Oligomers containing tracts of cytidine form hemiprotonated base pairs at acid pH and have been considered to be double-stranded. We have solved the structure of the DNA oligomer 5'-d(TCCCCC) at acid pH and find that it is a four-stranded complex in which two base-paired parallel-stranded duplexes are intimately associated, with their base pairs fully intercalated. The relative orientation of the duplexes is antiparallel, so that each base pair is face-to-face with its neighbours. The NMR spectrum displays only six spin systems, showing that the structure is highly symmetrical on the NMR timescale; the four strands are equivalent. A model derived by energy minimization and constrained molecular dynamics shows excellent compatibility with the observed nuclear Overhauser effects (NOEs) particularly for the very unusual inter-residue sugar-sugar NOEs H1'-H1', H1'-H2" and H1'-H4'. These NOEs are probably diagnostic for such tetrameric structures.

Pulsed nuclear magnetic resonance of exchangeable protons in H 2 0 solvent is most often accomplished by using Redfield's "2-1-4" weak-pulse sequence, which leaves the water proton magnetization unperturbed and provides second-order cancellation: the amplitude of a peak is a quadratic function of its distance to the frequency of cancel1ation.]s2 Hence, solvent suppression is insensitive to slight errors and drifts in frequency or field-gradient correction. The same features can be achieved by a sequence3 of strong, nonselective pulses (l5', T , 15', 15', T , 15'). The advantages of strong-pulse irradiation lie in the easier designing of sequences, in their insensitivity to long-term drift in pulse amplitude, and in the simpler generation of such pulses in an FT spectrometer.A defect of the sequences mentioned above is the generation of phase shifts, which combined with the existence of a residual but still intense solvent peak, lead to base-line ~n d u l a t i o n .~ We have searched for strong-pulse solvent-suppressing sequences that would produce no phase shift and hence no base-line undulation. Two sequences will be presented here. They share the following properties: (a) only strong, nonselective pulses (10 W) are used; (b) the pulse is at the solvent frequency; (c) they use only two phases, 0' and 180'; (d) no linear phase correction is required. The first sequence ("jump and return") provides first-order cancellation of the solvent peak. The phase is strictly constant across the spectrum, except for a step change of 180' at the solvent frequency. The second sequence ("set, jump, and return") gives second-order cancellation and only small phase shifts.The JR ("Jump and Return") Sequence (90°,, T , 9OO-J. The first pulse brings all spins along Ox in the rotating frame. During the waiting time they fan out in the xy plane; the solvent spins are at resonance and remain along Ox. The waiting time determines the variation of amplitude with frequency and the final longitudinal magnetization. The second pulse brings all spins back from the xOy plane to positions in the zOy plane, the solvent being returned to Oz. The free precession is then recorded. Transverse magnetizations start along Oy on one side of the solvent peak, along -0y on the other.Tuning is done while observing the free precession. It consists in (a) adjusting the pulse length to 90' (an error will only give small phase shifts, it will not affect solvent cancellation), (b) correcting for pulse-phase errors by a slight change in radiofrequency (rf) or by an adjustable phase differential, (c) correcting for pulse inequality (and/or radiation damping) by an adjustable pulse length differential. Figure 1 shows a spectrum of yeast tRNAPhe obtained with the JR sequence. It exhibits the characteristic 180' phase difference between the two sides of the solvent peak. Good phase control and shaping of the rf pulses are involved in these experiments. In our spectrometer, all rf manipulations are carried out at the (1) Redfield, A. G.; Kunz, S:D.; Ralph, E. K. (...

No abstract