The combined use of 2D-WATERGATE-ROESY and NOESY gradient experiments (5,6) results in a significant alleviation of the water resonance and excellent definition of the baseline in the NH region. This could allow, in principle, the estimation of differences in solvent accessibility of the NH protons on the basis of the ROE/NOE intensities and comparison with the classical methodology of d(NH)/DT temperature coefficients (9). Figure 1 illustrates the normal 1D ¹H NMR spectrum of Lys-vasopressin and the cross sections through the 2D NOESY (Fig. 1B) and 2D ROESY (Fig.1C), which contain the water-peptide cross peaks. The Lys-vasopressin signals show negative NOESY cross-peaks with the water line, corresponding to positive cross- relaxation s_NOE values. This can be attributed to peptide-water proton dip olar interactions which are modulated by higher frequency rate process. There are few exceptions of positive NOESY cross peaks which are the amide proton of Tyr2 (very probably its exchange rate is catalysed by the proximal a-⁺NH₃ group) and the side-chain amide protons of Gln4, where the chemical exchange with the water dominates over the NOE in the NOESY spectrum. The broad, due to chemical exchange, positive peak at 7.35 ppm can be attributed to the ⁺NH3^- amino group of Cys1.

In contrast to intermolecular water-peptide cross-peaks, all intramolecular NOESY cross- peaks between different protons of Lys-vasopressin are positive. This is to be expected for a molecule in the slow tumbling regime. ¹³C relaxation times of the a-carbon s of oxytocin in 90% H2O/10% D2O, pH ~3.5 at 279 K were interpreted by assuming that the overall rotational tumbling of oxytocin can be characterized by a rotational correlation time of an equivalent sphere of t_R ~ 2 ns (1). For Lys-vasopressin, which is expected to have si milar relaxation properties to that of oxytocin, an overall correlation time of \3052.4 ns would be expected taking into consideration the difference in the viscosity of the solution at 273 and 279 K.

Examination of the 2D ROESY cross section which contains the water proton-peptide NH proton cross-peaks (Fig. 1C) reveals reduced integral for Gln4 and reduced composite integral of Cys6 and Phe3 compared with that of Lys8 and Gly9. This result, in conjuction with reduced integrals of the respective NOESY water-peptide cross peaks can be interpreted by assuming limited solvent accessibility of the above NH protons. Previous solvent saturation transfer studies (11-12) of the accessibility of the peptide amide protons of the trans isomer to the aqueous solvent environment also suggest that, in oxytocin, the Cys6 amide proton is significantly shielded from solvent.

The resonances of the amide protons of the trans isomer and those which can be resolved for the cis isomer are shifted to low frequency with increasing temperature (9). This low frequency shift is usually attributed to the breaking of a fraction of its hydrogen bonds (10). The majority of the Dd/DT values are in the - 6 to -10 x 10^-3 ppm/K range, including those of both the cis and trans isomers of the reduced dithiol forms of oxytocin and AVP. This is consistent with open structures and solvent-amide proton hydrogen bonds. The most rotable exceptions are those of the Asn and Cys6 amide protons of the native disulfide forms of the peptides, which indicate that the disulfide bond imposes constraints on the conformation of the hexapeptide ring.