A theoretical and experimental investigation of the spectroscopic properties of a DNA-intercalator salphen-type ZnII complex

Article abstract of DOI:10.1002/chem.201304876

The photophysical and DNA-binding properties of the cationic zinc(II) complex of 5-triethylammonium methyl salicylidene ortho-phenylenediiminato (ZnL²⁺) were investigated by a combination of experimental and theoretical methods. DFT calculation

Full text of DOI:10.1002/chem.201304876

DOI: 10.1002/chem.201304876
Full Paper
&
DNA Intercalators
A Theoretical and Experimental Investigation of the Spectroscopic
Properties of a DNA-Intercalator Salphen-Type ZnII Complex
Alessandro Biancardi,*[a] Azzurra Burgalassi,[a] Alessio Terenzi,[b] Angelo Spinello,[b]
Giampaolo Barone,[b] Tarita Biver,*[a] and Benedetta Mennucci[a]
Abstract: The photophysical and DNA-binding properties of
the cationic zinc(II) complex of 5-triethylammonium methyl
salicylidene ortho-phenylenediiminato (ZnL2+) were investi-
gated by a combination of experimental and theoretical
methods. DFT calculations were performed on both the
ground and the first excited states of ZnL2+ and on its possi-
ble mono- and dioxidation products, both in vacuo and in
selected solvents mimicked by the polarizable continuum
model. Comparison of the calculated absorption and fluores-
cence transitions with the corresponding experimental data
led to the conclusion that visible light induces a two-elec-
tron photooxidation process located on the phenylenediimi-
nato ligand. Kinetic measurements, performed by monitor-
ing absorbance changes over time in several solvents, are in
agreement with a slow unimolecular photooxidation pro-
cess, which is faster in water and slower in less polar sol-
vents. Moreover, structural details of ZnL–DNA binding were
obtained by DFT calculations on the intercalation complexes
between ZnL and the d(ApT)2 and d(GpC)2 dinucleoside
monophosphate duplexes. Two main complementary bind-
ing interactions are proposed: 1) intercalation of the central
phenyl ring of the ligand between the stacked DNA base
pairs; 2) external electrostatic attraction between the nega-
tively charged phosphate groups and the two cationic trie-
thylammonium groups of the Schiff-base ligand. Such sug-
gestions are supported by fluorescence titrations performed
on the ZnL/DNA system at different ionic strengths and tem-
peratures. In particular, the values of the DNA-binding con-
stants obtained at different temperatures provided the en-
thalpic and entropic contributions to the binding and con-
firmed that two competitive mechanisms, namely, intercala-
tion and external interaction, are involved. The two mecha-
nisms are coexistent at room temperature under
physiological conditions.
have been also reported in the literature,[9] including their in-
teraction with DNA.[7,10–12] The interest in DNA binding by tran-
sition metal complexes is due to the potential discovery of
novel metallodrugs.[13] For example, it is well known that DNA
is the target of the most common clinically used platinum-
based drugs.[14]
Introduction
Salphen and salen[1] are among the oldest and most popular li-
gands in coordination chemistry. The zinc(II) ion is essential in
most biological systems, since it is present in a plethora of en-
zymes with structural (influencing both stability and conforma-
tion) and catalytic functions.[2] Zinc complexes are of current
interest for their applications in molecular sensors,[3] as emis-
sive materials in organic light-emitting diodes,[4] in homogene-
ous catalysis,[5] as building blocks for supramolecular chemis-
try,[6] and for their potential medical applications.[7] They are
also relevant as model systems to better understand the struc-
ture and the functions of zinc enzymes[2] and zinc-enzyme in-
hibition.[8] Several studies on other salen-type metal complexes
Recently, the cationic ligand 5-triethylammonium methyl sal-
icylidene ortho-phenylenediiminato has been used to prepare
first-row transition metal complexes with improved water solu-
bility and DNA-binding affinity.[12,15] UV/Vis absorption, circular
dichroism,[12] and linear dichroism[16] measurements indicate
that the zinc(II) complex of this Schiff-base ligand (ZnL2+, see
Figure 1) tightly binds DNA by intercalation. However, other as-
pects remain to be investigated, such as the role of the posi-
tive charge on the side-chain arms of the Schiff-base ligand.
Moreover, ZnL2+ shows marked enhancement of the fluores-
[a] A. Biancardi, Dr. A. Burgalassi, Dr. T. Biver, Prof. B. Mennucci
Dipartimento di Chimica e Chimica Industriale
Universitꢀ di Pisa, Via Risorgimento 35, 56126 Pisa (Italy)
E-mail: alessandro.biancardi@for.unipi.it
tarita.biver@unipi.it
[b] Dr. A. Terenzi, Dr. A. Spinello, Dr. G. Barone
Dipartimento di Scienze e Tecnologie Biologiche
Chimiche e Farmaceutiche, Universitꢀ di Palermo
Viale delle Scienze, Parco d’Orleans II
Edificio 17, 90128 Palermo (Italy)
Supporting information for this article is available on the WWW under
http://dx.doi.org/10.1002/chem.201304876.
Figure 1. Molecular structure of ZnL2+
.
Chem. Eur. J. 2014, 20, 7439 – 7447
7439 ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
cence intensity when its aqueous solutions are exposed to the
visible light from a tungsten source; interestingly, this is inhib-
ited when the complex is intercalated into native calf thymus
(ct) DNA.[10] By studying frozen solutions of a ZnII salphen com-
plex in organic solvents at 77 K, Germain et al.[17] first recog-
nized that 5 min exposure to visible light causes a photoin-
duced oxidation process giving rise to an EPR signal corre-
fects of the interaction with one or two explicit water mole-
cules coordinating to the zinc ion to give the pentacoordinate
complex [ZnL(H2O)]2+ and/or interacting with the phenoxyl
oxygen atoms. In the former case the explicit water molecule
forms two hydrogen bonds with the hydroxyl groups, whereas
in the latter case the second explicit water molecule is coordi-
nated to the metal center (see Figure S3 of the Supporting In-
formation). To include also the effects of outer solvent shells,
the clusters were embedded in a PCM solvent.
sponding to the formation of an S=1= radical.
2
The effect of light on complexes of d-block metals and the
subsequent effects on biomolecules like DNA is well-docu-
mented, for instance, as concerns photodynamic therapy.[18]
Metal compounds, such as ruthenium and osmium complexes,
were found to undergo a molecular light-switch effect that is
promising for selective sequence sensing.[19] The light-switch-
ing mechanism was also successfully investigated from the
theoretical point of view.[20,21]
Figure 2 compares the experimental spectrum of the com-
plex in water and the calculated spectra of the complex in
vacuo and in water. The visible absorption spectrum of ZnL2+
To clarify the photophysical properties of ZnL2+, we per-
formed a quantum-mechanical (QM) study of its absorbance
and fluorescence emission in environments of increasing com-
plexity and in its singlet, doublet, triplet, and mono- and dioxi-
dized forms. First we considered homogeneous solutions by
simulating their effects in terms of a polarizable continuum
model (PCM),[22] introduced the model photooxidation effects,
and finally analyzed intercalation into double-stranded DNA.
Detailed computational simulations of the intercalated struc-
tures were aimed at an in-depth understanding of their photo-
physical characteristics, which is still lacking. In particular, the
effects of different environments/binding modes on the ab-
sorption and fluorescence of these complexes have never
been considered in detail by using computational tools. How-
ever, accurate QM modeling (coupled with both atomistic and
continuum descriptions of the environment) of the absorption
and fluorescence properties of the intercalated probe is of
great help for understanding the microscopic details of the in-
teraction between small molecules and nucleic acids, and it
may be an important tool for the analysis of processes of bio-
chemical and biophysical significance, together with providing
a structural framework for new hypotheses.[23]
Figure 2. Comparison between the experimental absorption spectrum of
ZnL2+ in buffered aqueous solution (a) and the calculated TD-DFT absorp-
tion spectra in vacuo (b) and in water (c) by using PCM.
Results and Discussion
Gas phase and aqueous solution
To understand the important changes in the photophysical be-
havior of ZnL2+ under different environmental conditions, we
first analyzed the geometrical structures of both ground and
excited states of the isolated and solvated system using
a (TD)DFT approach combined with a PCM description of the
aqueous solvent. The ZnL2+ complex (Figure 1) is positively
charged and has several rotational degrees of freedom (the
main ones are shown in Figure S1 of the Supporting Informa-
tion).
in aqueous solution is mainly determined by a large absorp-
tion band at about 397 nm, which is attributed to a metal-per-
turbed intraligand electronic transition. This band is common
to both the free and bound ligand (see Figure S4 of the Sup-
porting Information). The UV absorption spectrum is composed
of strong bands in which the zinc atom has an effect. As
shown in Figure 2, the calculated absorption bands are system-
atically blueshifted compared with the experimental ones. The
calculated spectrum seems to be less affected by this shift on
moving from the visible to the UV region. Figure 2 shows
good agreement between experimental and calculated spec-
tra, especially when solvent effects are included. In particular,
inclusion of the solvent mainly modifies the relative transition
intensities. The spectrum of the free ligand (Figure S4 of the
The optimized ground- and excited-state structures (Fig-
ure S2 of the Supporting Information) indicate that the excita-
tion process involves planarization of the system (see also
Table S1 of the Supporting Information). To obtain a more re-
fined description of solvation in water, we also analyzed the ef-
Chem. Eur. J. 2014, 20, 7439 – 7447
www.chemeurj.org
7440
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
Supporting Information) shows that the zinc only has a shifting
effect on the first two transitions, but it has a key effect on the
part of the spectrum below 300 nm.
Table 1. Comparison between experimental and calculated absorption
(ABS) energies, emission (FLU) energies, and Stokes shifts (SS) of ZnL2+
.
The calculated TDDFT/PCM +1 water or +2 water data refer to ZnL2+
plus one and two explicit water molecules in a PCM solvent, as shown in
Figure S3 of the Supporting Information. All values are in electron volts.
The most important transitions in the spectra were analyzed
by the natural transition orbitals (NTO) method.[24] This is
a compact orbital representation of the electronic-transition
density matrix, mainly in term of a few single hole/particle
NTO pairs.[25] The dominant NTO pairs are shown in Figure 3.
Method
ABS
FLU
SS
exptl
in vacuo
TDDFT/PCM
TDDFT/PCM+1 water
TDDFT/PCM+2 water
3.36
3.80
3.80
3.84
3.85
2.58
2.66
2.75
2.77
2.66
0.78
1.14
1.05
1.07
1.19
a PCM description has negligible effects on the absorption en-
ergies, but it modifies the emission energies by inducing
a blueshift of 0.09 eV. The inclusion of one or two explicit
water molecules only slightly affects the absorption spectra,
whereas a redshift is observed for the emission in the complex
with a zinc-coordinated water ligand.
The photooxidation process
Recent studies indicate that exposure to a tungsten lamp
causes oxidation of ZnL2+ in aqueous solution, whereas this
process is inhibited when the metal complex is intercalated in
DNA.[10,17] To clarify such behavior, we performed a QM/PCM
study considering the mono-oxidized (ZnL3+, doublet) and di-
oxidized (ZnL4+, singlet and triplet) forms of ZnL2+ (see
Table S2 of the Supporting Information for a list of the main
geometrical parameters). The calculations showed that the oxi-
dation is mainly located on the aromatic rings of the ligand
(Figure S5 of the Supporting Information). Comparison be-
tween experiment and calculation (Figure 4 and Figure S6 of
the Supporting Information) seems to show that the experi-
mental spectrum of the oxidized complex corresponds to the
dioxidized triplet radical ZnL4+(T). This is also confirmed by the
relative energies calculated in water, which showed that ZnL4+
(T) is about 9 kcalmolꢀ1 more stable than ZnL4+(S).
The kinetics of the photooxidation process of ZnL2+ was ex-
perimentally monitored by measuring the change in absorb-
ance in time at different wavelengths (245, 293, and 373 nm).
The monoexponential trends obtained (Figure S7A of the Sup-
porting Information) were interpolated to obtain the time con-
stant 1/t, which was found to be independent of the ZnL2+
concentration (Figure S7B of the Supporting Information), in
agreement with a unimolecular reaction. The data show that
the kinetics of the process depend on the solvent, as shown in
Figure 5. In particular, the process is faster in water and the
rate monotonously decreases with decreasing solvent polarity.
On the other hand, for solvents of similar polarities (THF and
1,4-dioxane), the time constants are similar.
Figure 3. Dominant NTO pairs for the most important transitions of ZnL2+ in
vacuo (shown in Figure 2b). The NTO transition amplitudes are also shown.
The NTO analysis showed that the first two transitions are
mainly located on the ligand (Figure 3) and, as a result, this
band is practically the same for the free (Figure S4 of the Sup-
porting Information) and bound ligand (Figure 2b, c). On the
other hand, the presence of the zinc ion becomes more impor-
tant for transitions at wavelengths lower than 300 nm, as
shown in Figure 3 for the NTOs related to the S0!S4 and S0!
S6 transitions.
This behavior is difficult to simulate by calculations, as the
oxidation process of ZnL2+ is more likely driven by an excited-
state electron transfer, as shown by Germain et al.[17] However,
in the approximation of considering the solvent dependence
of the oxidation process to be similar in the ground and excit-
ed states, calculated vertical ionization energies [ZnL+2 >ZnL+
Table 1 compares the experimental and calculated absorp-
tion and emission energies and Stokes shifts of ZnL2+ in water.
The calculated absorption energies refer to the S0!S2 transi-
tion, whereas the emission energy refers to the transition from
the lowest excited state. Inclusion of the solvent by using
Chem. Eur. J. 2014, 20, 7439 – 7447
www.chemeurj.org
7441
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
root mean square deviation (RMSD) for all non-hydrogen
atoms is shown in Figure S8 of the Supporting Information,
which shows that the initial structure of both intercalation
complexes suddenly changes after about 1 ns, by about 3 ꢁ in
the RMSD. Then, oscillations of about 2 ꢁ of RMSD occur
around the equilibrium geometry, largely involving the tails of
the two dodecanucleotides and much less the structure of the
metal complex and the four stacked DNA bases of the interca-
lation pocket. In particular, the metal complex remains tightly
intercalated between the fifth and sixth bases for the whole
simulation time. The equilibrium geometry was used as start-
ing point for further geometry optimizations by hybrid two-
layer QM/molecular mechanics (MM) calculations with DFT as
QM method and the Amber99 force field as MM method.[26]
Because of the large size of the systems, here we modeled the
double-stranded DNA pockets using the d(ApT)2 and d(GpC)2
simplified models, which limit the interaction between ZnL2+
and the DNA to the first neighborhoods. As shown previous-
ly,[23] inclusion of at least the effects of the lateral sugar/phos-
phate chains in addition to the stacking bases is crucial to
obtain a reliable description of the short-range interactions be-
tween the DNA-bound molecule and the pocket; the selected
model systems are shown in Figure 6.
Figure 4. Comparison between calculated (b–d) absorption spectra of ZnL3+
(D spin state) and ZnL4+ (S and T spin states) in water (PCM) and the experi-
mental spectra of light-exposed species in water (a).
Figure 6. Optimized molecular structures of ZnL2+ intercalated in d(GpC)2
(left) and d(ApT)2 (right).
A straightforward method for the calculation of the transi-
tion energies relies on a fully QM strategy for both ZnL2+ and
the intercalation pocket. An alternative approach is based on
the use of
a
hybrid QM/polarizable MM (QM/MMpol)[27]
method describing the DNA pocket with a polarizable classical
force field given by point charges and induced dipoles. The
comparison between the fully QM and the QM/MMpol method
is useful in disentangling the different effects of the DNA
pocket. The solvent can be taken into account at the PCM
level by using a QM/PCM or a QM/MMpol/PCM[28] approach.
The optimized ground-state structures of ZnL2+ intercalated in
d(ApT)2 and d(GpC)2 are shown in Figure 6. Important opti-
mized dihedral angles are reported in Table S1 of the Support-
ing Information.
Figure 5. Dependence of the reciprocal time constant 1/t for ZnL2+ photo-
oxidation on the solvent polarity (ET(30) scale[50]), T=258C.
4(T)] decrease on increasing the polarity of the solvent, from
11.9 eV in the gas phase to 7.0 eV in THF to 6.3 eV in water, in
agreement with the experimental plot in Figure 5.
Figure 7 compares the experimental and calculated absorp-
tion spectra of ZnL2+ intercalated in the d(GpC)2 and d(ApT)2
pockets. This comparison shows that intercalation causes sig-
nificant changes in the absorption spectrum. Since we per-
formed a fully QM study on both ZnL2+ and the DNA pocket,
our calculated spectra explicitly describe the effects of the QM
DNA intercalation
Since no experimental intercalated structures are available for
ZnL2+, we obtained two possible pockets of the intercalated
system in alternating AT and GC dodecanucleotides from MD
simulations by following a recently reported procedure.[26] The
Chem. Eur. J. 2014, 20, 7439 – 7447
www.chemeurj.org
7442
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
Table 2. Comparison between experimental and calculated absorption
(ABS), emission (FLU) energies, and Stokes shifts (SS) for different ZnL2+
/polynucleotide systems. All values are in electron volts. The comparison
with the available experimental data in ctDNA is performed by averaging
the results calculated for d(ApT)2 and d(GpC)2, considering that ctDNA is
constituted by 60% AT and 40% GC base pairs, and assuming an equal
intercalation affinity.
Dinucleotide
d(ApT)2
Method
QM/QM
ABS
3.73
3.70
3.77
3.64
3.64
3.73
3.24
3.69
3.68
3.75
FLU
2.59
2.65
2.75
2.42
2.48
2.53
2.53
2.52
2.58
2.66
SS
1.14
1.05
1.02
1.22
1.16
1.20
0.71
1.17
1.09
1.09
QM/QM/PCM
QM/MMpol/PCM
QM/QM
QM/QM/PCM
QM/MMpol/PCM
exptl
d(GpC)2
ctDNA
hQM/QMi
hQM/QM/PCMi
hQM/MMpol/PCMi
Figure 7. Comparison between the experimental absorption spectrum of
ZnL2+ intercalated in ctDNA (a) and those calculated in water (PCM) for
ZnL2+ intercalated in d(GpC)2 (b) and d(ApT)2 (c). The corresponding spectra
of free ZnL2+ are also shown for each case (dashed lines). For the experi-
mental spectrum of the ZnL2+/DNA complexes (258C), low ionic strength
(1ꢂ10ꢀ3 m buffer) and excess polymer (CD =54 mm, CP/CD =10) were used to
ensure quantitative dye reaction; in the shown spectrum, the DNA contribu-
tion to absorbance was subtracted.
Figure 8. Dominant NTOs pairs for the most important transitions of ZnL2+
bound to d(ApT)2 (see Figure 7e). The NTO transition amplitudes are also
shown.
porting Information for AT and GC, respectively). This analysis
shows that all transitions (even those at lower energies) are
characterized by non-negligible charge-transfer contributions
between the base pairs and the DNA. The relevance of this
mixing in a different intercalated molecule was previously sug-
gested.[21,29] The comparison between the isodensity surfaces
of free and DNA-bound ZnL2+ shows that the intercalation
leads to breaking of the symmetry of the virtual orbitals (see
Figure 3, Figure 8, and Figure S9 of the Supporting Informa-
tion). Analysis of the orbitals involved in the different transi-
tions showed that the first two excitations are mainly located
on the ligand, whereas higher-energy transitions also involve
increasing contributions from both the zinc atom and the adja-
cent bases.
intercalation pocket. This is crucial for modeling the spectrum
at wavelengths longer than 300 nm, which is characterized by
strong mixing between ZnL2+ and the DNA pocket. Good qual-
itative agreement is found between the calculated and the ex-
perimental spectra (Figure 7). In particular, intercalation would
cause a decrease of the intensity of the dye bands and their
shift to higher wavelengths, which is fully reproduced in the
calculated spectra. The inclusion of the solvent by using a con-
tinuum PCM causes only a slight modification of the spectra.
Table 2 lists the experimental and calculated absorption and
emission energies of intercalated ZnL2+. As for the isolated
complex, the calculated absorption energies refer to the S0!
S2 transition, whereas the emission energies refer to transition
from the lowest excited state.
The binding was further analyzed by performing spectro-
fluorometric titrations to determine the variation in fluores-
cence of the system on addition of different amounts of DNA.
The most important transitions in the spectra were analyzed
by the NTO method (see Figure 8 and Figure S7 of the Sup-
Chem. Eur. J. 2014, 20, 7439 – 7447
www.chemeurj.org
7443
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
The interaction between ZnL2+ and natural DNA was repre-
sented by an apparent reaction [Eq. (1)]
P þ D Ð PD
ð1Þ
where D and PD denote the free and bound ZnL2+ complex
respectively, and P indicates a free site on the DNA polymer. In
agreement with previous findings,[10] the ZnL2+/DNA solutions
are stable to light.
One example of the obtained binding isotherms (DF=
FꢀfDCD, where fD =F8/CD and F8 is the fluorescence measured
in the absence of DNA) is shown in Figure S10A of the Sup-
porting Information. The data were analyzed according to
Equation (2) by using a procedure that first disregards the DF/
Df2 term (Df=fPDꢀfD is the amplitude of the isotherm) and
then calculates it iteratively.[30]
CPCD=DF þ DF=Df2 ¼ ðCP þ CDÞ=Df þ 1=KDf
ð2Þ
At convergence, the value of the equilibrium constant K for
Equation (1) is calculated by the slope/intercept ratio of the
straight line interpolating data points in the (CPCD/DF+DF/Df2)
versus (CP +CD) plot (Figure S10B of the Supporting Informa-
tion). At 258C and an ionic strength of 0.11m, K=(3.5ꢁ0.2)ꢂ
104 mꢀ1, in agreement with previous UV/Vis estimations.[12]
Figure 9. Dependence of the binding constant for the ZnL2+/DNA system
on salt content (a) and temperature (b). pH 7 (sodium cacodylate (NaCac
buffer), T=258C (a), I=0.01m (NaCac) (b).
The experiments were repeated under different conditions
of added salt (NaCl content) or temperature. Figure 9A is
a double-logarithmic plot showing the dependence of K on
the salt content of the medium. According to Record et al.[31]
the slope of the plot corresponds to 0.88m, where m is the
number of sodium ions displaced from the DNA double helix
due to the binding of one charged interacting molecule. It
turns out that m=0.7ꢁ0.1. This value is significantly lower
than the +2 total charge borne by ZnL2+, and lower than the
values obtained for other DNA-intercalating metal com-
plexes.[32] However, in the herein-analyzed system, the positive
charge is located quite far away from the central aromatic (i.e.,
intercalating) moiety. This result suggests that the protruding
lateral residues only partially participate in the binding.
presence of a binding mode with external features. On the
whole, the data suggest a complex binding mode in which
both features play a role. Other bifunctional intercalating mole-
cules with positively charged pendant arms were also found to
exhibit complex binding modes in which the prevalence of ex-
ternal binding or intercalation depends on experimental condi-
tions such as dye loading.[34]
Conclusion
Figure 9B shows the binding constants obtained by varying
the temperature between 5 and 458C. These data display a bi-
phasic trend. Possible complex bleaching was carefully
checked and ruled out; the fluorescence characteristics of the
systems remained unchanged in the whole analyzed tempera-
ture range. Therefore, this behavior should be ascribed to the
presence of a complex binding mechanism, in which at least
two binding modes coexist. From the slope (ꢀDH/R) and inter-
cept (DS/R) of the two linear trends shown in Figure 9B , the
values of the enthalpy and entropy variations (DH and DS, re-
spectively) of the two limiting binding modes can be obtained.
These data are collected in Table S3 of the Supporting Informa-
tion. According to Chaires,[33] the values of DH and DS provide
information on the binding mode of a small molecule to a poly-
nucleotide. The data at T>258C (highly negative DH, negative
DS) are in agreement with intercalation; on the other hand, for
Tꢂ258C, the small positive DH and positive DS reveal the
The kinetics of the photochemical reaction that occurs for the
cationic Schiff-base zinc(II) complex ZnL2+ on exposure to visi-
ble light from a tungsten lamp involves an unimolecular pro-
cess. This is faster in water and slows down in less polar sol-
vents. The process was interpreted by comparison of the ex-
perimental and calculated absorption/fluorescence spectra to
involve a two-electron photooxidation of the coordinated
Schiff-base ligand of ZnL2+ to give the triplet radical ZnL4+(T).
The calculated spectra of ZnL2+ intercalated in both d(ApT)2
and d(GpC)2 correctly reproduce the hypochromic effect exper-
imentally observed for ZnL2+ intercalated in natural DNA. The
intraligand band of the zinc(II) complex, which consists of two
transitions to the first and second excited electronic states, is
redshifted by about 0.12 eV on going from the free system (in
water) to the intercalated system, and the calculations gave
0.11 eV when the shift is averaged over the two pockets.
Chem. Eur. J. 2014, 20, 7439 – 7447
www.chemeurj.org
7444
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
The relaxed geometries were optimized by two-layer ONIOM calcu-
lations,[42] with the aim of performing a high-level calculation on
the intercalation pocket and to take account of the constraining ef-
fects of the double-helical structure at lower levels of theory. The
MPWB1K DFT functional and the dzvp basis set were used in the
higher layer to suitably model the hydrogen bonding and p–p
stacking interactions between the sixth and seventh Watson–Crick
base pairs. The Amber99 force field was used in the lower layer of
the ONIOM calculations.
The values of the experimental DNA-binding constants ob-
tained at different temperatures provided the enthalpic and
entropic contributions to the DNA binding and showed that
two interaction mechanisms are involved (intercalation and ex-
ternal interaction at high and low temperature, respectively)
and that the two mechanisms coexist at room temperature
under physiological conditions. Remarkably, this result is sup-
ported by the analysis of the structures of the intercalation
complexes, optimized by DFT. In fact, the latter are character-
ized by two main complementary binding interactions: 1) in-
tercalation in the stacked AT–AT or GC–GC base pairs; 2) exter-
nal electrostatic interaction between the negatively charged
phosphate groups and the triethylammonium cationic groups
of the Schiff-base ligand.
To evaluate the importance of a QM description of the intercalation
pockets constituted by the d(ApT)2 and d(GpC)2 dinucleoside
monophosphate duplexes in the determination of absorption and
fluorescence energies of ZnL2+, two models were compared. In the
first model, denoted QM/QM/PCM, a full QM description was used
for both the complex and the DNA fragments constituting the
pockets. As in the geometry optimizations, the DNA fragments
were described with a smaller basis set (6-31G). The second model,
denoted QM/MMPol/PCM,[28] instead used a QM description only
for ZnL2+, whereas the pockets were described with fixed charges
and induced dipoles. The fixed charges were obtained from a fit of
the electrostatic potential of fragment by using the Merz and Koll-
man method[43] with the same functional and the same basis set as
used in the QM calculations. The induced dipoles were obtained in
terms of isotropic polarizabilities placed on each MM atom. We
adopted the Thole model, which avoids problems of intramolecular
overpolarization by using a smeared dipole–dipole interaction
tensor.[44] Atomic isotropic polarizability values were taken from
the fit of experimental molecular polarizabilities performed by van
Duijnen and Swart by using the linear version of Thole dipole–
dipole tensor.[45] In both the QM/QM and QM/MMPol descriptions,
the effects of the rest of the DNA and of the solvent were simulat-
ed by using a PCM description. In the case of QM/MMPol/PCM, the
two classical parts (MMPol and PCM) were allowed to mutually po-
larize. The vertical ionization potential (VIP) was evaluated as the
difference between the energy of the optimized structure before
ionization and that of the ionized system at the same geometry.[46]
To effectively model the solvent effect, it is necessary to use the
non-equilibrium version of the PCM. In fact, the ionization process
can be seen as a vertical process in which the fast change in the
molecular charge density is coupled with a fast (mainly electronic)
and a slow (mainly orientational) response of the solvent mole-
cules, similar to the cases of electronic excitation and electron
transfer. All calculations were performed with the Gaussian 09
package.[47]
There is great interest in compounds that can interact with
light in a particular way, for instance in the framework of selec-
tive light-switch sensors for biomolecules. However, in-depth
mechanistic studies dealing with these systems are less abun-
dant. The salen and salphen ligands have been known for
a long time, but the possible photoinduced oxidation of their
metal complexes had not been elucidated. In this work the
combined use of experiments and theoretical calculations at
the (TD)DFT level of QM theory provided important informa-
tion both on the process of the metal complex oxidation and
on its interaction with DNA. This approach is not only interest-
ing for this particular system, but constitutes a promising tool
to be used in similar studies.
Experimental Section
Computational details
The molecular systems investigated in the calculations were
1) ZnL2+ and its oxidized derivatives in the gas phase and in water;
2) ZnL2+ intercalated in a GC or AT double-stranded DNA pocket.
The ground- and excited-state geometry optimizations and the ab-
sorption and emission calculations were performed at the (TD)DFT
level of QM theory by using the M05-2X functional[35] and the 6-
311+G(d,p) basis set. This functional was chosen for its ability to
describe both noncovalent interactions[36] and zinc complexes.[37]
Open-shell DFT/TDDFT calculations for triplet and doublet states
were carried out by using the unrestricted self-consistent field for-
malism. All the calculated absorption spectra were produced by
convolution of the vertical transitions energies by using Gaussian
functions of fixed full width at half-maximum of 0.24 eV. All the ge-
ometry optimizations were performed without imposing any con-
straints, whereas, in the case of the intercalated systems, the DNA
pocket structures were kept frozen and described by using the
smaller 6-31G basis set, analogously to previous works.[23] The inte-
gral equation formalism[38] version of the polarizable continuum
model (PCM)[22] was used to describe the effects of the solvent
both on the ground and the excited states. PCM cavities were built
up as a series of interlocking spheres centered on atoms with the
UFF radii. The intercalation pockets were obtained as described in
ref. [26] by combining a molecular dynamics (MD) simulation with
a following ONIOM calculation. The MD simulations were conduct-
ed for 20 ns by using the Gromacs 4.5.3 software package[39] on
[ZnL]/[dodeca(dA-dT)]2 and [ZnL]/[dodeca(dG-dC)]2 intercalated
complexes in the presence of explicit TIP3P water molecules with
the Amber99 force field[40] with Parmbsc0 nucleic acid torsions.[41]
Materials
The zinc(II) complex was synthesized by a recently reported proce-
dure,[10,15,16] by mixing 5-(triethylammoniummethyl) salicylaldehyde
chloride, 1,2 phenylendiamine, and zinc(II) perchlorate in a 2:1.1:1
molar ratio. A basic solution (NaOH) of salicylaldehyde in EtOH/H2O
was added dropwise to a previously prepared ethanolic solution of
the diamine and the metal perchlorate. The 5-(triethylammonium-
methyl) salicylaldehyde chloride ligand was prepared from 5-chlor-
omethyl salicylaldehyde and triethylamine in THF.[12] The product
was characterized by 1H NMR spectroscopy.[16]
Stock solutions of ZnL2+(perchlorate salt, M=807.075 gmolꢀ1
)
were prepared by dissolving weighed amounts of the solid in
MilliQ ultrapure water and kept in the dark at ꢀ208C. Working sol-
utions were obtained by dilution of the stocks, kept in the dark at
48C, and used within 2 d. ZnL2+ concentration was also checked
spectrophotometrically (e373 =1.52ꢂ104 mꢀ1 cmꢀ1[12]). Solutions in
nonaqueous solvents were prepared by dissolving weighed
Chem. Eur. J. 2014, 20, 7439 – 7447
www.chemeurj.org
7445
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
[6] a) S. J. Wezenberg, A. W. Kleij, Angew. Chem. 2008, 120, 2388–2399;
Angew. Chem. Int. Ed. 2008, 47, 2354–2364; b) M. S. Masar, N. C. Gianne-
schi, C. G. Oliveri, C. L. Stern, S. T. Nguyen, C. A. Mirkin, J. Am. Chem. Soc.
2007, 129, 10149–10158; c) S. S. Sun, C. L. Stern, S. T. Nguyen, J. T.
Hupp, J. Am. Chem. Soc. 2004, 126, 6314–6326.
[7] N. H. Campbell, N. H. Abd Karim, G. N. Parkinson, M. Gunaratnam, V. Pet-
rucci, A. K. Todd, R. Vilar, S. Neidle, J. Med. Chem. 2012, 55, 209–222.
[8] T. Tekeste, H. Vahrenkamp, Inorg. Chem. 2006, 45, 10799–10806.
[9] O. Alpturk, O. Rusin, S. O. Fakayode, W. H. Wang, J. O. Escobedo, I. M.
Warner, W. E. Crowe, V. Kral, J. M. Pruet, R. M. Strongin, P. Natl. Acad. Sci.
2006, 103, 9756–9760.
[10] G. Barone, A. Ruggirello, A. Silvestri, A. Terenzi, V. T. Liveri, J. Inorg. Bio-
chem. 2010, 104, 765–773.
[11] a) G. Barone, A. Terenzi, A. Lauria, A. M. Almerico, J. M. Leal, N. Busto, B.
Garcia, Coordin. Chem. Rev. 2013, 257, 2848–2862; b) N. H. Abd Karim,
O. Mendoza, A. Shivalingam, A. J. Thompson, S. Ghosh, M. K. Kuimova,
R. Vilar, RSC Adv. 2014, 4, 3355–3363.
amounts of the solid, kept in the dark at 48C, and used within 2 d.
The molar concentration of the dye is denoted CD.
DNA (from calf thymus, lyophilized sodium salt, highly polymer-
ized) was purchased from Sigma and its length was reduced to
about 700 base pairs by using a sonication procedure described
elsewhere.[48] Stock solutions of the polynucleotides were standar-
dized spectrophotometrically (e=13200mꢀ1 cmꢀ1 at 260 nm).[49]
The polynucleotide concentration was expressed in molarity of
base pairs and denoted CP. All solutions were prepared with Milli-
pore ultrapure water, which was also used as the reaction medium.
In the case of the DNA measurements, small amounts of sodium
cacodylate (Fluka, (CH3)2AsO2Na, 1.0ꢂ10ꢀ2 m) buffer were added to
the working solutions to keep the pH constant at 7.0. The nonaqu-
eous solvents THF (purity ꢃ99.8%, dried by distillation over potas-
sium) and 1,4-dioxane (spectrophotometric grade, purity 99%)
were from Sigma-Aldrich, whereas DMSO (purity ꢃ99.5%) was
from Carlo Erba. All other chemicals were of analytical grade from
Sigma and were used without further purification.
[12] A. Silvestri, G. Barone, G. Ruisi, D. Anselmo, S. Riela, V. T. Liveri, J. Inorg.
Biochem. 2007, 101, 841–848.
[13] K. D. Mjos, C. Orvig, Chem. Rev. 2014, DOI: 10.1021/cr400460s.
[14] A. V. Klein, T. W. Hambley, Chem. Rev. 2009, 109, 4911–4920.
[15] G. Barone, N. Gambino, A. Ruggirello, A. Silvestri, A. Terenzi, V. T. Liveri,
J. Inorg. Biochem. 2009, 103, 731–737.
Experimental methods
[16] A. Terenzi, C. Ducani, L. Male, G. Barone, M. J. Hannon, Dalton Trans.
2013, 42, 11220–11226.
[17] M. E. Germain, T. R. Vargo, B. A. McClure, J. J. Rack, P. G. Van Patten, M.
Odoi, M. J. Knapp, Inorg. Chem. 2008, 47, 6203–6211.
[18] N. J. Farrer, L. Salassa, P. J. Sadler, Dalton Trans. 2009, 10690–10701;
O. J. Stacey, S. J. A. Pope, RSC Adv. 2013, 3, 25550–25564.
[19] a) K. E. Erkkila, D. T. Odom, J. K. Barton, Chem. Rev. 1999, 99, 2777–2795;
b) A. J. McConnell, H. Song, J. K. Barton, Inorg. Chem. 2013, 52, 10131–
10136.
[20] A. Chantzis, T. Very, A. Monari, X. Assfeld, J. Chem. Theory Comput. 2012,
8, 1536–1541.
[21] T. Etienne, T. Very, E. A. Perpete, A. Monari, X. Assfeld, J. Phys. Chem. B
2013, 117, 4973–4980.
[22] J. Tomasi, B. Mennucci, R. Cammi, Chem. Rev. 2005, 105, 2999–3093.
[23] a) A. Biancardi, T. Biver, A. Marini, B. Mennucci, F. Secco, Phys. Chem.
Chem. Phys. 2011, 13, 12595–12602; b) A. Biancardi, T. Biver, F. Secco, B.
Mennucci, Phys. Chem. Chem. Phys. 2013, 15, 4596–4603.
[24] a) R. L. Martin, J. Chem. Phys. 2003, 118, 4775–4777; b) I. Mayer, Chem.
Phys. Lett. 2007, 437, 284–286; c) A. V. Luzanov, A. A. Sukhorukov, V. ffi.
Umanskii, Theor. Exp. Chem. 1976, 10, 354–361.
[25] A. W. Lange, J. M. Herbert, J. Am. Chem. Soc. 2009, 131, 3913–3922.
[26] A. Spinello, A. Terenzi, G. Barone, J. Inorg. Biochem. 2013, 124, 63–69.
[27] C. Curutchet, A. Munoz-Losa, S. Monti, J. Kongsted, G. D. Scholes, B.
Mennucci, J. Chem. Theory Comput. 2009, 5, 1838–1848.
[28] S. Caprasecca, C. Curutchet, B. Mennucci, J. Chem. Theory Comput.
2012, 8, 4462–4473.
Measurements of pH were made with a Metrohm 713 pH meter
equipped with a combined glass electrode. Spectrophotometric
measurements were performed on a PerkinElmer Lambda 35 spec-
trophotometer. Fluorescence measurements were carried out on
a PerkinElmer LS55 spectrofluorometer. Both apparatuses were
equipped with jacketed cell holders having temperature control
within ꢁ0.18C. Spectrofluorometric titrations were done by adding
increasing amounts of DNA directly to the cell containing the
ZnL2+ solution, with both solutions under the same solvent condi-
tions. In the experiments on the photochemical stability of ZnL2+
,
the illumination was done using a tungsten lamp (60 W) placed at
a fixed distance of 10 cm from the quartz cuvette (10 mm path
length) containing the sample. The solvent polarity was expressed
making use of the ET(30) scale (Dimroth–Reichardt ET parame-
ter[50]).
Acknowledgements
B.M. and A.B. acknowledge the European Research Council
(ERC) for financial support in the framework of the Starting
Grant (EnLight-277755). The financial support of “Obra Social
La Caixa” is also gratefully acknowledged.
[29] A. Chantzis, T. Very, C. Daniel, A. Monari, X. Assfeld, Chem. Phys. Lett.
2013, 578, 133–137.
[30] T. Biver, A. De Biasi, F. Secco, M. Venturini, S. Yarmoluk, Biophys. J. 2005,
89, 374–383.
[31] M. T. Record, C. F. Anderson, T. M. Lohman, Q. Rev. Biophys. 1978, 11,
103–178.
Keywords: density functional calculations · DNA recognition ·
intercalation · N,O ligands · photochemistry · photooxidation
[32] T. Biver, F. Secco, M. R. Tine, M. Venturini, A. Bencini, A. Bianchi, C.
Giorgi, J. Inorg. Biochem. 2004, 98, 1531–1538.
[33] J. B. Chaires, Arch. Biochem. Biophys. 2006, 453, 26–31.
[34] C. Bazzicalupi, A. Bencini, A. Bianchi, T. Biver, A. Boggioni, S. Bonacchi,
A. Danesi, C. Giorgi, P. Gratteri, A. M. Ingrain, F. Secco, C. Sissi, B. Valtan-
coli, M. Venturini, Chem. Eur. J. 2008, 14, 184–196.
[35] Y. Zhao, N. E. Schultz, D. G. Truhlar, J. Chem. Theory. Comput. 2006, 2,
364–382.
[36] E. G. Hohenstein, S. T. Chill, C. D. Sherrill, J. Chem. Theory. Comput. 2008,
4, 1996–2000.
[1] a) P. A. Vigato, S. Tamburini, Coordin. Chem. Rev. 2004, 248, 1717–2128;
b) P. Pfeiffer, E. Breith, E. Lubbe, T. Tsumaki, Justus Liebigs Ann. Chem.
1933, 503, 84–130.
[2] G. Parkin, Chem. Rev. 2004, 104, 699–767.
[3] a) S. J. Wezenberg, E. C. Escudero-Adan, J. Benet-Buchholz, A. W. Kleij,
Org. Lett. 2008, 10, 3311–3314; b) M. E. Germain, M. J. Knapp, J. Am.
Chem. Soc. 2008, 130, 5422–5423; c) M. Cano, L. Rodriguez, J. C. Lima,
F. Pina, A. Dalla Cort, C. Pasquini, L. Schiaffino, Inorg. Chem. 2009, 48,
6229–6235; d) A. D Cort, L. Mandolini, C. Pasquini, K. Rissanen, L. Russo,
L. Schiaffino, New J. Chem. 2007, 31, 1633–1638.
[4] K. Mꢃllen, U. Scherf, Organic Light Emitting Devices: Synthesis Properties
and Application, Wiley-VCH, Weinheim, 2006.
[5] a) N. Madhavan, C. W. Jones, M. Weck, Accounts Chem. Res. 2008, 41,
1153–1165b) D. J. Darensbourg, Chem. Rev. 2007, 107, 2388–2410; c) T.
Katsuki, Chem. Soc. Rev. 2004, 33, 437–444; d) P. G. Cozzi, Chem. Soc.
Rev. 2004, 33, 410–421.
[37] S. F. Sousa, E. S. Carvalho, D. M. Ferreira, I. S. Tavares, P. A. Fernandes,
M. J. Ramos, J. A. N. F. Gomes, J. Comput. Chem. 2009, 30, 2752–2763.
[38] E. Cances, B. Mennucci, J. Tomasi, Journal Chem. Phys. 1997, 107, 3032–
3041.
[39] B. Hess, C. Kutzner, D. van der Spoel, E. Lindahl, J. Chem. Theory.
Comput. 2008, 4, 435–447.
Chem. Eur. J. 2014, 20, 7439 – 7447
www.chemeurj.org
7446
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
[40] J. M. Wang, P. Cieplak, P. A. Kollman, J. Comput. Chem. 2000, 21, 1049–
1074.
[41] a) A. Perez, I. Marchan, D. Svozil, J. Sponer, T. E. Cheatham, C. A. Laugh-
ton, M. Orozco, Biophys. J. 2007, 92, 3817–3829; b) A. T. Guy, T. J.
Piggot, S. Khalid, Biophys. J. 2012, 103, 1028–1036.
[42] a) M. Svensson, S. Humbel, R. D. J. Froese, T. Matsubara, S. Sieber, K. Mo-
rokuma, J. Phys. Chem. 1996, 100, 19357–19363; b) T. Vreven, K. Moro-
kuma, J. Comput. Chem. 2000, 21, 1419–1432.
[43] a) U. C. Singh, P. A. Kollman, J. Comput. Chem. 1984, 5, 129–145; b) B. H.
Besler, K. M. Merz, P. A. Kollman, J. Comput. Chem. 1990, 11, 431–439.
[44] B. T. Thole, Chem. Phys. 1981, 59, 341–350.
M. Bearpark, J. J. Heyd, E. Brothers, K. N. Kudin, V. N. Staroverov, R. Ko-
bayashi, J. Normand, K. R. Raghavachari, T. Nakajima, Y. Honda, O. Kitao,
H. Nakai, T. Vreven, J. Montgomery, J. A., J. E. Peralta, F. Ogliaro, M. Bear-
park, J. J. Heyd, E. Brothers, K. N. Kudin, V. N. Staroverov, R. Kobayashi, J.
Normand, K. Raghavachari, A. Rendell, J. C. Burant, S. S. Iyengar, J.
Tomasi, M. Cossi, N. Rega, J. M. Millam, M. Klene, J. E. Knox, J. B. Cross, V.
Bakken, C. Adamo, J. Jaramillo, R. Gomperts, R. E. Stratmann, O. Yazyev,
A. J. Austin, R. Cammi, C. Pomelli, J. W. Ochterski, R. L. Martin, K. Moro-
kuma, V. G. Zakrzewski, G. A. Voth, P. Salvador, J. J. Dannenberg, S. Dap-
prich, A. D. Daniels, Ö. Farkas, J. B. Foresman, J. V. Ortiz, J. Cioslowski,
D. J. Fox, Gaussian Inc., Wallingford CT, 2009.
[45] P. T. van Duijnen, M. Swart, J. Phys. Chem. A 1998, 102, 2399–2407.
[46] a) P. Slavicek, B. Winter, M. Faubel, S. E. Bradforth, P. Jungwirth, J. Am.
Chem. Soc. 2009, 131, 6460–6467; b) K. C. Lau, C. Y. Ng, Accounts Chem.
Res. 2006, 39, 823–829; c) D. M. Close, J. Phys. Chem. A 2011, 115,
2900–2912; d) T. B. Truong, J. Phys. Chem. 1980, 84, 964–970.
[47] Gaussian 09, Revision H.11, M. J. Frisch, G. W. Trucks, H. B. Schlegel, G. E.
Scuseria, M. A. Robb, J. R. Cheeseman, G. Scalmani, V. Barone, B. Men-
nucci, G. A. Petersson, H. Nakatsuji, M. Caricato, X. Li, H. P. Hratchian,
A. F. Izmaylov, J. Bloino, G. Zheng, J. L. Sonnenberg, M. Hada, M. Ehara,
K. Toyota, R. Fukuda, J. Hasegawa, M. Ishida, T. Nakajima, Y. Honda, O.
Kitao, H. Nakai, T. Vreven, J. Montgomery, J. A., J. E. Peralta, F. Ogliaro,
[48] B. Garcia, J. M. Leal, R. Ruiz, T. Biver, F. Secco, M. Venturini, J. Phys. Chem.
B 2010, 114, 8555–8564.
[49] M. E. Reichmann, S. A. Rice, C. A. Thomas, P. Doty, J. Am. Chem. Soc.
1954, 76, 3047–3053.
[50] C. Reichardt, Angew. Chem. 1965, 77, 30–40; Angew. Chem. Int. Ed. Engl.
1965, 4, 29–40.
Received: December 13, 2013
Published online on May 14, 2014
Chem. Eur. J. 2014, 20, 7439 – 7447
www.chemeurj.org
7447
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim