Abstract
Background: High antimicrobial usage in swine production has the potential to create reservoirs of antimicrobial resistance (AMR) genes which are transferable to human pathogens via mobile genetic elements. Understanding microbial community responses to antibiotic use is central to unravelling transfer of such resistance genes. Our previous investigation revealed a scenario of optimal antibiotic activity associated with saturation of AMR genes on this farm. Here, we use amplicon and shotgun sequence data to investigate the microbiome signatures that underwrite such a phenomenon.Results: We generated 1.24 and 576 million high quality 16S rRNA gene amplicon and shotgun sequences from 24 porcine faecal samples, respectively. The ratio of taxa detection at genus level between the two methods was 1:24. Using shotgun sequence data, 235 unique AMR genes, 122 modes of action and 17 antibiotic classes were identified using the MEGARes AMR database. Antibiotic usage in growing pigs was significantly associated with microbial and AMR resistome structural and compositional changes detectable two weeks after antibiotic initiation. These were characterised by a down regulation of MDR efflux pumps and an up regulation of macrolide-specific efflux pumps in the growing pigs (treated-group) linked to lower abundance of Verrucomicrobiaeceae. In the sows (non-treated group), a potentially undetected infection, was characterised by a high abundance of pathogenic viral sequences, microbial structural changes i.e. family Alcaligenaceae, and an up regulation of beta-lactamases, including MDR efflux pumps. We assembled 682 near complete bacterial genomes revealing that a large proportion of the resistome is carried by Firmicutes and Proteobacteria, specifically multi-class gene carriage by Clostridium species and Escherichia coli, which occurred exclusively in the treatment group.Conclusion: Microbiome signatures i.e. microbial structure, composition and resistome carriage associated with antibiotic-use can be cost effectively screened with amplicon sequencing but their granularity unravelled using shotgun metagenomic data.