Accessory genome evolution in giant bacteria: antimicrobial resistance, secondary metabolism, and phage–host coevolution in free-living and symbiotic lineages

Giant sulfur-oxidizing bacteria of the family Beggiatoaceae and their symbiotic relatives (Candidatus Epulonipiscium and related taxa) represent some of the largest and most morphologically conspicuous prokaryotes, yet their accessory genomic repertoires beyond chemolithoautotrophy remain poorly characterized. These lineages have independently evolved gigantism and contrasting lifestyles —free-living chemolithoautotrophy versus host-associated fermentative symbiosis—providing a comparative framework to test how lifestyle shapes accessory genome evolution. I conducted a genomic survey of 82 assemblies representing eight taxa, including free-living Beggiatoa, Thioploca, Candidatus Thiomargarita, and Thiomargarita; and symbiotic Candidatus Epulonipiscium, Candidatus Epulonipiscioides, Candidatus Parepulonipiscium, and Epulopiscium. Putative antimicrobial resistance genes (ARGs) were identified using DeepARG and ABRicate; biosynthetic gene clusters (BGCs) with antiSMASH; defense systems with DefenseFinder and CRISPRCas++; plasmids and prophages with geNomad, PHASTEST, VirSorter2, and PhaMer followed by CheckV quality filtering; and phage protein co-evolution with HyPhy/GARD/BGM. Lifestyle-associated differences were tested with negative-binomial models using genome size as an offset. After curation, 214 high-confidence putative ARGs and 350 BGCs were retained. Symbiotic lineages were strongly enriched in ARGs (IRR = 4.60) but depleted in total BGC density (IRR = 0.55), driven by loss of NRPS, terpene, PKS, and arylpolyene clusters and partially offset by enrichment of RiPP clusters. Defense systems were heterogeneously distributed, and genus-level differences largely disappeared after genome-size adjustment. Defense ORFs were not enriched inside BGC cores but were modestly over-represented within a 50–100 kb neighbourhood. Ten plasmids and eleven curated prophage-like sequences were recovered, most showing limited similarity to public databases. Co-evolutionary signal was stronger in structural phage proteins than in lytic and tail-associated proteins. These findings indicate that giant sulfur bacteria reallocate accessory genomic investment between biosynthetic capacity and antimicrobial resistance according to lifestyle, and that their mobile genetic elements remain poorly represented in current databases. Not applicable.