NMR Study of the d(CTLGCTLTLCTLGC):d(GCAGAAGCAG)

Department of Chemistry, University of Southern Denmark, Odense University, DK-5230 Odense M, Denmark,. Center for Synthetic Bioorganic Chemistry, ...
0 downloads 0 Views 360KB Size
228

Bioconjugate Chem. 2000, 11, 228−238

Solution Structure of an LNA Hybridized to DNA: NMR Study of the d(CTLGCTLTLCTLGC):d(GCAGAAGCAG) Duplex Containing Four Locked Nucleotides Katrine E. Nielsen,† Sanjay K. Singh,‡ Jesper Wengel,‡ and Jens Peter Jacobsen*,† Department of Chemistry, University of Southern Denmark, Odense University, DK-5230 Odense M, Denmark, Center for Synthetic Bioorganic Chemistry, Department of Chemistry, University of Copenhagen, Universitetsparken 5, DK-2100 Copenhagen, Denmark. Received September 15, 1999; Revised Manuscript Received December 3, 1999

We have used two-dimensional 1H NMR spectroscopy at 750 MHz to determine a high-resolution solution structure of an oligonucleotide containing restricted nucleotides with a 2′-O, 4′-C-methylene bridge (LNA) hybridized to the complementary DNA strand. The LNA:DNA duplex examined contained four thymidine LNA modifications (TL), d(C1TL2G3C4TL5TL6C7TL8G9C10):d(G11C12A13G14A15A16G17C18A19G20). A total relaxation matrix approach was used to obtain interproton distance bounds from NOESY cross-peak intensities. These distance bounds were used as restraints in molecular dynamics (rMD) calculations. Forty final structures were generated for the duplex from A-form and B-form DNA starting structures. The root-mean-square deviation (RMSD) of the coordinates for the 40 structures of the complex was 0.6 Å. The sugar puckerings are averaged values of a dynamic interchange between N- and S-type conformation except in case of the locked nucleotides that were found to be fixed in the C3′-endo conformation. Among the other nucleotides in the modified strand, the furanose ring of C7 and G9 is predominatly in the N-type conformation whereas that of G3 is in a mixed conformation. The furanose rings of the nucleotides in the unmodified complementary strand are almost exclusively in the S-type conformation. Due to these different conformations of the sugars in the two strands, there is a structural strain between the A-type modified strand and the B-type unmodified complementary strand. This strain is relaxed by decreasing the value of rise and compensating with tip, buckle, and propeller twist. The values of twist vary along the strand but for a majority of the base pairs a value even lower than that of A-DNA is observed. The average twist over the sequence is 32 ( 1°. On the basis of the structure, we conclude that the high stability of LNA:DNA duplexes is caused by a local change of the phosphate backbone geometry that favors a higher degree of stacking.

INTRODUCTION

Scheme 1

A potential nucleotide-based gene inhibitor must possess an enhanced stability toward cellular nucleases, the ability to penetrate the cell membrane, and an efficient hybridization to the target RNA/DNA1 (He´le`ne and Toulme´, 1990; Uhlmann and Peyman, 1990). Several chemically modified DNA and RNA oligonucleotides have been synthesized, but only a few of them fulfill the requirements (He´le`ne and Toulme´, 1990; Uhlmann and Peyman, 1990; Nielsen, 1991; He´le`ne, 1991; Thuong and He´le`ne, 1993; De Mesmaeker et al., 1995). The LNA * To whom correspondence should be addressed. E-mail: [email protected]. † University of Southern Denmark. ‡ University of Copenhagen. 1 Abbreviations: AMBER, assisted model building with energy refinement; DQF-COSY, double quantum filtered correlation spectroscopy; dsDNA, double-stranded DNA; ssDNA, singlestranded DNA; DSS, 2,2-dimethyl-2-silapentane-5-sulfonate; E-BURP, excitation band-selective pulse with uniform response and pure phase; EDTA, ethylenediaminetetraacetic acid; LNA, locked nucleic acid; MARDIGRAS, matrix analysis of relaxation for discerning the geometry of an aqueous structure; NOE, nuclear Overhauser effect; NOESY, nuclear Overhauser effect spectroscopy; RANDMARDI, randomized MARDIGRAS; rMD, restrained molecular dynamics; RMSD, root-mean-square deviation; TOCSY, total correlation spectroscopy.

nucleotide monomer (Scheme 1) is a conformational restricted nucleotide analogue containing a 2′-O, 4′-Cmethylene bridge locked in an 3′-endo conformation (Singh et al., 1998; Singh and Wengel, 1998; Obika et al., 1997; Wengel, 1998; Koshkin et al., 1998ab,c). Ther-

10.1021/bc990121s CCC: $19.00 © 2000 American Chemical Society Published on Web 02/23/2000

Structure of an LNA:DNA Duplex by NMR

mal denaturation studies of duplexes formed between oligonucleotides containing the LNA modification and complementary DNA and RNA show that LNA recognizes both complementary DNA and RNA with remarkable thermal affinities. Thus, compared to the corresponding unmodified reference duplexes, the duplexes involving LNA (hybridized toward DNA or RNA) display unprecedented melting temperatures with increases between +4.0 and +9.3 °C obtained per modification (Singh et al., 1998; Singh and Wengel, 1998; Wengel, 1998; Koshkin et al., 1998a), and incorporation of a given number of LNA monomers into an oligonucleotide appears to be a very convenient and predictable way of upgrading the stability of duplexes toward complementary DNA or RNA. The LNA:LNA base pairing constitutes the most stable modified nucleic acid type duplex system hitherto discovered. Thus, melting temperatures greater than 93 °C have been observed for duplexes between two allmodified complementary LNAs (Koshkin et al., 1998c). Preliminary experiments (Wengel et al., 1999) suggest the LNA molecule so far to be the most promising candidate for efficient recognition of a given mixed sequence in a nucleic acid duplex. These results, togther with the demonstration that LNA is stable toward 3′exonucleolytic degradation (Singh et al., 1998) and that fluorescein-labeled LNA can be transferred into living MCF-7 breast cancer cells (Wengel et al., 1999), have stimulated the evaluation of LNA as antisense and/or antigene molecule. To obtain a detailed understanding of the foundation for the unprecedented enhanced thermal stability of LNA modified duplexes, we are currently investigating a series of LNA:DNA and LNA:RNA oligonucleotides using NMR spectroscopy. Recently, we reported the sugar conformations of a partly modified 9-mer single-stranded LNA as well as three 9- and 10-mer partly modified LNAs hybridized to unmodified DNA (Petersen et al., 1999). By use of selective DQF-COSY NMR spectra, we determined the ratio between the N-type (C3′-endo) and S-type (C2′endo) sugar conformations of the nucleotides. We found that the locked conformation of the LNA nucleotides both in ssLNA and in the duplexes organize the phosphate backbone in such a way as to introduce higher population of the N-type conformation of the neighboring unmodified nucleotides on the same strand. In general, the conformations of the flexible deoxyribose rings determine the overall structure of a (deoxy)ribo nucleic acid duplex. Duplexes in A-type conformation contain nucleotides with an N-type sugar conformations, while B-type duplexes contain nucleotides with an S-type sugar conformations (Dickerson et al., 1989; Saenger, 1984). In the attempt to investigate these structural properties of LNA:DNA duplexes, we have used 2D 1H NMR spectroscopy to determine the solution structure of the d(CTLGCTLTLCTLGC): d(GCAGAAGCAG) duplex (called the “LNA duplex” in the following) containing four thymidine LNA modifications as indicated by TL. A total relaxation matrix approach was used to obtain interproton distance bounds from NOESY cross-peak intensities (Borgias and James, 1990; Borgias et al., 1990; Liu et al., 1995). These distance bounds were used as restraints in molecular dynamics (rMD) calculations. Since many NOE contacts were observed, the resulting structure has high resolution and allowed the determination of local conformational features in the LNA duplex. EXPERIMENTAL PROCEDURES

Sample Preparation. The d(CTLGCTLTLCTLGC) oligonucleotide was synthesized as described elsewhere

Bioconjugate Chem., Vol. 11, No. 2, 2000 229

(Singh et al., 1998; Petersen et al., 1999; Koshkin et al., 1998a). The unmodified oligonucleotides were purchased from DNA Technology, Århus, Denmark. The oligonucleotides were purified by site exclusion on a Sephadex G15 column. The duplex was obtained by dissolving an equimolar amount of the two single strands in 0.5 mL of 10 mM sodium phosphate buffer (pH 7.0), 0.05 mM NaEDTA, 0.01 mM NaN3, and 0.1 mM DSS. The mixture was heated to 80 °C and slowly cooled to achieve hybridization. The numbering scheme used for the duplex was as follows:

For experiments carried out in D2O, the solid duplex was lyophilized three times from D2O and redissolved in 99.96% D2O (Cambridge Isotope Laboratories). A mixture of 90% H2O and 10% D2O (0.5 mL) was used for experiments examining exchangeable protons. The final concentration of the duplex was 2 mM. A similar sample was made of the unmodified duplex. NMR Experiments. NMR experiments were performed on a Varian UNITY 500 spectrometer or a Varian INOVA 750 spectrometer at 27 °C unless otherwise stated. NOESY spectra of the duplex at 500 MHz were acquired in D2O using 1024 complex points in t2 and a spectral width of 5000 Hz. NOESY spectra at 750 MHz were acquired using 2048 complex points in t2 and a spectral width of 7500 Hz. A total of 512 t1 experiments were recorded using the States phase cycling scheme. The residual signal from HOD was removed by low-power presaturation. The NOESY spectra in H2O at 750 MHz were acquired with a spectral width of 15.000 Hz using 4096 complex points and the WATERGATE pulse sequence (Stein et al., 1995). TOCSY spectra with mixing times of 90 ms were obtained in the TPPI mode at 500 MHz. An inversion recovery experiment to estimate the T1 relaxation rates was also obtained at 750 MHz. The NOE buildups were obtained at 750 MHz. These NOESY spectra were acquired in D2O using 2048 complex points in t2 and a spectral width of 7500 Hz. A total of 512 t1 experiments of 64 scan each was recorded using the States phase cycling scheme. NOESY spectra with mixing times of 50, 100, 150, and 200 ms were obtained sequentially without removing the sample from the magnet. 64 scans were acquired for each t1 value with a repetition delay of 4.0 s between each scan. The residual signal from HOD was removed by low-power presaturation. The acquired data were processed using FELIX (version 97.2, MSI, San Diego, Ca). All spectra were apodized by a skewed sine bell squared in F1 and F2. The four spectra used in the NOE buildups were zero-filled in F1 to 2K points and baseline corrected by a second-order polynomial fit in F2 and F1. The NOESY spectrum recorded in H2O was zero-filled in F1 to 4K points and baseline corrected by a second-order polynomial fit in F2 and F1. Distance Restraints. The upper and lower diagonal part of each of the four NOESY buildup spectra were integrated separately with FELIX, yielding a total of eight peak intensity sets that were corrected for minor saturation effects. The RANDMARDI procedure (Liu et al., 1995) of the complete relaxation matrix analysis method MARDIGRAS (Borgias and James, 1990; Borgias et al., 1990) was used to calculate interproton distance

230 Bioconjugate Chem., Vol. 11, No. 2, 2000

Nielsen et al.

bounds from the eight intensity sets. In the calculations, an absolute noise level of the same order of magnitude as the smallest integrated cross-peak was used. A relative noise level for each integrated peak was set at 5% of the integrated peak intensity. The dynamic range of observed cross-peak intensities at a given mixing time was 1000. In the RANDMARDI procedure, 30 different intensity sets were generated from each experimental data set based on the given noise levels, and MARDIGRAS calculations were performed on all of them. Resulting distances from all 8 × 30 intensity sets were combined into a single bounds file from which the rMD restraint file was generated. Upper and lower bounds in the bounds file were average interproton distances ( one standard deviation calculated from all of the MARDIGRAS runs. An additional 0.2 Å was added to the upper bounds. Restrained Molecular Dynamics. The obtained distance restraints were incorporated into an rMD procedure using DISCOVER (version 3.0, MSI, San Diego, CA) with the AMBER force field. The two starting models for structure refinement were either A-form or B-form DNA built in INSIGHTII (version 97, MSI, San Diego, CA), modified to include the LNA modification and followed by minimization using DISCOVER. All the distance restraints were incorporated into an rMD procedure for structure refinement. An initial energy minimization was followed by 28 ps of restrained molecular dynamics at 600 K for 4 ps, followed by cooling to 200 K in 50 K steps of 3 ps each. The final structure was then energy minimized to a maximum derivative of 0.01 Å. A distance-dependent dielectric constant ( ) r) was employed to account for solvent effects. The pseudo-energy term used to enforce the distance restraints was

{

k1 (r - r1)2 when r1 > r

ENOE ) 0 k2 (r - r2)

when r2 g r g r1 2

when r > r2

where r1 and r2 are the lower and upper distance bounds determined from cross-peak intensities and k1 and k2 are the force constants. A force constant of 50 kcal mol-1 Å-2 for lower and upper bounds was assigned to all distance restraints. DISCOVER uses a limit on the magnitude a distance restraint can contribute to the force field. This limit was set to 1000 kcal mol-1 Å-1. Helix parameters were calculated with the program CURVES 5.1 (Lavery and Sklenar, 1988; Lavery and Sklenar, 1989). RESULTS

Spectral Analysis. The 1D 1H NMR spectrum of the sample of the modified LNA:DNA duplex exhibits only lines from the expected LNA:DNA duplex. No signs of alternative hybridization were observed. The NOESY spectra of the LNA duplex exhibit the characteristic features of dsDNA sequential connectivities. The assignment of the nonexchangeable protons in the modified duplex was performed using standard methods (Wu¨thrich, 1986; Hare et al., 1983; Scheek et al., 1983; Feigon et al., 1983). Aromatic (H6, H5, H8, and H2) and sugar protons (H1′,H2′, H2′′,H3′, H4′, and H5′,H5′′) were assigned. The chemical shift values and the NOE connectivity pattern of the imino protons were observed to be in accordance with normal Watson-Crick base pairing. The NOESY spectrum with short mixing time (50 ms) allowed unambiguous assignments of the H2′ and

Figure 1. Parts of the 200 ms NOESY spectrum of the d(CTLGCTLTLCTLGC):d(GCAGAAGCAG) duplex showing the sequential connectivity pattern H6/8 and H2′ on the modified strand. Some of the strong cross-peaks in the lower region are between H3′ and H6/8 indicative of the C3′-endo conformations.

H2′′ resonances. The assignment of the exchangeable protons was obtained from the NOESY spectrum in H2O (Boelens et al., 1985). The chemical shift values are given in Table 1 and compared to the corresponding values of the unmodified duplex. Structure Calculations. More than 800 NOE crosspeaks were observed in the NOESY spectrum obtained with a mixing time of 200 ms. The accuracy of the integration of some cross-peaks was hampered by spectral overlap. These cross-peaks were therefore not included in the MARDIGRAS calculations. Integrations of the cross-peaks in the four NOESY spectra used were done separately for each side of the diagonal. This resulted in four sets of 325 NOE cross-peak intensities and four sets of 310 NOE cross-peak intensities, giving a total of 2540 integrated cross-peak intensities used in the RANDMARDI calculations. Some of the cross-peaks used in the structure calculations resulted predominantly from spin diffusion. This is taken into account during the statistical analysis of the results generated by the

Structure of an LNA:DNA Duplex by NMR

Bioconjugate Chem., Vol. 11, No. 2, 2000 231

Table 1. Chemical Shift Values (ppm) in the d(CTLGCTLTLCTLGC):d(GCAGAAGCAG) Duplexa C1 TL2 G3 C4 TL5 TL6 C7 TL8 G9 C10 G11 C12 A13 G14 A15 A16 G17 C18 A19 G20

H1′

H2′

H2′′

H3′

H4′

H5′

H5′′

H6/H8

H5/H2/CH3

5.97 (5.95) 5.43 (5.84) 6.08 (5.92) 5.96 (5.96) 5.35 (6.04) 5.52 (6.08) 5.69 (5.97) 5.28 (5.71) 6.04 (5.95) 6.15 (6.20) 6.04 (5.97) 6.02 (5.55) 6.20 (5.95) 5.72 (5.38) 6.13 (5.92) 6.02 (5.94) 5.84 (5.68) 5.83 (5.50) 6.08 (6.04) 6.07 (6.04)

2.59 (2.14) 4.95 (2.19) 2.62 (2.53) 2.54 (2.16) 5.19 (2.17) 5.09 (2.43) 2.73 (2.11) 5.03 (2.07) 2.40 (2.65) 2.14 (2.19) 2.74 (2.60) 2.42 (2.04) 2.73 (2.71) 2.46 (2.54) 2.64 (2.60) 2.57 (2.56) 2.43 (2.40) 2.10 (1.85) 2.62 (2.68) 2.37 (2.45)

2.59 (2.56) (2.53) 2.70 (2.73) 2.54 (2.55) (2.60) (2.57) 2.52 (2.48) (2.41) 2.60 (2.73) 2.24 (2.19) 2.81 (2.78) 2.65 (2.36) 2.92 (2.86) 2.67 (2.66) 2.90 (2.86) 2.75 (2.80) 2.63 (2.58) 2.47 (2.24) 2.83 (2.85) 2.27 (2.29)

4.71 (4.69) 4.67 (4.88) 4.81 (5.01) 4.76 (4.72) 4.52 (4.86) 4.58 (4.90) 4.59 (4.81) 4.66 (4.87) 4.79 (4.99) 4.45 (4.49) 4.84 (4.84) 4.89 (4.86) 5.00 (5.05) 4.90 (4.99) 4.98 (5.03) 4.97 (5.02) 4.89 (4.91) 4.83 (4.78) 4.98 (5.00) 4.61 (4.63)

4.13 (4.12) (4.20) 4.37 (4.40) 4.26 (4.27) (4.20) (4.21) 4.21 (4.20) (4.14) 4.21 (4.37) 4.02 (4.06) 4.26 (4.23) 4.32 (4.16) 4.41 (4.38) 4.35 (4.34) 4.43 (4.42) 4.40 (4.40) 4.37 (4.34) 4.21 (4.09) 4.37 (4.38) 4.18 (4.18)

3.80 (3.80) 4.36 (-) 4.33 (4.16) 4.14 (-) 4.27 (-) 4.29 (-) 4.30 (-) 4.31 (-) 4.11 (4.12) 4.15 (-) 3.81 (3.71) 4.23 (-) 4.22 (4.15) 4.23 (4.16) 4.24 (4.19) 4.24 (4.20) 4.21 (4.18) 4.09 (-) 4.09 (4.00) 4.13 (4.24)

3.83 (3.80) 4.31 (-) 4.15 (4.09) 4.19 (-) 4.27 (-) 4.23 (-) 4.11 (-) 4.21 (-) 4.31 (4.09) 4.15 (-) 3.81 (3.71) 4.23 (-) 4.25 (4.00) 4.23 (4.16) 4.24 (4.19) 4.24 (4.20) 4.21 (4.18) 4.09 (-) 4.17 (4.10) 4.22 (4.12)

7.86 (7.83) 7.39 (7.48) 7.94 (7.93) 7.43 (7.44) 7.31 (7.48) 7.54 (7.47) 8.00 (7.60) 7.42 (7.33) 7.89 (7.92) 7.63 (7.48) 8.03 (7.95) 7.61 (7.45) 8.12 (8.18) 7.33 (7.69) 7.89 (8.04) 7.77 (7.98) 7.43 (7.53) 7.34 (7.26) 8.10 (8.16) 7.58 (7.71)

5.93 (5.93) 1.70 (1.70) 5.25 (5.34) 1.56 (1.62) 1.64 (1.68) 5.63 (5.65) 1.61 (1.69) 5.43 (5.52)

H1/H3

H4

H4

8.05 (-)

7.08 (-)

8.35 (8.09)

6.73 (6.50)

8.31 (8.39)

6.81 (6.93)

8.23 (8.15)

6.90 (6.70)

8.65 (8.50)

6.75 (6.52)

8.40 (8.26)

6.38 (6.26)

13.81 (-) 12.66 (12.73) 14.26 (14.12) 13.91 (13.74) 13.54 (13.90) 12.73 (12.91) (-) (-)

5.42 (5.45) 7.12 (7.63) 12.46 (12.51) 7.07 (7.24) 7.30 (7.48) 12.73 (12.77) 5.15 (5.25) 7.43 (7.81)

(-)

a The value of the unmodified duplex are given in parentheses. The values are given at 25 °C relative to DSS. The protons in the C2′,C4′ linker have the following chemical shift values: TL2 H6′/H6′′, 4.00/4.15; TL5 H6′/H6′′, 4.00/4.10; TL6 H6′/H6′′, 4.02/4.10; TL8 H6′/H6′′, 3.93/4.09.

RANDMARDI procedure (Liu et al., 1995). Cross-peak integrals that corresponded to fixed distances in the dsDNA were used for internal calibrations in MARDIGRAS and, therefore, not converted into distance restraints for use in the rMD simulations. The MARDIGRAS calculations were performed by using a value of the correlation time of τc ) 2.5 ns. This value was justified from earlier work (Spielmann et al., 1995; Gotfredsen et al., 1996; Petersen and Jacobsen, 1998; Johansen and Jacobsen, 1998). Furthermore, it turns out that the results of the calculations are rather insensitive to the exact values of τc used. Calculation with τc ) 1.0 ns and τc ) 3.75 ns yielded exactly the same structure but with slightly larger RMSD deviations This is similar to what was found earlier (Mujeeb et al., 1993; Gotfredsen et al., 1996). The RANDMARDI calculations returned 339 interproton distances that are not covalently fixed in the modified DNA. The distance bounds used in RMD simulations were determined by combining the results from all of the individual MARDIGRAS calculations performed during the RANDMARDI procedure into one set. They were calculated individually for each proton pair corresponding to a NOESY cross-peak included in the RANDMARDI calculations. Each individual restraint was generated from the MARDIGRAS calculations. Table 2 gives a summary of the distance restraints used in the RMD calculations. The range of differences between the upper

and lower bounds for the 339 NOE-derived restraint bounds used (Table 2) was between 0.24 and 2.64 Å with an average flat well potential width of 0.77 Å. This range between the upper and lower bounds represents the precision of the restraint calculated from the NOESY data. Weak NOESY cross-peaks that are primarily from spin diffusion had the largest distance between the upper and lower distance bounds. Evaluation of the spectra obtained in H2O shows normal Watson-Crick hydrogen bonding, justifying the inclusion of 26 hydrogen bond distance restraints. Three hydrogen bonds were included for each of the six GC base pairs, and two hydrogen bonds were included for each of the four AT base pairs with upper and lower bounds of 1.74 and 2.10 Å, respectively. An additional 24 restraints with loose bounds involving labile protons were derived using the isolated spin pair approximation on the NOESY spectra recorded of the sample in H2O. Thus, a total of 389 interproton restraints was used in the RMD calculations. Twenty final structures for each of the two starting structures were generated. All the structures converged to one family of conformations. The average all atom rootmean-square (RMS) deviation of the coordinates of the 40 structures was 0.6 Å. The average sum of violations was 10.48 Å for the A-form starting structures and 10.09 Å for the B-form starting structures. We observed only six to eight violations greater than 0.2 Å and none greater

232 Bioconjugate Chem., Vol. 11, No. 2, 2000

Nielsen et al.

Table 2. Number of Distance Restraints Used in the rMD Calculations of the d(CTLGCTLTLCTLGC):d(GCAGAAGCAG) Duplexa intraresidue C1 TL2 G3 C4 TL5 TL6 C7 TL8 G9 C10 total

7 16 10 3 13 14 12 16 5 13 109

interresidue C1-TL2 TL2-G3 G3-C4 C4-TL5 TL5-TL6 TL6-C7 C7-TL8 TL8-G9 G9-C10

intraresidue 7 6 10 6 10 10 13 6 9

G11 C12 A13 G14 A15 A16 G17 C18 A19 G20

77

10 9 11 5 10 8 7 10 9 11 90

interresidue G11-C12 C12-A13 A13-G14 G14-A15 A15-A16 A16-G17 G17-C18 C18-A19 A19-G20

interstrand 7 5 9 5 6 3 3 5 6 49

TL2-A19 TL6-A15 TL6-A16 TL 8-A13 TL8-A15 C7-A15 G9-A13

1 2 2 2 1 3 3

14

a

The restraints were derived from MARDIGRAS calculations unless otherwise stated. The NOESY spectrum in H2O gave rise to 24 loose distance restraint of which 14 were interstrand distance restraints. The Watson-Crick base pairing were ensured by further including 26 distance restraints in the rMD calculations.

Figure 2. Stereoview of a superposition of stick plots of 20 calculated structures of the d(CTLGCTLTLCTLGC):d(GCAGAAGCAG) duplex. Deoxyribose protons have been omitted for clarity. Ten structure were randomly picked from each of the two sets of structures obtained from A- and B-DNA starting structures, respectively.

than 0.26 Å. Different views of the structure are presented in Figures 2 and 3. The reliability of the structure calculations was carefully evaluated. rMD calculations were performed where the force constants of the distance restraints changed to 25 and 100 kcal mol-1 Å-2, respectively. All the calculations generated structures that converged to the same family of structures as with the force constants of 50 kcal mol-1 Å-2. A lower force constant resulted in a higher RMSD and more violations. The structure obtained is not strongly dependent on the exclusion of any of the restraints. In particular, calculations performed without the Watson-Crick restraints yielded exactly the same structure but with slightly larger RMSD. Helix Parameters. Helical parameters for the 40 final structures were analyzed with the program CURVES 5.1 (Lavery and Sklenar, 1988; Lavery and Sklenar, 1989). Plots of some global helical parameters for the LNA duplex are shown in Figure 4. DISCUSSION

Spectral Data. The chemical shift values given in Table 1 demonstrate close agreement between the modified and unmodified duplex except for the protons near

the modification sites. As expected from the sugar conformation, the protons in the LNA nucleotides differ from those in the unmodified nucleotide. The protons on the other nucleotides show differences in chemical shift values between the modified and the unmodified duplex primarily at the A13, A19, C7, and G14 nucleotides. The largest effect is observed for the adenine H2 protons, especially those of A13 and A19 with downfield shifts of 0.51 and 0.38 ppm, respectively. A large effect is also observed at some of the other aromatic protons, e.g., C7H6 and G14H8, with an upfield shift of 0.40 ppm and a downfield shift of 0.36, respectively. The chemical shift values of the other aromatic proton are changed more moderately. The pronounced shifts of the resonances of the aromatic protons are most probably related to an enhanced stacking of the nucleobases. It is noteworthy that this effect is observed both on the base paired to the LNA nucleotide as well as the base neighboring the LNA nucleotide on the same strand. The changes of the chemical shift of the sugar protons of the modified nucleotides are clearly related to the chemical modification introduced whereas those of the unmodified nucleotides may, at least partly, be related to conformational changes of the deoxyribose ring.

Structure of an LNA:DNA Duplex by NMR

Figure 3. Basepair stacking in the d(CTLGCTLTLCTLGC): d(GCAGAAGCAG) duplex viewed from the minor groove side. The bases are shown in green and the sugar backbone in blue. The 2′-O,4′-C bridge is indicated in red. The (TL2-G3-C4):(G17C18-A19) step are shown at the bottom, the (C4-TL5-TL6):(G14A15-A16) in the middle and the (TL6-C7-TL8):(A13-G14-A15) at the top.

Description of the Structure. Qualitatively, the NOESY spectra of the LNA duplex indicate that it adopts a right-handed helix conformation. All bases are in the anti conformation and they all form normal WatsonCrick base pairs. Recently, we reported the use of a selective DQF-COSY spectrum to determine the sugar conformations (Petersen et al., 1999). The LNA nucleotides are locked in a C3′-endo (N-type) conformation. The unmodified nucleotide C7 is also exclusively in the N-type conformation whereas G3 and G9 are in mixed conformations with substantial N-type contributions. The nucleotides in the complementary strand are predominantly in the normal S-type configuration. This mixed situation forced us not to include any dihedral angle restraints in the rMD calculations. Doing that, it would have been necessary to include a description of the dynamic ring puckering motions of the sugars. Such a description is not available. Experimentally derived NMR distance restraints are distributed anisotropically and are short range in nature. This largely establishes local structures with some helical parameters better defined by the experimental data than others. We have used a number of NMR derived inter-

Bioconjugate Chem., Vol. 11, No. 2, 2000 233

strand restraints in the center of the helix originating from AH2 protons to H1′ protons and a number of loose bounds interstrand distance restraints that improved the ability to define the overall structure of the oligonucleotide (Weisz et al., 1994). We have shown representations of the structure of the duplex in Figures 2 and 3. Figure 2 demonstrates the high accuracy of this structure determination. The colored graphic representation in Figure 3 shows a picture of the base pairing and stacking of the bases close to the modification. Helical Parameters. There are three major categories of helical parameters: axis-base pair, intra-base pair, and inter-base pair parameters (Dickerson et al., 1989). These parameters for the duplex are shown in Figure 4. However, the parameters reported for the terminal base pairs of the complex are probably less reliable because of dynamic processes and a lower density of restraints and are only included for completeness. The global helix axis is slightly bent as shown in Figure 5. The width of the minor groove measured as the shortest interstrand phosphorus distance across the groove is included in Figure 6. Although local narrowing of the minor groove occurs, the width of the minor groove does not deviate much from what is observed in unmodified DNA duplexes. Values of the pseudorotational angles obtained from CURVES are averaged values of a dynamic interchange between N- and S-type conformation. The values for TL2, TL5, TL6, and TL8 were found by CURVES to be very close to 17° as reported earlier (Obika et al., 1997). The values obtained for the other nucleotides on the modified strand indicate that C7 and G9 are predominantly in the N-type conformation, whereas G3 is in a mixed conformation. The unmodified complementary strand was found to be almost exclusively in the S-type conformation. As mentioned earlier, structure calculation using NOE restraints do not include the necessary description of the dynamic ring puckering motion of the sugar. This implies that it is not expected that the values for the pseudorotation angle obtained from the CURVES analysis agree completely with the values obtained from the conformational study reported earlier (Petersen et al., 1999). However, we do find from the CURVES analysis that the obtained structure reproduces very well the values for the pseudorotation angles obtained earlier (Petersen et al., 1999). Axis-Base Pair Parameters. The helix parameters Xdisp, Ydisp, tip, and inclination deviate significantly from the values in B-DNA (Baleja et al., 1990; Mujeeb et al., 1993; Schmitz et al., 1992; Weisz et al., 1994). The values of Xdisp and tip show that the duplex is between B-type and an A-type DNA. There is a significant trend in the variation of the values of Ydisp and inclination along the strand. This is related to the slightly bended helix axis. Intra-Base Pair Parameters. The calculated values for shear, stretch, stagger, and opening do not show much deviation from B-type DNA. The buckle changes significantly along the strand. Large changes from -25 to 20° occur. This is connected to the change in sugar conformation of the nucleotides on the modified strand. The structural strain between the modified strand, having sugar conformations partly of the N-type, and the complementary strand, having almost exclusively S-type sugar conformations, is relaxed by large variation of values of the buckle. In conjunction with that, the values of propeller twist do also deviate from both A-type and B-type DNA. Especially, the large value of the propeller twist for the base pair TL5:A16 is noteworthy.

234 Bioconjugate Chem., Vol. 11, No. 2, 2000

Nielsen et al.

Structure of an LNA:DNA Duplex by NMR

Bioconjugate Chem., Vol. 11, No. 2, 2000 235

Figure 4. Helical parameters for the d(CTLGCTLTLCTLGC):d(GCAGAAGCAG) duplex (b) calculated using CURVES and compared with canonical A-DNA (- - -) and canonical B-DNA (ss).

Inter-Base Pair Parameters. The values for rise are markedly different in canonical B-DNA (3.4 Å) and A-DNA (2.6 Å). The values of rise in the LNA duplex vary from below 2.6 to 3.4 Å, but the average rise is below 2.8 Å, which is closer to the canonical A-type than to the canonical B-type DNA. This indicates increased base stacking and is connected to the sugar conformations of the modified strand that has much more N-type conformation than in normal B-DNA. However, since the sugar conformations on the base-paired nucleotides at the complementary strand are kept in the S-type conformation, the resulting strain is relaxed by decreasing the value of rise and compensated with tip, buckle, and propeller twist as described. The values of twist differ along the strand, but for a majority of the base pairs, a value even lower than that of A-DNA is observed. The average twist over the sequence is 32 ( 1°. This value is close to the one in the canonical A-type DNA. This shows in particular that the modification does not cause the helix to unwind. The roll, tilt, slide, and shift parameters do not deviate much from neither B- nor A-DNA and are within values normally observed in both A-type and B-type DNA (Mujeeb et al., 1993). Comparison with Related Structures. An impressive number of modified nucleotides for use in antisense or antigene strategy has appeared. Some of them have

been studied by structural techniques, but only a few high-resolution structures have been obtained of sugarmodified oligonucleotides, among these are the 2′-Omethylated RNA analogues. The substitution of one 2′O-CH3-modified ribose nucleotide into a DNA duplex has been studied by X-ray crystallography (Lubini et al., 1994). It shows that the presence of one 2′-O-CH3 ribonucleotide is sufficient to induce local transition from B-type DNA to A-type DNA with the 2′-O-CH3 group pointing into the minor groove. Blommers and co-workers used a fully modified 2′-O-CH3 ribonucleotide strand instead of an RNA strand to evaluate the effect of a single amide backbone modification in a complementary DNA strand (Blommers et al., 1994). Recently, a high-resolution structure of an all 2′-O-CH3-modified duplex 2′-OMe(CGCGCG)2 was published (Propenda et al., 1997; Adamiak et al., 1997). In this structure all residues have N-type sugar pucker with a very small rise and a high helical twist and inclination. Although, it may be somewhat difficult to justify a comparison between 2′-O-CH3modified oligonucleotides and the LNA duplex reported here, there is definitely the same trend of the structures in shifting from a B-type toward an A-type. Recently, we reported (Nielsen et al., 1999) the NMR structure of the modified d(CCGCTLAGCG): d(CGCTAGCGG) containing one modified LNA nucleotide. This

236 Bioconjugate Chem., Vol. 11, No. 2, 2000

Nielsen et al.

Figure 5. A schematic representation of the structure of the d(CTLGCTLTLCTLGC):d(GCAGAAGCAG) duplex showing the position of the helix axis as calculated by CURVES. To the left, the structure is viewed from the direction at which the helix axis is most straight. To the right, the structure has been turned 90°.

Figure 6. The structure of the d(CTLGCTLTLCTLGC):d(GCAGAAGCAG) duplex presented as a ribbon along the sugar backbone. The distances between the phosphate groups perpendicular to the two strand are indicated. The width of the minor groove is traditionally give by this distance reduced by 5.8 Å to account for the van der Waals’ radius.

structure was found to be close to a B-type helix, but shows some deviation from canonical B-DNA in the vicinity of the modification. These changes were most pronounced for the opening, buckle, tip rise, and twist completely in analogy to the one observed in the present study, but to a much lesser extent. The average twist observed for the d(CCGCTLAGCG): d(CGCTAGCGG) oligonucleotide was 35.0 ( 0.3°, which is close to the value in the canonical B-type DNA. A comparison between the d(CCGCTLAGCG):d(CGCTAGCGG) and d(CTLGCTLTLCTLGC):d(GCAGAAGCAG) duplexes shows that the small structural changes from B-type DNA toward A-type DNA in the former oligonucleotide con-

taining one modification have further increased in the latter oligonucleotide containing four modifications. Stability of LNA:DNA Duplexes. The stability of a DNA duplex can thermodynamically be described by use of Gibbs, free energy for duplex formation: ∆G ) ∆H T∆S. Searle and Williams (Searle and Williams, 1993) have in relation to that discussed the stability of nucleic acid structures in solution in terms of enthalpy-entropy compensations. Following their outline the effect of introducing a modification can be explained in terms of a change of the free energy of the “internal rotors” of the phosphate backbone (∆Gr) and the free energy related to the base stacking (∆Gs). The change of the free energy of the base stacking is dominated by an enthalpy change (∆Gs ≈ ∆Hs) whereas the change of the free energy of the internal rotors of the backbone is dominated by entropy (∆Gr ≈ ∆Sr) (Searle and Williams, 1993). Earlier, we found that the conformations of the nucleotides in the ssLNA d(CTLGATLATLGC) had larger amounts of N-type conformation than in the unmodified ssDNA and that this ssLNA oligonucleotide was preorganized for the formation of the d(CTLGATLATLGC):d(GCATATCAG) duplex (Nielsen et al., 1999). This implies that a smaller change of backbone entropy (∆Sr) has to be paid in the duplex formation. In this paper, we show that the organization of the backbone and the sugar rings into a C3′-endo fashion favors a more efficient stacking. This is equivalent to the structural changes from B-DNA (C2′endo conformation) to A-DNA (C3′-endo conformation) (Saenger, 1984). Thus, in LNA modified duplexes there is a local change of the phosphate backbone geometry that favors a higher degree of stacking and hence an increased value of -∆Hs. A straightforward consequence of the explanation given above is that the increase in the stability of LNA modified LNA:DNA duplexes should saturate as the number of modifications increase. This is exactly the observation. In the d(GTLGATLATLGC):d(GCATATCAC) duplex the melting temperature is increased by 5.3 °C per modification compared to the unmodified DNA duplex whereas in the fully modified d(GLTLGLALTLALTLGLmeCL):d(GCATATCAC) duplex the increase in the melting

Structure of an LNA:DNA Duplex by NMR

temperature was 4.5 °C per modification (Singh et al., 1998; Koshkin et al., 1998a). Thus, the largest effect per modification is observed in duplexes that are not fully modified. The combined contribution to the stability of the LNA: DNA duplexes from both preorganization and improved stacking was further investigated by the synthesis of an abasic LNA nucleotide and the inclusion of this abasic modified nucleotide in an oligonucleotide (Kværnø and Wengel, 1999). From melting temperature measurement, it was concluded that the base stacking are of crucial importance for the stability of LNA:DNA and LNA:RNA duplexes. It seems difficult to specify the amount of stabilization that arises from the ordering of the phosphate backbone and the part caused by increased stacking. Breslauer et al. (1986) have determined the nearest neighbor thermodynamic parameters related to the base pairing in DNA duplex formation and used these values to predict duplex stability. They found very convincing agreement between calculated values and experimentally determined ones. The values of ∆G°, ∆H°, and ∆S° for duplex formation of two nearest neighboring base pairs are very sequence specific with value of -∆G° between 4 and 15 kJ/mol, -∆H° between 23 and 50 kJ/mol, and -∆S° between 54 and 116 J/mol. Within these limits, the stability of LNA:DNA duplexes is easily explained by a slightly increased value of -∆H° and a slightly decreased value of -∆S°. ACKNOWLEDGMENT

This work was supported by grants from The Danish Natural Science Research Council and The Danish Technical Research Council. We thank The Instrument Center for NMR-Spectroscopy of Biological Macromolecules at The Carlsberg Laboratory, Copenhagen for providing spectrometer time at the 750 MHz spectrometer. Ms Britta M. Dahl is thanked for synthesis of the LNA oligonucleotides. LITERATURE CITED Adamiak, D. A., Milecki, J., Propenda, M., Adamiak, R. W., Dauter, Z., and Rypniewski, W. R. (1997) Crystal Structure of 2′-O-Me(CGCGCG)2, an RNA duplex at 1.30 Å Resolution. Hydration Pattern of 2′-O-methylated RNA. Nucleic Acids Res. 25, 4599-4607. Baleja, J. D., pon, R. T., and Sykes, B. D. (1990) Solution Structure of Phage Lamda Half-Operator DNA by Useof NMR, Restrained Molecular Dynamics, and NOE-Based Refinement. Biochemistry 29, 4828-4839. Blommers, M. J. J., Pieles, U., and De Mesmaeker, A. (1994) An Approach to the Structure Determination of Nucleic Acid Analogues Hybridized to RNA. NMR Studies of a Duplex Between 2′-O-Me RNA and an Oligonucleotide Containing a Single Backbone Modification. Nucleic Acids Res. 22, 41874194. Boelens, R., Scheek, R. M., Dijkstra, K., and Kaptein, R. (1985) Sequential Assigment of Imino- and Amino-Proton Resonances in 1H NMR Spectra of Oligonucleotides by TwoDimensional NMR Spectroscopy. Application to lac Operator Fragment. J. Magn. Reson. 62, 378-386. Borgias, B. A., and James, T. L. (1990) MARDIGRAS-A Procedure for Matrix Analysis of Relaxation for Discerning Geometry of an Aqueous Structure. J. Magn. Reson. 87, 475-487. Borgias, B. A., Gochin, M., Kerwood, D., and James, T. L. (1990) Relaxation Matrix Analysis of 2D NMR Data. Prog. NMR Spectrosc. 22, 83-100. Breslauer, Frank, R., Blu¨cker, H., and Marky, L. (1986) Predicting DNA Duplex Stability from Base Sequence. Proc. Natl. Acad. Sci. 83, 3746-3750.

Bioconjugate Chem., Vol. 11, No. 2, 2000 237 De Mesmaeker, A., Ha¨ner, R., Martin, P., and Moser, H. E. (1995) Antisense Oligonucleotides. Acc. Chem. Res. 28, 366374. Dickerson, R. E., Bansal, M., Calladine, C. R., Diekmann, S., Hunter, W. N., Kennard, O., von Kitzing, E., Nelson, H. C. M., Lavery, R., Olson, W. K., Saenger, W., Shakked, Z., Soumpasis, D. M., Tung, C.-S., Sklenar, H., Wang, A. H. J., and Zhurkin, V. B. (1989) Definitions and Nomenclature of Nucleic Acid Structure Parameters. EMBO J. 8, 1-4. Feigon, J., Leupin, W., Denny, W. A., and Kearns, D. R. (1983) Two-Dimensional Proton Nuclear Magnetic Resonance Investigation of the Synthetic Deoxyribonucleic Acid Decamer d(ATATCGATAT)2. Biochemistry 22, 5943-5951. Gotfredsen, C. H., Spielmann, H. P., Wengel, J., and Jacobsen, J. P. (1996) Structure of a DNA duplex Containing a Single 2′-O-Methyl-beta-D-araT: Combined Use of NMR, Restrained Molecular Dynamics and Full Relaxation Matrix Refinement. Bioconjugate Chem. 7, 680-688. Hare, D. R., Wemmer, D. E., Chou, S.-H., Drobny, G., and Reid, B. R. (1983) Assignment of the Non-Exchangeable Proton Resonances of d(C-G-C-G-A-A-T-T-C-G-C-G) Using TwoDimensional Nuclear Magnetic Resonance Methods. J. Mol. Biol. 171, 319-336. He´le`ne, C. (1991) The Anti-Gene Strategy: Control of Gene Expression by Triplex-Forming-Oligonucleotides. Anti-Cancer Drug Des. 6, 569-584. He´le`ne, C., and Toulme´, J.-J. (1990) Specific Regulation of Gene Expression by Antisense, Sense and Antigene Nucleic Acids. Biochim. Biophys. Acta 1049, 99-125. Johansen, F., and Jacobsen, J. P. (1998) Site Selective BisIntercalation of a Homodimeric Oxazole Yellow dye in DNA Oligonucleotides. J. Biomol. Struct. Dyn. 16, 205-222. Koshkin, A. A., Singh, S. K., Nielsen, P., Rajwanshi, V. K., Kumar, R., Meldgaard, M., Olsen, C. E., and Wengel, J. (1998a) LNA (Locked Nucleic Acids): Synthesis of the Adenine, Cytosine, Guanine, 5-Methylcytosine, Thymine and Uracil Bicyclonucleoside Monomers, Oligomerisation, and Unprecedented Nuclei Acid Recognition. Tetrahedron 54, 3607-3630. Koshkin, A. A., Rajwanshi, V. K., and Wengel, J. (1998b) Novel Convinient Syntheses of LNA [2.2.1]Bicyclo Nucleosides. Tetrahedron Lett. 39, 4381-4384. Koshkin, A. A., Nielsen, P., Meldgaard, M., Rajwanshi, V. K., Singh, S. K., and Wengel, J. (1998c) LNA (Locked Nucleic Acid): An RNA Mimic Forming Exceedingly Stable LNA:LNA Duplexes. J. Am. Chem. Soc. 120, 13252-13253. Kværnø, l., and Wengel, J. (1999) Investigation of Restricted Backbone Conformations as an Explanation for the Exceptional Thermal Stabilities of Duplexes Involving LNA (Locked Nucleic Acid): Synthesis and Evaluation of Abasic LNA. Chem. Commun. 1999, 657-658. Lavery, R., and Sklenar, H. (1988) The Definition of Generalized Helicoidal Parameters and of Axis Curvature of Irregular Nucleic Acids. J. Biomol. Struct. Dyn. 6, 63-91. Lavery, R., and Sklenar, H. (1989) Defining the Structure of Irregular Nucleic Acids: Conventions and Principles. J. Biomol. Struct. Dyn. 7, 655-667. Liu, H., Spielmann, H. P., Ulyanov, N. A., Wemmer, D. E., and James, T. L. (1995) Interproton Distances Bounds from 2DNOE Intensities: Effect of Experimental Noise and Peak Integrations Errors. J. Biol. NMR 6, 390-402. Lubini, P., Zu¨rcher, W., and Egli, M. (1994) Stabilizing Effects of the RNA 2′-Substituent: Crystal Structure of an Oligodeoxynucleotide Duplex Containing 2′-O-Methylated Adenosines. Chem. Biol. 1, 39-45. Mujeeb, A., Kerwin, S. M., Kenyon, G. L., and James, T. L. (1993) Solution Structure of a Conserved DNA Sequence from the HIV-1 Genome: Restrained Molecular Dynamics Simulation with Distance and Torsion Angle Restraints Derived from Two-Dimensional NMR Spectra. Biochemistry 32, 1341913431. Nielsen, C. B., Singh, S. K., Wengel, J., and Jacobsen, J. P. (1999) The Solution Structure of a Locked Nucleic Acid (LNA) Hybridized to DNA. J. Biomol. Struct. Dyn. 17, 175-191. Nielsen, P. E. (1991) Sequence-Selective DNA Recognition by Synthetic Ligands. Bioconjugate Chem. 2, 1-12.

238 Bioconjugate Chem., Vol. 11, No. 2, 2000 Obika, S., Nanbu, D., Hari, Y., Mori, K., In, Y., Ishida, T., and Imanishi, T. (1997) Synthesis of 2′-O, 4′-C-Methyleneuridine and -Cytidine. Novel Bicyclic Nucleosides Having a Fixed C3′endo Sugar Puckering. Tetrahedron Lett. 38, 8735 Petersen, M., and Jacobsen, J. P. (1998) Solution Structure of a DNA Complex with the Fluorescent Bis-Intercalator TOTO Modified in the Major Groove. Bioconjugate Chem. 9, 331340. 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., and Jacobsen, J. P. (2000) Conformation of Locked Nucleic Acid (LNA) Hybridized to DNA. J. Mol. Recognit. 13, 44-53. Propenda, M., Biala, E., Milecki, J., and Adamiak, R. W. (1997) Solution Structure of RNA Duplexes containing Alternating CG Base Pairs: NMR Study of r(CGCGCG)2 and 2′-O-Me(CGCGCG)2 under Low Salts Conditions. Nucleic Acids Res. 25, 4589-4598. Saenger, W. (1984) Principles of Nucleic Acid Structure, SpringerVerlag, New York. Scheek, R. M., Russo, N., Boelens, R., and Kaptein, R. (1983) Sequential Resonance Assigments in DNA 1H NMR Spectra by Two-Dimensional NOE Spectroscopy. J. Am. Chem. Soc. 105, 2914-2916. Schmitz, U., Sethson, I., Egan, W. M., and James, T. L. (1992) Solution Structure of a DNA Octamer Containing the Pribnow Box via Restrained Molecular Dynamics Simulation with Distance and Torsional Angle Constraints Derived from TwoDimensional Nuclear Magnetic Resonance Spectral Fitting. J. Mol. Biol. 227, 510-531. Searle, M. S., and Williams, D. H. (1993) On the Stability of Nucelic Acid Structures in Solution: Enthalpy - Entropy Compensations, Internal Rotations and Reversability. Nucleic Acid Res. 21, 2051-2056.

Nielsen et al. Singh, S. K., and Wengel, J. (1998) Universality of LNAMediated High-Affinity Acid Recognition. Chem. Commun. 1247-1248. Singh, S. K., Nielsen, P., Koshkin, A. A., and Wengel, J. (1998) LNA (Locked Nucleic Acids): Synthesis and High-Affinity Acid Recognition. Chem. Commun. 1998, 455-456. Spielmann, H. P., Wemmer, D. E., and Jacobsen, J. P. (1995) The Solution Structure of a DNA Complex with the Fluorescent TOTO Determined by NMR Spectroscopy. Biochemistry 34, 8542-8553. Stein, P. C., Jacobsen, J. P., and Spielmann, H. P. (1995) Design and Implementation of a Simple Shaped Notch Filter. J. Magn. Res., Ser. B 109, 93-96. Thuong, N. T., and He´le`ne, C. (1993) Sequence-Specific Recognition and Modification of Double-Helical DNA by Oligonucleotides. Angew. Chem., Int. Ed. Engl. 32, 666-690. Uhlmann, E., and Peyman, A. (1990) Antisense Oligonucleotides: A New Therapeutic Principle. Chem. Rev. 90, 543-584. Weisz, K., Shafer, R. H., Egan, W., and James, T. L. (1994) Solution Structure of the Octamer Motif in Immunoglobulin Genes via Restrained Molecular Dynamics Calculations. Biochemistry 33, 354-366. Wengel, J. (1998) Synthesis of 3′-C- and 4′-C-Branched Oligonucleotides and the Development of Locked Nucleic Acid (LNA). Acc. Chem. Res. 32, 301-310. Wengel, J., Koshkin, A., Singh, S. K., Nielsen, P., Meldgaard, M., Rajwanshi, V. K., Kumar, R., Skouv, J., Nielsen, C. B., Jacobsen, J. P., Jacobsen, N., and Olsen, C. E. (1999) LNA (Locked Nucleic Acids). Nucleosides Nucleotides 18, 13651370. Wu¨thrich, K. (1986) NMR of Proteins and Nucleic Acids, John Wiley & Sons, New York.

BC990121S