Author ORCID Identifier
https://orcid.org/0009-0007-2942-7488
Date of Award
Winter 11-13-2024
Document Type
Thesis (Ph.D.)
Department or Program
Microbiology and Immunology
First Advisor
Margaret E. Ackerman
Abstract
Vaccines are critical tools for preventing diseases. Although a number of experimental trials have been conducted to tackle resilient pathogens like Human Immunodeficiency Virus (HIV-1) and Mycobacterium tuberculosis (Mtb), the search for effective vaccines against these and other important pathogens is ongoing. In these efforts, establishing reliable relationships between immune markers and protection outcomes helps inform choices in future trials, and is the first step in defining correlates of protection (CoPs). Given the challenges confronting HIV and Mtb vaccines in particular, this thesis combines broad immune-profiling approaches enabled by systems serology experiments with novel analytical approaches supported by positive-unlabeled machine learning, and coupling this work to comparative investigations of preclinical modeling in-vitro and in nonhuman primates. In this thesis, we identified a novel CoP for a tuberculosis vaccine, that is consistent across vaccine regimens and biological systems. Further, we developed a Positive-Unlabeled Machine Learning pipeline to learn from vaccine efficacy field trials, and identify ‘hidden’ CoPs that are missed due to absence of protection status label. We also augmented HIV-1 antibodies to improve complement activity, and demonstrated that antibody combinations of these augmented antibodies show higher complement activity as compared to mutated antibody alone and wild type antibody cocktail, potentially by formation of hetero-hexamers, and plan to test them in non-human primates via passive immunization. Prior to performing this study, we compared complement system in humans and rhesus macaques, and demonstrated sex and species-associated differences between the two biological systems, highlighting limitations in the ability of macaque model to recapitulate human biology. Overall, this work demonstrates how a multi-faceted approach that uses multiplexed assays to measure immune features, machine learning to relate immune response features to protection outcomes, and passive immunization using Fc engineered antibodies, can help scan larger number of immune features in trials with limited sample size, and aid in development of an effective vaccine to combat “tough” pathogens.
Original Citation
- Kelkar, N.S., Goldberg, B.S., Dufloo, J., Bruel, T., Schwartz, O., Hessell, A.J. and Ackerman, M.E., 2024. Sex-and species-associated differences in complement-mediated immunity in humans and rhesus macaques. Mbio, 15(3), pp.e00282-24.
- Kelkar, N.S., Morrison, K.S. and Ackerman, M.E., 2023. Foundations for improved vaccine correlate of risk analysis using positive-unlabeled learning. Human Vaccines & Immunotherapeutics, 19(1), p.2204020.
- Xu, S., Kelkar, N.S. and Ackerman, M.E., 2024. Positive-unlabeled learning to infer protection status and identify correlates in vaccine efficacy field trials. iScience, 27(3).
Recommended Citation
Kelkar, Natasha S., "Empowering Vaccine Development: Integrating systems serology, machine learning and antibody engineering to tackle resilient pathogens" (2024). Dartmouth College Ph.D Dissertations. 326.
https://digitalcommons.dartmouth.edu/dissertations/326
Included in
Bioinformatics Commons, Immunity Commons, Immunology of Infectious Disease Commons, Systems Biology Commons
