In Silico Characterisation of Putative Prophages in Used in Probiotics for Vaginal Health.

While live biotherapeutics offer a promising approach to optimizing vaginal microbiota, the presence of functional prophages within introduced strains could impact their safety and efficacy. We evaluated the presence of prophages in 895 publicly available genomes using Phaster, Phigaro, Phispy, Prophet and Virsorter. Prophages were identified according to stringent (detected by ≥4 methods) or lenient criteria (detected by ≥2 methods), both with >80% reciprocal sequence overlap. The stringent approach identified 448 prophages within 359 genomes, with 40.1% genomes harbouring at least one prophage, while the lenient approach identified 1671 prophages within 83.7% of the genomes. To confirm our in silico estimates in vitro, we tested for inducible prophages in 57 vaginally-derived and commercial isolates and found inducible prophages in 61.4% of the isolates. We characterised the in silico predicted prophages based on weighted gene repertoire relatedness and found that most belonged to the or families. ResFam and eggNOG identified four potential antimicrobial resistance genes within the predicted prophages. Our results suggest that while prophages seldomly carry clinically concerning genes and thus unlikely a pose a direct risk to human vaginal microbiomes, their high prevalence warrants the characterisation of prophages in live biotherapeutics.