Transcriptomic and proteomic analysis of lycopene-overproducing Escherichia coli strains

Abstract

Systems biology represents a powerful method to describe and manipulate phenotypes of interest by incorporating biological information from various levels of cellular organization. Such an approach is illustrated from a library of both rationally-directed and combinatorial gene knockout strains of E. coli recombinantly producing the small molecule lycopene. Global genomic and proteomic expression changes associated with increased lycopene production of mutant E. coli constructs were discovered using whole-genome DNA microarrays and a novel LC-MS technique, respectively. While most genes and proteins showed few expression changes, key differences were identified, including targets distal to the non-mevalonate and precursorsupplying pathways. Based upon the expression data sets, it was hypothesized that the following may be associated with lycopene overproduction: histidine biosynthesis (hisH); the quinone pool (wrbA); acid resistance (ydeO and gadE); the glyoxylate pathway (iclR); NADPH redox balance (pntB); growth rate reduction; and membrane composition. In the pre-engineered background strain, deleting pntB (~20-25%) and ydeO (~30%) each led to moderately increased production; overexpressing wrbA led to 50-100% more production at 8 hours and 5-15% more production at later time points; deleting iclR caused small production increases (~5-10%); and supplementing media with histidine caused the parental and mutant strains to have similar production.

(cont.) From these observations, several themes emerged. First, reduced cellular growth and energy conservation appear to be important tradeoffs for increasing lycopene production. Second, reducing overflow metabolism to acetate and corresponding acid stress as well as providing a gluconeogenic flux to increase lycopene precursors appeared beneficial. Next, NADPH availability and balance seemed to be critical production factors. The sS factor is known to affect lycopene accumulation, and it was observed to have far-reaching effects on both the transcriptomic and proteomic data sets. While expression changes were not strictly additive between the five mutant strains examined in comparison to the pre-engineered background strain, a number of these common factors appear to be responsible for the high lycopeneproduction phenotype. This work serves as an important example of incorporating multiple layers of complementary biological information to define a basis for an observed phenotype, demonstrating a powerful paradigm for realizing production increases via systems metabolic engineering.