Large signal amplification in bacterial chemosensing due to a non-equilibrium phase transition with zero-order ultrasensitivity.*

Bacterial chemosensing is among the best studied sensory systems in biology. But two sets of experimental data are hard to reconcile with existing models. First, under certain conditions kinase activity measured via FRET undergoes all-or-none spontaneous switching. Second, while most models assume that kinase and receptor states are one and the same, past measurements of the receptor occupancy suggest that the binding of ligand is not cooperative and is unaffected by adaptation, in contrast to measured kinase activity. Motivated by the arrangement of receptors and kinases in a tight signaling array, we propose a new model where receptor and kinase states are distinct and live on edges and vertices of a lattice, respectively. Residing on the vertices, kinases spread both activity and inactivity to nearest neighbors through the edges of the lattice. Receptors, which sit on the edges between kinases, influence activity by modulating the ratio of these spreading rates. If this ratio is close to a critical value, kinase activity becomes sharp and switch like, mathematically analogous to zero order ultrasensitivity. This ultrasensitivity explains both the sharp dependence of kinase activity on the average receptor state and the large fluctuations necessary for coordinated all-or-none switching. Feedback from the kinase activity onto the spreading rates generates precise adaptation. Thus, our model explains these new results while remaining compatible with previous data.