Generation of human chiral metabolites of simvastatin and lovastatin by bacterial CYP102A1 mutants

Article abstract of DOI:10.1124/dmd.110.036392

Recently, the wild-type and mutant forms of cytochrome P450 BM3 (CYP102A1) from Bacillus megaterium were found to oxidize various xenobiotic substrates, including pharmaceuticals, of human P450 enzymes. Simvastatin and lovastatin, which are used to treat hyperlipidemia and hypercholesterolemia, are oxidized by human CYP3A4/5 to produce several metabolites, including 6′β-hydroxy (OH), 3″-OH, and exomethylene products. In this report, we show that the oxidation of simvastatin and lovastatin was catalyzed by wild-type CYP102A1 and a set of its mutants, which were generated by site-directed and random mutagenesis. One major hydroxylated product (6′β-OH) and one minor product (6′-exomethylene), but not other products, were produced by CYP102A1 mutants. Formation of the metabolites was confirmed by highperformance liquid chromatography, liquid chromatography-mass spectroscopy, and NMR. Chemical methods to synthesize the metabolites of simvastatin and lovastatin have not been reported. These results demonstrate that CYP102A1 mutants can be used to produce human metabolites, especially chiral metabolites, of simvastatin and lovastatin. Our computational findings suggest that a conformational change in the cavity of the mutant active sites is related to the activity change. The modeling results also suggest that the activity change results from the movement of several specific residues in the active sites of the mutants. Furthermore, our computational findings suggest a correlation between the stabilization of the binding site and the catalytic efficiency of CYP102A1 mutants toward simvastatin and lovastatin. Copyright

Full text of DOI:10.1124/dmd.110.036392

Supplemental Material can be found at:
http://dmd.aspetjournals.org/content/suppl/2010/10/20/dmd.110.036392.DC1.html
0090-9556/11/3901-140–150$20.00
DRUG METABOLISM AND DISPOSITION
Copyright © 2011 by The American Society for Pharmacology and Experimental Therapeutics
DMD 39:140–150, 2011
Vol. 39, No. 1
36392/3655240
Printed in U.S.A.
Generation of Human Chiral Metabolites of Simvastatin and
□
S
Lovastatin by Bacterial CYP102A1 Mutants
Keon-Hee Kim, Ji-Yeon Kang, Dong-Hyun Kim, Sun-Ha Park, Seon Ha Park, Dooil Kim,
Ki Deok Park, Young Ju Lee, Heung-Chae Jung, Jae-Gu Pan, Taeho Ahn, and Chul-Ho Yun
School of Biological Sciences and Technology, Chonnam National University, Gwangju, Republic of Korea (K.-H.K., J.-Y.K.,
D.-H.K., Su.-H.P., Se.H.P., C.-H.Y.); Systems Microbiology Research Center, Korea Research Institute of Bioscience and
Biotechnology, Daejeon, Republic of Korea (D.K., H.-C.J., J.-G.P.); Gwangju Center, Korea Basic Science Institute, Gwangju,
Republic of Korea (K.D.P., Y.J.L.); and Department of Biochemistry, College of Veterinary Medicine, Chonnam National
University, Gwangju, Republic of Korea (T.A.)
Received September 17, 2010; accepted October 20, 2010
ABSTRACT:
Recently, the wild-type and mutant forms of cytochrome P450 BM3 performance liquid chromatography, liquid chromatography-mass
(CYP102A1) from Bacillus megaterium were found to oxidize vari- spectroscopy, and NMR. Chemical methods to synthesize the me-
ous xenobiotic substrates, including pharmaceuticals, of human tabolites of simvastatin and lovastatin have not been reported.
P450 enzymes. Simvastatin and lovastatin, which are used to treat These results demonstrate that CYP102A1 mutants can be used to
hyperlipidemia and hypercholesterolemia, are oxidized by human produce human metabolites, especially chiral metabolites, of sim-
CYP3A4/5 to produce several metabolites, including 6ꢀ␤-hydroxy vastatin and lovastatin. Our computational findings suggest that a
(OH), 3ꢁ-OH, and exomethylene products. In this report, we show conformational change in the cavity of the mutant active sites is
that the oxidation of simvastatin and lovastatin was catalyzed by related to the activity change. The modeling results also suggest
wild-type CYP102A1 and a set of its mutants, which were gener- that the activity change results from the movement of several
ated by site-directed and random mutagenesis. One major hy- specific residues in the active sites of the mutants. Furthermore,
droxylated product (6ꢀ␤-OH) and one minor product (6ꢀ-exometh- our computational findings suggest a correlation between the sta-
ylene), but not other products, were produced by CYP102A1 bilization of the binding site and the catalytic efficiency of
mutants. Formation of the metabolites was confirmed by high- CYP102A1 mutants toward simvastatin and lovastatin.
Introduction
Simvastatin and lovastatin are well known hyperlipidemia and
hypercholesterolemia drugs that act as cholesterol-lowering agents
(Caron et al., 2007). Simvastatin (marketed under the trade names
Zocor, Simlup, Simcard, and Simvacor) is metabolized to at least four
primary metabolites, namely 6Ј␤-OH simvastatin, 6Ј-exomethylene
simvastatin, 6Ј␤-hydroxymethyl metabolite, and 3Љ-OH simvastatin
(Fig. 1). Although CYP3A4 is the main enzyme involved in the
primary metabolism of simvastatin, CYP2C8 (Tornio et al., 2005),
CYP2C9 (Transon et al., 1996), and CYP2D6 (Transon et al., 1996)
are also involved in the formation of simvastatin metabolites. The
This work was supported in part by the 21C Frontier Microbial Genomics
and the Application Center Program of the Ministry of Education, Science and
Technology of the Republic of Korea; the National Research Foundation [Grant
2010-0027685]; the Second Stage BK21 Project from the Ministry of Educa-
tion, Science and Technology of the Republic of Korea; and the Support
Program for the Advancement of National Research Facilities and Equipment
of the Ministry of Education, Science and Technology of the Republic of Korea
(2010).
K.-H.K., J.-Y.K., and D.-H.K. contributed equally to this work.
Parts of this work were previously presented at the following conference: Kang
JY, Kim KH, Kim DH, Jung HC, Pan JG, Ahn T, and Yun CH (2009) Generation of extensive oxidative metabolism of lovastatin (marketed under the
human metabolites of simvastatin and lovastatin by bacterial cytochrome P450
trade names Mevacor and Altoprev) in the human liver is primarily
mediated by CYP3A enzymes, particularly CYP3A4, to generate
three known metabolites, 6Ј␤-OH lovastatin, 3Љ-OH lovastatin, and
6Ј-exomethylene (García et al., 2003; Caron et al., 2007) (Fig. 2).
After oral ingestion, simvastatin and lovastatin, which are inactive
lactones, are hydrolyzed to the corresponding ␤-hydroxyacid form
BM3 enzymes [abstract]. Drug Metab Rev 41 (Suppl 3):47. 16th North American
Regional ISSX Meeting; 2009 Oct 18–22; Baltimore, MD. International Society for
the Study of Xenobiotics, Washington, DC.
Article, publication date, and citation information can be found at
http://dmd.aspetjournals.org.
doi:10.1124/dmd.110.036392.
(Vickers et al., 1990a). This molecule is a principal metabolite and an
inhibitor of HMG-CoA reductase. This enzyme catalyzes the conver-
□S The online version of this article (available at http://dmd.aspetjournals.org)
contains supplemental material.
ABBREVIATIONS: OH, hydroxy; FDA, U.S. Food and Drug Administration; MIST, metabolites in safety testing; WT, wild type; P450, cytochrome
P450; CYP102A1, cytochrome P450 BM3; HPLC, high-performance liquid chromatography; LC, liquid chromatography; MS, mass spectrometry;
␣NF, 7,8-benzoflavone; LOV, lovastatin; SIM, simvastatin; ESP, electrostatic potential; MM, molecular mechanics; PBSA, Poisson Boltzmann and
surface area; MD, molecular dynamics.
140

OXIDATION OF SIMVASTATIN AND LOVASTATIN BY CYP102A1 MUTANTS
141
FIG. 1. Structures of simvastatin and its human metabolites. The formation of metabolites is catalyzed by P450 enzymes in the presence of NADPH in the human liver.
sion of HMG-CoA to mevalonate, which is an early and rate-limiting
In February 2008, the U.S. Food and Drug Administration
step in the biosynthesis of cholesterol. In addition to the P450- (FDA) modified its standards for evaluating drug toxicity, partic-
mediated oxidation and ␤-oxidation processes, glucuronidation con- ularly with regard to the toxicity of drug metabolites when it issued
stitutes a common metabolic pathway for statins (Prueksaritanont et the Guidance for Industry: Safety Testing of Drug Metabolites
al., 2002). The metabolites resulting from microsomal oxidation of (http://www.fda.gov/downloads/Drugs/GuidanceComplianceRegulatory-
simvastatin (Vickers et al., 1990b) and lovastatin (Vyas et al., 1990) Information/Guidances/ucm079266.pdf). According to these guide-
by P450 enzymes are effective inhibitors of HMG-CoA reductase. lines, any human drug metabolites “formed at greater than 10% of
Therefore, it has been suggested that the metabolites may contribute parent drug systemic exposure at steady state” should be subject to
to the cholesterol-lowering effect of simvastatin and lovastatin. How- separate safety testing, which involves the synthesis and administra-
ever, systematic studies of the safety, efficacy, and toxicity of these tion of these metabolites to test animals (Guengerich, 2009 and
metabolites have not been performed.
references therein). The issue of human metabolites in safety testing
FIG. 2. Structures of lovastatin and its hu-
man metabolites. The formation of metab-
olites is catalyzed by P450 enzymes in the
presence of NADPH in the human liver.

142
KIM ET AL.
final organic solvent concentration of Ͻ1% (v/v). For the human CYP3A4
activity assay, a control experiment of 50 pmol of P450, 100 pmol of NADPH-
P450 reductase, 100 pmol of cytochrome b5, and 45 ␮M L-␣-dilauroyl-sn-
glycero-3-phosphocholine in 0.25 ml of 100 mM potassium phosphate buffer
(pH 7.4) was used instead of 50 pmol of CYP102A1.
Reactions were generally incubated for 10 min at 37°C and terminated with
a 2-fold excess of ice-cold dichloromethane. After centrifugation of the reac-
tion mixture, the supernatant was carefully removed, and the solvent was
evaporated under N2 gas as described previously (Vickers et al., 1990b). The
products were analyzed by HPLC using a Gemini C18 column (4.6 ϫ 150 mm,
5 ␮m; Phenomenex, Torrance, CA) with an acetonitrile-water (70:30, v/v)
mobile phase containing 2.5 mM formic acid. Eluates were detected by UV at
238 nm. Authentic metabolites of 6Ј␤-OH and 6Ј-exomethylene were obtained
by HPLC from reaction mixtures of simvastatin and lovastatin with human
CYP3A4 as described below.
To determine the total turnover numbers of several P450 enzymes, a 1.0
mM concentration of statin was used. The reaction in the presence of a 500 ␮M
concentration of substrate and 50 pmol of enzyme was initiated by the addition
of the NADPH-generating system in 0.25 ml of 100 mM potassium phosphate
buffer (pH 7.4) and incubated for 4 h at 37°C. After a 2-h incubation, a 500 ␮M
concentration of each substrate was added to the reaction mixture. The for-
mation rate of statin metabolites was determined by HPLC as described above.
The kinetic parameters (Km and kcat) were determined using nonlinear
regression analysis with GraphPad Prism software (GraphPad Software, Inc.,
San Diego, CA). The data were fit to the standard Michaelis-Menten equation:
v ϭ kcat[E][S]/([S] ϩ Km), where the velocity of the reaction is a function of
the turnover rate (kcat), which is the rate-limiting step, the enzyme concentra-
tion ([E]), the substrate concentration ([S]), and the Michaelis constant (Km).
LC-MS Analysis. CYP102A1 mutants and human CYP3A4 were incubated
with 100 ␮M lovastatin and simvastatin at 37°C for 30 min in the presence of
an NADPH-generating system. Reactions were terminated by the addition of a
2-fold excess of ice-cold dichloromethane. After centrifugation, the superna-
tant from each incubated reaction was removed and evaporated to dryness. The
(MIST) has presented a challenge at the early stages of drug devel-
opment for the pharmaceutical industry. Some metabolites of concern
can be prepared by chemical methods, but others may not be easily
prepared by these methods. In the latter case, human liver micro-
somes, heterologously expressed human P450 enzymes in bacteria
(Vail et al., 2005; Yun et al., 2006) and in insect cells (Rushmore et
al., 2000), and purified human P450 enzymes are potential candidates
for biocatalysts to prepare human drug metabolites. However, these
biocatalysts have several weaknesses, such as low catalytic activity
and poor stability for industrial use (Guengerich 2002; Julsing et al.,
2008).
The heme domain of P450 BM3 (CYP102A1) from Bacillus mega-
terium, which possesses monooxygenase activity, has strong similar-
ities to eukaryotic members of the CYP4A (fatty acid hydroxylase)
family. The P450 BM3 reductase domain, a mammalian-like diflavin
reductase, is fused to the C terminus of the P450 domain in a single
polypeptide. The fusion of these two enzymatic activities makes
soluble CYP102A1 an ideal model for mammalian, particularly hu-
man, P450 enzymes. In recent studies, CYP102A1 mutants engi-
neered through rational design or directed evolution were reported to
produce human drug metabolites (Kim et al., 2008b, 2009, 2010;
Sawayama et al., 2009; Park et al., 2010). In addition, CYP102A1 is
a versatile monooxygenase with a demonstrated ability to catalyze a
diversity of substrates and an established relevance to biotechnology
(Urlacher and Eiben, 2006; Yun et al., 2007; Julsing et al., 2008 and
references within).
In this study, we evaluated whether CYP102A1 mutants could be
used to produce the human metabolites of simvastatin and lovastatin.
To our knowledge, the human metabolites of these statins have not
been obtained by chemical synthesis and are not commercially avail-
able. Some of the tested mutations enabled the CYP102A1 enzyme to reaction residue was reconstituted into 100 ␮l of mobile phase by vortex
mixing and sonication for 20 s. An aliquot (10 ␮l) of this solution was injected
onto the LC column. LC-MS analysis was performed in electrospray ionization
(positive) mode on a Shimadzu LCMS-2010 EV system (Shimadzu, Kyoto,
Japan) having LC-MS solution software. The separation was performed on a
Shim-pack VP-ODS column (2.0 mm i.d. ϫ 250 mm; Shimadzu) using an
acetonitrile-water (70:30, v/v) mobile phase containing 2.5 mM formic acid at
a flow rate of 0.16 ml/min. To identify the metabolites, mass spectra were
recorded by electrospray ionization in positive mode. The interface and detec-
tor voltages were 4.4 and 1.5 kV, respectively. The nebulization gas flow was
set at 1.5 l/min. The interface, curve desolvation line, and heat block temper-
atures were 250°C, 230°C, and 200°C, respectively.
catalyze the oxidation of simvastatin and lovastatin to generate their
human metabolites. Furthermore, computational calculations and mo-
lecular modeling have been performed to provide further insight into
molecular recognition on the basis of free energy decomposition and
hydrogen bond analysis.
Materials and Methods
Chemicals. Simvastatin and lovastatin were obtained from Merck (Rahway,
NJ). NADPH was purchased from Sigma-Aldrich (St. Louis, MO). Other
chemicals were of the highest grade commercially available.
Identification of the Metabolites by NMR Spectroscopy. An Agilent
model 1100 HPLC system (Agilent Technologies, Santa Clara, CA) was used
for the isolation of the 6Ј␤-OH metabolites of simvastatin and lovastatin in the
reaction mixtures. Semipreparative columns were used for the isolation of
6Ј␤-OH simvastatin (SunFire Prep C18, 5 ␮m, 10 mm i.d. ϫ 150 mm; Waters,
Milford, MA) and 6Ј␤-OH lovastatin (Pursuit C18, 5 ␮m, 10 mm i.d. ϫ 250
mm; Varian, Inc., Palo Alto, CA) from the reaction mixtures. Each metabolite
mixture, which was approximately 10 mg in CH3OH, was injected onto the
semipreparative columns. The 6Ј␤-OH simvastatin was eluted with a linear
gradient (1.5%/min) of 30 to 90% CH3CN after elution with 30% CH3CN for
10 min. The metabolite fractions were collected at 18.2 min. The 6Ј␤-OH
lovastatin was eluted with a series of gradients: H2O-CH3CN (75:25, v/v) for
20 min; 25 to 45% CH3CN for 40 min (0.5%/min); 45 to 90% CH3CN for 40
min (4.5%/min); and 90% CH3CN for 10 min. The metabolite fractions were
collected at 63.7 min. The flow rate was 3 ml/min for both columns, and the
eluates were monitored at 240 nm.
Construction of P450 BM3 Mutants by Site-Directed Mutagenesis. The
CYP102A1 mutants used in this study were selected on the basis of earlier
work showing their increased catalytic activity toward several human sub-
strates (Kim et al., 2008b; Park et al., 2010 and references therein). Each
mutant bears amino acid substitution(s) relative to wild-type (WT) CYP102A1,
as summarized in Supplemental Table S1.
Expression and Purification of CYP102A1 Mutants. Wild-type and mu-
tant forms of CYP102A1 were expressed in Escherichia coli strain
DH5␣FЈ-IQ and purified as described previously (Kim et al., 2008b). The
CYP102A1 concentrations were determined from CO difference spectra using
␧ ϭ 91 mM/cm. For all of the wild-type and mutated enzymes, a typical culture
yielded 300 to 700 nM P450. The expression levels of wild-type and mutant
CYP102A1 forms were typically in the range of 1.0 to 2.0 nmol P450/mg
cytosolic protein.
Oxidation of Simvastatin and Lovastatin. Typical steady-state reactions
for the oxidation of simvastatin and lovastatin included 50 pmol of CYP102A1
in 0.25 ml of 100 mM potassium phosphate buffer (pH 7.4) along with a
NMR experiments were performed on a VNMRS 600-MHz NMR spec-
specified amount of substrate. To determine the kinetic parameters of several trometer (Varian) equipped with a carbon-enhanced cryogenic probe. Chloro-
CYP102A1 mutants, 2 to 200 ␮M concentrations of statins were used. An
form-d1 was used as a solvent, and chemical shifts for proton and carbon were
measured in parts per million relative to trimethylsilane. All of the one-
aliquot of an NADPH-generating system was used to initiate reactions; final
concentrations were 10 mM glucose 6-phosphate, 0.5 mM NADPϩ, and 1 dimensional and two-dimensional NMR experiments were performed with
IU/ml yeast glucose 6-phosphate. A stock solution of statins (20 mM) was
prepared in dimethyl sulfoxide and diluted into the enzyme reactions with a
standard pulse sequences in the VNMR (version 2.3) library and processed
with the same software. Spectral assignments were done with one-dimensional

OXIDATION OF SIMVASTATIN AND LOVASTATIN BY CYP102A1 MUTANTS
143
1H and 13C NMR spectroscopy along with two-dimensional NMR spectro- coupling. A SHAKE algorithm was applied to constrain all bonds containing
scopic techniques (double quantum correlation spectroscopy, heteronuclear
single quantum correlation, and heteronuclear multiple-bond correlation spec-
hydrogen atoms (van Gunsteren and Berendsen, 1977). The nonbonded cutoff
was kept at 10 Å, and long-range electrostatic interactions were treated by the
troscopy). The stereochemical configurations of the 6Ј␤-OH position of both particle mesh Ewald method with a fast Fourier transform grid having approx-
compounds were determined with 1-dimensional nuclear Overhauser effect imately 0.1 nm space (Darden et al., 1993). Trajectory snapshots were taken at
spectroscopy.
each 1 ps, which were then used for analysis. All simulations were performed
Effects of 7,8-Benzoflavone on the Oxidation Reactions of Simvastatin by the SANDER module of AMBER 10 with amber force field (ff03) (Case et
and Lovastatin Catalyzed by CYP102A1 Enzymes. Reaction mixtures con- al., 2008).
sisted of 50 pmol of P450, 100 mM potassium phosphate buffer (pH 7.4), an
NADPH-generating system, substrate (100 ␮M of simvastatin or lovastatin), Boltzmann and surface area (MM/PBSA) methods were used to calculate
and 200 ␮M 7,8-benzoflavone (␣NF) in 0.25 ml of 100 mM potassium
binding free energies for all of the CYP102A1 mutant complexes, using
phosphate buffer (pH 7.4). Products were analyzed by HPLC as described MM/PBSA implementation in AMBER 10 (Massova and Kollman, 2000).
Binding Free Energy Calculation. The molecular mechanics Poisson
above. In the case of the human CYP3A4 activity assay, a control experiment
of 50 pmol of P450, 100 pmol of NADPH-P450 reductase, 100 pmol of
cytochrome b5, and 45 ␮M L-␣-dilauroyl-sn-glycero-3-phosphocholine was
used instead of 50 pmol of CYP102A1.
Binding free energy (⌬Gbind) is given by the following equation: ⌬Gbind ϭ
⌬EMM ϩ ⌬GSol ϩ T⌬S, where ⌬Gbind is the binding free energy in solution,
⌬EMM is the molecular mechanics energy, which is composed of a van der
Waals and an electrostatic contribution, and ⌬GSol is the solvation energy,
which consists of electrostatic and nonpolar interactions. Entropy could be
calculated by either normal-mode analysis or quasi-harmonic approximation.
Estimation of entropy has been shown to be computationally expensive with a
low prediction accuracy. In this study, T⌬S was not considered for the
calculation, as we were mainly interested in protein flexibility and the relative
energy contributions of amino acids toward binding. ⌬EMM was calculated by
the following equation: ⌬EMM ϭ ⌬Eint_elec ϩ ⌬Eint_vdW, where ⌬Eint_elec and
⌬Eint_vdW are electrostatic and van der Waals interaction energies between
substrates and CYP102A1 mutants. These energies were computed with no
Spectral Binding Titrations. Spectral binding titrations were used to
determine dissociation constants (Ks) for substrates as described previously
(Kim et al., 2008a). Binding affinities of ligands to the CYP102A1 enzymes
were determined (at 23°C) by titrating 1.5 ␮M enzyme with the ligand in a
total volume of 1.0 ml of 100 mM potassium phosphate buffer (pH 7.4).
Final CH3CN concentrations were Ͻ2% (v/v). Spectral dissociation con-
stants (Ks) were estimated using GraphPad Prism software. Unless the
estimated Ks was within 5-fold of the P450 concentration, nonlinear re-
gression analysis was applied using the hyperbolic equation ⌬A ϭ Bmax[L]/
(Ks ϩ [L]), where A is the absorbance difference, Bmax is the maximum
absorbance difference extrapolated to infinite ligand concentration, and [L]
is the ligand concentration.
Molecular Dynamics and Docking Simulation. Coordinates of the
CYP102A1 structure (Protein Data Bank code 2UWH) (Huang et al., 2007)
from B. megaterium were taken from the Protein Data Bank. Crystallographic
water molecules were removed from all of the structures. The structure 2UWH
contains the mutated residue Phe82, which was corrected to Ala82. The
rectified CYP102A1 was then mutated to CYP102A1 mutant 16 (R47L/F81I/
F87V/E143G/L188Q/E267V) and CYP102A1 mutant 17 (R47L/E64G/F81I/
F87V/E143G/L188Q/E267V) using MODELLER (Sali and Blundell, 1993).
The substrates lovastatin (LOV) and simvastatin (SIM) were prepared for
molecular dynamics simulation with CYP102A1 mutants. The substrates were
then applied with partial atomic charges derived by fitting ESP obtained by
quantum electronic structure calculation. The substrates were geometry-opti-
mized with the 6-31G*(p,d) basis set by the B3LYP method using the Gaussian
03 program (Gaussian Inc., Wallingford, CT). The Antechamber module was
used to calculate the atom-centered restrained electrostatic surface potential by
fitting ESPs estimated from wave functions (Bayly et al., 1993). Parameters
that were missing for the ligands were generated using the general amber force
field and the Parmchk module of Antechamber (Wang et al., 2004). Docking
was performed using AutoDock (Morris et al., 1998) to determine the correct
position for the substrates in place of palmitate. The best position was selected
on the basis of the root-mean-square deviation and binding free energy. After
applying the force field, all hydrogen atoms were automatically added to all
systems. Structures were solvated with a TIP3P water model by creating an
isometric water box, in which the distance of the box is 8.5 Å from the
periphery of the protein (Jorgensen et al., 1983). Molecular systems were
neutralized by the addition of counterions. Systems were then energy-mini-
mized in two steps. In the first step, the CYP102A1 and substrates were kept
fixed, and only the water molecules were allowed to move; in the second step,
cutoff. The solvation energy ⌬GSol can be divided into two parts: ⌬GSol
ϭ
⌬Gsol_elec ϩ ⌬Gsol_npol. The electrostatic contribution to the solvation free
energy ⌬Gsol_elec was calculated using the generalized Born model (Gohlke
and Case, 2004). The hydrophobic contribution to the solvation energy ⌬Gsol_npol
was determined with a function of the solvent accessible surface area by the
following equation: ⌬Gsol_npol ϭ ␥SASA ϩ b, where ␥ and b are empirical
constants. The ␥ value was set to 0.0072 kcal/mol/Å2, whereas b was set to a
default value of 0. SASA is the solvent accessible surface area estimated using
the linear combination of pairwise overlaps model. Snapshots from 0.1 to 5 ns
molecular dynamics (MD) simulation trajectories were taken for the calcula-
tion of free energy. For PBSA and generalized Born surface area calculations,
dielectric constants for solute and solvent were taken as 1.0 and 80.0, respec-
tively. SASA was calculated using the molsurf program implemented in
AMBER 10 by taking the solvent probe as 1.4 Å. Free energy was decomposed
using MM/PBSA of AMBER 10 to estimate the contribution of each residue
in the binding process.
Results
Oxidation of Simvastatin and Lovastatin by Wild-Type P450
BM3 and Its Mutants. We examined whether CYP102A1 can oxi-
dize simvastatin and lovastatin. First, the ability of wild-type and a set
of CYP102A1 mutants to oxidize simvastatin and lovastatin was
measured at a fixed substrate concentration (100 ␮M) (Fig. 3).
Whereas simvastatin and lovastatin are known to produce at least four
and three metabolites, respectively (Figs. 1 and 2), CYP102A1 mu-
tants produced only two metabolites, one major (6Ј␤-OH statin)
metabolite and one minor (6Ј-exomethylene statin) metabolite. The
metabolites were analyzed by HPLC and compared with those of
human CYP3A4 (Figs. 4 and 5). Catalytically active mutants of
all atoms were allowed to move. For the first step, the energy minimization CYP102A1 produced one major product, which had a retention time
was performed in 500 and 1500 steps using the steepest descent and conjugate
gradient methods, respectively. For the second step, the energy minimizations
were performed in 1000 and 1500 steps using the steepest descent and
conjugate gradient methods, respectively. Heating was performed with an
NVT ensemble for 10 ps; the CYP102A1 mutants complexed with substrates
were restrained with a force constant of 10 kcal/mol/Å2. Equilibration was
performed for 50 ps on an NPT ensemble restraining the CYP102A1 mutants/
substrates complex by 1 kcal/mol/Å2. Final simulations, i.e., the production
phase, were performed for 5 ns on an NPT ensemble at a temperature of 310
K and 1 atm of pressure. Step size was 2 fs for the entire simulation. A
that exactly matched the 6Ј␤-OH statin and the minor metabolite
6Ј-exomethylene product. However, other products were not observed
under the experimental conditions used here.
The turnover numbers for the oxidation of statins (product forma-
tion) by the entire set of 18 mutants varied over a wide range.
CYP102A1 wild-type showed very low catalytic activity toward sim-
vastatin (0.0049 minϪ1 for 6Ј␤-OH simvastatin), and no other metab-
olites were observed for either substrate. (Fig. 3). Mutant 4 did not
show any apparent activity toward either statin (Ͻ 0.001 minϪ1).
Langevin thermostat and barostat was used for temperature and pressure Mutants 16 and 17 each showed higher activity than that of human

144
KIM ET AL.
FIG. 3. Rates of 6Ј␤-OH and 6Ј-exometh-
ylene product formation of simvastatin (A)
and lovastatin (B) by various CYP102A1
mutants. Assays were performed using 100
␮M simvastatin or lovastatin as described
under Materials and Methods. Values are
the mean Ϯ S.D. of triplicate determina-
tions.
CYP3A4. In the case of mutant 17, the turnover number (10 minϪ1
)
(Supplemental Figs. S3 and S4) with those of authentic compounds,
was 3.3- and 2040-fold higher than that of human CYP3A4 and which were produced by human CYP3A4. The production of 6Ј␤-OH
wild-type CYP102A1, respectively. and 6Ј-exomethylene statins by CYP102A1 mutants was confirmed
Characterization of the Chiral Metabolites of Simvastatin and by LC-MS analysis of the reaction mixture. The retention time and
Lovastatin. The identities of the major metabolite and substrate were fragmentation pattern of the CYP102A1 metabolites were exactly
verified by comparing the HPLC results (Figs. 4 and 5), the LC-MS matched to those of the authentic metabolites, which were produced
results (Supplemental Figs. S1 and S2), and the absorption spectra by human CYP3A4 (Supplemental Figs. S1 and S2).
FIG. 4. HPLC chromatograms of simvastatin metabolites produced by CYP102A1 mutants (13–17) and human CYP3A4. Peaks were identified by comparing the retention
times with those of metabolites produced by CYP3A4. The peaks of the substrate and two major products (hydroxylated product and exomethylene product) are indicated.
UV absorbance was monitored at 238 nm.

OXIDATION OF SIMVASTATIN AND LOVASTATIN BY CYP102A1 MUTANTS
145
FIG. 5. HPLC chromatograms of lovastatin metabolites produced by CYP102A1 mutants (13–17) and human CYP3A4. Peaks were identified by comparing the retention
times with those of substrate and authentic metabolites produced by CYP3A4. The peaks of the substrate and two major products are indicated. UV absorbance was
monitored at 238 nm.
Chemical structures of 6Ј␤-OH statins were identified by one- and simvastatin and lovastatin, respectively, in the case of mutant 17.
two-dimensional NMR experiments. Spectral assignments were ac-
Although other products were found to be metabolites of human liver
microsomes (Caron et al., 2007 and references therein), all of the
CYP102A1 enzymes, including wild-type and active mutants, pro-
duced only two metabolites, 6Ј␤-OH and 6Ј-exomethylene (Figs. 4
and 5). Human CYP3A4, the major enzyme for the hydroxylation
reactions of simvastatin and lovastatin in human liver, also shows a
preference for the 6Ј␤-hydroxylation reaction over the dehydrogena-
tion reaction at the 6Ј position (Figs. 4 and 5).
Kinetic Parameters of Simvastatin and Lovastatin Oxidation by
CYP102A1 Mutants. The two high-activity mutants 16 and 17 were
used to measure kinetic parameters for the 6Ј␤-hydroxylation of
simvastatin and lovastatin (Table 1; Supplemental Fig. S6). Wild-type
and some mutant forms of CYP102A1 did not exhibit any or appre-
ciable activities to determine reliable kinetic parameters (Fig. 3). Only
mutants 16 and 17 showed significantly elevated kcat values for the
6Ј␤-hydroxylation reaction, which yielded the highest kcat values of
Ͼ10 minϪ1 for simvastatin and lovastatin (Fig. 3). Human CYP3A4
exhibited kcat values of 6.6 and 4.2 minϪ1 for 6Ј␤-hydroxylation of
simvastatin and lovastatin, respectively. The overall range of Km
values for the CYP102A1 mutants was from 37 to 44 ␮M. Of note,
human CYP3A4 displayed a higher Km value of 130 ␮M for the
6Ј␤-hydroxylation of simvastatin. The catalytic efficiencies (kcat/Km)
1
complished with two-dimensional H NMR and 13C NMR spectros-
copy along with two-dimensional NMR spectroscopic techniques. The
results are shown in Supplemental Fig. S5 and Supplemental Tables
S2 to S5. The stereochemical configurations of the 6Ј␤-OH position
for both compounds were determined with one-dimensional nuclear
Overhauser effect spectroscopy. There have been reports for proton
chemical shifts in 6Ј␤-OH lovastatin in different NMR solvents (ace-
tonitrile-d4) (Vyas et al., 1990), and, to our knowledge, there have
been no publications of C13 chemical shift data for 6Ј␤-OH simva-
statin and 6Ј␤-OH lovastatin. The stereochemical configurations at the
6Ј␤-OH position of both compounds were determined with one-
dimensional double-pulsed field gradient spin echo nuclear Over-
hauser effect. When previously assigned proton peaks from H-6Ј␣ (␦
1.35 ppm for 6Ј␤-OH simvastatin and ␦ 1.34 ppm for 6Ј␤-OH lova-
statin) were selectively irradiated, proton NMR peaks from H-7Ј(eq)
(␦ 2.44 ppm for 6Ј␤-OH simvastatin and ␦ 2.43 ppm for 6Ј␤-OH
lovastatin) appeared out of two H-7Ј protons. Subsequent selective
irradiation of H-8Ј using a one-dimensional double-pulsed field gra-
dient spin echo nuclear Overhauser effect experiment confirmed the
6Ј␤-OH position of both compounds (Supplemental Figs. S5, E and F).
Simvastatin and lovastatin proved to be good substrates for
CYP102A1 enzymes, exhibiting high turnover numbers that ap-
proached 10 and 18 minϪ1 for 6Ј␤-OH product formation from of 6Ј␤-hydroxylation of simvastatin and lovastatin by mutant 17 were
TABLE 1
Kinetic parameters for the formation of the 6Јb-OH hydroxylated product by CYP102A1 mutants
Simvastatin
Lovastatin
P450 Enzyme
kcat
Km
kcat/Km
kcat
Km
kcat/Km
minϪ1
mM
minϪ1/␮M
minϪ1
mM
minϪ1/␮M
Mutant 16
Mutant 17
10 Ϯ 1
15 Ϯ 1
37 Ϯ 6
42 Ϯ 6
0.27 Ϯ 0.05
0.36 Ϯ 0.60
14 Ϯ 1
19 Ϯ 2
44 Ϯ 9
41 Ϯ 9
0.32 Ϯ 0.07
0.46 Ϯ 0.11
Human CYP3A4
6.6 Ϯ 0.9
130 Ϯ 30
0.051 Ϯ 0.014
4.2 Ϯ 0.5
54 Ϯ 15
0.078 Ϯ 0.024

146
KIM ET AL.
affinities of human CYP3A4 toward the simvastatin (Ks ϭ 14 Ϯ 1
␮M) and lovastatin (Ks ϭ 8.8 Ϯ 0.8 ␮M) substrates were determined
from the titration curves. However, in the case of the CYP102A1
mutants, the addition of the substrates tested here did not result in any
apparent spectral changes (data not known).
Modeling and Computational Calculation. In this study, we have
analyzed interactions in a dynamic environment that depend on time
to derive more insight into molecular recognition. Interactions were
analyzed on the basis of free energy decomposition and hydrogen-
bond analysis. For each complex, free energy was decomposed to
analyze the contributions of individual residues in binding. The hy-
drogen bonds that showed stable behavior during molecular dynamics
were analyzed. To fully investigate the influence of the CYP102A1
mutation on the interaction with substrates, the substrate-residue in-
teractions in each wild-type complex and the corresponding mutated
complex were also compared systematically.
FIG. 6. Total turnover numbers for 6Ј␤-OH product formation by CYP102A1
mutants. To determine the total turnover number of several CYP102A1 mutants, we
used 1.0 mM simvastatin or lovastatin. The reaction was initiated by the addition of
an NADPH-generating system and incubation for 4 h at 37°C. The formation rate of
metabolites was determined by HPLC as described in the legends to Figs. 4 and 5.
Docking and Molecular Dynamics Simulation Analysis. Binding
features observed in the simulation of the substrates docked into the
wild-type CYP102A1 (CYP102A1 WT) are schematically repre-
sented (Fig. 7, A and B) together with interatomic distances for the
0.36 and 0.46 minϪ1 ꢂ ␮MϪ1, which are more efficient than those of 6Ј␤-carbon of the substrates and the Fe atom of the heme. Coordina-
human CYP3A4 by 7- and 6-fold, respectively.
tion of the 6Ј␤-carbon of the substrates to the catalytic iron atom was
When the total turnover numbers (moles of product per mole of maintained for the entire simulation time, and none of the oxygen
catalyst) for the CYP102A1 mutants were determined, the overall atoms in the substrates formed a stable hydrogen bond to any residues
range was 150 to 210 (Fig. 6). Mutant 16 showed the highest activity, of CYP102A1 WT, suggesting that the substrate oxygen atoms are not
which was 5- to 6-fold higher than that of human CYP3A4 after a 4-h the primary determinant; this situation for binding of LOV/SIM into
incubation period. The wild-type CYP102A1 enzyme showed much CYP102A1 WT is unlike the case of other CYP102A1 mutant com-
lower catalytic activity toward simvastatin and lovastatin (0.77 and plexes. Because no direct hydrogen bond between CYP102A1 WT
1.9 nmol of product/nmol of P450, respectively).
and LOV/SIM was retained during the energy minimization stage of
Inhibitory Effects of ␣NF on the CYP102A1-Catalyzed Oxida- the initial complex model, the distance of the 6Ј␤-carbon of the
tion of Simvastatin and Lovastatin. Modulation of the catalytic substrates moved away from the catalytic iron atom during the equi-
activities of human CYP3A4 by ␣NF is known (Ueng et al., 1997). In librium stage of the MD simulation (8–9 Å). In the subsequent data
this work, we examined the effect of ␣NF on the catalytic activities of sampling stage, a direct hydrogen bond was also not observed. Sig-
CYP102A1 mutants that possessed human CYP3A4 activities. ␣NF nificantly, the movement of the substrates from their original position
inhibited the 6Ј␤-hydroxylation of simvastatin and lovastatin in a at the catalytic sites to the channel entrance was observed in the MD
concentration-dependent manner. When 50 ␮M ␣NF was added to simulation; this movement was found to be due to a collision between
incubated reaction mixtures with simvastatin and lovastatin (Supple- the fused ring connected to the oxooxane ring of LOV/SIM and
mental Fig. S7), the 6Ј␤-hydroxylation activities of human CYP3A4 flexible or large hydrophobic residues, such as Leu75 and Phe87. The
for simvastatin and lovastatin were inhibited by 42 and 69%, respec- simulations indicated that that the butanonate chain and the oxooxane
tively. The 6Ј␤-hydroxylation activities of CYP102A1 mutants 16 and ring could not fit into the space created by the Thr49, Met185, and
17 for simvastatin were inhibited by 44 and 77%, respectively. The Leu437 shift without inducing steric repulsion. Thus, a small increase
activities of CYP102A1 mutants 16 and 17 for lovastatin were inhib- in the hydroxylation potency for SIM caused by introducing a methyl
ited by 48 and 76%, respectively, in the presence of ␣NF. This result group on the LOV was only ascribed to the induced fit of SIM. Both
shows that ␣NF can bind to the active site of CYP102A1 mutants to simulations yield nearly identical mean values for the distance be-
change their catalytic activities.
tween the 6Ј␤-carbon of the substrates and the Fe iron, 9.07 and 8.62
A triple CYP102A1 mutant 10 (R47L/F87V/L188Q) has been Å for LOV and SIM (Fig. 7, A and B), respectively, which are
reported to metabolize typical mammalian P450 substrates, such as comparable to those observed in the experimental results described
amodiaquine, dextromethorphan, acetaminophen, testosterone, and above.
3,4-methylenedioxymethylamphetamine (van Vugt-Lussenburg et al.,
The replacement of several residues [mutant 16 (R47L/F81I/F87V/
2007). Although product formation from these chemicals by E143G/L188Q/E267V) and mutant 17 (R47L/E64G/F81I/F87V/
CYP102A1 mutant 10 was inhibited from 30 to 60% in the presence E143G/L188Q/E267V)] that led to quite potent monooxygenase ac-
of ␣NF, ␣NF did not have a significant effect on the metabolism of tivity induced significant changes in the overall shape of the catalytic
acetaminophen and 3,4-methylenedioxymethylamphetamine (van sites of CYP102A1, which were determined by the conformations of
Vugt-Lussenburg et al., 2007). This mutant could also oxidize sim- their side chains (Supplemental Fig. S9). The docking of substrates
vastatin and lovastatin to generate 6Ј␤-OH and 6Ј-exomethylene into CYP102A1 mutants (16 and 17) revealed that coordination of the
products at relatively low rates (6Ј␤-OH simvastatin, 0.046 Ϯ 0.003 6Ј␤-carbon of the substrates to the iron atom is possible only when a
minϪ1; 6Ј-exomethylene simvastatin, 0.00028 Ϯ 0.00068 minϪ1
6Ј␤-OH lovastatin, 0.22 Ϯ 0.17 minϪ1; and 6Ј-exomethylene lova- subsite around the heme. The simulations for CYP102A1 WT and its
statin, 0.057 Ϯ 0.008 minϪ1) (Fig. 3).
mutant complexes showed a significant difference in the buried sur-
;
hydrogen bond is made in the shallow cleft that acts as a catalytic
Titration. The addition of simvastatin and lovastatin to a solution face areas at the interface between the substrates and CYP102A1,
with human CYP3A4 produced a typical low- to high-spin conversion suggesting that interaction of the hydrophobic residues in the catalytic
(type I difference spectrum) (Supplemental Fig. S8). The binding subsites with the substrates is very extensive and consequently makes

OXIDATION OF SIMVASTATIN AND LOVASTATIN BY CYP102A1 MUTANTS
147
FIG. 7. Molecular binding surface area and binding site structural comparison of lovastatin and simvastatin in wild-type CYP102A1 and its mutants (16 and 17); the average
MD structures of WT/lovastatin (A), WT/simvastatin (B), mutant #16/lovastatin (C), mutant #16/simvastatin (D), mutant #17/lovastatin (E), and mutant #17/simvastatin
(F) are shown. The substrate, heme, and active sites are shown as stick diagrams.
an important contribution to the destabilization of substrate binding. possibly plays an important role in enhancing the affinity of the
We speculate that the mutation of CYP102A1 causes conformational oval-like pocket of CYP102A1 mutant 16. However, the hydroxyl of
changes in the active site. When coupled with structural information, Ser72 adopted a different conformation in the SIM complex with
these conformational changes can be used to interpret the experimen- CYP102A1 mutant 16 and failed to form a hydrogen bond with the
tal results.
hydroxyl group of SIM; it is present in only 17.0% of the computed
Hydrogen bonds play an essential role in stabilizing the complexes snapshots (Fig. 7D), indicating that SIM barely interacts with Ser72 in
of the CYP102A1 mutants (16 and 17)-LOV/SIM complexes. We the CYP102A1 mutant 16 active site model. The hydrophobic inter-
examined the geometry and stability of the hydrogen bond network action between the methyl butanoate group and Ile263/Ala264 was
formed by LOV/SIM in the binding pocket of mutants 16/17 on the shown to contribute to the complex stabilization, whereas the hydro-
basis of the trajectories of the MD simulations. Conformational gen bond to Ser72 was frequently observed to be cleaved in the
changes in the binding pocket at the active site caused by several simulation, giving a longer average distance than that for the complex
mutations affect the LOV and SIM binding (Fig. 7, C–F). Molecular with LOV.
surface visualization showed that there are relatively deep pockets in
In the model structure of CYP102A1 mutant 17 complex with
the CYP102A1 mutants. Interactions of these pockets with the oxoox- LOV, Ala330 and Leu437 make a hydrogen bond and an electrostatic
ane moiety of LOV/SIM could enhance their binding ability. The most interaction through their backbone carbonyl group with a hydroxyl
significant hydrogen bond contacts between the CYP102A1 mutants group and carbonyl group located in the oxooxane ring, respectively
and LOV/SIM were schematically identified and characterized in (95.3% occupied for Leu437 and 91.6% occupied for Ala330). These
terms of distances between heavy atoms and their percentage of interactions may stabilize the binding of LOV to CYP102A1 mutant
occurrence.
17. The averaged distances along the 5-ns MD trajectory between the
The hydroxyl group of the oxooxane ring in LOV complexed with 6Ј␤ carbon atom of LOV and the Fe atom of the heme is less than 2.77
CYP102A1 mutant 16 mainly participates in a hydrogen bond with Å (Fig. 7E). For the SIM complex with CYP102A1 mutant 17, no
the Ser72 hydroxyl group (87.3% occupied of the bonds being main- interaction exists between the main chain oxygen of the Ala330
tained during the MD simulation) (Fig. 7C). This hydrogen bond backbone and the carbonyl group of the oxooxane ring in SIM (Fig.

148
KIM ET AL.
7F), but another hydrogen bond forms between the main chain oxygen binding free energy. The calculated ⌬Gbind for the mutant 17/LOV
of the Leu437 backbone and the hydroxyl group of the SIM oxooxane complex (Ϫ19.49 kcal/mol) was the most negative among the ⌬Gbind
ring (86.0% occupied). This hydrogen bond may also result in the values for all complexes. This result agrees with the previous exper-
different conformation for bound SIM (Fig. 7F).
imental results, which showed that the energetic analysis performed
The hydrogen bond and the electrostatic interaction in the by MM-PB(generalized Born)SA is applicable to the current case
CYP102A1 mutant 17/LOV make significant contact with the sub- study.
strate compared with complexes of mutant 16/LOV, mutant 17/SIM,
The binding free energy of the mutant complexes 16-LOV, 17-
and mutant 16/SIM. According to our newly described binding mode, LOV, 16-SIM, and 17-SIM increased by Ϫ12.75, Ϫ16.76, Ϫ9.01, and
the carbonyl group of the Ala330 backbone forms a strong electro- Ϫ10.56 kcal/mol with respect to the corresponding WT-substrate
static interaction with the carbonyl group of the oxooxane ring in complex, which indicates that the two mutants 16 and 17 bind more
LOV, whereas a similar interaction does not exist in other complexes. strongly to LOV and SIM than WT does. This result also suggests that
The difference in the hydrogen bonding and electrostatic interaction mutants 16 and 17 exhibit substrate acceptance for LOV and SIM. The
reveals that LOV binds most strongly with CYP102A1 mutant 17, free energy components between the WT complex and each mutant
which explains the significantly higher catalytic activity of the complex were compared to elucidate the mechanism driving substrate
CYP102A1 mutant 17/LOV complex compared with that of other acceptance. The differences in the nonpolar solvation energies among
complexes. The large changes at position Leu437 are expected to the six complexes are relatively small, which indicates that better
account for the reduction in favorable interaction of LOV and SIM cavity packing is retained in the mutated complexes. There is a
with CYP102A1 mutant 16 and SIM with CYP102A1 mutant 17 decrease in electrostatic and van der Waals energies in the WT
compared with interaction of LOV with CYP102A1 mutant 17.
complex of approximately 35 to 70 and 15 to 22 kcal/mol for the LOV
Of interest, in CYP102A1 mutant 16 complexed with LOV and and SIM complexes relative to the respective CYP102A1 mutants,
SIM, the main chain oxygen atoms of Ala330 and Leu437 lose the which should be essential to the rejection of CYP102A1 WT for LOV
hydrogen bond and the electrostatic interaction compared with the and SIM.
CYP102A1 mutant 17 complexes. There is a particularly large dif-
Intermolecular van der Waals and electrostatic interactions are both
ference in orientations of the side chains of Ala330 and Leu437. important contributions to substrate binding. Nonpolar solvation terms,
Rotations occur at the side chains of Ala330/Leu437, which are which correspond to the burial of solvent accessible surface area upon
correlated with changes in the distances of the 6Ј␤-carbon in the binding of substrates to the active sites of mutants, contribute slightly
substrates from the heme iron. This alteration in the side-chain ori- favorably. The gas-phase electrostatic values, ⌬Eint_elec, for the four
entation of the active site residues also implies different interactions complexes show that electrostatic interactions are in favor of the binding;
between the substrates and the CYP102A1 mutants. Simulations also however, the overall electrostatic interactions energies, ⌬GSol (⌬Eint_elec ϩ
showed that the butanoate moiety did not form any hydrogen bonds to ⌬Gsol_elec) are positive and unfavorable for the binding, which is caused
the enzyme. Thus, the oxooxane ring can play an important role in by the large desolvation penalty for charged and polar groups that is not
binding the substrate not only by forming the hydrogen bond and the sufficiently compensated for upon complex formation. Comparing the
electrostatic interaction to the enzyme but also by properly directing van der Waals/nonpolar (⌬Eint_vdW ϩ ⌬Gsol_npol) contributions with the
the hydrophobic substituent to the catalytic subsite and enabling it to electrostatic contributions ⌬GElec, we find that the association between
enter the cavity more fully.
LOV and SIM and mutants 16 and 17 is mainly driven by more favorable
Free Energy Calculation. To evaluate the different contributions van der Waals/nonpolar interactions in the complex than in solution. The
to the binding free energy of LOV and SIM with CYP102A1 WT and electrostatic interactions between mutants 16 and 17 and the butanonate
the highly active mutants 16 and 17, absolute binding free energies chain are strong, but the electrostatic interactions between the solvent
were calculated for four complexes using the MM/PBSA method. The (water molecules) and the substrates are much stronger. Thus, when a
investigation into the forces involved in substrate binding can also be substrate transfers from the solvent to the binding pocket, the electrostatic
obtained by analyzing the MM/PBSA free energy contributions, contributions for the butanonate moiety are unfavorable to substrate
which are listed in Table 2 for six complexes. In the present study, we binding. The terminal butanyl chain of LOV and SIM is sandwiched
estimated the different components of interaction energy that contrib- between the side chains of Leu75 and Val78 and makes hydrophobic and
ute to substrate binding. These include van der Waals, electrostatic, van der Waals interactions with the side chains of Ala82 and Val87.
polar solvation, and nonpolar solvation interaction energies. The Overall, this finding suggests that the affinities of the two substrates for
entropy was not calculated as we are only interested in the relative the two mutants 16 and 17 are dominated by shape complementarity, but
TABLE 2
Individual energy components for the calculated binding free energies of lovastatin/simvastatin complexed with CYP102A1 wild-type and its two mutants, mutant 16
and mutant 17
All energies are averaged over 200 snapshots and are given in kilocalories per mole. Calculation of DGbind does not explicitly consider entropy contributions. The values represent the
mean Ϯ S.E.M.
Energy
WT/Lovastatin
WT/Simvastatin
M16/Lovastatin
M16/Simvastatin
M17/Lovastatin
M17/Simvastatin
kcal/mol
DEint_elec
DEint_vdW
DGsol_npol
DGsol_elec
DGSol
Ϫ6.79 Ϯ 0.65
Ϫ4.94 Ϯ 0.18
Ϫ0.55 Ϯ 0.17
9.55 Ϯ 0.25
9.76 Ϯ 0.16
2.76 Ϯ 0.02
Ϫ2.73 Ϯ 0.23
Ϫ7.20 Ϯ 0.28
Ϫ4.14 Ϯ 0.18
Ϫ0.62 Ϯ 0.04
9.05 Ϯ 0.19
8.42 Ϯ 0.25
1.84 Ϯ 0.06
Ϫ2.92 Ϯ 0.04
Ϫ38.48 Ϯ 0.26
Ϫ27.01 Ϯ 0.19
Ϫ3.16 Ϯ 0.02
54.17 Ϯ 0.23
55.34 Ϯ 0.24
15.68 Ϯ 0.21
Ϫ15.48 Ϯ 0.24
Ϫ80.12 Ϯ 0.57
Ϫ19.89 Ϯ 0.19
Ϫ4.57 Ϯ 0.01
92.65 Ϯ 0.43
88.07 Ϯ 0.42
12.54 Ϯ 0.22
Ϫ11.93 Ϯ 0.24
Ϫ48.03 Ϯ 0.28
Ϫ28.64 Ϯ 0.17
Ϫ4.16 Ϯ 0.09
60.33 Ϯ 0.22
56.18 Ϯ 0.27
12.31 Ϯ 0.15
Ϫ19.49 Ϯ 0.18
Ϫ73.92 Ϯ 0.43
Ϫ19.54 Ϯ 0.21
Ϫ3.66 Ϯ 0.02
83.64 Ϯ 0.38
79.99 Ϯ 0.31
9.72 Ϯ 0.27
DGElec
DGbind
Ϫ13.48 Ϯ 0.23
M16, mutant 16, R47L/F81I/F87V/E143G/L188Q/E267V; M17, mutant 17, R47L/E64G/F81I/F87V/E143G/L188Q/E267V; DGSol, polar/nonpolar (DGsol_elec ϩ DGsol_npol) contribution; DGElec
,
electrostatic (DEint_elec ϩ DGsol_elec) contribution.

OXIDATION OF SIMVASTATIN AND LOVASTATIN BY CYP102A1 MUTANTS
149
the total affinity arises from a more complex interplay between all of Several CYP102A1 mutants were generated by directed evolution
these energy components. The structural analysis demonstrates that the and rational design and were found to have much higher activity and
hydrophobic group in the upper part of the heme takes advantage of the stability than human P450 enzymes. These mutants could thus be
developed as industrial enzymes for cost-effective and scalable pro-
duction of human drug metabolites. The metabolites could also be
developed as lead compounds for new drugs.
favorable van der Waals and hydrophobic interactions with the binding
pocket of CYP102A1 mutants. In the binding pocket, Leu75, Val78,
Ala82, Val87, Ile263, and Leu437 residues constitute a relatively large
hydrophobic core, which can generate strong van der Waals and hydro-
phobic interactions with the fused ring of the substrates. The ⌬Eint_vdW
value for the complex of mutant 17 with LOV is the highest among the
complexes of mutants 16 and 17 with the two substrates because lova-
statin fits more snugly within the active site cavity and binds more tightly
to mutant 17 than to the other complexes by adding nonpolar packing to
the active site to a much greater extent.
Our computational findings suggest that a conformational change in
the cavity size of the mutant active sites is related to the activity
change. The modeling results further suggest that the changes in
activity result from the movement of several specific residues in the
mutant active sites (Fig. 7; Supplemental Fig. S9). The computational
results obtained from highly active mutants with the statins are con-
sistent with those of previous work for CYP102A1 mutants with
phenacetin/7-ethoxyresorufin (Kim et al., 2010).
We also found good correlation between in silico analyses and in
vitro determination of kcat values of CYP102A1 mutants. Mutant 17
showed higher kcat values than mutant 16 toward lovastatin and
simvastatin (Table 1). Calculated binding free energies of mutant
17/SIM and mutant 17/LOV complexes were greater than those of
mutant 16/SIM and mutant 16/LOV complexes, respectively (Table
2). When the average distances between the 6Ј␤ carbon atom of statins
and the Fe atom were compared, mutant 17 showed much shorter
average distances than those of mutant 16. However, we could not see
apparent differences in Km values between the complexes when
experimental error ranges were considered (Table 1).
In summary, this work involved a set of CYP102A1 mutants and
statin substrates, which are human P450 substrates, and revealed that
bacterial CYP102A1 enzymes catalyze the same reactions as human
CYP3A4 to generate the human metabolites 6Ј␤-OH and 6Ј-exom-
ethylene. The oxidation of simvastatin and lovastatin is catalyzed by
wild-type and some mutant forms of CYP102A1. One major hydroxy-
lated product, the 6Ј␤-OH product, was produced as a result of a
hydroxylation reaction. The other product, the exomethylene product,
was produced by a dehydrogenation reaction. Metabolite formation
was confirmed by HPLC and LC-MS by comparing the metabolites
with the authentic human metabolites produced by human CYP3A4.
Chemical structures of the major 6Ј␤-OH products were identified by
NMR. Thus, the CYP102A1 mutants efficiently produce human me-
tabolites of simvastatin and lovastatin.
Discussion
In this work, we have shown that bacterial CYP102A1 mutant
enzymes catalyze the same reactions as human CYP3A4 to generate
the human metabolites 6Ј␤-OH and 6Ј-exomethylene from simvasta-
tin and lovastatin. To our knowledge, the production of metabolites of
simvastatin and lovastatin by chemical synthesis has never been
reported. The 6Ј␤-OH products are chiral compounds, which are
usually difficult to synthesize by chemical methods. Therefore, an
alternative to chemical synthesis of these metabolites is to use
CYP102A1 enzymes to generate the metabolites of simvastatin and
lovastatin. Of interest, an unknown P450 from Nocardia autotropica
has been found to oxidize simvastatin to generate 6Ј␤-hydroxymethyl
simvastatin (Gbewonyo et al., 1991).
Human P450 enzymes are involved in the metabolism of most
(Ͼ80%) drugs currently available on the market (Guengerich, 2003).
We proposed previously that CYP102A1 from B. megaterium could
be developed as a biocatalyst with drug oxidation activities compa-
rable to those of human P450 (Yun et al., 2007 and references
therein). In recent studies, several lines of evidence supporting our
proposal have been reported. CYP102A1 mutants were found to
generate human drug metabolites by oxidizing various drugs, includ-
ing clozapine (Damsten et al., 2008), diclofenac (Damsten et al.,
2008), acetaminophen (Damsten et al., 2008), 3,4-methylenedioxym-
ethylamphetamine (Stjernschantz et al., 2008), dextromethorphan
(Stjernschantz et al., 2008), phenacetin (Kim et al., 2010), verapamil
(Sawayama et al., 2009), and astemizole (Sawayama et al., 2009).
CYP102A1 mutants can also oxidize several human P450 substrates,
including 7-ethoxycoumarin (Kim et al., 2008b), coumarin (Park et
al., 2010), chlorzoxazone (Park et al., 2010), and resveratrol (Kim et
al., 2009), to generate human metabolites.
Authorship Contributions
Participated in research design: K.H. Kim, D. Kim, Jung, Pan, Ahn, and
Yun.
Conducted experiments: K.H. Kim, Kang, D.H. Kim, Su.H. Park, Se.H.
Park, K.D. Park, and Lee.
Performed data analysis: K.H. Kim, Kang, D.-H. Kim, Su.H. Park, Se.H.
Park, D. Kim, K.D. Park, Lee, Jung, Pan, Ahn, and Yun.
Wrote or contributed to the writing of the manuscript: D. Kim, K.D. Park,
The issue of human MIST as outlined by the FDA has presented a
challenge at the early stages of drug development for the pharmaceu-
tical industry (Guengerich, 2009 and references therein). Metabolites
of concern should be prepared by chemical methods or with biocata- and Yun.
lysts, such as bacteria, yeast, and enzymes. Human drug metabolites
References
are required for studies of drug toxicity, especially to satisfy the MIST
Bayly CI, Cieplak P, Cornell WD, and Kollman PA (1993) A well-behaved electrostatic potential
based method using charge restraints for deriving atomic charges: the RESP Model. J Phys
Chem 97:10269–10280.
Caron G, Ermondi G, and Testa B (2007) Predicting the oxidative metabolism of statins: an
application of the MetaSite algorithm. Pharm Res 24:480–501.
Case DA, Darden TA, Cheatham TE III, Simmerling CL, Wang J, Duke RE, Luo R, Crowley M,
Walker RC, Zhang W, et al. (2008) AMBER 10, University of California, San Francisco.
Damsten MC, van Vugt-Lussenburg BM, Zeldenthuis T, de Vlieger JS, Commandeur JN, and
Vermeulen NP (2008) Application of drug metabolising mutants of cytochrome P450 BM3
(CYP102A1) as biocatalysts for the generation of reactive metabolites. Chem Biol Interact
171:96–107.
Darden T, York D, and Pedersen L (1993) Particle mesh Ewald: an N ꢂ log(N) method for Ewald
sums in large systems. J Chem Phys 98:10089–10092.
García MJ, Reinoso RF, Sa´nchez Navarro A, and Prous JR (2003) Clinical pharmacokinetics of
statins. Methods Find Exp Clin Pharmacol 25:457–481.
Gbewonyo K, Buckland BC, and Lilly MD (1991) Development of a large-scale continuous
issue, and efficacy during the process of drug discovery and devel-
opment. The “active metabolite” is also an important issue in drug
development (Johnson et al., 2004). In addition to the use of
CYP102A1 mutants for the production of the metabolites of human
P450-catalyzed reactions, we found other advantages to using
CYP102A1 mutants instead of human P450 enzymes. For example,
the human metabolite piceatannol can be generated from the antican-
cer agent resveratrol by CYP102A1 (Kim et al., 2009); however,
piceatannol cannot be made from its substrate, resveratrol, by human
P450 enzymes, as piceatannol is a potent inhibitor for the human
P450-catalyzed reactions.

150
KIM ET AL.
substrate feed process for the biotransformation of simvastatin by Nocardia sp. Biotechnol
Sali A and Blundell TL (1993) Comparative protein modelling by satisfaction of spatial
restraints. J Mol Biol 234:779–815.
Bioeng 37:1101–1107.
Sawayama AM, Chen MM, Kulanthaivel P, Kuo MS, Hemmerle H, and Arnold FH (2009) A
panel of cytochrome P450 BM3 variants to produce drug metabolites and diversify lead
compounds. Chemistry 15:11723–11729.
Gohlke H and Case DA (2004) Converging free energy estimates: MM-PB(GB)SA studies on the
protein-protein complex Ras-Raf. J Comput Chem 25:238–250.
Guengerich FP (2002) Cytochrome P450 enzymes in the generation of commercial products. Nat
Rev Drug Discov 1:359–366.
Guengerich FP (2003) Cytochromes P450, drugs, and diseases. Mol Interv 3:194–204.
Guengerich FP (2009) Introduction: Human metabolites in safety testing (MIST) issue. Chem Res
Toxicol 22:237–238.
Huang WC, Westlake AC, Mare´chal JD, Joyce MG, Moody PC, and Roberts GC (2007) Filling
a hole in cytochrome P450 BM3 improves substrate binding and catalytic efficiency. J Mol
Biol 373:633–651.
Johnson MD, Zuo H, Lee KH, Trebley JP, Rae JM, Weatherman RV, Desta Z, Flockhart DA, and
Skaar TC (2004) Pharmacological characterization of 4-hydroxy-N-desmethyl tamoxifen, a
novel active metabolite of tamoxifen. Breast Cancer Res Treat 85:151–159.
Jorgensen WL, Chandrasekhar J, Madura JD, Impey RW, and Klein ML (1983) Comparison of
simple potential functions for simulating liquid water. J Chem Phys 79:926–935.
Julsing MK, Cornelissen S, Bu¨hler B, and Schmid A (2008) Heme-iron oxygenases: powerful
industrial biocatalysts? Curr Opin Chem Biol 12:177–186.
Kim DH, Ahn T, Jung HC, Pan JG, and Yun CH (2009) Generation of the human metabolite
piceatannol from the anticancer-preventive agent resveratrol by bacterial cytochrome P450
BM3. Drug Metab Dispos 37:932–936.
Stjernschantz E, van Vugt-Lussenburg BM, Bonifacio A, de Beer SB, van der Zwan G, Gooijer
C, Commandeur JN, Vermeulen NP, and Oostenbrink C (2008) Structural rationalization of
novel drug metabolizing mutants of cytochrome P450 BM3. Proteins 71:336–352.
Tornio A, Pasanen MK, Laitila J, Neuvonen PJ, and Backman JT (2005) Comparison of
3-hydroxy-3-methylglutaryl coenzyme A (HMG-CoA) reductase inhibitors (statins) as inhib-
itors of cytochrome P450 2C8. Basic Clin Pharmacol Toxicol 97:104–108.
Transon C, Leemann T, and Dayer P (1996) In vitro comparative inhibition profiles of major
human drug metabolising cytochrome P450 isozymes (CYP2C9, CYP2D6 and CYP3A4) by
HMG-CoA reductase inhibitors. Eur J Clin Pharmacol 50:209–215.
Ueng YF, Kuwabara T, Chun YJ, and Guengerich FP (1997) Cooperativity in oxidations
catalyzed by cytochrome P450 3A4. Biochemistry 36:370–381.
Urlacher VB and Eiben S (2006) Cytochrome P450 monooxygenases: perspectives for synthetic
application. Trends Biotechnol 24:324–330.
Vail RB, Homann MJ, Hanna I, and Zaks A (2005) Preparative synthesis of drug metabolites
using human cytochrome P450s 3A4, 2C9 and 1A2 with NADPH-P450 reductase expressed
in Escherichia coli. J Ind Microbiol Biotechnol 32:67–74.
van Gunsteren WF and Berendsen HJC (1977) Algorithms for macromolecular dynamics and
constraint dynamics. Mol Phys 34:1311–1327.
Kim DH, Kim KH, Isin EM, Guengerich FP, Chae HZ, Ahn T, and Yun CH (2008a) Heterol-
ogous expression and characterization of wild-type human cytochrome P450 1A2 without
conventional N-terminal modification in Escherichia coli. Protein Expr Purif 57:188–200.
Kim DH, Kim KH, Kim D, Jung HC, Pan JG, Chi YT, Ahn T, and Yun CH (2010) Oxidation
of human cytochrome P450 1A2 substrates by Bacillus megaterium cytochrome P450 BM3.
J Mol Catal B Enzym 63:179–187.
Kim DH, Kim KH, Kim DH, Liu KH, Jung HC, Pan JG, and Yun CH (2008b) Generation of
human metabolites of 7-ethoxycoumarin by bacterial cytochrome P450 BM3. Drug Metab
Dispos 36:2166–2170.
Massova I and Kollman PA (2000) Combined molecular mechanical and continuum solvent
approach (MM-PBSA/GBSA) to predict ligand binding. Perspect Drug Discov Des 18:113–
135.
Morris GM, Goodsell DS, Halliday RS, Huey R, Hart WE, Belew RK, and Olson AJ (1998)
Automated docking using a Lamarckian genetic algorithm and an empirical binding free
energy function. J Comput Chem 19:1639–1662.
van Vugt-Lussenburg BM, Stjernschantz E, Lastdrager J, Oostenbrink C, Vermeulen NP, and
Commandeur JN (2007) Identification of critical residues in novel drug metabolizing mutants
of cytochrome P450 BM3 using random mutagenesis. J Med Chem 50:455–461.
Vickers S, Duncan CA, Chen IW, Rosegay A, and Duggan DE (1990a) Metabolic disposition
studies on simvastatin, a cholesterol-lowering prodrug. Drug Metab Dispos 18:138–145.
Vickers S, Duncan CA, Vyas KP, Kari PH, Arison B, Prakash SR, Ramjit HG, Pitzenberger SM,
Stokker G, and Duggan DE (1990b) In vitro and in vivo biotransformation of simvastatin, an
inhibitor of HMG CoA reductase. Drug Metab Dispos 18:476–483.
Vyas KP, Kari PH, Pitzenberger SM, Halpin RA, Ramjit HG, Arison B, Murphy JS, Hoffman
WF, Schwartz MS, and Ulm EH (1990) Biotransformation of lovastatin. I. Structure elucida-
tion of in vitro and in vivo metabolites in the rat and mouse. Drug Metab Dispos 18:203–211.
Wang J, Wolf RM, Caldwell JW, Kollman PA, and Case DA (2004) Development and testing of
a general amber force field. J Comput Chem 25:1157–1174.
Yun CH, Kim KH, Kim DH, Jung HC, and Pan JG (2007) The bacterial P450 BM3: a prototype
for a biocatalyst with human P450 activities. Trends Biotechnol 25:289–298.
Yun CH, Yim SK, Kim DH, and Ahn T (2006) Functional expression of human cytochrome P450
enzymes in Escherichia coli. Curr Drug Metab 7:411–429.
Park SH, Kim DH, Kim D, Kim DH, Jung HC, Pan JG, Ahn T, Kim D, and Yun CH (2010)
Engineering bacterial cytochrome P450 (P450) BM3 into a prototype with human P450
enzyme activity using indigo formation. Drug Metab Dispos 38:732–739.
Prueksaritanont T, Subramanian R, Fang X, Ma B, Qiu Y, Lin JH, Pearson PG, and Baillie TA
(2002) Glucuronidation of statins in animals and humans: a novel mechanism of statin
lactonization. Drug Metab Dispos 30:505–512.
Address correspondence to: Dr. Chul-Ho Yun, School of Biological Sciences
and Technology, Chonnam National University, Gwangju 500-757, Republic of
Korea. E-mail: chyun@jnu.ac.kr
Rushmore TH, Reider PJ, Slaughter D, Assang C, and Shou M (2000) Bioreactor systems in drug
metabolism: synthesis of cytochrome P450-generated metabolites. Metab Eng 2:115–125.