replying to Z. K. Barth Nature Communications https://doi.org/10.1038/s41467-025-63367-5 (2025)

In the accompanying Comment1, Barth raises concerns regarding our classification of genes encoding rifamycin resistance protein (rif), dihydrofolate reductase (dfr), and isoleucyl-tRNA synthetase (ileS) as antibiotic resistance genes (ARGs) in our study of nucleocytoplasmic large DNA viruses (NCLDVs)2. We appreciate Barth’s interest in our work and propose that these concerns primarily arise from some differing interpretations of our results and the imprecise use of certain terms.

Barth’s first concern focused on rif genes. He is right that the rif genes identified in our study encode the D13 protein, which is technically the target of rifampicin in poxviruses. We agree that cloning these genes into bacteria would not be expected to confer them with rifampicin resistance. We acknowledge that classifying D13 protein genes as antimicrobial resistance genes is more precise than categorizing them as ARGs, given that the conventional definition of ARGs specifically refers to genes that confer resistance to antibiotics in bacteria. However, it should be noted that traditional research and public health concerns focusing on antibacterial resistance have expanded to encompass resistance to a broader range of antimicrobial agents, including antibiotics, antifungals, antivirals, and antiparasitics3,4,5. In our article2, the ARGs carried by NCLDVs were predominantly classified into various families of antimicrobial resistance genes (e.g., in Table 1), while we specifically referred to rifampicin as an antiviral agent and discussed its potential role in the occurrence of D13 protein genes within the genomes of Poxiviridae. Previous studies6,7,8,9,10 have described both susceptible and resistant alleles of the D13 gene, differing at specific amino acid residues. In order to further investigate the potential presence of resistance-conferring mutations at specific positions within the D13 protein genes identified in our study, here we performed multiple sequence alignments. The analysis encompassed three distinct sequence groups: (1) the D13 gene variants identified in our study2, (2) the vaccinia virus D13 gene with reported resistance-conferring mutations8,9,10, and (3) putative rifampicin resistance protein sequences retrieved from the publicly accessible ARG database (DeepARG)11. Among the 42 NCLDV-encoded D13 sequences identified in our study, 15 exhibit amino acid substitutions at positions corresponding to known resistance-conferring mutations (Fig. 1). However, we cannot rule out the possibility that these putative resistance-conferring mutations may be largely attributable to natural variation in the D13 sequence across different giant viruses, and most of them might not confer phenotypic rifamycin resistance.

Fig. 1
figure 1

Multiple sequence alignment of the vaccinia virus D13 gene, publicly available “rifampicin resistance proteins” from the DeepARG database11, and the selected rif proteins annotated in our study2. The rifampicin resistance-conferring mutations previously reported8,9,10 for the vaccinia virus D13 gene are indicated above the alignment. Amino acids that match the rifampicin resistance-conferring mutations are highlighted in blue. For easier visualization and clarity, only the regions with mutations matched by the giant virus D13 sequences annotated in our study (e.g., 1–38, 308–321, and 438–490) are shown.

Full size image

Barth’s second concern concentrated on dfr genes. Whilst he concurs with our demonstration that two NCLDV-encoded dfr genes are able to confer trimethoprim resistance when cloned into Escherichia coli, he questions our interpretation of Muller et al’s findings regarding the performance of a marseillvirus dfr homolog in Saccharomyces cerevisiae12. We acknowledge that describing the marseillevirus dfr homolog’s performance in S. cerevisiae as “lacking trimethoprim susceptibility” provides greater precision than stating it “confers trimethoprim resistance”. Nevertheless, we propose that this semantic refinement does not invalidate our key point: certain NCLDV-encoded dfr homologs retain functionality across heterologous systems. Barth further posits that the comparable trimethoprim susceptibility profiles between NCLDV-encoded dihydrofolate reductases and eukaryotic homologs align with theoretical expectations. While we acknowledge this perspective, it is noteworthy that a virus predating on eukaryotic cells would be more likely to pick up eukaryotic genes than prokaryotic homologs of the same genes.

Barth’s third concern centered on ileS genes. Besides proposing that similar mupirocin susceptibility of NCLDVs’ and eukaryotic isoleucyl-tRNA synthetases encoded by ileS genes is theoretically anticipated, Barth believes that our phylogenetic analysis suggested that bacterial resistance-conferring ileS genes likely originated from eukaryotes rather than from NCLDVs. However, this interpretation differs from our relevant findings. We explicitly acknowledged2 that “Moreover, the ileS genes of giant viruses were shown to occupy an intermediate position between eukaryotic ileS and bacterial resistant ileS in the gene tree, implying that the ileS genes of giant viruses likely exhibit similar inherent resistance traits as those found in eukaryotes.”. It should be further pointed out that the possibility of direct or indirect transfers of ARGs between NCLDVs and bacteria cannot be excluded13,14, even though there is evidence suggesting that certain bacterial mupirocin-resistant ileS genes likely originated from eukaryotes15.

Barth also raises concerns regarding the genes encoding ABC-F proteins. Whilst he finds that our report of multiple types of ABC-F genes in the genomes of Phycodnaviridae and Pithoviridae is intriguing, he believes that the specific functions of these genes warrant further investigation. We agree with this view and recognize that certain ARGs may potentially exhibit multiple biological functions16.

Overall, Barth’s comments primarily stem from a disagreement with our definition of ARGs. We adopted the operational definition of ARGs, where an ARG is defined by its capacity to confer resistance to antimicrobial agents when it is present or increase susceptibility to antimicrobial agents when it is absent17,18. In contrast, Barth concurs with the clinical risk-focused definition of ARGs proposed by Martinez and colleagues, who believe that only those genes involved in clinical antibiotic resistance should be considered as true ARGs18. We would like to clarify that the putative ARGs found in giant viruses are not indicative of clinical risk per se. Nevertheless, some of Barth’s critique further underscores the importance of exploring the specific biological functions of the putative antimicrobial resistance determinants carried by NCLDVs, and we appreciate this valuable discussion.