Determination of sugar conformations by NMR in larger DNA duplexes using both dipolar and scalar data: application to d(CATGTGACGTCACATG)2.

Different methods for determining sugar conformations in large oligonucleotides have been evaluated using both J-coupling and NOE data. In order to stimulate COSY spectra, reliable estimates of line widths are required. We have measured T1p ( = T2) values for a large number of protons of the hexadecamer d(CATGTGACGTCACATG)2 using a new two-dimensional NMR experiment (T1RHOSY) to provide baseline information for the simulations. Both DQF-COSY and P.E.COSY cross-peaks have been systematically simulated as a function of line width, digitisation and signal-to-noise ratio. We find that for longer correlation times (tau > or = 5 ns), where line widths are comparable to or larger than active couplings, only sigma 1' (3J1'2'+3J1'2") is reasonably accurately determined (within +/- 1 Hz). Under these conditions, additional information is needed to determine the sugar conformation. We have used apparent distances H1'-H4' and H2"-H4', which provide a range of Ps over an interval of ca. 20 degrees. Complete analysis of time courses for intraresidue NOEs, with and without coupling constants, has also been evaluated for determining nucleotide conformations. Whereas Ps is poorly determined in the absence of both intrasugar NOEs and coupling constants, the range of solutions is decreased when intrasugar NOEs and sigma1' are also available. DQF-COSY, P.E.COSY and NOESY spectra at different mixing times of the hexadecamer d(CATGTGACGTCACATG)2 were recorded at three temperatures. A detailed analysis of the NOEs and coupling constants provided estimates of the sugar conformations in the hexadecamer. At 50 degrees C, the sugar conformations are well determined by the scalar and dipolar data, with pseudorotation phase angles of 126-162 degrees and mole fractions of the S conformation (fs) of 0.86 +/- 0.05. There was no statistically significant difference between fs for the purines and the pyrimidines, although there was a small tendency for Ps of the purines to be larger than those of the pyrimidines. At 25 degrees C, the sugar conformations were much less well determined, although the estimates of fs were the same within experimental error as at 50 degrees C. The experimental and theoretical results provide guidelines for the limits of conformational analysis of nucleic acids based on homonuclear NMR methods.