A simple metabolic architecture allows near-optimal adaptation to rapidly fluctuating environments

Bacteria live in environments that are continuously fluctuating and changing. Those fluctuations are usually not purely random but to some extent predictable. Exploiting this predictability can lead to an increased fitness. On longer timescales bacteria can "learn" the structure of these fluctuations through evolution. However, on shorter timescales, inferring the statistics of the environment and acting upon this information would need to be accomplished by physiological mechanisms. Here, we use a model of metabolism to show that a simple generalization of a ubiquitous regulatory motif (end-product inhibition) is sufficient both for learning complex continuous-valued features of the environment and for translating this information into predictive behavior; moreover, it accomplishes these tasks near-optimally. We discuss plausible genetic circuits that could instantiate this learning, and argue that the key ingredients required for such predictive behavior are readily accessible to bacteria.