Beyond rational-biosensor-guided isolation of 100 independently evolved bacterial strain variants and comparative analysis of their genomes

Abstract

Background: In contrast to modern rational metabolic engineering, classical strain development strongly relies on random mutagenesis and screening for the desired production phenotype. Nowadays, with the availability of biosensor-based FACS screening strategies, these random approaches are coming back into fashion. In this study, we employ this technology in combination with comparative genome analyses to identify novel mutations contributing to product formation in the genome of a Corynebacterium glutamicum L-histidine producer. Since all known genetic targets contributing to L-histidine production have been already rationally engineered in this strain, identification of novel beneficial mutations can be regarded as challenging, as they might not be intuitively linkable to L-histidine biosynthesis.

Results: In order to identify 100 improved strain variants that had each arisen independently, we performed > 600 chemical mutagenesis experiments, > 200 biosensor-based FACS screenings, isolated > 50,000 variants with increased fluorescence, and characterized > 4500 variants with regard to biomass formation and L-histidine production. Based on comparative genome analyses of these 100 variants accumulating 10-80% more L-histidine, we discovered several beneficial mutations. Combination of selected genetic modifications allowed for the construction of a strain variant characterized by a doubled L-histidine titer (29 mM) and product yield (0.13 C-mol C-mol^-1) in comparison to the starting variant.

Conclusions: This study may serve as a blueprint for the identification of novel beneficial mutations in microbial producers in a more systematic manner. This way, also previously unexplored genes or genes with previously unknown contribution to the respective production phenotype can be identified. We believe that this technology has a great potential to push industrial production strains towards maximum performance.

Keywords: Biosensor; Comparative genome analysis; Corynebacterium glutamicum; Fluorescence-activated cell sorting; Genome sequencing; High-throughput screening; L-histidine; Metabolic engineering; Single-nucleotide polymorphism.

Fig. 1

A Schematic representation of the plasmid-based pSenHis biosensor for intracellular detection of l-histidine in C. glutamicum. In the presence of elevated intracellular concentrations of l-histidine, the transcriptional activator LysG-A219L binds l-histidine and binds to its target promoter P_lysE to activate the expression of the eyfp gene encoding for the fluorescent reporter protein EYFP leading to fluorescent single cells. B Increased fluorescence response of single l-histidine-producing C. glutamicum CgHis2 cells during FACS upon supplementation of different His-Ala dipeptide concentrations. 0, 3, 10, and 30 mM l-His-l-Ala dipeptides were added to microtiter plate cultivations of C. glutamicum CgHis2 in defined CGXII medium with 2% d-glucose. In the middle of the exponential growth phase (after 6 h of cultivation), fluorescence intensities of 100,000 cells of each culture were analyzed by FACS. The depicted distributions represent standard deviations of the mean from three independent technical replicates showing the cell count at respective eYFP fluorescence (arbitrary units)

Fig. 2

l-histidine accumulation and biomass formation of 100 FACS-isolated C. glutamicum CgHis2 variants. All variants were independently generated by chemical mutagenesis, independently isolated in a biosensor-based FACS screening and characterized in detail. A l-histidine titer and biomass after 48 h (second characterization step). Error bars of variants represent three independent technical replicates. C. glutamicum CgHis2 reference strain (grey) is displayed as mean over ten independent cultivation rounds. Note: Due to technical variation of humidity parameters during the cultivation, variants were evaluated only in comparison to the respective triplicate of controls on the same microtiter plate. Therefore, comparison of variants from all cultivation rounds was performed by B percentage improvement of l-histidine titers in comparison to the specific C. glutamicum CgHis2 starting strain control. An increased product formation of 10% was set as threshold. Data represent average values and standard deviation of three independent technical replicates

Fig. 3

Total number of mutations per gene across 100 isolated C. glutamicum CgHis2 variants according to the position in the genome of C. glutamicum. High mutation rates of individual genes might indicate a link to the improved l-histidine production phenotype

Fig. 4

Growth and l-histidine production of C. glutamicum CgHis2 variants with different genomic modifications. C. glutamicum CgHis2 ΔfasB Δpyk served as starting variant for the construction of various strains combining three or four genomic modifications, which exhibited significant increase in l-histidine production (yellow). Δpyk always refers to the pyk1 gene. C. glutamicum CgHis2 ΔfasB Δcps served as starting variant for the construction of various strains with no improvement of product formation (Additional file 1: Fig. S9). Data represent average values and standard deviation of three independent technical replicates

Fig. 5

Lab-scale batch cultivations of A the C. glutamicum CgHis2 reference strain, B the FACS-isolated C. glutamicum CgHis2 12–10-5–6 variant, and C the reverse engineered C. glutamicum CgHis2 ΔfasB Δpyk1 NCgl2981-D735G pks-D1186N. Bioreactor cultivations were performed in duplicates using defined CGXII medium with 40 g L⁻¹ d-glucose as sole source of carbon and energy. Data points represent the average of two independent bioreactor cultivations