Abstract
A synthetic gene encoding human proinsulin, containing Escherichia coli preferred codons, with an additional N-terminal methionine, was used for the expression, of M-proinsulin and construction of nine derivatives. No improvement in expression was noted, relative to that of M-proinsulin, when the 5′- of the gene was appended to codons for seven amino acids of a well expressed E. coli protein (threonine dehydrogenase), or the constructs contained multiple copies of the proinsulin gene. That in the latter constructs only the gene adjacent to the prometer sequence is expressed, was shown by a construct containing a proinsulin gene followed by that for interferon α-2b. With the latter construct, the proinsulin was, predominantly, expressed. The availability of data on the constructs prompted, subjecting these to analysis by two models designed to predict the expression of proteins from the sequences, of putative mRNA, around the start of translation but no significant relationship was noted. In all cases the proteins were expressed as inclusion bodies, which were refolded to give products of desired masses and successfully converted into insulin derivatives. Of all the constructs containing a trypsin sensitive site before phenylalanine (F), the N-terminal sequence, MKR↓F, was most efficiently processed, by a cocktail of trypsin and buffalo carboxypeptidase B, to give insulin with the removal of the N-terminus linker as well as the C-peptide in a single step, without cleaving the trypsin sensitive K29T30 peptide bond.
•Free energy values for the unwinding of mRNA did not correlate with proinsulin expression.•In multi-copy constructs only the first gene, nearest to the promoter was expressed.•N-terminal MKR↓F was the optimal linker for the processing of proinsulin derivative.