Relationship Approaches to Absorption Distribution Metabolism and Excretion Predictions

Since 3D-QSAR approaches can account for specific interactions between ligands and a target, and suggest pharmacophore hypotheses, they have found several applications in metabolism prediction. Analogously, 3D-QSAR has been applied extensively to predict the interactions between ligands and transporters, since such targets, like enzymes, are biomacromolecules. Conversely, the prediction of permeation processes finds rarer applications in 3D-QSAR, as the behavior of fuzzy targets is less suitable for these computational methods. This section will consider, in turn, drug-metabolizing enzymes (Section 5.24.4.2.1), transporter proteins (Section 5.24.4.2.2), and permeation processes (Section 5.24.4.2.3).105

5.24.4.2.1 Applications of three-dimensional quantitative structure-activity relationship approaches to drug-metabolizing enzymes

5.24.4.2.1.1 Oxidoreductase

The cytochrome P450 enzymes (CYPs, EC 1.14.14.1) are membrane-bound proteins that catalyze oxidative reactions of xenobiotics and endobiotics with broad substrate specificity, playing a key role in drug metabolism. Briefly, CYP catalysis involves reduction of the heme iron, with resulting cleavage of the oxygen-oxygen bond. The electrons are donated by reductases, creating an 'electron transfer chain.'106'107

The CYP proteins are classified into families and subfamilies according to their sequence homology. More than 3000 CYP gene sequences are known. The first three families (CYP1, CYP2, and CYP3) are enzymes involved in drug metabolism, being responsible for about 90% of drug oxidations in humans, while other families play specific roles (e.g., CYP11, CYP17, CYP19, and CYP21 are implicated in steroid biosynthesis).

The 3D-QSAR techniques are suitable for studying CYP-catalyzed oxidations as they can analyze how ligands may bind to these enzymes as substrates and/or inhibitors. It is worth noting that in these studies, 3D-QSAR approaches are mainly used to build pharmacophore models rather than to yield predictive equations of limited interest, due to the complexity of these enzymes and the heterogeneity of their substrates. The soundness of the results was often confirmed by other computational approaches (e.g., homology modeling and molecular docking) and by experimental studies (x-ray crystallography and/or site-directed mutagenesis). This section will consider some relevant examples of 3D-QSAR

Figure 8 The CYP2B6 substrate pharmacophore illustrating three hydrophobic areas (cyan) and a HBA feature (green), with a vector in the direction of the putative hydrogen bond. (Reproduced from Ekins, S.; Bravi, G.; Ring, B. J.; Gillespie, T. A.; Gillespie, J. S.; VandenBranden, M.; Wrighton, S. A.; Wikel, J. H. J. Pharmacol. Exp. Ther. 1999, 288, 21-29, with permission from the American Society for Pharmacology and Experimental Therapeutics.)

Figure 8 The CYP2B6 substrate pharmacophore illustrating three hydrophobic areas (cyan) and a HBA feature (green), with a vector in the direction of the putative hydrogen bond. (Reproduced from Ekins, S.; Bravi, G.; Ring, B. J.; Gillespie, T. A.; Gillespie, J. S.; VandenBranden, M.; Wrighton, S. A.; Wikel, J. H. J. Pharmacol. Exp. Ther. 1999, 288, 21-29, with permission from the American Society for Pharmacology and Experimental Therapeutics.)

studies on CYP ligands. The relevance of these studies lies in the major role of CYP enzymes in human metabolism; understanding the basis of their activity is important to determine their role and predict effects on new substrates.108'109 Extensive analyses have been performed on CYP2B6 substrates using two 3D-QSAR approaches, namely WHIM and pharmacophore mapping.110 Both methods suggest the crucial relevance of three hydrophobic regions and one HBA, located at defined distances (Figure 8). These results are in agreement with classic QSAR studies that correlated binding affinities to CYP2B6 with log P and hydrogen-bonding descriptors.111 It is interesting to observe that both 3D-QSAR methods, even if conceptually different, yielded very similar results, suggesting some degree of mutual validation, although both methods failed to predict molecules not included in the training set. Docking analyses on homology-modeled CYP2B6 revealed that its binding site consists of three well-defined hydrophobic binding pockets adjacent to the catalytic heme. The size, shape, and position of these hydrophobic pockets are in agreement with pharmacophore models.112,113

The inability to predict CYP activity for substrates not included in training sets is explained by considering the broad specificity of these enzymes, which renders difficult or impossible the creation of a unique pharmacophore hypothesis. For example, the CYP2C family can bind both neutral and anionic ligands, and different models were developed to accounts for these two classes of molecules. All models show a hydrophobic/aromatic region between the hydroxylation site and an electron-rich moiety, which interacts with the heme iron. Extensive analyses of CYP2C9 substrates show that the highest affinity is shown by molecules having two negative regions equally separated from a central aromatic group.114 Homology models of CYP2C9 have indicated that Phe110 and Phe114 may be involved in aromatic interactions,115 while mutagenesis experiments have suggested that Arg97 and Arg108 may constitute the cationic binding sites.116 The key roles of aromatic and electrostatic interactions are also confirmed by classic QSAR analyses, which correlate binding affinities to CYP2C9 with pKa and log P values.111 A CoMFA analysis of a set of hydantoin and barbiturate substrates revealed that electrostatic features seem unimportant, while steric features and lipophilic properties are well correlated with binding data for these enzymes.117 This does not imply that electrostatic features of the hydantoin and barbiturate rings are not important in binding, simply that these features are not significant in describing the differences in binding of these compounds.

CYP2D6 is a polymorphic CYP member well known for its absence in 5-9% of the Caucasian population. 3D-QSAR studies on CYP2D6 substrates showed that these molecules are characterized by a basic group at about 5-7 A from the hydroxylation site, and coplanar aromatic rings.118 These observations are strongly confirmed by homology modeling and mutagenesis experiments, which showed that two acidic residues (Glu216 and Asp301) interact with the basic group in the ligand.119 The key role of a basic group was also confirmed by classic QSAR analyses, which correlated binding affinities to CYP2D6 with pKa values.111 Comparable analyses with CYP2D6 inhibitors underlined the major role of hydrogen bonding (as well as features already described for substrates) in accounting for differences in binding affinity.

The complexity and the very broad specificity of CYP enzymes find their greatest expression in the CYP3A subfamily, especially in CYP3A4. Indeed, many data suggest two or more binding modes due to the presence of at least two binding subpockets and one 'effector-binding' region. A more recent hypothesis proposes that the CYP3A4 can exist in multiple conformations regulated by allosteric effects.121 The CYP3A4 promiscuity has produced a number of theoretical models to predict its enzymatic activity, even if the predictions are often of modest statistical quality. To date, all QSAR analyses indicate the relevance of hydrophobic interactions and at least one hydrogen bond, and the crystal structure of CYP3A4 revealed the main residues involved in these interactions.122 Docking analyses confirmed the key role of multiple hydrophobic domains in the active site of CYP3A4.123

These few examples indicate the increasing interest in modeling human CYPs using 3D-QSAR analyses. Despite their modest ability to predict the enzymatic parameters for heterogeneous molecules, such models can predict qualitatively the ability of molecules to bind to CYPs. As such, they can be useful tools even if they fail to distinguish substrates from inhibitors. Furthermore, the increasing availability of crystal structures or homology models of relevant CYPs should facilitate docking analysis on these enzymatic targets.

To date, 3D-QSAR analyses have accounted for inhibitors and substrates, while the ability of molecules to induce CYPs are just beginning to be evaluated by such methods, due to the lack of sufficiently broad experimental data sets.124

Among the other oxidative enzymes involved in metabolism, monoamine oxidases (MAOs, EC 1.4.3.4), particularly MAO-B, have attracted great interest since their inhibitors play diverse roles in the pharmacological management of some neurological disorders. Recent studies applied CoMFA and GOLPE procedures to a large data set of MAO inhibitors, revealing that hydrophobicity and steric hindrance are the main factors involved in binding. Interestingly, docking studies on the crystal structure of human MAO-B have confirmed the CoMFA results and highlighted the role, essential for MAO-B activity and selectivity, of a hydrophobic cavity (the so-called entrance cavity) connecting the surface of the protein to the catalytic cavity.125 Other studies devoted to specific series of compounds (namely coumarins126 and indolylmethylamine127 derivatives) have led to similar results, and underlined the role of lipophilicity in both the affinity and selectivity of such inhibitors.

5.24.4.2.1.2 Transferases

UDP-glucoronosyltransferases (UGTs) are among the most important conjugating enzymes involved in drug-metabolism. Like CYPs, UGTs exist in multiple enzyme isoforms, differing in substrate specificity.128

The first studies applied 3D-QSAR techniques to inhibitor and substrates of rat and human UDPs.129 These studies highlighted the key role of electrostatic factors in inhibition and metabolism. The results are in agreement with those obtained with classic QSAR analyses, which correlated the rate of glucuronidation with the log P and pKa values of ligands.130 The log P term may reflect diffusion to the active site, and pKa the apparent lipophilicity.

More recent studies have applied the pharmacophore analysis to specific UGT isoforms. In particular, the enzymes in the UGT1A subfamily share common pharmacophore features, with the site of glucuronidation invariably adjacent to a hydrophobic region, and another hydrophobic domain located 6-8 A from the site of conjugation.131 Unfortunately, these pharmacophore models are useless to predict rates of glucuronidation, perhaps due to multiple binding modes. Indeed, a more comprehensive study, including substrates of 12 human UGTs and multiple pharmacophore models, has generated good predictability for all considered isoforms.132

A second class of conjugating enzymes that attracted great interest are the catechol O-methyltransferases (COMTs). These enzymes catalyze the methylation of various catechol derivatives such as catecholamines, catechol estrogens, and their metabolites, and several drug metabolites. There are two forms of COMTcoded by a single gene: the soluble, cytosolic form (s-COMT) and the membrane-bound form (mb-COMT), found in the rough endoplasmic reticulum. COMT inhibitors are used as drugs in the treatment of Parkinson's disease. CoMFA analyses133,134 have revealed the key role played by the electrostatic field in predicting the enzyme kinetic parameters of s-COMT. These studies underline how CoMFA results are sensitive to the method used to calculate atomic charge calculation. Semi-empirical charge calculations performed clearly better than fully empirical ones. The CoMFA results are in agreement with docking analyses of the s-COMT crystal structure, which have revealed the interaction between the catechol oxygen atoms and a Mg2 + ion involved in catalysis (Figure 9).

Catecholamine sulfotransferase (SULT1A3) is also involved in catechol metabolism. An extensive CoMFA study with 95 substrates has revealed the remarkable role that the electronic and steric fields play in enzyme interaction, as also confirmed by docking analysis.135 In particular, CoMFA results revealed the importance of an electrostatic interaction near Glu146 in SULT1A3, plus the fact that bulky substituents pra to the reacting hydroxy group are unfavorable.

Was this article helpful?

0 0
Lower Your Cholesterol In Just 33 Days

Lower Your Cholesterol In Just 33 Days

Discover secrets, myths, truths, lies and strategies for dealing effectively with cholesterol, now and forever! Uncover techniques, remedies and alternative for lowering your cholesterol quickly and significantly in just ONE MONTH! Find insights into the screenings, meanings and numbers involved in lowering cholesterol and the implications, consideration it has for your lifestyle and future!

Get My Free Ebook


Post a comment