On the structure and dynamics of duplex GNA

Article abstract of DOI:10.1021/jo201469b

Glycol nucleic acid (GNA), with a nucleotide backbone comprising of just three carbons and the stereocenter derived from propylene glycol (1,2-propanediol), is a structural analog of nucleic acids with intriguing biophysical properties, such as formation

Full text of DOI:10.1021/jo201469b

ARTICLE
pubs.acs.org/joc
On the Structure and Dynamics of Duplex GNA
Andrew T. Johnson,† Mark K. Schlegel,‡ Eric Meggers,*,‡ Lars-Oliver Essen,*,‡ and Olaf Wiest*,†,§
†Department of Chemistry and Biochemistry, University of Notre Dame, Notre Dame, Indiana 46556-5670, United States
‡Fachbereich Chemie, Philipps-Universit€at Marburg, Hans-Meerwein-Strasse, 35032 Marburg, Germany
§School of Chemical Biology and Biotechnology, Peking University, Shenzhen Graduate School, Shenzhen 518055, China
S Supporting Information
b
ABSTRACT: Glycol nucleic acid (GNA), with a nucleotide back-
bone comprising of just three carbons and the stereocenter derived
from propylene glycol (1,2-propanediol), is a structural analog of
nucleic acids with intriguing biophysical properties, such as forma-
tion of highly stable antiparallel duplexes with high WatsonꢀCrick
base pairing fidelity. Previous crystallographic studies of double
stranded GNA (dsGNA) indicated two forms of backbone con-
formations, an elongated M-type (containing metallo-base pairs)
and the condensed N-type (containing brominated base pairs). A
herein presented new crystal structure of a GNA duplex at 1.8 Å
resolution from self-complementary 30-CTCBrUAGAG-20 GNA oligonucleotides reveals an N-type conformation with alternating
gaucheꢀanti torsions along its (O30ꢀC30ꢀC20ꢀO20) backbone. To elucidate the conformational state of dsGNA in solution,
molecular dynamic simulations over a period of 20 ns were performed with the now available repertoire of structural information.
Interestingly, dsGNA adopts conformational states in solution intermediate between experimentally observed backbone
conformations: simulated dsGNA shows the all-gauche conformation characteristic of M-type GNA with the higher helical twist
common to N-type GNA structures. The so far counterintuitive, smaller loss of entropy upon duplex formation as compared to
DNA can be traced back to the conformational flexibility inherent to dsGNA but missing in dsDNA. Besides extensive interstrand
base stacking and conformational preorganization of single strands, this flexibility contributes to the extraordinary thermal stability
of GNA.
’ INTRODUCTION
Pioneering work from the Eschenmoser laboratory on the
chemical etiology of nucleic acid structure9 resulted in the dis-
covery of nucleic acids derived from a tetrose sugar backbone that
is capable of forming stable duplexes in an antiparallel fashion.10
Such L-α-threofuranosyl oligonucleotides (TNA) demonstrated
for the first time that the “six-bonds-per-backbone rule”11 can be
violated by having just five bonds in an arrangement of vicinal
phosphodiester groups instead. Encouraged and inspired by
Eschenmoser’s TNA, we envisioned that the removal of the 40-
CH2O group of TNA would result in a much simplified acyclic
propylene glycol backbone. Indeed, such simplified glycol nucleic
acids (GNA), shown in Scheme 1, form highly stable antiparallel
duplexes in a WatsonꢀCrick fashion which surprisingly surpass
the thermal and thermodynamic stabilities of the analogous du-
plexes of DNA and RNA.12 Interestingly, the R- and S-enantio-
mers, (R)-GNA and (S)-GNA, do not cross-pair with each
other.13 Thus, GNA uniquely combines atom economy with
structural simplicity and high duplex stability. It constitutes the
simplest scaffold for a chemically stable phosphodiester-bond-
containing nucleic acid backbone. Given the combination of
structural simplicity, straightforward synthetic accessibility, and
For about the last 20 years, synthetic organic chemists have
extensively modified the structure of DNA and RNA oligonu-
cleotides to alter either their properties, e.g., for antisense tech-
nology, or to learn about the principles of genetic information
storage and retrieval. The initial model of nucleic acid structure
dating back to Watson and Crick1 emphasized the importance of
stacking the hydrophobic nucleobases in combination with specific
hydrogen bonds within the base pairs A:T and G:C. Today, we
know that the (deoxy)ribose phosphate backbone is not just a
simple connecting unit for the nucleobases but is in fact crucial
for duplex structure and stability. Many backbone replacements
have been explored but, unexpectedly, only a few have been
found to match or improve the duplex stability compared to
natural DNA or RNA. Simplified acyclic (deoxy)ribose analogs
are especially appealing from a synthetic perspective and as
potential building blocks for primitive genetic molecules during
the evolution of life on Earth.2 Previous experiments with acyclic
nucleotides in DNA revealed that adding one or more acyclic
nucleotides to oligodeoxynucleotides resulted in strong thermal
duplex destabilizations. This was attributed to lacking conforma-
tional preorganization of the more flexible acyclic nucleotides in
single strands resulting in a significant entropic penalty upon
duplex formation.3ꢀ8
Received: July 15, 2011
Published: August 12, 2011
r
2011 American Chemical Society
7964
dx.doi.org/10.1021/jo201469b J. Org. Chem. 2011, 76, 7964–7974
|

The Journal of Organic Chemistry
ARTICLE
high duplex stability of GNA duplexes, GNA comprises a promising
nucleic acid scaffold for applications in biotechnology and nano-
technology and reinforces its candidacy as one of the initial genetic
molecules formed during the origins of life on Earth.14
Information, for a comparison with DNA). Interestingly, the two
modified GNA structures display significant differences among
each other that may or may not be a consequence of the different
chemical modifications introduced into the duplex for phasing
the crystallographic data. First, the 6-mer-BrGNA duplex is con-
siderably compressed along the z-axis relative to the 8-mer-
CuGNA. Although this compression is accompanied by very
little change in the helix diameter, it causes the 6-mer-BrGNA to
adopt a significantly shorter helical pitch of 26 Å with 10 residues
per turn, in comparison to 60 Å with 16 residues per turn for the
8-mer-CuGNA helix. This shorter helical pitch results from the
much stronger twist in the 6-mer-BrGNA, which subsequently
brings the phosphate groups of opposing strands in closer
contact. Second, the conformations of the propylene glycol
backbone differ, especially the torsional angles between the
vicinal C30ꢀO and C20ꢀO bonds. In the 6-mer-BrGNA duplex
structure, the nucleotides adopt strictly alternating gauche and
Two recently determined crystal structures of (S)-GNA
homoduplexes, an 8-mer duplex containing two Cu(II)-coordi-
nated hydroxypyridone homobase pairs (8-mer-CuGNA)15 and
a brominated 6-mer duplex (6-mer-BrGNA),16 revealed that the
overall GNA double helix differs significantly from canonical A-
and B-form nucleic acids. Although it shares some similarities
with RNA, it might be best described as a helical ribbon loosely
wrapped around the helix axis (Figure 1a,b; see Table 1 for
sequences and abbreviations). Within the backbone, the propy-
lene glycol nucleotides can adopt two different conformations,
gauche (in this paper gauche refers to the g-conformation)
and anti, with respect to the torsional angles formed by
O30ꢀC30ꢀC20ꢀO20 (Figure 1d). A strikingly large tilt of the
backbone with respect to the helical axis is present which results
in extensive, zipper-like interstrand interactions and reduced in-
trastrand baseꢀbase interactions (Figures 1aꢀc and S8, Supporting
Table 1. List of Abbreviations and Sequences Used
throughout the Papera
6-mer-BrGNA
8-mer-BrGNA
16-mer-BrGNA
8-mer-GNA
30-GBrCGCGC-20
30-CTCBrUAGAG-20
30-CTCBrUAGAGCTCBrUAGAG-20
30-CGAATTCG-20
Scheme 1. Constitutions of DNA, RNA, TNA, and the S- and
R-Enantiomers of GNA
16-mer-GNA
8-mer-CuGNA
16-mer-CuGNA
8-mer-DNA
30-CGAATTCGCGAATTCG-20
30-CGHATHCG-20
30CGHATHCGCGHATHCG-20
50-CGAATTCG-30
16-mer-DNA
50-CGAATTCGCGAATTCG-30
a H = hydroxypyridone, BrC = 5-bromocytidine, BrU = 5-bromouracil.
Figure 1. Crystal structures of (S)-GNA duplexes. (a) 8-mer-CuGNA derived from 30-CGHATHCG-20 (2JJA) in the presence of Cu2+, with H =
hydroxypyridone nucleobase. (b) 6-mer-BrGNA derived from 30-GBrCGCGC-20 (2WNA), with BrC = 50-bromocytosine nucleobase. (c) 8-mer-BrGNA
derived from 30-CTCBrUAGAG-20 (2XC6, this work), with BrU = 50-bromouracil nucleobase. Gauche and anti refer to the torsional angles between the
vicinal C20ꢀO and C30ꢀO bonds. (d) Newman projections of the gauche (g-) and anti conformations adopted by the torsional angle
O30ꢀC30ꢀC20ꢀO20. (e) Newman projection of the anti conformation adopted by the torsional angle N1ꢀC10ꢀC20ꢀO20.
7965
dx.doi.org/10.1021/jo201469b |J. Org. Chem. 2011, 76, 7964–7974

The Journal of Organic Chemistry
ARTICLE
anti conformations (type N), whereas in the 8-mer-CuGNA
duplex structure the nucleotides forming conventional Watsonꢀ
Crick base pairs adopt a gauche conformation (type M)16 with
respect to the torsional angles between C20ꢀO and C30ꢀO
(Figure 1). In the latter, only the synthetic hydroxypyridone
glycol nucleotides adopt anti conformations, presumably to
compensate for the increased C10ꢀC10 distance within the
artificial metal base pair.
Table 2. Crystallographic Data and Refinement Statistics of
8-mer-BrGNA Duplex (30-CTCBrUAGAG-20, 2XC6)
data collection statistics
wavelength (Å)
resolution (Å)
space group
0.919 85
15.8ꢀ1.83
I4122
completeness (%)
redundancy
a,b
99.0 (98.3)
17.4 (17.9)
0.054 (0.571)
32.2 (8.9)
34.2
A recent report provides experimental insight into the origins
of the high thermal stability observed for GNA.17 Analysis of the
thermodynamic parameters shows that while formation of GNA
duplexes is less exothermic than its DNA counterpart, duplex
formation is entropically not as costly for GNA. This surprising
finding is counterintuitive considering that the GNA backbone is
acyclic, consisting solely of rotatable bonds and would be ex-
pected to be much more flexible than the cyclic deoxyribose
backbone. Loss of these degrees of freedom during duplex for-
mation should therefore be entropically less favorable for GNA
than for DNA, yet the opposite is observed. This could be due to
the fact that single-stranded GNA (ssGNA) is less flexible than
expected from its molecular structure, that GNA duplexes exhibit
higher flexibilities than their DNA counterpart, or a combination
of both. In the crystal structure of GNA, in both its duplex as well
as single nucleotide form,14 a strong preference is observed for
the backbone to adopt an anti conformation with respect to the
nucleobase (N1ꢀC10ꢀC20ꢀO20) (Figure 1e), while the vicinal
CꢀO bonds (O30ꢀC30ꢀC20ꢀO20) in single nucleotide crystal
structures16 favor a gauche conformation (Figure 1d). This,
along with the presence of strong Cotton effects indicating the
loss of helical structure in ssGNA upon heating, point toward
preorganization of ssGNA prior to annealing as the main factor in
the increased entropic stability in relation to DNA.
Rmerge
I/σ(I)a,c
Bwilson (Å2)
refinement statistics
resolution (Å)
8.00ꢀ1.83
0.234/0.271
26.6
d
R-factor/Rfree
average B-factor (Å2)
water molecules
rmsd bonds (Å)
rmsd angles (Å)
20
0.011
1.715
a Values in parentheses correspond to the highest resolution shell.
b Rmerge = ((∑∑|Ij(h) ꢀ ÆI(h)æ|/(∑∑Ij(h)). c As calculated with the
program SCALA. d R = ∑||Fo| ꢀ k|Fc||/∑|Fo|, with k as a scaling factor;
R
free calculated with test set (7.1% of all data).
from regular B-DNA to investigate backbone conformations of
GNA and to rationalize the counterintuitive thermodynamic para-
meters that have been observed for GNA.
’ RESULTS AND DISCUSSION
Molecular dynamics (MD) simulations of nucleic acids have
significantly contributed to the understanding of structure and
dynamics of both natural18ꢀ22 and unnatural nucleic acids.23ꢀ31
MD simulation can complement crystallography in the study of
nucleic acids in a variety of ways. One obvious advantage is the
ability to study species that do not lend themselves easily to
crystallization. Second, MD simulations allow the investigation
of the dynamics of a structure in the nanosecond time regime,
whereas X-ray crystallography gives a snapshot for an otherwise
time-averaged structure with little information about the under-
lying dynamics of the nucleic acid. Finally, in the absence of an
experimental structure of unmodified GNA, model building and
MD simulations can be used to understand the structure and
dynamics of these interesting species.
Crystal Structure of an 8-mer GNA Duplex (8-mer-BrGNA).
Previous work revealed significant structural differences between
an 8-mer metallo-GNA duplex (8-mer-CuGNA, Figure 1a) that
harbored a Cu(II)-bridged hydroxypyridone homobase pair and
a 6-mer GNA duplex containing a 5-bromocytosine (Figure 1b).
For example, the elongated double helix of the former required
16 residues per turn, whereas the latter was condensed to only 10
residues per turn. At this point, it is not clear if these observations
giving rise to the classification as M- and N-type GNA duplexes16
are caused by the unique interstrand metal-coordination imple-
mented in 8-mer-CuGNA, sequence differences, or the ability of
GNA to adopt multiple conformations that depend on crystal
packing.
In this paper, we present the first combined experimental and
computational study of the structure and dynamics of parent
GNA duplexes using different sequences and lengths (Table 1)
to derive generalizable properties of GNA. We will start with a
discussion of a new brominated 8-mer GNA (8-mer-BrGNA)
duplex. With a GC content of 50% and six types of base steps, its
sequence is more complex than the shorter 6-mer previously
disclosed that comprised only repetitive GC and CG base steps.
This will then, together with the recently determined structures
of the Cu(II)ꢀhydroxypyridone homobase-pairs-containing GNA
duplex (8-mer-CuGNA) and the brominated 6-mer GNA (6-mer-
BrGNA), serve as a starting point for the modeling of parent
GNA (containing only natural nucleobases) and to deconvolute
the effects of modifications that were introduced to phase their
crystal structures by SAD and MAD techniques. The structure
and dynamics of the GNA models are then compared with those
We hence determined a GNA crystal structure which had a
different sequence and utilized here only a 5-bromouracil (BrU)
modification for phasing at the bromine edge by multiple anomalous
dispersion (MAD). The self-complementary sequence 30-CTCBr-
UAGAG-20 gave rod-shaped tetragonal crystals in sitting drop
setups after 2 weeks at 4 ꢀC in 10% 2-methyl-2,4-pentanediol,
40 mM sodium cacodylate, pH 5.5, 20 mM cobalt hexamine,
40 mM LiCl, 20 mM MgCl2. Crystals diffracted up to a resolution
of 1.8 Å at the synchrotron and allowed structure solution by
standard MAD techniques (Table 2). The refined GNA duplex
structure contains one single strand, 19 H2O molecules, and two
sodium ions. Unlike in the two previous structures of GNA
duplexes, no electron density was assigned to cobaltꢀhexamine
complexes that were so far found to be essential for GNA
crystallization by associating to the phosphate moieties of the
GNA backbone.
7966
dx.doi.org/10.1021/jo201469b |J. Org. Chem. 2011, 76, 7964–7974

The Journal of Organic Chemistry
ARTICLE
Table 3. Comparison of Average Helical Parameters for GNA Duplex Structures to B-DNA and A-DNAa
8-mer-BrGNA (N-type)
6-mer-BrGNA (N-type)
8-mer-CuGNA (M-type)
B-DNA
A-DNA
helical sense
right
10
right
10
right
16
right
10
right
12
residues per turn
helical pitch (Å)
helical rise (Å)b
x-displacement (Å)c
tilt (deg)d
Roll (deg)d
twist (deg)d
29
26
60
34
34
2.9
2.6
3.8
3.4
0.1
ꢀ0.1
0.6
36
2.9
ꢀ5.8
0.0
ꢀ6.0
0.5
ꢀ5.4
0.0
ꢀ4.2
0.1
3.6
6.4
ꢀ2.8
22.9
ꢀ3.4
5.4
8.0
35.8
ꢀ3.2
5.5
35.7
ꢀ3.4
5.4
31
slide (Å)d
PꢀP distance (Å)e
0.2
ꢀ1.5
5.9
7.0
a Data for GNA were calculated using the program CURVES (ref 12). Data for B- and A-DNA were taken from refs 10 and 11. b Global interbase pair
parameter. c Global base pair-axis parameter. d Local interbase pair step parameters. e Average intrastrand PꢀP distances.
All bases form standard WatsonꢀCrick hydrogen bonds with
the 5-bromouracil nucleotides having no apparent distorting effect.
While having a completely unrelated sequence, this new duplex
structure (8-mer-BrGNA) shows strikingly similar helix para-
meters like the previous 6-mer-BrGNA (Table 3). The backbone
around the vicinal C30ꢀO30 and C20ꢀO20 bonds adopts the
strictly alternating gauche and anti conformations of N-type
GNA16 so that each base pair contains one nucleotide in the
gauche and one in the anti conformation. As a consequence of
the alternating O30ꢀC30ꢀC20ꢀO20 torsions, the phosphateꢀ
phosphate distances within the same strand of the GNA back-
bone of 8-mer-BrGNA alternate between 5.79 Å (anti) and
5.17 Å (gauche); similar values were observed for the 6-mer-
BrGNA structure (5.58 Å, 5.09 Å).
From the structural data of the two determined brominated
duplexes with very different sequences, 30-GBrCGCGC-20 versus
30-CTCBrUAGAG-20, it can be concluded that N-type GNA
duplexes with their highly twisted and alternating gaucheꢀanti
backbone conformations can be formed independent of their
sequences. Interestingly, the GNA nucleotides adopting gauche
backbone conformations were found to comprise exclusively
purine bases in the 6-mer-BrGNA with its track of alternating
purineꢀpyrimidine nucleotides. In contrast, the new structure of
the self-complementary 8-mer-BrGNA, where a stretch of four
pyrimidine nucleotides is followed by four consecutive purine
nucleotides, demonstrates that there is actually no intrinsic
preference for purines to adopt gauche conformations. The lack
of a specific backbone preference by GNA duplexes is corrobo-
rated by the copper(II)-containing duplex that adopts an all-
gauche M-type conformation and is only disturbed at the Cu(II)-
bridged hydroxypyridone base pair. Here, the C10ꢀC10-distance
(12.7 Å) is widened by about 2.0 Å as compared to standard
WatsonꢀCrick base pairs and triggers thereby at this position a
change to the anti-conformation of the O30ꢀC30ꢀC20ꢀO20
torsion. Overall, the crystal structures of GNA duplexes indicate a
high variability of their backbone conformations that can cur-
rently not be traced back to unique interactions with counterions
from the crystallization solution or restraints given by different
packing arrangements.
free solution, where context-free dynamics may indicate an intrinsic
preference for backbone conformations. Using the structural in-
formation of the three different crystal structures of GNA duplexes,
we performed 20 ns MD simulations for each (see Table S1 of the
Supporting Information for system parameters). Additional starting
structures for the simulation of longer GNA duplexes, which
served as models for longer strands currently not experimentally
accessible, were generated by linking symmetry-mates in the
corresponding crystal structures, which we found to be already
suitably packed in a quasi-continuous manner. The analysis of
MD simulations was performed by writing the coordinates every
20 ps throughout the production run to obtain a total of 1000
snapshots. The coordinates of the non-natural nucleic acids were
reformatted (scripts available in the Supporting Information) to
generate input for a CURVES analysis of each structure. The
results were averaged over all snapshots, and the helicoidal twist
was plotted over time.
Analysis of MD Simulations Starting from N-type GNA
Duplexes. Bromination of GNA bases in the two experimentally
determined structures of 6-mer-BrGNA and 8-mer-BrGNA was
not found to induce obvious perturbation of the GNA duplex
structure, so that MD simulations could be directly based on the
available structures. Table 4 lists the average value of helicoidal
parameters obtained from both MD simulations as well as
the used N-type GNA crystal structures. It can be seen that
the results are in excellent agreement. For example, the local rise
between neighboring base pairs remains almost unaffected by the
simulation of GNA duplexes in solution as well as the base pair
tilt and roll parameters. The only major difference here is the
helical twist. It is not uncommon for solution simulations of
nucleic acids to report different values for helical twist as well as
several other backbone parameters due to the different environ-
ments caused by crystal packing. The crystallographic analysis of
short oligonucleotides has the shortcoming of a high ratio of end-
standing base pairs vs the number of base pairs in between.
Accordingly, the end-to-end alignment of duplexes, often a
prerequisite for the formation of ordered crystal lattices, needs
some distortion of twist as compared to duplexes in free solution.
It is also known that the AMBER force field tends to under-
estimate average base pair twist in DNA by ≈3ꢀꢀ4ꢀ and is also
likely to factor in this observation.32
MD Simulations of GNA Duplexes. To study the similarities
and differences between GNA and its DNA counterparts, an
understanding of the helicoidal parameters that describe their
duplex structures is needed. Moreover, the ability of GNA
duplexes to adopt gauche and anti conformations along their
backbone in crystalline form needs analysis of their behavior in
In the crystal structures the average helical twist is found to be
≈37ꢀ while in the MD simulation it is found to be 31.8ꢀ ( 2.4ꢀ
and 30.9ꢀ ( 2.0ꢀ for the 6-mer- and 8-mer-BrGNA structures,
respectively. The average base pair roll of the crystal structure is
7967
dx.doi.org/10.1021/jo201469b |J. Org. Chem. 2011, 76, 7964–7974

The Journal of Organic Chemistry
ARTICLE
Table 4. Average Helical Parameters for DNA and GNA Duplex Structures Obtained from Experiment and MD Simulations
Starting from N-type Duplexesa,b
8-mer-DNA
MD
6-mer-BrGNA
MD
8-mer-BrGNA
MD
exp.
16-mer-DNA MD
exp.
exp.
16-mer-BrGNA MD
shift (Å)c
rise (Å)
0.1 ((0.2)
3.5((0.1)
0.5((1.9)
4.3((2.8)
32.9((2.0)
ꢀ0.6((0.4)
0.0((0.2)
3.5((0.1)
0.0
3.2
0.0((0.2)
0.0
3.3
0.0((0.2)
0.0((0.1)
3.3((0.1)
0.1((1.0)
2.2((1.6)
30.6((1.5)
ꢀ3.0((0.2)
3.4
ꢀ0.1
0.6
3.3((0.1)
0.0((1.6)
4.5((2.8)
31.8((2.4)
ꢀ3.2((0.3)
3.3((0.1)
0.2((1.5)
2.5((2.9)
30.9((2.0)
ꢀ3.0((0.3)
tilt (deg)
roll (deg)
twist (deg)
slide (Å)
ꢀ0.7((2.6)
4.2((2.3)
0.0
0.0
5.9
3.1
36.0
0.2
33.3((2.5)
ꢀ0.8((0.3)
37.2
ꢀ3.3
36.8
ꢀ3.2
a Data for nucleic acids were calculated using the program CURVES.33,34 Data for B-DNA were taken from Olson et al.35 b Standard deviations (in
parentheses) given for MD data. c Local interbase pair parameters.
Figure 2. Comparison of backbone conformations of (a) the 16-mer-BrGNA MD with (b) the 6-mer-BrGNA crystal structure. (c) Overlay of the
crystal structure of the 6-mer-BrGNA duplex (red) with the average structure from MD simulation (green). Anti and gauche refers to the torsional angle
formed by O30ꢀC30ꢀC20ꢀO20.
found to be approximately 1.4ꢀ higher than what is observed in
the ensemble of molecular dynamics structures. An overlay of the
crystal structure and average MD structure can be seen in Figure 2,
showing the similarity of the structures with an average backbone
rmsd of 1.4 Å. Apart from the average helical twist, the experi-
mental and theoretical structures of 8-mer-BrGNA are also in
excellent agreement.
structures that is accompanied by a relaxed twisting of the overall
duplex. The otherwise excellent agreement between the experi-
mental and computed structures for 8-mer-BrGNA validates the
approach chosen and makes a computational model building for
parent GNA promising.
Analysis of MD Simulations Starting from Elongated
M-type GNA Duplexes. We will continue the discussion with
an analysis of the dsGNA crystal structure 8-mer-CuGNA that
already adopts an M-type conformation in its crystalline state.
Using this structure as a start point for MD simulations may give
some indication whether the less twisted structure of the M-type
duplexes obtained before by MD simulation of BrGNA duplexes
is a feature of context-free behavior in solution or just one of
various possibilities for GNA to adopt relaxed M-type structures.
In addition, the two hydroxypyridone base pairs of 8-mer-CuGNA
have to be considered which were engineered into the GNA
strands to complex a copper ion.15 While this metalꢀGNA hybrid
allowed for the acquisition of the first GNA crystal structure, the
interstrand complexation might not be representative of GNA
duplexes containing only regular nucleobases. We hence studied
the strands containing the hydroxypyridone base pairs as well as
an analog where the two hydroxypyridone base pairs are replaced
with AꢀT base pairs, resulting in the GNA sequence 30-
CGAATTCG-20 (8-mer-GNA).
One other major difference between the GNA duplex struc-
tures from crystals and MD simulations in solution is given
by the significant conformational differences adopted by the
O30ꢀC30ꢀC20ꢀO20 backbone torsional angles. In the crystal
structures of both the 6-mer- and 8-mer-BrGNA, this torsional
angle is found to alternate between gauche and anti along the
backbone. In the MD simulations, there is a strong preference for
the gauche conformation and very few (less than 1%) structures
adopt the anti conformation (Figure 2). Crystal structures of
single GNA nucleotides show that the gauche conformation is
indeed favored,17 while the 8-mer-CuGNA structure shows a
preference for the gauche conformation in areas away from the
metal chelation site.15 Close inspection of the crystal structures
does not reveal any specific interactions that could rationalize the
observed conformational behavior of the O30ꢀC30ꢀC20ꢀO20
torsion. The possibility that it is caused by a combination of
crystal packing and an accessible energy barrier concerning
rotation about the C30ꢀC20 bond needs therefore to be con-
sidered. Obviously the MD simulation of N-type GNA duplexes
in solution allows a facile NfM transition of their backbone
Overall, the helicoidal parameters provided by CURVES are
quite similar between experimental and simulated CuGNA
duplex structures.33 The simulated CuGNA duplex structures
7968
dx.doi.org/10.1021/jo201469b |J. Org. Chem. 2011, 76, 7964–7974

The Journal of Organic Chemistry
ARTICLE
Table 5. Average Helical Parameters for GNA Duplex Structures Obtained from Experiment and MD Simulations Starting from
M-type Duplexesa,b
experiment
16-mer-CuGNAd
MD simulation
16-mer-CuGNA
8-mer-CuGNA
8-mer-CuGNA
8-mer-GNA
16-mer-GNA
shift (Å)c
rise (Å)
0.0
3.5
0.0
3.4
ꢀ0.1((0.2)
3.5((0.1)
0.0((0.1)
3.5((0.1)
0.0((0.2)
3.5((0.1)
0.0((1.5)
0.9((2.7)
28.7((2.2)
ꢀ3.2((0.3)
0.0((0.1)
3.4((0.1)
0.0((2.2)
1.2((1.7)
29.5((1.4)
ꢀ3.5((0.2)
tilt (deg)
roll (deg)
twist (deg)
slide (Å)
0.0
0.0
ꢀ0.8((1.6)
ꢀ0.4((2.4)
24.1((1.9)
ꢀ2.9((0.2)
0.1((0.9)
ꢀ2.7
23.5
ꢀ3.5
ꢀ2.8
22.9
ꢀ3.4
ꢀ2.4((1.8)
23.9((1.6)
ꢀ2.9((0.2)
a Data for nucleic acids were calculated using the program CURVES.33,34 b Standard deviations (in parentheses) given for MD data. c Local interbase pair
parameter. d 16bpꢀCuGNA created by including an additional asymmetric unit along the z-axis to create a quasi-continuous strand of twice the length.
remain in their elongated state with an almost unaffected twist of
24ꢀ, i.e. ∼15ꢀ16 base pairs are required per helix turn. In
contrast, an exchange of the synthetic Cu(II)/hydroxypyridone
base pairs against natural ones induces a significantly increased
average twist between successive base pairs that is calculated to
29ꢀ ( 2.2ꢀ for the 8-mer-GNA MD simulation and 30ꢀ for the
structural ensemble of the 16-mer-GNA simulation (Table 5).
Accordingly, the latter M-type structures show a similar helical
twist as observed before for BrGNA duplexes of ≈12 residues per
turn. Apparently, the synthetic Cu(II)/hydroxypyridone base
pairs, which make up 25% of all base pairs present in 8-mer-
CuGNA, exert a strong unwinding effect on the overall GNA duplex
structure. Interestingly, a similar effect of unwinding was recently
Figure 3. View of a 20ꢀ20 purine interstrand base stack in 16-mer-GNA.
observed in the NMR structure of a synthetic DNA hybrid which
comprised three consecutive Ag(I)/imidazole base pairs. There the
helical B-form DNA twist was reduced from 36ꢀ to ∼28ꢀ.36
the results by Darden and co-workers and point to relatively
strong effects of crystal packing on GNA duplex conformation.
Finally, in the absence of accurate experimental solution struc-
tures of DNA or GNA addressing this point, incorrect force field
parameters in both cases can, of course, not be excluded.
Comparison of GNA and B-DNA Duplexes. With the success
of using different forms of GNA crystal structures to perform
model building and MD for generating undisturbed and conver-
ging dsGNA structures, a model of the elusive parent GNA struc-
ture containing only natural nucleobases can now be proposed
on the basis of the 16-mer-GNA. As expected, the structures of
dsDNA and dsGNA differ substantially. Most notably, the
average base pair slide of GNA is calculated to be ca. ꢀ3.2 to
ꢀ3.5 Å, whereas DNA possesses an average base pair slide of ca.
ꢀ0.6 Å. This large difference causes GNA to base stack in an
interstrand fashion as opposed to the conventional intrastrand
base stacking seen in canonical dsDNA.38 Hydrophobic overlap
can be seen when a base stack contains two 20-purines (Figure 3).
These purine/purine contacts could help stabilize the GNA
duplex. One of us had tested this hypothesis by using over-
hanging bases on either the 30 or 20 termini.17 It was observed that
overhanging nucleotides at the 20-termini lead to strong stabiliza-
tion of the duplex. The addition of overhanging adenines on the
20-termini leads to an increase of melting temperature of 14.6 ꢀC
and an increased thermodynamic stability of 4.0 kcal molꢀ1. This
overlap of 20-nucleobases is captured by all MD simulations
presented, as indicated by the large, negative average base pair
slide (ꢀ2.9 to ꢀ3.5 Å). When overhanging bases are added to the
30-end, it results in only a slight increase in melting temperature
(0ꢀ1.2 ꢀC). Also, addition of nucleotide overhangs on DNA
produced a less pronounced effect.
A structural reason for the difference between M-type du-
plexes adopted by CuGNA and others is certainly the conforma-
tions adopted by the phosphate backbone. It can be seen in the
crystal structure of CuGNA that the O20ꢀC20ꢀC30ꢀO30 tor-
sion belonging to the hydroxypyridone base pairs adopts an anti
conformation that remains stable during MD simulations, while
the same torsion belonging to natural base pairs adopts a gauche
conformation. An obvious explanation is that the backbone is
rather flexible and capable to convert rapidly between the gauche
and anti conformations. In CuGNA the metalloꢀbase pairs are
stiffening the duplex structure to some extent so that it adopts the
observed elongated form of ∼16 base pairs per turn. Taken
together, these results indicate that duplex GNA with natural
base pairs generally prefers to adopt a more condensed N-type
backbone conformation of just 12 base pairs per turn.
However, another possibility for the obtained different helix
turn heights and twists has been proposed by Darden and co-
workers.37 They observed that in MD simulations of duplex
DNA, where the dimensions of the crystal lattice are fixed to the
experimentally determined lattice size, the helicoidal parameters
matched well. It should be noted that they used the Parm94 force
field known to underestimate the B-DNA helical repeat compar-
ing to the more recent Parm98/99 parametrizations. However,
several differences could be discerned when comparing an un-
constrained MDs simulation in solution and MD simulation of
the crystal lattice. Most notably, the helicoidal parameters concern-
ing backbone conformation as well as the helical twist parameters
differed substantially between structural ensembles in solution
and in the crystal lattice. The differences in the backbone torsional
angles observed in our simulations and the X-ray structure match
7969
dx.doi.org/10.1021/jo201469b |J. Org. Chem. 2011, 76, 7964–7974

The Journal of Organic Chemistry
ARTICLE
Table 6. Results from Normal Mode Calculations of Single- and Double-Stranded 16-mer-DNA and GNAa
single stranded
TStotal
double stranded
TStotal
single strand ꢀ double strand
TSvib
TSvib
TΔSvib
TΔStotal
DNA (ε = 4)
GNA (ε = 4)
DNA (ε = 80)
GNA (ε = 80)
405.1
361.4
402.6
354.6
435.8
391.9
433.3
384.8
786.1
705.2
778.5
687.7
818.2
737.1
810.5
719.2
24.0
17.6
26.7
21.4
53.3
46.6
56.1
50.3
a All results are in kcal molꢀ1. ε denotes the dielectric constant used in the nmode calculations.
The simulations showed that dsGNA duplexes comprising
only standard nucleobases combines structural elements from all
experimental GNA structures. The average base pair slide of the
BrGNA crystal structures is most similar to the simulated dsGNA
structure by possessing an average base pair slide of ꢀ3.3 Å as
compared to ꢀ3.2 Å. The average base pair roll of dsGNA is
predicted to be a hybrid between the experimental structures
possessing a roll of ≈0.9ꢀ compared to ꢀ2.7ꢀ, 5.9ꢀ, and 3.1ꢀ for
the 8-mer-CuGNA, 6-mer-BrGNA, and 8-mer-BrGNA structures,
respectively. dsGNA is expected to possess an average helical
twist of ≈29ꢀ ( 2.2, which lies approximately halfway between
the ≈23ꢀ helical twist found in the 8-mer-CuGNA and the ≈37ꢀ
helical twist observed in the 8-mer-BrGNA crystal structures.
Dynamics of the GNA Duplex. In order to explore the relative
entropic effects between DNA and GNA annealing, we per-
formed frequency analysis to get a more quantitative information
on the entropy from the simulations. Twenty nanosecond trajec-
tories of 16-mer duplex and single stranded DNA and GNA were
obtained, and 400 snapshots were taken from each trajectory and
subjected to vibrational mode analysis using the nmode module of
AMBER. Previous experimental work has shown that GNA
incurs substantially less entropic penalty upon duplex formation
in comparison to DNA (Table 6). The ΔΔS observed between
DNA and GNA is ≈14 kcal molꢀ1. Results obtained from
vibrational analysis predict the ΔΔS between DNA and GNA
to be ≈5.8ꢀ6.7 kcal molꢀ1. In agreement with previous analyses
of entropic contributions to dimerization,39 the absolute magni-
tude of ΔΔS obtained using nmode is underestimated by ∼50%,
possibly because the solvent molecules and counterions are not
considered in the normal mode calculation. Nevertheless, the
results clearly show that GNA duplex formation is entropically
more favorable than in DNA. These calculations also suggest that
the largest contribution to ΔΔS is due to vibrational entropy.
This affirms the hypothesis that the observed twisting/untwist-
ing motions present in dsGNA are likely responsible for the
entropic effects previously explained. Another interesting ob-
servation is that in the simulations of ssGNA no hairpin forma-
tion occurs throughout the trajectory. Hairpin formation is typical of
ssDNA40 but is not observed in the 20 ns simulations of ssGNA.
Therefore, the GNA single strand itself may not be as flexible as
its DNA counterpart, contributing to the differences in entropic
penalty observed between the two species. Of course, other
possibilities such as solvation effects cannot be excluded, but they
are beyond the scope of this study.
The analysis of the MD trajectories of dsGNA helps to
rationalize the counterintuitive observation that duplex forma-
tion in GNA is entropically less unfavorable than for DNA2 as
well as the high stability of the GNA complexes. The results
clearly show that the structure is quite flexible, as demonstrated by
the rmsd plots of the backbone atoms (P, C20, C30, O20, and O30)
Figure 4. Backbone rmsd and average base pair twist versus time from
the 16-mer-GNA simulation. The black curves highlight periodical
variations and correspond to a moving average of fluctuations using a
window size of 2 ns.
with periodic fluctuations on the 8ꢀ10 ns time scale for 16-mer-
GNA, shown in Figure 4 (results for 8-mer-GNA are shown in
the Supporting Information), which is associated with a helicoi-
dal twisting and untwisting mode which is observed along the
entire 100 ns trajectory. It is interesting to note that this motion is
quasi-periodic with distances between the minima ranging from
∼5 to ∼15 ns. This is in line with the previously described
function of DNA as a “molecular spring”41 and suggests that GNA
has an even larger amplitude of the twisting and untwisting mode.
This conformational flexibility allows access to a large number of
microstates for the backbone torsions and is consistent with the
7970
dx.doi.org/10.1021/jo201469b |J. Org. Chem. 2011, 76, 7964–7974

The Journal of Organic Chemistry
ARTICLE
found for dsGNA formation. Accordingly, for dsDNA the duplex
structure is more rigid and thermal energy tends to rupture
hydrogen bonds and melting of the duplex. On the other hand, in
GNA it appears that thermal energy leads to enhanced global
molecular motions rather than the rupturing of hydrogen-bonded
base pairs. The flexibility of the GNA backbone adapts to the
thermal motion and resists duplex melting. While end-fraying
effects cannot be used as a predictor of duplex melting, it is of
note that with the enhanced molecular motions present in GNA
that no end-fraying can be observed in our 100 ns simulation.
These observations from the simulations are fully consistent with
the experimental results suggesting a decreased entropic penalty
for the formation of the GNA duplex.
’ CONCLUSIONS
The reported new crystal structure of a brominated 8-mer-
GNA duplex, together with previous structures of a 6-mer-
BrGNA duplex of different sequence and a copper(II)-hydro-
xypyridone base pair containing 8-mer GNA duplex, allow us to
identify some general features of GNA duplexes. First, a large
backbone-base inclination results in a significant slide between
neighboring base pairs and thus the formation of extensive inter-
strand base stacking. Second, the individual GNA nucleotides
have the flexibility to adopt two different conformations with
respect to the torsional angles between C20ꢀO and C30ꢀO. The
alternating gaucheꢀanti backbone observed in two brominated
GNAduplexes lead to ahighlytwistedduplex andmightconstitutea
general feature of GNA duplexes, although crystal packing effects
cannot be ruled out.
Figure 5. Plot of hydrogen-bond distance versus time of a terminal G-C
base pair: red = DNA duplex, blue = GNA duplex. The hydrogen bond of
interest is shaded red in the structure.
The MD simulations of dsGNA explain the high duplex stability
observed in duplex melting experiments through a combination
of enthalpic and entropic factors. Backbone rmsd plots show
periodic fluctuations that are associated with a twisting/untwist-
ing mode while the enthalpic (hydrogen bonds and base stacking)
contributions to free energy remain intact. This observation is
consistent with the thermodynamic analyses put forth previously
by one of us17 and provides an explanation for the unique properties
of dsGNA. While preorganization of ssGNA may play a part in
the more entropically favorable duplex formation, the new finding of
a twisting/untwisting mode inherent to duplex GNA (Figure 6)
may additionally contribute to entropic stabilization. The flex-
ible, yet energetically favorable nature of dsGNA is unique when
compared to that of dsDNA. Most likely, dsGNA trades off base
stacking in favor of an increased number of microstates leading to
an overall increase in entropy. This increase in entropy more than
makes up for what GNA loses in enthalpic contributions to free
energy and hence generates a more stable and fluid structure than
dsDNA. The MD simulations used in this work capture the main
structural features found in the experimental structures. The
major difference between the crystal structures and theoretical
predictions is in the average helical twist that is affected by the
gauche/anti distribution of the vicinal CꢀO backbone torsions.
While the N-type crystal structures of BrGNA alternate between
gauche and anti backbone configurations, the MD simulations
show a strong preference for an all-gauche conformation that has
previously been observed for the single-nucleotide GNA and
double-stranded M-type CuGNA. In contrast to the latter, the
relaxed form of M-type GNA structure has only 12 base pairs per
turn and is significantly condensed. As stated above, this phe-
nomenon is likely caused by the absence of disturbing metallo-
base pairs and crystal packing artifacts that have been shown to
Figure 6. Uncoiled and coiled structures observed in the simulation of
the 16-mer-GNA duplex.
decreased entropic penalty relative to dsDNA observed in the
thermodynamic analyses of duplex formation. Although modes
of dsDNA also have a significant degree of torsional plasticity,42
including twisting and untwisting, the modes observed for dsGNA
are quite different from that based on the large changes in the
persistence length of the dsGNA and the winding and unwinding
of the ribbon helix.
The MD trajectories may also shed light on the inherent
stability found in dsGNA. While the backbone rmsd fluctuates
throughout the simulation time, hydrogen bonding across the
duplex remains intact for dsGNA (blue in Figure 5), but not for
dsDNA (red in Figure 5). This indicates that much of the
enthalpic contribution to duplex formation found in dsDNA is
retained in the backbone while the ability of dsGNA to adopt
variable backbone conformations leads to the increased entropy
7971
dx.doi.org/10.1021/jo201469b |J. Org. Chem. 2011, 76, 7964–7974

The Journal of Organic Chemistry
ARTICLE
silica gel using 3:2:0.01 hexanes:acetone:Et3N to afford phosphorami-
dite 1 (see Figure 7) as a white foam (1.05 g, 78%). 31P NMR (162 MHz,
CDCl3) δ (ppm): 150.6, 150.2. HRMS: calcd for C37H45N4O7BrP
(M + H)+ 767.2204, found (M + H)+ 767.2206.
GNA Oligonucleotide Synthesis and Purification. GNA
oligonucleotides were prepared by an DNA/RNA synthesizer on a 1 μmol
scale. GNA phosphoramidites were used at a concentration of 100 mM
with a standard protocol for 2-cyanoethyl phosphoramidites, except that
the coupling was extended to 3 min. After the trityl-on synthesis, the
resin was incubated with concentrated aqueous ammonia at room tem-
perature overnight in the dark. The entire solution was then applied
directly to a Sep-Pak Classic reverse phase column (Waters, 360 mg) and
washed sequentially with 3% NH4OH (15 mL), water (10 mL), 1.5%
aqueous TFA (10 mL), and water (10 mL), and the oligonucleotides
finally eluted with 20% aqueous acetonitrile. The oligonucleotides were
further purified by HPLC eluting over a Waters Xterra column (MS C18,
4.6 ꢁ 50 mm, 2.5 μM particle size) at 60 ꢀC with a linear gradient (flow =
1.0 mL/min) of acetonitrile and aqueous triethylammonium acetate
buffer (50 mM, pH = 7.0). Identities were confirmed by MALDI-
TOF MS.
X-ray Crystallography. The 8-mer-BrGNA duplex used for
structural analysis had the sequence 30-CTCBrUAGAG-20, where the
5-bromouracil at position 4 allowed phasing by multiple anomalous
dispersion (MAD) at the bromine edge. The 8-mer-BrGNA duplex gave
rodlike crystals in sitting-drop setups after 2 weeks at 4 ꢀC in 10%
2-methyl-2,4-pentanediol, 40 mM sodium cacodylate, pH 5.5, 20 mM
cobalt hexamine, 40 mM LiCl, 20 mM MgCl2. Accordingly, MAD
diffraction data up to a resolution of 1.8 Å were recorded at synchrotron
beamline ID23-1, ESRF, Grenoble. After data reduction and scaling the
crystal structure was solved by SHELXE in the tetragonal space group
I4122 (a = b = 56.93 Å, c = 28.97 Å) and refined by alternating cycles of
REFMAC544 and COOT45 using stereochemical parameters derived
from the previously determined atomic resolution structure of a 6-mer-
BrGNA duplex. Given the data quality, structure refinement with one
molecule per asymmetric symmetry unit converged at apparently rather
high but acceptable R-factor/R-free values of 23.4%/27.1%. Indexing
and refinement of the 8-mer-BrGNA duplex structure was repeated in
alternative, lower symmetry space groups, but that did not alleviate this
problem. Closer inspection of the difference electron density indicated
helical but distorted features in the intervening, large solvent channels
(calculated solvent content 65%). These density features might corre-
spond to GNA duplexes highly disordered in the crystal lattice and were
hence not amenable for structural modeling. The asymmetric unit of the
refined GNA duplex structure contains one single strand, 19 H2O
molecules, and two Na+ ions. The 30-terminal cytosine nucleotide is
partly disordered as indicated by an occupancy of only 70%.
Figure 7. Synthesis of the phosphoramidite 1.
affect previously studied nucleic acids, even though other explana-
tions, e.g., different ion strengths, can currently not be excluded.37
In conclusion, the combined effort of theory and experiment has
shed light on both the structure and inherent thermodynamic
stability and dynamics of unmodified GNA. The discovery of a
pronounced twisting/untwisting mode of dsGNA could not only
play an important role in the high thermodynamic stability of
GNA but also give a hint as to why the simplest form of a duplex-
forming nucleic acid currently known was not chosen by life.
Obviously, the large degree of backbone variations would
hamper trivial modes of sequence-specific recognition by biolo-
gical macromolecules.
’ EXPERIMENTAL SECTION
Synthesis of the 5-Bromouracil Phosphoramidite for
Automated Solid-Phase Synthesis. General Procedures and
Reagents. NMR spectra were recorded on a 300 MHz spectrometer.
High-resolution mass spectra were obtained with a LTQ-FT instrument
using ES ionization. Reactions were performed under an atmosphere of
argon unless otherwise specified. Compound 2 was prepared as de-
scribed previously43 and its identity and purity were confirmed using
1H NMR.
Compound 3. To a suspension of 5-bromouracil (1.04 g, 5.5 mmol)
in anhydrous DMF (11 mL) under argon was added NaH (44 mg,
1.1 mmol, 60% in mineral oil) and the solution was allowed to stir under
argon for 1 h. A solution of compound 2 (1.95 g, 5.2 mmol) in DMF
(11 mL) was added to the first solution and then heated to 85 ꢀC
overnight. The next morning, the solution was cooled, all solvent
removed, and the resulting oil coevaporated with toluene, redissolved
in ethyl acetate, and concentrated again. The product was purified via
flash chromatography over silica gel starting with 2:1:0.01 hexanes:
acetone:Et3N, then 3:2:0.01 hexanes:acetone:Et3N, and finally eluting
with 1:1:0.01 hexanes:acetone:Et3N to afford compound 3 as a white
foam (1.32 g, 43%). 1H NMR (300 MHz, CDCl3) δ (ppm): 7.65
(s, 1H), 7.46 (d, J = 7.4 Hz, 2H), 7.39ꢀ7.17 (m, 7H), 6.85 (d, J = 8.8 Hz,
4H), 4.14 (m, 2H), 3.79 (s, 6H), 3.64 (dd, J = 14.5, 8.6 Hz, 1H), 3.21 (d,
J = 4.5 Hz, 2H). 13C NMR (75 MHz, CDCl3) δ (ppm): 160.0, 158.6,
151.0, 145.8, 144.6, 135.7, 130.0, 128.07, 128.01, 127.0, 113.3, 95.8, 86.5,
68.7, 64.6, 55.3, 52.3. IR (solid) ν (cmꢀ1): 3439 (br), 3168 (br), 3059,
2931, 2835, 1675, 1606, 1506, 1443, 1347, 1300, 1245, 1174, 1070, 1029,
906, 826, 727, 701, 621, 582, 526, 423. HRMS: calcd for C28H27N2O6Br-
Na (M + Na)+ 589.0945, found (M + Na)+ 589.0956.
Computational Methodology. The sequences of GNA shown in
Table 1 were built starting from the X-ray structures of the hydroxypyr-
idone copper complex15 or the bromo-GNA structures in xleap. Charges
for all GNA nucleosides were derived using RESP fitting on DFT-
optimized and modified GNA residues. Geometry optimizations were
performed using Gaussian 0346 at the B3LYP/6-31G* level while the
backbone torsional angles were restrained to those found in the experi-
mentally observed structure. Electrostatic potentials were calculated at
the HF/6-31G* level of theory by fitting of the RESP charges using
antechamber and setting intermolecular equivalencies as needed.47 All
MD simulations and postprocessing was performed using the AMBER 9
suite of programs48, and in analogy to our previous studies of non-natural
oligonucleotides, the structures were loaded into xleap and edited as
needed.23ꢀ26
Phosphoramidite 1. To an argon-purged solution of compound 3
(1.00 g, 1.8 mmol) and N,N-diisopropylethylamine (1.85 mL, 10.6
mmol) in methylene chloride (30 mL) was added 2-cyanoethyl N,N-
diisopropylchlorophosphoramidite (0.59 mL, 2.6 mmol) dropwise and
the solution stirred for 2 h at room temperature under argon. The
solution was diluted with methylene chloride and washed once with
saturated aqueous NaHCO3, dried over Na2SO4, and finally concen-
trated. The crude product was purified by flash chromatography over
For simulations of natural nonmetalated GNA sequences the corre-
sponding hydroxypyridone base pairs were removed and replaced with
either A or T. This was done by removing the entire nucleobase except
for the nitrogen linkage to the backbone and flanking carbons. Atoms
7972
dx.doi.org/10.1021/jo201469b |J. Org. Chem. 2011, 76, 7964–7974

The Journal of Organic Chemistry
ARTICLE
were then renamed to correspond to the new nucleobases, and xleap was
allowed to automatically add in the missing atoms. Unlike the case of
DNA and RNA, the effect of ion strength in MD simulations has not yet
been studied in detail and only the number of Na+ counterions necessary
to balance the negative charge of the phosphate backbone were included
in the simulation. Finally, the solute was placed in a pre-equilibrated box
of TIP3P waters extending 8 Å beyond the solute in each dimension.
Table 1 provides the sequences of all the nucleic acids studied (AMBER
atom types used for the GNA backbone can be found in Figure S9 in the
Supporting Information). 16-mer-GNA, 16-mer-CuGNA, and 16-mer-
BrGNA were built by including an additional duplex from the crystal
structure. PyMol 0.99 was used to generate symmetry-mates and retain
an additional duplex stacked end to end against the original 8-mer
structure as well as to generate the figures. These quasi-continuous forms
were then subsequently edited as described above. All starting geome-
tries and additional instructions on how to perform these simulations are
available in the Supporting Information.
’ ASSOCIATED CONTENT
S
Supporting Information. Details of the X-ray crystal-
b
lographic and computational studies as well as analytical infor-
mation. This material is available free of charge via the Internet at
http://pubs.acs.org. The coordinates and structure factors of 8-mer-
BrGNA are deposited in the RCSB database under accession code
2XC6.
’ AUTHOR INFORMATION
Corresponding Author
*E-mail: owiest@nd.edu, essen@chemie.uni-marburg.de, meggers@
chemie.uni-marburg.de.
’ ACKNOWLEDGMENT
After the initial setup was complete, minimization was performed first
by restraining the solute and allowing sodium ions and water to relax. For
the case of unmodified GNAs, an additional minimization step was
included where restraints were added between hydrogen bonds of the
replaced nucleotides. Finally, the entire system was allowed to relax prior
to the MD simulations. The particle mesh Ewald molecular dynamics
(pmemd) module of AMBER 9 was used to perform all MD simulations
in order to increase the efficiency of the parallel simulations. Parameters
for the simulations were adopted from the Cornell et al. forcefield, along
with the adjustments by Wang et al. (Parm99).49,50 A bonded model was
used to describe the base pairs in the hydroxypyridoneꢀcopper com-
plex, and parameters were fit to reproduce the coordinates from the
crystal structure. Equilibration and heating of the system including the
GNA and a water box as shown in Figure S4 (Supporting Information)
was performed using a constant volume (NVT) ensemble by adding a
10 kcal molꢀ1 restraint on the solute while the system was heated from
0 to 300 K over 20 ps. Solute restraints are then removed and the system
is equilibrated for 200 ps in a constant pressure (NPT) ensemble at
1 atm. Production runs of 20 ns (100 ns for 16-mer-GNA) were sub-
sequently performed. Two femtosecond time steps were used for DNA,
GNA, and BrGNA while 1 fs time steps were required for system stability
in the hydroxypyridone GNA examples. SHAKE was used in all simulations
to restrain bonds to hydrogen.
Parameters for unmodified GNA bases and glycol backbone were
taken directly from the Cornell et al. force field.49 It should be noted that
the crucial backbone torsional parameters in this force field were fit to
reproduce experimental and quantum mechanical results for methyl
ethyl ether and similar acyclic models. It therefore should be expected
that this force field will perform quite well in describing the backbone
parameters in GNA, as the backbone is actually more similar to the training
sets used to create the force field than the more widely studied DNA is.
A more limited study of GNA using the parameter set gave indeed results in
excellent agreement with experiment.31 To further test the applicability of
the AMBER force field to GNA, MP2/6-31G* calculations of OꢀCꢀCꢀO
torsional energies were compared to results using AMBER. Two model
systems were used for this study, one for GNA and one for DNA. It is found
that the AMBER force field performs similarly for both systems, successfully
displaying that the force field is as effective in describing the backbone
torsions present in GNA as well as what is observed in DNA.
We would like to thank the Center of Research Computing at
the University of Notre Dame and Teragrid (Grant CHE090124)
for the generous allocation of computer resources used in this
project. We also gratefully acknowledge a Peter J. Grace Fellow-
ship from the University of Notre Dame to A.T.J. The authors
thank Sandor Brockhauser for support at ESRF beamline ID23-1,
Grenoble, and the reviewers for valuable suggestions.
’ REFERENCES
(1) Watson, J. D.; Crick, F. H. C. Nature 1953, 171, 737.
(2) Appella, D. H. Curr. Opin. Chem. Biol. 2009, 13, 687.
(3) Schneider, K. C.; Benner, S. A. J. Am. Chem. Soc. 1990, 112, 453.
(4) Azymah, M.; Chavis, C.; Lucas, M.; Morvan, F.; Imbach, J. L.
Nucleosides Nucleotides 1992, 11, 1241.
(5) Vandendriessche, F.; Augustyns, K.; Vanaerschot, A.; Busson, R.;
Hoogmartens, J.; Herdewijn, P. Tetrahedron 1993, 49, 7223.
(6) Nielsen, P.; Kirpekar, F.; Wengel, J. Nucleic Acids Res. 1994,
22, 703.
(7) Nielsen, K. D.; Kirpekar, F.; Roepstorff, P.; Wengel, J. Bioorg.
Med. Chem. 1995, 3, 1493.
(8) Peng, L.; Roth, H. J. Helv. Chim. Acta 1997, 80, 1494.
(9) Eschenmoser, A. Science 1999, 284, 2118.
(10) Schoning, K. U.; Scholz, P.; Guntha, S.; Wu, X.; Krishnamurthy,
R.; Eschenmoser, A. Science 2000, 290, 1347.
(11) Wippo, H.; Reck, F.; Kudick, R.; Ramaseshan, M.; Ceulemans,
G.; Bolli, M.; Krishnamurthy, R.; Eschenmoser, A. Bioorg. Med. Chem.
2001, 9, 2411.
(12) Zhang, L.; Peritz, A.; Meggers, E. J. Am. Chem. Soc. 2005,
127, 4174.
(13) Schlegel, M. K.; Peritz, A. E.; Kittigowittana, K.; Zhang, L.;
Meggers, E. Chembiochem 2007, 8, 927.
(14) Schlegel, M. K.; Xie, X.; Zhang, L.; Meggers, E. Angew. Chem.
Int. Ed. 2009, 48, 960.
(15) Schlegel, M. K.; Essen, L.-O.; Meggers, E. J. Am. Chem. Soc.
2008, 130, 8158.
(16) Schlegel, M. K.; Essen, L.-O.; Meggers, E. Chem. Commun.
2010, 46, 1094.
(17) Schlegel, M. K.; Xie, X.; Zhang, L.; Meggers, E. Angew. Chem.
Int. Ed. 2009, 48, 960.
(18) Cheatham, T. E.; Kollman, P. A. Annu. Rev. Phys. Chem. 2000,
51, 435.
(19) Cheatham, T. E.; Srinivasan, J.; Case, D. A.; Kollman, P. A.
J. Biomol. Struct. Dyn. 1998, 16, 265.
(20) Seibel, G. L.; Singh, U. C.; Kollman, P. A. Proc. Natl. Acad. Sci.
U. S. A. 1985, 82, 6537.
Postprocessing was performed using the ptraj module of AMBER 9.
C10ꢀC10 intrastrand distances, PꢀP intrastrand distances, and all
hydrogen-bond distances were extracted for analysis. Structures were
extracted every 20 ps to give a total of 1000 structures per simulation.
Curves 5.1 was then used to calculate the helical parameters for each pdb
and results were either averaged or viewed over time.33,34
(21) Dixit, S. B.; Beveridge, D. L.; Case, D. A.; Cheatham, T. E.;
Giudice, E.; Lankas, F.; Lavery, R.; Maddocks, J. H.; Osman, R.; Sklenar,
H.; Thayer, K. M.; Varnai, P. Biophys. J. 2005, 89, 3721.
Structures of the model systems used and graphs of the torsional
energies can be found in the Supporting Information.
7973
dx.doi.org/10.1021/jo201469b |J. Org. Chem. 2011, 76, 7964–7974

The Journal of Organic Chemistry
ARTICLE
(22) Orozco, M.; Noy, A.; Perez, A. Curr. Opin. Struct. Biol. 2008,
18, 185.
(23) O’Daniel, P. I.; Jefferson, M.; Wiest, O.; Seley-Radtke, K. L.
J. Biomol. Struct. Dyn. 2008, 26, 283.
(24) O’Neil, L. L.; Grossfield, A.; Wiest, O. J. Phys. Chem. B 2007,
111, 11843.
(49) Cornell, W. D.; Cieplak, P.; Bayly, C. I.; Gould, I. R.; Merz,
K. M.; Ferguson, D. M.; Spellmeyer, D. C.; Fox, T.; Caldwell, J. W.;
Kollman, P. A. J. Am. Chem. Soc. 1995, 117, 5179.
(50) Wang, J. M.; Cieplak, P.; Kollman, P. A. J. Comput. Chem. 2000,
21, 1049.
(25) O’Neil, L. L.; Wiest, O. J. Phys. Chem. B 2008, 112, 4113.
(26) O’Neil, L. L.; Wiest, O. Org. Biomol. Chem. 2008, 6.
(27) Ivanova, A.; Rosch, N. J. Phys. Chem. A 2007, 111, 9307.
(28) He, W.; Hatcher, E.; Balaeff, A.; Beratan, D. N.; Gil, R. R.;
Madrid, M.; Achim, C. J. Am. Chem. Soc. 2008, 130, 13264.
(29) Lynch, S. R.; Liu, H. B.; Gao, J. M.; Kool, E. T. J. Am. Chem. Soc.
2006, 128, 14704.
(30) De Winter, H.; Lescrinier, E.; Van Aerschot, A.; Herdewijn, P.
J. Am. Chem. Soc. 1998, 120, 5381.
(31) Zhou, H.; Johnson, A. T.; Wiest, O.; Zhang, L. Org. Biomol.
Chem. 2011, 13, 2840.
(32) Olson, W. K.; Zhurkin, V. B. Curr. Opin. Struct. Biol. 2000,
10, 286.
(33) Lavery, R.; Sklenar, H. J. Biomol. Struct. Dyn. 1988, 6, 63.
(34) Lavery, R.; Sklenar, H. J. Biomol. Struct. Dyn. 1989, 6, 655.
(35) Olson, W. K.; Bansal, M.; Burley, S. K.; Dickerson, R. E.;
Gerstein, M.; Harvey, S. C.; Heinemann, U.; Lu, X.; Neidle, S.; Shakked,
Z.; Sklenar, H.; Suzuki, M.; Tung, C. S.; Westhof, E.; Wolberger, C.;
Berman, H. M. J. Mol. Biol. 2001, 313, 229.
(36) Johannsen, S.; Megger, N.; B€ohme, D.; Sigel, R. K. O.; M€uller, J.
Nat. Chem. 2010, 2, 229.
(37) Bevan, D. R.; Li, L. P.; Pedersen, L. G.; Darden, T. A. Biophys.
J. 2000, 78, 668.
(38) Mura, C.; McCammon, J. A. Nucleic Acids Res. 2008, 36, 4941.
(39) Li, L.; Uversky, V. N.; Dunker, A. K.; Meroueh, S. O. J. Am.
Chem. Soc. 2007, 120, 15668.
(40) Johnson, A. T.; Wiest, O. J. Phys. Chem. B 2007, 111, 14398.
(41) Girard, E.; Prange, T.; Dhaussy, A.-C.; Migianu-Griffoni, E.;
Lecouvey, M.; Chervin, J.-C.; Mezouar, M.; Kahn, R.; Fourme, R. Nucleic
Acids Res. 2007, 35, 4000.
(42) Bomble, Y. J.; Case, D. A. Biopolymes 2008, 89, 722.
(43) Zhang, L.; Peritz, A. E.; Carroll, P. J.; Meggers, E. Synthesis
2006, 645.
(44) Collaborative Computing Project Number 4 Acta Crystallog.
Sect. D 1994, 50, 760.
(45) Emsley, P.; Cowtan, K. Acta Crystallogr. Sect. D 2004, 60, 2126.
(46) Frisch, M. J.; Trucks, G. W.; Schlegel, H. B.; Scuseria, G. E.;
Robb, M. A.; Cheeseman, J. R.; Montgomery, J. A., Jr.; Vreven, T.;
Kudin, K. N.; Burant, J. C.; Millam, J. M.; Iyengar, S. S.; Tomasi, J.;
Barone, V.; Mennucci, B.; Cossi, M.; Scalmani, G.; Rega, N.; Petersson,
G. A.; Nakatsuji, H.; Hada, M.; Ehara, M.; Toyota, K.; Fukuda, R.;
Hasegawa, J.; Ishida, M.; Nakajima, T.; Honda, Y.; Kitao, O.; Nakai, H.;
Klene, M.; Li, X.; Knox, J. E.; Hratchian, H. P.; Cross, J. B.; Bakken, V.;
Adamo, C.; Jaramillo, J.; Gomperts, R.; Stratmann, R. E.; Yazyev, O.;
Austin, A. J.; Cammi, R.; Pomelli, C.; Ochterski, J. W.; Ayala, P. Y.;
Morokuma, K.; Voth, G. A.; Salvador, P.; Dannenberg, J. J.; Zakrzewski,
V. G.; Dapprich, S.; Daniels, A. D.; Strain, M. C.; Farkas, O.; Malick,
D. K.; Rabuck, A. D.; Raghavachari, K.; Foresman, J. B.; Ortiz, J. V.; Cui,
Q.; Baboul, A. G.; Clifford, S.; Cioslowski, J.; Stefanov, B. B.; Liu, G.;
Liashenko, A.; Piskorz, P.; Komaromi, I.; Martin, R. L.; Fox, D. J.; Keith,
T.; Al-Laham, M. A.; Peng, C. Y.; Nanayakkara, A.; Challacombe, M.;
Gill, P. M. W.; Johnson, B.; Chen, W.; Wong, M. W.; Gonzalez, C.;
Pople, J. A. Gaussian 03; Gaussian: Wallingford, CT, 2004.
(47) Cieplak, P.; Cornell, W. D.; Bayly, C.; Kollman, P. A. J. Comput.
Chem. 1995, 16, 1357.
(48) Case, D. A.; Darden, T. A.; Cheatham, T. E., III; Simmerling,
C. L.; Wang, J.; Duke, R. E.; Luo, R.; Merz, K. M.; Pearlman, D. A.;
Crowley, M.; Walker, R. C.; Zhang, W.; Wang, B.; Hayik, S.; Roitberg,
A.; Seabra, G.; Wong, K. F.; Paesani, F.; Wu, X.; Brozell, S.; Tsui, V.;
Gohlke, H.; Tan, C.; Mongan, J.; Hornak, V.; Cui, G.; Beroza, P.;
Mathews, D. H.; Schafmeister, C.; Ross, W. S.; Kollman, P. A. AMBER 9;
University of California: San Francisco, 2006.
7974
dx.doi.org/10.1021/jo201469b |J. Org. Chem. 2011, 76, 7964–7974