Interaction with 8-anilino-1-naphthalenesulfonate (ANS) is widely used to detect molten globule states of proteins. We have found that even with stable partially folded states, the development of the fluorescence enhancements resulting from such interactions can be relatively slow and kinetically complex. This is probably because initial binding of the dye can induce subsequent changes in the protein structure, so that the ultimate resulting fluorescence enhancement is not necessarily a good, nonperturbing probe of the preexisting state of the protein. When ANS is used to study folding mechanisms the problem is compounded by the difficulty of distinguishing effects due to the development of dye interactions from those due to the changing populations of folding intermediates.Many of these complications can be avoided by experiments where the ANS is introduced only after folding has been allowed to proceed for a variable time. The initial fluorescence intensity after mixing, resulting only from rapid and therefore hopefully relatively nonperturbing interactions with the protein, can be monitored at different refolding times to provide a better reflection of the progress of the reaction, uncomplicated by dye interaction effects. Such studies of the folding of carbonic anhydrase and a-lactalbumin have been compared with conventional single-mix experiments and large discrepancies observed. When ANS was present throughout refolding, time-dependent changes attributed to the formation or reorganization of protein-ANS complexes were clearly superimposed on those associated with the actual progress of refolding, and the folding kinetics and population of intermediates were also substantially perturbed by the dye. Thus, it is clear that the pulse method, though cumbersome, should be used where refolding reactions are to be probed by dye binding.The results emphasize that fluorescence enhancement tends to be greatest in early intermediates, in contrast to what, for carbonic anhydrase at least, might appear to be the case from the more conventional experiments. Later intermediates in the folding of both of these proteins actually induce little fluorescence enhancement and therefore may be quite different in nature from equilibrium molten globule states. sitive to side-chain clustering, but they are not necessarily very representative of global structure and the total fluorescence intensity, which is the parameter most easily followed, is difficult to interpret in detail. Combinations of such simple methods with protein engineering offer a more general approach (Matthews, 1993;Serrano, 1994), but here too there are potential problems if the clustering is rather tenuous or nonspecific. It is clear neither whether such behavior would necessarily be detected by simple spectroscopic methods nor how mutations ought to be analyzed in the case of nonspecific interactions. One technique that appears to go some way to addressing the problem uses 8-anilino-1-naphthalene-sulfonate (ANS) as an extrinsic probe. This dye tends to b...