Spin-orbital coupling effects on the photophysical properties and photocytotoxicity of merocyanine dyes

Article abstract of DOI:10.1039/a701638g

A series of merocyanine dyes possessing different heteroatoms in the terminal aromatic and barbiturate units has been synthesized The nature of the heteroatom does not affect the all-trans alignment of the central polyenic bridge, as demonstrated by ¹H NMR spectroscopy. There are pronounced changes, however, in the photophysical properties of the individual molecules due to structural, electronic and spin-orbital coupling effects induced by the heteroatoms. In particular, it has been possible to vary the quantum yield for formation of the lowest-energy triplet excited state by more than two orders of magnitude. Quantum yields and lifetimes for fluourescence and photoisomerization are also affected by the choice of heteroatom. Despite the major changes in triplet yield that take place upon insertion of heteroatoms displaying strong spin-orbital coupling effects (e.g., Se) the efficacy for light-induced cytotoxicity towards cancer cells shows but a modest dependence on the structure of the dye. There is a reasonably good correlation between triplet yield and the rate of light-induced cell killing as might be expected for a mechanism based on intermediate generation of singlet molecular oxygen. However, equally good correlations exist between cell-killing performance and both relative lipophilicity and the rate at which the dye penetrates into intact biological cells.

Full text of DOI:10.1039/a701638g

View Article Online / Journal Homepage / Table of Contents for this issue
Spin–orbital coupling e†ects on the photophysical properties and
photocytotoxicity of merocyanine dyes
Andrew C. Benniston,a* Kirpal S. Gulliyab and Anthony Harrimanc*
a Chemistry Department, University of Glasgow, Glasgow, UK G12 8QQ
b Baylor Research Institute, Baylor Medical Center, 3812 Elm St., Dallas, T exas 75226, USA
c L aboratoire de Photochimie, Ecole de Chimie, Polymeres et Materiaux, Universite L ouis
Pasteur, 1 rue Blaise Pascal, F-67008 Strasbourg, France
A series of merocyanine dyes possessing di†erent heteroatoms in the terminal aromatic and barbiturate units has been synthe-
sized. The nature of the heteroatom does not a†ect the all-trans alignment of the central polyenic bridge, as demonstrated by 1H
NMR spectroscopy. There are pronounced changes, however, in the photophysical properties of the individual molecules due to
structural, electronic and spinÈorbital coupling e†ects induced by the heteroatoms. In particular, it has been possible to vary the
quantum yield for formation of the lowest-energy triplet excited state by more than two orders of magnitude. Quantum yields
and lifetimes for Ñuourescence and photoisomerization are also a†ected by the choice of heteroatom. Despite the major changes
in triplet yield that take place upon insertion of heteroatoms displaying strong spinÈorbital coupling e†ects (e.g., Se) the efficacy
for light-induced cytotoxicity towards cancer cells shows but a modest dependence on the structure of the dye. There is a
reasonably good correlation between triplet yield and the rate of light-induced cell killing as might be expected for a mechanism
based on intermediate generation of singlet molecular oxygen. However, equally good correlations exist between cell-killing
performance and both relative lipophilicity and the rate at which the dye penetrates into intact biological cells.
Cyanine dyes, such as merocyanines and kryptocyanines, have
been studied extensively as potential photosensitizers for
photodynamic therapy1 and as radiation sensitizers2 for treat-
ment of solid tumours. Merocyanine 540 (1), in particular, has
attracted considerable attention3 and its photophysical
properties have been investigated by several research
groups.4h10 This dye is poorly soluble in water, where exten-
sive aggregation occurs, but readily soluble in alcohols, micel-
les and lipid membranes.11 Fluorescence is moderately intense
and, in common with most cyanine dyes, intersystem crossing
to the triplet manifold is extremely inefficient, at least in Ñuid
solution.12 The dominant photoprocess involves isomerization
from the Ðrst-excited singlet state, which possesses an all-trans
conformation,13 to form a long-lived cis-isomer. Because pho-
toisomerization involves structural disruption, the photo-
physical properties are dependent on the extent of frictional
forces between the excited state and the surrounding
medium.14 Furthermore, since the ground-state structure
involves contributions from polar resonance forms, the effi-
ciency for light-induced isomerization is also dependent on the
dielectric and polarization characteristics of the local environ-
ment.12
achieved without detracting from the preferential uptake of
these dyes into infected cells. One simple means for realizing
this goal could involve changing the nature of the hetero-
atoms resident in the aromatic and/or barbiturate residues
since these atoms are expected to inÑuence the overall spinÈ
orbital coupling properties of the chromophore. Indeed, it
might be anticipated that careful optimization of the spinÈ
orbital coupling characteristics, by way of the internal heavy-
atom e†ect,17 could provide
a convenient route for
enhancement of the triplet yield and we now report on the
photophysical properties of several such derivatives. It is
shown that this strategy induces signiÐcant changes in the rate
and yield of isomerization and in the quantum yield for for-
mation of the triplet excited state. In fact, by judicious selec-
tion of heteroatoms the triplet yield can be varied over several
orders of magnitude and the bis-selenium derivative is shown
to be a particularly potent photosensitizer. These compounds
have been studied also for their ability to destroy leukemia
cells under illumination but correlations between cell-killing
efficacy and triplet yield must be regarded with suspicion
because of the inherent changes in physicochemical properties
that accompany substitution.
The major attraction of 1 as a photo- or radio-sensitizer
concerns its known selective assimilation into certain types of
cancer cells; a property which presumably arises because of
the particular molecular architecture. Indeed, we recently
described the synthesis and properties of novel merocyanine
540 derivatives for which the relative rate of assimilation into
Daudi cells was controlled by the length of the alkyl substit-
uents appended to the thiobarbiturate subunit.15 In this
manner, the in situ solubility and localization of the dye could
be varied over a wide range, with the performance of the dye
as a phototoxin reÑecting the in situ concentration. Corre-
sponding studies have been reported by Gunther et al.16 in
which it was demonstrated that the efficacy for light-induced
destruction of leukemia cells and for photoinactivation of
viruses depends to some extent on the structure of the dye.
It is essential, however, that improvements in the photo-
sensitizing efficacy, and in particular the triplet yield, are
Experimental
Materials
Merocyanine 540 (1) was obtained from Eastman-Kodak and
was puriÐed by column chromatography on silica using
acetoneÈmethanol (98/2) as eluent; the absorption maximum
(j ) in dilute ethanol solution was located at 560 nm with a
max
molar absorption coefficient (e ) of 187 000 dm3 mol~1
max
cm~1. The basic methodology used for preparation of com-
pounds 2È5 has been reported previously.15 The selenium-
containing barbiturate required for synthesis of merocyanine 6
was prepared as before.18 Raw materials were purchased from
Aldrich Chemicals or Eastman-Kodak and were used as
received. Water was double-distilled and deionized with a
Millipore puriÐcation system. Ethanol (Aaper, Absolute grade)
J. Chem. Soc., Faraday T rans., 1997, 93(15), 2491È2501
2491

View Article Online
was used as received. All other solvents were spectroscopic
grade, used as received, or were redistilled under vacuum.
length with 50 individual laser shots being averaged and
analysed by computerized non-linear, least-squares iterative
procedures. Where appropriate, the laser intensity was attenu-
ated with crossed-polarizers. In several cases it was necessary
to restrict the intensity of the monitoring beam to a low level
in order to avoid photolysis of the photoisomer. Such attenu-
ation was achieved by placing appropriate neutral density
Ðlters before the sample cell. For all experiments, laser inten-
sities were calibrated by reference to zinc meso-
2: 1H NMR ([2H ]dimethyl sulfoxide): d 0.88È0.94 (t, 9H,
6
J \ 7 Hz), 1.15È1.19 (t, 6H, J \ 7.5 Hz), 1.27È1.34 (m, 4H),
1.51È1.68 (m, 4H), 1.99È2.12 (m, 2H), 2.58È2.63 (t, 2H, J \ 7
Hz), 3.06È3.17 (q, 6H, J \ 5 Hz), 4.29È4.40 (m, 4H), 4.58È4.70
(m, 2H), 6.89È6.90 (d, 1H, J \ 12 Hz), 7.42È7.50 (t, 1H, J \ 7.4
Hz), 7.58È7.60 (t, 1H, J \ 7.2 Hz), 7.75È7.90 (m, 3H), 7.92È7.9
(d, 1H, J \ 12 Hz), 8.06È8.16 (d, 1H, J \ 14 Hz). UVÈVIS
(C H OH): j \ 592 nm (e \ 126 000 dm3 mol~1 cm~1).
tetraphenylporphyrin in N -saturated benzene.19,20
2
5
max
max
2
3: 1H NMR ([2H ]dimethyl sulfoxide): d 0.88È0.94 (t, 9H,
Di†erential absorption coefficients for the triplet excited
6
J \ 7.2 Hz), 1.14È1.19 (t, 6H, J \ 7.2 Hz), 1.26È1.32 (m, 4H),
states were measured by the energy transfer method using
anthracene as donor. A solution of anthracene in ethanol was
prepared so as to possess an absorbance of 0.3 at 355 nm and
1.64 (m, 4H), 1.98È2.10 (m, 2H), 2.55È2.61 (t, 2H, J \ 6.7 Hz),
3.04È3.15 (q, 6H, J \ 5 Hz), 4.27È4.37 (m, 6H), 6.50È6.56 (d,
1H, J \ 14 Hz), 7.22È7.28 (t, 1H, J \ 7.6 Hz), 7.37È7.43 (t, 1H,
J \ 7.6 Hz), 7.49È7.52 (d, 1H, J \ 7.8 Hz), 7.58È7.61 (d, 1H,
J \ 7.2 Hz), 7.77È7.87 (t, 1H, J \ 12 Hz), 8.08È8.14 (d, 1H,
J \ 14 Hz), 8.18È8.29 (t, 1H, J \ 13 Hz). UVÈVIS (C H OH):
deoxygenated by purging with N . The triplet absorption
2
signal was monitored at the peak21 of 422 nm and the laser
intensity was attenuated until the decay proÐle could be Ðt to
a single-exponential process. Various concentrations of a
merocyanine dye were added to the solution and the lifetime
of triplet anthracene was measured at 422 nm. For each solu-
tion, the rate of formation of the triplet state of the merocya-
nine dye was monitored at 680 nm and compared with the
rate of decay of triplet anthracene. The concentration of
merocyanine dye was increased until the dye began to
compete with anthracene for absorption of the excitation laser
pulse; with at least Ðve di†erent dye concentrations being used
for each study. At moderately high dye concentrations, triplet
energy transfer was quantitative and the concentration of dye
triplet was determined by reference to triplet anthracene pos-
sessing a di†erential absorption coefficient22 of 52 000 dm3
mol~1 cm~1 at 422 nm. The averaged triplet absorption coef-
Ðcient, appropriate for each member of the series, was
42 000 ^ 4000 dm3 mol~1 cm~1.
2
5
j
\ 595 nm (e \ 120 000 dm3 mol~1 cm~1).
max
max
4: 1H NMR ([2H ]dimethyl sulfoxide): d 0.88È0.94 (t, 9H,
6
J \ 7.2 Hz), 1.14È1.20 (t, 6H, J \ 7.2 Hz), 1.26È1.35 (q, 4H,
J \ 7.3 Hz), 1.52È1.68 (m, 4H), 1.98È2.12 (m, 2H), 2.59È2.63 (t,
2H, J \ 7.2 Hz), 3.07È3.20 (q, 6H, J \ 7.3 Hz), 4.36 (m, 4H),
4.59 (m, 2H), 6.97È7.02 (d, 1H, J \ 13 Hz), 7.39È7.43 (t, 1H,
J \ 7.3 Hz), 7.78È7.92 (m, 5H), 8.12È8.15 (d, 1H, J \ 8 Hz).
UVÈVIS (C H OH): j \ 604 nm (e \ 115 000 dm3
2
5
max
max
mol~1 cm~1).
5: 1H NMR ([2H ]dimethyl sulfoxide): d 0.86È0.89 (t, 9H,
J \ 7.3 Hz), 1.14È1.18 (t, 6H, J \ 7.2 Hz), 1.23È1.29 (q, 4H,
6
J \ 7.3 Hz), 1.46È1.49 (m, 4H), 1.98È2.10 (m, 2H), 2.59È2.63 (t,
2H, J \ 7.2 Hz), 3.03È3.10 (q, 6H, J \ 7.3 Hz), 3.76È3.79 (t,
4H), 4.45 (m, 2H), 6.77 (d, 1H, J \ 13 Hz), 7.27È7.31 (t, 1H,
J \ 7.3 Hz), 7.46È7.51 (t, 1H), 7.69È7.72 (m, 3H), 7.88È7.92 (d,
1H, J \ 12.6 Hz), 8.01È8.03 (d, 1H, J \ 8 Hz). UVÈVIS
(C H OH): j \ 579 nm (e \ 162 000 dm3 mol~1 cm~1).
Di†erential absorption coefficients for the unstable isomers
formed under illumination were determined by the complete
2
5
max
max
6: 1H NMR ([2H ]dimethyl sulfoxide): d 0.88È0.94 (t, 9H,
conversion method in O -saturated ethanol. A dilute solution
6
2
J \ 7.2 Hz), 1.15È1.19 (t, 6H, J \ 7.2 Hz), 1.25È1.34 (q, 4H,
of the merocyanine dye was prepared so as to possess an
J \ 7.3 Hz), 1.52È1.68 (m, 4H), 1.98È2.10 (m, 2H), 2.59È2.63 (t,
2H, J \ 7.2 Hz), 3.06È3.17 (q, 6H, J \ 7.3 Hz), 4.36 (m, 4H),
4.59 (m, 2H), 6.97È7.00 (d, 1H, J \ 13 Hz), 7.37È7.43 (t, 1H,
J \ 7.3 Hz), 7.78È7.91 (m, 5H), 8.12È8.15 (d, 1H, J \ 8 Hz).
UVÈVIS (C H OH): j \ 602 nm (e \ 135 000 dm3
absorbance at the peak maximum of ca. 0.10 and was irradi-
ated with a single pulse delivered from a mode-locked
Nd : YAG pumped dye laser (40 mJ, 200 ps). The excitation
wavelength was matched to the peak of the absorption spec-
trum of the merocyanine dye. The monitoring beam was
passed through a high-radiance monochromator tuned to the
absorption peak of the merocyanine dye. The laser intensity
was varied with crossed-polarizers and the di†erential absorp-
tion signal due to the conversion of the ground-state species
into the unstable isomer was measured. The photomultiplier
tube was protected from scattered laser light with an optical
shutter that was opened 10 ns after the excitation pulse.
Approximately 100 individual laser shots were averaged and
the decay proÐles extrapolated to zero time. The laser inten-
sity was calibrated using zinc meso-tetraphenylporphyrin in
benzene as standard.19,20 For each compound, the size of the
bleaching signal increased with increasing laser intensity
before reaching an optimum value, after which the signal
began to decrease with increasing laser power. This latter
e†ect arises because the unstable isomer is photolabile and
absorbs the laser pulse. The experimental conditions were
optimized so as to minimize the signiÐcance of this e†ect and
the absorption coefficients were calculated from the maximum
bleaching signal.
2
5
max
max
mol~1 cm~1).
Methods
1H NMR spectra were recorded with a Bruker AM360
FT-NMR instrument with SiMe as internal standard.
4
Absorption spectra were recorded with a Hitachi U3210 spec-
trophotometer and Ñuorescence spectra were recorded with a
fully-corrected Perkin-Elmer LS5 spectroÑuorimeter. Solu-
tions for Ñuorescence studies were adjusted to possess an
absorbance of \0.05 at the excitation wavelength. Singlet
excited state lifetimes were measured with a Hamamatsu
single-shot streak camera following excitation by a 30 ps laser
pulse at 532 nm. A narrow bandpass Ðlter was used to isolate
Ñuorescence over the spectral range 620 ^ 10 nm and the laser
intensity was attenuated so as not to saturate the streak
camera. Approximately 100 individual laser shots were aver-
aged and the laser proÐle was deconvoluted from the experi-
mental decay record prior to data analysis. The time window
of the streak camera was from 50 ps to 10 ns.
The di†erential absorption spectrum for the Ðrst-excited
singlet state was recorded by picosecond laser Ñash photolysis
Flash photolysis studies were made with a frequency-
doubled, mode-locked Quantel YG402 Nd : YAG laser (pulse
width 30 ps; pulse energy 25 mJ). Solutions were adjusted to
possess an absorbance of ca. 0.1 at 532 nm and were purged
with N , O , or air according to the needs of the experiment.
techniques.
A frequency-doubled, mode-locked Nd : YAG
laser (25 mJ, 30 ps) was used as excitation source and a white
light continuum was used for the monitoring beam. The two
pulses were directly almost collinearly through the sample cell
and transmitted light was collected far from the sample cell in
order to minimize contamination by Ñuorescence. The moni-
toring pulse was split before entering the sample cell so as to
2
2
Transient di†erential absorption spectra were recorded point-
by-point with Ðve individual laser shots being averaged at
each wavelength. Kinetic studies were made at a Ðxed wave-
2492
J. Chem. Soc., Faraday T rans., 1997, V ol. 93

View Article Online
provide reference and excitation beams. Subtraction of these
two beams, after dispersion with a Spex spectrograph and
recording with a Princeton image-intensiÐed, dual-diode array
spectrometer, provided the required transient di†erential
absorption spectra. Approximately 300 individual spectra
were averaged at each delay time. The repetition rate of the
laser was kept to 1 Hz so as to avoid build-up of the unstable
isomer.
RMPI-1640 culture medium, re-suspended in growth medium
and incubated overnight at 37 ¡C. After 22 h of incubation,
cell viability was determined by the Trypan Blue exclusion
method.26 Comparative experiments were made with dye only
and light only controls. The cell-killing efficiency is reported
in terms of the log [reduction] factor, which is deÐned as the
logarithm of the number of surviving cells divided by the
number of cells exposed to irradiation. For each irradiation
period, three separate aliquots of photolysed solution were
used for cell counting and the cellular content of each aliquot
was counted Ðve times. Each experiment was repeated three
times.
Quantum yields for formation of singlet molecular oxygen
were measured by time-resolved, near-IR luminescence
spectroscopy23 using meso-tetrakis(3-hydroxyphenyl)porphy-
rin in [2H ]methanol as standard.24 Solutions were prepared
4
in [2H ]methanol to possess an absorbance of 0.25 at 532 nm
4
General considerations
and were saturated with O . The laser intensity was varied
over a wide range using crossed-polarizers and 50 individual
2
Merocyanine dyes such as 1 have potential biomedical appli-
cations with regard to both diagnosis and subsequent treat-
ment of leukemia cells.27 Before medical exploitation of these
dyes can be augmented, however, improvements must be
made in the selectivity with which the dyes recognize leukemia
cells and in the efficiency with which intracelluar dye acts as a
phototoxin to destroy the host cell under illumination. This
paper is concerned with the latter requisite and, in particular,
with an attempt to render the photophysical characteristics of
the dye more amenable for its use as a cytological photo-
sensitizer. Although the detailed mechanism by which merocy-
anine dyes kill leukemia cells under illumination with visible
light has not been adequately clariÐed27 it is likely that the
long-lived triplet excited state of the dye plays some role in the
overall process. Since the inherent triplet quantum yields for
merocyanine dyes structurally related to 1 are extremely low
(e.g., 0.003) it seems likely that major improvements in pho-
toactivity could be gained by employing dyes with higher
triplet yields but which display similar recognition for infected
cells.
As a logical starting point for the design of improved pho-
toreagents, therefore, it seems opportune to consider ways to
increase the triplet yield without signiÐcantly altering of the
molecular architecture. In previous work we attempted to
increase the triplet yield by curbing the competing photoiso-
merization process with the incorporation of bulky groups
into the molecular framework.15 A more reÐned strategy
might be to use the internal heavy-atom e†ect as a means by
which to increase spinÈorbital coupling within the excited
singlet state manifold.17 This could be done by substitution of
halogen atoms into the benzoxazole subunit or simply by
replacing the existing heteroatoms with heavier ones. We fol-
lowed the latter approach.
The normal synthetic route28 used to prepare 1, and its
close analogues, permits replacing heteroatoms in both ben-
zoxazole and thiobarbiturate units. Following this approach,
we replaced the oxygen atom in the benzoxazole residue
present in 1 with carbon (2), sulfur (3) and selenium (4) while
retaining all other structural attributes. Of these compounds,
preliminary screening showed the benzselenazole derivative 4
to be the most potent phototoxin towards leukemia cells and,
consequently, a second set of dyes was prepared so as to vary
the nature of the terminal heteroatom Y but while keeping the
benzselenazole residue. The usual sulfur atom was replaced
with oxygen (5) and selenium (6). The compounds were iso-
lated and puriÐed as sodium salts. Because of our concern
that these systematic alterations to the chemical composition
laser shots were averaged for each determination. The relative
yield of singlet oxygen was determined by computer extrapo-
lation of the decay proÐle to the centre of the laser pulse.
Measurements were made under conditions where the
observed luminescence yield exhibited a linear dependence on
incident laser intensity. Fresh aliquots of solution were used
for each determination to avoid problems due to photo-
degradation of the chromophore.
Di†usion coefficients were measured in 50% aqueous glyc-
erol using the Ñuorescence photofading method developed by
Axelrod et al.25 The solution was examined with a Zeiss Avio-
vert 35 inverted microscope and excitation was provided by a
frequency-doubled Nd : YAG laser (10 ns, 70 mJ) operated in
single-shot mode and focussed through the optics of the
microscope. Recovery of the initial Ñuorescence signal due to
Brownian motion of unbleached dye molecules was monitored
continuously at 610 nm. Partition coefficients (P) were deter-
mined by dissolution of 1 mg of dye in a mixture of chloro-
form (10 cm3) and 2% w/w aqueous solution of human serum
albumin (10 cm3). After sonication and standing for 24 h, a
further 10 cm3 of chloroform was added and the layers
separated. The concentration of dye in each layer was mea-
sured by absorption spectroscopy and the partition coefficient
was deÐned as the molar ratio of dye in water relative to that
dissolved in chloroform.
For Ñuorescence microscopy studies, a suspension of Daudi
cells (1 ] 106 cells cm~3) in growth medium was incubated for
precisely 30 min at 37 ¡C with a Ðxed concentration of
merocyanine dye in 50% aqueous ethanol containing 1%
serum protein. Parallel studies were made with dye concentra-
tions of 0, 10, 20 and 50 lmol dm~3. After incubation, the
cells were isolated by centrifugation, washed three times with
culture medium, and re-suspended in fresh growth medium.
The cells were counted and the suspension diluted to give a
Ðnal concentration of 1 ] 104 cells cm~3 in the growth
medium. Aliquots of these suspensions were placed on the
microscope slides, protected with a cover slip, and examined
by Ñuorescence spectroscopy. The excitation wavelength was
selected so as to coincide with the maximum of the corrected
excitation spectrum and the integrated Ñuorescence yield was
measured. The derived Ñuorescence yield, in arbitrary units,
was divided by the known Ñuorescence quantum yield of the
dye in absolute ethanol to give the relative intracellular dye
concentration. The various dyes were compared at each con-
centration of staining solution.
A suspension of Daudi cells (1 ] 106 cells cm~3) in growth
medium was mixed with merocyanine dye (5 lg cm~3). After
30 min incubation at 37 ¡C, the suspension was placed in
Falcon petri dishes (35 ] 10 mm) and exposed to light
delivered with an argon-ion pumped dye laser tuned to the
absorption peak of the dye. Output from the dye laser was
attenuated to give a constant light Ñux at each excitation
wavelength and was defocussed with an optical Ðbre so as to
provide a spherical illumination area of 2 cm diameter. After
irradiation for varying times, the cells were washed twice with
J. Chem. Soc., Faraday T rans., 1997, V ol. 93
2493

View Article Online
triplet pattern (d 7.71) and its coupling constant of 11.0 Hz.
For the alkenic protons of merocyanine dyes, it has been
established15 that coupling constants lying in the region of 12
Hz can be considered to be indicative of an all-trans ground-
state conformation. This is clearly the case for merocyanine 5.
Furthermore, close examination of the 1H NMR spectra
recorded at di†erent temperatures shows that only a single
isomer is present, at least within the observable limit (i.e., ca.
95%) of 1H NMR spectroscopy. The 1H NMR spectral simi-
larity exhibited by dyes 1È6 allows us to conclude that each of
the dyes studied here adopts an all-trans conformation in the
ground state.
Photophysical properties of 4 in ethanol solution
Fig. 1 1H NMR spectrum (360 MHz) of
5
recorded in
The major goal of this study is to evaluate how the nature of
the heteroatoms a†ects the photophysical properties of
merocyanine dyes with a view to identifying the most prom-
ising photosensitizer for subsequent use in photodynamic
therapy. In fact, the photophysical properties of 1 and 3 have
been described previously15,29 but data for the other com-
pounds have not been reported before. During the course of
this investigation it was observed that the photophysical
behaviour of dyes 1È6 followed a common pattern, but with
the derived values di†ering markedly between the various
compounds. The characteristic behaviour that follows from
light absorption by these dyes is detailed below for the benz-
selenazole derivative 4, while photophysical data collected for
all the compounds are given in Table 1. Measurements were
made in deoxygenated ethanol solution at 22 ¡C, unless stated
otherwise.
[2H ]dimethyl sulfoxide showing the atom labelling
6
might induce serious modiÐcations in structure or important
physicochemical properties additional studies were under-
taken with a view to elucidation of the ground-state structure
and of the lipophilicity.
Results
Ground-state conformation of merocyanine 5
On the basis of resonance-Raman2 and 1H NMR15 spectral
studies, the ground-state conformation of 1, and of the related
N-alkylated derivatives, is known to comprise an all-trans
arrangement of the central polyenic bridge. To ensure that
incorporation of di†erent heteroatoms into the aromatic or
barbiturate subunits does not perturb the ground-state con-
formation of the newly synthesized dyes detailed 1H NMR
The absorption spectrum recorded for 4 in dilute ethanol
solution (Fig. 2) shows an intense band centred at 604 nm for
which the peak maximum (j ) is also the (0,0) transition.
max
The molar absorption coefficient measured at the peak
studies were made in [2H ]dimethyl sulfoxide at 25 ¡C. We
maximum (e \ 115 000 ^ 7000 dm3 mol~1 cm~1) was used
6
max
describe here the results pertaining to merocyanine 5 but note
to calculate the oscillator strength ( f ) for the transition
that the other dyes showed identical behaviour with respect to
the magnetic properties of the polyene. 1H NMR spectral data
are provided for all the new dyes in the Experimental section
of this paper. Although some unfortunate spectral overlap
according to30
4.39 ] 10~9
P
f \
e(l) dl
(1)
n
occurred in [2H ]dimethyl sulfoxide, studies carried out in
6
where e(l) refers to the molar absorption coefficient at wave-
number l (in cm~1) and n is the refractive index of the sur-
[2H ]acetone or [2H ]chloroform were less successful owing
6
1
to the poor solubility of the dyes in these solvents.
Despite accidental equivalence of some aromatic and
alkenic proton resonances in 5, assignment of most signals
was possible using 1HÈ1H spin-decoupling techniques. The 1H
NMR spectrum (360 MHz) recorded for 5 in [2H ]dimethyl
6
sulfoxide at 25 ¡C is shown in Fig. 1, together with an atom
labelling scheme. Chemical shifts for the varous protons and
their coupling constants, where appropriate, are compiled in
the Experimental section. Although each of the four alkenic
protons could not be deÐnitively assigned, it was possible to
unambiguously identify H(8) and H(11) by way of their char-
acteristic chemical shift [d \ 6.77 H(8), 7.89 H(11)], doublet
pattern, and spin-decoupling spectra (not shown). The coup-
ling constants (J) were found to be 11.5 and 12.6 Hz, respec-
tively, for H(8) and H(11). On the basis of spin-decoupling
experiments, the alkenic proton H(9) could be identiÐed by its
Fig. 2 Absorption and Ñuorescence spectra recorded for merocya-
nine 4 in dilute ethanol solution. The excitation wavelength used for
the Ñuorescence spectrum was 540 nm.
Table 1 Photophysical properties measured for the various merocyanine dyes in dilute ethanol solution at 22 ¡C
compound
j
a/nm
log e
b
D c/cm~1
k d/108 s~1
f e
U f
q g/ps
U f
q h/ls
U f
q i/ms
U f
max
560
max
5.27
ss
F
F
s
T
T
I
I
*
1
2
3
4
5
6
600
550
600
600
650
650
4.8
2.6
2.4
2.2
4.4
3.3
0.88
0.59
0.56
0.52
0.80
0.61
0.16
0.18
0.18
0.18
0.19
0.16
410
660
780
700
430
480
0.003
0.003
0.008
0.100
0.020
0.380
800
820
670
135
180
95
0.40
0.32
0.31
0.07
0.28
0.03
6.8
1.0
11.5
2.4
3.1
3.5
n.d.j
n.d.
n.d.
0.020
0.005
0.045
592
595
604
579
602
5.10
5.08
5.06
5.21
5.13
a Absorption maximum, ^2 nm. b Molar absorption coefficient (dm3 mol~1 cm~1), ^8%. c Stokes shift, ^100 cm~1. d Radiative rate constant,
^15%. e Oscillator strength, ^12%. f ^10%. g ^25 ps. h ^25 ls. i ^0.1 ms. j n.d. \ not detected.
2494
J. Chem. Soc., Faraday T rans., 1997, V ol. 93

View Article Online
rounding solvent. The derived value ( f \ 0.52) indicates that
the transition is highly allowed. Consequently, Ñuorescence is
readily detected following excitation into the lowest-energy
absorption transition (Fig. 2), with the Ñuorescence excitation
spectrum giving a good match to the absorption proÐle over
the entire visible spectral range. The Stokes shift,31 measured
in dilute ethanol solution, was 600 ^ 50 cm~1, indicating the
absence of substantial geometric changes upon promotion to
the Ðrst-excited singlet state. The Ñuorescence quantum yield
transient spectrum shows pronounced absorption bands to
the blue and red of the bleaching region and stimulated emis-
sion centred around 630 nm. Decay of the stimulated emission
signal corresponds to a lifetime for the Ðrst-excited singlet
state of 680 ^ 40 ps.
Laser Ñash photolysis studies carried out in N -saturated
2
ethanol solution indicate that there are two transient species
present after decay of the Ðrst-excited singlet state [Fig. 3(b)].
The short-lived transient, which absorbs predominantly
around 700 nm, is assigned to the triplet excited state of the
dye on account of its reactivity towards molecular oxygen.
The long-lived transient, which absorbs weakly at 625 nm, did
not react with oxygen and, on the basis of related studies
made with various merocyanine dyes,34 this species is believed
to be a geometric isomer formed by rotation around one of
the central double bonds. Its di†erential absorption spectrum
[Fig. 3(b)] indicates that the isomer absorbs slightly to the red
of the ground state. The isomer was observed to be extremely
photolabile. Under low illumination conditions, the lifetime of
(U ), measured relative to Rhodamine 101 in ethanol,32 was
F
found to be 0.18 ^ 0.02 and the Ñuorescence lifetime (q ) was
S
measured to be 700 ^ 20 ps. The Ñuorescence decay proÐle
was strictly single exponential, regardless of excitation or
detection wavelengths, and the measured lifetime was unaf-
fected by the presence of dissolved oxygen. The radiative rate
constant was calculated from the StricklerÈBerg expression33
P
F(l) dl
P
e(l)
l
k \ 2.88 ] 10~9n2
dl
(2)
F
the isomer (q ) was (2.4 ^ 0.1) ms, with decay following Ðrst-
P
F(l)
l3
1
dl
order kinetics and restoring the pre-pulse baseline. The di†er-
ential molar Ñuorescence coefficient for the isomer at 625 nm
was derived to be 6750 ^ 1250 dm3 mol~1 cm~1, as measured
by complete conversion. Using this value, the quantum yield
where F(l) is the Ñuorescence intensity at wavenumber l. The
derived value (k \ 2.2 ] 108 s~1) was found to be in excellent
F
agreement with that estimated from the experiment data
for formation of the isomer (U ) was derived to be 0.07 ^ 0.02
I
(k \ U /q ).
following excitation with a 200 ps laser pulse. Isomerization
F
F S
Excitation of 4 in N -saturated ethanol with a 30 ps laser
occurs exclusively from the Ðrst-excited singlet state, as
demonstrated by time-resolution of its appearance at 625 nm
(which is an isosbestic point for the triplet state), and com-
petes with population of the triplet manifold.
2
pulse at 532 nm resulted in formation of the Ðrst-excited
singlet state of the dye. The di†erential absorption spectrum
recorded under such conditions is displayed in Fig. 3(a). The
lifetime of the singlet state measured by transient absorption
techniques was found to be 650 ^ 30 ps, compared with a
value of 700 ps measured by time-resolved Ñuorescence spec-
troscopy. In addition to bleaching of the ground state, the
The triplet state decayed with a lifetime (q ) of 135 ^ 15 ls
T
in deaerated ethanol and was quenched by O with a
2
bimolecular rate constant of (1.3 ^ 0.3) ] 109 dm3 mol~1 s~1.
The triplet state could also be populated via energy transfer
from the triplet anthracene in ethanol following laser excita-
tion at 355 nm. From these studies, the bimolecular rate con-
stant for triplet energy transfer was found to be (4 ^ 1) ] 109
dm3 mol~1 s~1 and the di†erential molar Ñuorescence coeffi-
cient for the triplet at the peak of 680 nm was 42 000 ^ 3000
dm3 mol~1 cm~1. The triplet state, as formed by energy trans-
fer, exhibited the same di†erential absorption spectrum as that
generated by direct excitation of 4 and decayed cleanly to the
pre-pulse baseline. On the basis of actinometric laser Ñash
photolysis studies made with a 200 ps laser pulse, the
quantum yield for population of the triplet excited state (U )
T
was determined to be 0.10 ^ 0.01.
It is assumed that the triplet arises directly from the Ðrst-
excited singlet state and that it retains the all-trans conforma-
tion. As seen also for the Ðrst-excited singlet state, the
lowest-energy excited triplet state exhibits a broad and some-
what structured absorption proÐle [Fig. 3(b)]. Since it can be
populated from triplet anthracene, the triplet state of 4 must
lie at relatively low energy. In fact, we can set the triplet
energy (E ) as being between that of anthracene (E \ 178 kJ
T
T
mol~1)35 and the energy of singlet molecular oxygen, O (1* ),
2
g
(E \ 94 kJ mol~1).36 It is interesting to note that this crude
assignment places the triplet energy well above the energy
T
content of the cis-isomer (*H \ 75 ^ 5 kJ mol~1)15,37
although photoisomerization does not take place from the
triplet manifold. Attempts to employ the external heavy-atom
e†ect to induce spin-forbidden absorption17 to the triplet
manifold were unsuccessful. Similarly, it was not possible to
detect low-temperature phosphorescence from 4, even in the
presence of excess iodomethane.
Fig. 3 (a) Di†erential transient absorption spectrum recorded imme-
diately after excitation of 4 in deoxygenated ethanol with a 30 ps laser
pulse at 532 nm. The spectrum shows simulated emission and absorp-
tion bands due to the Ðrst-excited singlet state, in addition to blea-
ching of the ground state. (b) Di†erential transient absorption spectra
recorded 25 ns (…) and 1 ms (>) after excitation of 4 in deoxygenated
ethanol solution with a 30 ps laser pulse at 532 nm. The longer-lived
species is believed to be the cis-isomer formed by rotation around one
of the polyenic double bonds while the shorter-lived species is the
triplet excited state.
Quenching of the triplet state by O in [2H ]methanol
2
4
solution resulted in formation of singlet molecular oxygen,
O (1* ), as detected by time-resolved luminescence spectros-
2
g
copy. The quantum yield for formation of singlet oxygen (U )
*
was found to be 0.020 ^ 0.005 in O -saturated solution. The
2
product attacks ground state dye,38 as evidenced by complete
J. Chem. Soc., Faraday T rans., 1997, V ol. 93
2495

View Article Online
and that no new deactivation routes are introduced. This
approach facilitates isolation of the e†ects induced by incorp-
oration of speciÐc heteroatoms in terms of the simpliÐed
scheme given in Scheme 1. Here, k and k refer, respectively,
F
IC
to rate constants for radiative and nonradiative return of the
excited singlet state to the ground state, k is the rate con-
ISC
stant for intersystem crossing to the triplet manifold, while k
TC
and k refer, respectively, to rate constants for formation and
CT
decay of the unstable isomer. The term k refers to non-
T
radiative deactivation of the lowest-energy triplet state.
Light-induced cytotoxicity studies
Merocyanine dyes 1È6 provide a set of compounds of compa-
rable structure but with disparate triplet yields in ethanol
solution. It is interesting, therefore, to investigate what e†ect
the nature of the heteroatom has on the ability of the dyes to
induce photo-inactivation of leukemia cells. Some of the com-
pounds described here have been studied earlier with respect
to their ability to kill certain biological cells15 and/or
viruses16 but such experiments need to be performed under
identical conditions if critical comparison of individual dyes is
to be attempted. Furthermore, it is important that other
factors, such as in vivo concentrations, internal mobilities, and
cellular distribution patterns, are assessed since these can
make major contributions to the overall efficacy for light-
induced cytotoxicity.
Di†usion coefficients (D) were measured for each dye in
aqueous glycerol using laser-induced photofading micro-
scopy.25 The derived values fall within a narrow range (Table
2) suggesting that there were no major changes in the internal
mobility of the compounds. In marked contrast, the various
dyes exhibit surprisingly di†erent lipophilic character, as
quantiÐed by the partition coefficient (P) measured for equili-
bration of the dyes between human serum albumin and
Scheme 1
bleaching of the dye which occurs under visible light illumi-
nation in the presence of O . The bimolecular rate constant
2
for interaction between ground-state dye and O (1* ) was
2
g
measured by time-resolved, near-IR luminscence spectroscopy
to be (1.0 ^ 0.3) ] 109 dm3 mol~1 s~1. This overall process
includes contributions from both physical and chemical quen-
ching e†ects.
Comparison of the photophysical properties of merocyanine
dyes 1–6
Photophysical measurements made with the other merocya-
nine dyes in ethanol solution showed very similar behaviour
and, in particular, both triplet and isomer were readily
observed. The nature of the heteroatom did not a†ect the
absorption or Ñuorescence spectral proÐles but caused signiÐ-
cant changes in the positions of the respective peak maxima
and also in the molar absorption coefficient measured at the
absorption peak (e ) (Table 1). It is further seen that the
nature of the heteroatom exerts a strong e†ect on the magni-
tude of the derived quantum yields and lifetimes. Indeed, there
max
CHCl (Table 2). Except for 1, the dyes demonstrate a clear
3
preference for organic media, especially the selenium-
are small changes in both U and q , without a†ecting to any
F
S
containing merocyanines. The amount of dye assimilated into
the cell after incubation at 37 ¡C for exactly 30 min with a
great extent the magnitude of the Stokes shift (D ). There are
SS
more signiÐcant changes in the quantum yield for formation
Ðxed concentration of dye ([dye] ) reÑects the partition coef-
of the triplet excited state (U ) and in the triplet lifetime (q ) in
cell
T
T
Ðcient of that dye. Lipophilic dyes enter the cell more easily,
deoxygenated ethanol solution (Table 1). The quantum yield
where they accumulate in the plasma membrane as demon-
strated by confocal Ñuorescence microscopy. The cell-killing
capability of the dyes towards Daudi cells is illustrated by the
percentage of cells surviving after exposure to total illumi-
nation of 40 and 90 J cm~2 at the absorption maximum
(Table 2). The results are expressed in the form of log
[reduction] factors, which are deÐned as the logarithm of the
number of surviving cells divided by the number of cells
exposed to irradiation.
There is a large variation in cell-killing efficacy, with the
more lipophilic dyes giving the better performance. In report-
ing the various log [reduction] factors we have followed
normal practice in that no attempt has been made to correct
the experimental data for in situ dye concentrations or for the
number of photons absorbed by a particular suspension. It
must be realized, therefore, that the quoted log [reduction]
values will be dependent on several factors and, in particular,
on the ability of the dye to penetrate inside the cell. Both
bound dye and light are needed to e†ect efficient cytotoxicity
at the dye concentrations used here. Dark cytotoxicity39 can
be induced with very high loading (i.e., [50 lg per cm~3) of
dye. During illumination, coloration due to bound dye fades
and does not recover after standing in the dark.
for generation of the unstable isomer (U ) and its lifetime (q )
I
I
were also found to be dependent on the nature of the hetero-
atom (Table 1). Marked changes were also apparent in the
di†erential spectra recorded for the cis-isomers, especially the
di†erential absorption coefficients. Clearly, the most impor-
tant perturbation of the photophysical properties of these dyes
occurs upon introduction of selenium atoms into aromatic
and/or barbiturate subunits.
Because of the similarity of spectral proÐles and overall
photophysical behaviour, we consider that the heteroatoms
merely a†ect the rates of inherent processes (i.e., Ñuorescence,
isomerization, internal conversion and intersystem crossing)
Table 2 Partition coefficients, di†usion coefficients, in situ dye con-
centrations, and the efficacy for light-induced cytotoxicity towards
Daudi cells measured for the various merocyanine dyes
log [reduction]d
compound log Pa Db/107 cm2 s~1 [dye]
c
40 J
90 J
cell
100
1
2
3
4
5
6
]1.10
[0.02
[0.13
[0.71
[0.60
[1.24
2.2
1.9
2.2
2.1
2.0
1.9
0.19
0.43
0.62
0.80
0.72
0.96
1.42
1.88
2.45
3.36
2.96
3.75
140
170
200
195
225
Discussion
Electronic and steric considerations
a Partition coefficient, ^10%. b Di†usion coefficient, ^10%. c In situ
dye concentration in arbitrary units, ^15%. d Cell-killing efficacy
measured exposure to 40 and 90 J of light energy at the appropriate
absorption maximum, ^7%.
The ground-state structure and the general photophysical
behaviour of merocyanines 1È6 remain similar. Thus, each dye
exists exclusively in the all-trans conformation in its ground
2496
J. Chem. Soc., Faraday T rans., 1997, V ol. 93

View Article Online
state but, upon illumination with visible light, efficient photoi-
somerization to a cis-isomer takes place. Photoisomerization
occurs in competition to Ñuorescence and intersystem crossing
to the triplet manifold. Examination of the results collected in
Table 1 shows, however, that the nature of the heteroatoms in
aromatic and/or barbiturate subunits exerts a pronounced
inÑuence on the photophysical properties, as measured in
ethanol solution at room temperature. This e†ect becomes
clear simply by considering the various absorption and emis-
sion maxima. Since the primary objective of the present study
is to identify the most appropriate photosensitizer for use as a
phototoxin towards leukemia cells, it is important to under-
stand the reasons why the heteroatom perturbs these proper-
ties.
For simplicity, merocyanines 1È6 can be divided into cate-
gories; namely, those in which the aromatic heteroatom X is
varied whilst retaining a Ðxed barbiturate residue (1È4) and
dyes for which the aromatic subunit remains unaltered but the
heteroatom Y, present as a terminal group for the barbiturate
residue, is changed systematically (4È6). In the Ðrst instance,
the terminal heteroatom Y is sulfur while for the second set of
compounds the aromatic heteroatom X is selenium. We have
retained the other structural attributes, including the water-
solubilizing anchor and the lipophilic butyl chains incorpor-
ated into the barbiturate unit. In evaluating individual e†ects,
it is assumed that modiÐcations in one part of the molecule do
not signiÐcantly a†ect structural or electronic properties of the
other residue.
Although the N atom of the aromatic nucleus functions as the
primary electron donor and the carbonyl groups on the barbi-
turate subunit are the principal acceptors it is clear that other
functions can contribute towards the dipole moment (Scheme
2). In particular, the CwN amide bonds should be considered
to possess partial double-bond character such that conjuga-
tion extends throughout the entire molecule. According to
extended Huckel molecular-orbital calculations, the electron
density localized on the amido N atoms, and hence their ten-
dency to enter into ionic resonance forms, is inÑuenced by the
electronegativity of the heteroatom Y. In the special case of
Y \ O, additional polar resonance structures can be drawn
that involve ionization of the terminal carbonyl group and that
serve to augment the polarity of this dye. The latter e†ect
is apparent from the absorption spectrum since, relative to 4
and 6, the maximum moves to higher energy while there
is an increase in the oscillator strength ( f ) for the transition
(Table 1).
Similarly, the heteroatom X might be expected to a†ect the
molecular dipole moment in one of two ways. Firstly, intro-
duction of the larger heteroatoms could distort the NwCwC
bond angle and thereby restrict formation of zwitterionic reso-
nance forms involving the N atom as a donor. Molecular
mechanics calculations made using AM3 parameters,
however, indicated that this bond angle was not particularly
sensitive to the nature of X and remained around 123¡. Sec-
ondly, the heteroatom could modify the degree of electron
density localized on the neighbouring N atom, which, in turn,
would a†ect its ability to function as a donor for polar reso-
nance forms. Extended HMO calculations conÐrm that the
nature of X inÑuences the electron density resident on the N
atom, probably by way of an inductive e†ect via the connect-
ing aromatic ring. The calculated electron densities (s) are
listed in Table 3 and it can be seen that the absorption
maximum and oscillator strength vary systematically with this
parameter. Thus, the electron density is highest for X \ 0 and,
consequently, merocyanine 1, which is expected to possess the
highest dipole moment, absorbs at higher energy with the
largest oscillator strength (Table 1). The selenazole derivative
4 is expected to be the least polar dye in this series since its N
atom is the weakest donor. Consequently, the absorption
maximum is situated at lower energy and the transition is less
allowed than found for the other members.
From the above discussion it becomes clear that the pres-
ence of polar resonance structures must increase the energy of
the absorption transition and raise the total oscillator
strength. This e†ect has been noted previously for merocya-
nine dyes having di†erent terminal acceptor functions.41
Because the dipole moments are small it seems likely that
dipolar resonance structures make only a modest contribution
to the total electronic structure of the molecule. Minor
changes in the fraction of these dipolar forms, therefore, may
A detailed understanding of the optical properties of these
merocyanine dyes is rendered difficult by the incongruous
nature of their electronic properties. The dyes are quite polar,
having dipole moments of ca. 10 D (1 D B 3.336 ] 10~30
Cm),40 due to the presence of zwitterionic resonance forms.
Scheme 2
J. Chem. Soc., Faraday T rans., 1997, V ol. 93
2497

View Article Online
Table 3 Derived rate constants for isomerization and intersystem crossing for the various merocyanine dyes together with the stereoelectronic
properties induced by changing the nature of the heteroatoms
compound
X
Y
sa
mb/cm~1
V
c/cm3
E
d/cm~1
k
e/108 s~1
k
f/s~1
k
e/106 s~1
k f/103 s~1
mol
135
OP
TC
CT
ISC
T
1
2
3
4
5
6
O
C
S
Se
Se
Se
S
S
S
S
O
Se
0.340
0.326
0.312
0.307
0.307
0.307
519
461
730
2024
1813
3318
17 860
16 890
16 810
16 560
17 270
16 610
9.8
4.9
4.0
1.0
6.5
0.6
150
1000
90
420
320
285
7.3
4.6
10.3
145
46.5
785
1.25
1.20
1.50
7.40
5.55
10.55
160
141
144
144
144
a Calculated electronic charge resident on the N atom in the aromatic residue, ^0.005. b SpinÈorbital coupling constant taken from ref. 43.
c Molar volume of the rotor calculated using the parameters given in ref. 46, ^5 cm3. d Optical energy of the absorption transition, ^50 cm~1.
e ^12%. f ^5%.
cause large perturbations to the optical properties. In this
way, the heteroatoms modify the electronic and optical
properties of these merocyanine dyes and, in turn, have
important consequences for the photophysical characteristics.
spinÈorbital coupling constants (m) for these two heteroatoms
(Fig. 4), as might be expected for a non-radiative process con-
trolled exclusively by spinÈorbit coupling.42 Allowing for the
paucity of data, the rate constant can be expressed as follows
k
\ A(c c )2ma
(3)
ISC
X Y
Photophysical properties
where A is the corresponding rate constant in the absence of
spinÈorbital coupling, m is the sum of the spinÈorbital coup-
ling constants43 for heteroatoms X and Y, and the constant a
is expected to have a value close to 2. The terms c and c
For each merocyanine derivative studied here, the Ñuores-
cence spectral proÐle was observed to be a reasonable mirror
image of the relevant portion of the absorption spectrum
while the latter was in good agreement with the corrected
Ñuorescence excitation spectrum. Small Stokes shifts are
apparent for each dye, indicating that the geometry of the
excited singlet state is similar to that of the ground state. In
fact, the excited singlet state shows an absorption spectrum
that is not unlike that of the ground state, there being con-
siderable overlap between the two sets of transitions. Fluores-
X
Y
refer, respectively, to atomic orbital coefficients at atoms X
and Y and allow for their relative interaction with the
chromophore, which might di†er between the two sites.17 On
the basis of Fig. 4, a \ 2.3 ^ 0.2 and it is apparent that inter-
system crossing would be essentially negligible in the absence
of the spinÈorbital coupling e†ects induced by these hetero-
atoms. It appears that the rates of intersystem crossing can be
explained in terms of an internal heavy-atom e†ect whilst a
similar correlation is observed for deactivation of the triplet
state (Fig. 4), where a \ 1.2 ^ 0.1. Since the nature of the het-
eroatom a†ects the energy of the Ðrst-excited singlet state it is
likely that there is a variation in the triplet energies of these
dyes which would also contribute towards the rate of non-
radiative deactivation of the triplet state. However, the form
of Fig. 5, and in particular the derived value for a, suggests
that the di†erences in triplet energy are either small or that
the energy-gap law44 is less e†ective at controlling the triplet
lifetime than is the internal heavy-atom e†ect for these dyes.
Internal conversion from the Ðrst-excited singlet state to the
ground state plays an important, if not dominant, role in the
photophysics of merocyanine dyes and accounts for much of
the overall photon balance. In part, internal conversion results
in formation of an unstable isomer which we presume to arise
via twisting around a double bond in the polyenic chain to
give the cis conformation. Photoisomerization requires tor-
sional motion of at least one of the terminal groups and pre-
vious work has established15 that it is the benzoxazole
cence quantum yields (U ) are relatively insensitive to the
F
nature of the heteroatom and there is only a modest variation
in the lifetime of the Ðrst-excited singlet state (q ). Comparable
S
singlet lifetimes are obtained by both transient absorption and
by time-resolved Ñuorescence spectroscopies. The radiative
rate constants (k \ U /q ) vary somewhat throughout the
F
F S
series (Table 1) with values derived from the StricklerÈBerg
expression33 being in good accord with those determined from
experimental data. The perturbation of k originates from
F
changes in the spectral maxima and in the molar absorption
coefficients.
Deactivation of the excited singlet state results in popu-
lation of the triplet manifold, although in all cases the sum of
quantum yields for Ñuorescence and triplet formation is con-
siderably less than unity (i.e., U ] U \ 1). Absorption
F
T
spectra and molar absorption coefficients recorded for the
lowest-energy triplet states remain comparable and consist of
a
relatively broad transition centred around 680 nm.
However, quantum yields for formation of the triplet excited
state (U ), as measured in deoxygenated ethanol, vary over a
T
wide range. There is a corresponding disparity in the triplet
lifetimes measured in deoxygenated solution at low laser
intensity where tripletÈtriplet annihilation is unimportant. The
triplet state decays to restore the ground state, without
involvement of a long-lived isomer, and the properties of the
triplet state are insensitive as to whether that state is popu-
lated via intersystem crossing or by way of triplet energy
transfer from anthracene. Each triplet reacts with molecular
oxygen with a bimolecular rate constant of ca. 1 ] 109 dm3
mol~1 s~1 to generate singlet molecular oxygen, O (1* ).
2
g
Quantum yields for generation of O (1* ) in O -saturated
2
g
2
ethanol (U ) could be measured for a few of the dyes using
time-resolved, near-IR luminescence spectroscopy but in most
*
cases the yields were too low for proper determination by this
method.
Rate constants for formation of the triplet excited state via
Fig. 4 Double-logarithm plot showing the correlation between the
rate constants for intersystem crossing from excited singlet state (…)
and from the excited triplet state (>) with the combined spinÈorbit
coupling constant for heteroatoms X and Y. The solid lines drawn
through the datapoints correspond to non-linear, least-squares com-
puter Ðts to eqn. (3) with the parameters given in the text.
direct intersystem crossing (k \ U /q ) increase with increas-
ISC
T S
ing atomic number of the heteroatoms X and Y (Table 3). In
fact, there is a clear correlation between k and the combined
ISC
2498
J. Chem. Soc., Faraday T rans., 1997, V ol. 93

View Article Online
studies made with related merocyanine dyes,15 has a value of
ca. 1013 s~1. According to the simple potential-energy
diagram given as Fig. 6, the total activation energy for inter-
nal conversion can be estimated from
E \ E [ E
(5)
A
B
op
where E is the potential energy at the top of the barrier,
assuming this latter term remains constant for all members of
B
the series. The validity of this latter assumption is conÐrmed
by the realization that the rate constants for internal conver-
sion and photoisomerization show an exponential dependence
on the energy of the absorption transition (Fig. 5). Thus,
increasing the potential energy of the excited singlet state by
raising its [equilibrated] excitation energy provides for a
faster rate of isomerization since the barrier is lowered. The
Fig. 5 Correlation between the rate constants for internal conver-
sion (…) and for light-induced isomerization (>) with the optical
energy of the absorption transition. The solid lies drawn through the
data points correspond to non-linear, least-squares computer Ðts to
eqn. (4) and (5).
average value for E extracted from the Ðts corresponds to
B
20 100 ^ 200 cm~1. As expected, the rate of internal conver-
subunit that undergoes displacement in 1. The structure of the
so-called cis-isomer is unknown but we assume that it arises
by way of rotation around the central double bond. Absorp-
tion spectra recorded for the cis-isomers show that this species
absorbs slightly to the red of the corresponding trans-isomer
and has a somewhat smaller molar absorption coefficient at
the peak. Such spectral changes would be consistent with
rotation of the aromatic head-group around the Ðrst of the
double bonds in the polymethine backbone:
sion shows a signiÐcantly better correlation with the energy of
the excited state than does the rate of isomerization.
It is likely that similar considerations hold for thermal
isomerization of the cis-isomer but it is more difficult to assess
the energy of this species. There is no reason to suppose that
this energy remains constant throughout the series of dyes
studied here and, therefore, the relative k values might be
CT
considered to reÑect the size of the barrier to isomerization.
Photocytotoxicity studies
There can be no doubt that merocyanine dyes, such as 1, are
potent phototoxins for certain types of cancer cells under
visible light illumination.27 The reaction mechanism remains
obscure, although many reports have attributed cell killing to
the long-lived triplet excited state of the dye. In particular,
reaction between the triplet state and molecular oxygen to
give singlet molecular oxygen O (1* ), has been invoked as
2
g
Quantum yields for formation of the cis-isomer (U ) depend
I
upon the nature of the heteroatom in merocyanines 1È4 but
the rate constants for photoisomerization (k \ U /q ) show
TC
I S
no obvious dependence on the molar volume of the rotor
(V ).45 The latter parameter (Table 3) is deÐned as the molar
ROT
volume of the aromatic head-group,15 including the water-
solubilizing chain, and the heteroatom X makes only a small
contribution to the total.46 Similarly, there seems to be no
signiÐcant correlation between V
and the rate constant
ROT
(k ) with which the cis-isomer reverts to the original trans
CT
form. These Ðndings suggest that isomerization is only weakly
dependent on the size of the rotor but with the variation in
rate constants being set by some other factor.
In fact, both the rate of internal conversion [k \ (1 [ U
IC
F
[ U )/q ] and the rate of photoisomerization are seen to
T
S
depend exponentially on the energy of the ground-state
absorption transition (E \ 1/j ), as illustrated by Fig. 5.
op
max
This situation arises because the absorption maximum reÑects
the relative dipole moment of the molecule which, in turn, is a
statement about the bond order at the isomerizing bond.
Polar dyes have lower bond order for the polyenic chain and
are characterized by higher-energy absorption maxima. Fur-
Fig. 6 SimpliÐed potential-energy diagram proposed to account for
light-induced and thermal isomerization steps. According to this
model there is a barrier to each isomerization step (E ). For photo-
A
isomerization the barrier is taken as the energy di†erence between the
thermore, isomerization involves an activation energy (E )
A
potential at the top of the barrier (E ) and the excitation energy for
such that the rate constant can be expressed as follows:
B
the Ðrst-excited singlet state (E ). We assume that E remains Ðxed
OP
B
for all members of the series but that E depends on the dipole
moment of the individual dye. In principle, similar considerations
k
\ B exp([E /RT )
(4)
OP
IC
A
Here, the pre-exponential factor B refers to the activationless
could be made for thermal isomerization of the cis-isomer to the more
stable trans-isomer.
rate constant for internal conversion which, from previous
J. Chem. Soc., Faraday T rans., 1997, V ol. 93
2499

View Article Online
being the primary process responsible for initiation of cell
death.27 This is somewhat surprising because of the low triplet
yield found for 1 and also because of preferential residence in
the plasma membrane13 where damage by O (1* ) is mini-
mized. In an e†ort to explore further the role of the triplet
excited state in light-induced cytotoxicity we have quantiÐed
the performance of merocyanines 1È6 as phototoxins under
identical conditions. Since these compounds exhibit markedly
disparate triplet quantum yields, it was anticipated that the
efficacy for photodestruction of Daudi cells could be related to
the triplet yield measured in Ñuid solution. At the concentra-
tions used here none of the compounds were abnormally toxic
in the dark.
eter, in turn, inÑuences the rate of assimilation of reagent into
intact cells. Indeed, it is well established that replacement of
oxygen or sulfur atoms in barbiturates by selenium provides
an important route for the preparation of lipophilic drugs.48
Furthermore, earlier work has reported15 a direct correlation
between the rate of uptake of sensitizer and the log
[reduction] factor for merocyanine dyes having similar triplet
yields. On this basis, it is not possible to explain the improved
performance of the lipophilic sensitizers in terms of their
enhanced triplet yields.
It was observed that the nature of the heteroatom did not
a†ect the selective assimilation of merocyanine dye into leuke-
mia cells. By measuring the rate of uptake of 1 into infected
and normal cells we conclude that the preference for leukemia
cells corresponds to a factor of ca. 27-fold. Within statistical
limits, all the derivatives studied here exhibit the same prefer-
ence for infected cells. This level of cellular speciÐcity is con-
sidered to be too low for devising a Ñuorescence-based
diagnostic test for leukemia cells present at low concentration
in blood supplies.
2
g
Fig. 7(a) shows the correlation between triplet quantum
yield measured in deoxygenated ethanol and the efficacy for
light-induced cell killing as expressed in terms of the log
[reduction] factor. At two di†erent irradiation doses, there is
a good correlation between the log [reduction] factor and the
measured triplet quantum yield. On the assumption that the
triplet state reacts quantitatively with molecular oxygen to
produce O (1* ) it is possible that similar correlations hold
2
g
between log [reduction] and the quantum yield for formation
Conclusion
of O (1* ), although we were unable to measure the latter
2
g
values in many cases. The quantitative structureÈreactivity
correlation (QSRC) shown in Fig. 7(a) could be used to argue
in favour of a triplet-based cytotoxicity mechanism.47
It has been shown that large variations in triplet quantum
yield can be engineered by replacement of the heteroatoms in
these merocyanine dyes. Such modulations in triplet yield are
achieved by way of a cooperative e†ect in which the rate of
intersystem crossing is enhanced by the internal heavy-atom
e†ect while there is a concomitant attenuation in the rate of
isomerization due to a decrease in dipole moment. It becomes
clear, therefore, that these heteroatoms play extremely impor-
tant roles in determining both photophysical and physico-
chemical properties of the dyes, due to subtle electronic e†ects.
The ability to control the photophysical properties in this way
is useful in that it allows optimization of the performance of
the dye for selected tasks. However, the mutual correlation
between photophysical and physicochemical properties, espe-
cially the relationship between triplet yield and molecular
polarity, prevents meaningful discussion about the mechanism
for the observed light-induced cytotoxicity. It is quite unreal-
istic to attribute the improved performance of the selenium-
containing dyes as phototoxins to their higher triplet quantum
yields.47 Whilst this parameter probably contributes towards
the overall efficacy of these compounds it is more reasonable
to suppose that their higher intracellular solubility and faster
rate of uptake into infected cells provides the real impetus for
improved biocidal activity.
Unfortunately, equally good QSRCs exist between log
[reduction] and the partition coefficient [Fig. 7(b)] or the in
situ concentration of dye [Fig. 7(c)]. These latter correlations
arise because the nature of the heteroatom exerts a pro-
nounced e†ect on the lipophilicity of the dye and this param-
This work was supported by the National Institutes of Health
(A.H.), the North Texas Leukemia Society (K.S.G.), the Royal
Society of London (A.H.) and the University of Glasgow
(A.C.B.). The award of a J.W.T. Jones Travel Fellowship
(A.C.B.) is also gratefully acknowledged.
References
1
2
C. J. Gomer, Photochem. Photobiol., 1991, 54, 1093.
A. Harriman, L. C. T. Shoute and P. Neta, J. Phys. Chem., 1991,
95, 2415.
F. Sieber, Photochem. Photobiol., 1987, 46, 1035.
B. Kalyanaraman, J. B. Feix, F. Sieber, J. P. Thomas and A. W.
Girotti, Proc. Natl. Acad. Sci. USA, 1987, 84, 2999.
J. Davila, A. Harriman and K. S. Gulliya, Photochem. Photobiol.,
1991, 54, 1.
3
4
Fig. 7 Relationship between the efficacy of the dye to function as a
phototoxin towards Daudi cells and various photophysical or physi-
cochemical properties that vary according to the nature of the hetero-
atoms X and Y. In each case, data are for exposure to 40 J cm~2 (…)
and 90 J cm~2 (>). The plots refer to correlations between log
[reduction] factors and (a) logarithm of the quantum yield for forma-
tion of the excited triplet state, (b) logarithm of the partition coeffi-
cient and (c) the in situ concentration of dye. The solid lines drawn
through the data points are non-linear, least-squares computer Ðts to
an expression of the type y \ a ] bx and have no theoretical signiÐ-
cance.
5
6
7
B. Franck and U. Schneider, Photochem. Photobiol., 1992, 56, 271.
J. Davila, A. Harriman and K. S. Gulliya, J. Chem. Soc., Chem.
Commun., 1989, 1215.
8
9
P. H. Aramendia, M. Krieg, C. Nitsch, E. Bittersman and S. E.
Braslavsky, Photochem. Photobiol., 1989, 48, 187.
N. S. Dixit and R. A. Mackay, J. Am. Chem. Soc., 1983, 105, 2928.
10 R. J. Singh, B. Feix, T. J. Pintar, A. W. Girotti and B. Kalyanara-
man, Photochem. Photobiol., 1991, 53, 345.
2500
J. Chem. Soc., Faraday T rans., 1997, V ol. 93

View Article Online
11 A. S. Waggoner and A. Grinvald, Anal. NY Acad. Sci., 1997, 303,
217.
30 W. Kauzmann, Quantum Chemistry, Academic Press, New York,
1957.
12 M. Hoebeke, J. Piette and A. Van der Vorst, J. Photochem. Pho-
31 J. B. Birks, Photophysics of Aromatic Molecules, Wiley-
tobiol. B: Biol., 1991, 9, 281.
Interscience, New York, 1970.
13 P. R. Dragsten and W. W. Webb, Biochemistry, 1978, 17, 5228.
14 K. S. Gulliya, J. Davila and A. Harriman, Cancer J., 1990, 3, 360.
15 A. C. Benniston, K. S. Gulliya and A. Harriman, J. Chem. Soc.,
Faraday T rans., 1994, 90, 953.
32 S. R. Meech and D. Phillips, J. Photochem., 1983, 23, 193.
33 S. J. Strickler and R. A. Berg, J. Chem. Phys., 1962, 37, 814.
34 A. C. Benniston and A. Harriman, J. Chem. Soc., Faraday T rans.,
1994, 90, 2627.
16 W. H. H. Gunter, R. Searle and F. Sieber, Semin. Hematol., 1992,
29, 88.
35 L. K. Patterson, G. Porter and M. R. Topp, Chem. Phys. L ett.,
1970, 7, 612.
17 R. Bonnett, A. Harriman and A. N. Kozyrev, J. Chem. Soc.,
Faraday T rans., 1992, 88, 763.
36 P. R. Ogilby and C. S. Foote, J. Am. Chem. Soc., 1982, 104, 2069.
37 P. H. Aramendia, R. Duckowicz, L. Scha†ardi and J. O. Tocho,
J. Phys. Chem., 1990, 94, 1389.
18 W. H. H. Gunter, R. Searle and F. Sieber, Phosphorus, Sulfur,
Silicon, 1992, 67, 417.
38 R. J. Singh, J. B. Feix and B. Kalyanaraman, Photochem. Photo-
biol., 1992, 55, 483.
19 L. Pekkarinen and H. Linschitz, J. Am. Chem. Soc., 1960, 82,
2407.
39 K. S. Gulliya, S. Pervaiz and R. B. Dowben, Photochem. Photo-
biol., 1990, 51, 831.
20 J. K. Hurley, N. Sinai and H. Linschitz, Photochem. Photobiol.,
1983, 38, 9.
21 G. Porter and M. W. Windsor, Discuss. Faraday Soc., 1954, 17,
178.
40 L. M. Kushner and C. P. Smyth, J. Am. Chem. Soc., 1949, 71,
1401.
41 L. G. S. Brooker, Frontiers in Chemistry, V ol. III, Interscience,
22 I. Carmichael and G. L. Hug, J. Phys. Chem. Ref. Data, 1996, 15,
New York, 1945, p. 129.
1.
42 S. P. McGlynn, T. Azumi and M. Kinoshita, Molecular Spectros-
copy of the T riplet State, Prentice Hall, Englewood Cli†s, NJ,
1969.
23 M. A. J. Rodgers and P. T. Snowden, J. Am. Chem. Soc., 1982,
104, 5541.
24 R. Bonnett, D. J. McGarvey, A. Harriman, E. J. Land, T. G. Tru-
scott and U-J. WinÐeld, Photochem. Photobiol., 1988, 48, 271.
25 D. Axelrod, D. E. Koppel, J. Schlessinger, E. Elson and W. W.
Webb, Biophys. J., 1976, 16, 1055.
26 K. S. Gulliya, J. L. Mathews, J. W. Fay and R. M. Dowben,
Cancer Chemother. Pharmacol., 1988, 22, 211.
27 K. S. Gulliya, Novel Chemotherapeutic Agents: Preactivation in
the T reatment of Cancer and Aids, R. G. Landes, Austin, TX,
1966.
28 F. M. Hamer, T he Chemistry of Heterocyclic Compounds. T he
cyanine Dyes and Related Compounds, Interscience, New York,
1963.
43 S. L. Murov, Handbook of Photochemistry, Marcel Dekker, New
York, 1973.
44 R. Englman and J. Jortner, Mol. Phys., 1970, 18, 145.
45 P. Gautam and A. Harriman, J. Chem. Soc., Faraday T rans.,
1994, 90, 697.
46 A. Bondi, J. Phys. Chem., 1964, 68, 441.
47 R. W. Redmond, M. B. Srichai, J. M. Bilitz, D. D. Schlomer and
M. Krieg, Photochem. Photobiol., 1994, 60, 348.
48 H. G. Mautner and E. M. Clayton, J. Am. Chem. Soc., 1959, 81,
6270.
29 A. Harriman, J. Photochem. Photobiol. A: Chem., 1992, 65, 79.
Paper 7/01638G; Received 7th March, 1997
J. Chem. Soc., Faraday T rans., 1997, V ol. 93
2501