The immune system evolves across the host's lifetime to protect against the wide array of threats in nature. While the immune system is capable of evolving and adapting to a single antigen (Ag), it becomes a challenging task to defend against diverse antigenic targets, including mutants of the same pathogen or a wide spectrum of pathogen species. First, the rapid intra-host diversification of highly-mutable pathogens such as human immunodeficiency virus (HIV) or hepatitis C virus (HCV) creates a coevolutionary arms race with the immune system. As a result, viruses persist into a chronic infection in most subjects and are only cleared in rare cases. In addition, The vaccination trials so far to elicit broadly-neutralizing antibodies (bnAbs) against highly-mutable viruses have met with failure. Second, the immune system has to allocate its finite amount of adaptive immune cells against the wide spectrum of pathogens in the environment. As memory cells accumulate from each pathogen encounter, the host's immune repertoire gradually becomes skewed: more adaptive immune cells are dedicated to the frequent pathogens while fewer are reserved for the rare pathogens. The skewed repertoire in the elderly has been correlated with immune risk phenotype and a chronic inflammatory response even in the absence of pathogens, but whether there is a mechanistic connection remains unknown.

My dissertation aims to address the following questions regarding the evolution of the immune system against diverse antigenic targets: (1): What are the mechanisms and deciding factors behind the distinct coevolutionary outcomes observed in different subjects? (2): What makes a viable vaccine design strategy to guide immune system evolution towards bnAbs? (3): What are the side effects of a skewed immune repertoire, as a result of adapting to different pathogens encountered during hosts' lifetime? Also, how can human intervention alleviate these side effects?

My dissertation shows that (1): different coevolutionary outcomes are decided by the timing and efficacy of successive narrow and broad antibody (Ab) responses, which in turn are determined by the conservation level and initial diversity of Ag. (2): A viable vaccine strategy to elicit bnAbs should balance suppression of strain-specific B cells and preserving cross-reactive B cells. The corresponding optimal selection strength should increase in time as driven by the evolution of B cell cross-reactivity. (3): Mediated by adaptive-innate feedback, repeated pathogen encounters during host lifetime and resulting memory inflation may trigger a fragility, in which any encounter with a novel pathogen will cause the system to irreversibly switch from health to chronic inflammation (CI). In addition, the onset of CI strongly depends on the history of encountered pathogens; the timing of onset can be delayed drastically when the same set of infections is encountered in a specific order.

The immune system is a complex system of cells and molecules that work cooperatively to protect us against pathogenic organisms. It can perform complicated tasks such as pattern recognition, learning, and memory, all of which require dynamical coordination among a large number of components across multiple scales. Nevertheless, the multitude of different components makes it challenging to unveil the mechanistic principles that give rise to these remarkable functions.

My thesis focuses on how our immune system learns from infections and improves specificity of pathogens recognition on the fly. This process is known as affinity maturation, where the affinity of B cell receptor improves through Darwinian evolution. Although recent progresses in experiments revealed many details, what remains is a first-principle and quantitative understanding of how different elements come together to achieve the goal. Using statistical physics tools and computational modeling, I study various aspects of the maturation process, including molecular interactions, information extraction, and evolutionary dynamics.

To understand how B cells with different affinities are discriminated during affinity maturation, we investigate the process of antigen extraction, where B cells use cytoskeleton forces to extract antigen molecules from other presenting cell surface. We show this process allows a B cell to infer its receptor affinity by measuring the number of extracted antigens.Our model highlights the regulatory role of mechanical force: Application of a constant force with proper magnitude can enhance discrimination fidelity, and usage of a dynamical force that introduces negative feedback can improve discrimination robustness with respect to fluctuations in antigen concentration.

To illustrate how molecular interactions influence cellular evolution, we couple the physical theory of antigen extraction to a minimal model of affinity maturation and simulate ensembles of cell populations under different conditions. The multiscale model predicts that the affinity ceiling stems from the physical limit of antigen tether strength and identifies strategies to alleviate the constraint.

Lastly, we present a study on the long-term coevolution between evolving pathogen and adaptive immune response. Our work reveals that the asymmetric reaction range between immunogenicity (the ability of pathogens to induce an immune response) and antigenicity (the ability of pathogens to interact with antibodies) is critical in determining the dynamics of coevolution.