We have been interested in whether three proteins that share a five-stranded β-barrel "OB-fold" structural motif but no detectable sequence homology fold by similar mechanisms. Here we describe native-state hydrogen exchange experiments as a function of urea for SN (staphylococcal nuclease), a protein with an OB-fold motif and additional nonconserved elements of structure. The regions of structure with the largest stability and unfolding cooperativity are contained within the conserved OB-fold portion of SN, consistent with previous results for CspA (cold shock protein A) and LysN (anticodon binding domain of lysyl tRNA synthetase). The OB-fold also has the subset of residues with the slowest unfolding rates in the three proteins, as determined by hydrogen exchange experiments in the EX1 limit. Although the protein folding hierarchy is maintained at the level of supersecondary structure, it is not evident for individual residues as might be expected if folding depended on obligatory nucleation sites. Rather, the site-specific stability profiles appear to be linked to sequence hydrophobicity and to the density of long-range contacts at each site in the threedimensional structures of the proteins. We discuss the implications of the correlation between stability to unfolding and conservation of structure for mechanisms of protein structure evolution.The concept of a protein fold as a recurrent unit of structure is central to modern structural biology. One of the goals of The Protein Structure Initiative is to determine representative examples of all folds (1), estimated to number between 1000 (2) and 10 000 (1,3,4). The difficulty of protein structure prediction targets and the approaches for modeling unknown structures are often predicated on whether the protein is likely to have a novel fold (5). Inferences about evolutionary relationships are often made on the basis of protein folds, because structure can be conserved even when sequences have diverged to levels where sequence homology is no longer detectable (6). While the concept of a protein fold is generally accepted, the classification of folds in practice remains somewhat subjective and ambiguous.The OB-fold 1 (Figure 1) was originally described as an oligonucleotide/oligosaccharidebinding fold (7) and consists of a five-stranded β-barrel structure. It is represented well in both sequence and structure databases (3,(8)(9)(10). The October 2006 release of the SCOP database (11) classifies some 95 protein domains as OB-fold structures. These are grouped into 27 function families with activities ranging from molybdate-binding proteins to RNA chaperones. † This work was supported by NSF Grant MB 0236316 to A.T.A., a Richard Crain Jr. Memorial Fellowship to W.M.M., and a Todd M.Schuster undergraduate summer fellowship to E.W. * To whom correspondence should be addressed: Department of Molecular and Cell Biology, University of Connecticut, 91 N,. Eagleville Rd., U-3125, Storrs, CT 06269-3125., Telephone: (860) Fax: (860) ., E-mail: andrei@uconn.edu.. ‡...