|dc.description.abstract||Biotechnology has the potential to play a key role in the transition from an economy based on non-renewable fossil fuels to a greener economy in which renewable feedstocks can be used to generate a wide range of products. Fermentation monitoring in real-time can provide detailed biological and engineering information about the bioprocess, which is a very valuable asset for process development and is an instrumental tool for product manufacturing, allowing to quickly detect and react to deviations from desired process specifications, potentially preventing bioprocess failure and the loss of capital and resources.
While many different parameters can be monitored in a bioprocess, the monitoring of metabolites is particularly useful, for these provide the best representation of the phenotypic state of a living organism at any specific time. The aim of this thesis is to explore the use of metabolomics to monitor the metabolites present in the liquid phase of a bioreactor in real-time. Metabolomics offers the possibility to simultaneously detect hundreds of compounds without the need to develop complex and laborious chemometric models. To study this technology, a fluidics system was developed coupling a bench-top bioreactor to a mass spectrometer, allowing automatic on-line metabolomics analysis every five minutes.
The on-line metabolomics analysis system was implemented with an Escherichia coli (E. coli) fermentation process of succinate production and 886 different ions were monitored in an untargeted fashion using an Exactive Orbitrap mass spectrometer. This bioprocess was also analysed by liquid chromatography - mass spectrometry (LC-MS) in order to better characterise the signals monitored by untargeted on-line metabolomics, find process biomarkers and identify potential strain engineering targets. Being able to monitor hundreds of metabolites with a high time resolution can be a very powerful tool for process development, allowing the comparison of different strains and cell lines, as well as identifying by-products and bottlenecks without requiring extensive analytical calibration.
Based on the on-line metabolomics results and the LC-MS analysis, metabolites of interest for the bioprocess were identified and this information was used to develop a targeted on-line metabolomics method. Forty-one metabolites were monitored on-line in a targeted fashion for three E. coli succinate fermentation replicates and univariate linear regression models were built to correlate the mass spectrometry signal to the concentration of 12 metabolites and the biomass in the bioreactor, demonstrating how this technology can be used to monitor the concentration of multiple bioprocess metabolites of interest on-line. These regression models were built by splitting the triplicate fermentation data into a training set and a validation set, and model performance was statistically assessed. Importantly, regression models for the 41 metabolites monitored could have been developed with further analytical calibration. Therefore, metabolomics allows the monitoring of a significantly larger amount of compounds compared to alternative technologies, such as vibrational spectroscopy.
Finally, a kinetic model was built to describe the aerobic batch phase of the succinate bioprocess, allowing improved process understanding. Furthermore, the model was employed in combination with targeted on-line metabolomics to forecast the future evolution of key process compounds at different stages of the process, showing how on-line metabolomics can be used as a forecasting tool during product manufacturing.
In summary, this thesis demonstrates the use of on-line metabolomics for bioprocess improvement in two main ways. Firstly, on-line metabolomics can be used to monitor hundreds of metabolites at a very high time resolution without requiring prior calibration or process knowledge (untargeted method). This is a very powerful tool for bioprocess development, allowing the comparison of different strains, cell lines and process conditions at a much faster throughput than alternative technologies, particularly when LC-MS is used to complement the on-line metabolomics data. Secondly, on-line metabolomics can be used to monitor the concentration of specific metabolites (targeted method), which has great potential in improving the performance of product manufacturing in the biotechnology industry, for example by using predictive kinetic models in real-time.||en