The bacterial pangenome contains a vast diversity of antiphage systems, whose overall extent is still unknown. In this study, we developed complementary machine learning approaches to systematically predict antiphage function from genomic context, protein sequence, or their combination, achieving up to 99% precision and 92% recall. We validated these models experimentally in Escherichia and Streptomyces with the discovery of 12 antiphage systems. Applied to over 32,000 bacterial genomes, these models expand the predicted antiphage repertoire, with ~1.5% of bacterial genomes devoted to defense and more than 85% of predicted protein families remaining uncharacterized. We provide an interactive catalog of more than 19,000 candidate operon families for experimental follow-up. Together, these findings show that most molecular diversity in bacterial immunity remains uncharacterized and provide a foundation for its systematic exploration.

Phages are ubiquitous in nature, and bacteria with very different genomics, metabolisms, and lifestyles are subjected to their predation. Yet, the defence systems that allow bacteria to resist their phages have rarely been explored experimentally outside a very limited number of model organisms. Actinobacteria (Actinomycetota) are a phylum of GC-rich Gram-positive bacteria, which often produce an important diversity of secondary metabolites. Despite being ubiquitous in a wide range of environments, from soil to fresh and sea water but also the gut microbiome, relatively little is known about the anti-phage arsenal of Actinobacteria. In this work, we used DefenseFinder to systematically detect 131 anti-phage defence systems in 22803 fully sequenced prokaryotic genomes, among which are 2253 Actinobacteria of more than 700 species. We show that, like other bacteria, Actinobacteria encode many diverse anti-phage systems that are often encoded on mobile genetic elements. We further demonstrate that most detected defence systems are absent or rarer in Actinobacteria than in other bacteria, while a few rare systems are enriched (notably gp29-gp30 and Wadjet). We characterize the spatial distribution of anti-phage systems on Streptomyces chromosomes and show that some defence systems (e.g. RM systems) tend to be encoded in the core region, while others (e.g. Lamassu and Wadjet) are enriched towards the extremities. Overall, our results suggest that Actinobacteria might be a source of novel anti-phage systems and provide clues to characterize mechanistic aspects of known anti-phage systems.

Iron-sulfur (Fe-S) clusters are ancient and ubiquitous protein cofactors and play irreplaceable roles in many metabolic and regulatory processes. Fe-S clusters are built and distributed to Fe-S enzymes by dedicated protein networks. The core components of these networks are widely conserved and highly versatile. However, Fe-S proteins and enzymes are often inactive outside their native host species. We sought to systematically investigate the compatibility of Fe-S networks with non-native Fe-S enzymes. By using collections of Fe-S enzyme orthologs representative of the entire range of prokaryotic diversity, we uncovered a striking correlation between phylogenetic distance and probability of functional expression. Moreover, coexpression of a heterologous Fe-S biogenesis pathway increases the phylogenetic range of orthologs that can be supported by the foreign host. We also find that Fe-S enzymes that require specific electron carrier proteins are rarely functionally expressed unless their taxon-specific reducing partners are identified and co-expressed. We demonstrate how these principles can be applied to improve the activity of a radical S-adenosyl methionine(rSAM) enzyme from a Streptomyces antibiotic biosynthesis pathway in Escherichia coli. Our results clarify how oxygen sensitivity and incompatibilities with foreign Fe-S and electron transfer networks each impede heterologous activity. In particular, identifying compatible electron transfer proteins and heterologous Fe-S biogenesis pathways may prove essential for engineering functional Fe-S enzyme-dependent pathways.

The bacterial pangenome encodes an immense array of antiphage systems, yet much of their diversity remains uncharted. In this study, we developed language models to predict novel antiphage proteins in two ways: first via fine-tuning ESM2, a protein language model capable of detecting distant homology to known defense proteins, second via a genomic language model with ALBERT architecture which predicts defensive function based on genomic context. We demonstrate that applying these approaches to Actinomycetota - a phylum largely unexplored for antiphage defenses, can accurately predict previously unknown functional defense mechanisms, leading to the discovery and experimental validation of six defense systems with novel antiphage proteins. Analysis of over 30,000 bacterial genomes predicted more than 45,000 uncharacterized protein families potentially involved in antiphage defense, underscoring the vast, untapped diversity of these systems.