Inverse Metabolic Engineering to Improve N-Glycosylation Efficiency In E. Coli Cells

INVERSE METABOLIC ENGINEERING TO IMPROVE N-GLYCOSYLATION EFFICIENCY IN E. COLI CELLS.

Topic: Genomic, Proteomic and Metabolomic Approaches for Modifying Cellular Processes

Authors:

Jagroop Pandhal: j.pandhal@sheffield.ac.uk

Phillip Wright: p.c.wright@sheffield.ac.uk

A major goal of recombinant protein therapeutics is to increase overall production with the aim to alleviate demand associated with high/multiple doses. There are a number of choices of expression systems that can be used to express recombinant proteins, ranging from bacteria to plants to mammalian cells. Of these, bacterial hosts are particularly attractive since these organisms tend to grow rapidly and on relatively inexpensive media. The molecular biology workhorse, E. coli, is a very popular bacterial host strain as it comes with a wealth of biological information and it is facile to manipulate genetically. For these reasons it has been used to produce human therapeutic protein insulin. However, unlike insulin, more than 75% of proteins with human therapeutic importance in the pharmaceutical industry (either released or in clinical and preclinical development) are post-translationally modified (Sethuraman and Stadheim 2006). Since the modification is often essential for biological activity, alternative expression systems capable of this process such as yeasts, filamentous fungi, plants, mammalian and insect cells are chosen.

Glycosylation is the most common and complex post-translational protein modification. The process of covalently attaching glycan groups to polypeptide chains has several implications for the role of the protein final product including functional structure, half-life and cellular locality (Varki 1993). Presently, the preferred cellular host for therapeutic glycoprotein production, and in particular antibodies, are Chinese hamster ovary cells.

Between 1998 and 2003 several studies reviewed the then recent findings that prokaryotes, archaea and bacteria, performed glycosylation of their proteins. Long considered to be restricted to eukaryotes, this discovery led to scarce but exciting information of how widespread and important this modification was across all three domains of life. In 2002, a glycosylation pathway (pgl), characterised in Campylobacter jejuni, was transformed into Escherichia coli together with a vector (pET24b-AcrA) harbouring a glycosylation target from the same source organism. Although the nature of the glycans assembled were different to human glycans, these studies allowed for the prospect of using E. coli as a host for the production of human therapeutic proteins.

In this study, this pgl glycosylation pathway operon and glycosylation target protein AcrA (from C jejuni) are expressed in E.coli CLM24 cells on separate vectors (12 modification genes on pACYC vector and AcrA gene on arabinose inducible pEC vector). AcrA is a subunit of a multidrug transporter and contains two N-glycosylation sites. Because this periplasmic lipoprotein is the best characterised glycoprotein in C. jejuni (Nita-Lazar et al. 2005) and has previously been expressed and glycosylated in E. coli (Wacker et al. 2002), it was chosen for recombinant glycoprotein production in this study.

Increasing the titres of recombinant protein produced in host cell line fermentation processes is a challenge. The most popular strategy to date has been to introduce the gene encoding the therapeutic protein into the host cell and subsequently screen for increased production by: a) perturbing the growth conditions, b) varying vector characteristics, e.g. promoter/enhancer regions, codon usage etc. and, c) random mutagenesis of genes or incorporation of homologous genes. Collectively, these methods have been applied widely but success has been limited. The latter strategy can be more directed in classical metabolic engineering (Stephanopoulos and Vallino 1991) if a rate-determining step in a pathway can be identified using transcriptomics, proteomics or metabolomics. The identified bottleneck can subsequently be alleviated by over-expressing or knocking down/out the expression of relevant enzymes. In a previous study, we applied high throughput isobaric tags for proteome quantitation in cells over-producing glycosylated AcrA (gAcrA). Nearly 300 proteins were quantified and using a Bayesian approach statistically significant network changes were highlighted. For example, by increasing expression of isocitrate lyase the flux through the glyoxylate pathway was enhanced and therefore the TCA cycle was largely bypassed. This carbon conserving route was expected to contribute to the 300% relative increase in gAcrA although total AcrA did not change significantly. It is expected that carbon was used for glycan precursor synthesis and didn’t affect cellular growth significantly. Although a successful forward engineering experiment was demonstrated here, indirect effects such as counter-balancing regulation or a previously unknown rate-limiting step often come into play.

In the present work, an inverse metabolic engineering approach is applied to improve the E. coli system of glycosylating AcrA. Inverse metabolic engineering follows the strategy:-

1)      Constructing a desired phenotype using gene libraries

2)      Determining the genetic factor conferring this phenotype

3)      Bestowing the phenotype on another strain by directed genetic manipulation

We therefore constructed 4 gene libraries from E. coli DNA (1, 2, 4 and 8 kb libraries) and transformed these into the N-glycosylation capable E. coli CLM24 (Pgl, AcrA) cells. A screen was developed to select colonies with statistically significant increases in glycoslayed AcrA (gAcr). This involved soybean agglutinin lectin binding due to the nature of the terminal glycan N-acetylgalactosamine on AcrA, and chemiluminescence detection. To ensure maximum genome coverage the number of colonies screened was based on the following equation

N = ln (1 - P)/ln (1 - I/GS)

Where GS: Genome size (4,639,675), P: Probability of full coverage (0.99 -because 99% coverage desired of entire genome to 128 bp resolution, I: Insert size (two different restriction 4-cutters used with each either A,T,G,C so (4*4*4*4)/2), N: Number of transformants needed.

The DNA library inserts from high-producing clones were subsequently sequenced. Repeated gene (or gene segments) sourced from different libraries were weighted with more significance and proposed for future gene manipulations. Several targets were tested by forward metabolic engineering and the gAcrA production compared to control cells. Absolute quantitations of gAcrA were achieved by generating isotopically labelled (¹⁵N) gAcrA protein and spiking samples with of known concentrations of a chymotryptic peptide. Both normal (light) and labelled (heavy) gAcrA peptides were targeted by pseudo-SRM on an HCT Ultra PTM Discovery ESI-Ion Trap MS/MS (Bruker Daltonics, Coventry UK). Five genes were expressed in E. coli CLM24 and gAcrA production was increased by up to 10-fold (up to 10 mg L^-1). These genes were involved in sugar acquisition (uptake and synthesis), central carbon metabolism and energy generation.

Associated work aimed at creating human-type glycosylations in E. coli will be benefit by the improved glycosylation capability demonstrated here. The overall aim is to use this bacterial host as a factory for the production of glycosylated therapeutic human proteins by making production levels and costs competitive with current mammalian cells.

References

Nita-Lazar M, Wacker M, Schegg B, Amber S, Aebi M. 2005. The N-X-S/T consensus sequence is required but not sufficient for bacterial N-linked protein glycosylation. Glycobiology 15(4):361-367.

Sethuraman N, Stadheim TA. 2006. Challenges in therapeutic glycoprotein production. Curr. Opin. Biotechnol. 17(4):341-346.

Stephanopoulos G, Vallino JJ. 1991. Network rigidity and metabolic engineering in metabolite overproduction. Science 252(5013):1675-81.

Varki A. 1993. Biological roles of oligosaccharides: all of the theories are correct. Glycobiology 3(2):97-130.

Wacker M, Linton D, Hitchen PG, Nita-Lazar M, Haslam SM, North SJ, Panico M, Morris HR, Dell A, Wren BW and others. 2002. N-linked glycosylation in Campylobacter jejuni and its functional transfer into E. coli. Science 298(5599):1790-3.