In Vivo Reporting and Modeling of N-Linked Glycosylation In Escherichia Coli

Protein glycosylation is important for folding, solubility, serum half-life, and efficacy of many therapeutic proteins.  Currently, the ability to produce recombinant protein therapeutics is limited to eukaryotic hosts, since high-titer-producing prokaryotic hosts such as Escherichia coli do not naturally glycosylate proteins.  However, the recent transfer of N-linked glycosylation machinery from Campylobacter jejuni to E. coli has opened the door to engineered protein glycosylation in prokaryotic hosts.  In these newly engineered E. coli, glycosylation occurs in the periplasm.  We recently reported the creation of a folding and solubility reporter for proteins in the bacterial periplasmic space.  Briefly, the reporter uses the unique co-translational translocation activity of the SRP pathway, together with a β-lactamase (BLA) fusion to the protein of interest, to couple ampicillin resistance to the protein's folding and solubility in the periplasm.  Furthermore, we have expanded this technique to develop an assay for protein-protein interactions in the periplasm using a split-β-lactamase reporter system to measure the interactions of several known interacting domains.

We have since extended the functions of the reporters to measure properties beyond intrinsic protein folding, such as the effect of periplasmic chaperone activity, disulfide bond formation, and engineered N-linked protein glycosylation.  In eukaryotic cells, production of many proteins has been shown to be largely dependent on glycosylation. We hypothesized that if folding in E. coli were dependent on glycosylation, then the reporter would couple protein glycosylation to ampicillin resistance.  We tested this protein in our folding reporter and found that only when co-expressed with functional glycosylation machinery were cells able to grow on antibiotic-selective media. By plating on ampicillin, the folding reporter allows us to select for glycosylated protein over aglycosylated protein with high enrichment factors.  This finding could allow us to engineer proteins to be more efficiently glycosylated either by engineering the machinery or the protein of interest by directed evolution.  Finally, we present the development of a metabolic model of N-linked protein glycosylation in E. coli to determine cellular factors which may contribute to increased protein glycosylation.  The model will help guide decisions about strains, gene knockouts, and growth conditions that can maximize protein glycosylation.