Mechanistic modeling of adaptation and memory in natural killer (NK) cells

In the innate immune system, natural killer (NK) cells recognize and eliminate aberrant cells without binding specific antigens. Deciding to disengage or kill the target requires integrating signals from activating and inhibitory receptors. Since NK cells can repeatedly kill many targets, a robust, tightly regulated activation threshold must be maintained to prevent autoimmunity. Previous experimental work demonstrated that NK cells undergo an adaptive process known as "education" in which the local environment influences NK cell sensitivity. While existing statistical models capture experimentally observed behaviors, a mechanistic understanding of NK cell education is still lacking. Here, we propose a mechanistic model for receptor dynamics that describes decision-making and adaptation in NK cells. Our model synthesizes experimental observations on the role of receptor-mediated endocytosis in NK cell activation and regulation. Through numerical simulations, we show that endocytosis can introduce an additional signaling timescale that allows for robust decision-making. Additionally, we demonstrate that NK cells can store memory of previous target cell encounters in surface receptor levels.