A statistical mechanics approach to vaccination against HIV

Abstract

Most vaccines stimulate the production of antibodies that provide a potent defense upon reinfection by the same strain of pathogen. The key process in antibody development is a stochastic process known as affinity maturation (AM) which generates strain-specific antibodies upon immunization by one antigen. A highly mutable virus like HIV evades recognition by these strain-specific antibodies via the emergence of new mutant strains within the patient. In some chronically infected patients, antibodies that can bind diverse antigens and thus protect against many HIV strains arise naturally; they are named broadly-neutralizing antibodies (bnAbs). A vaccine that elicits bnAbs could prevent HIV infections. This vaccine is expected to contain several different antigens. However, because bnAbs rarely appear in HIV patients, the complex mechanisms by which they emerge are not well understood. Theoretical models of AM could help identify promising vaccination strategies and shed light on a previously ignored problem in basic immunology; meaning how AM works with several antigens. For my thesis I investigated two pathways for breadth evolution. First, motivated by experimental findings that bnAbs have many mutations that may modify the flexibility of the binding region, I examined how flexibility influences breadth. A flexible binding region is expected to enable different conformations and therefore to allow binding to diverse antigens. Towards that goal, I developed a theoretical model of AM which, combined with Molecular Dynamics simulations, suggests that eliciting flexibility-affecting mutations is not essential for the evolution of bnAbs if proper germline B cells are first activated. This is significant as it simplifies the task of immunogen design. For my second project, I studied how separating the different antigens in time and mutational distance affects breadth of binding and antibody titers. The main observation is that introducing the antigens at different times is key to generating breadth. Furthermore, sequentially introducing one antigen per injection yields the greatest breadth and antibody titers. We also devised a prediction tool for breadth given a set of antigens and an immunization protocol. My results suggest optimal vaccination strategies which are expected to guide future in vivo investigations by our collaborators.