Recognition of G-quadruplex DNA by triangular star-shaped compounds: With or without side chains?

Article abstract of DOI:10.1002/chem.201002810

We report the synthesis of two new series of triangular aromatic platforms, either with three aminoalkyl side chains (triazatrinaphthylene series, TrisK: six compounds), or without side chains (triazoniatrinaphthylene, TrisQ). The quadruplex-DNA binding b

Full text of DOI:10.1002/chem.201002810

FULL PAPER
DOI: 10.1002/chem.201002810
Recognition of G-Quadruplex DNA by Triangular Star-Shaped Compounds:
With or Without Side Chains?
Hꢀlꢁne Bertrand,[a] Anton Granzhan,[a] David Monchaud,[a, b] Nicolas Saettel,[a]
Rꢀgis Guillot,[c] Sarah Clifford,[d] Aurore Guꢀdin,[e, f] Jean-Louis Mergny,[e, f] and
Marie-Paule Teulade-Fichou*[a]
Abstract: We report the synthesis of
two new series of triangular aromatic
platforms, either with three aminoalkyl
side chains (triazatrinaphthylene series,
TrisK: six compounds), or without side
chains (triazoniatrinaphthylene, TrisQ).
The quadruplex–DNA binding behav-
ior of the two series, which differ es-
sentially by the localization of the cat-
ionic charges, was evaluated by means
of FRET-melting and G4-FID assays.
For the trisubstituted triazatrinaphthy-
lenes (TrisK), the length of the sub-
stituents and the presence of terminal
hydrogen-bond-donor groups (NH2)
were shown to be crucial for ensuring a
high quadruplex affinity (DT1/2 values
of up to 208C at 1 mm for the best can-
didate, TrisK3-NH) and selectivity
versus duplex DNA. Subsequently,
comparison of data collected on both
the telomeric- and c-myc-quadruplex
showed that the nonsubstituted TrisQ
is even more efficient than TrisK3-NH,
both in terms of quadruplex affinity
(DT1/2 =268C in K+ buffer) and selec-
tivity versus duplex DNA. Structural
considerations conducted with the c-
myc quadruplex indicate that both
TrisK3-NH and TrisQ stack well onto
the G-quartet but in an offset position,
which might be influenced by the for-
mation of multiple hydrogen bonds
with the target in the former case. Fi-
nally, the nonsubstituted TrisQ displays
a binding profile very similar to some
of the best quadruplex binders,
BRACO-19 and bisquinolinium 360A,
used herein as references, and thereby
represents a highly promising novel
molecular design for quadruplex recog-
nition.
Keywords: DNA recognition · G-
quadruplexes · heterocycles · ligand
design · polycycles
Introduction
[a] Dr. H. Bertrand, Dr. A. Granzhan, Dr. D. Monchaud, Dr. N. Saettel,
Dr. M.-P. Teulade-Fichou
It is recognized that DNA sequences containing repeats of
bases are highly susceptible to aberrant replication and per-
turbation of other DNA-related processes such as recombi-
nation and transcription.[1] These dysfunctions may lead ulti-
mately to modifications of the genetic material (i.e., muta-
tions, genomic instability) and may have a role in explaining
mechanisms linked to cancer development or more largely,
be involved in pathogenic rearrangements genome-wide.[2]
At the molecular level, the peculiar behavior of repeat se-
quences is attributed to the formation of secondary struc-
tures through intrastrand Watson–Crick base pairing (e.g.,
folded back hairpins) or inter- or intrastrand non-canonical
base-assemblies (e.g., mismatched base-pairs, base-triplets
or quartets). Amongst all the nucleic bases, guanine exhibits
the highest propensity to self-assemble, thus forming GG
pairs, G-ribbons, and four-base associations called G-quar-
tets.[3] G-quartet motifs may coordinate to alkaline cations
(K+, Na+) to form consecutive p stacks that initiate the for-
mation of tetrahelical structures named quadruplexes
(Figure 1).[3–4] Consequently, single-stranded DNA-domains
containing G repeats are predicted to form guanine–quadru-
plex secondary structures.[5] Quadruplex structures exhibit a
high thermodynamic stability in biological conditions, repre-
senting a fascinating example of supramolecular self-assem-
bly. Consequently, these peculiar nucleic acid structures are
Institut Curie, Centre de Recherche, CNRS UMR176
Centre Universitaire Paris XI, Bꢀt. 110, 91405 Orsay (France)
Fax : (+33)169075381
E-mail: marie-paul.teulade-fichou@curie.fr
[b] Dr. D. Monchaud
Current address: Institut de Chimie Molꢁculaire
CNRS UMR5260, Universitꢁ de Bourgogne (ICMUB)
Facultꢁ des Sciences Mirande, 9, Avenue Alain Savary
21000 Dijon (France)
[c] Dr. R. Guillot
Institut de Chimie Molꢁculaire et des Matꢁriaux dꢂOrsay
CNRS UMR8182, Universitꢁ Paris Sud XI
Bꢀt. 420, 91405 Orsay (France)
[d] Dr. S. Clifford
Dꢁpartement de chimie minꢁrale, analytique et appliquꢁe
Universitꢁ de Genꢃve, quai Ernest-Ansermet 30
1211 Genꢃve 4 (Switzerland)
[e] A. Guꢁdin, Dr. J.-L. Mergny
Laboratoire des Rꢁgulations et Dynamique du Gꢁnome
INSERM U565, CNRS UMR5153
Musꢁum National dꢂHistoire Naturelle, 43, Rue Cuvier
75005 Paris (France)
[f] A. Guꢁdin, Dr. J.-L. Mergny
INSERM U869, Universitꢁ de Bordeaux
Institut Europꢁen de Chimie et Biologie, 2, Rue Robert Escarpit
33607 Pessac (France)
Supporting information for this article is available on the WWW
under http://dx.doi.org/10.1002/chem.201002810.
Chem. Eur. J. 2011, 17, 4529 – 4539
ꢄ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
4529

Figure 1. Schematic representation of various forms of the human telo-
meric quadruplex in Na+ and K+ conditions and of the c-myc quadru-
plex, identified by NMR and X-ray studies.
currently the focus of intense attention with the aim to deci-
pher their biological roles[6] and to evaluate their potential
for building DNA-based nanoobjects.[7]
Interestingly, quadruplexes may bind small molecules,
which in turn stabilize these structures and may act as mo-
lecular silencers by shifting a functional DNA conformation
(duplex) to a nonfunctional one (quadruplex). Over the past
decade, the search for small molecules acting as quadruplex
ligands has developed dramatically with essentially two
major goals: finding probes to sense quadruplex formation
in various biological contexts and discovering new DNA-in-
teracting agents of potential therapeutic interest. Thus, G-
quadruplex binders have great potential as drugs for anti-
cancer research, especially in view of the fact that quadru-
plex-forming sequences are present in the regions involved
in tumorogenesis (oncogenes)[8] or essential for the mainte-
nance of genome stability (telomeres). Thus, these two
DNA domains have been the major targets for quadruplex
drug design, whereas more recently, targeting quadruplexes
in messenger RNA is attracting increasing attention.[9]
However, G-rich sequences with the potential to form
quadruplexes are widely distributed in the genome and have
also been found in mRNA, therefore the correlation of in
vivo effects and in vitro data requires ligands that are capa-
ble of specific binding only to certain quadruplex motifs (te-
lomeric vs. oncogenic, DNA vs. RNA, etc.) Currently this is
the challenge, since in terms of molecular recognition the
quadruplex structures have not delivered all their secrets.
Indeed, although hundreds of quadruplex ligands have been
reported,[10] only a handful of structures of quadruplex–
ligand complexes have been resolved by X-ray or NMR
analysis.[11] Most structures have been obtained with bimo-
lecular[12] and tetramolecular[13] quadruplexes that were used
as models. Only recently information has been provided
from the X-ray analysis of the naphthalene diimide deriva-
tive NDI interacting with the intramolecular telomeric
quadruplex[14] and from the NMR-derived structure of the
tetracationic porphyrin TMPyP4 complexed with the intra-
molecular quadruplex of the c-myc oncogene.[15] On the
whole, structural studies have established that large aromat-
ic platforms stack on external G-quartets and in some cases,
might be sandwiched at the junction of two quadruplex
units.[13a,14] The presence of chains, (for example, aminoalkyl
substituents) might be a determinant of selectivity, since in
principle they allow supplementary interactions with the
target DNA.[12b] The role of the side chains was initially
thought to favor the anchorage in the grooves;[14,16] however,
this has been reconsidered since the tetrasubstituted NDI
drug was shown to interact only with loops at the interface
of two quadruplex units[14] or inside a quadruplex RNA.[17]
In particular, the introduction of side chains terminated with
hydrogen-bond-donor groups seems to favor the hydrogen-
bonding network with loops, either by direct contact or
more often, mediated by water molecules, which could pro-
vide clues for the specific recognition of a given quadruplex.
Alternatively, several classes of ligands, devoid of side
chains but endowed with large p surfaces that are likely to
maximize p–p interactions with G-quartets, have been
shown to interact strongly with quadruplexes. Amongst
these ligands we will only cite RHPS4, for which an NMR-
derived structure with a tetramolecular quadruplex is availa-
ble,[13c] and the well-studied telomestatin.[18] Notably, in the
latter case, although there are few doubts that this molecule
stacks well on terminal G-quartets, no structural data is
available besides a recent molecular-modeling study.[19]
4530
www.chemeurj.org
ꢄ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2011, 17, 4529 – 4539

Recognition of G-Quadruplex DNA
FULL PAPER
In view of the structural diversity of p systems interacting
with G-quartets, one question that remains to be addressed
is the optimal geometry for a ligand to maximize p–p inter-
actions with a G-quartet. At first glance, the ligands of four-
fold symmetry, matching that of a G-quartet and of compa-
rable size, represent an obvious choice. However, the por-
phyrin TMPyP4 that fulfills these criteria has so far not
been shown to establish direct contact with G-quartets as
seen from two independent studies with c-myc and the telo-
meric quadruplex,[12c,15] and possesses a very poor quadru-
plex versus duplex selectivity.[20,21] The NDI ligands that
have a limited p surface (four cycles) are more potent quad-
ruplex binders than their perylenediimide counterparts,
which display a much larger p surface (seven cycles).[22] Fi-
nally, a pursuit towards an extended, fourfold-symmetric
design may render the ligand nonplanar, as in the case of
tetraazoniapentaphenopentaphene previously reported by
Ihmels et al.,[21] and eventually disfavor its quadruplex-
stacking properties. Thus, it is still difficult to define the
structural features of the ligand that govern the quadruplex
recognition; however, based on the structural data available,
it appears that the presence of cationic charges and their lo-
calization (terminal amino groups vs. aromatic core), as well
as the topology of the aromatic core, are crucial for a strong
and selective interaction with the quadruplex structures.[17]
Nonetheless, due to the diversity of the systems developed
so far, it is difficult to dissociate contributions due to aro-
matic–aromatic interactions and side-chain effects. In an at-
tempt to address this question, we report herein on the syn-
thesis of two new series of triangular aromatic compounds,
either with or without side chains, and on the comparison of
their binding to two different quadruplex structures (the
human telomeric quadruplex and the oncogene c-myc quad-
ruplex) using FRET-melting, fluorescent intercalator dis-
placement (G4-FID) assays, and molecular modeling stud-
ies.
TrisK, with respect to the Celtic symbol Triskele meaning
“three-legged”. In contrast, TrisQ is devoid of any substitu-
ents and contains three quinolizinium units with quarternary
nitrogen atoms in the bridgehead positions of the polyaro-
matic core. This compound represents a C3-symmetric ana-
logue of the previously reported four-branched tetraazonia-
pentaphenopentaphene[21] and is expected to have a perma-
nent charge (3+), localized at the core unit and hence a
strong p-deficient character. Interestingly, in the course of
our study,[25] other C3-symmetric compounds designed for
quadruplex recognition have emerged. However, these mo-
lecular systems display either moderate affinity for the
quadruplex (Ka =105 mꢀ1, as determined by mass spectrome-
try)[26] or no significant quadruplex binding.[27]
Results
Synthesis of the ligands
TrisK series: The synthetic strategy for the 5,11,17-triazatri-
naphthylene derivatives was based on our previously pub-
lished synthesis of the parent trimethyl-TrisK[24] and quina-
cridines.[28] This pathway enabled structural diversity by the
introduction of substituent chains in the final step through
reaction of the common intermediates, 6,12,18-trichloro-
5,11,7-triazatrinaphthylene derivatives 3a–3b with primary
amines (Scheme 1). The synthesis of 3a–3b was readily per-
formed through a high-yielding three-step procedure from
commercially available materials, 1,3,5-tribromobenzene
and methyl esters of anthranilic or 5-methylanthranilic acids.
After the highly efficient palladium-catalyzed triple amina-
tion of 1,3,5-tribromobenzene by methyl anthranilates, basic
hydrolysis of the resulting triester derivatives 1a–1b enabled
the quantitative formation of the triacids 2a–2b, which after
cyclization at reflux with POCl3, afforded the trichloro de-
rivatives 3a–3b in high yield. In the final step, the alkylami-
no side chains were introduced by nucleophilic substitution
of chlorine with primary amines upon heating in solvent-
free conditions. The chains introduced were namely 1-
amino-2-methoxyethane, to give TrisK2-OMe (“2” for the
number of carbon atoms in the chain and OMe for the me-
thoxy terminal group), N,N-dimethyl-1,2-diaminoethane to
give TrisK2-NMe (NMe for the N,N-dimethylamino termi-
nal group), N,N-dimethyl-1,3-diaminopropane to give
TrisK3-NMe; 1,2-diaminoethane to give TrisK2-NH (NH for
the NH2 terminal function), and 1,3-diaminopropane to give
TrisK3-NH. The trimethyl analogue of the latter, MeTrisK3-
NH, was also synthesized. After purification by column
chromatography, the alkylamino-substituted derivatives
For a systematic study aimed at establishing the struc-
ture–property relationships we synthesized two series of C3-
symmetric heterocyclic derivatives with a trinaphthylene-
like polyaromatic core, which offers convenient access to
structural diversity. Among these compounds, the TrisK de-
rivatives represent an extension of the pentacyclic quinacri-
dine series of quadruplex ligands developed by our group
over the past years.[16,23] These ligands feature the neutral
5,11,17-triazatrinaphthylene core, substituted with three al-
kylamino substituents. In its neutral form, this motif was
previously used for 2D self-assembly studies[24] and termed
1
TrisK were obtained as free bases and characterized by H
and 13C NMR spectroscopy and HRMS analysis.
TrisQ: The synthesis of 4a,10a,16a-triazoniatrinaphthylene
(TrisQ) was performed through the cyclodehydration path-
way previously established for the other polycyclic quinolizi-
nium derivatives (Scheme 2).[21,29] The triacetal precursor
Chem. Eur. J. 2011, 17, 4529 – 4539
ꢄ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
4531

M.-P. Teulade-Fichou et al.
Scheme 1. Synthesis of tris(aminoakyl)-TrisK compounds. Reagents and conditions: i) Pd
tone, 16 h, room temp.; iii) POCl3, 16 h, reflux; iv) R2NH2, 3 d, reflux.
ACHUTGTNNERNUG(OAc)2, PACHTUGNTRENNNUG
enable a reliable comparison
between the two series of li-
gands, it was necessary to deter-
mine the protonation state of
the TrisK compounds under the
experimental conditions. To this
end, potentiometric titrations of
TrisK2-NH were performed in
MeOH/H2O (50:50 v/v). The
Scheme 2. Synthesis of TrisQACHTNUGRTNEUNG
[BF4ꢀ]3. Reagents and conditions: i) N-methylpyrrolidone (NMP), 7 d, RT;
ii) NaPF6 (aq); iii) MeSO3H, 72 h, 2108C; iv) NaBF4 (aq).
results allowed us to determine
the pKa values of 10.87, 10.03,
8.57, 5.88, 5.09, 3.78, and 3.66
corresponding to seven of its
ꢀ
4
N
nine nitrogen atoms (see the Supporting Information, Fig-
ure S3, A); two pKa values could not be determined, most
likely due to low pKa values (<1). The three highest pKa
values were assigned to the primary amino groups of the
aminoalkyl side chains (N3), with the remaining assigned to
the aromatic secondary amino groups (N2) and the hetero-
cyclic core (N1). This assignment was made by analogy with
related systems, that is, quinacridines and phenanthrolines,
which possess ring nitrogens with low pKa values.[16,30] The
analysis of the charged-species distribution curve (see the
Supporting Information, Figure S3, B) reveals that at pH 7.2
(conditions used for the DNA-binding studies), TrisK2-NH
exists predominantly as a tricationic species with the charges
localized on the side chains, whereas its heteroaromatic core
is not protonated. This fully validates the direct comparison
between the two series, as TrisQ is tricationic independently
of the pH except at pHꢁ9, at which decomposition of the
quinolizinium moieties may occur.[31] Finally, it may be as-
sumed that most derivatives of the TrisK series (TrisK3-NH,
MeTrisK3-NH, TrisK2-NMe, and TrisK3-NMe) have a pro-
tonation profile similar to that of TrisK2-NH and thereby
exist as tricationic species at pH 7.2, except for TrisK2-OMe,
benzene with 2-(1,3-dioxolan-2-yl)pyridine, was subjected to
cyclodehydration upon heating in methanesulfonic acid to
give TrisQ
ꢀ
[PF6 ]3, which was converted by ion metathesis
ꢀ
into the more stable tetrafluoroborate salt TrisQ
A
characterized by H and 13C NMR spectroscopy, mass spec-
1
trometry, and elemental analysis. The structure of
TrisQ
ꢀ
[BF4 ]3 was supported by single-crystal X-ray diffrac-
tion analysis data (see the Supporting Information, Fig-
ure S1). The compound crystallized from aqueous solution
in the P212121 space group with the inclusion of one mole-
cule of water; notably, no p–p stacking between the aromat-
ic cations in the solid state was observed.
Potentiometric titrations and evaluation of the charge state
of the ligands at pH 7.2: In the TrisK structure, up to three
types of nitrogen atom can be identified (see the Supporting
Information, Figure S2); namely intracyclic (N1), aromatic
amino (N2, located in the para position to N1), and aliphatic
amino nitrogen atoms located at the end of the side chains
(N3). Compounds TrisK can consequently exhibit variable
cationic charge distribution depending on the pH. Thus, to
4532
www.chemeurj.org
ꢄ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2011, 17, 4529 – 4539

Recognition of G-Quadruplex DNA
FULL PAPER
which is assumed to be neutral at physiological pH given
the lack of protonable groups on its side chains.
Interaction with human telomeric quadruplex
Evaluation of quadruplex-DNA affinity and selectivity by
FRET-melting: The quadruplex–DNA affinity of the six
TrisK compounds and of TrisQ was evaluated by using the
FRET-melting assay.[32] This assay is based on the thermal
denaturation of the quadruplex-forming oligonucleotide
F21T mimicking the human telomeric repeat (F21T: FAM-
G3ACHTUNGTRENNUNG[T2AG3]3-TAMRA, in which FAM is 6-carboxyfluorescein
and TAMRA is 6-carboxytetramethylrhodamine), moni-
tored by a fluorescence resonance energy transfer (FRET)
effect. A semi-quantitative evaluation of the ligand binding
affinity is obtained by measuring the increase in melting
temperature (DT1/2) induced by the ligand (DT1/2
=
T1/2ACHTUNGTRENNUNG(+ligand)ꢀT1/2(no ligand), in which T1/2 is the tempera-
ture at which 50% of the oligonucleotide is unfolded). In
view of the polymorphism of the human telomeric quadru-
plex as a function of the cation present in solution,[33]
FRET-melting experiments have been carried out both in
Na+- and K+-rich buffers. A great advantage of the FRET-
melting assay is the direct insight into the quadruplex-
versus duplex-DNA selectivity, which is a critical issue for
the design of specific ligands. To this end, FRET-melting ex-
periments are performed in the presence of various amounts
of competitive duplex-DNA ds26 (self-complementary se-
quence d(CA2TCG2ATCGA2T2CGATC2GAT2G)). If the
stabilization of the quadruplex imparted by the ligand is af-
fected by the excess of duplex-DNA, a drop in the DT1/2
value is observed. The extent of this drop is inversely pro-
portional to the selectivity and can be expressed by the ratio
Figure 2. FRET-melting results of TrisK derivatives; experiments were
carried out with F21T (0.2 mm), ligand (1 mm), without (black bars) or
with ds26 (3 mm: horizontally striped bars; 10 mm: white bars); concentra-
tion expressed in strand in lithium cacodylate buffer (10 mm), pH 7.2,
with A) NaCl (100 mm) or B) LiCl (90 mm)+KCl (10 mm).
S
FRET =DT
G
(ꢀds26), which reflects the prefer-
ential binding of a compound to the quadruplex-DNA (S!
measured to be between 6.2 and 13.98C) compared with
compounds that have terminal NH2 groups (TrisK2-NH and
TrisK3-NH, with DT1/2 values measured to be between 13.0
and 19.88C); 4) the additive effects of these two features are
seen from comparing TrisK2-NMe and TrisK3-NH; and
5) the presence of three methyl groups on the TrisK core is
clearly deleterious to the interaction, since a drop of almost
118C is observed for MeTrisK3-NH compared with TrisK3-
NH (DT1/2 =9.3 vs. 19.98C, respectively).
This data implies firstly that the three positive charges,
due to the three terminal amines, promote the ligand–quad-
ruplex interaction on the basis of electrostatic attraction;
however, it is not yet known if this is mediated by direct
contact with the DNA phosphates. In this regard the chain
length could play a role for the interaction, because a five-
membered chain would enable better flexibility and DNA
accessibility. Secondly, the results indicate that terminal pri-
mary amines have a crucial influence that could be a result
of their hydrogen-bond-donating ability, thereby suggesting
that hydrogen bonding may be involved between the ligand
and the quadruplex grooves or loops. Finally, tight p-stack-
ing interactions are certainly involved between the planar
TrisK core and the external G-quartets since the stabiliza-
1) or a complete lack of selectivity (S!0).[34]
As depicted in Figure 2 and Table S1 (Supporting Infor-
mation), low (DT1/2 =0.28C for TrisK2-OMe) to very high
(DT1/2 =19.88C for TrisK3-NH) levels of stabilization were
obtained with TrisK derivatives. The comparison of the per-
formances of the various compounds in Na+ and K+ condi-
tions indicates that globally the series is rather insensitive to
the cationic nature of the buffer because the results ob-
tained were similar in either buffer (average difference:
3.68C). For this reason, only values obtained in K+ condi-
tions will be used hereafter for comparison purposes. In con-
trast, the various side chains have a dramatic effect and the
following trends can be inferred: 1) the neutral TrisK2-OMe
does not stabilize the G-quadruplex, which implies that ter-
minal amino groups are required for F21T stabilization as
shown by the DT1/2 values measured for all the amino termi-
nated derivatives; 2) the length of the side chains has a sig-
nificant influence because 5-membered chains lead to a
better F21T stabilization compared with 4-membered ones
(typically 5–78C) both for NMe2- or NH2-terminated arms;
3) stabilization from compounds with terminal NMe2 groups
are modest (TrisK2-NMe and TrisK3-NMe, with DT1/2 values
Chem. Eur. J. 2011, 17, 4529 – 4539
ꢄ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
4533

M.-P. Teulade-Fichou et al.
tion is highly sensitive to the enhancement of the steric hin-
drance around the ligand aromatic core.
If we examine the effect of the TrisQ compound (Figure 3
and Table S1 in the Supporting Information), we note a very
high stabilization in K+ conditions (DT1/2 =26.28C), making
Figure 4. Quadruplex- versus duplex-DNA selectivity estimated by
FRET-melting (SFRET =DT1/2(+10 mM)/DT1/2
) for TrisK3-NH, TrisQ, and
BRACO-19 in Na+ (black) and in K+ (white) conditions.
which features a very similar profile but with an even lower
selectivity in Na+ buffer (SFRET =0.23 in Na+ vs. 0.82 in K+
buffer). In conclusion, although TrisK3-NH is worth consid-
ering due to its excellent recognition properties irrespective
of the cation used, TrisQ may be a more promising candi-
date because of its high efficiency in K+, rivaling the FRET-
melting performances of BRACO-19, one of the most
widely studied G-quadruplex ligands.[38]
Figure 3. Comparison of FRET-melting results in Na+ and K+ buffers for
TrisK3-NH, TrisQ and BRACO-19. The same experimental conditions
were used as in Figure 2.[35]
this compound a more potent binder than TrisK3-NH, which
is the best candidate from the TrisK series. However, their
comparison is not straightforward since, in contrast to the
TrisK series, TrisQ shows a remarkable dependence on the
cationic nature of the buffer, displaying strikingly modest re-
sults in the Na+ buffer (DT1/2 =10.28C). Interestingly, this
situation is reminiscent of that of BRACO-19, (Figure 3), in
which DT1/2 =16.5 and 27.38C in Na+ and K+ buffers, re-
spectively (i.e., DDT1/2 =10.88C) but with an even higher
amplitude (DDT1/2 =16.48C for TrisQ). This preference for
K+ over Na+ has been reported for several classes of quad-
ruplex binders[32] without being well understood. Nonethe-
less, this trend may reflect the greater accessibility of the
terminal quartets inside the K+-promoted conformation(s),
which is in agreement with the structures determined in the
solid state and in solution. However, the possibility for ex-
ternal binding out of the quartets observed for certain li-
gands[12c,14] and the coexistence of several differently folded
quadruplex conformations[33,36] make this trend difficult to
be fully rationalized.
Evaluation of quadruplex-DNA affinity and selectivity by the
G4-FID assay: To further evaluate the quadruplex-binding
properties of the two best candidates TrisK3-NH and TrisQ,
fluorescent indicator displacement titrations (G4-FID assay)
were performed. This assay enables evaluation of ligand af-
finity for DNA structures on the basis of its ability to dis-
place the fluorescent probe thiazole orange (TO) from both
quadruplex- and duplex-DNA matrices. The displacement
efficiency that reflects the affinity is quantified by G4DC50
and dsDC50 representing the concentrations required to dis-
place 50% of TO from the given DNA matrix.[20b,39] Herein,
we used two DNA matrices previously used to calibrate this
assay:[39] the telomeric quadruplex 22AG (AG
3ACHTNUGTRNE[UGN T2AG3]3),
which differs from F21T used in FRET-melting only by one
supplementary adenine at the 5’-end, and the duplex-DNA
ds26. Finally, since the FRET-melting results highlight the
strong impact of the cation on quadruplex recognition, the
experiments performed with 22AG have been carried out
both in Na+- and K+-rich conditions.
As depicted in Figure 5A and Table 1, in Na+ conditions
TrisK3-NH displaces TO from 22AG with a much better ef-
ficiency (G4DC50 =0.58 mm) than from ds26 (dsDC50 >2.5 mm).
Unexpectedly, the displacement of TO from 22AG is signifi-
cantly less efficient in K+ conditions (G4DC50 =2.05 mm), in
contrast to FRET-melting results. Currently, we do not have
any satisfactory explanation for the discrepancy between the
two methods, however, the existence of allosteric or external
binding sites in the K+ form(s), which would induce stabili-
zation but not displacement of the TO probe, might be rea-
soned. This situation has already been observed with the
side-arm substituted quinacridines.[20b] In contrast, the G4-
FID curves (Figure 5B) show that TrisQ is a much better
TO displacer from 22AG in K+ than in Na+ conditions
In terms of selectivity, the excess of ds26 affects the quad-
ruplex stabilization to various extents, depending on the
ligand and the cationic conditions, but generally the resist-
ance to the competition is better in K+ than in Na+ condi-
tions (Figure 3 and Figure 4). In the TrisK series, compound
TrisK3-NH clearly emerges as the most selective ligand be-
cause it exhibits a moderate to high selectivity both in Na+
and in K+ buffers (SFRET =0.49 and 0.70, Figure 4). In the
case of TrisQ, an even higher selectivity is observed in K+
buffer (SFRET =0.80), which makes this compound one of the
strongest and most selective quadruplex binders reported so
far.[37] On the other hand, it elicits a rather low selectivity in
Na+ buffer, in line with its low stabilizing effect (SFRET
0.41). Again, this situation is reminiscent of BRACO-19,
=
(
G4DC50 =0.3 and 2.3 mm respectively); the ratio of the two
4534
www.chemeurj.org
ꢄ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2011, 17, 4529 – 4539

Recognition of G-Quadruplex DNA
FULL PAPER
G-quartet of the quadruplex, and furthermore stabilized by
the side chains through additional electrostatic interactions
and hydrogen bonding. In contrast, TrisQ cannot interact
with loops or any other surrounding structural features, for
example, phosphates or walls of the grooves. It was thus of
interest to determine if the presence and the nature of the
loops that surround the quartets could influence significantly
the binding behavior of the two compounds. We decided to
study the interaction of the two ligands with another well-
defined G-quadruplex structure, the quadruplex formed in
the sequence of the c-myc promoter, which has been exten-
sively studied since dysregulation of this element is involved
in many cancers and formation of quadruplexes has been
shown to promote repression of its transcription.[40] In addi-
tion, a number of ligands have been reported recently for
their interaction with c-myc.[41] The interaction was evaluat-
ed by the two methods used previously. Firstly, the competi-
tive FRET-melting assay was performed in the presence of
F21T and unlabeled c-myc (T2GAG3TG3TAG3TG3TA2) as
competitor. The results shown in Figure 6A indicate a very
Figure 5. G4-FID titration curves obtained with A) TrisK3-NH and
B) TrisQ, for experiments carried out with 22AG (Na+ =filled square;
K
+ =filled circle) and ds26 (K+ =open triangle). Buffer: lithium cacody-
late (pH 7.2, 10 mm) and NaCl or KCl (100 mm).
Table 1. G4-FID results: DC50 and selectivity values for TrisK3-NH and
TrisQ.
[b]
Ligand
G4DC50[a] [mm]
dsDC50 [mm]
SG4ꢀFID
22AG
22AG
A
ds26
22AG
22AG
ACHTUNGTRENNUNG
(Na+)
(K+)
(K+)
TrisK3-NH
TrisQ
2.0
0.3
0.6
2.3
>2.5
>2.5
2
17
7
5
[a] Experimental errors are estimated at ꢂ5%. [b] Estimated selectivity:
SG4ꢀFID
=
dsDC50/G4DC50; however, the dsDC50 value cannot be determined
when the interaction with duplex-DNA is very weak. In this case, the se-
lectivity is estimated as follows: the TO displacement obtained with
2.5 mm of ligand on duplex DNA is determined and the concentration C
of ligand required to induce the same fluorescence decrease on quadru-
plex-DNA is determined. The estimated selectivity value corresponds to
the ratio of these two concentrations: SG4ꢀFID =2.5/C.
Figure 6. A) Competitive FRET-melting with F21T and c-myc as compet-
itor and various compounds. Experiments were carried out with F21T
(0.2 mm), ligand (1 mm) without (black bars) or with c-myc (3 mm: horizon-
tally striped bars; 10 mm: white bars), with the concentration expressed in
strand. Buffer: lithium cacodylate (pH 7.2, 10 mm), with NaCl (100 mm)
or LiCl (99 mm)+KCl (1 mm); B) FRET-melting stabilization obtained
with FmycT in the same buffer conditions (black: Na+; white: K+).
G4DC50 (Na+/K+) indicates an eightfold preference. In addi-
tion, TrisQ hardly displaces TO from the duplex matrix ds26
(dsDC50 >2.5 mm). In this case, the preference for binding the
telomeric quadruplex in K+ conditions and the high quadru-
plex versus duplex selectivity (SG4ꢀFID =17) are fully consis-
tent with the FRET-melting results. Of note, the displace-
ment of TO is incomplete and levels off at 80%, which
again may indicate the existence of allosteric binding sites
both for the ligand and the fluorescent probe resulting in in-
direct competition.
efficient competition since, for both ligands, the stabilization
of F21T is dramatically decreased at the lowest concentra-
tion in c-myc competitor (3 mm) and totally removed at the
highest concentration (10 mm). This phenomenon, which in-
dicates redistribution of ligands from the telomeric to the c-
myc quadruplex, seems to be rather independent of the
cation present in the medium although the resistance is
slightly stronger in K+ for TrisQ. The effects of the two li-
gands are very similar to that of the bisquinolinium 360A,
one of the best quadruplex ligands,[42] although the latter
shows a slightly higher resistance to the competition in K+
conditions.
Interaction with the c-myc quadruplex
It can be assumed that in case of TrisK3-NH, the association
is mainly driven by p–p stacking interactions between the
broad aromatic surface of the ligand and the most accessible
These first results indicate that the two ligands have a
strong affinity for c-myc and hence a poor ability to discrim-
Chem. Eur. J. 2011, 17, 4529 – 4539
ꢄ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
4535

M.-P. Teulade-Fichou et al.
inate between the two quadruplexes. This was confirmed by
FRET-melting using the doubly labeled c-myc sequence
FmycT (FAM-T2GAG3TG3TAG3TG3TA2-TAMRA). In Na+
and K+ conditions, remarkably strong stabilization is in-
duced, with DT1/2 values (Na+/K+) ranging from 23/21.5 to
32.3/28.78C (Figure 6B). Again, the TrisQ compound ap-
pears slightly more efficient than the three-armed TrisK3-
NH, but altogether the effect of the two ligands compares
favorably with that of the reference compound 360A
that an NMR-derived structure of its complex with TMPyP4
has been recently reported (PDB entry: 2A5R).[15] This
structure, obtained with the 24-nucleotide modified se-
quence Pu24I (TGAG3TG2IGAG3TG4A2G3, with the gua-
nine in position 10 being substituted by an inosine), is one
of the few examples of a solution structure of a complex
with a small molecule bound (within a more or less defined
pocket), to an intramolecular quadruplex of biological rele-
vance. Manual docking within Maestro[45] was then per-
formed (minimizing steric clashes and maximizing the p–p
overlap and hydrogen bonding when applicable), resulting
in a few selected positions shown in Figure 8.
ACHTUNGTRENNUNG(DT1/2ACHTUNGTRENNUNG
(Na+/K+)=35.5/22.48C), thereby confirming that the
two compounds are strong c-myc binders.[40,41c–d,43] Of note,
due to the high stability of c-myc, the FRET-melting was
performed at lower KCl concentration than that used for
F21T (1 vs. 10 mm KCl) and the values cannot be directly
compared with those of Table 1 (see the Supporting Infor-
mation, Tables S2A and B).
The G4-FID assay was also performed with TrisK3-NH,
TrisQ, and the c-myc quadruplex and the resulting titrations
curves are shown in Figure 7. A significant difference is ob-
Figure 7. G4-FID titrations curves obtained with TrisK3-NH (circles) and
TrisQ (squares) for experiments carried out with c-myc quadruplex in
lithium cacodylate buffer (pH 7.2, 10 mm), in the presence of KCl
(100 mm) at 208C.
Figure 8. Superposition onto G-tetrad of A) TrisK3-NH; B) and C) TrisQ
(two possible orientations). Molecular graphics images were produced
using the UCSF Chimera package.[46]
served between the two ligands, indicating that TrisQ displa-
ces the fluorescent probe with a much higher efficiency than
TrisK3-NH (G4DC50 <0.1 mm for TrisQ vs. 1.3 mm for TrisK3-
NH). Although the higher binding affinity of TrisQ was sug-
gested by the FRET-melting measurements, this large differ-
ence was somewhat unexpected. Again, this may indicate an
indirect competition with the TO probe due to external
binding in case of TrisK3-NH, thereby suggesting a clear dif-
ference in the binding mode of the two compounds in the
concentration range examined. Finally, the data demon-
strates that a combination of the two assays is necessary to
gain detailed information on ligand–quadruplex interactions.
In the case of TrisK3-NH (Figure 8, A), at least one hy-
drogen bond between one amino terminated chain and a
phosphate group and/or the deoxyribose can always be pres-
ent, maintaining an optimal overlap with the quartet and in-
dicating the ability of the molecule to establish supplemen-
tary interactions with the residues surrounding the quartet.
This model is in agreement with recent observations that
the interaction of drugs with cationic side-arms with struc-
tural features surrounding the quartets can be mediated by
hydrogen bonding[12b,14,17] and suggests that a combination of
the ligandꢂs side-chain–DNA interactions and p–p accessi-
bility governs the interaction. In the case of TrisQ, p stack-
ing between the electron-rich G-tetrad and the electron-
poor TrisQ core is achieved through at least two guanine
bases and two quinolizinium branches of TrisQ (Figure 8, B
and C). However, due to its smaller steric hindrance and the
lack of potential hydrogen bonding, this p system could
adopt several positions, resulting in various geometries of
Structural considerations: The nature of the binding interac-
tion is difficult to rationalize if the appropriate geometry of
interaction is not available. To gain further insights into the
ligand–quadruplex interaction, manual docking was carried
out with minimized structures of TrisK3-NH and TrisQ and
the c-myc quadruplex (ChemAxon Calculator Plugins)[44]
This quadruplex structure provides a clear advantage given
.
4536
www.chemeurj.org
ꢄ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2011, 17, 4529 – 4539

Recognition of G-Quadruplex DNA
FULL PAPER
the interaction. The third quinolizinium branch could either
position itself on the center of the G-quartet (Figure 8B) or
adopt a clearly offset position pointing outside the G-quar-
tet, towards the entrance of the groove (Figure 8C).
Notably, in both cases, the cofacial arrangement does not
imply a maximum overlap between the aromatic three-
branched core and the quartet, as would have been the case
if van der Waals and solvophobic effects were dominating.
More likely, the relative orientation of the ligand and the
quartet is guided by minimizing the electrostatic repulsions
between the two p systems and external constraints (e.g.,
steric hindrance of the loop residues in proximity).
ty in K+. Consistent with this, the quadruplex- versus
duplex-DNA selectivity is better in K+ conditions. These
observations tend to demonstrate that the interaction is
stronger in K+ and suggest that the Na+ form is not well-
recognized by the trisquinolizinium scaffold, thus revealing
steric hindrance and/or a reduced accessibility of the exter-
nal quartets. This trend has been observed for a number of
ligands including BRACO-19, but there is currently no satis-
factory rational explanation, although understanding this
difference would be of importance for the development of
ligands able to discriminate between different quadruplexes.
Conversely, TrisK3-NH shows a relatively similar level of
stabilization in Na+ and K+ conditions, whereas the G4-FID
data indicate a significant preference for the Na+ form.
Using the same methods, the interaction of the two li-
gands with the c-myc quadruplex was evaluated. Both TrisQ
and TrisK3-NH show exceptional stabilization effects (DT1/2
values of up to 328C), which are close to that of the refer-
ence compound 360A. Again, the G4-FID measurements
helped to confirm the superior binding ability of the nonsub-
stituted TrisQ scaffold, which seems to have tighter binding
likely to be attributable to a unique binding mode.
Globally, the data obtained for TrisQ shows good accord-
ance between the two methods used (FRET-melting and
G4-FID). In contrast, for TrisK3-NH, divergences are
found; G4-FID data showed differences that were not re-
vealed by FRET-melting or at least less clearly. This sug-
gests that the trisubstituted compound has a more complex
binding behavior; the side chains may be responsible for
multiple binding modes (stacking on G-quartets, external
binding to loops, and external binding to duplex-DNA),
which makes the data more difficult to evaluate because it
represents several contributions. These observations confirm
the importance of using at least two or ultimately several
methods of evaluation based on different principles, with
the aim to better understand the contributions that come
into play. Our study also shows the necessity of comparing
the performance of new ligands with well-studied reference
compounds to help place newly proposed molecular designs
in the very dense context of quadruplex ligands.
Molecular modeling indicates that TrisK3-NH has a good
capacity for quartet overlapping that might explain its high
quadruplex-stabilization activity and in this respect, this
ligand seems more efficient than its two-branched counter-
part, that is, the trisubstituted quinacridine developed previ-
ously.[16] In addition, the presence of free amino groups may
promote hydrogen bonding with the target either directly or
through water molecules.[44] This would be consistent with
experimental measurements, which have shown the impor-
tance of chain length in optimizing the properties in the
TrisK series (at least a four-membered chain is required).
The potential of TrisK3-NH for establishing hydrogen bond-
ing with residues surrounding the quartets could also explain
the preference for the Na+ form of the telomeric quadru-
plex, which exhibits accessible diagonal and edgewise loops.
However, whereas the side chains might enable supplemen-
tary interactions with the quadruplex, they appear not to be
Discussion
Although several molecular guidelines have been estab-
lished and used to enhance the binding selectivity of small
molecules to quadruplexes, the question of ligand structural
predictability still remains with respect to the large structur-
al diversity of quadruplex binders, the conformational heter-
ogeneity of quadruplexes, and the small amount of structur-
al data available for drug–quadruplex complexes. Herein,
we have chosen to explore the potential of an aromatic
three-branched scaffold of C3-symmetry to compare com-
pounds possessing similar geometrical features but differing
by the presence or the absence of side chains in an attempt
to evaluate the various contributions due to, on one hand,
aromatic–aromatic interactions and on the other hand, side-
chain–DNA interactions. To this end, two synthetic schemes
have been devised; the first scheme gave access to the TrisK
series (six analogues) possessing a neutral core endowed
with three ammonium-terminated side chains and the
second involved the preparation of a unique compound
TrisQ, an electron-poor system featuring three permanently
charged bridgehead nitrogen atoms.
Evaluation by FRET-melting and G4-FID assays led to
the identification of two compounds (TrisK3-NH and TrisQ)
that display binding efficiencies close to that of the best
binders reported so far, such as 360A and BRACO-19,
thereby fully validating the choice of the molecular design.
Clearly, the three-branched scaffold is a structural advantage
that provides not only a large seven-ring p-surface favorable
to a strong interaction with G-quartets, but most important-
ly, creates a certain steric hindrance that prevents duplex
binding. This is evidenced in the case of TrisQ, which has an
excellent binding selectivity for quadruplex- versus duplex-
DNA in spite of the absence of side chains. More mitigated
results were obtained with TrisK3-NH, which exhibits a
more modest quadruplex- versus duplex-DNA selectivity, al-
though it remains at a satisfactory level relative to many re-
ported quadruplex binders.
Another remarkable point is the large difference in the
binding performances of TrisQ with respect to the cation
present in the solution. Indeed, TrisQ binds the human telo-
meric quadruplex with a very poor efficiency in the presence
of Na+, which is in stark contrast to its high stabilizing abili-
Chem. Eur. J. 2011, 17, 4529 – 4539
ꢄ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
4537

M.-P. Teulade-Fichou et al.
a strong determinant of selectivity in the present case. In
the case of TrisQ, the relative position of the aromatic plat-
form on the quartet may vary significantly thus providing
more possibilities in the geometry of the arrangement than
for the trisubstituted compound. The partial overlap of the
aromatic surface area in contact is in line with the basic
principle of aromatic–aromatic interactions[47] and suggests
that the notion of shape complementarity and optimal aro-
matic overlap between a ligand and a G-quartet should be
taken with care and is not necessarily a successful guide for
optimizing interaction.[48] It is also likely that a simple p-
stacking interaction mode is also dependent on the external
constraints and susceptible to influences of the surrounding
structural features and of their local dynamics.
Finally, if we consider that ligands may interact at the
junction of two quadruplex units through p-stacking and
eventually dimerization,[12b,14] it is likely that TrisK3-NH and
TrisQ will not display the same ability to interact with a
quadruplex junction, since the former is able to dimerize by
analogy with the behavior of NDI, whereas TrisQ should
not be prone to dimerization due to strong electrostatic re-
pulsion as shown by the X-ray analysis of its structure (Fig-
ure S1 in the Supporting Information). This might explain
the differences in their binding behavior but a full evalua-
tion requires further investigation.
Acknowledgements
The authors would like to thank the Centre National de la Recherche
Scientifique (CNRS) and the Commissariat ꢅ lꢂEnergie Atomique (CEA)
for funding H.B., the Institut National de la Santꢁ et de la Recherche
Mꢁdicale (INSERM), the Rꢁgion Aquitaine, Fondation pour la Recher-
che Mꢁdicale (FRM) (J.L.M.), and a grant ANR-09-BLAN-0355 from
the “G4Toolbox” (M.P.T.F. and J.L.M.). Anne De Cian (MNHN) is grate-
fully acknowledged for preliminary FRET-melting experiments and Eric
Largy (Institut Curie) for help with the G4-FID experiments. Prof. Heiko
Ihmels (Universitꢆt Siegen, Germany) is greatly acknowledged for fruit-
ful discussions.
[1] a) R. D. Wells, R. Dere, M. L. Hebert, M. Napierala, L. S. Son, Nu-
cleic Acids Res. 2005, 33, 3785–3798; b) J. Eddy, N. Maizels, Mol.
Carcinog. 2009, 48, 319–325.
[2] F. Bena, S. Gimelli, E. Migliavacca, N. Brun-Druc, K. Buiting, S. E.
Antonarakis, A. J. Sharp, Hum. Mol. Genet. 2010, 19, 1967–1973.
[3] J. T. Davis, Angew. Chem. 2004, 116, 684–716; Angew. Chem. Int.
Ed. 2004, 43, 668–698.
[4] a) D. J. Patel, A. T. Phan, V. Kuryavyi, Nucleic Acids Res. 2007, 35,
7429–7455; b) S. Burge, G. N. Parkinson, P. Hazel, A. K. Todd, S.
Neidle, Nucleic Acids Res. 2006, 34, 5402–5415; c) A. N. Lane, J. B.
Chaires, R. D. Gray, J. O. Trent, Nucleic Acids Res. 2008, 36, 5482–
5515.
[5] a) A. Piazza, J. B. Boule, J. Lopes, K. Mingo, E. Largy, M. P. Teu-
lade-Fichou, A. Nicolas, Nucleic Acids Res. 2010, 38, 4337–4348;
b) A. Mendez-Bermudez, M. Hills, H. A. Pickett, A. T. Phan, J. L.
Mergny, J. F. Riou, N. J. Royle, Nucleic Acids Res. 2009, 37, 6225–
6238.
[6] H. J. Lipps, D. Rhodes, Trends Cell Biol. 2009, 19, 414–422.
[7] a) N. Sreenivasachary, J. M. Lehn, Proc. Natl. Acad. Sci. USA 2005,
102, 5938–5943; b) T. Giorgi, S. Lena, P. Mariani, M. A. Cremonini,
S. Masiero, S. Pieraccini, J. P. Rabe, P. Samori, G. P. Spada, G. Got-
tarelli, J. Am. Chem. Soc. 2003, 125, 14741–14749; c) D. Gonzalez-
Rodriguez, P. G. A. Janssen, R. Martin-Rapun, I. De Cat, S. De
Feyter, A. Schenning, E. W. Meijer, J. Am. Chem. Soc. 2010, 132,
4710–4719; d) M. Nikan, J. C. Sherman, Angew. Chem. 2008, 120,
4978–4980; Angew. Chem. Int. Ed. 2008, 47, 4900–4902.
Conclusion
Although TrisK3-NH and TrisQ share common features
(shape, size, and triscationic charge under the experimental
conditions) and can stack on G-quartets, they clearly inter-
act with quadruplexes with different strengths and differ by
their capacity to discriminate one quadruplex from another
and quadruplex- from duplex-DNA. TrisK3-NH interacts
certainly in a dual mode involving quartet and loops, where-
as the interaction of TrisQ appears most exclusively driven
by efficient p-stacking forces with the quartet surface. How-
ever, a comparison of data collected on both telomeric and
c-myc quadruplexes suggests that the naked, nonsubstituted
platform is more efficient both in terms of quadruplex-bind-
ing strength and in terms of quadruplex- over duplex-DNA
selectivity (at least in K+, the physiologically relevant condi-
tions). It is likely that the presence of flexible amino-termi-
nated side chains represents a double-edged sword which,
on one hand, holds a great potential for establishing hydro-
gen bonding interactions with the residues surrounding the
quartets and on the other hand, favors nonspecific external
electrostatic interactions with both quadruplex and duplex
architectures. The trisusbstituted TrisK scaffold remains
nonetheless an attractive design, which retains a high affini-
ty and selectivity for the telomeric form in Na+ and may be
optimized by chemical engineering of the pendant side
chains.
[8] a) S. Neidle, G. Parkinson, Nat. Rev. Drug Discovery 2002, 1, 383–
393; b) L. H. Hurley, Nat. Rev. Cancer 2002, 2, 188–200.
[9] a) M. Wieland, J. S. Hartig, Chem. Biol. 2007, 14, 757–763; b) S.
Kumari, A. Bugaut, J. L. Huppert, S. Balasubramanian, Nat. Chem.
Biol. 2007, 3, 218–221; c) D. Gomez, A. Guedin, J. L. Mergny, B.
Salles, J. F. Riou, M. P. Teulade-Fichou, P. Calsou, Nucleic Acids Res.
2010, 38, 7187–7198.
[10] a) D. Monchaud, M. P. Teulade-Fichou, Org. Biomol. Chem. 2008, 6,
627–636; b) M. Franceschin, Eur. J. Org. Chem. 2009, 2225–2238;
c) D. Monchaud, A. Granzhan, N. Saettel, A. Guꢁdin, J.-L. Mergny,
M.-P. Teulade-Fichou, J. Nucleic Acids 2010, DOI: 10.4061/2010/
525862; d) S. N. Georgiades, N. H. Abd Karim, K. Suntharalingam,
R. Vilar, Angew. Chem. 2010, 122, 4114–4128; Angew. Chem. Int.
Ed. 2010, 49, 4020–4034.
[11] a) S. Neidle, Curr. Opin. Struct. Biol. 2009, 19, 239–250; b) A. T.
Phan, FEBS J. 2010, 277, 1107–1117.
[12] a) S. M. Haider, G. N. Parkinson, S. Neidle, J. Mol. Biol. 2003, 326,
117–125; b) N. H. Campbell, G. N. Parkinson, A. P. Reszka, S.
Neidle, J. Am. Chem. Soc. 2008, 130, 6722–6724; c) G. N. Parkinson,
R. Ghosh, S. Neidle, Biochemistry 2007, 46, 2390–2397.
[13] a) G. R. Clark, P. D. Pytel, C. J. Squire, S. Neidle, J Am Chem. Soc
2003, 125, 4066–4067; b) E. Gavathiotis, R. A. Heald, M. F. G. Ste-
vens, M. S. Searle, Angew. Chem. 2001, 113, 4885–4887; Angew.
Chem. Int. Ed. 2001, 40, 4749–4751; c) E. Gavathiotis, R. A. Heald,
M. F. G. Stevens, M. S. Searle, J. Mol. Biol. 2003, 334, 25–36; d) S.
Cosconati, L. Marinelli, R. Trotta, A. Virno, S. De Tito, R. Romag-
noli, B. Pagano, V. Limongelli, C. Giancola, P. G. Baraldi, L. Mayol,
E. Novellino, A. Randazzo, J. Am. Chem. Soc. 2010, 132, 6425–
4538
www.chemeurj.org
ꢄ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2011, 17, 4529 – 4539

Recognition of G-Quadruplex DNA
FULL PAPER
6433; e) M. J. Cocco, L. A. Hanakahi, M. D. Huber, N. Maizels, Nu-
cleic Acids Res. 2003, 31, 2944–2951.
[14] G. N. Parkinson, F. Cuenca, S. Neidle, J. Mol. Biol. 2008, 381, 1145–
1156.
[15] A. T. Phan, V. Kuryavyi, H. Y. Gaw, D. J. Patel, Nat. Chem. Biol.
2005, 1, 167–173.
[16] C. Hounsou, L. Guittat, D. Monchaud, M. Jourdan, N. Saettel, J. L.
Mergny, M. P. Teulade-Fichou, ChemMedChem 2007, 2, 655–666.
[17] G. W. Collie, S. M. Haider, S. Neidle, G. N. Parkinson, Nucl. Acids
Res. 2010, 38, 5569–5580.
[18] a) K. Shin-ya, K. Wierzba, K. Matsuo, T. Ohtani, Y. Yamada, K.
Furihata, Y. Hayakawa, H. Seto, J. Am. Chem. Soc. 2001, 123, 1262–
1263; b) T. Tauchi, K. Shin-ya, G. Sashida, M. Sumi, A. Nakajima, T.
Shimamoto, J. H. Ohyashiki, K. Ohyashiki, Oncogene 2003, 22,
5338–5347; c) N. Temime-Smaali, L. Guittat, A. Sidibe, K. Shin-ya,
C. Trentesaux, J.-F. Riou, PLoS ONE 2009, 4, e6919.
[19] S. Agrawal, R. P. Ojha, S. Maiti, J. Phys. Chem. B 2008, 112, 6828–
6836.
[20] a) A. De Cian, L. Guittat, K. Shin-ya, J.-F. Riou, J.-L. Mergny, Nu-
cleic Acids Symposium Series 2005, 49, 235–236; b) D. Monchaud,
C. Allain, M. P. Teulade-Fichou, Bioorg. Med. Chem. Lett. 2006, 16,
4842–4845.
[34] A. De Cian, P. Grellier, E. Mouray, D. Depoix, H. Bertrand, D.
Monchaud, M. P. Teulade-Fichou, J. L. Mergny, P. Alberti, ChemBio-
Chem 2008, 9, 2730–2739.
[35] A. De Cian, PhD Thesis, Universitꢁ Pierre et Marie Curie, Paris VI
(France) 2007.
[36] a) K. W. Lim, S. Amrane, S. Bouaziz, W. X. Xu, Y. G. Mu, D. J.
Patel, K. N. Luu, A. T. Phan, J. Am. Chem. Soc. 2009, 131, 4301–
4309; b) G. N. Parkinson, M. P. H. Lee, S. Neidle, Nature 2002, 417,
876–880; c) A. T. Phan, D. J. Patel, J. Am. Chem. Soc. 2003, 125,
15021–15027; d) A. Ambrus, D. Chen, J. X. Dai, T. Bialis, R. A.
Jones, D. Z. Yang, Nucleic Acids Res. 2006, 34, 2723–2735.
[37] A. Arola-Arnal, J. Benet-Buchholz, S. Neidle, R. Vilar, Inorg.
Chem. 2008, 47, 11910–11919.
[38] a) 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; b) S. Neidle, FEBS J.
2010, 277, 1118–1125.
[39] D. Monchaud, C. Allain, H. Bertrand, N. Smargiasso, F. Rosu, V.
Gabelica, A. De Cian, J. L. Mergny, M. R. Teulade-Fichou, Bio-
chimie 2008, 90, 1207–1223.
[40] V. Gonzalez, L. H. Hurley, Annu. Rev. Pharmacol. Toxicol. 2010, 50,
111–129.
[21] A. Granzhan, H. Ihmels, K. Jꢆger, Chem. Commun. 2009, 1249–
1251.
[41] a) P. Wu, D. L. Ma, C. H. Leung, S. C. Yan, N. Y. Zhu, R. Abagyan,
C. M. Che, Chem. Eur. J. 2009, 15, 13008–13021; b) E. A. Owen,
M. A. Keniry, Aust. J. Chem. 2009, 62, 1544–1549; c) J. Dash, P. S.
Shirude, S. Balasubramanian, Chem. Commun. 2008, 3055–3057;
d) T. Lemarteleur, D. Gomez, R. Paterski, E. Mandine, P. Mailliet,
J. F. Riou, Biochem. Biophys. Res. Commun. 2004, 323, 802–808.
[42] a) A. De Cian, L. Lacroix, C. Douarre, N. Temime-Smaali, C. Tren-
tesaux, J. F. Riou, J. L. Mergny, Biochimie 2008, 90, 131–155; b) C.
Granotier, G. Pennarun, L. Riou, F. Hoffschir, L. R. Gauthier, A.
De Cian, D. Gomez, E. Mandine, J. F. Riou, J. L. Mergny, P. Mailliet,
B. Dutrillaux, F. D. Boussin, Nucleic Acids Res. 2005, 33, 4182–4190.
[43] a) P. Wang, C.-H. Leung, D.-L. Ma, S.-C. Yan, C.-M. Che, Chem.
Eur. J. 2010, 16, 6900–6911; b) J. E. Reed, A. J. P. White, S. Neidle,
R. Vilar, Dalton Trans. 2009, 2558–2568; c) J. Alzeer, B. R. Vummi-
di, P. J. C. Roth, N. W. Luedtke, Angew. Chem. 2009, 121, 9526–
9529; Angew. Chem. Int. Ed. 2009, 48, 9362–9365; d) T. Agarwal, S.
Roy, T. K. Chakraborty, S. Maiti, Bioorg. Med. Chem. Lett. 2010, 20,
4346–4349.
[22] a) J. T. Kern, P. W. Thomas, S. M. Kerwin, Biochemistry 2002, 41,
11379–11389; b) V. Casagrande, A. Alvino, A. Bianco, G. Ortaggi,
M. Franceschin, J. Mass Spectrom. 2009, 44, 530–540.
[23] a) J. L. Mergny, L. Lacroix, M. P. Teulade-Fichou, C. Hounsou, L.
Guittat, M. Hoarau, P. B. Arimondo, J. P. Vigneron, J. M. Lehn, J. F.
Riou, T. Garestier, C. Helene, Proc. Natl. Acad. Sci. USA 2001, 98,
3062–3067; b) M. P. Teulade-Fichou, C. Carrasco, L. Guittat, C.
Bailly, P. Alberti, J. L. Mergny, A. David, J. M. Lehn, W. D. Wilson,
J. Am. Chem. Soc. 2003, 125, 4732–4740.
[24] N. Saettel, N. Katsonis, A. Marchenko, M. P. Teulade-Fichou, D.
Fichou, J. Mater. Chem. 2005, 15, 3175–3180.
[25] H. Bertrand, PhD Thesis, Universitꢁ Pierre et Marie Curie, Paris VI
(France), 2008.
[26] L. Ginnari-Satriani, V. Casagrande, A. Bianco, G. Ortaggi, M. Fran-
ceschin, Org. Biomol. Chem. 2009, 7, 2513–2516.
[27] J. E. Moses, D. J. Ritson, F. Zhang, C. M. Lombardo, S. Haider, N.
Oldham, S. Neidle, Org. Biomol. Chem. 2010, 8, 2926–2930.
[28] R. Lartia, H. Bertrand, M. P. Teulade-Fichou, Synlett 2006, 610–614.
[29] a) S. Arai, M. Hida in Advances in Heterocyclic Chemistry, Vol. 55
(Ed.: A. R. Katritzky), Academic Press, 1992, pp. 261–358; b) H.
Ihmels in Science of Synthesis, Vol. 15: Six-Membered Hetarenes
with One Nitrogen or Phosphorus Atom (Ed.: D. Black), Georg
Thieme Verlag, Stuttgart, 2004, pp. 907–946.
[44] Marvin, version 5.2, ChemAxon, 2009.
[45] Maestro, version 9.0.211, Schrçdinger, LLC, New York, NY, 2009.
[46] E. F. Pettersen, T. D. Goddard, C. C. Huang, G. S. Couch, D. M.
Greenblatt, E. C. Meng, T. E. Ferrin, J. Comput. Chem. 2004, 25,
1605–1612.
[47] C. A. Hunter, K. R. Lawson, J. Perkins, C. J. Urch, J. Chem. Soc.
Perkin Trans. 2 2001, 651–669.
[30] O. Baudoin, M. P. Teulade-Fichou, J. P. Vigneron, J. M. Lehn, J. Org.
Chem. 1997, 62, 5458–5470.
[31] C. K. Bradsher, J. P. Sherer, J. Org. Chem. 1967, 32, 733–737.
[32] A. De Cian, L. Guittat, M. Kaiser, B. Sacca, S. Amrane, A. Bour-
doncle, P. Alberti, M. P. Teulade-Fichou, L. Lacroix, J. L. Mergny,
Methods 2007, 42, 183–195.
[48] J. Debray, W. Zeghida, M. Jourdan, D. Monchaud, M.-L. Dheu-An-
dries, P. Dumy, M.-P. Teulade-Fichou, M. Demeunynck, Org.
Biomol. Chem. 2009, 7, 5219–5228.
Received: September 30, 2010
Revised: January 3, 2011
Published online: March 17, 2011
[33] R. D. Gray, L. Petraccone, J. O. Trent, J. B. Chaires, Biochemistry
2010, 49, 179–194.
Chem. Eur. J. 2011, 17, 4529 – 4539
ꢄ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
4539