Direct quantification of loop interaction and π-π Stacking for G-quadruplex stability at the submolecular level

Article abstract of DOI:10.1021/ja503585h

The well-demonstrated biological functions of DNA G-quadruplex inside cells call for small molecules that can modulate these activities by interacting with G-quadruplexes. However, the paucity of the understanding of the G-quadruplex stability contributed

Full text of DOI:10.1021/ja503585h

Article
pubs.acs.org/JACS
Direct Quantification of Loop Interaction and π−π Stacking for
G‑Quadruplex Stability at the Submolecular Level
Chiran Ghimire,† Soyoung Park,‡,§ Keisuke Iida,∥ Philip Yangyuoru,† Haruka Otomo,‡,§ Zhongbo Yu,†
Kazuo Nagasawa,*,∥ Hiroshi Sugiyama,*,‡,§ and Hanbin Mao*,†
†Department of Chemistry and Biochemistry and School of Biomedical Sciences, Kent State University, Kent, Ohio 44242, United
States
‡Department of Chemistry, Graduate School of Science, Kyoto University, Kitashirakawa-oiwakecho, Sakyo-ku, Kyoto 606-8502,
Japan
§Institute for Integrated Cell Material Sciences (iCeMS), Kyoto University, Yoshida-ushinomiyacho, Sakyo-ku, Kyoto 606-8501,
Japan
∥Department of Biotechnology and Life Science Faculty of Technology, Tokyo University of Agriculture and Technology (TUAT),
Koganei, Tokyo 184-8588, Japan
S
* Supporting Information
ABSTRACT: The well-demonstrated biological functions of DNA G-quadruplex inside
cells call for small molecules that can modulate these activities by interacting with G-
quadruplexes. However, the paucity of the understanding of the G-quadruplex stability
contributed from submolecular elements, such as loops and tetraguanine (G) planes (or G-
quartets), has hindered the development of small-molecule binders. Assisted by click
chemistry, herein, we attached pulling handles via two modified guanines in each of the
three G-quartets in human telomeric G-quadruplex. Mechanical unfolding using these
handles revealed that the loop interaction contributed more to the G-quadruplex stability
than the stacking of G-quartets. This result was further confirmed by the binding of
stacking ligands, such as telomestatin derivatives, which led to similar mechanical stability
for all three G-quartets by significant reduction of loop interactions for the top and bottom
G-quartets. The direct comparison of loop interaction and G-quartet stacking in G-
quadruplex provides unprecedented insights for the design of more efficient G-quadruplex-
interacting molecules. Compared to traditional experiments, in which mutations are employed to elucidate the roles of specific
residues in a biological molecule, our submolecular dissection offers a complementary approach to evaluate individual domains
inside a molecule with fewer disturbances to the native structure.
guanine residues through Hoogsteen base pairs (Figure 1a,
inset). It has been generally accepted that the stacking between
G-quartets provides stability to a G-quadruplex, whereas the
size and the orientation of the loops determine the type and the
flexibility of the structure.7 The importance of the G-quartet
stacking to the structural stability has been supported by the
binding geometry of many ligands that stack onto G-quartet
planes.8 Recent findings, however, have suggested that the loop
interaction is also important for the structural stability.9,10 Due
to the lack of appropriate techniques, however, the relative
contribution of G-quartet stacking and loop interaction to the
G-quadruplex stability has not been resolved unambiguously.
We reasoned that by submolecular dissection of a G-quadruplex
in a quartet-by-quartet fashion, we may directly distinguish the
contribution of the G-quartet stacking and loop interaction to
the stability of G-quadruplex.
INTRODUCTION
■
With firm demonstration of in vivo formation of DNA G-
quadruplexes,1 research focus in the field has now shifted to
their biological functions. It has been shown that G-quadruplex
formed in the single-stranded telomeric region can inhibit
telomerase, an enzyme overexpressed in 80% of cancer cells to
elongate telomere overhang, which leads to cell immortality.2,3
The inhibition of telomerase can be tuned by G-quadruplex-
binding ligands, such as telomestatin.4 Due to potential
pharmaceutical applications, this finding has attracted an
intensive research interest. As a motor protein, telomerase
can be stopped by a roadblock whose mechanical stability is
higher than the stall force of the enzyme.5 Elucidation of the
mechanical property of the G-quadruplex/ligand complex is
therefore instrumental to design more effective molecules that
can strengthen G-quadruplex as a mechanical blocker not only
to telomerase but also to other motor proteins, such as RNA
and DNA polymerases.
G-quadruplex has a rather unique topology.6 It is composed
of a stack of G-quartet planes, each of which is joined by four
Received: April 10, 2014
Published: October 8, 2014
© 2014 American Chemical Society
15537
dx.doi.org/10.1021/ja503585h | J. Am. Chem. Soc. 2014, 136, 15537−15544

Journal of the American Chemical Society
Article
Figure 1. Dissection of DNA structures at the submolecular level. (a) Experimental setup for the quartet-by-quartet dissection of G-quadruplex. A G-
quartet structure and a click chemistry linkage (triazole) are shown in the left and right insets, respectively. (b) Human telomeric sequences with
alkyne-modified guanine residues (underlined) in the top, middle, or bottom G-quartet of an expected hybrid-1 G-quadruplex. (c) Synthesis of
alkyne-modified telomeric sequences. An alkyne-modified guanine residue for solid-phase synthesis (II) was prepared from compound I. After
incorporation of II into oligodeoxynucleotides (III) via solid-phase synthesis, a dsDNA handle terminated with an azide group (IV) was attached to
III through click chemistry (see Materials and Methods for details).
Previously, we have probed the mechanical property of DNA
G-quadruplexes with and without ligands using optical
tweezers.11,12 Assisted by click chemistry,13 we have been able
to attach pulling handles to one or two loops, from which the
unfolding and refolding of telomeric G-quadruplex can be
investigated along specific geometries. To dissect a G-
quadruplex quartet-by-quartet, pulling handles are designed to
attach to the 3′-phosphate of the particular pair of guanine
residues in the top, middle, or bottom G-quartet separately
(Figure 1c). The mechanical unfolding experiments via these
pulling handles revealed that unfolding forces of the overall G-
quadruplex are smaller than those through loops, suggesting
that loops contribute more to the structural stability than the
G-quartets. Among the three G-quartets in a telomeric G-
quadruplex, the middle quartet presents the weakest mechanical
stability. This is consistent with the least interaction between
the loops and the middle G-quartet, confirming that the quartet
stacking is weaker than the loop interaction. Binding of
telomestatin derivatives corroborates this finding. When a
telomestatin dimer binds to G-quadruplex by stacking on both
the top and bottom G-quartets, it leads to similar mechanical
stabilities for all three G-quartets, probably due to the
prevention of loop−quartet interactions. Whereas a telomesta-
tin monomer and dimer demonstrate different mechanical
stabilities for the bottom G-quartet, they share the same
stability for the top G-quartet. This result is consistent with the
NMR structure in which the monomer ligand prefers to stack
on the top G-quartet. The submolecular dissection method,
therefore, can be used conveniently to probe the ligand binding
sites of biological molecules.
15538
dx.doi.org/10.1021/ja503585h | J. Am. Chem. Soc. 2014, 136, 15537−15544

Journal of the American Chemical Society
Article
digoxigenin at the 5′-ends. Two 2028 bp dsDNA handles were mixed
with alkyne-modified G4-G22, G5-G23, or G6-G24 in equimolar ratio
(0.5 μM) in 10 μL aqueous solution, followed by adding freshly
prepared 3 μL of a solution that contains DMS/t-ButOH 3:1 v/v with
33 mM CuBr and 67 mM TBTA (tris[(1-benzyl-1H-1,2,3-triazol-4-
yl)methyl]amine, Sigma).14,15 The reaction mixture was incubated
overnight in the absence of light. The CuBr was removed by addition
of equimolar amounts of EDTA, followed by ethanol precipitation. As
a result of cyclic addition via click chemistry,13 the ssDNA sequence
was sandwiched between the two dsDNA handles.
Synthesis of Telomestatin Derivatives. Synthesis and structural
characterizations of the derivatives were performed according to the
precedent report.16 All derivatives were prepared as 10 mM stock
solutions in dimethyl sulfoxide (DMSO). Further dilutions to working
concentrations were performed using 10 mM Tris buffer (pH 7.4)
with 100 mM KCl.
Surface Plasmon Resonance (SPR) Binding Assay. The 5′-
biotin-AGGG(TTAGGG)3-3′ telomere DNA sequence used in the
SPR experiments was purchased from Sigma Genosis (HPLC grade).
The SPR binding experiments were performed with Biacore T-200
(GE Healthcare). The binding experiments were carried out according
to the report by Hurley and co-workers as follows.17−19 The biotin-
labeled DNA was bound to a streptavidin-coated sensor chip (Series S
Sensor Chip SA). One flow cell was used to immobilize the DNA [400
resonance units (RU)], while a second cell was left blank as a control.
The binding experiments were performed in the sterile, filtered, and
degassed HEPES buffer [0.01 M HEPES (pH 7.4), 3 mM EDTA, and
0.05% surfactant P20 with 100 mM KCl]. For binding experiments, an
experimental solution of L2H2-6OTD monomer and its dimer was
prepared in the HEPES buffer by serial dilutions from stock solution
(10 mM). The experimental solutions at concentrations from 7.8 to
62.5 nM were injected through the DNA and blank flow cells at a rate
of 100 μL/min at 25 °C until a constant steady-state response was
obtained (300 s, Figure S1, Supporting Information). Compound
solution flow was then replaced by buffer flow, resulting in dissociation
of the complex (350 s). To remove any remaining bound compound
after the dissociation phase of the sensorgram, a low-pH glycine
regeneration buffer was used (10 mM glycine at pH 2). The reference
response from the blank cell was subtracted from the response of the
sample flow cell to give an instrument response (RU) that is directly
proportional to the amount of L2H2-6OTD monomer and its dimer
bound to the immobilized DNA. Reference-subtracted sensorgrams for
each concentration were analyzed using the kinetics 1:1 binding
program using BIAevaluation software. The dissociation constant (KD)
between G4 ligands and telomeric G4 is defined according to KD = kd/
ka, where ka and kd represent the kinetic constants for association and
dissociation, respectively (see Table S1 of the Supporting Information
for values).
Circular Dichroism (CD). Five micromolar DNA samples were
dissolved in 10 mM Tris buffer (pH 7.4) with 100 mM KCl with (100
nM) or without telomestatin derivatives for CD spectra. Each
spectrum was collected with a JASCO-810 spectropolarimeter (Easton,
MD) using a quartz cuvette with a 1 mm optical path length at room
temperature. The average spectra of three scans over the wavelength
range of 220−320 nm were acquired with a scan rate of 100 nm/min.
The background signals from the buffer were subtracted from the
spectra of DNA samples and smoothed using the Savitzky−Golay
function.
Single Molecule Force Ramp Assay. Single molecule force
ramping was performed with home-built dual-trap optical tweezers20 at
23 °C in 10 mM Tris buffer containing 100 mM KCl at pH 7.4, with
and without telomestatin derivatives. Two kinds of polystyrene beads
coated either with streptavidin or digoxigenin antibody were separately
trapped by laser tweezers in a microfluidic chamber. The antibody-
coated bead was incubated with the DNA construct prior to the laser
trapping. One of the traps was fixed while the other was controlled by
a steerable mirror. The DNA was tethered between the two
polystyrene beads through affinity interactions between digoxigenin
antibody and biotin−streptavidin complexes. When the two beads
were moved apart, the tension in the DNA tether increased, which was
MATERIALS AND METHODS
■
Materials. Enzymes and plasmids were purchased from New
England Bioabs (NEB). Nucleotides were purchased from Integrated
DNA Technology (IDT). Polystyrene beads coated with streptavidin
or anti-digoxigenin antibody were purchased from Spherotech (Lake
Forest, IL). All other chemicals (>99% purity, unless specified) were
purchased from VWR.
Synthesis of 5′-O-Dimethoxytrityl-N2-tert-butylphenoxyace-
tyl-2′-deoxyriboguanosine 3′-O-Ethynylphosphinoamidite.
Bis(N,N-diisopropylamino)ethynylphosphine (compound II in Figure
1c) was prepared by Grignard reaction using bis(N,N-
diisopropylamino)chlorophosphine and ethynylmagnesium bromide
as reported previously.14 The solution of 5′-DMT-N2-tert-butylphe-
noxyacetyl-2′-deoxyguanosine (0.82 mmol, 1.0 equiv, compound I in
Figure 1c) and 1H-tetrazole was prepared in anhydrous dichloro-
methane (10 mL) under argon. Bis(N,N-diisopropylamino)-
ethynylphosphine (1.64 mmol, 2.0 equiv) in anhydrous dichloro-
methane (20 mL) was placed in a 100 mL two-neck round-bottom
flask with a 50 mL addition funnel. To a solution of bis(N,N-
diisopropylamino)ethynylphosphine was dropwise added 2′-deoxy-
riboguanosine and 1H-tetrazole in dichloromethane over 15 min and
the mixture stirred for 30 min at room temperature under argon. After
completion of the reaction, triethylamine was added to neutralize the
reaction mixture and the solvent was removed in a vacuum. The
residue was purified by column chromatography using a 50−100%
gradient of ethyl acetate in hexane containing 1% triethylamine to
afford an amorphous, ivory solid (341 mg, 45% yield). 31P NMR
1
4
(CDCl3): δ 97.2, 96.0. H NMR (CDCl3): δ 7.83 (d, JHH = 1.4 Hz,
1H), 7.41−7.37 (m, 4H), 7.30−7.24 (m, 7H), 7.19 (td, 3JHH = 7.1 Hz
and 4JHH = 1.4 Hz, 1H), 6.92 (d, 3JHH = 8.9 Hz, 2H), 6.79 (dd, 3JHH
=
5.8 Hz and 4JHH = 2.7 Hz, 4H), 6.32 (ddd, 2JHH = 13.6 Hz, 3JHH = 6.1
4
Hz and JHH = 1.3 Hz, 1H), 4.70−4.66 (m, 1H), 4.64−4.63 (m, 2H),
4.30−4.26 (m, 1H), 3.77 (s, 6H), 3.37−3.28 (m, 2H), 3.07 (dd, 3JHH
=
4
11.9 Hz and JHH = 1.7 Hz, 1H), 2.66−2.53 (m, 2H), 1.32 (s, 9H),
1.19−1.17 (m, 9H), 1.12 (t, 3JHH = 6.8 Hz, 3H). 13C NMR (CDCl3): δ
169.9, 158.8, 155.6, 154.4, 147.9, 146.3, 146.2, 144.6, 137.2, 135.8,
130.2, 128.3, 128.1, 127.1, 127.0, 122.3, 114.6, 113.4, 92.3, 86.8, 86.3,
86.1, 85.6, 83.9, 76.4, 67.4, 63.7, 55.4, 34.5, 31.6, 24.5. HRMS (ESI):
calcd for C51H59N6NaO8P [M + Na]+ 937.4030, found 937.4024.
Oligodeoxynucleotide (ODN) Synthesis. ODNs (compound III
in Figure 1c) were synthesized on solid supports using alkyne-
modified phosphinoamidite and commercially available O5′-dimethox-
ytrityl-2′-deoxyribonucleoside-O3′-phosphoramidites. Solid-phase oli-
gonucleotide synthesis was performed on an ABI DNA synthesizer
(Applied Biosystem, Foster City, CA). The alkyne-modified
phosphinoamidite was chemically synthesized as described above
and incorporated without purification into oligonucleotide through
coupling reactions for 10 min. Coupling yields with alkyne-modified
phosphinoamidites were equal to the ones obtained with standard
phosphoramidite building blocks. Cleavage from the solid support was
performed with 50% of anhydrous ethylenediamine in toluene at room
temperature for 2 h. The cleavage mixture was discarded and the solid-
support was dried using a SpeedVac. The crude oligonucleotides were
obtained by washing the support with 5% acetonitrile in water (1 mL)
and purified by RP-HPLC using a linear gradient of 0 to 100%
acetonitrile over 40 min at a flow rate 1.5 mL/min (50 mM TEAA
solution was used). The trityl-on fractions were collected and dried
using a SpeedVac. The deprotection of trityl group was carried out
with an 80% acetic acid solution for 2 h. After a second purification by
RP-HPLC, alkyne-modified oligonucleotide was collected by freeze-
drying. DNA concentrations were determined by using the Nano drop
ND-1000 (Nanodrop Technologies, Wilmington, DE).
Synthesis of DNA Constructs for Mechanical Unfolding
Experiments. To mechanically unfold and refold structures formed in
telomeric DNA, the ethynyl-modified ssDNA fragments [the G4-G22
(top), G5-G23 (middle), and G6-G24 (bottom) constructs] prepared
above were sandwiched between the two dsDNA strands of 2028 bp
length. These two 2028 bp dsDNA handles were prepared separately
by PCR from a pBR322 plasmid using two sets of primers. One set of
primers contains azide and biotin, while the other contains azide and
15539
dx.doi.org/10.1021/ja503585h | J. Am. Chem. Soc. 2014, 136, 15537−15544

Journal of the American Chemical Society
Article
Figure 2. Mechanical unfolding of a hybrid-1 G-quadruplex. (a) Circular dichroism spectra of 5 μM wild type (WT) and the top, middle, and bottom
telomeric sequences (see Figure 1b) in 10 mM Tris (pH 7.4) buffer with 100 mM KCl. (b−d) Single-molecule force ramp experiments for the top,
middle, and bottom telomeric sequences in the same buffer. (b) Typical force−extension (F−X) curves. Scale bars in the blowup insert represent 10
nm. (c) Change in contour length (ΔL) of unfolded features. (d) Rupture force of unfolded features. (e) Root mean square difference (RMSD) of
the appropriate inter-residue distances between known PDB structures and experimental measurements.
recorded in the force−extension (F−X) curves using the Labview
program (National Instruments Corp., Austin, TX). We recorded the
force range from 0 to 60 pN at 1000 Hz with a loading rate of 5.5 pN/
s. The single molecular nature of the DNA was confirmed by the
observation of the 65 pN plateau in the F−X curve.
The F−X curves were filtered using the Savitzky−Golay function
with a time constant of 10 ms in the Matlab program (The Math
Works, Nattick, MA). The change in extension (Δx) at a given force
was calculated from the difference between stretching and relaxing
curves at that force. The change in contour length (ΔL) was calculated
from Δx using the wormlike chain model (WLC).21
Stability. The submolecular dissection of DNA G-quadruplex
structure was carried out by laser tweezers, with which the
mechanical unfolding of a biomacromolecule was carried out
along specific trajectories defined by two residues that serve as
pulling handles15 (Figure 1a). To clearly differentiate different
G-quartet layers, handle residues must be selected from the
known G-quadruplex conformations. The human telomeric G-
quadruplex structure can form a variety of different
conformations dependent on solvent conditions or flanking
sequences.7 In our previous investigations, we have determined
that hybrid-1 G-quadruplex is the most likely conformation
formed in the same sequence used here under single-molecular
conditions.22 Assuming this conformation, we chose G4 and
G22 to attach the pulling handles for the top G-quartet, G5 and
G23 for the middle G-quartet, and G6 and G24 for the bottom
G-quartet (Figure 1b). We modified these G-residues with
terminal alkynes through phosphate groups,14 which allow the
attachment of double-stranded DNA as pulling handles via click
chemistry (see Materials and Methods and Figure 1c). Since
the distance between the modified phosphate and the 5′-end
guanine base in a specific G-quartet is shorter than that
between the same phosphate and the 3′-end guanine base in
another G-quartet (see Figure S2 and Table S2 in the
Supporting Information), the tension felt by the G-quadruplex
likely propagates through the 5′-end guanine pairs. By choosing
the specific guanine pairs discussed above, we are able to unfold
desired G-quartets in the hybrid-1 G-quadruplex.
kBT
FP
Δx
ΔL
1
2
F
S
= 1 −
+
where L is the contour length, kB is the Boltzmann constant, T is the
absolute temperature, P is the persistence length (50.8 nm),22 and S is
the stretching modulus (1243 pN).22
Wherever appropriate, three sets of experiments were performed to
obtain standard deviations, which are reported in the main text or as
error bars in figures.
Calculation of Change in the Free Energy of Unfolding. The
change in free energy of unfolding a G-quadruplex (ΔGunfold) was
calculated using Jarzynski’s equality equation under nonequilibrated
conditions23
N
⎛
⎞
⎟
⎠
w
k T
B
1
N
i
ΔGunfold = −k T ln
exp −
⎜
∑
B
⎝
i=1
where N is the number of repetitions and wi is the nonequilibrium
work done to unfold the G-quadruplex. See the literature11 for a
detailed calculation of the work.
To probe the G-quadruplex structures after the modification,
we performed CD experiments (Figure 2a). A valley at 240 nm
and two peaks at 260 and 290 nm suggested that the hybrid-1
G-quadruplex conformation25 is maintained for all three
mutants. Further confirmation of the hybrid-1 structure came
from the single-molecule structural identification method based
on the end-to-end distance measurements15,22 (see below).
The bias for ΔGunfold was estimated from the histogram of work
distribution according to the literature.24
RESULTS AND DISCUSSION
Quartet-by-Quartet Dissection Suggests That Loop
Interaction is a Predominant Factor for G-Quadruplex
■
15540
dx.doi.org/10.1021/ja503585h | J. Am. Chem. Soc. 2014, 136, 15537−15544

Journal of the American Chemical Society
Article
Figure 3. Loop interaction is stronger than the G-quartet stacking. (a) Rupture force for different unfolding geometries of human telomeric G-
quadruplex. “Top”, “middle”, or “bottom” indicates unfolding through a particular G-quartet (see Figure 1a). “L1”, “L2”, and “L3” depict T9 in the
first loop, T15 in the second loop, and T21 in the third loop, respectively, counted from the 5′-end of the G-quadruplex. The sequence of the G-
quadruplex-forming sequence is 5′-T1T2A3G5G6G7T8T9A10G11G12G13T14T15A16G17G18G19T20T21A22G23G24G25T26T27A28-3′. (b)
Schematic drawing depicts that the loop−loop interaction is stronger than the quartet stacking in a hybrid-1 telomeric G-quadruplex.
Figure 4. Effect of telomestatin derivatives on the stability of G-quadruplex. The left panel shows rupture force histograms from different G-quartets
in the presence of either telomestatin monomer (blue) or dimer (green). Rupture force histograms of free G-quadruplexes are shown in red. The
right panel depicts structures of telomestatin derivatives.
With this structural clarification, each set of the alkyne-
modified telomeric sequences was reacted with two dsDNA
handles labeled with azide at one of their ends through click
chemistry15 (see Materials and Methods). The attachment of
the DNA constructs to two optically trapped polystyrene
particles was achieved through the affinity linkage between the
digoxigenin antibody and biotin−streptavidin complexes
(Figure 1a). Using the single molecular force-ramp assay (see
Materials and Methods), we observed rupture events that
represent the unfolding of G-quadruplexes (Figure 2b). From
each unfolding event, we measured the change in contour
length (ΔL, Figure 2c) and the rupture force (Frupture, Figure
2d), which depict the size and the mechanical stability of a G-
quadruplex, respectively. The ΔL is in agreement with expected
value of ∼6.6 nm from all three unfolding geometries (PDB
code: 2HY925). On the basis of these ΔL values, we calculated
the end-to-end distance (x) between the two handle residues by
the function x = L − ΔL.11 Here L is the contour length of the
structure. After combining these distances with those between
different pairs of residues measured previously,15 we compared
the corresponding distances obtained from the known human
telomeric G-quadruplex structures. Evaluation using the RMSD
(Figure 2e) between these two sets of data allowed us to
conclude that the best matching conformation is the hybrid-1
G-quadruplex, therefore, confirming the CD results.
The Frupture histograms for the three G-quartets revealed
(Figure 2d) that the rupture forces for the top and bottom G-
quartets are similar, while that for the middle quartet is the
15541
dx.doi.org/10.1021/ja503585h | J. Am. Chem. Soc. 2014, 136, 15537−15544

Journal of the American Chemical Society
Article
smallest. This result suggested that the loop interaction is a
predominant factor for the stability of G-quadruplex. Since the
top and the bottom quartets interact more extensively with the
loops than the middle quartet, the rupture forces are higher for
the top and the bottom G-quartets. When we summarized the
rupture force for each pair of handle residues, we found that
indeed, unfolding through loop residues has a higher rupture
force than that through G-quartet residues (Figure 3a; see
Table S3 of the Supporting Information for sequences). In fact,
the highest rupture force is from the unfolding via two loop
residues, T9 in loop1 and T21 in loop3. These results strongly
supported that loop interactions have stronger mechanical
stability than the stacking between G-quartets (Figure 3b).
These loop interactions could have conceivably originated from
hydrogen bonding between loop bases26 and the stacking
interaction of the bases in the loop and the G-quartet.25,26
Dissection of Telomestatin-Bound G-Quadruplex
Confirmed That the Loop Interaction is Stronger Than
the G-Quartet Stacking. To provide further evidence that the
loop interaction is a predominant stabilization factor for human
telomeric G-quadruplex, we proceeded to investigate the effect
of ligand binding on the G-quadruplex stability. We chose
telomestatin derivatives as model ligands, which are known to
bind tightly with G-quadruplexes by stacking with terminal
quartets.4 NMR structures have revealed that a telomestatin
monomer, L2H2-6OTD (Figure 4, blue), prefers to stack on
the top G-quartet in a hybrid-1 telomeric G-quadruplex.8 On
the other hand, we designed an L2H2-6OTD dimer (Figure 4,
green) with a flexible linker that matches with the height of the
human telomeric G-quadruplex. This ligand is expected to stack
on both the top and bottom G-quartets.27,28 Due to this
stacking, we reasoned that top or bottom G-quartet(s) may be
shielded from loop interactions, which may result in similar
rupture force as that for the middle G-quartet.
telomestatin dimer (∼14 vs ∼9 kcal/mol for free G-
quadruplex).
When we used 100 nM L2H2-6OTD monomer (KD = 10.8
nM, Figure S1a and Table S1,Supporting Information), we
found that G-quadruplex assumes the expected hybrid-1
conformation from CD signatures8 (Figure S3a, Supporting
Information). Comparison of the rupture force revealed that
the top G-quartet (37
1 pN) showed similar mechanical
stability as that for the dimer binding (38 3 pN, Figure 4
top). In contrast, the bottom G-quartet had a lower rupture
force (34 2 pN) than that bound with the dimer (36.8 0.4
pN, Figure 4, bottom). This result is consistent with the fact
that the L2H2-6OTD monomer prefers to bind to the top,
instead of the bottom, G-quartet,8 which reinforces the
mechanical stability of the top G-quartet only. Similar to the
dimer binding, the rupture forces for the monomer-bound
structures are higher than for free G-quadruplexes along the
same unfolding geometry, indicating again that ligand binding
increases the overall stability of G-quadruplex.12 The increased
ΔGunfold upon monomer binding (1−2 kcal/mol; see Table 1)
confirmed this stability variation.
The overall ΔGunfold trend (Table 1) revealed that G-
quadruplex bound with telomestatin dimer is strongest in
stability, while the free G-quadruplex is weakest. Such a trend
agrees qualitatively with the binding constant measurement in
which L2H2-6OTD dimer (KD = 8.3 nM) shows stronger
binding than the monomer (KD = 10.8 nM). Interestingly,
ΔGunfold is similar for different G-quartets incubated with the
same ligand, which reflects the fact that ΔGunfold is the state
function independent of unfolding pathways.15
Quartet-by-Quartet Unfolding of G-Quadruplex In-
volves Two Energy Barriers. It is interesting that the
difference in the rupture force between the G-quadruplex/
telomestatin monomer and the G-quadruplex/telomestatin
dimer complexes is more obvious from the unfolding via the
middle G-quartet than those through terminal G-quartets
(Figure 4). Such a difference can be rationalized by an
unfolding model that involves two energy barriers (Figure 5).
The first (inner) barrier involves the unfolding of the G-
quartet that is directly subject to mechanical stress. During the
unfolding from the top or bottom G-quartet, due to the weaker
contribution of the quartet stacking with respect to the loop
interaction, the middle G-quartet is not as strong as the
terminal quartets that interact more extensively with loops. As a
result, the second (outer) barrier in these cases is dominated by
the unfolding of the distally located terminal G-quartet (Figure
5, top and bottom panels, solid curves). Given that telomestatin
dimer stacks to the two terminal quartets with similar strength,
both the first and the second energy barriers increase their
magnitudes to a similar level (Figure 5, top and bottom panels,
green solid curves). On the other hand, due to the preferential
binding of the telomestatin monomer to the top G-quartet,8
only the top quartet increases the energy barrier. Therefore,
during the unfolding from the top G-quartet, the first energy
barrier (top quartet) is higher than the second (bottom quartet;
see the blue solid curve in Figure 5, top panel), whereas this
trend is reversed during the unfolding from the bottom G-
quartet (see the blue solid curve in Figure 5, bottom panel)
when monomer is bound.
This scenario was exactly observed. To ensure the complete
stacking of the dimer, we used a saturation concentration of
100 nM (KD for the dimer is 8.3 nM; see Materials and
Methods and Figure S1 and Table S1, Supporting Informa-
tion). CD spectrum confirmed that the hybrid-1 G-quadruplex
conformation is maintained after ligand binding (Figure S3b,
Supporting Information). Comparison of the rupture forces
revealed that they increase to the same value for all three G-
quartets (38
3, 37
2, and 36.8
0.4 pN for the top,
middle, and bottom G-quartets, respectively, Figure 4),
confirming that due to the stacking of the telomestatin dimer,
the loops are shielded from interacting with both top and
bottom G-quartets. The increased rupture force is a result of
ligand binding, which stabilizes the G-quadruplex.12 The
stabilization has been corroborated by the measurement of
the change in free energy of unfolding (ΔGunfold) (Table 1; see
Materials and Methods for calculations), which showed
increased stability of G-quadruplex after binding with the
Table 1. Change in Free Energy of Unfolding Telomeric G-
Quadruplex [ΔGunfold SD (bias), kcal/mol] with and
without Telomestatin Derivatives
no ligand
monomer
dimer
During the unfolding through the middle G-quartet, the first
energy barrier corresponds to the unfolding of the middle
quartet, while the top and the bottom G-quartets represent the
second barrier (Figure 5, middle). This geometry leads to a
top
9.7 0.4 (−0.4)
9.0 0.6 (−0.5)
9.9 0.9 (1.4)
11 4 (−0.3)
13 2 (−0.3)
11 2 (−0.1)
15 2 (0.1)
14 2 (3.5)
13 2 (−0.3)
middle
bottom
15542
dx.doi.org/10.1021/ja503585h | J. Am. Chem. Soc. 2014, 136, 15537−15544

Journal of the American Chemical Society
Article
In contrast, during the unfolding from the middle G-quartet,
due to the much reduced distance between the two energy
barriers, the outer energy barrier remains predominant, even
after application of the force (Figure 5, middle panel; compare
solid and dotted curves). Since the outer energy barrier
represents both the top and bottom quartets in this geometry
(see above), it leads to well-separated unfolding forces of the
free G-quadruplex (weakest), the monomer-bound G-quad-
ruplex, and the dimer-bound G-quadruplex (strongest). Finally,
being the G-quadruplex with the longest distance (X) from the
folded state to the highest energy barrier (see above and Figure
5), the unfolding force of the middle G-quadruplex is lowest
among three G-quadruplexes due to the largest reduction of the
energy barrier by a value of FX (see above). Such a prediction is
well-validated by experimental observation in Figure 2b−d.
CONCLUSIONS
■
Using submolecular dissection of individual G-quartet planes in
human telomeric DNA G-quadruplexes, we found that loop
interaction contributes more to the G-quadruplex stability than
the G-quartet stacking. These results suggest that in the design
of ligands, molecules that interact with loops perhaps affect the
structural stability more than those that intercalate with top or
bottom G-quartet. As stacking modes of telomestatin
derivatives can be well-differentiated at the submolecular
level, we anticipate that this dissection method provides an
alternative approach to probe ligand binding sites of
biomacromolecules that are recalcitrant to conventional
structural characterization methods.
ASSOCIATED CONTENT
* Supporting Information
■
Figure 5. Proposed energetic diagram for the unfolding of G-
quadruplex from the top, middle, or bottom G-quartet.
S
Figures S1−S3, illustrating the SPR determination of ligand
binding with GQ, the G-quartet to phosphate distance, and CD
data for ligand binding with GQ; Table S1, summarizing ka, kd,
and KD values of ligand binding with GQ; Table S2, outlining
distance measurements for G-quartet to phosphate; and Table
S3, compiling sequences of different unfolding geometries. This
material is available free of charge via the Internet at http://
pubs.acs.org.
reduced distance between the two energy barriers. Because the
terminal G-quartets are stronger than the middle quartet due to
the loop interaction (see above), the second (outer) energy
barrier is higher than the first (inner). Such an energy landscape
renders the longest distance from the folded state to the highest
energy barrier for the middle G-quadruplex among three G-
quadruplexes (Figure 5). Since both top and bottom G-quartets
bound with the dimer must be overcome before the whole G-
quadruplex can be unfolded in this geometry, the second
energy barrier in the presence of the telomestatin dimer is
significantly higher than for the monomer, which binds to the
top quartet only.
AUTHOR INFORMATION
Corresponding Authors
knaga@cc.tuat.ac.jp
■
hs@kuchem.kyoto-u.ac.jp
hmao@kent.edu
When force (F) is applied through a particular G-quartet, it
reduces the free energy landscape with a magnitude of FX
(solid lines in Figure 5; the slope of each line represents force
F),5 where X is the reaction coordinate of the unfolding. During
the unfolding through the top or bottom G-quartet, due to the
relatively large distance between the two energy barriers (ΔX),
the outer energy barrier is more reduced than the inner barrier
by FΔX. This leads to a predominant inner energy barrier after
application of force (Figure 5, top and bottom panels, dotted
curves). As a result, while unfolding through the top G-quartet,
similar rupture forces exist between the telomestatin dimer and
monomer, as both species stack to strengthen the top quartet,
which is the inner barrier for this geometry. The difference in
rupture forces becomes larger during the unfolding through the
bottom G-quartet, since this quartet is the predominant inner
barrier, which is strengthened by the telomestatin dimer but
not the monomer.
Notes
The authors declare no competing financial interests.
ACKNOWLEDGMENTS
■
This research work was supported by NSF CHE-1026532 to
H.M. It was supported in part by a Grant-in-Aid for Scientific
Research (B) from JSPS (No. 23310158). K.N. is grateful for
financial support from the Mukai Science and Technology
Foundation, Tokyo, Japan, and Mochida Memorial Foundation
for Medical and Pharmaceutical Research, Tokyo, Japan. K.I. is
grateful for financial support in the form of JSPS Predoctoral
Fellowships for Young Scientists. H.S. expresses sincere thanks
for the CREST grant from the Japan Science and Technology
Corporation (JST), grants from the WPI program (iCeMS,
Kyoto University), and for the global COE program from the
Ministry of Education, Culture, Sports, Science and Technol-
15543
dx.doi.org/10.1021/ja503585h | J. Am. Chem. Soc. 2014, 136, 15537−15544

Journal of the American Chemical Society
Article
ogy (MEXT), Japan. We would also like to acknowledge Dr.
Michael Zagorski and Colin Boyle from Case Western Reserve
University for allowing us to use the CD instrument in their lab.
REFERENCES
■
(1) Biffi, G.; Tannahill, D.; McCafferty, J.; Balasubramanian, S. Nat.
Chem. 2013, 5, 182.
(2) Hurley, L. H.; Wheelhouse, R. T.; Sun, D.; Kerwin, S. M.; Salazar,
M.; Fedoroff, O. Y.; Han, F. X.; Han, H.; Izbicka, E.; Von Hoff, D. D.
Pharmacol. Ther. 2000, 85, 141.
(3) Guittat, L.; Alberti, P.; Gomez, D.; De Cian, A.; Pennarun, G.;
Lemarteleur, T.; Belmokhtar, C.; Paterski, R.; Morjani, H.; Trentesaux,
C.; Mandine, E.; Boussin, F.; Mailliet, P.; Lacroix, L.; Riou, J. F.;
Mergny, J. L. Cytotechnology 2004, 45, 75.
(4) Kim, M.-Y.; Vankayalapati, H.; Shin-ya, K.; Wierzba, K. a.;
Hurley, L. H. J. Am. Chem. Soc. 2002, 124, 2098.
(5) Koirala, D.; Yangyuoru, P. M.; Mao, H. Rev. Anal. Chem. 2013,
32, 197.
(6) Williamson, J. R. Annu. Rev. Biophys. Biomol. Struct. 1994, 23, 703.
(7) Burge, S.; Parkinson, G. N.; Hazel, P.; Todd, A. K.; Neidle, S.
Nucleic Acids Res. 2006, 34, 5402.
(8) Chung, W. J.; Heddi, B.; Tera, M.; Iida, K.; Nagasawa, K.; Phan,
A. T. J. Am. Chem. Soc. 2013, 135, 13495.
(9) Guedin, A.; Gros, J.; Alberti, P.; Mergny, J. L. Nucleic Acids Res.
́
2010, 38, 7858.
(10) Fujimoto, T.; Nakano, S.-i.; Sugimoto, N.; Miyoshi, D. J. Phys.
Chem. B 2012, 963.
(11) Yu, Z.; Schonhoft, J. D.; Dhakal, S.; Bajracharya, R.; Hegde, R.;
Basu, S.; Mao, H. J. Am. Chem. Soc. 2009, 131, 1876.
(12) Koirala, D.; Dhakal, S.; Ashbridge, B.; Sannohe, Y.; Rodriguez,
R.; Sugiyama, H.; Balasubramanian, S.; Mao, H. Nat. Chem. 2011, 3,
782.
(13) Kolb, H. C.; Finn, M. G.; Sharpless, K. B. Angew. Chem., Int. Ed.
2001, 40, 2004.
(14) Krishna, H.; Caruthers, M. H. J. Am. Chem. Soc. 2012, 134,
11618.
(15) Yu, Z.; Koirala, D.; Cui, Y.; Easterling, L. F.; Zhao, Y.; Mao, H. J.
Am. Chem. Soc. 2012, 134, 12338.
(16) Iida, K.; Majima, S.; Nakamura, T.; Seimiya, H.; Nagasawa, K.
Molecules 2013, 18, 4328.
(17) Redman, J. E. Methods 2007, 43, 302.
(18) Rezler, E. M.; Seenisamy, J.; Bashyam, S.; Kim, M.-Y.; White, E.;
Wilson, W. D.; Hurley, L. H. J. Am. Chem. Soc. 2005, 127, 9439.
(19) Wang, P.; Ren, L.; He, H.; Liang, F.; Zhou, X.; Tan, Z.
ChemBioChem 2006, 7, 1155.
(20) Mao, H.; Luchette, P. Sens. Actuators B 2008, 129, 764.
(21) Baumann, C. G.; Smith, S. B.; Bloomfield, V. A.; Bustamante, C.
Proc. Natl. Acad. Sci. U.S.A. 1997, 94, 6185.
(22) Dhakal, S.; Cui, Y.; Koirala, D.; Ghimire, C.; Kushwaha, S.; Yu,
Z.; Yangyuoru, P. M.; Mao, H. Nucleic Acids Res. 2013, 41, 3915.
(23) Jarzynski, C. Phys. Rev. Lett. 1997, 78, 2690.
(24) Palassini, M.; Ritort, F. Phys. Rev. Lett. 2011, 107, 060601.
(25) Dai, J.; Punchihewa, C.; Ambrus, A.; Chen, D.; Jones, R. A.;
Yang, D. Nucleic Acids Res. 2007, 35, 2440.
(26) Zhu, H.; Xiao, S.; Liang, H. PLoS One 2013, 8, e71380.
(27) Iida, K.; Tera, M.; Hirokawa, T.; Shin-ya, K.; Nagasawa, K.
Chem. Commun. 2009, 6481.
(28) Amemiya, Y.; Furunaga, Y.; Iida, K.; Tera, M.; Nagasawa, K.;
Ikebukuro, K.; Nakamura, C. Chem. Commun. 2011, 47, 7485.
15544
dx.doi.org/10.1021/ja503585h | J. Am. Chem. Soc. 2014, 136, 15537−15544