Skip to main content
Premium Trial:

Request an Annual Quote

Large-Scale Unbiased Antibody Capture Method Links Enteroviruses to Acute Flaccid Myelitis


NEW YORK – A team of researchers has provided the most definitive evidence to date that the enteroviruses EV-D68 and EV-A71 — which typically cause outbreaks of respiratory illness among children every other year — cause acute flaccid myelitis (AFM), a rare polio-like illness afflicting children.

Using an a high-throughput immunoprecipitation and sequencing method called VirScan, the team was able to detect significantly higher levels of antibodies to enteroviruses in samples of cerebrospinal fluid from AFM patients.

The research team included collaborators from the University of San Francisco, the Chan Zuckerberg Biohub, Brigham and Women's Hospital, Massachusetts General Hospital, the US Centers for Disease Control and Prevention, the National Institutes of Health, and others.

As described in Nature Medicine this week, antibodies to enteroviruses proved to be the only targets significantly enriched between 42 samples of CSF from AFM patients and 58 samples of CSF from pediatric patients who had other neurological disease. This was particularly remarkable considering that VirScan uses a phage display library covering approximately 3,000 viruses, expressing 481,966 peptides from all known vertebrate viruses and arboviruses.

Efforts prior to VirScan to link the EV-D68 and EV-A71 infections to AFM had been unfulfilling, although the infections were clearly temporally linked to the syndrome's characteristic limb weakness: The viruses had been detected from non-sterile sites in kids with AFM, and also detected in CSF from four patients out of 590 confirmed cases over the past five years.

In 2014, the CDC developed a PCR-based EV-D68 assay to process a high volume of samples their labs were receiving during a large outbreak of severe enterovirus respiratory infections. That was also the first year the CDC began investigating AFM cases, but almost all positive enterovirus detections in kids with limb weakness came from non-sterile sites.

Recently developed animal models have lent credence to the hypothesis that enteroviruses are causing AFM. For example, EV-D68 administered in the blood or gut of mice can make its way to the nervous system and causes paralysis. In people, the virus also seems to selectively damage the neurons in the anterior horn of the spinal cord, much like poliovirus, which also happens to be an enterovirus.

Still, just last year disease detectives at the CDC described the continued challenges of uncovering what precisely was causing the polio-like symptoms. "We keep thinking, what else could we do that's different, what else could we look for, or could we look in a different way? — honestly, there aren't a lot of brilliant new ideas coming forward at this time," Steve Oberste, the Polio and Picornavirus Laboratory Branch chief at CDC, said at the time.

Kevin Messacar, a co-author on the Nature Medicine study and pediatric infectious disease specialist at the University of Colorado, noted in an email that evidence is continuing to accumulate supporting the hypothesis that enteroviruses are the driving factor behind "the seasonal, biennial spikes" of AFM in the US. The study shows, "Although the virus may no longer be present in the spinal fluid when a patient presents with weakness, the footprint of antibodies responding to the virus are present to say the virus had been there before," Messacar said.

The fact that the VirScan method is unbiased is what makes it so powerful, said Ryan Schubert, the first author of the study and a neurologist at UCSF. "It's as much about what it finds as what it doesn't find," Schubert said in an interview.

For example, in the Nature Medicine study the VirScan results showed that patient samples did not have significantly higher levels of antibodies against any of the other 3,000 or so viruses represented in the library.

And, there were similar amounts of antibodies to enteroviruses present in CSF of children in which the virus had previously been detected in other non-sterile sites, such as in stool samples or nasopharyngeal swabs, as there were in samples from children who had no previous evidence of EV infection.

In contrast, the study also evaluated the samples using unbiased metagenomic next-generation RNA sequencing as well as targeted sequencing for enteroviruses. These methods only detected EV-A71 in a single case, at 71.31 reads per million sequences, from a sample that had previously been known to be positive by qPCR and Sanger sequencing. The group also deployed a CRISPR-Cas9-based enrichment technique called FLASH (finding low-abundance sequences by hybridization) but it provided no advantage.

In the study, approximately 7 percent of the control patients also had EV antibodies, but Schubert emphasized that the control cohort consisted of samples of CSF from kids with unexplained neurological illness.

The baseline levels of antibodies to viruses in the CSF of healthy people are not really known, Schubert said, in part because lumbar punctures aren't routinely done on healthy individuals. The healthy samples that have been examined so far have had very low levels of antibodies, if they have any at all, he said, and it is important to consider that low levels of antibodies are occasionally known to move between the blood and the spinal fluid.

The VirScan method uses a technique called phage display that was originally developed in the 1980s, and was deemed worthy of a Nobel prize in Chemistry last year.

"Now with advances in DNA sequencing, we're able to do new things with phage display," Schubert explained.

For example, genetics researcher Steve Elledge and his collaborators at Harvard University originally forged phage display into VirScan, publishing an initial study in 2015 describing its use in blood samples to attempt to determine every virus a person had been exposed to in their lifetime. The Elledge team has since focused VirScan on different areas, including kidney and hematopoietic cell transplantation monitoring.

Elledge's group had also envisioned CSF as an application for the method, he noted in an email, and it published a study comparing CSF and blood for autoantibodies using the original PhIP-Seq technology from which VirScan was created as a specific application.

Meanwhile, "We made our own version of it," Schubert said, specifically as a discovery tool for neurological disease, like multiple sclerosis. "We are interested in autoimmune disease and other diseases where there isn't a clear answer as to what went wrong," he explained.

To create the modified VirScan tool for neurology research, Schubert said he and his colleagues downloaded the sequences of, essentially, all known viruses. They then generated fragments, removed redundancies, made the fragments into DNA, and then cloned all of the pieces into T7 bacteriophage and did phage display. Each phage then had "a viral piece on its head," he said, which was used as "bait" to pull out antibodies from the CSF samples.

Targeted DNA sequencing of the insert subsequently enabled identification and quantification of the viral pieces, and the generation of a fingerprint each patient's CSF immune history.

The nervous system is by and large a sterile site that does not harbor microbes the way the gut or respiratory system do, so CSF is an unusually clean sample type with respect to noise from other, benign, pathogens.

Once the enterovirus signal was discovered, researchers at the Chan Zuckerberg Biohub then made enterovirus protein to confirm the presence of antibodies using ELISA, Schubert said.

The study notes that an attempt was made to identify signatures specific to EV-A71 or EV-D68 using both VirScan and ELISA in blood samples but both assays yielded cross-reactivity in patients with known EV infections, and CSF from AFM cases was commonly enriched for antibodies targeting more than one EV species.

There had been a great urgency to uncover the cause of AFM. "People had tried for so long to find a virus," Schubert said, adding, "It turns out that looking for antibodies against the virus works much better than looking for the virus directly."

Indeed, the AFM data proved to be remarkably clear-cut. "We were all very excited – we all thought that the enteroviruses were responsible for this, so to have it staring right back like that was good to see," he said.

Schubert said the team was motivated in part by first-hand experience treating AFM patients — typically children who present to the hospital with sudden-onset limb weakness a few weeks after an upper-respiratory infection. "Most of us have taken care of patients or know people who have been affected by this," Schubert said, noting that one team member, a neurologist at UCSF, also has a child who had been afflicted with AFM.

Whether VirScan could ever be used for diagnostic purposes remains to be seen. It will likely take more time to develop it beyond a research and discovery tool, but eventually an unbiased screen might be useful, Schubert said. He noted that the UCSF Center for Next-Gen Precision Diagnostics, which offers metagenomic sequencing for diagnostics, could be the type of lab that could bring VirScan on as a testing service. It might also be useful in the future for cases of viral meningitis or encephalitis cases — of which currently about half never have a causative agent identified — potentially helping to improve outcomes for patients.

In an email Elledge said that his group is also continuing to work on new applications for VirScan, for example exploring how the immune system is restructured upon infection with viruses such as measles. The group has also developed a method called T-Scan for discovery of T-cell epitopes, and VirScan has been extended to encephalitis by a former member of Elledge's lab who recently used it to uncover a chronic dengue infection in a patient whose progressive dementia and paralysis had been a mystery.

In terms of potential future diagnostics use, he also noted a method the group is currently working on called ParalleL Analysis of Translated ORFs, or PLATO, that combines in vitro display of full-length proteins with analysis by high-throughput DNA sequencing.