E. coli selection of human genes encoding secreted and membrane proteins based on cDNA fusions to a leaderless beta-lactamase reporter.

Although several signal peptide-trapping methods have been devised and used to detect signal sequences, none have relied on using E.coli to identify eukaryotic proteins with signal peptides. Here, we describe a system for selecting human secreted and membrane proteins in E. coli followed by the direct validation of secretion in human cells. The method is based on cDNA fusions to a leaderless beta-lactamase reporter gene to isolate clones encoding signal peptides of human genes. We found that beta-lactamase fusion proteins carrying a eukaryotic signal peptide at its N-terminus were able to direct their export into the periplasm in E. coli to confer survival upon challenge with carbenicillin. When libraries constructed from 5' end-enriched cDNAs fused to beta-lactamase were screened in E.coli, approximately 0.5%-1% of the cDNAs are selected, and over half of the surviving clones were found to encode for secreted fusion proteins when tested in human cells. These clones were sequenced and shown to represent human genes encoding signal peptides of secreted and membrane proteins. We conclude that this is an efficient and effective strategy to easily enrich cDNA libraries for the identification of novel genes likely to encode secreted enzymes, growth factors, and receptors.