Phototransformation of the drug rivastigmine: Photoinduced cleavage of benzyl-nitrogen sigma bond

Article abstract of DOI:10.1016/j.jphotochem.2012.04.015

The photochemical behaviour of rivastigmine has been investigated under UV-B irradiation by HPLC and NMR. Kinetic data have been obtained irradiating aqueous solutions (5 × 10^-5 M). For mechanistic purposes and products studies irradiation has

Full text of DOI:10.1016/j.jphotochem.2012.04.015

Journal of Photochemistry and Photobiology A: Chemistry 239 (2012) 1–6
Contents lists available at SciVerse ScienceDirect
Journal of Photochemistry and Photobiology A:
Chemistry
journal homepage: www.elsevier.com/locate/jphotochem
Phototransformation of the drug rivastigmine: Photoinduced cleavage of
benzyl-nitrogen sigma bond
Fabio Temussia, Monica Passanantia, Lucio Previteraa, Maria Rosaria Iescea,∗, Marcello Briganteb,c
,
Gilles Mailhotb,c, Marina DellaGrecaa
a UdR Napoli 4 INCA, IC-REACH, Dipartimento di Scienze Chimiche, Università di Napoli Federico II, Complesso Universitario Monte S. Angelo, Via Cinthia 4, I-80126 Napoli, Italy
b Clermont Université, Université Blaise Pascal, Institut de Chimie de Clermont-Ferrand (ICCF)-ENSCCF, BP 10448, F-63000 Clermont-Ferrand, France
c CNRS, UMR 6296, ICCF, BP 80026, F-63171 Aubière, France
a r t i c l e i n f o
a b s t r a c t
Article history:
The photochemical behaviour of rivastigmine has been investigated under UV-B irradiation by HPLC and
NMR. Kinetic data have been obtained irradiating aqueous solutions (5 × 10−5 M). For mechanistic pur-
poses and products studies irradiation has also been carried out on concentrated solutions (1 × 10−3 M) in
water/acetonitrile, acetonitrile, methanol, methanol/acetonitrile, in the presence and absence of oxygen.
Photoproducts, isolated by chromatographic processes, have been identified by spectroscopic means.
Photochemical cleavage of the benzyl-nitrogen bond occurs and leads to radical-derived and, mainly,
ion-derived products. Mechanistic pathways have been hypothesised on the basis of both steady-state
irradiation and laser flash photolysis experiments.
Received 16 March 2012
Received in revised form 17 April 2012
Accepted 21 April 2012
Available online 27 April 2012
Keywords:
Drug
Benzylic amines
Photooxidation
Electron transfer
Laser flash photolysis
© 2012 Elsevier B.V. All rights reserved.
1. Introduction
at 254 nm. Hence, extension of photochemical investigation to
bioactive compounds is useful to evaluate their photobiologi-
The behaviour of a substance under irradiation is a concern of
increasing attention in various scientific fields from chemistry to
biochemistry, from engineering to environmental sciences. Pho-
tochemical transformations may be usefully employed in organic
synthesis, often avoiding drastic conditions [1], or may represent
an environmental problem [2]. Indeed a substance introduced into
the environment can convert to products that may be persistent
or noxious, sometimes even more than the parent compound.
Sunlight combined with oxygen, and/or natural or synthetic sen-
sitizers, is one of the main factors for these transformations. In
this context, our researches are addressed to investigate the pho-
tochemistry of bioactive compounds, especially drugs [3] that
represent emerging environmental contaminants [4,5]. The inter-
est for drugs is also due to the photobiological risk associated
with some of them [6]. Given the complexity and heterogeneity
of this broad class of compounds, it is often difficult to predict
or rationalize the photochemical behaviour of drugs. Moreover
the knowledge of factors (substituents, solvents, light) that affect
the photochemistry of organic compounds are generally limited
to relatively simple molecules, in organic solvents and irradiation
cal risk and also to confirm known photochemical processes
or, sometime, to evidence new photoinduced routes [3,7]. Here
we analyse the photochemical behaviour of rivastigmine, (S)-3-
[(1-dimethylamino)ethyl]phenyl N-ethyl-N-methyl carbamate, an
inhibitor of acetylcholinesterase that is used for the treatment of
mild to moderate Alzheimer’s disease in adults [8]. Introduced
in EU in 1998 and in USA in 2000, rivastigmine is relatively a
new drug, and attention has been generally focused to analytical
methods for its detection [9] or determination of optical purity
[10].
Previous studies have examined the behaviour of rivastigmine
hydrogen tartrate bulk drug under forced degradation conditions
[0.5 N hydrochloric acid, 0.5 N sodium hydroxide, 3% hydrogen
peroxide, heat (60 ◦C), and UV light (254 nm)] [11]. Rivastigmine
degradation was observed only under basic conditions yielding,
after 48 h, trace amount of (S)-3-(1-dimethylaminoethyl) phenol,
which is also the major human metabolite known as NAP 226-90
[11].
The aim of this work is to explore the behaviour of free
rivastigmine with particular attention to mimic environmental
conditions (water, ꢀ > 300 nm). Various steady state irradia-
tion and laser flash photolysis experiments have been carried
out in order to understand the mechanisms leading to its
photodegradation.
∗
Corresponding author. Tel.: +39 081 674334; fax: +39 081 674393.
E-mail address: iesce@unina.it (M.R. Iesce).
1010-6030/$ – see front matter © 2012 Elsevier B.V. All rights reserved.
http://dx.doi.org/10.1016/j.jphotochem.2012.04.015

2
F. Temussi et al. / Journal of Photochemistry and Photobiology A: Chemistry 239 (2012) 1–6
(calibrated using a DH-2000-CAL Deuterium Tungsten Halogen ref-
erence lamp) and normalized to the actinometry results using
PNA/pyridine actinometer [12]. The incident photon flux in solution
was 6.10 × 1019 photons m−2 s−1
.
2.3.1.1. Phototransformation kinetics and quantum yield calculation.
Rivastigmine solutions (5 × 10−5 M) were irradiated as above. At
fixed time intervals an aliquot (200 ␮L) was withdrawn and ana-
lysed by HPLC. The time evolution of rivastigmine could be fitted
with a pseudo-first order equation C0 = Ct exp(−kt) where C0 was
the initial rivastigmine concentration, Ct the concentration at time
t and k the pseudo-first order degradation rate constant.
The kinetic constant calculated is k = (1.08 0.02) × 10−6 s−1 and
the half-life time is t1/2 = 177.5 h. Assuming rivastigmine as the only
absorbing species present in water, the polychromatic quantum
yield degradation (ꢁ) of the drug was calculated, in the overlap
range 288–320 nm with lamp emission (see Fig. 1), as follows:
Fig. 1. Molar absorption coefficient of rivastigmine (1) aqueous solution (solid line)
and emission spectrum of radiation reaching the solution inside the pyrex reactor
(dashed line).
Rriv
ꢁ =
Ia
2. Materials and methods
where Rriv is the rivastigmine degradation rate (M s−1) and Ia is the
absorbed photon flux per unit of surface and unit of time. The latter
2.1. Chemicals
was calculated from:
ꢀ
ꢀ
2
Rivastigmine tartrate was purchased by Kemprotec, ana-
lytical standard grade (99%). p-Nitroanisole (PNA) and 9,10-
anthraquinone were supplied by Aldrich and used without further
purification. Milli-Q water (Millipore) was used to prepare aqueous
solutions. Acetonitrile and methanol were of HPLC grade (Sigma-
Aldrich). Rivastigmine (1, Fig. 1) was obtained by dissolving the salt
in NaHCO3 (sat.) and extracting with ethyl acetate.
Ia
=
I0(ꢀ)(1 − 10−ε(ꢀ)d[Riv]
)
ꢀ
1
where I0 is the incident photon flux, ε the molar absorption coeffi-
cient of rivastigmine, d the optical path length inside the cells and
[Riv] the initial rivastigmine concentration.
Polychromatic quantum yield ꢁriv is 2.61 × 10−3
.
2.3.2. Sunlight irradiation of rivastigmine
2.2. Apparatus
A solution of rivastigmine in water (1 × 10−5 M) was exposed
to sunlight in a closed pyrex tube in September–October 2010 in
Naples and analysed by HPLC after 6, 12 and 20 days.
HPLC experiments were carried out on an Agilent 1100 HPLC
system, equipped with an UV detector, using a RP-18 column (Gem-
ini, 5 ␮m, 110 A, 250 mm × 4.6 mm). The flow was set to 0.8 mL/min.
The detector lamp was set at 254 nm. The column was equilibrated
with a mixture of A (H2O containing 1% formic acid)–B (acetoni-
trile) 9:1 (v/v). The run followed this programme: isocratic B 10%
from the start to 7 min, an increase of B up to 90% from 7 to 12 min,
isocratic B 90% for 11 min, return to B 10% in 2 min.
2.3.3. Irradiation experiments for mechanistic and preparative
purposes
Irradiations were performed in quartz tubes (20 cm × 1 cm,
25 mL) by means of a Helios Italquartz photoreactor (system II)
equipped with six 15 W UV-B lamps with a maximal output at
ca. 310 nm (1 mW cm−2) or four 15 W UV-A lamps with a maximal
output at ca. 366 nm (1 mW cm−2).
Analytical and preparative TLC were made on Kieselgel 60 F254
plates with 0.2 mm and 0.5 or 1 mm layer thickness, respectively
(Merck).
2.3.3.1. Irradiation in different solvents. Four 1 × 10−3 M solu-
tions of rivastigmine in H2O/CH3CN (9:1, v/v), CH3CN, CH3OH,
CH3OH/CH3CN (9:1, v/v) were prepared by dissolving 5 mg in
20 mL. Each solution was irradiated in open quartz tubes with UV-B
lamps and analysed by HPLC and 1H NMR at selected times. Similar
irradiation experiments were performed as above in closed quartz
tubes after saturating with argon.
A further series of solutions treated as above were irradiated for
1 h. Then, the solvents were evaporated and each residue was care-
fully analysed by 1H NMR. The photoproducts in each mixture were
identified by comparing the NMR signals with those of standard
compounds, which were isolated and characterized by performing
preparative photochemical experiments (see Section 2.3.3.3, Fig. 2,
Table 2 and Supporting Content).
NMR spectra were recorded on a Varian Inova-500 instrument
operating at 499.6 and 125.6 MHz for 1H and 13C, respectively, and
referenced with deuterated solvents (CDCl3).
Electronic impact mass spectra (EI-MS) were obtained with a
GC–MS QP5050A (Shimadzu) equipped with a 70 eV EI detector.
IR spectra were recorded on a Jasco FT/IR-430 instrument
equipped with single reflection ATR using CHCl3 as solvent.
UV–Vis spectra were recorded with a Varian Cary 300 UV–Vis
spectrophotometer.
A Varian Cary Eclipse fluorescence spectrofluorimeter was used,
adopting a 5 nm bandpass on both excitation and emission.
2.3. Steady-state irradiation experiments
All the experiments were performed in triplicate.
Samples of all solutions were kept in the dark and analysed as
above showing no degradation after 48 h.
2.3.1. Irradiation in water
Rivastigmine water solutions (5 × 10−5 M) were irradiated
in a photochemical setup (system I). System I consisted of a
thermostated pyrex cylindrical reactor (total volume 100 mL) sur-
rounded by six Duke GL20E 20 W lamps. The temperature was
held at 293 2 K. Emission spectrum reaching solution (Fig. 1) was
measured with an Ocean Optics SD 2000 CCD spectrophotometer
2.3.3.2. Photoproducts isolation. A solution of rivastigmine (32 mg)
was dissolved in 128 mL of Milli-Q H2O/CH3CN (9:1, v/v,
1 × 10−3 M) and irradiated in open tubes with UV-B lamps for

F. Temussi et al. / Journal of Photochemistry and Photobiology A: Chemistry 239 (2012) 1–6
3
5
5
9
O
O
O
11
11
OH
7
8
7
8
N
O
N
O
1
3
1
3
10
N
O
N
O
O
10
12
12
3
1
13
13
2
O
O
O
OCH3
N
O
N
N
O
6
4
5
Fig. 2. Rivastigmine and its photoproducts.
5 h. Then, the solvents were evaporated under vacuum and the
residue (20 mg) was separated by preparative TLC. Elution with
CH2Cl2/AcOEt (8:2, v/v) gave a mixture of 4 and 5 (<1 mg), ketone 2
(4 mg), alcohol 3 (7 mg), an intractable material (5 mg) and rivastig-
mine (2 mg) at decreasing Rfs (Fig. 2).
Rivastigmine (15 mg) dissolved in 60 mL of acetonitrile
(1 × 10−3 M) was exposed in open tubes to UV-B lamps for 1 h. After
evaporation of the solvent, preparative TLC of the residue as above
gave ketone 2 (10 mg) and the drug (1 mg).
2.4. Laser flash photolysis studies
For 266/355 nm excitation, experiments were carried out using
the fourth/third harmonic of a Quanta Ray GCR 130-01 Nd:YAG
laser system instrument, used in a right-angle geometry with
respect to the monitoring light beam. The single pulses were ca. 9 ns
in duration, with energy of ∼60 mJ/pulse. Individual cuvette sam-
ples (V = 3 mL) were used for a maximum of four consecutive laser
shots. The transient absorbance at the pre-selected wavelength
was monitored by a detection system consisting of a pulsed xenon
lamp (150 W), monochromator and a photomultiplier (1P28). A
spectrometer control unit was used for synchronising the pulsed
light source and programmable shutters with the laser output. The
signal from the photomultiplier was digitised by a programmable
digital oscilloscope (HP54522A). A 32 bits RISC-processor kinetic
spectrometer workstation was used to analyse the digitised signal.
The pseudo-first order decay constants of transient species were
obtained by fitting the absorbance vs. time data with single or dou-
ble exponential equations. The error was calculated as 3ꢃ from the
fit of the experimental data. All experiments were performed at
ambient temperature (295 2 K) in aerated and argon-saturated
solutions.
A
solution of rivastigmine (15 mg) in methanol (60 mL,
1 × 10−3 M) was saturated with argon and irradiated in closed tubes
to UV-B lamps for 5 h. After evaporation of the solvent, preparative
TLC of the residue as above gave a fraction (1 mg) composed of 4
and 5 (in ca. 1:1 molar ratio), ether 6 (3 mg) and rivastigmine (9 mg)
(Fig. 2).
Products 2, 3 and 6 were fully characterized. Alkene 4 and alkane
5 were recovered as a 1:1 mixture in small amounts, therefore only
1H NMR data could be obtained and refer to this mixture.
2.3.3.3. Spectroscopic data of photoproducts 2–6. 3-Acetylphenyl N-
ethyl-N-methyl carbamate (2): UV ꢀmax (CH3OH) nm: 282 (log ε
3.2); EI-MS m/z 221, 177, 86, 58; ꢂmax (CHCl3) 1717, 1693, 1607,
1406 and 1168 cm−1 1H and 13C NMR, see Table 1.
;
3. Results and discussion
3-(1-Hydroxyethyl)phenyl N-ethyl-N-methyl carbamate (3):
UV ꢀmax (CH3OH) nm 252 (log ε 2.9); EI-MS m/z 223, 86, 77, 58;
Firstly, the drug stability was checked in aqueous solution in the
dark. It was stable after 48 h, even when tested in acidic (pH 4) and
alkaline (pH 9) solutions. These pH ranges are milder than those
previously employed [11] and are usually considered environmen-
tally relevant [13].
The absorption spectrum of rivastigmine (1) in water shows a
band centred at 260 nm with a shoulder at 270 nm and a weak tail
extending to ꢀ > 290 nm (Fig. 1). Overlap of this spectrum with UV-B
lamp emission spectrum is reported in Fig. 1.
ꢂmax (CHCl3) 3626, 3501, 1716, 1600, 1172 and 1014 cm−1 1H and
;
13C NMR, see Table 1.
3-Vinylphenyl N-ethyl-N-methyl carbamate (4) and 3-
ethylphenyl N-ethyl-N-methyl carbamate (5): (1:1 mixture);
selected 1H NMR signals for alkene 4: ıH (CDCl3) 7.30 (1H, t,
J = 7.8 Hz, H-5), 7.22 (1H, d, J = 7.8 Hz, H-6), 7.16 (1H, brs, H-2), 6.68
(1H, dd, J = 17.6, 11.0 Hz, H-7), 5.74 and 5.26 (2H, d, J = 17.6 Hz and
J = 11.0 Hz, H-8); selected 1H NMR signals for alkane 5: ı 7.25 (1H,
t, J = 8.0 Hz, H-5), 7.02 (1H, d, J = 8.0 Hz, H-6), 6.95 (1H, brs, H-2),
2.65 (2H, q, J = 7.0 Hz, H-7), 1.23 (3H, t, J = 8.0 Hz, H-8); overlapping
1H NMR signals: ıH 6.92 (2H, brs, H-4), 3.47 and 3.40 (4H, 2q,
J = 6.0 Hz, H-11), 3.06 and 2.99 (6H, 2 s, H-13), 1.22 and 1.17 (6H,
2t, J = 6.0 Hz, H-12).
Kinetics experiments were made irradiating a dilute water solu-
tion of rivastigmine (5 × 10−5 M) in a pyrex photoreactor with UV-B
light. Under these conditions the drug exhibited a half-life time of
177.5 h and a polychromatic quantum yield of 2.61 × 10−3
.
A 1 × 10−5 M solution of the drug in water was exposed to
sunlight in quartz tubes in Naples in September–October 2010.
Drug changes were monitored by HPLC. After 12 days rivastigmine
decreased by approximately 50% and converted to ketone 2 and,
in small amount, alcohol 3. Ketone 2 was identified by comparing
HPLC retention time with an authentic sample (see below) and by
characteristic 1H NMR signals identified in the mixture. Due to the
low amount, alcohol 3 was detected only by 1H NMR comparing its
signals with those of an authentic sample (see below).
3-(1-methoxyethyl)phenyl N-ethyl-N-methyl carbamate (6):
UV ꢀmax (CH3OH) nm: 261 (log ε 2.7); EI-MS m/z 237, 236, 206,
86, 58; ꢂmax (CHCl3) 1715, 1608, 1236 and 1168 cm−1 1H and 13
; C
NMR, see Table 1.
2.3.3.4. Irradiation experiments with sensitizers. Three solutions of
rivastigmine (5 mg) in 100 mL of H2O/CH3CN (9:1, v/v, 5 × 10−4 M)
in the presence of 0.3, 0.5, 0.7 equiv. of ketone 2 were irradiated in
closed quartz tubes with UV-B lamps and analysed by HPLC.
Two solutions of rivastigmine (5 mg) in 100 mL of H2O/CH3CN
(1:1, v/v, 5 × 10−4 M) with and without 0.5 equiv. of 9,10-
anthraquinone were irradiated in closed quartz tubes with UV-A
lamps (centred at ꢀ 360 nm) and analysed by HPLC.
With the purpose of isolating the photoproducts more concen-
trated drug solutions (1 × 10−3 M) were used. Since rivastigmine
(1) and its photoproducts are slightly soluble in water, acetoni-
trile was chosen as cosolvent to obtain clear solutions. In order to
accelerate the photoreaction, irradiation of the 1 × 10−3 M solution

4
F. Temussi et al. / Journal of Photochemistry and Photobiology A: Chemistry 239 (2012) 1–6
Table 1
1H and 13C NMR Data of compounds 1, 2, 3 and 6 (CDCl3, 500 MHz).a
No.b
1
2
3
6
ıH
ıC
ıH
ıc
ıH
ıC
ıH
ıC
1
2
3
4
5
6
7
8
145.4
120.7*
151.4
120.2*
128.8
124.2
65.5
19.9
43.0
138.4
121.6
151.8
126.7
129.4
125.0
197.2
19.9
147.4
118.8
151.6
120.7
129.2
122.1
70.0
145.1
119.5
151.7
120.8
129.2
122.8
79.3
7.05 brs
7.70 brs
7.14 brs
7.06 brs
7.00 brd (7.8)
7.27 t (7.8)
7.10 d (7.8)
3.26 q (6.7)
1.35 d (6.7)
2.19 s
7.34 brd (7.0)
7.47 t (8.0)
7.79 d (8.0)
7.01 brd (8.0)
7.32 t (8.5)
7.18 d (8.5)
4.88 q (6.4)
1.48 d (6.4)
7.04 brd (7.8)
7.33 t (7.8)
7.13 d (8.5)
4.29 q (6.5)
1.42 d (6.5)
2.60 s
25.0
23.8
9
10
11
12
13
OMe
154.5; 154.3
44.0
13.2; 12.4
34.1; 33.7
154.5; 154.3
44.1
13.2; 12.4
154.6; 154.4
44.1
13.2; 12.5
34.2; 33.8
154.6; 154.3
44.1
13.2; 12.5
34.2; 33.8
56.6
3.45 and 3.39 q (7.0)
1.22 and 1.17 t (7.0)
3.05 and 2.97 s
3.49 and 3.41 q (7.0)
1.27 and 1.20 t (7.0)
3.09 and 3.00 s
3.48 and 3.41 q (6.9)
1.24 and 1.19 t (6.9)
3.07 and 2.99 s
3.48 and 3.41 q (6.9)
1.25 and 1.19 t (6.9)
3.07 and 2.99 s
3.23 s
a
J values are in parentheses and are reported in Hz; chemical shifts are given in ppm; assignments were confirmed by COSY, HSQC, and HMBC experiments.
Numbering of carbons of compounds 1 and 2 is reported in Fig. 2; it has been used also for compounds 3–6.
b
(water/acetonitrile 9:1, v/v) was carried out in quartz tubes using
UV-B lamps. Drug changes were monitored by HPLC and 1H NMR
confirming the trend observed in dilute conditions. Products 2 and
3 were isolated by TLC and spectroscopically characterized (Fig. 2,
Table 1). Furthermore, chromatography gave a fraction consisting
of two products in very small amounts that were successively iden-
tified as alkene 4 and alkane 5 (Fig. 2).
With the aim of verifying the role of water and oxygen in the
drug photodegradation and for products studies, UV-B irradiation
experiments were carried out in different solvents (acetonitrile,
methanol, methanol/acetonitrile 9:1, v/v) and in the presence and
absence of oxygen. Table 2 reports product distribution after 1 h
irradiation. The product percentages were deduced by 1H NMR by
integration of characteristic signals.
All products isolated retain the carbamate moiety while they
present a new function instead of N(Me)2 function (Fig. 2). As shown
in Table 2, in all solvents ketone derivative 2 is the main photoprod-
uct, in water even under deareated conditions. Vinyl 4 and ethyl 5
derivatives are secondary products. Alcohol derivative 3 is present
only in water while in methanol ether 6 is found. These observa-
tions were confirmed by irradiation experiments at longer times by
both 1H NMR and HPLC analyses.
Product structures indicate that, despite the presence in the
drug of two photolabile functions, carbamate and benzylamine,
only the latter is involved in the formation of the photoproducts.
In particular, the primary photochemical event is the cleavage
of benzyl-nitrogen sigma bond. This breakdown pattern is also
observed in the mass spectrum of rivastigmine that exhibites a
base peak at m/z 206 [M−NMe2]+ [14]. As reported for various ben-
zylic compounds [15,16], the photochemical cleavage may occur
Fig. 3. Normalized fluorescence spectra (emission and excitation) of rivastigmine
aqueous solution at 295 2 K.
homolytically and/or heterolytically from the excited state S1. In
the case of rivastigmine, a value of 434 kJ mol−1 for the transi-
tion energy S0 → S1 was estimated from the intersection between
normalized emission and excitation spectra (Fig. 3). A plausi-
ble interpretation of the photochemical events from excited 1 is
depicted in Fig. 4. If the cleavage occurs homolytically a radical pair
7 is formed, and by oxygen trapping it leads to ketone 2 via per-
oxidic species 8. As minor route a disproportion occurs affording
alkene 4 and alkane 5. Radical pair 7 can also convert to ionic pair
Table 2
Product distribution after 1 h of UV-B irradiation of rivastigmine solutions (1 × 10−3 M) under different conditions.
Solvent
Condition
Yield (%)a
1
2
3
4
5
6
H2O/CH3CNb
CH3CN
In air
Under argon
In air
Under argon
In air
Under argon
In air
70
70
<5
90
<95
>95
<90
<80
20
20
95
–
5
–
10
10
–
–
–
–
–
–
Trace
Trace
–
Trace
Trace
–
–
–
–
–
5
5
CH3OH
Trace
Trace
Trace
10
Trace
Trace
Trace
10
Trace
<5
–
CH3OH/CH3CNc
10
–
Under argon
Trace
a
By 1H NMR.
H2O/CH3CN 9:1 (v/v).
CH3OH/CH3CN 9:1 (v/v).
b
c

F. Temussi et al. / Journal of Photochemistry and Photobiology A: Chemistry 239 (2012) 1–6
5
4 and 5
N
R =
O
O2
hν
1*
2
1
O
N
RO
O
RO
7
8
et
H2O
RO
MeOH
6
3
N
9
Fig. 4. Suggested phototransformation pathways.
e (hydr)
-H.
Fig. 7. Transient absorption spectra obtained after 266 nm excitation of 1.66 mM
rivastigmine in water (filled circles) and acetonitrile (empty circles) solutions at
295 2 K.
H2O
hν/Η2Ο
2
N
N
RO
RO
1
10
11
N
R =
O
3.1. Laser flash photolysis data
Fig. 5. Photoionization of rivastigmine in water.
In order to support our mechanistic hypotheses transient
absorption spectroscopy experiments were performed. Fig. 7 shows
the transient absorption spectra upon LFP excitation (266 nm) of
rivastigmine (1.66 mM) in pure water and acetonitrile immedi-
ately after the laser pulse (0.2 ␮s) at 295 K in open cuvettes. For
the absorption band centred at 310 nm, present in both cases, a
fast decay of a first species (probably due to the triplet state) and
formation and consequent decay of a long lived transient were
observed. Observing the trace at 310 nm (data not shown) it is
possible to see the parallel decay and growth from two unrelated
transients having different rate constants. From the fit of the sec-
ond species absorption vs. time decay we obtained a correlation
between the pseudo-first order decay and the water/acetonitrile
percentage ranging from (3.24 0.01) × 105 (100% water) up to
(6.18 0.06) × 106 (100% acetonitrile). On the basis of literature
data, the methyl-benzyl radicals typical absorption band is in the
range 315–320 nm [21,22]. Thus, it is reasonable to attribute this
signal to the formation of the resonance-stabilized benzylic radi-
cal 7. Isolation of typical radical-derived products as alkene 4 and
alkane 5 confirms this assignment. The second absorption band,
centred at 700 nm and detected only in water, could be attributed
to the presence of the solvated electron [23]. This supports the
proposed mechanism reported in Fig. 5 although the absence of
absorption band of radical cation 10. In an attempt to evidence the
formation of this intermediate we tried to selectively form it via
electron transfer from rivastigmine to the triplet excited chloranyl
at 355 nm [24] (data not shown). Under such conditions it was pos-
sible to discern only the band at 450 nm due to the chloranil radical
anion, but no transient relative to radical cation 10. It is probable
that this species is not detectable owing to its low absorption and/or
short life-time.
via an electron-transfer with the formation of the well stabilized
benzylic cation 9. This process is supported by the presence of ben-
zyl alcohol 3 in aqueous media and benzyl ether 6 in methanol
deriving from the nucleophilic solvent trapping of the charged
positive species [15]. The benzylic carbocation formation could be
partly favoured by the presence of the oxygen-substituent in meta-
position of the aromatic ring. In the photochemically excited state,
meta-alkoxy groups are known to be particularly efficient in stabi-
lization of a developing positive charge at a benzylic site [17].
Under aqueous conditions it is also possible that the drug under-
goes a photoionization leading to radical cation 10 that could decay
to 2 via reaction with oxygenated species (O2, superoxide anion)
or, in the absence of oxygen, via hydrolysis of iminium ion 11,
as reported in Fig. 5. This hypothesis accounts for the findings of
ketone 2 in deaerated aqueous conditions.
It is well known that the photochemical single electron-transfer
reaction (SET) of amines can be promoted by ketones [18,19].
Really, we observed that the photodegradation rate of rivastig-
mine increased (i) in time as well as ketone 2 was formed and (ii)
in the presence of increasing concentrations of 2, suggesting that
ketone 2 itself could act as a photosensitizer. The tendency of the
drug to give a radical cation via SET-promoted photochemical reac-
tion (Fig. 6) was confirmed using a well-known electron-transfer
sensitizer as 9,10-anthraquinone [20]. Irradiation at 360 nm of a
water/acetonitrile solution of rivastigmine in the presence of 0.5
equiv. of this compound converted the drug to ketone 2 within
20 min while no trace was detected in the blank experiment.
- H+
O2
hν
SET
N
2 + Me2NH
N
+ ket
1 + ket
RO
RO
10
N
ket = antraquinone or 2
R =
O
Fig. 6. Single electron-transfer (SET) of rivastigmine in the presence of aromatic ketones.

6
F. Temussi et al. / Journal of Photochemistry and Photobiology A: Chemistry 239 (2012) 1–6
4. Conclusion
[5] M. Isidori, A. Parrella, P. Pistillo, F. Temussi, Effects of ranitidine and its pho-
toderivatives in the aquatic environment, Environment International 35 (2009)
821–825.
The work gives further information on the photochemical
behaviour of benzylic amines confirming the tendency to undergo
␤-cleavage and to give an electrofugal group in the presence
of suitable N-substituents [18]. Indeed, the photodegradation of
rivastigmine in water involves the tertiary amino site and leads
mainly to ion-derived products characterized by departure of
Me2N-group. It is reported that amines bearing N-substituents that
are capable of stabilizing the formed aminium radicals have lower
oxidation potentials as compared to those with N-electron with-
drawing substituents as amides and carbamates [18]. This could
account for the unreactivity of the carbamic-N function of rivastig-
mine as compared with the reactivity of the benzylaminic function
[18].
The drug photodegradation is slow by direct irradiation
with UV-B light or sunlight, but it is promoted by ketones
or photoelectron-transfer sensitizers such as anthraquinone. It
is noteworthy that the main photoproduct is a ketone, and
it influences the photodegradation rate of the drug acting as
photosensitizer. This poses a problem in a possible indirect pho-
tosensitive effect of rivastigmine.
[6] S. Onoue, Y. Tsuda, Analytical studies on the prediction of photosen-
sitive/phototoxic potential of pharmaceutical substances, Pharmaceutical
Research 23 (2006) 156–164.
[7] F. Cermola, M. DellaGreca, M.R. Iesce, L. Previtera, M. Rubino, F. Temussi, Pho-
toreactivity of triazolopyridinones, including the drug trazodone, in aqueous
solution, Journal of Photochemistry and Photobiology A 206 (2009) 198–204.
[8] P. Bar-On, C.B. Millard, M. Harel, H. Dvir, A. Enz, J.L. Sussman, I. Silman, Kinetic
and structural studies on the interaction of cholinesterases with the anti-
Alzheimer drug rivastigmine, Biochemistry 41 (2002) 3555–3564.
[9] M.Y. Salem, A.M. El-Kosasy, M.G. El-Bardicy, M.K. Abd El-Rahman, Spec-
trophotometric and spectrodensitometric methods for the determination of
rivastigmine hydrogen tartrate in presence of its degradation product, Drug
Test Analysis 2 (2010) 225–233.
[10] Z.-Y. Wang, X.-X. Xu, Z.-D. Hu, J.-W. Kang, Sensitive method for enantiosepa-
ration of rivastigmine with highly sulfated cyclodextrin as chiral selector by
capillary electrophoresis, Chinese Journal of Chemistry 24 (2006) 1384–1387.
[11] B.M. Rao, M.K. Srinivasu, K.P. Kumar, N. Bhradwaj, R. Ravi, P.K. Mohakhud, G.O.
Reddy, P.R. Kumar, A stability indicating LC method for rivastigmine hydro-
gen tartrate, Journal of Pharmaceutical and Biomedical Analysis 37 (2005)
57–63.
[12] D. Dulin, T. Mill, Development and evaluation of sunlight actinometers, Envi-
ronmental Science and Technology 16 (1982) 815–820.
[13] T.W.J. Valenti, P. Perez-Hurtado, C.K. Chambliss, B.W. Brooks, Aquatic toxicity
of sertraline to Pimephales promelas at environmentally relevant surface water
pH, Environmental Toxicology and Chemistry 28 (2009) 2685–2694.
[14] F. Pommier, R. Frigola, Quantitative determination of rivastigmine and its
major metabolite in human plasma by liquid chromatography with atmo-
spheric pressure chemical ionization tandem mass spectrometry, Journal of
Chromatography B 784 (2003) 301–313.
Acknowledgement
[15] S.A. Fleming, J.A. Pincock, Photochemical cleavage reactions of benzyl-
heteroatom sigma bonds, in: V. Ramamurthy, K.S. Schanze (Eds.), Organic
Molecular Photochemistry, Marcel Dekker, New York, 1999, pp. 211–281.
[16] N.J. Turro, V. Ramamurthy, J.C. Scaiano, Modern Molecular Photochemistry of
Organic Molecules, University Science Books, Sausalito, FL, 2010.
[17] J.A. Pincock, P.J. Wedge, The photochemistry of methoxy-substituted benzyl
acetates and benzyl pivalates: homolytic vs heterolytic cleavage, Journal of
Organic Chemistry 59 (1994) 5587–5595.
NMR spectra have been performed at Centro di Metodologie
Chimico-Fisiche of University of Naples on a Varian Inova-500
instrument of Consortium INCA.
Appendix A. Supplementary data
[18] U.C. Yoon, Z. Su, P.S. Mariano, The dynamics and photochemical consequences
of aminium radical reactions, in: W. Horspool, F. Lenci (Eds.), CRC Handbook of
Photochemistry and Photobiology, CRC Press, Boca Raton, 2004, pp. 101–120.
[19] S.G. Cohen, N.M. Stein, Kinetics of photoproduction of benzophenone by
amines. Deamination and dealkylation of amines, Journal of the American
Chemical Society 93 (1971) 6542–6551.
Supplementary data associated with this article can be
found, in the online version, at http://dx.doi.org/10.1016/
j.jphotochem.2012.04.015.
[20] A. Galadi, M. Julliard, Photosensitized oxidative degradation of pesticides,
Chemosphere 33 (1996) 1–15.
References
[21] R.F.C. Claridge, H. Fischer, Self-termination and electronic spectra of substituted
benzyl radicals in solution, Journal of Physical Chemistry 87 (1983) 1960–1967.
[22] D. Vialaton, C. Richard, Direct photolyses of thiobencarb and ethiofencarb in
aqueous phase, Journal of Photochemistry and Photobiology A 136 (2000)
169–174.
[23] F.-Y. Jou, G.R. Freeman, Shapes of optical spectra of solvated electrons. Effect of
pressure, Journal of Physical Chemistry 81 (1977) 909–915.
[24] L.J. Johnston, N.P. Schepp, Reactivities of radical cations: characterization of
styrene radical cations and measurements of their reactivity toward nucle-
ophiles, Journal of the American Chemical Society 115 (1993) 6564–6571.
[1] A.G. Griesbeck, J. Mattay, Synthetic Organic Photochemistry, Marcel Dekker,
New York, 2005.
[2] P. Boule, Environmental Photochemistry, Springer, Berlin, 1999.
[3] F. Temussi, G. Bassolino, F. Cermola, M. DellaGreca, M.R. Iesce, S. Montanaro, L.
Previtera, M. Rubino, Investigation on the phototransformation of tadalafil in
aqueous media. 6-Epimerization vs. solvent trapping reaction, Photochemical
& Photobiological Sciences 9 (2010) 1139–1144.
[4] J. Fick, R.H. Lindberg, M. Tysklind, D.G.J. Larsson, Predicted critical environ-
mental concentrations for 500 pharmaceuticals, Regulatory Toxicology and
Pharmacology 58 (2010) 516–523.