Systems optimisation of host cell tRNA usage and codon decoding for the improvement of bioprocessing parameters

Lead Research Organisation: University of Kent
Department Name: Sch of Biosciences


The decoding of genes during protein synthesis is a complex process that must occur with great accuracy in order for cells and organisms to remain healthy. Accurate protein synthesis is achieved through the interplay of many different molecules, including ribosomes (the molecular machines that actually achieve protein synthesis), tRNAs (adapter molecules that transport amino acids to the ribosome), and translation factors (helper proteins that establish the correct contact between ribosomes and tRNAs). In order to achieve accurate protein synthesis it is critical that the levels of each of these elements are matched exactly to the frequency with which they are used: if cells contain too much or too little of any of these elements, protein synthesis errors occur more frequently and cellular health declines. In normal cells that only produce proteins from their own genes, the protein synthesis system and levels of the molecules described above are optimised to achieve the required low error rates and high translational speed. However, in industrial applications additional genes are often introduced into cells with the aim of producing specific proteins that are not naturally produced by them. This strategy is used in hte pharmaceutical industry to produce the latest generation drugs against cancer, multiple sclerosis and arthritis. When cells make proteins from foreign or artificial genes, the protein synthesis machinery must deal with a situation for which it has not been optimised. We predict that this will increase error rates during the production of the relevant proteins. Protein synthesis errors have negative effects for the ease with which protein-based drugs can be purified and formulated following synthesis in the host cells, and may also adversely affect the potency of the final product. A second prediction we make is that, if we understood the principles of optimisation in detail, we might develop strategies that restore optimal protein synthesis and reduce error rates. Both predictions follow logically from existing knowledge of the translational machinery, although to date they have not yet been experimetnally tested and therefore we can not be completely sure whether they are true. Because our predictions on the relationship between optimised protein synthesis and expression of foreign proteins have important consequences for our ability to make high-quality protein-based drugs, we wish to test them in a small pilot study. We will develop computational models of protein synthesis that will help us to understand the principles of optimisation in protein synthesis. We will then use thes models to suggest strategies for achieving optimisation under conditions of foreign protein synthesis in a simple yeast-based expression system. Lastly, we will test experimentally whether these strategies do indeed improve the quality of proteins, and facilitate their processing following synthesis. If this pilot study confirms our predictions, we will use this as basis for a larger study in which we develop optimisation strategies for the various protein synthesis systems used in the pharmaceutical industry.

Technical Summary

A major research challenge in bioprocessing is to understand the relationship between host cell health, product quality and product performance during downstream processing, and how these parameters can be optimised. Decoding of the genetic code occurs through multiple interacting elements including tRNAs, ribosomes and mRNA codons, and these elements and their interactions are heavily optimised to maximise the speed of translation and cell fitness, while minimising errors in the decoding process. The introduction of additional genes expressed at very high levels, such as recombinant protein (rP) genes in modern expression systems, disrupts optimisation of the decoding system. This negatively affects cell health, rP quality, and rP behaviour during purification. Conversely, a detailed understanding of the principles of optimisation during protein synthesis enables us to re-establish optimisation under conditions of rP expression, thus improving all of the parameters listed above. Although connections between optimal decoding of the genetic code and rP performance parameters are strongly suggested by the current state of academic knowledge, there are no experimental data avilable that conclusively prove such connections. We therefore propose to use BRIC Enabling Fund resources to conduct a pilot study, where we will use a highly focussed set of effective whole bioprocess-modelling exercises combined with experimental work to investigate for a representative yeast-based expression system: a) whether loss of optimisation in the tRNA dependent decoding system during high-level rP expression is a source of problems with product performance, and b) whether model-based re-optimisation of the decoding system under conditions of rP expression can improve rP quality and rP homogeneity. If this pilot study was successful, a larger study would be proposed to develop more broadly applicable strategies (eg for mammalian and insect cell expression systems).

Planned Impact

This project will have both economic and societal impacts as well as training impact. The expected results of this and any possible follow-on project will have direct impact on improved health and well being. The direct aim of the proposed project is to improve the efficiency of the production of pharmaceutical drugs. The methods we propose to develop here are directly aimed at improving industrial processes in this area and we expect that they will lead to improved cost structures in the industry. This identifies two groups of beneficiaries: (1) The UK industrial/pharmaceutical sector which benefits from this research through an increased competitiveness. (2) Users of drugs (patients) will be indirect beneficiaries through reduced costs of drugs and potentially through the resulting improvements in availability. A direct aim of this project is to transfer skills and new methods to the industrial sector. Should this pilot project be successful, then we will place great emphasis on developing knowledge transfer strategies that enable industry to apply and extend the optimisation methods we propose. Specifically, we will seek use contacts and resources of BRIC but also of the Enterprise office of the University of Kent to identify concrete industrial users of the results of this project and transfer knowledge to them. Hence our project will enhance the research capacity of private and third sector organisations. The project does not limit itself to solving specific industrial problems but to develop general solving strategies with direct impact on the competitiveness of the individual industrial users. In this sense it is a direct contribution to the knowledge economy and increases the to the general wealth and wellbeing of the UK. The results of this project also have scope for commercialisation. Specifically, we anticipate that the intellectual property generated in this project will lend itself for to commercialisation. This could take the shape of a spin-off consulting company specialising on advising industry on modelling techniques in the context of rP expression. The project will employ a postdoctoral researcher (Radu Zabet) who will through his activites receive further competence in cross and inter-disciplinary work. While Mr. Zabet is currently receiving intense training within his chose field of study (modelling transcription processes) this additional expose to a related but still different set of research problems will be an additional training opportunity that will increase his overall skill. Hence, even though this project does not have an explicit training component, it contributes to the training of highly skilled researchers and helps to develop a workforce that is crucial for the UK knowledge economy. This project is highly interdiciplinary in that it uses both wet-lab experiments and mathematical/computational modelling to achieve a tangible practical goal. The development of innovative techniques and cross-disciplinary approaches is therefor crucial to the project and as such it will improve methodology and cross disciplinary approaches.
Description Many modern drugs are comprised of proteins. These include current frontline drugs against diseases like cancer, cardiovascular disease, multiple sclerosis, psoriasis and others. The efficient production of such drugs is both important for the health of patients and as an economic factor, since the market for protein-based drugs was estimated at nearly 150 billion dollars in 2010.

Protein drugs are large and complex chemical structures that differ strongly from older, so-called "small molecule" drugs. One of the main differences lies in their production: while small molecule drugs are produced via purely chemical processes, protein drugs are produced by introducing artificial genetic material into microbial or animal cells grown in culture. In response to the new genetic material, these cells then start to produce the new protein, which can be purified and given to patients.

Although this strategy works generally well, some poorly characterised consequences arise from the fact that host cells containing additional genetic material make more protein than they would normally do. This additional demand on the protein synthesis machinery could potentially lead to sick cells, and in turn result in drugs with undesirable properties and less than the best possible quality.

One of the key properties of genetic material is that slightly differing variants of a gene can produce the same protein. At the outset of our study, our hypothesis was that different genes making the same drug should affect host cells in different ways. Such differences could be exploited to minimise negative effects arising from the additional demand on the protein synthesis machinery in cells that produce protein drugs.

In our study, we initially constructed computer models of the translational apparatus of a particular type of host cell (yeast cells). By comparing predictions from our computer models to experimental measurements of protein synthesis in yeast cells, we ensured that these models were an accurate representation of protein synthesis in real cells. We then analysed simulations performed with the models, and found that upon introduction of additional genetic material, a particular resource called the ribosomes becomes limiting. This limitation can strongly impair the health of drug-producing host cells. Our simulations also predicted that the different genetic variants that produce the same drug impair cell health to different degrees, thus validating our initial hypotheses. All of the predictions from our computer simulations were then verified by targeted experiments.

In summary, we were able to identify mechanisms by which the production of proteins from artificial genetic material can negatively impact cells. We were also able to develop strategies for minimising this negative impact, thus enabling the generation of protein-based drugs with improved properties.
Exploitation Route Our computational models will be useful tools for improving sequence design strategies for recombinant protein encoding sequences.
Sectors Pharmaceuticals and Medical Biotechnology,Other

Description We have had several collaborative projects with industrial partners following this grant. These projects generally involved optimisation of test sequences for the respective partners. We have had proof-of-concept projects funded through our local commercialization department, through further BBSRC (BioProNET) funding and through direct industrial funding to adapt findings from this project to industrial recombinant protein expression systems.
First Year Of Impact 2011
Sector Manufacturing, including Industrial Biotechology,Pharmaceuticals and Medical Biotechnology
Impact Types Economic

Description Leverhulme Trust Research Project Grant
Amount £168,860 (GBP)
Funding ID RPG-2014-032 
Organisation The Leverhulme Trust 
Sector Charity/Non Profit
Country United Kingdom
Start 10/2014 
End 09/2017
Description BRAIN Biotech 
Organisation Biotechnology Research and Information Network AG
Country Germany 
Sector Public 
PI Contribution This partnership evaluated whether the sequence optimisation tools developed during the grant can be adapted to optimise the interaction of signal sequences of secreted proteins. We suggested sequence designs based on our computational models.
Collaborator Contribution The partner synthesized the corresponding sequences, tested expression levels, and shared results with us.
Impact No outputs to date.
Start Year 2015
Description Novozymes 
Organisation Novozymes Biopharma UK ltd
Country United Kingdom 
Sector Private 
PI Contribution Confidential Disclosure Agreement signed with Novozymes Biopharma UK Ltd, for a short project testing some ideas arising from the BRIC project.
Collaborator Contribution The partner synthesized DNA constructs.
Impact No outputs.
Start Year 2012
Description Resource allocation in the yeast translation system during recombinant protein expression 
Form Of Engagement Activity A talk or presentation
Part Of Official Scheme? No
Geographic Reach International
Primary Audience Professional Practitioners
Results and Impact Invited presentation at Novozymes Biopharma, Nottingham, UK. Invited presentation at Novozymes Biopharma, resulting from initial contacts made at the BRIC dissemination meeting in October 2011.

no actual impacts realised to date
Year(s) Of Engagement Activity 2011
Description Systems optimisation of host cell tRNA usage and codon decoding for the improvement of bioprocessing parameters 
Form Of Engagement Activity A talk or presentation
Part Of Official Scheme? No
Geographic Reach National
Primary Audience Professional Practitioners
Results and Impact Close-out presentation at the BRIC dissemination event

no actual impacts realised to date
Year(s) Of Engagement Activity 2011