Identifying the Molecular Triggers of Allergic Immunity: A Systematic Approach to Allergen and Adjuvant Discovery

Abstract

Allergic diseases reflect a misdirected immune response toward innocuous environmental or dietary components. While the major food allergens and their molecular identities are known, few systematic efforts have identified unique allergens or small-molecule adjuvants that influence allergic immunity. In this dissertation, I integrate protein language models and neuronal screens to identify protease allergens and small-molecule adjuvants that modulate allergic inflammation. Immunodominant allergens from plants, animals, and fungi are enzymatically active serine or cysteine proteases that share a conserved catalytic triad in their active site. Despite their diverse origins, these enzymes appear to have convergently evolved similar proteolytic functions that initiate type-2 immune responses. As a result the enzymatic activity of protease allergens is required for their functional detection by the immune system. Given long-standing associations between microbiome dysbiosis and allergic diseases, we hypothesized that the commensal microbiome represents an untapped source of allergenic proteases. To identify candidates, we developed a protein language model-based framework to learn shared features of conserved catalytic domains in known allergenic proteases and applied it to human gut and oral microbiome gene catalogs, predicting thousands of putative serine and cysteine protease allergens. Two high-confidence candidates from Staphylococcus schleiferi and Tannerella forsythia were validated in vivo where they activated hallmark type-2 responses consisting of CD301b+ dendritic cell (DC) activation and Th2 cell differentiation, confirming our classifier can predict allergenicity through protease activity. The identity of the adjuvants that shape allergic immunity are also unknown. We previously showed that sensory neurons directly respond to protease allergens, raising the possibility that neuronal responsiveness may reveal substances with allergenic potential. Thus, we screened a subset of plant derived small molecules using an in vitro primary neuronal plate-based assay. We found that while plant derived saponins (PDSs), a diverse group of immunomodulatory molecules, activated sensory neurons in vivo, they did not induce the release of the neuropeptides Substance P and/or calcitonin gene-related peptide (CGRP) in vitro. One subclass, the Ginsenosides, altered responses to the protease allergen papain in vivo through distinct effects on DC migration and T helper cell differentiation, while Polyphyllins, broadly activated DCs and promoted CD8⁺ T cell responses in vivo. Overall, this dissertation showcases two strategies for allergen and adjuvant discovery across biological systems.