Complex exchanges among plasmids and clonal expansion of lineages shape the population structure and virulence of Borrelia burgdorferi

Abstract

The many plasmids (commonly >20 per strain) of Borrelia burgdorferi (Bb) pose challenges for studies of Lyme disease. The genetic content of most plasmids cannot be resolved using short-read sequencing. We generated long-read assemblies (LRA) of 183 isolates of human-derived Bb and analyzed patterns of genome variation in the Lyme spirochete. LRAs confirm that populations consist of strongly-structured genotypes with nearly-clonal structure and tight-knit blocks of accessory genome elements. Notably, these patterns of linkage were statistical but not physical, with linkage blocks distributed across multiple plasmids. A consequence of this structure is that plasmid name and/or plasmid subtype does not capture strain-specific genetic content. We used network methods to characterize patterns of genetic linkage. We demonstrate that co-occurring gene networks, here termed genetic modules, are the fundamental unit of genome variation in Bb and designate genetic modules consisting of co-occurring genes. We linked genetic modules to Bb phenotype by identifying modules that influence dissemination in humans. This modular decomposition clarifies previously observed associations between strain and virulence. For example, virulent RST1/OspC type A strains are distinguished by the presence of virulence-associated modules containing gene content from lp28-1, lp56, and the chromosome along with the absence of gene content on lp28-1 and lp28-4 associated with localized disease. LRAs also demonstrate that the well-established statistical linkage between physical unlinked genetic markers (e.g. RST and OspC) is a general pattern among accessory genome elements. In summary, genetic modules containing genes linked across multiple replicons, rather than strain-defining plasmids, organize the Bb accessory genome and the strain-specific variation responsible for differences in human virulence.