Effect of BSA binding on photophysical and photochemical properties of triarylmethane dyes

Article abstract of DOI:10.1021/jp981185n

We have employed a combination of steady-state and time-resolved spectroscopic techniques to explore the effect of protein binding on the photophysical and photochemical properties of three triarylmethane dyes: ethyl violet, crystal violet, and malachite green. Our results indicate that the binding sites of bovine serum albumin (BSA) are very efficient in preventing fast nonradiative relaxation processes that occur via rotational motion of the aromatic rings of these triarylmethanes. As a result, remarkable enhancements in fluorescence quantum yield and lifetime, intersystem crossing efficiency, and photoreactivity are observed upon protein binding. The 532 nm laser-induced photobleaching of ethyl violet noncovalently bound to BSA yields leuco ethyl violet and 4,4a?2-bis(diethylamino)benzophenone as reaction products. The former was more prominent in nitrogen-purged samples and the latter in air-equilibrated samples. The time-resolved transient spectra of the ethyl violet complex show superimposed elements of the spectroscopic signatures of both ethyl violet triplet and the semireduced dye radical. Based on the nature of the reaction photoproducts and transient intermediates, the first step of the bleaching process is postulated to be an electron or hydrogen atom transfer from the protein to the dye moiety. An analogous reaction mechanism was observed for protein-bound crystal violet.

Full text of DOI:10.1021/jp981185n

4678
J. Phys. Chem. B 1998, 102, 4678-4688
Effect of BSA Binding on Photophysical and Photochemical Properties of Triarylmethane
Dyes
Mauricio S. Baptista and Guilherme L. Indig*
School of Pharmacy, UniVersity of Wisconsin, Madison, Wisconsin 53706
ReceiVed: February 17, 1998; In Final Form: April 14, 1998
We have employed a combination of steady-state and time-resolved spectroscopic techniques to explore the
effect of protein binding on the photophysical and photochemical properties of three triarylmethane dyes:
ethyl violet, crystal violet, and malachite green. Our results indicate that the binding sites of bovine serum
albumin (BSA) are very efficient in preventing fast nonradiative relaxation processes that occur via rotational
motion of the aromatic rings of these triarylmethanes. As a result, remarkable enhancements in fluorescence
quantum yield and lifetime, intersystem crossing efficiency, and photoreactivity are observed upon protein
binding. The 532 nm laser-induced photobleaching of ethyl violet noncovalently bound to BSA yields leuco
ethyl violet and 4,4′-bis(diethylamino)benzophenone as reaction products. The former was more prominent
in nitrogen-purged samples and the latter in air-equilibrated samples. The time-resolved transient spectra of
the ethyl violet complex show superimposed elements of the spectroscopic signatures of both ethyl violet
triplet and the semireduced dye radical. Based on the nature of the reaction photoproducts and transient
intermediates, the first step of the bleaching process is postulated to be an electron or hydrogen atom transfer
from the protein to the dye moiety. An analogous reaction mechanism was observed for protein-bound crystal
violet.
Introduction
The most prevalent reaction paths pertinent to biological
photooxidation processes are electron or hydrogen atom transfer
between the excited sensitizer and the substrate (reaction type
I; eq 1) and energy transfer from the triplet sensitizer to
molecular oxygen (reaction type II; eq 2).1,5 In the photosen-
sitization reaction type I, the radical pair formed through the
primary electron or hydrogen atom transfer event can subse-
quently react with dissolved molecular oxygen to produce a
variety of cytotoxic reactive oxygen species such as peroxyl,
superoxide, and hydroxyl radicals. In the photosensitization
reaction type II the reactive species formed is singlet oxygen
(1O2*).
The use of photosensitizing agents and light for the treatment
of neoplastic diseases, generally known as photodynamic therapy
(PDT), has become a topic of increasing medical interest.1-4
This new modality of cancer treatment has been the subject of
clinical trials involving a remarkable variety of tumors, and some
of the most successful procedures have already been approved
by regulatory agencies of several countries.3,4 The PDT strategy
is aimed at promoting neoplastic cell death and tumor eradication
through the photochemical generation of highly reactive (cy-
totoxic) species in situ. The combined involvement of a
photosensitizer, light, and dissolved molecular oxygen has been
considered as a requirement for the phototoxic effects to
develop.1-5 These oxygen-dependent deleterious effects are
ordinarily designated as photodynamic action.
kI
O2
3Sens* + S
98 radical pair
98 reactive oxygen species (1)
kII
3Sens* + O2 98 1O2*
(2)
In a classical discussion on mechanisms of photosensitized
oxidation, Foote5 considered two categories of photoreactions
that are prevalent in aerobic solutions and of essential impor-
tance for PDT. The first category takes into account processes
in which the primary interaction of the electronically excited
sensitizer occurs with a molecule of substrate (S). The second
category describes reaction pathways in which the primary
interaction involves a molecule of oxygen. Photosensitized
oxidations are predominantly mediated by the first excited triplet
state of the sensitizer (3Sens*). Most often this occurs because
the triplet lives much longer than its respective singlet, and its
long lifetime favors the occurrence of the primary sensitizer-
substrate and sensitizer-oxygen interactions. Long-lived triplets
are especially important when considering interactions between
species that are found free in solution.
For species free in solution, the relative importance of the
type I and II reactions depends on the respective reaction rates,
kI and kII, and the concentrations of substrate and oxygen. In
complex biological systems the noncovalent binding of photo-
sensitizers to proteins, nucleic acids, and other biological
substrates is likely to be a widespread phenomenon.6-11 When
the ground-state photosensitizer is physically attached to a
substrate, the variables regulating the competition between
reaction type I and type II are no longer those that occur for
free species. The type I mechanism becomes much more
competitive, especially when the excited photosensitizer is bound
to a macromolecular domain that, for topological reasons, offers
some protection against diffusional oxygen quenching.12,13 The
binding of photosensitizers to biopolymers has other important
consequences. Within the framework of host-guest complexes
the dynamics of separation of the primary photochemical
* Corresponding author: phone 608/265-6664; e-mail glindig@
facstaff.wisc.edu.
S1089-5647(98)01185-7 CCC: $15.00 © 1998 American Chemical Society
Published on Web 05/15/1998

Triarylmethane Dyes
J. Phys. Chem. B, Vol. 102, No. 23, 1998 4679
products are altered, and thus reaction paths not observed for
species free in solution may become more competitive in
restricted reaction spaces. The photophysical and photochemical
properties of photosensitizers also may be greatly affected by
the nature of the microenvironment in which they are accom-
modated in the biopolymer,6-11 and because diffusion no longer
precedes the interaction of the excited photosensitizer with the
substrate, short-lived excited singlet states are more likely to
engage in the photosensitization process.
In this report we describe a study on mechanisms of
photoreaction of triarylmethane dyes noncovalently bound to a
model biological host (bovine serum albumin). Triarylmethane
dyes are known to bind efficiently to natural and synthetic
anionic biopolymer polyelectrolytes in solution6,7,9,10 and are
among the families of dyes previously considered for employ-
ment in PDT.14,15 We are especially interested in the identifica-
tion of reaction routes facilitated by the protein binding event
that are distinct from oxygen-dependent type I and type II
processes.
where the subscripts X and R refer to the unknown (TAM dye)
and reference (quinine sulfate) samples, respectively. A is the
absorbance of the sample at the wavelength of excitation, B is
the integrated area under the corrected emission spectrum, I is
the relative intensity of the exciting light at the wavelength of
excitation, and n is the refractive index of the solution. The
samples were excited at 570 nm (EV+ and CV+), 429 nm
(MG+), or 350 nm (quinine sulfate), and the corrected fluores-
cence spectra were integrated from 590 to 800 nm (TAM dyes)
or from 360 to 650 nm (reference). The lifetime measurements
were performed by exciting the samples at 310 nm (aqueous
and glycerol solutions; N2 filled arc lamp) or 550 nm (sucrose
octaacetate glass; H2-filled arc lamp). Fluorescence emission
was detected at 90° to the excitation beam using cutoff filters
(450 and 575 nm for excitation at 310 and 550 nm, respectively)
to prevent scattered light from reaching the detection photo-
multiplier. An aqueous solution of colloidal silica was used as
a standard scatterer for the acquisition of lamp profiles. The
fluorescence lifetime data were analyzed using single-, double-,
and triple-exponential functions, and the best fit was identified
on the basis of ø2 values, randomness of residuals, and
autocorrelation function patterns.21 The Timemaster Strobe-
master software package also provided for the simultaneous
analysis of several independent measurements (global data
analysis).22
Experimental Section
Materials. Ethyl violet (EV+, chloride salt; Aldrich), crystal
violet (CV+, chloride salt; Sigma), and malachite green (MG+,
oxalate salt; Eastman Kodak) were recrystallized from methanol,
and their purity was assessed by thin-layer chromatography
(TLC; silica gel, MetOH-acetic acid 95:5, v:v). Quinine sulfate
monohydrate (Aldrich) was recrystallized three times from
water. Potassium ferrioxalate was prepared as previously
described16 and recrystallized three times from water. Leuco
ethyl violet (leuco-EV) was prepared following procedure
described elsewhere17 and recrystallized two times from 1:1
hexanes-methanol. Benzophenone was recrystallized from
ethanol. Sucrose octaacetate was treated with activated carbon
(Darco G-60, Aldrich) in ethanol and recrystallized two times
from ethanol. All recrystallized compounds were dried under
vacuum. Bovine serum albumin (BSA; initial fractionation by
cold alcohol precipitation, globulin and fatty acid free) from
Sigma, leuco crystal violet (leuco-CV), 4,4′-bis(dimethylamino)-
benzophenone (Michler’s ketone), 4,4′-bis(diethylamino)ben-
zophenone, tetrachloro-1,4-benzoquinone, glycerol, and colloidal
silica (Ludox) from Aldrich, ethyl acetate, methanol, and
acetonitrile from Fisher, ethanol from Pharmco, and hexanes
from EM Science were all of high-purity grade and used as
supplied. Water was distilled, deionized, and filtered prior to
use (Millipore Milli-Q system; resistivity, 18 MΩ cm). Unless
otherwise stated, the experiments were carried out at 20 °C in
0.01 M phosphate buffer (sodium salts, from Sigma), pH 5.8
(MG+) or pH 7.3 (EV+ and CV+).
Instruments and Methods. The spectrophotometric studies
were performed with a Shimadzu UV-2101PC spectrophotom-
eter. The molar extinction coefficients employed in the analyti-
cal procedures were the following (ꢀmax × 10-4, M-1 cm-1):
in phosphate buffer, ꢀ617 ) 9.5 (MG+), ꢀ595 ) 9.1 (EV+), ꢀ590
) 9.0 (CV+); in ethyl acetate, ꢀ350 ) 4.3 (4,4′-bis(diethylamino)-
benzophenone), ꢀ345 ) 3.9 (Michler’s ketone), ꢀ273 ) 5.8 (leuco-
EV), and ꢀ266 ) 5.6 (leuco-CV). Fluorescence quantum yield
(φ) and lifetime (τf) measurements were carried out on a
Timemaster Strobemaster fluorometer from Photon Technology
International, Inc. For measurement of fluorescence quantum
yields, quinine sulfate in 0.5 M sulfuric acid was used as
fluorescence standard (φf ) 0.546).18,19 Fluorescence quantum
yield calculations were performed using the equation20
Scatchard plots23 for the binding of EV+ and CV+ to BSA
were obtained using fluorescence data as described elsewhere.24
The binding constants were determined by the nonlinear fitting
of the experimental data to a Langmuir type binding isotherm
(eq 4),23,25 assuming the existence of two classes of BSA binding
sites for EV+ and CV+ (as suggested by the Scatchard analysis)
and assuming that the BSA binding sites are independent,
noninteracting sites.26
[Dye]b/[BSA] ) n1K1[Dye]f /(1 + K1[Dye]f) + n2K2[Dye]f /
(1 + K2[Dye]f) (4)
In eq 4 [Dye]b and [Dye]f are the concentrations of bound and
free dye, respectively. [BSA] represents the total protein
concentration, n1 and n2 are the number of bindings sites type
1 and 2, respectively, and K1 and K2 are the corresponding
binding constants.
For the dye bleaching experiments the samples were placed
in standard 10 mm (path length) quartz cells at a distance of
approximately 10 cm from the light source. The samples were
irradiated using the 532 nm line of a ND:YAG laser (Continuum
model 7010) operating at 10 Hz. The defocused laser beam
(circular profile with diameter of about 5 mm) was directed to
the center of the quartz cell. The absolute photolysis energy
was kept constant over the course of any specific experiment.
The temperature-controlled cell holder allowed continuous
magnetic stirring of the samples during photolysis. The
photobleaching quantum efficiencies were measured considering
only the first 10% decrease in dye concentration and using
potassium ferrioxalate as chemical actinometer.16,27
Our laser flash photolysis equipment is similar to previously
described systems.28,29 The major assembly components are a
nanosecond dye laser (Continuum, ND6000) pumped by a Nd:
YAG laser (Continuum, 7010), used as the excitation light
source, a 300 W xenon arc lamp system (Oriel, 66084), which
provides the analysis beam, a monochromator (CVI, CM110),
a red-sensitive photomultiplier tube (Hamamatsu, R446), and a
dual-channel 600 MHz digital oscilloscope (LeCroy, 9360).
Detailed description of our laser flash photolysis system will
be reported elsewhere.30
φX ) φR(AR/AX)(BX/BR)(IR/IX)(nX/nR)2
(3)

4680 J. Phys. Chem. B, Vol. 102, No. 23, 1998
Baptista and Indig
SCHEME 1: Triarylmethane Dyes
Qualitative analysis of low molecular weight reaction pho-
toproducts was performed by thin-layer chromatography (TLC).
The photolyzed samples were extracted with ethyl acetate
(EtOAc), and the TLC analysis was run on silica gel plates
(Eastman 13181, plates with fluorescence indicator) using
hexanes:EtOAc 2:1 (v:v) as eluent. The photoproducts were
characterized on the basis of RF values of photolyzed and
authentic (standard) samples. The observed RF values were 0.59
( 0.01 (Leuco-CV), 0.44 ( 0.01 (Michler’s ketone), 0.65 (
0.02 (leuco-EV), and 0.51 ( 0.02 (4,4′-bis(diethylamino)-
benzophenone). MK and 4,4′-bis(diethylamino)benzophenone
were detected by the inspection of the chromatogram with
ultraviolet (366 nm) light. Leuco-CV and leuco-EV were
visualized in their oxidized forms, CV+ and EV+, respectively.
The oxidation of the leuco compounds on the TLC plates was
performed by spraying the plates with an acetonitrile solution
of tetrachloro-1,4-benzoquinone. Quantitative analysis of the
reaction photoproducts was performed using UV-visible spec-
troscopy. The yields of the low molecular weight photoproducts
were estimated on the basis of their concentration in the EtOAc
extracts. Calibration curves were obtained by mixing authentic
photoproducts with BSA in the same range of concentrations
as was produced in the bleaching experiments and subjecting
the mixture to the same process of extraction as the photolyzed
samples.
Mass spectra of high molecular weight photoproducts were
obtained using a matrix-assisted laser desorption/ionization
(MALDI) time-of-flight (TOF) mass spectrometer (Bruker
model Reflex II). After photolysis of the protein-dye samples
in phosphate buffer using 532 nm laser light, the low molecular
weight photoproducts were extracted from the aqueous solution
with ethyl acetate, and the aqueous phase dried under a stream
of nitrogen. The resulting solid was dissolved in 3 mL of water
and dialyzed to infinite dilution (>1010-fold), initially against
5% acetic acid solution and subsequently against deionized
(Milli-Q) water. The protein concentration of the solutions used
in the preparation of the MALDI-TOF samples was ap-
proximately 20 mM. The matrix used to facilitate sample
desorption was [2-(4-hydroxyphenylazo)benzoic acid] (HABA).
quantum yields ∼10-5). In more viscous media or in microen-
vironments that render steric hindrance to rotational relaxation
processes, fluorescence and intersystem crossing become more
competitive events, and photoreactivity tends to increase.6,7,9,10
The effect of BSA on the fluorescence of CV+ in 0.01 M
phosphate buffer at 20 °C is shown in Figure 1. The increase
in CV+ fluorescence as a function of protein concentration
reaches a plateau, in agreement with the behavior expected for
TAM dyes when the rigidity of the surrounding microenviron-
ment increases.10,33,34 At the plateau region of the binding
isotherm essentially all dye molecules are noncovalently bound
to the protein, thereby experiencing loss of rotational degrees
of freedom. The BSA binding sites are very effective in
preventing “free rotor” motions in the CV+ moiety. This
decreases the efficiency of radiationless deactivation of CV+,
a process that involves a low-lying twisted intramolecular
charge-transfer state (TICT),32 and leads to a remarkable
enhancement in its fluorescence quantum yield (Table 1) and
lifetime (Table 2). EV+ shows a similar binding profile (Figure
1, inset).
For the case of MG+, the TAM dye displaying the smallest
molecular volume and largest dipole moment, the binding to
BSA is less efficient. In this case, the plateau region of the
binding isotherm is not reached even in the presence of BSA
concentrations 2-fold higher than those required for the binding
isotherms of EV+ and CV+ to reach a plateau (Figure 1, inset).
To minimize the interference of the formation of MG+ carbinol
base during data acquisition, all studies involving MG+ were
performed at pH 5.8. At pH 7.3 the alkaline hydrolysis of MG+
becomes more competitive, precluding the acquisition of reliable
binding information. Because the flattest region of the BSA
titration curve is located between pH 5 and pH 8, no significant
changes in structure and charge distribution within the protein
binding sites should be expected as a result of changes in pH
in the 5.8-7.3 range.35 Indeed, control experiments carried out
with EV+ and CV+ have demonstrated that the BSA binding
isotherms are analogous at both pHs (Figure 1, inset).
The effect of protein binding on the fluorescence quantum
yield of EV+ and CV+ is similar to that observed when the
viscosity36,37 of the surrounding environment increases (at 20
°C) from 1.00 cP (water) to 1410 cP (glycerol), but not as large
as what is observed when these dyes are placed in a rigid
amorphous matrix (sucrose octaacetate glass; Table 1). This
suggests that the steric restriction imposed by the BSA binding
sites on the rotational relaxation of EV+ and CV+ is comparable
to that experienced by these dyes in a microenvironment as
cohesive as glycerol. Assuming the same sort of behavior is
observed for the case of MG+, the fluorescence quantum yield
for this TAM dye bound to BSA is expected to be of the order
of 3 × 10-3 (compare glycerol data in Table 1). The low
Results and Discussion
I. Formation of Ground-State Complexes of BSA with
TAM Dyes. We have employed a combination of photophysi-
cal techniques to characterize the formation of ground-state
complexes of TAM dyes (Scheme 1) with BSA in phosphate
buffer. TAM dyes show poor photoreactivity in low-viscosity
media due to fast relaxation processes that occur via rotational
motions of their aromatic rings.31-33 Accordingly, in aqueous
solutions the fluorescence lifetime of free TAM dyes is very
short, typically in the picosecond range31,33 (fluorescence

Triarylmethane Dyes
J. Phys. Chem. B, Vol. 102, No. 23, 1998 4681
Figure 1. Effect of BSA concentration on the fluorescence spectrum of CV+. From the bottom, in order of increasing fluorescence intensity at 632
nm, the BSA concentrations were (in µM): 0.0, 0.25, 0.62, 1.11, 2.10, 3.31, 5.68, 9.13, 13.50, 19.62, 28.79, 41.02, 49.43, and 74.67. Inset: BSA
effect on the integrated fluorescence of EV+ (squares), CV+ (circles), and MG+ (diamonds). Open symbols, pH 5.8; solid symbols, pH 7.3. Phosphate
buffer 10 mM; [dye] ) 10 µM; T ) 20 °C. λexc ) 520 nm (EV+ and CV+) or 429 nm (MG+).
TABLE 1: Medium Effect on the Fluorescence Quantum
binding constants associated with the high- and low-affinity BSA
binding sites were estimated via nonlinear fitting of the
experimental data to Langmuir-type binding isotherms23,25
(Figure 2, inset). The computed values of binding constants
were 1.4 × 106 and 3.5 × 104 M-1 for the case of EV+ and 3.4
× 105 and 2.1 × 104 M-1 for the case of CV+.39
Yield (Φf) of TAM Dyes
Φf
buffera
ACN
BSAa,b
glycerol
SOAGc
EV+ 9.6 × 10-5 4.8 × 10-5 1.7 × 10-2 2.2 × 10-2 0.17
CV+ 5.6 × 10-5 3.0 × 10-5 7.5 × 10-3 9.2 × 10-3 0.14
MG+ 3.7 × 10-5 2.1 × 10-5
3.7 × 10-3 0.024
It is not necessarily appropriate to interpret the double-
exponential fluorescence decay data by assuming the existence
of two distinct BSA microenvironments that exert discrete
influences on the emission properties of bound TAM molecules.
Although differences in the amino acid composition or other
binding site characteristic may lead to differences in the
photophysical and photochemical behavior of bound TAM dyes,
the fluorescence decay of EV+ and CV+ free in solution also
follows double-exponential functions (Table 2). Therefore, it
is reasonable to presume that even when bound to a single type
of macromolecular binding site the fluorescence decay profiles
of EV+ and CV+ will still typically follow double-exponential
functions.
a [buffer] ) 10 mM, pH 7.3 for EV+ and CV+ and pH 5.8 for MG+.
b [BSA] ) 40 µM. c Assuming the refractive index of the sucrose
octaacetate glass as 1.5.
binding efficiency of MG+ precluded the precise assessment
of its fluorescence quantum yield when bound to BSA. Also
shown in Table 1 are values of fluorescence quantum yield for
the TAM dyes in acetonitrile. The viscosity of acetonitrile at
20 °C is only 0.38 cP, and as expected, the fluorescence quantum
yields of EV+, CV+, and MG+ in acetonitrile are smaller than
their respective values in aqueous solution.
The fluorescence decay of CV+ and EV+ bound to BSA, a
typical multiple-binding-site carrier,26,35,38 follows double-
exponential functions. The global analysis of three independent
determinations yielded fluorescence lifetimes of 0.4 and 2.9 ns
(ø2 ) 1.032) for BSA-bound EV+ (Table 2). For the case of
BSA-bound CV+ the respective lifetimes were 0.4 and 3.6 ns
(ø2 ) 1.056). These are remarkably high values if compared
with the lifetimes of CV+ and EV+ free in low-viscosity media
(5-10 ps31,33). The fluorescence emission of BSA-bound MG+
was not strong enough to permit reliable time-resolved analysis.
The characterization of two populations of excited EV+ and
CV+ might suggest the existence of two classes of binding sites
for these dyes in the protein environment, and indeed, the
Scatchard analysis23,38 of EV+ and CV+ fluorescence data
suggested the existence of two distinct BSA binding sites
capable of accommodating these TAM dyes (Figure 2). The
The double-exponential decay of excited singlet TAM dyes
free in solution is rationalized in terms of the existence of two
main populations of TAM ground-state conformers in equilib-
rium with each other.40-42 Upon photoexcitation, the population
of TAM molecules reaching the conformation of the TICT state
faster is the one displaying the shorter fluorescence lifetime.31
Based on the dynamics of ground-state bleach recovery, the
existence of two excited singlet (S1) TAM populations has been
characterized even in low-viscosity media. For the case of CV+
in ethanol, the ground-state bleach recovery follows double-
exponential functions with decay times of 1.6 and 4.0 ps.33
Ground-state conformers of TAM dyes are observed in
solution, owing to a peculiar characteristic of their molecular
structures.31,40-42 As consequence of steric restrictions, the
aromatic rings of cationic TAM dyes are forced out of the

4682 J. Phys. Chem. B, Vol. 102, No. 23, 1998
Baptista and Indig
TABLE 2: Fluorescence Lifetimes for EV+ and CV+ in Several Environmentsa
BSAb,c
glycerolc
SOAGd
PMAAe
EV+
EV+
CV+
EV+
CV+
EV+
CV+
τ1 (ns)
τ2 (ns)
0.38 ( 0.02
(0.39)
3.0( 0.3
(2.9)
0.38 ( 0.01
(0.39)
3.2 ( 0.4
(3.6)
0.32 ( 0.01
(0.33)
4.3 ( 1.0
(4.3)
0.30 ( 0.01
(0.34)
4.6 ( 0.7
(4.8)
0.46 ( 0.06
(0.44)
2.76 ( 0.04
(2.70)
0.3 ( 0.1
0.5
(0.25)
2.39 ( 0.09
(2.36)
2.5
f1
0.74 ( 0.02
0.26 ( 0.02
1.08 ( 0.06
(1.03)
0.83 ( 0.05
0.17 ( 0.05
0.97 ( 0.02
(1.06)
0.86 ( 0.03
0.14 ( 0.03
1.10 ( 0.05
(1.12)
0.82 ( 0.02
0.18 ( 0.02
1.14 ( 0.07
(1.18)
0.11 ( 0.03
0.89 ( 0.02
1.00 ( 0.02
(0.984)
0.18 ( 0.04
0.82 ( 0.02
1.01 ( 0.07
(0.991)
0.63
0.37
f2
ø2
a Values from the convoluted “best fit” of fluorescence decay curves to double-exponential functions. In parentheses are shown results of the
global analysis of three to five independent measurements. b [BSA] ) 40 mM, phosphate buffer 10 mM, pH ) 7.3. c [Dye] ) 8 µM, λexc ) 310
nm. d Sucrose octaacetate glass; [Dye] ) 6 µM, λexc ) 550 nm. e Aqueous solution of poly(methacrylic acid), pH 3.6; data from ref 10.
Figure 2. Scatchard plots for the binding of EV+ (circles) and CV+ (squares) to BSA in phosphate buffer 10 mM pH 7.3 at 20 °C. [TAM]b )
concentration of dye molecules bound to BSA; [TAM]f ) concentration of dye molecules free in solution, and [BSA] ) total concentration of BSA
(0.32 µM). The biphasic profile of the Scatchard plots suggests the existence of distinct binding sites in the BSA molecule capable of accommodating
TAM molecules. The extrapolation of the dotted straight lines to the horizontal axis gives values of 0.87 and 1.88 for EV+and 0.78 and 1.81 for
CV+. These values suggest the existence of two classes of BSA binding sites for these TAM molecules, each class containing only one binding
site.23,25 Inset: Langmuir-type binding isotherm. The best fit of the experimental data to eq 4 yielded the following values of binding constants:
1.4 × 106 and 3.5 × 104 M-1 for EV+ and 3.4 × 105 and 2.1 × 104 M-1 for CV+.
molecular plane that contains the central carbon atom. The out-
of-plane angle is modulated by both the repulsion between ortho
hydrogen atoms in adjacent aromatic rings and the restoration
force that opposes rotation, because of the double-bond character
of the twisted chemical bond. For the case of symmetric
molecules, such as the triple-substituted CV+ and EV+, the
repulsion and restorative forces acting on each distinct ring are
expected to be equivalent. When the aromatic rings rotate in
the same sense and the angles between them are held even, the
resulting three-blade propeller is symmetric (represented by a
D3 point group), while when one of the aromatic rings rotates
in the opposite sense with regard to the other two rings, an
unsymmetric propeller is generated (represented by a C2 point
group). The existence of two TAM ground-state rotational
isomers in equilibrium in solution was originally contemplated
by Lewis and co-workers40 as the reason for the appearance of
the overlapped doublet in the absorption spectrum of CV+.
The values of the relative integrated fluorescence intensities
(f values; Table 2) provided by the double-exponential analysis
of the fluorescence decay of EV+ and CV+ in glycerol and in
BSA are comparable, suggesting that the population distributions
of emitting species in the protein environment are similar to
the respective distributions in glycerol. This is in keeping with
the observation that the fluorescence quantum yields of EV+
and CV+ bound to BSA are comparable to their respective
fluorescence yields in glycerol (Table 1). In both cases the
population of excited molecules that relax faster are the ones
accounting for the majority of the emitting fluorophores
(respective values of relative integrated fluorescence intensities
around 0.80). In glass of sucrose octaacetate, the population

Triarylmethane Dyes
J. Phys. Chem. B, Vol. 102, No. 23, 1998 4683
TABLE 3: Photobleaching Efficiencies for EV+ and CV+ a
BSAb,c
BSAb,c
bufferb
(N2-purged)
(air-equilibrated)
(air-equilibrated)
EV+
CV+
5.5 × 10-5
3.3 × 10-5
2.5 × 10-6
3.3 × 10-5
1.5 × 10-5
8.4 × 10-7
a Measured using potassium ferrioxalate as chemical actinometer.
Photolysis carried out at 532 nm. b Phosphate buffer 10 mM, pH 7.3;
[Dye] ) 10 µM. c [BSA] ) 40 µM.
BSA-TAM noncovalent complexes. Under photolysis condi-
tions, essentially all TAM molecules are bound to the biopoly-
mer (large [BSA]/[TAM] ratio; plateau region of the binding
isotherms, see inset of Figure 1), thereby experiencing loss of
rotational degrees of freedom. The quantum efficiency of
bleaching was found to be higher in nitrogen-purged than in
air-equilibrated samples (Table 3). The increased bleaching
efficiency of BSA-bound EV+ and CV+ in nitrogen-purged
samples as compared to air-equilibrated samples is an indication
of the involvement of the first excited triplet state of these TAM
dyes in the photobleaching process. The difference between
the photobleaching efficiency of EV+ and CV+ free in solution
and protein bound is more prominent. For the case of CV+ an
18-fold increase is observed as a consequence of BSA binding,
while for the case of EV+ a 13-fold increase is observed (air-
equilibrated samples; Table 3). Enhanced intersystem crossing
(ISC) efficiency for dyes bound to macromolecules, as compared
to the respective unbound dye, has been observed for a large
variety of families of dyes, including triarylmethanes.19 We
were unable to detect any transient signal (nanosecond-
microsecond range) following nanosecond laser excitation of
EV+ and CV+ free in buffer solution, although well-resolved
transient spectra were obtained upon excitation of these TAM
dyes noncovalently bound to BSA (see below).
Figure 3. Effect of photolysis time on the maximum absorption of
the BSA-CV+ complex (595 nm, main panel) and the BSA-EV+
complex (602 nm, inset). Open circles, air-equilibrated samples; solid
circles, nitrogen-purged samples. Laser excitation at 532 nm, 10 Hz,
75 mJ/pulse for the CV+ complex (50 µM BSA) and 87 mJ/pulse for
the EV+ complex (40 µM BSA). Phosphate buffer 10 mM, pH 7.3.
of emitting fluorophores displaying the longer lifetime is larger
than the short-lived (subnanosecond) population, and a sub-
stantial enhancement in fluorescence quantum yield is observed
for both dyes, EV+ and CV+.
Although fluorescence spectroscopy has been identified as a
powerful technique for the detection and characterization of
TAM-protein complexes, absorption spectroscopy can also
provide evidence of binding. The binding of EV+, CV+, and
MG+ to BSA in 0.01 M phosphate buffer at 20 °C led to a
small bathochromic shift in the wavelength of the maximum
dye absorption. The profiles of the respective binding isotherms
(λmax vs [BSA]) resemble those obtained through the analysis
of the protein effect on the integral fluorescence of these TAM
dyes (Figure 1, inset). The wavelength of maximum MG+
absorption is 617 nm in phosphate buffer and 621 nm in the
presence of 80 µM BSA ([MG+] ) 10 µM). For the case of
EV+ the wavelength of maximum absorption shifts from 595
nm in phosphate buffer to 602 nm when bound to BSA. The
respective values for CV+ are 590 nm (buffer) and 595 nm
(BSA-bound). It must be emphasized that the values shown
above were obtained in the presence of a large excess of protein
with respect to dye concentration. They represent spectroscopic
characteristics of TAM dyes bound as monomers to BSA. Note,
however, that several TAM dyes form aggregates upon binding
to a variety of natural and synthetic polyelectrolytes under
conditions of high polymer loading ([dye] . [polymer]).10,39,43,44
Although self-association of TAM dyes leads to remarkable
increases in their photoreactivity,10,44 this report focuses on the
photophysical and photochemical properties of TAM dyes bound
to the protein in their monomeric form.
II. Photochemistry of Protein-Dye Noncovalent Com-
plexes. The observed increase in fluorescence lifetime and
quantum yield of BSA-bound TAM dyes as compared to the
free dyes in buffer solution suggests that the efficiency of
intersystem crossing (ΦISC) might also undergo substantial
increase as consequence of protein binding. We have explored
the binding effect on the photochemistry of EV+ and CV+, the
dyes thermally stable at 20 °C in phosphate buffer pH 7.3. Figure
3 shows the bleaching of BSA-bound CV+ and EV+ upon laser
excitation at 532 nm, as measured by the decrease in absorbance
at the wavelength of maximum absorption of the respective
The analysis of low molecular weight reaction photoproducts
was performed by extraction of photolyzed samples with ethyl
acetate and characterization of the extracted compounds by TLC
and UV-visible spectroscopy. 4,4′-Bis(diethylamino)benzophe-
none and leuco-EV were identified as products of laser
photolysis of protein-bound EV+. For the case of the BSA-
CV+ complex, 4,4′-bis(dimethylamino)benzophenone (Michler’s
ketone) and leuco-CV were characterized as photoproducts. The
benzophenone-type photoproducts were more prominent in air-
equilibrated samples and the leuco dyes in nitrogen-purged
samples.
The reduction of a TAM dye to its leuco form is a
two-electron process (formally 2e- + H+ or H + e-), initiated,
in this case, by a photoinduced electron or hydrogen atom
transfer from the protein to the dye moiety. It should be noted
that 78 out of the 583 BSA amino acid residues are readily
susceptible to photosensitized oxidation5,7 (2 Trp, 4 Met, 17
His, 20 Tyr, and 35 Cys).35,45 After initiation, and following
the reaction coordinates of the bleaching process, the semire-
duced carbon-centered dye radical can either react with dissolved
molecular oxygen to produce a benzophenone-type photoproduct
or accommodate a second electron to form the leuco derivative
of the TAM dye (Scheme 2). Because the electronic excitation
of the dye moiety occurs within a ground-state protein-dye
complex, the first excited singlet state of the dye is created
physically attached to the protein. Therefore, it is conceivable
that not only the TAM’s triplet but also the relatively short-
lived excited singlet populations may, to some extent, engage
in the initial electron-transfer process that leads to the formation
of the semireduced dye radical (no diffusion needed for the
reaction to occur). Scheme 2 illustrates these reaction pathways.

4684 J. Phys. Chem. B, Vol. 102, No. 23, 1998
Baptista and Indig
SCHEME 2
Scheme 2 also shows reaction pathways that generate singlet
oxygen and superoxide radical and simultaneously preserve the
photosensitizer from photodecomposition. Under aerobic condi-
tions the collisional quenching of triplet TAM dyes by dissolved
molecular oxygen prevents the triplets from engaging in the
initial electron-transfer process. The singlet oxygen generated
by triplet-triplet energy transfer46 from triplet TAM to ground-
state oxygen might, in principle, add to the TAM molecules to
generate an unstable dioxetane intermediate whose thermal
cleavage would produce benzophenone-type photoproducts. This
parallel route of formation of the benzophenone-type photo-
products is, however, unlikely to play any meaningful role
here.47,48 Analogously, the oxidation of the semireduced
(neutral) dye radical back to the TAM cation by molecular
oxygen protects the dye from bleaching and generates another
reactive oxygen species5 as reaction product, in this case the
superoxide radical.
In addition to the oxygen quenching of the TAM triplet, the
oxidation of the semireduced TAM dye radical by oxygen is
another reaction pathway that may contribute to the decrease
in photobleaching efficiency in air-equilibrated samples as
compared to nitrogen-purged samples. With the employment
of spin-trapping techniques and electron paramagnetic resonance
(EPR) spectroscopy, the formation of superoxide radicals during
photolysis of CV+ in air-equilibrated aqueous solutions has been
observed in the presence of reducing agents such as NADH
and glutathione.49 Although the photochemical formation of
singlet oxygen and superoxide radicals is of fundamental
importance for the aerobic mechanisms of action of photosen-
sitizers,1,5 the central focus of this discussion will be kept on
the mechanisms of dye photobleaching rather than emphasizing
reaction routes capable of protecting the TAM dyes from
photodecomposition.50
The competitive reaction pathways demonstrated by the
characterization of reaction products in air-equilibrated and
nitrogen-purged samples is also made evident through direct
absorption spectroscopy measurements. Figure 4 shows that
upon laser excitation of BSA-bound EV+ in nitrogen-purged
solution the photoproduct that absorbs in the 380 nm region,
4,4′-bis(diethylamino)benzophenone, is not formed in significant
amounts, while the differential absorption spectrum in the 250-
280 nm region clearly indicates the formation of the dye’s leuco
(reduced) form as photoproduct (Figure 4, inset). The BSA-
CV+ complex displayed equivalent behavior.
Control experiments have demonstrated that when CV+
samples are photolyzed in the presence of the primary reaction
photoproducts, both leuco-CV and Michler’s ketone are removed
from solution through secondary reaction mechanisms. The
engagement of leuco-CV and Michler’s ketone in secondary
processes explains, at least in part, why the primary photolysis
products do not fully accumulate in solution over the course of
the reaction. The amounts of leuco-CV and Michler’s ketone
found in solution after prolonged photolysis are typically in the
10% range of the respective amount of photobleached CV+.
The samples containing EV+ displayed analogous behavior. The
low yields found for the primary low molecular weight reaction
photoproducts upon prolonged photolysis may also be a
consequence of the existence of competitive photodecomposition
pathways leading to the formation of dye-protein heteroadducts.
The formation of dye-protein heteroadducts through the
covalent linkage of dye radical intermediates with BSA was
investigated with the employment of matrix-assisted laser

Triarylmethane Dyes
J. Phys. Chem. B, Vol. 102, No. 23, 1998 4685
observed when a molecular weight between 7.9 and 8.7 kDa is
reached is Gly-Cys-Glu-Lys-Ser-Leu-His.35 This sequence
contains two amino acid residues readily susceptible to photo-
sensitized oxidation,5,7 Cys62 (8.1 kDa peptide) and His67 (8.7
kDa peptide). The oxidative transformation of histidine residues
has been identified as the primary cause of deactivation of a
variety of enzymes.5 A similar computation starting at the BSA
carboxy terminus give the amino acid sequence Ile-Gln-Lys-
Gly-Thr-Asp-Pro. We hypothesize that the BSA fragmentation
occurs at least predominantly at the Cys62 and or His67 amino
acid residues, because the 7.9-8.7 kDa amino acid sequence
from the BSA carboxy terminus does not contain any amino
acid residues readily susceptible to photosensitized oxidation.
Laser flash photolysis experiments provided information on
the nature and spectral signatures of reaction intermediates. The
time-resolved transient spectra following laser photolysis of the
BSA-EV+ noncovalent complex in air-equilibrated and nitrogen-
purged samples are presented in Figure 5. In both cases the
appearance of a broad transient absorption band in the 320-
520 nm spectral region with a maximum around 410 nm
correlates with the bleaching of ground-state EV+. At early
times after the laser pulse, the absorption maximum around 410
nm is slightly more prominent in air-equilibrated samples as
compared to nitrogen-purged samples. In this last case a broader
absorption envelop is observed initially, with the 410 nm peak
becoming more noticeable at ca. 100 ns after the laser pulse.
For wavelengths longer than 620 nm significant values of optical
densities are observed only in the submicrosecond time domain.
This fast component of the transient spectra with absorption
beyond 620 nm also contributes to the broad absorption profile
observed in the 320-520 nm region in the submicrosecond
range (Figure 5). The decay of the BSA-bound EV+ transients
follow triple-exponential decay functions. The average lifetimes
obtained for nitrogen-purged samples were approximately 0.04,
5, and 140 µs. Only the species displaying the longer lifetime
was efficiently quenched by oxygen in the protein environment.
In air-equilibrated samples the lifetime of this species was
approximately 50 µs. A more rigorous analysis of the transient
data to include population distributions was precluded by the
weak spectroscopic signals characteristic of the systems under
consideration and wide distribution of lifetimes.
Figure 4. Changes in absorption at 380 nm upon photolysis of EV+
noncovalently bound to BSA. The samples are the same as shown in
the inset of Figure 3. The 4,4-bis(diethylamino)benzophenone is the
identified reaction product which absorbs in the 380 nm region. Inset:
differential spectrum obtained after 16 min of photolysis of the nitrogen-
purged sample, i.e., the spectrum recorded at t ) 16 min minus the
spectrum recorded at t ) 0 (solid line), and the absorption spectrum of
leuco-EV in ethanol (dashed line).
TABLE 4: MALDI-TOF-MS Data for BSA and BSA-TAM
Complexes Photolyzed at 532 nma
m/z, Da
sampleb
BSA (air)
peak 1
peak 2
peak 3
66 448
67 076
67 210
66 997
67 065
66 986
BSA-EV+ (air)
BSA-EV+ (N2)
BSA-CV+ (air)
BSA-CV+ (N2)
BSA-CV+ (N2)
58 238
59 319
58 595
58 831
58 392
8520
8427
8543
8575
8643
a Values of m/z for monocharged species. b Air, air-equilibrated
samples; N2, nitrogen-purged samples. Samples photolyzed in the
presence of phosphate buffer 10 mM, pH 7.3.
desorption/ionization time-of-flight mass spectrometry (MALDI-
TOF-MS).51-53 Samples with essentially all dye molecules
bound to BSA with 1:1 stoichiometry were photolyzed to probe
whether dye-protein heteroadducts are formed as a consequence
of phototransformation of BSA-EV+ and BSA-CV+ nonco-
valent complexes. An increase in the mass of the protein peak
was observed upon photolysis of TAM-BSA complexes in both
air-equilibrated and nitrogen-purged samples (Table 4). The
shift of the BSA peak is an indication of the photochemical
formation of heteroadducts between BSA and the TAM dyes.
The measured increases in BSA mass were slightly higher than
those expect for the formation of 1:1 BSA:TAM heteroadducts.
The average mass excess (0.22 kDa) is essentially identical to
the mass of HABA (FW 242.23), the matrix material used to
facilitate sample desorption. The formation of covalent adducts
between proteins and the matrix material is an event known to
interfere in MALDI-TOF-MS resolution.53
The mass spectrometry data also show that the laser photolysis
of BSA-TAM noncovalent complexes induces the fragmenta-
tion of the biopolymer. The mass of the smaller fragmentation
product is approximately 8.5 kDa (Table 4). If we assume that
this smaller fragment represents a polypeptide covalently linked
to a TAM molecule, then the mass of the polypeptide itself is
approximately 8.1 kDa. Interestingly, by adding up the mo-
lecular weight of the amino acids of the BSA primary structure
starting from its amino terminus, the amino acid sequence
Because the 320-520 nm spectral range represents a window
between intense absorption bands in the EV+ ground-state
absorption spectrum, it is not surprising that EV+ transients are
repeatedly observed in this spectral region. Accordingly,
following subpicosecond 606 nm excitation of EV+ in organic
solvents, the rise and decay of a picosecond (singlet) transient
was noticed in the 320-520 nm region.54 The 320-520 nm
spectral window is also observed for the case of CV+. The
triplets of ethyl violet10 and crystal violet,55 as well as the
CV• 56-60 and EV• 10 semireduced dye radicals, are all known
to absorb light in the 320-520 nm region. However, the
spectroscopic discrimination of EV+ and CV+ triplets from their
respective free radicals is still a subject of some debate.60
To characterize more rigorously the formation of both the
EV• semireduced radical and the EV+ triplet upon photolysis
of BSA-bound EV+ (Scheme 2), two experiments were carried
out to determine the spectral fingerprints of each of these
transient species individually. The spectral signature of the
transient species formed upon laser photolysis of EV+ in a
sucrose octaacetate glass is shown in Figure 6. The transient
spectra of EV+ in the glassy medium show a broad absorption
envelop in the 320-520 nm region and also considerable
(positive) changes in absorption beyond 620 nm. The glassy

4686 J. Phys. Chem. B, Vol. 102, No. 23, 1998
Baptista and Indig
Figure 5. Transient absorption spectra of EV+ noncovalently bound to BSA. Upper panel, air-equilibrated sample; lower panel, nitrogen-purged
sample. Inset (upper panel): transient decay traces for air-equilibrated and nitrogen-purged samples recorded at 410 nm. [EV+] ) 13 µM, [BSA]
) 40 µM, phosphate buffer 10 mM, pH 7.3, λexc ) 560 nm, 108 mJ/pulse.
medium facilitates the formation of EV+ triplets by providing
a highly viscous environment, but the formation of the semire-
duced EV‚ dye radical is minimized in this medium because
sucrose octaacetate is a poor electron/H atom donor. A transient
species displaying a broad absorption band in the 320-520 nm
region and also displaying absorption beyond 620 nm has been
observed before for EV+ noncovalently bound to poly-
(methaacrylic acid) in aqueous solution and postulated as being
triplet EV+.10 The contribution of a transient species displaying
a broad absorption band in the 320-520 nm range along with
significant absorption beyond 620 nm is clearly noticed in the
transient spectra of BSA-bound EV+, especially in the case of
nitrogen-purged samples and at early reaction times (Figure 5).
To assess the spectral fingerprint of the semireduced EV•
radical, a mixture containing EV+ and benzophenone in
acetonitrile was photolyzed at 355 nm under conditions in which
the excitation light was overwhelmingly absorbed by benzophe-
none. The quenching of triplet benzophenone (3BP*) by a series
of triarylmethane cations, including EV+ and CV+, occurs at
diffusion-controlled rates in acetonitrile.60 Triplet benzophenone
and other triplet carbonyls (3n,π* states) are regularly quenched
in solution through triplet-triplet energy-transfer processes or
via electron- or hydrogen atom-transfer reactions involving
molecules of solvent or other substrates present in solution.46
The reductive electron-transfer quenching process, namely the
electron or H atom transfer from a substrate to 3BP*, produces
the biphenyl ketyl radical (BPH•) and the corresponding
electron-deficient substrate radical. The oxidative quenching
of 3BP* generates the benzophenone cation radical (BP•+) and
the reduced substrate radical. Oxidative quenching has been
3
postulated as the dominant mechanism of quenching of BP*
by CV+ in acetonitrile.56,59,61 As a consequence of the low
viscosity of the reaction medium, a significant fraction of the
initially formed CV• and BP•+ radicals diffuse away from each

Triarylmethane Dyes
J. Phys. Chem. B, Vol. 102, No. 23, 1998 4687
virtually identical transient spectrum was observed upon reduc-
tion of triplet EV+ with ascorbic acid, and this last spectrum is
also presumed to represent the semireduced EV• radical.10 This
hypothesis is supported by what has been observed for the case
of CV+ in the presence of benzophenone.56,59
The contribution of a species displaying absorption maximum
around 410 nm is clearly observed in the transient spectra of
BSA-bound EV+ (Figure 5). Therefore, the time-resolved
spectra observed upon laser photolysis of the BSA-EV+
complex show superimposed elements of the spectroscopic
signatures of both the EV+ triplet and the semireduced EV•
radical. The contribution of the triplets in broadening the
absorption envelop in the 320-520 nm region is more prominent
at early reaction times in nitrogen-purged samples, while the
contribution of the semireduced EV• radical is associated with
the contribution of the longer-lived species.
Figure 6. Transient absorption spectra for EV+ in glass of sucrose
octaacetate. [EV+] ) 11 µM, λexc ) 560 nm, 105 mJ/pulse.
It is reasonable to presume that the short-lived component (τ
∼ 40 ns) of the broad transient signal represents a population
of triplets that are efficiently quenched by electron-transfer
events to generate the semireduced EV• radical. Based on the
MALDI-TOF-MS data, the Cys62 and His67 BSA amino acid
residues were identified as the likely quenchers of this short-
lived triplet population. Because the transient decay traces
follow triple-exponential functions and because the longer-lived
transient shows the spectroscopic characteristics of the semire-
duced EV• radical, we hypothesize that the transient species
displaying the intermediate lifetime represent a second popula-
tion of EV+ triplets, possibly bound to the second (less reactive)
BSA binding site.
The time resolution of our laser flash photolysis equipment
does not permit any direct inference on the involvement of EV+
singlet populations in the formation of the semireduced EV•
radical. However, because the singlet population is generated
while the dye is physically attached to the same protein
environment that appears to be very efficient in quenching the
respective triplet population, it is conceivable that semireduced
EV• radicals may also be produced through the reduction of
the dye’s singlet. At the end of the laser pulse (ca. ap-
proximately 20 ns after the flash), significant absorption around
410 nm is already observed, but whether this early population
of semireduced EV• radicals are formed only via efficient EV+
triplet quenching or also through the dye’s singlet manifold
remains an open question.
Figure 7. Transient absorption spectra for a nitrogen-purged acetonitrile
solution containing BP (1.0 µM) and EV+ (20 µM). Spectra recorded
at 10 and 140 µs after the laser flash. The inset shows the respective
spectra recorded at 50 ns, 300 ns, and 3 µs after the flash. λexc ) 355
nm, 150 mJ/pulse.
other before the back electron transfer takes place.61 The
semireduced CV• radical displays a strong bell-shaped absorp-
tion band with a maximum at 405 nm.56 The formation of the
semireduced CV• radical upon laser excitation of benzophenone
in BP/CV+ mixtures in acetonitrile was confirmed by electron
spin resonance spectroscopy. Upon excitation of the reaction
mixture inside the resonator’s cavity, the appearance of the well-
known spectrum of the CV• radical (g ) 2.0033) was observed.56
Figure 7 shows the transient spectra obtained upon 355 nm
laser excitation of a nitrogen-purged solution of benzophenone
and EV+ in acetonitrile. In the absence of benzophenone no
transients were detected in the nanosecond to millisecond time
domain. The decay of benzophenone triplets occur concomi-
tantly with both the ground-state bleaching of EV+ and the
appearance of a new absorption band around 410 nm (Figure
7, inset). The EV+ ground-state bleaching can be easily detected
at short time intervals after the laser pulse, indicating that EV+
Concluding Remarks
The formation of noncovalent complexes of TAM dyes with
BSA has remarkable effects on the singlet lifetime, intersystem
crossing efficiency, and photoreactivity of these photosensitizers.
The laser-induced photobleaching of protein-bound triaryl-
methane dyes is initiated by an electron or hydrogen atom
transfer from the protein to the dye moiety. The striking
increase in triplet and semireduced dye radical yield caused by
the binding phenomena indicates that in the presence of oxygen
these photosensitizers are much more likely to engage in
photosensitized oxidation processes type I and type II when
bound to biopolymers than when free in solution. As suggested
by the oxygen quenching effect on the longer-lived ethyl violet
transient, we expect the type I process to be particularly favored
in the BSA environment. When free in low-viscosity media
the photoreactivity of TAM dyes is extremely poor, and no
significant photodynamic action should be expected from these
photosensitizers under such circumstances. Interestingly, the
photochemical efficiency of BSA-bound dye bleaching increases
in nitrogen-purged samples as compared to air-equilibrated
3
efficiently reacts with BP* (see small depletion around 600
nm in the upper curve shown in the inset of Figure 7). At 10
µs after the laser pulse 3BP* no longer contributes to the
observed transient spectra, and the well-defined bell-shaped
absorption band with maximum around 410 nm can be attributed
solely to the ethyl violet transient species. This species does
not show any significant absorption beyond 620 nm. Because
the spectral fingerprint of this transient is distinct from the one
obtained in glass of sucrose octaacetate, we hypothesize that
the transient spectra shown in Figure 7 (main panel) represents
the absorption spectra of the semireduced EV• radical.
A

4688 J. Phys. Chem. B, Vol. 102, No. 23, 1998
Baptista and Indig
(25) Blanchard, J.; Fink, W. T.; Duffy, J. P. J. Pharm. Sci. 1977, 66,
1470.
samples. Independent of the bleaching route that is dominant
under anaerobic conditions (presumably, sequential two-electron
abstraction from the protein or one-electron abstraction followed
by the formation of protein-dye covalent adducts), upon laser
excitation of protein-TAM complexes the photoinduced dam-
age of the host protein proceeds via reaction paths that do not
require oxygen to operate, and consequently are not of the
ordinary photodynamic type.5 The formation of BSA-TAM
covalent adducts and BSA fragmentation represent peculiar
oxygen-independent reaction courses that are well-suited for the
treatment of hypoxic or poorly perfused tumor areas. The
mechanistic steps of BSA-bound dye bleaching were found to
be very similar between EV+ and CV+, suggesting that these
oxygen-independent reaction paths may be rather common when
considering the photochemistry of cationic photosensitizers
noncovalently bound to biopolymers. TAM dyes are photo-
sensitizers known to accumulate in the mitochondria and disrupt
mitochondrial activity.14,62 Therefore, it is likely that their
phototoxicity develops upon noncovalent binding to mitochon-
drial proteins or other mitochondrial biopolymers. Since a large
variety of families of extensively conjugated cationic dyes are
known to naturally accumulate in the mitochondria,63 as well
as to bind very efficiently to natural and synthetic anionic
biopolymer polyelectrolytes,6-11 the oxygen-independent pro-
cesses identified here for the case of TAM dyes are of relevance
for the design of new photosensitizers specifically tailored for
anaerobic phototherapy.
(26) Klotz, M. L.; Huston D. L. Trends Pharmacol. Sci. 1988, 4, 253.
(27) Hamai, S.; Hirayama, F. J. Phys. Chem. 1983, 87, 83-89.
(28) Scaiano, J. C. J. Am. Chem. Soc. 1980, 102, 7747.
(29) Liao, Y.; Bohne, C. J. Phys. Chem. 1996, 100, 734.
(30) Kavalauskas, M. P.; Indig, G. L. To be published.
(31) Sundstro¨m, V.; Gillbro, T.; Bergstro¨m, H.Chem. Phys. 1982, 73,
439.
(32) Vogel, M.; Rettig, W. Ber. Bunsen-Ges. Phys. Chem. 1985, 89,
962-968.
(33) Ben-Amotz, D.; Harris, C. B. Chem. Phys. Lett. 1985, 119, 305.
(34) Kemnitz, K.; Yoshihara, K. Chem. Lett. 1990, 1789.
(35) Peters, T., Jr. In All About Albumin. Biochemistry, Genetics, and
Medical Applications; Academic Press: New York, 1996.
(36) Viswanath, D. S.; Natarajan, G. In Data Book of the Viscosity of
Liquids; Hemisphere: New York, 1989.
(37) Gray, D. E. In American Institute of Physics Handbook, 3rd ed.;
McGraw-Hill: New York, 1972.
(38) Das, R.; Mitra, S.; Nath, D.; Mukherjee, S. J. Phys. Chem. 1996,
100, 14514.
(39) The formation of dye aggregates under conditions of high BSA
loading ([TAM] . [BSA]) may also contribute to the biphasic character
of the Scatchard plots. However, the spectral evidence for the formation of
dye aggregates is extremely faint, even when considering the highest BSA
loading employed in this analysis. Under conditions of high [TAM]/[BSA]
ratios the most pronounced changes in the TAM’s absorption spectra induced
by BSA binding are observed in the Amax/Ashoulder ratio (represented by the
A
595/A550 ratio in the case of EV+). The maximum change observed in this
spectral parameter over the course of the Scatchard analysis was ap-
proximately 5%, and it would be very difficult to determine unambiguously
the contributions of dye aggregate formation and dye monomer-BSA
complex formation, respectively, to this small change. Further discusion
on this subject can be found in ref 44.
(40) Lewis, G. N.; Magel, T. T.; Lipkin, D. J. Am. Chem. Soc. 1942,
64, 1774.
(41) Ishikawa, M.; Maruyama, Y. Chem Phys. Lett. 1994, 219, 416.
(42) Maruyama, Y.; Ishikawa, M.; Satozono, H. J. Am. Chem. Soc. 1996,
118, 6257.
(43) Pal, M. K.; Ghosh, J. K. Spectrochim. Acta 1994, 50A, 119.
(44) Duxbury, D. F. Chem. ReV. 1993, 93, 381.
(45) Hirayama, N.; Akashi, S.; Furuya, M.; Fukuhara, K. Biochem.
Biophys. Res. Commun. 1990, 173, 639.
Acknowledgment. This work was supported by the Ameri-
can Cancer Society (Grant No. IRG-35-37-3) and by the
Burroughs Wellcome Fund-American Foundation of Pharma-
ceutical Education (AACP Grant Program for New Investiga-
tors). The University of Wisconsin at Madison is also acknowl-
edged for startup funds and continuous financial support. The
authors thank Dr. T. D. Heath for editorial assistance, and Ms.
J. Grant for her help with mass spectrometry measurements.
(46) Gilbert, A.; Baggott, J. In Essentials of Molecular Photochemistry;
Blackwell Publications: London, 1991.
(47) Kuramoto, N.; Kitao, T. Dyes Pigm. 1982, 3, 49.
(48) Denman, S.; Jameel, S.; Hay, J.; Sugden, J. K. Dyes Pigm. 1996,
30, 67.
(49) Fischer, V.; Harrelson, W. G., Jr.; Chignell, C. F.; Mason, R. P.
Photobiochem. Photobiophys. 1984, 7, 111.
(50) The back electron transfer from the semireduced dye radical to the
protein backbone is another reaction route capable of protecting the TAM
dyes from photobleaching.
(51) Muddiman, D. C.; Bakhtiar, R.; Hofstadler, S. A.; Smith, R. D. J.
Chem. Educ. 1997, 74, 1288.
(52) Hillenkamp, F. AdV. Mass. Spectrom. 1995, 13, 95.
(53) Weinberger, S. R. Spectroscopy 1992, 7, 54.
(54) Martin, M. M.; Plaza, P.; Meyer, Y. H. Chem. Phys. 1991, 153,
297.
(55) Jones, G. II; Goswami, K. J. Phys. Chem. 1986, 90, 5414.
(56) Jockusch, S.; Timpe, H. J.; Fischer, C.-H.; Schnabel, W. J.
Photochem. Photobiol. A: Chem. 1992, 63, 217.
(57) Bhasikuttan, A. C.; Shastri, L. V.; Sapre, A. V.; Rama Rao, K. S.
V.; Mittal J. P. J. Photochem. Photobiol. A: Chem. 1994, 84, 237.
(58) Bhasikuttan, A. C.; Sapre, A. V.; Rama Rao, K. S. V.; Mittal, J. P.
Photochem. Photobiol. 1995, 62, 245.
(59) Jockusch, S.; Timpe, H. J.; Schnabel, W.; Turro, N. J. J. Photochem.
Photobiol. A: Chem. 1996, 96, 129.
(60) Naguib, Y. M. A.; Steel, C.; Cohen, S. G.; Young, M. A. J.
Photochem. Photobiol. A: Chem. 1996, 96, 149.
(61) In addition to the 3BP* oxidative quenching by CV+, other routes
of formation of the semireduced CV• radical are likely to operate in this
system. The free radicals generated upon H atom abstraction from
acetonitrile or CV+ molecules by triplet benzophenone can also promote
the reduction of the crystal violet cation to its semireduced radical. Further
discussion on this mechanistic route can be found in refs 56 and 57.
(62) Gadelha, F. R.; Moreno, S. N. J.; De Souza, W.; Cruz, F. S.;
DoCampo, R. Mol. Biochem. Parasitol. 1989, 34, 117.
(63) Chen, L. B. Methods Cell. Biol. 1989, 29, 103.
References and Notes
(1) Henderson, B. W.; Dougherty, T. J. Photochem. Photobiol. 1992,
55, 145.
(2) Levy, J. G.; Obochi, M. Photochem. Photobiol. 1996, 64, 737.
(3) Dougherty, T. J. Photochem. Photobiol. 1993, 58, 895.
(4) Kessel, D. Drugs Today 1996, 32, 385.
(5) Foote C. S. Science 1968, 162, 963.
(6) Oster, G.; Bellin, J. S. J. Am. Chem. Soc. 1957, 79, 294.
(7) Bellin, J. S.; Yankus, C. A. Arch. Biochem. Biophys. 1968, 123,
18.
(8) Yariv, S.; Ghosh, D. K.; Hepler, L. G. J. Chem. Soc., Faraday
Trans. 1991, 87, 1201.
(9) Pal, M. K.; Ghosh J. K. Spectrochim. Acta 1994, 50A, 119.
(10) Jones, G. II; Oh, C.; Goswami, K. J. Photochem. Photobiol. A:
Chem. 1991, 57, 65.
(11) Jones, G. II; Rahman, M. A. J. Phys. Chem. 1994, 98, 13028.
(12) Cilento, G. Experientia 1988, 44, 572.
(13) Cilento, G.; Adam, W. Free Radicals Biol. Med. 1995, 19, 103.
(14) Fiedorowicz, M.; Pituch-Noworolska, A.; Zembala, M. Photochem.
Photobiol. 1997, 65, 855.
(15) Viola, A.; Hadjur, C.; Jeunet, A.; Julliard, M. J. Photochem.
Photobiol. B: Biol. 1996, 32, 49.
(16) Hatchard, C. G.; Parker, C. A. Proc. R. Soc. London, Ser. A 1956,
235, 518-536.
(17) Petersen, W. C. U.S. Pat. 4,394,314, 1983.
(18) Melhuish, W. H. J. Phys. Chem. 1961, 65, 229.
(19) Meech, S. R.; Phillips, D. J. Photochem. 1983, 23, 193.
(20) Demas, J. N.; Crosby, G. A. J. Phys. Chem. 1971, 75, 991-1024.
(21) Grinvald, A.; Steinberg, I. Z. Anal. Biochem. 1974, 59, 583.
(22) Knutson, J. R.; Breechem, J. M.; Brand, L. Chem. Phys. Lett. 1983,
102, 501.
(23) Scatchard, G. Ann. N. Y. Acad. Sci. 1949, 51, 600.
(24) Naik, D. V.; Paul, W. L.; Threatte, R. M.; Schulman, S. G. Anal.
Chem. 1975, 47, 267-270.