Modeling and biological investigations of an unusual behavior of novel synthesized acridine-based polyamine ligands in the binding of double helix and G-quadruplex DNA

Article abstract of DOI:10.1002/cmdc.201000332

Three novel 2,7-substituted acridine derivatives were designed and synthesized to investigate the effect of this functionalization on their interaction with double-stranded and G-quadruplex DNA. Detailed investigations of their ability to bind both forms of DNA were carried out by using spectrophotometric, electrophoretic, and computational approaches. The ligands in this study are characterized by an open-chain (L1) or a macrocyclic (L2, L3) framework. The aliphatic amine groups in the macrocycles are joined by ethylene (L2) or propylene chains (L3). L1 behaved similarly to the lead compound m-AMSA, efficiently intercalating into dsDNA, but stabilizing G-quadruplex structures poorly, probably due to the modest stabilization effect exerted by its protonated polyamine chains. L2 and L3, containing small polyamine macrocyclic frameworks, are known to adopt a rather bent and rigid conformation; thus they are generally expected to be sterically impeded from recognizing dsDNA according to an intercalative binding mode. This was confirmed to be true for L3. Nevertheless, we show that L2 can give rise to efficient π-π and H-bonding interactions with dsDNA. Additionally, stacking interactions allowed L2 to stabilize the G-quadruplex structure: using the human telomeric sequence, we observed the preferential induction of tetrameric G-quadruplex forms. Thus, the presence of short ethylene spacers seems to be essential for obtaining a correct match between the binding sites of L2 and the nucleobases on both DNA forms investigated. Furthermore, current modeling methodologies, including docking and MD simulations and free energy calculations, provide structural evidence of an interaction mode for L2 that is different from that of L3; this could explain the unusual stabilizing ability of the ligands (L2>L3>L1) toward G-quadruplex that was observed in this study.

Full text of DOI:10.1002/cmdc.201000332

DOI: 10.1002/cmdc.201000332
Modeling and Biological Investigations of an Unusual
Behavior of Novel Synthesized Acridine-Based Polyamine
Ligands in the Binding of Double Helix and G-Quadruplex
DNA
Carla Bazzicalupi,*[a] Matteo Chioccioli,[b] Claudia Sissi,*[c] Elena Porcꢀ,[c] Claudia Bonaccini,[b]
Claudia Pivetta,[c] Andrea Bencini,[a] Claudia Giorgi,[a] Barbara Valtancoli,[a] Fabrizio Melani,[b]
and Paola Gratteri*[b]
Three novel 2,7-substituted acridine derivatives were designed
and synthesized to investigate the effect of this functionaliza-
tion on their interaction with double-stranded and G-quadru-
plex DNA. Detailed investigations of their ability to bind both
forms of DNA were carried out by using spectrophotometric,
electrophoretic, and computational approaches. The ligands in
this study are characterized by an open-chain (L1) or a macro-
cyclic (L2, L3) framework. The aliphatic amine groups in the
macrocycles are joined by ethylene (L2) or propylene chains
(L3). L1 behaved similarly to the lead compound m-AMSA, effi-
ciently intercalating into dsDNA, but stabilizing G-quadruplex
structures poorly, probably due to the modest stabilization
effect exerted by its protonated polyamine chains. L2 and L3,
containing small polyamine macrocyclic frameworks, are
known to adopt a rather bent and rigid conformation; thus
they are generally expected to be sterically impeded from rec-
ognizing dsDNA according to an intercalative binding mode.
This was confirmed to be true for L3. Nevertheless, we show
that L2 can give rise to efficient p–p and H-bonding interac-
tions with dsDNA. Additionally, stacking interactions allowed
L2 to stabilize the G-quadruplex structure: using the human te-
lomeric sequence, we observed the preferential induction of
tetrameric G-quadruplex forms. Thus, the presence of short
ethylene spacers seems to be essential for obtaining a correct
match between the binding sites of L2 and the nucleobases
on both DNA forms investigated. Furthermore, current model-
ing methodologies, including docking and MD simulations and
free energy calculations, provide structural evidence of an in-
teraction mode for L2 that is different from that of L3; this
could explain the unusual stabilizing ability of the ligands
(L2>L3>L1) toward G-quadruplex that was observed in this
study.
Introduction
The activity of many antimalarial, antibacterial, and anticancer
agents is based on their interaction with helical double-strand-
ed DNA (dsDNA).[1–3] As a consequence, much effort has been
devoted over the past few decades to the design and synthe-
sis of new molecules that can reversibly bind and/or react with
dsDNA; the aim of these research efforts has been to use
these compounds as novel drugs or as probes to better under-
stand the mechanism of action of drugs that are already avail-
able.
minimum is reached.[18] In contrast, cancer cells have evolved
mechanisms to maintain telomere length, the most common
of which is based on the activation of a reverse transcriptase
called telomerase; this occurs in 80–90% of tumor cells.[19–21]
Novel antitumor strategies are aimed at interfering with the
mechanisms of telomere maintenance by targeting telomerase
[a] Dr. C. Bazzicalupi, Prof. A. Bencini, Dr. C. Giorgi, Prof. B. Valtancoli
Department of Chemistry “Ugo Schiff”, University of Florence
Via della Lastruccia 3-13, 50019 Sesto Fiorentino, Florence (Italy)
Fax: (+39)055-457-3364
More recently, interest has also focused on non-canonical
DNA structures. For example, a great deal of attention has
been devoted to clarifying the structural features of telomeric
DNA.[4–8] In fact, the telomeric sequence has been found to be
closely related to the immortalization process of cancer
cells[9–14] and genetic stability.[15–17] Therefore, it represents a
potentially suitable target for anticancer therapy. Telomeres
consist of guanine-rich sequence repeats (in humans, the hexa-
nucleotide motif d(TTAGGG) for example) located at the end of
chromosomes, where their function is to preserve chromo-
some integrity. Telomeric DNA is gradually shortened in normal
cells as a function of the replication cycle; this leads to cell-
cycle arrest and eventually apoptosis when a critical length
E-mail: carla.bazzicalupi@unifi.it
[b] Dr. M. Chioccioli, Dr. C. Bonaccini, Prof. F. Melani, Prof. P. Gratteri
Laboratory of Molecular Modeling Cheminformatics & QSAR
Department of Pharmaceutical Sciences, University of Florence
Via Ugo Schiff 6, 50019 Sesto Fiorentino, Florence (Italy)
Fax: (+39)055-457-3780
E-mail: paola.gratteri@unifi.it
[c] Prof. C. Sissi, Dr. E. Porcꢀ, Dr. C. Pivetta
Department of Pharmaceutical Sciences, University of Padova
Via Marzolo 5, 35100 Padova, (Italy)
Fax: (+39)049-827-5366
E-mail: claudia.sissi@unipd.it
Supporting information for this article is available on the WWW under
http://dx.doi.org/10.1002/cmdc.201000332.
ChemMedChem 2010, 5, 1995 – 2005
ꢁ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
1995

C. Bazzicalupi, C. Sissi, P. Gratteri, et al.
MED
directly according to an enzyme recognition process, or indi-
rectly by means of telomere-interacting agents.[22]
The human telomeres are largely double-stranded DNA se-
quences, but the terminal 100–200 nucleotides at the 3’ end
are single stranded. This portion can easily fold into a G-quad-
ruplex arrangement, a DNA secondary structure that consists
of four guanines (G-quartet or G-tetrad) stacked in coplanar
cyclic arrays associated by eight Hoogsteen hydrogen bonds.
Optimal telomerase activity requires an unfolded single-strand-
ed DNA as substrate. Therefore, ligands that selectively bind to
and stabilize G-quadruplex structures may interfere with the
telomerase enzymatic process.
The number of identified G-quadruplex ligands has grown
rapidly over the past few years. Extensive efforts have been
made to establish reliable structure–activity relationships with
the aim of identifying effective and selective telomerase inhibi-
tors. As common structural features, they generally share a
large, flat, aromatic surface, and the presence of protonatable
side chains.[22]
Classic dsDNA intercalators are likely to correspond to this
pharmacophoric model. Indeed, several derivatives of DNA
binders have also been considered as G-quadruplex binders.
An interesting example is represented by acridine derivatives.
Effective anticancer drugs containing this heterocyclic moiety
(such as amsacrine, m-AMSA) are generally able to efficiently
recognize DNA according to an intercalation binding mode.
However, several studies have shown that by modulating the
substitution pattern on this aromatic ring system, it is possible
to preferentially direct such ligands toward G-quadruplex
structures. In particular, a series of 3,6-disubstituted[23–25] and
3,6,9-trisubstituted acridines,[26–29] as well as some compounds
characterized by a macrocyclic skeleton containing the acridine
moiety[30,31] or the related quinacridine system,[32] have been re-
ported as efficient G-quadruplex binders endowed with phar-
macological anticancer properties.
Scheme 1. Reagents and conditions: a) K2CO3, CH3CN, reflux, 5 h; b) HBr,
CH3COOH, 908C, 24 h.
On the other hand, different substitution patterns, for exam-
ple at the 2,7-position, have, to date, been poorly investigated
as G-quadruplex binders, because they apparently do not pro-
vide relevant G-quadruplex recognition.[33]
groups in hydrogen bromide/acetic acid in the presence of
phenol as an antioxidant. Similarly, the reaction of 1 with
1,4,7,10-tetratosyl-1,4,7,10-tetraazadecane (4)[35] and 1,5,9,13-
tetratosyl-1,5,9,13-tetraazatridecane (5)[36] afforded the respec-
tive tosylated macrocycles 6 and 7, which were then depro-
tected in HBr/CH3COOH/phenol, and further isolated as hydro-
bromide salts (L2 and L3).
Herein we report the synthesis of a series of polyaza ligands
featuring an acridine moiety inserted in an open-chain (L1) or
macrocyclic (L2 and L3) aliphatic polyamine framework via
functionalization of the 2- and 7-positions of the heteroaro-
matic system. These ligands enable the exploitation of the
effect that a cyclic organization of the aliphatic polyamine
chain has on the recognition properties for various DNA struc-
tures; thus, we decided to carry out a comprehensive study on
the binding properties of L1, L2, and L3 toward telomeric G-
quadruplex structures, and to compare them with their respec-
tive dsDNA binding profiles.
All new ligands under investigation feature one or two poly-
amine chains linked to the acridine moiety, and facile protona-
tion of the amine groups occurs in aqueous solution. Ligand
1
protonation was studied by potentiometric and H NMR meas-
urements in aqueous solution (see figures S1, S2, and a de-
tailed description in the Supporting Information). The stepwise
basicity constants (logK) potentiometrically determined at
298 K, are listed in Table 1.
Results and Discussion
Ligand L1 was synthesized by reaction of 2,7-dibromomethyl-
acridine (1 in Scheme 1)[34] with the ditosylated amine 1-methyl-
1,4-di(4-toluenesulfonyl)-1,4-diazabutane (2) in the presence of
potassium carbonate, followed by removal of the tosyl (Ts)
All ligands can bind up to five acidic protons in the 2.5–10.5
pH range investigated, effectively affording various protonated
species at different alkaline to acidic pH values. In all three
cases, the diprotonated form [H2L]2+ (L=L1, L2, or L3) is the
1996
www.chemmedchem.org
ꢁ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemMedChem 2010, 5, 1995 – 2005

Acridine-Based Polyamine DNA Binding Ligands
ters for the binding process of L1 and L2 to calf thymus DNA,
evaluated according to the McGhee and Von Hippel formal-
ism.[37] Data for m-AMSA are also included for comparison.
The data in Table 2 indicate that our novel compounds have
slightly higher binding affinity than m-AMSA. The exclusion pa-
rameter (n), indicating the number of DNA bases involved in
Table 1. Protonation constants of ligands L1–L3.[a]
Equilibrium
LogK
L2
L1
L3
L+H+ =[HL]+
9.76(1)
9.24(1)
5.96(1)
5.26(1)
3.21(2)
10.2(1)
8.3(1)
5.6(1)
3.5(2)
3.0(1)
10.1(1)
8.8(2)
7.3(1)
6.3(1)
3.3(2)
[HL]+ +H+ =[H2L]2+
[H2L]2+ +H+ =[H3L]3+
[H3L]3+ +H+ =[H4L]4+
[H4L]4+ +H+ =[H5L]5+
[a] Determined potentiometrically in Me4NCl (0.1m) at 298 K; values in pa-
rentheses are standard deviations.
Table 2. Thermodynamic binding parameters describing ligand–ctDNA
complex formation.[a]
Ligand
Ka [10ꢀ5 mꢀ1
]
n [bases]
1
most abundant species at physiological pH. H NMR investiga-
L1
L2
m-AMSA
1.07ꢁ0.11
1.55ꢁ0.20
0.36ꢁ0.03
5.2ꢁ0.1
3.1ꢁ0.1
3.2ꢁ0.3
tions indicated that the two first protonation steps occur on
the methylated nitrogen atoms in the case of L1, and on the
central nitrogen atoms of the tetra-amine chain in the case of
L2 and L3.
[a] Carried out in 10 mm Tris, 20 mm KCl, pH 7.5, 258C; Ka is the binding
constant, and n is the exclusion number.
Acridine derivatives binding to dsDNA: solution studies
ligand recognition, although slightly less in the case of L2, is
close to the theoretical value of 4 required for an intercalative
binding mode. This leads us to assume a similar mode of DNA
interaction for m-AMSA and the two novel compounds.
To clarify this point, we performed topoisomerase I unwind-
ing assays. The assay involves incubation of supercoiled plas-
mid DNA with excess topoisomerase I and increasing concen-
trations of test compound. Topoisomerase I converts super-
coiled into relaxed plasmid DNA. After removal of the drug
and enzyme, re-supercoiling of the DNA occurs if the ligand
had been bound by an intercalative mode.[38]
All derivatives were characterized spectroscopically, and their
binding to ctDNA was evaluated quantitatively by monitoring
the change in absorption properties of the ligands in the pres-
ence of increasing concentrations of nucleic acid. Unfortunate-
ly, under the experimental conditions used, it was not possible
to quantitatively monitor the DNA binding process for L3. In
fact, this ligand was very effective at inducing ctDNA precipita-
tion at even the lowest binding ratios examined. Spectra re-
corded for L1 are shown as examples in Figure 1a. L2 exhibits
analogous behavior. Table 2 lists the thermodynamic parame-
A summary of these results is shown in Figure 2. Derivatives
L1 and L2 behave similarly to the reference compound m-
AMSA, thus confirming their ability to promote DNA unwind-
ing.[39] In contrast, L3 did not exhibit any re-supercoiling effect
across the whole drug concentration range tested.
Figure 1. a) Spectral changes of L1 with increasing ctDNA concentration and
b) binding isotherms for the process of L1 binding to dsDNAs of different
base composition (indicated). Titrations were performed in 10 mm Tris,
20 mm KCl, pH 7.5, 258C.
Figure 2. Topoisomerase I-mediated unwinding assays for compounds a) L1
and b) L2; sc: supercoiled DNA; r: DNA relaxed by topoisomerase I. Products
were resolved on a 1% agarose gel in 1ꢂ TAE buffer.
ChemMedChem 2010, 5, 1995 – 2005
ꢁ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemmedchem.org
1997

C. Bazzicalupi, C. Sissi, P. Gratteri, et al.
MED
Because the affinity of a ligand for DNA can be affected by
the DNA sequence, we monitored the binding process of the
three ligands to oligonucleotides of defined base composition,
namely poly(dA–dT) and poly(dG–dC) (Figure 1b, showing data
for L1). Binding isotherms in this figure show how the binding
of L1 and L2 to the tested DNA sequences are essentially
equal, and thus they confirm a lack of sequence selectivity. For
comparison, the thermodynamic binding parameters for the
complex formation of L1 and L2 with poly(dG–dC) are listed in
Table 3.
Table 3. Thermodynamic binding parameters for complex formation of
L1 and L2 with poly(dG–dC).[a]
Ligand
Ka [10ꢀ5 mꢀ1
]
n [bases]
L1
L2
0.89ꢁ0.07
1.26ꢁ0.18
4.9ꢁ0.1
2.8ꢁ0.1
[a] Carried out in 10 mm Tris, 20 mm KCl, pH 7.5, 258C; Ka is the binding
constant, and n is the exclusion number.
Finally, to further evaluate the dsDNA binding properties of
the novel tested ligands, we monitored the thermal stability of
a dsDNA sequence in the presence of increasing drug concen-
trations. The target DNA was prepared by annealing two com-
plementary oligonucleotides, one 5’-end labeled with fluores-
cein, and the other 3’-end labeled with a quencher (DABCYL).
In the double-stranded form, the two labeling groups are in
close proximity, and thus the fluorescence signal is quenched;
upon dissociation (melting), the fluorescence signal is en-
hanced. In Figure 3a, the variation of the DNA melting temper-
ature (Tm) induced by increasing drug concentrations is report-
ed. Under our experimental conditions, a good correlation be-
tween unwinding data and dsDNA stabilization emerged.
Indeed, L1 and L2 turned out to be effective in stabilizing
dsDNA, with DTm values similar to those recorded with m-
AMSA. Conversely, L3 did not promote duplex stabilization.
Altogether, these results allowed us to draw a picture of the
dsDNA binding pattern for these acridine derivatives. Although
all ligands efficiently interact with DNA, the nature of the side
chains plays a key role in the binding mode. Experimental re-
sults seem to suggest a preferential, non-intercalative binding
mode for L3. In this case, the driving force is mainly the elec-
trostatic interaction of the charged ligand with the nucleic
acid, which promotes DNA precipitation. In contrast, L1 and L2
showed similar behavior, resembling that of m-AMSA, thus
suggesting a more favorable balance of the p–p interactions
between the acridine moiety and the base pairs, on the one
hand, and the H-bond and salt bridge interactions, supported
by the charged substituents, on the other.
Figure 3. Variation of the melting temperature (DTm) of oligonucleotides ar-
ranged into a) double-stranded or b) G-quadruplex structures upon addition
of increasing concentrations of acridine derivatives in phosphate buffer con-
taining 50 mm KCl, pH 7.4; heating rate: 0.28Cminꢀ1
.
macrocycles L2 and L3 than for L1. Surprisingly, whereas no
such intercalation seems to be present with L3, L2, which con-
tains an overall shorter cyclic aliphatic chain, is still able to in-
teract with dsDNA in a manner similar to the acyclic L1. Thus,
to gain a more in-depth insight, we decided to carry out a
modeling investigation.
Acridine derivatives binding to dsDNA: modeling studies
The binding mode of the diprotonated species of L1, L2, and
L3 toward dsDNA was investigated by docking procedures fol-
lowed by molecular dynamics (MD) simulations, with an explic-
it treatment of the aqueous environment.
The results shown in Figure 4 and in figure S3 (Supporting
Information) largely corroborate the solution studies, as the
three ligands feature different binding modes toward the poly-
G–poly-C double helix. For the open-chain ligand L1, the inter-
action can be described as properly intercalative both in the
case of the CG site, as well as for the GC site. In all situations
the adduct was stabilized by additional H-bond contacts in-
volving the terminal protonated nitrogen atoms and the car-
bonyl oxygen atom belonging to the cytosine and guanine
units (Figure 4a), or the phosphate groups (figure S3a, Sup-
porting Information). In fact, the planar acridine moiety of L1
can be easily lodged between the base pairs, unhindered by
the pendant arms, which, on the contrary, can be set in the
grooves. Moreover, the RMSD value evaluated as a function of
time during the recorded MD trajectory (figure S4, Supporting
It is well known that small macrocycles such as L2 and L3,
which contain a rigid aromatic moiety in their structures, often
adopt bent conformations.[40] As a consequence, one could
easily propose that the insertion of the acridine moiety be-
tween DNA base pairs would be more difficult in the case of
1998
www.chemmedchem.org
ꢁ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemMedChem 2010, 5, 1995 – 2005

Acridine-Based Polyamine DNA Binding Ligands
the binding sites of L2 and the nucleobases. The relative free
energies for these complexes were estimated through the mo-
lecular mechanics Poisson–Boltzmann surface area (MM-PBSA)
approach. Importantly, the MM-PBSA results are in agreement
with the thermodynamic parameters obtained by the solution-
phase studies, showing the higher intercalative ability of L1
and L2 than that of L3 (Table 4).
Table 4. Gibbs free energy values calculated by using the MM-PBSA ap-
proach for adducts formed between ligands L1–L3 and DNA.
Ligand
DG [kcalmolꢀ1
]
dsDNA[b]
dsDNA[a]
G-tetrad DNA
L1
L2
L3
ꢀ5.20
ꢀ9.00
8.85
ꢀ5.06
2.08
5.65
ꢀ9.60
ꢀ16.54
ꢀ13.47
[a] CG intercalative site. [b] GC intercalative site.
Acridine derivatives binding to G-quadruplex forming
sequences: solution studies
The potential interaction of tested ligands with G-quadruplex
structures was evaluated by monitoring both the stabilization
and the induction of G-quadruplex structures assumed by telo-
meric sequences. The former were assayed by fluorescence
melting experiments: in this case an oligonucleotide (HTS) was
used that contains four repeats of the human telomeric se-
quence and which is labeled at the 5’ end with DABCYL and at
the 3’ end with fluorescein. Again, upon folding into a G-quad-
ruplex, these two labels are brought into close proximity, and
fluorescence is quenched. With an increase in temperature, the
DNA melts and the fluorescein signal can be detected.
Figure 4. Results of the modeling procedures (docking and MD) for the ad-
ducts formed by diprotonated forms of a) L1, b) L2, or c) L3 and poly-G-poly-
C dsDNA. Results were obtained by starting from docking in a CG intercala-
tive binding site; hydrogen bonds are indicated by green lines.
As shown in Figure 3b, all our acridine derivatives can shift
the G-quadruplex melting temperature to higher values. The
values determined herein are lower than those reported for
other G-quadruplex-selective acridine derivatives (DTm >308C
was reported for BRACO-19 at 1 mm), but are relevantly higher
than those observed toward dsDNA.[33] These data clearly illus-
trate how the efficiency of our novel ligands is modulated by
the nature of the side chains. In particular, L2 appeared to be
the most active in stabilizing a G-quadruplex structure, fol-
lowed by L3 and L1. Notably, m-AMSA was confirmed to be a
poor G-quadruplex binder. Thus, the similarity between L1 and
L2 in the dsDNA binding properties is not conserved when the
target sequence assumes a G-quadruplex structure.
Information) denotes that the L1–poly-G–poly-C adduct is
quite stable, and it is not disrupted during the 10 ns simulation
time.
On the other hand, L3, most likely hindered by the bent
conformation adopted by its macrocyclic structure, seems to
fail at intercalation and gives rise to groove binding, interact-
ing mainly through hydrogen bonds between its protonated
nitrogen atoms and the polyphosphate DNA backbone (Fig-
ure 4c and figure S3c, Supporting Information). After 10 ns MD
simulation, the DNA double helix is almost completely rebuilt,
and no trace of the CG or GC intercalative site remains.
As far as L2 is concerned, it shows an intermediate behavior
between L1 and L3. In spite of its rather rigid and strained
bent conformation, which is generally considered unsuitable
for the stereochemical requirement of a classic DNA intercala-
tor, modeling studies give evidence of stable conformations
for the L2–DNA adduct (Figure 4b), in which the macrocycle is
able to give rise to simultaneous p–p and H-bonding interac-
tions. Indeed, L2 behaves as a multifunctional ligand toward
DNA by virtue of its small size. The presence of the shorter eth-
ylene spacers in place of propylene units featured in L3 ap-
pears to be essential for obtaining a correct match between
To determine if this interaction reflects any ability of our
new tested compounds to induce the formation of G-quadru-
plex structures, we performed electrophoretic mobility shift
assays (EMSA), using an oligonucleotide containing only two
human telomeric repeats (2GGG). This sequence can form G-
quadruplex structures only by pairing two or four strands, thus
leading to dimeric or tetrameric structures resolvable by poly-
acrylamide gel electrophoresis (Figure 5). Among the tested ac-
ridine derivatives, only L2 was able to induce the folding of
2GGG into a tetrameric G-quadruplex structure, thus confirm-
ing it as the most efficient G-quadruplex binder.
ChemMedChem 2010, 5, 1995 – 2005
ꢁ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemmedchem.org
1999

C. Bazzicalupi, C. Sissi, P. Gratteri, et al.
MED
by H-bonding with a phosphate
group of the other dimer (Fig-
ure 6b and 6c). However, com-
pared with L3, L2 causes a less
significant disruption of the TATA
tetrad, also present at the 3’-
and 5’-end interface in the bio-
logical unit of the BRACO-19–
DNA complex. In particular, only
one adenine residue is consider-
ably shifted from its original po-
sition, and, at the same time, the
carbonyl group of a thymine res-
idue gives rise to an additional
Figure 5. Effect of increasing of concentrations of a) L2 or b) L3 on the assembly of 2GGG (1 mm strand concentra- H-bond interaction with a pro-
tion) into dimeric and tetrameric G-quadruplex structures in 10 mm Tris, 1 mm EDTA, 100 mm KCl, pH 8.0, at 378C
tonated nitrogen of the ligand,
for 24 h. Reaction products were resolved on a 16% native polyacrylamide gel in 0.5ꢂ TBE containing 20 mm KCl;
thus exerting a significant role in
M=monomeric oligonucleotide, D=dimeric oligonucleotide, T=tetrameric G-quadruplex.
the stabilization of the adduct
(Figure 6b). However, in the case
The induction of the tetrameric DNA arrangement for the
2GGG telomeric sequence correlates well with recent crystallo-
graphic data showing a related trisubstituted acridine bound
to a G-quadruplex[7] with two dimeric units held together by
one drug molecule stacked on the terminal G-quartets. Thus
this crystallographic structure was considered a suitable model
for computational studies.
Acridine derivatives binding to G-quadruplex-forming
sequences: modeling studies
The orientations resulting from MD simulation carried out on
selected poses of the acridine derivatives in complex with the
G-quadruplex obtained from docking are shown in Figure 6.
All these adducts remained quite stable during the simulation
(figure S5, Supporting Information).
All ligands insert into the tetrameric form of the G-quadru-
plex, between the 3’-end G-tetrad belonging to one G-quadru-
plex dimer and the 5’-end of the other, in a manner similar to
that shown by the very effective G-quadruplex ligand BRACO-
19 present in the X-ray crystallographic complex structure.[7]
However, the acridine moieties of L2 and L3 give rise to stron-
ger p–p interactions with the 3’-end G-tetrad than in the case
of L1, the acridine system of which shows the most significant
deviation from the expected parallel position (168 between the
involved planes versus 38 and 3.58 for L2 and L3, respectively).
Moreover, all ligands cause, to varying degrees, a rotation of
one dimeric G-quadruplex unit relative to the other. Using the
BRACO-19–G-quadruplex complex as a reference, and consider-
ing the dihedral angle formed between 3’- and 5’-end G-tet-
rads, a 18, 38, and 98 twist (figure S6, Supporting Information)
is observed for L2, L3, and L1, respectively.
Analysis of the L1–DNA complex shows that both p-stacking
and H-bonding interactions involve only one dimer, whereas
Figure 6. Results of MD simulations of the adducts formed by the diproto-
both the G-quadruplex units are simultaneously bound by L2
nated species a) L1, b) L2, and c) L3 with the G-quadruplex structure formed
and L3 (Figure 6). As far as L2 and L3 are concerned, their
by the 2GGG oligonucleotide. Hydrogen bonds are indicated by green lines;
interacting and non-interacting DNA monomers (tube representation) are
colored in light blue and gray, respectively.
binding modes seem similar, with both ligands interacting by
p-stacking with a guanidine residue of the 5’-end G-tetrad, and
2000
www.chemmedchem.org
ꢁ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemMedChem 2010, 5, 1995 – 2005

Acridine-Based Polyamine DNA Binding Ligands
of the L3 complex, both the aforementioned adenine and thy-
mine residues are in a different position, and no additional H-
bond contacts are established.
poorly recognizes G-quadruplex structures. Ligand L2 features
a small macrocyclic framework which is generally expected to
adopt a rather bent and rigid conformation, usually not prone
to intercalative interactions with DNA. However, in spite of
this, L2 binds to dsDNA almost as well as the open-chain
ligand L1, and it shows the best performance toward a G-
quadruplex. Distinctly, ligand L3, characterized by the same
cyclic structure of L2, but with longer and more flexible propyl-
ene spacers, should, in principle, be able to adopt a planar
conformation more readily. Nevertheless, it is unable to interca-
late into dsDNA, nor can it induce G-quadruplex structures.
As pointed out by the molecular modeling studies, the
better performance of L2 relative to that of L3 and the open-
chain ligand L1 can be attributed entirely to its particular con-
formation and dimensions, which establish an optimal match
with both dsDNA and the tetrameric G-quadruplex structure.
The distinct DNA binding modes characterized for our novel
ligands could explain their effect on telomerase, with L2 being
a good inhibitor of this enzyme, a property that is coupled
with favorably low short-term cytotoxicity. These results indi-
cate how important it is to set up simultaneous combinations
of interactions of various kinds between DNA and ligand in
order to reach an optimal binding efficiency, thus effectively
opening new possibilities for the design of novel selective
DNA binders.
Altogether, the observations derived from the modeling pro-
cedures support the quite surprising results obtained from the
solution studies, which show evidence that the bent conforma-
tions of L2 and L3 do not prevent G-quadruplex recognition.
The binding free energy values obtained by the MM-PBSA cal-
culations indicate that binding affinity decreases in the order
L2>L3>L1 (Table 4), in agreement with the reported solution
studies.
Enzymatic and cellular activity
From the results discussed above, L2 turns out to be a poten-
tial telomerase inhibitor. Therefore, we decided to determine
whether L2 is able to efficiently interfere with the enzyme-
mediated elongation process of a proper template. Indeed, we
observed that L2 can efficiently inhibit telomerase activity with
an IC50 value of 1.7 mm, well below the drug concentration re-
quired to interfere with the Taq polymerase amplification pro-
cess (10 mm; figure S7, Supporting Information). For compari-
son, L1 was tested in the same way, and it was confirmed to
be unable to decrease telomerase activity.
Finally, the toxicity of tested compounds was evaluated with
the HeLa cancer cell line, and the results are summarized in
Table 5. Our data show that after drug exposure for 72 h, all
novel derivatives are less cytotoxic than m-AMSA. The most cy-
totoxic compound is derivative L3, with an IC50 value of 32 mm,
whereas for L2 and L4, IC50 values are >100 mm.
Experimental Section
Materials
The human telomeric sequence HTS 5’-[DABCYL]-AGG-GTT-AGG-
GTT-AGG-GTT-AGG-GT-[FAM]-3’ was synthesized and purified by
ATDBIO (Southampton, UK). Oligonucleotides 2GGG (5’-TAC-AGA-
TAG-TTA-GGG-TTA-GGG-TTA-3’), 1GGG (5’-TAC-AGA-TAG-TTA-GGG-
TTA-GAC-TTA-3’), poly(dA–dT)25, poly(dG–dC)25, QMup (5’-[FAM]-
GTG-AGA-TAC-CGA-CAG-AAG-3’), QMdown (5’-CTT-CTG-TCG-GTA-
TCT-CAC-[DABCYL]-3’), TS (5’-AAT-CCG-TCG-AGC-AGA-GTT-3’), ACX
(5’-GTG-CCC-TTA-CCC-TTA-CCC-TTA-CCC-TAA-3’), Tup (5’-TGA-GGA-
TCC-GCC-TGG-ACA-GCA-TGG-3’), and Tdown (5’-GTC-GAA-TTC-TCG-
GCG-AGA-AGC-AGG-3’) were provided by Eurogentec (Belgium).
QMup and QMdown were mixed at equimolar concentrations,
heated at 958C for 5 min, and then cooled to room temperature
overnight before use. Calf thymus DNA (ctDNA) and plasmid
pBR322 were purchased from Sigma–Aldrich (USA) and Fermentas,
respectively, and used with no further purification. Reagents and
solvents for synthesis were purchased from Aldrich and used with-
out further purification.
Table 5. IC50 values for all tested compounds obtained in HeLa cells.
Compd
IC50 [mM]
m-AMSA
2.0ꢁ0.4
>100
>100
32ꢁ6.2
L1
L2
L3
This result is an interesting starting point. Indeed, a canoni-
cal telomerase inhibitor can be devoid of any toxic effect after
such short time exposures. Thus the properties associated with
the G-quadruplex interaction profile of L2, as revealed through
our work, indicate this compound to be a suitable lead candi-
date worthy of further optimization.
Ligand synthesis
Ligands L1, L2, and L3 were synthesized by reaction of 2,7-dibro-
momethylacridine (1, Scheme 1)[34] with 1-methyl-1,4-ditosyl-1,4-di-
azabutane (2), 1,4,7,10-tetratosyl-1,4,7,10-tetraazadecane (4),[35] and
1,5,9,13-tetratosyl-1,5,9,13-tetraazatridecane (5).[36] The resulting to-
sylated products 3, 6, and 7 were deprotected in a mixture of HBr/
CH3COOH.
Conclusions
The new 2,7-substituted acridine derivatives L1, L2, and L3
show an unusual trend in their binding patterns toward canon-
ical double-helical DNA and non-canonical G-quadruplex struc-
tures. The DNA recognition profile of L1 is in line with the ex-
pectation, as it parallels the behavior shown by m-AMSA: it
binds to dsDNA through a pure intercalative process and
Bis[N,N’-bistosyl-2-methylaminoethylaminomethyl]-(2,7)-acridine
(3): A suspension of 1 (1.91 g, 5.233 mmol) in dry CH3CN (170 mL)
was added dropwise over a period of 4 h to a and vigorously
stirred suspension of 2 (5.21 g, 13.6 mmol) and K2CO3 (13.6 g,
ChemMedChem 2010, 5, 1995 – 2005
ꢁ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemmedchem.org
2001

C. Bazzicalupi, C. Sissi, P. Gratteri, et al.
MED
0.1 mol) in dry CH3CN (80 mL) at reflux. After the addition was
completed, the solution was held at reflux for an additional 5 h.
The resulting suspension was filtered, and the solution was evapo-
rated under vacuum to give a crude oil, which was purified by
column chromatography on neutral alumina (activity II/III, CH2Cl2/
EtOAc, 8:1 v/v as eluent). The eluted fractions were collected and
evaporated to dryness to afford 3 as a white solid. Yield: 1.98 g
(2.04 mmol, 39%); elemental analysis: calcd (%) for C49H53N5S4O8: C
60.79, H 5.52, N 7.23, found: C 61.1, H 5.4, N 7.3.
(d, 2H), 7.93 (d, 2H) 4.59 (s, 4H), 3.06 (m, 4H), 2.71 (m, 4H), 2.23
(m, 4H), 1.66 (m, 4H), 1.44 ppm (m, 2H); 13C NMR (D2O, pH 4.26):
d=149.31, 141.33, 138.39, 132.76, 130.04, 126.44, 122.42, 49.69,
44.64, 44.40, 42.66, 22.23, 21.94 ppm.
Potentiometric
measurements.
Potentiometric
titrations
(ꢀlog[H+]) were carried out with 0.1m N(CH3)4Cl (pKw =13.83) at
25.0ꢁ0.18C by using equipment and procedures previously de-
scribed.[41] The computer program HYPERQUAD[42] was used to cal-
culate the protonation constants of the ligands from emf data. Dis-
tribution diagrams were calculated by using the Hyss program.[43]
Bis(2-methylaminoethylaminomethyl)-(2,7)-acridine pentahydro-
bromide (L1·5HBr·H2O): Compound 3 (0.97 g, 1 mmol) and phenol
(10.5 g, 0.112 mol) were dissolved in a mixture of HBr/CH3COOH
(33%, 100 mL). The reaction was stirred at 908C for 24 h until a
precipitate was formed. The solid was filtered out and washed sev-
eral times with CH2Cl2. The pentahydrobromide salt was recrystal-
lized from EtOH/H2O (3:1) to yield 0.41 g (0.525 mmol, 52.5%). Ele-
mental analysis: calcd (%) for C21H29N5·5HBr·H2O: C 32.59, N 9.05, H
NMR spectroscopic measurements. 1H (300.07 MHz) and 13C
(75.46 MHz) NMR spectra were recorded at 298 K in CDCl3 and D2O
solutions at various pH values on a Varian Gemini 300 spectrome-
ter. Small amounts of 0.01m NaOD and DCl were added to the sol-
utions to adjust the pD. The pH was calculated from the measured
pD values by using the relationship: pH=pDꢀ0.40.[44]
DNA binding studies. Ligand–DNA spectroscopic titrations were
performed at 258C in 10 mm Tris, 20 mm KCl, pH 7.5 with a Perki-
nElmer Lambda 20 apparatus equipped with a Haake F3-C thermo-
stat. Binding was monitored by recording the signal in the ligand
absorption range after the addition of scalar amounts of DNA to a
freshly prepared drug solution. For each drug/DNA ratio, the frac-
1
4.69, found: C 32.6, N 9.1, H 4.7; H NMR (D2O, pH 2.50): d=9.60 (s,
1H), 8.50 (s, 2H), 8.38 (d, 2H), 8.18 (d, 2H) 4.65 (s, 4H), 3.64 (m,
4H), 3.53 (m, 4H), 2.84 ppm (s, 6H); 13C NMR (D2O, pH 2.50): d=
150.6, 139.8, 138.5, 131.9, 130.3, 125.9, 120.65, 50.6, 44.3, 42.9,
33.3 ppm.
2,4,7,10-Tetratosyl-2,4,7,10-tetraaza[12]-(2,7)-acridinophane (6):
A suspension of 1 (1.40 g, 3.83 mmol) in dry CH3CN (140 mL) was
added dropwise over a period of 4 h to a vigorously stirred sus-
pension of 4 (3.5 g, 4.59 mmol) and K2CO3 (9.82 g, 71.0 mmol) in
dry CH3CN (70 mL) at reflux. After the addition was completed, the
solution was held at reflux for an additional 5 h. The resulting sus-
pension was filtered, and the solution was evaporated under
vacuum to give a crude oil, which was purified by column chroma-
tography on neutral alumina (activity II/III, CH2Cl2/EtOAc, 8:1 v/v as
eluent). The eluted fractions were collected and evaporated to dry-
ness to afford 6 as a white solid. Yield: 1.55 g (1.60 mmol 41.8%);
elemental analysis: calcd (%) for C49H51N5S4O8: C 60.91, H 5.32, N
7.25, found: C, 60.8; H, 5.3; N, 7.2.
tion of bound ligand was calculated [n=(eꢀe0)/(e ꢀe0), for which
1
e0 and e are the extinction coefficients of the free and DNA-
1
bound ligand, respectively]. Data were evaluated according to the
equation derived by McGhee and Von Hippel.[37]
Topoisomerase I DNA unwinding assay. Supercoiled pBR322 plas-
mid DNA (0.15 mg) was incubated with 1 U topoisomerase I (Invi-
trogen) and increasing concentrations of tested compounds for
24 h at 378C. Reactions were terminated by two extractions with
one volume of phenol/CHCl3/isoamyl alcohol (25:24:1). Samples
were loaded on an agarose gel (1%) and run at 40 V for 3 h. Gels
were stained with ethidium bromide and photographed.
Fluorescence melting studies. The melting temperature of fluores-
cein-labeled DNAs in the presence or absence of ligands was de-
termined by fluorescence melting experiments performed in a
Roche LightCycler, using an excitation source at l 488 nm, and
reading the fluorescence emission at l 520 nm. Melting experi-
ments were performed in a total volume of 20 mL containing
0.25 mm DNA and variable concentrations of tested derivatives in
LiP buffer (10 mm LiOH, 50 mm KCl, pH 7.4 with H3PO4). Reaction
mixtures were first denatured by heating at 958C for 5 min, and
then cooled to 308C at a rate of 0.58C minꢀ1. The temperature was
then slowly increased (0.28Cminꢀ1) up to 908C and again lowered
at the same rate to 308C. Recordings were taken during both the
melting and annealing processes to check for hysteresis. Tm values
were determined from the first derivatives of the melting profiles
using the Roche LightCycler software. Each curve was repeated at
least three times, and errors were ꢁ0.48C.
Electrophoretic mobility shift assay. Single-stranded oligomer
2GGG was 5’-end labeled with 32P. DNA was allowed to fold over-
night, and then increasing ligand concentrations in 10 mm Tris-HCl,
1 mm EDTA, 50 mm KCl, pH 8.0 were added. Reaction mixtures
were incubated at 258C for 30 min and then loaded on native
(non-denaturing) 16% polyacrylamide gels. Electrophoresis pro-
ceeded for 3 h in 0.5ꢂ TBE supplemented with 20 mm KCl. Gels
were dried, and the resolved bands were visualized and quantified
on a PhosphorImager (Amersham).
2,4,7,10-Tetraaza[12]-cyclo(2,7)-acridinophane pentahydrobro-
mide (L2·5HBr): Compound 6 (1.00 g, 1 mmol) and phenol (10.5 g,
0.112 mol) were dissolved in a mixture of HBr/CH3COOH (33%,
100 mL). The reaction was stirred at 908C for 24 h until a precipi-
tate was formed. The solid was filtered out and washed several
times with CH2Cl2. The pentahydrobromide salt was recrystallized
from EtOH/H2O (3:1) to yield 0.51 g (0.68 mmol, 68%). Elemental
analysis: calcd (%) for C21H27N5·5HBr: C 33.45, N 9.29, H 4.28,
1
found: C 33.3, N 9.2, H 4.3; H NMR (D2O, pH 4.10): d=9.89 (s, 1H),
8.62 (s, 2H), 8.36 (d, 2H), 8.24 (d, 2H) 4.61 (s, 4H), 3.51 (m, 4H),
3.41 (m, 4H), 3.29 ppm (s, 4H); 13C NMR (D2O, pH 4.10): d=149.30,
141.32, 138.29, 132.66, 131.05, 126.35, 122.41, 49.66, 44.56, 44.40,
42.63 ppm.
2,6,10,14-Tetratosyl-2,6,10,14-tetraaza[15]-(2,7)-acridinophane
(7): This compound was synthesized in dry CH3CN from 1 (1.00 g,
2.74 mmol), 5 (2.64 g, 3.28 mmol), and K2CO3 (7.03 g, 50.9 mmol)
by following the procedure reported above for 6. Yield: 1.40 g
(1.39 mmol, 51%); elemental analysis: calcd (%) for C52H57N5S4O8: C
61.94, H 5.70, N 6.95, found: C 61.8, H 5.6, N 6.8.
2,6,10,14-Tetraaza[15]-cyclo(2,7)-acridinophane pentahydrobro-
mide (L3·5HBr·H2O): This compound was synthesized in a mixture
of HBr/CH3COOH (33%) from 7 (1.00 g, 1 mmol) in the presence
phenol (10.5 g, 0.112 mol) by following the procedure described
for L2. Yield: 0.61 g (0.75 mmol, 75%); elemental analysis: calcd
(%) for C24H33N5·5HBr·H2O: C 35.41, N 8.60, H 4.95, found: C 35.5, N
Taq polymerase assay. To meet proper working conditions, com-
pounds were assayed against Taq polymerase activity by using
1
8.6, H 5.0; H NMR (D2O, pH 4.26): d=9.09 (s, 1H), 8.31 (s, 2H), 8.16
2002
www.chemmedchem.org
ꢁ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemMedChem 2010, 5, 1995 – 2005

Acridine-Based Polyamine DNA Binding Ligands
pBR322 (2.5 ng) as a DNA template and appropriate primer se-
quences Tup and Tdown (0.5 mm) to amplify the 906–1064 se-
quence of the plasmid by PCR. The reaction was carried out in an
Eppendorf thermocycler performing 25 cycles of: 30 s at 948C, 30 s
at 658C, and 30 s at 728C. The reaction products were resolved on
a 2% agarose gel in TBE (89 mm Tris base, 89 mm boric acid, 2 mm
Na2EDTA) and stained by ethidium bromide.
truncated octahedral box, the edges of which were located 10 ꢃ
from the closest atom of the DNA fragments, and which contains
~4700 water molecules for the poly(dG–dC) dsDNA oligonucleo-
tides, and ~5800 water molecules for the G-quadruplex complexes.
To maintain neutrality in the system, 16 and 38 K+ counterions
were added to the solvent bulk of the dsDNA–water complexes
and to the solvent bulk of the G-quadruplex–water complexes, re-
spectively. In the case of the G-quadruplex structure, according to
X-ray data,[7] 4 K+ ions were placed along the axis within the cen-
tral core of the complex, midway between each G-tetrad. Before
starting the MD simulations, an energy minimization of the com-
plexes was performed by setting a convergence criterion on a gra-
dient of 0.01 kcalmolꢀ1 ꢃꢀ1. Water shells and counterions were
then equilibrated for 40 ps at 300 K, after which 10 ns of MD simu-
lation in an isothermal-isobaric ensemble was performed without
any restraint on each complex. The ff03 version of the AMBER
force field was used for the DNA fragments and the counterions,[51]
whereas the TIP3P model[52] was employed to explicitly represent
water molecules. In the production runs, the ligand–DNA fragment
systems were simulated in periodic boundary conditions. The
van der Waals and short-range electrostatic interactions were esti-
mated within a 10 ꢃ cutoff, whereas the long-range electrostatic
interactions were assessed by using the particle mesh Ewald
method,[53] with 1 ꢃ charge grid spacing interpolated by fourth-
Telomerase activity (TRAP) assay. An aliquot of 5ꢂ106 HeLa cells
in exponential growth phase was pelleted and lysed for 30 min on
ice using 100 mL 0.5% CHAPS, 1 mm EGTA, 25% 2-mercaptoetha-
nol, 1.74% PMSF, and glycerol (10% w/v). The lysate was centri-
fuged at 13000 rpm (21500ꢂg) for 30 min at 48C, and the super-
natant was collected, stored at ꢀ808C, and used as a telomerase
source. Telomerase activity was assayed by using a modified telo-
mere repeat amplification protocol (TRAP) assay.[45] Briefly, a proper
substrate TS (100 ng) was elongated by telomerase by incubation
of the reaction mixture with 1 mg protein extract at 378C for
30 min in the presence or absence of increasing drug concentra-
tions. Product amplification was then performed by the addition of
100 ng return primer ACX and 2 U Taq polymerase and PCR am-
plification (33 cycles of: 30 s at 928C, 30 s at 588C, and 45 s at
728C). The reaction products were resolved by a 10% polyacryl-
amide gel (19:1) in TBE and visualized after staining with Sybr
Green I. Telomerase inhibition values are expressed as percent of
telomerase inhibition relative to control (no drug) lanes.
order B-spline, and by setting the direct sum tolerance to 10ꢀ5
.
Bonds involving hydrogen atoms were constrained by using the
SHAKE algorithm[54] with a relative geometric tolerance for coordi-
nate resetting of 0.00001 ꢃ. Berendsen’s coupling algorithms[55]
were used to maintain constant temperature and pressure with
the same scaling factor for both solvent and solutes, and with the
time constant for heat-bath coupling maintained at 1.5 ps. The
pressure for the isothermal-isobaric ensemble was regulated by
using a pressure relaxation time of 1 ps in the Berendsen’s algo-
rithm. The simulations of the solvated complexes were performed
using a constant pressure of 1 atm and a constant temperature of
300 K. A time step of 2 fs was used in the simulations, which were
carried out with the AMBER9 program suite.[50]
Cell toxicity assay. The HeLa (human epithelial) cell line was main-
tained in DMEM supplemented with 10% heat-inactivated fetal calf
serum, 50 UmLꢀ1 penicillin G, and 50 mgmLꢀ1 streptomycin at 378C
under a humidified atmosphere and 5% CO2. For the MTT assay,
cells were plated in 96-well plates at 10000 cells per well, and cul-
tured overnight. Afterward, compounds were added in triplicate,
and plates were incubated in the presence of the drug for 72 h. At
the end of this period, MTT [3-(4,5-dimethylthiazol-2-yl)-2,5-diphe-
nyltetrazolium bromide, a yellow tetrazole] was added to a final
concentration of 0.8 mgmLꢀ1, and incubation was continued for an
additional 2 h. After removal of the medium, DMSO was added
(150 mL per well). Soluble formazan salts formed by living cells
were homogenized by manual pipetting, and the absorbance at
l 540 nm was read. Results were analyzed as sigmoidal dose–re-
sponse curves.
MM-PBSA calculations. Free energies were calculated by using the
MM-PBSA method. The electrostatic contribution to the solvation
free energy was calculated with the nonlinear Poisson–Boltzmann
method as implemented in the adaptive Poisson–Boltzmann solver
(APBS)[56] program through the AMBER/iAPBS interface. The hydro-
phobic contribution to the solvation free energy was determined
with terms dependent on solvent-accessible surface area. In these
calculations, we used a solvent probe radius of 1.4 ꢃ to define the
dielectric boundary, a physiological salt concentration of 0.154m to
calculate the effect of salt on the free energies, and dielectric con-
stants of 1.0 and 80.0, respectively, for the solute and surrounding
solvent. Atomic charges for DNA fragments and ligands are the
same as those employed in the MD simulations. For atomic radii,
we applied the PARSE[57] parameter set. In the case of the G-quad-
ruplex structures, the cations (K+) present within the negatively
charged central channel were also explicitly included in the calcu-
lation. The K+ ion radius was kept at 2.025 ꢃ, based on a previous
study.[58] Free energies were estimated by collecting snapshots
every 40 ps during the last 4 ns of the MD simulations.
Computational methods. The binding capacity of H2L12+, H2L22+
,
and H2L32+ toward the telomeric G-quadruplex structure of se-
quence d(TAG-GGT-TAG-GGT) and toward double-helical DNA oli-
gonucleotides (10-mers) of base composition poly(dG–dC) was in-
vestigated. Ligand molecules and dsDNA containing CG or GC in-
tercalative binding sites were built by the Build module of Maestro
v. 8.5.[46] Starting coordinates for the tetrameric G-quadruplex were
obtained from the biological unit of the BRACO-19–d(TAG-GGT-
TAG-GGT) complex X-ray crystal structure (PDB ID: 3CE5).[7]
Docking calculations were performed using Glide[47] with the DNA
structures kept fixed in their original conformation throughout the
docking procedures. Selected poses of the ligand–target (both
dsDNA and G-quadruplex) complexes were submitted to MD simu-
lation for 10 ns in explicit solvent, and the RMSD values of the
complexes were monitored as a function of simulation time.
The atomic electrostatic charges of the ligands were calculated by
means of the RESP procedure,[48] that is, fitting them to an electro-
static potential calculated at the HF/6-31G* level of theory using
Gaussian 09 software.[49] General Amber force field (GAFF) parame-
ters were then assigned to the ligands by the antechamber module
implemented in AMBER9 suite.[50] Each complex was immersed in a
The entropic contribution was estimated with the normal mode
analysis (Nmode[59] module of AMBER9 suite); the snapshots were
minimized in the gas phase for a maximum of 1ꢂ105 cycles to
give an energy gradient of 1ꢂ10ꢀ4 kcalmolꢀ1 ꢃꢀ1. Because of the
extensive computational requirement, the normal mode analyses
were performed by considering snapshots collected every 200 ps
ChemMedChem 2010, 5, 1995 – 2005
ꢁ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemmedchem.org
2003

C. Bazzicalupi, C. Sissi, P. Gratteri, et al.
MED
[31] M. Kaiser, A De Cian, M. Sainlos, C. Renner, J. L. Mergny, M. P. Teulade-
Fichou, Org. Biomol. Chem. 2006, 4, 1049–1057.
during the last 4 ns of the MD simulations. Molecular graphics
were produced with Visual Molecular Dynamics (VMD)[60] and Mer-
cury.[61]
[32] C. Hounsou, L. Guittat, D. Monchaud, M. Jourdan, N. Saettel, J. L.
Mergny, M. P. Teulade-Fichou, ChemMedChem 2007, 2, 655–666.
[33] R J. Harrison, A. P. Reszka, S. M. Haider, B. Romagnoli, J. Morrell, M. A.
Read, S. M. Gowan, C. M. Incles, L. R. Kelland, S. Neidle, Bioorg. Med.
Chem. Lett. 2004, 14, 5845–5849.
[34] S. Bartoli, C. Bazzicalupi, S. Biagini, L. Borsari, A. Bencini, E. Faggi, C.
Giorgi, C. Sangregorio, B. Valtancoli, Dalton Trans. 2009, 1223–1230.
[35] A. Bencini, A. Bianchi, E. Garcia-Espana, M. Giusti, M. Micheloni, P. Pao-
letti, Inorg. Chem. 1987, 26, 681–684.
Acknowledgements
We are grateful to MIUR and Ministero dell’Istruzione, dell’Univer-
sitꢁ e della Ricerca, (PRIN 2007) for financial support.
[36] C. Bazzicalupi, A. Beconcini, A. Bencini, V. Fusi, C. Giorgi, A. Masotti, B.
Valtancoli, J. Chem. Soc. Perkin Trans. 2 1999, 1675–1682.
Keywords: acridine derivatives · DNA · G-quadruplexes · MM-
PBSA · molecular dynamics
[37] J. D. McGhee, P. H. Von Hippel, J. Mol. Biol. 1974, 86, 469–489.
[38] S. M. Zeman, D. M. Crothers, Methods Enzymol. 2001, 340, 51–68.
[39] Y. Pommier, J. M. Covey, D. Kerrigan, J. Markovits, R. Pham, Nucleic Acids
Res. 1987, 15, 6713–6731.
[40] For some recent examples see Ref. [34] and: a) C. Bazzicalupi, A. Bencini,
S. Biagini, E. Faggi, S. Meini, C. Giorgi, A. Spepi, B. Valtancoli, J. Org.
Chem. 2009, 74, 7349–7363; b) M. C. Aragoni, M. Arca, A. Bencini, S. Bia-
gini, A. J. Blake, C. Caltagirone, F. Demartin, G. De Filippo, F. A. Devilla-
nova, A. Garau, K. Gloe, F. Isaia, V. Lippolis, B. Valtancoli, M. Wenzel,
Inorg. Chem. 2008, 47, 8391–4303; c) C. Cruz, R. Delgado, M. G. B. Drew,
V. Felix, J. Org. Chem. 2007, 72, 4023–4034; d) C. Bazzicalupi, A. Bencini,
A. Bianchi, L. Borsari, A. Danesi, C. Giorgi, C. Lodeiro, P. Mariani, F. Pina,
S. Santarelli, A. Tamayo, B. Valtancoli, Dalton Trans. 2006, 4000–4010.
[41] C. Lodeiro, A. J. Parola, F. Pina, C. Bazzicalupi, A Bencini, A Bianchi, C.
Giorgi, A. Masotti, B. Valtancoli, Inorg. Chem. 2001, 40, 2968–2975
[42] P. Gans, A. Sabatini, A Vacca, Talanta 1996, 43, 1739–1753.
[43] L. Alderighi, P. Gans, A. Ienco, D. Peters, A. Sabatini, A. Vacca, Coord.
Chem. Rev. 1999, 184, 311–318.
[1] S. Neidle, Prog. Med. Chem. 1979, 16, 151–221.
[2] H. M. Berman, P. R. Yuong, Annu. Rev. Biophys. Bioeng. 1981, 10, 87–114.
[3] W. P. Wilson, R. L. Jones, Adv. Pharmcol. Chemother. 1981, 18, 177–222.
[4] S. Haider, G. N. Parkinson, M. A. Read, S. Neidle in DNA and RNA Binders.
From Small Molecules to Drugs (Eds.: M. Demeunynck, C. Bailly, W. D.
Wilson), Wiley-VCH, Weinheim, 2003, pp. 337–359.
[5] J. E. Reed, S. Neidle, R. Vilar, Chem. Commun. 2007, 4366–4368.
[6] G. N. Parkinson, R. Ghosh, S. Neidle, Biochemistry 2007, 46, 2390–2397.
[7] N. H. Campbell, G. N. Parkinson, A. P. Reszka, S. Neidle, J. Am. Chem. Soc.
2008, 130, 6722–6724.
[8] J. Dai, M. Carver, D. Yang, Biochimie 2008, 90, 1172–1183, and referen-
ces therein.
[9] J. L. Mergny, C. Helene, Nat. Med. 1998, 4, 1366–1367.
[10] D. Y. Sun, L. H. Hurley in Methods in Enzymology, Drug–Nucleic Acid Inter-
actions, Vol. 340 (Eds.: J. B. Chaires, M. J. Waring), Academic Press Inc.,
San Diego, 2001, pp. 573–592.
[44] A. K. Covington, M. Paabo, R. A. Robinson, R. G. Bates, Anal. Chem.
1968, 40, 700–706.
[45] N. W. Kim, F. Wu, Nucleic Acids Res. 1997, 25, 2595–2597.
[46] Maestro v. 8.5, Schrçdinger, LLC, New York, NY (USA), 2007.
[47] Glide, v. 4.5, Schrçdinger, LLC, New York, NY (USA), 2007.
[48] C. I. Bayly, P. Cieplak, W. D. Cornell, P. A. Kollman, J. Phys. Chem. 1993,
97, 10269–10280.
[11] L. H. Hurley, Nat. Rev. Cancer 2002, 2, 188–200.
[12] S. Neidle, G. N. Parkinson, Nat. Rev. Drug Discovery 2002, 1, 383–393.
[13] A. G. Bodnar, M. Ouellette, M. Frolkis, S. E. Holt, C. P. Chiu, G. B. Morin,
C. B. Harley, J. W. Shay, S. Lichtsteiner, W. E. Wright, Science 1998, 279,
349–352.
[49] Gaussian 09 (Revision A.01), M. J. Frisch, G. W. Trucks, H. B. Schlegel,
G. E. Scuseria, M. A. Robb, J. R. Cheeseman, G. Scalmani, V. Barone, B.
Mennucci, G. A. Petersson, H. Nakatsuji, M. Caricato, X. Li, H. P. Hratch-
ian, 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. A., Jr., Montgomery, J. E. Peralta,
F. Ogliaro, M. Bearpark, J. J. Heyd, E. Brothers, K. N. Kudin, V. N. Starover-
ov, 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. Morokuma, V. G. Zakrzewski, G. A. Voth, P. Salvador, J. J.
Dannenberg, S. Dapprich, A. D. Daniels, ꢆ. Farkas, J. B. Foresman, J. V.
Ortiz, J. Cioslowski, D. J. Fox, Gaussian Inc., Wallingford, CT (USA), 2009.
[50] D. A. Case, T. A. Darden, T. E. Cheatham, C. L. Simmerling, J. Wang, R. E.
Duke, R. Luo, K. M. Merz, D. A. Pearlman, M. Crowley, R. C. Walker, W.
Zhang, B. Wang, S. Hayik, A. Roitberg, G. Seabra, K. F. Wong, F. Paesani,
X. Wu, S. Brozell, V. Tsui, H. Gohlke, L. Yang, C. Tan, J. Mongan, V.
Hornak, G. Cui, P. Beroza, D. H. Mathews, C. Schafmeister, W. S. Ross,
P. A. Kollman, AMBER9, University of California, San Francisco, CA (USA),
2006.
[14] C. B. Harley, Mutat. Res. 1991, 256, 271–282.
[15] J. A. Hackett, D. M. Feldser, C. W. Greider, Cell 2001, 106, 275–286.
[16] H. Sun, J. K. Karow, I. D. Hickson, N. Maizels, J. Biol. Chem. 1998, 273,
27587–27592.
[17] H. Sun, R. J. Bennett, N. Maizels, Nucleic Acids Res. 1999, 27, 1978–1984.
[18] C. B. Harley, A. B. Futcher, C. W. Greider, Nature 1990, 345, 458–460.
[19] C. W. Greider, E. H. Blackburn, Cell 1985, 43, 405–413.
[20] G. B. Morin, Cell 1989, 59, 521–529.
[21] N. W. Kim, M. A. Piatyszek, K. R. Prowse, C. B. Harley, M. D. West, P. L. C.
Ho, G. M. Coviello, W. E. Wright, S. L. Weinrich, J. W. Shay, Science 1994,
266, 2011–2015.
[22] A. De Cian, L. Lacroix, C. Douarre, N. Temime-Smaali, C. Trentesaux, J.-F.
Riou, J.-L. Mergny, Biochimie 2008, 90, 131–155.
[23] R. J. Harrison, S. M. Gowan, L. R. Kelland, S. Neidle, Bioorg. Med. Chem.
Lett. 1999, 9, 2463–2468.
[24] M. A. Read, A. A. Wood, J. R. Harrison, S. M. Gowan, L. R. Kelland, H. S.
Dosanjh, S. Neidle, J. Med. Chem. 1999, 42, 4538–4546.
[25] C. M. Schultes, W. Guyen, J. Cuesta, S. Neidle, Bioorg. Med. Chem. Lett.
2004, 14, 4347–4351.
[26] M. Read, R. J. Harrison, B. Romagnoli, F. A. Tanious, S. H. Gowan, A. P.
Reszka, W. D. Wilson, L. R. Kelland, S. Neidle, Proc. Natl. Acad. Sci. USA
2001, 98, 4844–4849.
[27] S.M Gowan, J. R. Harrison, L. Patterson, M. Valenti, M. A. Read, S. Neidle,
L. R. Kelland, Mol. Pharmacol. 2002, 61, 1154–1162.
[51] Y. Duan, C. Wu, S. Chowdhury, M. C. Lee, G. Xiong, W. Zhang, R. Yang, P.
Cieplak, R. Luo, T. Lee, J. Comput. Chem. 2003, 24, 1999–2012.
[52] W. L. Jorgensen, J. Chandrasekhar, J. D. Madura, R. W. Impey, L. M. Klein,
J. Chem. Phys. 1983, 79, 926–936.
[53] U. Essmann, L. Perera, M. L. Berkowitz, T. Darden, H. Lee, L. G. A. Peder-
sen, J. Chem. Phys. 1995, 103, 8577–8593.
[28] R. J. Harrison, J. Cuesta, G. Chessari, M. A. Read, S. K. Basra, A. P. Reszka,
J. Morrell, S. M. Gowan, C. M. Incles, F. A. Tanious, W. D. Wilson, L. R. Kel-
land, S. Neidle, J. Med. Chem. 2003, 46, 4463–4476.
[29] C. Martins, M. Gunaratnam, J. Stuart, V. Makwana, O. Greciano, A. P.
Reska, L. R. Kelland, S. Neidle, Bioorg. Med. Chem. Lett. 2007, 17, 2293–
2298.
[30] P. Alberti, J. Ren, M. P. Teulade-Fichou, L. Guittat, J. F. Riou, J. B. Chaires,
C. Hꢄlꢅne, J. P. Vigneron, J. M. Lehn, J. L. Mergny, J. Biomol. Struct. Dyn.
2001, 19, 505–513.
[54] J. P. Ryckaert, G. Ciccotti, H. J. C. Berendsen, J. Comput. Phys. 1977, 23,
327–341.
[55] H. J. C. Berendsen, J. P. M. Postma, W. F. Van Gunsteren, A. Di Nola, J. R.
Haak, J. Chem. Phys. 1984, 81, 3684–3690.
[56] N. A. Baker, D. Sept, S. Joseph, M. J. Holst, J. A. McCammon, Proc. Natl.
Acad. Sci. USA 2001, 98, 10037–10041.
2004
www.chemmedchem.org
ꢁ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemMedChem 2010, 5, 1995 – 2005

Acridine-Based Polyamine DNA Binding Ligands
[57] D. Sitkoff, K. A. Sharp, B. Honing, J. Phys. Chem. 1994, 98, 1978–1988.
[58] P. Hazel, G. N. Parkinson, S. Neidle, Nucleic Acids Res. 2006, 34, 2117–
2127.
[61] C. F. Macrae, I. J. Bruno, J. A. Chisholm, P. R. Edgington, P. McCabe, E.
Pidcock, L. Rodriguez-Monge, R. Taylor, J. van de Streek, P. A. Wood, J.
Appl. Crystallogr. 2008, 41, 466–470.
[59] J. Kottalam, D. A. Case, Biopolymers 1990, 29, 1409–1421.
[60] W. Humphrey, A. Dalke, K. Schulten, J. Mol. Graphics 1996,14, 27–28.
Received: August 5, 2010
Revised: September 18, 2010
Published online on October 18, 2010
ChemMedChem 2010, 5, 1995 – 2005
ꢁ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemmedchem.org
2005