Angewandte
Chemie
[
18–21,26–29]
solutions.
In addition, the debate is fuelled by a
studies (carried out with spin labeled quadruplexes) indicate
considerable number of contradictory results: CD spectra of
the sequence (GGGTTA) GGG and A(GGGTTA) GGG
that the resulting distances of the minimized structures are
not representative for the thermodynamic ensemble of the
highly flexible loops. Therefore the distances used for assign-
ment were derived from the 5-methyl carbon atoms of the
marked positions in the nucleotide (positions 5 and 11) taken
from the protein data bank (PDB). Intermolecular quadru-
plex formation was excluded by measuring distances in single-
labeled sequences. These single-labeled sequences were also
used to assess the local mobility of the spin labels, which again
indicated the high flexibility of the trinucleotide loops (see
Supporting Information).
3
3
are interpreted as exclusively the antiparallel basket form
+
+
[24]
with both Na and K ions. Cross-linking experiments using
a platinum complexes indicate that for the sequences A-
(
GGGTTA) GGG and (TTAGGG) the antiparallel basket
3 4
[
28]
form dominates for both ions.
biophysical study of (TTAGGG) in K suggests only the
3
On the other hand a
+
4
[
33]
+1 hybrid species as dominant form. Taken together it
remains unclear which structure really contributes to the
observed mixture since an unequivocal assignment by CD
spectra alone is difficult and other methods are at least to
some extent inconclusive. Herein we introduce spin-label
quadruplex EPR for investigating distance in the nanometer
range and demonstrate the presence of a 1:1 mixture of the
+
It is generally accepted that in Na ion containing
solutions the antiparallel basket form (Figure 1b) is the only
[
10]
conformation.
In accordance with the distance (2.9 nm)
between the spin labels as determined from the PDB
coordinates, we found exclusively a single species with a
distance of (3.0 Æ 0.1) nm by DEER, agreeing with the
expectation of solely the basket conformation present (Fig-
+
parallel propeller and antiparallel basket structures in K ion
solution.
Several approaches were used to incorporate spin labels
[
34–37]
+
into DNA or RNA sequences.
Besides mobility studies
ure 2a and b). In contrast, in K ion solution we found a
on DNA folding by conventional continuous wave (cw)-
[
38]
EPR, pulsed DEER experiments were utilized to measure
[
14,34,37,39,40]
distances in DNA model systems.
We have synthe-
[
41–43]
sized oligonucleotides spin labeled with TEMPA.
We
chose to incorporate the spin labels at nucleotides 5 and 11 as
depicted in Figure 1a because measurements of the distance
between these positions allows the different quadruplexes to
be discriminated (for a comparison of predicted distances in
all iterations of deoxythymidine positions see Supporting
Information). CD spectroscopy was used to compare spin-
+
+
labeled with unmodified sequences both in Na and K ion
solutions. Importantly, identical spectra were obtained for the
labeled and unlabeled oligonucleotides, showing that the
incorporation of the TEMPA groups did not change the
quadruplex topologies (see Supporting Information Fig-
ure S1).
For DEER-experiments, the DNA samples (50 mm oligo-
+
+
nucleotide concentration) were annealed in Na or K
Figure 2. DEER spectroscopy data. Dipolar evolution after background
solution. After shock-freezing in liquid nitrogen to trap the
annealed conformations they were measured at 45 K (for
details see Supporting Information). Data were analyzed
correction for the A(GGGTTA) GGG sequence in the presence of
3
+ +
Na (a) and K (c) solutions, respectively, with fitting curves assuming
one or two Gaussians, respectively. Corresponding distance distribu-
+
[44]
tions reveal one conformer in Na (b) and two distinct species in
using DeerAnalysis2008. Model-free analysis revealed that
the spin-label distance distributions can be characterized well
by one or two Gaussian curves. Tikhonov regularization (see
Supporting Information, Figure S4) as well as considering the
root-mean-square (RMS) deviation of the Gaussian models
enabled us to unambiguously distinguish whether one or two
conformations are present. The presence of two conforma-
+
K (d) containing solutions.
mixture of two species. Fitting with only one broader
Gaussian results in a significantly worse agreement with the
experimental data, whilst using two Gaussians for Figure 2b
results in a single conformation as well. The two forms of
+
+
tions in K ion also agrees with the findings frequently
quadruplexes in K are present in relative intensities of 55
[
8,18–32]
reported.
and 45% (Æ 6%). The first species is characterized by a
distance of (1.8 Æ 0.2) nm corresponding to the expected
value in the parallel propeller form (Figure 1d) whereas the
second species has a distance of (3.0 Æ 0.1) nm identical to the
The width of the distribution (half width at half maximum
1 nm) is caused by the flexibility of the spin-labeled
ꢀ
trinucleotide loops. The resulting distances discussed below
correspond to the center of the Gaussian curves. The nitro-
xide-labeled structures were modeled by energy-minimizing
in explicit solvents using specialized force-field parameters
for the label based on the AMBER force field. The distances
derived from this modeling procedure are similar to the
distances shown in Figure 1. However, molecular dynamics
+
results obtained in Na ion solution (Figure 2c and d). Our
measurements thus agree with the presence of the propeller
and the basket quadruplex. To exclude a mistaken assignment
of the distance measurements with the respective topologies,
we synthesized an additional control oligonucleotide. The
sequence TT(GGGTTA) GGGA slightly deviates from the
3
Angew. Chem. Int. Ed. 2009, 48, 9728 –9730
ꢀ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim