Introduction

Top Abstract Introduction Materials and Methods Results Discussion References

The cytochromes P450 (P450) are a superfamily of hemoproteins that are involved in oxidative, peroxidative, and reductive metabolic biotransformation of a wide variety of endogenous compounds, such as steroids, prostaglandins, and fatty acids, and exogenous compounds, such as drugs and carcinogens (Ortiz de Montellano, 1995; Nelson et al., 1996). Individual P450s exhibit unique substrate specificity and regio- and stereoselectivity profiles that reflect different tertiary structures. The CYP2B subfamily contains some of the first P450s to be cloned and purified and has served as a model system for mammalian P450 structure-function analysis through the use of allelic variants, site-directed mutagenesis, and computer-aided molecular modeling (Lewis et al., 1999; Domanski and Halpert, 2001). The CYP2B1 substrate recognition site residues 103, 114, 115, 206, 209, 290, 294, 297, 298, 302, 362, 363, 367, 477, 478, and 480 have been demonstrated to be very important for substrate metabolism (Domanski and Halpert, 2001; Domanski et al., 2001). Human CYP2B6 metabolizes about 3% of drugs in clinical use, and structure-function analyses have only recently begun as new clinically relevant substrates have been identified (Rendic and Di Carlo, 1997) and heterologous expression has been accomplished (Hanna et al., 2000). More work is necessary to understand the structure and specificity of CYP2B6 and its role in metabolism.

Recently efforts have intensified to use the quantitative structure-activity relationship (QSAR) for understanding P450 active sites (Ekins et al., 2001). Such approaches are based on the idea of combining chemical knowledge about small compounds with experimental data obtained from in vitro systems. A considerable number of current QSAR models have been generated for P450s, especially for CYP2B6, 2C9, 2D6, and 3A4 (Koymans et al., 1992; Strobl et al., 1993; Jones et al., 1996; Rao et al., 2000; Ekins et al., 1999a,b,c,d, 2000). The QSAR approaches used include comparative molecular-field analysis, Catalyst, and molecular surface-weighted holistic invariant molecular analysis. Comparative molecular-field analysis is dependent on the alignment of small compounds, whereas Catalyst is alignment independent. In some cases, the pharmacophores were combined with homology models of P450s and provide a more powerful tool to investigate the active-site features and substrate interactions (de Groot et al., 1999a,b; Afzelius et al., 2001). A novel and selective CYP2D6 substrate, which is suitable for high-throughput screening, has already been successfully designed with the help of one of these models (Onderwater et al., 1999).

Two 3D-QSAR models were generated for CYP2B6 substrates (Ekins et al., 1999c). One is a pharmacophore model built by Catalyst and consisting of three hydrophobes and one hydrogen bond acceptor region. The other partial least-squares model was generated using molecular surface-weighted holistic invariant molecular descriptors. Molecular size, positive electrostatic potential, hydrogen bond acceptors, and hydrophobicity were found to be important for CYP2B6 substrate binding. Although both 3D-QSAR models predicted satisfactory K_m values for the majority of the test set molecules, the pharmacophore generated by Catalyst does not overlay the oxidation site of the substrates in the training set. This may be a real limitation for a pharmacophore made for substrates rather than inhibitors.

Only one mammalian P450 (rabbit CYP2C5) has been crystallized to date (Williams et al., 2000). This is a revolution in the study of the structures of these membrane-bound enzymes, but generation of further structures is impeded by the difficulty in obtaining diffraction quality crystals. In the absence of experimental data, homology modeling becomes an important tool to predict the three-dimensional structures of P450 enzymes. Several reviews have recently addressed the homology modeling of P450s (de Groot and Vermeulen, 1997; Szklarz and Halpert, 1997a; Peterson and Graham, 1998). Early models were based on one or more of four bacterial crystal structures (Szklarz and Halpert, 1997b). However, recently models have been built based on CYP2C5 structures, including CYP2B1, 2B4, and 2B5 (Spatzenegger et al., 2001) and CYP2C9 (Afzelius et al., 2001). These homology models provide a structural basis for understanding the mechanism of P450-mediated drug metabolism and drug-drug interactions.

In this study, we used Catalyst to generate pharmacophores for CYP2B6 substrates with the reaction site of each substrate overlaid. In addition, a model of CYP2B6 was constructed by homology modeling based on the crystal structure of CYP2C5. The pharmacophores were combined with the CYP2B6 model by docking the substrates in the active site of CYP2B6. Finally, the pharmacophores were used to predict the K_m value for test set molecules in conjunction with the CYP2B6 model. The combined model holds increased potential to study and predict CYP2B6-mediated drug metabolism.

	Materials and Methods

Top Abstract Introduction Materials and Methods Results Discussion References

Pharmacophore Generation for CYP2B6 Substrates. Pharmacophore generation was done using Catalyst (Molecular Simulations, Inc., San Diego, CA). The structures of CYP2B6 substrates were built interactively using the View Compounds Workbench. The conformational search was performed on each structure using the best conformer generation method in Catalyst. The number of conformers for each substrate was limited to a maximum of 250 with an energy range of 20 kcal/mol. The three-dimensional pharmacophore models were generated based on the K_m values and the conformers of the substrates by selecting the default chemical functions in the Feature Dictionary in Catalyst, including hydrogen bond donor or acceptor, hydrophobic, negative or positive ionizable features, etc. The substrates were then fit to the generated pharmacophore models. The results showed that the sites of oxidation of the substrates were not overlaid when they were fit to the pharmacophore models because there is no default function in the Feature Dictionary that can recognize the reaction sites of the substrates and then overlay them together when a pharmacophore is generated.

View larger version (25K): [in this window] [in a new window]   Fig. 1.   The structures of CYP2B6 substrates in the training set and test set. *, denotes a chiral carbon.

Homology Modeling of CYP2B6. The three-dimensional structure of human CYP2B6 was built based on the crystal structure of CYP2C5 (Williams et al., 2000) using Insight II/Homology and Insight II/Discover-3 modules (Molecular Simulations, Inc.). The sequence of CYP2B6 was obtained from SwissProt (accession number P20813). The sequence alignment of CYP2B6 and 2C5 was done by GCG (Wisconsin Package Version 10.0; Genetics Computer Group, Madison, WI). The crystal structure of CYP2C5 lacks the N-terminal residues 1 to 29 and F-G loop residues 212 to 222. Correspondingly, the CYP2B6 was modeled from residues 31 to 491. The segment in CYP2B6 corresponding to residues 212 to 222 in CYP2C5 was constructed based on the coordinates of CYP2C5 containing one of two alternative models for density corresponding to the F-G loop (E. F. Johnson, unpublished observation). For residues 276 to 278, the only segment not considered to be conserved, the coordinates were generated by searching the PDB databank to find the regions of proteins that meet the geometric criteria of the loop. The coordinates of the other residues were assigned based on CYP2C5. The heme group was copied from CYP2C5 into the CYP2B6 model.

1

1

Docking. Benzyloxyresorufin and 7-ethoxy-4-trifluoromethylcoumarin (7-EFC) were docked automatically in the active site of the CYP2B6 model using the Insight II/Affinity module. The conformer of each docked substrate that fits the pharmacophore the best was exported from Catalyst to be the initial structure for docking.

Pharmacophore Validation by a Test Set of Substrates. A series of substrates in the test set were used to validate the pharmacophore models. A conformational search was also performed on the test set substrates. Each substrate was mapped to the pharmacophore by the Best Fit Compare method in Catalyst. All conformers of the substrate were checked during the mapping process. Therefore, a number of mappings to the pharmacophore were obtained for each substrate. The predicted activity, which indicates how active the molecule is estimated to be and how well the molecule matches to the pharmacophore, was estimated for each mapping of the substrate. The mapping in which the substrate is located in the orientation for metabolism was selected with the help of the CYP2B6 model. In this manner, the final predicted K_m for each substrate was determined.

	Results

Top Abstract Introduction Materials and Methods Results Discussion References

Pharmacophores of CYP2B6 Substrates. Pharmacophore models were generated for a training set of 16 structurally diverse substrates of CYP2B6 (Fig. 1) with K_m values shown in Table 1. The K_m values were obtained from the literature and were generated with CYP2B6 expressed in -lymphoblastoid cells. Eight hypotheses (Table 2) were generated by Catalyst. All the hypotheses consist of two hydrophobes and one hydrogen bond acceptor, which are located in different relative positions in each instance.

View this table: [in this window] [in a new window]   TABLE 1 Training set substrates The Km values were generated using CYP2B6 expressed in -lymphoblastoid cells.

View larger version (107K): [in this window] [in a new window]   Fig. 2.   Pharmacophores A and B for CYP2B6 substrates. The R regions (orange) represent the overlaid reaction sites of the substrates. The cyan regions H1, H2, H3, H4 are hydrophobes. The green regions HBA1 and HBA2 stand for hydrogen bond acceptors with the vectors in the directions of the putative hydrogen bond. H1RH2 = 88.6°; H1RHBA1 = 46.1°; H2RHBA1 = 56.8°; H3RH4 = 19.9°; H3RHBA2 = 43.4°; H4RHBA2 = 36.3°.

View larger version (119K): [in this window] [in a new window]   Fig. 3.   A, mapping of benzyloxyresorufin with pharmacophore A; B, mapping of 7-EFC with pharmacophore B.

Homology Modeling. To determine how well the pharmacophores fit the active site of CYP2B6, a three-dimensional structure of human CYP2B6 was built based on the crystal structure of CYP2C5. The sequence alignment is shown in Fig. 4. The amino acid residues that have been determined previously to be important for substrate metabolism in the CYP2B subfamily (Domanski and Halpert, 2001; Domanski et al., 2001) are shown in Fig. 5.

View larger version (43K): [in this window] [in a new window]   Fig. 4.   Sequence alignment between CYP2B6 and 2C5 from residue 31 to 491.  The GAP method in GCG (Genetics Computer Group) was used for sequence alignment. The sequence identity between CYP2B6 and CYP2C5 is 49%, and the similarity is 61%.

View larger version (69K): [in this window] [in a new window]   Fig. 5.   Active site of a CYP2B6 homology model constructed based on the crystal structure of CYP2C5. The heme group is shown in red. Part of the backbone in the active site is shown as gray ribbons. The substrate recognition site residues determined to be important for substrate metabolism in CYP2B1 are shown in purple.

Combination of the Pharmacophores and the CYP2B6 Model by Docking. The pharmacophores of the substrates should help one to infer complementary structural features of the CYP2B6 active site. To gain confidence in the pharmacophores, we assessed how well the pharmacophores fit the active site by automatic docking of benzyloxyresorufin and 7-EFC into CYP2B6. The docking results are shown in Figs. 6 and 7. The substrates are oriented with the metabolic site pointing to the heme.

View larger version (40K): [in this window] [in a new window]   Fig. 6.   Benzyloxyresorufin docked in the active site of CYP2B6. The heme group is shown in red. Benzyloxyresorufin is shown in orange, and residues within 5 Å are shown in green. The reaction site R and hydrophobes H1 and H2 of pharmacophore A are located according to the orientation and position of benzyloxyresorufin in the active site when compared with the mapping in Fig. 3A. The distance between H1 (the center of the phenyl ring matching H1) and R is 3.9 Å, and the angle H1RH2 equals 90.3°. The values are very similar to the corresponding distance (4.0 Å) and angle (88.6°) in pharmacophore A (Fig. 2A). The distance between R and H2 does not vary.

View larger version (61K): [in this window] [in a new window]   Fig. 7.   7-EFC docked in the active site of CYP2B6. Two possible orientations were found for 7-EFC (orange) due to its small molecular size. The enzyme is oriented in the same direction in the two pictures. The heme group is shown in red, and the residues within 5 Å of 7-EFC are shown in green. The reaction site R and hydrophobes H3 and H4 of pharmacophore B are located in the active site according to the orientations and positions of 7-EFC when compared with the mapping in Fig. 3B. The structure of 7-EFC is rigid, requiring that R, H3, and H4 are located in almost the same relative position as in pharmacophore B.

Predictions of the Pharmacophores in Conjunction with the CYP2B6 Model. The significance of a pharmacophore is determined by its ability to predict new compounds. Pharmacophores A and B were used to predict the K_m values for substrates in the test set (Table 3) in conjunction with the CYP2B6 model. The conformers were calculated for each substrate in the test set. A number of mappings were obtained when each substrate was fit to the pharmacophore by Catalyst because each substrate has a number of conformers. Knowing the orientations and positions of pharmacophores in the active site of CYP2B6 helps to select which mapping is the one most suitable for metabolism by CYP2B6. Table 3 gives the experimental and predicted K_m values for the test set molecules. Of the total substrates in the test set, the K_m values for most were successfully predicted with at least one of the two pharmacophores. It should be noted that the K_m values reported for bupropion and verapamil refer to the racemic mixtures. Since the R- and S-enantiomers may bind differently to the enzyme, K_m values were predicted for both forms. For both enantiomers of bupropion, pharmacophores A and B give satisfactory predicted K_m values, indicating that bupropion may have more than one binding position in the active site of CYP2B6. For verapamil, the predicted K_m values are close to the experimental data except for R-verapamil and pharmacophore B. An exception to the predictive power of the pharmacophores is lidocaine. There are two ethyl groups attached to the N atom (Fig. 1). Although lidocaine fits pharmacophores A and B very well (the predicted K_m is low; Table 3), the two ethyl groups may hinder the access of the N atom to the heme iron. This may explain why the predicted K_m is lower than the experimental K_m (Table 3).

	Discussion

Top Abstract Introduction Materials and Methods Results Discussion References

Two pharmacophores were determined for CYP2B6 substrates in this study, both of which include two hydrophobic regions and one hydrogen bond acceptor. The correlation coefficients of the two pharmacophores are 0.84 and 0.82. Both pharmacophores complement the active site of a CYP2B6 homology model based on CYP2C5 and gave satisfactory estimated K_m values for the majority of the test set molecules in conjunction with the protein model. Pharmacophore A is located in one part of the active site, and pharmacophore B is located in another position, with some overlap between the two pharmacophores, indicating that different possible binding locations exist within the active site of the enzyme. The results are consistent with previous QSAR studies on CYP2B proteins that suggested the importance of hydrophobic and electronic properties in the binding of substrates (Lewis et al., 1999; Ekins et al., 1999c).

Initially, we sought to investigate the active-site features of CYP2B6 by combining our new three-dimensional homology model with the existing substrate pharmacophore of CYP2B6 generated by Catalyst (Ekins et al., 1999c). Following the procedures in that study we recreated the pharmacophore, which gave satisfactory predicted K_m values for the majority of the test set molecules (data not shown). However, we realized that the oxidation sites of the substrates in the training set had not been overlaid when they were mapped to the pharmacophore. Since the site of oxidation of the substrates should point toward the heme iron in the enzyme, a new function was made and saved in the Feature Dictionary in Catalyst to force the reaction site of each substrate to be overlaid when the pharmacophores were generated.

In constructing the new pharmacophores, we made some modifications to the initial training set of 16 compounds. In particular, antipyrine was deleted due to questions about the validity of the very high K_m value (17.7 mM). In addition, benzphetamine and cinnarizine were moved to the test set (Table 1). These three compounds were replaced in the training set by arteether, amitriptyline, and propofol. Five compounds with different metabolic pathways are thus included in the test set (Table 3) in this study to assess pharmacophore A and B. In addition, the R- and S-enantiomers of bupropion and verapamil in the test set were considered. The reported K_m values of the two compounds correspond to the racemic mixtures. However, the R- and S-forms differ in their binding with the CYP2B6 model. Therefore, the K_m values were predicted for both R- and S-enantiomers of the two compounds. Consideration of the stereochemistry had relatively little impact with bupropion or with pharmacophore A and verapamil. However, pharmacophore B predicted a K_m in the millimolar range for R-verapamil. In the article by Ekins et al. (1999c), the R- and S-forms were not distinguished when the K_m values were predicted for bupropion and verapamil.

The pharmacophore model made by Ekins consists of three hydrophobes and one hydrogen bond acceptor region. The pharmacophores generated in this study include two instead of three hydrophobes, one hydrogen bond acceptor, and an additional reaction site. If the mapping of 7-EFC with the pharmacophore in the article by Ekins et al. (1999c) is compared with that (Fig. 3B) in this study, we can see that the phenyl ring to which the ethoxy group attaches matches a hydrophobe in both studies. The trifluoromethyl group attached to position 4 of the other ring also fits a second hydrophobe in both studies. Therefore, the hydrophobicity of the trifluoromethyl group and the phenyl ring may be very important for 7-EFC binding. If the structures of 7-ethoxy-4-trifluoromethylcoumarin, 4-chloromethyl-7-ethoxycoumarin, and 7-ethoxycoumarin (Fig. 1) are compared, only the first two compounds have a group in position 4 of the phenyl ring, and their K_m values (1.7 and 33.7 µM; Table 1) are lower than that of the third compound (115 µM; Table 1). This suggests that 4-substituent contributes to substrate binding in the active site of CYP2B6.

Lewis et al. (1999) constructed a three-dimensional homology model of CYP2B6 based on bacterial CYP102. A series of substrates were docked into the active site of the constructed model. The substrate template was then obtained by superimposing the docked substrates. Although the template yielded some important features for the substrates, the low sequence similarity between CYP2B6 and CYP102 is a major drawback. In our study, we had the advantage of having a closely related mammalian CYP2C5 crystal structure to construct the homology model CYP2B6. The higher sequence identity and similarity between the CYP2C5 and CYP2B6 make the constructed model more reliable.

Several recent studies have focused on the combination of pharmacophores of small compounds and homology models of P450 enzymes. Afzelius et al. (2001) generated a 3D-QSAR model for 29 structurally diverse competitive CYP2C9 inhibitors. The 3D-QSAR model was constructed with the help of the homology model of CYP2C9. The CYP2C9 inhibitors were docked into the active site of the homology model, and then the active conformers of the inhibitors were selected and used in the 3D-QSAR analysis. The 3D-QSAR model was able to predict the entire test set molecules within 0.5 log units of the experimental K_i values. The active site residues of the homology model of CYP2C9 complement the features of the 3D-QSAR model. Furthermore, de Groot et al. (1999a,b) have also published a combined protein and pharmacophore model for CYP2D6. The pharmacophore model was constructed by overlaying 40 substrates using some distance criteria. Then the derived pharmacophore model was oriented into the active site of the CYP2D6 homology model. The pharmacophore model complemented the protein model perfectly, although the two models were derived independently, thereby validating each other. Normally, pharmacophore or 3D-QSAR models are used to study substrate/inhibitor binding site features and design new compounds without knowledge of the target protein structure. However, if pharmacophore or 3D-QSAR models are combined with three-dimensional structures of proteins, the reliability and the predictive power of the model should be enhanced. In this study, the two pharmacophores were also combined with the CYP2B6 model to deduce the orientations and positions of the pharmacophores in the active site of the enzyme. In the future, the pharmacophores, in conjunction with the protein model, may be used as queries to search a database to predict substrates for CYP2B6. For a compound in which the reaction site is known (N- or O-dealkylation), the two pharmacophores, along with knowledge of their orientations and positions in the active site, can be used directly to predict K_m values and evaluate the possibility of metabolism by CYP2B6. For compounds that undergo C-hydroxylation, the reaction sites could be multiple. The homology model can also be used to predict the reaction site by docking the compound into the active site. Then the K_m values can be predicted based on the knowledge of the orientations and positions of the pharmacophores in the active site.