Rapid development of small-molecule producing microorganisms based on metabolite sensors

Small-molecules made by microorganisms, such as amino acids, vitamins, organic acids or antibiotics are industrially important substances. However, there are two major limitations in microbial strain development. First, laborious plasmid constructions are usually involved in strain development. Second, no general high-throughput screening methodology exists to identify a producer at the single-cell level. In the present doctoral thesis these problems are addressed and applied to Corynebacterium glutamicum. The metabolite sensor pSenLys was constructed. It uses the transcriptional regulator LysG of C. glutamicum, as well as the promotor of its target gene lysE. Fusion with eyfp resulted in a graded fluorescence output in response to the cytosolic L-lysine concentration, which increases in strains with higher productivity. Turning the inconspicuous metabolite L-lysine into a conspicuous one enabled the high-throughput screening of producing cells via fluorescent activated cell sorting (FACS). A screening accuracy exceeding 91 % was determined by isolation of fluorescent cells out of a population consisting of non-producing cells in a 10.000 fold excess over producing cells. Furthermore, metabolite sensors were developed for the detection of L-serine and O-acetyl- L-serine in C. glutamicum and L-arginine in E. coli. Single-cell analysis using metabolite sensors and FACS of Llysine producing strains is demonstrated, opening up a number of different possibilities for microbial population analysis. The established screening routine was used to isolate 270 fluorescent cells from a randomly mutagenized population of C. glutamicum. 185 clones accumulated L-lysine in the range of 0.2 to 37 mM. Targeted sequencing of six genes from 40 of the 185 mutants resulted in 24 strains carrying known mutations, or mutations in known genes, whereas in 16 mutants no known gene was mutated. Sequencing the genomes of 10 mutants revealed that they carry between 36 and 268 SNPs. In one strain the UDP-MurNac-tripeptide synthetase was mutated resulting in MurE-G81E. Introduction of this, so far unknown, mutation into the genome of the wild type, and also into defined L-lysine producers, caused an increased L-lysine production in all strains. Consequently, murE is now the third gene in addition to lysC and hom which, when mutated alone, causes an increased Llysine production. Thus, the principle of the use of a metabolite sensor for high-throughput isolation of new producers and identification of new targets has been demonstrated successfully.