In Silico Analysis of Gardnerella Genomospecies Detected in the Setting of Bacterial Vaginosis

Abstract BACKGROUND Gardnerella vaginalis is implicated as one of the causative agents of bacterial vaginosis, but it can also be isolated from the vagina of healthy women. Previous efforts to study G. vaginalis identified 4 to 6 clades, but average nucleotide identity analysis indicates that G. vaginalis may be multiple species. Recently, Gardnerella was determined to be 13 genomospecies, with Gardnerella piottii, Gardnerella leopoldii, and Gardnerella swidsinkii delineated as separate species. METHODS We accessed 103 publicly available genomes annotated as G. vaginalis. We performed comprehensive taxonomic and phylogenomic analysis to quantify the number of species called G. vaginalis, the similarity of their core genes, and their burden of their accessory genes. We additionally analyzed publicly available metatranscriptomic data sets of bacterial vaginosis to determine whether the newly delineated genomospecies are present, and to identify putative conserved features of Gardnerella pathogenesis. RESULTS Gardnerella could be classified into 8 to 14 genomospecies depending on the in silico classification tools used. Consensus classification identified 9 different Gardnerella genomospecies, here annotated as GS01 through GS09. The genomospecies could be readily distinguished by the phylogeny of their shared genes and burden of accessory genes. All of the new genomospecies were identified in metatranscriptomes of bacterial vaginosis. CONCLUSIONS Multiple Gardnerella genomospecies operating in isolation or in concert with one another may be responsible for bacterial vaginosis. These results have important implications for future efforts to understand the evolution of the Gardnerella genomospecies, host–pathogen interactions of the genomospecies during bacterial vaginosis, diagnostic assay development for bacterial vaginosis, and metagenomic investigations of the vaginal microbiota.