Stochastic processes in T-cell signaling
Author(s)Yang, Ming, Ph. D. Massachusetts Institute of Technology
Massachusetts Institute of Technology. Dept. of Chemical Engineering.
Arup K. Chakraborty.
MetadataShow full item record
T cells are orchestrators for adaptive immunity. Antigen recognition by T cells is mediated by the interactions between T-cell receptors (TCRs) and peptide-MHC (pMHC) molecules. How T cells can translate stimulatory external cues (e.g., TCRpMHC interactions where the peptides are derived from foreign proteins) to functional responses (e.g., proliferation), while not responding to self-pMHC has been a puzzle for decades. The ability to discriminate foreign antigens from self antigens demands extraordinary intricacy for the design and operation of signaling pathways. This problem is a special challenge since fluctuations at the microscopic level are ubiquitous in biochemical networks, due to stochastic nature of reactions and uncertainties in protein expressions. The prevalence of noise imposes further challenges for T cells to deliver biological functions reliably. The overarching theme of this thesis is to understand the role of stochasticity in T-cell signaling. Four problems have been selected for presentation in this thesis: 1. Fluctuation-driven transitions can drive cellular systems out of stable states and lead to spurious responses. We proposed a theoretical and computational framework to identify key reactions or species that are responsible for regulating such stochastic transitions. The identification of these critical components for network stability not only pinpoints key deleterious protein mutations, but also helps intelligently select drug targets. The semi-analytical method we derived using large-deviation theory and calculus of variations agrees well with computational costly brute-force simulations. Additionally, our framework unveils qualitative characteristics of key reactions regulating stochastic transitions. We believe that we have developed the first method to carry out fully stochastic sensitivity analyses using analytical calculations. 2. In collaboration with Dr. Jeroen Roose's lab at UCSF, we investigated the roles of RasGRP and SOS in the activation of ERK and P38 MAP kinase pathways. We extended established computational models developed in our lab and predicted that SOS' allosteric pocket is important for the magnitude and bimodal pattern of ERK activation, which was confirmed experimentally. The synergy between computational modeling and experimental studies enabled us to propose mechanistic models that incorporate features such as co-operativity and non-linearity thresholding to study P38 activation. These models were consistent with the experimental findings that SOS is preferentially more important than RasGRP for Rac-P38 activation and SOS' allosteric pocket has little effect on Rac-P38 pathways, and also generated numerous experimentally testable hypotheses. 3. While rare events, such as escapes from stable basins, take a long time (waiting time) to occur, they take little time to complete once they have started. We showed that for Markov processes characterized by detailed balance, successful transitions, on average, complete exactly as quickly as transitions in the opposite (non-rare) direction. We first provide a general proof by invoking time reversibility, and then elaborate the proof by considering two specific dynamics, namely, continuous-time Markov Chains with detailed balance and one-dimensional Langevin Dynamics. We employ ideas from measure theory and stochastic calculus. We conclude that rare events, once they happen, happen quickly, and speculate about extensions to nonequilibrium systems, such as viral escape. 4. While microscopic fluctuations complicate reliable functioning of biochemical networks, stochastic noise also offers enormous information about the underlying network that generates such noise. We present an effort to exploit the non-random structure of random noise for network topology identification. In particular, we applied linear noise approximations to two three-node network motifs, namely, incoherent feed-forward loop (IFF) and negative feedback loop (NFB), and obtained correlation functions governing the fluctuations of species copy numbers at steady state. We identified two signatures that can be used to discriminate IFF from NFB. This endeavor represents a first step toward understanding how, and to what extent, time-series data with fine time and length resolutions can be used to infer network structures.
Thesis (Ph. D.)--Massachusetts Institute of Technology, Dept. of Chemical Engineering, 2012.Cataloged from PDF version of thesis.Includes bibliographical references (p. 89-93).
DepartmentMassachusetts Institute of Technology. Dept. of Chemical Engineering.
Massachusetts Institute of Technology