Search

summary of the identified exchange peaks observed to- gether with their chemical shifts is reported in Table 1. Their assignment was achieved assuming that each resonance of the native state is in exchange with the resonance of the correspond- ing proton in the molten globule state (e. g. the H2,6 resonance will interconvert directly with the corresponding H2’,6’ reso- nance, where the prime notation refers to protons in the molten globule state). A comparison between the chemical shifts of the molten globule species and the corresponding native resonances shows that, as previously observed by Shimizu et al. (1993), the former move towards the random coil values (Table l), with H4 of Trpl04 as the only exception. Exchange is observed for Trp26, Phe31, Tyr103, Trpl04 and Trpll8. The observation of these exchange peaks has also been reported by Baum et al. (1989) in a two-dimensional exchange spectrum of guinea pig a-lactalbumin and by Shimizu et al. (1993) for bovine a-lactal- bumin. The presence of exchange involving other residues, such as Tyrl8 and Phe53, could not be excluded from the ROESY, but could not be detected because of peak overlap. An estimate of the exchange rate was derived for the well separated resonances of Tyr103 H2,6/H2’,6’, Phe3 1 H2,6/H2’,6’ and Trpl04 H4/H4’ from the ratio between cross and diagonal peaks in the NOESY spectrum at 40 ms mixing time (Emst et al., 1987). Simplified equations were used based on the following assumptions: (a) the two sites are assumed equally populated; (b) both cross and spin lattice relaxation terms are equal for the two species; (c) an initial rate approximation is assumed. While the first two assumptions are rather arbitrary in a general case, the last one is expected to be reliable at a mixing time of 40 ms. Comparable interconversion rates are shown by Phe31 and Trpl04 (1.5 s-’ and 1.7

We have suggested elsewhere the use of surface mapping by spin label probes (Esposito et al., 1992). According to this approach, soluble nitroxides are added to a protein solution. Resonances of protons that are accessible to the nitroxide are broadened and bleached out of the spectrum, while resonances in the protein interior remain unaffected. This approach is, in principle, complementary to another technique, photochemically induced dynamic nuclear polarization, which maps the position of aromatic protons on the protein surface. A detailed comparison between the two techniques is necessary for a confident use of the more recent suggested nitroxide perturbation approach. In the present study, we show that the results obtained by the two techniques for the native state of bovine alpha-lactalbumin are fully consistent and may therefore be combined for the study of protein surfaces.