CYP2D6 allelic variants *34, *17-2, *17-3, and *53 and a Thr309Ala mutant display altered kinetics and NADPH coupling in metabolism of bufuralol and dextromethorphan and altered susceptibility to inactivation by SCH 66712

Article abstract of DOI:10.1124/dmd.117.079871

Metabolic phenotype can be affected by multiple factors, including allelic variation and interactions with inhibitors. Human CYP2D6 is responsible for approximately 20% of cytochrome P450–mediated drug metabolism but consists of more than 100 known varian

Full text of DOI:10.1124/dmd.117.079871

DMD Fast Forward. Published on May 21, 2018 as DOI: 10.1124/dmd.117.079871
This article has not been copyedited and formatted. The final version may differ from this version.
DMD #797871
ꢀ
CYP2D6 allelic variants *34, *17-2, *17-3, and *53 and a Thr309Ala mutant display
altered kinetics and NADPH coupling in metabolism of bufuralol and dextromethorphan
and altered susceptibility to inactivation by SCH 66712
Sarah M. Glass, Cydney M. Martell, Alexandria K. Oswalt, Victoria Osorio-Vasquez,
Christi Cho, Michael J. Hicks, Jacqueline M. Mills, Rina Fujiwara, Michael J. Glista,
Sharat S. Kamath and Laura Lowe Furge
Department of Chemistry, Kalamazoo College, Kalamazoo, MI, USA
ꢀ
1

DMD Fast Forward. Published on May 21, 2018 as DOI: 10.1124/dmd.117.079871
This article has not been copyedited and formatted. The final version may differ from this version.
DMD #797871
Running Title: Inactivation of allelic variants of CYP2D6
Corresponding Author: Laura Lowe Furge, Department of Chemistry, Kalamazoo
College, 1200 Academy Street, Kalamazoo, MI 49006. Tel: 269-337-7020; Fax: 269-
3
37-7251. Email: Laura.Furge@kzoo.edu.
Number of Text Pages: 49
Number of Tables: 5
Number of Figures: 6
Number of References: 56
Number of words in Abstract: 198
Number of words in Introduction: 741
Number of words in Discussion: 1454
Abbreviations used: SCH 66712, 5-fluoro-2-[4-[(2-phenyl-1H-imidazol-5-yl)methyl]-1-
piperazinyl]pyrimidine; CYP, cytochrome P450 enzyme; HLM, human liver microsomes;
HPLC, high performance liquid chromatography; ESI, electrospray ionization; LC, liquid
chromatography; MS, mass spectrometry; m/z, mass to charge ratio; CID, collision
induced dissociation; TFA, trifluoroacetic acid; GSH, glutathione; NAC, N-acetyl
cysteine; CuOOH, cumene hydroperoxide; XIC, extracted ion chromatogram; MBI,
mechanism-based inactivator.
ꢀ
2

DMD Fast Forward. Published on May 21, 2018 as DOI: 10.1124/dmd.117.079871
This article has not been copyedited and formatted. The final version may differ from this version.
DMD #797871
Abstract
Metabolic phenotype can be affected by multiple factors including allelic variation and
interactions with inhibitors. Human CYP2D6 is responsible for approximately 20% of
CYP-mediated drug metabolism but consists of over 100 known variants with several
variants being commonly found in the population and others quite rare. Four CYP2D6
allelic variants - three with a series of mutations distal to the active site (*34, *17-2, *17-
3
) and one ultra-metabolizer with mutations near the active site (*53), along with
reference *1 and an active site mutant of *1 (Thr309Ala) were expressed, purified, and
studied for interactions with the typical substrates dextromethorphan and bufuralol and
the inactivator SCH 66712. *34, *17-2, and *17-3 displayed reduced enzyme activity and
NADPH coupling while producing the same metabolites as *1 suggesting a possible role
for Arg296 in NADPH coupling. Though a higher activity variant, *53 displayed similar
NADPH coupling to *1 but was less susceptible to inactivation by SCH 66712. The
Thr309Ala mutant showed similar activity as *1 but with greatly reduced NADPH
coupling. Overall, these results suggest that kinetic and metabolic analysis of individual
CYP2D6 variants is required to understand their possible contributions to variable drug
response and the complexity of personalized medicine.
ꢀ
3

DMD Fast Forward. Published on May 21, 2018 as DOI: 10.1124/dmd.117.079871
This article has not been copyedited and formatted. The final version may differ from this version.
DMD #797871
Introduction
CYP2D6 is responsible for approximately 20% of CYP-mediated drug
metabolism though it only makes up about 2-5% of the total P450 content in the human
liver (Guengerich, 2015; Rendic and Guengerich, 2015). Examples of CYP2D6
substrates include tamoxifen, codeine, bufuralol, debrisoquine, and dextromethorphan
among many others (Ingelman-Sundberg, 2005).
There are over 100 different allelic variants of CYP2D6 with metabolic profiles
ranging from poor to ultra-rapid (Zanger and Schwab, 2013). Each allelic variant is
described by star allele nomenclature where variants are numbered in the order of their
discovery (Robarge et al., 2007). Using this naming system, 2D6*1 is the reference form
to which all other sequences are compared. Identified mutations and possible metabolic
effects
associated
with
each
variant
can
be
found
on
the
PharmacogeneVariation Consortiumat www.PharmVar.org.ꢀ
Previously our group investigated structural fluctuations using molecular
dynamics of four allelic variants - *34, *17-2, *17-3, and *53, in comparison with the
reference *1 (de Waal et al., 2014). *34 was chosen because it has an Arg296Cys
alteration that is commonly found in allelic variants of CYP2D6 (Table 1). *17 was
chosen for its clinical relevance since it is found in 9-34% of Africans and African-
Americans (Ingelman-Sundberg, 2005; Zhou, 2009a). In addition, *17-2 and *17-3
provide a series of sequential amino acid changes that include the Arg296Cys change in
*
34 (Table 1). All three of these allelic variants are associated with reduced activity
(Oscarson et al., 1997; Zhou, 2009a; Sridar et al., 2011). *53 was included in our study
ꢀ
4

DMD Fast Forward. Published on May 21, 2018 as DOI: 10.1124/dmd.117.079871
This article has not been copyedited and formatted. The final version may differ from this version.
DMD #797871
because it may cause an ultra-metabolizer phenotype without a gene-duplication event
(Sakuyama et al., 2008; Muroi et al., 2014).
Allelic variants of CYP2D6 can lead to altered drug responses since CYP2D6
variants have been shown to affect the pharmacokinetics of ~50% of 2D6-metabolized
drugs (Ingelman-Sundberg, 2005). Personalized medicine is a way to combat differences
in drug responses through adjusting doses according to genotype and phenotype
(
Kirchheiner and Rodriguez-Antona, 2009) or prescribing different medications
Chicago, 2016). As the demand grows for personalized medicine, there will be a need
(
for understanding the effects of allelic variants, especially on major drug metabolizing
enzymes like CYP2D6 (Yamazaki, 2017).
Inhibitors can also affect CYP2D6 phenotype. Inhibition can be reversible or
irreversible. When inhibition is irreversible and is caused by a substrate that is
metabolized into a reactive intermediate that covalently modifies the enzyme, it is
classified as a mechanism-based inactivator (MBI). MBIs can alter extensive and
intermediate metabolizer phenotypes to become poor or null metabolizer phenotypes.
Mechanism-based inactivation can occur through three mechanisms: covalent
modification of an amino acid in the enzyme, coordination to the heme, or causing the
heme to become covalently bound to the enzyme (Correia and Hollenberg, 2015).
Inactivation is NADPH-, time-, and concentration-dependent and creates a loss of
enzyme activity for an extended period of time since the enzyme is rendered completely
inactive and must be synthesized anew.
The most potent MBI of CYP2D6 is SCH 66712 (Palamanda et al., 2001; Nagy et
al., 2011; Livezey et al., 2012). SCH 66712 was a lead compound as a human dopamine
ꢀ
5

DMD Fast Forward. Published on May 21, 2018 as DOI: 10.1124/dmd.117.079871
This article has not been copyedited and formatted. The final version may differ from this version.
DMD #797871
receptor antagonist but was dropped after it was identified as a CYP2D6 inactivator
Palamanda et al., 2001). Our group previously determined that SCH 66712 inactivates
D6 through protein adduction, though the site of adduction has yet to be identified
Nagy et al., 2011). Molecular docking experiments suggested that Thr309, a nucleophile
(
2
(
in the active site, might be the possible target (Nagy et al., 2011). Thus, in the current
study we have also included a CYP2D6 active site mutant, Thr309Ala. Thr309 is thought
to be involved in the oxidation mechanism, regioselectivity, and determining heme spin
equilibrium (Keizers et al., 2005a; Bonifacio et al., 2006). Thr309 may also serve as a
nucleophile for reactive inactivating electrophiles as the corresponding threonine has
been the site of inactivation for MBIs with other CYPs (Gay et al., 2011; Lin et al., 2011;
Nagy et al., 2011; Zhang et al., 2011).
Allelic variants and MBIs can cause adverse drug-drug interactions and lead to
serious side effects and increase the cost of healthcare (Classen et al., 1997). Gaining a
better understanding of mechanism-based inactivators and allelic variants is one step that
can be taken to reduce this toxicity (Yamazaki, 2017).
The current study aims to characterize the activity of a few CYP2D6 variants. The
results presented with the allelic variants allow for a better understanding of enzyme
structure/function of CYP2D6 and the effects of amino acid changes on substrate
catalysis and inactivation.
ꢀ
6

DMD Fast Forward. Published on May 21, 2018 as DOI: 10.1124/dmd.117.079871
This article has not been copyedited and formatted. The final version may differ from this version.
DMD #797871
Materials and Methods
Chemicals. SCH 66712 was obtained from Schering-Plough Research Institute
(now Merck & Co., Inc., Kenilworth, NJ) and reconstituted in water for use in assays
described below. Ultra-pure solvents (water, acetonitrile, and methanol) for MS were
purchased from EMD Chemicals, Inc. (Gibbstown, NJ). All other solvents were HPLC
grade and purchased from Sigma-Aldrich (St. Louis, MO). Bufuralol (mixture of
enantiomers) and hydroxyl-bufuralol were from Toronto Research Chemical (Ontario,
Canada). Nickel-NTA agarose was from Qiagen (Germantown, MD). NuPAGE Novex
1
0% Bis-Tris Mini Gels, SimplyBlue, NuPAGE MES SDS Running Buffer, and
NuPAGE LDS Sample Buffer were from Invitrogen (now Thermo Fischer Scientific,
Waltham, NJ). Cumene hydroperoxide (CuOOH) was purchased from TCI America
(Portland, OR) and used after washing and extraction (vide infra). Potassium phosphate,
NADPH, 1,2-didodecanoyl-sn-glycero-3-phosphocholine phospholipids, acetonitrile,
catalase, super oxide dismutase, dextromethorphan, dextrorphan, chymotrypsin, trypsin,
wide range SigmaMarker molecular weight markers, and all other reagents were
purchased from Sigma-Aldrich.
Preparation of CuOOH. Caution! Strong oxidant and can be explosive. CuOOH
used in metabolite experiments was prepared by diluting 5 mL of CuOOH containing
~
20% aromatic hydrocarbons (as purchased) with 20 mL of CH
was twice washed with 25 mL of 2 M NaOH followed by two washes with water. The
organic layer containing the CuOOH was collected and dried with Na SO . After filtering
layer was removed by roto-evaporation leaving purified CuOOH
2
Cl2. The diluted CuOOH
2
4
2 4 2 2
of Na SO , the CH Cl
ꢀ
7

DMD Fast Forward. Published on May 21, 2018 as DOI: 10.1124/dmd.117.079871
This article has not been copyedited and formatted. The final version may differ from this version.
DMD #797871
that was stored neat at 4 °C. Aqueous dilutions of CuOOH (5 mM) were prepared for use
in metabolite studies.
Enzymes. The starting human *1 plasmid (the DB11 construct containing an N-
terminal truncation to MARQVHSSWNLPP (as previously reported by Gillam et al.,
1
2
995) with a four-His tag on the C-terminal end (as previously described by Hanna et al.,
001) and a rat NADPH-reductase plasmid (Gillam et al., 1993) were gifts from
Professor Fred Guengerich (Vanderbilt University) (Gillam et al., 1993; Gillam et al.,
995; Hanna et al., 2001). Allelic variants *34, *17-2, *17-3 and *53 as well as the
1
CYP2D6_Thr309Ala mutant were prepared by site-directed mutagenesis of the *1
plasmid by Genewiz (Southfield, NJ) and resulting mutations were confirmed by plasmid
sequencing. Recombinant human *1, CYP2D6_Thr309Ala, and allelic variants *34, *17-
2
, *17-3, and *53 were expressed in E. coli along with the pGro7 plasmid for expression
of the chaperone proteins GroEL and GroES (Takara Bio Inc., Shiga, Japan) as
previously described (Gillam et al., 1995; Wang et al., 2012). Briefly, purification
involved preparation of spheroplasts followed by nickel-NTA agarose column
chromatography and de-salting using Pierce Zeba™ spin columns due to the tendency of
CYP2D6 to aggregate upon dialysis. No stabilizing CYP2D6 ligands were included in the
expression or purification steps. P450 reduced CO difference spectra were measured
using the methods of Omura and Sato to calculate active P450 concentration (Omura and
Sato, 1964a).ꢀ Levels of P420 were also determined using the method of Omura and Sato
as summarized by Guengerich (Guengerich, 2014), but only P450 values were used in
calculations of enzyme concentrations for assays described below. Purified CYP2D6
enzymes were also examined for extent of purification by SDS-PAGE on 10%
ꢀ
8

DMD Fast Forward. Published on May 21, 2018 as DOI: 10.1124/dmd.117.079871
This article has not been copyedited and formatted. The final version may differ from this version.
DMD #797871
acrylamide gels. Recombinant P450 NADPH-reductase (reductase) was purified from E.
coli as previously described and analyzed for purity and mass by whole protein MS (Shen
et al., 1989; Holmans et al., 1994).
Determination of K
m
and vmax with CYP2D6 enzymes. Purified CYP2D6
enzymes (0.2 µM) were individually reconstituted with reductase (0.4 µM) and freshly
sonicated 1,2-didodecanoyl-sn-glycero-3-phosphocholine phospholipids (30 µM) for 10
minutes at room temperature. Bufuralol (0-600 µM) or dextromethorphan (0 - 1000 µM)
and potassium phosphate (pH 7.4, 100 mM, final) were added to achieve a final reaction
volume of 100 µL. The reaction mixture was incubated in a 30 °C shaking bath for 3
minutes and then initiated with NADPH generating system (5 mM glucose 6-phosphate,
+
0
.5 mM NADP , and 0.5 units/ml glucose-6-phosphate dehydrogenase). Incubation
temperature of 30 °C rather than 37 °C was selected to ensure more reproducible activity
of the all the variants. Reactions were quenched with 20 µL acetonitrile and placed on ice
after a time previously determined to be in the linear product formation range. Samples
were then centrifuged (16,100 x g) for five minutes and the supernatant was injected onto
a Kinetex 100 Å C18 column (2.6 µM, 4.6 x 100 mm, Phenomenex, Torrance, CA) with
a guard column attached to a Waters e2965 HPLC (Milford, MA) paired with a Waters
4
74 fluorescence detector (Milford, MA). Isocratic mobile phases were used in both
bufuralol and dextromethorphan analysis (30% acetonitrile, 70% water, 1 mM perchloric
acid for bufuralol and 20% acetonitrile, 0.8% acetic acid, 0.05% triethylamine for
dextromethorphan). The excitation and emission wavelengths for bufuralol assays were
2
52 and 302 nm, respectively, and 280 and 310 nm for dextromethorphan. Velocity of
product formed per minute per pmol 2D6 was calculated based on the concentration of
ꢀ
9

DMD Fast Forward. Published on May 21, 2018 as DOI: 10.1124/dmd.117.079871
This article has not been copyedited and formatted. The final version may differ from this version.
DMD #797871
dextrorphan or 1’-hydroxybufuralol product by comparison to a standard curve.
Substrate Depletion Analysis. Assays were completed by the method of Obach
and Reed-Hagen (Obach and Reed-Hagen, 2002). Briefly, reactions contained
dextromethorphan (0.5 – 1000 µM) and *1, *53 or the Thr309Ala mutant (0.6 µM)
reconstituted with reductase (1.2 µM) and lipids (90 µM) in 100 mM potassium
phosphate buffer. Reactions were initiated with NADPH (1 mM) and at 0, 1, 2, 5, 10, 20,
and 30 min aliquots (20 µL) were removed and quenched with acetonitrile (20 µL).
Samples (10 µL) were analyzed by HPLC as described above except the
dextromethorphan peak was quantitated instead of dextrorphan. The depletion rate
constant was then determined by plotting the log percentage dextromethorphan remaining
versus time fitted to a first order decay function. The depletion rates for each
concentration were then plotted against a log scale of dextromethorphan concentration
and the K
m
was determined with the following equation described by Obach and Reed-
)], where kdep is the depletion rate constant,
dep([S]=0) is the theoretical maximum consumption rate at an infinitesimally low substrate
concentration, [S] is the substrate concentration, and K is the Michaelis constant. The
was determined by the inflection point for the resulting curve.
NADPH Coupling to Metabolite Formation. P450 2D6 variants (final
Hagen: kdep = kdep([S]=0) * [1- ([S]/[S] + K
m
k
m
K
m
concentration 0.1 µM) were reconstituted with lipids (30 µM) and reductase (0.2 µM) on
ice for 20 minutes before potassium phosphate buffer (pH 7.4, 100 mM) and bufuralol
(100 µM) or dextromethorphan (100 µM or 600 µM) were added. Prior to the initiation
of the reaction, each sample was heated in a 30 °C water bath with shaking for three
minutes. The reaction was then initiated with NADPH (0.3 mM), transferred to a quartz
ꢀ
10

DMD Fast Forward. Published on May 21, 2018 as DOI: 10.1124/dmd.117.079871
This article has not been copyedited and formatted. The final version may differ from this version.
DMD #797871
microcuvette, and monitored at 340 nm for 10 minutes at 30 °C in a spectrophotometer
Varian Cary-300, now Agilent) (Santa Clara, CA). Final reaction volume was 333 µL.
(
The reactions were quenched with acetonitrile once removed from the spectrophotometer.
The steady state rate of NADPH consumption (nmol NADPH/min/nmol CYP2D6) was
found using the slope of the linear portion of the resulting spectra for the entire 10
minutes. The products were analyzed by HPLC as described above and peaks quantitated
by comparison to the appropriate standard curves to calculate pmol product
formed/min/pmol CYP2D6. Product formation for each reaction was quantified following
quenching of the NADPH consumption reactions. The rate of product formation divided
by the rate of NADPH consumption was used to determine the extent of coupling.
Metabolites. Metabolites formed in separate assays with dextromethorphan and
bufuralol with each variant as well as the Thr309Ala mutant were analyzed using LC-
n
ESI-MS detection and MS . P450 2D6 variants (final concentration 1 µM) were
reconstituted with lipids (30 µM) and reductase (2 µM) on ice for 20 minutes before
potassium phosphate buffer (pH 7.4, 100 mM), substrate (30 µM for dextromethorphan
and bufuralol) and water were added. Samples were placed in a water bath at 30 °C with
shaking for three minutes, and then initiated with either an NADPH generating system
(final reaction volume was 100 µL) or with CuOOH (500 µM). The NADPH generating
+
system consisted of glucose 6-phosphate (5 mM), NADP (0.5 mM), and glucose-6-
phosphate dehydrogenase (0.5 units/ml). The reactions proceeded for 30 minutes before
each sample was quenched with acetonitrile, centrifuged, and analyzed using a Waters
Alliance 2690 HPLC system (Milford, MA) in conjunction with a Thermo-Fisher
Scientific LXQ MS (Waltham, MA). Separation was achieved with a Kinetex C18
ꢀ
11

DMD Fast Forward. Published on May 21, 2018 as DOI: 10.1124/dmd.117.079871
This article has not been copyedited and formatted. The final version may differ from this version.
DMD #797871
column (2.5 µm, 100 Å, 100 x 2.10 mm; Phenomenex, Torrance, CA) and flow rate of
.1 ml/min. The mobile phase for isocratic separation of the dextromethorphan samples
0
was as described above - 20% acetonitrile, 0.8% acetic acid, and 0.05% triethylamine in
MS grade water. For bufuralol samples, the HPLC conditions were modified to a gradient
separation as follows: Solvent A, 10% acetonitrile, 1 mM perchloric acid in MS grade
water; Solvent B: 30% acetonitrile, 1 mM perchloric acid in MS grade water; t = 0 min,
1
00% A; t = 5 min, 100% A; t = 25 min, 0% A; t = 35 min, 0% A; t = 40 min, 100% A; t
55 min, 100% A. The column effluent was directed into an LXQ mass analyzer
=
operated in the positive ion mode using the Xcalibur software package. The ESI
conditions were as follows: sheath gas, 19 arbitrary units; auxiliary gas, 10 arbitrary
units; spray voltage, 5.5 kV; capillary temperature, 275 °C; capillary voltage, 10 V, and
tube lens offset, 15 V.
Spectral Binding Titrations. Spectral binding titrations studies were carried out
with recombinant purified CYP2D6 enzymes (1 µM) in potassium phosphate buffer (100
mM, pH 7.4). The solution was evenly divided between two cuvettes and the experiments
were performed at room temperature using a Cary 300 dual-beam spectrophotometer
(Varian, Inc., Walnut Creek, CA). A baseline correction was recorded (350-500 nm).
Substrate, typically 1-300 µM, was then titrated into the sample cuvette and the
equivalent volume of water was added to the reference cuvette; the spectra were recorded
(350-500 nm) after each addition. The difference between the absorbance maximum and
minimum was plotted against substrate concentration, and the data were analyzed by a
nonlinear regression using KaleidaGraph (Synergy Software, Reading, PA). The
s
dissociation constant, K , was determined using the following quadratic velocity equation
ꢀ
12

DMD Fast Forward. Published on May 21, 2018 as DOI: 10.1124/dmd.117.079871
This article has not been copyedited and formatted. The final version may differ from this version.
DMD #797871
2
1/2
(
Eqn. 1): [CYP2D6 • substrate] = 0.5 (K
represents substrate concentration, E
spectral dissociation constant for the reaction CYP2D6
CYP2D6•substrate].
SCH 66712 Metabolite Analysis. Metabolites of SCH 66712 formed by all
purified CYP2D6 enzymes were determined by LC-ESI-MS as previously described
Nagy et al., 2011; Bolles et al., 2014). Briefly, purified 2D6 enzymes (1 µM), lipids (30
s
+ E
is the total enzyme concentration, and K
substrate
t t s t t t t
+ S ) – [0.25 (K + E + S ) - E S ] , where
S
t
t
s
is the
+
↔
[
(
µM), and reductase (2 µM) were reconstituted for 10 minutes at room temperature
followed by the addition of potassium phosphate (100 mM, pH 7.4), SCH 66712 (100
µM), and ultrapure water in a volume of 200 µL. The mixture was pre-incubated for 3
minutes at 37 °C in a water bath with shaking followed by the addition of NADPH
generating system (30 µL) or CuOOH (500 µM). The reaction proceeded for 40 minutes
at 37 °C with shaking in a water bath. The reaction was quenched with acetonitrile (30
µL) and was placed on ice. The sample was then centrifuged for 5 minutes at 13,800 x g
and the supernatant was collected without disrupting the pellet. The samples were
analyzed using LC-ESI-MS as previously described (Nagy et al., 2011).
Determination of K
individually reconstituted with reductase (6 µM) and freshly sonicated lipids (90 µM) for
0 minutes on ice. To this, potassium phosphate (pH 7.4, 100 mM), superoxide dismutase
80 U/mL), catalase (5 mg/mL), and SCH 66712 (0-48 µM) were added. The final
I
and kinact. Purified CYP2D6 enzymes (3 µM) were
2
(
reaction volume was 200 µL. This primary reaction mixture was incubated in a 30 °C
shaking bath for 3 minutes and then initiated with NADPH (1 mM). Aliquots (10 µL) of
the primary reaction were removed at 0, 1, 2, 5, or 9 minutes and added to the secondary
ꢀ
13

DMD Fast Forward. Published on May 21, 2018 as DOI: 10.1124/dmd.117.079871
This article has not been copyedited and formatted. The final version may differ from this version.
DMD #797871
reaction containing NADPH (1 mM) and dextromethorphan or bufuralol (100 µM) in
potassium phosphate, pH 7.4 (100 mM) (final volume 200 µL). Reactions were quenched
with 30 µL of acetonitrile and placed on ice after 10 minutes. Samples were then
centrifuged (16,100 x g) for five minutes and an aliquot of the supernatant (10 µL) was
analyzed by HPLC as described above for analysis of 1’-hydroxybufuralol or
dextrorphan.
Trapping Agents. *53 (3 µM) was reconstituted with reductase (6 µM) and lipids
(90 µM) for 20 min on ice. After reconstitution, potassium phosphate buffer (100 mM,
pH 7.4) and freshly prepared solutions of NAC (2 mM) or GSH (2 mM) were added in a
final reaction volume of 200 µL. Reactions were pre-incubated for 3 min at 30 °C and
then initiated with NADPH (1 mM), except the no NADPH control reaction that received
an equal volume of water. After 45 min, reactions were quenched with 30 µL of
acetonitrile and placed on ice. Samples were then centrifuged and the supernatant
collected for analysis by LC/MS conditions as described above with a solvent system
consisting of A (0.05% TFA, 10% methanol, 90% water) and B (0.05% TFA, 10%
methanol, 90% acetonitrile). After an initial 5 min hold at 95% A, a linear gradient of
9
5% A to 50% A over 20 min was applied for resolution of reaction components
followed by a 1 min hold at 50% A and then return to the initial conditions.
Partition Ratio. Primary reaction mixtures contained purified *1 (3 µM), the
Thr309Ala mutant (3 µM), or *53 (1 µM), reductase (6, 6, or 2 µM, respectively), lipids
(
30 µM), potassium phosphate buffer (100 mM, pH 7.4), and varying concentrations of
SCH 66712 (0-150 µM) in a final volume of 100 µL. The primary reaction mixtures were
pre-incubated in a shaking water bath at 30 ºC for 3 min and then all primary reactions
ꢀ
14

DMD Fast Forward. Published on May 21, 2018 as DOI: 10.1124/dmd.117.079871
This article has not been copyedited and formatted. The final version may differ from this version.
DMD #797871
except the control were initiated with NADPH (1 mM). The control received an equal
volume of water. The primary reactions were then incubated for 30 min to ensure
inactivation was complete. Aliquots of 10 µL were added to the secondary reactions and
the samples were incubated and analyzed as described above. From a plot of the percent
remaining activity versus [inactivator]/[CYP2D6], the partition ratio was calculated by
the method of Silverman using the intersection of the linear regression of the steeper
slope of the high ratios with the x-axis (Silverman, 1988). All experiments were done in
triplicate.
Molecular Visualization and Docking. Substrates were docked in to the active
site of CYP2D6 (PDB ID 3QM4) using AutoDock Vina (http://autodock.scripps.edu)
(Morris et al., 1998; Huey et al., 2007). Only chain A of 3QM4 and its corresponding
heme cofactor were used; crystallographic waters and prinomastat were removed prior to
docking studies. All molecular structures were produced using PyMOL Molecular
Graphics System, Version v1.7.6.6 (Schrödinger, Portland, OR).
ꢀ
15

DMD Fast Forward. Published on May 21, 2018 as DOI: 10.1124/dmd.117.079871
This article has not been copyedited and formatted. The final version may differ from this version.
DMD #797871
Results
Expression and Purification of CYP2D6 Enzymes. Each CYP2D6 enzyme
expression culture, except *1, showed ~230-350 nmols of P450 per liter of culture as
determined from reduced CO-difference spectra of whole E. coli cells at the time of cell
harvest; *1 showed higher expression at ~650 nmols/liter. Final purification indicated
greater than ~90% homogeneity (Supplemental Fig. 1). Reduced CO-difference spectra at
the end of purification showed 3 – 14 µM P420 compared to 17 – 96 µM P450 (Fig. 1
and Supplemental Table 1). For *17-2 and *17-3, the levels of P420 were prominent and
accounted for ~25-40% of the enzyme present. For use in assays, the concentration of
active enzyme for each variant was determined using reduced CO-difference spectra at
4
50 nm and P420 was not used to calculate enzyme concentration.
Michaelis-Menten Analysis. Purified enzymes were analyzed with regards to
steady-state enzyme kinetics with two typical CYP2D6 substrates – bufuralol and
dextromethorphan. Both substrates are widely used in the characterization of CYP2D6
activity and both possess aromatic rings and a basic nitrogen that are characteristic of
CYP2D6 substrates. However, dextromethorphan is more rigid and has fewer degrees of
rotation than bufuralol.
All CYP2D6 enzymes were active in metabolism of bufuralol and
dextromethorphan with the greatest metabolic efficiency observed with *53 (Fig. 2, Table
2
). The overall efficiency, as calculated by vmax/K
magnitude greater than *1 when bufuralol was the substrate and only ~1.5x higher than
1 with dextromethorphan as substrate, though this later value is an underestimate due to
the presence of additional and quantitatively significant metabolites of dextromethorphan
m
, of *53 was nearly an order of
*
ꢀ
16

DMD Fast Forward. Published on May 21, 2018 as DOI: 10.1124/dmd.117.079871
This article has not been copyedited and formatted. The final version may differ from this version.
DMD #797871
formed by *53 samples (vide infra). Likewise, the Thr309Ala mutant had similar
efficiency as *1 with bufuralol as substrate (89%), but the efficiency was reduced when
dextromethorphan was the substrate (54%). Again, this was due to the presence of
additional products formed by Thr309Ala that were not formed by *1 (or by *53). The
*
34, *17-2, and *17-3 enzymes had lower overall activity compared to *1 regardless of
substrate (Fig. 2, Table 2).
To gain additional understanding of K
53 and the Thr309Ala mutant that produced multiple products in addition to
m
with dextromethorphan as substrate for
*
dextrorphan, substrate depletion analysis via the method of Obach and Reed-Hagen were
completed with *53, the Thr309Ala mutant, and, for comparison, *1 (Obach and Reed-
Hagen, 2002) (Supplemental Fig. 2). The value for K
4 µM - lower, but comparable, to the value determined by Michaelis-Menten analysis
120 ± 30 µM). The K for *53 as measured using substrate depletion was 270 ± 76 µM
and similar to the value reported in Table 5 (300 ± 80 µM). With the Thr309Ala mutant
the K measured by substrate depletion was 21 ± 7 µM. The substrate depletion value of
was much lower in this case than the value determined by Michaelis-Menten because
m
with *1 using this method was 30
±
(
m
m
K
m
the Thr309Ala mutant favored production of N-oxide and N-demethylation products and
produced less dextrophan than *1 and *53 (vide infra). Finally, using substrate depletion
m
analysis with dextromethorphan revealed that the K values for *1 and the Thr309Ala
mutant metabolism of dextromethorphan were similar (30 and 21 µM, respectively), as
they were with bufuralol metabolism where both enzymes produced the same major
product. The substrate depletion method does not allow for calculation of vmax
.
ꢀ
17

DMD Fast Forward. Published on May 21, 2018 as DOI: 10.1124/dmd.117.079871
This article has not been copyedited and formatted. The final version may differ from this version.
DMD #797871
Metabolism of Bufuralol and Dextromethorphan. Metabolites of bufuralol and
2
dextromethorphan, identified based on MS fragmentation patterns (not shown), were
examined with each variant. In reactions containing bufuralol (262 m/z) as substrate,
product distributions among 1’-, 4-, and 6-hydroxybufuralol eluting at 25, 28, and 35
min, respectively, were similar among variants and additional, novel mono-oxygenated
metabolites of 278 m/z were not observed (Fig. 3A; note that chromatograms are on the
1
’,2’
same scale with normalized level indicated on the figure). Dehydrogenated (Δ
)
bufuralol eluted later than the hydroxylated products at 41 min (labeled f) and was
formed by all variants as well, but at an order of magnitude lower level than the mono-
oxygenated products (Fig. 3B). The peak at 25 minutes (labeled d) in the 260 m/z XIC
was a contaminant that was present in the standard as well as in all in the reactions with
NADPH and CuOOH (Fig. 3B and Supplemental Fig. 3). Also, an additional metabolite
with 260 m/z formed was formed by the Thr309Ala mutant and eluted at 28 min (labeled
e, Fig. 3B). The identity of the metabolite is unknown, but the CID fragmentation is
consistent with dehydrogenation at possibly the amine (Supplemental Fig. 3).
With the more rigid substrate dextromethorphan (272 m/z), all variants produced
the expected dextrorphan metabolite (labeled a at 6 min with 258 m/z) (Fig. 4A).
However, a more abundant 258 m/z peak in reactions with Thr309Ala appeared at a later
retention time (22 min, labeled b) and had a mass spectrum and retention time consistent
with formation of 3-methoxymorphinan (N-demethylation; Fig. 4A). Some 3-
methoxymorphinan (labeled b) was formed by all the variants, but it was greatest with the
Thr309Ala mutant. All of the variants also produced some low levels of 3-
ꢀ
18

DMD Fast Forward. Published on May 21, 2018 as DOI: 10.1124/dmd.117.079871
This article has not been copyedited and formatted. The final version may differ from this version.
DMD #797871
hydroxymorphinan (both O- and N-demethylation, labeled c), though *17-2 and *17-3
produced levels at the limit of detection (Fig. 4B).
Mono-oxgenated dextromethorphan products (288 m/z) were detected in assays
with the variants with the most notable peaks in reactions with *53 and the Thr309Ala
mutant (Fig. 4C). With *53, there were two mono-oxygenation peaks eluting at ~5.5 and
~
7.5 min (labeled d and e). One of the mono-oxygenation peaks is likely a 7-
hydroxylation product that was observed by Flanagan et al. with a Phe120Val mutant
Flanagan et al., 2004). The other mono-oxygenation peak showed the same
(
n
2
fragmentation pattern upon MS analysis and, based on MS , might represent a
hydroxylation at the neighboring carbon-6 position (Supplemental Fig. 4). In addition, all
variants, but particularly the Thr309Ala mutant, produced a 288 m/z product with a much
later retention time (~33 min, labeled h) (Fig. 4C). The CID for the later peak showed a
2
72 m/z fragment consistent with loss of an oxygen. Furthermore, upon treatment with
TiCl , the peak was no longer detectable, supporting the designation of the metabolite as
3
an N-oxide (data not shown). The levels of the mono-oxygenation peaks formed by *53
and the Thr309Ala mutant are similar to the levels of dextrorphan and explain why the
Michaelis-Menten analysis with dextromethorphan as substrate produces lower enzyme
efficiency than with bufuralol since only the dextrorphan metabolite was quantitated in
the Michaelis-Menten analysis. Peaks labeled f and g were not identified; peak g was
only seen in reactions with the Thr309Ala mutant (Fig. 4C).
When CuOOH was used to support metabolism of bufuralol, only the 1’-
hydroxybufuralol peak was detected among the hydroxlated bufuralol products
1
’,2’
(
Supplemental Fig. 5). Conversely, dehydrogenated (Δ ) bufuralol (labeled f) was
ꢀ
19

DMD Fast Forward. Published on May 21, 2018 as DOI: 10.1124/dmd.117.079871
This article has not been copyedited and formatted. The final version may differ from this version.
DMD #797871
retained in reactions with CuOOH and, for *1, *34 and *53, higher levels of the
dehydrogenated product were observed than in reactions with NADPH (Supplemental
Fig. 5). In reactions with dextromethorphan and CuOOH as an oxygen surrogate,
dextrorphan was the only observed major metabolite (Supplemental Fig. 6). In addition,
we observed decreased enzyme activity in reactions with CuOOH (data not shown)
consistent with inactivation of CYP2D6 by CuOOH as has been reported with other
CYPs such as 2B1, 3A4, 1A1, etc. (Yao et al., 1993; Yu et al., 1996; He et al., 1998).
Coupling of NADPH Consumption to Substrate Metabolism. Since
differences in Michaelis-Menten kinetics and enzyme efficiencies for the CYP2D6
variants may be at least partially due to differences in the ability of the variants to pass
electrons from the reductase to the active site of 2D6, a set of NADPH-metabolite
coupling experiments with all variants were completed using bufuralol as substrate.
Bufuralol was chosen for the experiments with all variants due to the negligible levels of
competing products formed; that is, 1’- hydroxybufuralol was the major metabolite
formed by all the variants and therefore readily quantitated in relationship to NADPH
consumption (vs. dextromethorphan products were quite different among the variants).
The rate of NADPH consumption with bufuralol as substrate ranged from ~7.5 –
-
1
-1
3
1.6 min and the rate of metabolite formation from ~0.50 – 5.2 min with the overall
percent coupling from ~4 – 18% (Table 3). *53 and *1 demonstrated similar coupling of
17-18%. While the Thr309Ala mutant produced the same amount of product as *1, the
~
rate of NADPH consumption was approximately quadruple that of *1 leading to a much
lower coupling at only ~4.5% − lower than that of *34 and *17-3, variants that displayed
ꢀ
20

DMD Fast Forward. Published on May 21, 2018 as DOI: 10.1124/dmd.117.079871
This article has not been copyedited and formatted. The final version may differ from this version.
DMD #797871
much lower enzyme activity than the Thr309Ala mutant in Michaelis-Menten analysis
Fig. 2).
Since Michaelis-Menten analysis showed a greater vmax for *34, *17-2, and *17-3
(
when dextromethorphan was used as the substrate versus bufuralol, NADPH coupling
experiments were also completed for these variants with dextromethorphan as substrate.
NADPH coupling was similar for dextromethorphan metabolism as it was for bufuralol
metabolism by these variants (Table 3). There was no difference in NADPH coupling in
reactions with dextromethorphan at 100 µM vs. 600 µM (data not shown).
Spectral Binding with SCH 66712. SCH 66712 displayed Type I substrate
binding upon titration with all of the purified CYP2D6 enzymes in this study
(
Supplemental Fig. 7). Fit of the binding data for *1 with the quadratic equation yielded a
of 35 ± 8 µM (Supplemental Fig. 7). Fits for all CYP2D6 enzymes in this study
yielded similar K values between 15-60 µM (Table 4).
K
s
s
Inactivation of CYP2D6 enzymes by SCH 66712. We have previously shown
potent inactivation by SCH 66712 of CYP2D6 in Supersomes™ (Nagy et al., 2011). In
the present study we examined the susceptibility to time- and concentration-dependent
inactivation of purified, recombinant CYP2D6 variants by SCH 66712 using Kitz-Wilson
analysis with either bufuralol or dextromethorphan as reporter substrates (Fig. 5, Table 5,
Supplemental Fig. 8).
With bufuralol as reporter substrate, purified, recombinant *1 showed a kinact of
-
1
0
.14 ± 0.01 min and K
I
of 10 ± 3 µM. For the allelic variant *34, the inactivation
-
1
kinetics were similar to *1 with a kinact of 0.096 ± 0.004 min and K
I
of 6.6 ± 0.9 µM
(
Fig. 5, Table 5). Results from inactivation assays with dextromethorphan as reporter
ꢀ
21

DMD Fast Forward. Published on May 21, 2018 as DOI: 10.1124/dmd.117.079871
This article has not been copyedited and formatted. The final version may differ from this version.
DMD #797871
substrate were comparable (Table 5, Supplemental Fig. 8). The activity of allelic variants
*
17-2 and *17-3 was too low for reliable measurement of inactivation kinetics.
The allelic variant *53 showed much greater enzyme activity and product
formation than *1 in Michaelis-Menten analysis with both bufuralol and
dextromethorphan. However, in Kitz-Wilson analysis *53 was less susceptible to
-
1
inactivation. In assays with bufuralol as substrate, kinact was 0.074 ± 0.003 min and K
I
was 5.7 ± 0.9 µM (Fig. 5, Table 5). With dextromethorphan as substrate, the kinact was
-
1
0
.05 ± 0.02 min and K
to low levels of inactivation (Table 5).
values from Kitz-Wilson analysis of the Thr309Ala mutant were
.042 ± 0.004 min and 1.0 ± 0.4 µM, respectively, with bufuralol as substrate and 0.034
I I
was ~3 µM. Accurate measurement of K was more difficult due
The kinact and K
I
-
1
0
±
-
1
0.002 min and 1.7 ± 0.5 µM, respectively, with dextromethorphan as substrate (Table
). While the rates of inactivation, kinact, for the Thr309Ala mutant were lower than for
1 and *53, the lower K values resulted in little change in overall inactivation (kinact / K ).
5
*
I
I
Inclusion of the nucleophilic trapping agents NAC or GSH in inactivation assays did not
prevent or reduce the amount of inactivation observed in reporter reactions (data not
shown). Furthermore, no NAC or GSH adducts of SCH 66712 or SCH 66712 metabolites
were observed in mass spectral analysis (data not shown).
Partition Ratio. Partition ratio is a measure of the number of turnovers on
average an enzyme completes prior to an inactivation reaction that halts the enzyme’s
activity. (Silverman, 1988). The lower the partition ratio, the fewer turnovers completed
before inactivation. Previously we showed the partition ratio for *1 in Supersomes™ to
be ~3 (Nagy et al., 2011). The partition ratio measured for purified, recombinant *1 in the
ꢀ
22

DMD Fast Forward. Published on May 21, 2018 as DOI: 10.1124/dmd.117.079871
This article has not been copyedited and formatted. The final version may differ from this version.
DMD #797871
current study was ~3 while the partition ratio for *53 was ~24 as estimated from the
intercept of the linear regression line obtained from the lower ratios of SCH 66712 to
CYP2D6 and the line obtained from higher ratios of SCH 66712 to CYP2D6 for each
enzyme inactivation assay (Fig. 6). With the Thr309Ala mutant, a partition ratio could
not be determined due to the low rate of inactivation (no more than 20-40% loss of
activity in reactions with up to 150 µM SCH 66712 and in incubations for 60 min) (data
not shown).
Metabolites of SCH 66712. We previously reported the metabolites of SCH
6
6712 formed by CYPs 2D6, 2C9, 2C19, and 3A4 (Nagy et al., 2011). In similar
metabolism experiments with CYP2D6 allelic variants in this study, mono-oxygenation
of SCH 66712 at two different positions was apparent by the presence of four distinct m/z
3
55 ions in the mass spectral analysis (M+1) as seen previously with other CYPs
(Supplemental Fig. 9). The *53 variant and the Thr309Ala mutant produced more of the
first eluting peak that corresponds to the SCH 66712 metabolite with mono-oxygenation
on the phenyl ring [Supplemental Fig. 9, peak a, and (Nagy et al., 2011; Bolles et al.,
2
014)]. No other metabolites such as dehydrogenation or dealkylation were observed.
When CuOOH was used as oxygen surrogate, only the *53 variant and the Thr309Ala
mutant retained formation of the first mono-oxygenation product (mono-oxygenation on
the phenyl ring of SCH 66712, labeled a) (Supplemental Fig. 9).
Digestion and Mass Spectral Analysis of Adducted CYP2D6 *1. In whole
protein mass analysis of *53 and the Thr309Ala mutant after incubation with SCH 66712
and NADPH, evidence for adducted whole protein was not observed (data not shown).
We have previously observed in mass spectral analysis whole protein adduction of
ꢀ
23

DMD Fast Forward. Published on May 21, 2018 as DOI: 10.1124/dmd.117.079871
This article has not been copyedited and formatted. The final version may differ from this version.
DMD #797871
CYP3A4 and CYP2D6 by SCH 66712 (Bolles et al., 2014) and we have postulated that
Thr309 of CYP2D6 might be the site of inactivation by SCH 66712.
To investigate the role of Thr309 in inactivation of *1 by SCH 66712, inactivated
*
1 was purified by gel, digested with chymotrypsin and/or trypsin, and resulting peptides
analyzed by mass spectrometry (n=5). Protein sequence coverage in all analyses varied
from 50-90%. While there was some supporting evidence for adduction in the peptide
3
04
316
fragment
SAGMVTTSTTLAW
that contains Thr309, the results were not
conclusive (data not shown).
ꢀ
24

DMD Fast Forward. Published on May 21, 2018 as DOI: 10.1124/dmd.117.079871
This article has not been copyedited and formatted. The final version may differ from this version.
DMD #797871
Discussion
The activity of CYP2D6 (*34, *17-2, *17-3, and *53) as well as reference
CYP2D6 (*1) and an active site mutant of *1 (Thr309Ala) were characterized by
Michaelis-Menten analysis (Fig. 2, Table 2). As in previous in vitro studies (Oscarson et
al., 1997; Marcucci et al., 2002; Yu et al., 2002; Shen et al., 2007; Sridar et al., 2011;
Kim et al., 2013), the *34, *17-2, and *17-3 all displayed decreased enzyme efficiency in
m
comparison with *1. The K values determined by Michaelis-Menten analysis were much
higher for dextromethorphan than for bufuralol with all the enzymes; it has been
suggested that the rigidity of dextromethorphan contributes to this effect when Phe120 is
mutated (Flanagan et al., 2004). Indeed, in substrate depletion analysis the K
m
with
dextromethorphan as substrate remains high for *53 that lacks Phe120, but is lower for
*
1 and the Thr309Ala mutant metabolism of dextromethorphan.
The *53 variant had a higher enzyme efficiency compared to *1 with both
dextromethorphan (154% of *1) and bufuralol (778% of *1) as substrate. Other groups
have observed similar kinetic effects with *53 (Sakuyama et al., 2008; Muroi et al., 2014)
and with Phe120Ala mutants (Flanagan et al., 2004; Keizers et al., 2004). The percentage
enzyme efficiency of dextromethorphan metabolism by *53 was much lower than that of
bufuralol partially due to the fact that the analysis only accounted for dextrorphan
product formation though *53 produces additional hydroxylated metabolites (Fig. 4).
Estimation of K
methods yielded a K
higher than K for *1 and the Thr309Ala mutant (Table 2 and Supplemental Fig. 2).
These findings demonstrate the importance of Phe120 for determining substrate
m
for *53 metabolism of dextromethorphan by substrate depletion
m
value similar to the one determined by traditional methods and
m
ꢀ
25

DMD Fast Forward. Published on May 21, 2018 as DOI: 10.1124/dmd.117.079871
This article has not been copyedited and formatted. The final version may differ from this version.
DMD #797871
orientation and product regioselectivity. In docking experiments with dextromethorphan
and *1 and *53 models, Phe120 provides pi-pi stacking interactions with the aromatic
ring of dextromethorphan in an orientation consistent with O-demethylation in *1.
However, that favorable interaction is missing in *53 (Supplemental Fig. 10).
Though *53 produced much greater amounts of product over time than *1, *53
had approximately the same coupling of NADPH consumption to product formation as
*
2
1 (Tables 2, 3). The amino acid changes in *53 are along substrate access/egress channel
c [nomenclature of (Cojocaru et al., 2007)]. The smaller size of isoleucine compared to
phenylalanine at position 120 and the greater polarity of serine as compared to alanine at
position 122 may facilitate metabolite egress and contribute to more rapid turnovers of
the *53 enzyme. Previous molecular dynamics simulations by our group showed that *53
had more open egress and access channels in time evolution studies; it may be that faster
egress and access supports higher metabolic rates and greater degrees of freedom for
substrate interaction with the heme (de Waal et al., 2014).
The Thr309Ala active site mutant displayed similar kinetic values as *1 with
bufuralol as a substrate. With dextromethorphan as substrate, greater differences in K
and vmax were noted using traditional Michaelis-Menten analysis in part due to additional
metabolites formed by the mutant. By substrate depletion methods, the K values for
metabolism of dextromethorphan by *1 and the mutant were similar (Supplemental Fig.
). NADPH coupling studies revealed that while the Thr309Ala mutant had similar rates
of product formation as *1, it consumed ~4 times as much NADPH in the process (Table
). Thr309, or the conserved equivalent in other CYPs, has been shown to be involved in
dioxygen activation to form the oxyenoid species (Compound I) by proton relay (Keizers
m
m
2
3
ꢀ
26

DMD Fast Forward. Published on May 21, 2018 as DOI: 10.1124/dmd.117.079871
This article has not been copyedited and formatted. The final version may differ from this version.
DMD #797871
et al., 2004; Keizers et al., 2005b; Zhang et al., 2011). Our results support previous
findings that Thr309 is important for product regio-selectivity and oxygenation
mechanisms.
All three lower activity variants (*34, *17-2, and *17-3) have the Arg296Cys
amino acid change. Arg296 is located on the N-terminal end of helix I and points outward
to the surface of CYP2D6 along channel 2c. Arg296 is mutated to cysteine in more than
3
0 allelic variants of CYP2D6 usually in combination with other mutations. To our
knowledge, there are no identified roles of Arg296 in the mechanism of CYP2D6. It is
possible that Arg296 may be involved with binding reductase and coupling product
formation to NADPH consumption. *34, *17-2, and *17-3 all displayed lower NADPH
coupling than *1 that might explain, in part, the decreased activity of theses variants -
especially for *34. *17-2 and *17-3, unlike *34, both displayed increased P420 and lack
of stability during purification that likely contributed more significantly to their reduced
activity.
Given the differences in metabolite formation and enzyme kinetics with bufuralol
and dextromethorphan, we analyzed the interaction of the variants with the known potent
inactivator, SCH 66712 (Nagy et al., 2011). Results of spectral binding suggest that SCH
6
6712’s ability to act as a substrate was not impacted by any particular amino acid
changes in the allelic variants (Table 4). We also observed that all the CYP2D6 variants
in this study produced the same monoxygenated metabolites of SCH 66712 as shown
previously with *1 and with CYP3A4 (Nagy et al., 2011; Bolles et al., 2014).
Due to the increased activity of *53, we initially thought that the *53 variant
would be more susceptible to mechanism-based inactivation by SCH 66712; however,
ꢀ
27

DMD Fast Forward. Published on May 21, 2018 as DOI: 10.1124/dmd.117.079871
This article has not been copyedited and formatted. The final version may differ from this version.
DMD #797871
this was not the case (Figs. 5 and 6). It is possible that *53 could be adducted in reactions
with SCH 66712, but that the replacement of Phe120 by Ile allows substrates to still
access and egress from the active site even with SCH 66712 bound; however, no protein
adduct was observed in whole protein mass spectrometry with *53. Alternatively, the
reactive metabolite may be able to move out of the active site without adducting the
enzyme. Since the Phe120 has been shown to control orientation of aromatic rings in
substrates (Keizers et al., 2004), it is also possible that in *53 the SCH 66712 can no
longer readily take the orientation that allows for the production or reaction of the
reactive electrophile.
Inactivation constants for *34 were similar to *1 (Table 5) and the values for
inactivation constants were similar to those reported previously for CYP2D6 in HLM and
Supersomes™ with dextromethorphan as reporter substrate (Palamanda et al., 2001).
Both *53 and the Thr309Ala mutant displayed reduced rates of inactivation though the
Thr309Ala mutant had overall similar inactivation (kinact/K
Supplemental Fig. 8). The observed inactivation with the Thr309Ala mutant may be the
result of excessive production of H by the mutant due to lowered coupling of NADPH
I
) as *1 (Fig. 5, Table 5,
2
O
2
consumption to product formation. Similar effects have been reported in the literature for
analogous threonine mutants, e.g. Thr252 of P450cam and others (Martinis et al., 1989;
Yeom et al., 1995). Also, CYP2D6 (and other P450s) have been shown to be sensitive to
2 2
H O mediated oxidation and inactivation (Albertolle et al., 2018; Vaz et al., 1996). Our
group has also observed that CYP2D6 is sensitive to inactivation by CuOOH. In fact, we
had hoped to see if CuOOH could modulate SCH 66712 inactivation and possible
formation of reactive electrophiles, but inactivation by CuOOH in the absence of SCH
ꢀ
28

DMD Fast Forward. Published on May 21, 2018 as DOI: 10.1124/dmd.117.079871
This article has not been copyedited and formatted. The final version may differ from this version.
DMD #797871
6712 was too great to allow for meaningful analysis (unpublished observations). To
6
address inactivation further, both whole protein mass spectrometry and proteomic
analysis with the Thr309Ala mutant and the *53 variant were attempted. However,
interpretation was complicated because after the reaction of enzyme with NADPH,
ionization of CYP2D6 decreased and spectral noise increased, consistent with protein
adduction (data not shown).
We previously suggested that the reactive electrophile for inactivation of
CYP2D6 by SCH 66712 could be a quinone resulting from mono-oxygenation of SCH
6
6712 (+354 amu) (Nagy et al., 2011). A quinone would be a soft electrophile, while a
secondary alcohol such as on Thr309 would be a hard nucleophile, making this
n
combination less likely. However, we have also observed in our MS data of metabolites
the formation of a carbinolamine (hydroxylation on one of the carbons of the piperazine
ring, peak d in Supplemental Fig. 9) that could allow for ring opening to an aldehyde that
could serve as a reactive electrophile, or reduce to an alcohol (with the ring open). The
aldehyde would be a hard electrophile and therefore might be more likely to interact with
the secondary alcohol of Thr309.
Findings in the present study further support the idea that substrate interactions
with the enzyme, not just the ability of the enzyme to make/break chemical bonds,
influences metabolite formation and inactivation as proposed by others (Vandenbrink et
al., 2012) and that allelic variants can alter access channels and active site architecture
(Maekawa et al., 2017). In conclusion, the examination of naturally occurring allelic
variants of CYP2D6 allows for analysis of structure/function of CYP2D6 and
understanding of mechanism and drug metabolism profiles.
ꢀ
29

DMD Fast Forward. Published on May 21, 2018 as DOI: 10.1124/dmd.117.079871
This article has not been copyedited and formatted. The final version may differ from this version.
DMD #797871
Acknowledgement. We thank Drs. F. P. Guengerich, P. Hollenberg, H. Zhang, I.
Denisov, S. Sligar, and A. Venter for helpful comments and Mr. Greg Cavey, Dr. Samuel
Mackintosh, and Dr. Alan Tackett for proteomic analysis of CYP2D6. We are indebted to
Dr. Mark Hail of Novatia for assistance with ProMass software.
ꢀ
30