Anda di halaman 1dari 16

Research article

Received: 4 December 2013

Revised: 18 February 2014

Accepted: 10 March 2014

Published online in Wiley Online Library

(wileyonlinelibrary.com) DOI 10.1002/jms.3359

Explanation through density functional theory


of the unanticipated loss of CO2 and differences
in mass fragmentation proles of ritonavir and
its rCYP3A4-mediated metabolites
Shalu Jhajra,a Tarun Handa,a Sonam Bhatia,b P. V. Bharatamb and
Saranjit Singha*
In the present study, the metabolism of ritonavir was explored in the presence of rCYP3A4 using a well-established strategy
involving liquid chromatographymass spectrometry (LCMS) tools. A total of six metabolites were formed, of which two were
new, not reported earlier as CYP3A4-mediated metabolites. During LCMS studies, ritonavir was found to fragment through
six principal pathways, many of which involved neutral loss of CO2, as indicated through 44-Da difference between masses
of the precursors and the product ions. This was unusual as the drug and the precursors were devoid of a terminal carboxylic
acid group. Apart from the neutral loss of CO2, marked differences were also observed among the fragmentation pathways of
the drug and its metabolites having intact N-methyl moiety as compared to those lacking N-methyl moiety. These unusual
fragmentation behaviours were successfully explained through energy distribution proles by application of the density
functional theory. Copyright 2014 John Wiley & Sons, Ltd.
Additional supporting information may be found in the online version of this article at the publishers web site.
Keywords: ritonavir; metabolism; rCYP3A4; mass fragmentation; CO2 neutral loss; group substitution effect; density functional theory

Introduction

452

Mass spectrometry is an indispensable tool for drug metabolite


identication (Met-ID) studies. In particular, high-resolution
(HRMS), multi-stage (MSn) and hydrogen deuterium exchangemass spectrometry (HDXMS) investigations on the drug and
its biotransformation products help to obtain comprehensive
information on the structures of the latter. A strategy for Met-ID,
involving establishment of fragmentation proles of the parent
and its metabolites, and their comparison, has been proposed
earlier by us.[13] The metabolites, being similar to the parent drug,
usually follow the same or parallel mass fragmentation prole, with
some of the resulting product ions differing in mass in accordance
with the changes in structural motifs where biotransformation had
taken place. A critical comparison among various product ions
yields vital information regarding the type and site of the metabolic
change(s) involved. For example, an increase in the mass of a
metabolite by 16 Da owing to metabolic oxidation would usually lead
to a constant mass difference of 16 Da during fragmentation in all the
steps where an altered moiety remains intact. The fragments originating from the unaltered part of the structural framework generally tend
to yield product ions identical to those of the parent.
The purpose of the present study was to investigate rCYP3A4-mediated metabolism of ritonavir (1,3-thiazol-5-ylmethyl-N-[(2S,3S,5S)
-3-hydroxy-5-[(2S)-3-methyl-2-{[methyl({[2-(propan-2-yl)-1,3-thiazol4-yl]methyl})carbamoyl]amino}butanamido]-1,6-diphenylhexan-2yl]carbamate, Fig. 1). We employed commercial recombinant
humanized CYP3A4 (rCYP3A4) isozyme and detected a total of
six metabolites. Incidentally, a near overlapping study was reported

J. Mass Spectrom. 2014, 49, 452467

lately by Lin et al.,[4] in which only four (C-hydroxylated ritonavir, Ndealkylated ritonavir, N-demethylated and deacylated ritonavir) out
of six metabolites were detected when metabolic reaction was carried
out using CYP3A4 puried from expressed Escherichia coli TOPP3 cells.
While carrying out metabolite characterization using mass
spectrometry tools,[2] we conducted investigations in positive
electrospray ionization (+ESI) mode. We found that ritonavir
was dissociated through six principal mass fragmentation routes,
wherein neutral loss of 44 Da was observed in many fragmentation steps. The exact mass of the losses, calculated from the
difference between accurate (observed) masses of the precursor
and the product ions involved, matched best to CO2
(43.9898 Da). The same observation was made even with
metabolites, where loss of 44 Da occurred during MS2, MS3 or
MS4 transitions. This observation was unusual, as the parent drug
and its metabolites were devoid of a terminal carboxylic acid
group. Apart from the neutral loss of CO2, marked differences
were even observed in fragmentation pathways of the drug
and its metabolites, particularly those that had intact N-methyl

* Correspondence to: Saranjit Singh, Department of Pharmaceutical Analysis,


National Institute of Pharmaceutical Education and Research (NIPER), Sector
67, S.A.S. Nagar 160 062, Punjab, India. E-mail: ssingh@niper.ac.in
a Department of Pharmaceutical Analysis, National Institute of Pharmaceutical
Education and Research (NIPER), S.A.S. Nagar, Punjab, India
b Department of Medicinal Chemistry, National Institute of Pharmaceutical
Education and Research (NIPER), S.A.S. Nagar, Punjab, India

Copyright 2014 John Wiley & Sons, Ltd.

Explanation for unanticipated loss of CO2


followed by the determination of exact (theoretical) masses of
their corresponding protonated species.
In vitro metabolite generation

Figure 1. Chemical structure of ritonavir. The structure is divided into 3


parts (A, B and C) for easy understanding of drug fragmentation pattern
and metabolic changes.

moiety in contrast to N-dealkylated metabolites. To explain both


these odd observations, in silico energy distribution proles were
established by applying density functional theory (DFT).

rCYP3A4 (nal conc. 10 ppm) was added to the culture tubes


containing 40-l phosphate buffer (100 mM, pH 7.4). This mixture
was pre-incubated for 5 min at 37 1 C, followed by addition of
ritonavir (nal conc. 10 M). Subsequently, NADPH solution (a cofactor,
nal conc. 1.3 mM) was added to initiate the metabolic reaction.
The volume was made up to 200 l with 100 mM phosphate buffer.
From each of the reaction mixtures, 20-l aliquot was collected at
0 min and quenched with an equal volume of chilled ACN. The
remaining sample was incubated at 37 1 C for 1 h for metabolic
reactions to occur. Two other sets, a control devoid of NADPH
and a blank without rCYP3A4, were also prepared. These solutions
were simultaneously subjected to incubation at 37 1 C for 1 h.
Mass fragmentation studies on ritonavir

Experimental
Chemicals and materials
Pure ritonavir was obtained as a gratis sample from Ranbaxy
Pharmaceuticals Ltd. (Gurgaon, India). Nicotinamide adenine dinucleotide phosphate-reduced tetrasodium salt (NADPH), testosterone, dipotassium hydrogen phosphate and potassium dihydrogen
phosphate were purchased from HIMEDIA Laboratories Pvt. Ltd.
(Mumbai, India). rCYP3A4 was acquired from BD Gentest, BD
Biosciences (California, USA). A Zorbax Eclipse XDB, C18 column
(250 mm 4.6 mm, 5 ) from Agilent Technologies (Wilmington,
DE, USA) was used for the separation of the drug and its metabolites.
For liquid chromatographymass spectrometry (LCMS) analyses,
high performance liquid chromatography grade acetonitrile (ACN)
from J.T. Baker (Mexico, USA), purchased locally from RFCL Limited,
Mohali, India was used. Ultra pure water was obtained from a water
purication unit (ELGA, Bucks, England).
Instruments
For liquid chromatographyhigh resolution mass spectrometry
(LCHRMS) studies, a 1100 series LC from Agilent Technologies
(Waldronn, Germany) connected to a time-of-ight MS (MicrOTOFQ, Bruker Daltonics, Bremen, Germany) was used. The entire unit
was operated using Hystar (ver. 3.1) and MicrOTOF control (ver. 2.0)
software. Liquid chromatographymultiple stage mass spectrometry
(LCMSn) data were acquired using Accela LC (Thermo Electron
Corporation, San Jose, USA), which was connected to a linear ion trap
quadrupole mass spectrometer (LTQXL, Thermo Electron Corporation). Instrument control and data collection were carried out by
using Xcalibur software (ver. 2.0.7 SP1). The mass studies were essentially carried out in +ESI mode after optimization of mass operation
parameters. A 5-mM solution of sodium formate (Sigma-Aldrich
Chemicals, Bangalore, India) was used as a calibrant in HRMS studies.
HDXMS studies were conducted on drug using the MS/TOF system.
Methodology
In silico metabolite prediction

J. Mass Spectrom. 2014, 49, 452467

LCMS (LCHRMS and LCMS ) studies on the metabolites

An LC method was rst developed for the separation of the metabolites and their characterization by mass tools. To achieve optimal
separation of the analytes from the polar matrix components, a gradient mode was employed, using a mobile phase composed of ACN
and 10 mM ammonium acetate (pH 4.7). The ow rate and column
temperature were 0.4 ml/min and 25 C, respectively. The initial mobile phase ratio of ACN:buffer was 10:90, which was maintained for
rst 5 min and changed to 90:10 in 40 min. It was held at this composition till 45 min, again brought back to initial ratio in 48 min
and was equilibrated for the next 12 min. This gradient LC method
was employed further in LCHRMS and LCMSn studies, which were
conducted on prepared in vitro incubated samples. Positive ionization mode was selected for LC-MS studies on the metabolites.
The metabolites were not detected in negative ionization mode,
probably due to lack of suitable functional groups. The parent itself
also had very low ionization in the negative mode.
Strategy used for structure elucidation of the metabolites

A generic strategy for the characterization of the metabolites, proposed earlier from our laboratory,[2] was followed in the present
study. The initial step involved the generation of total ion chromatogram (TIC) on rCYP3A4-treated drug sample (including zero
min sample) using LCMS/TOF. It was followed by prediction of
CYP3A4-mediated metabolites by stand-alone computational tool,
like MetaSite. The exact mass of each predicted metabolite and of
those previously reported in the literature were used to generate
post-acquisition extracted ion chromatograms (EICs). The metabolites present in the TIC were identied based on EICs, while the
peaks due to matrix components were segregated by comparison
of the TIC of the sample with the blank (0-min sample). Additionally,
analyses of accurate mass data and monitoring of the characteristic

Copyright 2014 John Wiley & Sons, Ltd.

wileyonlinelibrary.com/journal/jms

453

Before initiating experimental studies, molecular masses of


CYP3A4-mediated metabolites were predicted using MetaSite
software (Molecular Discovery, Pinner, Middlesex, UK). This was

For HRMS studies on the drug, a 2 g/ml solution was directly


injected in the MS/TOF system at a ow rate of 6 l/min, using
an integrated syringe pump. The MSn studies were similarly
conducted by directly injecting a 5 g/ml solution of the drug
into the LTQXL system at a ow rate of 3 l/min. The HDXMS
study was carried out by preparing drug solution in D2O:ACN
(50:50) and directly injecting it into the MS/TOF system. Critical
HRMS and MSn parameters (Supplementary Table 1) were
optimized in all these experiments, and the same were used for
all subsequent LCMS studies on the metabolites.

S. Jhajra et al.
formed from the molecular ion [M + H]+. Among them, the fragments
of m/z 426 and 296 were most abundant. Some of the fragments were
seen in MSn studies only. The most probable molecular formula for the
ions observed in HRMS studies was calculated from their
accurate masses with the help of an elemental composition
calculator (available from; http://www.wsearch.com.au/Tools/
elemental_composition_calculator.htm). Additionally, HDXMS
studies provided information about the number of labile
hydrogen atoms present in their structures. The whole data are
compiled in Table 1, which also includes HDXMS values, accurate
m/z values for the losses and their probable molecular formulae.
The MSn data, which helped in establishing link among various
observed ions, are listed separately in Table 2. A comprehensive
fragmentation pattern of ritonavir was delineated taking into
account all available data. The same is shown in Fig. 3. Such
comprehensive mass fragmentation of ritonavir is not previously
reported in the literature.
Apparently, the molecular ion of ritonavir ([M + H]+ = m/z 721)
underwent fragmentation through six principal routes to form both
different and common product ions (Fig. 3), highlighting the
protonation at different positions in the drug molecule. Critical
comparison among these routes also revealed that many steps
were essentially similar to each other, with some of them differing
in sequence of losses. The same are individually discussed below:

common fragments produced from the drug and the metabolites[1]


helped in identication of the latter in the samples.
The accurate mass data helped in calculation of the difference in
observed mass and thus elemental composition of the metabolites
with respect to the drug. This allowed determination of ring plus
double bonds (RDBs), elemental composition and metabolic
change in each metabolite, including metabolic events involved.
Finally, help was taken of MSn data, and the fragmentation
pathways generated therefrom were employed to justify the
proposed structure of each metabolite.
DFT calculations

Geometry optimizations without symmetry constraints were


performed on the model structure of ritonavir and its metabolites
(by taking relevant parts of the structures). In the case of metabolites, the portions where metabolic changes resulted into differential
mass fragmentation pattern were selected as the model structures.
DFT calculations[5] were carried out using the B3LYP method[6,7] with
6-31 + G(d) basis set as implemented in Gaussian 09 suit of
packages.[8] Frequencies were computed analytically at B3LYP/
6-31 + G(d) level for all the optimized species to characterize each
stationary point as a minimum or a transition state, and to estimate
the zero-point energies (ZPE). The calculated ZPE values (at 298.15 K)
were scaled by factors of 0.9806 for B3LYP level calculations.[9]
Further, the partial atomic charges on various atomic centres were
estimated by natural bond orbital analysis[10,11] using the internal
module of Gaussian 09 at the B3LYP/6-31 + G(d) level of theory.
All the transition states were characterized by single negative
imaginary frequency.

Route 1: The rst route involved loss of H2O from the parent to
form a product ion of m/z 703, which was attributed to the
protonation of secondary alcoholic group in the drug structure.
The latter underwent further fragmentation through the pathway m/z 703 533 489 347, 250.
Route 2: This route involved cleavage of C30N31 urea amide bond
present in part C of the parent ion to form product of m/z 551,
accompanied with a neutral loss of 1-(2-isopropylthiazol-4-yl)
-N-methylmethanamine, due to the charge present on the tertiary
nitrogen. In this case, the product ion of m/z 533 was formed from
m/z 551 upon the loss of H2O. The latter further followed a

Results and discussion


Mass fragmentation pattern of ritonavir
Figure 2 shows the MS/TOF line spectrum of ritonavir in ESI positive
mode. Evidently, a total of 17 fragments (as labelled in Fig. 2) were

Intens.

+MS2(721.0000), 4.2-4.5min

x10 5
2.5

h
426.1848

2.0

268.1489

q
1.5

0.0

300

500

600

b
677.3316

e d

703.3098
721.3221

[M+H]+
400

606.3119

200

200

533.2236
551.2338

489.2332
507.2438

408.1740

l
n

365.1657
382.1936

296.1434

311.1745

197.0722

250.1560

1.0

0.5

347.1577

700 m/z

454

Figure 2. HRMS spectrum of ritonavir in ESI positive ionization mode.

wileyonlinelibrary.com/journal/jms

Copyright 2014 John Wiley & Sons, Ltd.

J. Mass Spectrom. 2014, 49, 452467

Explanation for unanticipated loss of CO2


Table 1. HRMS and HDXMS data for ritonavir and its fragments in ESI positive ionization mode
Peak no.

HRMS
data

Best possible Exact mass of the Error in


molecular
most probable
mmu
formula
structure

RDB

Possible
parent ion

721.3221
703.3098
677.3316
606.3119
551.2338
533.2236
507.2438
489.2332
426.1848
408.1740
392
382.1936
365.1657
347.1577
311.1745

C37H49N6O5S2
+
C37H47N6O4S2
+
C36H49N6O3S2
+
C33H44N5O4S
+
C29H35N4O5S
+
C29H33N4O4S
+
C28H35N4O3S
+
C28H33N4O2S
+
C23H28N3O3S
+
C23H26N3O2S
+
C24H30N3O2
+
C22H28N3OS
+
C22H25N2OS
+
C22H23N2S
+
C19H23N2O2

721.3200
703.3095
677.3302
606.3109
551.2323
533.2217
507.2424
489.2319
426.1846
408.1740
392
382.1948
365.1682
347.1576
311.1754

2.1
0.3
1.4
1.1
1.5
1.9
1.4
1.3
0.2
0.0

1.2
2.5
0.1
0.9

16.5
17.5
15.5
14.5
14.5
15.5
13.5
14.5
11.5
12.5

10.5
11.5
12.5
9.5

+
[M + H]
+
[M + H]
+
[M + H]
+
[M + H]
a/d
b/d
e/f
+
[M + H]
d/h
f
b/h
k
g/i/l
f/h/c

o
p*
q
q*
r

296.1434
285
268.1489
268
250.1560

C14H22N3O2S
+
C18H25N2O
+
C13H22N3OS
+
C18H22NO
+
C18H20N

296.1427
285
268.1478
268
250.1590

0.7

1.1

3.0

5.5

4.5

9.5

[M + H]
b
o
h
g/l/q

s
t*

197.0722
171

C9H13N2OS
+
C9H15N2S

197.0743
171

2.1

4.5

q
q

[M + H]
a
b
c
d
e
f
g
h
i
j*
k
l
m
n, n

Losses

H2O
CO2
C4H5NOS
C8H14N2S
C8H14N2S/H2O
C8H14N2S/CO2
CO2/H2O
C14H21N3O2S
C6H13NO2/H2O
C4H5NOS
C14H21N3O2S/CO2
NH3
C6H10N2O2/CH3NO2S/H2O
C9H12N2OS/C4H5NOS/
C14H21N3O2S
C23H27N3O3S
C18H24N4O2S2
CO
C5H6N2O2S
C10H13N3O2S/C4H5NOS/
H2O
C4H9N
C5H9NO

HDXMS
data

Number
of labile
hydrogen(s)

726.3460
706.3179
682.3561

555.2522
535.2304
511.2644
491.2270
431.2131

387.2185
367.1770
348.1622
315.1950

5
3
5

4
2
4
2
5

5
2
1
4

297.1459

269.1510

251.1641

197.0759

* Observed only during MS study (Table 2), hence data other than molecular formulae were not available.

further lost H2O to form a product of m/z 489, which subsequently


followed the same sequence as in Route 1, viz., m/z 489 347, 250.

Table 2. MS data of ritonavir in ESI positive ionization mode


n

MS stage

Precursor
ion(s)*

721

703
677
606
551
426
296
533
507
408
382
e
268
f
268
489
365

MS

MS

MS

MS

Product ion(s)*
703, 677, 606, 551, 533, 507, 489,
#
#
#
426, 408, 382, 365 , 347 , 311 ,
#
296, 268, 250
533, 489
#
#
#
507, 489, 392, 382, 365 , 285 , 250
#
#
#
311 , 296 , 268
#
533, 507, 489, 408 , 347, 311, 250
e
408, 382, 365, 347, 311, 268 , 250
f
268
#
#
489, 347 , 250
#
#
#
#
489, 392 , 347 , 311 , 250
#
#
347 , 250
#
#
#
365, 347 , 268 , 250
#
250
#
#
197 , 171
#
#
347 , 250
#
#
347 , 250

* All values in m/z, Ions could not be captured for further MS


a b c
studies, superscripts , , ........ are notations used for product ions
n
that were subsequently subjected to MS analysis. Ions of m/z
n
392, 285 and 197 were appeared only during MS fragmentation.

J. Mass Spectrom. 2014, 49, 452467

Route 4: The fourth route involved the generation of a protonated ion


of m/z 426 upon the loss of part C (Fig. 1) from the parent. This route
was observed due to charge on the secondary nitrogen of part B of
the drug. The ion of m/z 426 further fragmented to ions of m/z 408,
382, 365, 347, 311, 268 and 250. Among these, the ion of m/z 408 on
MS4 generated fragments of m/z 347 and 250, again similar to Routes
13. The ion of m/z 382 followed the same fragmentation pathway as
in Route 3. Apart from the ions of m/z 408 and 382, the products of
m/z 311 and 268 were also formed directly from m/z 426. Among
the remaining ones, the fragment of m/z 268 further dissociated to
form a product ion of m/z 250, which, however, was not observed
from the same mass ion in Route 3, probably due to lower abundance of precursor ion in the latter case.
Route 5: The fth route resulted into formation of a product of m/z
606 with a loss of thiazol-5-ylmethanol from part A of the drug

Copyright 2014 John Wiley & Sons, Ltd.

wileyonlinelibrary.com/journal/jms

455

pathway m/z 533 489 347, 250, same as Route 1. Along with the
ion of m/z 533, other fragments generated parallely from m/z 551
were of m/z 507, 408, 392, and 311. Of these, the ion of m/z 507

Route 3: In the third route, the parent fragmented to product ion of


m/z 677, with an observed accurate mass difference of 43.9905 Da
(error <1.5 mmu), which corresponded to CO2. This CO2 neutral loss
was unusual in the absence of a carboxyl moiety in the parent (Fig. 3).
The fragment of m/z 677 further formed multiple ions in MS3 step,
including one of m/z 507, which followed the same fragmentation
substeps as in Route 2, viz., m/z 507 489 347, 250. Other
fragments formed on MS3 were of m/z 392, 382, 365 and 285. In
MS4 step, the ion of m/z 382 was further reduced by following the
pathways: m/z 382 365 347, 250, and m/z 382 268. The
product ions of m/z 347 and 250 were the same as in Routes 1 and 2.

S. Jhajra et al.

Figure 3. Proposed mass fragmentation prole of ritonavir in ESI positive ionization mode.

(Fig. 1), attributed to charge on the carbamate oxygen. In this


fragmentation route, the product ion of m/z 606 further
fragmented into ions of m/z 311, 296 and 268. The fragment of
m/z 296 sequentially cleaved by following the pathway: m/z
296 268 197, 171.
Route 6: The sixth route involved direct fragmentation of the parent molecular ion to a carbonyl cation (m/z 296) equivalent to
part C of the drug structure (Fig. 1). The product ion of m/z 296
followed exactly the same pathway as in Route 5, where it was
formed from the precursor of m/z 606, instead of the parent.

On the whole, the unusual loss of CO2 was observed from


fragments of m/z 551, 533 and 426, leading further to ions of
m/z 507, 489 and 382, respectively.

Generation of TICs of control and actual samples


A method separating the drug, its metabolites and matrix components was optimized on LCMS/TOF by using the actual sample.
The resultant TIC for both control and actual samples are shown in
Fig. 4.

lntens.
M4

5000

4000

3000

M6

M1

2000

M2

M3

1000

M5

0
32

34

36

38

40

42

44

46 Time [min]

456

Figure 4. TIC of ritonavir metabolites after incubation with rCYP3A4. The metabolites were identied based on difference in peaks from the matrix
chromatogram, which is also shown in the gure in red.

wileyonlinelibrary.com/journal/jms

Copyright 2014 John Wiley & Sons, Ltd.

J. Mass Spectrom. 2014, 49, 452467

Explanation for unanticipated loss of CO2


Consideration of in silico predicted and previously known
metabolites
In total, 15 metabolites were predicted using MetaSite software
for CYP3A4-dependent metabolism of ritonavir. Additionally,
the metabolites known in the literature were also taken into
account. The theoretical exact mass of their corresponding protonated species were searched for checking their presence in
the actual samples.
Identication of metabolites in TIC of actual samples
As evident in Fig. 4, there were multiple peaks in TIC, which were
due to both the metabolites and the matrix. The comparison of
TIC against the control, and consideration of other approaches
outlined in the experimental section, helped in identifying the
peaks due to the metabolites. A total of six peaks, which could be
attributed to be the metabolites, are labelled in Fig. 4 as M1M6.
The consideration of EICs (Supplementary Fig. 1) for the exact
masses of predicted and previously known metabolites further
helped in substantiation that all these six peaks were in fact of
the metabolites. HRMS data and corresponding derived information for the metabolites are given in Table 3. As shown in Table 3,
a few ions with low abundance had higher error in mmu, which
was likely due to interference of matrix ions with them. The generated MSn data are listed in Table 4.
Characterization of metabolites based on MS data
Metabolite M1

J. Mass Spectrom. 2014, 49, 452467

The Q-TOF line spectrum of this metabolite is shown in Supplementary Fig. 2. Its HRMS and MSn fragmentation data are listed
in Tables 3 and 4, respectively. This metabolite had a molecular
ion of m/z 582.2816, with the difference of 139.0368 Da from
the parent drug corresponding to C7H9NS, which indicated loss
from Part C of the drug. This was supported by the appearance
of the ion of m/z 426 (Route 4) and also by the lack of the
fragment of m/z 296 (Route 6), which originated from Part A
and B, and C, respectively in the case of the drug. The MS3 and
MS4 fragmentation analysis of another major fragment of m/z
525, which resulted from [M + H]+ after loss of C2H3NO, also
revealed that the metabolism involved Part C of the drug. The
difference of 99 Da between m/z 525 and 426 corresponded to
2-amino-3-methylbutanal, a partial motif of Part C, further conrming
that the site of metabolism was Part C only. The postulated fragmentation pattern of the metabolite is shown in Fig. 6.
Just like the parent drug, the loss of CO2 in this metabolite was
also observed in the sequences: m/z 525481 and m/z 426382
(shown in Fig. 6). These ions of m/z 525 and 426 did not have any
free caboxlic group.
Metabolite M3
This metabolite had an accurate mass of m/z 723.2901 (Table 3),
with the difference of 1.9717 Da from the parent drug
(m/z 721.3184). The possibility of the removal of two hydrogen
atoms was eliminated, as that would have meant a difference
of 2.0156 Da, almost 43.9 mDa away from the observed value.
Thus an alternate possibility was considered, which involved
demethylation followed by hydroxylation, or vice versa. The only
facile site for demethylation existed in Part C of the structure of
drug. The same was supported by HRMS (Table 3) and MSn
fragmentation data (Table 4), wherein all the fragments
containing Part C also had the same mass difference of
~1.9717 Da. For example, the line of m/z 296 (Route 6) seen in
the drug was replaced in the spectrum of the metabolite by line
of m/z 298, highlighting the involvement of hydroxylation and
demethylation. Due to this reason only, the fragment of m/z
298 showed a facile loss of H2O to generate a product ion of
m/z 280. On other hand, the ion of m/z 426 (Route 4) that
originated from Parts A and B remained same as the drug. The
structure and fragmentation pathway of this metabolite are
shown in Fig. 7.
Again, similar to the parent drug, this metabolite also showed
loss of CO2, which happened during conversion of ion of m/z 426
to product of m/z 382 (shown in Fig. 7).
Metabolite M4
The accurate mass difference of metabolite M4 against the drug
was 15.9922 Da (~16 Da), which pointed towards its generation
upon metabolic oxidation or hydroxylation of the drug. The Q-TOF
line spectrum of this metabolite is shown in Supplementary Fig. 2.
The corresponding HRMS and MSn data are listed in Tables 3 and
4, respectively. The [M + H]+ of the metabolite fragmented to give
ions of m/z 426 and 312, as compared to m/z 426 and 296 (Route
6) in the case of the drug (Fig. 3). The ion of m/z 426 (Route 4) in
the metabolite had a fragmentation pattern similar to that observed
in the drug, revealing no metabolic changes in Parts A and B. The
appearance of an ion of m/z 312, with a mass of 16 Da higher than
the characteristic ion of m/z 296 in the drug, clearly indicated that
Part C of the drug was involved in the oxygenation process.

Copyright 2014 John Wiley & Sons, Ltd.

wileyonlinelibrary.com/journal/jms

457

M1 was characterized as the deacylated metabolite involving


removal of Part A from ritonavir (Fig. 1). Its Q-TOF line spectrum is
shown in Supplementary Fig. 2. The structure was supported
by HRMS (Table 3) and MSn fragmentation data (Table 4). The
metabolite had accurate mass of m/z 580.3307 with a difference
of 140.9877 Da (~141) from the drug, which implied a loss of
C5H3NO2S corresponding to Part A of the drug. A careful comparison of its fragmentation pattern with the drug revealed a
difference of 141 Da throughout the sequence of fragmentation
observed in the drug. For example, Route 2 in the drug viz. m/z
721 551 533 (Fig. 3) was replaced in the metabolite by the
sequence m/z 580 410 392 (Fig. 5). The presence of the
fragment of m/z 296 (Route 6) and m/z 285 (Route 3), similar to
the drug, pointed towards the presence of unaltered Part B
and C, respectively. Moreover, considering together both these
fragments completely justied the structure of this metabolite. For
further conrmation, ions of m/z 296 and 285 were subjected to
MS3 analysis. The former followed the same fate as in the drug by
yielding the product ions of m/z 268, 197 and 171. On the other
hand, m/z 285 cleaved to yield fragments of m/z 268 and 250, while
the same in the case of the drug was generated in MS3 step from a
minor product of m/z 677, in which case further fragmentation was
not possible due to its lower abundance. On subsequent fragmentation, the ion of m/z 285 reduced to form a product of m/z 268,
which dissociated further into the ions of m/z 250 and 233. On the
other hand, the ion of m/z 268, which was also formed from m/z
296, fragmented to form ions of m/z 197 and 171. This nding
established that despite the two fragments having the same m/
z of 268, they were actually different in structure. The fragment
of m/z 250 generated from the sequence m/z 285 268,
indicated that Part B of the drug also remained intact during
the biotransformation reaction.

Metabolite M2

458

wileyonlinelibrary.com/journal/jms

Copyright 2014 John Wiley & Sons, Ltd.

M4

M3*

M2

M1

Meta-bolite

[M + H]
a
b
c
d
e
f
g
h
h
i
+
[M + H]
a
b
c
d
e
f
g
h
i
j
k
l
m
+
[M + H]
a
b
c
d
e
f
g
+
[M + H]
a
b
c
d
e

Line No.
580.3307
463.2680
410.2438
392.2273
375.2089
347.2138
296.1397
285.1928
268.1527
268.1527
250.1611
582.2816
525.2553
507.2478
481.2506
463.2543
426.1866
408.1709
382.1847
365.1572
347.1428
311.1678
285.1851
268.1664
250.1611
723.2901
525.2553
507.2478
426.1867
311.1790
298.1117
280.1203
250.1510
737.3106
719.2957
551.2352
533.2241
507.2478
426.1857

MS/TOF data
+

C32H46N5O3S
+
C27H35N4O3
+
C24H32N3O3
+
C24H30N3O2
+
C24H27N2O2
+
C23H27N2O
+
C14H22N3O2S
+
C18H25N2O
+
C18H22NO
+
C13H22N3OS
+
C18H20N
+
C30H40N5O5S
+
C28H37N4O4S
+
C28H35N4O3S
+
C27H37N4O2S
+
C27H35N4OS
+
C23H28N3O3S
+
C23H26N3O2S
+
C22H28N3OS
+
C22H25N2OS
+
C22H23N2S
+
C19H23N2O2
+
C18H25N2O
+
C18H22NO
+
C18H20N
+
C36H47N6O6S2
+
C28H37N4O4S
+
C28H35N4O3S
+
C23H28N3O3S
+
C19H23N2O2
+
C13H20N3O3S
+
C13H18N3O2S
+
C18H20N
+
C37H49N6O6S2
+
C37H47N6O5S2
+
C29H35N4O5S
+
C29H33N4O4S
+
C28H35N4O3S
+
C23H28N3O3S

Best possible molecular


formula

Table 3. Interpretation of LCMS/TOF data of ritonavir metabolites

580.3316
463.2704
410.2438
392.2333
375.2067
347.2118
296.1427
285.1961
268.1696
268.1478
250.1590
582.2745
525.2530
507.2424
481.2632
463.2526
426.1846
408.1740
382.1948
365.1682
347.1576
311.1754
285.1961
268.1696
250.1590
723.2993
525.2530
507.2424
426.1846
311.1754
298.1220
280.1114
250.1590
737.3150
719.3044
551.2323
533.2217
507.2424
426.1846

Exact mass of the most probable


structure
0.9
2.4
0.0
6.0
2.2
2.0
3.0
3.3
16.9*
4.9
2.1
7.1
2.3
5.4
12.6
1.7
2.0
3.1
10.1
11.0
14.8
7.7
11.0
3.2
2.1
9.2
2.3
5.3
2.1
3.6
10.3
8.9
8.0
4.4
8.7
2.9
2.4
5.3
1.1

Error in mmu
12.5
12.5
10.5
11.5
12.5
11.5
5.5
7.5
8.5
4.5
9.5
13.5
12.5
13.5
11.5
12.5
11.5
12.5
10.5
11.5
12.5
9.5
7.5
8.5
9.5
16.5
12.5
13.5
11.5
9.5
5.5
6.5
9.5
16.5
17.5
14.5
15.5
13.5
11.5

RDB

[CH2]

+[O]

+[O],

[C7H9NS]

[C5H3NO2S]

Metabolic change

S. Jhajra et al.

J. Mass Spectrom. 2014, 49, 452467

Copyright 2014 John Wiley & Sons, Ltd.

f
g
h
i
j
+
[M + H]
a
b
c
+
[M + H]
a
b
c
d
e
f
g
h
i
j

Line No.
382.1848
312.1349
294.1275
284.1394
250.1610
737.3057
442.1728
296.1387
268.1463
707.3042
525.2550
507.2477
481.2662
426.1865
382.1970
365.1571
311.1677
282.1271
268.1559
250.1610

MS/TOF data
C22H28N3OS
+
C14H22N3O3S
+
C14H20N3O2S
+
C13H22N3O2S
+
C18H20N
+
C37H49N6O6S2
+
C23H28N3O4S
+
C28H35N4O3S
+
C13H22N3OS
+
C36H47N6O5S2
+
C28H37N4O4S
+
C28H35N4O3S
+
C27H37N4O2S
+
C23H28N3O3S
+
C22H28N3OS
+
C22H25N2OS
+
C19H23N2O2
+
C13H20N3O2S
+
C18H22NO
+
C18H20N

Best possible molecular


formula
382.1948
312.1376
294.1271
284.1427
250.1590
737.3150
442.1795
296.1427
268.1478
707.3044
525.2530
507.2424
481.2632
426.1846
382.1948
365.1682
311.1754
282.1271
268.1696
250.1590

Exact mass of the most probable


structure

10.0
2.7
0.4
3.3
2.0
9.2
6.7
4.0
1.5
0.2
2.0
5.3
3.0
2.0
2.2
11.0
7.7
0.0
13.7
2.0

Error in mmu

* These metabolites were conrmed to be second generation based on HR-MS data; all other metabolites were rst generation products.
indicates addition, while ( ) indicates loss of the respective moieties and mass. Unpredicted represents more than one metabolic changes.

M6

459

J. Mass Spectrom. 2014, 49, 452467

M5

Meta-bolite

Table 3. (Continued)

10.5
5.5
6.5
4.5
9.5
16.5
11.5
5.5
4.5
16.5
12.5
13.5
11.5
11.5
10.5
11.5
9.5
5.5
8.5
9.5

RDB

[CH2]

+[O]

Metabolic change

Explanation for unanticipated loss of CO2

wileyonlinelibrary.com/journal/jms

S. Jhajra et al.
n

Table 4. MS fragmentation data for ritonavir metabolites


Metabolite

MS : precursor
ion(s)*
2

M1

MS : 580

M2

MS : 410
3
MS : 296
3
MS : 285
4
MS : 392
4
a
MS : 268
4
b
MS : 268
2
MS : 582

M4

MS : 525
3
MS : 426
4
MS : 507
4
MS : 481
4
MS : 382
4
c
MS : 311
4
MS : 285
4
MS : 268
2
MS : 723
3
MS : 525
3
MS : 426
3
MS : 298
4
MS : 311
2
MS : 737
3

M5

MS :
3
MS :
3
MS :
4
MS :
2
MS :

551
426
312
533
737

567
523
442
296
268
707
525
426
282

M3

M6

MS :
3
MS :
3
MS :
3
MS :
4
MS :
2
MS :
3
MS :
3
MS :
3
MS :

Product ion(s)*
#

562 , 463 , 410, 392, 375, 296,


#
#
#
285, 268, 250 , 197 , 171
#
#
#
392, 375 , 347 , 295 , 250
a
#
#
268 , 197 , 171
b
#
#
268 , 250 , 233
#
#
#
375 , 267 , 250
#
#
197 , 171
#
#
250 , 233
#
525, 507, 481, 426, 382, 365 , 311,
#
285, 268, 250
#
#
507, 481, 311 , 285, 250
#
#
c
#
408 , 382, 365 , 311 , 268, 250
#
#
#
#
446 , 408 , 311 , 250
#
#
463 , 250
#
#
#
#
365 , 347 , 268 , 250
#
#
#
294 , 267 , 250
#
#
268 , 250
#
250
#
#
#
525, 507 , 426, 311, 298, 280 , 250
#
507
#
#
#
#
408 , 382 , 365 , 311, 250
#
280
#
250
#
#
#
719 , 551, 533, 507 , 426, 382 ,
#
#
311 , 284, 250
#
#
#
#
#
533, 507 , 489 , 408 , 311 , 250
#
#
#
#
#
408 , 382 , 365 , 268 , 250
#
#
284 , 213
#
507
#
#
#
#
719 , 567, 549 , 523 , 442, 311 ,
296, 268
#
#
#
549 ,523 , 505
#
#
392 , 311
#
#
424, 311 , 250
#
#
268, 197 , 171
#
#
197 , 171
#
#
525, 507 , 481 , 426, 282
#
#
507 ,481
#
#
#
#
#
408 , 382 , 311 , 268 , 250

* All values in m/z; Ions could not be captured for further MS


a b
studies; superscripts , : Ions of same mass that on subsequent
n
MS analysis yielded different fragments.

However, the exact position of this change could not be proposed


due to low abundance of the product ions, obverting their further
fragmentation. The fragmentation pattern of the metabolite is
postulated in Fig. 8.
In this case too, CO2 loss occurred during fragment sequences:
m/z 551 507 and m/z 426 382 (shown in the Fig. 8).
Evidently, both prescursors of m/z 551 and 426 did not have a
free caboxlic group.
Metabolite M5

460

This metabolite (m/z 737.3073) also had an accurate mass


difference of 15.9889 Da (~16 Da) from the parent molecule

wileyonlinelibrary.com/journal/jms

(m/z 721.3184). This difference again indicated towards oxidation


of the drug. The Q-TOF line spectrum of the metabolite
(Supplementary Fig. 1) and its corresponding HRMS (Table 3)
and MSn (Table 4) data showed the presence of product ions of
m/z 442 and 296, as compared to m/z 426 and 296 in the case
of the drug (Fig. 3). The common product ion of m/z 296
(Route 6) revealed that metabolic change did not occur on Part
C of the drug. Oppositely, the appearance of a product ion of
m/z 442 in the metabolite against characteristic ion of m/z 426
(Route 4) observed in the drug clearly indicated that either Part
A or B was the site of oxygenation. However, the appearance of
a fragment of m/z 250 highlighted that Part B of the drug
remained unchanged. So it could be proposed that this monooxygenated metabolite appeared upon hydroxylation of part
A of the drug molecule. The exact position of change in Part A
could not be delineated, as the key product ions were formed in
a very low abundance, obverting their further fragmentation. The
fragmentation pathway of this metabolite is proposed in Fig. 9.
In this case, CO2 loss occurred during fragment sequence: m/z
567 523 (shown in the Fig. 9), despite that the precursor was
devoid of a free caboxlic group.
Metabolite M6
Both HRMS (Table 3) and MSn data (Table 4) were used to
propose the structure of metabolite M6. This metabolite
(m/z 707.3042) had an accurate mass difference of 14.0142 Da
from the drug (m/z 721.3184), indicating elemental loss of CH2.
The Q-TOF line spectrum for the metabolite is shown in Supplementary Fig. 2. The comparison of the fragmentation pattern of
the metabolite (Fig. 10) with the drug (Fig. 3) showed that the
characteristic product ion of m/z 296 (Route 6), representing Part
C in the drug, was replaced by ion of m/z 282, again with a difference
of 14 Da. This justied the probability of demethylation from Part C
of the drug. Also, the characteristic ion of the drug of m/z 426
(Route 4) and subsequent fragments remained the same in the
metabolite, which provided support that there were no changes in
Parts A and B of the drug during the process of metabolism.
Similar to the parent drug and metabolite M2, neutral CO2 loss
occurred in this metabolite during the fragment sequences: m/z
525 481 and m/z 426 382 (shown in the Fig. 10), despite
that the precursor ions of m/z 525 and 426 did not have a free
caboxlic group.
Investigation of CO2 elimination during fragmentation of
ritonavir using DFT:
The fragmentation pattern of ritonavir and its metabolites
(except in metabolite M1 in which carbamate part was absent
due to metabolic changes) showed neutral loss of CO2 with an
accurate mass difference of ~43.9898 Da in MS2, MS3 or MS4 steps
(shown in Figs. 812). For example, during MS2 fragmentation of
the drug, the parent of m/z 721 formed product ion of m/z 677
involving loss of 44 Da. The same mass loss was also observed
in a few other transitions in the case of the drug and even the
metabolites, viz., m/z 551 507, m/z 533 489 and m/z
426 382. In the absence of a free terminal carboxylate group
present in the structures of ions of m/z 721, 551, 533 and 426
(Fig. 3), the likely logical explanation for the said loss was either
rearrangement of the carbamate moiety to a terminal
carboxylate group or the formation of an ion-neutral complex.
A critical study of the literature revealed that similar type of
neutral loss of CO2 was previously observed by Yu et al.[12] in

Copyright 2014 John Wiley & Sons, Ltd.

J. Mass Spectrom. 2014, 49, 452467

Explanation for unanticipated loss of CO2

Figure 5. Proposed mass fragmentation pattern of metabolite M1 of ritonavir in ESI positive ionization mode.

J. Mass Spectrom. 2014, 49, 452467

was inuenced by factors like 3D conguration, stability and


internal energy during its dissociation in the collision cell.[2528]
Also, DFT had been reported to be very well applicable to
gas-phase chemistry that prevailed in the MS system and could
be used to calculate electron density over a neutral molecule,
and redistribution of the same, and even its inuence on bond
strengths once the protonation has occurred.[23,29] Further, it is
helpful in the optimization of structural geometries of molecules
based on energy minimization calculations and even assists in
the prediction of the preferable sites of protonation amongst
several possibilities (relative proton afnities), including the
determination of energy redistribution and bond labilities.[23,29]
For DFT calculations, a structural part common to ritonavir and
mass fragments of m/z 551, 533 and 426 was selected as a
precursor ion (PI1+) model system. Figure 11 shows possible
pathways A and B for the CO2 loss involving rearrangement

Copyright 2014 John Wiley & Sons, Ltd.

wileyonlinelibrary.com/journal/jms

461

carbamate compounds. Also, several reports were found, which


mentioned of unusual neutral losses like SO, SO2, H2SO2, CO, H2O,
NH3, etc. from the central part of protonated or deprotonated
ions.[1322] Many of these investigations also proposed mechanisms
of occurrence of such unusual losses. In most studies involving loss
of SO, SO2, H2SO2, CO and H2O, the proposition was a gaseous
phase rearrangement, which was hypothesized to be energetically
favoured.[1321] The elimination of NH3 was justied through an
ionneutral complex formation.[22]
Therefore, the mechanism during the mass fragmentation
process in our case was explored considering both, the
rearrangement and the ionneutral complex formation. This
was done by quantum chemical analysis through DFT, an in silico
tool that had been employed by many researchers for the same
purpose, but in different situations.[23,24] The basic contention
employed was that the fragmentation pattern of the parent ion

S. Jhajra et al.

Figure 6. Proposed mass fragmentation pattern of metabolite M2.

462

and ionneutral complex, respectively. The potential energy of


PI1+ was arbitrarily considered as zero for the calculation of
relative energies of the intermediates (I1+I4+), transition states
(TS1TS4, Fig. 12) and the product ion (Pr+). The corresponding
potential energy diagrams of both the pathways A and B are
shown in Fig. 13.
As the parent had multiple potential sites (O3, O4, N2, N48) for
protonation, the proton afnity values at these centres were
estimated. The results (see Supplementary Table 2) indicated that
N48 of thiazole ring was the most preferred site for protonation,
in comparison to others. Hence, N48+ ion (PI1+) was chosen as a
precursor to develop the reaction pathway. Incidentally, PI1+
could exist in iminol tautomeric state represented by I1+, for
which further two conformations (extended and closed) were
possible, as shown in pathway A in Fig. 11. Evidently, the closed
conformation was stable by 9.11 kcal/mol, and its closed geometry assisted in the rearrangement, forming TS1 (3D structure in
Fig. 12) through simultaneous bond breaking (C45O4) and
bond formation (C45N2), with a transition state barrier of
54.68 kcal/mol. TS1 further formed intermediate (I2+) with a
potential energy of 6.37 kcal/mol with reference to PI1+. This
intermediate converted to product ion (Pr+) with an activation
energy of 35.17 kcal/mol through TS2 (3D structure in Fig. 12)
by losing CO2 through cleavage of C1N2 bond (dC1N2 = 1.56 )
and simultaneous shift of the proton from O4 to N2. As shown in
Fig. 13, this step represented the lowest energy point in the
entire reaction pathway A, with potential energy value of 5.31
kcal/mol, highlighting that CO2 loss was a thermodynamically
driven process.
The ionneutral complex is proposed to form upon transfer of
proton from ring nitrogen in CID to O4 via a through space
interaction in PI1+ or perhaps pre-existence of charge on O4, as

wileyonlinelibrary.com/journal/jms

part of the possible multiple protonation sites in the parent, P


(Pathway B). The resultant ion is shown as PI2+ in Fig. 11, which
evidently was less stable as compared to PI1+ by 34.73 kcal/mol
(see Table 2, supporting information). In PI2+, the C45O4 bond
(bond dissociation energy (BDE) of 8.71 kcal/mol) underwent
cleavage to form I3+, an ionneutral complex between thiazole
methylium cation and carbonic acid. This complex after surpassing energy barrier of 15.81 kcal/mol converted to intermediate
I4+ by the formation of C45N2 bond due to the attack of N2
centre on the electron decient carbon (C45) of methylium
cation as shown in TS3 (3D structure in Fig. 12). The intermediate I4+ is characterized by quaternary nitrogen centre carrying a
COOH moiety. The elimination of CO2 from this intermediate
happened through proton migration, justied through attraction of the latter (during the loss of CO2) towards the -cloud
of the aromatic thaizole ring. The result was the formation of
another transition state, TS4. The transfer of carbamate COOH
proton with concomitant cleavage of the N-CO2 bond
eventually led to the formation of product ion (Pr+). This pathway
B, thus, is also thermodynamically driven.
A critical comparison of pathways A and B revealed that due to
less stability of PI2+ than PI1+ by 34.73 kcal/mol, the pathway A
was favoured initially. However, the consideration of overall
energy barrier highlighted that during pathway A, there was
subsequent hurdle of ~73 kcal/mol, while it was a lower value
of ~41 kcal/mol in pathway B. This meant that pathway B was
the more likely mechanism involved. Other reasons dictating
the preference for pathway B included the possibility of delocalization of cation on the methyl thiazole ion, thus improving its
stability, and subsequent interaction between thiazole
methylium ion and the phenyl ring of Part B of drug (Fig. 1) in
the neutral moiety after loss of CO2.

Copyright 2014 John Wiley & Sons, Ltd.

J. Mass Spectrom. 2014, 49, 452467

Explanation for unanticipated loss of CO2

Figure 7. Proposed mass fragmentation pattern of metabolite M3 in ESI positive ionization mode.

463

Figure 8. Proposed mass fragmentation pattern of metabolite M4 in ESI positive ionization mode.

J. Mass Spectrom. 2014, 49, 452467

Copyright 2014 John Wiley & Sons, Ltd.

wileyonlinelibrary.com/journal/jms

S. Jhajra et al.

Figure 9. Proposed mass fragmentation pattern of metabolite M5 in ESI positive ionization mode.

464

Figure 10. Proposed mass fragmentation pattern of metabolite M6 in ESI positive ionization mode.

wileyonlinelibrary.com/journal/jms

Copyright 2014 John Wiley & Sons, Ltd.

J. Mass Spectrom. 2014, 49, 452467

Explanation for unanticipated loss of CO2

Figure 11. Plausible reaction pathways for the loss of CO2 through the rearrangement (pathway A) and ionneutral mechanism (pathway B). The 3D
optimized geometries for whole reaction pathway are given in Supplementary Figs. 45. All values are in kcal/mol.

TS2

TS1

TS4

TS3

Figure 12. 3D optimized geometries of transition states involved in reaction pathways A and B. Dashed lines represents bond distances in unit.

(Pathway B)

(Pathway A)

80
70

Energy (kcal/mol)

60
50
40
30
20
10
0
-10
-20
+

Figure 13. Potential energy surface for the reaction pathways A and B involving the loss of CO2 from the precursor ion. Energy of I1 in closed con+
formation was taken under consideration. The energies are in kcal/mol relative to the precursor ion (PI1 ).

Investigational analysis of differential fragmentation behaviour


of ritonavir and its metabolites by using DFT

J. Mass Spectrom. 2014, 49, 452467

Copyright 2014 John Wiley & Sons, Ltd.

wileyonlinelibrary.com/journal/jms

465

A critical appraisal of the fragmentation pathways of ritonavir


(Fig. 3) and its metabolites M1M6 (Figs. 5 to 10) showed a differential behaviour. Ritonavir and its metabolites, M1, M4 and M5,
which had intact tertiary N31, underwent cleavage of C30N31
bond, as exemplied through conversion of parent to product
ion of m/z 551 in the case of the drug (Route 2). On the other
hand, metabolites M2, M3 and M6, in which N31 was converted
to secondary form as a result of demethylation or deacylation,
followed cleavage of N29C30 bond. In order to understand this
differential behaviour during fragmentation, quantum chemical

calculations were again performed by using DFT. The model


structures taken for DFT calculations are shown in Figs. 14a and
b. The results are given in Table 5. The minor difference in relative
energy values under gas phase condition suggested that protonated ions N29+ and N31+ were isoenergetic to each other. In the
case of implicit solvent analysis, which was done by taking ACN
as a solvent (major component in the mobile phase), the relative
stability of the two protonated ions differed by 9.68 kcal/mol,
with N31+ being considerably more stable. It highlighted that
the fragmentation of C30N31 bond was energetically preferred
for ritonavir and its metabolites M1, M4 and M5, in line with the
experimental observation. The same was also supported by calculated proton afnity values in gas and solvent phase conditions

S. Jhajra et al.

Summary

Figure 14. Model structures used for analysis of differential fragmentation behaviour of ritonavir and its metabolites by using DFT.
Table 5. Relative energy (RE) and proton afnity (PA) values (kcal/mol)
of different ions calculated using B3LYP/6-31 + g(d) level of theory
RE

Fig.
Fig.
Fig.
Fig.

14a-N29
+
14a-N31
+
14b-N29
+
14b-N31

PA

Gas
phase

Solvent
phase*

Gas
phase

Solvent
phase*

0.57
0.00
0.00
10.68

9.68
0.00
0.00
7.83

220.83
221.45
220.53
209.68

242.75
252.47
251.04
243.27

* Acetonitrile was chosen as solvent as it was the major


component of the mobile phase.

466

(Table 5). The respective values were 220.83 kcal/mol and


242.75 kcal/mol for N29+, and 221.45 kcal/mol and 252.47 kcal/
mol for N31+. Hence, this data also supported better stability of
the latter protonated ion. On the contrary, in the case of M2, M3
and M6 metabolites, N29+ ion was signicantly stable even under
the gas phase condition by 10.68 kcal/mol, and it also maintained
its thermodynamic stability in the investigated solvent condition
(Table 5). The predominance of N29+ ion over N31+ was also
supported by the higher proton afnity value for the attack of
proton on N29 centre, which supported the dissociation of
N29C30 bond in demethylated metabolites of ritonavir.
In the literature, a study of the effect of N-substitution on NC
bond strength of amide group by using DFT and a comparison
with experimentally calculated BDE was reported by Marochkin
et al.[30] They found that electron-donating substituents like C
(O), NHC(O), O and N3 on carbonyl of the amide
bond had stabilizing effect and resulted in increase in BDE.
However, the substitution of groups like methyl and phenyl on
the amide nitrogen resulted in decrease in BDE by weakening
of the NC bond. Substituted methyl and phenyl decreased
BDE by 1090 kJ/mol on moving from primary to tertiary amides.
Therefore, the corroboration of the above-discussed differential
behaviour was done through BDE studies, considering that dissociation of C30N31 or N29C30 bond was also governed by the
stability of resultant fragments after bond cleavage. In the case of
ritonavir and its metabolites, M1, M4 and M5, the fragments that
were formed owing to the dissociation of C30N31 bond were
characterized by charge stabilization due to delocalization on
N29, C30 and O32 atomic centres. The difference in the charge
density between N29 and N31 centres was accordingly found
to be 0.206 e. However, in demethylated analogues, the difference in the charge density (N29 and N31) was greatly reduced to
0.074 e, making both the sites equally electron dense. Thus,
proton afnity values at N29 and N31 (Table 5) were decisive of
preferable protonation site, as discussed above.

wileyonlinelibrary.com/journal/jms

The present study provides extensive biotransformation-related information on ritonavir, along with extensive mass fragmentation
pathways of the drug and its all six CYP3A4 generated metabolites.
The metabolites formed were deacylated ritonavir (M1), N-dealkylated
ritonavir (M2), N-demethylated C-hydroxylated ritonavir (M3),
C-hydroxylated ritonavir (M4 and M5) and N-demethylated ritonavir
(M6) of which two were new. The unusual CO2 loss during mass fragmentation of ritonavir was analyzed for possibility of rearrangement
as well as ion-neutral complex mechanism through DFT. Although
the initial step for protonation appeared to favour pathway A
(rearrangement mechanism), favourable activation barriers found on
the reaction pathway of ionneutral mechanism suggest that this
pathway is credible for CO2 loss, as justied by DFT. In addition, the
difference in fragmentation pathways of ritonavir and its metabolites
(M1 to M6) could be ascribed to difference in the proton afnity
values on N29 and N31 centres, again justied by DFT modelling.
Acknowledgements
Authors want to acknowledge Bristol Myers Squibb, Bangalore,
India for providing fellowship and nancial assistance to one of
the authors (Shalu Jhajra). Authors also acknowledge Ninad
Varkhede (Biocon-Bristol Myers Squibb Research Center (BBRC,
Bangalore)), Ravi Shah (BBRC, Bangalore) and reviewers of this
manuscript for their valuable suggestions.

References
[1] B. Prasad, S. Singh. Identication of rat urinary metabolites of rifabutin
n
using LCMS and LCHR-MS. Eur. J. Pharm. Sci. 2010, 41, 173.
[2] B. Prasad, A. Garg, H. Takwani, S. Singh. Metabolite identication by
liquid chromatography-mass spectrometry. TrAC Trends Anal. Chem.
2011, 30, 360.
[3] B. Prasad, S. Singh. In vitro and in vivo investigation of metabolic fate
of rifampicin using an optimized sample preparation approach and
modern tools of liquid chromatographymass spectrometry.
J. Pharm. Biomed. Anal. 2009, 50, 475.
[4] H.-l. Lin, J. DAgostino, C. Kenaan, D. Calinski, P. F. Hollenberg. The effect of ritonavir on human CYP2B6 catalytic activity: Heme modication contributes to the mechanism-based inactivation of CYP2B6
and CYP3A4 by ritonavir. Drug Metab. Dispos. 2013, 41, 1813.
[5] R. G. Parr, W. Yang. Density-functional theory, Oxford university
press: New York, 1989.
[6] A. D. Becke. Density-functional thermochemistry. III. The role of exact
exchange. J. Chem. Phys. 1993, 98, 5648.
[7] C. Lee, W. Yang, R. G. Parr. Development of the Colle-Salvetti
correlation-energy formula into a functional of the electrondensity. Phys. Rev. B. 1988, 37, 785.
[8] M. J. Frisch, G. W. Trucks, H. B. Schlegel, G. E. Scuseria, M. A. Robb, J. R.
Cheeseman, G. Scalmani, V. Barone, B. Mennucci, G. A. Petersson, H.
Nakatsuji, M. Caricato, X. Li, H. P. Hratchian, A. F. Izmaylov, J. Bloino,
G. Zheng, J. L. Sonnenberg, M. Hada, M. Ehara, K. Toyota, R. Fukuda,
J. Hasegawa, M. Ishida, T. Nakajima, Y. Honda, O. Kitao, H. Nakai,
T. Vreven, J. A. Montgomery Jr., J. E. Peralta, F. Ogliaro, M. Bearpark,
J. J. Heyd, E. Brothers, K. N. Kudin, V. N. Staroverov, R. Kobayashi,
J. Normand, K. Raghavachari, A. Rendell, J. C. Burant, S. S. Iyengar,
J. Tomasi, M. Cossi, N. Rega, J. M. Millam, M. Klene, J. E. Knox,
J. B. Cross, V. Bakken, C. Adamo, J. Jaramillo, R. Gomperts, R. E. Stratmann,
O. Yazyev, A. J. Austin, R. Cammi, C. Pomelli, J. W. Ochterski, R. L. Martin,
K. Morokuma, V. G. Zakrzewski, G. A. Voth, P. Salvador, J. J. Dannenberg,
S. Dapprich, A. D. Daniels, . Farkas, J. B. Foresman, J. V. Ortiz,
J. Cioslowski, D. J. Fox. Gaussian 09: EM64L-G09RevB.01, Gaussian Inc.:
Wallingford CT, 2010.
[9] A. P. Scott, L. Radom. Harmonic vibrational frequencies: An evaluation of Hartree-Fock, Mller-Plesset, quadratic conguration interaction, density functional theory, and semiempirical scale factors.
J. Phys. Chem. 1996, 100, 16502.

Copyright 2014 John Wiley & Sons, Ltd.

J. Mass Spectrom. 2014, 49, 452467

Explanation for unanticipated loss of CO2


[10] A. E. Reed, L. A. Curtiss, F. Weinhold. Intermolecular interactions
from a natural bond orbital, donor-acceptor viewpoint. Chem.
Rev. 1988, 88, 899.
[11] A. E. Reed, R. B. Weinstock, F. Weinhold. Natural population analysis.
J. Chem. Phys. 1985, 83, 735.
[12] D. Yu, Q. Feng, Z. Yu. An unusual rearrangement process of neutral
loss of CO2 in carbamates by electrospray ionization-multistage
mass spectrometry. J. Chin. Mass Spectrom. Soc. 2005, 26, 6.
[13] M. Sun, W. Dai, D. Q. Liu. Fragmentation of aromatic sulfonamides in
electrospray ionization mass spectrometry: Elimination of SO2 via
rearrangement. J. Mass Spectrom. 2008, 43, 383.
[14] K. Klagkou, F. Pullen, M. Harrison, A. Organ, A. Firth, G. J. Langley.
Fragmentation pathways of sulphonamides under electrospray tandem mass spectrometric conditions. Rapid Commun. Mass Spectrom.
2003, 17, 2373.
[15] N. Hu, P. Liu, K. Jiang, Y. Zhou, Y. Pan. Mechanism study of SO2 elimination from sulfonamides by negative electrospray ionization mass
spectrometry. Rapid Commun. Mass Spectrom. 2008, 22, 2715.
[16] S. Crotti, L. Stella, I. Munari, F. Massaccesi, L. Cotarca, M. Forcato, P.
Traldi. Claisen rearrangement induced by low-energy collision of
ESI-generated, protonated benzyloxy indoles. J. Mass Spectrom.
2007, 42, 1562.
[17] F. Wang. Collision-induced gas-phase Smiles rearrangement in
phenoxy-N-phenylacetamide derivatives. Rapid Commun. Mass
Spectrom. 2006, 20, 1820.
[18] Y. Zhou, Y. Pan, X. Cao, J. Wu, K. Jiang. Gas-phase smiles rearrangement
reactions of deprotonated 2-(4, 6-dimethoxypyrimidin-2-ylsulfanyl)-Nphenylbenzamide and its derivatives in electrospray ionization mass
spectrometry. J. Am. Soc. Mass Spectrom. 2007, 18, 1813.
[19] X. Chen, J. Xing, D. Zhong. Rearrangement process occurring in
the fragmentation of adefovir derivatives. J. Mass Spectrom.
2004, 39, 145.
[20] Z. X. Zhao, H. Y. Wang, C. Xu, Y. L. Guo. Gas-phase synthesis of
hydrodiphenylcyclopropenylium via nonclassical Favorskii rearrangement
from alkali-cationized , -dibromodibenzyl ketone. Rapid Commun. Mass
Spectrom. 2010, 24, 2665.

[21] Y. Chen, Y. Yin, X. Sun, X. Liu, H. Wang, Y. Zhao. A novel


rearrangement in electrospray ionization multistage tandem mass
spectrometry of amino acid ester cyclohexyl phosphoramidates of
AZT. J. Mass Spectrom. 2005, 40, 636.
[22] J. Bialecki, J. Ruzicka, A. B. Attygalle. An unprecedented
rearrangement in collision-induced mass spectrometric fragmentation of protonated benzylamines. J. Mass Spectrom. 2006, 41, 1195.
[23] A. Alex, S. Harvey, T. Parsons, F. S. Pullen, P. Wright, J. A. Riley.
Can density functional theory (DFT) be used as an aid to a
deeper understanding of tandem mass spectrometric
fragmentation pathways? Rapid Commun. Mass Spectrom.
2009, 23, 2619.
[24] Z. Xiang. Mechanism of SO2 elimination from the aromatic
sulfonamide anions: A theoretical study. Comput. Theor. Chem.
2012, 991, 74.
[25] A. K. Shukla, J. H. Futrell. Tandem mass spectrometry: Dissociation of
ions by collisional activation. J. Mass Spectrom. 2000, 35, 1069.
[26] E. De Hoffmann. Tandem mass spectrometry: A primer. J. Mass
Spectrom. 1996, 31, 129.
[27] V. Gabelica, E.D. Pauw. Internal energy and fragmentation of
ions produced in electrospray sources. Mass Spectrom. Rev.
2005, 24, 566.
[28] C. Collette, L. Drahos, E. D. Pauw, K. Vkey. Comparison of the internal energy distributions of ions produced by different electrospray
sources. Rapid Commun. Mass Spectrom. 1998, 12, 1673.
[29] M. Alcam, O. Mo, M. Yez. Computational chemistry: A useful
(sometimes mandatory) tool in mass spectrometry studies. Mass
Spectrom. Rev. 2001, 20, 195.
[30] I. I. Marochkin, O. V. Dorofeeva. Amide bond dissociation enthalpies:
Effect of substitution on NAC bond strength. Comput. Theor. Chem.
2012, 991, 182.

Supporting information
Additional supporting information may be found in the online
version of this article at the publishers web site.

467

J. Mass Spectrom. 2014, 49, 452467

Copyright 2014 John Wiley & Sons, Ltd.

wileyonlinelibrary.com/journal/jms

Anda mungkin juga menyukai