(469e) Designing a Synthetic Golgi Reactor

Royle, K., Imperial College London
Jimenez del Val, I., Imperial College London
Polizzi, K. M., Imperial College London
Kontoravdi, C., Imperial College London

Recombinant gene expression has revolutionised the pharmaceutical and healthcare industry; in 2011, 16 of the top 50 selling drugs worldwide were derived from recombinant technology [1] with sales of $53.4 billion in the US [2].

Glycosylation, the post-translational covalent addition of oligosaccharide chains to the protein structure, is a critical quality attribute of biopharmaceuticals. Indeed, glycosylation affects protein folding, stability, immunogenicity and functionality. Of the 212 currently approved biotherapeutics, 78 are glycosylated [3]. There are two prominent forms of glycosylation, N-linked and O-linked glycosylation, of which the former is a paramount concern in biotherapeutics. In essence, it stems from the addition of a core glycan GlcNAc2Man9Glc3 onto a sequence of Asn-X-Ser/Thr residues in the polypeptide. Subsequently, the core structure goes through a variety of trimming and addition reactions to produce the particular glycan structure [4].

As with all biological processes, heterogeneity occurs in the structures produced. Variations occur in the addition and trimming processing reactions, resulting in moderately different glycan structures, affecting the pharmacokinetic behaviour of the product. The latter is extremely important, such that the consistent glycosylation of biotherapeutics is a real challenge to industry.

To target this, we are designing and developing a synthetic Golgi reactor which will disengage the trimming and addition reactions from the cell. First and foremost, this allows protein production in a microbial host, greatly enhancing yield. Secondly, extracting the process from the cell allows greater control thereby reducing heterogeneity. Finally, it provides flexibility, such that therapeutics with optimised glycosylation profiles for enhanced functionality and stability can be designed.

The initial glycan structure will be transferred in vivo, by a partially ‘humanised’ strain of Pichia pastoris. The early glycosylation reactions are common between yeast and mammalian systems, up to the production of a Man8GlcNAc2 structure. There the two pathways diverge: yeast glycoproteins undergo substantial mannosylation while the human profile requires removal of mannose and addition of N-acetylglucosamine residues. In the ‘humanised’ strain of P. pastoris, the initial enzyme of the mannosylation cascade (α-1-6-mannosyltransferase) has been knocked out. In its place, a mouse α1,2-mannosidase IA has been inserted. Consequently, recombinant proteins produced with this system contain N-linked glycans with the core, human Man5GlcNAc2 structure.

To complement the in vivo core glycosylation, we are designing an in vitro synthetic Golgi reactor which links upstream protein production with downstream processing. The reactor was initially conceived of as a series of modular columns, each containing a glycosylation enzyme immobilised in a peptide hydrogel. The three enzymes selected were N-Acetylglucosaminyltransferase-1 (GnTI), Alpha Mannosidase 2 (MsnII) and N-Acetylglucosaminyltransferase-2 (GnTII) as these are the first three enzymes in the human glycosylation profile.

The dimensions and process parameters presented an optimisation problem. For instance, longer columns provide a higher residence time, increasing the percentage of oligosaccharide conversion. Long columns, however, encounter problems with pressure drop. Similarly, low flow rates can also increase oligosaccharide conversion but such a strategy would negatively impact yield and some reactions are affected by co-product inhibition. In contrast to a trial and error experimental optimisation of parameters, we have implemented the kinetics from a previously validated model of glycosylation in the Golgi apparatus published by this group [6] in the reactor scenario. This allowed the optimisation of flow rate, column length and enzyme concentration.

The in silico model provided a first approximation of the reactor design; to start to develop the reactor in practice, we have expressed the three glycosylation enzymes. These enzymes have been cloned into a commercial P. pastoris expression vector with an N-terminal truncation removing the transmembrane region. In its place, we have incorporated the specific peptide sequence for incorporation into a peptide hydrogel [7]. Preliminary characterisation of the enzymes has been carried out, using a combination of fluorogenic substrates assays, HPLC and MALDI TOF/TOF MS confirmation of product structure. Indeed, as the kinetics of the glycosylation enzymes are derived from in solution studies, it was important to analyse their behaviour when immobilised to hydrogels. For proof-of-concept, we are using the K3 Kringle fragment as a substrate ‘therapeutic’ for the tethered glycosylation enzymes.


[1] EvaluatePharma (2012) World Preview 2018: Embracing the Patent Cliff.

[2] Aggarwal, S. (2012) Nat Biotechnol, 30, 1191-1197.

[3] Kyriakopoulos, S. & Kontoravdi, C. (2013) Eur J Pharm Sci, 48, 428-441.

[4] Kornfeld, R. & Kornfeld, S. (1985) Annu Rev Biochem, 54, 631-664.

[5] Christen, M. T., Frank, P., Schaller, J. & Llinas, M. (2010) Biochemistry, 49, 7131-7150.

[6] Jimenez del Val, I., Nagy, J. M. & Kontoravdi, C. (2011) Biotechnol Prog, 27, 1730-1743.

[7] Roberts, D., Rochas, C., Saiani, A. & Miller, A. F. (2012) Langmuir, 28, 16196-16206.