Nucleic acids targeted to drugs: SELEX against a quadruplex ligand

Article abstract of DOI:10.1016/j.biochi.2011.05.022

A number of small molecules demonstrate selective recognition of G-quadruplexes and are able to stabilize their formation. In this work, we performed the synthesis of two biotin-tagged G4 ligands and analyzed their interactions with DNA by two complementa

Full text of DOI:10.1016/j.biochi.2011.05.022

Biochimie 93 (2011) 1357e1367
Contents lists available at ScienceDirect
Biochimie
journal homepage: www.elsevier.com/locate/biochi
Research paper
Nucleic acids targeted to drugs: SELEX against a quadruplex ligand
,
b
,
,
,
,b
Amandine Renaud de la Faverie a 1, Florian Hamon c 1, Carmelo Di Primo a b, Eric Largy c, Eric Dausse a
,
Laurence Delaurière a b, Corinne Landras-Guetta c, Jean-Jacques Toulmé a
**
,
,
,b,
***
Marie-Paule Teulade-Fichou c , Jean-Louis Mergny
,
a,b,
*
a Université de Bordeaux, Laboratoire ARNA, F-33000 Bordeaux, France
b INSERM, U869, IECB, F-33600 Pessac, France
c Institut Curie, Section Recherche, CNRS UMR176, Centre Universitaire Paris XI, Bât. 110, F-91405 Orsay, France
a r t i c l e i n f o
a b s t r a c t
Article history:
A number of small molecules demonstrate selective recognition of G-quadruplexes and are able to
stabilize their formation. In this work, we performed the synthesis of two biotin-tagged G4 ligands and
analyzed their interactions with DNA by two complementary techniques, FRET and FID. The compound
that exhibited the best characteristics (a biotin pyridocarboxamide derivative with high stabilization of
an intramolecular quadruplex and excellent duplexequadruplex specificity) was used as bait for in vitro
selection (SELEX). Among 80 DNA aptamer sequences selected, only a small minority (5/80) exhibited
G4-prone motifs. Binding of consensus candidates was confirmed by SPR. These results indicate that G4
ligands that appear highly specific when comparing affinities or stabilization for one quadruplex against
one duplex, do not only bind quadruplex sequences but may also recognize other nucleic motifs as well.
This observation may be relevant when whole genome or transcriptome analysis of binding sites is
seeked for, as unexpected binding sites may also be present.
Received 6 April 2011
Accepted 24 May 2011
Available online 31 May 2011
Keywords:
Quadruplexes
Unusual nucleic acids structure
Aptamer
SELEX
SPR
DrugeDNA interactions
Ó 2011 Elsevier Masson SAS. All rights reserved.
1. Introduction
treatment [9]. Since their initial discovery in 1997 [10], the number
of known G-quadruplex ligands has grown rapidly over the past
Quadruplex ligands bind to a family of DNA secondary struc-
tures called G-quadruplexes, which result from the stacking of
several G-quartets (typically 2e5), each quartet being composed of
four coplanar guanines [1,2]. The level of interest in these structures
has increased due to the recent demonstration that G-quadruplex
structures play roles in key biological processes [3e6]. Quad-
ruplexes have now been studied in organisms from Escherichia coli
to humans. It is estimated that over 376,000 G-quadruplexes are
present in the human genome [7,8].
Small molecules that stabilize G-quadruplex structures often
exhibit interesting biological activities such as antiproliferative
properties. These ligands may interfere with telomere conforma-
tion and telomere elongation, opening new possibilities for cancer
few years [11]. While first generation compounds exhibited a rela-
tively limited duplex vs. quadruplex selectivity, a variety of scaffolds
lead to an exquisite selectivity toward G4 structures over duplexes
(for ex [12]). On the other hand, a still underachieved objective is to
design a ligand exhibiting specific binding for one quadruplex
topology over the others.
To understand the specificity of recognition, one approach is to
quantify binding to a number of independent sequences, either in
parallel or in succession. Besides the now classical competition
dialysis test [13], we have recently developed two assays that allow
testing 20 or more sequences ([14], and Tran et al., published in the
same issue). However, if scaling up the process would perhaps
allow screening a hundred or so motifs, this number is far from the
possible sequence diversity of G4-forming sequences e and even
less for random sequence motifs. To obtain a more general picture
of the sequence preference of these compounds, we reasoned that
in vitro evolution of nucleic acid sequences (the so called SELEX
method) could provide interesting information. Indeed up to 1014
different sequences (i.e., structures) can be screened at once
through SELEX for their specific binding property to a given target.
Aptamers have been raised against a wide range of molecules,
including nucleic acid hairpins [23]. By starting from a random pool
* Corresponding author. INSERM, U869, IECB, 33600 Pessac Cedex, France.
Tel.: þ33 (0) 540 003 034; fax: þ33 (0) 557 571 015.
** Corresponding author. CNRS UMR 176 ; Bat. 110, Institut Curie, Orsay, France.
*** Corresponding author. Université de Bordeaux, Laboratoire ARNA, F-33000
Bordeaux, France.
E-mail addresses: jean-jacques.toulme@inserm.fr (J.-J. Toulmé), mp.teulade-fichou@
curie.u-psud.fr (M.-P. Teulade-Fichou), jean-louis.mergny@inserm.fr (J.-L. Mergny).
1
These authors contributed equally.
0300-9084/$ e see front matter Ó 2011 Elsevier Masson SAS. All rights reserved.
doi:10.1016/j.biochi.2011.05.022

1358
A. Renaud de la Faverie et al. / Biochimie 93 (2011) 1357e1367
of sequences (in here, a random window of 30 nucleotides was
chosen), one may hope to select the motifs that bind more tightly to
a G4 ligand. Such selection would be facilitated by working with
a biotin-labeled G4 ligand, as [15].
In this work, we selected two unrelated G4-binding scaffolds i.e.,
pyridodicarboxamide and a copper terpyridine, prepared their
biotin derivatives and analyzed their quadruplex stabilization and
selectivity via FRET melting and FID experiments. The most
promising compound, PDC-biotin, was then used as a bait to fish
DNA ligands via the SELEX approach. A number of candidate
sequences were identified, and binding to the ligand (with or
without biotin) was confirmed by SPR.
concentrated to dryness under vacuum. The residue was taken back
in 50 ml of CH2Cl2 and 50 ml of H2O. The white precipitate was
filtered, washed with CH2Cl2 and H2O (3x each). This purification
afforded biotin N-hydrosuccinimidyl ester 1 (2.10 g, 6.15 mmol, 75%
yield) as a white powder (Scheme 3).
m.p.: 203e204 ꢀC; 1H NMR (300 MHz, CDCl3 þ 1 drop DMSO-
d6):
d
¼ 6.43 (s, 1H), 6.37 (s, 1H), 4.33e4.27 (m, 1H), 4.17e4.13
(m, 1H), 3.05 (q, J ¼ 11.4 Hz, J ¼ 6.9 Hz, 1H), 2.86e2.79 (m, 5H), 2.68
(t, J ¼ 7.2 Hz, 2H), 2.58 (d, J ¼ 12.6 Hz, 1H), 1.70e1.59 (m, 3H),
1.56e1.36 (m, 3H) ppm; 13C NMR (75 MHz, DMSO-d6):
d
¼ 170.3,
169.0, 167.7, 61.0, 59.2, 55.3, 30.0, 27.9, 27.6, 25.5, 24.3 ppm; LRMS
(ESI-MS): 364.1 [M þ Na]þ.
2.1.2. Tert-butyl-4,7,10-trioxa-1,13-tridecanamine carbamate 2
2. Material and methods
In
a
250 mL round-bottomed flask, 4,7,10-trioxa-1,13-
tridecanamine (4.00 ml, 18.34 mmol, 1.0 eq.) was dissolved in
ethanol (100 ml) to afford a colorless solution. Tert-butyl phenyl
carbonate (3.39 ml, 18.34 mmol, 1.0 eq.) was added and the
mixture was refluxed for 20 h. Reaction mixture was concentrated
under vacuum and dissolved in H2O (150 ml). The solution was
slightly acidified (pH w 4e5) with 1 M HCl and extracted with
CH2Cl2 (3 ꢁ 150 ml). The aqueous phase was strongly basified with
concentrated aq. NaOH (pHw12e13) and extracted with CH2Cl2
(3 ꢁ 150 ml). The organic phases were combined, dried and
concentrated under vacuum. The residue was purified by flash
chromatography on silica gel (CH2Cl2/MeOH/NH4OH: 80/17.5/2.5)
affording 2 (2.88 g, 9.00 mmol, 49% yield) as an oil.
2.1. Synthesis description
Firstly, the N-hydrosuccinimide ester of biotin 1 was prepared
in presence of coupling agent (EDCi and DMAP) in DMF and
then precipitated from a dichloromethaneewater mixture (1:1)
with
a 75% yield. The monoprotection of 4,7,10-trioxa-1,1,3-
tridecanamine was achieved in presence of tert-butyl phenyl
carbonate in ethanol under reflux. After workup, 2 was isolated
with 49% yield. 1 and 2 were efficiently coupled in DMF and the
BOC-protecting group was removed in acidic conditions affording
compound 3 with a 92% yield over the 2 steps (Scheme 1).
On the other hand, 4-chloropyridine-2,6-dicarboxylic acid 4
was obtained with a 75% yield from chelidamic acid by reaction
in phenylphosphonic acid followed by a slow hydrolysis. 4 was
coupled with quinolin-3-amine in DMF in presence of EDCi and
HOBt to afford PDC-Cl 5 with a 78% yield by precipitation from
the reaction mixture. The nucleophilic substitution between 3
and 5 in DMSO afforded 6 with 63% yield. Methylation with
iodomethane in DMF, afforded PDC-biotin derivative with 87%
yield (Scheme 2).
The original Cu-ttpy-biotin was prepared in two steps starting
from bromethyl derivative Br-ttpy [16]. First, nucleophilic substi-
tution of Br-ttpy by 3 in presence of triethylamine in DMF, afforded
ttpy-biotin with 19% yield. Then, ttpy-biotin previously dissolved in
dichloromethane, was deposited in a vial and a solution of copper
nitrate in acetonitrile was carefully layered [16]. A slow precipita-
tion at the interface provided Cu-ttpy-biotin, which was recovered
by filtration with 50% yield (Scheme 3).
1H NMR (300 MHz, CDCl3):
3.24e3.20 (m, 2H), 2.82 (m, J ¼ 6.5 Hz, 2H), 1.75e1.70 (m, 4H), 1.43
(s, 9H) ppm; 13C NMR (75 MHz, CDCl3):
d
¼ 5.12 (s, 1H), 3.65e3.55 (m, 12H),
d
¼ 156.2, 79.0, 70.7
(2 peaks), 70.3 (2 peaks), 69.6 (2 peaks), 39.7, 38.7, 33.3, 29.7,
28.6 ppm; LRMS (ESI-MS): 321.3 [M þ H]þ.
2.1.3. N-biotin-4,7,10-trioxa-1,13-tridecanamine 3
In a 100 ml flask, 1 (461 mg, 1.35 mmol, 1.0 eq.) and 2 (433 mg,
1.35 mmol, 1 eq.) were dissolved in DMF (15 ml) and stirred at RT
for 16 h. The mixture was concentrated under vacuum. The residue
was dissolved in HCl/MeOH ([HCl] 3 M) and stirred at RT for 16 h.
After basification with NaOH (3 M), mixture was slowly concen-
trated under vacuum. The residue was purified by flash chroma-
tography on silica gel (CH2Cl2/MeOH/NH4OH (25%): 80/17.5/2.5)
affording 3 (555 mg, 1.24 mmol, 92% yield) as a colorless oil, that
solidifies at standing.
1H NMR (300 MHz, CDCl3):
d
¼ 7.01 (s, 1H), 6.64 (s, 1H), 5.84 (s,
1H), 4.50 (m, 1H), 4.31 (m, 1H), 3.64e3.56 (m, 12H), 3.33 (m, 2H),
3.14 (m, 1H), 2.92e2.72 (m, 4H), 2.52 (d, J ¼ 12.6 Hz, 2H), 2.21
(t, J ¼ 6.5 Hz, 2H), 1.78e1.67 (m, 8H), 1.44 (m, 2H) ppm; 13C NMR
2.1.1. Biotin N-hydrosuccinimidyl ester 1
In a 100 mL round-bottomed flask, biotin (2.00 g, 8.19 mmol, 1.0
eq.) and N-hydroxysuccinimide (1.04 g, 9.00 mmol, 1.1 eq.) were
dissolved in hot DMF (30 ml) affording a colorless solution. DMAP
(5 mg) and EDCi (1.73 g, 9.00 mmol, 1.1 eq.) were added and
solution was stirred overnight at RT. Reaction mixture was
(75 MHz, CDCl3):
d
¼ 173.5, 163.9, 70.5, 70.3, 70.0, 69.8, 69.7, 69.5,
62.0, 60.1, 55.9, 39.5, 37.2, 35.7, 31.2, 28.8, 28.1, 25.6 ppm; LRMS
(ESI-MS): 447.2 [M þ H]þ.
Scheme 1. a) N-hydrosuccinimide, EDC, DMAP, DMF, RT, 16 h, 75%; b) tert-butyl-phenyl-carbonate, ethanol, reflux, 20 h, 49%; c) DMF, RT, 16 h; d) HCl, MeOH, RT, 16 h, 92% over 2 steps.

A. Renaud de la Faverie et al. / Biochimie 93 (2011) 1357e1367
1359
Scheme 2. a) Phenylphosphonic dichloride, 120 ꢀC, 2 h; b) H2O, RT, 1 h, 75% over 2 steps; c) quinolin-3-amine, EDCi, HOBt, CH2Cl2/DMF, RT, 16 h, 78%; d) 3, NEt3, DMSO, 90 ꢀC, 16 h,
63%; e) CH3I, DMF, 40 ꢀC, 16 h, 87%.
2.1.4. 4-chloropyridine-2,6-dicarboxylic acid 4
m.p.: 211e212 ꢀC; 1H NMR (300 MHz, DMSO-d6):
d
¼ 8.23 (s, 2H)
In a 50 mL round-bottomed flask, phenylphosphonic dichloride
(3.06 ml, 21.8 mmol, 4.0 eq.) was added to 4-oxo-1,4-
dihydropyridine-2,6-dicarboxylic acid (1.0 g, 5.46 mmol, 1.0 eq.)
and the reaction mixture was heated at 120 ꢀC for 2 h under an inert
atmosphere. After cooling, 5 ml H2O were added dropwise. The
mixture was stirred vigorously for 1 h and concentrated to dryness
under vacuum. The residue was suspended in water and filtered,
washed with water (2x), CH2Cl2 (2x) and Et2O (3x) affording 4
(831 mg, 4.08 mmol, 75% yield).
ppm; 13C NMR (75 MHz, DMSO-d6):
d
¼ 164.6, 150.0, 145.3, 127.3;
LRMS (ESI-MS): 224.1 [M þ Na]þ.
2.1.5. PDC-Cl 5
In a 100 ml flask, 4-chloropyridine-2,6-dicarboxylic acid 4
(275 mg, 1.351 mmol, 1.0 eq.) and quinolin-3-amine (487 mg,
3.38 mmol, 2.5 eq.) were suspended in CH2Cl2/DMF. HOBt (41.3 mg,
0.338 mmol, 0.25 eq.) and EDCi (647 mg, 3.38 mmol, 2.5 eq.) were
added successively and the mixture was stirred for 16 h at RT. The
Scheme 3. a) 3, NEt3, DMF, 60 ꢀC, 4d, 19%; b) CuNO3, CH3CN, CH2Cl2, 4 ꢀC, 24 h, 50%.

1360
A. Renaud de la Faverie et al. / Biochimie 93 (2011) 1357e1367
yellow mixture was homogeneously turning pink with formation of
a precipitate. The reaction mixture was filtered and solid was
carefully washed with H2O (3x) and CH2Cl2 (3x). The solid was
dried under vacuum affording 5 (480 mg, 1.06 mmol, 78% yield) as
a pale yellow powder.
(t, J ¼ 6.0Hz, 2H), 1.78e1.67 (m, 8H), 1.44 (m, 2H) ppm; LRMS (ESI-
MS): 768.5 [M þ H]þ.
2.1.9. Cu-ttpy-biotin
In a vial, copper nitrate (24 mg, 0.128 mmol, 3.0 eq.) in aceto-
nitrile (1 ml) was carefully added dropwise over ttpy-biotin 8
(33 mg, 0.043 mmol, 1.0 eq.) in CH2Cl2 (1 ml) to form two immis-
cible layers, that was then kept at 4 ꢀC for 4 days. The green solid
which formed at the interface was filtered and carefully washed
with acetonitrile, CH2Cl2 and Et2O affording Cu-ttpy-biotin (19 mg,
0.021 mmol, 50% yield) as a green powder.
m.p.: >230ꢀC; 1H NMR (300 MHz, DMSO-d6):
d
¼ 11.48 (s, 2H),
9.37 (s, 2H), 8.98 (s, 2H), 8.46 (s, 2H), 8.07e8.04 (m, 4H), 7.74
(t, J ¼ 7 Hz, 2H), 7.65 (t, J ¼ 7 Hz, 2H) ppm; LRMS (ESI-MS): 452.1
[M ꢂ H]ꢂ.
2.1.6. PDC-PEG-biotin 6
In a 50 ml flask, 5 (50 mg, 0.11 mmol, 1.0 eq.), 3 (98 mg,
0.022 mmol, 2.0 eq.) and triethylamine (0.15 ml, 0.11 mmol, 1.0 eq.)
were dissolved in anhydrous DMSO (1.5 ml) and reaction mixture
was stirred overnight at 90 ꢀC under argon atmosphere. After the
addition of H2O, a precipitate appeared which was filtered off. The
solid was suspended in a mixture of absolute EtOH and Et2O which
was evaporated to dryness. The resulted product was purified by
chromatography on neutral alumina using CH2Cl2/EtOH (98/2) as
mobile phase affording 6 (60 mg, 0.069 mmol, 63% yield) as a white
powder.
LRMS (ESI-MS): 845.3 [MeNO3 þ OH]þ; HRMS (ESI-MS): m/z
calculated for C42H53N8O8SCu: 892.3003; found 892.3022. HPLC: 1
peak at 4.8 min in a multistep gradient H2O (TFA 0.5%)/CH3CN.
2.1.10. Characterization
1H and 13C spectra were recorded at 25 ꢀC on a Bruker Avance 300
using TMS as internal standard. Deuterated CDCl3 and DMSO-d6 were
purchased from SDS. The following abbreviations are used: singlet
(s), doublet (d) and multiplet (m). Low resolution mass spectrometry
(ESI-MS) was recorded on a micromass ZQ 2000 (waters). High
resolution mass spectrometry was provided by the I.C.S.N. (Gif/
Yvette, France). Melting points were taken on a Kofler melting point
apparatus and are uncorrected. Silica gel chromatography was
m.p.: 225e226 ꢀC; 1H NMR (300 MHz, CDCl3):
d
¼ 11.29 (s, 2H),
9.36 (s, 2H), 8.96 (s, 2H), 8.04 (d, J ¼ 8.1 Hz, 4H), 7.77e7.48 (m, 8H),
6.42 (s, 1H), 6.36 (s, 1H), 4.28 (m, 1H), 4.10 (m, 1H), 3.58e3.46 (m,
10H), 3.41e3.34 (m, 2H), 3.05 (m, 3H), 2.79 (dd, J ¼ 12.3 Hz,
J ¼ 4.8 Hz, 1H), 2.55 (d, J ¼ 12.9 Hz, 1H), 2.03 (t, J ¼ 7.2 Hz, 2H), 1.85
(t, J ¼ 6 Hz, 2H), 1.64e1.42 (m, 6H), 1.28 (m, 2H) ppm; 13C NMR
carried out with Merck silica gel (Si 60, 40e63 mm). Reagents and
chemicals were purchased from SigmaeAldrich unless otherwise
stated. Solvents were purchased from SDS. Alumina gel chromatog-
raphy was carried out with Merk aluminiumoxid (70e230 mesh
ASTM) 90 aktiv neutral which was deactivated with addition of water
(7%) corresponding approximately to an activity grade III.
(75 MHz, CDCl3):
d
¼ 171.8, 163.1, 162.7, 156.3, 146.1, 144.2, 131.9,
128.6, 128.3, 127.9, 127.7, 127.1, 124.1, 69.8, 69.6, 69.5, 68.1, 67.6, 61.0,
59.1, 55.4, 35.7, 35.2, 29.4, 28.5, 28.2, 28.0, 25.3 ppm; LRMS (ESI-
MS): 864.6 [M þ H]þ.
2.2. Absorbance measurements
2.1.7. PDC-biotin
In a 50 ml flask, 6 (30 mg, 0.035 mmol, 1.0 eq.) was dissolved in
anhydrous DMF (0.13 ml) and iodomethane (0.4 ml) was added
dropwise. The reaction mixture was stirred at 40 ꢀC overnight
under argon atmosphere. After concentration under vacuum,
a precipitate was induced by addition of absolute EtOH. The
precipitate was filtered off and dried with Et2O affording PDC-
biotin (35 mg, 0.030 mmol, 87% yield) as an orange powder.
Experiments are performed on a Uvikon XL spectrophotometer.
2.2.1. Thermal difference spectra (TDS)
By measuring the difference between the absorbance spectra
(between 220 and 340 nm) at high temperature (oligonucleotide is
totally unfolded) and the one at low temperature (oligonucleotide
is folded) one can obtain a thermal difference spectra (TDS) which
provides information on the nature of the folded form [17]. Thermal
difference spectrum of G4 has two positive peaks around 249 and
270 nm and one negative at 295 nm. A 1 cm path length quartz cell
m.p.: 139e140 ꢀC; 1H NMR (300 MHz, CDCl3):
d
¼ 11.82 (s, 2H),
10.15 (s, 2H), 9.68 (s, 2H), 8.54 (d, J ¼ 8.1 Hz, 4H), 8.24 (m, 2H), 8.08
(m, 2H), 7.73 (m, 2H), 7.59 (bs, 1H), 6.41 (m, 1H), 4.77 (s, 6H), 4.29
(s, 1H), 4.11 (s, 1H), 3.57e3.35 (m, 17H), 3.05 (m, 3H), 2.78 (m, 1H),
2.01 (s, 2H), 1.86 (m, 2H), 1.66e1.34 (m, 6H), 1.27 (s, 2H) ppm;
LRMS (ESI-MS): 446.9 [M ꢂ 2I]2þ; 1020.7 [M þ H ꢂ I]þ; HRMS
(ESI-MS): m/z calculated for C47H59N9O7SI: 1020.3303; found
1020.335, HPLC: 1 peak at 14.9 min in a multistep gradient H2O
(TFA 0.5%)/CH3CN.
is used in a reaction volume of 580
m
L. Oligonucleotides are
prepared as a 4
m
M solution in 10 mM lithium cacodylate pH 7.2,
100 mM NaCl or KCl buffer and annealed by heating to 90 ꢀC for
2 min, followed by cooling to 20 ꢀC. Scans are performed over
a wavelength range of 220 (or 240)e295 nm.
2.2.2. Tm determination
2.1.8. Ttpy-biotin 8
Absorbance is measured at 240 and 295 nm as previously
described [18,19].
In a 50 ml flask, 40-(4-bromo-methyl-phenyl)-2,20:60,200-terpyr-
idine (90 mg, 0.22 mmol, 1.0 eq.), triethylamine (0.315
mL,
2.24 mol, 0.1 eq.) and 3 (100 mg, 0.22 mmol, 1.0 eq.) in DMF
m
2.2.3. Circular dichroism (CD)
(minimum amount) were heated to 60 ꢀC for 4 days. The solution
was diluted in CH2Cl2 and washed twice with aq. NH4Cl (2 ꢁ 20 ml).
The organic phases were dried over MgSO4, filtrated and evapo-
rated. The residue was purified on silica gel with CH2Cl2/MeOH/
NH4OH (80/17.5/2.5) as eluent affording ttpy-biotin 8 (32.8 mg,
0.043 mmol, 19% yield) as a white powder.
CD spectra are recorded on a JASCO-J815 spectropolarimeter as
previously described [20] with a scanning speed of 500 nm minꢂ1
,
a response time of 1 s, 1 nm data pitch and 1 nm bandwidth.
2.3. Fluorescence
1H NMR (300 MHz, CDCl3):
d
¼ 8.73 (bs, 4H), 8.67 (d, J ¼ 8.1 Hz,
2.3.1. FRET-melting experiments
2H), 7.89 (t, J ¼ 7.5 Hz, 4H), 7.57 (d, J ¼ 7.8 Hz, 2H), 7.36 (t, J ¼ 6 Hz, 2H),
6.96 (s, 1H), 6.21 (s, 1H), 5.36 (s, 1H), 4.45 (m, 1H), 4.28 (m, 1H), 4.03
(s, 2H), 3.64e3.53 (m, 12H), 3.32 (m, 2H), 3.11 (m, 1H), 2.92e2.84
(m, 2H), 2.74 (d, J ¼ 12.9 Hz, 1H), 2.20 (t, J ¼ 7.2 Hz, 2H), 1.93
FRET-melting measurements are performed as described previ-
ously [21,22]. Experiments are carried out in 96-well plates on a real
time PCR apparatus Mꢁ3005P Stratagene as follow: 5 min at 25 ꢀC,
then increase of 1 ꢀC every minute until 95 ꢀC. Each condition is

A. Renaud de la Faverie et al. / Biochimie 93 (2011) 1357e1367
1361
tested in duplicate, in a volume of 25
mL for each sample. F21T is
2.4.2. Amplification, cloning and sequencing
prepared at 0.2 M, ligands are at 1 M and competitors at 3 or 10
m
m
m
M
Candidates are replicated using polymerase chain reaction for
20 cycles in a final volume of 1 mL, using 100 units of AmpliTaq
Gold DNA polymerase (Applied Biosystems) and P3 and P5 primers
(50 d-GGGAGACAAGAATAAACGCTCAA and 50 d-GCCTGTTGTGAGC
CTCCTGTCGAA respectively). PCR amplification of the recovered
candidates and production of single-stranded DNA for the next round
were performed according to [23,24]. The 50 extremity of the primer
is made heavier with a succession of six C3 links extended with a DNA
stretch of 20 nucleotides. During the PCR, this six C3 region cannot be
amplified by the Taq DNA polymerase. A PCR product with two
strands of unequal length is thus synthesized. Purification and
separation of both strands are performed using Nanosep 10K Omega
(PALL Life Sciences) devices and polyacrylamide gel, followed by
10 mM Tris HCl (pH 8)/25 mM NaCl/1 mM EDTA extraction and
sodium acetate and ethanol precipitation. Cloning is performed
following TOPO TA Cloning kit (Invitrogen) with E. coli One shot
TOP10 (Invitrogen) and using electroporation. PCR products from
positives colonies are send for sequencing to the Centre de Génomique
Fonctionnelle, Bordeaux (France).
final concentration. All are in a buffer of 10 mM lithium cacodylate pH
7.2, 100 mM NaCl or 10 mM KCl (completed by 90 mM LiCl) then
heated at 90 ꢀC during 2 min and finally put in ice.
2.3.2. HT-G4-FID (High-throughput quadruplex fluorescent
intercalator displacement assay)
G4-FID measurements are performed as described previously
[45]. TO (thiazole orange) and cacodylic acid were purchased
from Aldrich and used without further purification. Stock solutions
of TO (2 mM in DMSO), PDC-biotin and Cu-ttpy-biotin (500 mM in
DMSO) were used for G4-FID assay. Oligonucleotides purified by
reverse phase HPLC were purchased from Eurogentec (Belgium).
Sequences: 22AG corresponds to the human telomeric repeat:
[50-AG3(T2AG3)3-30]; c-kit1 [50-G3AG3CGCTG3AG2AG3-30] and c-kit2
[50-G3CG3(CG)2AG3AG4-30] are two sequences of the c-kit oncogene
promoter. TBA is the Thrombin Binding Aptamer sequence
[50-GGTTGGTGTGGTTGG-30]. ds26 is the duplex obtained with the self-
complementary sequence [50-CA2TCG2ATCGA2T2CGATC2GAT2G-30].
G4-FID measurements were performed on a FLUOstar Omega
microplate reader (BMG Labtech) with a 96 wells black quartz
microplate (Hellma). A temperature of 25 ꢀC is kept constant inside
the microplate reader. Every ligandeoligonucleotide combination is
tested in duplicate. The microplate is filled with i) Kþ or Naþ-buffer
2.5. Surface plasmon resonance (SPR) binding assays
The SPR experiments were carried out on a BiacoreÔ 3000
apparatus.
solution (qs for 200
nucleotides (5 M) and TO (10
and iii) an extemporaneously prepared 5
Naþ-buffer (0e100
L along the line of the microplate, i.e., from
column A to column H: 0, 0.125, 0.25, 0.375, 0.5, 0.625, 0.75,1.0,1.25,
1.5, 2.0 and 2.5 M). After 10 min of orbital shaking at 500 rpm,
fluorescence is measured using the following experimental
parameters; positioning delay: 0.5 s, 20 flashes per well, emission/
excitation (bandwidth) filters 485/520. The percentage of
displacement is calculated from the fluorescence intensity (FA),
using: percentage of displacement ¼ 100 ꢁ [(FA/FA0) ꢁ 100], FA0
being the fluorescence from TO bound to DNAwithout added ligand.
The percentage of displacement is then plotted as a function of the
concentration of added ligand. The DNA affinity was evaluated by
the concentration of ligand required to decrease the fluorescence of
TO by 50%, noted DC50, and determined after nonlinear fitting of the
displacement curve.
m
L) ii) 10
m
m
L of a solution of pre-folded oligo-
M for G4-DNA or 15 M for ds26)
M ligand solution in Kþ or
m
m
2.5.1. Evaluation of the SELEX populations
m
Around 250 RU of biotinylated PDC was immobilized on
a streptavidin-coated SAD200m sensor chip (XanTec, Dusseldorf)
prepared as previously described [25]. Binding experiments were
performed at 23 ꢀC in SE buffer (used for SELEX) containing 0.005%
P20 surfactant. Free biotin was injected at 50 nM after ligand
immobilization to saturate free streptavidin and to improve surface
stability. The SELEX populations in SE buffer were injected at
m
m
500 nM during 3 min across the sensor chip surface, at a 20 mL/min
flow rate. The regeneration of the functionalized surface was ach-
ieved with a 1 min pulse of 5 M LiCl followed by a 1 min pulse of
a solution containing 40% formamide, 3.6 M urea and 30 mM EDTA.
The surface was then washed with a 1 min pulse of SE buffer. The
sensorgrams were double-referenced using BiaEval 4.1 software
(Biacore) to remove instrument noise and buffer contribution to the
signal [26].
2.5.2. Binding assays with consensus candidates
2.4. SELEX
Around 30 RU of biotinylated oligonucleotide candidates 30a,
33a and 40a were immobilized on a SA streptavidin-coated
sensor chip (Biacore GE Healthcare, Uppsala), prepared accord-
ing to the manufacturer’s instructions. Binding experiments were
performed at 23 ꢀC in P buffer (20 mM sodium phosphate pH 7,
140 mM KCl and 20 mM NaCl). Free biotin was injected to satu-
rate free streptavidin. PDC prepared in P buffer was injected at
increasing concentrations during 1 min across the sensor surface,
In vitro selection was carried out as previously described using
a DNA library with a 30 nt random region [23]. First, DNA library is
heated at 70 ꢀC for 5 min, cooled at 4 ꢀC for 2 min and placed at
room temperature for 15 min. Then 300 pmol of DNA library (or
decreasing amounts in the successive rounds) are mixed with
streptavidin beads (50
m
g of Streptavidin MagneSphere Para-
magnetic Particles from Promega or 500
mg of Dynabeads M-280
at a 20 mL/min flow rate. The baseline level is reached after 3 min
from Invitrogen Dynal) previously equilibrated in SE buffer (20 mM
HEPES pH 7.4, 140 mM potassium acetate, 20 mM sodium acetate
and 3 mM magnesium acetate) for 30 min.
or less (not shown) which is inferior to the delay between each
injection cycle (injection starts 320 s after the beginning of the
cycle).
2.4.1. Selection steps
3. Results
Candidates not retained by the beads are mixed with 10 pmol of
the ligand (or decreasing amounts in the successive rounds) at room
3.1. Design and synthesis of PDC-biotin and Cu-ttpy-biotin
temperature in a final volume of 100 mL of SE buffer for 30 min. The
ligand is beforehand mixed with beads for 15 min at room
temperature. Unbound DNA is removed and the beads are washed
3.1.1. Choice of the G4 recognition scaffold
and derivatization strategy
with SE buffer and resuspended in 100
m
L water. Candidates are
A number of independent families of G4 ligands have been
synthesized and/or studied by our groups (to name a few:
eluted in 80
m
L from the target by heating for 1 min at 75 ꢀC.

1362
A. Renaud de la Faverie et al. / Biochimie 93 (2011) 1357e1367
Table 1
dibenzophenanthrolines/quinacridines [27], pyridodicarboxamides
[28], terpyridine metallo-organic complexes [16], cryptolepin [29],
meridine [30] or ethidium analogs [31]). Among this plethora of
compounds, the pyridodicarboxamide (PDC) bisquinolinium and ter-
pyridine metallo-organic complexes appeared as the most attractive
scaffolds due to their high affinity and selectivity, along with their
rapid and convenient synthetic access [32]. We therefore decided to
synthesize the biotin derivative of one representative member of each
series. Previous studies on these series indicated that (at least) one
position was available for functionalization. A classical design con-
sisted in linking the biotin to the ligand via a flexible linker. The chosen
linker was relatively long, to allow biotinestreptavidin interaction in
the presence of a DNA bound to the quadruplex ligand.
Binding properties of PDC and PDC-biotin.
b
Compound
ΔTm (Kþ)a
S (Kþ)b
ΔTm (Naþ)a
S (Naþ
)
PDC
PDC-biotin
Cu-ttpy
34.1
30.8
15.3
8.6
1.02
1.09
0.78
0.73
24.9
26.2
16
0.88
0.93
0.78
0.61
Cu-ttpy-biotin
6.9
a
Stabilization, in ꢀC, of the F21T apparent melting temperature, at
1 mM
compound concentration in either 10 mM KCl þ 90 mM LiCl (Kþ conditions) or
100 mM NaCl (Naþ).
b
S is the selectivity index, corresponding to the ratio between ΔTm obtained in
presence and absence of ds26, a double-stranded competitor.
contrast, Cu-ttpy-biotin is significantly less active than Cu-ttpy
(ΔTm ¼ 8.6 ꢀC and 15.3 ꢀC, at 1
mM ligand, respectively), demon-
3.1.2. Synthesis
strating that the functionalization of this ligand has a somewhat
detrimental effect. Increasing concentrations of a non-labeled
double-stranded competitor, ds26, were then added to the reaction
mixture. As shown in Fig. 1B, ds26 had little or no effect on the
quadruplex stabilization induced by PDC and PDC-biotin, even at
Both syntheses are extensively described in the experimental
section. The original PDC-biotin derivative was prepared through
a converging synthetic route. The original Cu-ttpy-biotin was
prepared in two steps starting from Br-ttpy [16].
the highest concentration tested (10 mM; i.e., a 50-fold excess over
the quadruplex F21T sequence. S z 1 for both compounds in KCl).
In contrast, Cu-ttpy-biotin exhibited somewhat degraded
properties as compared to Cu-ttpy, which is already less specific
than PDC (S < 1): the presence of ds26 led to a decrease in F21T
stabilization by the biotinylated compound, indicating that this
double-stranded sequence could act as a bait for the quadruplex
ligand. Results are summarized in Table 1.
3.2. Evaluation of the biotin derivatives
An essential condition to use these biotin derivatives for SELEX
experiments was to demonstrate that the biotin label did not
significantly affect quadruplex affinity and selectivity. To confirm
this hypothesis, we analyzed the binding of the two-biotin deriv-
atives and compared them with the non-biotinylated controls
(formulas shown in Fig. 1A; FRET results summarized in Fig. 1B).
FRET-melting experiments demonstrated that PDC-biotin retained
3.2.1. HT-G4-FID experiments
HT-G4-FID experiments confirmed that PDC-biotin retains
a high quadruplex affinity over a number of G4 structures as
a
quadruplex stabilization potential equivalent to PDC
(ΔTm ¼ 30.8 ꢀC and 34.1 ꢀC, at 1
mM ligand, respectively). In
Fig. 1. Binding properties of the quadruplex ligands (A) Chemical formula of PDC, PDC-biotin, Cu-ttpy and Cu-ttpy-biotin. (B) Quantitative analysis of the FRET-melting competition
experiments. The stabilization in potassium is indicated for the four compounds in the absence (black bars) or presence of double-stranded (ds26) at 3
gray bars).
mM (gray bars) or 10 mM (light

A. Renaud de la Faverie et al. / Biochimie 93 (2011) 1357e1367
1363
Table 2
G4-FID results.
Table 3
Base content of the selected sequences.a
Compound
22AG.K 22AG.Na TBA
c-kit1 c-kit2 ds26
Base
A
T
C
G
Totala
Cu-ttpy-biotin DC50
0.48
1.19
4
0.34
21
0.84
32
0.56
20
n.d.
n.d.
n.d.
n.d.
0.47
57
0.24
49
n.d.
n.d.
n.d.
n.d.
0.52
51
0.19
62
n.d.
n.d.
n.d.
n.d.
5.23
Number
%
540
22.5
699
29.1
541
22.6
619
25.8
2399
100
Ratioa 11
Cu-ttpy
PDC-biotin
360A
DC50
0.30
7.04b
a
80/92 Sequences were retained for analysis. Only the random sequence window
is analyzed (Most sequences are 30 nucleotide long, a few are 29 or 31 nt-long).
Ratioa 23
DC50
0.41
0.42 26.70
64
0.18 11.74b
65
Ratioa 65
DC50
0.39
Ratioa 30
amplification (experiment schematized in Supplementary Fig. S2).
In order to minimize the capture of undesired candidates, coun-
terselection steps are performed at every SELEX round to eliminate
non-specific binders interacting with naked beads. Candidates not
retained by the beads are mixed with the biotinylated ligand at
room temperature. Unbound DNA is removed and the beads are
washed, resuspended in water, then eluted by heating (the full
protocol is presented in the experimental section).
DC50 is the ligand concentration (in
mM) required for half displacement of TO.
Different structures (five quadruplexes and the ds26 duplex) were tested.
a
This ratio expresses the selectivity vs. ds26 and is calculated as follows
G4DC50
/
dsDC50
.
b
Estimation; n.d.: not determined.
compared to 360A (DC50 around 0.5 mM, except for 22AG.Na, see
Surface plasmon resonance experiments were performed to
follow the evolution of the populations after each selection round
(Fig. 2). Minimal binding was observed for the sequences selected
after the first round. In contrast, starting with round 3, the pop-
ulations bound to the PDC-functionalized sensor chip. The highest
signals were observed when populations of rounds 7 and 8 were
injected across the surface. A candidate from an in vitro selection
against a different target, using the same DNA library, was used as
a negative control. As expected, the resulting sensorgram (Fig. 2)
shows that the candidate behaves as the populations of the initial
Table 2 and Supplementary Fig. S1) together with a remarkable
selectivity vs. duplex DNA (DC50 > 25 M). Conversely, Cu-ttpy-
biotin appeared to have a significantly lower affinity and moreover
to be less selective than Cu-ttpy [16]. HT-G4-FID confirmed that
PDC-biotin was the best candidate for the SELEX study.
m
3.3. SELEX experiments
In order to perform SELEX experiments with PDC-biotin, we
synthesized a DNA library containing 30 randomized nucleotides
flanked by fixed regions complementary to primers for
rounds. In contrast, when
a well-characterized G4-forming
Fig. 2. Analysis by surface plasmon resonance of the SELEX populations binding to PDC. SELEX populations prepared at 500 nM in SE buffer were injected at 20 mL/min during 3 min
across a PDC-functionalized surface. The regeneration of the surface was achieved with a 1 min pulse of 5 M LiCl followed by a 1 min pulse of a solution containing 40% formamide,
3.6 M urea and 30 mM EDTA. The first and second arrows indicate the beginning and the end of the injection, respectively.

1364
A. Renaud de la Faverie et al. / Biochimie 93 (2011) 1357e1367
Table 4
structure, we kept that sequence within the “likely” category
Quadruplex forming potential of the selected sequence.a
(Table 4)). Nevertheless, not only G4-likely sequences were rela-
tively rare (5/80 ¼ 6.25%), they were of modest stability, and none of
the sequences are capable of forming motifs of high thermal
stability, as none of them contained 4 blocks of 3 guanines. Besides
these “likely” sequences, 15/80 sequences could be considered as
marginally compatible with quadruplex formation, meaning that
they contained clusters of 2þ guanines, but not in sufficient number
(4þ) to allow intramolecular quadruplex formation. Finally, 60/80
(75%) of the sequences were very unlikely to form quadruplexes.
In order to identify common traits between the sequences
binding to PDC-biotin, the sequences were then aligned using
Clustal [33]. Four major families were identified (Fig. 3); while the
top sequences (“A” family) were theoretically compatible with G4
formation, the consensus motifs found for the three other families
were clearly not susceptible to form G-quadruplexes. Mfold
predictions [34] of some of these sequences suggested that intra-
molecular WatsoneCrick interactions of intermediate to high
stability could be formed with some candidates (ΔG between ꢂ6
and ꢂ14.75 kcal/mol; examples of Mfold predicted stable
secondary motifs provided in supplementary information e note
that quadruplex structures are not supported by Mfold).
From the alignments presented in Fig. 3 and the Mfold predic-
tions (examples provided in Supplementary Fig. S5), 12 consensus
sequences were synthesized; their sequences are shown in Fig. 4,
top. Their lengths varied between 30 and 60 nucleotides, as they
may encompass the random region (in red) and parts of the 50 and
30 tails used for PCR, when these motifs were likely involved in
secondary structures. First, using classical methods used to
evidence quadruplex formation (CD and TDS) we confirmed that
they were not capable of forming G-quadruplexes, except for
10.13e30 and 7.7e30. We then confirmed that these sequences
were capable of binding PDC, using SPR (Fig. 4, bottom) or fluo-
rescence titration (not shown). Unfortunately, while the SPR
profiles provided in Fig. 4 unambiguously demonstrate that these
motifs interact with the PDC ligand, they did not allow a precise
determination of the Kd of the complex. In the range of concen-
G4 potential
Likelyb
Possiblec
Unlikely
Total
Number
%
5
6.25
15
18.75
60
75
80
100
a
80/88 Sequences were retained for analysis (the 8 remaining sequences were
discarded as they did not pass QC). Only the random sequence window is analyzed
(the terminal 50 and 30 motifs used for amplification do not contain G-blocks).
b
Fits the general consensus for G4 formation (four blocks of 2þ guanines).
c
The sequence contains 2e3 G-blocks of guanines, potentially allowing inter-
molecular quadruplex formation.
sequence, 32B3 Kras, was injected a high binding response was
observed (Fig. 2). This could suggest that the SELEX populations
contain candidates that are able to fold as G4 structures.
3.4. Analysis of the candidates
Ninety-two sequences were cloned in E. coli using Topo TA
cloning. We isolated sequences following rounds 7 and 10. PCR
products from positives colonies were sent for sequencing to the
Centre de Génomique Fonctionnelle, Bordeaux, France. Eighty-eight
sequences were retrieved (shown in Supplementary Fig. S3);
80 sequences of them were of excellent quality (perfect match
for both primer sequences with a central region of 29e31
nucleotides).
We then performed a global analysis of base content. Surpris-
ingly, we did not find an overrepresentation of guanines in the
initially random region (G representing 25.8% of the bases, i.e., very
close to the expected ¼ initial composition; Table 3). Had quad-
ruplex-prone sequences emerged as a dominant motif for the
selected sequences, one would have expected an enrichment of this
base. We nevertheless analyzed each sequence individually to check
for G4-forming potential. Out of the 80 selected sequences, only 5
exhibited a canonical quadruplex-prone motif, i.e., four blocks of at
least two guanines, compatible with the formation of an intra-
molecular G4 structure (note that for one of these sequences, the
distance between two G-blocks was larger than the general 7-nt
consensus accepted for bioinformatic searches: as we previously
demonstrated that longer loops could be tolerated within a G4
trations assayed (up to 50 mM) no saturation of the binding of PDC
to the immobilized oligonucleotides was observed. This suggests
either that the PDC ligand, due to its chemical nature, is able to
Fig. 3. Alignment of selected sequences. Full Clustal analysis [33] is provided as Supplementary Fig. S4; this figure illustrates four interesting conserved motifs (labeled AeD) within
the random region. Family A corresponds to two related G4-prone sequences (runs of two consecutive guanines are boxed). Identity is indicated: for example, 7.01 corresponds to
the first sequence isolated after the 7th selection round. Highly conserved nucleotides are shown in different colors for each motif. (For interpretation of the references to colour in
this figure legend, the reader is referred to the web version of this article.)

A. Renaud de la Faverie et al. / Biochimie 93 (2011) 1357e1367
1365
Fig. 4. Analysis by surface plasmon resonance of PDC binding to candidates. Top: Sequences chemically synthesized with a biotin at the 50 end and assayed by SPR. When applicable,
the corresponding family (AeD) defined in Fig. 3 is indicated on the left. Parts of the oligonucleotide corresponding to the initial random window are shown in red. Length (in nt) is
indicated after the clone reference (for ex, 10.12e40 corresponds to a 40 mer sequence taken from clone #12 isolated after round 10). Bottom: Interaction analysis of three
candidates with PDC. Biotinylated candidates were immobilized as described in Materials and Methods. PDC prepared in P buffer was injected at different concentrations (5e50 mM)
during 1 min across the sensor chip surface. No regeneration step was required between each injection. No binding was observed using a surface functionalized with a TTC control
sequence (50 d-TTCTTTCTTTCTTTCTTTCT). (A): 7.7e30 candidate. (B): 7.17e33 candidate. (C): 10.12e40 candidate. (For interpretation of the references to colour in this figure legend,
the reader is referred to the web version of this article.)
stack on itself once bound to the DNA motifs or that the Kd of the
complex is above 50 M.
proteins or small molecules [35e44]. On the other hand directed
in vitro evolution relies on the selection of shapes allowing
maximal interactions with the offered target. A successful SELEX
will lead to the identification of a folded oligomer that establishes
every possible hydrogen bond with donor or acceptor sites, and
m
4. Discussion and conclusion
The results presented here demonstrate that one can achieve
functionalization of a G4 ligand with biotin ([15] and this study)
or other labels [32]. Nevertheless, one needs to check if the tag
affects the properties of the ligand. As shown here, the biotin
label significantly altered the properties of the Cu-ttpy moeity.
Similar results were found with the fluorescent thiazole orange
label [32], confirming that such modification of a small molecule
may alter the properties of the G4-binding core. One neverthe-
less may conclude that the neutral biotin tag has a lower detri-
mental effect on selectivity e if any e than the positively charged
TO group.
Having validated the PDC-biotin ligand, we performed a SELEX
experiment to identify DNA sequences that bind to this derivative.
It initially came as a surprise (and for some of us, a disappoint-
ment!) that so few of the selected motifs were actually G4 prone.
Analysis of individual sequences by SPR confirmed that non G4-
forming sequences directly interact with the G4 ligand. Despite
having confirmed that PDC-biotin has a strong affinity for the
human telomeric quadruplex motif in vitro, we failed to isolate
clones with similar sequences. This illustrates that the SELEX
strategy we use is by no mean exhaustive. On the one hand one
may propose that there is a counterselection against G4-prone
sequences during each amplification step even though four-
stranded aptamers have been identified against a number of
electrostatic as well as
pep interaction with the available sites of
the target molecule. Very generally this involves a conformational
change of the aptamer upon interaction with its ligand. Indeed
some aptamers are wrapped around the target molecule, that is no
longer accessible in the case of small molecules like tobramycin.
From this view point quadruplexes are probably not the best
ligands for small molecules as they exhibit a limited flexibility.
Imperfect hairpins with loops and bulges allow for better induced
fit and might actually surpass four-stranded structures as far as
binding is concerned. Therefore, even if a G4 ligand appears
specific when challenged by specific double-stranded or single-
stranded sequences (ds26 in the FRET melting assay), this does
not necessarily mean that these ligands only bind to G-quad-
ruplexes! If limited strong sites are available in perfect double-
stranded DNA it is likely that a large number of competing sites
are offered by tertiary RNA structures. Finally, one may argue that
the method used to elute candidates from the target (heating at
75 ꢀC in water) could not be harsh enough to dissociate apta-
mereligand complexes of highest stability. Nevertheless, heating
at a higher temperature for elution favors the non-specific back-
ground (residual candidates directed against the streptavidin and
the magnetic beads surface could be eluted). Moreover, after using
higher temperature for the elution, we have observed an elution
of the streptavidin.

1366
A. Renaud de la Faverie et al. / Biochimie 93 (2011) 1357e1367
[13] P. Ragazzon, J.B. Chaires, Use of competition dialysis in the discovery of
One way to overcome this problem is to perform the elution by
G-quadruplex selective ligands, Methods 43 (2007) 313e323.
[14] L. Lacroix, A. Séosse, J.L. Mergny, Fluorescence-based duplexequadruplex
competition test to screen for telomerase RNA quadruplex ligands, Nucleic
Acids Research 39 (2011) e21.
competition with the free target (unbiotinylated target). The limi-
tation to use this protocol is the possibility that the free target at
high concentration would inhibit the PCR amplification of the
candidates. Other selection experiments will be performed using
different elution schemes in order to compare the population of
sequences recovered.
Site-specific substitutions will be necessary to define precisely
the binding pocket and confirm that the bulges and loops pre-
dicted by Mfold are relevant for recognition (one should note
that Mfold does not currently support quadruplex structures and
misses all G4-prone motifs). Finally it will be interesting to i)
investigate whether these DNA motifs recognize other quad-
ruplex ligands and ii) determine the outcome of a selection with
a RNA library.
[15] S. Muller, S. Kumari, R. Rodriguez, S. Balasubramanian, Small-molecule-
mediated G-quadruplex isolation from human cells, Nature Chemistry
2
(2010) 1095e1098.
[16] H. Bertrand, D. Monchaud, A. De Cian, R. Guillot, J.L. Mergny, M.P. Teulade-
Fichou, The importance of metal geometry in the recognition of G-quadruplex
DNA by metal-terpyridine complexes, Organic & Biomolecular Chemistry 5
(2007) 2555e2559.
[17] J.L. Mergny, J. Li, L. Lacroix, S. Amrane, J.B. Chaires, Thermal difference spectra:
a specific signature for nucleic acid structures, Nucleic Acids Research 33
(2005) e138.
[18] J.L. Mergny, L. Lacroix, Analysis of thermal melting curves, Oligonucleotides 13
(2003) 515e537.
[19] J.L. Mergny, L. Lacroix, UV melting of G-quadruplexes, Current Protocols in
Nucleic Acid Chemistry 17 (2009) 17.11.11e17.11.15.
[20] A. Guédin, A. De Cian, J. Gros, L. Lacroix, J.L. Mergny, Sequence effects in single
base loops for quadruplexes, Biochimie 90 (2008) 686e696.
[21] J.L. Mergny, J.C. Maurizot, Fluorescence resonance energy transfer as a probe
Acknowledgments
for G-quartet formation by
a telomeric repeat, Chembiochem 2 (2001)
124e132.
[22] A. De Cian, L. Guittat, M. Kaiser, B. Saccà, S. Amrane, A. Bourdoncle,
P. Alberti, M.P. Teulade-Fichou, L. Lacroix, J.L. Mergny, Fluorescence-based
melting assays for studying quadruplex ligands, Methods 42 (2007)
183e195.
[23] E. Dausse, C. Cazenave, B. Rayner, J.J. Toulmé, In vitro selection procedures for
identifying DNA and RNA aptamers targeted to nucleic acids and proteins,
Methods in Molecular Biology 288 (2005) 391e410.
[24] K.P. Williams, D.P. Bartel, PCR product with strands of unequal length, Nucleic
Acids Research 23 (1995) 4220e4221.
[25] Di Primo, C., Dausse, E., Toulmé, J.J. Surface plasmon resonance investigation
of RNA aptamereRNA ligand interactions, in: J. Goodchild (Ed.), Therapeutic
Oligonucleotides, Methods in Molecular Biology, 764. doi:10.1007/978-1-
61779-188-8_19.
We thank members of the Mergny, Toulmé and Teulade-Fichou
laboratories for helpful discussions. This work was supported by an
ANR grant (G4-TOOLBOX, ANR-Blan-09-355) to J.L.M., M.P.T.F. and
J.J.T. A.R.F. is the recipient of an Université Bordeaux Segalen e
MESR PhD studentship. J.L.M. acknowledges support from the
Région Aquitaine, the Fondation pour la Recherche Médicale
(F.R.M.), and INCa and wishes to dedicate this article to the memory
of G. Cherruault e a friend reaped by a Katana we still cannot
always parry.
[26] D.G. Myszka, Improving biosensor analysis, Journal of Molecular Recognition
12 (1999) 279e284.
Appendix. Supplementary data
[27] C. Hounsou, L. Guittat, D. Monchaud, M. Jourdan, N. Saettel, J.L. Mergny,
M.P. Teulade-Fichou, G-quadruplex recognition by quinacridines: a SAR, NMR
and biological study, ChemMedChem 2 (2007) 655e666.
Supplementary data associated with this article can be found, in
the online version, at doi:10.1016/j.biochi.2011.05.022.
[28] 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, Preferential binding of a G-quadruplex ligand to human chro-
mosome ends, Nucleic Acids Research 33 (2005) 4182e4190.
[29] L. Guittat, P. Alberti, F. Rosu, S. Van Miert, E. Thetiot, L. Pieters, V. Gabelica,
E. De Pauw, A. Ottaviani, J.F. Riou, J.-L. Mergny, Interactions of cryptolepine
and neocryptolepine with unusual DNA structures, Biochimie 85 (2003)
535e547.
References
[1] S. Neidle, S. Balasubramanian, Quadruplex Nucleic Acids. RSC Biomolecular
Sciences, Cambridge, 2006, pp. 301.
[2] D. Sen, W. Gilbert, Formation of parallel four-stranded complexes by guanine-
rich motifs in DNA and its applications for meiosis, Nature 334 (1988)
364e366.
[3] L.A. Cahoon, H.S. Seifert, An alternative DNA structure is necessary for
pilin antigenic variation in Neisseria gonorrhoeae, Science 325 (2009)
764e767.
[4] P. Sarkies, C. Reams, L.J. Simpson, J.E. Sale, Epigenetic instability due to
defective replication of structured DNA, Molecular Cell 40 (2010)
703e713.
[30] L. Guittat, A. DeCian, F. Rosu, V. Gabelica, E. DePauw, E. Delfourne, J.L. Mergny,
Ascididemin and meridine stabilise G-quadruplexes and inhibit telomerase
in vitro, Biochimica et Biophysica Acta
e General Subjects 1724 (2005)
375e384.
[31] F. Koeppel, J.F. Riou, A. Laoui, P. Mailliet, P.B. Arimondo, D. Labit, O. Petigenet,
C. Hélène, J.L. Mergny, Ethidium derivatives bind to G-quartets, inhibit telo-
merase and act as fluorescent probes for quadruplexes, Nucleic Acids
Research 29 (2001) 1087e1096.
[32] P. Yang, A. De Cian, M.P. Teulade-Fichou, J.L. Mergny, D. Monchaud, Engi-
neering bisquinoliniumethiazole orange conjugates for fluorescent sensing of
G-quadruplex DNA, Angewandte Chemie International Edition 48 (2009)
2188e2191.
[33] R. Chenna, H. Sugawara, T. Koike, R. Lopez, T.J. Gibson, D.G. Higgins,
J.D. Thompson, Multiple sequence alignment with the Clustal series of
programs, Nucleic Acids Research 31 (2003) 3497e3500.
[34] M. Zuker, Mfold web server for nucleic acid folding and hybridization
prediction, Nucleic Acids Research 31 (2003) 3406e3415.
[35] B. Gatto, M. Palumbo, C. Sissi, Nucleic acid aptamers based on the G-quad-
ruplex structure: therapeutic and diagnostic potential, Current Medicinal
Chemistry 16 (2009) 1248e1265.
[36] P.J. Bates, D.A. Laber, D.M. Miller, S.D. Thomas, J.O. Trent, Discovery and
development of the G-rich oligonucleotide AS1411 as a novel treatment for
cancer, Experimental and Molecular Pathology 86 (2009) 151e164.
[37] D. Levesque, J.D. Beaudoin, S. Roy, J.P. Perreault, In vitro selection and char-
acterization of RNA aptamers binding thyroxine hormone, Biochemical Jour-
nal 403 (2007) 129e138.
[38] A.T. Phan, V. Kuryavyi, J.B. Ma, A. Faure, M.L. Andreola, D.J. Patel, An inter-
locked dimeric parallel-stranded DNA quadruplex: a potent inhibitor of HIV-1
integrase, Proceedings of the National Academy of Sciences of the United
States of America 102 (2005) 634e639.
[5] M.J. Law, K.M. Lower, H.P. Voon, J.R. Hughes, D. Garrick, V. Viprakasit,
M. Mitson, M. De Gobbi, M. Marra, A. Morris, A. Abbott, S.P. Wilder, S. Taylor,
G.M. Santos, J. Cross, H. Ayyub, S. Jones, J. Ragoussis, D. Rhodes, I. Dunham,
D.R. Higgs, R.J. Gibbons, ATR-X syndrome protein targets tandem repeats and
influences allele-specific expression in a size-dependent manner, Cell 143
(2010) 367e378.
[6] C. Ribeyre, J. Lopes, J.-B. Boulé, A. Piazza, A. Guédin, V.A. Zakian, J.-L. Mergny,
A. Nicolas, The yeast Pif1 helicase prevents genomic instability caused by G-
quadruplex-forming CEB1 sequences in vivo, PLoS Genetics
e1000475.
5 (2009)
[7] A.K. Todd, M. Johnston, S. Neidle, Highly prevalent putative quadruplex
sequence motifs in human DNA, Nucleic Acids Research 33 (2005) 2901e2907.
[8] J.L. Huppert, S. Balasubramanian, Prevalence of quadruplexes in the human
genome, Nucleic Acids Research 33 (2005) 2908e2916.
[9] A. De Cian, L. Lacroix, C. Douarre, N. Temime-Smaali, C. Trentesaux, J.F. Riou,
J.L. Mergny, Targeting telomeres and telomerase, Biochimie 90 (2008)
131e155.
[10] D. Sun, B. Thompson, B.E. Cathers, M. Salazar, S.M. Kerwin, J.O. Trent,
T.C. Jenkins, S. Neidle, L.H. Hurley, Inhibition of human telomerase by
a G-quadruplex-interactive compound, Journal of Medicinal Chemistry 40
(1997) 2113e2116.
[11] D. Monchaud, M.P. Teulade-Fichou, A hitchhiker’s guide to G-quadruplex
ligands, Organic & Biomolecular Chemistry 6 (2008) 627e636.
[12] I.M. Dixon, F. Lopez, A.M. Tejera, J.P. Esteve, M.A. Blasco, G. Pratviel,
[39] L. Petraccone, G. Barone, C. Giancola, Quadruplex-forming oligonucleotides as
tools in anticancer therapy and aptamers design: energetic aspects, Current
Medicinal Chemistry e Anti-cancer Agents 5 (2005) 463e475.
B. Meunier,
A G-quadruplex ligand with 10000-fold selectivity over
[40] D.J.F. Chinnapen, D. Sen, Hemin-stimulated docking of cytochrome
c to
duplex DNA, Journal of the American Chemical Society 129 (2007)
1502e1503.
a hemin-DNA aptamer complex, Biochemistry 41 (2002) 5202e5212.

A. Renaud de la Faverie et al. / Biochimie 93 (2011) 1357e1367
1367
[41] D.M. Tasset, M.F. Kubik, W. Steiner, Oligonucleotide inhibitors of human
thrombin that bind distinct epitopes, Journal of Molecular Biology 272 (1997)
688e698.
[42] D.E. Huizenga, J.W. Szostak, A DNA aptamer that binds adenosine and ATP,
Biochemistry 34 (1995) 656e665.
[43] P. Schultze, R.F. Macaya, J. Feigon, 3-Dimensional solution structure of the
thrombin-binding DNA aptamer d(GGTTGGTGTGGTTGG), Journal of Molecular
Biology 235 (1994) 1532e1547.
[44] R.F. Macaya, P. Schultze, F.W. Smith, J.A. Roe, J. Feigon, Thrombin-binding DNA
aptamer forms a unimolecular quadruplex structure in solution, Proceedings
of the National Academy of Sciences of the United States of America 90 (1993)
3745e3749.
[45] E. Largy, F. Hamon, M.-P. Teulade-Fichou, Development of a high-throughput
G4-FID assay for screening and evaluation of small molecules binding quad-
ruplex nucleic acid structures, Analytical and Bioanalytical Chemistry (2011).
doi:10.1007/s00216-00011-05018-z.