Multimodal reasoning for broadly neutralizing antibody discovery from label-free human B cell repertoires across virus families

2026-10-02

AI Summary

This study addresses the extreme scarcity and unclear cellular origins of broadly neutralizing antibodies, as well as the limited cross-viral-family generalizability of existing discovery tools. To overcome these challenges, this work proposes ImmuneAgent, a closed-loop artificial intelligence system that integrates multimodal reasoning with continual meta-learning and incorporates wet-lab feedback mechanisms to efficiently screen candidate antibodies from unlabeled B-cell repertoires. The proposed system achieves a neutralization rate of 55% and a broad-spectrum antibody yield of 11%. Notably, five candidate antibodies confer 100% protection against influenza in murine models and successfully generalize to human metapneumovirus (hMPV) and human papillomavirus (HPV). Collectively, this research establishes a universal intelligent paradigm for cross-viral-family antibody discovery.

Abstract

Discovering broadly neutralizing antibodies (bnAbs) from human natural immune repertoires remains a fundamental challenge in immunology, hindered by: the extreme rarity of bnAb, incomplete understanding of their cellular origins across pathogens, and the inability of existing computational tools to generalize across emerging viral threats. Here we present ImmuneAgent, a closed-loop AI system that integrates multimodal reasoning with continual meta-learning and wet-lab feedback to overcome these barriers. Applied to screen the natural BCR repertoires from vaccinated or infected cohorts, the system achieves a ~55% neutralization antibody discovery rate (60 of 110 cloned candidates) and a ~11% bnAb yield (12 of 110), substantially outperforming a state-of-the-art sequence-based neutralization predictor or cofolding models evaluated at the same cloning budget. Five ImmuneAgent-discovered antibodies conferred 100% in vivo protection against lethal influenza challenge, comparable to the clinical-stage therapeutic MEDI8852. The system recovered the cellular and structural determinants of bnAb activity and identified FCRL5+CD27+ atypical memory B cells as a conserved bnAb reservoir and hydrophobic interface enrichment as a cross-viral structural signature, which generalized to unseen antigens, discovering human metapneumovirus (hMPV) cross-neutralizing and human papillomavirus (HPV)-neutralizing antibodies without antigen-specific sorting. These results validate that ImmuneAgent is a generalizable framework for rapid therapeutic antibody discovery against emerging viral threats.

Problem

• broadly neutralizing antibodies
• B cell repertoires
• multimodal reasoning
• antibody discovery
• viral threats

Innovation

• Multimodal reasoning
• Continual meta-learning
• Broadly neutralizing antibodies
• Closed-loop AI system
• B cell repertoires