The complex, nonlinear mechanisms connecting early T cell receptor (TCR) activation to diverse T cell fates have not been fully elucidated. Understanding these processes quantitatively is however crucial to fine-tune immunotherapy treatments.

Grégoire Altan-Bonnet's lab has developed a robotic platform to track over days the dynamics of cytokines and cell surface markers involved in the T cell response processes. In collaboration with them, we found a latent space representation of high-dimensional cytokine dynamics in which trajectories are ordered according to antigen quality, in a variety of immunological settings. We termed this property "antigen encoding". A channel capacity calculation revealed how T cells can encode, in this latent space, significantly more information than just a binary distinction between self and non-self.

Guided by this insight, we went back to mechanisms. Recent experiments on T cells with modified receptor ITAMs from Paul E. Love's lab (phosphorylation sites) suggested that ITAM multiplicity modulates the negative feedback mediated by the phosphatase SHP-1. Including this hypothesis in a phenomenological model of TCR activation, we predicted non-trivial ligand antagonism effects, which were observed in subsequent experiments.