Modeling of Phase 2 Metabolism Enzymes

Whereas Phase 1 metabolism involves hydroxylation, oxidation, and reduction pathways, phase 2 metabolism involves primarily conjugation (of the phase 1 metabolites) with a variety of groups (e.g., sulfation, glucuronidation, glutathione conjugation, acetylation, amino acid conjugation, and methylation).67 With the exceptions of methylation and acetylation, phase 2 biotransformation reactions result in a large increase in hydrophilicity.68 Enzymes involved in phase 2 metabolism include, GST, MT, NAT, SULT, and UGT.67'68 Modeling of phase 2 metabolism has not received as much attention as the modeling of phase 1 metabolism for various reasons, as described below. A short overview of 3D modeling (homology modeling, pharmacophore modeling, and 3D-QSAR) will be presented in the following sections for the main enzymes involved in phase 2 metabolism.

5.34.3.1 Glutathione S-Transferases

GSTs catalyze the nucleophilic attack of the thiol of the tripeptide glutathione (GSH) on electrophilic substrates, a reaction usually resulting in addition or substitution, depending on the nature of the substrate.68 Many diverse compounds, including toxic xenobiotics and reactive products of intracellular processes such as lipid oxidation, act as GST substrates.69 The primary function of GST isoenzymes is generally considered to be detoxification of both endogenous and xenobiotic compounds.70,71 However, GSTs can also lead to the formation of more reactive intermediates, either directly or following conversion of glutathione conjugates by other biotransformation enzymes. In some cases the glutathione conjugates formed are labile, and dissociate to the parent electrophile in other tissues.71 Substrates for GSTs share three common features: they are hydrophobic, contain an electrophilic atom, and also react nonenzymatically with GSH.68

Mammalian cytosolic GST isoenzymes have been grouped in eight families, alpha, mu, pi, theta, kappa, sigma, zeta, and omega,68,72-75 which are distinguished by molecular masses, isoelectric points, and other properties. The primary sequence homology between subunits of the same family is > 70%, while interfamily subunit homologies are approximately 30-40%.68'70'76'77 In humans, multiple members of each subfamily class are now known.68 Cytosolic GSTs are found as dimeric proteins comprised of two subunits.78 Each isoenzyme subunit contains an active site consisting of two binding sites, one for the cofactor GSH (G-site) and one for the electrophilic substrate (H-site).78,79 Furthermore, one nonsubstrate-binding site (binding compounds that are not conjugated but transported by GSTs) is present in the dimer.

Comparison of cytosolic GST crystal structures reveals that all cytosolic GST isoenzyme subunits share a number of general structural features, though they differ considerably in detail. The tertiary structure and a large part of the secondary structure are generally maintained between cytosolic GSTs from different families and different species.

Microsomal GSTs are membrane bound, and, in spite of the fact that their catalytic function is similar to soluble cytosolic GSTs, they share no sequence identity with the latter GSTs, as was shown by a two-dimensional projection structure of a microsomal membrane-bound GST determined using electron crystallography.80 Microsomal GSTs occur as a trimer, which forms a pore-like structure with six a-helices delineating a region with the local symmetry axis in the center, suggesting a membrane transport function for these GSTs.80 This indicates that for comparable catalytic functions of microsomal and cytosolic GSTs, structurally different enzymes exist.80

The first models for a small-molecule bound to a GST isoenzyme suggested both hydrophobic interactions and hypothetical hydrogen bonds to polar groups in the active site to be important for binding and orientation of (a-bromoisovaleryl)urea for conjugation.81

Another small-molecule model was derived for rat GST 4-4 (rGST M2-2; nomenclature adapted from Mannervik etal.82 - the first letter indicates the species (h, human; m, mouse; p, pig; r, rat), the capital following 'GST' indicates the class, and the numbers indicate the specific GST), as a surrogate for the polymorphic hGST M1-1.83 The derived substrate model incorporated information on regio- and stereoselective product formation of 20 substrates covering three chemically and structurally different classes (aromatic diol epoxides, aromatic chlorides, pyrene oxides, and (aza)phenanthrene oxides). The model highlighted three interaction sites responsible for Lewis acid-Lewis base interactions as well as a region responsible for aromatic interactions.83 This substrate molecule model successfully predicted the conjugation to GSH of 11 substrates of GST 4-4 (representing three classes of compounds) that were not used to construct the model.83 This small-molecule model was subsequently improved with another series of substrates, which extend beyond the original model in one specific area.84 Based on these data, additional steric restrictions imposed by the protein in this region have been incorporated into the original small-molecule model.83

A small-molecule model based upon GSH analogs was constructed for hGST P1-1, which is expressed at elevated levels in tumor cells. A large number of GSH analogs were used to construct a pharmacophore for the G-site of GST P1-1, overlapping charged and hydrophobic centers.85 The pharmacophore was consequently docked into the crystal structure of GST P1-1,86 which indicated complementarity of all the major motifs.85

In the 1990s, when crystal structures were not available for all GSTs, several homology models were constructed. As there was generally at least one crystal structure available for each GST family, alignment between amino acid

Figure 2 Homology model of cytosolic rGST M2-2 showing the dimeric (top/bottom) character of the enzyme.89

sequences of the crystal structure and a GST to be modeled was relatively easy when using a GST from the same family as a template for the model, due to the high homology. Furthermore, as the overall tertiary and secondary structures are largely maintained in all GSTs, the quality of the resulting alignments was generally excellent.

Cachau etal.,87 Hsiao etal.,88 and De Groot et al.,89 used the crystal structure of rat mu class GST 3-3 (rGST M1-1)90 as the template for the construction of homology models for, respectively, human mu class GSTs M1b-1b, M2-2, and M3-3,87 chicken theta class GST,88 and rat mu class GST 4-4 (rGST M2-2, shown in Figure 2).89 All models closely resembled the crystal structures upon which the models were based. The modeled GST M2-2 structure87 agreed remarkably well with the crystallographic structure of human muscle GST M2-2.91 The orientation of four GSH conjugates docked into the GST 4-4 model appeared to be similar to the orientation of these substrates in a small-molecule model for GST 4-4.83 The homology model of GST 4-4 (combined with the small-molecule model83) was used successfully to identify amino acids that could be involved in binding or activation of substrates in the active site.89

Several homology models have been generated for the GST theta family, for example a model for hGSTT1, hGST T2,93 and for the mouse isoforms mGST T1, mGST T2, and mGST T3.94 The model for hGST T293 was used to consolidate the role of specific amino acids in the active site. The mouse models used the crystal structures of hGST T2 as a template, and were used to compare the active sites and explain the inability of mGST T1 and mGST T3 to metabolize 1-menaphthyl sulfate.94

The crystal structure of hGST A1-1 has been used to generate homology models for rat GSTs of the alpha family, rGST A3 and rGSTA5.95 The resulting models were again used to investigate the importance of specific amino acids in the active site for activity toward conjugation of Aflatoxin B1, supported by site-directed mutagenesis.95

GSTs from the pi family have so far only been used for docking studies (hGST P1a, hGST P1b, and hGST P1c)96 and docking algorithm evaluation.97

Due to the number of crystal structures currently available for GSTs from all families, pharmacophore and homology modeling for these enzymes is not receiving as much attention as that for the cytochromes P450, compared to the efforts expended in the 1990s.

5.34.3.2 Methyltransferases

Methylation is a common but minor pathway for phase 2 metabolism of xenobiotics, and is catalyzed by various enzymes, such as thioether methyltransferase, phenyl O-methyltransferase, catechol O-methyltransferase (COMT), phenylethanolamine N-methyltransferase (PNMT), and histamine N-methyltransferase (HNMT).68 Several modeling studies have been performed on MTs involved in endogenous pathways (e.g., DNA and RNA methylation), as well as studies in various plants. However, modeling of the human counterparts involved in drug metabolism has received relatively little attention.

A homology model of human HNMTwas developed, and used to determine the location of the amino acid linked to polymorphism for this enzyme. This model supports experimental observations with regards to protein stability.98 A set of homology models for rabbit and human indolethylamine N-methyltransferases was obtained, and used successfully to explain the observed species differences and polymorphisms.99 These homology models were based on the rat crystal structure of COMT. This crystal structure was also used for a docking study that, in combination with experiments using the human enzyme, resulted in various suggestions for mechanisms controlling the methylation by human COMT.101

A series of comparative molecular field analysis (CoMFA) studies on PNMTand the a2-adrenoceptor was performed in order to investigate the factors that determine potency and selectivity for a series of PNMT inhibitors. The results of the CoMFA studies were similar to those of previous QSAR analysis. COMT inhibitors are used in the treatment of Parkinson's disease, but their application is limited due to adverse side effects. Docking in the rat crystal structure, CoMFA analysis, and GRID/GOLPE models were used to examine a series of COMT inhibitors. The three models were in agreement with each other, and suggested that an interaction between the catechol oxygen atoms and the magnesium ion in COMT is important, while several hydrogen-bonding and hydrophobic contacts influence inhibitor binding.104

5.34.3.3 N-Acetyltransferases

N-Acetylation is a major route of metabolism for compounds containing an aromatic amine or hydrazine group, which are converted to aromatic amides and hydrazides, respectively.68 The reaction is catalyzed by NATs, which are found in most mammals, and require the cofactor acetyl-CoA.68 Humans, rabbits, and hamsters express only two NATs (NAT1 and NAT2), whereas mice express a third (NAT3).68 NAT1 and NAT2 have different but overlapping substrate specificities, and both show genetic polymorphisms.68 Homology models have been constructed for both human NAT1 and NAT2, based on the crystal structure of NAT from Salmonella typhimurium (StNAT). The model for NAT1 was used to predict that human NATs have adopted a common catalytic mechanism from cysteine proteases to accommodate acetyl transfer reactions,105 while the NAT2 model was used to explain the polymorphism of this enzyme.106 The crystal structures of S. typhimurium and Mycobacterium smegmatis NATs were used to generate a homology model of Mycobacterium tuberculosis NAT (MtNAT), and were used for specific inhibitor design against MtNAT (which is responsible for resistance against the antitubercular drug isoniazid).107

5.34.3.4 Sulfotransferases

Sulfate conjugation generally produces a highly water-soluble sulfuric acid ester. The reaction is catalyzed by SULTs, a large gene family of soluble (cytosolic) enzymes, and involves transfer of a sulfonate (SO^). The required cofactor for the reaction is 3'-phosphoadenosine 5'-phosphosulfate (PAPS).68'108 Although the most common substrates for this enzyme are phenols and aliphatic alcohols, metabolism is not limited to those classes.68 In mammals, sulfation is involved in the detoxification of therapeutic, dietary, and environmental xenobiotics, and contributes to the homeostasis and regulation of numerous biologically active endogenous chemicals such as steroids, iodothyronines, bile acids, and neurotransmitters.109,110 In addition, for a large number of procarcinogens sulfation is the terminal step in the bioactivation pathway, and is necessary to reveal their mutagenic/carcinogenic activity. Multiple SULTs have been identified in all mammalian species, and are members of five gene families (SULT1 to SULT5). Crystal structures are now available for seven human SULTs: the cytosolic SULT1A1, SULT1A3, SULT1E1, SULT2A1, SULT2B1A, and SULT2B1B and the Golgi-resident NDST-1.108 Docking can be used to predict binding modes within these crystal structures, as was done for dopamine in SULT1A3. The PAPS binding pocket is highly conserved, while the substrate-binding site of each SULT reflects the differences in substrate specificity in these crystal structures: the cytosolic SULTs typically recognize small hydrophobic molecules, whereas the Golgi-resident SULTs recognize hydrophilic carbohydrates and tyrosine residues in peptides.108 The design of SULT inhibitors has focused on exploring environmental toxins and dietary agents, rather than drug development.108 The crystal structure of mouse Sult1e1 has been used to build a homology model of human SULT1E1 before the human crystal structure was solved, to identify the potential (allosteric) binding pocket using a series of hydroxylated polychlorinated biphenyls.112

Similarly, a model of rat aryl sulfotransferase IV (AST IV) was constructed, based on mouse Sult1e1.113 Docking of a series of N-hydroxyarylamines in the homology model of AST IV highlighted specific steric constraints within the active site, which could explain the change from substrate to competitive inhibitor within this series of compounds.113 CoMFA has been performed on Km values of a series of 35 substrates of rat AST IV, and could explain the activities of six more substrates.114 The CoMFA results114 showed a good fit with the predicted active site in the homology model of AST IV113

Four CoMFA models were derived from the Km values for 95 substrates of SULT1A3. The different models used subsets of the data set or the entire data set and different molecular alignment rules.115 All four models were statistically significant, and highlighted factors affecting binding in the SULT1A3 binding site.

Recently, the crystal structures of numerous human SULTs have been obtained by the Structural Genomics Consortium.116

5.34.3.5 Uridine Diphosphate (UDP)-Glucuronosyltransferases

Glucuronidation is a major pathway of xenobiotic biotransformation in most mammalian species, and requires the cofactor uridine diphosphate-glucuronic acid.68,117 The reaction is metabolized by UGTs (also called glucuronyl-transferases), which are present in many tissues.68,117 The site of glucuronidation is generally an electron-rich nucleophilic heteroatom (oxygen, nitrogen, or sulfur).68 Human UGTs are a family of enzymes that detoxify many hundreds of compounds by their conjugation to glucuronic acid, rendering them harmless, more water-soluble, and, hence, excretable. Genetic inheritance, age, and environmental factors largely determine the different profiles of the inducible hepatic UGTs. Variation in the complement of these UGTs may result in dramatic differences in the safe elimination of toxic metabolites.

UGTs are classified into two distinct gene families: UGT1 and UGT2, the latter showing genetic polymorphisms.68 In humans, up to 16 different functional UGT isoforms belonging to subfamilies 1A and 2B have been characterized, showing that these isoforms possess distinct but frequently overlapping substrate specificities.117 Certain unexpected adverse reactions and decreased efficiency of drug therapy may be due to interindividual variations in the expression of UGTs.

Human UGTs have been studied extensively using pharmacophore and 3D-QSMR approaches.69,118 Pharmacophores for UGT1A1,69,119-121 UGT1A3,121 UGT1A4,118,119,121,122 UGT1A6,121 UGT1A7,121 UGT1A8,121 UGT1A9,69,121 UGT1A10,121 UGT2B4,121 UGT2B7,121 UGT2B15,121 and UGT2B17121 have been reported. Pharmacophore modeling of UGTs is hampered by the ability of UGT isoforms to accommodate multiple substrate-binding modes, and requires the construction of a 'glucuronidation feature' for use in the computational algorithm.69,122

Substrates of UGT1A1120 and UGT1A4122 share two key hydrophobic regions 3 A and 6-7 A from the site of glucuronidation.119,120,122 These results were confirmed by pharmacophore-based 3D-QSAR and molecular field-based 3D-QSAR (self-organizing molecular field analysis) methods.118,120,122 An aromatic ring attached to the nucleophilic group was found to increase the likelihood of glucuronidation by UGT1A6, UGT1A7, and UGT1A9.121 The pharmacophore of UGT1A9 consists of two hydrophobic regions at similar distances to the site of glucuronidation, as for UGT1A1 and UGT1A4,119,120,122 but also contains a hydrogen bond acceptor feature close to the most distant hydrophobic feature.118 A large hydrophobic region close to the site of glucuronidation and a hydrogen bond acceptor 10 A from the site of glucuronidation were common in most UGT2B4 substrates.

Detoxify the Body

Detoxify the Body

Need to Detoxify? Discover The Secrets to Detox Your Body The Quick & Easy Way at Home! Too much partying got you feeling bad about yourself? Or perhaps you want to lose weight and have tried everything under the sun?

Get My Free Ebook


Post a comment