Designing structural-motifs for the preparation of acylated proinsulin and their regiospecific conversion into insulin modified at Lys (K).

Eight proinsulin encoding genes were prepared and their translation products, when treated with a cocktail of trypsin and carboxypeptidase B, analyzed for the following features. One, their ability to undergo facile removal of the N-terminal linker, generating the phenylalanine residue destined to be the N-terminal of the B-chain of insulin, at a rate similar to that involved in the removal of the C-peptide. Two, processing of diarginyl insulin, produced in the latter process, by carboxypeptidase B then needed to be rapid to remove the two arginine residues, Three, both these operations were to be efficient whether the N-terminal methionine was acylated or not. Four, the proinsulin constructs needed to contain a minimum number of sites for acylation. The aforementioned features were monitored by mass spectrometry and the proinsulin derivative containing MRR at the N-terminal and K mutated to Q, designated as MRR-(Q) human proinsulin [MRR-(Q) hpi] optimally fulfilled these requirements. The derivative was smoothly acylated with reagents of two chain lengths (acetyl and dodecanoyl) to give acetyl/dodecanoyl MRR-(Q) hpi. Acetyl MRR-(Q) hpi, using the cocktail of the two enzymes, was smoothly converted into, acetyl insulin. However, when dodecanoyl MRR-(Q) hpi was processed with the above cocktail, carboxypeptidase B (whether from animal pancreas or recombinant) showed an unexpected specificity of acting on the K-T bond of the insulin derivatives when K contained a large hydrophobic acyl group, generating dodecanoyl des-30 insulin.