Giant virus genomes are unlikely to be reservoirs of antibiotic resistance genes

arising from X. Yi et al. Nature Communications https://doi.org/10.1038/s41467-024-51936-z (2024)

In their recent article ‘Giant viruses as reservoirs of antibiotic resistance genes’, Yi et al.¹ claim that 39.5% of viruses in the phylum Nucleocytoviricota encode antibiotic resistance genes (ARGs). In my opinion, this claim comes from the overclassification of several gene groups as ARGs. Many nucleocytoviruses (also known as NCLDV) are remarkable for having large (>500 kb) genomes that can encode genes ubiquitous in cellular life, but absent from most viruses². While some of these gene classes are homologous to ARGs that provide certain bacteria with resistance to certain antibiotics, antibiotic-insensitive homologs are also conserved in eukaryotes. For the vast majority of nucleocytovirus-encoded putative ARGs presented by Yi et al. antibiotics are unlikely to have factored into their evolutionary history.

The putative nucleocytovirus ARGs presented in Yi et al. fall primarily within two groups: a conserved poxvirus capsid scaffold, and families of genes that are widely conserved in eukaryotes and prokaryotes and include some antibiotic targets. Just over half (51.2%) of the ARGs reported in nucelocytovirus isolates by Yi et al. are poxvirus ‘rifampicin-resistance genes’. These genes actually encode a scaffold protein that facilitates capsid assembly. This protein is known as D13 in vaccinia virus, and for historical reasons, is also commonly labeled as a ‘rifampicin-resistance’ protein³. Rifampicin, also known as rifampin, is a synthetic derivative of the antibiotic rifamycins, originally isolated from an actinomycetes bacterium⁴. Rifampicin’s antibacterial activity comes from its inhibition of bacterial RNA polymerases⁵. As a strange quirk of biochemistry, rifampicin also impedes capsid assembly of vaccinia virus and some other poxviruses through interactions with the capsid scaffold³. Long before D13 was known to function as a capsid scaffold, researchers discovered that vaccinia could evolve resistance to rifampicin through mutation of the locus encoding D13⁶. Thus, D13 became known as the ‘rifampicin-resistance’ protein, although it is technically the poxvirus target of rifampicin.

The ‘rifampicin-resistance’ annotation is unfortunately widespread and predominates in publicly available poxvirus genomes. This annotation is used for susceptible and resistant alleles, as well as for many homologs for which there is no rifampicin sensitivity data. This scaffold gene is universally conserved in poxviruses, and the role of poxviruses as human pathogens has likely led to their overrepresentation among sequenced nucleocytovirus isolates. These factors may contribute to the scaffold’s dominance among putative ARGs reported by Yi et al. It should be stressed that the capsid scaffold has no functional equivalent in bacterial cells, and cloning the scaffold into bacteria would not be expected confer them with rifampicin resistance. Thus, it is dubious to call D13 an ARG no matter how commonly it may be annotated as such.

Most of the other ‘ARGs’ found in Yi et al.’s survey are genes conserved within bacteria and eukaryotes that are present in some nucleocytovirus genomes. Two notable examples of these genes are dihydrofolate reductase (dfr) genes and isoleucyl-tRNA synthetase (ileS) genes. These are the most common and third most common putative ARGs amongst sequenced nucleocytovirus genomes in the Yi et al. survey, respectively. The synthetic antibiotic trimethoprim has a broad spectrum of activity against bacterial dihydrofolate reductases⁷. Yi et al. demonstrate that two nucleocytovirus-encoded dfr genes are able to provide trimethoprim resistance when cloned into E. coli. To provide context for this observation, the authors cite previous work by Mueller et al.⁸ and claim that a marseillvirus dfr homolog was able to confer resistance to trimethoprim and pyrimethamine when cloned into Saccharomyces cerevisiae. However, while Mueller et al. claim that the viral dfr was resistant to trimethoprim, they do not frame their experiment as demonstrating conferral of trimethoprim resistance to S. cerevisiae. Rather, heterologous complementation of the viral dfr in a S. cerevisiae dfr-deficient strain showed that the marseillvirus homolog did not have trimethoprim susceptibility. This is not a gain of resistance because wild type S. cerevisiae is already resistant to trimethoprim⁸. In general, trimethoprim is much less inhibitory towards eukaryotic dihydrofolate reductases than it is towards bacterial homologs⁹. This difference in inhibitory activity is one factor that allows trimethoprim to be used for the treatment of bacterial infections. Given that nucleocytoviruses are eukaryote infecting viruses, it should not be surprising that their dihydrofolate reductases might show a similar level of susceptibility to trimethoprim as eukaryote-encoded homologs.

This same problem also applies to labeling Nucleocytoviricota ileS homologs as ARGs. The ileS gene encodes isoleucyl-tRNA synthetase. This protein is the drug target of mupirocin. Mupirocin was originally isolated from the bacterium Pseudomonas fluorescens and has broad spectrum antibiotic activity¹⁰. Mupirocin does not generally affect eukaryotic isoleucyl-tRNA synthetase homologs. Some bacterial species have acquired a eukaryotic-like ileS homolog that confers mupirocin resistance. These eukaryote-like ileS genes occur on mobile plasmids in pathogenic Staphylococcus aureus, and within the mupirocin synthesis cluster of P. fluorescens¹⁰. Notably, Yi et al. claims the phylogenetic positioning of mupirocin-resistant bacterial ileS genes “indicated a potential bridging role of NCLDVs in the possible transmission between eukaryotic and resistant bacterial ileS genes¹.” This argument is flawed, because there is no nesting of mupirocin-resistant bacterial ileS among nucleocytovirus or eukaryotic sequences. If resistant bacterial ileS genes were directly acquired from nucleocytoviruses or from eukaryotes with or without a nucleocytovirus intermediate, then the resistant ileS should be nested within the eukaryotic or Nucleocytoviricota clades. Recent work has suggested the horizontal transfer of ileS between Nucleocytoviricota and proto-eukaryotes¹¹. Yi et al.’s own results for ileS appear consistent with this hypothesis along with an additional transfer of proto-eukaryotic ileS into a bacterial lineage. Thus, the acquisition of muciprocin resistant ileS by bacteria is likely ancient and would not have involved modern nucleocytoviruses.

Removing the poxvirus capsid scaffold, dfr, and ileS genes eliminates most of the ARGs that Yi et al. report in nucleocytoviruses. However, there is one more gene class that warrants discussion. The second most common ARG type listed by Yi et al. is the ATP binding cassette F (ABC-F) protein encoding genes. Unlike the other prominent ARGs listed by Yi et al. ABC-F proteins are not known targets of antibiotics. ABC-F proteins interact with the ribosome, and some of these proteins are able to provide resistance to antibiotics by occluding the antibiotic’s binding site on the ribosome¹². However, many ABC-F proteins are thought to have functions unrelated to antibiotic resistance with proposed functions in E-site tRNA release, ribosome recycling, translation initiation, and ribosome biogenesis¹³. Given the vast difference between eukaryotic and bacterial ribosomes, it seems unlikely that Nucleocytoviricota encoded ABC-F proteins would be functional in both contexts. Nevertheless, it is intriguing that Yi et al. report that Phycodnaviridae and Pithoviridae genomes encode multiple types of ABC-F genes. These genes could be involved in manipulation of the host ribosome as a means of controlling gene expression, and would be an interesting subject for further investigation.

In their expansive classification of ARGs, Yi et al. use an operational definition for ARGs, where a gene is an ARG if expression of that gene renders a microbe resistant to an antibiotic it would otherwise be vulnerable to¹⁴. While there is appeal in this definition’s simplicity, it does not provide insight into the clinical, epidemiological, and ecological significance of such genes¹⁴. This results in a much broader definition of ARGs than many ARG researchers use. Additionally, Yi et al. classify the homologs of antibiotic targets as ARGs even in cases where there is no data demonstrating the presence or absence of antimicrobial sensitivity for these homologs and little risk of these homologs being transferred to antibiotic sensitive species. These problems are further compounded by the loose parameters Yi et al. use to define ARG homologs (>25% sequence identity over >80% of target and query coverage)¹. Overall, it is inappropriate to call nucleocytoviruses reservoirs of ARGs, unless one is willing to argue that humans with our rifampicin-resistant RNA-polymerases, trimethoprim-resistant dihydrofolate reductases, and mupirocin-resistant isoleucyl-tRNA synthetases, are also reservoirs of ARGs.

The viruses of Nucleocytoviricota have remarkable genomes with many functional gene classes not seen elsewhere in the viral world. Often among these are core metabolism genes conserved in both bacteria and eukaryotes¹. Given the critical role these genes play in cell survival, and the vast evolutionary distance between bacterial and eukaryotic paralogs, these gene classes often become the targets of therapeutic antibiotics intended to eliminate bacterial infections while maintaining the host’s survival. While it is reasonable to expect that nucleocytovirus-encoded homologs of these genes would be resistant to bacteria-specific antibiotics, there is no evidence that nucleocytoviruses have facilitated the transfer of such genes to bacteria, or that the gene’s function within nucleocytoviruses is at all related to antibiotic resistance. At present, there is no evidence that nucleocytoviruses act as reservoirs of antibiotic resistance in nature.