Structure-based similarity network accelerates the discovery of lysins as oral microbiome modulators targeting periodontal pathogens

Abstract

Microorganisms significantly influence human health, and dysbiosis of the oral microbiome plays a critical role in the development and progression of both oral and systemic diseases. This highlights the urgent need for novel therapeutics targeting specific pathogens. Here, we presented a structure-based pipeline to efficiently identify potential phage-derived periodontal lysins (LysPds) from nearly one million proteins. We predicted the structures of candidate lysins using AlphaFold2 and developed an innovative structure-based similarity network to classify them into distinct clusters, each with unique functional properties. A systematic characterization of 16 representative LysPds from 11 superfamilies revealed that over 90% demonstrated potent antibacterial activity against key periodontal pathogens. Among these, LysPd078 was identified as a promising preclinical drug candidate, effectively reconfiguring microbiome communities while demonstrating significant efficacy and safety in mouse models of periodontitis and calvarial infection. Our findings highlight the effectiveness of structure-based similarity networks in exploring vast protein spaces and underscore the potential of LysPd078 as a targeted modulating agent for the oral microbiome.