The use of 18O-exchange and base-catalyzed n-dealkylation with liquid chromatography/tandem mass spectrometry to identify carbinolamide metabolites

Article abstract of DOI:10.1002/rcm.4595

Oxidation of N-alkyl-substituted amides is a common transformation observed in metabolism studies of drugs and other chemicals. Metabolism at the alpha carbon atom can produce stable carbinolamide compounds, which may be abundant enough to require complet

Full text of DOI:10.1002/rcm.4595

RAPID COMMUNICATIONS IN MASS SPECTROMETRY
Rapid Commun. Mass Spectrom. 2010; 24: 2151–2161
Published online in Wiley InterScience (www.interscience.wiley.com) DOI: 10.1002/rcm.4595
18
The use of O-exchange and base-catalyzed
N-dealkylation with liquid chromatography/tandem
mass spectrometry to identify carbinolamide metabolites
*
Andrew J. Bessire , Alfin D. N. Vaz, Gregory S. Walker, Wei Wei Wang
and Raman Sharma
Department of Pharmacokinetics, Dynamics, and Metabolism, Pfizer Global R&D, Groton, CT 06340, USA
Received 27 January 2010; Revised 21 April 2010; Accepted 22 April 2010
Oxidation of N-alkyl-substituted amides is a common transformation observed in metabolism
studies of drugs and other chemicals. Metabolism at the alpha carbon atom can produce stable
carbinolamide compounds, which may be abundant enough to require complete confidence in
structural assignments. In a drug discovery setting, rapid structural elucidation of test compounds is
critical to inform the compound selection process. Traditional approaches to the analysis of
carbinolamides have relied upon the time-consuming synthesis of authentic standards or purification
of large enough quantities for characterization by nuclear magnetic resonance (NMR). We describe a
simple technique used in conjunction with liquid chromatography/tandem mass spectrometry (LC/
MS/MS) which demonstrates the chemical identity of a carbinolamide by its distinctive ability to
reversibly exchange [18O]water through an imine intermediate. A key advantage of the technique is
that the chromatographic retention times of metabolites are preserved, allowing direct comparisons
of mass chromatograms from non-treated and [18O]water-treated samples. Metabolites susceptible to
the treatment are clearly indicated by the addition of 2 mass units to their original mass. An
additional test which can be used in conjunction with 18O-exchange is base-catalyzed N-dealkylation
of N-(a-hydroxy)alkyl compounds. The use of the technique is described for carbinolamide metab-
olites of dirlotapide, loperamide, and a proprietary compound. Copyright # 2010 John Wiley & Sons,
Ltd.
The N-dealkylation of amines is a commonly encountered
metabolic transformation of drugs and xenobiotics.1 The
reaction is effected by multiple forms of cytochrome P450
enzymes2 and involves the hydroxylation of the alkyl carbon
atom alpha to the nitrogen atom to generate an unstable
carbinolamine intermediate. Dissociation of this intermedi-
ate yields the N-dealkylated amine and an aldehyde or
ketone from the alkyl portion of the molecule.3,4 Alkyl sub-
stituents on amide nitrogens can also undergo CYP-
catalyzed hydroxylations adjacent to the amide nitrogen
atom. For example, the enzyme peptidylglycine a-amidating
mono-oxygenase catalyzes the formation of an a-hydro-
xyglycine intermediate as a first step in peptide amidation.5
In contrast to carbinolamines, the carbinolamides thus
formed are often stable and can be isolated.6–15 The stability
of such structures has been attributed to intramolecular
hydrogen bonding between the carbonyl oxygen and the
hydroxyl hydrogen atoms, possibly with the participation of
a solvent water molecule.16 Due to the isobaric relationship
between carbinolamides and metabolites hydroxylated at
other positions, the use of mass spectrometry to distinguish
between the two species requires diagnostic ions to be
present in the tandem mass (MS/MS) spectrum. Alterna-
tively, structural characterization of a carbinolamide metab-
olite relies on matching the chromatographic retention time
and mass spectrum to an authentic synthetic standard, or on
nuclear magnetic resonance (NMR) analysis of the purified
metabolite. Either of these approaches is resource- and time-
intensive.
In the course of our studies on the metabolic profiles of
dirlotapide (CP-742,033; Slentrol) in dogs17 and rats, we
identified a major metabolite that was determined by liquid
chromatography/tandem mass spectrometry (LC/MS/MS)
to be a debenzylated product with additional oxidation on
either the amide nitrogen atom or N-methyl carbon atom.
Due to the extreme lability of other bonds in the molecule,
the characteristic loss of formaldehyde, which would be
expected from an N-hydroxymethyl metabolite, could not be
observed in the mass spectrum. Consequently, the structure
could not be determined unambiguously from its mass
spectral fragmentation pattern. The metabolite was then
purified from dog feces and its structure was established by
proton NMR to be the stable carbinolamide metabolite.
As a consequence of the intense effort required to
determine the structure of the dirlotapide metabolite by
traditional approaches, we have used this metabolite to
*Correspondence to: A. J. Bessire, Pharmacokinetics, Dynamics,
and Metabolism, Pfizer Global Research and Development,
Groton, CT 06340, USA.
E-mail: andrew.j.bessire@pfizer.com
Copyright # 2010 John Wiley & Sons, Ltd.

2152 A. J. Bessire et al.
18OH
CH2
O
C
16OH
CH2
O
C
O
C
H+
[
18O]Water
CH2
H2O+
R
N
R
N
R
N
H
H
Imine
intermediate
Scheme 1.
develop a method using two simple chemical reactions in
conjunction with LC/MS/MS that enables an unambiguous
(specific activity 5.65 mCi/mmol, radiochemical purity
>96% by HPLC).
distinction between
a stable carbinolamide and other
possible hydroxylated species. As shown in Scheme 1 we
took advantage of the chemistry of the proposed metabolite,
reasoning that an N-a-hydroxyalkyl structure should
reversibly dehydrate under acidic conditions, thus allowing
the incorporation of an atom of 18O when the reaction is
conducted in 18O-enriched water. This species could then be
observed as a peak in the mass spectral chromatogram at the
same retention time as the original metabolite, but now with
a mass addition of 2 u. We further reasoned that, under basic
conditions, deprotonation of the hydroxymethyl group
should promote the irreversible loss of the alkyl group to
produce the N-dealkylated metabolite.
In vitro metabolism studies
Loperamide and compounds C and D were incubated with
rat and monkey liver microsomes at substrate concentrations
of 20 mM and microsomal protein concentrations of 1 mg/
mL. Compound A was incubated with human liver micro-
somes and human cryopreserved hepatocytes at substrate
concentrations of 10 mM. Hepatocyte incubations with com-
pound A were conducted in Williams’ E media containing
750 000 viable cells/mL and were maintained under an
atmosphere of 5:95 CO2/O2. Compound B was incubated
with rat and human liver microsomes at 10 mM substrate
concentrations and microsomal protein concentrations of
2 mg/mL.
The method has been used to identify a proposed carbi-
nolamide metabolite of loperamide, and to characterize
metabolites from two proprietary compounds in discovery
programs at Pfizer. We suggest the use of this technique as a
rapid, unequivocal means to identify hydroxylations on alkyl
groups alpha to the nitrogen atom of amides, which avoids
the time-consuming purification of a metabolite as would be
required for structural elucidation by NMR. An advantage
of using [18O]water for the derivatization procedure is that
complex samples, such as crude plasma, urine, or fecal ex-
tracts, can be analyzed by comparison with control samples
without the complication of shifted HPLC retention times.
Microsomal systems contained 10 mM MgCl2 and 100 mM
potassium phosphate buffer at pH 7.4 in final volumes of 2 to
4 mL, and were initiated by the addition of NADPH (1.3 mM
final concentration). All reaction mixtures were maintained
at 378C and were gently shaken for the duration of the
incubation period. At the end of the reaction timecourse
incubations were terminated by the addition of four volumes
of cold acetonitrile with thorough mixing. Following centri-
fugation for 5 min at 1100 g, supernatants were evaporated in
a vacuum centrifuge, and residues were reconstituted in
HPLC mobile phase for analysis.
EXPERIMENTAL
HPLC, mass spectrometry, and radioactivity
monitoring
Chemicals
[14C]Dirlotapide was prepared by Perkin Elmer Life Sciences
(Boston, MA, USA). The material was diluted with non-
radiolabeled dirlotapide to an appropriate specific activity
(see below) by the Pfizer Radiochemical Synthesis Group
(Groton, CT, USA). Unlabeled dirlotapide, N-desbenzyl-N-
desmethyl dirlotapide (M4), and compound A were syn-
thesized by Pfizer (Groton, CT, USA). Loperamide was
purchased from Sigma-Aldrich (St. Louis, MO, USA).
Acetonitrile, sodium hydroxide, and concentrated sulfuric
acid were purchased from J.T. Baker (Phillipsburg, NJ, USA).
Water enriched in 18O (97 atom %) was purchased from
Isotech (Sigma-Aldrich). All other solvents and reagents
were of the highest grade commercially available, and were
used without further purification.
For [14C]dirlotapide analyses, feces or plasma samples were
extracted with 5 mL of acetonitrile by mixing thoroughly for
10 min, followed by centrifugation for 5 min at 1100 g. This
procedure was conducted three times, then for each sample
the supernatants were combined and the solvent was evap-
orated under vacuum in a Genevac vacuum centrifuge.
Samples were analyzed by LC/MS/MS with offline radio-
activity profiling using an LC-ARC stopped-flow system
(AIM Research, Hockessin, DE, USA). The HPLC column
effluent was mixed with Stop-Flow AQ scintillation counting
fluid, and radiolabeled sample components were measured
in 12 s fractions with a counting time of 30 s. For both LC/
MS/MS and LC-ARC analyses, the HPLC systems consisted
of an Agilent 1100 pump, membrane degasser, and auto-
sampler. Chromatography was performed using a 150 mm ꢀ
4.6 mm Zorbax Stable-bond CN column packed with 5
micron particles. The mobile phase consisted of 5 mM
ammonium acetate, pH 6.8 (solvent A), and acetonitrile
(solvent B) at a flow rate of 0.7 mL/min. The mobile phase
composition was 20% B from 0–3 min, then increased to 40%
B by 43 min, then to 90% B by 68 min, held isocratically at 90%
B until 73 min, then decreased to 20% B by 74 min, after
Animal studies
Feces, bile, urine, and blood for preparation of plasma were
collected from male and female Sprague-Dawley rats admini-
stered approximately 50 mCi of [14C]dirlotapide (specific
activity 6.74 mCi/mmol, radiochemical purity >99% by
HPLC).
Similar samples were collected from male and female bea-
gle dogs given approximately 17 mCi/kg of [14C]dirlotapide
Copyright # 2010 John Wiley & Sons, Ltd.
Rapid Commun. Mass Spectrom. 2010; 24: 2151–2161
DOI: 10.1002/rcm

LC/MS/MS to identify carbinolamide metabolites 2153
which the column was re-equilibrated at 20% B until 79 min.
Dirlotapide eluted at approximately 58 min under these
conditions, and all metabolites eluted earlier.
before the next injection. For compound B, a Synergi Hydro
RP (Phenomenex, Torrance, CA, USA) 4.6 ꢀ 150 mm, 4
micron column was used for separation. The gradient pro-
gram was 5% B from 0–3 min, 60% B by 25 min, 80% B by
35 min, 95% B by 38 min, then held at starting conditions for a
further 10 min.
For in vitro metabolite studies of loperamide and com-
pounds C and D, acetonitrile-quenched incubation mixtures
were centrifuged for 5 min at 1100 g, then supernatants were
evaporated to dryness under vacuum. Residues were recon-
stituted in 20:80 acetonitrile/water prior to analysis by
LC/MS/MS. The HPLC system was as described above, but
the HPLC column was an Xterra RP18 3.0 ꢀ 150 mm column
(Waters, Milford, MA, USA) packed with 5 micron particles.
The mobile phase consisted of ammonium acetate, pH 6.8
(solvent A), and acetonitrile (solvent B). The column was
held at 20% B for 3 min, then ramped to 70% B over the next
25 min, washed at 90% B until 34 min, then was changed back
to starting conditions at 35 min and re-equilibrated for an
additional 5 min. Loperamide eluted at 24.7 min, and all
metabolites eluted earlier.
Metabolites of [14C]dirlotapide, loperamide, and com-
pounds C and D were identified using a TSQ Quantum
triple-quadrupole mass spectrometer (ThermoFisher Scien-
tific, Waltham, MA, USA) operated in positive-ion mode. The
electrospray ionization (ESI) source was operated at 4200 V
with a capillary temperature setting of 2758C. Q1 scans,
precursor-ion scans and product-ion scans were used for
structural elucidation. For dirlotapide and its metabolites,
precursor-ion scans for precursors of m/z 249 and 421
fragment ions at collision energies (CE) of 60 and 40 V,
respectively, were used for metabolite monitoring. Product-
ion scans were collected at collision energies ranging from
20–60 V, and utilized a collision gas pressure ranging from
0.7–1.5 mTorr of argon.
Nuclear magnetic resonance
For NMR studies, dirlotapide metabolite M2 was extracted
with acetonitrile from approximately 200 g of dog feces and
purified using semi-preparative conditions on a model 215
fraction collector (Gilson, Middletown, WI, USA). A Mono-
chrom CN 10 ꢀ 150 mm column (MetaChem, Torrance, CA,
USA) was used with the gradient described above at a
flow rate of 2.5 mL/min. An approximately 4 min window
of column eluent was isolated which contained several
metabolites, including M2. After several injections of the
crude fecal extract, the HPLC fractions were combined and
the solvent was evaporated. The residue was reconstituted in
20:80 acetonitrile/H2O, and a further purification step was
performed on an analytical scale using an Ace 4.6 ꢀ 150 mm
phenyl column (Advanced Chromatography Technologies,
Aberdeen, UK) packed with 3 micron particles. The solvent
system for this separation consisted of water (solvent A)
and acetonitrile (solvent B) at a flow rate of 0.6 mL/min. The
mobile phase gradient began at 40% B and changed to 60% B
by 40 min, then ramped to 90% B at 41 min and was held
there until 51 min, then changed back to starting conditions
at 52 min, with a 5 min re-equilibration period before the next
injection. This purification step allowed for good separation
of metabolite M2 from other metabolites in the fraction.
Manual collection of the HPLC fractions containing M2 was
performed, and these were combined and evaporated as
before. Several cycles of reconstitution in acetonitrile-d3
followed by evaporation under vacuum were performed to
remove residual protonated solvents. Finally, samples were
dissolved in 0.15 mL of acetonitrile-d3 ‘100%’ (Cambridge
Isotope Laboratories, Andover, MA, USA). Based on liquid
scintillation analysis of the purified metabolite, approx-
imately 40 mg of M2 were isolated.
Accurate mass measurements of dirlotapide metabolites
were obtained using a Q-TOF-2 quadrupole-time-of-flight
mass spectrometer (Waters, Milford, MA, USA) equipped
with an ESI source operated in positive-ion mode. An
internal lock mass (Leu enkephalin, þm/z 556.2766) was
used throughout the analysis via a lockspray, allowing the
calibrant to be introduced into the mass spectrometer every
5 s. For loperamide metabolites, elemental formulae were
generated by analysis on an LTQ-Orbitrap XL (ThermoFisher
Scientific, San Jose, CA, USA) at a resolution setting of 30 000
(specified at m/z 400).
All NMR spectra were recorded on a Bruker Avance
600 MHz (Bruker BioSpin Corporation, Billerica, MA, USA)
controlled by XWIN-NMR V3.5 and equipped with a 2.5 mm
BBI probe. 1D spectra were recorded using a sweep width of
12 000 Hz and a total recycle time of 3.6 s. The resulting time-
averaged free induction decays were transformed using an
exponential line broadening of 0.3 Hz to enhance signal to
noise. All spectra were referenced using residual acetonitrile-
d3 (d ¼ 1.94 ppm relative to TMS, d ¼ 0.00). The multiplicity
edited HSQC data were recorded using the standard pulse
sequence provided by Bruker. A 1K ꢀ 128 data matrix was
acquired using a minimum of 128 scans and 16 dummy scans
with a spectral width of 8000 Hz in the f2 dimension. The
data was zero-filled to a size of 1K ꢀ 1K. A relaxation delay of
1.5 s was used between transients.
Metabolite determinations from incubations with com-
pounds A and B were performed with an LTQ Orbitrap XL
mass spectrometer using an ESI source in positive-ion
mode. The HPLC system consisted of an Accela pump,
autosampler, and diode-array detector. Full scan mass
spectra were acquired at a resolution setting of 15 000
(specified at m/z 400) with data-dependent MSn at 30%
(compound A) or 65% (compound B) normalized collision
energy for metabolite fragmentation studies. The mobile
phase consisted of 5 mM ammonium formate adjusted to pH
3 with formic acid (solvent A) and acetonitrile (solvent B)
flowing at 1 mL/min. For compound A, separation was
achieved with a 4.6 ꢀ 150 mm Kromasil C18 column (Varian,
Torrance, CA, USA) packed with 5 micron particles. The
gradient profile was 5% B from 0–3 min, 50% B by 28 min and
held isocratically for 1 min, then back to starting conditions
by 30 min, followed by 5 min of column re-equilibration
Substitution with 18O-enriched water
Carbinolamide metabolites, either as purified HPLC frac-
tions or as crude sample extracts, were dried under vacuum
or a stream of N2. Residues were then redissolved in 15 mL of
Copyright # 2010 John Wiley & Sons, Ltd.
Rapid Commun. Mass Spectrom. 2010; 24: 2151–2161
DOI: 10.1002/rcm

2154 A. J. Bessire et al.
1800
1600
1400
1200
1000
800
600
400
200
0
temperature for 1 h. The sample was neutralized by the
addition of 10 mL of 10 N H2SO4 and centrifuged before
analysis by LC/MS/MS.
M2
Dirlotapide
M21
RESULTS AND DISCUSSION
Dirlotapide and numerous metabolites were identified in
feces, bile, and plasma of Sprague-Dawley rats and beagle
dogs after oral administration of [14C]dirlotapide (Fig. 1). The
proposed carbinolamide metabolite M2 was observed in
significant quantities in all rat and dog matrices examined,
and was the most abundant metabolite present in rat fecal
samples. M21, a ring-hydroxylated analogue of M2, was also
observed in rat and dog feces. Metabolites M4 and M24, the
products of amide N-dealkylation from M2 and M21,
respectively, were observed in plasma and/or feces as well.
A scheme depicting the relevant metabolism of dirlotapide is
shown in Fig. 2.
[14C]Dirlotapide had an HPLC retention time of approxi-
mately 57 min. Full scan MS spectra consisted of an [MþH]þ
ion at m/z 675, with some unavoidable source fragmentation
which produced a less intense ion at m/z 554. The product-ion
mass spectrum of m/z 675 at 20 V CE contained ions at m/z 91,
122, 421, 483, 526, 554, and 585 (Fig. 3(A)). Loss of the benzyl
group produced the m/z 91 and 585 ions, while loss of N-
methyl-N-benzylamine produced the acylium ion at m/z 554,
0
5
10
15 20
25
30
35
40
45 50
55
60
65
Time (min)
Figure 1. Representative HPLC radiochromatogram of an
extracted rat fecal sample following a single oral dose of
14C]dirlotapide.
[
acetonitrile and 30 mL of 18O-enriched water was added.
Concentrated H2SO4 (5 mL) was added to catalyze water
exchange, and after mixing well the tubes were capped and
left at room temperature for 1 h. The 18O reaction mixture
was centrifuged and then analyzed by LC/MS/MS without
any pH modification.
N-Dealkylation by base treatment
Aqueous mixtures (100 mL) containing carbinolamide metab-
olites were alkalinized with the addition of 10 mL of 10 N
NaOH and the solutions were allowed to stand at room
F
F
F
O
N
O
N
N
O
N
Dirlotapide
CH3
[M+H]+ = 675
F
F
F
F
F
F
OH
O
N
N
O
N
N
O
O
N
N
O
O
HN
HN
M21
M2
CH2OH
CH2OH
[M+H]+ = 617
[M+H]+ = 601
In vivo,
or
Rxn with strong base
F
F
F
F
F
F
OH
O
N
N
O
N
N
O
O
N
N
O
O
H2N
H2N
M24
M4
[M+H]+ = 587
[M+H]+ = 571
Figure 2. Schematic diagram of the relevant dirlotapide metabolic pathway. Some
intermediate metabolites have been omitted for clarity.
Copyright # 2010 John Wiley & Sons, Ltd.
Rapid Commun. Mass Spectrom. 2010; 24: 2151–2161
DOI: 10.1002/rcm

LC/MS/MS to identify carbinolamide metabolites 2155
421.1
100
80
(A)
122.0
F
F
F
526
554
O
N
N
O
60
N
O
421
N
40
122
91
20
91.1
675.4
483.2
553.6
0
100
421.1
(B)
F
F
F
80
60
O
N
526
554
N
O
N
O
421
HN
40
20
0
CH2OH
526.2
554.2
483.3
100
200
300
400
500
600
m/z
Figure 3. MS/MS spectra of dirlotapide (A) and dirlotapide metabolite M2 (B).
which upon loss of CO produced the m/z 526 ion. The m/z 483
ion appears to be formed by elimination of HNCO from m/z
526 with recombination of the resulting tropyllium ion
and the fragment containing the biphenyl and indole ring
systems. The m/z 421 fragment is the acylium ion containing
the biphenyl and indole ring systems. At higher collision
energy (CE), a m/z 249 fragment was observed due to the
acylium ion containing the biphenyl ring system, which
yielded m/z 221 and 201 ions after successive neutral losses of
CO and HF.
[14C]Dirlotapide metabolite M2 was observed at a reten-
tion time of 48.5 min, and had an [MþH]þ ion at m/z 601.
The collision-induced dissociation (CID) spectrum of M2 had
ions at m/z 249, 421, 483, 526, and 554 (Fig. 3(B)) in common
with dirlotapide, suggesting debenzylation of dirlotapide
and hydroxylation on either the N atom or N-methyl group.
Accurate mass measurements on M2 yielded m/z 601.2083,
differing from the mass calculated for C33H28N4O4F3 (m/z
601.2063) by 3.4 ppm. These data supported the assignment
of oxidation on either the amide N atom or methyl group.
No distinction could be made between the two by mass
spectrometry. The m/z 554 ion resulting from in-source
fragmentation occurred so readily that the diagnostic loss of
formaldehyde expected from the CID of an N-hydroxy-
methyl structure, which in the case of M2 would produce an
m/z 571 ion, could not be observed. Incubation of an isolated
sample of M2 with 20% TiCl3 for 1 h produced no change in
the spectrum or retention time, suggesting that M2 was not
an N-oxide metabolite.18
sample of M2 from dog or rat fecal extracts. Therefore, we
attempted derivatizations designed to take advantage of the
pH-dependent dehydration/rehydration equilibrium of the
proposed N-hydroxymethyl compound. An HPLC fraction
containing M2 was collected and the solvent was evaporated
under a stream of N2. The residue was reconstituted in
ACN, and [18O]water was added with a small amount of
concentrated sulfuric acid as a catalyst to promote pro-
duction of the imine. After 1 h, an HPLC peak at the retention
time of M2 (Figs. 4(A) and 4(B)) was observed at m/z 603
(2 mass units higher than before 18O exchange), by Q1 scans
and precursor-ion scans for precursors of the m/z 421
fragment ion. The product-ion spectrum of m/z 603 was
identical to that of M2 (m/z 601), suggesting that a single
18O atom was incorporated into the molecule at the position
49.10
[
16O]M2
48.99
100
(A)
(B)
(C)
80
60
40
20
50.71
0
49.01
100
[
18O]M2
80
60
40
20
50.61
50.64
0
100
M4
80
60
40
20
0
50.81
32.68
39.13
30 32
34
36
38
40
42
44
46
48
50
52
54
56
58
60
Time (min)
Due to the abundance of M2 in the circulation of the rat and
dog, it was deemed necessary to better characterize its
structure. NMR spectroscopy could provide an unambigu-
ous determination of the site of oxidation; however, this
would require the laborious purification of a concentrated
Figure 4. Precursor-ion (m/z 421 fragment) chromatograms
of an isolated sample of dirlotapide metabolite M2: (A) untreated
control, (B) after treatment with [18O]water and H2SO4, and (C)
after treatment with 2 N NaOH.
Copyright # 2010 John Wiley & Sons, Ltd.
Rapid Commun. Mass Spectrom. 2010; 24: 2151–2161
DOI: 10.1002/rcm

2156 A. J. Bessire et al.
sample did not incorporate 18O into its structure. In the crude
extract, however, in contrast to the purified sample of M2,
the addition of 5 mL of 2 N NaOH was insufficient to raise
the pH high enough to promote N-dealkylation. A check of
the pH of the sample after this treatment indicated a pH of
approximately 9, which after 1 h produced no change in the
radiochromatogram. However, after the addition of 10 mL of
10 N NaOH to a 100 mL aliquot of the crude extract, the pH
was approximately 13. After 1 h the disappearance of M2 and
M21 peaks in the radiochromatogram was observed, with a
proportionate increase in the size of peaks corresponding to
the N-dealkylated analogs M4 and M24, respectively (Fig. 5).
These results are consistent with observations of reduced N-
dealkylation rates for N-(hydroxymethyl)benzamide com-
pounds at pH values below 10.5.19
1000
800
M4
M2
M24
600
M21
400
200
0
30 32 34 36 38 40 42 44 46 48 50 52 54 56 58 60
Time (min)
Control
Base treated
Figure 5. Radiochromatograms of a crude rat fecal extract
before (Control) and after (Base treated) treatment with 10 N
NaOH. The conversion of M21 into M24 and of M2 into M4 can
be observed.
of the amide N-methyl group, and demonstrating the ability
of M2 to form an imine intermediate. Additionally, we
surmised that in strong base solutions, the deprotonation of
the alcoholic O atom would drive M2 to N-dealkylate by
the loss of formaldehyde. In fact, incubation of an HPLC
fraction containing M2 with 2 N NaOH for 1 h at room
temperature produced N-desbenzyl-N-desmethyl-dirlotap-
ide (M4, Fig. 4(C)), which was verified by comparison of
LC/MS/MS retention times and CID spectra with an
authentic analytical standard of M4.
The isotopic distribution of [MþH]þ and [MþNH4]þ ions
of M2, and [MþH]þ ions of M21, both before and after
treatment with [18O]water and acid (Fig. 6) demonstrated the
increase in abundance of the respective Mþ2 isotopes after
18O-exchange. The mass peaks due to the 16O isotopes which
remained after treatment (m/z 601 and 617, respectively, for
M2 and M21, Figs. 6(B) and 6(D)) were due either to residual
[16O]water in the sample at the time of the exchange
experiment, or to the back-exchange of 16O during liquid
chromatography. The Mþ2 peaks in the untreated control
sample (m/z 603 and 619, respectively, for M2 and M21,
Figs. 6(A) and 6(C)) were due primarily to the 14C label
(specific activity of 6.74 mCi/mmol) at the carbonyl carbon
atom nearest the biphenyl ring system.
A crude rat fecal extract was then dried and mixed
with [18O]water and H2SO4 as described above, and the
incorporation of 18O into M2 and M21, a proposed N-
hydroxymethyl metabolite with hydroxylation on the benzyl
ring adjacent to the indole system, was easily observed.
A previously identified carboxylic acid metabolite in this
To investigate the possibility of re-incorporation of 16O into
metabolites from the aqueous mobile phase component during
601.4
618.4
100
(A)
619.4
620.4
50
602.4
623.3
617.4
604.4
604.3
0
100
620.4
603.3
(B)
621.4
622.4
50
601.3
618.4
618.3
605.3
625.4
0
100
617.3
(C)
(D)
50
619.4
620.4
623.6
0
100
619.3
50
0
620.3
617.3
622.5
625.5
625
610.4
610
614.0
601.4
611.4
600
605
615
620
m/z
Figure 6. Isotopic distribution of dirlotapide metabolites M2 and M21 before and after treatment with
18O]water and acid: (A) [MþH]þ and [MþNH4]þ ions of M2 before treatment; (B) [MþH]þ and [MþNH4]þ
[
ions of M2 after treatment; (C) [MþH]þ ions of M21 before treatment; and (D) [MþH]þ ions of M21 after
treatment.
Copyright # 2010 John Wiley & Sons, Ltd.
Rapid Commun. Mass Spectrom. 2010; 24: 2151–2161
DOI: 10.1002/rcm

LC/MS/MS to identify carbinolamide metabolites 2157
the course of liquid chromatography, especially with com-
monly used acidic mobile phases as low as pH 3, MS spectra
of M2 were compared after 18O-exchange using HPLC
separations at both pH 3 and pH 6.8 conditions. The retention
time (tR ¼ 48.5 min) of dirlotapide metabolite M2 was un-
changed by varying the pH, and no changes in the isotopic
distributions of the M and Mþ2 mass peaks at m/z 601 and
603 (due to 16O and 18O incorporation, respectively) could be
discerned by inspection (Fig. 7). Additionally, by monitoring
and integrating the multiple reaction monitoring (MRM)
transitions for non-radiolabeled M2 of m/z 601 > 421 and
603 > 421, corresponding to [16O]M2 and [18O]M2, respec-
tively, a quantitative measure of the relative amount of
16O incorporation due to back-exchange during chromatog-
raphy was established. The MRM peak area ratios ([18O]M2/
[16O]M2) after use of the 18O-exchange technique described
above were 5.35 and 5.96, respectively, at pH 3.0 and pH 6.8,
reflecting a 10.2% reduction in peak area for [18O]M2 after
chromatography at pH 3. By comparison, the [18O]M2/
[16O]M2 ratio obtained for a control sample not treated with
[18O]water was 0.015 with chromatography performed at pH
6.8. These results suggested that re-equilibration of an 18O-
exchanged product with [16O]water in the mobile phase
during the course of liquid chromatography was minimal
using either pH 3 or pH 6.8 analytical conditions, and did not
preclude the use of the technique even for metabolites with
HPLC retention times approaching 1 h.
the methyl group of the benzylmethylamino moiety in
dirlotapide was absent in the 1H spectrum of the isolate.
Furthermore, the inequivalent methylene pair at d 4.49 and
4.76 which were assigned as the terminal benzyl methylene
moiety of dirlotapide appeared as a broad singlet at d 4.64 in
1
the H spectrum of the M2 isolate (Fig. 8). In addition, the
multiplicity edited HSQC data contained a cross peak from
the d 4.64 resonance with a carbon chemical shift of 63.1 ppm
and a phase that indicated a methylene. All of the NMR data
was consistent with the proposed N-hydroxymethyl struc-
ture of M2.
In an effort to demonstrate the universality of the 18O-
exchange technique, we applied it to a reported carbinola-
mide metabolite of loperamide, and to metabolites of four
proprietary compounds (compounds A–D). Loperamide
contained an N,N-dimethylamide structure, while com-
pound A was a cyclic N-alkylamide compound with multiple
possible oxidation sites on the aliphatic moiety. Compound B
contained an N,N-dimethyl amide moiety, while compounds
C and D contained N-benzylamide moieties.
As described previously,20 the metabolism of loperamide
in rat liver microsomes yielded N-desmethyl loperamide
and N-desmethyl-N-hydroxymethyl loperamide as the
major reaction products (Fig. 9). The MS/MS spectrum of
loperamide consisted of a single prominent fragment ion at
m/z 266 due to loss of the N,N-dimethyl-a,a-diphenylbutyr-
amide moiety, and had minor ions at m/z 432 and 238
(Fig. 10(A)). The m/z 432 ion was due to loss of dimethyl-
amine from loperamide, while the m/z 238 ion was due to
the loss of ethylene from the m/z 266 ion. In contrast
to N-desbenzyl-N-hydroxymethyl dirlotapide (M2), the
identity of the carbinolamide metabolite of N-desmethyl
loperamide (m/z 479) could be convincingly determined by
analysis of CID mass spectra (Fig. 10) by virtue of a
Although we believed these observations to give unam-
biguous evidence for the N-hydroxymethyl structures of M2
and M21, we performed the isolation and purification of a
sufficient amount of M2 for NMR analysis to firmly establish
the validity of the technique. The spectrum of the M2 isolate
contained three fewer downfield resonances when compared
to similar data on dirlotapide. The singlet assigned as
603.4
100
(A)
80
60
40
604.4
605.3
20
602.4
600.6
606.4
607.5
596.2
0
603.3
100
(B)
80
60
604.3
40
601.3
605.4
20
0
602.3
606.3
607.4
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
m/z
Figure 7. Isotopic envelopes of the [MþH]þ ions of M2 at (A) pH 6.8 and (B) pH 3.0, demonstrating that back-
exchange of 16O from the aqueous mobile phase during the course of liquid chromatography did not occur to a
significant extent. The HPLC retention time of M2 under both conditions was approximately 49 min.
Copyright # 2010 John Wiley & Sons, Ltd.
Rapid Commun. Mass Spectrom. 2010; 24: 2151–2161
DOI: 10.1002/rcm

2158 A. J. Bessire et al.
F
F
F
O
N
O
NH
NH
O
HN
HO
8.5
8.0
7.5
7.0
6.5
6.0
5.5
5.0
4
.5
4
.0
3.5
3.0
ppm
F
F
F
O
N
O
NH
NH
O
N
8.5
8.0
7.5
7.0
6.5
6.0
5.5
5.0
4
.5
4
.0
3.5
3.0
ppm
Figure 8. 1H NMR spectra of dirlotapide metabolite M2 (top) and of dirlotapide (bottom) in
ACN-d3.
prominent ion at m/z 449 due to the neutral loss of
[37Cl16O]N-desmethyl-N-hydroxymethyl loperamide; and
[37Cl18O]N-desmethyl-N-hydroxymethyl loperamide follow-
ing 18O-exchange (Figs. 10(B)–10(D)) were indicative of
18O substitution at the amide N-hydroxymethyl position, as
shown by the loss of 32 u due to [18O]formaldehyde in the
sample incubated with [18O]water, and by loss of unlabeled
formaldehyde (30 u) in the untreated sample.
formaldehyde.
To illustrate the 18O-exchange technique with loperamide,
an aliquot of the crude microsomal reaction mixture was
dried and treated with [18O]water and acid as described
above. Analysis of the carbinolamide peak eluting at 18.2 min
had [M]/[Mþ2]/[Mþ4] isotope ratios consistent with
the addition of 18O to a chlorine-containing compound.
The CID spectra of m/z 479, 481, and 483 at tR ¼ 18.2 min,
corresponding to [35Cl16O]N-desmethyl-N-hydroxymethyl
loperamide; [35Cl18O]- (CID spectrum not shown) and
Initially, several attempts to demonstrate the N-deal-
kylation of N-desmethyl-N-hydroxymethyl loperamide
using triethylamine (TEA) as the base were unsuccessful,
as no change in the mass chromatogram of a TEA-treated rat
20.49
100
90
80
70
60
N-desmethyl loperamide
+m/z 463
Loperamide
+m/z 477
22.19
50
N-desmethyl-N-hydroxymethyl
loperamide
40
+m/z 479
30
18.22
20
10
23.11
24.91
35.35
38.28
36.68
4.10
25.90
33.59
0.08
17.84
4.57
13.62
0
0
5
10
15
20
25
30
35
Time (min)
Figure 9. Base peak chromatogram of an incubation mixture of loperamide in rat liver
microsomes after 30 min.
Copyright # 2010 John Wiley & Sons, Ltd.
Rapid Commun. Mass Spectrom. 2010; 24: 2151–2161
DOI: 10.1002/rcm

LC/MS/MS to identify carbinolamide metabolites 2159
266.2
268.2
100
50
(A)
OH
266
Cl
N
Ph
Ph
N
O
477.2
479.2
0
100
OH
238
H2CO
268
Cl
(B)
(C)
(D)
238.1
N
Ph
Ph
H
50
N
NH3
432
CH2OH
O
449
449.2
252.2
268.1
432.2
434.1
211.1
0
100
OH
238
268
37Cl
481.2
483.2
H2CO
N
238.1
Ph
Ph
50
H
N
NH3
CH2OH
OH
252.2
O
451.2
451.2
434
451
269.2
270.2
211.2
211.1
0
100
238
270
37Cl
238.1
H2C18
O
N
Ph
Ph
H
50
0
N
NH3
C
18OH
O
H2
451
434
239.0
250
434.2
272.2
196.0
150
200
300
350
400
450
m/z
Figure 10. MS/MS spectra of (A) loperamide; (B) [35Cl16O]N-desmethyl, N-hydroxymethyl
loperamide from an untreated sample; (C) [37Cl16O]N-desmethyl, N-hydroxymethyl loper-
amide from an untreated sample; and (D) [37Cl18O]N-desmethyl, N-hydroxymethyl loper-
amide from a sample treated with [18O]water.
liver microsomal sample could be observed. In subsequent
careful analysis of the CID fragment ions contained in the
MS/MS spectra of these three metabolites suggested that one
was an N-oxide compound (losing oxygen, not water, during
CID) on the piperidine moiety; and two were hydroxylations
on the aliphatic ring system (due to facile dehydration
during CID).21 The 18O-exchange technique was employed
to distinguish hydroxylations a to the amide N atom from
oxidations at other positions. A sample of compound A
incubated with human hepatocytes was evaporated to
dryness under vacuum, then reconstituted in ACN to which
[18O]water and H2SO4 were added. After approximately 1 h
at room temperature, the sample was analyzed by LC/MSn
without further processing. Of the five separate singly oxidi-
zed metabolites observed, only one (retention time 20.8 min)
was found to have incorporated an atom of 18O after treat-
ment by virtue of the presence of an enhanced M þ 2 isotope
peak at m/z 487. The MS/MS spectrum of this metabolite
contained the m/z 195 ion both before and after treatment
(Figs. 11(B) and 11(C)), in common with the parent com-
pound, indicating that the metabolic modification was not on
the R3 portion of the molecule. Incorporation of 18O was
clearly indicated by dehydration of the m/z 293 ion (2 u
higher than the corresponding ions before treatment), which,
upon loss of H128O, produced an m/z 273 ion, common to both
pre- and post-treatment MS/MS spectra of this metabolite.
Thus, of five isobaric singly oxidized metabolites, the one
containing a hydroxyl modification alpha to the amide N
atom was distinctly identified using the 18O-exchange tech-
nique described here. Due to the cyclic nature of compound
A and its carbinolamide metabolites, treatment with strong
experiments, however, 10 mL of 10 N NaOH was substituted
for TEA in the mixture. After 1 h at room temperature, the
mixture was neutralized with an equimolar amount of H2SO4
prior to analysis by LC/MS/MS. The loss of the m/z 479 peak
(N-desmethyl-N-hydroxymethyl loperamide) was easily
observed by comparison with a control sample. Other
metabolites were unchanged by base treatment, with the
exception of a quaternary pyridinium metabolite resulting
from oxidation of the piperidine ring. The latter metabolite
was distinct chromatographically as well as by virtue of its
mass and fragmentation properties, so its loss due to the base
treatment did not complicate the analysis. The inability of
TEA to drive the N-dealkylation of N-desmethyl-N-hydroxy-
methyl loperamide demonstrated the necessity for high pH
to accomplish the reaction. Compound A is a proprietary
molecule containing a cyclic amide with multiple possible
sites of oxidation by drug-metabolizing enzymes. The MS2
spectrum of compound A was sparse, containing only ions at
m/z 195, 275, and 321 (Fig. 11(A)). The m/z 275 ion contained
the piperidine amide moiety and residue R1, while the m/z
321 ion resulted from cleavage of the amide CꢁN bond,
producing a fragment which contained the piperidine moiety
and residues R2 and R3.
Incubations of compound A with human liver microsomes
or hepatocytes produced five singly oxygenated metabolites,
in addition to several secondary metabolites. Of these, three
primary singly oxidized metabolites (m/z 485, þ 16 u relative
to the parent compound) were shown by MS/MS spectra to
be modified on the piperidine portion of the molecule. A
Copyright # 2010 John Wiley & Sons, Ltd.
Rapid Commun. Mass Spectrom. 2010; 24: 2151–2161
DOI: 10.1002/rcm

2160 A. J. Bessire et al.
SM: 5B
275.1755
100
(A)
195
275
80
60
40
O
R
R
N
2
3
R
1
321
195.0917
20
321.2076
470.2640
0
100
273.1596
291.1703
- H216O
(B)
195
80
60
40
20
273
291
16OH
O
R2 R3
- H216O
N
195.0915
R1
218.1174
337
319
255.1489
273.1597
319.1915
467.2438
0
100
- H218O
(C)
273
293
18OH
195
80
60
40
20
0
293.1745
O
R2 R3
- H218O
N
R1
195.0916
218.1175
319
339
255.1493
250
319.1918
300
467.2436
450
200
350
400
500
m/z
Figure 11. MS/MS spectra of compound A (A), and of a carbinolamide metabolite of compound A before (B)
and after (C) treatment with [18O]water and acid.
base would not be expected to produce N-dealkylation as
was the case with dirlotapide and loperamide carbinolamide
metabolites, and thus these reactions were not performed.
A glucuronide conjugate was also observed at 18.2 min at
m/z 647 in the same hepatocyte incubation with compound A.
Partial in-source fragmentation of this conjugate produced
the aglycone at m/z 471 (þ2 u relative to the parent
compound), and MS/MS spectra of each of these ions were
consistent with demethylation of the R1 residue and oxi-
dation on the piperidine portion of the molecule. After
treatment with [18O]water and H2SO4, Mþ2 peaks were
observed for both the aglycone and glucuronide conjugate at
m/z 473 and 649, respectively, each with an elemental for-
mula consistent with the incorporation of 18O into the
structure. Thus, the glucuronide metabolite corresponded to
a demethylated analogue of the carbinolamide metabolite
just described, with conjugation on the R1 moiety.
Compounds C and D, structural analogs containing an
N-benzyl amide moiety, each produced an abundant metab-
olite in rat and monkey liver microsomes for which mass
spectrometry indicated a net loss of 2 u relative to the parent.
Analysis of the metabolite of compound C by 1H NMR
spectroscopy (data not shown) demonstrated that under
neutral conditions the metabolite existed as a hydroxylated
structure (þ16 u relative to parent) at the benzyl C atom a to
the amide N atom. Dehydration reactions for compounds C
and D occurred in the ESI source of the mass spectrometer to
such an extent that no trace of the benzyl-hydroxylated
metabolites could be observed. For compounds such as these,
irreversible dehydration of metabolites under acid con-
ditions, or loss of the label in the ion source of the mass
spectrometer, would preclude use of the 18O-exchange
technique presented here.
Compound B, containing an N,N-dimethylamide moiety,
was observed by high-resolution MS to have produced a
hydroxylated metabolite after incubation with rat and hu-
man liver microsomes, presumably on one of the amide N-
methyl groups. Upon treatment with [18O]water and H2SO4
as described, this metabolite N-dealkylated completely to
the monomethyl amide derivative (data not shown), under-
scoring the high lability of some carbinolamides. This
observation is consistent with a previous report,6 in which
the stability of N-hydroxymethyl-N-methylbenzamide was
much less than that of N-hydroxymethylbenzamide.
CONCLUSIONS
A simple and rapid technique has been developed for
conclusively distinguishing carbinolamide compounds
from other possible isobaric N-oxygenated species. A key
advantage of the technique as compared to other derivatiza-
tions which could be employed for the same purpose is the
fact that chromatographic retention times are not altered as
a result of chemical modification, thus allowing a direct
comparison of mass chromatograms before and after
treatment. The technique has been used with multiple
Copyright # 2010 John Wiley & Sons, Ltd.
Rapid Commun. Mass Spectrom. 2010; 24: 2151–2161
DOI: 10.1002/rcm

LC/MS/MS to identify carbinolamide metabolites 2161
isobaric metabolites of compound A to distinguish between
those in which hydroxylation occurred a to the amide N
atom, and those at other positions on the same aliphatic ring
in which dehydration-rehydration equilibrium does not
occur. In particular, the technique has shown utility for cases
in which mass spectrometry alone was insufficient for
complete refinement of the oxidation site in compounds of
this type, or for which NMR spectroscopy was not feasible
due to its relatively high sample mass requirements. The
method is limited to structures in which the reversible
formation of an imine intermediate can produce a stable
18O-exchanged hydration product. The use of the technique
with benzyl-hydroxylated amides may not be feasible
due to the high rate of dealkylation or dehydration under
the conditions used for analysis; however, this should be
evaluated on a case-by-case basis.
3. Beckett AH. Xenobiotica. 1971; 1: 365.
4. Testa B, Jenner P. Drug Metabolism, Chemical and Biochemical
Aspects, pp. 82–84. Marcel Dekker: New York, 1976.
5. Young SD, Tanburini PP. J. Am. Chem. Soc. 1989; 111:
1933.
6. Ross D, Farmer PB, Gescher A, Hickman JA, Threadgill MD.
Biochemical Pharmacology. 1982; 31: 3621.
7. Ross D, Farmer PB, Gescher A, Hickman JA, Threadgill MD.
Biochemical Pharmacology. 1983; 32: 1773.
8. Koike M, Norikura R, Futaguchi S, Yamaguchi T, Sugeno K,
Iwatani K, Ikenishi Y, Nakagawa Y. Drug Metab. Dis. 1987;
15: 426.
9. Kedderis G, Argenbright LS, Walsh JS, Smith JL, Stearns RA,
Arison BA, Miwa GT. Drug Metab. Dis. 1989; 17: 606.
10. Slatter JG, Mutlib AE, Abbott FS. Biomed. Environ. Mass.
Spectrom. 1989; 18: 690.
11. McMahon RE, Sullivan HR. Biochem. Pharm. 1965; 14: 1085.
12. Li Y, Gorrod JW. Xenobiotica. 1994; 24: 409.
13. Pekol T, Daniels JS, Labutti J, Parsons I, Nix D, Baronas E,
Hsieh F, Gan L-S, Miwa G. Drug Metab. Dis. 2005; 33:
771.
14. Mutlib AE, Abbott FS. Drug Metab. Disp. 1992; 20: 451.
15. Tajima M, Iida T, Yoshida S, Komatsu K, Namba R, Yanagi
M, Noguchi M, Okamoto H. J. Biol. Chem. 1990; 265:
9602.
16. Tanaka FS, Swanson HR, Frear DS. Phytochem. 1972; 11: 2701.
17. Merritt DA, Bessire AJ, Vaz AD, Sams JP, Lynch MP. J. Vet.
Pharmacol. Therap. 2007; 30 (Suppl 1): 17.
Acknowledgements
The authors acknowledge the work of Kim A. Johnson,
Thomas O’Connell, and Chandra Prakash for the identifi-
cation of the benzyl-hydroxylated metabolite of compound C
1
by H NMR.
18. Seaton QF, Lawley CW, Akers HA. Analyt. Biochem. 1984;
138: 238.
19. Tenn WJ III, Murphy JL, Bim-Merle JK, Brown JA, Junia AJ,
Price MA, Nagorski RW. J. Org. Chem. 2007; 72: 6075.
20. Kalgutkar AS, Nguyen HT. Drug Metab. Disp. 2004; 32: 943.
21. Ramanathan R, Su A-D, Alvarez N, Blumenkrantz N,
Chowdhury SK, Alton K, Patrick J. Anal. Chem. 2000; 72:
1352.
REFERENCES
1. Hucker HB. Drug Metabolism Rev. 1973; 2: 33.
2. Coutts RT, Su P, Baker GB. J. Pharmacological and Toxicological
Methods. 1994; 31: 177.
Copyright # 2010 John Wiley & Sons, Ltd.
Rapid Commun. Mass Spectrom. 2010; 24: 2151–2161
DOI: 10.1002/rcm