Introduction

Top Abstract Introduction Materials and Methods Results Discussion References

Enzymatic O-methylation of aromatic hydroxyl groups is an important step in the metabolism of endogenous catecholamines as well as of many xenobiotics. Catechol-O-methyltransferase (COMT¹, EC 2.1.1.6) plays a central role in the metabolic inactivation of catecholic neurotransmitters, such as dopamine, norepinephrine, and epinephrine. Catecholestrogens and neuroactive drugs possessing a catechol structure, such as L-3,4-dihydroxyphenylalanine (L-DOPA) are also O-methylated by COMT (Guldberg and Marsden, 1975; Männistö et al., 1992). The latter is of particular importance since L-DOPA remains the principal drug used in the therapy of Parkinson's disease.

Selective inhibition of COMT reduces the level of O-methylation of L-DOPA to 3-O-methyl-L-DOPA and increases the amount of L-DOPA gaining access to the brain (Männistö et al., 1992; Bonifati and Meco, 1999). This observation has prompted interest in the development of potent COMT inhibitors, to be used as adjuncts in L-DOPA therapy of Parkinson's disease (Männistö and Kaakkola, 1989; 1990).

COMT requires Mg²⁺ ions for catalysis and uses the ubiquitous cofactor S-adenosyl-L-methionine (AdoMet) as the methyl donor. The catalytic mechanism of COMT has been extensively detailed on the basis of structural (Vidgren et al., 1994) and theoretical studies (Vidgren and Ovaska, 1997; Zheng and Bruice, 1997; Lau and Bruice, 1998; Kahn and Bruice, 2000; Kuhn and Kollman, 2000; Lautala et al., 2001). Binding of the catechol moiety to the catalytic site positions one of the catechol hydroxyls in close proximity to the activated methylsulfonium group of AdoMet. The reaction involves the transfer of the methyl group from the cofactor to the proximal hydroxyl group, generating the corresponding mono-O-methylated catechol and S-adenosyl-L-homocysteine (Guldberg and Marsden, 1975).

Nitro-substituted catechol derivatives are among the most potent COMT inhibitors. They possess the same binding motif as the catechol substrates, but the presence of the strong electron-withdrawing nitro function hinders their reactivity toward O-methylation (Bäckström et al., 1989; Borgulya et al., 1989). 1-(3,4-Dihydroxy-5-nitrophenyl)-2-phenyl-ethanone (BIA 3-202; Fig. 1) is a recently developed nitrocatechol-type COMT inhibitor (Parada et al., 2001; Learmonth et al., 2002), the potential benefits of which for the treatment of Parkinsonian patients are presently under clinical evaluation.

View larger version (25K): [in this window] [in a new window]   Fig. 1.   COMT ligands and compounds studied throughout this work. Atom numbering used in the text is indicated at selected positions.

The complexes between several nitrocatechol inhibitors and the soluble form of rat COMT (S-COMT) have been studied by X-ray crystallography. The first crystal structure reported (Vidgren et al., 1994) shows the enzyme complexed with the inhibitor 3,5-dinitrocatechol (3,5-DNC, also OR-486). Later on, the crystallization of the enzyme with another nitrocatechol-type inhibitor (OR-1840) was mentioned in the literature (Vidgren et al., 1999), although no detailed description or atomic coordinates were made available. More recently, two other structures were finally disclosed, showing the rat S-COMT complexed with a new tight-binding nitrocatechol inhibitor (BIA 3-335) (Bonifácio et al., 2002) and a bisubstrate inhibitor (Lerner et al., 2001).

In vitro, COMT catalyzes the mono-O-methylation of substituted catechol substrates at either the meta (3-O-methylation) or para (4-O-methylation) hydroxyl functions, relative to the substituent at C1 (Fig. 2). The ratio of meta/para O-methylation varies with the nature of the catechol substituent and the experimental conditions (Männistö et al., 1992) However, in vivo, the regioselectivity of O-methylation of those substrates generally shifts toward much higher meta/para ratios, with lesser or null amount of the p-O-methylated products being formed (Daly et al., 1960; Frère and Verly, 1971).

View larger version (13K): [in this window] [in a new window]   Fig. 2.   Two opposite bound orientations of substituted catechols, in the active site of COMT. R, variable substituent. Steric impediments determine that O-methylation is only possible at the hydroxyl oxygen closer to the AdoMet cosubstrate.

The nitrocatechol inhibitors, on the other hand, are only negligibly O-methylated, in spite of the fact that they interact with the catalytic site of COMT. Nevertheless, after administration of the nitrocatechol inhibitor tolcapone (RO 40-7592) to human volunteers, small amounts of the m-O-methyltolcapone derivative are formed, which accumulates for several days in plasma (Da Prada et al., 1994; Dingemanse et al., 1996). However, no detectable p-O-methyltolcapone is reported. The same qualitative results are obtained with the new nitrocatechol inhibitor BIA 3-202, with which only marginal O-methylation of the m-hydroxyl group is observed in the rat and human (P. Soares-da-Silva, unpublished data).

Explanation of the apparently different regioselectivities of O-methylation by COMT in vivo versus in vitro conditions poses an intriguing challenge that has not been solved to date. Moreover, the apparently exclusive in vivo m-O-methylation of BIA 3-202 (and tolcapone) cannot be explained through analysis of the crystal structures of the complexes between COMT and this class of inhibitors. By simple observation of the X-ray structures, it is clear that in every case the catechol hydroxyl group in position meta (relative to the C1 substituent) is always sterically inaccessible to the methylsulfonium group of AdoMet.

In the present work, we study the nature of the molecular interaction of BIA 3-202 with the active site of rat S-COMT and propose a plausible explanation for the apparent paradox of the regioselectivity of its O-methylation in vivo. A putative structural model of the enzyme-inhibitor complex is presented, and the implications for the design of new inhibitor molecules, as well as for the understanding of their metabolism via O-methylation are discussed.

	Materials and Methods

Top Abstract Introduction Materials and Methods Results Discussion References

Chemicals and Enzyme. BIA 3-202, the individual meta-O- (BIA 3-270) (Learmonth et al., 2002) and para-O- (BIA 3-449) methylated derivatives (Learmonth and Freitas, 2002), tolcapone and its derivatives m-O-methyltolcapone and p-O-methyltolcapone were synthesized in the Laboratory of Chemistry (BIAL, Mamede do Coronado, Portugal). The recombinant rat soluble COMT was produced in Escherichia coli and was purified as previously described (Bonifácio et al., 2001; Rodrigues et al., 2001).

Molecular Modeling. The structures of the catechol-type ligands were built and optimized at the PM3 semiempirical level with Spartan Pro v1.03 program (Wavefunction, Inc., Irvine, CA). The atomic coordinates of rat S-COMT were obtained from the Protein Data Bank (1vid.pdb), water molecules were deleted (except HOH400 coordinated to Mg²⁺), and all hydrogens added to the protein. The catechol was docked with both hydroxyls protonated, and Lys144 was considered deprotonated. Unrestrained flexible-docking of the catechol ligands and the COMT active site was performed with the program GOLD v1.1 (Jones et al., 1995a,b, 1997b) using a genetic algorithm optimizer. Default parameters were employed except that the torsion constants for rotation about the C.ar-N.pl3 and C.ar-C.2 bond types were increased to 12.0 kcal/mol to hinder unrealistic free rotations of the catechol-nitro and catechol-carbonyl bonds, respectively. For the genetic algorithm used in the conformation space exploration, a population of 100 individuals (conformations) was subjected to 10⁵ mutational generations with a selection pressure of 1.1. All atoms at COMT molecular surface within a radius of 14.0 Å from the Mg²⁺ ion were used as the target-binding site. Catechol coordination to the Mg²⁺ ion is implicitly treated by GOLD as a special hydrogen bond, where the metal behaves as a hydrogen bond donor. Therefore, no constraints were used to force the catechol into the catalytic position. The fit of a given protein-ligand interaction configuration is evaluated by a scoring function, which includes the intramolecular torsion energy of the ligand and both the intermolecular van der Waals and hydrogen-bonding potentials. As with all genetic algorithm optimizers there is no way to ascertain that the global minimum energy solution is obtained, therefore 20 independent docking runs were performed for each ligand, starting from random configurations, and the twenty optimized structures were analyzed as an ensemble of alternative complexes.

Metabolic Studies: in Vitro O-Methylation. Methylation of BIA 3-202 and tolcapone was evaluated by incubating 10 µM of the nitrocatechol compound with 1 µM COMT in enzyme reaction mixture. The reaction was terminated with one volume 1% formic acid in acetonitrile. Reaction products were analyzed by LC-(AP-ESI)MS (HP 1100 Series, Agilent Technologies, Palo Alto, CA) with negative ion detection. The separation was performed on a Lichrospher 100 RP-18 column (LiChroCART 250-3, 5 µm, Merck, Darmstadt, Germany). The mobile phases used were A, water/formic acid 1% (v/v) and B, acetonitrile/formic acid 1% (v/v). The gradient conditions were at 0 to 3 min 50% A and 50% B; at 10 min 47% A and 53% B; at 10 to 11 min 47% A and 53% B; and at 15 min 40% A and 60% B. The flow rate was 0.8 ml/min, and injection volume was 30 µl. Selected ion monitoring with detection set for the molecular ion of each compound of interest was used for quantification. The analytical range used extended from 20 to 500 ng/ml for the standards of meta-O- and para-O-methylated BIA 3-202 and tolcapone.

Metabolic Studies: Microsomal Metabolism. BIA 3-270, BIA 3-449, m-O-methyltolcapone, and p-O-methyltolcapone (100 µM) were independently incubated with rat liver microsomes (prepared by differential centrifugation) containing 0.8 mg of total protein in 100 mM phosphate buffer, pH 7.4, 60 mM MgCl₂ as described elsewhere (Busse et al., 1995). The reactions, started by the addition of 4 mg/ml NADPH, were carried out at 37°C and were terminated with one volume 1% formic acid in acetonitrile. Reaction products were analyzed by LC-(AP-ESI)-MS as described above. Control experiments were established by testing the same reactions in the absence of added drug or the microsomes.

	Results

Top Abstract Introduction Materials and Methods Results Discussion References

Molecular Interactions: Docking of BIA 3-202. The structure of the functional complex formed between BIA 3-202 and COMT was studied by unrestrained flexible-docking simulations. In addition, 3,5-dinitrocatechol and L-DOPA (Fig. 1) were also independently docked to the active site of the enzyme to assess the performance of the docking procedure and its ability to predict experimental data. The docked configurations of 3,5-DNC were compared with that of the crystallographic complex (1vid.pdb) and those of L-DOPA used to explain the observed regioselectivity (meta/para-O-methylation ratio) of S-COMT with this catechol substrate.

View larger version (22K): [in this window] [in a new window]   Fig. 3.   Docking results of BIA 3-202. Twenty independently optimized docked structures are represented in decreasing order of their interaction scores (dashed line). The distances between the methylsulfonium carbon (CE) of AdoMet and the catechol pO () and mO () are shown for every solution structures and the corresponding binding modes are sketched.

View larger version (84K): [in this window] [in a new window]   Fig. 4.   Three most representative docked configurations of BIA 3-202 in the active site of COMT. The two hydroxyl groups of the catechol moiety are coordinated to the Mg2+ ion and form hydrogen bonds to Glu199 and Lys144. Amine nitrogens and carboxylic oxygens are represented in dark blue and red, respectively. Solution structures A and B represent binding in para orientation whereas in solution C the inhibitor binds in the meta configuration.

Validation. As mentioned above, the COMT inhibitor 3,5-dinitrocatechol and the substrate L-DOPA were also docked to the enzyme under the same conditions as for BIA 3-202, to assess the validity of the docking procedure. These two compounds were chosen since they are structurally related to BIA 3-202, and there is sufficient structural and functional information available to which the results of the simulation could be correlated.

View larger version (17K): [in this window] [in a new window]   Fig. 5.   Superposition of 20 independently docked configurations of 3,5-DNC (thin lines) on the crystallographic coordinates of the complex (1vid.pdb, thick lines).

View larger version (28K): [in this window] [in a new window]   Fig. 6.   Docking results of 3,5-DNC (A) and L-DOPA (B). In each panel, 20 independently optimized docked structures are represented in decreasing order of their interaction scores (dashed line). The distances between the methylsulfonium carbon (CE) of AdoMet and the catechol pO () and mO () are shown for each solution structure.

Metabolic Studies: in Vitro O-Methylation and O-Demethylation. To verify the hypothesis raised by the docking simulations, we investigated whether BIA 3-202 could, in fact, undergo in vitro enzymatic O-methylation at both pO and mO positions (see Fig. 1). Nitrocatechols are known to be extremely poor substrates of COMT, even though they bind to the catalytic site. However, after incubation of 1.0 µM purified recombinant rat COMT (soluble form) with an excess of BIA 3-202 (10 µM), small amounts of both m-O- (BIA 3-270) and p-O- (BIA 3-449) methylated products were formed in a time-dependent fashion (Fig. 7A). Control experiments with no enzyme added revealed no traces of the methylated products.

View larger version (26K): [in this window] [in a new window]   Fig. 7.   A, O-methylation of BIA 3-202 into BIA 3-270 and BIA 3-449, by the recombinant S-COMT; error bars represent the S.E.M. from three replicates; B, O-demethylation of BIA 3-270 and BIA 3-449 by rat liver microsomes; curves represent the evolution of BIA 3-202 formed from each of the two O-methylated metabolites; error bars represent the S.E.M. from four replicates.

View larger version (29K): [in this window] [in a new window]   Fig. 8.   A, O-methylation of tolcapone into m-O-methyltolcapone and p-O-methyltolcapone derivatives, by the recombinant S-COMT; error bars represent the S.E.M. from three replicates; B, O-demethylation of m-O- and p-O-methyltolcapone by rat liver microsomes; curves represent the formation of tolcapone from each of the two O-methylated metabolites; error bars represent the S.E.M. from four replicates.

	Discussion

Top Abstract Introduction Materials and Methods Results Discussion References

The observed conversion of BIA 3-202 (P. Soares-da-Silva, unpublished data) or tolcapone (Da Prada et al., 1994; Dingemanse et al., 1996) into the respective m-O-methylated (but not the p-O-methylated) products, in the plasma of humans and rats, constitutes an intriguing paradox with respect to the pool of crystallographic information available. In all X-ray structures of complexes between COMT and related nitrocatechol inhibitors that have been reported (Vidgren et al., 1994, 1999; Lerner et al., 2001; Bonifácio et al., 2002), the meta-hydroxyl oxygen of the nitrocatechol motif is sterically inaccessible to the methylsulfonium group of AdoMet. Therefore, the available crystallographic data may not be strictly extended to the case of BIA 3-202 (or tolcapone). Instead, alternative modes of binding between the inhibitor and the catalytic site of the enzyme must occur in solution. In this work, the molecular basis of the interaction of BIA 3-202 with catechol-O-methyltransferase was characterized by molecular modeling and in vitro enzymatic experiments.

The unrestrained flexible-docking simulations, herein presented, predict that BIA 3-202 interacts competitively to the substrate of the enzyme, as is generally reported for the nitrocatechol COMT inhibitors (Vidgren and Ovaska, 1997). According to the obtained models, the enzyme-inhibitor complex is stabilized by the coordination bonds formed between the two catechol hydroxyl oxygens and the Mg²⁺ ion and by hydrogen bonds between the same hydroxyls and the residues Glu199 and Lys144. In addition, extensive van der Waals contacts also contribute to the stabilization of the complex. The benzyl carbonyl group of the 1-(2-phenyl-ethanone) substituent may have an important role in binding thermodynamics, as indicated by the dramatic decrease in potency of the inhibitor upon replacement of the carbonyl group (CO) by methylene (CH₂) (Learmonth et al., 2002). Our docking results suggest that this group may have a 2-fold role in the stabilization of the complex. On one hand it may contribute to reducing the desolvation free energy penalty upon formation of the complex. On the other hand, fitting into the narrow catalytic site forces the substituent atoms attached to the catechol C1 to remain in the same plane of the catechol ring. Since the carbonyl group and the catechol aromatic ring tend to be coplanar in solution because of conjugation of the respective  systems, the association process is expected to proceed with minimal loss of conformational entropy. This is thought to be a relevant characteristic of BIA 3-202. Interestingly, catechol substrates with a planar conjugated substituent or a fused unsaturated ring meta to the hydroxyl group, such as 3,4-dihydroxybenzoic acid ethyl ester, caffeic acid, or 6,7-dihydroxycoumarin show higher affinity for COMT than equivalent nonconjugated compounds (Lautala et al., 2001). Finally, inspection of the catalytic site clearly shows that when the nitrocatechol motif is bound to the catalytic site, only one of the catechol hydroxyls is sterically accessible to the methylsulfonium of AdoMet.

Regarding the multiple possibilities of interaction, the solution structure of the complex between COMT and BIA 3-202 is better described as an ensemble of alternative bound configurations, rather than one unique form. The two main predicted bound configurations - meta and para - are related to each other by a 180-degree rotation about the catechol symmetry axis and force either pO or mO hydroxylates, respectively, into catalytically competent positions. The implication of this conclusion is that, if there is any O-methylation of BIA 3-202, it should be able to occur at either meta or para hydroxyl oxygens.

Indeed, our in vitro experiments showed that in the presence of an excess of BIA 3-202, the recombinant enzyme could catalyze the formation of small but equivalent quantities of both p-O- and m-O-methylated derivatives. These results constitute direct evidence that the inhibitor must be able to bind in at least two alternative orientations that enable the formation of the two possible O-methylated products.

The reasons why the in vivo metabolism of BIA 3-202 (P. Soares-da-Silva, unpublished data) or tolcapone (Da Prada et al., 1994; Dingemanse et al., 1996) leads to the accumulation of their respective m-O-methylated but not the p-O-methylated products remains an open question. However, the results of this work indicate that p-O-methylated derivatives of both inhibitors are subjected to further microsomal metabolism, whereas no significant conversion is observed with the respective meta-isomers. Whereas the major conversion routes undertaken by the methylated compounds are still to be disclosed, we have shown that at least part of the disappearance of p-O-methylated BIA 3-202 and p-O-methyltolcapone is because of their O-demethylation, catalyzed by a microsomal enzyme system. Moreover, this demethylation is shown to be regioselective toward the para-methylated metabolites. Indeed, regioselective O-demethylation of compounds possessing single or conjugated aromatic rings with adjacent methoxy groups has been previously reported. Such is the case of the calcium channel blocker verapamil, which possesses a 3,4-dimethoxyphenyl motif and is shown to be selectively O-demethylated at the para position by a cytochrome P450 of the 2C subfamily (Busse et al., 1995). Furthermore, other para-substituted methoxyphenyl compounds also undergo O-demethylation by different cytochromes P450 isoforms (Jones et al., 1997a; Kuriya et al., 2000).

It is proposed therefore that BIA 3-202 may in fact undergo in vivo O-methylation by COMT at either meta or para catechol hydroxyl groups but that subsequent metabolic routes would lead to preferential reutilization of the p-O-methylated derivative (BIA 3-449). The overall balance is the observed accumulation of the m-O-methylated metabolite (BIA 3-270) over the para-isomer. A similar mechanism is suggested for the related COMT inhibitor, tolcapone.

Additional details of the docking experiments are discussed in the following paragraphs. The validity of the theoretical approach was assessed by performing additional docking experiments between 3,5-DNC or L-DOPA and COMT. The crystallographic structure of the complex COMT/3,5-DNC is correctly predicted, as is the regioselectivity of O-methylation of L-DOPA by the enzyme.

The comparison of the docking results of BIA 3-202 with those of 3,5-dinitrocatechol provide some insights into the reasons for the different preferences of meta/para binding modes of the two ligands. We have shown that the nitrocatechol motif is able to adopt two opposite orientations in the catalytic pocket. Preliminary docking simulations with 3-nitrocatechol (data not shown) indicate, however, that the nitro group itself has a clear tendency to dock exclusively on the side of the AdoMet cosubstrate (corresponding to para geometry). The same behavior is predicted herein for the related 3,5-DNC. Hence, the additional nitro group at position C5 (see Fig. 1) does not appear to counteract that preference. Indeed, it stacks well between the two hydrophobic residues Trp38 and Pro174, maximizing van der Waals contacts (Fig. 5). The 1-(2-phenyl-ethanone) substituent of BIA 3-202 has a different effect, although. Being larger and more flexible than the nitro group, the 1-(2-phenyl-ethanone) substituent is able to explore additional favorable interactions with neighboring residues. Such is the case illustrated in Fig. 4C, where extensive hydrophobic interactions between the phenyl side chain of the inhibitor and the indole ring of Trp143 apparently overcomes the steric constraints resulting from binding the nitrocatechol moiety in meta configuration. This means that the 5-nitro and the 1-(2-phenyl-ethanone) substituents of BIA 3-202 must compete with each other to drive the formation of the complex in either para or meta orientation, respectively. The observed overall result of such competition is a mixture of the two binding configurations.

Despite the efforts undertaken to co-crystallize COMT with BIA 3-202, no crystals have been obtained. Naturally, the coexistence of multiple forms of the complex is one possible explanation for the lack of success in the co-crystallization. It should be noted, on the other hand, that the inhibitor 3,5-DNC, the bound structure of which has been determined by X-ray, is predicted by our approach to have only one highly preferential binding mode.

With regard to the regioselectivity of COMT with its natural substrates, the polar substituents (R) have been shown to generally favor O-methylation of catechols at meta rather than para position relative to R (Creveling et al., 1972; Lotta et al., 1995).

Previous molecular dynamics studies have suggested that ionized carboxylic or amine groups would guide the catechol substituent into the meta position (relative to the methylation site), to avoid unfavorable electrostatic interactions with active site hydrophobic residues (Lotta et al., 1995; Lau and Bruice, 1998; Kuhn and Kollman, 2000). Although electrostatics typically play an important role in molecular interactions, it is interesting to note that in the case of L-DOPA, the regioselectivity can be adequately explained with the present results, without taking into account explicit electrostatic terms. In GOLD scoring function, van der Waals and hydrogen-bonding potentials are the only intermolecular terms used to guide the structure optimization.

The binding pocket of COMT if flanked, on one side (opposite to the methylation site), by the two hydrophobic and bulky residues Trp38, Pro174, and Leu198, which increase locally the height of the cavity and narrow the entrance space. As a consequence, the amine and carboxylic groups attached to the sp³ -carbon of L-DOPA can hardly fit between those three residues and tend to direct the catechol substituent into the opposite direction (against Trp143), where the ionized groups become extensively solvated. Such steric constraints contribute to populate catechol binding-modes that favor O-methylation at position meta (mO). Hence, the current results, although strictly qualitative, suggest an alternative explanation for the observed COMT regioselectivity with L-DOPA.

In general terms, the information herein presented provides evidence of the role of molecular modeling as complimentary tools to classical experimental techniques. In particular, the current findings are expected to provide important insights for the design of new COMT inhibitors and for better understanding their metabolism by COMT. Despite the obvious similarities between the core structure of BIA 3-202 and that of other nitrocatechol inhibitors already co-crystallized with COMT, the actual solution structures of these complexes may be much more diverse than those that have been provided by X-ray crystallography. This knowledge could expand the possibilities of exploring additional intermolecular interactions between COMT and new inhibitors.