Spin equilibrium and O2-binding kinetics of Mycobacterium tuberculosis CYP51 with mutations in the histidine-threonine dyad

Article abstract of DOI:10.1016/j.jinorgbio.2014.03.017

The acidic residues of the "acid-alcohol pair" in CYP51 enzymes are uniformly replaced with histidine. Herein, we adopt the Mycobacterium tuberculosis (mt) enzyme as a model system to investigate these residues' roles in finely tuning the heme conformation, iron spin state, and formation and decay of the oxyferrous enzyme. Properties of the mtCYP51 and the T260A, T260V, and H259A mutants were interrogated using UV-Vis and resonance Raman spectroscopies. Evidence supports that these mutations induce comprehensive changes in the heme environment. The heme iron spin states are differentially sensitive to the binding of the substrate, dihydrolanosterol (DHL). DHL and clotrimazole perturb the local environments of the heme vinyl and propionate substituents. Molecular dynamics (MD) simulations of the DHL-enzyme complexes support that the observed perturbations are attributable to changes in the DHL binding mode. Furthermore, the rates of the oxyferrous formation were measured using stopped-flow methods. These studies demonstrate that both HT mutations and DHL modulate the rates of oxyferrous formation. Paradoxically, the binding rate to the H259A mutant-DHL complex was approximately four-fold that of mtCYP51, a phenomenon that is predicted to result from the creation of an additional diffusion channel from loss of the H259-E173 ion pair in the mutant. Oxyferrous enzyme auto-oxidation rates were relatively constant, with the exception of the T260V-DHL complex. MD simulations lead us to speculate that this behavior may be attributed to the distortion of the heme macrocycle by the substrate.

Full text of DOI:10.1016/j.jinorgbio.2014.03.017

Spin equilibrium and O2-binding kinetics of Mycobacterium tuberculosis
CYP51 with mutations in the histidine-threonine dyad
Gareth K. Jennings, Anuja Modi, Justin E. Elenewski, Caroline M. Ritchie,
Thuy Nguyen, Keith C. Ellis, John C. Hackett
PII:
DOI:
Reference:
S0162-0134(14)00094-4
doi: 10.1016/j.jinorgbio.2014.03.017
JIB 9499
To appear in:
Journal of Inorganic Biochemistry
Received date:
Revised date:
Accepted date:
3 January 2014
25 March 2014
28 March 2014
Please cite this article as: Gareth K. Jennings, Anuja Modi, Justin E. Elenewski,
Caroline M. Ritchie, Thuy Nguyen, Keith C. Ellis, John C. Hackett, Spin equi-
librium and O2-binding kinetics of Mycobacterium tuberculosis CYP51 with muta-
tions in the histidine-threonine dyad, Journal of Inorganic Biochemistry (2014), doi:
1
0.1016/j.jinorgbio.2014.03.017
This is a PDF file of an unedited manuscript that has been accepted for publication.
As a service to our customers we are providing this early version of the manuscript.
The manuscript will undergo copyediting, typesetting, and review of the resulting proof
before it is published in its final form. Please note that during the production process
errors may be discovered which could affect the content, and all legal disclaimers that
apply to the journal pertain.

ACCEPTED MANUSCRIPT
Spin equilibrium and O -binding kinetics of
2
Mycobacterium tuberculosis CYP51 with mutations
in the histidine-threonine dyad
‡
Gareth K. Jennings, Anuja Modi , Justin E. Elenewski, Caroline M. Ritchie, Thuy Nguyen, Keith
C. Ellis and John C Hackett*
Department of Physiology and Biophysics and The Massey Cancer Center, Virginia
Commonwealth University School of Medicine, Richmond, Virginia 23219
‡
Current Address: Department of Chemistry, University of South Carolina, Columbia, South
Carolina.
*
Corresponding Author: John C Hackett, Goodwin Research Laboratory, VCU Massey Cancer
Center, 401 College St., Richmond, Virginia 23219.
Telephone: +1.804.828.5679
Email: jchackett@vcu.edu
1

ACCEPTED MANUSCRIPT
Abstract
The acidic residues of the “acid-alcohol pair” in CYP51 enzymes are uniformly replaced with
histidine. Herein, we adopt the Mycobacterium tuberculosis (mt) enzyme as a model system to
investigate these residues roles in finely tuning the heme conformation, iron spin state, and
formation and decay of the oxyferrous enzyme. Properties of the mtCYP51 and the T260A,
T260V, and H259A mutants were interrogated using UV-Vis and resonance Raman
spectroscopies. Evidence supports that these mutations induce comprehensive changes in the
heme environment. The heme iron spin states are differentially sensitive to the binding of the
substrate, dihydrolanosterol (DHL). DHL and clotrimazole perturb the local environments of the
heme vinyl and propionate substituents. Molecular dynamics (MD) simulations of the DHL-
enzyme complexes support that the observed perturbations are attributable to changes in the
DHL binding mode. Furthermore, the rates of the oxyferrous formation were measured using
stopped-flow methods. These studies demonstrate that both HT mutations and DHL modulate the
rates of oxyferrous formation. Paradoxically, the binding rate to the H259A mutant-DHL
complex was approximately four-fold that of mtCYP51, a phenomenon that is predicted to result
from the creation of an additional diffusion channel from loss of the H259-E173 ion pair in the
mutant. Oxyferrous enzyme auto-oxidation rates were relatively constant, with the exception of
T260V-DHL complex. MD simulations lead us to speculate that this behavior may be attributed
to distortion of the heme macrocycle by the substrate.
Keywords: CYP51 · Cytochrome P450 · Stopped-flow spectroscopy · Molecular Dynamics ·
Resonance Raman Spectroscopy
2

ACCEPTED MANUSCRIPT
1
. Introduction
The cytochromes P450 are among the most flexible enzymatic catalysts, using O and
2
reducing equivalents from nicotinamide cofactors to afford a breadth of oxidative
transformations[1-5]. Entry into the catalytic cycle occurs with a substrate binding event that
often, though not always displaces a water molecule from the hexacoordinate ferric heme center,
inducing a low (S = 1/2) to high (S = 5/2) spin transition. The corresponding increase in the
redox potential facilitates reduction to the ferrous state and subsequent O binding. The resulting
2
oxyferrous intermediate undergoes reduction to a ferric peroxo species that, in turn, is dually
protonated by a putative hydrogen-bonded network. Protonation cleaves the O-O bond to afford
an oxoferryl  -cation radical, also known as Compound I, that is generally accepted to be the
hydroxylating species in canonical cytochrome P450 catalysis[6, 7]. Two amino acids, typically
aspartate and threonine, are broadly conserved among these enzymes and have been established
as critical components of the proton delivery system[8-21]. However, this axiom is largely based
on extensive experimental and theoretical studies of cytochrome P450cam (CYP101). By
contrast, residues replacing the “acid-alcohol” pair have been identified in other cytochrome
P450 enzymes, although their roles in catalysis remained undefined[22].
A distinguishing feature of one P450 family, the sterol 14α-demethylases (CYP51), is
that the acidic residues of the acid-alcohol pair are uniformly replaced with histidine [23]. The
active sites of mtCYP51[24] and CYP101 including the relevant regions of the I-helix are
illustrated in Figure 1. This superimposition highlights the structural equivalence of the HT dyad
3

ACCEPTED MANUSCRIPT
and acid-alcohol pair. Summarized in Scheme 1, CYP51s demethylate 14α-methylated steroids
such as dihydrolanosterol (DHL)[25]. This transformation requires three revolutions of the P450
catalytic cycle, each requiring O and two electrons. The first two steps are apparently
2
Compound I-mediated hydroxylations that yield a 14α-carboxaldehyde. These are followed by a
mechanistically enigmatic third step that removes the carboxaldehyde as formate with
concomitant loss of the 14α-hydrogen. Biochemical experiments utilizing isotopically labeled
substrates have concluded that the ferric peroxo intermediate mediates the third step, but the
precise mechanism remains elusive[26]. To catalyze these two mechanistically distinct reactions,
the enzyme must switch between the effective use of Compound I and peroxo intermediates.
This catalytic strategy is generally considered to require reorganization of the proton delivery
machinery prior to the third step. Molecular dynamics (MD) simulations and hybrid quantum
mechanics/molecular mechanics studies in our group support this notion[27, 28]. While these
results confirm the value of theory to distinguish physically permissible mechanisms from
unrealistic ones, in this case complementary experiments are necessary to confirm whether
redundant proton-transfer pathways are operative in CYP51 catalysis.
To attain a comprehensive understanding of CYP51 catalysis, a complete picture of the
catalytic cycle including the mechanisms of oxygen binding, electron and proton transfer, and
identification of the elusive intermediates in the deformylation step are of substantial importance.
The first reactive oxygen intermediate and bifurcation point of the canonical catalytic cycle is the
oxyferrous species, which may proceed to productive oxidation of the substrate or auto-oxidize
to yield the ferric enzyme and superoxide anion. Like the acid-alcohol pair in other P450
enzymes, the HT dyad of the CYP51 family may similarly tune the spin state of the heme iron,
modulate the rate of oxygen binding, and stabilize the resulting oxyferrous complex from auto-
4

ACCEPTED MANUSCRIPT
oxidation[29-32]. In this study, we adopt the Mycobacterium tuberculosis (mt) enzyme as a
model system and take the first steps toward dissecting the relationships between conserved
active site residues and enzyme mechanism. UV-Visible (UV-Vis) absorption and resonance
Raman (rR) spectroscopy were used to probe both the heme environment and electronic structure
of the ferric enzyme in response to ligand binding. Furthermore, the impact of mutations and
substrate on the rates of O binding to the ferrous enzymes and subsequent auto-oxidation of the
2
oxyferrous species are investigated. Changes in these parameters are interpreted in light of
molecular dynamics simulations and computational predictions of O migration pathways into
2
the active sites.
2
. Materials and Methods
2
.1.Synthesis of 24,25-Dihydrolanosterol and Cyclodextrin Encapsulation. Lanosterol (55-65%;
Steraloids, Newport, RI) was purified by silica gel chromatography using ethyl acetate:hexane
(
1:4) as a solvent. To lanosterol (2 mmol, 0.85 g) in dichloromethane (40 ml) 5% Pd/C [20%
w/w), 0.17 g] was added and the mixture hydrogenated at room temperature at 50 psi for 24 h.
(
The resulting mixture was filtered through celite, and the filter cake was washed thrice with 5
mL of dichloromethane. The solvent was evaporated to afford 24,25-dihydrolanosterol (DHL) as
a white solid in quantitative yield. To improve the aqueous solubility, DHL was encapsulated in
2
4
-hydroxypropyl-β-D-cyclodextrin (HPCD). To 9.5 mL of a stirring and refluxing solution of
5% (w/v) HPCD in 100 mM potassium phosphate buffer (pH 7.5), 2.1 mg of DHL was added
and allowed to dissolve. After cooling to room temperature, the solution was diluted to 500 μM
DHL. HPCD-encapsulated DHL (HPCD-DHL; DHL hereafter) was aliquoted and lyophilized
for storage and reconstituted with deionized water prior to use.
5

ACCEPTED MANUSCRIPT
2
.2. Construction of mtCYP51 mutants. The mtCYP51-pET-17b expression plasmid was a gift
from Larissa Podust (University of California-San Francisco)[25]. The CYP51 mutants were
constructed using the Quikchange (Stratagene) mutagenesis method where DpnI (New England
Biolabs) was used to digest the parental DNA. Transformation of the remaining DNA resulted in
colonies that were screened for the desired mutation by DNA sequencing.
2
.3. Protein expression and purification. mtCYP51 and its mutants were over expressed and
isolated from HMS174(DE3) E. coli cells. A single colony was used to inoculate 100 mL of LB
media with ampicillin (100 μg/mL) and grown overnight at 37 C with shaking at 200 rpm. The
overnight culture (15 mL) was used to inoculate 1 L of LB media containing ampicillin (100
μg/mL), potassium phosphate (100 mM, pH 7.4) and glycerol (5% v/v). The expression culture
was incubated at 37 C with shaking at 180 rpm. When the A600 reached 0.4 to 0.6, the
temperature was decreased to 20 C (18 C in the case of the T260V mutant) and the shaking to
1
50 rpm. When the A600 reached 0.6 to 0.8, the culture was induced with 0.2 mM isopropyl 1-
thio--D-galactopyranoside and 0.5 mM δ-aminolevulinic acid. After 20 h of growth, the cells
were harvested by centrifugation at 5000  g and 4 C for 10 minutes. Cell pellets were
resuspended in lysis buffer (3 mL/g cell pellet) containing 50 mM Tris-HCl (pH 8.0), 500 mM
NaCl, 10% (v/v) glycerol, and 1 mM phenylmethylsulfonyl fluoride. The cell suspension was
lysed by two passes through a high-pressure homogenizer at 25000 psi. The lysate was
2
+
centrifuged at 10000  g for 45 minutes. The supernatant was loaded onto Ni -NTA agarose
resin (Qiagen) that had been previously equilibrated with the lysis buffer. The column was
washed with 1.5 - 2 L of the lysis buffer and proteins were then eluted with the same buffer
containing 50 mM imidazole. Red-colored fractions were pooled and concentrated with Amicon
filtration units (Millipore) with a molecular weight cutoff of 30 kDa. Concentrated protein was
6

ACCEPTED MANUSCRIPT
loaded onto a Superdex-75 column previously equilibrated with 50 mM Tris-HCl (pH 8.0), 100
mM NaCl, and 5% (v/v) glycerol. Proteins were eluted with the same buffer at a flow rate of 0.2
mL/min. Red fractions with a A /A ratio of 1.4 or greater were pooled and the purity of these
419 280
preparations were confirmed by SDS-PAGE. CO binding assay was performed by monitoring
the absorbance at 450 nm after complete reduction with sodium dithionite and bubbling with CO.
Spectra of carbonmonoxyferrous mtCYP51 and mutants consistently have Soret, , and  bands
at 449 nm, 542 nm, and 574 nm, respectively. Cytochrome P450cam was expressed and purified
as previously described [33]. Proteins were stored at -80 C until required.
2
.4. UV-Vis Spectroscopy Ligand Binding Titrations. Binding of DHL and clotrimazole (CLO) to
mtCYP51 and mutants were monitored in 50 mM Tris-HCl buffer (pH 8.0) containing 100 mM
NaCl and 5% (v/v) glycerol using an Olis Cary 14 Conversion double-beam UV-Vis
spectrometer. After recording a baseline with ferric protein, CLO dissolved in ethanol or DHL
aqueous solution was incrementally added to the sample cell, while the corresponding volume of
solvent was added to the reference cell. The total volume of additional solvent did not exceed 2%
(v/v). The difference in absorbance between the wavelength minima and maxima (A) were
plotted against the inverse of the DHL or CLO concentrations and the K values were determined
d
by a linear regression fit to the equation 1/ A  (K / A )(1/[L])1/ A . The percent of
d
max
max
ferric enzymes in the high-spin state (%HS) were computed using the relative absorbances at 646
nm ( A646 ) and 419 nm ( A419 ) as well as the corresponding extinction coefficients for the low-
spin
(
419
)
and high-spin components
(
419
)
using the equation
%
HS 100A /(( / )A  A )[32, 34].
646
646
419
419
419
7

ACCEPTED MANUSCRIPT
2
.5. Resonance Raman Spectroscopy. RR spectra were obtained following excitation using the
4
06.7 nm line of a Coherent Innova 302C krypton ion laser. Laser powers at the sample were 35
mW. Spectra were collected using a f/9.7 single grating monochromator (Acton SP2750,
Princeton Instruments) at a 15 micron slit width using 1800 grooves/mm gratings, and imaged
using a 1340 x 400 pixel back-illuminated CCD camera with UV optimized coatings (PyLoN
4
00BR eXcelon, Princeton Instruments). Reference calibrations were performed with respect to a
Hg vapor lamp. The nonlinear fluorescence background of resonance Raman spectra were
removed using asymmetric least squares smoothing [35] in MATLAB.
2
.6. Stopped-flow UV-Vis Spectroscopy Kinetic Studies. Buffers used in the stopped-flow kinetic
experiments were thoroughly degassed using multiple rounds of a freeze-pump-thaw procedure
in Schlenk tubes attached to a gas/vacuum manifold. In brief, buffer-filled Schlenk tubes were
submerged in liquid N until completely frozen. The remaining headspace was thoroughly
2
evacuated by opening it to vacuum and then closed. The buffer was allowed to fully thaw and the
procedure was repeated with at least three additional cycles. When the degassing procedure was
complete, the buffers were stored under an atmosphere of N . mtCYP51 proteins in 50 mM Tris-
2
HCl (pH 8.0) buffer were purged with N on a sealed gas manifold for 20 minutes at 4 C and
2
transferred to an N -filled glovebox. The proteins were reduced by the addition a few crystals of
2
sodium dithionite and complete reduction were monitored by UV-Vis absorption spectroscopy.
Excess dithionite was removed using a Sephadex G-25 HiTrap desalting column previously
equilibrated with degassed buffer. Degassed buffer was oxygenated by bubbling O for 30 min.
2
Dissolved O concentrations were determined using the azide modification of the Winkler
2
method.[36]
8

ACCEPTED MANUSCRIPT
Kinetic data were acquired using an Applied Photophysics SX-18MV stopped flow UV-
Vis spectrophotometer equipped with a xenon lamp and photodiode array detector. The dead
time of the instrument was determined to be 1.2 ms using the method described by Matsumura.
[
37] After flushing with degassed buffer, the two syringes were respectively loaded with
oxygenated buffer and ferrous mtCYP51 or mutants in anaerobic buffer. Concentrations
provided in the Figures are those after mixing in the 350 μL flow cell. The photodiode array
collected 800 spectral scans over two seconds. The kinetic data sets were fit to the reaction
k1
k2
2

3

2
3

2
model Fe Fe O Fe O using the Applied Photophysics Pro-K software. Singular-
value decomposition (SVD) and global fitting routines were used to derive basis spectra for the
individual species and to compute rate constants.
2
.7. Cryogenic UV-Vis Spectroscopy. Ferric enzyme solutions were purged with N at 4 ºC and
2
transferred to an N -filled glovebox for manipulation and preparation of the oxyferrous
2
intermediate. All buffers used in these procedures were thoroughly degassed using the
aforementioned freeze-pump-thaw procedure. Anaerobic ferric mtCYP51 was reduced with a
few crystals of sodium dithionite and incubated at 4 ºC for 15 minutes. The UV-Vis absorption
spectra from 350 to 700 nm were recorded in a sealed cuvette to confirm complete reduction as
supported by the appropriate shift in the Soret band. After returning the sample to the glove box,
excess sodium dithionite was removed using a Sephadex G-25 column previously equilibrated
with anaerobic buffer. 0.25 mL Anaerobic ferrous enzyme was added to a polymethacrylate
cuvette containing 0.75 mL of glycerol and thoroughly mixed. The final buffer composition was
5
0 mM Tris-HCl (pH 8.0) with 75% glycerol. Following a one h incubation at 4 ºC, O was
2
bubbled into the cold glycerol/protein mixture for three minutes. The samples were subsequently
frozen in liquid nitrogen vapors followed by flash freezing in the liquid. UV-Vis spectra were
9

ACCEPTED MANUSCRIPT
measured using an Olis Cary-14 spectrophotometer conversion instrument custom-modified to
accommodate a Janis Research STVP-100 optical cryostat cooled to 80 K with liquid N2.
2
.8. Molecular Dynamics Simulations. Explicit-solvent MD simulations of mtCYP51, as well as
the H259A and T260V mutants were performed as previously described [27, 28] using the
CHARMM22 force field [38] and the NAMD 2.9 code[39]. Before initiating simulations, a 5000
step conjugate gradient (CG) minimization was performed to eliminate unfavorable inter-atomic
contacts. Water and ions were then equilibrated around the protein using NVT simulation for
5
00 ps with protein atoms fixed at their initial positions. The simulation was minimized for an
additional 5000 CG steps and switched to the NPT ensemble to equilibrate protein with the
solvent for a further 500 ps. Protein C atoms were restrained near their initial positions during
α
-1
-2
this process using a 5.0 kcal·mol ·Å harmonic potential. Finally, all constraints were released
and an NPT equilibration was performed for 30 ns to solvate and relax the system from its
crystallographic conformation.
2
.9. Oxygen Diffusion Calculations. The potentials of mean force (PMFs) for gas diffusion were
obtained using the implicit ligand sampling (ILS) method[40, 41]. This technique has previously
produced physically plausible PMFs for monatomic and diatomic gas migration in systems
including hydrogenases [42], oxygenases [43], and heme proteins such as myoglobin [40].
During ILS sampling, a small ligand (gas) molecule is placed at a series of regular grid points
{r} encapsulating the protein and the interaction energy E(r) is computed, assuming only van
der Waals interactions contribute meaningfully through a Lennard-Jones interaction term. This
process is repeated at each point for a total of M ligand conformations and sampled over N
1
0

ACCEPTED MANUSCRIPT
trajectory frames. The potential of mean force A (r) at r can then be computed (Equation 1) at
ILS
a temperature T as where k is Boltzmann’s constant. In practice, this affords a volumetric map
from which isosurfaces may be produced at a given value of the PMF for a gas molecule.
N
M
exp[E (r)]
m,n
AILS (r)  kT log

NM
n1 m1
(1)
Gas diffusion calculations were performed using data from the last 5 ns of NPT equilibration at
3
2
00 K for each system. Frames were sampled every 2.0 ps and ligand configurations sampled in
0 different orientations at each point on a 1.0 Å grid with 23 subsamplings at each grid point.
CHARMM parameters were utilized to treat van der Waals interactions and ligand parameters
for O were assumed from prior, validated calculations [40]. PMF surfaces were constructed
2
using the volmap utility as implemented in Visual Molecular Dynamics [44].
3
. Results and Discussion
While the cognate substrate of mtCYP51 is unknown, it is nonetheless critical to have identified
a model substrate of the enzyme for frutiful mechanistic studies. Two 14α-methylated sterols,
DHL and obtusifoliol, have been identified as substrates for the enzyme. With support from
spinach ferredoxin reductase and a mt ferredoxin to provide electrons, Bellamine and coworkers
demonstrated that mtCYP51 removes the C14 methyl group from both of these sterols,
confirming that they both enter the enzyme active site and are subject to catalytic turnover [25].
While obtusifoliol is demethylated at least five-fold more rapidly by mtCYP51 than DHL, this
substance must be extracted from plant sources and is not readily available. In contrast, DHL can
be prepared by simple and efficient hydrogenation of lanosterol (vide supra), which is
1
1

ACCEPTED MANUSCRIPT
commercially available at relatively low cost. We have therefore adopted DHL as the model
substrate for the studies described herein.
Hallmarks of DHL and other ligands that do not coordinate to the heme iron of mtCYP51
are their common failure to induce a complete low- to high-spin transition of the heme iron that
is measurable as a type-I spectroscopic response. In previous studies of mtCYP51, the fraction of
enzyme converted to the high-spin state did not exceed 18% following titration with HPCD-
encapsulated DHL[45]. X-Ray crystallography studies confirm that two additional ligands,
estriol [46] and 4,4’-dihydroxybenzophenone [24], bind in the active site and their difference
spectra support provocation of a more substantial spin transition than that observed for DHL.
The observed spin transitions are nonetheless incomplete and likewise result in a mixture of the
high- and low-spin species. While many compounds that provoke more substantial spin
transitions are more convenient ligand binding studies, they must be confirmed as substrates
before they can have potential as mechanistic probes.
3
.1. UV-Vis Spectral binding titrations. The Soret bands for all of the ligand-free, ferric CYP51
proteins occured at 419 nm at 25 ºC, consistent with low-spin, hexacoordinate aquaferric resting
states of these enzymes. UV-Vis absorption difference spectra were used to monitor the low- to
high-spin transition of the heme iron as the water ligand is displaced. Partial spin transitions were
observed to depend on the concentration of DHL in all proteins with the exception of T260A.
Titration of the equivalent concentrations of DHL-free HPCD does not result in a spin shift of
any of the reported proteins. Despite the presence of 50 μM DHL, the highest concentration used
in these studies, T260A difference spectra provided no evidence for a change in the spin
equilibrium. Analysis of the absolute spectra of ferric T260A revealed that the fraction of high-
spin enzyme in the presence and absence of 50 μM DHL were 27% and 29%, respectively. The
1
2

ACCEPTED MANUSCRIPT
apparent lack of spin shift or perturbation of the high-spin fraction by DHL supports that T260A
is unlikely capable of accommodating DHL in the active site. On the other hand, mtCYP51,
T260V, and H259A difference spectra displayed classic type-I difference spectra having troughs
at 421, 422, and 422 nm and peaks at 388, 389, and 390 nm, respectively. The K determined for
d
mtCYP51 was 0.20 μM, somewhat lower than the 1 μM value reported by Bellamine and
coworkers[25]. Furthermore, with 50 μM DHL present 15% of mtCYP51 was in the high-spin
state, also similar to previous results [45]. The T260V and H259A mutants bind to DHL with
somewhat lower affinity, having K values of 0.43 and 1.2 μM, respectively. H259A behaved
d
similar to mtCYP51, where at most 18% of H259A could be found in the high-spin state. This
observation is not particularly surprising considering that the distance between H259 and the
heme iron likely precludes its direct perturbation of the iron coordination sphere. The 36% high-
spin T260V is expected since the critical T260 hydroxyl group capable of stabilizing the six-
coordinate aquaferric enzyme is replaced by a methyl group. What was unexpected was that the
spin shift depends on DHL. Thus, the origins of the change in spin equilibrium are more complex
than the loss of a hydrogen bond from T260. The equilibrium shift is more likely attributable to a
ligand induced conformational shift of the valine residue and/or perturbation of the DHL binding
mode.
Titration of CLO into all proteins results in a classic “type II” spectroscopic response,
characterized by a red-shift of the Soret bands consistent with replacement of the water ligand
with the CLO azole heterocycle. In difference spectra obtained by titration of CLO into
mtCYP51, T260A, T260V, and H259A the Soret bands are shifted to 433, 429, 432, and 433 nm,
respectively. Difference spectra of mtCYP51, T260V, and H259A share nearly identical troughs
at 413-414 nm, consistent with depletion of the low-spin aquaferric enzyme by CLO. In contrast,
1
3

ACCEPTED MANUSCRIPT
this trough occurs at 392 nm in the T260A mutant demonstrating that CLO preferably depletes
the high-spin enzyme population. K values for CLO were calculated by fitting the inverses of
d
the absorbance changes and CLO concentrations. The K values of mtCYP51 (3.3 μM), H259A
d
(4.9 μM), and T260A (4.1 μM) are in approximate agreement with the 5 μM value reported by
Bellamine and coworkers.(25) This value was somewhat lower for the T260V (1.1 μM) mutant.
3
.2. Resonance Raman Spectroscopy of Ferric mtCYP51 and Mutants. RR spectra were
measured to investigate the changes in heme electronic structure resulting from HT mutations,
DHL, and azole inhibitor binding. The high-frequency regions of the rR spectra from 406.7 nm
laser excitation are illustrated in Figure 2 and the frequencies of the major markers are compiled
in Table 1. The high frequency region of the spectra contains bands that are sensitive to the
oxidation (ν ), and spin state (ν ) those that are sensitive to the spin state and coordination
4
3
environment (ν and ν ), and the stretching modes of the heme vinyl substituents (νC=C). The
2
10
-1
oxidation state vibrations, ν , occur at 1375 cm in the spectra of mtCYP51 and T260A and at
4
-1
1
374 cm in the spectra of T260V and H259A. These values are as expected for a P450 enzyme
in the ferric oxidation state [33]. Binding of CLO whose imidazole group coordinates directly to
-1
the heme iron, shifts ν at most three cm . The spin-state sensitive bands measured for ligand-
4
-
1
-1
free and DHL-bound mtCYP51 are identical, each appearing at 1505 cm (ν ), 1586 cm (ν ),
3
2
-1
and 1642 cm (ν ). Collectively, these markers are characteristic of the low-spin,
10
hexacoordinate aquaferric enzyme. The high-spin component of mtCYP51-DHL complex
observed in the UV-Vis spectra was not visible in the rR, possibly due to a smaller Raman cross
sections of the high-spin markers. CLO binding to mtCYP51 consistently red shifts of the ν , ν ,
3
2
-1
and ν bands. The largest of these is the eight cm shift of ν , possibly attributable to ligand
1
0
10
charge transfer. Charge transfer from the CLO ligand would weaken the C -Cmeso π bonds of the
α
1
4

ACCEPTED MANUSCRIPT
heme, thereby decreasing the frequency of ν . In mtCYP51, DHL has a negligible effect on
1
0
-1
νC=C, while CLO shifts these to 1619 cm . Regardless of the their initial positions in the spectra,
CLO likewise shifts the νC=C band to the same position in the spectra of the mutants.
The high frequency rR spectra of the mutants confirm the results of UV-Vis absorption studies
that lesions of the HT dyad alter the spin equilibrium of the ligand-free enzyme and change the
sensitivity of the spin equilibrium to ligand binding. T260A exemplifies the former scenario in
which the enzyme is partially in the high-spin configuration in the ligand free form. Furthermore,
the spectrum remains essentially unchanged following the addition of DHL. This observation
underscores the conclusions of the difference spectra that T260A does not bind to DHL. The
-1
predominant ν band in the T260A spectra occurs at 1504-1505 cm , consistent with the low-
3
-1
spin form of the enzyme; however these bands have a shoulder centered at 1489 and 1490 cm
revealing the high-spin subpopulation of the enzyme observed in the UV-Vis spectra. By
contrast, binding of CLO abolishes the ν shoulder from the high-spin component and shifts the
3
remainder of the spectrum to mirror that of the low-spin mtCYP51-CLO complex. While the rR
spectrum of T260V in its unbound form reflects that of ligand-free mtCYP51, the T260V spin
state markers and νC=C bands undergo substantial shifts on binding to DHL. The ν , ν , and ν
3
2
10
-1
bands in the DHL-free enzyme occur at 1504, 1587, and 1641 cm , respectively and are
consistent with the low-spin enzyme. In the presence of DHL however, there are marked shifts of
-1
the ν and ν bands to 1487 and 1570 cm , respectively, both characteristic of the high-spin
3
2
form. Thus, the T260V mutant behaves somewhat similarly to P450cam [33], where substrate
binding largely displaces the axial-coordinated water molecule resulting in the pentacoordinate,
high-spin enzyme. Furthermore, the νC=C band of this mutant is the most sensitive to DHL, which
-1
induces a 7 cm shift in accord with change in the environment surrounding the heme vinyls.
1
5

ACCEPTED MANUSCRIPT
DHL binding to H259A similarly induces a partial low- to high-spin transition, though it is less
pronounced than that observed with T260V. The CLO-bound forms of both T260V and H259A
reflect those of the mtCYP51-CLO complex and are similarly in the low-spin state.
The low frequency regions of the rR spectra are illustrated in Figure 3 and the
frequencies of the major markers compiled in Table 1. The low frequency region contains the
heme skeletal stretching modes as well as bending vibrations of the peripheral vinyl and
propionate moieties. As is typical in the rR spectra of P450 enzymes, the heme stretching modes
-1
ν and ν , centered at 675 and 343 cm are resistant to mutations and ligand binding. In
7
8
-
1
mtCYP51, the propionates bending mode (δpropionates) appears at 380 cm and the peripheral
-
1
vinyls bending mode (δvinyl) appears at 409 cm . These assignments are made based on the
complete vibrational assignments of the heme in myoglobin and cytochrome c [47, 48]. Changes
in the chemical environments surrounding these functional groups are reflected in shifts of the
corresponding frequencies in the rR spectrum. The δpropionates frequency positions provide insight
into the hydrogen bonding and/or ionic interactions surrounding the heme propionates that
anchor the cofactor into the active site. The δvinyl frequencies are sensitive to the coplanarity of
the vinyl functional group with the heme. Out-of-plane vinyls generally have higher δvinyl than
those that are coplanar with the heme macrocycle. The δvinyl frequency decrease is due to the
enhanced withdraw of electron density from the vinyl when it is in the heme plane, as opposed to
an increase in this frequency when the overlap between the vinyl and -electron systems are
attenuated when the vinyls are nearly perpendicular to this plane [49,50]. In mtCYP51, DHL
-1
-1
only induces a one cm shift of δ
and a three cm shift of δpropionates, supporting that this
vinyl
ligand does not induce marked conformational or environmental changes of the heme peripheral
substituents. Mirroring the νC=C band shifts observed in the high frequency region, CLO binding
1
6

ACCEPTED MANUSCRIPT
-1
-1
shifts the δvinyl vibrations by 11 cm to 420 cm . This observation is in accord with the previous
-1
results of Matsuura and coworkers who observed a 13 cm shift of the δ
vibration with
vinyl
ketoconazole, 4-phenylimidazole, and fluconazole[51]. Matsuura and coworkers did not observe
a corresponding shift of the δpropionates and thus concluded that the azoles did not interfere with the
heme propionates. For comparison, we measured rR spectra of mtCYP51 with 300 μM
-1
fluconazole and ketoconazole. The measured values of δpropionates and δvinyls at 385 cm and 426
-1
cm respectively were identical to the previously reported measurements [51]. Shifts observed in
the δvinyl vibrations of mtCYP51 spectra uniformly support that the azole ligands shift of the
heme vinyls into conformations out of the heme plane. There are corresponding shifts of the νC=C
to higher frequencies in the high-frequency region of these spectra; however there is evidence
opposing that there is a correlation between νC=C and the orientation of the vinyl group relative to
the heme plane [52] and so these are excluded from further analyses of vinyl conformations.
Moreover, these spectra similarly lacked evidence of perturbations to the heme propionate
-1
environments. However, in spectra of CLO-mtCYP51 the dominant δpropionates band at 379 cm
-1
has an obvious shoulder at 367 cm . Thus, the observation of a second, overlapping band
confirms that CLO imparts conformational heterogeneity to the heme propionates. Furthermore,
-1
the magnitude (12 cm ) of this partial shift supports that the number and/or strength of the
hydrogen bonds differs among the accessible configurations of the mtCYP51-CLO complexes. A
-1
1
0 cm shift is typical for the cleavage of a propionate hydrogen bond as demonstrated for
thromboxane synthase and myoglobin [53, 54]. In summary, DHL does not induce perceptible
changes in the conformations and/or environments of the heme peripheral substituents, both
fluconazole and ketoconazole only perturb the heme vinyls, and CLO uniquely perturbs both the
vinyls and the propionates.
1
7

ACCEPTED MANUSCRIPT
RR studies of CLO and DHL binding to the mutants revealed that while many of the
effects on the heme periphery were similar to the corresponding mtCYP51 complexes, some
notable differences were observed. The effects of CLO on both the δvinyls and δpropionates are
-1
-1
-1
ubiquitous. CLO red shifts the δvinyl vibrations by 17 cm , 10 cm , and 12 cm in the T260A,
T260V, and H259A mutants, respectively. The dominant δpropionates bands appears at 377, 379,
-1
and 379 cm in the T260A, T260V, and H259A mutants, respectively, in accord with this ligand
uniformly inducing out-of-plane conformations of the heme vinyl groups. Like mtCYP51, these
-1
have shoulders at 364, 369, and 367 cm , so the apparent conformational heterogeneity induced
by CLO is conserved among all of the CLO-enzyme complexes. In response to DHL binding,
changes to the low frequency are subtle though nonetheless significant. Compared to ligand-free
-1
-1
mtCYP51, only the δpropionates of T260A shift 4 cm to 384 cm . In contrast, the vinyl and
propionate vibrational markers of T260V are remarkably sensitive to DHL binding. On binding
-1
-1
DHL, the T260V δpropionates intensity at 381 cm decreases and shifts to 378 cm while a new
-1
band appears at 364 cm . Apparently, the δ
of T260V behave similarly to when CLO is
propionates
present by shifting to a frequency telling of substantial changes in the hydrogen bonding
environments surrounding the propionates. DHL binding to T260V likewise shifts the δvinyl
-1
-1
marker from a relatively sharp band at 410 cm to a broad δ
centered at 424 cm consistent
vinyl
with DHL inducing conformational heterogeneity into the vinyls with concurrent loss of
coplanarity between the vinyls and the heme plane. Taken together with spectra of the high
frequency region, both the spin state and heme conformation are sensitive to DHL in the active
site. Finally, the low frequency rR spectrum of the H259A mutant reflects that of the wild type;
-1
-1
however, there are minor two cm and three cm shifts of the δ
and δvinyl modes
propionates
supporting that this mutation causes only minor conformational or environmental changes of the
1
8

ACCEPTED MANUSCRIPT
heme peripheral groups. It appears then that the mutants each respond similarly to mtCYP51 on
binding CLO, however in contrast HT mutations result in a continuum of sensitivity of the heme
peripheral environment to DHL.
3
.3. MD Simulations. MD simulations were performed in explicit water to explore the
conformational landscapes of the amino acid side chains bounding the active site, DHL, and the
heme cofactor to understand the structural underpinnings for the rR observations. On the
timescale of the simulations, these proteins maintain their overall fold and no major
conformational changes were observed. There are however visible differences in the orientation
of DHL and the environments of the heme propionates. Illustrated in Figure 4A, the final
trajectory point from the 30 ns simulations were superimposed to illustrate the relative position
of DHL and the chemical environments of the heme propionates following MD equilibration. All
simulations illustrate that the 17β-aliphatic chain of DHL is oriented above the propionates while
the steroid skeleton lies above the porphyrin. The binding modes of DHL in mtCYP51 and
H259A are quite similar, differing primarily in the conformation of the 17β-chain. In T260V
however, DHL is shifted away from the propionates along the extended axis of the molecule and
upward toward the roof of the active site. Our expectation was that the valine side chain would
occupy a similar volume and spatial orientation as that of threonine. To our surprise, the MD
simulations showed that the probable orientation of the valine methyl is oriented away from the
iron and toward the heme periphery. This orientation places the additional methyl group adjacent
to the heme 2-vinyl, resulting in a migration of the vinyl group below the heme plane. This
-1
prediction is validated in the rR spectra of this mutant, in which DHL binding induces a 14 cm
shift of the δvinyl frequencies, consistent with departure of at least one of the heme vinyls from the
heme plane. Illustrated in Figure 4B, simulations also predict that HT mutations result in
1
9

ACCEPTED MANUSCRIPT
environmental changes around one or both of the heme propionates. In the T260V-DHL
complex, hydrogen bonds between the 7-propionate, Q72 and H392 are disrupted that are
otherwise present in mtCYP51 or H259A. Rearrangement of the amino acids observed in the
MD simulations are corroborated by the magnitude of shifts observed in rR spectra. In the
T260V mutant, simulations predict the disruption of two hydrogen bonds, consistent with the
large shift observed in the propionate vibrations. In the H259A-DHL complex, only H101 moves
away from the 6-propionate. The minor reorientation of residues observed in the H259A
simulation are consistent with the correspondingly small vibrational shifts observed in the rR
spectra.
3
.4. Oxyferrous formation and decay kinetics. The oxyferrous species is the first important
branch point in P450 catalysis from which the catalytic cycle can proceed to productive
monooxygenation of substrate or uncouple to afford the ferric enzyme and superoxide anion. The
oxyferrous has been characterized in a few of cytochromes P450 and only in those bearing the
conserved acid-alcohol pair. To establish the CYP51 mechanism, it is critical that we first
understand the impact of the active site on the stability of the oxyferrous complex. To this end,
we have measured the rates of oxygen binding (k ) and auto-oxidation (k ) of oxyferrous
1
2
mtCYP51 and the HT mutants. These rate constants are compiled in Table 1.
For comparison to existing literature and to validate the approach in our hands, these
rates were also measured for CYP101 in the presence of 1 mM camphor. Spectra collected after
mixing ferrous CYP101 with oxygen-saturated buffer and the basis spectra resulting from SVD
analysis are illustrated in the Figure 5A,B. Oxyferrous CYP101 displays a Soret band at 421 nm
and single peak at 555 nm in the vicinity of the α and β absorption bands. The decay of these
spectroscopic features correspond to a shift of the Soret band to 391 nm, consistent with the
2
0

ACCEPTED MANUSCRIPT
formation of the high-spin, camphor-CYP101 complex and the superoxide anion. In experiments
with CYP101, the ferrous enzyme was rapidly depleted within the dead time of the instrument
precluding the calculation of a reliable rate constant for O binding. The rate constant measured
2
−
1
for auto-oxidation of the oxyferrous enzyme was 0.0067 s at 25 ºC. This value is nearly
−
1
identical to the 0.0074 s value measured by Eisenstein and coworkers in their stopped-flow
kinetic analysis at 26 ºC. Moreover, these CYP101 values are consistent with the literature
−1
−1
values of 0.004 s measured at 20 ºC [55] and 0.0043 s measured by Peterson and coworkers
at 23 ºC [56].
In the mtCYP51 stopped flow experiments, spectra of ligand-free ferrous enzyme were
collected over 0.5 s (400 scans/s) after mixing the anaerobic, ferrous enzyme with O -saturated
2
buffer. Unlike the ferric enzymes that are represented by a mixture of spin states, UV-Vis spectra
support the presence of a single ferrous spin state. Both mtCYP51 and mutants have a single
Soret band at 424-425 nm and α- and β-bands at 528 and 558 nm, respectively regardless of
whether DHL is present. Spectroscopic evidence for the single ferrous spin species justifies the
k1
k2
2

3

2
3

2
use of the Fe Fe O Fe O kinetic model. Spectra for wild-type mtCYP51 in the
presence and absence of DHL are illustrated in Figure 5C-F. After mixing, the Soret band shifts
from 424 nm (at 2.5 ms) to 418 nm (0.5 s), consistent with conversion of the ferrous to the low-
spin ferric enzyme. In addition, SVD analysis reveals a third basis vector with a Soret band at
4
22 nm whose accumulation and decay coincide with dissipation of the ferrous and appearance
of the ferric enzyme. This basis vector represents the transient accumulation of the oxyferrous
intermediate. The pseudo-first order rate constant for formation of oxyferrous mtCYP51 is 63
-1
-1
-1 -1
μM ·s , while the presence of DHL attenuates k to 15 μM ·s . The results of previous MD
1
simulations published by our group(28) support that DHL largely occludes the putative substrate
2
1

ACCEPTED MANUSCRIPT
entrance channel [57]. We speculate that the rate constant for DHL-free mtCYP51 corresponds
to the fully “open” form of the enzyme. This assertion is founded in the crystal structures of
ligand-free ferric mtCYP51, where the BC-loop that bounds the access channel is disordered [46,
5
8] Assuming this channel is also the predominant pathway for O diffusion into the active site,
2
it is not surprising that the presence of DHL reduces the O binding rate. The presence of DHL
2
has little impact on the rate of oxyferrous mtCYP51 auto-oxidation. This behavior is quite
different from that exhibited by other P450 enzymes. For example, removal of camphor from
CYP101 increased the auto-oxidation rate constant ~ 3.5-fold at 26º C [55]. Likewise, Denisov
and coworkers reported that auto-oxidation of nanodisc-incorporated CYP3A4 increased 60-fold
following the removal of testosterone and 150-fold following the removal of bromoergocryptine
[
59].
HT mutations do change the O -binding and oxyferrous oxidation kinetics in a ligand
2
dependent manner. In accord with the previous evidence that DHL does not bind to T260A, the
-1
-1
rate constants for O binding are ~31 μM ·s , regardless of whether DHL is present. This rate is
2
only half that determined for mtCYP51, therefore this mutation must alter the primary O2
diffusion pathway. Since DHL is expected to occlude the primary O diffusion pathway in these
2
enzymes, increases in O binding rates for both T260V-DHL and H259A-DHL seemed
2
counterintuitive. The rate of O binding to the T260V is reduced approximately four fold relative
2
to that of the mtCYP51. Interestingly, the rate for T260V-DHL O binding is twice that
2
determined for mtCYP51-DHL. The O binding rate constant to H259A is attenuated by half
2
relative to mtCYP51; however, the rate in the presence of DHL is among the largest measured in
these studies and comparable to DHL-free mtCYP51. We speculate that these mutations stymie a
DHL-induced conformational shift that would otherwise partially occlude the primary O2
2
2

ACCEPTED MANUSCRIPT
diffusion pathway. Nevertheless, we use the implicit ligand sampling method, described below,
to glean insight into the structural bases for these seemingly paradoxical rates. Unlike those for
-1
O binding, the rates of oxyferrous auto-oxidation are mostly ~2-3 s . The single exception in
2
these studies is the T260V-DHL complex. In this case, the observed rate is nearly three-fold
greater than that measured for the mtCYP51-DHL complex. It is tempting to attribute the relative
instability of this oxyferrous species to the loss of the stabilizing effect imparted by a T260
hydrogen bond and MD simulations support that this effect may be at least partially due
perturbation of the heme geometry. Such changes in the electronic structure of the heme iron
could weaken the Fe-O bond of the incipient oxyferrous species, explaining the elevated auto-
2
oxidation rate observed in T260V-DHL. Since HT mutations undoubtedly affect the architecture
of the active site to afford changes in the rates of O binding and auto-oxidation, it is not prudent
2
to interpret these results based on local property changes induced by replacing amino acids in the
mtCYP51 sequence alone. Taken together with the rR measurements, these results support that
there are widespread changes in the active architecture induced by HT mutations.
3
.5. Cryogenic UV-Vis Absorption Spectroscopy. To verify that the SVD-derived absorption
spectra correspond to authentic intermediates, we measured the UV-Vis spectra of the ferric,
ferrous, and the oxyferrous species at 80 K in aqueous-glycerol matrices. While spectra of the
ferric and ferrous enzymes can be easily measured at ambient temperature, the half-life of the
oxyferrous species is at most 0.38 s at 25 ºC (H259A; Table 2), which precludes using the
standard measurement techniques. The oxyferrous intermediates were prepared by bubbling O2
into buffer containing 75% glycerol at 4 ºC, followed by freezing in liquid N . The large glycerol
2
content affords an optically transparent glass that when frozen is ideal for optical spectroscopy at
low temperatures[60, 61]. If the cold oxyferrous samples were allowed to anneal to room
2
3

ACCEPTED MANUSCRIPT
temperature, they rapidly auto-oxidized to the ferric oxidation state. Ferrous samples could be
annealed to room temperature without appreciable oxidation to the ferric state. The Soret band
positions for each species with and without DHL are compiled in Table 3. The complete
cryogenic spectra of mtCYP51 and mtCYP51-DHL are illustrated in Figure 6, while those of the
mutants are available in Figure S5 of the Supporting Information. In general, spectra of the
ferric, ferrous, and oxyferrous species in frozen aqueous glycerol matrices are very similar to
those assigned to the basis vectors resulting from the SVD calculations. In close agreement with
the kinetic analysis, oxyferrous, ferrous, and ferric mtCYP51 have Soret bands at 424, 422, and
4
18 nm, respectively while those for the mtCYP51-DHL complex are at 420, 422, and 418 nm.
The minor differences between the cryogenic spectra and SVD basis vectors may be attributed to
the high glycerol content of the buffer and the reduction in temperature broadening of the Soret
band at low temperature. The behavior of the ferric state of the T260V mutant in the presence of
DHL deviates from this behavior. At 80 K, the Soret band of the T260V mutant with DHL is
split, having maxima at 395 and 415 nm characteristic of a mixture of the high- and low-spin
states of the heme iron. Since the high-spin component is not visible at ambient temperature, this
observation supports that in the presence of DHL that the low-spin configuration is an electronic
excited state of the heme core. The percentage of T260V in the high-spin state at 80 K is 55%.
The increase in high-spin content as the temperature is lowered indicates that the high-spin state
is actually the ground electronic state of the T260V-DHL complex. Hence, the energetic ordering
of the high- and low-spin states is inverted in the T260V mutant and this observation may be
attributable to the predicted distortion of the heme macrocycle in this mutant. The T260V mutant
could therefore prove useful as a model system in future studies of P450 spin equilibria.
2
4

ACCEPTED MANUSCRIPT
3
.6. Oxygen Diffusion Calculations. To reconcile the observations made in the stopped-flow
kinetic experiments measuring O binding, we computed three-dimensional PMFs for O
2
2
diffusion using the ILS approach and the final 5 ns of the MD simulations. This method provides
a comprehensive overview of how mutation-induced side chain dynamics affect the O diffusion
2
pathways and their energetic barriers. Owing to the computational expense of the ILS sampling
method for these systems and our primary interest in pathways available for O diffusion into the
2
DHL complexes, we limited our simulations to the mtCYP51-, T260V-, and H259A-DHL
complexes. Illustrated in Figure 7, the ILS simulations indicate that there is a common,
predominant path for O transport into each of the CYP51 proteins.(Figure 7, A, C, and F) This
2
common channel extends from the protein surface adjacent to A73 in the loop immediately N-
terminal to the B’-helix, above the heme propionates, and terminating at the heme iron under the
C and D rings of DHL. While continuous pathways are evident in simulations of both the WT
and H259A mutants, only three disconnected regions consistent with such a pathway are visible
-1
at a contour value of 2.5 kcal∙mol in simulations of the T260V mutant. A second channel begins
at the protein surface and terminates prematurely between the H259 and T260 residues of
mtCYP51 and T260V mutant. (Figure 7 B and F) In the case of the T260V mutant, it is possible
that the length of the simulation was not sufficient to capture the opening of a second channel to
support the observed O binding rate. Conversely, in the H259A mutant, this second channel
2
extends deep into the active site and terminates near the heme iron flanked by the A/B rings of
the substrate and T260. (Figure 7D) The presence of this second O migration pathway with a
2
comparable free energy barrier for diffusion as the common pathway explains the accelerated
rate observed in the H259A-DHL complex. The opening of this channel is afforded by the loss of
an ion pair between H259 and E173 by replacing the former residue with alanine. Moreover, in
2
5

ACCEPTED MANUSCRIPT
H259A this second channel is populated by a hydrogen-bonded network of water molecules that
extends from bulk solvent at the protein surface to the heme (Figure 7E), reinforcing that
molecules of similar size to O can exploit this channel to gain entry in to the enzyme active site.
2
4
. Summary and Conclusions
We have characterized mtCYP51 and three variants bearing mutations in the HT dyad and
demonstrate the utility of concomitant MD simulations to interpret disparate spectroscopic
observations in the context of active site structure. Evidenced by similar binding affinities for
DHL and CLO, the mutations are well tolerated. RR demonstrates changes the heme’s electronic
structure, conformations of the peripheral groups, propionate environment, and DHL binding
orientation. The greatest perturbations are observed in the T260V mutant. T260V active site
changes are evident in the relevant rR vibrational signatures and UV-Vis spectra at 80K. The
conclusions of these spectroscopic studies are supported by the results of molecular dynamics
simulations. Despite the apparent conservation of volume when mutating threonine to valine, in
this case the result is a comprehensive change in the heme environment.
This study also describes the kinetic studies of O binding to the ferrous enzymes as well
2
as auto-oxidation of the corresponding oxyferrous intermediates. The decrease in O binding rate
2
in mtCYP51 in the presence of DHL is not surprising, since the ligand is expected to occlude a
prominent diffusion pathway. Paradoxically, T260V-DHL and H259A-DHL demonstrate two-
and four-fold larger rate constants for O binding than mtCYP51-DHL. ILS of the respective MD
2
trajectories revealed the probable structural bases for these observations. These clearly illustrated
alternative O diffusion pathways into the T260V and H259A active sites. This application of the
2
2
6

ACCEPTED MANUSCRIPT
ILS approach demonstrates its value in supporting the interpretation of kinetic studies of
diatomic molecule diffusion into proteins.
2
7

ACCEPTED MANUSCRIPT
Abbreviations
CCD
CLO
CG
Charge-coupled device
Clotrimazole
Conjugate gradient
CYP51
Sterol 14α-demethylase
CYP101 Cytochrome P450cam
DHL
24,25-Dihydrolanosterol
-hydroxypropyl-β-D-cyclodextrin
HPCD
2
FLU
ILS
KET
LB
Fluconazole
Implicit Ligand Sampling
Ketoconazole
Luria-Bertani
MD
mt
Molecular Dynamics
Mycobacterium tuberculosis
NAMD Nanoscale Molecular Dynamics
NPT
PMF
NVT
rR
Isobaric-Isothermal Ensemble
Potential of Mean Force
Canonical Ensemble
Resonance Raman
SVD
Singular-Value Decomposition
2
8

ACCEPTED MANUSCRIPT
Acknowledgements
Computational resources were provided by the Ohio Supercomputer Center. The authors are
grateful to Ilia Denisov (University of Illinois Urbana-Champaign) for helpful advice guiding the
preparation of P450 enzymes in frozen aqueous glycerol matrices.
Supporting Information
Site-directed mutagenesis primers, complete difference spectra for CLO and DHL, stopped flow
scans and SVD analysis of the mutants, and low-temperature spectra of ferric, ferrous, and
oxyferrous mutants are available.
References
[
[
[
1]
2]
3]
M. Sono, M. P. Roach, E. D. Coulter, J. H. Dawson, Chem. Rev. 96 (1996) 2841-2888.
B. Meunier, S. P. de Visser, S. Shaik, Chem. Rev. 104 (2004) 3947-3980.
I. G. Denisov, T. M. Makris, S. G. Sligar, I. Schlichting, Chem. Rev. 105 (2005), 2253-
2
277.
[
[
4]
5]
F. P. Guengerich, Curr. Drug Metab. 2 (2001) 93-115.
Ortiz de Montellano, P. R. (2005) Cytochrome P450 : structure, mechanism, and
biochemistry, Vol. 3, Kluwer Academic/Plenum Publishers, New York.
J. Rittle, M. T. Green, Science, 330 (2010) 933-937.
[
[
6]
7]
C. M. Krest, E. L. Onderko, T. H. Yosca, J. C. Calixto, R. F. Karp, J. Livada, , J. Rittle,
M. T. Green, J. Biol. Chem. 288 (2013) 17074-17081.
[
[
8]
9]
S. Shaik, D. Kumar, S. P. de Visser, A. Altun, W. Thiel, Chem. Rev. 105 (2005) 2279-
2
328.
V. Guallar, R. A. Friesner, J. Am. Chem. Soc., 126 (2004) 8501-8508.
2
9