Acta Cryst. (2009). F65, 809–812doi:10.1107/S1744309109026281
Acta Crystallographica Section F
Crystallization and X-ray diffraction analysis of an
‘all-locked’ nucleic acid duplex derived from a
Katja Behling,aAndre ´ Eichert,a
Jens P. Fu ¨rste,aChristian Betzel,b
Volker A. Erdmannaand
Charlotte Fo ¨rstera*
aInstitute of Chemistry and Biochemistry, Free
University Berlin, Thielallee 63, 14195 Berlin,
Germany, andbInstitute of Biochemistry and
Food Chemistry, University of Hamburg,
Notkestrasse 85, Building 22a, c/o DESY,
22603 Hamburg, Germany
Received 8 June 2009
Accepted 6 July 2009
Modified nucleic acids are of great interest with respect to their nuclease
resistance and enhanced thermostability. In therapeutical and diagnostic
applications, such molecules can substitute for labile natural nucleic acids that
are targeted against particular diseases or applied in gene therapy. The so-called
‘locked nucleic acids’ contain modified sugar moieties such as 20-O,40-C-
methylene-bridged ?-d-ribofuranose and are known to be very stable nucleic
acid derivatives. The structure of locked nucleic acids in single or multiple LNA-
substituted natural nucleic acids and in LNA–DNA or LNA–RNA hetero-
duplexes has been well investigated, but the X-ray structure of an ‘all-locked’
nucleic acid double helix has not been described to date. Here, the
crystallization and X-ray diffraction data analysis of an ‘all-locked’ nucleic acid
helix, which was designed as an LNA originating from a tRNASermicrohelix
RNA structure, is presented. The crystals belonged to space group C2, with unit-
cell parameters a = 77.91, b = 40.74, c = 30.06 A˚, ? = 91.02?. A high-resolution
and a low-resolution data set were recorded, with the high-resolution data
showing diffraction to 1.9 A˚resolution. The crystals contained two double
helices per asymmetric unit, with a Matthews coefficient of 2.48 A˚3Da?1and a
solvent content of 66.49% for the merged data.
The stabilization of nucleic acids by introducing modified nucleotides
is an ongoing subject in therapeutic and diagnostic applications, as
natural macromolecules such as RNA and DNA are labile towards
nuclease digestion and have low thermal stability. ‘Locked’ nucleic
acids (LNAs) and the related LNA families contain modified sugar
moieties such as 20-O,40-C-methylene cross-linked ?-d-ribofuranose,
in contrast to the naturally occurring ribose/deoxyribose in RNA/
DNA. LNAs were first synthesized by the groups of T. Imanishi
(Obika et al., 1997) and J. Wengel (Kumar et al., 1998) and they bind
to complementary RNA or DNA via standard Watson–Crick base
pairing (Vester & Wengel, 2004).
It has been well documented that LNAs show a greatly increased
thermostability in comparison to other modified nucleic acids. An
example of a comparative study of different modifications in nucleic
acids with respect to their stability has been reported for the tenascin
C-binding aptamer TTA-1 after substitution with different modified
nucleic acid blocks (Schmidt et al., 2004). The in vitro thermostability
was described to be in the following order using different common
modifications of nucleic acids: 20-F/20-OMe < RNA/RNA ? 20-OMe/
20-OMe < 20-F/LNA < RNA/LNA = LNA/RNA < 20-OMe/LNA <
LNA/LNA. An explanation of the enhanced stability of LNAs has
been proposed by the research groups of P. Jacobsen and J. Wengel
(Petersen et al., 2000) based on the investigation of an LNA–DNA
duplex structure: the ?-d-ribofuranose LNA is ‘locked’ in the 30-endo
conformation. This directs the phosphate backbone into a confor-
mation with a decreased loss of entropy upon helix formation, in
which the duplex favours a more efficient stacking of the nucleobases.
This implicates a loss in enthalpy upon helix formation. The authors
report that the formation of an LNA–DNA duplex is favoured by
both enthalpy and entropy.
# 2009 International Union of Crystallography
All rights reserved
The structure and conformation of locked nucleic acids has been
well investigated and has been analyzed extensively using single or
multiple LNA-substituted natural nucleic acids or heteroduplexes
such as LNA–DNA or LNA–RNA complexes. Structural studies
using ?-d- or ?-l-LNA/DNA mix-mers hybridized to RNA or DNA
showed the following. ?-d-LNA (LNA/DNA mix-mer)–RNA
duplexes adopt the A-type conformation, whereas ?-l-LNA (LNA/
DNA mix-mer)–DNA helices adopt the B-type conformation
(Petersen et al., 2002; Vester & Wengel, 2004). Investigations using
fully modified LNA strands associated with RNA or DNA forming a
heteroduplex showed that the ?-d-ribofuranose LNA binds to RNA
adopting the A-RNA conformation and the binding of ?-d-ribo-
furanose to DNA induces a mixed N- and S-type sugar puckering
(Nielsen et al., 2004), whereas the ?-l-ribofuranose LNA binds to
DNA in a B-DNA type conformation (Nielsen et al., 2002). In
summary, the ?-d-ribofuranose LNA substitutions induce an A-type
nucleic acid conformation and the ‘locked’ 30-endo conformation
seems to increase the thermostability of the duplex to a great extent.
An understanding of the LNA tertiary structure is of great interest
in order to explain the enhanced thermostability of these modified
nucleic acids. The application of nucleic acids in diagnostic and
therapeutic medicine, in gene therapy and in drug design is an
ongoing research field and opens a broad field of new medications.
Nevertheless, the nuclease-sensitivity and the low stability of nucleic
acids is a great problem which often prevents successful applications.
Approaches involving the application of modified nucleic acids which
possess enhanced thermostability and nuclease resistance will facil-
itate the use of such molecules in therapy and diagnostics. LNAs are
known to be very stable nucleic acids and the structure of an ‘all-
locked’ nucleic acid duplex will provide insight into the detailed local
geometric parameters, which may help in finding further explanations
for their increased stability. LNAs have a great potential for use in
drug development and for application in diagnostics and therapy
(Kaur et al., 2007; Petersen & Wengel, 2003).
To our knowledge, the structure of an ‘all-locked’ nucleic acid
homoduplex has not yet been described. The question of how the
conformation of the modified LNA sugar influences the phosphate
backbone and the stacking of base pairs in a completely ‘all-locked’
?-d-ribofuranose nucleic acid led us to the idea of undertaking a
comparative X-ray structure analysis. We have recently solved the
1.2 A˚resolution crystal structure of an Escherichia coli tRNASer
microhelix, which resembles the aminoacyl stem of the tRNA
(Eichert et al., 2009). We therefore decided to focus our interest on
crystallizing an ‘all-locked’ LNA duplex with a sequence corre-
sponding to this RNA. We designed the helix as a completely ‘all-
locked’ nucleic acid by maintaining the base sequence of the RNA.
Here, we present the crystallization and preliminary X-ray diffraction
analysis of the LNA homoduplex. These data could lead to the first
X-ray structure of an ‘all-locked’ nucleic acid duplex which can be
directly compared with the corresponding RNA structure.
2. Materials and methods
2.1. Crystallization of the ‘all-locked’ LNA tRNASermicrohelix
The sequence of the 7-mer LNA helix was derived from the E. coli
tRNASeraminoacyl stem microhelix which has been crystallized
previously (Eichert et al., 2009) and originated from the tRNA
isoacceptor with database ID RS 1661 (Sprinzl & Vassilenko, 2005).
The LNA was designed to contain exclusively locked nucleic acid
building blocks by maintaining the base sequence of the RNA for
further comparative studies, except for the U to T exchange in
standard LNA synthesis. Since we have previously evaluated the
quality of commercially available LNAs in functional studies of
aptamers (Schmidt et al., 2004) and crystallization experiments
(Fo ¨rster et al., 2006), we again employed locked oligonucleotides
from commercial sources for the present study. Chemically synthe-
sized single strands with sequences 50-LNG-LNG-LNT-LNG-LNA-
purchased from IBA (Go ¨ttingen, Germany) at HPLC purification
grade. No further purification was undertaken as we routinely crys-
tallize chemically synthesized oligonucleotides after HPLC purifica-
tion. For hybridization, both LNA single strands were annealed in
distilled water, heated to 363 K and subsequently cooled to room
temperature within 3–4 h. The resulting LNA duplex was concen-
trated in a speed vac (SpeedVac SC 110, Savant, Minnesota, USA) to
a final concentration of 0.5 mM. This sample was used for all subse-
quent crystallization setups.
For the initial crystallization screening experiments, we used two
different crystallization kits. The first was the Natrix Nucleic Acid
Crystallization Kit (HR2-116; Hampton Research, California, USA)
consisting of 48 different conditions. Within this screen, each solution
was used in a 100 ml reservoir well and 1 ml of the reservoir solution
was added to the LNA in the droplet for crystallization experiments
as follows. Setups were prepared at 294 K using the sitting-drop
vapour-diffusion technique with CrystalQuick Lp plates from
Greiner Bio-One (Germany). 1 ml of a 0.5 mM unbuffered solution of
LNA in distilled water was mixed with 1 ml reservoir solution and
equilibrated against 100 ml reservoir solution; the plates were directly
covered with a VIEWseal foil (Greiner Bio-One, Germany). As a
second screen, we used the Nucleic Acid Mini Screen from Hampton
Research (HR2-118; Hampton Research, California, USA) with 24
different conditions using Linbro plates (ICN Biomedicals Inc., Ohio,
USA). In contrast to the first screen, 40%(v/v) MPD (2-methyl-2,4-
pentanediol) in distilled water pH 7.4 was used as the reservoir
solution in all setups. Here, the hanging-drop vapour-diffusion tech-
nique was applied, with the crystal droplets hanging on cover slides.
For crystallization trials, 1 ml of a 0.5 mM unbuffered solution of LNA
in distilled water was used and combined with 1 ml crystallization
solution from the screen. Equilibration took place at 294 K against
1 ml 40%(v/v) MPD in distilled water pH 7.4 as described above.
Crystals appeared after 3–4 d using the following conditions from
the second crystallization screen: 40 mM sodium cacodylate pH 5.5,
20 mM cobalt hexammine, 80 mM sodium chloride, 20 mM magne-
sium chloride, 10%(v/v) MPD with equilibration against 35%(v/v)
MPD in distilled water pH 7.4. Optimization of crystal growth was
performed by variation of the aqueous MPD concentration in the
reservoir between 30 and 42%. The best crystals appeared within the
range 33–41%(v/v) MPD in Linbro plates using the hanging-drop
2.2. Acquisition of X-ray diffraction data and data-processing
Following our previous experience with nucleic acid crystals grown
in MPD, the crystals were directly flash-frozen in the crystallization
solution. In the presence of roughly 20%(v/v) MPD, which is the
estimated concentration in the droplet after equilibration, no further
cryoprotectant reagent was needed. X-ray diffraction data were
recorded at the Elettra Synchrotron (Trieste, Italy) on beamline
XRD1 at a wavelength of 1.000 A˚. Two data sets were collected: a
high-resolution data set in the resolution range 80.0–1.90 A˚and a
subsequent low-resolutiondata setfrom 80.0 to2.70 A˚resolution. The
crystallographic data were processed and merged using the programs
Behling et al.
? ‘All-locked’ nucleic acid duplex
Acta Cryst. (2009). F65, 809–812
DENZO and SCALEPACK from the HKL-2000 package (Otwi-
nowski & Minor, 1997). The Matthews coefficient and solvent content
was calculated according to Matthews (1968). Molecular-replacement
calculations were calculated with the program Phaser (McCoy et al.,
2005) from the CCP4 program suite (Collaborative Computational
Project, Number 4, 1994) using different nucleic acids as models, such
as an artificially constructed LNA model that was built of LNA
building blocks which correspond to the sequence of the LNA helix or
the natural tRNASermicrohelix.
3. Results and discussion
We focused on crystallizing a locked nucleic acid duplex containing
exclusively LNA nucleotides derived from a tRNASermicrohelix, the
structure of which we have solved recently (Eichert et al., 2009), in
order to perform a comparative structure analysis between LNA and
RNA. The LNA 7-mer duplex, with a base sequence corresponding to
that of the E. coli tRNASermicrohelix isoacceptor RS 1661 (Sprinzl &
Vassilenko, 2005), crystallized in 40 mM sodium cacodylate pH 5.5,
20 mM cobalt hexammine, 80 mM sodium chloride, 20 mM magne-
sium chloride, 10%(v/v) MPD with equilibration against various
concentrations of aqueous MPD. The crystal used in the measure-
ment was equilibrated against 40%(v/v) MPD. Representative crys-
tals had approximate dimensions of 0.2 ? 0.2 ? 0.1–0.05 mm and are
shown in Fig. 1.
3.2. Crystallographic data
The ‘all-LNA’ duplex (Fig. 2) crystallized in space group C2 with
two helices per asymmetric unit. We collected two data sets. The high-
resolution data contained diffraction data to 1.9 A˚resolution with
high completeness and a low R value. The low-resolution data were
collected from 80 to 2.7 A˚resolution. The data sets were merged and
the following crystallographic statistics were calculated (Table 1). The
unit-cell parameters were a = 77.91, b= 40.74, c= 30.06 A˚, ?= 91.02?,
with a Matthews coefficient of 2.48 A˚3Da?1, which corresponds to a
solvent content of 66.5% and two LNA duplexes per asymmetric unit.
The overall Rmergevalue was 7.3% (21.7% for the last resolution shell;
1.93–1.90 A˚) and the overall completeness was 98.0% (97.2% for the
highest resolution shell). The structure was solved by molecular
replacement and we are presently running refinement calculations.
By determining the structure of an ‘all-locked’ ?-d-ribofuranose
LNA homoduplex, we wish to contribute to a more detailed under-
standing of the increased thermostability of locked nucleic acids.
Such an understanding will enhance specific drug design and the
application of these molecules in diagnostics and gene therapy. The
Acta Cryst. (2009). F65, 809–812Behling et al.
? ‘All-locked’ nucleic acid duplex
Representative crystals of the ‘all-locked’ LNA duplex with the base sequence
originating from the E. coli tRNASermicrohelix. The crystals show approximate
dimensions of 0.2 ? 0.2 ? 0.1–0.05 mm. Pictures of two different setups are shown,
with equilibration using 33%(v/v) MPD (a) or 35%(v/v) MPD (b). Details are
described in the text.
(a) The aminoacyl stem from E. coli tRNASerwas used for the design of an ‘all-locked’ nucleic acid which was crystallized in this study. The sequence is presented in the figure
and the numbering corresponds to the tRNASeracceptor stem taken from the tRNA database (Sprinzl & Vassilenko, 2005). (b) A guanosine nucleotide is shown as an LNA
with the 20-O,40-C methylene-bridged ?-d-ribofuranose (indicated by an arrow; left) and shown as RNA containing the natural ribose with the 20,30-cis-diol group (right).
tRNASermicrohelix serves as a model structure for the investigata-
tion of the detailed local geometric parameters of an ‘all-locked’
nucleic acid compared with those of the natural RNA, as we have
already solved the corresponding RNA structure (Eichert et al.,
2009). A comparative study between the natural ‘all-RNA’ duplex
and the modified ‘all-LNA’ duplex should provide new insights into
LNA conformation and stability, as to our knowledge this will be the
first X-ray structure of an ‘all-locked’ nucleic acid helix. These
extremely stable modified nucleic acids have potential for use in
medical applications and gene therapy in the future (Kaur et al., 2007;
Petersen & Wengel, 2003).
The work was supported within the RiNA network for RNA
technologies by the Federal Ministry of Education and Research, the
City of Berlin and the European Regional Development Fund. We
gratefully acknowledge the Elettra synchrotron facility, Trieste, Italy
for providing beam time.
Collaborative Computational Project, Number 4 (1994). Acta Cryst. D50,
Eichert, A., Fu ¨rste, J. P., Schreiber, A., Perbandt, M., Betzel, C., Erdmann,
V. A. & Fo ¨rster, C. (2009). Biochem. Biophys. Res. Commun. 386, 368–373.
Fo ¨rster, C., Brauer, A. B. E., Brode, S., Schmidt, K. S., Perbandt, M., Meyer,
A., Rypniewski, W., Betzel, C., Kurreck, J., Fu ¨rste, J. P. & Erdmann, V. A.
(2006). Acta Cryst. F62, 665–668.
Kaur, H., Babu, B. R. & Maiti, S. (2007). Chem. Rev. 107, 4672–4697.
Kumar, R., Singh, S. K., Koshkin, A. A., Rajwanshi, V. K., Meldgaard, M. &
Wengel, J. (1998). Bioorg. Med. Chem. Lett. 8, 2219–2222.
Matthews, B. W. (1968). J. Mol. Biol. 33, 491–497.
McCoy, A. J., Grosse-Kunstleve, R. W., Storoni, L. C. & Read, R. J. (2005).
Acta Cryst. D61, 458–464.
Nielsen, K. E., Rasmussen, J., Kumar,R., Wengel, J., Jacobsen, J. P. & Petersen,
M. (2004). Bioconjug. Chem. 15, 449–457.
Nielsen, K. M., Petersen, M., Hakansson, A. E., Wengel, J. & Jacobsen, J. P.
(2002). Chemistry, 8, 3001–3009.
Obika, S., Nanbu, D., Hari, Y., Morio, K.-I., In, Y., Ishida, T. & Imanishi, T.
(1997). Tetrahedron Lett. 38, 8735–8738.
Otwinowski, Z. & Minor, W. (1997). Methods Enzymol. 276, 307–326.
Petersen, M., Bondensgaard, K., Wengel, J. & Jacobsen, J. P. (2002). J. Am.
Chem. Soc. 124, 5974–5982.
Petersen, M., Nielsen, C. B., Nielsen, K. E., Jensen, G. A., Bondensgaard, K.,
Singh, S. K., Rajwanshi, V. K., Koshkin, A. A., Dahl, B. M., Wengel, J. &
Jacobsen, J. P. (2000). J. Mol. Recognit. 13, 44–53.
Petersen, M. & Wengel, J. (2003). Trends Biotechnol. 21, 74–81.
Schmidt, K. S., Borkowski, S., Kurreck,J., Stephens, A. W., Bald, R., Hecht, M.,
Friebe, M., Dinkelborg, L. & Erdmann, V. A. (2004). Nucleic Acids Res. 32,
Sprinzl, M. & Vassilenko, K. S. (2005). Nucleic Acids Res. 33, D139–D140.
Vester, B. & Wengel, J. (2004). Biochemistry, 43, 13233–13241.
Behling et al.
? ‘All-locked’ nucleic acid duplex
Acta Cryst. (2009). F65, 809–812
X-ray diffraction data and processing statistics for the ‘all-LNA’ tRNASer
A high-resolution and a low-resolution data set were recorded and the data were merged;
details are described in the text. The statistics of the merged data are shown in the table.
Values in parentheses are for the highest resolution shell.
Unit-cell parameters (A˚,?)
a = 77.91, b = 40.74,
c = 30.06, ? = 91.02
Duplexes per ASU
Solvent content (%)
Resolution range (A˚)
is the sum over the individual measurements of a reflection with indices hkl andP
ijIiðhklÞ ? hIðhklÞij=P
iIiðhklÞ, where Ii(hkl) and hI(hkl)i are the
observed individual and mean intensities of a reflection with indices hkl, respectively,P
the sum over all reflections.