Optimal non-equilibrium decision making to store immune memory

Our adaptive immune system consists of highly diverse immune receptors to mount specific responses against a multitude of pathogens. During an infection, a fraction of these receptors forms a memory for later encounters. Here, we address a key question: which of the immune receptors should be kept as memory so that they can mount a response against evolved forms of the original pathogen in future infections? To do so, we have developed a theoretical framework, where memory storage is a non-equilibrium decision-making process between an adaptive exploration to mount a specific response and exploitation of existing yet suboptimal memory that can be utilized immediately to suppress an infection. To achieve a long-term benefit for the host, we show that memory generation should involve feedback from receptors’ affinity and should favor cross-reactive receptors with a moderate affinity over high-affinity receptors against the infecting pathogen. The recipe for memory generation should be tuned over the host’s evolutionary timescale based on the pathogenic evolutionary rates. Our results are consistent with recent experiments that suggest cell fate decisions during memory generation are highly regulated to balance the affinity and cross-reactivity of immune receptors.