC (open bar), phenobarbital (dotted bar), or clofibrate (gray bar), as indicated under Materials and Methods. Results are the means ± S.D. (n = 3 animals). The asterisk denotes results significantly different (p < 0.05) from controls. 1, pyrocatechol; 2, 4-nitrocatechol; 3, 4-methylcatechol; 4, pyrogallol; 5, 3,5-dinitrocatechol; 6, tetrachlorocatechol; 7, 2,3-dihydroxybenzaldehyde; 8, 4-tert-butylcatechol; 9, 4-tert-butyl-5-metoxycatechol; 10, 3-fluorocatechol; 11, octylgallate; 12, methylgallate.

• Download figure
• Open in new tab
• Download powerpoint

Figure 2

Effect of inducers on the glucuronidation of catechols by rat liver microsomes.

Rats were treated with 3-MC (open bar), phenobarbital (dotted bar), or clofibrate (gray bar), as indicated under Materials and Methods. Results are the means ± S.D. (n = 3 animals). The asterisk denotes results significantly different (p < 0.05) from controls. Neurotransmitters: 1, l-dopa; 2, dopamine; 3, 5-hydroxydopamine; 4, 6-hydroxydopa; 5, 6-hydroxydopamine; 6, noradrenaline; 7, adrenaline. Dopaminergic drugs: 1, apomorphine; 2, carbidopa; 3, dihydrexidine; 4, dobutamine. Cyanoacetate derivatives: 1, tyrphostine A23; 2, ethyl-3,4-dihydroxybenzylidenecyanoacetate; 3, 2(1-thienyl)-ethyl-3,4-dihydroxybenzylidenecyanoacetate. Catechol estrogens: 1, 2-hydroxyestrone; 2, 4-hydroxyestradiol; 3, 2-hydroxyestradiol. Catechins: 1, catechin; 2, epicatechin, 3, epicatechin gallate; 4, epigallocatechin; 5, epigallocatechin gallate. Carboxylic acids: 1, protocatechuic acid; 2, 3,4-dihydroxyphenylacetic acid; 3, 2,3-dihydroxybenzoic acid; 4, 3,4-dihydroxymandelic acid; 5, gallic acid. COMT inhibitors: 1, nitecapone; 2, entacapone; 3, tolcapone.

Partial Least-Squares Model.

PLS analysis was carried out to model the variation of glucuronidation activity based on the structural variation of the series of catechols considered. Glucuronidation activities determined with liver microsomes of noninduced and 3-MC-induced rats were used as y-variables (Table 2). The x-variable block consisted of three physicochemical descriptors (log P, pK_a,V_m) (Table 2) and 14 substructure descriptors. PLS modeling of the whole set of data from the 42 catechols resulted in a three-component cross-validated model explaining 69% of the response variation. For two compounds, 6-hydroxydopa and 3,4-dihydroxyacetic acid, the activity predicted by the PLS model deviated more than 1 log unit from the measured value. After excluding these two compounds, a three-component model was obtained explaining 82% (R² = 0.82; goodness of fit) and predicting 61% (Q² = 0.61; goodness of prediction) of the response variation. The experimental versus the predicted activities are shown in Fig. 3. Linear relationships were obtained using the regression coefficients indicated in Table 3. IndividualR² andQ² values showed that the glucuronidation activities with nontreated rats were modeled and predicted better (R² = 0.88;Q² = 0.70) than those obtained upon 3-MC induction (R² = 0.75;Q² = 0.52; Fig. 3). The glucuronidation activities in control and 3-MC-treated rats could be calculated by inserting the coefficients of Table 2 in the following regression equations:log RLM=0.472 (log P)/V+0.773 pKa−0.427 Spo−0.058 Upo−0.054 Sp+0.011 Sd+0.114 Up−0.030 Ud−0.332 Hdo−0.398 HDp− Equation 10.069 HDd−0.674 N+−1.967 O−p−0.243 O−d−0.045 pKa2−3.294 log 3­MC=0.480 (log P)/V+0.708 pKa−0.255 Spo−0.063 Upo−0.020 Sp+0.035 Sd+0.104 Up−0.014 Ud−0.266 Hdo−0.422 HDp− Equation 20.024 HDd−0.696 N+−1.801 O−p−0.358 O−d−0.041 pKa2−2.852 The PLS weight plots (w*c) shown in Fig.4 indicate the relationships between the responses (control rats, Fig. 4A; 3-MC-treated rats, Fig. 4B) and the structural variables. The two sets of activity values obtained with nontreated and treated rats were strongly correlated (close to each other in the weight plots), thus justifying their processing as a single model. Structural variables contributing to the final PLS model were a combination of the descriptors' lipophilicity, molar volume (log P)/V_m, pK_a of the catechol group and its square term, and of 12 substructure descriptors. O⁻p (indicator for the carboxyl group on the catechol ring) and log P/V_m had a strong influence (scattered at the greatest distance from the plot origin) in all three dimensions of the model and were positively and negatively correlated with the glucuronidation activities, respectively. The pK_a value of catechol influenced mostly the second component of the model. The two responses (log RLM, log 3-MC) were mainly separated in the third dimension of the model, which was influenced by hydrogen bond donors in ortho or distal positions (Fig. 4B).

• Download figure
• Open in new tab
• Download powerpoint

Figure 3

PLS analysis of glucuronidation activity of catechols by liver microsomes of nontreated (A) and 3-MC-treated (B) rats.

A linear relationship between experimental values versus calculated activity was obtained with R² = 0.88 and Q² = 0.70 for nontreated rats and with R² = 0.75 andQ² = 0.52 for rats treated with 3-MC. The numbers refer to the catechols listed in Table 2.

• Download figure
• Open in new tab
• Download powerpoint

Figure 4

PLS weight plots (w*c) showing the influence of different structural variables on the glucuronidation rate by rat liver microsomes.

A, plot of the first component (w*c[1]) against the second component (w*c[2]); B, plot of the first component against the third component (w*c[3]).

Glucuronidation of Catechols by the Recombinant UGT1A6 and UGT2B1.

We previously reported that UGT1A6 exhibited a strict substrate specificity toward short phenols (Harding et al., 1988), whereas UGT2B1 was very active in glucuronidating both phenol and carboxyl substances, thus representing a vast panel of structures. These isoforms were tested for their ability to glucuronidate catechols. Our results indicated that only small catechols were glucuronidated by the two isoforms to an appreciable extent (Table 2). The other catechols tested, particularly the carboxyl catechols and the neurotransmitters, were not substrates of these isoforms, except dobutamine for UGT2B1. The apparent kinetic constants K_m andV_max for the glucuronidation of small catechols by the UGT1A6 and UGT2B1 were calculated to further predict which isoform is likely to be involved in the glucuronidation in vivo (Table 4). For the selected catechols, the catalytic efficiency of UGT1A6 was higher than that of UGT2B1. Catechol itself was glucuronidated by UGT1A6. Tetrachlorocatechol was a good substrate for both isoforms. 4-Methylcatechol was preferentially glucuronidated by UGT1A6, whereas octylgallate and dobutamine were substrates of UGT2B1 only (Table 4). For dobutamine, two glucuronides could be separated by the HPLC method. The apparent kinetic constants show that one glucuronide was more efficiently formed than the other (dobutamine 2; Table 4). Due to the limited number of catechols that were substrates of the recombinant enzymes used in this study and the low activity measured for some of them, no PLS model could be developed on the series.

Discussion

The metabolism of catechols has not yet been well characterized, although they are widely represented in the living organisms and environment, and that some of them present potential valuable therapeutic properties. In this study, we investigated the potency of rat liver UGTs to glucuronidate a large variety of catechols from endogenous and exogenous origins. Liver microsomes that contain all expressed UGT isoforms were used as a first approach to evaluate their relative rates of conjugation in vitro. Furthermore, an attempt to determine the isoforms involved in that process was performed with the aid of recombinant UGT1A6 and 2B1 stably expressed into V79 cells and induction studies of the animals with standard inducers.

The glucuronidation rates of catechols by rat liver microsomes and recombinant UGT1A6 and 2B1 varied considerably according to the different structural classes and also within each class. To predict the glucuronidation activity of rat liver microsomes toward catechols, a PLS model was developed taking into account 17 different parameters that characterize the physicochemical properties of the molecules. The PLS modeling could not be attempted for activities measured with the recombinant UGT1A6 and 2B1; this was due to a strict substrate specificity of these isoforms toward small catechols and to low activities measured for other many catechols.

The modeling of the glucuronidation activity with rat liver microsomes resulted in a three-component cross-validated model explaining 69% of the response variation. When two compounds (6-hydroxydopa and 3,4-dihydroxyacetic acid) were excluded from the series, the model was markedly improved, explaining 82% of the variations. Three physicochemical parameters seemed to influence glucuronidation of catechols. The data were best modeled using the ratio log P/V_m for the effect of lipophilicity and size and including the square term of pK_a for the effect of the ionization of the catechol group. An appropriate series of selected substrate sets would be required to confirm the nonlinearity of the effects. However, both effects may be also related to the structure and function of UGTs. Lipophilicity/size effect may reflect mainly the transport of substrates to the active site, whereas the pK_a effect is more likely related to the catalytic mechanism or electrostatic interactions close to the reaction center. UGTs are integral endoplasmic reticulum glycoproteins with the substrate binding site on the luminal side of the membrane. Hydrophobicity is responsible for the accumulation of the catechols within the hydrophobic phospholipid bilayer of the endoplasmic reticulum. N-Terminal domains of UGTs that strongly interact with membrane phospholipids have been proposed to favor access of the hydrophobic substrates, such as tetrachlorocatechol, to the active site (Ouzzine et al., 1999). By contrast, the hydrophilic carboxyl catechols and some catechins or neurotransmitters cannot easily cross the hydrophobic barrier to be glucuronidated.

The model also suggests a parabolic relationship between pK_a and glucuronidation activity, with a maximum between 8 and 9. Partially ionized catechols were among the substrates glucuronidated at the highest rates, whereas completely ionized catechol (entacapone, tolcapone, nitecapone, and 3,5-dinitrocatechol) were glucuronidated to a lower extent. This result can be related to the catalytic reaction mechanism accounting for the glucuronidation of phenols. Recent results obtained in our laboratory strongly suggest a general base (histidine or aspartic/glutamic acid residues) for activation of the nucleophilic acceptor species by deprotonation of the phenolic substrate according to a S_N2 mechanism (Ouzzine et al., 2000). This process promotes its transfer to glucuronic acid, leading to the release of UDP. A mechanistic explanation for the nonlinear pK_a effect may be that an optimum of the electron-withdrawing substituent effect may balance between the stabilization of the transition-state charge and the decrease of the nucleophilicity of the attacking oxygen.

On the catechol ring, the most acidic hydroxyl group is vicinal to the nitro group (NO₂ in ortho position). In this condition, we have found that glucuronidation preferentially occurred on the other less acidic hydroxyl group (Luukkanen et al., 1999). Another explanation may be that the nitro group or the ionized neighboring hydroxyl group hinders optimal binding. However, the effect seems to be isoform-dependent. High glucuronidation activity was found for 3,5-dinitrocatechol after 3-MC induction. This suggests that the inducible UGT isoforms could accept both the nitro group and the ionized catechol group. Human UGT1A9 has been reported to glucuronidate entacapone at a high activity, and human UGT1A1 has been found to glucuronidate the two hydroxyl groups of entacapone, although at a low activity (Lautala et al., 2000).

Some of the catechols have other functional groups, such as alcohol, phenol, and carboxyl groups, that can be potentially acceptors for glucuronic acid instead of the catechol moiety. Most of these compounds, however, presented a very low glucuronidation rate. Therefore, this possible multisite glucuronidation does not interfere too much with the interpretation of the results. Two exceptions were observed for 2-hydroxyestradiol and dobutamine, which were glucuronidated at a high rate. However, glucuronidation activities with 2-hydroxyestradiol and 2-hydroxyestrone were similar, thus suggesting that glucuronidation takes place mainly on the catechol group. In the case of dobutamine, liquid chromatography/mass spectrometry analysis indicated that the catechol moiety of the molecule was more glucuronidated than the phenol end (H. Keski-Hynnilä, M. Kurkela, E. Elovaara, L. Antonio, J. Magdalou, L. Luukkanen, J. Taskinen, and R. Koistiainen, manuscript submitted).

Interestingly, from the standard inducers tested, only the polycyclic aryl hydrocarbons 3-MC and Aroclor 1254 markedly enhanced the glucuronidation rate of catechols. On the other hand, phenobarbital had no significant effect. 3-MC and related polycyclic aryl hydrocarbons, such as dioxin, are known to specifically increase the expression of UGT1A6 in rat (Bock et al., 1998). Such induction could account for the stimulation effect of 3-MC found in rat liver microsomes. Indeed, UGT1A6 was found to catalyze the glucuronidation of catechols, especially small catechols. This isoform presents strict substrate specificity toward planar and short phenols (Harding et al., 1988). Larger molecules, such as neurotransmitters, dopaminergic drugs, catechins, catechol estrogens, and COMT inhibitors, were, as expected, either not substrates or poor substrates of the enzyme. Concerning the COMT inhibitors, Lautala et al. (1997, 2000) recently reported that entacapone and tolcapone were not glucuronidated by the human orthologous UGT1A6. On the other hand, the phenobarbital-inducible UGT2B1 presents a broader substrate specificity when compared with UGT1A6 since structurally unrelated alcohols, phenols, carboxylic acids of the series of 2-phenylpropionic acid, and fatty acids were glucuronidated by this isoform (Pritchard et al., 1994). Among the series of catechols considered in this study, UGT2B1 competes with UGT1A6 for the glucuronidation of small catechols. As for UGT1A6, the carboxyl catechols, neurotransmitters, catechins, catechol estrogens, or COMT inhibitors were not substrates of the UGT2B1 isoform. An exception was found for octylgallate, dobutamine, and ethyl-3,4-dihydroxybenzylidenecyanoacetate, which were specifically glucuronidated by UGT2B1. Phenobarbital was unable to stimulate the glucuronidation of many catechols. This was the case of some catechols, such as octylgallate and 4-tert-butyl-5-metoxycatechol, which are substrates of the phenobarbital-inducible UGT2B1. On the other hand, glucuronidation of 4-tert-butylcatechol, a good substrate of UGT2B1, was indeed enhanced 2-fold upon phenobarbital treatment (Fig. 1; compound 8). It is likely that these compounds are substrates of several other isoforms. Clofibrate stimulated the glucuronidation of a restricted number of catechols, among those was octylgallate. Clofibrate and other peroxisome proliferators are known to enhance specifically the glucuronidation of bilirubin, supported by the UGT1A1 isoform. Octylgallate has been reported to be a surrogate substrate of UGT1A1 in humans (Senafi et al., 1994). This result would explain why octylgallate glucuronidation was increased upon treatment of rats with clofibrate. By contrast, the catechols themselves, and particularly the COMT inhibitors, did not present any inducing effect on UGT isoforms implicated in the glucuronidation of catechols, indicating that no autoinduction occurred.

All catechols in which glucuronidation was strongly enhanced by 3-MC were not effective substrates of UGT1A6. This was particularly observed for bulky catechols, such as apomorphine, 4-hydroxyestradiol, octylgallate, or epicatechin gallate. The results suggest that other UGTs that are 3-MC-inducible could be involved in this process. A potential candidate is UGT1A7, which has been recently characterized. The mRNA expression of this isoform has been shown to be increased by 3-MC (Emi et al., 1995). The enzyme catalyzes the glucuronidation of bulky phenols chemically related to benzo [a]pyrene phenols (Grove et al., 1997). In rabbit, UGT1A7 was reported to also glucuronidate some catechols (octylgallate and catechol estrogens) (Bruck et al., 1997). Our data showing that glucuronidation of octylgallate or 4-hydroxyestradiol was increased upon 3-MC treatment would support the involvement of the rat UGT1A7. Work is in progress to investigate this point.

The PLS model was able to predict the glucuronidation values for many catechols in rats treated with 3-MC, although the model was better for activities from control animals. Glucuronidation of six compounds was increased more than 10-fold upon 3-MC treatment. This induction could be predicted for five of them. The PLS model implies that 3-MC may induce UGT isoforms that can tolerate larger structures and hydrogen bond donors, especially in the ortho and distal positions.

In conclusion, large differences in the glucuronidation rates were observed in the series of catechols considered in this study. Glucuronidation seems to be an efficient biotransformation pathway for selected catechols that are hydrophobic and noncharged molecules. Sulfotransferases and COMTs may relay the UGTs for the metabolism of other catechols.