An ‘internal” signal sequence directs secretion and processing of proinsulin in bacteria

Abstract

Most secreted proteins, both eukaryotic and prokaryotic, contain an amino-terminal extension that is removed some time during transport (see ref. 1 for a review). The signal hypothesis^2,3 states that these amino-terminal extensions (presequences or signal sequences) serve to bind the protein to the membrane and then to lead the protein through. We recently constructed a series of plasmids in which parts of the Escherichia coli prepenicillinase signal sequence⁴ were contiguous with the gene for rat preproinsulin, containing 21 of the 24 codons that code for the preproinsulin signal sequence. When Escherichia coli was transformed with each of these hybrid signal sequence constructions, an insulin antigen was secreted⁵ and correctly processed to proinsulin⁶ in every case. One of these constructions fused the first half of the prepenicillinase signal sequence codons to the preproinsulin gene; as insulin antigen was not secreted when the same codons were fused to the DNA coding only for proinsulin and hence lacking the eukaryotic signal sequence, we hypothesized that the eukaryotic signal sequence was sufficient to direct transport in bacteria. We could not, however, eliminate some minor but specific role for the amino terminus donated by the bacterial signal sequence. One way to do this is to replace the prepenicillinase sequences with that of a nonsecreted E. coli protein, for example β-galactosidase. We show here that E. coli harbouring a plasmid encoding a β-galactosidase–preproinsulin fusion protein efficiently secretes insulin antigen and processes the protein to proinsulin despite the fact that the signal sequence is internal to the amino terminus.