Mathematical modelling of activation-induced heterogeneity reveals cell state transitions underpinning macrophage responses to LPS

Despite extensive work on macrophage heterogeneity, the mechanisms driving activation induced heterogeneity (AIH) in macrophages remain poorly understood. Here, we use two in vitro cellular models of LPS-induced tolerance (bone marrow-derived macrophages or BMDMs and RAW 264.7 cells), single-cell protein measurements, and mathematical modelling to explore how AIH underpins primary and secondary responses to LPS. We measure expression of TNF, IL-6, pro-IL-1{beta}, and NOS2 and demonstrate that macrophage community AIH is dependent on LPS dose. We show that altered AIH kinetics in macrophages responding to a second LPS challenge underpin hypo-responsiveness to LPS. These empirical data can be explained by a mathematical 3-state model including negative, positive, and non-responsive states (NRS), but they are also compatible with a 4-state model that includes distinct reversibly NRS and non-responsive permanently states (NRPS). Our mathematical model, termed NoRM (Non-Responsive Macrophage) model identifies similarities and differences between BMDM and RAW 264.7 cell responses. In both cell types, transition rates between states in the NoRM model are distinct for each of the tested proteins and, crucially, macrophage hypo-responsiveness is underpinned by changes in transition rates to and from NRS. Overall, our findings provide support for a critical role for phenotypically negative macrophage populations as an active component of AIH and primary and secondary responses to LPS. This reveals unappreciated aspects of cellular ecology and community dynamics associated with LPS-driven training of macrophages.