Photoprocesses of naphthalene imide and diimide derivatives in aqueous solutions of DNA

Article abstract of DOI:10.1021/ja992332d

Despite the growing number of redox-active chromophores utilized to photoinduce oligonucleotide cleavage, detailed correlations between the degree of ground-state complexation and product yields have not been developed. To elucidate the specific role of s

Full text of DOI:10.1021/ja992332d

J. Am. Chem. Soc. 2000, 122, 427-436
427
Articles
Photoprocesses of Naphthalene Imide and Diimide Derivatives in
Aqueous Solutions of DNA
Joy E. Rogers, Sharon J. Weiss, and Lisa A. Kelly*
Contribution from the Department of Chemistry and Biochemistry, UniVersity of Maryland,
Baltimore County, 1000 Hilltop Circle, Baltimore, Maryland 21250
ReceiVed July 6, 1999. ReVised Manuscript ReceiVed NoVember 17, 1999
Abstract: Despite the growing number of redox-active chromophores utilized to photoinduce oligonucleotide
cleavage, detailed correlations between the degree of ground-state complexation and product yields have not
been developed. To elucidate the specific role of singlet and triplet excited states in nucleotide photooxidation,
the photochemical reactivities of N-(2-(N-pyridinium)ethyl)-1,8-naphthalene imide (NI) and N,N′-bis-[2-(N-
pyridinium)ethyl]-1,4,5,8-naphthalene diimide (NDI) with calf-thymus DNA have been explored as a function
of ground-state complexation with the DNA polymer. Upon addition of calf-thymus DNA to a phosphate
buffered solution of the naphthalene imide derivatives, distinct changes in the UV absorption spectrum of the
chromophores, along with single isosbestic points, are observed. Analysis of these changes using the
noncooperative model of McGhee and von Hippel yield association constants of (2.46 ( 0.42) × 104 M-1 and
(7.78 ( 0.11) × 105 M-1 for NI and NDI, respectively. Pulsed 355 nm excitation of either NI or NDI in the
presence of calf-thymus DNA produced the reduced NI•- and NDI•- species that absorbed maximally at 400
and 480 nm, respectively, from the triplet excited states. For both compounds, the yield of radical anion from
•-
self-quenching processes was substantial (φI ) 0.11 ( 0.01 and 0.25 ( 0.01 for NI and NDI, respectively).
However, pulsed excitation of NI in the presence of DNA resulted in the production of radical species that
were not attributed to self-quenching processes. For both compounds, the fraction of associated imide was
systematically varied between 0 and 1. The intersystem crossing yield was found to decrease linearly with the
fraction bound to DNA from 0.71 to 0.08 for NI and 0.35 to 0.004 for NDI.
Introduction
versatility may be achieved by modification of a specific
chemical system to (a) favor site-specific association and
reactivity or (b) preferentially produce nonselective and diffus-
ible intermediates (e.g. hydroxyl radicals). The naphthalene-
derived imide and diimide chromophores have been shown to
be a class of chromophores capable of generating a multitude
of reactive intermediates.2-5 In a previous report, we have shown
that the triplet excited states of 1,8-naphthalimide and 1,4,5,8-
naphthaldiimide derivatives are photoreduced by the individual
nucleotides and that the kinetics of nucleotide oxidation are
tunable, over ∼2 orders of magnitude, depending upon the
nucleotide employed.6 This observation, coupled with the
prospects of readily functionalizing (on the imide nitrogen) the
naphthalene-derived imides with specific recognition groups,
makes them ideal candidates for sequence-specific photooxi-
dation and cleavage of oligonucleotide polymers.
The enormous interest in the development of chemical
systems as structural probes in biological macromolecules, as
well as therapeutic agents, has led to the design of a growing
number of studies on synthetic “reagents” that can be activated
by UV or visible light. The ability to deliver chemical reactivity
on demand has been used in photodynamic therapy, where the
localized interaction of a chromophore and light results in tumor
destruction. Recently, the development of synthetic chemical
systems that can photocleave nucleic acid or protein polymers
has emerged into a growing database of molecular systems
capable of photoinitiating specific or nonspecific lesions. This
work is largely driven by the desire to create molecular
chromophores as (a) nonsequence-specific reagents for utiliza-
tion in photofootprinting experiments, or (b) sequence-specific
structural probes and sequencing agents.
Of the chromophores employed, hydroxyl radical generators,
singlet oxygen generators, electron transfer agents, and excited
states capable of hydrogen atom abstraction from the nucleotide
ribose unit have been most studied. The multitude of synthetic
systems used for photoinitiated oligonucleotide cleavage has
been recently reviewed.1 Desirable chemical systems are those
that demonstrate a versatile range of photochemical reactivity
that can be tuned for specific or nonspecific interactions. This
Specifically, L-lysine derivatives of 1,8-naphthalimide have
been previously reported to nick both supercoiled DNA, as well
(2) Aveline, B. M.; Matsugo, S.; Redmond, R. W. J. Am. Chem. Soc.
1997, 119, 11785-11795.
(3) Matsugo, S.; Kawanishi, S.; Yamamoto, K.; Sugiyama, H.; Matsuura,
T.; Saito, I. Angew. Chem., Int. Ed. Engl. 1991, 30, 1351-1353.
(4) Saito, I.; Takayama, M.; Kawanishi, S. J. Am. Chem. Soc. 1995, 117,
5590-5591.
(5) Saito, I.; Takayama, M.; Sugiyama, H.; Nakatani, K.; Tsuchida, A.;
Yamamoto, M. J. Am. Chem. Soc. 1995, 117, 6406-6407.
(6) Rogers, J. E.; Kelly, L. A. J. Am. Chem. Soc. 1999, 121, 3854-
3861.
* To whom correspondence should be addressed.
(1) Armitage, B. Chem. ReV. 1998, 98, 1171-1200.
10.1021/ja992332d CCC: $19.00 © 2000 American Chemical Society
Published on Web 01/06/2000

428 J. Am. Chem. Soc., Vol. 122, No. 3, 2000
Rogers et al.
as double-stranded DNA fragments, upon UV irradiation and
treatment with hot piperidine.4 Specific cleavage at the 5′-side
of 5′-GG-3′ steps was observed. Upon nitration of the naph-
thalimide ring, photocleavage at thymine sites was also ob-
served. Variable selectivity between these two modes of
cleavage was obtained, depending upon the position of the nitro
substituent. In a subsequent publication, Saito et al. demonstrated
that irradiation of this L-lysine 1,8-naphthalimide derivative in
the presence of a duplex hexamer selectively produced piperi-
dine-labile damage at the 5′-side of 5′-GG-3′ sites.5 Using laser
flash photolysis, the one-electron reduced imide was directly
detected. Production of this species was via bimolecular
quenching of the naphthalimide triplet state. The quenching rate
constant was estimated to be 5.3 × 107 M-1 s-1. When the
photoinduced reaction is not base-specific and not reversible,
site-specific damage is rendered only when the chromophore
is tightly associated with the target of interest. This prerequisite
localizes the site of damage and results in rapid (since the
process is not diffusion controlled) photochemical reactions.
Moreover, rapid excited state quenching by the associated target
precludes the formation of diffusible species (e.g., singlet
oxygen). Although the naphthalimide derivatives are flat planar
species, likely to intercalate between DNA base pairs, no
evidence of ground-state complexation was indicated in the
reports of Saito et al.4,5
In a series of separate reports, the noncovalent interactions
of a series of N-alkylamine-substituted naphthalene imides with
DNA were assessed.7-10 Since the pendant alkylamine is
protonated in neutral aqueous solution, the binding constants
were shown to be strongly dependent upon ionic strength.9 For
a given DNA substrate, imide substituent, and ionic strength
(0.10 M NaCl), the binding constants for the naphthaldiimide
derivatives (K ) (3.0-4.6) × 105 M-1) were ∼10-fold larger
than those of the corresponding naphthalene monoimide.7
The ability of functionalized naphthalene imides and diimides
to strongly associate with DNA and, in certain cases, exhibit a
binding preference for GC-rich regions,10 makes them viable
candidates to explore the photoredox chemistry while associated
to DNA. A detailed understanding of the photochemical
mechanisms must first consider the nature of the excited states
responsible for damage. Under conditions where the chromo-
phore is not associated with the DNA polymer, photodamage
initiated by triplet states will predominate, owing to their long
lifetimes. However, upon association of the chromophore to the
target, rapid nucleotide oxidation by the singlet excited-state
becomes competitive. Triplet-state quenching of a naphthalene
imide by oligonucleotide duplexes has been demonstrated.5
However, the partitioning of singlet and triplet state reactivity,
while associated with the oligonucleotides, has not been
explored.
state remains unclear in cases where the chromophore is
associated with the DNA polymer. Moreover, the photoredox
activity of the excited states of the corresponding naphthaldi-
imide chromophore with DNA has not been characterized.
Although a number of organic and inorganic chemical systems
have been employed to photochemically cleave oligonucleotide
polymers, the photochemical efficiency has not been systemati-
cally correlated with extent of ground-state complexation. In
this paper, we (a) report the synthesis of a novel pair of cationic
naphthalene imide and diimide derivatives, (b) characterize the
noncovalent interactions of these two photoredox reagents with
calf-thymus DNA, and (c) demonstrate the first systematic
investigation of the relative roles of singlet and triplet excited
states in nucleotide oxidation.
Experimental Section
Materials. 2-Bromoethylamine hydrobromide, 1,4,5,8-naphthalene-
tetracarboxylic dianhydride, ethanolamine (99+ %), p-toluenesolfonyl
chloride (99+%), methyl viologen dichloride hydrate (98%), 1,4-
diazobicyclo [2.2.2]-octane (DABCO) (98%) and benzyl viologen
dichloride (97%) were used as received from Aldrich Chemicals
(Milwaukee, WI). 1,8-naphthalic anhydride (97%, Acros) was recrystal-
lized from N,N-dimethylacetamide (DMA) and dried in vacuo prior to
use. Sodium hydrogen phosphate (99%) was obtained from Acros and
used as received. Solutions of calf-thymus DNA (highly polymerized
sodium salt (Sigma, St. Louis, MO)) were sonicated at 25 °C for 1 h.
The resulting solution was filtered through a 0.45 µM Millipore filter.
DNA concentrations (in base pairs) were determined spectroscopically
using ꢀ ) 13 200 M-1 cm-1 at 260 nm.11 DNA stock solutions were
stored at 4 °C and discarded after one week.
Water was deionized and freshly passed through an Ion-Pure Reverse
Osmosis system. The system utilizes a point of use cartridge system,
followed by UV irradiation to provide > 18 MΩ ultrapure bacteria-
free water. Other materials were obtained from commercial sources.
N-(2-(N-Pyridinium)ethyl)-1,8-naphthalene Imide (NI, 1). Etha-
nolamine (5 mL) was added dropwise over 15 min to a solution of
1,8-naphthalic anhydride (3.46 g, 17.4 mmol) in 40 mL of DMA. The
mixture was heated at 100 °C for 2 h. The crude reaction mixture was
concentrated on a rotary evaporator,and the product washed with 95%
ethanol. The resulting N-(2-ethanol)-1,8-naphthalimide (3.16 g, 13.1
mmol) and p-toluenesulfonyl chloride (2.75 g, 14.4 mmol) were stirred
in pyridine at room temperature for 20 h and then refluxed for 5 h.
After conversion to the chloride salt (DOWEX 1 × 8-200 ion-exchange
resin), the crude product was recrystallized in methanol under an
atmosphere of acetone. 1H NMR (DMSO-d6): 9.20 (d, 2H, pyr), 8.60
(t, 1H, pyr), 8.40 (d+d, 4H, napht), 8.05 (t, 2H, pyr), 7.80 (t, 2H, napht),
4.95 (t, 2H, -CH2), 4.60 (t, 2H, -CH2). Anal. Calcd C, 67.36, H, 4.43,
N, 8.27. Found C, 66.57, H, 4.43, N, 8.15. UV max (10 mM phosphate
buffer; pH 7.00) 344 nm (ꢀ ) 13 500 M-1 cm-1); 264 nm (4800 M-1
cm-1); 232 nm (ꢀ ) 38 900 M-1 cm-1); 214 nm (ꢀ ) 21 700 M-1
cm-1).
N,N′-bis-[2-(N-Pyridinium)ethyl]-1,4,5,8-naphthalene Diimide (NDI,
2). Ethanolamine (10 mL) was added dropwise (over a period of 15
min) to a solution of 1,4,5,8-naphthalenetetracarboxylic dianhydride
(5.00 g, 18.6 mmol) in 40 mL N,N-dimethylformamide (DMF). The
reaction mixture was heated at 90 °C for 2 h. The resulting N,N′-(2-
ethanol)-1,4,5,8-naphthalene diimide (3) was filtered, and the precipitate
was washed with acetone. Compound (3) (2.3 g, 9.02 mmol) and
p-toluenesulfonyl chloride (3.78 g, 19.84 mmol) were stirred in pyridine
at room temperature for 20 h and then refluxed for 5 h. The product
was filtered, and the precipitate was washed with acetone and
recrystallized from methanol. 1H NMR (DMSO-d6): 9.24 (d, 4H, pyr),
8.61 (t + s, 6H, pyr + napht), 8.09 (t, 4H, pyr), 4.96 (t, 4H, -CH2),
4.62 (t, 4H, -CH2). Anal. Calcd (for NDI‚2H2O) C, 57.44, H, 4.44,
N, 9.57. Found C, 57.10, H, 4.41, N, 9.51. UV max (10 mM phosphate
buffer; pH 7.00): 382 nm (ꢀ ) 25 600 M-1 cm-1), 362 nm (ꢀ ) 21 000
M-1 cm-1), 260 nm (ꢀ ) 9720 M-1 cm-1), 234 nm (ꢀ ) 35 200 M-1
cm-1).
We have previously demonstrated that the triplet states of
naphthalene imide and diimide systems do oxidize individual
nucleotides.6 However, nucleotide oxidation in DNA polymers
by these imide excited states has not yet been demonstrated.
The reports by Saito et al. provide convincing evidence that
the triplet excited states of 1,8-naphthalimide derivatives does,
indeed, produce oxidized nucleotide, with subsequent oligo-
nucleotide cleavage. However, the role of the singlet excited
(7) Yen, S.-F.; Gabbay, E. J.; Wilson, W. D. Biochemistry 1982, 21,
2070-2076.
(8) Hopkins, H. P.; Stevenson, K. A.; Wilson, W. D. J. Solution Chem.
1986, 15, 563-579.
(9) Tanious, F. A.; Yen, S.-F.; Wilson, W. D. Biochemistry 1991, 30,
1813-1819.
(10) Liu, Z.-R.; Hecker, K. H.; Rill, R. L. J. Biomol. Struct. Dyn. 1996,
14, 331-339.
(11) Armitage, B.; Yu, C.; Devadoss, C.; Schuster, G. B. J. Am. Chem.
Soc. 1994, 116, 9847-9859.

Photoprocesses of ND and NDI DeriVatiVes
J. Am. Chem. Soc., Vol. 122, No. 3, 2000 429
Ground-State Complexation Studies. A stock solution of sonicated
calf thymus DNA (1-4 mM base pairs in H2O) was added in 10 µL
aliquots (not to exceed 200 µL) to 2.00 mL of NI or NDI solution
(10-30 µM in 10 mM phosphate buffer at pH ) 7.0). Ground-state
spectral changes were monitored as a function of added DNA. Spectral
changes were analyzed as described below at 344 and 354 nm for NI
or 362 and 382 nm for NDI. Data from a total of three (for NI) or five
(for NDI) experiments were analyzed.
The extinction coefficient of the free imide (ꢀf) was determined
through a Beer’s Law plot, and the extinction coefficient of the bound
complex (ꢀb) was estimated by dividing the limiting absorbance by the
imide concentration. From the spectral changes (∆A ) (Ao - A/Ao))
and extinction coefficient data, the binding density ν (concentration
of bound imide/total DNA concentration) was calculated according to
eq 1.12
Figure 1. Structures of NI (1) and NDI (2) naphthalene imide
derivatives employed in this work.
are the structures of the naphthalene imide (NI, 1) and diimide
(NDI, 2) derivatives employed in this study. The pyridinium
derivatives of the chromophores were prepared to provide water
soluble photooxidizing agents that can noncovalently associate
to double stranded DNA, either by an intercalative or electro-
static mode. In both cases, the ground-state absorption spectra
of NI and NDI exhibit the S0 - S1 vibronic progression at
wavelengths longer than 300 nm. Higher energy π-π* transi-
tions are observed at wavelengths shorter than 250 nm. The
one-electron reduction potentials were determined using cyclic
voltammetry. The half-wave potentials for the reduction of NI
and NDI were determined to be -0.84 and -0.11 (V vs NHE),
respectively.
Upon 355 nm nanosecond pulsed excitation of a deaerated
solution of 1 or 2, ground-state bleaching is observed concomi-
tant with the formation of transients absorbing at longer
wavelengths (Figure 2). In both cases, the transients were
quenched by molecular oxygen. The transient absorption spectra
are identical to the T1 - Tn spectra reported previously for other
naphthalene imide and naphthalene diimide derivatives.2,6 At
low excitation energies, the triplet decay kinetics are first-order.
At higher excitation energies, triplet-triplet annihilation is
observed. Moreover, at high NI or NDI concentration, self-
quenching was observed (kself ≈ 1.2 × 109 M-1 s-1 and 2 ×
109 M-1 s-1 for NI and NDI, respectively). Under conditions
of low excitation intensities and small ground-state concentra-
tions, the triplet state lifetimes of NI and NDI were found to be
162 and 100 µs, respectively in deoxygenated 10 mM pH 7.00
phosphate buffer (Figure 2 insets). The triplet-state lifetimes in
aqueous buffer solutions were somewhat longer than those
reported previously for neutral NI and NDI excited states in
1:1 CH3CN:H2O.6
LT
ꢀb - ꢀf
∆A
)
ν
(1)
(M ) (
)
ꢀf
T
where LT and MT are the total concentrations of imide and DNA,
respectively. The concentration of “free” imide (Lf) was calculated
according to eq 2.
Lf ) LT - νMT
(2)
For each wavelength that was analyzed, plots of ν/Lf vs ν were
constructed, and the curves were fit to the noncooperative model of
McGhee and von Hippel.13 Values of K (equilibrium constant in M-1
)
and n (number of base pairs associated with each bound imide) were
determined by a nonlinear least-squares regression. The average and
standard deviation for each set of data were calculated. At each imide/
DNA ratio employed, the fraction of imide molecules not associated
with the DNA polymer (f1 ) Lf/LT) was evaluated from the ground-
state spectral changes associated with the DNA addition.
General Techniques. Laser flash photolysis kinetic investigations
were carried out at 22.0 ( 0.2 °C using a water-jacketed sample holder
connected to a Lauda RM6-B circulating water bath. Ground-state UV/
vis absorption spectra were measured using a JASCO V-570 double-
beam spectrophotometer. Proton NMR spectra were obtained using
either a GE QE-300 or a Varian Mercury 200 MHz NMR spectrometer.
Fluorescence spectra were measured using a SPEX Fluoromax-2
fluorescence spectrometer. Excitation wavelengths of 368 nm and 390
nm were used for NI and NDI, respectively.
Nanosecond transient absorption measurements employed the tech-
nique of laser flash photolysis. The third-harmonic (355 nm) of a
Q-switched Nd:YAG laser (Continuum Surelight II, pulse width ∼8
ns) was used for laser flash excitation. Pulse energies of up to 25 mJ
cm-2 pulse-1 were typically employed. A detailed description of the
laser flash photolysis apparatus has been previously published.6
One-electron reduction potentials of the imide and diimide were
measured in 0.10 M Na2SO4 aqueous solution containing ∼5 mM imide.
Solutions were bubbled with nitrogen prior to measurement. The cyclic
voltammograms were obtained using a BAS CV-1B CV controller that
was interfaced to a pentium PC for data acquisition. For these
measurements, platinum auxiliary and glassy-carbon working electrodes
were employed, with a silver/silver chloride (approximately 3 M KCl)
reference electrode. The electrochemical half-wave potentials of the
imide, measured with respect to the Ag/AgCl reference electrode, were
Spectra of Reduced NI and NDI. To characterize the UV/
vis spectrum of the one-electron reduced form of the naphthalene
imide (1) and diimide (2) derivatives, DABCO was used as a
reductive quencher of the imide and diimide triplet states. Under
conditions where the imide triplet state is exclusively and quan-
titatively quenched by DABCO, reduced NI or NDI is observed
concomitant with triplet-state decay. Production of NI•- or
NDI•- in the presence of a secondary electron acceptor resulted
in the growth of the reduced acceptor.15 In these investigations,
methyl viologen and benzyl viologen were employed as secon-
dary electron acceptors from NI•- and NDI•-, respectively. Since
the one-electron reduction potential of methyl viologen in water
converted to a NHE reference using E° ) 0.36 V vs NHE for the
1/2
Fe(CN)63-/4- couple in ferrocyanide.14 A scan rate of 100mV/sec was
typically employed.
Results
(E° ) -0.450 V vs NHE)16 is more negative than that for the
1/2
Photophysical and Redox Properties. Shown in Figure 1
NDI/NDI•- couple, benzyl viologen was used as a better elec-
(12) Lohman, T.; Bujalowski, W. Thermodynamic Methods for Model-
Independent Determination of Equilibrium Binding Isotherms for Protein-
DNA Interactions: Spectroscopic Approaches to Monitor Binding. In
Methods in Enzymology; Sauer, R. T., Ed.; Academic Press: New York,
1991; Vol. 208, pp 258-290.
(13) McGhee, J. D.; von Hippel, P. H. J. Mol. Biol. 1974, 86, 469-489.
(14) Sawyer, D. T.; Sobkowiak, A.; Roberts, J. L. Electrochemistry for
Chemists, 2nd ed.; Wiley-Interscience: New York, 1995; p 27.
(15) In these experiments, the concentration of the secondary electron
acceptor was kept small enough (50 µM for methyl and benzyl viologen)
in relation to DABCO ([DABCO] ) 6.2 mM and 10 mM for NI and NDI,
respectively) to preclude direct quenching of the imide triplet state by the
bipyridine derivatives. Under these conditions, the imide radical is
quantitatively “scavenged” by the viologen.
(16) Wardman, P. J. Phys. Chem. Ref. Data 1989, 18, 1637-1723.

430 J. Am. Chem. Soc., Vol. 122, No. 3, 2000
Rogers et al.
Figure 3. Transient absorption spectra of imide radical anions observed
upon 355 nm excitation of (a) 5 µM NI containing 520 µM DABCO
and (b) 11 µM NDI containing 1.67 mM DABCO. All solutions were
argon-saturated and contain 10 mM pH 7.0 phosphate buffer.
Figure 2. T1 - Tn absorption spectra observed after pulsed (8 ns pulse)
355 nm excitation of (a) NI ([NI] ) 5 µM) and (b) NDI ([NDI] ) 1
µM) in 10 mM phosphate buffer (pH 7.0) under argon-saturated
conditions. Times shown are (a) 5, 50, and 100 µs after the laser pulse
and (b) 0, 50, 100, and 300 µs after the laser pulse. Insets: Triplet
decay traces with first-order fits measured at (a) 480 nm and (b) 400
nm.
∆A •- ) ∆ꢀ •-[3I*] f φ •-
(3)
I
I
o q I
•-
In eq 3, ∆AI is the transient absorbance from the one-electron
reduced imide, obtained by extrapolating the decay component
•-
tron-accepting agent (E° ) -0.360 V vs NHE).16 The
to time zero, ∆ꢀI is the extinction coefficient of the radical,
1/2
[3I*]o is the triplet state concentration produced by the laser
pulse,19 and fq ) (kself[I]/kself[I] + kd) is the fraction of triplet
states quenched by the ground state of the imide. The absorption
change observed at 400 and 480 nm, respectively, for NI and
NDI was plotted as a function of [3I*]o fq. Fits of the data to eq
concentration of reduced methyl or benzyl viologen was deter-
mined from the absorbance change (extrapolated to time zero)
at 395 or 603 nm, respectively.17,18 The extinction coefficients
of the imide radical anions were then determined from their
respective absorption changes (at time zero after the laser pulse)
following quenching of the triplet precursor by DABCO. The
spectra of NI•- and NDI•- are shown in Figure 3.
•-
3 (see Supporting Information) yield φI of 0.11 ( 0.01 and
0.25 ( 0.01 for NI and NDI, respectively.
Ground-State Interactions with Calf Thymus DNA. Upon
addition of calf-thymus DNA to a buffered solution of NI or
NDI, a decrease in the long-wavelength absorption band
intensity and slight red-shift are observed (Figure 4). In a and
b of Figure 4, a single clean isosbestic point is observed at all
imide/DNA ratios, suggesting that only one spectrally distinct
imide/DNA complex is present. The spectral changes were
analyzed and fit to the noncooperative model of McGhee and
von Hippel,13 to obtain the binding constant and average number
of occupied sites in the DNA polymer. Analysis of the observed
spectral changes with added calf thymus DNA yielded binding
constants of (2.46 ( 0.42) × 104 M-1 and (7.78 ( 0.11) × 105
M-1 for NI and NDI, respectively. In both cases, the average
occupied site size was determined to be 2.7 base pairs.
In both cases, a ∼60% hypochromicity20 is observed upon
complexation to DNA. Moreover, a 5-10 nm red-shift is
Redox Products Produced from Self-Quenching. To elu-
cidate the products from the self-quenching process, transient
absorption spectra were measured under conditions where
3
3
significant quenching of NI* and NDI* by their respective
ground states does occur. The transient spectral features of the
products are identical with those of NI•- and NDI•-. Our
observation of single-electron transfer from self-quenching is
consistent with that reported by Aveline et al. when a naph-
thalene diimide derivative was used.2 Although the oxidized
species are necessarily produced, the similarity of the self-
quenching spectra to those shown in Figure 3 suggests that these
species absorb weakly in the wavelength range that was probed.
Using the extinction coefficients of NI•- and NDI•- from
Figure 3, the yields of these radical anions produced from self-
•-
quenching (φI ) were determined from eq 3.
(17) Watanabe, T.; Honda, K. J. Phys. Chem. 1982, 86, 2617-2619
(19) The triplet-state concentration was evaluated from the T - T
absorption change at 480 nm for both NI and NDI.
(20) Percent hypochromicity calculated from the decrease in molar
extinction coefficient upon binding to DNA ((ꢀf - ꢀb/ꢀf)).
v•+
395nm
(ꢀ
) 42 100 M-1 cm-1).
(18) Tsukahara, K.; Wilkins, R. J. Am. Chem. Soc. 1985, 107, 2632-
2635 (ꢀ
v•+
) 14 000 M-1 cm-1).
395nm

Photoprocesses of ND and NDI DeriVatiVes
J. Am. Chem. Soc., Vol. 122, No. 3, 2000 431
Figure 5. (a) Transient absorption spectra observed after pulsed 355
nm excitation of 42 µM solution of NI in the presence of 202 µM calf
thymus DNA. Times shown are 0, 23, 53, and 152 µs after the laser
pulse (decreasing ∆A at 480 nm). Inset: Top trace shows exponential
growth of NI•- (monitored at 400 nm) in the presence of DNA. Bottom
trace shows the biexponential growth and decay of NI•- produced from
self-quenching (no added DNA). In this case, the concentration of NI
) 5.5 µM in buffer corresponds to the concentration of “free” imide
when DNA is present. (b) Transient absorption spectra observed after
pulsed 355 nm excitation of 30 µM solution of NDI in the presence of
30 µM calf thymus DNA. Times shown are 0, 50, and 300 µs after the
excitation pulse (decreasing ∆A at 485 nm). From the measured ground-
state absorption spectra, along with eq 1, data shown in (a) and (b)
correspond to 87 and 46% of NI and NDI, respectively, associated with
the DNA All solutions are argon-saturated and in 10 mM phosphate
buffer (pH 7.0). Inset: Biexponential formation and decay of NDI•-
(monitored at 430 nm) in the presence of DNA (top trace). Bottom
trace shows the biexponential formation and decay of NDI•- produced
from self-quenching (no added DNA). In this case, the concentration
of NDI ) 16 µM and corresponds to the concentration of “free” imide
when DNA is present.
Figure 4. Ground-state spectral changes observed upon addition of
calf thymus DNA to (a) 32 µM NI (0-340 µM DNA) and (b) 21 µM
NDI (0-110 µM DNA) in 10 mM pH 7.0 phosphate buffer. Insets:
Data from (a) 344 nm and (b) 382 nm analyzed according to eqs 1 and
2 and fitted to the noncooperative model of McGhee and von Hippel.
observed for both compounds. Both observations are qualita-
tively consistent with those observed for intercalating com-
pounds. Using cationic naphthalene imide and diimide deriva-
tives that are nearly identical to those employed in this work,
the mode of association with calf thymus DNA has been
previously investigated and found to be exclusively intercala-
tive.21 The ground-state spectral changes observed in our work
are identical with those reported by Yen et al.. Thus, we assume
that the primary mode of DNA interaction with NI and NDI is
predominantly that of intercalation.
Excited-State Interactions with Calf Thymus DNA. Upon
355 nm laser flash excitation of a dearated aqueous buffered
solution of NI (1) containing calf thymus DNA, a transient is
observed immediately after the laser pulse that absorbs maxi-
mally at 480 nm (Figure 5a). The spectral features of this
transient are consistent with those of the NI triplet state shown
in Figure 2a. With time, the triplet state evolves to produce a
second, long-lived transient. By comparison of the spectrum of
this species with that of the radical anion shown in Figure 3a,
we have identified this species as the one-electron reduced form
of NI.
observed could be arising from dynamic self-quenching.22 To
test this hypothesis, a buffered solution of NI was prepared at
a concentration corresponding to that of the nonassociated
(“free”) imide in the solution containing DNA. Shown in the
inset of Figure 5a is the absorption from NI•-, measured at 400
nm, exclusively from self-quenching. Under these conditions
(87% of the imide molecules are bound to DNA), it is clear
that more NI•- is produced when DNA is present. From the
kinetic trace shown in Figure 5a, it is apparent that NI•- decays
on a significantly slower time scale ( ∼1 ms). We have ruled
out the possibility that the observed NI•- is formed from bound
NI-forming triplet states by estimating the absorbance change
that would be observed if NI•- were forming by this process.
Using the extinction coefficient of NI•-, we estimate the
Since self-quenching of nonassociated 3NI* molecules by NI
does occur, we considered the possibility that the radical anion
(21) The mode of binding of cationic naphthalene imide and diimide
derivatives (the pendant groups on the imide nitrogens are a quaternary
alkylammonium substituents) to DNA has been reported in ref 7. Using a
combination of viscometric titrations, along with circular dichroism and
UV/vis spectroscopies, Yen et al. have shown that the compounds interact
strongly with DNA through intercalative interactions.
(22) We thank the reviewer for bringing this possibility to our attention.

432 J. Am. Chem. Soc., Vol. 122, No. 3, 2000
Rogers et al.
maximum concentration of NI•- formed in the presence of DNA
that does not come from self-quenching in “bulk” aqueous
solution is 0.5 µM. Given the concentration of NI not associated
with DNA (5.5 µM), the self-quenching rate constant (1.4 ×
109 M-1 s-1), and the triplet state lifetime in the absence of
quenching (162 µs), we estimate that, in the absence of any
quenching by DNA, 56% of the triplet states are quenched by
this process. With an estimate of the concentration of bound
triplet states (2 µM), and the cage escape efficiency following
self-quenching (0.11), we estimate that the maximum concen-
tration of NI•- that can be formed from self-quenching is 0.1
µM.
At lower concentrations of DNA, where the concentration
of naphthalimide in the bulk aqueous phase is higher, NI•-
appears to be produced exclusively from self-quenching. A
detailed analysis of the partitioning of NI•- produced from self-
quenching only (no added DNA) and from nucleotide oxidation
has been carried out. From our study (results shown in Figure
2a of the Supporting Information), we conclude that when the
concentration of nonassociated or “free” NI is less than ∼10
µM, 50-100% (depending on the [NI]) of the total NI•-
observed by laser flash photolysis arises from electron transfer
from DNA.23 At higher concentrations (low ν (with DNA) or
[NI] (in the absence of DNA)), self-quenching products are
dominant and imide radical anion produced from DNA oxidation
cannot be discerned.
ation increases. Under conditions where NI is predominantly
bound, 3NI* is clearly observed as the sole transient immediately
after the 8 ns laser pulse (Figure 5a). The spectrum evolves to
that characteristic of the one-electron reduced NI. This result is
initially surprising, given the fact that diffusion is not required
prior to the electron transfer process. However, we have
previously shown that 3NI* reacts primarily with isolated
guanine nucleobases, and with a rate constant that is nearly 2
orders of magnitude slower than that expected for a diffusion-
controlled reaction.6 Thus, our observation of slow electron
transfer within the NI/DNA complex is not surprising. Although
the analogous experiment was carried out with NDI, it is
apparent that NDI•- produced comes predominantly from self-
quenching processes in the bulk aqueous solution.
To investigate the role of the singlet excited state in nucleotide
oxidation, fluorescence emission spectra were recorded as calf
thymus DNA was titrated into pH 7.0 phosphate buffered
solution (10 mM) of NI and NDI. In all cases, excitation was
at the isosbestic point in the ground-state absorption spectrum
(368 and 390 nm for NI and NDI, respectively). As calf thymus
DNA was added to a buffered solution of NI, the fluorescence
emission at 395 nm was diminished by a factor of 29 when
99% of the chromophrore was associated to the DNA. In an
analogous experiment, the fluorescence intensity at 412 nm of
NDI was also strongly quenched by a factor of 270 (99% of
NDI bound to DNA) with the addition of DNA. In both cases,
emission spectral shifts were not observed.
In the range of DNA concentrations where self-quenching is
minimized, the observed rate constants for 3NI* decay or NI•-
growth, measured at 480 and 400 nm, respectively, were, within
experimental error, independent of the concentration of CT DNA
base pairs added. Due to our limitations in the range of DNA
concentrations that could be employed (at low DNA concentra-
tions, self-quenching dominates), this is not surprising.
The identical experiments were carried with NDI (2). In the
presence of DNA, a transient was observed immediately after
pulsed excitation (Figure 5b). By comparison with Figure 2b,
we assign this transient to the triplet state of the chromophore.
This transient evolves to a species that has a spectrum identical
to that of NDI•- shown in Figure 3b. To assess the role of self-
quenching in production of this radical, we systematically varied
the concentration of nonassociated NDI by DNA addition. At
each addition, the amount of NDI•- produced from self-
quenching was determined by pulsed excitation of an aqueous
buffered solution of NDI only. The concentration of NDI in
this solution corresponded to that in the bulk aqueous phase of
the DNA-containing sample. Shown in the inset of Figure 5b,
the yield of long-lived NDI radical is identical whether DNA
is present. The concentration of NDI in the bulk aqueous phase
was varied by the addition of DNA. The results of this study
are shown in the Supporting Information. In all cases, the yield
of NDI•- with and without DNA were identical, suggesting that
the yield of NDI•- from redox interactions with DNA is
negligible compared to that from self-quenching.
Discussion
Nucleotide Oxidation by Naphthalimide Excited States.
The design of chemical systems as potential redox-active oligo-
nucleotide cleavage agents must consider the reduction and oxi-
dation potentials of the chromophore excited state and nucleotide
units, respectively. Steenken and Jovanovic have determined
the one-electron oxidation potentials of the G, A, T, and C
nucleosides at pH 7 using pulse radiolytic techniques.24 In order
of decreasing ease of oxidation, the reduction potentials of the
nucleoside radicals were determined to be 1.29 (G) < 1.42 (A)
< 1.6 (C) < 1.7 (T) (V vs NHE at pH 7). Moreover, recent
calculations have shown that -GG- steps are more easily oxi-
dized than single guanine residues, with the HOMO localized
almost exclusively on the 5′-side of the step.25 Using the oxi-
dation potentials of the individual nucleosides as estimates in
DNA polymers,26 it can be concluded that nucleotide oxidation
will be exergonic when initiated by excited states whose
reduction potential is more positive than ∼1.3-1.7 V vs NHE.
Using the half-wave reduction potentials of 1 and 2 determined
in this work (see above), along with the zero-zero excitation
energies of the T1 and S1 states of these chromophores,6 we
estimate the potentials to be (a) 1.5 and 2.6 V (vs NHE in
aqueous solution) for reduction of 3NI* and 1NI*, respectively
and (b) 1.9 and 3.1 V (vs NHE in aqueous solution) for reduction
3
1
of NDI* and NDI*, respectively. Thus, we anticipate that
forward electron transfer from all of the nucleotide bases to the
singlet excited state of NI and NDI will be exergonic by at least
1 eV. Likewise, forward electron transfer from some of the
nucleotide bases to the triplet excited states of NI and NDI will
be only slightly exergonic. Redox tunability offers the prospect
As additional amounts of DNA are added, the degree of
imide/DNA complexation is increased. The yield of photoin-
duced transients was elucidated under varying degrees of
complexation. For both imides, the magnitude of the nanosecond
laser-induced changes associated with (a) ground-state bleaching
at 340 and 380 nm (for NI and NDI, respectively) and (b) triplet-
state absorbance at 480 and 485 nm (for NI and NDI,
respectively) is significantly reduced as the degree of complex-
(24) Steenken, S.; Jovanovic, S. V. J. Am. Chem. Soc. 1997, 119, 617-
618.
(25) Sugiyama, H.; Saito, I. J. Am. Chem. Soc. 1996, 118, 7063-7068;
Prat, F.; Houk, K. N.; Foote, C. S. J. Am. Chem. Soc. 1998, 120, 845-846.
(26) Although the oxidation potentials for only the individual nucleosides
are known, they remain undetermined in DNA polymers, where ion solvation
is potentially reduced.
(23) This fraction refers to the fraction of NI•- produced in the presence
of DNA that is over and above that from self-quenching (see Supplemental
Information Figure 2). It does not imply anything about the fraction of triplet
states that are quenched.

Photoprocesses of ND and NDI DeriVatiVes
J. Am. Chem. Soc., Vol. 122, No. 3, 2000 433
of targeting the initial photochemistry to specific bases or subsets
of bases. However, site-specific cleavage requires that the
thermal events leading to strand cleavage can compete ef-
fectively with other events, such as “hole-hopping,” that can
rapidly move the oxidative damage to a thermodynamic sink,
such as guanine nucleobases. Several recent reports have
demonstrated that oxidative damage does indeed occur at sites
far removed from the site of initial electron transfer.1,27-29 These
reports support hole-hopping and/or long-distance interactions
to damage duplex DNA at remote guanine sites.
employed in Figure 5a (202 µM). We estimate this rate constant
to be ∼2 × 107 M-1 s-1 31
.
Saito et al. have reported the
bimolecular quenching of the triplet state of an L-lysine
derivative of 1,8-naphthalimide by duplex DNA.5 The value
estimated by us is consistent with the value of 5.3 × 107 M-1
s-1 reported by these workers. However, in our investigations,
the kinetics cannot be studied over a widely varying range of
nucleotide concentrations to conclusively show that the process
is a diffusional one. From our previous kinetic studies with
individual nucleotides,6 it is clear that the NI triplet state
undergoes electron transfer almost exclusively with guanine
nucleotides since reactivity with the other nucleotides is more
than 30 times slower. Moreover, the guanine reactivity is more
than 2 orders of magnitude slower than that expected for a
diffusion-controlled process. Since our data do not conclusively
show that a dynamic process is occurring, the possibility of slow
reactivity of a bound triplet state by a nearby guanine residue
cannot be ruled out.
From a comparison of the transient spectra of NI•- in Figure
3 with the spectra measured upon laser flash excitation of NI
in the presence of calf thymus DNA (Figure 5a), it is apparent
that photoreduction of NI by the native DNA is occurring to
facilitate nucleotide oxidation. Since the one-electron reduction
3
potential of NI* is only slightly higher than the oxidation
potential of the purine bases (G and A), the most likely oxidative
target is expected to be guanine-rich regions. Thus, using
naphthalene imide triplet states, it will be difficult to discern
between nucleobase-specific and hole-hopping mechanisms.
Singlet- vs Triplet-State Electron Transfer. The role of
excited singlet and triplet states of the naphthalene imides in
DNA oxidation was elucidated using a combination of absorp-
tion, fluorescence, and transient absorption experiments. The
observation of significant fluorescence quenching as DNA is
titrated into aqueous buffered solutions of NI or NDI suggests
the availability of an efficient singlet-state quenching pathway
in the presence of DNA. From picosecond pump-probe kinetic
measurements, the lifetimes of the 1NI* and 1NDI* states have
been determined to be 2.4 ns and 280 ps, respectively.30 Thus,
at the DNA concentrations employed (up to 800 µM and 180
µM nucleotide base pairs for NI and NDI, respectively), dynamic
singlet-state quenching cannot be responsible for the observed
fluorescence quenching. Thus, we attribute the strongly dimin-
ished fluorescence intensity to rapid electron transfer from the
DNA nucleotides to the singlet excited state of NI or NDI within
the preformed imide/DNA ground-state complex. Energy trans-
fer from the singlet excited states of NI or NDI is not likely,
since their excited singlet-state energies (3.4 and 3.2 eV, respec-
tively6) are ∼1 eV lower than those of the nucleotide bases.
As shown in Figure 5a, the formation of reduced NI is
concomitant with triplet state decay. Although self-quenching
does occur and also produces NI•-, it is clear that the amount
of the radical species increases in the presence of DNA (Figure
5a inset). The results clearly suggest that nucleotide oxidation
can be facilitated by the naphthalene imide triplet state. In the
limited range of DNA concentrations (150-440 µM) where NI•-
is predominantly formed from nucleotide oxidation, the observed
rate of radical production is, within experimental error, inde-
pendent of DNA concentration. Our observation of the slow
formation (kobs ≈ 104 s-1) of NI•- from the triplet precursor
state does suggest that the process is a dynamic one resulting
from reaction of nonassociated triplet states with the nucleotides.
With the assumption that the process is dynamic, a bimolecular
rate constant can be estimated from the observed rate constant
of NI•- formation (kobs ) 104 s-1), the known triplet state
lifetime (kd ) 6.2 × 103 s-1), and the nucleotide concentration
The roles of singlet excited states of the chromophores in
nucleotide oxidation must also be considered when ground-state
complexes are present. If significant concentrations of long-
lived reduced imide or diimide were produced from rapid
singlet-state quenching within the imide/DNA complex, it would
be discernible in the post-nanosecond laser pulse spectrum.
Under conditions where NI and NDI are predominantly associ-
ated with the native DNA, the laser-induced T1 - Tn absorption
changes are substantially diminished (see below and Figure 6).
The detailed relationship between singlet states that are “stati-
cally” quenched, triplet states that are formed, and reduced imide
radicals produced is addressed below.
In attempts to time-resolve the reduced imide radical/oxidized
nucleotide charge-separated state, picosecond pump-probe
experiments were carried out. Aqueous buffered solutions of
compounds 1 or 2 with calf thymus DNA were prepared where
the imide or diimide was predominantly (>98%) bound to the
DNA polymer. Immediately after 355 nm excitation with a 30
ps laser pulse, only trace amounts of imide or diimide radical
anion were observed and persisted after 10 ns. Since charge-
separated states were not observed in picosecond pump-probe
experiments, forward charge transfer and recombination is
occurring on time scales significantly shorter than 30 ps.
Relative Production of Triplet States with Added DNA.
From (a) the decrease in apparent triplet quantum yield and (b)
significant fluorescence quenching observed upon the addition
of calf thymus DNA to an aqueous buffered solution of NI or
NDI, we propose that rapid electron transfer, to 1NI* and 1NDI*
from associated nucleotides of calf thymus DNA, is occurring.
The ground-state equilibrium and excited-state interactions with
calf thymus DNA are depicted in Scheme 1.
In Scheme 1, triplet excited states of the free imide (I) are
produced upon excitation and intersystem crossing (pathway
(a)). Near unity intersystem crossing yields (φISC) have been
previously reported for 1,8-naphthalimide and N,N-bis(2,2-
dimethylpropyl)-1,4,5,8- naphthaldiimide in deoxygenated aceto-
nitrile, but intersystem crossing of 1,8-naphthalimide is some-
what diminished in aqeuous solution due to hydrogen-bonding
to solvent.2,34 Using the T1 - Tn extinction coefficients of NI
(27) Candeias, L. P.; Steenken, S. J. Am. Chem. Soc. 1993, 115, 2437-
(31) Since the nucleotides in CT DNA are not homogeneously distributed,
this bimolecular rate constant represents only an “average” bimolecular rate
constant for quenching by the collection of nucleotides.
(32) (a) The concentration of imide excited states was determined by
comparative actinometry using benzophenone in deaerated acetonitrile (φISC
) 1 and ꢀT-T ) 6200 M-1 cm-1 at 520 nm);30b (b) Bensasson, R.; Gramain,
J. C. J. Chem. Soc., Faraday Trans. 1980, 76, 1801-1810.
(33) Brun, A. M.; Harriman, A. J. Am. Chem. Soc. 1991, 113, 8153-
8159.
2440.
(28) Meggers, E.; Michel-Beyerle, M. E.; Glese, B. J. Am. Chem. Soc.
1998, 120, 12950-12955.
(29) Barton, J. K. Pure Appl. Chem. 1998, 70, 873-879; Hall, D. B.;
Kelley, S. O.; Barton, J. K. Biochemistry 1998, 37, 15933-15940; Nunez,
M. E.; Hall, D. B.; Barton, J. K. Chem. Biol. 1999, 6, 85-97 and references
therein.
(30) Rogers and Kelly, unpublished data.

434 J. Am. Chem. Soc., Vol. 122, No. 3, 2000
Rogers et al.
Scheme 1. Ground-State Association of the Imide (I) and Calf Thymus DNAa
a
Pathways (a) and (b) show the formation of charge-separated states that are observed on the nanosecond laser flash photolysis time scale.
and NDI derivatives that were previously measured in 1:1 CH3-
CN:H2O,6 we estimate φISC to be 0.71 and 0.35 for NI and NDI,
respectively. Of the “free” triplet states formed via pathway (a),
a fraction of them will be diffusionally quenched to form
solvated ion pairs, and some of the ion pairs will diffusionally
separate, in competition with geminate recombination, to form
I•-. We have previously determined that the efficiency of I•-
production following quenching of NI and NDI triplet states
by the four individual nucleotides ranged from 19 to 37%,
depending upon the imide and nucleotide employed.
At varying concentrations of calf thymus DNA, a fraction of
the imide chromophores becomes associated with the DNA
polymer and singlet-state quenching within the complex may
occur. Since this process is in competition with intersystem
crossing within the excited-state manifold of DNA-bound imide,
the apparent intersystem crossing yield (φAPP) is reduced.
Moreover, in the case of both NI and NDI, the earliest resolvable
transient (>30 ps) has spectral features characteristic of triplet-
state only, indicating that radical products from singlet state
quenching are not significant. From Scheme 1, the total triplet-
state yield, observed from free and associated imide is given as
(φISC + φAPP), the sum of that from pathway (a) and (b).
Accordingly, the total T1 - Tn absorption change (∆AT1), ob-
served immediately after the 8 ns laser pulse, is given by eq 4
∆AT1 ) ∆ꢀT1[1I*]TOT{φISCf1 + φAPP(1 - f1)} )
∆ꢀT1[1I*]TOT{(φISC - φAPP)f1 + φAPP} (4)
where, ∆ꢀT1 is the T1 - Tn extinction coefficient, [1I*]TOT is
the total concentration of electronically excited states produced,
Figure 6. T1 - Tn absorbance changes ((9) extrapolated to time ) 0
after the laser pulse) induced upon 355 nm excitation of a solution of
(a) NI and (b) NDI at 480 and 485 nm, respectively, as a function of
the fraction of ground states that are not associated (f1 ) fraction of
“free” imides) with calf thymus DNA. All solutions are prepared in
deoxygenated phosphate buffer solution (10 mM, pH 7.0). Data have
been normalized for changes in ground-state absorbance at 355 nm
upon complexation. Superimposed on these graphs are measured
fluorescence intensities (O) at (a) 395 nm and (b) 412 nm for NI and
NDI as calf thymus DNA was added. In both (a) and (b) fluorescence
excitation was at the isosbestic point in the ground-state spectrum.
6a vs 6b.36 The large error on the latter value is a a result of
the large uncertainty in the near-zero intercept in Figure 6b.
Correspondingly, the φISC values derived from this treatment
(φISC ) 0.79 ( 0.06 and 0.41 ( 0.05 for NI and NDI,
respectively) are in good agreement with those estimated from
pulsed photolysis of aqueous solutions of NI and NDI in the
absence of DNA. The fraction of triplet states that is produced
in competition with rapid singlet state quenching (φAPP) is
significantly larger in the case of NI than NDI.
and f1 is the fraction of imide molecules that are not associated
with the calf thymus DNA (e.g., “free” imide). Assuming that
the T1 - Tn extinction coefficients of “free” and “associated”
imide are the same, and that the total number of excited states
produced stays constant, eq 4 predicts that the laser-induced T1
- Tn absorption change should decrease linearly with decreasing
f1.The T1 - Tn absorbance changes, extrapolated to time zero,
as a function of the fraction of “free” imide (f1), are shown in
a and b of Figure 6 for NI and NDI, respectively. Data have
been corrected for the change in ground-state absorbance at 355
nm induced by binding to the calf thymus DNA. As predicted
from eq 3, both plots are linear. Extrapolation of the data to f1
) 0 yields the “limiting” intersystem crossing (φAPP) of the
imide while associated with calf thymus DNA. The T1 - Tn
extinction coefficients (∆ꢀT1) have been previously determined
for NI and NDI derivatives.35 The concentration of excited states
([1I*]TOT) was determined using a benzophenone actinometer
solution. Using these values, along with linear least-squares
analysis of the data shown in a and b of Figure 6, we estimate
φAPP to be 0.08 ( 0.02 and 0.004 ( 0.02 for NI and NDI. The
results are consistent with the clearly nonzero intercept in Figure
From the intersystem crossing quantum yields of NI and NDI,
along with the measured singlet-state lifetimes (τS1 ) 1/kS1
)
(34) Samata, A.; Ramacharndram, B.; Saroja, G. J. Photochem. Photo-
biol., A 1996, 101, 29-32.
(35) The determinations of φAPP in aqueous solution were carried out
using the assumption that the triplet absorption coefficients of NI and NDI
were identical to those reported earlier (ref 6) in 50:50 CH3CN:H2O.
(36) The apparent “scatter” in the absorption changes corresponding to
small f1 values is a consequence of the small transient absorption signals in
this range. Thus, the analysis in Figure 6, where the ISC efficiency is
extrapolated to fully bound conditions, provides a more accurate determi-
nation of φAPP
.

Photoprocesses of ND and NDI DeriVatiVes
J. Am. Chem. Soc., Vol. 122, No. 3, 2000 435
2.4 ns (NI) and 280 ps (NDI)), the rate constants for intersystem
crossing (kISC) are 3.3 × 108 s-1 and 1.5 × 109 s-1 for NI and
NDI, respectively. In both cases, the decrease in ISC quantum
yield suggests a competing pathway for S1 deactivation when
the chromophores are bound to the DNA. The rate constants
(ket) for this competitive pathway are evaluated from eq 5.
Moreover, the ground-state concentrations used by these workers
are sufficiently high (100 µM) that radical anions from self-
quenching could certainly arise. From our studies (see above),
significant concentrations of NI•- were produced at concentra-
tions that were an order of magnitude smaller. In published
reports of DNA oxidation and cleavage using naphthalene imide
chromophores,4,5 this mechanism was not discussed.
φISC
ket
In a second report, strand scission of supercoiled DNA in
the presence of bis(substituted)-1,4,5,8-naphthaldiimide was in-
vestigated.3 A bis(hydroperoxy) derivative of the naphthaldi-
imide was shown to spontaneously nick supercoiled DNA.
However, in control experiments, naphthaldiimide derivatives
that lacked the pendant hydroperoxide substituent did not cleave
the DNA, even after piperidine treatment. The result is perplex-
ing, considering the fact that derivatives of 1,4,5,8-naphthal-
imides are more powerful oxidizing agents in their electronically
excited singlet and triplet states. Since, on time scales slower
than 30 ps, singlet-derived products are not observed with either
compound, charge recombination is apparently rapid. Under
conditions where NDI derivatives are predominantly associated
with the DNA and such “static” processes can occur, the photo-
cleavage efficiency will be severely diminished. Thus, it is criti-
cal to consider the partitioning of “free” and “bound” excited
states when correlating results from separate studies. From our
observation of the data presented herein, the contrasting photo-
reactivities of naphthalimide and naphthaldiimide may be a re-
sult of differing degrees of association (and thus differing
degrees of static quenching processes) within a ground-state
complex.
) 1 +
(5)
φAPP
kS1
From eq 5 and the quantum yields given above, we estimate
the rate constants for this competive process to be 3.7 (( 0.07)
× 109 s-1 and 3.6 (( 15) × 1011 s-1 for reaction of the imide
singlet states with DNA.37 Assuming that the competitive
process is electron transfer, it is clear that, in both cases, electron
transfer competes very effectively with intersystem crossing.
Although the rate constant for intersystem crossing in NDI is
higher than that of NI, the lower apparent triplet yield of the
former can be attributed to the more rapid photoinduced electron
transfer to the singlet state of NDI. This result is consistent with
NDI being a better electron acceptor than NI.
Our observation of fluorescence quenching with added DNA
is consistent with Scheme 1. Owing to the rapid singlet-state
quenching within the imide/DNA complex, the fluorescence
intensity is significantly diminished. Consistent with Scheme
1, fluorescence is predominantly observed only from electroni-
cally excited imide singlet states that are free in solution. Thus,
it is anticipated that the fluorescence intensity will increase
linearly with f1. These data are superimposed in a and b of Figure
6 for NI and NDI, respectively. The linear dependence is
consistent with the model shown in Scheme 1.
From our systematic correlation of the relative yield of long-
lived naphthalimide and naphthaldiimide radical with the
fraction of bound chromophore, it is apparent that the compari-
son of photocleavage quantum yields (and yields of the radical
intermediates) must consider the relative role of singlet- and
triplet-state electron transfer. As the data in Figure 6 clearly
show, rapid singlet-state quenching within the imide/DNA
complex results in diminished product yield. In the case of
compound 2, extrapolation of the transient absorption data to
the “fully bound” conditions leads to insignificant product
formation, suggesting that rapid charge-recombination, within
the complex, is occurring. In contrast, intersystem crossing in
compound 1, even when predominantly bound, is apparently
competitive with singlet-state quenching. This is evidenced by
the nonzero intercept in Figure 6a. The finite amounts of reduced
NI formed as a result of competitive intersystem crossing is
likely a result of slower charge recombination within the triplet
vs singlet radical ion pair.
The issue of competitive electron transfer and intersystem
crossing for intercalated photocleavage agents has been ad-
dressed for other compounds. In related experiments, Brun and
Harriman investigated the photochemistry of intercalated qua-
ternary diaazaaromatic salts.38 Consistent with the observations
reported herein, these workers observed significant fluorescence
o
quenching (φf /φf ≈ 20 or 50, depending upon the chromophore
employed) upon intercalation of the dye. The reduced chro-
mophore was observed by picosecond pump-probe spectros-
copy. First-order charge recombination occurred rapidly (τrec
< 100 ps) but apparently on a somewhat slower time scale than
singlet ion-pair recombination in our work. In other work, singlet
vs triplet ion pairs have been considered in the cleavage of
duplex DNA by intercalated anthraquinone.39 It was postulated
that efficient cleavage of DNA does not occur from singlet-
state precursors, since rapid charge recombination occurs before
strand cleavage and hole hopping takes place.
Significance to Efficient DNA Photocleavage. One report
has addressed oligonucleotide cleavage resulting from irradiation
of an L-lysine derivative of 1,8-naphthalimide.5 In the former
case, photocleavage of a duplex hexamer was observed under
conditions of steady-state irradiation. Using laser flash, the
reduced naphthalimide was observed as a product of the
bimolecular quenching of the triplet excited state. Ground-state
association by the planar 1,8-naphthalimide with the oligo-
nucleotide was not addressed, and the relative yield of redox
products, as a function of complexation, was not determined.
Conclusions
In summary, we have carried out detailed investigations to
characterize the photophysical and photochemical properties of
naphthalene imide and diimide derivatives in the presence of
DNA. The systematic investigations correlate the extent of
chromophore/DNA complexation with triplet-state and radical
production following pulsed photolysis. Significantly, the rela-
tive yield of long-lived triplet excited states produced decreases
linearly as the fraction of associated chromophores increases.
In the case of the 1,8-naphthalimide derivative (1) employed
in this work, the intersystem crossing efficiency within the
ground-state complex is 0.08. In contrast, intersystem crossing
within the ground-state complex of NDI and calf thymus DNA
is negligible (0.004), and only trace amounts of radical products
(37) The electron-transfer rate constants from the intersystem crossing
efficiencies (eq 5) were estimated assuming that T1 - Tn extinction
coefficients of “free” and “bound” imides are identical.
(38) Brun, A. M.; Harriman, A. J. Am. Chem. Soc. 1991, 113, 8153-
8159.
(39) Ly, D.; Kan, Y.; Armitage, B.; Schuster, G. B. J. Am. Chem. Soc.
1996, 118, 8747-8748.

436 J. Am. Chem. Soc., Vol. 122, No. 3, 2000
Rogers et al.
are observed. The design of chemical systems that can initiate
site-specific damage in biological polymers requires efficient
and selective association with the target of interest. Hence, it is
critical to understand and circumvent the factors responsible
for “short-circuiting” the photochemical pathway. Strategies for
increasing the intersystem crossing efficiency and spatial charge
separation are currently being explored.
produced from DNA-bound molecules is consistent with very
rapid electron transfer and charge recombination processes
initiated by the electronically excited singlet state.
Acknowledgment. We thank the American Cancer Society,
Maryland Division, for financial support. J.E.R. and S.J.W. are
grateful for Research Assistantship support from the UMBC
Graduate School (DRIF).
For both compounds employed, radical anion production from
self-quenching processes is significant. In the case of NI,
additional radical anion is produced when DNA is present,
suggesting that nucleotide oxidation is occurring. Conversely,
when NDI is used, all of the NDI radical anion that is observed
by transient absorption can be accounted for by self-quenching
interactions. The observation that additional NDI•- is not
Supporting Information Available: Transient absorption
data fit to eq 3. Production of radical anion from self-quenching
and DNA oxidation as a function of imide and diimide concen-
trations (PDF). This material is available free of charge via the
Internet at http://pubs.acs.org.
JA992332D