Tracking antibiotic resistance

See allHide authors and affiliations

Science  19 Sep 2014:
Vol. 345, Issue 6203, pp. 1454-1455
DOI: 10.1126/science.1260471

According to the World Health Organization (1), “Antimicrobial resistance…threatens the effective prevention and treatment of an ever-increasing range of infections caused by bacteria, parasites, viruses and fungi….A post-antibiotic era—in which common infections and minor injuries can kill—far from being an apocalyptic fantasy, is instead a very real possibility for the 21st Century.” Often, antibiotic-resistance genes acquired by bacteria are associated with mobile genetic elements that mediate their exchange between pathogens, or between commensal and pathogenic bacterial populations. Thus, the genetic context in which an antibiotic-resistance gene is placed can reflect its mobility within the bacterial population. Although the presence of resistance genes in clinical isolates can be detected with either polymerase chain reaction technology or high-throughput short–read length DNA sequencing, their precise genetic context may be overlooked because of the limitations of these methodologies. As Conlan et al. (2) demonstrate, the use of single-molecule long-read DNA sequencing can address this limitation and opens the way to track the transmission of antibiotic-resistant bacteria in a health care environment.

The transfer of genetic elements between bacteria, known as horizontal transfer (3), can occur through plasmids, small DNA molecules that are separate from bacterial chromosomal DNA. Conlan et al. applied long-read genome sequencing to identify plasmids harboring the gene blaKPC, which encodes a carbapenemase that hydrolyzes carbapenem antibiotics. This analysis was carried out after an outbreak in 2011 at the U.S. National Institutes of Health Clinical Center in which six infected patients died from carbapenem-resistant Klebsiella pneumoniae (4). Conlan et al. surveyed more than 1000 patients for carbapenem-resistant bacteria. The bacterial genomes from such infected patients were analyzed with long-read sequencing (single-molecule real-time sequencing). Sixty-three plasmid sequences were completed from patient or hospital environmental isolates, providing a unique snapshot of plasmid complexity from a single geographical site. The sequences allowed elucidation of the fine detail of plasmid diversity that would not have been obvious otherwise. Two patients infected with near-identical carbapenemase-resistant Enterobacter cloacae strains harbored blaKPC on the same plasmid (pKPC-47e), triggering increased surveillance measures. Shared carbapenemase-encoding plasmids were found in three different bacterial genera epidemiologically linked with hospital environments (sinks). Of five patients who presented with carbapenemase-resistant Klebsiella sp., long-read genome sequencing associated one of these patients to the original 2011 outbreak, indicating nosocomial transmission. Long-read genome sequencing unequivocally ruled out a link with the 2011 outbreak in the remaining cases, despite epidemiological evidence suggesting otherwise. Carriage of shared plasmids was observed between a patient isolate and isolates of other bacterial species present in the ward environment. For several cases, compelling evidence to exclude nosocomial transmission may have been gathered from short-read sequencing, but long-read sequencing not only produced the required diagnostic result, but additionally provided reference sequences for all chromosomes and plasmids to aid future surveillance. Plasmid comparisons also disproved the hypothesis that transfer of a blaKPC plasmid had occurred between K. pneumoniae and E. cloacae within a patient from the original outbreak (4); long-read genome sequencing showed discrete genetic contexts for the two blaKPC alleles, indicating independent acquisition. Long-read genome sequencing also proved critical for determining the precise genomic context of blaKPC alleles in cases where more than one blaKPC gene was present in a single isolate.


Long-read sequencing can resolve complete plasmids from bacterial genome assemblies. Plasmid comparisons can help to track the dissemination of antimicrobial-resistance genes among pathogenic bacteria in a clinical setting. By contrast, short-read sequencing will generally produce collapsed repeats at repeated locations such as insertion sequences (black) that are dispersed throughout the plasmid backbone (white), making the context of antibiotic-resistance genes (pink, blue, yellow) difficult to discern.


Plasmids play a key role in disseminating antimicrobial-resistance genes among pathogenic bacteria. Resistance genes often aggregate in modular clusters (5) that enable resistance to persist even in the absence of direct selection. Some insertion sequences (IS) (6), such as IS26, sandwich resistance genes into a composite transposon that can mobilize elsewhere. Other elements, such as ISEcp1, mobilize adjacent genes, leading to multiple copies of resistance genes within the genome. The multiplicity of insertion sequences and other repetitive elements is a major confounding factor in standard short-read genome assembly, which can lead to the fragmentation of chromosome and plasmid sequence assemblies. In this sense, plasmids may be viewed as the “dark matter” of short-read bacterial genome assemblies, with many large-scale genomic studies conspicuously avoiding the complexities of plasmid structure. Genomic comparisons such as that described by Conlan et al. reveal how the dynamism in the structure and arrangement of resistance elements can only be realized by “closing” plasmid genomes with long-read sequencing (see the figure).

Traditional Sanger sequencing is the gold standard for the analysis and assembly of complete plasmid sequences from antibiotic-resistant strains of bacteria. This approach may suffer from the need to isolate or subclone individual high molecular weight plasmids before sequencing (7), which is often technically difficult, time-consuming, and costly, and may be intractable for multiple plasmids. Short-read sequencing technologies can affordably produce an assembly of a bacterial genome that contains nonrepetitive sequences typically in hundreds of “contigs” separated by “collapsed repeats” indicative of multiple copies of the same sequence located in several different locations within the genome. These repeats are often mobile elements such as insertion sequences that may be found in multiple copies on plasmids, thus making it difficult to assemble plasmid sequences.

Cataloging the collection of antibiotic-resistance genes in any particular bacterium is relatively straightforward, but determining how these genes fit together within plasmids, which is critical for understanding how these elements spread in clinical settings, can be more difficult. By contrast, the genome sequences produced through long-read sequencing offer a complete picture of the plasmid content of a bacterium, including the number, position, and context within mobile elements of every acquired antibiotic-resistance gene. Long-read genome assembly offers clear advantages for the resolution of complete plasmid sequences that can discriminate plasmid diversity, antimicrobial-resistance gene context, and multiplicity. Such information will enhance our understanding of plasmid carriage, transfer, epidemiology, and evolution.


View Abstract

Stay Connected to Science

Navigate This Article