Stochastic and spatiotemporal effects in T-cell signaling

Abstract

T lymphocytes are key orchestrators of the adaptive immune response in higher organisms. This thesis seeks to apply different techniques from engineering and the physical sciences to understand how T cells balance the risks of autoimmunity and infection. (1) What features of proteins do T cells search for that correlate with pathogenicity, distinguishing self from foreign? Two contrasting theories have emerged that attempt to describe T cell ligand potency, one based on the half-life (tv12) of the interaction between T cell receptors (TCR) and peptide-MHC complexes (pMHC), the second on the equilibrium affinity (KD). We study an extensive set of TCR-pMHC interactions in CD4+ T cells which have differential KD and kinetics of binding. The data indicate that ligands with short t1/2 can be highly stimulatory if they have fast on-rates. Simple models suggest these fast-kinetic ligands are stimulatory because the pMHC bind and rebind the same TCR several times. Accounting for rebinding, ligand potency is KD-based when ligands have fast on-rates and t1/2-based when they have slow on-rates, unifying previous theories. (2) How do T cells make optimal responses with the imperfect information they receive through their receptors? Recent experiments suggest that T cells sometimes make stochastic decisions. Biological systems without sensors and genetic diversity, such as some bacteria, make stochastic decisions to diversify responses in uncertain environments, thereby optimizing performance (e.g. growth). T cells, however, can draw on considerable environmental and genetic diversity to diversify their responses. Using T cell biology as a guide, we identify a new role for noise in such systems: it helps systems achieve complex goals with simple signaling machinery. With decision-theoretic techniques, we suggest necessary conditions for noise to be useful in this way. (3) How can biological systems, like T cells, maintain desired responses in the presence of molecular noise, suppressing it or exploiting it as needed? We develop a semianalytical technique to determine how small changes in the rate constants of different reactions or in the concentrations of different species affect the rate at which biological systems escape stable cellular states. A single deterministic simulation yields the sensitivities with respect to all reactions and species in the system. This helps to predict those species or interactions that are most critical for regulating molecular noise, suggesting those most promising as drug targets or most vulnerable to mutation. These projects and others discussed in this thesis recruit techniques from random walks, statistical inference, and large deviation theory to understand problems ranging in scale from individual molecular interactions to the population of T cells acting in concert.