Prediction of ¹³C chemical shifts in nucleic acids

P. Novák, V. Sklenář, and R. Fiala

National Centre of Biomolecular Research

Masaryk University

Kotlářská 2, 611 37 Brno, Czech Republic

Chemical shifts are sensitive probes of biomolecular structure [1]. The structural information contained in chemical shifts is, however, very different from the distance and dihedral angle constraints provided by NOE intensities and scalar coupling constants, respectively. The value of chemical shift of a specific nucleus is a sum of many contributions whose sources are not readily identifiable. Therefore, a necessary first step in the studies of chemical shifts in relation to the biomolecular structure is establishing a reliable procedure of calculating the chemical shifts for known structural elements.

As quantum chemical approaches at sufficiently high level remain beyond the applicable limit for molecules of biological interest, the chemical shift calculations rely mostly on classical or semiclassical approaches considering long-range effects arising from the magnetic susceptibilities, electric field effects and close contact effect (mainly hydrogen bonds) as well as the effect of local geometry. In interpreting macromolecular chemical shifts it is usually assumed that the individual contributions are approximately independent and additive. Since hydrogen atoms are bound to only one other atom, the local geometric effect tend to be reasonably constant for a particular type of nucleus and proton chemical shifts can be adequately characterized considering the long-range effects only. On the other hand, the local geometries need to be generally taken into account for ¹³C shifts.

For proteins, a number of programs exists to relate the proton and heteronuclear chemical shifts to structural features. Much less is known about the chemical shift – structure relationships in nucleic acids. It has been shown that ¹H shifts in both DNA [2] and RNA [3] can be adequately represented by just three contributions, namely by ring current, magnetic anisotropy and electric field effects, with the electric field effect playing a relatively small role. Carbon and nitrogen chemical shifts for different sugar puckers and base orientations were studied by DFT on the level of DNA and RNA dinucleotides [4].

The systematic studies of ¹³C chemical shifts of nucleic acids are hindered by inadequate database of available shifts and structures in BioMagResBank. In this situation, we decided to concentrate on a few molecules for which reliable ¹³C chemical shift data and accurate structures are available, namely Dickerson-Drew dodecamer (PDB code 1NAJ), d(GCGAAGC) hairpin (PDB code 1PQT), DNA duplex (part of protein-DNA complex PDB code 1AHD).

Chemical shifts were calculated for ¹³C in nucleic acid bases. The structural dependent part of the ¹³C chemical shifts was calculated as a sum of the following contributions: ring current shift and magnetic anisotropy (calculated by program NUCHEMICS), and electric field effect (calculated by program MOLMOL). The calculated shifts were added to the experimental shift values of free dNTP which are assumed to be devoid of the effects of stacking and base-pairing interactions.

The resulting chemical shifts were found to be rather sensitive to even slight structural changes. This confirms that ¹³C shifts are sensitive monitors of structural features, on the other hand, it makes obtaining good agreement with experimental shift values difficult. Tight correlation with the experimental shift data were observed for well defined regions of the molecules, especially for double helices. Larger errors in less regular regions can be ascribed partially to inaccurate structural information and in part to deficiencies in the method used to calculate the chemical shifts.

References

1. Wishart, D.S. & Case, D. A. (2001), Methods Enzymol. 338, 3-34.

2. Wijmenga et al. (1997) J. Biomol. NMR, 10, 337-350.

3. Cromsigt et al. (2001) J. Biomol. NMR, 21, 11-29.

4. Xu and Au-Yeung (2000) J. Phys. Chem. B, 104, 5641-5650.