Introduction

Top Abstract Introduction Materials and Methods Results and Discussion References

The quantitative description of structural elements controlling substrate specificities of enzymes catalyzing metabolism of xenobiotics is a major objective of predictive toxicology and drug metabolism. Because the metabolism of virtually all xenobiotics is catalyzed by more than one enzyme or at least by multiple isoforms whose expression can vary with species, tissue, and development, such a comprehensive analysis is extremely difficult. However, the broad specificities of some drug metabolizing enzymes suggests that selective binding to the catalytic site is not essential for catalysis, and isoform specificity depends on other more readily defined parameters. For instance, numerous structure activity studies with purified and microsomal flavin-containing monooxygenases (FMOs)² suggest that the overall size and shape of a nucleophilic xenobiotic is a major factor responsible for differences in the specificities of FMO isoforms [cf. review by Poulsen and Ziegler (1995) for references to original literature supporting this conclusion]. To what extent this will be true for all mammalian FMO isoforms is not known, but because rate constants for the oxidation of sulfur compounds by the enzyme-bound and protein-free 4a-hydroperoxyflavin are essentially the same (Jones, 1985), this assumption is probably valid for xenobiotics bearing nucleophilic sulfur or selenium functional groups. On the other hand, although enzyme protein has a large effect on the oxidation of amines, substrate size was still a major factor responsible for the differences in substrate specificities of pig FMO1 and rabbit FMO2 for a homologous series of tricyclic xenobiotics bearing the same N,N-dimethyl functional group (Nagata et al., 1990). Thiocarbamides with increasingly bulky substituents on the one and three positions are also apparently capable of serving as selective substrate probes for FMO isoforms in mammalian microsomes (Guo et al., 1992). The studies described in this report were undertaken to probe size restrictions for thiocarbamides and, by extension, other nucleophilic xenobiotic substrates for human FMO1.

	Materials and Methods

Top Abstract Introduction Materials and Methods Results and Discussion References

N-Benzylimidazole, acetylcholine chloride, and all thiocarbamides were purchased from Aldrich Chemical Co. (Milwaukee, WI). Methimazole, NADP⁺, glucose 6-phosphate, L-mesenteroides glucose 6-phosphate dehydrogenase were obtained from Sigma Chemical Co. (St. Louis, MO). All other reagents were of the highest grades available commercially. Thiocholine chloride was prepared by the methanolysis of acetylthiocholine chloride as described (Guo et al., 1992). Phenylthiourea, 1,3-diphenylthiourea, and larger thiocarbamides were recrystallized from ethanol and stored at 0-5°C. Fifty millimolar solutions of these thiocarbamides in ethanol were prepared daily. Methimazole and thiourea are more stable, and aqueous solutions (0.1 M) of these substrates can be stored at 5-10°C for 2 to 3 weeks without any detectable changes. A small sample (2.5 mg of protein) of human FMO1 Supersomes was purchased from Gentest (Woburn, MA)

Construction and Expression of Recombinant FMO1. FMO1 in baculovirus were carried out by minor modifications of the procedure described by Dolphin et al. (1998). The insert was excised by incubation with XbaI and HindIII, then gel-purified and ligated into pFastBac1 (Life Technologies, Rockville, MD) to give FMO1/pFastBac1. FMO1 cDNA was a generous gift from Dr. Karen S. Browning of our department. FMO1 recombinant baculovirus was generated by transfection of Spodoptera frugiperda (Sf)9 cells with the corresponding recombinant bacmid DNA, obtained via site-specific transposition using the Bac-to-Bac Baculovirus Expression System (Life Technologies, Rockville, MD). For expression, 500 ml of S. frugiperda (Sf)9 cells, grown to a density of 1 × 10⁶ cells/ml in a 2-liter spinner flask, were infected with virus at a multiplicity of infection of 7, and incubated by stirring at 100 rpm in a stirring platform at 27° for 72 h. Cells were pelleted and resuspended in 50 ml of lysis buffer, consisting of 50 mM Tris-HCl (pH 7.4), 0.154 M KCl, 0.2 mM phenylmethylsulfonyl fluoride. The cells were broken by two passes through a precooled French press cell disrupter at 800 psi. The lysed cells were centrifuged at 3000g for 15 min at 4°C. The supernatant was saved and microsomal fraction was obtained by centrifugation of the resulting supernatant at 100,000g for 1 h at 4°C. Microsomal pellets were resuspended in 10 ml of 10 mM HEPES (pH 7.5) buffer containing 0.154 M KCl, 1 mM EDTA, and 20% (v/v) glycerol. The preparation was stored in aliquots at 80° until use. Protein concentration was determined by the method of Bradford with Bio-Rad (Hercules, CA) reagents using BSA as a standard.

Activity Measurements. Microsomal thiocarbamide S-oxygenase activities were measured by following substrate-dependent thiocholine oxidation as described (Guo et al., 1992). The reaction medium contained 0.1 M phosphate (pH 7.4), 0.25 mM NADP⁺, 2.5 mM glucose 6-phosphate, 1.5 U glucose 6-phosphate dehydrogenase, 2 mM N-benzylimidazole, and 150 µM thiocholine. Aliquots withdrawn at 3-min intervals were deproteinized and analyzed for thiocholine exactly as described previously (Guo et al., 1992). Substrate activities of trimethylamine, N,N-dimethylaniline, and chlorpromazine were estimated by their inhibition of 10 mM methimazole S-oxygenation.

	Results and Discussion

Top Abstract Introduction Materials and Methods Results and Discussion References

Initial studies carried out with commercial human Supersomes and microsomes isolated in this laboratory from Sf9 cells-infected FMO1 baculovirus indicated that both preparations catalyzed methimazole-dependent oxidation of thiocholine. In the absence of methimazole, there was no detectable loss of thiocholine for up to 12-min incubation with the complete reaction medium. Microsomes isolated from control Sf9 cells were also completely devoid of methimazole S-oxidase activity. In addition, neither 1.0 mM trimethylamine nor phenothiazine at the highest concentration soluble in the assay medium had any detectable effect on the methimazole S-oxidase activity of microsomes bearing expressed human FMO1. In contrast, N,N-dimethylaniline and chlorpromazine at 1.0 mM inhibited the oxidation of 10 µM methimazole 50 and 70%, respectively. Although oxidation products were not isolated, the pronounced inhibition of methimazole oxidation suggests that these amines can serve as alternate substrates for human FMO1. This interpretation is consistent with earlier studies of Lemoine et al. (1990), who reported that human kidney, but not human liver microsomes catalyzed the N-oxidation of imipramine. The enzymatic basis for this difference was subsequently shown to be due to the expression of different FMO isoforms in these tissues in adults (Dolphin et al., 1996; Lang et al., 1998). Although nucleophiles not much bigger than methimazole are readily accepted by essentially all FMO isoforms, in humans, only FMO1 catalyzes the oxidation of the much larger tricyclic drugs. Rettie has proposed that the N-oxidation of imipramine or orphenadrine can serve as marker activities for human FMO1 (A. Rettie)³. However, this leaves unanswered the question whether this isoform, like pig FMO1, can catalyze the oxidation of nucleophiles bearing even more bulky substituents.

This question is addressed by the data summarized in Table 1, which indicate that the tricyclic drugs approach the upper size of compounds that can enter the active site of human FMO1. Although the K_m values for methimazole, thiourea, and phenylthiourea are virtually the same (3-5 µM), the larger 1,3-diphenylthiourea is virtually excluded from the enzyme-bound 4-hydroperoxyflavin in human FMO1. These data suggest that pig FMO1, which accepts thiocarbamides much larger than 1,3-diphenylthiourea (Guo et al., 1992), would be a poor model for human FMO1. The size limits for substrates accepted by human FMO1 appear more similar to those observed for rat or rabbit FMO1.

View this table: [in this window] [in a new window]   TABLE 1 Km and Vmax values for the oxidation of thiocarbamides catalyzed by human FMO1 expressed in insect cells Activities were measured by following substrate-dependent thiocholine oxidation as described in the text. Also see Guo et al. (1992). Values are means ± S.E. of triplicate measurements for each preparation. Kinetic constants were obtained from reciprocal plots of velocity of substrate-dependent thiocholine oxidation versus substrate concentration above and below the concentration required to give half maximal velocity.

The observation that the K_m and V_max values for methimazole and thiocarbamides that can access the enzyme-bound oxidant of human FMO1 are essentially the same suggest that tight binding enzyme-substrate complexes are not required for catalysis, and size restrictions measured with thiocarbamides can be used to predict activity with other xenobiotic nucleophiles.

For instance, all of the compounds in Fig. 1, except for 1,3-diphenylthiourea, are substrates for human FMO1. Although they differ considerably in structure, the space filling models, Fig. 2, reveal some common features. First, in the projections shown, the width of the smaller substrates (thiourea, phenylthiourea, methimazole, and N,N-dimethylaniline) is no more than 4.2 Å, and other substituents on the heteroatoms are limited to methyl groups or hydrogens. This suggests that these substrates could readily contact the 4a-hydroperoxy flavin located at the bottom of a substrate channel (or cleft) at least 5 Å in diameter. The depth of the substrate site cannot be estimated with any precision, but it appears to accommodate the side chains of imipramine, orphenadrine, and chloropromazine. Although phenothiazine is S-oxygenated by pig FMO1 (Nagata et al., 1990), activity with human FMO1 could not be detected at the highest concentrations soluble in the assay medium. The sulfur atom in phenothiazine, like that of 1,3-diphenylthiourea, is apparently excluded from the catalytic site by bulky groups adjacent to the sulfur atom. On the other hand, the side chain tertiary amine group of chlorpromazine is accepted by human FMO1 as are those of the structurally similar imipramine and orphenadrine. From the dimensions in Fig. 2, it appears that the hydrophenoxyflavin in human FMO1 lies no more than 5 Å below the surface of amino acid side chains that control access of nucleophiles into the substrate site.

View larger version (18K): [in this window] [in a new window]   Fig. 1.   Molecular structures of selected xenobiotics.

View larger version (50K): [in this window] [in a new window]   Fig. 2.   Space filling models of the compounds in Fig. 1. The heteroatom, susceptible to attack by FMO, is drawn facing the bottom of the page, and each compound was rotated to show its widest profile. The dimensions, calculated with the Spartan program, are in Å from the center of the atom on the outer edge of the widest part of the compound. The length of the side chains of imipramine and chlorpromazine is the distance from the center of the ring nitrogen to the center of the side chain nitrogen. That of orphenadrine is from the center of the side chain oxygen to the terminal nitrogen. The length of substituents on sulfur of ranidine is from the center of the sulfur atom to the ends of the molecule. The drawings and dimensions are courtesy of Dr. L. L. Poulsen (Department of Chemistry and Biochemistry, University of Texas at Austin).

Although activity of ranitidine with human FMO1 was not measured directly, the dimensions of the molecule shown in Fig. 2 suggest that human FMO1 can catalyze N-oxygenation but not S-oxygenation of ranitidine. The substituents on sulfur in ranitidine like those adjacent to sulfur on 1,3-diphenylthiourea would prevent contact of sulfur in these compounds with the enzyme-bound 4a-hydroperoxy flavin. The dimensions of the side chain bearing the N_,N-dimethyl group suggest that ranitidine should be a substrate not only for human FMO1 but also for FMO3 or any other isoform that accepts methimazole as a substrate.

Although size restrictions limiting access of nucleophiles into the catalytic sites of FMO isoforms can serve as a guide for predicting substrate activity, considerable caution must be exercised. Except for relatively rigid molecules such as sulindac sulfide (Light et al., 1982), most xenobiotics are flexible molecules that can assume different shapes in solution, and it is quite possible that only a very minor form would be responsible for substrate activity.

Despite these obvious limitations, the studies reported in this article indicate that size limitations may serve as an initial guide for predicting substrate activity in humans of a specific xenobiotic nucleophile bearing bulky substituents. The data also show that human FMO1 only catalyzes the oxidation of xenobiotics considerably smaller than those accepted by pig or guinea pig FMO1.