Modeling heterogeneity, commitment, and memory of bacterial spore germination.

Spores of bacteria are metabolically dormant, resistant to microbicides, and vectors of food spoilage and diseases and survive for years in their dormant state. Upon exposure to nutrient germinants, spores can rapidly return to life through germination, losing their resistance and becoming easy to kill. Despite extensive research on germination heterogeneity, commitment, and memory, many mechanisms underlying germination of Bacillota spores remain unclear, as a comprehensive mathematical model describing germination characteristics of individual spores is lacking. Woese et al. (PNAS, 59:869, 1968) developed a simple model predicting that time-to-germination of a spore with active enzymes is proportional to 1/. Here, we present a novel approach inspired by artificial neural networks to model spore germination, treating it as a decision-making process upon the activation produced by binding of germinants to germinant receptors. Major findings include the following: (i) using a sigmoid activation function to model germination thresholds allows predictions of distributions in time-to-commitment and kinetic germination to be well fitted to experimental observations; (ii) modeling spore commitment and memory after two separate germinant pulses fits well to experimental data of spores germinated with L-alanine, and a zero fraction of germinated spores by a second pulse is predicted by loss of memory; and (iii) modeling kinetic CaDPA release from individual spores through SpoVA channels and fitting experimental data of spores. This work enhances our understanding of unexplored biophysical intricacies of spore germination, and the use of the model may generate new data.IMPORTANCESpore germination is a crucial process through which spores of bacteria return to life when triggered by germinants, and some spore species cause food spoilage, human diseases, and bioterrorism. Understanding and theoretical predictions of spore germination could facilitate the development of "germinate to kill" strategies as spores lose their resistance upon germination. Here, we developed a novel mathematical model to describe the characteristics of spore germination including heterogeneity, commitment, memory, and kinetic CaDPA release using an artificial neural network. This model predicts new aspects of germination such as the retention and loss of memory and the effect of GRs' distribution on germination rate and could be useful in data-driven discoveries to enhance our understanding of germination's biophysical intricacies.