NEW YORK (GenomeWeb) – Australian researchers have developed a deep sequencing-based strategy for characterizing resistance-related mutations in mixed bacterial populations.
The approach, known as resistance mutation sequencing (RM-seq), builds on the rationale behind low error amplicon sequencing (LEA-seq) — a targeted sequencing method that involves molecular barcoding. By tweaking that method to develop RM-seq, the team is able to look at several samples simultaneously.
The researchers also came up with a new bioinformatic pipeline to reliably find resistance mutations, even those present at low frequency, explained first author Romain Guérillot, a microbiology and immunology researcher at the University of Melbourne.
"In a biological sample, you can have a small population of resistant clones that may not be detected by traditional antibiotic resistance testing," he said. "This method, because it's based on deep sequencing, allows us to identify, accurately, very small subpopulations of resistant clones."
That, in turn, may help in detecting resistance earlier than traditional techniques, Guérillot added. Although the researchers have not yet demonstrated that the method is applicable to culture-free bacterial analyses, he noted that RM-seq's sensitivity may make it well suited for clinical applications in the future.
In their proof-of-principle experiments, published last week in Genome Medicine, he and his colleagues used RM-seq to detect previously unappreciated resistance mutations in Staphylococcus aureus. They also applied the approach to cultured Mycobacterium tuberculosis from clinical sputum samples and to a mouse model of S. aureus infection in an effort to start teasing out its clinical applicability, as well as to study the interplay between distinct resistance mutations and infection spread in the body.
"There's no doubt [antimicrobial resistance] is going to lead to a lot more complications in the way we practice medicine and a lot more deaths over the coming decades," said co-corresponding author Benjamin Howden, an applied microbial genomics researcher affiliated with the University of Melbourne, the Peter Doherty Institute for Infection and Immunity, and Austin Health's infectious diseases department.
Howden noted that "one of the key mechanisms of microbial resistance is microbial species developing mutations in key genes that allow them to resist antibiotics."
But while the identities of many resistance genes have been documented, the authors explained, the precise nature of the mutations, the mechanisms by which they arise, their impacts on host-pathogen interactions, and their dynamics in bacterial sub-populations found in a single patient are often far murkier.
"The repertoire of resistance genes (resistome) is now well defined and there are several curated databases and software prediction tools for resistance genes detection," they wrote. "In contrast, comprehensive lists of mutations that confer antibiotic resistance are lacking, despite equivalent clinical relevance."
With that in mind, the researchers set out to establish a reliable method to systematically tally specific resistance mutations. When working out the protocol, Guérillot "was thinking about techniques to better understand how antibiotics drive mutations and how those mutations drive the way bacteria and other microbes behave," Howden recalled.
In contrast to whole-genome sequencing or a more general targeted sequencing strategy, he explained, RM-seq provides a quantitative look at resistance mutations in a high-throughput manner across many microbes.
The researchers first select for microbes containing a range of resistance mutations through exposure to a given antibiotic or drug in vitro. From there, they use PCR to randomly barcode and amplify specific sites in the bacterial genomes, which are then assessed through deep paired-end sequencing and analyzed to annotate the collection of resistance mutations.
"Sequencing reads sharing identical barcodes are grouped to create consensus sequences of the genetic variants initially present in the population," the authors explained, which enhances the accuracy of the approach and offers a quantitative look at resistance allele frequencies.
In their study, for example, the researchers characterized resistance alleles in S. aureus clones selected through growth at different rifampicin concentrations and searched for mutations spelling cross-resistance to a second antibiotic in that bug.
They also tallied RpoB resistance alleles in S. aureus cultured from different mouse organs following infection in vivo, and retrospectively used the amplicon-based approach to search for resistant subpopulations in M. tuberculosis cultures from a patient with chronic pulmonary, multi-drug resistant tuberculosis.
In pools of S. aureus selected on the antibiotic rifampicin, for example, the team identified 72 specific mutations in a known "rifampicin resistance-determining region," including quantifiable mutations in the rpoB gene. Just a handful of specific rifampicin resistance alleles were documented in the region previously, Guérillot noted.
"This illustrates that we lack a good understanding of the mutations that lead to resistance. And this is only for one antibiotic and one particular bug," he said. "If you consider all of the bacteria — different species that acquire different mutations — I think we really need a better understanding of the repertoire of these resistance mutations."
Such alleles can have important impacts on treatment outcomes, even when found at low frequency, Howden explained. During tuberculosis infection, for example, the presence of even a very small subpopulation of drug-resistant M. tuberculosis clones could be amplified after treatment with a given drug.
But resistance alleles found in just a few microbes can be difficult to detect through random sequencing on a bacterial population, Howden explained, calling RM-seq "a very precise quantification of specific resistance alleles in a population."
The RM-seq method has been developed using Illumina short-read sequencing instruments, which currently have an advantage when it comes to the sequencing depth needed to accurately detect and quantify the mutations, Guérillot said. Even so, he noted, it could be adapted to other sequencing technologies, including long read approaches, provided the error rate is sufficiently low.
From their results so far, the study's authors concluded that RM-seq "enables the unbiased quantification of resistance alleles from complex in vitro-derived clone libraries, selectable under any experimental condition, allowing identification and characterization of mutational resistance and its consequences."
The general strategy behind RM-seq should be applicable to any microbe, provided researchers have a sense of the target sequence or genes where resistance mutations might turn up. For previously uncharacterized species or strains, Guérillot noted, it may be beneficial to sequence the genome of the bug first to find resistance hotspots before trying to quantify specific resistance mutations.
"You need to know what you are looking for with the targeting," he said. Once resistance hotspots are known, though, "you can use RM-seq to identify the repertoire of mutations."
Similarly, the strategy does not appear to be influenced by microbial DNA sequence biases or structure.
"We used Staphylococcus aureus, which has low GC [content] and Mycobacterium tuberculosis, which has high GC content," Guérillot said. Usually, differences in GC content can be limiting when performing PCR on bacterial DNA, he said, but"you can do this PCR on any bacteria."
The researchers noted that the RM-seq method may be useful not only for detecting resistance in the clinic and for genomic surveillance to see the advent and spread of resistance mutations in epidemiological studies, but also for understanding and prioritizing resistance mutation drivers in the lab.
"It can go into the clinical front where we use it in patients to … rapidly characterize resistance mutations and how we treat patients," Howden said. "The other thing we're interested in doing is understanding a bit more about how certain mutations provide advantages in [certain] infection sites."
In the interest of applying the approach in clinical settings down the road, the team is in the process of applying RM-seq directly to biological samples, rather than to the types of bacterial cultures used in the current study. It is also interested in developing a suite of primers, or an RM-seq kit, for targeting all of the known resistance hotspots in a given microbe in a multiplexed manner.
The researchers have already made the bioinformatics pipeline that accompanies RM-seq available to the broader research community through Github.