The apparent clock-like evolution of Escherichia coli in glucose-limited chemostats is reproducible at large but not at small population sizes and can be explained with Monod kinetics.

To follow and model evolution of a microbial population in the chemostat, parameters are needed that give an indication of the absolute extent of evolution at a high resolution of time. In this study the evolution of the maximum specific growth rate ( micro (max)) and the residual glucose concentration was followed for populations of Escherichia coli K-12 under glucose-limited conditions at dilution rates of 0.1 x h(-1), 0.3 x h(-1) and 0.53 x h(-1) during 500-700 h in continuous culture. Whereas micro (max) improved only during the initial 150 h, the residual glucose concentration decreased constantly during 500 h of cultivation and therefore served as a convenient parameter to monitor the evolution of a population at a high time resolution with respect to its affinity for the growth-limiting substrate. The evolution of residual glucose concentrations was reproducible in independent chemostats with a population size of 10(11) cells, whereas no reproducibility was found in chemostats containing 10(7) cells. A model based on Monod kinetics assuming successive take-overs of mutants with improved kinetic parameters (primarily K(s)) was able to simulate the experimentally observed evolution of residual glucose concentrations. Similar values for the increase in glucose affinity of mutant phenotypes (K(s(mutant)) approximately equal 0.6 x K(s(parent))) and similar mutation rates per cell per generation leading to these mutant phenotypes (1-5 x1 0(-7)) were estimated in silico for all dilution rates. The model predicts a maximum rate of evolution at a dilution rate slightly below micro (max)/2. With increasing and decreasing dilution rates the evolution slows down, which also explains why in special cases a selection-driven evolution can exhibit apparent clock-like behaviour. The glucose affinity for WT cells was dependent on the dilution rate with highest values at dilution rates around micro (max)/2. Below 0.3 x h(-1) poorer affinity was mainly due to the effects of rpoS.