Characterizing the full complement of antimicrobial resistance genes and linking the resistance genes and plasmid to source bacteria

Abstract

Antimicrobial resistance (AMR) is a global public health threat. Selective pressure exerted by antimicrobial use has been the major driving force, more than 2.8 million AMR infections occur yearly in the United States. The intestinal microbiome is an important reservoir of antimicrobial resistance genes. Yet, little is known about the true diversity of the antimicrobial resistance genes in food animal intestinal microbiomes. We employed deep shotgun metagenomic sequencing and proximity ligation (Hi-C) library sequencing to characterize the resistome, assemble genomes from metagenomic samples, and accurately attribute antibiotic resistance genes and plasmids to host bacteria cells. We randomly selected 21 cecal samples from food animal sources (cattle n= 6, swine n= 6, chicken n= 3, and turkey n= 6). We generated more than 75 million reads/sample for Hi-C and more than 100 million reads/sample for shotgun metagenomic sequence reads. Bioinformatics analysis revealed over 200 bins containing metagenome assembled genomes (MAGs) with different levels of completeness, novelty scores, and contamination based on CheckM. A total of 245 previously uncharacterized genomes were reconstructed with high level of confidence (>90% Completeness, >90% Novelty, < 5% contamination). Of the 245 newly reconstructed MAGs, 24 were at bacteria taxonomic rank level, 5 at phyla (Actinobacteria; 11 genomes. Firmicutes; 3 genomes. Bacteroidetes; 17 genomes. Euryarchaeota; 2 genome), 4 at class (Bacilli; 1 genome. Clostridia; 9 genomes. Deltaproteobacteria; 3 genomes. Gammaproteobacteria; 1 genome), 5 at order level (Actinomycetales; 2 genomes. Bacteroidales; 25 genomes. Clostridiales; 114 genomes. Lactobacillales; 2 genomes. Selenomonadales; 2 genomes), and 3 at family level (Lachnospiraceae, 24 genomes. Spirochaetaceae; 1 genome. Spirochaetaceae; 2 genomes). We identified over 400 antimicrobial resistance genes representing 22 antimicrobial classes including: aminoglycoside (40 gene variants), beta-lactams (37 gene variants), bleomycin (2 gene variants), colistin (3 gene variants), fosfomycin (4 gene variants), glycopeptide (6 gene variants), lincosamide (9 gene variants), lincosamide/streptogramin (2 gene variants), macrolide (16 gene variants), macrolide/lincosamide/streptogramin (4 gene variants), nitroimidazole (1 gene variant), phenicol (9 gene variants), phenicol/oxazolidinone (1 gene variant), phenicol/quinolone (2 gene variants), pleuromutilin ( 1 gene variant), quinolone (5 gene variants), streptogramin (1 gene variant), streptothricin (3 gene variants), sulfonamide (3 gene variants), tetracycline (25 gene variants), and trimethoprim (8 gene variants). Plasmid characterization using Hi-C proximity ligation and shotgun metagenomics allowed the identification of 146 plasmids (>= 85% completeness, >= 90% reference sequence similarity), and over 13000 plasmid-contigs (<85% completeness, < 90% reference sequence similarity). Shotgun metagenomics provide valuable insights into the diversity and identity of the resistome present in a microbiome, while Hi-C generates millions of paired-end reads linking DNA fragments in close proximity. When shotgun metagenomics is coupled with the Hi-C proximity ligation approach it shows a great capability in genome binning and simultaneous retrieval of high-quality MAGs from a single sample, thusly enabling the link of resistance genes and plasmids to host bacterial cells and facilitating the public health management decisions aimed at reducing the source and exposure routes of AMR to humans.