Studies on 16α-Hydroxylation of Steroid Molecules and Regioselective Binding Mode in Homology-Modeled Cytochrome P450-2C11

We investigated the 16α-hydroxylation of steroid molecules and regioselective binding mode in homology-modeled cytochrome P450-2C11 to correlate the biological study with the computational molecular modeling. It revealed that there was a positive relationship between the observed inhibitory potencies and the binding free energies. Docking of steroid molecules into this homology-modeled CYP2C11 indicated that 16α-hydroxylation is favored with steroidal molecules possessing the following components, (1) a bent A-B ring configuration (5β-reduced), (2) C-3 α-hydroxyl group, (3) C-17β-acetyl group, and (4) methyl group at both the C-18 and C-19. These respective steroid components requirements were defined as the inhibitory contribution factor. Overall studies of the male rat CYP2C11 metabolism revealed that the above-mentioned steroid components requirements were essential to induce an effective inhibition of [3H]progesterone 16α-hydroxylation. As far as docking of homology-modeled CYP2C11 against investigated steroids is concerned, they are docked at the active site superimposed with flurbiprofen. It was also found that the distance between heme iron and C16α-H was between 4 to 6 Å and that the related angle was in the range of 180 ± 45°.


Introduction
Cytochrome P450 (P450) constitutes a large superfamily of heme-containing enzymes capable of oxidizing a variety of substrates, both of endogenous (such as steroids) and exogenous (xenobiotics) origins [1][2][3][4][5][6][7]. Although a variety of P450s are able to metabolize a broad range of substrates, the enzymes often exhibit strict regio-and stereoselectivity towards pertinent compounds, such as various steroids [1]. One of the most active and versatile P450 is rat CYP2C11, a microsomal P450 isoform catalyzing more than 90% of steroid 16α-hydroxylations [8][9][10]. It is well-known that several 3-keto-4-ene steroids such as progesterone and testosterone are metabolized in a gender-specific and predominant manner by the adult rat liver microsomes. In the male, these steroids are primarily metabolized into two oxidized (16α-hydroxyl and 6β-hydroxyl) products mainly by the respective, male-specific cytochrome P450 subforms, CYP2C11 and CYP3A2, while they are primarily metabolized into the 5α-reduced products by female predominant 5αreductase [11]. Most of P450 structures reveal that the heme group is buried deep within the protein matrix, indicating that residues outside of the active site may also be required to guide the substrate into the heme pocket by recognizing substrates at the protein surface and/or comprising part of a substrate access channel [12].
In recent years, homology modeling has become an important tool to study the P450 function, especially in conjunction with experimental approaches [13]. A large amount of work has been directed to elucidating the substratebinding sites of various P450s, and the understanding of this field is now becoming increasingly important, mainly 2 International Journal of Medicinal Chemistry using the two powerful techniques, site-directed mutagenesis and computational molecular modeling of the relevant P450s [11,12,[14][15][16]. In homology modeling, a 3-dimensional (3D) model of the protein is constructed based on its amino acid sequence and on the crystal structure of one or more reference proteins. This mainly involves a sequence alignment between the protein and the template(s) [17].
A challenge remains still for the development of a precise 3D-crystal structure of CYP2C11. Therefore, in this study, an investigation was carried out on the docking mode of 71 different steroid molecules against a computationally homology-modeled 3D-structure of CYP2C11, so as to see a correlation of the biologically obtained results with the AutoDock computational results.
The pairwise sequence alignments of the target sequence with that of template was carried out and the sequence identity of templates with the target sequence is shown in Table 1. The amino acid sequence of the aligned protein templates of chain A of P450 2C9-flurbiprofen (1r9o), P450 2C9-warfarin (log5), and P450 2C9 (log2) exhibited the highest percentage of identity with that of CYP2C11 in the range of 83.5%, 75.9%, and 75.9%, respectively. The chain A of the templates of CYP51-estriol (1x8vA), CYP51 (1h5zA), C37L/C151T/C442A (1u13A), CYP51-4phenylimidazole (1e9xA), and CYP51-fluconazole (1ea1A), whose percentages of identity were 23.3%, 23.9%, 23.9%, 23.9%, and 23.9%, respectively, had been rejected due to their too low similarities with the target sequence.
The amino acid sequence of the aligned protein templates of chain A of lr9o.pdb, log5.pdb, and log2.pdb exhibited the highest percentage of identity with that of CYP2C11 in the range of 83.5%, 75.9%, and 75.9%, respectively, as shown in Figure 1 and Table 1. The above two findings strongly support the hypothesis that the key amino acid residues of CYP2C11 are identical, for the most part, to that of CYP2C9. However, this finding must be further verified experimentally. Figure 2 illustrates the ribbon schematic presentations of the homology model of CYP2C11 in sequence alignment with the warfarin-bound CYP2C9 (PDB code, log5) [19], with CYP2C9 (PDB code, log2) [19], and with the flurbiprofen-bound CYP2C9 (PDB code, 1r9o). The details of these sequence views are shown in Figure 1, including both the proposed key amino acid residues and the different and similar residues of the aligned protein sequences.

Hydroxylation of Steroids and Their Docking
Conformation within the Active Site. The ideal conformation of the steroid molecules within their binding site in varieties of P450s was proposed by many investigator [28][29][30][31][32][33]. They reported that the respective substrates for prokaryotic P450s cam, and eryF are positioned in such a way that a substrate is hydroxylated at a distance of 4.5 and 4.8Å from the heme Fe to the hydroxylated atom [34]. These substrates were also oriented in such a way that the hydrogen, which is abstracted during the reaction, be located within 2Å of the oxygen of the oxy-preferryl intermediate [13]. It is also reported that the docked substrates should be located with the distance between their oxidation site (C16) and the heme iron being 6Å and with the C-H-Fe angle at C16 being 180 • [28]. The C-H bond in C-H-Fe sequence should be

F K K S DY FMP F S AGKR I C A GE A LA R T E L F L F F T T I L QN F N L K S L VD V K D I D F K K S KY FMP F S AGKR I CV GE
The first four symbols represent PDB code and the last symbol, A, B, C, or D, represents the amino acid chain involved in sequence alignments.

FLP SWF
Heme and Hec molecules perpendicular to the heme surface. The substrate was usually placed at a position equivalent to that of camphor in the P450cam crystallographic structure, which gives a distance of about 4.2-4.9Å between the oxidation sites and the heme iron. However, molecular dynamics simulations of camphorbound P450cam suggests that the average distance between the carbon atom, at which hydroxylation takes place, and the heme iron is 5.3Å.
Szklarz et al. [8] proposed that for catalysis to occur the following conditions must be met: (1) the distance between the hem iron and the carbon, at which the hydroxylation takes place, must be 5.6-6Å to allow room for the active oxygen, which results in the carbon to active oxygen distance of 3.9-4.2Å, and the hydrogen to oxygen distance of 2.3-3.1Å and (2) the angle between the carbon, the hydrogen, and the heme iron (or active oxygen) should be close to 180 • (180 ± 45 • ) to promote hydrogen bond formation. Therefore, the analysis of our docking study revealed that the results met the above-mentioned requirements for catalysis, (1) and (2) proposed by Szklarz et al. [8]. That is, the binding orientation would place a potential site for C-16αhydroxylation within 5-6Å of the heme iron and the angle between the carbon, the hydrogen and the heme iron (or active oxygen) should be as close as possible to 180 • (180 ± 45 • ).
Analysis of the docking results revealed that there were a considerable number of conformations flexibilities of the docked substrates oriented in order to meet the above-mentioned conditions, and it was noticed that many conformations were docked within the required distance (4-6Å), but not by the required angle (180 ± 45 • ).

The Docking Energy of Binding and the Experimentally Observed Inhibitory Potency.
Inhibitor docking studies revealed that there was a reasonable positive relationship between their observed inhibitory potencies against [3H]PROG16α-hydroxylation and the number of conformations met with the above mentioned condition (Table 2). In this type of a comparative study between biological potency and computational simulation, it is of our primary concern 11β-Hydroxyprogesterone >10 μM  whether the correlation coefficient is positive or not. In order to examine this relationship, the correlation between the AutoDock inhibition constant (Ki) of steroid substrates and the inhibition potency (IC 40 ) against [ 3 H]progesterone 16αhydroxylation of the rat liver microsome was plotted. As shown is Figure 5, the correlation coefficient was positive and it showed that our model and described enzymatic mechanism were valid.

Docking Mode of the Most Potent
Inhibitor, 3α-Hydroxy-5β-Pregnan-20-One (33). The steroid molecule 3α-hydroxy-5β-pregnan-20-one (33), as shown in Table 2, exhibited the highest number of the conformations met with the abovementioned conditions with the lowest binding free energy (ΔGb) of −10.09 kcal/mol, and the minimum inhibition constant (Ki) of 4.03 × 10 −8 , that is, with the highest binding affinity (IC 40 ; = 0.24 × 10 −7 M) within the CYP2C11 binding site pocket. The docked inhibitor 33, as shown in Figure 4, was located within 5.7Å between the C16-carbon atom, where the proposed 16α-hydroxylation takes place, and the heme iron, and the angle between C16-carbon, C16-α-hydrogen, and the heme iron was 150.6 • . The RMSD International Journal of Medicinal Chemistry 7 (distance inÅ, measured between the centeroid of the docked substrate and that of the bound ligand, flurbiprofen) was 1.03Å. Also, the inhibitor 33 showed a bent A-B ring configuration within the binding site pocket, as shown in Figure 3(b).  Figure 3 with planar A-B ring configuration, whereas inhibitor 34 with 5β-reduced A-B ring exhibited bent A-B ring configuration within the binding site.

Conclusion
Computer simulated automated docking studies were performed using AutoDock 3.05. Docking results revealed that there was a variety of conformations of the docked inhibitors meeting the confirmation of the reported orientation requirements of steroids within their binding sites [7,13,21]. The docked inhibitors were shown to be positioned so that the site of hydroxylation (C16-carbon) resides within 5-6Å from the heme iron, which is consistent with the distances seen in the case of other P450 substrate complex, with the angle between C16-carbon, C16α-hydrogen, and the heme iron being 180 ± 45.0 • . It was noticed that steroids were docked exactly overlapped with the flurbiprofen, as their average RMSD was 1.98Å. Also a positive correlation was obtained between the observed inhibitory potencies against [ 3 H]PROG 16α-hydroxylation and the binding free energies of the docked steroids. The correlation between the observed inhibitory potencies and AutoDock inhibition constants (ki), exhibited also a positive correlation coefficient. Steroid molecule 33 exhibited the lowest binding free energy, that is, the highest affinity within the binding site of CYP2C11, and with the highest number of conformations meeting the reported requirements. This agrees well with the biologically observed results; its observed inhibitory potency index against [ 3 H]PROG 16α-hydroxylation was 31.46 (IC 40 ;: 3α-hydroxy-5β-pregnan-20-one 33, 0.24 × 10 −7 M, vs. progesterone 1, 7.55 × 10 −7 M).
As a whole, the results of the present docking investigation revealed that many amino acid residues responsible for binding of the flurbiprofen-bound CYP2C9 (1r9o), were also essential for the interaction between CYP2C11 and inhibitors. Moreover, docking of steroid molecules within the 3-D homology model of CYP2C11 based on that of warfarinbound CYP2C9 (log5), CYP2C9 (log2), and flurbiprofenbound CYP2C9 (log5), were in a fair agreement with the observed biological data.

Preparation of Adult Male Rat Liver Microsomes.
Approximately 95-day-old male Wistar rats, castrated on the 70th day after birth, were used. The liver microsomes were prepared as previously described [29,31]. The experiments were performed according to instrumental guidelines for the care and use of laboratory animals.

[3H]PROG Metabolism by Rat Liver Microsomes-Inhibitory Effects of Various Unlabeled Steroids.
The metabolism by rat liver microsomes were examined, according to our previously described procedure [23][24][25]. Briefly, the microsomal suspension (400-600 μg of protein/2.2 mL, total volume of the reaction mixture) was preincubated with [3H]PROG (20 nM) under the absence or presence of an unlabeled steroid (0.01-10 μM) at 36 • C for 30 min. Then NADH (3.16 μM) was added, and the reaction mixture was incubated further for 5 min. After the incubation, two identical samples were mixed and extracted with toluene. The toluene-extractable [3H]PROG metabolites (more than 90%) were isolated by various paper chromatographic systems and then identified by recrysallization method [26]. Other miscellaneous procedures are described in our previous papers [29,31].

Protein Homology Modeling.
Since the crystal structure of CYP2C11 is not available, the three dimensional (3D) model of CYP2C11 used in the present simulation was constructed based on a homology modeling method. The homology modeling procedure and the sequence alignment were performed with the cooperation of Swiss-Model (Swiss-Model version 36.0003) [17,18] [33], which was utilized for the study of binding mode of inhibitors within CYP2C11. This program addresses automatically the flexible docking of the ligands into a known protein structure. In contract, flexibility of the target protein is not taken into account. AutoDock 3.05 scans the active site for low energy binding models and for suitable orientations of the probe molecule, using a modified genetic algorism that employs a local search (GALS) and precomputed grids for the evaluation of the interaction energy. The target homologymodeled protein CYP2C11 was separated alone by using DS modeling 1.1 software (DS modeling 1.1; Accelrys inc., San Diego, CA (2003)) and representative amino acids of the ligand-binding site were selected within 5Å neighborhood surrounding the embedded ligand, flurbiprofen. A 120Å 120Å 120Å grid size (x, y, z) with a spacing of 0.300Å centered at-18.44, 86.67, and 30.89Å that encompassed the active site where the ligand, flurbiprofen, was embedded, was used to guide the docked inhibitors. The results of 250 randomly seeded runs were analyzed for each of the docked inhibitors. The docked inhibitors were assigned to a cluster if the atomic coordinates of the docked inhibitors exhibited a root-mean-square deviation (RMSD) of less than 0.5Å difference from each other (RMSD-tolerance of 0.5Å). The clusters were ranked from the averaged lowest energy obtained for members of the cluster to the highest. The analysis was carried out for the top 10 docking clusters. Each of the clusters that exhibited significant negative interaction energies was examined by DS modeling program to determine their binding orientations.

Evaluation of Docked
Results. DS modeling 1.7 was utilized for the molecular modeling and the evaluation of H-bonds in ligand-receptor interaction and for the measurement of RMSD, which was computed and expressed in angstrom (Å) as a locational comparison of two relevant molecules of interest. In the actual sense, it was measured as a distance between the centroid of the docked inhibitor and the bound-ligand, flurbiprofen (FLP).