Skip to main content
Premium Trial:

Request an Annual Quote

CRISPR-Based Antimicrobial Resistance Dx Tool Makes Metagenomic Sequencing More Accessible


NEW YORK (GenomeWeb) – Researchers at the Chan-Zuckerberg Biohub and their colleagues in the US, the UK, and Namibia have developed a CRISPR-based method for the detection of antimicrobial resistance (AMR) genes that has the potential to allow smaller labs with less expensive instruments to conduct metagenomic sequencing.

Pathogens that are resistant to currently available medications are an increasing threat — drug-resistant infections currently cause 700,000 deaths each year and are predicted to cause 10 million deaths annually by 2050, according to a 2016 report by the UK's Review on Antimicrobial Resistance. The problem is reaching such proportions that many companies are working to develop systems that can rapidly identify both the pathogens that are causing disease and the genes that are causing those pathogens to stop responding to even the harshest treatments.

Currently, most clinical labs look to isolate pathogens in culture in order to then ascertain antimicrobial susceptibility. But this process can take anywhere from 48 hours to six weeks, depending on the organism in question, and doesn't always result in the successful growth of bacteria.

"You can't always culture every bacterial strain that's causing an infection. That's especially true when a patient will present to the E.R. having a serious bacterial infection, and the very first thing that the doctors there will do is give them broad spectrum antibiotics," said Emily Crawford, a researcher at the University of California, San Francisco and a project leader at the CZ Biohub. Crawford is the senior author on the new study in Nucleic Acids Research that describes the CRISPR-based AMR detection method.

Crawford explained that although broad-spectrum antibiotics are typically helpful in clearing up nasty infections, they also cause a decrease in the presence of a pathogen, even if they don't wipe it out completely. "Even if the pathogen is resistant, it can be such a big hit to the number of viable bacteria in the sample that you'll end up having a culture-negative sample after that," she added. "You won't be able to actually culture anything out of it, so you've lost that chance to look at resistance."

Another popular method for the detection of AMR genes, which is much faster than culturing, is PCR. But while those kinds of assays typically take 24 hours to yield results, many of them look for individual AMR genes or mutations, Crawford said, which isn't very useful if a patient comes in with an uncharacterized infection.

The next step up, metagenomic sequencing, is great for detecting pathogens in clinical samples. But when it comes to finding AMR genes, the relatively low abundance of target DNA in high levels of background human genetic or human microbiome DNA can lead to suboptimal results. Such sequencing also requires the use of expensive instruments, which can be out of reach for many labs.

The combination of multiplex PCR with next-generation sequencing — such as 16S rRNA gene profiling or Ion Torrent's AmpliSeq technology — provides effective enrichment of low-abundance targets and a quick turnaround, but such methods are also expensive to use, and are therefore less accessible to labs and hospitals in countries and settings that may have fewer resources.

For Crawford and her colleagues, CRISPR-Cas9 presented the perfect solution to all these problems — their FLASH method (Finding Low Abundance Sequences by Hybridization) is a next-generation CRISPR-Cas9 diagnostic approach that uses Cas9 to enrich for a programmed set of sequences. It has the same 24-hour turnaround time as the PCR assays and, like the AmpliSeq technology, it can target all known AMR sequences at once.

"FLASH is completely programmable in the sense that all you have to do is look for CRISPR-Cas9 target sites in any genes you want, and you can easily switch out or add in new AMR targets. That's much harder to do with multiplex PCR," Crawford said. "FLASH can have a 24-hour turnaround under good conditions. The main focus of my group right now is in looking at drug-resistant tuberculosis — that's a pathogen that grows so slowly that phenotypic testing can take on the order of six weeks-plus to get a phenotypic answer. If we can get a genotypic answer within 24 hours, that makes a huge difference in treatment."

FLASH uses a set of Cas9 guide RNAs designed to cleave sequences of interest into fragments that can then be sequenced. Input genomic DNA or cDNA is first blocked by phosphatase treatment and then digested with Cas9 complexed to this set of gRNAs. To choose optimal gRNA targets for FLASH, the team also developed a flexible computational tool called FLASHit which first defines targetable 20-mer Cas9 sites and then designs a relatively small set of gRNAs that provides a relatively high sequence coverage.

While a single FLASH-derived fragment is sometimes sufficient to uniquely identify an AMR gene, the guides were designed to cut each gene into multiple fragments, both to increase the probability of detection.

"Some Cas9 sites don't work very efficiently in this method, and because of that, we felt it was important to emphasize the use of our software tool FLASHit to build redundancy into a FLASH experiment," Crawford said.

FLASH is able to target hundreds or thousands of AMR genes at once. But more significantly, the amplification of the target genes that results from the use of Cas9 effectively depletes background DNA sequences, including any human or human microbiome DNA that may be in the sample.

This is especially important in light of how and where Crawford and her colleagues would like to see FLASH deployed.

"AMR is a big problem in the developed world but an even bigger problem in developing countries where there's poor public health use of antibiotics and where there's also a lot of antibiotics given to livestock," she said. "There are a growing number of labs around the world — and that includes both academic labs in low-resource countries as well as public health labs across the United States and other developed countries — that have access to next-generation sequencing, but they tend to have fairly limited access, and usually it will be an Illumina MiSeq instrument or an iSeq instrument. When we think about metagenomic sequencing, which is essentially culture-free or pre-culture sequencing, that's something that's really been only accessible to a large research labs that have access to an Illumina NovaSeq or HighSeq where you might be paying $10,000 or more for a single sequencing run."

She also noted that most metagenomic samples such as blood, respiratory fluid, or stool are quite complex and usually contain a lot of human genetic and microbiome material.

"You need to sequence really deep to get at the actual pathogen and even deeper to get at the AMR genes. FLASH basically eliminates all of that background, and you can get the same answers as far as the presence and identity of AMR sequences, with a thousand-fold fewer sequencing resources," Crawford added. "It takes the accessibility of metagenomic sequencing from being only accessible if you have a million-dollar NovaSeq instrument at your big research institution to being equally as accessible to a small lab that would just have an iSeq or a MiSeq."

Further, because of its ability to drown out the background noise, the method is fairly sample-type agnostic. In their paper, Crawford and her colleagues described using the method to test for AMR genes in respiratory fluid from four patients with respiratory infections. They also used FLASH to look for drug-resistance genes in dried blood spots from patients infected with the malaria parasite Plasmodium falciparum.

The team has also experimented with FLASH in cerebrospinal fluid samples and a student in Crawford's lab is currently analyzing its effectiveness in stool samples. "There's no particular reason why any sample type shouldn't be able to work," she said, adding that the method is also disease agnostic, and will most likely detect "anything that's got nucleic acid."

In fact, the authors concluded in their study, FLASH could be used to detect rare mosaic alleles, for targeted transcriptomics from clinical samples, for enrichment of microbiome components from complex mixtures, and for recovery of targeted transcripts from single-cell sequencing libraries.

It could even be used to detect cancer mutations. If the method proves effective in stool samples, companies looking to compete with Exact Sciences for the colorectal cancer at-home stool testing market may come knocking. Crawford said she's happy to talk to pharmaceutical or biotech companies that may want to partner in commercializing FLASH as a diagnostic test or assay. So far, the Biohub researchers aren't engaged in any such talks, but that option is on the table.

What she's most excited about, she said, "is that there are many labs — public county health departments and small hospitals around the world — that are just now starting to use these [sequencing] instruments. They're now going to be able to actually use them to do some more advanced things like look at AMR genes."