QUICK SEARCH:   [advanced] Author: Keyword(s): Year:  Vol:  Page:

This Article Abstract Full Text (PDF) Data Supplement Erratum All Versions of this Article: dmd.106.013888v1 35/3/493    most recent Submit a response Alert me when this article is cited Alert me when eLetters are posted Alert me if a correction is posted Services Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Ekins, S. Articles by Harwood, H. J. Search for Related Content PubMed PubMed Citation Articles by Ekins, S. Articles by Harwood, H. J., Jr.

Three-Dimensional Quantitative Structure-Activity Relationship Analysis of Human CYP51 Inhibitors

Pfizer Global Research and Development, Groton Laboratories, Groton, Connecticut

(Received November 13, 2006; accepted December 26, 2006)

	Abstract

Top Abstract Materials and Methods Results Discussion References

 
CYP51 fulfills an essential requirement for all cells, by catalyzing three sequential mono-oxidations within the cholesterol biosynthesis cascade. Inhibition of fungal CYP51 is used as a therapy for treating fungal infections, whereas inhibition of human CYP51 has been considered as a pharmacological approach to treat dyslipidemia and some forms of cancer. To predict the interaction of inhibitors with the active site of human CYP51, a three-dimensional quantitative structure-activity relationship model was constructed. This pharmacophore model of the common structural features of CYP51 inhibitors was built using the program Catalyst from multiple inhibitors (n = 26) of recombinant human CYP51-mediated lanosterol 14-demethylation. The pharmacophore, which consisted of one hydrophobe, one hydrogen bond acceptor, and two ring aromatic features, demonstrated a high correlation between observed and predicted IC₅₀ values (r = 0.92). Validation of this pharmacophore was performed by predicting the IC₅₀ of a test set of commercially available (n = 19) and CP-320626-related (n = 48) CYP51 inhibitors. Using predictions below 10 µM as a cutoff indicative of active inhibitors, 16 of 19 commercially available inhibitors (84%) and 38 of 48 CP-320626-related inhibitors (79.2%) were predicted correctly. To better understand how inhibitors fit into the enzyme, potent CYP51 inhibitors were used to build a Cerius² receptor surface model representing the volume of the active site. This study has demonstrated the potential for ligand-based computational pharmacophore modeling of human CYP51 and enables a high-throughput screening system for drug discovery and data base mining.

CYP51 is one of the most conserved P450s across phyla (Aoyama et al., 1996; Yoshida et al., 1997), with a 93% amino acid sequence identity between rat and human and 39 to 42% identity between mammalian and fungal enzymes (Aoyama et al., 1996; Yoshida et al., 1997). In humans there are three CYP51 genes, including two pseudogenes and a functional gene on chromosome 7 (Rozman et al., 1996). Substrates for CYP51 from mammals, plants, and fungi are lanosterol, obtusifoliol, and 24-methylene-dihydroxylanosterol, respectively (Lamb et al., 1998). Such substrate specificities can be explained by a limited number of amino acid substitutions in substrate recognition sites 1, 2, and 5 between mammalian and fungal CYP51, and it is likely that these locations are responsible for conferring selectivity for sterol metabolism (Yoshida et al., 1997). Before the sequences of fungal and human CYP51 were known, this enzyme was identified as an antifungal therapeutic target that could be inhibited by many imidazole, triazole (Vanden Bossche et al., 1987, 1995; Hartman and Sanglard, 1997; Kelly et al., 1997), and nonazole compounds (Aoyama et al., 1983, 1987; Yoshida and Aoyama, 1985; Gebhardt et al., 1994; Hartman and Sanglard, 1997). Many of these marketed compounds are specific toward a single fungal species, whereas others have widespread application (Hartman and Sanglard, 1997). The selectivity of these antifungals is also variable, with some exhibiting no interaction with the mammalian enzyme. Many of these azole antifungals are also potent inhibitors of other mammalian P450s involved in drug metabolism and there is thus the potential for clinically significant drug-drug interactions when they are coadministered with substrates for the same enzymes (Strolin Benedetti and Bani, 1999).

There have been many studies using azole, oxysterol, and mechanism-based inactivator-type inhibitors of fungal or mammalian CYP51 (Trzaskos et al., 1986, 1993, 1995; Frye et al., 1993, 1994). In addition, the purification (Sonoda et al., 1993) and expression of human CYP51 have allowed the determination of its regulation by oxysterols (Stromstedt et al., 1996). However, because of the membrane-bound nature of mammalian P450s, there is as yet no crystal structure for human CYP51. Therefore, our understanding of the structural requirements of the human CYP51 active site are limited to homology models constructed using various bacterial P450s (Ishida et al., 1988; Tuck et al., 1991, 1992; Tafi et al., 1996; Talele et al., 1997; Holtje and Fattorusso, 1998; Lewis et al., 1999), including Mycobacterium tuberculosis CYP51 (Matsuura et al., 2005; Rupp et al., 2005) and the crystal structure of M. tuberculosis CYP51 (Podust et al., 2001).

There have also been few quantitative structure-activity relationship (QSAR) models of CYP51 to date and they have dealt exclusively with the fungal CYP51s (Tafi et al., 1996; Fujita, 1997). The utility of more classic QSAR models has also been reviewed for the production of azole-type agricultural fungicides (Fujita, 1997). However, approaches to modeling the common features of inhibitors of other human P450s have been widely shown using a Catalyst pharmacophore approach (Ekins et al., 2001), comparative molecular field analysis (Jones et al., 1996), and other QSAR methods (Ekins et al., 2003). These techniques have also been used to predict a test set of molecules excluded from the training set (Ekins et al., 1999).

Recently a series of dual-acting hypoglycemic and hypocholesterolemic agents were found to inhibit glycogen phosphorylase and human CYP51 (Harwood et al., 2005). The prototype of this series, CP-320626, inhibited glycogen phosphorylase and human CYP51 with IC₅₀ values of 155 nM and 4 µM, respectively, reduced glycogenolysis and cholesterologenesis in cultured cells, and lowered plasma glucose and plasma cholesterol in experimental animals (Harwood et al., 2005). To understand the characteristics of these CP-320626-related CYP51 inhibitors and other inhibitors of this human CYP51 enzyme, we have used a computational approach to produce a predictive Catalyst pharmacophore. This model was constructed using a training set of structurally diverse commercially available molecules and CP-320626-related CYP51 inhibitors with varying IC₅₀ values. The validity of the predictions made by this model was then evaluated using test sets of molecules not in the training set. These test sets contained both commercially available molecules and structurally diverse CP-320626-related CYP51 inhibitors for which in vitro data were subsequently generated.

	Materials and Methods

Top Abstract Materials and Methods Results Discussion References

 
Materials. Human cytochrome P450 reductase was obtained from Panvera (now Invitrogen, Madison, WI). Rat microsomal lipid was prepared as described by Sundin et al. (1987). Triarimol, CP-320626, and all other related CYP51 inhibitors denoted as A-P and 1–48 (see supplemental Table for structures) in the tables were synthesized as described previously (Hoover et al., 1998; Martin et al., 1998; Rath et al., 2000; Treadway et al., 2001; Harwood et al., 2005; Yu et al., 2006) and in the patents cited in these publications. Lovastatin was purchased from Calbiochem (La Jolla, CA). Azaconazole, itraconazole, and terconazole were obtained from Research Diagnostics (Flanders, NJ). Sulfaphenazole was purchased from Gentest Corp. (now BD Gentest, Woburn, MA). Dioleoyl phosphatidyl choline, lanosterol (97% pure), ergosterol, and all other chemicals were obtained from Sigma-Aldrich (St. Louis, MO) or Steraloids (Newport, RI).

CYP51 Lanosterol Demethylase Assay. Human CYP51 expressed in Topp 3 cells was partially purified using the method of Stromstedt et al. (1996). Protein concentrations were measured using a BCA Protein Assay kit (Pierce Chemical, Rockford, I) with bovine serum albumin as a standard as described by the manufacturer. Cytochrome P450 content was measured by the method of Omura and Sato (1964). Initial rate conditions for the formation of 4,4-dimethylcholesta-8,14,24-trien-3ß-ol (triene) were determined in preliminary studies with the CYP51 reconstitution system. In brief, 25 µl of lanosterol (1 mM, suspended in a mixture of 50:50 tyloxapol-acetone, dioleoyl phosphatidyl choline micelles (5 mg/ml), methanol, and 100 mM potassium phosphate buffer, pH 7.4) was added to tubes to provide a final lanosterol concentration of 50 µM (approximate K_m). Inhibitors (ranging from 0.01 to 200 µMin methanol) or methanol (control) were then added to the tubes, and the lanosterol-inhibitor mixture was allowed to dry under nitrogen for 10 min. To dried substrate and inhibitor tubes, 20 pmol of CYP51, 125 pmol of human reductase, and 50 µl of rat lipid were added and allowed to sit at room temperature for 10 to 15 min for enzyme reconstitution. Buffer (100 mM potassium phosphate, pH 7.4, with 20% glycerol, 0.1 mM dithiothreitol, 0.1 mM EDTA, and 0.5 mM potassium cyanide) was added to the tubes after the reconstitution period for a final incubation volume of 0.5 ml. Tubes were preincubated for 2 min at 37°C, and the reaction was initiated by the addition of 50 µl of a NADPH regenerating system (final incubation concentration; 10 mM MgCl₂, 0.54 mM NADP, 6.2 mM DL-isocitric acid, and 0.5 U/ml isocitric dehydrogenase). Reactions were terminated at 60 min with the addition of 25 µl of ergosterol (1 mg/ml in ethyl acetate), followed by extraction with 5 ml of ethyl acetate. Tubes were then vortexed before centrifugation followed by removal of 3 to 4 ml of supernatant and transfer to fresh tubes and evaporation under nitrogen at 50°C. The evaporated samples were reconstituted in a 150-µl mobile phase, and 25 to 50 µl was injected onto the high-performance liquid chromatography system (see below).

Triene formation was determined using a modification of the method of Strömstedt et al. (1996). In brief, triene metabolite and ergosterol were separated on a Waters Novapak C₁₈ column (4.0 µM; 150 mm x 3.9 mm; Waters, Bedford, MA) with a mobile phase consisting of methanol-acetonitrile-high-performance liquid chromatography grade water (45:45:10), at a flow rate of 1.5 ml/min. The triene and ergosterol were monitored by UV detection at 248 nm using a SpectroMonitor variable wavelength detector (LDC Analytical, Riviera Beach, FL) and the Multichrom data acquisition system (version. 2.11; Fisons Instruments, Beverly, MA) was used for data collection, analysis, and reporting. Approximate retention times for triene metabolite and ergosterol were 25 and 30 min, respectively.

Interpretation of CYP51 Lanosterol Demethylase Assay Results. The percent inhibition of triene formation with methanol controls was determined with each inhibitor at three to six concentrations in triplicate and the average was used as described previously (Harwood et al., 2005). The IC₅₀ was calculated from a graph of percent CYP51 activity remaining versus inhibitor concentration using DeltaGraph (version 4.0; Monterey, CA).

Molecular Modeling. The computational molecular modeling studies were carried out using a Silicon Graphics Octane workstation and based on a pharmacophore methodology described previously for other P450s (Ekins et al., 1999, 2001). The pharmacophore method described below represents the features on the ligands used and cannot be related easily to features in the protein binding site. This method has also been applied to other proteins of relevance to absorption, distribution, metabolism, excretion, and toxicity (Ekins and Swaan, 2004). The receptor surface modeling method also described below attempts to provide information relating to the volume requirements of the enzyme active site for the heme and nonheme binding compounds with highest affinity.

Modeling with Catalyst. The three-dimensional structures of CYP51 inhibitors from the present study or literature data sets were built interactively using Catalyst (version 4.0; Accelrys Software, Inc., San Diego, CA). The number of conformers generated for each inhibitor was limited to a maximum of 255 with an energy range of 0 to 20 kcal/mol. A training set of 26 molecules was selected and used to build the pharmacophore (Table 1). Ten hypotheses were generated using these conformers for each of the molecules, and IC₅₀ values were generated after selection of the following features for the inhibitors: hydrogen bond donor, hydrogen bond acceptor, hydrophobic, positive ionizable, and ring aromatic. After assessment of all 10 hypotheses generated, the lowest energy cost hypothesis was considered the best.

The goodness of the structure activity correlation was estimated by means of the correlation coefficient r value. Catalyst also calculates the total energy cost of the generated pharmacophores from the deviation between the estimated activity and the observed activity, combined with the complexity of the hypothesis (i.e., the number of pharmacophore features). A null hypothesis, which presumes that there is no relationship in the data and that experimental activities are normally distributed about their mean, is additionally calculated. Hence, the greater the difference between the energy cost of the generated hypothesis and the energy cost of the null hypothesis, the less likely it is that the hypothesis reflects a chance correlation. This criterion is then used as an assessment of the pharmacophore model selected.

Catalyst CYP51 Pharmacophore Validation Using Test Sets of IC₅₀ Values. Two test sets of 19 commercially available molecules and 48 proprietary compounds, respectively, were subjected to the fast-fit algorithm for the Catalyst hypothesis to predict an IC₅₀ value. None of these molecules was included in the initial training set for the CYP51 IC₅₀ pharmacophore. The fast-fit algorithm refers to the method of finding the optimal fit of the substrate to the hypothesis among all the conformers of the molecule without performing an energy minimization on the conformers of the molecule (Catalyst tutorials release 3.0; Accelrys Software, Inc.). The computational predictions derived using the fit to the CYP51 Catalyst pharmacophore were compared with the observed in vitro values. We selected an acceptable arbitrary cutoff for potency of inhibition as accepted of 10 µM.

Catalyst CYP51 Pharmacophore Validation Using Permuting of Activity Data. The statistical significance of the pharmacophore hypothesis used was tested by permuting (randomizing) the structures and the activities and then repeating the Catalyst hypothesis generation procedure 10 times.

Modeling with Cerius². The three-dimensional structures of CYP51 inhibitors generated using Catalyst as described previously were imported into Cerius² (Accelrys Software, Inc.). Inhibitors with an IC₅₀ 10 µM were selected, and a receptor surface model (Hahn and Rogers, 1995) was generated. Additional molecules were then fitted to this model to discover which features were outside of the volume occupied by the active molecules.

	Results

Top Abstract Materials and Methods Results Discussion References

 
CYP51 in Vitro Inhibition Data. Partially purified CYP51 was characterized for protein and carbon monoxide difference spectra (data not shown). Total P450 was determined as 0.22 nmol/mg of protein. The assay conditions were also developed to provide optimal lanosterol turnover using a final ratio of P450/reductase of 1:6. The K_m determined for lanosterol with the reconstituted human CYP51 was 30 µM (data not shown), in agreement with previously determined values (Lamb et al., 1998). Using a standard concentration of 50 µM lanosterol, 93 commercially available molecules including the competitive inhibitor CP-320626 (Harwood et al., 2005) and related CYP51 inhibitors were screened in the IC₅₀ assay (Tables 1, 2, 3).

Catalyst CYP51 Pharmacophore Model. Catalyst uses a collection of molecules with inhibitory activity over multiple orders of magnitude for the enzyme of interest to construct a model of the structural features (pharmacophore) necessary for interaction of the molecules with the active site of the enzyme. The resultant pharmacophore hypotheses explain the variability of the potency of inhibition with respect to the geometric localization of these features of the molecules. The IC₅₀ values for structurally diverse inhibitors of CYP51 generated in this study cover >3 orders of magnitude (0.13–>200 µM) (Table 1). These values were used to build a pharmacophore, with the lowest energy model yielding four structural features (Fig. 1) necessary for inhibitor binding to the active site of CYP51: one hydrophobe, one hydrogen bond acceptor, and two ring aromatic features (additional hydrophobes). The Catalyst pharmacophore demonstrated a good correlation of observed versus estimated IC₅₀ values (r = 0.92) (Fig. 2).

View larger version (60K): [in this window] [in a new window]   FIG. 1. The human CYP51 pharmacophore for n = 26 inhibitors generated using Catalyst (see Materials and Methods) illustrating one hydrophobe (cyan), one hydrogen bond acceptor (green), and two ring aromatic features (orange). The interbond angles and the distances between features are also annotated.

View larger version (16K): [in this window] [in a new window]   FIG. 2. Correlation of observed and estimated inhibition IC50 values (log units) for the CYP51 Catalyst model. The central line corresponds to the regression of the data, the dashed lines represent the 95% confidence interval for the regression, and the outer lines are the 95% confidence interval for the population.

Catalyst CYP51 Pharmacophore Validation Using Test Sets of IC₅₀ Values. After constructing the Catalyst 3D-QSAR model for a training set of 26 IC₅₀ values generated by this laboratory in vitro (Table 1), the model was used to predict IC₅₀ values of a test set of 19 commercially available molecules for which in vitro IC₅₀ values were also generated (Table 2) and not included in the initial Catalyst model. In 16 cases, the IC₅₀ values were correctly predicted to be either active (IC₅₀ <10 µM) or inactive molecules (IC₅₀ >10 µM). The poorly predicted molecules were made up of 5.3% false negatives (n = 1, bifonazole) and 10.5% false positives (n = 2, triarimol and azaconazole).

A second test set of IC₅₀ values was composed of 48 CP-320626-related CYP51 inhibitors of which 38 were correctly predicted using the Catalyst 3D-QSAR model following the criteria as described above (Table 3). The poorly predicted molecules were made up of 4.2% false negatives (n = 2, molecules 1 and 37) and 16.6% false positives (n = 8, molecules 7, 12, 38, 39, 44, 45, 46, and 47).

Modeling with Cerius². Fifteen molecules with in vitro IC₅₀ values <10 µM produced a volume using the receptor surface model software present in Cerius² (Fig. 3A). When fluconazole, an inhibitor with a low affinity for human CYP51 was fitted to this model, part of the molecule lies outside of the volume (Fig. 3B).

View larger version (32K): [in this window] [in a new window]   FIG. 3. A, opaque Cerius2 receptor surface model of active CYP51 inhibitors showing hydrogen bond acceptor regions (blue coloring) (front view). This model was generated with training set molecules with IC50 values 10 µM (see Materials and Methods). B. fluconazole fitted to Cerius2 receptor surface model (side view) for human CYP51, showing an area of the molecule outside of the surface boundary for this volume (see Materials and Methods).

	Discussion

Top Abstract Materials and Methods Results Discussion References

Because there is no crystal structure for the human membrane-bound CYP51, alternative techniques must be considered to understand whether a novel molecule will bind to it. It is possible, however, to begin to understand the specificity of an enzyme without possessing information regarding its crystal structure. The most widely used computational method for modeling in the cytochrome P450 field is homology modeling based on crystallized bacterial P450s. The initial models of the active site for CYP51 from Saccharomyces cerevisiae and an inactive mutant (Ishida et al., 1988) were constructed using alignment with P450cam. By using these, novel substrate analogs were developed as competitive inhibitors for this enzyme (Tuck et al., 1991). The formation of a fungal CYP51 biphenyl-iron complex also yielded information as to the approximate height of the cavity (10 Å) above the porphyrin ring in the active site (Tuck et al., 1992). Further work studying the I helix of the fungal CYP51 with lanosterol has shown that Gly-310 is an important residue, although interactions at His-317 and Met-313 may also occur (Vanden Bossche et al., 1995). It has been suggested that whereas ketoconazole and itraconazole have multiple interactions with the CYP51 I helix, fluconazole has fewer and may result in its lower antifungal activity (Vanden Bossche and Koymans, 1998). Comparative homology models have identified differences in the active sites of Candida albicans and S. cerevisiae (Boscott and Grant, 1994), in agreement with earlier work (Yoshida and Aoyama, 1987). When the human CYP51 I helix sequence is aligned with those from C. albicans and S. cerevisiae (Vanden Bossche and Koymans, 1998), there are six amino acids that are different between the fungal and the human P450s. It is therefore likely that the surface volume of inhibitors for fungi and human CYP51 could be markedly different due to the alteration in the active site caused by these amino acid substitutions. Numerous studies have used homology modeling and docking for CYP51 to show lanosterol hydrogen bonding interactions (Holtje and Fattorusso, 1998), triazole and azole inhibitors interacting with the hydrophobic active site and the heme iron (Talele et al., 1997), and functional group modifications that alter enzyme-inhibitor interactions (Tsukuda et al., 1998). CYP51 homology models have been used in de novo design (Ji et al., 2000) and in understanding azole binding to human and crystallized M. tuberculosis CYP51 enzymes (Matsuura et al., 2005; Rupp et al., 2005). The latter study indicated that the human enzyme may have a larger and differently shaped active site cavity due to the replacement of Phe-255 with Leu-310 (Matsuura et al., 2005; Rupp et al., 2005).

Although there are data sets in the literature describing the fungicidal activity of CYP51 inhibitors (Stehmann and de Waard, 1995), there is little information relating to human disease. The present study has described, to our knowledge, the first 3D-QSAR relationship using in vitro data for inhibitors of human CYP51. By using 26 structurally diverse molecules with a range of inhibitory activity, we constructed a predictive pharmacophore model that was used to score the activity of test molecules and predict whether new chemical entities might interact with this enzyme. The model for the data set generated in this study suggested a pharmacophore with four features, indicating the predominance of hydrophobic features (Fig. 1) in agreement with previous reports (Matsuura et al., 2005; Rupp et al., 2005) and a high correlation (r = 0.92) (Fig. 2) between predicted and observed IC₅₀ values.

One of the best ways to test the validity of a 3D-QSAR is to predict the activity of molecules excluded from the training set and then compare these to observed values. In addition to prospectively assessing many proprietary compounds, several commercially available molecules were evaluated in vitro as they had either important endogenous roles (cholesterol, arachidonic acid, progesterone, pregnenolone, and anandamide), were known P450 substrates (erythromycin, tolbutamide, lovastatin, chlorpropamide, and propafenone), or were known P450 inhibitors (sulfaphenazole, fluconazole, azaconazole, econazole, bifonazole, terconazole, itraconazole, triarimol, and sulfinpyrazone). It was thought that these molecules would provide a structurally diverse test set to validate the pharmacophore constructed from a similarly structurally diverse training set including heme-binding azoles and non-heme-binding CP-320626-related compounds (Harwood et al., 2005). In vitro IC₅₀ values were generated for CYP51 inhibition for 19 commercially available molecules. Sixteen of the 19 (84%) molecules fulfilled the criterion (Table 2). Non-heme-binding CP-320626-related CYP51 inhibitors were also successfully predicted using the same pharmacophore model with an 80% prediction rate (Table 3). These results suggest that the CYP51 pharmacophore has utility for both classification of novel molecules that are either known heme binders or nonheme binders with an acceptable level of accuracy for early screening.

By aligning the most potent CYP51 inhibitors (azoles and CP-320626-related compounds) with the pharmacophore, it was possible to generate a receptor surface model of their volume. The receptor surface model of these active inhibitors forms an elongated shape (Fig. 3A), not too dissimilar from that of lanosterol and the volume generated by Rupp et al. (2005). This receptor surface model can also be used to explain the apparent lack of inhibitory activity toward human CYP51 by inactive molecules, by first mapping them to the pharmacophore and then placing them inside this receptor surface model. Features outside of the volume could suggest poor steric interactions within the active site of the enzyme that would be unfavorable for binding and therefore inhibition. As an example, we have fitted fluconazole to the pharmacophore and shown that a triazole ring is outside of the surface volume generated by the active compounds (Fig. 3B). This may explain the lack of potency of fluconazole toward the human CYP51 but subsequent activity toward the fungal P450. Indeed this is in agreement with a recent study that has also shown that fluconazole is considerably less active than ketoconazole with human CYP51 with K_d values of 120 and 8 µM, respectively. The differential potency between these two compounds may also result in part because fluconazole possesses a hydrophilic triazole and hydroxyl groups that would decrease affinity for the hydrophobic binding site relative to the more hydrophobic ketoconazole (Matsuura et al., 2005).

The computational techniques described and applied in this article may allow elucidation of areas within the human CYP51 active site that can be targeted by directed synthesis to enable production of more selective antifungal agents. Likewise, if human CYP51 is pursued as a therapeutic target, for example, in cancer (Downie et al., 2005; Kumarakulasingham et al., 2005), this pharmacophore model enables mining of compound data bases for inhibitors and may help in understanding the affinity of inhibitors for the human CYP51 active site. Specific inhibitors of human CYP51 (Harwood et al., 2005) might also be applicable, for example, as agents for lipid lowering, either alone or in combination with other agents (e.g., 3-hydroxy-3-methylglutaryl-CoA reductase inhibitors) currently marketed and/or in development (Harwood and Hamanaka, 1998) for use as therapeutic interventions in areas with substantial medical needs such as coronary heart disease, obesity, and diabetes.

Although the pharmacophore technique could represent an approach for identifying modulators of human CYP51, possibly by data base screening, we have taken the same approach with in vitro data for antifungal activity (Tafi et al., 1996), to create a second pharmacophore for fungal CYP51 (data not shown). Therefore, it is possible to differentiate between the human and fungal CYP51 enzymes in silico to develop selective modulators of this enzyme across phyla. The publication of X-ray structures of CYP51 in M. tuberculosis with bound inhibitors (Podust et al., 2001) suggests that we will require selective inhibitors if we are to capitalize on CYP51 as a therapeutic target; the human CYP51 pharmacophores may allow this in combination with the protein models (de Groot and Ekins, 2002).

In conclusion, this study has described a large number of molecules tested as potential CYP51 inhibitors and has enabled the identification of nonazoles with low micromolar potency. These molecules have been used to build and test the first 3D-QSAR for human CYP51, which can be used to predict the inhibitory potential of other molecules toward this enzyme. We suggest that this CYP51 pharmacophore may be used to garner information regarding the interactions with the active site of this P450 and visually describe the selectivity of antifungal agents. This model could also enable rapid discovery of further modulators of this enzyme for therapeutic use.

	Acknowledgments

 
We gratefully acknowledge Professor Michael R. Waterman (Department of Biochemistry, Vanderbilt University School of Medicine, Nashville, TN) for supplying clones expressing recombinant CYP51, Joanne Jeffries-Griffor (Molecular Sciences, Pfizer Global Research and Development, Groton, CT) for partial purification of CYP51 from these clones, and Dr. Patrick Dorr (Pfizer, Sandwich, UK) for supplying several of the compounds evaluated in these studies.

	Footnotes

 
Article, publication date, and citation information can be found at http://dmd.aspetjournals.org.

doi:10.1124/dmd.106.013888.

ABBREVIATIONS: P450, cytochrome P450; QSAR, quantitative structure-activity relationship; 3D, three-dimensional.

The online version of this article (available at http://dmd.aspetjournals.org) contains supplemental material.

¹ Current affiliation: ACT LLC, Jenkintown, Pennsylvania.

² Current affiliation: Department of Pharmacology, University of Connecticut Health Center, Farmington, Connecticut.

Address correspondence to: Dr. Sean Ekins, Vice President, Computational Biology, ACT LLC, 601 Runnymede Ave., Jenkintown, PA 19046. E-mail ekinssean{at}yahoo.com

	References

Top Abstract Materials and Methods Results Discussion References

 


Aoyama Y, Noshiro M, Gotoh O, Imaoka S, Funae Y, Kurosawa N, Horiuchi T, and Yoshida Y (1996) Sterol 14-demethylase P450 (P45014DM) is one of the most ancient and conserved P450 species. J Biochem (Tokyo) 119: 926–933.[Abstract/Free Full Text]

Aoyama Y, Yoshida Y, Hata S, Nishino T, and Katsuki H (1983) Buthiobate: a potent inhibitor for yeast cytochrome P-450 catalyzing 14-demethylation of lanosterol. Biochem Biophys Res Commun 115: 642–647.[CrossRef][Medline]

Aoyama Y, Yoshida Y, Sonoda Y, and Sato Y (1987) 7-Oxo-24,25-dihydrolanosterol: a novel lanosterol 14-demethylase (P-450(14DM)) inhibitor which blocks electron transfer to the oxyferro intermediate. Biochim Biophys Acta 922: 270–277.[Medline]

Boscott PE and Grant GH (1994) Modeling cytochrome P450 14 demethylase (Candida albicans) from P450cam. J Mol Graph 12: 185–192, 195.[CrossRef][Medline]

Chan JR, Phillips LJ 2nd, and Glaser M (1998) Glucocorticoids and progestins signal the initiation and enhance the rate of myelin formation. Proc Natl Acad Sci USA 95: 10459–10464.[Abstract/Free Full Text]

de Groot MJ and Ekins S (2002) Pharmacophore modeling of cytochromes P450. Adv Drug Del Rev 54: 367–383.[CrossRef][Medline]

Downie D, McFadyen MC, Rooney PH, Cruickshank ME, Parkin DE, Miller ID, Telfer C, Melvin WT, and Murray GI (2005) Profiling cytochrome P450 expression in ovarian cancer: identification of prognostic markers. Clin Cancer Res 11: 7369–7375.[Abstract/Free Full Text]

Ekins S, Berbaum J, and Harrison RK (2003) Generation and validation of rapid computational filters for CYP2D6 and CYP3A4. Drug Metab Dispos 31: 1077–1080.[Abstract/Free Full Text]

Ekins S, Bravi G, Ring BJ, Gillespie TA, Gillespie JS, VandenBranden M, Wrighton SA, and Wikel JH (1999) Three dimensional-quantitative structure activity relationship (3D-QSAR) analyses of substrates for CYP2B6. J Pharmacol Exp Ther 288: 21–29.[Abstract/Free Full Text]

Ekins S, de Groot M, and Jones JP (2001) Pharmacophore and three dimensional quantitative structure activity relationship methods for modeling cytochrome P450 active sites. Drug Metab Dispos 29: 936–944.[Abstract/Free Full Text]

Ekins S and Swaan PW (2004) Development of computational models for enzymes, transporters, channels and receptors relevant to ADME/TOX. Rev Comp Chem 20: 333–415.[CrossRef]

Frye LL, Cusack KP, and Leonard DA (1993) 32-Methyl-32-oxylanosterols—dual-action inhibitors of cholesterol biosynthesis. J Med Chem 36: 410–416.[CrossRef][Medline]

Frye LL, Cusack KP, Leonard DA, and Anderson JA (1994) Oxolanosterol oximes—dual-action inhibitors of cholesterol biosynthesis. J Lipid Res 35: 1333–1344.[Abstract]

Fujita T (1997) Recent success stories leading to commercializable bioactive compounds with the aid of traditional QSAR procedures. Quant Struct-Act Relat 16: 107–112.[CrossRef]

Gebhardt R, Beck H, and Wagner KG (1994) Inhibition of cholesterol biosynthesis by allicin and ajoene in rat hepatocytes and HepG2 cells. Biochim Biophys 1994: 57–62.

Hahn M and Rogers D (1995) Receptor surface models 2: application to quantitative structure-activity relationships studies. J Med Chem 38: 2091–2102.[CrossRef][Medline]

Hartman PG and Sanglard D (1997) Inhibitors of ergosterol biosynthesis as antifungal agents: recent success stories leading to commercializable bioactive compounds with the aid of traditional QSAR procedures. Curr Pharm Des 3: 177–208.

Harwood H and Hamanaka E (1998) Modulators of dyslipidemia. Emerging Drugs 3: 147–172.[CrossRef]

Harwood HJ Jr, Petras SF, Hoover DJ, Mankowski DC, Soliman VF, Sugarman ED, Hulin B, Kwon Y, Gibbs EM, Mayne JT, et al. (2005) Dual-action hypoglycemic and hypocholesterolemic agents that inhibit glycogen phosphorylase and lanosterol demethylase. J Lipid Res 46: 547–563.[Abstract/Free Full Text]

Hiroi T, Imaoa S, and Funae Y (1998) Dopamine formation from tyramine by CYP2D6. Biochem Biophys Res Commun 249: 838–843.[CrossRef][Medline]

Holtje HD and Fattorusso C (1998) Construction of a model of the Candida albicans lanosterol 14--demethylase active site using the homology modelling technique. Pharm Acta Helv 72: 271–277.[CrossRef][Medline]

Hoover DJ, Lefkowitz-Snow S, Burgess-Henry JL, Martin WH, Armento SJ, Stock IA, McPherson RK, Genereux PE, Gibbs EM, and Treadway JL (1998) Indole-2-carboxamide inhibitors of human liver glycogen phosphorylase. J Med Chem 41: 2934–2938.[CrossRef][Medline]

Ishida N, Aoyama Y, Hatanaka R, Oyama Y, Imajo S, Ishiguro M, Oshima T, Nakazato H, Noguchi T, Maitra US, et al. (1988) A single amino acid substitution converts cytochrome P450(14DM) to an inactive form, cytochrome P450(SG1): complete primary structures deduced from cloned DNAs. Biochem Biophys Res Commun 155: 317–323.[CrossRef][Medline]

Ji H, Zhang W, Zhou Y, Zhang M, Zhu J, Song Y, Lu J, and Zhu J (2000) A three dimensional model of lanosterol 14-demethylase of Candida albicans and its interaction with azole antifungals. J Med Chem 43: 2493–2505.[CrossRef][Medline]

Jones JP, He M, Trager WF, and Rettie AE (1996) Three-dimensional quantitative structure-activity relationship for inhibitors of cytochrome P4502C9. Drug Metab Dispos 24: 1–6.[Medline]

Kelly SL, Lamb DC, Baldwin BC, Corran AJ, and Kelly DE (1997) Characterization of Saccharomyces cerevisiae CTP61, sterol 22-desaturase, and inhibition by azole antifungal agents. J Biol Chem 272: 9986–9988.[Abstract/Free Full Text]

Kumarakulasingham M, Rooney PH, Dundas SR, Telfer C, Melvin WT, Curran S, and Murray GI (2005) Cytochrome P450 profile of colorectal cancer: identification of markers of prognosis. Clin Cancer Res 11: 3758–3765.[Abstract/Free Full Text]

Lamb DC, Kelly DE, and Kelly SL (1998) Molecular diversity of sterol 14--demethylase substrates in plants, fungi and humans. FEBS Lett 425: 263–265.[CrossRef][Medline]

Lewis DFV, Wiseman A, and Tarbit MH (1999) Molecular modelling of lanosterol 14-demethylase (CYP51) from Saccharomyces cerevisiae via homology with CYP102, a unique bacterial cytochrome P450 isoform: quantitative structure-activity relationships (QSARs) within two related series of antifungal azole derivatives. J Enzyme Inhib 14: 175–192.[Medline]

Martin WH, Hoover DJ, Armento SJ, Stock IA, McPherson RK, Danley DE, Stevenson RW, Barrett EJ, and Treadway JL (1998) Discovery of a human liver glycogen phosphorylase inhibitor that lowers blood glucose in vivo. Proc Natl Acad Sci USA 95: 1776–1781.[Abstract/Free Full Text]

Matsuura K, Yoshioka S, Tosha T, Hori H, Ishimori K, Kitagawa T, Morishima I, Kagawa N, and Waterman MR (2005) Structural diversities of active site in clinical azole-bound forms between sterol 14-demethylases (CYP51s) from human and Mycobacterium tuberculosis. J Biol Chem 280: 9088–9096.[Abstract/Free Full Text]

Miettinen TA (1988) Cholesterol metabolism during ketoconazole treatment in man. J Lipid Res 29: 43–51.[Abstract]

Nelson DR, Koymans L, Kamataki T, Stegeman JJ, Feyereisen R, Waxman DJ, Waterman MR, Gotoh O, Coon MJ, Estabrook RW, et al. (1996) P450 superfamily: update on new sequences, gene mapping, accession numbers and nomenclature. Pharmacogenetics 6: 1–42.[Medline]

Omura T and Sato R (1964) The carbon monoxide-binding pigment of liver microsomes. I. Evidence for its hemoprotein nature. J Biol Chem 239: 2370–2378.[Free Full Text]

Podust LM, Poulos TL, and Waterman MR (2001) Crystal structure of cytochrome P450 14-sterol demethylase (CYP51) from Mycobacterium tuberculosis in complex with azole inhibitors. Proc Natl Acad Sci USA 98: 3068–3073.[Abstract/Free Full Text]

Rath VL, Ammirati M, Danley DE, Ekstrom JL, Gibbs EM, Hynes TR, Mathiowetz AM, McPherson RK, Olson TV, Treadway JL, et al. (2000) Human liver glycogen phosphorylase inhibitors bind at a new allosteric site. Chem Biol 7: 677–682.[CrossRef][Medline]

Rozman D, Stromstedt M, Tsui LC, Scherer SW, and Waterman MR (1996) Structure and mapping of the human lanosterol 14-demethylase gene (CYP51) encoding the cytochrome P450 involved in cholesterol biosynthesis; comparison of exon/intron organization with other mammalian and fungal CYP genes. Genomics 38: 371–381.[CrossRef][Medline]

Rupp B, Raub S, Marian C, and Holtje HD (2005) Molecular design of two sterol 14-demethylase homology models and their interactions with the azole antifungals ketoconazole and bifonazole. J Comput Aided Mol Des 19: 149–163.[CrossRef][Medline]

Sonoda Y, Endo M, Ishida K, Sato Y, Fukusen N, and Fukuhara M (1993) Purification of a human cytochrome P-450 isozyme catalyzing lanosterol 14-demethylation. Biochim Biophys Acta 1170: 92–97.[Medline]

Stehmann C and de Waard M (1995) Relationship between chemical structure and biological activity of triazole fungicides against Botrytis cinerea. Pestic Sci 44: 183–195.[CrossRef]

Strolin Benedetti M and Bani M (1999) Metabolism-based drug interactions involving oral azole antifungals in humans. Drug Metab Rev 31: 665–671.[CrossRef][Medline]

Stromstedt M, Rozman D, and Waterman MR (1996) The ubiquitously expressed human CYP51 encodes lanosterol 14--demethylase, a cytochrome P450 whose expression is regulated by oxysterols. Arch Biochem Biophys 329: 73–81.[CrossRef][Medline]

Stromstedt M, Waterman MR, Haugen TB, Tasken K, Parvinen M, and Rozman D (1998) Elevated expression of lanosterol 14-demethylase (CYP51) and the synthesis of oocyte meiosis-activating sterols in postmeiotic germ cells of male rats. Endocrinology 139: 2314–2321.[Abstract/Free Full Text]

Sundin M, Warner M, Haaparanta T, and Gustafsson JA (1987) Isolation and catalytic activity of cytochrome P-450 from ventral prostate of control rats. J Biol Chem 262: 12293–12297.[Abstract/Free Full Text]

Tafi A, Anastassopoulou J, Theophanides T, Botta M, Corelli F, Massa S, Artico M, Costi R, Di Santo R, and Ragno R (1996) Molecular modeling of azole antifungal agents active against Candida albicans. 1. A comparative molecular field analysis study. J Med Chem 39: 1227–1235.[CrossRef][Medline]

Talele TT, Hariprasad V, and Kulkarni VM (1997) Docking analysis of a series of cytochrome P-450(14) DM inhibiting azole antifungals. Drug Des Discov 15: 181–190.

Treadway JL, Mendys P, and Hoover DJ (2001) Glycogen phosphorylase inhibitors for treatment of type 2 diabetes mellitus. Expert Opin Investig Drugs 10: 439–454.[CrossRef][Medline]

Trzaskos J, Kawata S, and Gaylor JL (1986) Microsomal enzymes of cholesterol biosynthesis: purification of lanosterol 14-methyl demethylase cytochrome P-450 from hepatic microsomes. J Biol Chem 261: 14651–14657.[Abstract/Free Full Text]

Trzaskos JM, Ko SS, Magolda RL, Favata MF, Fischer RT, Stam SH, Johnson PR, and Gaylor JL (1995) Substrate-based inhibitors of lanosterol 14-methyl demethylase: I. Assessment of inhibitor structure-activity relationship and cholesterol biosynthesis inhibition properties. Biochemistry 34: 9670–9676.[CrossRef][Medline]

Trzaskos JM, Magolda RL, Favata MF, Fischer RT, Johnson PR, Chen HW, Ko SS, Leonard DA, and Gaylor JL (1993) Modulation of 3-hydroxy-3-methylglutaryl-CoA reductase by 15--fluorolanost-7-en-3-ß-Ol: a mechanism-based inhibitor of cholesterol biosynthesis. J Biol Chem 268: 22591–22599.[Abstract/Free Full Text]

Tsukuda T, Shiratori Y, Watanabe M, Ontsuka H, Hattori K, Shirai M, and Shimma N (1998) Modeling, synthesis and biological activity of novel antifungal agents (1). Bioorg Med Chem Lett 8: 1819–1824.[CrossRef][Medline]

Tuck EF, Robinson CH, and Silverton JV (1991) Assessment of the active-site requirements of lanosterol 14-demethylase: evaluation of novel substrate analogues as competitive inhibitors. J Org Chem 56: 1260–1266.[CrossRef]

Tuck SF, Aoyama Y, Yoshida Y, and Ortiz de Montellano PR (1992) Active site topology of Saccharomyces cerevisiae lanosterol 14-demethylase (CYP51) and its G310D mutant (cytochrome P-450_SG1). J Biol Chem 267: 13175–13179.[Abstract/Free Full Text]

Vanden Bossche H and Koymans L (1998) Cytochromes P450 in fungi. Mycoses 41: 32–38.[Medline]

Vanden Bossche H, Koymans L, and Moereels H (1995) P450 inhibitors of use in medical treatment: focus on mechanisms of action. Pharmacol Ther 67: 79–100.[CrossRef][Medline]

Vanden Bossche H, Marichal P, Gorrens J, Bellens D, Verhoeven H, Coene M-C, Lauwers W, and Janssen PAJ (1987) Interaction of azole derivatives with cytochrome P-450 isozymes in yeast, fungi, plants and mammalian cells. Pestic Sci 21: 289–306.[CrossRef]

Waterman MR and Lepesheva GI (2005) Sterol 14-demethylase, an abundant and essential mixed-function oxidase. Biochem Biophys Res Commun 338: 418–422.[CrossRef][Medline]

Wrighton SA, Ring BJ, and VandenBranden M (1995) The use of in vitro metabolism techniques in the planning and interpretation of drug safety studies. Toxicol Pathol 23: 199–208.[Abstract/Free Full Text]

Yoshida Y and Aoyama Y (1985) Effects of buthiobate, a fungicide, on cytochrome P-450 of rat liver microsomes. J Pharmacobiodyn 8: 432–439.[Medline]

Yoshida Y and Aoyama Y (1987) Interaction of azole antifungal agents with cytochrome P-45014DM purified from Saccharomyces cerevisiae microsomes. Biochem Pharmacol 36: 229–235.[CrossRef][Medline]

Yoshida Y, Noshiro M, Aoyama Y, Kawamoto T, Horiuchi T, and Gotoh O (1997) Structural and evolutionary studies on sterol 14-demethylase P450 (CYP51), the most conserved P450 monooxygenase: II. Evolutionary analysis of protein and gene structures. J Biochem (Tokyo) 122: 1122–1128.[Abstract/Free Full Text]

Yu LJ, Chen Y, Treadway JL, McPherson RK, McCoid SC, Gibbs EM, and Hoover DJ (2006) Establishment of correlation between in vitro enzyme binding potency and in vivo pharmacological activity: application to liver glycogen phosphorylase a inhibitors. J Pharmacol Exp Ther 317: 1230–1237.[Abstract/Free Full Text]

This article has been cited by other articles:

				A. N. Eddine, J. P. von Kries, M. V. Podust, T. Warrier, S. H. E. Kaufmann, and L. M. Podust X-ray Structure of 4,4'-Dihydroxybenzophenone Mimicking Sterol Substrate in the Active Site of Sterol 14{alpha}-Demethylase (CYP51) J. Biol. Chem., May 30, 2008; 283(22): 15152 - 15159. [Abstract] [Full Text] [PDF]

				S. Ekins, C. Chang, S. Mani, M. D. Krasowski, E. J. Reschly, M. Iyer, V. Kholodovych, N. Ai, W. J. Welsh, M. Sinz, et al. Human Pregnane X Receptor Antagonists and Agonists Define Molecular Requirements for Different Binding Sites Mol. Pharmacol., September 1, 2007; 72(3): 592 - 603. [Abstract] [Full Text] [PDF]

This Article Abstract Full Text (PDF) Data Supplement Erratum All Versions of this Article: dmd.106.013888v1 35/3/493    most recent Submit a response Alert me when this article is cited Alert me when eLetters are posted Alert me if a correction is posted Services Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Ekins, S. Articles by Harwood, H. J. Search for Related Content PubMed PubMed Citation Articles by Ekins, S. Articles by Harwood, H. J., Jr.