PROTEIN STRUCTURE 



On balance, it is obvious that more direct experimental 

 evidence is needed before the dominant features of the insulin 

 structure are completely established, although it seems reason- 

 able to suppose that regions of a-helix configuration are prob- 

 ably a feature of the molecular stereochemistry. Extensive 

 x-ray crystal structure studies of insulin are being carried out 

 by Low and her associates here at Harvard. Even for insulin, 

 therefore, in which the chain sequences and cross links are com- 

 pletely known, the exact spatial configuration of the molecule 

 still remains in doubt. For all other proteins the uncertainties 

 are far greater, but there are some for which great progress 

 has been made, and we turn finally to a consideration of two of 

 these — ribonuclease and serum albumin. 



Ribonuclease 



This relatively simple protein from pancreas has been inten- 

 sively studied both by chemists (3,48) and by x-ray crystallog- 

 raphers (22,23). In addition to the x-ray work already pub- 

 lished the extensive and important work of David Harker and 

 his associates at the Protein Structure Project holds promise of 

 greatly increasing insight into the structure of this molecule. 

 However, it would be premature at the moment to draw ex- 

 tensive conclusions from the x-ray work, although far-reaching 

 developments may emerge shortly. Studies of sedimentation, 

 diffusion, and viscosity, in addition to the x-ray studies, all 

 indicate that the molecule is a highly compact structure. The 

 chemical studies have produced a preliminary but illuminating 

 picture. The evidence appears decisive (3) that the molecule 

 consists of a single polypeptide chain, probably 124 residues long 

 (48) with an A^- terminal lysyl group and a C- terminal valine. 

 No sulfhydryl groups are found, but there are four disulfide 

 bridges, which must therefore cross-link different portions of 

 the chain into loops. It was earlier suggested that the four pro- 

 line residues present could provide loci for the folding of the 

 peptide chains ; a model based on this assumption was published 

 by Anfinsen et al. (3). However, this model has now been aban- 



425 



