N-heterocyclic carbene capture by cytochrome P450 3A4

Article abstract of DOI:10.1124/mol.116.103721

Cytochrome P450 3A4 (CYP3A4) is the dominant P450 enzyme involved in human drug metabolism, and its inhibition may result in adverse interactions or, conversely, favorably reduce the systemic elimination rates of poorly bioavailable drugs. Herein we describe a spectroscopic investigation of the interaction of CYP3A4 with N-methylritonavir, an analog of ritonavir, widely used as a pharmacoenhancer. In contrast to ritonavir, the binding affinity of N-methylritonavir for CYP3A4 is pH-dependent. At pH < 7.4, the spectra are definitively type I, whereas at pH ≥7.4 the spectra have split Soret bands, including a red-shifted component characteristic of a P450- carbene complex. Variable-pH UV-visible spectroscopy binding studies with molecular fragments narrows the source of this pH dependence to its N-methylthiazolium fragment. The C2 proton of this group is acidic, and variable-pH resonance Raman spectroscopy tentatively assigns it a pK_a of 7.4. Hence, this fragment of N-methylritonavir is expected to be readily deprotonated under physiologic conditions to yield a thiazol-2- ylidene, which is an N-heterocyclic carbene that has highaffinity for and is presumed to be subsequently captured by the heme iron. This mechanism is supported by time-dependent density functional theory with an active site model that accurately reproduces distinguishing features of the experimental UV-visible spectra of N-methylritonavir bound to CYP3A4. Finally, density functional theory calculations support that this novel interaction is as strong as the tightest-binding azaheterocycles found in P450 inhibitors and could offer new avenues for inhibitor development.

Full text of DOI:10.1124/mol.116.103721

Supplemental material to this article can be found at:
http://molpharm.aspetjournals.org/content/suppl/2016/06/08/mol.116.103721.DC1.html
1
521-0111/90/1/42–51$25.00
http://dx.doi.org/10.1124/mol.116.103721
Mol Pharmacol 90:42–51, July 2016
MOLECULAR PHARMACOLOGY
Copyright ª 2016 by The American Society for Pharmacology and Experimental Therapeutics
N-Heterocyclic Carbene Capture by Cytochrome P450 3A4 s
Gareth K. Jennings, Caroline M. Ritchie, Lisa S. Shock, Charles E. Lyons,
and John C. Hackett
Department of Physiology and Biophysics and the Massey Cancer Center, Virginia Commonwealth University School of
Medicine, Richmond, Virginia
Received February 26, 2016; accepted April 27, 2016
ABSTRACT
Cytochrome P450 3A4 (CYP3A4) is the dominant P450 enzyme proton of this group is acidic, and variable-pH resonance
involved in human drug metabolism, and its inhibition may Raman spectroscopy tentatively assigns it a pK of 7.4. Hence,
a
result in adverse interactions or, conversely, favorably reduce this fragment of N-methylritonavir is expected to be readily
the systemic elimination rates of poorly bioavailable drugs. deprotonated under physiologic conditions to yield a thiazol-2-
Herein we describe a spectroscopic investigation of the in- ylidene, which is an N-heterocyclic carbene that has high-
teraction of CYP3A4 with N-methylritonavir, an analog of affinity for and is presumed to be subsequently captured by the
ritonavir, widely used as a pharmacoenhancer. In contrast to heme iron. This mechanism is supported by time-dependent
ritonavir, the binding affinity of N-methylritonavir for CYP3A4 is density functional theory with an active site model that accurately
pH-dependent. At pH ,7.4, the spectra are definitively type I, reproduces distinguishing features of the experimental UV-visible
whereas at pH $7.4 the spectra have split Soret bands, spectra of N-methylritonavir bound to CYP3A4. Finally, density
including a red-shifted component characteristic of a P450- functional theory calculations support that this novel interac-
carbene complex. Variable-pH UV-visible spectroscopy bind- tion is as strong as the tightest-binding azaheterocycles found
ing studies with molecular fragments narrows the source of this in P450 inhibitors and could offer new avenues for inhibitor
pH dependence to its N-methylthiazolium fragment. The C2 development.
Ritonavir is a high-affinity, type II P450 ligand that binds to
Introduction
CYP3A4 through hydrophobic interactions dominated by
The cytochrome P450 (P450) enzymes are involved in the
phenyl side chains and coordination of the thiazole nitrogen
biosyntheses of hormones and the disposition of xenobiotics
to the heme iron (Sevrioukova and Poulos, 2010). Structural
(
Guengerich, 2001; Denisov et al., 2005; Ortiz de Montellano,
modifications abolishing the phenyl groups substantially
lower the affinity of these molecules for CYP3A4 and result
in the observation of multiple ligand-binding orientations in
their crystal structures (Sevrioukova and Poulos, 2013; Kaur
et al., 2015). Nonetheless, thiazole coordination to the heme
iron and orientation of the isopropyl thiazole unit near the
protein surface remain consistent features of these structures.
Moreover, CYP3A4 is capable of hydroxylating the isopropyl
thiazole substituent, so the inverted binding mode orienting
this end of the molecule toward the heme is indeed possible,
though it has not been observed (Koudriakova et al., 1998).
Owing to the observed hydroxylation, it has been suggested
that the nitrogen atom in the isopropyl thiazole fragment does
not appreciably coordinate to the heme (Sevrioukova and
Poulos, 2010).
Akin to ritonavir, the majority of heme-coordinating P450
inhibitors are azaheterocycles where nitrogen atoms directly
coordinate to iron; however, there are a few fragments where
the coordinating atom is carbon. In these examples, the
ligands are generally accepted to be carbenes generated by
reductive or oxidative catalysis. For example, the red-shifted
Soret band (450–470 nm) resulting from catalytic reduction of
2
005). Inhibition of the dominant P450 in human drug
metabolism, namely, cytochrome P450 3A4 (CYP3A4), has
the potential to result in adverse drug-drug interactions by
reducing the systemic elimination rates of coadministered
drugs (von Moltke et al., 1998; Hollenberg, 2002; Lin et al.,
2
002; Rendic, 2002). Conversely, pharmacoenhancers ex-
ploit this effect and target CYP3A4 for therapeutic benefit
by overcoming the rapid clearance, high dosing frequency,
and subtherapeutic plasma concentrations characteristic of
CYP3A4 substrates. The most successful application of this
strategy has been applied in the treatment of human immu-
nodeficiency virus and hepatitis C virus infections whereby
ritonavir (Fig. 1) inhibits the clearance of antiviral therapeu-
tics through inhibition of CYP3A4 (Kempf et al., 1995, 1997;
Koudriakova et al., 1998; Hirsch et al., 2008; Klibanov et al.,
2
015).
This work was supported by the National Institutes of Health National
Institute of General Medical Sciences [Grants R01GM092827, R01GM114168]
and the Office of Naval Research [Grant N000141210773].
dx.doi.org/10.1124/mol.116.103721.
This article has supplemental material available at molpharm.
aspetjournals.org.
s
ABBREVIATIONS: CHAPS, 3-[3-(cholamidopropyl)dimethylammonio]-1-propanesulfonate; CYP3A4, cytochrome P450 3A4; DMSO, dimethylsulf-
oxide; HS, high spin; MDP, methylenedioxyphenyl-2-ylidene; MITV, N-[2-Isopropylthiazol-4-ylmethyl(methyl)carbamoyl]-L-valine; NMeR,
N-methylritonavir; NMTI, N-methylthiazolium iodide; P450, cytochrome P450; rR, resonance Raman; SVD, singular value decomposition.
4
2

Thiazol-2-Ylidene Coordination to CYP3A4
43
(
Pittsburgh, PA). All other reagents of the highest available grade
were purchased from Sigma-Aldrich (St. Louis, MO) or Fisher
Scientific (Pittsburgh, PA). N-methylthiazolium iodide (NMTI) was
synthesized as previously described by Kena Diba et al. (2010). To a
solution of thiazole (2 ml, 2.4 g, 28.2 mmol) in 6 ml of methanol,
iodomethane (2.18 ml, 5 g, 35.2 mmol) was added, and the solution was
stirred for 48 hours at room temperature in the dark. The resultant
white crystals were vacuum filtered and washed with methanol
followed by ethyl acetate. The crystals were dried under vacuum
overnight to yield 2.05 g of NMTI (9.05 mmol, 32%). For mass
1
spectrometry, the calculated value for C
00.04 a.m.u. was found.
CYP3A4 Expression and Purification. CYP3A4 lacking resi-
4 6
H NS [M ] is 100.02 a.m.u.,
1
dues 3–23 and modified to have a C terminal 4ꢀHis tag was expressed
in DH5a Escherichia coli cells using the pCW-CYP3A4 plasmid
generated in our laboratory. We used 100 ng of pCW-CYP3A4 plasmid
to transform chemically competent DH5a cells. Two colonies were
used to inoculate 100 ml of terrific broth (100 mg/ml ampicillin), and
they were grown for 16 hours at 37°C. One liter batches of terrific broth
supplemented with 100 mg/ml ampicillin were inoculated with 20 ml of
starter culture and grown at 200 rpm and 37°C until the o.d. at 600 nm
Fig. 1. Structures of ritonavir, NMeR, and MITV. The CYP3A4 heme-
coordinating nitrogen atom is highlighted with a dashed circle, and the site
of hydroxylation is indicated with an arrow. The acidic C2 proton in NMeR
is indicated as labeled.
polyhalogenated methanes has been attributed to ferrous
dihalocarbene complexes (Wolf et al., 1977; Ruf et al., 1984).
The methylenedioxphenyl substituent that is present in reached 0.4–0.6. The incubation temperature was then decreased to
therapeutic drugs (i.e., paroxetine and tadalafil) (Ring et al., 24°C until the o.d. reached 0.8 at which time the rotations were
2
(
005), drugs of abuse (i.e., methylenedioxymethamphetamine) decreased to 150 rpm. The cultures were then supplemented with
0
.5 mM of 5-aminolevulinic acid and induced with 1 mM isopropyl b-D-
Heydari et al., 2004), and insecticide synergists (i.e., piperonyl
butoxide) (Casida, 1970) undergoes NADPH and O -dependent
2
1
-thiogalactopyranoside. After 24 hours the bacteria were harvested
by centrifugation at 4000g for 10 minutes.
oxidation to generate the carbene methylenedioxyphenyl-2-
2
1
ylidene (MDP), which strongly coordinates to the Fe21-heme
The cell pellets were resuspended (1.5 ml∙g cells) in lysis buffer
(
100 mM potassium phosphate, pH 7.4, 500 mM KCl, 0.5% CHAPS,
iron and gives rise to a distinct split Soret band with maxima
at ∼427 and 455 nm (Hodgson and Philpot, 1974).
10% glycerol, and 2 mM b-mercaptoethanol). On ice, the resuspension
was incubated for 45 minutes with lysozyme (1.5 mg/ml) and sub-
sequently sonicated for 15 minutes. The insoluble material was
Our search for a high-affinity type I ligand led us to
N-methylritonavir (NMeR, Fig. 1), an analog wherein pelleted by centrifugation at 30,000g for 1 hour. The supernatant
21
the heme-coordinating thiazole nitrogen has been blocked was loaded onto a Ni -NTA column pre-equilibrated with wash
by methylation. Herein, we report a detailed spectroscopic buffer (100 mM potassium phosphate, pH 7.4, 500 mM KCl, 0.2%
CHAPS, 20 mM imidazole 10% glycerol, and 2 mM b-mercaptoetha-
characterization of the NMeR complex with recombinant
nol). The column was washed with 15 column volumes of wash buffer
CYP3A4. UV-visible absorption and resonance Raman (rR)
and eluted with elution buffer (100 mM potassium phosphate, pH 7.4,
spectroscopy demonstrate that the spin-state populations of
500 mM KCl, 0.2% CHAPS, 100 mM imidazole 10% glycerol, 2 mM
NMeR-CYP3A4 are strongly pH dependent. There is a sub-
stantial high-spin population at low pH that shifts to low spin
at high pH. UV-visible difference spectra of the low-spin
complexes have split Soret bands with a strongly red-shifted
b-mercaptoethanol).
Red-colored fractions were pooled and exchanged into dilution
buffer (25 mM potassium phosphate, pH 6.8, 10% glycerol, and
2
mM b-mercaptoethanol) with several rounds of concentration and
component similar to those observed in the spectra of carbene dilution using Amicon centrifugal concentrators (MWCO 30 kDa).
complexes. Recognizing the acidity of the C2 carbon of Buffer-exchanged protein was then loaded onto a CM-Sepharose
N-methylthiazolium ions, we propose that the hydrophobic column equilibrated in dilution buffer. CYP3A4 was subsequently
eluted with a linear gradient of 0 to 500 mM KCl in dilution buffer.
a
environment of the CYP3A4 active site suppresses the pK of
NMeR’s N-methylthiazolium C2 hydrogen such that it is
readily deprotonated to a N-methyl-1,3-thiazol-2-ylidene
Fractions with an A417/A280 ratio .1.3 were pooled and dialyzed into
storage buffer (50 mM potassium phosphate, pH 7.4, 50 mM KCl, 20%
glycerol, and 2 mM b-mercaptoethanol) and stored at 280°C until
required.
UV-Visible Absorbance Analysis of Ligand Binding. Mea-
surement of UV-visible absorbance spectra was performed on a dual-
beam Olis Cary-14 Spectrophotometer Conversion (Olis, Bogart, GA).
(
thiazol-2-ylidene hereafter), an N-heterocyclic carbene that
has high affinity for iron. Density functional theory calcula-
tions using a truncated model of the Fe3 -P450 active site
support this scenario. To our knowledge, this represents the
1
first example of a transition metal-N-heterocyclic carbene Samples of purified CYP3A4 in 25 mM bis-tris propane/25 mM bicine
complex within an enzyme.
buffer (pH 6.4–8.4) containing 50 mM KCl, 0.1 mM EDTA, and 20%
glycerol were split equally into two cuvettes, and difference spectra
(
300–700 nm) were collected after a baseline and after the addition of
ligands dissolved in dimethylsulfoxide (DMSO). Samples were
allowed to equilibrate for 15 minutes after the addition of each aliquot
before recording the spectra. The final volume of DMSO did not exceed
Materials and Methods
Materials. NMeR bicarbonate and N-[2-Isopropylthiazol-4-
ylmethyl(methyl)carbamoyl]-L-valine (MITV) were purchased from 1% of the total sample volume.
Toronto Research Chemicals (Toronto, Canada). 5-Aminolevulinic
Due to the extreme variation in the absorbance minima and
acid was purchased from Chem-Impex International (Wood Dale, maxima in difference spectra with pH, particularly when NMeR is
IL), isopropyl b-D-1-thiogalactopyranoside from LabScientific (High-
present, an approach based on singular-value decomposition (SVD)
lands, NJ), and 3-[3-(cholamidopropyl)dimethylammonio]-1-propane (Henry and Hofrichter, 1992) was applied to determine the dissocia-
(
CHAPS) from AG Scientific (San Diego, CA). Chromatogra-
tion constants from sets of spectra generated by the ligand titrations.
phy columns were purchased from GE Healthcare Bio-Sciences
Spectra for each ligand concentration (n) were arranged in columns of

4
4
Jennings et al.
a data matrix [A(l,n)] and subjected to the SVD method that populations in the high-spin configuration (%HS) were computed
T
HS
IS
HS
decomposes the data matrix into the product of matrices A 5 USV .
using the equation %HS 5 (100I )/(RI 1 I ), where R 5 1.24 is the
After SVD analyses of the raw data, spectra were reconstructed using
intensity ratio determined by Mak et al. (2013) for CYP3A4 to account
a set of truncated matrices, wherein only the first column vectors from
each of U, S, and V derived from the raw spectra were retained while and low-spin states.
3
for differences in the Raman cross-sections of the n band in the high-
T
the remaining matrix elements were set to zero, thereby removing the
Density Functional Theory Calculations. A truncated model
consisting of a pristine iron porphyrin and a thiolate (HS ) ligand was
2
experimental noise partitioned into those elements.
T
In turn, the spectral amplitudes a 5 SV calculated using the used to model the essential elements of ligand coordination to the
truncated matrices were used to determine dissociation constants, K
d
,
P450 heme. The energetics of the reaction to replace water by several
by fitting the first column vector of a to the corresponding total ligand azaheterocycles and MDP were computed using density functional
concentrations [L
t
] by minimizing
theory. Geometry optimizations and vibrational frequency analyses
were performed using the M06 density functional (Zhao and Truhlar,
2
008) and def2-TZVP (Weigend and Ahlrichs, 2005) basis sets with the
Â
Ã
amax Lf
Gaussian 09 suite of programs (Gaussian, Wallingford, CT). Calcula-
tions using the B3LYP density functional (Becke, 1988, 1993a,b)
including the D3 dispersion correction (Grimme et al., 2011) and def2-
f ða; ½Ltꢁ; ½EꢁÞ 5 a 2
Â
Ã
(1)
Kd 1 Lf
using fsolve in MATLAB, where ½Lf ꢁ 5 ½Ltꢁ 2 ½ is the concentration of TZVP basis set were performed using the Turbomole 6.3 suite of
Eꢁa
am
free ligand, amax is the spectral amplitude at ligand saturation, and [E] programs (Ahlrichs et al., 1989). All reported energies include zero-
is the enzyme concentration. In the case of ritonavir, the spectral
amplitudes were better described by the cooperative model
point energy corrections derived from harmonic vibrational frequency
analyses. Vertical excitation energies were performed with time-
dependent density functional theory using the TPSSh (Tao et al.,
2
003) functional, def2-TZVPP basis sets, RI-J COSX approximation
n
(
Neese et al., 2009), and the M06/def2-TZVP optimized geometries
amax½Ltꢁ
d
a 5
(2),
n
n
with the ORCA 3.02 program (Neese, 2012).
K 1 ½Ltꢁ
where n is the Hill coefficient.
Results
The effects of oxidation and reduction on the difference spectra were
performed using potassium hexachloroiridate and sodium dithionite,
respectively. In the oxidation experiments, samples containing
UV-Visible Absorbance Analysis of Ligand Binding.
In the crystal structure of CYP3A4 bound to ritonavir, the
thiazole nitrogen is clearly coordinated to the heme iron, and
its titration into the enzyme results in a type II difference
spectrum (Sevrioukova and Poulos, 2010). Hence, we expected
3–5 mM CYP3A4 and 200 mM NMeR were prepared and allowed to
equilibrate as described previously for the ligand titrations. After
equilibration of the samples in the spectrophotometer, potassium
hexachloroiridate was added to a final concentration of 50 mM, and the
spectra were recorded every 5 minutes for 2.5 hours. For reduction that methylating the nitrogen ligand would disrupt coordina-
experiments, 1.5 mM CYP3A4 samples (1 ml) in buffer (pH 7.4) were tion and facilitate displacement of the bound water to induce
purged with N
00 mM NMeR to the first cuvette and an equal amount of DMSO to
the second cuvette, and both were allowed to equilibrate for
2
and split equally between two cuvettes. We added a high-spin transition (Fig. 2, A–C). To our surprise, the ad-
2
dition of NMeR at pH 7.4 shifts the Soret band of CYP3A4
to 421 nm like that observed for ritonavir (Sevrioukova and
Poulos, 2010) (Fig. 3A). Recognizing the acidity of the
N-methylthiazolium C2 proton(Breslow, 1958), we considered
the possibility that this proton in NMeR could likewise be
deprotonated to the corresponding thiazol-2-ylidene, which in
turn could replace water as the sixth axial ligand (Fig. 2,
A→B→D). To test this hypothesis, the UV-visible spectra of
31
1
5 minutes. A baseline spectrum was measured with NMeR-Fe
-
31
CYP3A4 and Fe -CYP3A4 in the sample and the reference positions
of the spectrophotometer, respectively. To measure the NMeR-Fe
2
1
-
21
CYP3A4/Fe -CYP3A4 difference spectrum, equal volumes of a
saturated solution of sodium dithionite were added to each cuvette,
and the spectra were repeatedly measured over several minutes until
the spectrum stabilized.
Resonance Raman Spectroscopy. The rR spectra were ob- CYP3A4 were measured at pH 6.0, 6.4, 7.0. 7.4, 8.0, and 8.4 in
tained after excitation with the 406.7 nm line of a Coherent Innova a mixed bis-tris propane/bicine buffer to maintain uniform
3
02C krypton ion laser (Coherent, Santa Clara, CA). Laser powers at
buffering capacity across this range (Fig. 3A). At pH 6.0, the
the sample were 35 mW. Samples (100–200 ml) were maintained in
spinning NMR tubes during data collection. Spectra were collected
using a f/9.7 single-grating monochromator (Acton SP2750; Princeton
Instruments, Trenton, NJ) with a 100 mm slit width, using 2400
grooves/mm gratings, and were imaged using a 1340 ꢀ 400 pixel back-
illuminated charge-coupled device camera with UV optimized coat-
ings (PyLoN 400BR eXcelon; Princeton Instruments). The reported
Soret band maximum is at 418 nm, consistent with the
presence of the low-spin, Fe -H O enzyme. However, there
2
31
is a distinct shoulder present at 390 nm and a weak band at
37 nm, both consistent with an admixture of the penta-
6
coordinate, high-spin state of the enzyme, as initially expected.
As the pH is incrementally increased, the 390 nm component of
spectra are the mean of three 10-minute scans and are unsmoothed. the Soret and the 637 nm bands disappear. At pH 8.4, the high-
Reference calibrations were performed with respect to a mercury spin features of the spectrum are no longer visible, and the
vapor lamp. The nonlinear fluorescence background of rR spectra were Soret band maximum has shifted to 422 nm. The dramatic
removed using asymmetric least squares in MATLAB (http://zanran_
changes in these spectral features support a pH-dependent
storage.s3.amazonaws.com/www.science.uva.nl/ContentPages/
change in the NMeR-CYP3A4 interaction.
4
43199618.pdf).
Difference spectra of NMeR at pH 6.4, 7.4, and 8.4 were
measured to determine the binding affinities and identify any
unique distinguishing features in the spectra (Fig. 4, A–C). At
pH 6.4, titration of NMeR into CYP3A4 results in a classic type
I spectrum with a maximum at 382 nm and minimum at
The concentrations of CYP3A4 were ∼8 mM in 25 mM bis-tris
propane/25 mM bicine buffer (pH 6.0–8.4) containing 50 mM KCl,
.1 mM EDTA, and no ligand, 200 mM NMeR, or 200 mM ritonavir. To
determine the relative populations of the high- and low-spin CYP3A4
from the corresponding positions of the n bands in the spectra, these
0
3
were fit to Gaussian line shapes, and their intensities were calculated 412 nm. Repetition of the titration experiment at pH 7.4
by numeric integration. Calculation of the percentage of the enzyme results in a difference spectrum with a distinct split-Soret

Thiazol-2-Ylidene Coordination to CYP3A4
45
Fig. 2. Putative proton-dependent equilibrium between (A) the
N-methylthiazolium unit and (B) thiazol-2-ylidene. (C) The N-methylthiazolium
is expected to bind above the heme plane, displace the coordinated water
molecule, and induce a low- to high-spin transition of the heme iron. (D)
Thiazol-2-ylidene is a Lewis base that replaces the water ligand to form a
hexa-coordinate low-spin complex.
band, which has maxima at 373, 433, and 467 nm and a
minimum at 412 nm. At pH 8.4, the 467 nm maximum and
4
12 nm minimum are retained while the remaining maxima
shift to 432 nm and 369 nm. Because the extrema of the
difference spectra change dramatically between pH 6.4 and
7
.4, the method relying on SVD with fitting of the spectral
d d
amplitudes to determine K values was used. The K values
for NMeR are likewise pH dependent, undergoing a 14-fold Fig. 3. (A) UV-visible absorption spectra of Fe3+-CYP3A4 (2 mM) in the
presence of 200 mM NMeR at pH 6.0, 6.4, 7.0, 7.4, 8.0, and 8.4. The Soret and a/b
decrease between pH 6.4 and 8.4. These observations support
the increased contribution of a strong stabilizing component of
the CYP3A4-NMeR interaction with increasing pH.
bands are in insets. (B) Representative Fe2 -CYP3A4-NMeR/Fe -CYP3A4
difference spectrum. Samples contained 1.5 mM CYP3A4 and 200 mM
NMeR.
+
2+
To determine whether the UV-visible spectral changes and
K decrease observed with NMeR were the result of pH effects
d
on the protein, a parallel set of titrations and difference determine whether the observed UV-visible spectra could
spectra were measured with ritonavir (Fig. 4, D–F). In the alone be attributed to the interaction of this heterocycle or
Soret region, all spectra have troughs and peaks at 407 nm whether the interactions with the remainder of the ritonavir
and 426 nm, respectively. In the a/b region, peaks were molecule tuned the spectra of the complex. Indeed, the Soret
consistently identified at 544 nm and 584 nm, with a trough and a/b regions of the thiazole-CYP3A4 spectra are identical
at 568 nm. Unlike the other ligands investigated in this study, to that of the ritonavir complex, supporting that the ritona-
the spectral amplitudes fit poorly to a single-binding site vir scaffold does not influence the spectrum (Supplemental
model. Alternatively, these data were best described by Fig. 2). The K of thiazole is 3.4 6 1.0 mM at pH 7.4, more
d
the Hill cooperative model. Within the reported error, the than 4800-fold higher than that measured for ritonavir.
K
d
values determined at pH 6.4 and 7.4 were the same Additionally, the K
d
of thiazole is nearly one and two orders
values measured for
(0.45 mM and 0.42 mM), while this value slightly decreased at of magnitude higher than the K
d
pH 8.4 (0.32 mM). Furthermore, ritonavir binding was simi- 1,2,4-triazole and imidazole (Conner et al., 2012), and, as
larly positively cooperative in each of these experimental supported by density functional theory calculations, it is
conditions.
among the weakest azaheterocyclic ligands for thiolate-
The crystal structure of the ritonavir-CYP3A4 complex ligated Fe3 -heme.
1
clearly illustrates the coordination of the thiazole frag-
NMTI was synthesized and titrated into the CYP3A4 to lend
ment to the CYP3A4 heme iron. Nevertheless, we sought to further support to the concept that the unique features of the

4
6
Jennings et al.
Fig. 4. CYP3A4 (2 mM) absorbance difference spectra and binding isotherms (inset) for (A–C) NMeR, (D–F) ritonavir, and (G–I) MITV, each at pH 6.4,
.4, and 8.4 (left to right). Reported K values and Hill coefficients for ritonavir are the mean of three ligand titration experiments 6 S.D. Inset binding
7
d
isotherms illustrate data from a single representative titration experiment with the corresponding data fitting. Spectra of MITV in buffer show that the
strong band at 360 nm is attributable to the ligand and not the complex with CYP3A4.
difference spectra were attributed to the deprotonation and expected NMeR to assume a similar orientation in the
coordination of the N-methylthiazolium unit of NMeR. NMTI active site.
did not produce difference spectra with the conditions used for
Nevertheless, we sought to investigate the possibility that
NMeR. However, incubation of 20 mM NMTI with 4 mM pH change partially inverts NMeR’s binding mode such that
CYP3A4 at pH 8.4 over 2.5 hours resulted in a difference the isopropyl unit is oriented toward the heme iron and
spectra having a split Soret band with maxima at 378 nm and that this scenario could underlie the unusual features of the
4
64 nm, similar to the features obtained with NMeR (Supple- difference spectra. To test this hypothesis, titration experi-
mental Fig. 1). This result both underscores the importance of ments were performed with MITV (Fig. 1), a truncated analog
the ritonavir scaffold to achieve reasonable binding affinity of ritonavir representing only the left-hand side of the
and supports the notion that the N-methylthiazolium frag- molecule and lacking the heme-coordinating thiazole. MITV
ment contributes the unusual spectral features to the NMeR- should only produce type II spectra if the isopropyl thiazole
CYP3A4 complex.
fragment interacts with the CYP3A4 heme, so it was used as a
Crystal structures clearly demonstrate that analogs of tool to investigate the pH dependence of isopropyl thiazole
ritonavir lacking the phenyl side chains maintain the fragment coordination.
“
thiazole-first” binding mode. In cases when the thiazole has
As illustrated in Fig. 4, G–I, the features of the difference
been replaced with another azaheterocycle, coordination is spectra are pH-independent, and they lack the distinctive
generally maintained. Additionally, in examples wherein the 467 nm band observed in NMeR difference spectra. Further-
azaheterocycle is absent, the orientation of the remaining more, MITV is a comparatively weak ligand for CYP3A4 (K
d
5
scaffold remains consistent with that of ritonavir (Sevrioukova 1.1–2.4 mM) and exhibits only a 2-fold decrease in K with
d
and Poulos, 2012, 2013; Kaur et al., 2015). Moreover, the increasing pH. These differences preclude the possibility that
observation that deaza-ritonavir (the heme coordinating the unusual difference spectra are the result of pH-dependent
nitrogen is replaced by C-H) produces a partial type I spectral NMeR inversion in the CYP3A4 active site.
change is similarly consistent with “thiazole-first” binding.
Hence, evidence supports that the “thiazole-first” binding characteristic typically observed in the spectra of Fe
The strongly red-shifted component at 467 nm is a
2
1
-
mode is very tolerant to scaffold changes, and we therefore carbene complexes (Hodgson and Philpot, 1974; Wolf et al.,

Thiazol-2-Ylidene Coordination to CYP3A4
47
1
977; Ruf et al., 1984). Although there is no obvious electron
1
The high-frequency regions of CYP3A4 rR spectra are
3
source capable of reducing the Fe -thiazol-2-ylidene in illustrated in Fig. 5, A–C. The oxidation-state sensitive
Fig. 2D to a Fe -thiazol-2-ylidene species under the pre- marker bands (n ) are positioned at 1374 cm in all spectra
sent experimental conditions, the spectra of the NMeR regardless of pH or ligand and are consistent with the Fe
2
1
21
4
3
1
-
CYP3A4 complex was treated with both oxidizing and state of the heme in these experiments. Spin-state sensitive
reducing agents with the expectation that spectral changes
would provide insight into the dominant iron oxidation
state before treatment with these reagents. In the presence
bands in the spectra of ligand-free CYP3A4 occur at
2
1
21
21
1
502 cm (n ), 1584 cm (n ), and 1638 cm (n10), all of
3 2
1
3
which are consistent with a predominately low-spin Fe
-
2
1
of the reducing agent sodium dithionite, the Fe -NMeR-
H O enzyme. In the penta-coordinate high-spin state, these
2
2
1
CYP3A4/Fe -CYP3A4 difference spectra revealed notable
changes. Spectra of the reduced species (Fig. 3B) displayed
a prominent Soret band at 447 nm. Conversely, measure-
ment of difference spectra in the presence of the strong
oxidant potassium hexachloroiridate did not reveal any
changes in the difference spectra with the exception of
moderate increases in the intensities of the 433 nm and
21
21
3
markers are expected to shift to ∼1490 cm (n ), ∼1570 cm
(n ), and ∼1630 cm (n10). Due to its minimal overlap with
2
21
3
other bands in the spectra, the intensity of n is routinely
used to determine the relative contributions of the spin
states. In the ligand-free enzyme, a high-spin component of
21
n3 is represented as a small shoulder 1490 cm . Fitting of
the high- and low-spin components of the bands to two
Gaussians and determination of their relative intensities
reveal only a small pH-dependent variation from 8.6% 6
4
67 nm bands over 3 hours. However, an equivalent in-
crease was likewise observed in the control experiments
lacking potassium hexachloroiridate. The observation that
the CYP3A4-NMeR complex is readily reduced to a spec-
troscopically distinct species, possibly the corresponding to
0
.5% at pH 6.0 to 3.6% 6 0.5% at pH 8.4. At pH 7.4, the
high-spin contribution was determined to be 4.9% 6 0.5%, in
agreement with the 5% contribution previously determined
for the ligand-free enzyme (Mak et al., 2013). The remaining
rR spectral features nearly overlap over the entire pH range,
supporting that these changes do not appreciably influence
the heme or its immediate environment.
2
1
Fe -CYP3A4-NMeR complex, while being resistant to a
strong one-electron oxidant supports that the CYP3A4-
3
1
NMeR complex is a Fe -carbene complex.
Resonance Raman Spectroscopy of CYP3A4. Reso-
nance Raman spectroscopy permits measurement of the
symmetry-allowed vibrational transitions of the heme chro-
mophore in the P450 active site environment. Because the
vibrational manifold is sensitive to changes in electronic
structure and conformation of the heme, we have used this
At pH 6.0, NMeR-bound CYP3A4 has strong contributions
2
1
21
from the high-spin marker bands at 1489 cm (n
(n
2
3
), 1569 cm
), and 1632 cm (n10). As the pH increases, their intensities
decrease, and they coalesce into their corresponding low-spin
to quantitate the contributions
2
1
approach to interrogate the effects of pH and ligand binding on components. Focusing on n
3
these characteristics of the CYP3A4 active site. To this end, rR of each spin state, the high-spin contribution was found
spectra of ligand-free CYP3A4 as well as the ritonavir and to be maximal at pH 6.0 (30.5% 6 0.5%), decreasing with
NMeR complexes were measured (Fig. 5).
the increasing pH to a minimum at pH 8.4 (4.3% 6 0.5%).
Fig. 5. The rR spectra of (A, D) ligand-free, (B, E) NMeR-bound, and (C, F) ritonavir-bound Fe3+-CYP3A4 at pH 6.0, 6.4, 7.0, 7.4, 8.0, and 8.4. The high-
and low-frequency regions are illustrated in panels A–C and D–F, respectively. Protein concentrations were 8–10 mM and those of NMeR and ritonavir
21
4 8
were 200 mM. The high- and low- frequency spectra were normalized to the intensities of the n and n bands, respectively. Bands at 721 cm and 1421
21
cm are present in samples containing ligands alone (without CYP3A4) and are therefore not attributable to the CYP3A4-NMeR or CYP3A4-ritonavir
complexes. Arrows indicate the direction of the intensity changes occurring in relevant bands with increasing pH.

4
8
Jennings et al.
Nonlinear least-squares fitting of the %HS and pH values to dependent. Because this band consistently appears in the
the equation
ritonavir spectra and is pH dependent in NMeR spectra, we
speculate that this increase in Raman activity is caused by a
conformational change in the propionates and/or the C/D
(3) rings of the heme that accompanies coordination by the
ligand.
%
HSmin 2 ð%HSmax 2 %HSminÞ
%
HS 5
pH 2 pKa
1
2 10
resulted in an apparent pK
a
of 7.4 for the high- to low-spin
Density Functional Theory Calculations. To assess
transition (Fig. 6). In contrast to NMeR, features of the rR the geometries and strength of the iron-carbon bond
spectra with ritonavir are resistant to changes in pH, and formed between the thiazol-2-ylidene and the P450 heme
they indicate that the high-spin content does not exceed relative to other azaheterocyles and MDP, we used density
3
%. Considering the pH resistance of rR spectra of the functional theory calculations employing the B3LYP-D3
ligand-free and ritonavir-bound enzymes and the near and M06 density functionals. The former functional was
identical chemical structures of these ligands, it appears selected because it has been widely applied to study
that the observed pH dependence of the spin shift with P450 chemistry (Shaik et al., 2005, 2010); the latter was
NMeR is attributable to its N-methylthazolium unit. selected for its improved performance for the treat-
Hence, the apparent pKa of 7.4 is tentatively assigned to ment of transition metal complexes (Cramer and Truhlar,
the NMeR C2 proton, and the loss of the high-spin enzyme 2009).
is attributed to capture of the thiazol-2-ylidene by the
CYP3A4 heme.
The geometry of a model thiazol-2-ylidene coordinated to
the ferric iron of our truncated model optimized at the M06/
The corresponding low-frequency regions of the rR def-TZVP level of theory is illustrated in Fig. 7. Both density
spectra are illustrated in Fig. 5, D–F. Spectra of the functionals predict a stable, low-spin (S 5 1/2) complex with
ligand-free and ritonavir-bound enzymes are similarly the Fe-C and Fe-S bond lengths comparable to the corre-
resistant to changes in pH. Furthermore, their spectra sponding Fe-N (∼2.1 Å) and Fe-S (∼2.3 Å) bond lengths
are very similar with the exception of the so-called pro- observed in high-resolution crystal structures of P450-
pionate bands, which generally have components posi- imidazole complexes (Verras et al., 2006; Sugimoto et al.,
2
1
21
21
tioned at ∼370 cm , ∼380 cm , and ∼390 cm
.
2008). Attempts to optimize the geometry of a complex with
The positions of these bands can be affected by changes the sulfur atom coordinated to the iron atom resulted in
in interactions between the propionates and the protein, dissociation of the complex. Geometry optimizations at both
but recent evidence obtained with hemes containing levels of theory with other azaheterocycles (Table 1) predict
isotopically-labeled methyl groups support that these that these distances vary at most by ∼0.06 Å. However, the
modes are more adequately characterized as out-of- Fe-C bond in the MDP complex is notably shorter by 0.05–
plane distortions of the heme C and D pyrroles (Mak 0.10 Å compared with those predicted for the azaheterocycle
et al., 2004; Podstawka et al., 2006). The propionates complexes.
envelope of the ligand-free enzyme has bands positioned
In an effort to determine the relative binding energy of the
at 371 cm and 380 cm . These plus an additional band thiazol-2-ylidene in comparison with other azaheterocyles and
2
1
21
2
1
at 387 cm
enzyme.
2
appear in spectra of the ritonavir-bound MDP, energy changes for displacement of H O from a low-spin
2
1
A similar band positioned at 390 cm also appears in the
spectra of the NMeR complex, although its intensity is pH
2
2
3+
Fig. 7. M06/def2-TZVP optimized geometry of the [(Porphyrin )(Fe
)
2
(
HS )(N-methyl thiazol-2-ylidene)] complex. The Fe-C2 and Fe-S dis-
Fig. 6. Titration curve illustrating the pH-dependence of spin-state in
NMeR bound CYP3A4 derived from the n bands of rR spectra.
3
tances are labeled in Angstroms. Distances of the complex optimized at the
B3LYP-D3/def2-TZVP level of theory are in parentheses.

Thiazol-2-Ylidene Coordination to CYP3A4
49
TABLE 1
21
Density functional theory energetics (kcal∙mol ) for replacement of water as the proximal ligand in a
truncated model of the P450 active site by several heterocycles and relevant bond distances in the
complexes
M06/def2-TZVP
Fe-X
B3LYP-D3/def2-TZVP
Fe-X
Ligand (X)
Thiazol-2-ylidene
DE
Fe-S
DE
Fe-S
216.1
215.2
214.8
210.7
210.3
28.5
2.13
2.10
2.11
2.05
2.10
2.17
2.11
2.13
2.16
2.24
2.21
2.21
2.23
2.22
2.22
2.21
2.21
2.21
215.4
214.0
213.6
210.3
29.8
2.12
2.12
2.11
2.00
2.10
2.16
2.14
2.13
2.13
2.26
2.22
2.22
2.24
2.23
2.22
2.21
2.22
2.22
1
,2,4-Triazole (N2)
,2,4-Triazole (N4)
1
Methylenedioxyphenyl-2-ylidene
Imidazole
Pyridine
29.0
1
,2,3-Triazole (N2)
,2,3-Triazole (N1)
26.9
26.6
1
26.4
25.4
Thiazole
24.6
27.5
H
2
O-coordinated model were calculated for each of the re- of the Fe21-CO and Fe21-MDP model systems. The Fe21-CO
spective fragments:
complex was predicted to have a strong Soret band at 448.7 nm
(
f 5 0.0126), whereas the Fe2 -MDP was predicted to have two
nearly degenerate bands in the Soret region at 454.3 nm (f 5
.0229) and 454.6 nm (f 5 0.0230). These predictions are in
(4) excellent agreement with the measured spectra of the syn-
1
ÂÀ
ÁÀ
ÁÀ
Á
Ã
Porphyrin2
Porphyrin2
2
2
Fe3 1 ðHS ÞðH2OÞ 1 L→
2
0
ÂÀ
Á
Ã
Fe3 1 ðHS ÞðLÞ 1 H2O
2
21
22
12
21
thetic complexes [(Fe )(PPIX )(CH
(
4
3
S
)(CO)] and [(Fe
TPP )(nBuS )(MDP)] that have Soret bands at 450 nm and
59 nm, respectively (Collman and Sorrell, 1975; Mansuy
)
Displacement of H O by the fragments is consistently exo-
22
12
2
thermic. With the exception of water displacement by N1 of
1
,2,3-triazole and thiazole, the predicted energetic trends are
et al., 1979). Recognizing the accuracy of the approach for
the same for both the M06 and B3LYP-D3 functionals. Taking
the magnitude of the exothermicity as an index of ligand
strength, the trend follows as: thiazol-2-ylidene . 1,2,4-
triazole . MDP . imidazole . 1,2,3-triazole . thiazole. These
results differ in some respects to those described by Conner et al.
31
these systems, we applied it to the Fe -thiazol-2-ylidene
complex. Vertical excitations of 430.4 nm (f 5 0.0026) and
4
61.5 nm (f 5 0.0010) were predicted, consistent with those
31
observed in the NMeR-Fe -CYP3A4 spectra.
(2012) whose study evaluated the energetics of water exchange
by imidazole, 1,2,4-triazole, and 1,2,3 triazole using an identical
model system, M06 functional, and the spin-state corrected
Discussion
6
-31G basis set (Swart et al., 2010). This study reported
Herein we describe a detailed spectroscopic investigation of
the energetic trend: imidazole (25.3 kcal×mol ) . 1,2,4-triazole the complex between an N-methylthiazolium analog of
23.2 kcal×mol ) . 1,2,3 triazole-N2 (22.8 kcal×mol ) . 1,2,3 ritonavir and CYP3A4. Despite the obstruction of the heme-
triazole-N1 (21.7 kcal×mol ). Notable differences between coordinating nitrogen atom by methylation, it fails to consis-
our results and those described in Conner et al. (2012) include tently produce UV-visible spectra demonstrative of the
the 2- to 5-fold differences in exothermicities as well as the expected high-spin transition of the heme iron. In view of
prediction that coordination of imidazole is more exothermic the acidity of the N-methylthiazolium C2 proton and the
21
21
21
(
2
1
than 1,2,4-triazole.
a
sensitivity of its pK to environment, we executed variable-pH
We speculate that these differences are attributable to the spectroscopy studies. Difference spectra confirmed that the
moderately-sized basis set used in their calculations. Owing to contribution of the high-spin enzyme decreases with increas-
our use of a larger, more flexible basis set in these studies as ing pH, having definitively type I characteristics at pH 6.4 and
well as the close agreement between the results obtained with split Soret bands at pH $7.4 that include a red-shifted
two density functionals with different theoretical underpin- component at 467 nm. The red-shifted component is similar
nings, we are confident that our calculations are more reliable. to those observed in “metabolic intermediate complexes” that
Nonetheless, these calculations support the idea that, of the result from the bioactivation of dihalomethanes and methyl-
fragments that frequently occur in type II inhibitors of P450 enedioxyphenyl compounds to carbenes (Hodgson and Philpot,
enzymes, thiazol-2-ylidene is among the strongest ligands for 1974; Wolf et al., 1977; Ruf et al., 1984). Changes in UV-visible
3
1
Fe -P450, closely followed by the two possible binding modes spectra are accompanied by an increase in the affinity of
of 1,2,4-triazole.
NMeR for CYP3A4, indicative of further stabilization of the
To complement experimental evidence that NMeR binding interaction under these conditions. Although the affinity of
to CYP3A4 results in a thiazol-2-ylidene complex, we used NMTI for CYP3A4 was too low to be measured by difference
time-dependent density functional theory calculations to pre- spectroscopy, extended incubations with large concentrations
dict vertical excitation energies and oscillator strengths (f) at high pH likewise produced spectra with the distinct red-
that are comparable to experimental UV-visible absorption shifted band observed in the NMeR-CYP3A4 complex, sup-
spectra. To confirm the validity of this approach, calculations porting that this fragment constitutes the heme-coordinating
were initially performed using models of systems with exten- element. Conversely, variable pH difference spectroscopy
sively documented UV-visible spectra. Specifically, we calcu- demonstrated that changes in this condition do not apprecia-
lated the vertical excitation energies and oscillator strengths bly affect the binding affinities or characteristics of the spectra

5
0
Jennings et al.
obtained with the closely related compounds ritonavir or hydrophobic interactions between the active site and the
MITV. remaining ritonavir scaffold. Hence, for N-methylthiazolium
Taken together with the established acidity of the to bind, the energetic destabilization resulting from placing a
N-methylthiazolium C2 proton, these data point to the cation in the nonpolar environment must be offset by hydro-
possibility that this fragment is deprotonated to a thiazol-2- phobic interactions between CYP3A4 and the scaffold. By
ylidene, an N-heterocyclic carbene that in turn coordinates compelling the N-methylthiazolium unit to enter this envi-
to the heme iron and is responsible for the unusual differ- ronment, the thermodynamic preference to accommodate the
ence spectra. Density functional theory calculations using a neutral thiazol-2-ylidene is enhanced and thereby drives the
minimal model of the putative thiazol-2-ylidene complex, equilibrium toward deprotonation. This phenomenon is man-
3
1
22
12
[
(Fe )(Porphyrin )(HS )(thiazol-2-ylidene)], revealed that ifested as a measurable pK
a
suppression.
it is indeed stable and that electronic excitation energies Calculations support that the thiazol-2-ylidene interaction
calculated using time-dependent density functional theory are with the P450 heme is at least as strong as the azaheterocycles
in excellent agreement with UV-visible spectra of the NMeR- found in the tightest binding competitive inhibitors of P450s.
bound CYP3A4. This remarkable agreement bolsters the assign- Conversely, thiazole is predicted to be among the weakest
3
1
ment of this interaction as a thiazol-2-ylidene-coordinated ligands for the Fe -P450. However, ritonavir binds to
heme and supports that the unusual spectra are directly CYP3A4 with nearly 8-fold higher affinity at pH 8.4. This
attributable to electronic transitions of the heme and this apparent discrepancy is logical for two reasons. First, because
unusual ligand.
NMeR is a quaternary cation, there is a larger energetic
A more detailed view of the heme electronic structure and penalty associated with stripping bound water molecules
the influence of pH on its immediate environment was (desolvation) from the ligand before it enters the active site.
provided by rR spectroscopy. Only rR spectra of the NMeR- Second, in ritonavir the thiazole N-atom is optimally posi-
CYP3A4 complex demonstrated pH dependence, and only tioned for coordination to the heme iron, although this is not
those bands corresponding to changes in spin state and out-of- the case for the C2 in NMeR. Hence, positioning of the C2 and
plane distortions of the C/D pyrroles were affected. Because accommodation of the methyl group likely coincide with a
the binding mode of NMeR is expected to closely resemble that substantial rearrangement of this region relative to the
of ritonavir, whose rR spectra are not influenced by pH, these binding mode observed for ritonavir, thereby adding to the
experiments likewise support that the pH-dependence is energetic penalty. Indeed, NMeR is not an improvement over
attributable to the N-methylthiazolium fragment. Hence, we ritonavir with regard to affinity. However, it does demonstrate
a
tentatively assign the apparent pK of 7.4 to the NMeR C2 a novel carbene–P450 interaction that has the potential for
proton in the active site of CYP3A4. This value is at best an optimization and possible applications in the development of
approximation of this pK for two reasons. First, the rela- inhibitors for CYP3A4 and other P450s. Finally, unlike other
a
tive spin state contributions only provide an indirect mea- carbene ligands, oxidative or reductive bioactivation of the par-
sure of a protonation/deprotonation equilibrium. Second, the ent molecule is unnecessary to generate the iron-coordinating
deprotonated species is readily captured by the heme and element.
a
thereby distorts the equilibrium. Nevertheless, this pK is in
line with the estimates made for the C2 proton in thiamine-
dependent enzymes.
Deprotonation of an alkylthiazolium ion to a thiazol-2-
ylidene likewise occurs in the catalytic cycles of thiamine-
dependent enzymes. After the initial observation of Breslow
Acknowledgments
The authors thank the Ohio Supercomputer Center for a generous
allocation of computational resources.
Authorship Contributions
Participated in research design: Jennings, Hackett.
Conducted experiments: Jennings, Ritchie, Shock, Hackett.
Contributed new reagents or analytic tools: Lyons.
Performed data analysis: Jennings, Ritchie, Shock, Hackett.
Wrote or contributed to the writing of the manuscript: Jennings,
Hackett.
(
1958) that the C2 proton of N-methylthiazolium exchanges
with D O with a half-life of 20 minutes, deprotonation of this
2
position remains generally accepted as the initial step in
thiamine-dependent catalysis. The first solution pKa mea-
surements of thiamine and N-methylthiazolium yielded val-
ues of 12.7 and $ 13.5, respectively (Hopmann and Brugnoni,
References
1
973). Later measurements for thiamine assigned the C2
Ahlrichs R, Baer M, Haeser M, Horn H, and Koelmel C (1989) Electronic structure
calculations on workstation computers: the program system turbomole. Chem Phys
Lett 162:165–169.
Becke AD (1988) Density-functional exchange-energy approximation with correct
asymptotic behavior. Phys Rev A Gen Phys 38:3098–3100.
Becke AD (1993a) Density-functional thermochemistry. III. The role of exact ex-
change. J Chem Phys 98:5648–5652.
Becke AD (1993b) A new mixing of Hartree–Fock and local density-functional theo-
ries. J Chem Phys 98:1372–1377.
Breslow R (1958) On the mechanism of thiamine action. IV.1 Evidence from studies
on model systems. J Am Chem Soc 80:3719–3726.
Casida JE (1970) Mixed-function oxidase involvement in the biochemistry of in-
secticide synergists. J Agric Food Chem 18:753–772.
Collman JP and Sorrell TN (1975) Letter: A model for the carbonyl adduct of ferrous
cytochrome P450. J Am Chem Soc 97:4133–4134.
Conner KP, Vennam P, Woods CM, Krzyaniak MD, Bowman MK, and Atkins WM
proton a pK of 18.0. Despite their high solution values, the
a
active sites of these enzymes are known to suppress the C2
proton pKa such that it is readily deprotonated under
physiologic conditions. This idea is supported by the NMR
studies of C2 deprotonation kinetics in pyruvate decarbox-
ylase and transketolase, which showed that the protein5
environment accelerated these rates by as much as 2 ꢀ 10
(
Kern et al., 1997). There is also evidence that pyruvate
a
decarboxylase suppresses the pK of its enamine intermediate
from 15.4 in water to ∼6 in the enzyme environment, a change
that corresponds to a billion-fold rate enhancement for
enamine deprotonation (Jordan et al., 1999).
Spectroscopic measurements of thiazole and NMTI binding
to the enzyme reveal they are very poor ligands, thus the high
affinities of ritonavir and NMeR (at high pH) are conferred by
(
2012) 1,2,3-Triazole-heme interactions in cytochrome P450: functionally compe-
tent triazole-water-heme complexes. Biochemistry 51:6441–6457.
Cramer CJ and Truhlar DG (2009) Density functional theory for transition metals
and transition metal chemistry. Phys Chem Chem Phys 11:10757–10816.
Denisov IG, Makris TM, Sligar SG, and Schlichting I (2005) Structure and chemistry
of cytochrome P450. Chem Rev 105:2253–2277.

Thiazol-2-Ylidene Coordination to CYP3A4
51
Grimme S, Ehrlich S, and Goerigk L (2011) Effect of the damping function in dis-
persion corrected density functional theory. J Comput Chem 32:1456–1465.
Guengerich FP (2001) Uncommon P450-catalyzed reactions. Curr Drug Metab 2:
Neese F (2012) The ORCA program system. WIRES: Comp Mol Sci 2:73–78.
Neese F, Wennmohs F, Hansen A, and Becker U (2009) Efficient, approximate and
parallel Hartree–Fock and hybrid DFT calculations. A ‘chain-of-spheres’ algorithm
for the Hartree–Fock exchange. Chem Phys 356:98–109.
9
3–115.
Henry ER and Hofrichter J (1992) Singular value decomposition: application to
Ortiz de Montellano PR (2005) Cytochrome P450: structure, mechanism, and bio-
chemistry, 3rd ed, Kluwer Academic/Plenum, New York.
Podstawka E, Mak PJ, Kincaid JR, and Proniewicz LM (2006) Low frequency reso-
nance Raman spectra of isolated alpha and beta subunits of hemoglobin and their
deuterated analogues. Biopolymers 83:455–466.
Rendic S (2002) Summary of information on human CYP enzymes: human P450
metabolism data. Drug Metab Rev 34:83–448.
Ring BJ, Patterson BE, Mitchell MI, Vandenbranden M, Gillespie J, Bedding AW,
Jewell H, Payne CD, Forgue ST, and Eckstein J et al. (2005) Effect of tadalafil on
cytochrome P450 3A4-mediated clearance: studies in vitro and in vivo. Clin
Pharmacol Ther 77:63–75.
Ruf HH, Ahr H, Nastainczyk W, Ullrich V, Mansuy D, Battioni JP, Montiel-Montoya
R, and Trautwein A (1984) Formation of a ferric carbanion complex from halothane
and cytochrome P-450: electron spin resonance, electronic spectra, and model
complexes. Biochemistry 23:5300–5306.
Sevrioukova IF and Poulos TL (2010) Structure and mechanism of the complex be-
tween cytochrome P4503A4 and ritonavir. Proc Natl Acad Sci USA 107:
18422–18427.
analysis of experimental data. Methods Enzymol 210:129–192 DOI:10.1016/0076-
6
879(92)10010-B.
Heydari A, Yeo KR, Lennard MS, Ellis SW, Tucker GT, and Rostami-Hodjegan A
2004) Mechanism-based inactivation of CYP2D6 by methylenedioxymethamphet-
(
amine. Drug Metab Dispos 32:1213–1217.
Hirsch MS, Günthard HF, Schapiro JM, Brun-Vézinet F, Clotet B, Hammer SM,
Johnson VA, Kuritzkes DR, Mellors JW, and Pillay D et al.; International AIDS
Society-USA (2008) Antiretroviral drug resistance testing in adult HIV-1 infection:
2
1
008 recommendations of an International AIDS Society-USA panel. Top HIV Med
6:266–285.
Hodgson E and Philpot RM (1974) Interaction of methylenedioxyphenyl (1,3-benzo-
dioxole) compounds with enzymes and their effects on mammals. Drug Metab Rev
3:231–301.
Hollenberg PF (2002) Characteristics and common properties of inhibitors, inducers,
and activators of CYP enzymes. Drug Metab Rev 34:17–35.
Hopmann RF and Brugnoni GP (1973) pK of thiamine C(2)H. Nat New Biol 246:
1
57–158.
Jordan F, Li H, and Brown A (1999) Remarkable stabilization of zwitterionic inter-
mediates may account for a billion-fold rate acceleration by thiamin diphosphate-
dependent decarboxylases. Biochemistry 38:6369–6373.
Kaur P, Chamberlin R, Poulos TL, and Sevrioukova IF (2015) Structure-based in-
hibitor design for evaluation of a CYP3A4 pharmacophore model. J Med Chem
DOI:10.1021/acs.jmedchem.5b01146 [published ahead of print].
Kempf DJ, Marsh KC, Denissen JF, McDonald E, Vasavanonda S, Flentge CA, Green
BE, Fino L, Park CH, and Kong XP et al. (1995) ABT-538 is a potent inhibitor of
human immunodeficiency virus protease and has high oral bioavailability in hu-
mans. Proc Natl Acad Sci USA 92:2484–2488.
Sevrioukova IF and Poulos TL (2012) Interaction of human cytochrome P4503A4
with ritonavir analogs. Arch Biochem Biophys 520:108–116.
Sevrioukova IF and Poulos TL (2013) Dissecting cytochrome P450 3A4-ligand in-
teractions using ritonavir analogues. Biochemistry 52:4474–4481.
Shaik S, Cohen S, Wang Y, Chen H, Kumar D, and Thiel W (2010) P450 enzymes:
their structure, reactivity, and selectivity-modeled by QM/MM calculations. Chem
Rev 110:949–1017.
Shaik S, Kumar D, de Visser SP, Altun A, and Thiel W (2005) Theoretical perspective
on the structure and mechanism of cytochrome P450 enzymes. Chem Rev 105:
2279–2328.
Kempf DJ, Marsh KC, Kumar G, Rodrigues AD, Denissen JF, McDonald E, Kukulka
MJ, Hsu A, Granneman GR, and Baroldi PA et al. (1997) Pharmacokinetic en-
hancement of inhibitors of the human immunodeficiency virus protease by co-
administration with ritonavir. Antimicrob Agents Chemother 41:654–660.
Kena Diba A, Noll C, Richter M, Gieseler MT, and Kalesse M (2010) Intramolecular
stereoselective protonation of aldehyde-derived enolates. Angew Chem Int Ed Engl
Sugimoto H, Shinkyo R, Hayashi K, Yoneda S, Yamada M, Kamakura M,
Ikushiro S, Shiro Y, and Sakaki T (2008) Crystal structure of CYP105A1
(P450SU-1) in complex with 1alpha,25-dihydroxyvitamin D3. Biochemistry 47:
4017–4027.
Swart M, Güell M, Luis JM, and Solà M (2010) Spin-state-corrected Gaussian-type
orbital basis sets. J Phys Chem A 114:7191–7197.
Tao J, Perdew JP, Staroverov VN, and Scuseria GE (2003) Climbing the density
functional ladder: nonempirical meta-generalized gradient approximation
designed for molecules and solids. Phys Rev Lett 91:146401.
Verras A, Alian A, and de Montellano PR (2006) Cytochrome P450 active site plas-
ticity: attenuation of imidazole binding in cytochrome P450(cam) by an L244A
mutation. Protein Eng Des Sel 19:491–496.
von Moltke LL, Greenblatt DJ, Grassi JM, Granda BW, Duan SX, Fogelman SM,
Daily JP, Harmatz JS, and Shader RI (1998) Protease inhibitors as inhibitors of
human cytochromes P450: high risk associated with ritonavir. J Clin Pharmacol
38:106–111.
Weigend F and Ahlrichs R (2005) Balanced basis sets of split valence, triple zeta
valence and quadruple zeta valence quality for H to Rn: design and assessment of
accuracy. Phys Chem Chem Phys 7:3297–3305.
Wolf CR, Mansuy D, Nastainczyk W, Deutschmann G, and Ullrich V (1977) The
reduction of polyhalogenated methanes by liver microsomal cytochrome P450. Mol
Pharmacol 13:698–705.
Zhao Y and Truhlar DG (2008) Density functionals with broad applicability in chem-
istry. Acc Chem Res 41:157–167.
4
9:8367–8369.
Kern D, Kern G, Neef H, Tittmann K, Killenberg-Jabs M, Wikner C, Schneider G,
and Hübner G (1997) How thiamine diphosphate is activated in enzymes. Science
2
75:67–70.
Klibanov OM, Gale SE, and Santevecchi B (2015) Ombitasvir/paritaprevir/ritonavir
and dasabuvir tablets for hepatitis C virus genotype 1 infection. Ann Pharmacother
4
9:566–581.
Koudriakova T, Iatsimirskaia E, Utkin I, Gangl E, Vouros P, Storozhuk E, Orza D,
Marinina J, and Gerber N (1998) Metabolism of the human immunodeficiency
virus protease inhibitors indinavir and ritonavir by human intestinal microsomes
and expressed cytochrome P4503A4/3A5: mechanism-based inactivation of cyto-
chrome P4503A by ritonavir. Drug Metab Dispos 26:552–561.
Lin HL, Kent UM, and Hollenberg PF (2002) Mechanism-based inactivation of cy-
tochrome P450 3A4 by 17 alpha-ethynylestradiol: evidence for heme destruction
and covalent binding to protein. J Pharmacol Exp Ther 301:160–167.
Mak PJ, Podstawka E, Kincaid JR, and Proniewicz LM (2004) Effects of systematic
peripheral group deuteration on the low-frequency resonance Raman spectra of
myoglobin derivatives. Biopolymers 75:217–228.
Mak PJ, Zhu Q, and Kincaid JR (2013) Using resonance Raman cross-section data to
estimate the spin state populations of cytochromes P450. J Raman Spectrosc 44:
1
792–1794.
Address correspondence to: John C Hackett, Department of Physiology and
Biophysics and the Massey Cancer Center, Virginia Commonwealth Univer-
sity School of Medicine, 401 College St. Richmond, Virginia, 23219. E-mail:
john.hackett@vcuhealth.org
Mansuy D, Battioni JP, Chottard JC, and Ullrich
V
(1979) Preparation of a
porphyrin-iron-carbene model for the cytochrome P 450 complexes obtained upon
metabolic oxidation of the insecticide synergists of the 1,3-benzodioxole series.
J Am Chem Soc 101:3971–3973.