Machine Learning Accelerated Antibody Ranking and Extracellular Vesicle Engineering for Antiviral and Anticancer Immunotherapeutics

Abstract

One of the most pressing challenges in modern medicine is the development of therapeutics that can effectively counter rapidly evolving biological threats, including infectious pathogens and cancer cells. In response to the growing need for adaptable and potent treatments, we present a series of innovations that leverage machine learning and extracellular vesicle (EV) engineering to overcome therapeutic resistance and enhance immune targeting. To address the challenge of viral escape in fast-mutating pathogens like SARS-CoV-2, we have developed a machine learning-assisted antibody generation pipeline (AbGen) powered by an antibody language model (AbLM). AbGen enables high-throughput screening and redesign of IgGs with broad neutralization capacity against wildtype and emerging variants, including Delta and Omicron. To further overcome variant-mediated antibody evasion, we have engineered EV- presented IgG (evIgG), an innovative EV-based therapeutic platform that displays anti-spike IgG antibodies on the EV surface, achieving exceptionally high loading efficiency. evIgG shows over 150-fold enhanced neutralization efficacy against Omicron BA.5 compared to soluble IgG through maintaining ultra-high binding affinity and resisting dissociation due to multivalent cooperative interactions. Notably, evIgG effectively blocks pulmonary BA.5 viral amplification in hACE2- transgenic mice, highlighting its therapeutic potential in vivo. Extending this EV-based engineering strategy to oncology, we have explored the use of tumor-derived extracellular vesicles (TEVs) as cancer vaccines. TEVs, enriched with tumor- specific neoantigens, elicit robust anti-tumor immune responses in a murine melanoma model, significantly reducing tumor burden and prolonging survival. These findings highlight the potential of TEVs as a promising platform for anticancer immunotherapies. Together, these studies establish a transformative framework for therapeutic development combining AI-driven prioritizations with EV-based delivery systems to combat immune evasion in infectious diseases and cancer. This integrative approach offers precision-targeted solutions to some of the most complex challenges in clinical medicine.