Insulin has contains 51 amino acids, split into two chains held together by disulfide bonds. The two chains come from a single gene and a single mRNA. Clearly a lot of processing is required. There is an A chain containing 21 amino acids and a B chain containing 30.
Mutations of phenylalanine at position #24 on the B chain results in MODY (Maturity Onset Diabetes of the Young) in which not enough insulin is made. Every known vertebrate insulin contains phenylalanine at B#24.
A fascinating paper [ Proc. Natl. Acad. Sci. vol. 117 pp. 29618 – 29628 ’20 ] explains why.
The reason is that having phenylalanine at B#24 appears to be crucial in folding of the insulin into its final form. We have 20 amino acids, and changing phenylalanine at B#24 to any of the other 19 amino acids results in poor insulin production.
Well we can now make any protein this long by automated peptide synthesis. Which amino acid is closest in shape and structure to phenylalanine? Tyrosine clearly. So the authors made insulin with tyrosine at B#24 (outside the cell).
Guess what — insulin synthesized (outside the cells) B#24 tyrosine bound to the insulin receptor almost as well (20 fold less well), but in terms of biological activity there was no difference. The 3 dimensional structures of B#24 tyrosine and B#24 phenylalanine were nearly identical.
The problem was in the processing of the parent protein (proinsulin) with something other than B#24 phenylalanine to insulin, which involves breaking the chain and forming 3 disulfide bonds between 6 cysteines. So it isn’t structures which evolution is conserving by B#24 phenylalanine but the ability to be processed and folded correctly.
Time to let the authors speak for themselves “Our results suggest that sequences required for insulin’s bioactivity (similar in all vertebrates) are frozen at the edge of nonfoldability.”
Chemiotics II 