BARBARA W. LOW AND JOHN T. EDSALL 



To fit the experimental data some deviations are necessary 

 from this simple structure — that is, chains 30 to 25 A in length 

 with an a-helix configuration cannot all be stacked parallel; 

 irregularities must occur. Several different kinds of irregularities 

 can be postulated: (7) short lengths of chain in an a-helix con- 

 figuration ('^l 5 A) may be imperfectly aligned so that as a group 

 the coiled chains zigzag about the over-all axial direction; (2) 

 short lengths of a-helix may be followed by regions in which the 

 specific local molecular configuration is not associated with any 

 regular peptide chain configuration. 



These conclusions underline a further difference between 

 the a fibrous protein model and the globular protein model. In 

 globular proteins the stable discrete unit is a single molecule; 

 here, the third level of molecular organization described earlier 

 is intramolecular. Ordered and disordered regions of chain 

 configurations are confined within and establish the unique 

 three-dimensional molecular structure. Globular proteins are 

 almost always shorter than the model structure which would 

 be derived if peptide chains of average length (calculated from 

 end group determinations) were each coiled into a single 

 cylindrical a-helix structure and then stacked together in close- 

 packed array. The peptide chains must therefore turn corners 

 and reverse direction. In horse hemoglobin there are only 6 

 A^-terminal end groups per molecule, that is, three per (globin) 

 half-molecule. Even in the simple idealized structure (Figure 

 7) there are 17 "crystallographic" ends, that is, seventeen close- 

 packed cylindrical regions of coiled configuration. The absolute 

 intensity studies demand even more numerous discontinuities. 

 We referred earlier to the stereochemical discontinuity which a 

 proline residue introduces into regular helical chain configura- 

 tions which have the maximum number of strong and equivalent 

 hydrogen bonds. In the a-helix proline residues cause different 

 problems in left- and right-handed structures. This difference 

 may be understood by considering the closely similar 7r-helix. 

 In the drawing (Figure 5), of a right-handed helix a reasonably 

 planar proline ring can be constructed by joining the nitrogen 



410 



