High-Throughput Screening in Metabolic Engineering for Improved Production Strains

Abstract

Living cells are endowed with biochemical reaction networks that define the attainable in vivo conversions of substrate molecules to reaction products. By analyzing and modifying these reactions, mainly catalyzed by enzymes, the cellular features can be intentionally improved for a particular reaction pathway or even new features are added by recombinant DNA technology. This metabolic engineering champions for example the production of chemicals by microorganisms and lays the foundation for the development of bioprocesses for products naturally not made at economic titers or purities. The vast and ongoing progress made in the fields of systems and synthetic biology underpins novel metabolic engineering strategies and ultimately fuels the development of urgently required environmentally benign bioprocesses for replacing processes utilizing limited fossil resources in unsustainable ways. In this thesis, metabolic engineering of two industrially relevant microorganisms for the showcase production of isopentanol, an alcohol currently derived from petrochemistry is addressed. To this end Escherichia coli, arguably the incumbent workhorse in biotechnology, and Pseudomonas putida, an up-and-coming microorganism for metabolic engineering, were equipped with a pathway for alcohol production originating from yeast. For both microbes substantial product titers made from glucose as input substrate were found. However, further analysis showed that with P. putida also isovaleric acid was made as a major side product. This could be attributed to the versatile metabolism of this bacterium, which is per se a feature, but needs further optimization for improved isopentanol titers. To address the side product formation at the process level, a two-stage fed-batch protocol was developed. Limiting oxygen supply during the production phase allowed to improve isopentanol titers as well as curbing the short-chain fatty acid production. Second, with the development of a protocol for constructing sRNA libraries we investigated a potential solution for side product formation at the molecular level. sRNAs allow the knock-down of gene expression by specifically limiting translation of a target mRNA to the corresponding protein. As the precise enzyme(s) responsible for a side reaction are often unknown, for instance here the ones catalyzing the reaction to the unwanted fatty acid, naïvely targeting all potential targets and then screening for improved strain properties is a valid option. To this end, a simple computational workflow was developed for the conversion of genome annotations into DNA oligomers encoding sRNAs against these annotated genes. Due to increasingly cheap on-chip, pooled DNA synthesis such vast libraries can economically be produced. However, as this genetic engineering side is increasingly powerful, screening the vast corresponding strain libraries for improved product titers is to an increasing degree throughput-limiting. Small molecule products, including isopentanol, are analytically relatively inconspicuous and therefore traditionally require chemical analyses for detection. These analyses, usually by liquid or gas chromatography, quickly become the limiting factor when dealing with large strain libraries. This frequently occurring problem was addressed by the development of a biosensor circuit, allowing the formation of an easily read-out signal as a proxy for the actual alcohol titer. Here, the underlying genetic blueprint consisted of a transcription factor, which is activated by the product molecule of interest and subsequently binds to its cognate promoter for expression of a green fluorescent protein. As there are few transcription factors available encoding this functionality for isopentanol, a related transcription factor (AlkS) was chosen as a starting point for engineering this functionality by means of directed evolution. Libraries of alkS variants were screened by fluorescence assisted cell sorting and variants encoding the sought after specificity were found. The corresponding biosensor variants were characterized at the single-cell level for the detection of multiple industrially relevant alcohols. Besides, the sensor system was successfully applied to real-life detection of isopentanol produced by microbes. First, a detailed automation-based protocol relying on liquid handling robots was established. Subsequently, this protocol allowed for successful screening of an overexpression pathway library for improved isopentanol titers in an E. coli strain equipped with the biosensor circuit. Second, for a P. putida isopentanol production strain co-cultivated with an orthogonal E. coli biosensor strain, product dependent biosensor output was demonstrated. In conclusion, this thesis work evaluated two industrially relevant microbes for their potential in isopentanol production and concomitantly developed methods for strain library generation as well as for automated library screening enabled by biosensor circuits.