Studies on bioprocesses for protein secretion with Corynebacterium glutamicum

After being discovered 60 years ago, Corynebacterium glutamicum is a major industrial workhorse for the production of amino acids like L-glutamate and L-lysine at several million tons per year. Intense and still ongoing basic and applied research fueled this great biotechnological success story. Currently, C. glutamicum is increasingly getting into focus as production host for heterologous proteins of both technical and clinical interest. Consequently, research is shifting towards the use of C. glutamicum microbial cell factory for protein production.To facilitate rapid and reliable characterization of newly constructed strain variants from metabolic engineering, microbioreactor (MBR) systems emerged as versatile tools. Such systems provide an increased experimental throughput with the ability to control environmental cultivation parameters, as well as monitoring of culture dynamics like biomass formation. In this study, methods for MBR systems are developed to suite the demands of a specific screening objective, namely the quantitative phenotyping of C. glutamicum strains secreting green fluorescent protein (GFP) and cutinase as heterologous model proteins. This involves the careful selection of standard operating conditions with respect to physiological demands like oxygen-unlimited metabolization of glucose as main carbon source, as well as determination of accuracy and imprecision of online biomass monitoring. Furthermore, to obtain time-resolved data on secretory protein formation and substrate consumption, an improved method is presented and validated that does not accompany a loss in MBR cultivation throughput. The need for (semi-)automated data processing from high-throughput MBR cultivations is also discussed on the example of derived performance indicators (PIs) that represent condensed evaluation metrics for rapid evaluation of whole cultivation experiments. As application example, a method for automated growth rate determination is presented in detail, since this PI is probably the most often applied characteristic in biological fitness testing of mutant strain libraries. In addition, on the example of maximizing secreted GFP titer, it is shown how MBR systems, integration of laboratory automation and Kriging-based Design of Experiments (DoE) complement each other in a synergistic way. As a result, an iterative workflow is presented that serves as blueprint for development of further biotechnological applications. Unexpectedly, secreted GFP titer could be doubled, showing that routinely applied nutrition media designed for amino acid production with C. glutamicum need to be carefully adapted and optimized with changing screening objectives, that is here secretion of heterologous proteins. To complement the current knowledge on how to select the optimal signal peptide (SP) for different expression hosts and different target proteins of choice, the interrelation of bioprocess control strategy and choice of SP to optimize cutinase secretion with C. glutamicum is investigated in detail. Since the envisaged degree of process control could not be realized with the available MBR systems, a consistent data set was generated relying on more than 150 bench scale bioreactor runs. Furthermore, the results are discussed and interpreted in the light of changing bioprocess optimization objectives, which again highlights the need for careful definition of optimization objectives. Representing a typical application example of MBR systems, the quantitative microbial phenotyping of a library of genome reduced C. glutamicum strains for heterologous cutinase secretion was conducted. The collected data comprise growth rates and cutinase yields as extracellular phenotypes, as well as detailed analysis at the transcriptome and proteome level for a small subset of strains. Next to surprising phenotypes due to specific genomic deletions, as well as differential analysis of phenotypes from strains with overlapping genomic deletions, attempts were made to explain the metabolic perturbations from observed significantly differential regulation at the protein level. Also, by incorporating all data on extracellular phenotypes, a data-driven, phenomenological multiple regression approach was used to identify the minimum set of genomic deletions needed in terms of improved cutinase secretion. Finally, a few future aspects outreaching the scope of this work are presented as outlook. These aspects concern the application of C. glutamicum as potential alternative host for heterologous protein production, the demand for further development of microbioreactor systems and the need for smart solutions for warehousing, (re-)processing and interpretation of heterogeneous data sets to cope with the foreseeable increase of information output generated from high-throughput experimentation in combination with powerful analytical methods