Comparison of Steroid Hormone Hydroxylations by and Docking to Human Cytochromes P450 3A4 and 3A5.

PURPOSE
Hydroxylation activity at the 6β-position of steroid hormones (testosterone, progesterone, and cortisol) by human cytochromes P450 (P450 or CYP) 3A4 and CYP3A5 and their molecular docking energy values were compared to understand the catalytic properties of the major forms of human CYP3A, namely, CYP3A4 and CYP3A5.


METHODS
Testosterone, progesterone, and cortisol 6β-hydroxylation activities of recombinant CYP3A4 and CYP3A5 were determined by liquid chromatography. Docking simulations of these substrates to the heme moiety of reported crystal structures of CYP3A4 (Protein Data Bank code ITQN) and CYP3A5 (6MJM) were conducted.


RESULTS
Michaelis constants (Km) for CYP3A5- mediated 6β-hydroxylation of testosterone and progesterone were approximately twice those for CYP3A4, whereas the value for cortisol 6β-hydroxylation mediated by CYP3A5 was similar to the value for that by CYP3A4. Maximal velocities (Vmax) of the three steroid hormones 6β-hydroxylation catalyzed by CYP3A5 were 30%-63% of those by CYP3A4. Thus, Vmax/ Km values of these hormones for CYP3A5 resulted in 22%- 31% of those for CYP3A4. The differences in the docking energies between CYP3A4 and CYP3A5 for steroid hormones were slightly correlated to the logarithm of CYP3A5/CYP3A4 ratios for Km values (substrate affinity).


CONCLUSIONS
The Vmax, rather than Km values, for CYP3A5-mediated 6β-hydroxylation of three steroid hormones were different from those for CYP3A4. Molecular docking simulations could partially explain the differences in the accessibility of substrates to the heme moiety of human CYP3A molecules, resulting in the enzymatic affinity of CYP3A4 and CYP3A5.


INTRODUCTION
Cytochrome P450 (P450 or CYP) 3A is the most important human P450 subfamily, due to its high abundance in the human hepatic microsomes, which accounts for approximately 30% of total P450s enzymes (1,2). Furthermore, CYP3A enzymes are responsible for 50% of all pharmaceuticals oxidation (3). Inducible CYP3A4 is generally recognized as the predominant CYP3A form expressed in human livers and intestines. On the other hand, polymorphically expressed CYP3A5 contributes as much as 50% of hepatic CYP3A in a third of Caucasians and about half of African-Americans (4). It was thought that the substrate specificity of CYP3A5 is similar to that of CYP3A4, because CYP3A4 and CYP3A5 are 83% homologous in terms of amino acid sequences. However, some differences in catalytic properties have been reported (5,6). We previously reviewed the reported values for (a) Michaelis-Menten constants (Km), maximal velocities (Vmax), and intrinsic clearance (CLint, Vmax/Km) values for reactions mediated by CYP3A4 and/or CYP3A5, (b) inhibition constants (Ki) and 50% inhibitory concentrations (IC50), and (c) maximum inactivation rate constants (kinact), and demonstrated that the substrate specificities of CYP3A4 and CYP3A5 generally overlap, but some differences can be identified. (7,8).
It is well established that some steroid hormones are metabolized by not only steroidogenic P450s, such as CYP11B1, CYP11B2, CYP17A1, CYP19A1, _________________________________________ Corresponding Author: Prof. Toshiro Niwa, PhD, School of Pharmacy, Shujitsu University , 1-6-1 Nishigawara, Naka-ku, Okayama 703-8516, Japan; E-mail: tniwa@shujitsu.ac.jp CYP27A1, but also by drug-metabolizing P450s including CYP2C, CYP2D, and the CYP3A subfamily (9,10). In terms of hydroxylation at the C6β position, testosterone 6β-hydroxylation is predominantly catalyzed by the CYP3A subfamily (9,10) and is recommended as a preferred reaction of CYP3A4/5 for in vitro experiments in the guidelines for new drug applications regarding in vitro studies of drug metabolism and drug-drug interactions regulated by the US Food and Drug Administration (FDA), European Medicines Agency (EMA), and Japanese Ministry of Health, Labour and Welfare (11)(12)(13). The ratio of 6β-hydroxycortisol to cortisol, a stress hormone, in plasma or urine has been used as an in vivo endogenous marker of CYP3A4/5 metabolic activity (14)(15)(16). Progesterone is hydroxylated by drug-metabolizing P450s including CYP2D6 and CYP3A4 at the C2β, C6α, C6β, C16α, and C21 positions, as well as by the steroidogenic CYP21A1 at the C21 position (9,10). However, there are few reports that compare, in detail, the kinetic parameters for the hydroxylation of these steroid hormones mediated by CYP3A4 and CYP3A5.
We previously conducted docking simulation studies for substrates and inhibitors of P450 proteins, including CYP3A4 and CYP3A5, and demonstrated that some of the results were consistent with the enzymatic characteristics observed in in vitro studies, although the primary sequence of CYP3A5 was aligned with that of CYP3A4 using threedimensional modeling software (17,18).
In the present study, we compared the CYP3A4 and CYP3A5-mediated 6β-hydroxylation of three steroid hormones: testosterone, progesterone, and cortisol. In addition, we carried out docking simulations for the binding of CYP3A4 and CYP3A5 with steroid hormones using a docking simulation program and compared the results with previously reported compounds (18). We report herein that the differences in accessibility of substrates to the heme of human CYP3A molecules resulted in the apparent enzymatic affinity of CYP3A4 and CYP3A5.

Materials
CYP3A4 and CYP3A5 expressed in recombinant Escherichia coli (Bactosomes) were obtained from Cypex Ltd (Dundee, UK). Testosterone and cortisol were purchased from Nakarai Tesque (Kyoto, Japan) and Fujifilm Wako Pure Chemicals (Osaka, Japan), respectively. Metabolites 6β-hydroxytestosterone and 6β-hydroxycortisol were obtained from Sigma-Aldrich (St Louis, MO, USA). Progesterone and 6βhydroxyprogesterone were purchased from Tokyo Chemical Industry (Tokyo, Japan) and Steraloids Inc. (Newport, RI USA), respectively. All other reagents and organic solvents used were of the highest purity commercially available.
All data were analyzed using the average of triplicate determinations. In the preliminary experiments, the linearity of the reaction with P450 concentration and incubation time was confirmed for each CYP3A subfamily. Km and Vmax values were calculated by performing Michaelis-Menten kinetics using nonlinear least squares regression by means of MULTI (23).

Docking Simulation of Steroid Hormones with Human P450 Enzymes
The crystal structures of CYP3A4 (Protein Data Bank code ITQN) and CYP3A5 (6MJM) were used (24,25). Prior to the docking simulations, the energies of the CYP3A4 and CYP3A5 structures were minimized using the CHARMM22 force field. Docking simulations were carried out for the binding of substrates to P450 enzymes using the MMFF94x force field distributed in the Molecular Operating Environment (MOE) ASEDock software (Chemical Computing Group, Montreal, Canada) (18). Twenty solutions were generated for each docking experiment and then ranked according to total interaction energy (U value).

Literature Search for Kinetic Parameters
For the correlation analysis, kinetic parameters such as Km, Vmax, and Vmax/Km for several compounds, including estradiol and estrone, were obtained from the literature (10,18,26). Correlation analysis was carried out using Prism (GraphPad Software, La Jolla, CA).

RESULTS
The kinetic parameters for the 6β-hydroxylation of three steroid hormones by CYP3A4 and CYP3A5 were estimated by fitting the values into the Michaelis-Menten plots ( Table 1). The Km values for CYP3A5-mediated 6β-hydroxylation of testosterone and progesterone were approximately twice those for CYP3A4, whereas the value for cortisol 6βhydroxylation by CYP3A5 was similar to that of CYP3A4. The Vmax values for CYP3A5-mediated 6β-hydroxylation of the three steroid hormones were 30-63% of those for CYP3A4. Thus, Vmax/ Km values for the 6β-hydroxylation of these hormones by CYP3A5 were 22-31% of those for CYP3A4. For both CYP3A4 and CYP3A5, Km values for progesterone were lowest, followed by testosterone. Vmax values for testosterone were highest among the three steroid hormones, and Vmax/ Km values for cortisol were lowest.
Molecular docking simulations were conducted on six steroid hormones with CYP3A4 and CYP3A5 ( Table 2). For estrone, the Km value for CYP3A5mediated 4-hydroxylation is reported to be higher than that for CYP3A4 (10,26). As such, ligand-P450 interaction energies (U values) were found to be lower for CYP3A4 than for CYP3A5. However, U values for CYP3A5 in other steroid hormones were lower than those for CYP3A4. In all cases, the docking distance was between 3.82 and 11.3Å. Several compounds, other than these steroid hormones, were investigated in order to confirm the present docking results.  CYP3A5 (8, 18), had a lower U value for CYP3A4 than that for CYP3A5. We investigated the relationship between the ratios of mean Km, Vmax, and Vmax/Km values obtained in the present study and previously reviewed for CYP3A4 and CYP3A5 (18), and the differences between the U values for these P450s (Fig. 1). The difference in U values between CYP3A5 and CYP3A5 were slightly correlated with the logarithm of CYP3A5/CYP3A4 ratios for Km (r = 0.62, p = 0.14, Fig. 1A) and Vmax values (r=0.64, p = 0.12, Fig.  1B), but not with the logarithm of CYP3A5/CYP3A4 ratios for Vmax/Km values (Fig.  1C). U: interaction energy in kcal/mol. Seven reactions of five steroid hormones as shown in Table  2, except for DHEA, were investigated. Dotted lines show the regression curve between the logarithm of the CYP3A5/CYP3A4 ratios of Km, Vmax, and Vmax/Km values and the differences between U values for CYP3A4 and CYP3A5 for steroid hormones.

DISCUSSION
Testosterone 6β-hydroxylation is recommended as a preferred reaction of CYP3A4/5 for in vitro experiments (11)(12)(13) and the 6βhydroxycortisol/cortisol ratio is used as an in vivo endogenous marker of CYP3A4/5 metabolic activities (14)(15)(16). In addition, cortisol, a stress hormone, is known to be one of the many hormones responsible for the physiological changes involved in response to stress, as well as a glucocorticoid hormone that is produced by the adrenal glands and released for a variety of reasons (29). Progesterone 6β-hydroxylation is catalyzed by drug-metabolizing P450s, including CYP3A4 and CYP2D6 (9, 10).    Thus, the addition of other substrates, such as midazolam, might be necessary for the assessment of CYP3A4/5 activity as recommended by FDA and PMDA (11,13). Furthermore, the effect of CYP3A5 polymorphism on the plasma or urine ratio of 6βhydroxycortisol/cortisol in vivo as a marker for stress might be of minor importance. Further clinical studies are necessary to elucidate the problem. We previously carried out docking simulations for the binding of CYP3A4 and CYP3A5 with 13 substrates using docking simulation program MOE and found that the docking energy ratio between CYP3A5 and CYP3A4 were not significantly correlated with the CYP3A5/CYP3A4 ratios of Vmax and Vmax/Km values, nor with molecular weight, volume, or log P values of substrates. Furthermore, neither the differences in the enzyme source (baculovirus infected insect cells or Escherichia coli engineered to express P450s) nor the supplementation or co-expression of cytochrome b5 were correlated with the Km, Vmax, and Vmax/Km values for CYP3A4 or CYP3A5 (18). In our previous study, the primary sequence of CYP3A5 was aligned with that of CYP3A4 using threedimensional modeling software (18), while in the present study, we used the recently reported crystal structures of CYP3A5 (Protein Data Bank code 6MJM) (25). Vincristine, a typical substrate of CYP3A5, but not of CYP3A4, has higher Vmax and Vmax/Km values and a lower Km value for reactions mediated by CYP3A5 than by CYP3A4 (27). In the present study, vincristine was shown to dock closer to the heme region of CYP3A5 (10.81Å) with a lower U value than to that of CYP3A4 (18.50Å); the findings were similar to the previous in silico analysis (18). In the case of midazolam, for which Km value for CYP3A5 was higher than for CYP3A4, the docking energies for CYP3A4 were lower than those for CYP3A5. Although U values for CYP3A5 in steroid hormones, with the exception of estrone, were lower than those for CYP3A4, the difference in Km values as well as Vmax values, but not Vmax/Km values and docking results, between CYP3A4 and CYP3A5 for steroid hormones were slightly corelated (Fig. 2). However, because it's highly unlikely that the reaction with higher Vmax values would have higher U values, we suggest that molecular docking simulations could partially explain the differences in substrate affinity (the Km values), as demonstrated in the previous study (18). As CYP3A4 and CYP3A5 are 83% homologous in terms of amino acid sequences, the most plausible explanation for the differences between CYP3A4 and CYP3A5 is that conformational changes, caused by amino acid differences, lead to altered accessibility of substrates and inhibitors to the heme iron of CYP3A4 and CYP3A5. However, it would be difficult to determine the substrates for which Km and Vmax values for CYP3A5 are different from those for CYP3A4, by using the differences in U values (18). Further quantitative analytical studies with additional substrates and inhibitors, including those with a preference for either CYP3A4 or CYP3A5, are required. In the future, more advanced simulations such as molecular dynamics simulations are of interest, in order to explain the induced-fit effects caused by different ligands binding.
In conclusion, Vmax values, rather than Km values, for 6β-hydroxylation of three steroid hormones by CYP3A5 were different from those for CYP3A4, and molecular docking simulations might partially explain the differences in the affinity of substances for CYP3A4 and CYP3A5, based on the accessibility of substrates to the heme moiety of CYP3A molecules.