Accelerated metabolomics for bioprocess development

Reiter, Alexander Marc Christopher; Oldiges, Marco (Thesis advisor); Blank, Lars M. (Thesis advisor); Wiechert, Wolfgang (Thesis advisor)

Aachen : RWTH Aachen University (2022)
Dissertation / PhD Thesis

Dissertation, RWTH Aachen University, 2022

Abstract

Microbial phenotyping and bioprocess development have been significantly accelerated by the use of small scale and highly parallelized cultivation platforms embedded in laboratory robotics, resulting in a bottleneck in bioanalytical bioprocess analytics. While microscale cultivation platforms allow the monitoring of typical process parameters, without comprehensive analytics, only limited information about product- and by-product formation is provided. Although liquid chromatography mass spectrometry can provide such a comprehensive and quantitative insight, it is often constrained by analysis runtime and throughput. For a quantitative or targeted approach, six methods for amino acid determination were developed and evaluated in this study, using two strong cation exchanger columns and a dilute-and-shoot flow-injection-analysis (DS-FIA-MS/MS) approach in hyphenation with either a triple-quadrupole or a quadrupole time-of-flight mass spectrometer. To compensate matrix effects, isotope dilution mass spectrometry with 13C15N labeled amino acids was utilized. A detailed method validation study confirms the versatility of the methods for metabolite profiling studies of microbial cultivation supernatants. For single analytes, the methods using chromatography columns demonstrated a linear range of approximately 4 orders of magnitude, sufficient response factors, and low quantification limits (7-443 nM). Overall, relative standard deviation for all analytes was comparable, with 8 % and 11 % for unbuffered and buffered media, respectively. The DS-FIA-MS/MS methods with an analysis time of 1 min per sample showed comparable performance but up to 35 times higher throughput. Metabolic footprinting is a holistic approach to gather large scale metabolomic information of a biological system, and it is thus a driving force in systems biology and bioprocess development. To keep up with the cultivation throughput, the ongoing development of automated cultivation platforms necessitates the development of a comprehensive and rapid metabolite profiling tool. For a semi-targeted approach, a workflow was developed to automatically provide and select relevant metabolite information from a genome scale model in order to build an organism-specific comprehensive metabolome analysis method. The deduced metabolite set was partitioned into methods for a chromatography-free multi-reaction-monitoring profiling approach with DS-FIA-MS/MS based on in-house, literature, and predicted metabolite information. The workflow was used to develop a method for Saccharomyces cerevisiae that allows to screen 252 metabolites in 7 minutes per sample. The method was validated using a commercially available yeast metabolome standard, and correctly identified up to 74.2 % of the annotated metabolites. In a first case study, three commercially available yeast extracts were screened, with 118 metabolites passing quality control thresholds for statistical analysis, allowing for the identification of distinguishing metabolites. The presented methodology performs metabolite screening in a time-efficient manner by scaling analysis time to metabolite coverage and is adaptable to other microbial systems by starting with genome scale model information. Corynebacterium glutamicum (C. glutamicum) is an industrial platform organism for the production of amino acids. Previously, the organism was used to produce L-Histidine, with research focusing on metabolic engineering methods to increase titer and yield. Only a few studies that provide information on bioprocess development have been published in this area, with medium optimization and fed-batch cultivation procedures being particularly promising areas for further research. As a result, bioprocess development for a C. glutamicum mutant that produces L-Histidine was carried out in order to assess the performance and applicability of the DS-FIA-MS/MS methods. The performance and applicability of the targeted DS-FIA-MS/MS method are demonstrated by high-throughput screening of a C. glutamicum strain library obtained through random mutagenesis. Medium sensitivity analysis in a parallelized and automated cultivation platform identified potential medium bottlenecks in the form of phosphate and magnesium availability. The process was transferred to laboratory scale and subsequently optimized by medium and feed enrichment in a fed-batch cultivation procedure, improving the L-Histidine titer by factor 5.8. The semi-targeted DS-FIA-MS/MS method allowed to identify accumulating L-Histidinol as a precursor for L-Histidine formation and intermediates of the methionine biosynthesis. Statistical analysis indicated that, in addition to the previously reported purine and glycine metabolism, methionine biosynthesis might be involved in L-Histidine formation via tetrahydrofolate cofactor regeneration. The developed and validated tools in form of cultivation protocols, robotic workflows and data processing pipelines allowed for the identification of the best producer, assessment of medium limitations, the transfer of the process to the laboratory scale, and the identification of potential targets for future metabolic engineering approaches in high-throughput.