Recombinant proteins are used to treat a variety of diseases including infections, autoimmune diseases and cancer. Most of these proteins require a range of post-translational modifications to ensure their correct function, safety and stability, and are therefore produced in mammalian cells. Compared to bacteria and yeast however, mammalian cell culture is expensive and productivities are low. Increasing the productivities of mammalian cell cultures would decrease the costs associated with the production of recombinant protein drugs, lowering the price of medical treatments and increasing their accessibility to the general public.
While significant improvements in product titres have been achieved in the last two decades through optimization of media composition and process conditions, attempts to improve mammalian cells via genetic engineering have been less successful (Chapter 2). One reason for this is, that while our knowledge of the specific cellular functions required for protein production (e.g., transcription, translation, folding, and secretion) is quite detailed, our understanding of how the individual parts of the global cellular machinery interact and give rise to a given phenotype is very limited.
An improved understanding of cellular interactions is best acquired by employing a systems biology approach combining a set of global measurements such as transcriptomics, proteomics, metabolomics and fluxomics. Each omics technique provides a different snapshot of the cellular machinery, and their combined use has the potential to elucidate the complex relationship between genotype and phenotype. Transcriptomics and proteomics are well established techniques and are routinely applied to investigate mammalian cells (Chapter 3). Due to post-transcriptional/post-translational modifications and regulation however, neither transcript nor protein abundance necessarily correlate with phenotype. Metabolite levels and fluxes on the other hand are the end-products of all cellular regulation, and accordingly are more representative of the phenotype. To allow evaluation of the effect of mRNA and protein changes on phenotype, transcriptomics and proteomics should ideally be combined with metabolomics and/or fluxomics. Unfortunately, metabolomics and fluxomics technologies are not well established and robust protocols need to be developed.
Two key objectives of this thesis were to develop experimental protocols for fluxomics and metabolomics. Robust protocols for the measurement of a range of extracellular metabolites (e.g., hypoxanthine, GlutaMAX, glucose, lactate), biomass composition, and cell dry weight of mammalian cells were developed (Chapter 4, Appendix B, C, D, E), enabling routine application of fluxomics to mammalian cells. To ensure the validity of these protocols, a fluxomics approach was used to investigate the underlying metabolic changes in CHO cells before and after a metabolic switch from lactate production to consumption. The results highlighted that no gross measurement error was present in the acquired data, and that fluxomics is a useful tool for the investigation of metabolic differences in mammalian cells. It was shown that TCA cycle fluxes during lactate production and consumption were similar, and as such, lactate consuming cells were metabolically more efficient than lactate producing cells (Chapter 5).
Metabolomics is one of the most demanding omics techniques with regards to sample preparation. Metabolites are chemically very diverse, which poses a substantial challenge for quantitative extraction. In addition, metabolites have a rapid intracellular turnover and it is necessary to quench metabolism in order to avoid changes in intracellular metabolite concentrations during sampling. In order to address these challenges, a range of quenching and extraction solutions were quantitatively evaluated for their suitability to the extraction of intracellular metabolites from industrially relevant mammalian cell cultures (Chapter 6). Quenching with ice-cold 0.9 % NaCl and extraction with 50 % acetonitrile was shown to be superior to all other investigated protocols.
Practical assessment of the metabolite extraction protocol was performed by investigating differences in the metabolite profile of CHO cells cultivated in three commercial media (Chapter 7). The media displayed different capabilities to support cell growth, with maximum viable cell densities and culture duration varying from medium to medium. Principal component analysis of intracellular metabolite concentrations resulted in three clusters corresponding to the different media, highlighting that differences in growth were reflected in the metabolite profile. A possible bottleneck at UDP-dehydrogenase was identified in the media with the slowest growth rate. Notably, for cells cultivated in the best performing medium, a unique metabolite was detected.
Finally, a combination of transcriptomics, metabolomics and fluxomics was employed to characterise the cellular mechanics of recombinant protein production in Hek293 cells. In this study, a protein producing Hek293 cell line was compared to its parental cell line in batch bioreactor cultures (Chapter 8). The protein producer had a lower glucose uptake rate than the non-producer, suggesting an improved metabolic efficiency as the same growth rate was observed. The majority of metabolic differences identified by flux analysis were mirrored at the mRNA and metabolite level. Gene expression profiles revealed many differences between the producer and non-producer, although the majority of these were small (FC < 1.5). A large number of differentially regulated genes belonged to general biological functions (e.g., cell cycle, proliferation, cell death), although no phenotypic differences were observed in this regard. Increased expression of XBP-1 suggested activation of the unfolded protein response in the producer cell line, potentially reflecting increased demand on the protein processing machinery and consequent adaptive response of the cells.
The protocols developed in this thesis pave the way for the application of fluxomics and metabolomics to mammalian cells. The practical applications of these two omics techniques were demonstrated through the characterisation of CHO cell lines exhibiting different metabolic and growth phenotypes. In addition, it has been demonstrated that the combined use of transcriptomics, metabolomics and fluxomics provides a better understanding of productivity in Hek293 cells.