Antibiotic resistance genes (ARGs) are pervasive in gut microbiota, but it remains unclear how often ARGs are transferred, particularly to pathogens. Traditionally, ARG spread is attributed to horizontal transfer mediated either by DNA transformation, bacterial conjugation or generalized transduction. However, recent viral metagenome (virome) analyses suggest that ARGs are frequently carried by phages, which is inconsistent with the traditional view that phage genomes rarely encode ARGs. Here we used exploratory and conservative bioinformatic strategies found in the literature to detect ARGs in phage genomes, and experimentally assessed a subset of ARG predicted using exploratory thresholds. ARG abundances in 1181 phage genomes were vastly overestimated using exploratory thresholds (421 predicted vs 2 known), due to low similarities and matches to protein unrelated to antibiotic resistance. Consistent with this, four ARGs predicted using exploratory thresholds were experimentally evaluated and failed to confer antibiotic resistance in Escherichia coli. Reanalysis of available human- or mouse-associated viromes for ARGs and their genomic context suggested that bona fide ARG attributed to phages in viromes were previously overestimated. These findings provide guidance for documentation of ARG in viromes, and reassert that ARGs are rarely encoded in phages.

Antibiotic resistance genes (ARG) are pervasive in gut microbiota, but it remains unclear how often ARG are transferred, particularly to pathogens. Traditionally, ARG spread is attributed to horizontal transfer mediated either by DNA transformation, bacterial conjugation or generalized transduction. However, recent viral metagenome (virome) analyses suggest that ARG are frequently carried by phages, which is inconsistent with the traditional view that phage genomes rarely encode ARG. Here we used exploratory and conservative bioinformatic strategies found in the literature to detect ARG in phage genomes, and experimentally assessed a subset of ARG predicted using exploratory thresholds. ARG abundances in 1,181 phage genomes were vastly over-estimated using exploratory thresholds (421 predicted vs 2 known), due to low similarities and matches to protein unrelated to antibiotic resistance. Consistent with this, 4 ARG predicted using exploratory thresholds were experimentally evaluated and failed to confer antibiotic resistance in Escherichia coli. Re-analysis of available human-or mouse-associated viromes for ARG and their genomic context suggested that bona fide ARG attributed to phages in viromes were previously over-estimated. These findings provide guidance for documentation of ARG in viromes, and re-assert that ARG are rarely encoded in phages.

Enterococcus faecalis is an opportunistic pathogen that has emerged as a major cause of nosocomial infections worldwide. Many clinical strains are indeed resistant to last resort antibiotics and there is consequently a reawakening of interest in exploiting virulent phages to combat them. However, little is still known about phage receptors and phage resistance mechanisms in enterococci. We made use of a prophageless derivative of the well-known clinical strain E. faecalis V583 to isolate a virulent phage belonging to the Picovirinae subfamily and to the P68 genus that we named Idefix. Interestingly, most isolates of E. faecalis tested—including V583—were resistant to this phage and we investigated more deeply into phage resistance mechanisms. We found that E. faecalis V583 prophage 6 was particularly efficient in resisting Idefix infection thanks to a new abortive infection (Abi) mechanism, which we designated Abiα. It corresponded to the Pfam domain family with unknown function DUF4393 and conferred a typical Abi phenotype by causing a premature lysis of infected E. faecalis. The abiα gene is widespread among prophages of enterococci and other Gram-positive bacteria. Furthermore, we identified two genes involved in the synthesis of the side chains of the surface rhamnopolysaccharide that are important for Idefix adsorption. Interestingly, mutants in these genes arose at a frequency of ~10−4 resistant mutants per generation, conferring a supplemental bacterial line of defense against Idefix.