Design and application of metabolite sensors for the FACS-based isolation of feedback-resistant enzyme variants

Microbes are used since decades for the industrial conversion of sugar feedstock and other renewable carbon sources to important bulk or fine chemicals. Nowadays, food and feed ingredients like amino acids and vitamins are biotechnologically produced in ever increasing quantities. Microorganisms like $\textit{Escherichia coli}$ or $\textit{Corynebacterium glutamicum}$ are used in these production processes, but– except of fermentation products – they are not naturally evolved for profitable metabolite formation. Thus, strain development is an essential prerequisite for the establishment of microbial production processes and can be performed by a number of approaches, such as metabolic engineering or undirected mutagenesis and screening. In the latter approach, large clone libraries are easily generated, but the bottleneck are efficient screening strategies to identify the desired variants with improved metabolite production. Such screening techniques should preferably rely on an optical signal that reflects increased product formation and enables the use of fluorescence activated cell sorting(FACS), which currently allows the highest throughput with more than one billion clones per assay. To generate the desired optical signal, intracellular metabolite sensors have been developed, which transform intracellular product concentrations into a graded optical output. These sensors are basedeither on RNA sequences or on transcription factors, which bind specifically to small molecules and inresponse to that, drive transcription of a reporter gene. In this study, new transcription factor-based metabolite sensors for the monitoring of intracellular concentrations of different amino acids in $\textit{C. glutamicum}$ or $\textit{E. coli}$ were constructed. The sensor pSenLys is based on the transcriptional regulator LysG of $\textit{C. glutamicum}$ and activates transcription of its target gene $\textit{lysE}$ in the presence of elevated intracellular concentrations of L-lysine, L-arginine, and L-histidine. This sensor was used to screen plasmid libraries generated by either undirected or sitedirected mutagenesis of ArgB (N-acetylglutamate kinase), HisG (ATP phosphoribosyl transferase) and LysC (aspartate kinase), representing feedback-inhibited key enzymes in arginine, histidine, and lysine biosynthesis. Using FACS, productive variants of these enzymes were isolated and characterized $\textit{in vitro}$ and $\textit{in vivo. C. glutamicum}$ strains carrying the best variants secreted up to 45 mM lysine, 34mM arginine and 17 mM histidine into the medium. LysG of $\textit{C. glutamicum}$ belongs to family of LysR-type transcriptional regulators (LTTR). As for LysG, transcriptional activation of most LTTRs depends on the presence of a co-inducer, which is often the product or an intermediate of the regulated pathway. To date, no structure of a full-length LTTR cocrystallized with its co-inducer is available. In this study, LysG was crystallized without ligand and cocrystallizedwith its effector L-arginine and structural models were generated. The overall structureof LysG corresponds to known LTTR structures consisting of an N-terminal DNA-binding domain, an α-helical linker, and two C-terminal effector-binding domains. Four LysG molecules are present in the unit cell, assembled into a tetramer (dimer of dimers). The structural model of the LysG-arginine complex enabled the identification of side-chains in the effector binding domains, which are involved in arginine binding. Structural differences between the models for LysG and LysG-Arg were observed, which disclosed a communication pathway from the effector-binding to the DNA-binding domain, whose binding affinity to DNA could be modulated by effector-interaction