L-nucleotides and 8-methylguanine of d(C₁^m8G₂C₃G₄C_5LG_6LC₇G₈C₉G₁₀)₂ act cooperatively to promote a left-handed helix under physiological salt conditions

Abstract

The structure and thermal stability of a hetero chiral decaoligodeoxyribonucleotide duplex d(C₁^m8 G₂C₃G₄C_5LG_6LC₇G₈C₉G₁₀)d(C₁₁^m8G₁₂C₁₃G₁₄C_15LG_16LC₁₇G₁₈C₁₉G₂₀) (O₁) with two contiguous pairs of enantiomeric 2′-deoxy-l-ribonucleotides (C_5LG_6L/C_15LG_16L) at its centre and an 8-methylguanine at position 2/12 was analysed by circular dichroism, NMR and molecular modelling. O₁ resolves in a left-handed helical structure already at low salt concentration (0.1 M NaCl). The central l₂-sugar portion assumes a B* left-handed conformation (mirror-image of right-handed B-DNA) while its flanking d₄-sugar portions adopt the known Z left-handed conformation. The resulting Z₄–B₂*–Z₄ structure (left-handed helix) is the reverse of that of B₄–Z₂*–B₄ (right-handed helix) displayed by the nearly related decaoligodeoxyribonucleotide d(^mC₁G₂^mC₃G₄C_5L G_6L^mC₇G₈^mC₉G₁₀)₂, at the same low salt concentration (0.1 M NaCl). In the same experimental conditions, d(C₁^m8G₂C₃G₄C₅G₆C₇G₈C₉G₁₀)₂ (O₂), the stereoregular version of O₁, resolves into a right-handed B-DNA helix. Thus, both the 8-methylguanine and the enantiomeric step C_LpG_L at the centre of the molecule are needed to induce left-handed helicity. Remarkably, in the various heterochiral decaoligodeoxyribonucleotides so far analysed by us, when the central C_LpG_L adopts the B* (respectively Z*) conformation, then the adjacent steps automatically resolves in the Z (respectively B) conformation. This allows a good optimisation of the base–base stackings and base–sugar van der Waals interactions at the ZB*/B*Z (respectively BZ*/Z*B) junctions so that the Z₄–B₂*–Z₄ (respectively B₄–Z₂*–B₄) helix displays a T_m (∼65°C) that is only 5°C lower than the one of its homochiral counterpart. Here we anticipate that a large variety of DNA helices can be generated at low salt concentration by manipulating internal factors such as sugar configuration, duplex length, nucleotide composition and base methylation. These helices can constitute powerful tools for structural and biological investigations, especially as they can be used in physiological conditions.

INTRODUCTION

The d→l reversal of deoxyribose chirality (Fig. 1) represents an interesting modification of DNAs that can be useful for the improvement of our understanding of the structure and dynamics of DNA molecules. It can also help in the design of molecules more resistant to nucleases in the context of antisense and aptamer strategies. Previous studies have shown that homochiral l-oligonucleotides could assume reverted helicity (left-handed helices) relative to natural DNAs (1). However, the properties of heterochiral l-d oligonucleotides still remain largely unknown, although enantiomeric deoxy-l-nucleotides have been used to gain insight into the relationship between the local DNA conformation and the helical sense of a double helix (2). For instance, double-stranded oligonucleotides bearing a single l-residue are still able to form stable Watson–Crick base pairs (1,3). This is also true for the DNA duplex d(^mC₁G₂^mC₃G₄C_5LG_6L^mC₇G₈^mC₉G₁₀)₂ with two contiguous l-residues in the middle of each strand (4). At neutral pH and in mild salt conditions, similar to physiological conditions, this molecule adopts an all-right-handed helix resulting from d-residues assuming the B-DNA form and l-residues the Z*-DNA form (mirror-image of Z-DNA). The practical conclusion of this study was that l-residues can cooperate with d-residues to form a heterochiral right-handed helix.

Figure 1.

Structures of the d- and l-deoxyribose sugars.

The stability of heterochiral right-handed helices is intimately related to an efficient base–base stacking and sugar–base van der Waals interactions. The absence of disruption in the chain of interactions contributes to maintain the helix stability at the BZ*/Z*B junctions (4). l-Enantioresidues in the Z* conformation allow such interactions to continue within the same groove of the DNA helix. Actually, the melting temperature (T_m) of the heterochiral oligonucleotide has been found to be only 5°C lower than that of its natural counterpart (4).

To learn more about the range of structures which can be assumed by heterochiral oligonucleotides, we now focuse on the d(C₁^8mG₂C₃G₄C_5LG_6LC₇G₈C₉G₁₀)₂ (O₁) molecule. Relative to the previously studied d(^mC₁G₂^mC₃G₄C_5L G_6L^mC₇G₈^mC₉G₁₀)₂ oligonucleotide (4), the new oligonucleotide, O₁: (i) still presents contiguous enantiomeric C_L and G_L at its centre; (ii) has no methylation of its cytosines; and (iii) guanine at position 2/12 now bears a C8-methylgroup that has been shown to markedly stabilise the Z-conformation of short oligonucleotides, even at low salt concentrations (5).

The properties of O₁ were analysed by circular dichroism (CD), UV melting and NMR experiments, as well as molecular modelling. In the CD and UV studies, the stereoregular d(C₁^m8G₂C₃G₄C₅G₆C₇G₈C₉G₁₀)₂ (O₂) was used in parallel to O₁, as a reference molecule. Under mild salt conditions (≤0.2 M NaCl), O₁ resolves into a left-handed helix, Z₄–B₂*–Z₄, while O₂ keeps the right-handed helix (B₁₀). Thus, the methylation of guanine at position 2/12 and the replacement of two contiguous d-residues by enantiomeric l-residues at the centre of the molecule are both needed to induce the Z-conformation in physiological salt conditions.

MATERIALS AND METHODS

Oligonucleotide synthesis

2′-Deoxy-l-nucleosides were prepared according to slightly modified procedures (6,7). Introduction of a methyl group at the C8 position of 2-N-isobutyryl-2′-deoxyguanosine was performed by a free radical methylation method as described by Sugiyama et al. (5). Base-protected deoxynucleosides were dimethoxytritylated and further converted to the corresponding β-cyanoethylphosphoramidites according to standard procedures. Oligonucleotides were synthesised by means of a Model 381A (Applied Biosystems Inc.) DNA synthesiser (10 µmol scale). After deprotection with 30% aqueous ammonia for 5 h at 55°C, the crude oligonucleotides were purified by anion-exchange chromatography on a DEAE–Sephadex A25 (Pharmacia) column using triethylammonium hydrogen carbonate buffer (pH 7.5, linear gradient from 0.5 to 1.5 M) as eluent. Fractions were analysed by C₁₈-reverse phase or anion-exchange HPLC, and those having a purity >95% were combined, co-evaporated with water several times and converted to the sodium salt with 2× Dowex 50 W (Na⁺ form). The final purity was >98% as determined by HPLC analysis.

CD and UV melting experiments

CD spectra of oligonucleotide solutions (70 µM in 10 mM Tris–HCl pH 7.0) were recorded in a 1 mm path length cell on a Jasco J-810 spectrophotometer equipped with a refrigerated water bath. Spectra at different temperatures were recorded after a 10 min equilibration period.

UV melting experiments were carried out on a Uvikon 931 (Kontron) spectrophotometer. The temperature of 1 cm path length cells was controlled by a Huber PD 415 temperature programmer connected to a refrigerated ethylene glycol/water bath (Huber Ministat). Annealing was performed by heating samples (7.5 µM oligonucleotide in 2 M NaCl, 10 mM Tris–HCl pH 7.0) at 90°C for 2 h and gradually cooling them at room temperature. During experiments, the sample temperature was increased at a rate of 30°C/h and the absorbance was measured at 260 nm. The melting temperature was read at the maximum of the first derivative of the UV melting curves.

NMR spectroscopy

NMR experiments were performed on a Bruker Avance- 500 MHz spectrometer. Data were processed on a Silicon Graphics Workstation with the Felix 2000 (Accelrys) and XwinNMR (Bruker) software.

Samples were dissolved in 0.4 ml of potassium dihydrogen phosphate–disodium hydrogen phosphate buffer at ionic strength 0.1 pH 7.1, and contained 1 mM EDTA. Concentrations were nearly 1.5 mM in duplex.

For assignments of exchangeable protons, samples were dissolved in H₂O buffer. Phase-sensitive NOESY experiments in TPPI mode using the WATERGATE pulse sequence (8) were recorded at the four different mixing times of 40, 80, 200 and 300 ms and acquired with 4096 complex data points in t₂ and 800 points in t₁, with a relaxation delay of 1.5 s, at both 20 and 30°C.

For assignments of non-exchangeable protons, samples were lyophilised several times in ²H₂O buffer and finally dissolved in 99.996% ²H₂O. NOESY experiments were recorded at different mixing times (40, 80, 120, 200 and 300 ms) with a relaxation delay of 2 s at 20 and 30°C. A total of 2048 complex points were acquired in t₂ and 512 real points in t₁.

Two-dimensional (2D) P-COSY and 2D TOCSY were also used for assignments of both exchangeable and non-exchangeable protons. ¹H–³¹P experiments (9,10) were used for the phosphorus assignments, and PFG-PEP HSQC (11) for the carbon assignments. A table of the chemical shifts is given in the Supplementary Material available at NAR Online.

NMR restraints

Interproton distance restraints used in structure calculations were measured from NOESY experiments recorded at various mixing times in H₂O and ²H₂O. Adequate references for distance calibrations in H₂O and ²H₂O were taken into account (12,13). Distances were classified as strong (1.8–3 Å), medium (1.8–4 Å), weak (1.8–5 Å) and very weak (1.8–6 Å) according to cross-peak intensities. Rotamer domains for backbone torsion angles α, β, γ, δ, ε and ζ were determined using ¹H- and ³¹P-NMR data, as previously described (13–15).

Watson–Crick hydrogen bond restraints were added for the stem residues which were experimentally determined as canonically paired. The criteria were chemical shifts of imino protons and the network of nuclear Overhauser effects (NOEs) in the base pair. Hydrogen bond restraints are the distances from a standard Watson–Crick base pair between the two heavy atoms of the hydrogen bond with an uncertainty of ±0.2 Å. For the terminal base pairs solely, an additional restraint was added, corresponding to the distance between the proton and the hydrogen bond acceptor. During the torsion angle and Cartesian dynamics steps, planarity restraints on the base pairs were added with a weight of 25 kcal/mol/Ų as described in Kuszewski et al. (16). These restraints permitted a significant propeller twisting of the base pairs to take place unhindered.

Molecular dynamics protocol

A molecular dynamics simulated annealing protocol driven by experimental restraints was used to solve the solution structure of the molecule. This protocol uses torsion angle molecular dynamics implementation in the CNS program (17). We started with DNA strands under extended conformations, using the CNS protocol generate_extended.inp, to avoid bias of a predetermined form (17,18).

The dynamics was carried out in three stages (18). In the first stage, we used dynamics in the torsion angle space: molecules were equilibrated at 20 000 K over 90 ps, and cooled slowly at 1000 K over 90 ps. Non-bonded interactions were described by a repulsive quadratic term and calculations were performed under NOE, hydrogen bond (50 kcal/mol/Ų) and dihedral (5 kcal/mol/°²) restraints. At the beginning of the cooling step, planarity restraints were added with a weight of 25 kcal/mol/Ų. The second stage involved Cartesian molecular dynamics consisting first of a cooling step from 1000 to 300 K in 15 ps, followed by a 15 ps equilibration period at 300 K. The non-bonded interactions were described by Lennard–Jones potentials, and electrostatic terms were turned on. The force constants on dihedral angle, NOE, hydrogen bond and planarity restraints were the same as previously. In a third step, dihedral angle force constants were slowly increased from 5 to 50 kcal/mol/°², during a Cartesian dynamics stage where the system evolved from 1000 to 300 K in a 9 ps period; this is followed by an equilibration period of 9 ps at 300 K. A total of 3000 steps of powell minimisation was then performed. Calculations were made 50 times; for each run, a different set of initial velocities was used.

Charge calculation with ab initio method

Charges of methylated nucleosides were calculated by fitting of the electrostatic potential, through the Merz Kollman procedure (19). This involved the comparative DFT geometry optimisations of guanines, 8-methylguanine, 9-methylguanine, 8,9-dimethylguanine, guanosine and 8-methylguanosine, using the B3LYP functional set (20–23) along with the 6–31G* basis set (24). Amino groups of nucleic bases were checked to be pyramidalised and not planar. The two nucleosides were optimised in South and North conformations, as in Hocquet et al. (25). All calculations were made with Gaussian 98 (26).

RESULTS AND DISCUSSION

CD analysis

CD spectra of heterochiral d(C₁^8mG₂C₃G₄C_5LG_6LC₇G₈C₉G₁₀) (O₁) and of homochiral d(C₁^8mG₂C₃G₄C₅G₆C₇G₈C₉G₁₀) (O₂) were recorded in buffers of varying NaCl concentrations (Fig. 2). The spectral features of O₁ (Fig. 2A) are specific for Z-DNA molecules (1,27) whatever the salt concentration. Even in the absence of NaCl, the O₁ spectrum displays two relatively weak bands of similar magnitudes at ∼260 nm (positive) and ∼295 nm (negative). The bands demonstrate a gradual symmetrical increase as a function of NaCl concentration reflecting an increase of Z-DNA. The spectrum of the stereoregular O₂ at low salt concentrations (0–0.2 M NaCl) (Fig. 2B) contains a large negative band at ∼255 nm accompanied by a smaller positive band at ∼280 nm. These are features of the right-handed B-DNA. At high salt concentration (2 M NaCl), the O₂ spectrum is typical of the left-handed Z-DNA.

Figure 2.

Effects of NaCl on the CD spectra of (A) d(C₁^m8G₂C₃G₄C_5LG_6LC₇G₈C₉G₁₀) (O₁) and (B) d(C₁^m8G₂C₃G₄C₅G₆C₇G₈ C₉G₁₀)₂ (O₂), at a 70 µM concentration in 10 mM Tris–HCl buffer pH 7, 12°C.

Sugiyama et al. (5) have concluded that ^m8G is a good inducer of Z-DNA, after having determined a mid-point at 0.03 M NaCl for the B→Z transition of a ^m8G-containing hexa-nucleotide d(C₁G₂C₃^m8G₄C₅G₆)₂. With the homochiral O₂ decaoligonucleotide, that incorporates ^8mG at the 2/12 positions, the mid-transition is observed at much higher NaCl concentration (largely >0.2 M), indicating that the length of the oligonucleotide and the position of the methylation in the sequence could influence the transition. Actually, the comparison of O₁ and O₂ proves that in long oligonucleotides, l-nucleotides are needed in addition to ^m8G to produce the B→Z transition at low NaCl concentration.

Temperature effects

Temperature melting effects on O₁ and O₂ were studied by combining CD and UV spectroscopies. We have seen above (Fig. 2) that both oligonucleotides present CD spectra typical of left-handed DNAs in 2 M NaCl, 12°C. In Figure 3A and B, the DNA thermal denaturation is reflected by a gradual reduction of the CD signal at ∼295 nm. The UV melting curves recorded at ∼260 nm for oligonucleotide O₁ and O₂ at concentrations nearly 10 times lower than those used in CD experiments provide T_ms of 64.7 and 69.7°C, respectively (Fig. 3C). Actually, the same 5°C difference has already been found between the T_m of the heterochiral molecule d(^mC₁G₂^mC₃G₄C_5LG_6L^mC₇G₈^mC₉G₁₀)₂ and that of its homochiral counterpart d(^mC₁G₂^mC₃G₄C₅G₆^mC₇G₈^mC₉G₁₀)₂ (4), indicating that the heterochirality does not disturb the helix stability too much.

Figure 3.

CD spectra of (A) O₁ and (B) O₂ at 70 µM concentration in a 10 mM Tris–HCl buffer pH 7, 2 M NaCl recorded as a function of temperature. (C) UV melting curves at 260 nm of O₁ and O₂ at 7.5 µM concentration in the same buffer as in CD.

Thus, a global picture of the DNA double helix highly sensitive to punctual changes in the DNA chemistry and to experimental conditions can already be deduced from CD experiments. More information on the shape of the helix and on its local structures, including the ZB*/B*Z junctions, can be gleaned from NMR and molecular modelling studies.

NMR analysis

Particular NOEs and chemical shifts observed in ²H₂O. We used NOESY, COSY and TOCSY spectra to assign the resonances of the non-exchangeable protons of O₁ in ²H₂O. These spectra are not shown. Identification of the ^m8G methyl group, the only methyl present in the oligonucleotide, permits in turn the assignments of the ^m8G deoxyribose-H1′ proton, and using through-bond experiments for the other sugar protons. Chemical shifts are listed in the Supplementary Material.

In the NOESY base–sugar H6/H8–H3′ fingerprint region, all the H6C–H3′G connectivities are strong, while those of H8G–H3′C, except H8G₆–H3′C₅, are very weak or absent. In the base–sugar H6/H8–H1′ fingerprint region, the GpC and the CpG steps display very distinct patterns: base–sugar H1′ NOEs are found only for GpCs and inter-residue base–sugar H2′/H2″ NOEs are systematically present in GpCs and absent in CpGs (except in C_5LpG_6L). The only interesidue NOE observable in the non-central CpGs concerns H8G_i–H4′C_(i-1).

A noticeable feature in the NOESY spectra is the appearance of very strong intranucleotide base–sugar cross-peaks for the guanine residues G₄, G₈, G₁₀ (H8–H1′) and ^m8G₂ (Me8–H1′). Intensities are similar to those exhibited by H5–H6 in the cytosine aromatic rings. Such cross-peaks are not observed for the guanine with inverted chirality, G_6L. This residue is the only guanine to adopt the glycosyl trans conformation in O₁, all the other guanines assuming the glycosyl syn conformation (12,13).

Another remarkable feature concerns the unusually high-field shifts displayed by the H2′ and H5′ resonances (∼1.7 and ∼2.6 p.p.m., respectively) of all the cytosines (except C_5L). The high-field shift probably results from the particular van der Waals interactions taking place between the sugars of cytosines and the bases of the succeeding guanines in O₁ (5).

Particular NOEs and base pairings observed in H₂O. The fingerprint region of imino protons connected to amino, aromatic and sugar H1′ protons is presented in Figure 4. Chemical shifts are typical for imino protons involved in Watson–Crick base pairs (28) at 13.29 (G_L6), 13.15 (G₈), 13.09 (^m8G₂) and 12.82 p.p.m. (G₄). At the same time, imino–imino interbase NOEs are detected for GpC steps, but not for the CpG steps.

Figure 4.

Expanded NOESY contour plot at 500 MHz of O₁ at 20°C in H₂O solution, showing the connectivities between the imino and the amino, the aromatic and the sugar 1′ proton resonances. Vertical lines provide the chemical shifts and the cross-peaks of the four imino protons with the other protons, the names of which are generally noted.

Together with the NOEs typical of Watson–Crick base pairs, we observe the three following inter-residue NOEs: H₁(G₈)–H₁′C₇, H₁G₂–H₁′C₁ and H₁G₄–H₁′C₃. Their presence features the existence of large deviations from standard B-DNA at the corresponding steps.

Cross-peaks arising between the C₁₉/C₉ geminal amino protons at 8.24 and 6.88 p.p.m. and the ^m8G₂/^m8G₁₂ imino proton at 13.09 p.p.m. are also detected. Amino protons of ^m8G give rise to two separate resonances, probably because the rotation of the amino group around the C₂–N₂ bond is considerably slowed down by the adjacent bulky methyl on the ring.

All together, the CpG and GpC steps are involved in very distinct networks of connectivities, and the enantio-C_LpG_L step displays spectral features which clearly deviate from those of the other CpG steps.

A sketch of the main NOE connectivities in oligonucleotide O₁ is presented in Figure 5. These illustrate the differences between the CpG and the GpC steps and the particular behaviour of the C_5LpG_6L/G_16LpC_15L step.

Figure 5.

Main NOE connectivities in the helix O₁. Arrows indicate NOEs between the non-exchangeable protons (intra-strand, black); imino and sugar/base protons (intra-strand, red); imino–imino and imino–amino protons (inter-strand, blue); amino–amino protons (inter-strand, green); and base–base protons (inter-strand, pink). See also the text.

Phosphodiester and sugar conformations in the backbone. Analysis of P-COSY and NOESY spectra provides information about sugar conformation (12,28). In the COSY spectra, we found that ³J₁′₂′ > ³J₁′₂″ (implying a C2′-endo sugar conformation) for cytosine sugars (except C₅) and the reverse for guanine sugars G₂, G₄, G₈ and G₁₀ (implying an unusual C3′-endo conformation). This observation is confirmed by the presence of large J_3′4′ and J_2″3′ cross-peaks for the G₂, G₄, G₈ and G₁₀ residues, while these cross-peaks are absent or weak for pyrimidines (except C₅). The NOESY data are in full agreements with these data. Accordingly, G₂, G₄, G₈ and G₁₀ were restrained to the C3′-endo conformation and the G₆, C₁, C₃, C₇ and C₉ residues to the C2′-endo conformation. The C₅ residue presents signs of conformational averaging and thus was not restrained.

According to the ¹H–³¹P correlation spectra, the differences detected on the base structures extend to the backbone phosphates. A fingerprint region (³¹P–H3′, H4′, H5′5″) is presented in Figure 6. The chemical shifts of phosphorus atoms are listed in the Supplementary Material. Among the nine phosphorus resonances, those six arising from the five CpG steps and the G₄pC_5L step are found in a rather classical range (between –3.5 and –4.5 p.p.m.). The three remaining phosphorus resonances (G₂pC₃, G₈pC₉ and G_6LpC₇) are at low field (at about –2.5 p.p.m.), which is usually explained by an α or a ζ angle assuming the trans conformation (12,13). However, the observation of H4′–³¹P cross-peaks (besides the common H3′–³¹P cross-peak) for the GpC steps (except that of G₄pC_5L) and the C_5LpG_6L step indicates that the β angles and γ angles of these steps occupy the trans domain and the gauche + domain, respectively (12,29).

Figure 6.

A HETCOR ¹H–³¹P spectrum of O₁ at 500 MHz in deuterated phosphate buffer pH 7 at 20°C. H3′–P correlations are found in the left part of the spectrum and those of H4′–P in the right part.

Molecular modelling

Structure calculations. Starting with the two oligonucleotide strands in extended conformation, we performed 50 simulations applying different initial velocities. The 10 best structures selected on energetic grounds are presented in Table 1. Seventy-two percent (36 out of 50) of the structures have an overall energy better than –600 kcal/mol and do not present NOE violations >0.2 Å. The r.m.s.d. for the 10 base pairs in the 10 best structures was 0.83 Å (relative to the average structure, Table 1), which represents a rather good structural definition given the conservative restraints used in the calculations.

Table 1. Statistics of the NMR restraints and structures.

NMR restraints
Intra-residue distances	166
Inter-residue distances	186
Hydrogen bond restraints	36
Dihedral angle restraints	84
Planarity restraints	10
Total restraints	482
Restraints/residue	24.1
Structure analysis of the 10 best conformers
Average deviation from ideal covalent geometry
Bond length	0.0018 ± 0.00006
Bond angles	0.4359 ± 0.0073
Improper angles	0.263 ± 0.01136
NOE violations
Number	0
r.m.s.d of the violation (Å)	0.0193 ± 0.00106
Dihedral angle violations
Number	0.5
r.m.s.d of the violations (°)	0.02 ± 0.0269
r.m.s.d of the structuresa	0.83 ± 0.06

^ar.m.s.d relative to the average structure.

The minimised averaged helix taking into account the 10 best O₁ structures combines a Z-DNA portion (base pairs 1–20 to 4–17), a B*-DNA portion (base pairs 5–16 and 6–15) and again a Z-DNA portion (base pairs 7–14 to 10–11). We obtain, therefore, a left-handed helix with a Z₄–B₂*–Z₄ alternation (Fig. 7). In the two Z portions, the sugars alternate the directions of their O4′ atoms, thus generating a zigzag characteristic of the Z-DNA (27). In contrast, the sugars of G₄/C₁₇ and of the enantiomeric C_5L/G_16L and G_6L/C_15L base pairs belonging to the B* portion, have their sugar O4′ atoms pointing in a unique and same direction, a feature belonging to B-DNA (here, B*-DNA).

Figure 7.

Stereo view of the mean minimised structure of O₁ obtained by molecular dynamics using the NMR data. Sugars of strand 1 are presented in a stick mode with the O4′ atoms indicated in red.

In the Z₄–B₂*–Z₄ helix, the Z and B* portions can also be distinguished from the differences of base–base stackings and base–sugar interactions in CpG steps. The Z-DNA allows the interstrand stacking of the cytosine bases and the intrastrand van der Waals interaction of cytosine bases and guanine sugars. For the central enantio-C_LpG_L step, a B-type (B*-type) stacking of bases is obtained.

The torsion angle values of the best structures are given in Table 2. The guanines G2, G4, G8 and G10 adopt the C3′-endo syn conformation, while the other residues are in the C2′-endo-anti range.

Table 2. Mean phosphodiester angles for the 10 best conformers of O₁ obtained with NMR data.

Mean	γ (°)	δ (°)	ε (°)	ζ (°)	α (°)	β (°)	χ (°)
C1	–177 ± 58	148 ± 1	–141 ± 17	123 ± 22	174 ± 99	–166 ± 53	–144 ± 10
G2	176 ± 88	91 ± 1	–164 ± 16	–80 ± 21	–150 ± 10	–121 ± 5	93 ± 4
C3	48 ± 5	145 ± 1	–145 ± 8	108 ± 12	162 ± 108	–165 ± 60	–145 ± 3
G4	165 ± 81	97 ± 3	–151 ± 21	–142 ± 66	133 ± 73	–179 ± 83	82 ± 17
C5	–176 ± 32	–136 ± 13	137 ± 20	103 ± 13	164 ±143	162 ± 25	131 ± 7
G6	–96 ± 111	–149 ± 3	–158 ± 27	122 ± 50	152 ± 47	–179 ± 16	103 ± 8
C7	78 ± 30	141 ± 4	–167 ± 13	139 ± 27	157 ± 93	–141 ± 49	–159 ± 2
G8	143 ± 37	90 ± 3	–151 ± 12	–97 ± 66	–157 ± 15	–128 ± 12	70 ± 7
C9	42 ± 4	146 ± 1	–154 ± 14	136 ± 31	147 ±160	–170 ± 74	–146 ± 4
G10	–124 ± 59	91 ± 1					76 ± 25

On each line, the α and β angles are those implicating the phosphorus of the following residue (i.e. for C₁, α and β angles are related to the C₁pG₂ step).

Description of the left-handed Z₄-B₂*-Z₄ helix and its comparison with the right-handed B₄–Z₂*–B₄ helix. The previously reported chimeric molecule d(^mC₁G₂^mC₃G₄C_5L G_6L^mC₇G₈^mC₉G₁₀)₂ has been found to resolve in a right-handed helix B₄–Z₂*–B₄ at low salt concentration (4). The B₄–Z₂*–B₄ helix is the reverse image of the left-handed helix Z₄–B₂*–Z₄ assumed by O₁ in the same experimental conditions. In these oligonucleotides, the conformation of the central enantio-C_LpG_L step is dictated by the conformation adopted by flanking DNA portions, the conformation of this latter being itself highly sensitive to subtle chemical modifications, such as guanine methylation (O₁). In the B₄–Z₂*–B₄ helix, the base–base stacking and base–sugar interactions taking place in the B₄ portions propagate to the whole helix thanks to the inversion of the sugar polarity at C_5L and the passage of the G_6L glycosyl bond into a syn orientation in the central enantio-C_LpG_L portion. These two operations that generate the Z* helix are needed to compensate the effects introduced by the chirality inversion at the centre of the molecule and to keep the base orientation in the grooves. In O₁, the methylated G₂/G₁₂ residue and the enantio-C_LpG_L step act together to induce the left-handed helix Z₄–B₂*–Z₄, at low salt concentration (Fig. 8). In the Z-DNA portion, the base positions are inverted relative to the right-handed helix. This occurs in response to both the inversion of polarity of the cytosine sugars and the adoption of the syn form by the guanine residues, the latter transition being strongly favoured by methylation of the guanine at position 8 (1). These two operations are needed to preserve the base stacking in the convex major groove of Z-DNA; they are not necessary in the central B₂* portion, as the change of the sugar chiralities results in an effect similar to that produced by the inversion of the sugar polarity at cytosine and the anti to syn transition of the glycosyl bond at guanine. Thus, at the Z–B* junction, the C_5L residue can adjust its base stacking with the preceding 5′G₄ residue without changing its sugar polarity, while the G_6L residue keeps the glycosyl anti conformation in order to optimise its interactions with its succeeding C₇ residue. The arrangement of the G₄pC_5L step at the Z–B* junction of the Z₄–B₂*–Z₄ helix is presented in Figure 9, together with the structure of the GpC step in B-DNA and Z-DNA helices. The particular arrangement adopted by the G₄pC_5L step in O₁ allows the continuation of the base–base stackings and of the van der Waals interactions along the whole left-handed helix. This continuation does not need the inversion of the sugar polarity and the glycosyl anti to syn conversion at the succeeding C_5LpG_6L/G_16LpC_15L step.

Figure 8.

Stereo views of the mean minimised structures of (A) O₁ and (B) d[^mC₁G₂^mC₃G₄C_5LG_6L^mC₇G₈^mC₉G₁₀]₂ (4). The backbone is drawn as a ribbon.

Figure 9.

Structure of the GpC step in three different DNA conformations: (A) Z-DNA; (B) B-DNA; (C) at the ZB* junction of the O₁ structure. Bases usually in the major groove are coloured in red, as are the sugar O4′ atoms. The base edges in the major groove are oriented forward in (B) and are opposite in (A) and (C).

CONCLUSION

The most salient feature of these studies is that heterochiral oligonucleotides can easily stabilise into right-handed or left-handed helices under physiological pH and salt conditions. Contiguous enantiomeric l-residues at the centre of self-complementary oligonucleotides can resolve into Z*-DNA (right-handed helix) or B*-DNA (left-handed helix), the result depending on the conformation of their flanking d-residues. If these assume the B (respectively Z) conformation, then the central C_LpG_L portion will adopt the Z* (respectively B*) conformation.

Although results demonstrate that in our chimeric oligonucleotides, the conformation of the flanking portions dictates the conformation of the central enantio-C_LpG_L, and therefore the whole helical conformation, the problem of the helical sense of such molecules at low salt concentration is not yet solved. We have seen, for instance, that a simple guanine methylation is sufficient to induce a left-handed helix, thus highlighting the high efficiency of subtle chemical modifications in the conformational change of chimeric oligonucleotides. Yet, how the effects are transmitted from the chemically changed position to the distant l-residues requires a detailed study with numerous heterochiral duplexes combining l- and d-DNA domains of different lengths and bearing guanine methylations at various positions.

Finally, we believe that chimeric oligonucleotides can constitute good tools in the search for molecules targeting DNAs and RNAs and to decipher the local structures and binding mechanisms of DNAs. The left-handed strands formed at low salt concentrations can be used as aptamers to detect proteins that bind tightly to Z-regions in DNAs. More importantly, they can also be useful in the preparation of antibodies that specifically recognise left-handed helices.

SUPPLEMENTARY MATERIAL

Chemical shifts of the O₁ compound are provided in a table in the Supplementary Material available at NAR Online.