Enhanced heterologous protein production in Streptomyces lividans by metabolic flux optimization.
White biotechnology has demonstrated a vast growth during the last decade, mainly thanks to its sustainable advantages compared to classical chemical processes. To maintain this growth, different micro-organisms are being studied as potential cell factories. Among these micro-organisms, Streptomyces lividans has received increasing attention as a host for the production of heterologous proteins. To enable rational strain improvement strategies, a thorough understanding of the impact of introducing heterologous genes in S. lividans is indispensable.
This dissertation focusses on understanding the metabolic burden in S. lividans TK24 caused by the introduction of a heterologous gene. This is achieved by comparing a 13C-metabolic flux analysis of a cellulase A producing strain and an empty plasmid bearing strain, yielding a quantitative characterisation of the central carbon metabolism during exponential growth. This dissertation reports the first implementation of 13C-metabolic flux analysis in the research group as well as the first 13C-analysis of a heterologous protein producing S. lividans overall.
Since flux estimation accuracy in 13C-metabolic flux analysis is heavily influenced by the used 13C-tracer (mixture), wet lab experiments are preceded by an optimal experimental design study. A complete framework for cost-effective design of 13C-tracer experiments is built and illustrated on two different networks to demonstrate its generic nature. Firstly, the search for the optimal design is implemented based on a linearised and a non-linear approach. In both cases, validation is based on a single scalar value, i.e the D-criterion and the S-criterion, respectively. It is demonstrated that both approaches lead to the same optimal design. Secondly, this optimal design is further complemented with a simultaneous minimisation of experimental cost. Optimal mixtures are obtained which pose a noticeable decrease in cost as compared to 100% 1,2-GLC, which was identified in previous research as the most effective tracer. This deviation from previous defined optimal mixtures highlights the importance for an a priori experimental design when studying a new organism or strain. The multi-objective approach also offers a way to evaluate the information loss when budget is constrained.
Flux distributions in the central carbon metabolism are finally calculated for a recombinant S. lividans strain, producing a thermostable cellulase, and a S. lividans reference strain. A comprehensive, stationary 13C-metabolic flux analysis is applied on both strains, including the application of the implemented optimal experimental design framework. Replicate batch tracer experiments are performed with the optimally determined glucose tracer mixture. Mass isotopomer data are obtained by setting up GC-MS protocols for the analysis of proteinogenic amino acids. Highest flux estimation accuracy is finally obtained by fitting the experimental data of two tracer experiments simultaneously, thereby accounting for biological variations. This framework now forms the basis for any future work on 13C-metabolic flux analysis with S. lividans.
A clear metabolic shift is observed in the recombinant protein producing strain compared to the reference strain, clearly indicating a metabolic burden. Most prominent differences are the increased secretion of organic acids, suggesting a possible imbalance in the production of precursor metabolites, and the increased flux through the pentose phosphate pathway and the citric acid cycle, leading to an excess cofactor generation.
The quantified impact of introducing a heterologous gene in S. lividans can now be applied in rational strain improvement strategies. An increased flux through the pentose phosphate pathway during enhanced protein production is confirmed to be a recurring observation in heterologous protein producing micro-organisms. Overexpression of these genes offers a first rational strain improvement strategy for S. lividans based on fluxomic analysis. In addition, the found flux map can be used as additional constraints in genome-wide genetic modification simulations. When implemented, the envisioned modifications of these engineered strains can easily be quantified due to availability of an in- house multi-objective framework for 13C-MFA.