Evolutionary optimization of bacterial motility and chemotaxis pathways

Information storage and processing in bacteria are mediated by cellular networks. While molecular details underlying the functioning of many bacterial networks are well known, the constraints under which their organization and regulation were optimized during evolution remain less understood. In this study, we approached the problem from two related angles. First, we investigated the limits of proteomic resource allocation to a cellular function, using the Escherichia coli (E. coli) motility pathway as a model system. Second, we explored the limits of network robustness to changes in the relative abundance of its components (i.e., stoichiometry), with the E. coli chemotaxis pathway serving as a benchmark example of a simple, well-characterized signal transduction network. To comprehensively characterize the motility-growth trade-off over a wide range of flagellar expression levels, we quantified the relation between the expression of flagellar genes and motile behavior, as well as the impact of motility on the growth fitness of E. coli. We further compared strategies of resource investment used by a number of laboratory strains and natural isolates of E. coli. We demonstrate that major limitations on resource investment in motility, at both high and low levels of gene expression, arise from hydrodynamic constraints on bacterial swimming. Although bacterial motility initially increases with the investment in the expression of flagellar genes, it saturates at high levels of expression despite further elevation in the number of flagella. Mathematical modelling suggests that this saturation is due to the physical limitation on E. coli’s ability to enhance its swimming by adding more than a certain number of flagella. Together with the fitness cost of flagellar biosynthesis and operation, this creates the physiologically relevant range within which the expression level of flagellar genes can enhance motility at the cost of growth fitness. Within this range, strategies of resource allocation towards motility depend on the medium, growth rate, and E. coli isolate. In the nutrient-rich medium, E. coli MG1655 and other tested K-12 strains appear to maximize their motility up to the physical limit while avoiding an excessive cost of expression beyond this limit. Motility-growth trade-off is also observed during carbon-limited growth, where motility is not maximized, but instead varies inversely with growth rate and appears to correlate with the level of motility at which the benefit of chemotaxis toward additional nutrients saturates. Furthermore, the expression of flagellar genes becomes bimodal at very low expression levels, possibly to avoid the production of poorly motile cells. Finally, the motility of natural E. coli isolates is limited to the same range as observed in K-12 derivatives, although its levels vary between isolates and, in some cases, are sensitive to the mechanical properties of the growth environment – probably reflecting the niche-specific selection. To address the second question, we constructed a library of the E. coli chemotactic pathway, in which we systematically varied the expression levels of individual genes using the ribosome binding sites (RBSs) of different strength, and estimated the functionality of multiple pathway variants. By connecting operon compositions of the candidates to their phenotypes, we revealed several prominent patterns underlying their performance (i.e., the ability to spread in soft agar). In particular, the expression levels of the regulatory proteins CheR and CheZ, which modulate the activity of the input (CheA) and output (CheY) modules of the network – appeared to have the strongest impact on pathway functionality. Besides, while the coordinated expression of antagonistic components – previously shown to be critical for the pathway functionality – did play a role, it was not essential, suggesting that overall pathway performance may involve higher-order interactions that remain to be explored.