Substrate utilization and cross-feeding synergistically determine microbiome resistance to pathogen invasion

Abstract Understanding how microbiomes resist pathogen invasion remains a key challenge in natural ecosystems. Here, we combined genome-scale metabolic models with synthetic community experiments to unravel the mechanisms driving pathogen suppression. We developed curated genome-scale models for each strain, incorporating 48 common resource utilization profiles to fully capture their metabolic capacities. Trophic interactions inferred from models accurately predicted pathogen invasion outcomes, achieving an F1 score of 96% across 620 invasion tests involving diverse microbial communities and nutrient environments. Importantly, considering both substrate and metabolite features provided a more holistic understanding of pathogen suppression. In particular, cross-feeding metabolites within the native community emerged as crucial yet often overlooked predictors of community resistance, disproportionally favoring native species over invaders. This study lays the foundation for designing disease-resistant microbiomes, with broad implications for mitigating pathogen exposure in diverse environments.