NEW YORK (GenomeWeb) – Researchers from Harvard University, Brigham and Women's Hospital, and elsewhere have developed a high-throughput immunoprecipitation and sequencing method that can gauge interactions between protein sequences from all known viruses and the antibodies present in an individual's blood sample.
In a proof-of-principle study appearing in Science today, the team used the approach — known as VirScan — to look at around 106 million different viral peptide-antibody interactions per person in more than 500 individuals from the US, Peru, Thailand, and South Africa.
Along with age-, geography-, and HIV-status-related differences in apparent viral exposure histories, the researchers uncovered so-called "public epitopes" — sets of amino acids from each virus that tend to be recognized by antibodies from many individuals.
The latter finding suggests different people often produce comparable antibodies to a given viral invader — an insight that's expected help in designing vaccines and future versions of VirScan.
The investigators are interested in using VirScan to search for viruses that may contribute to common, complex diseases, though they noted that the test could eventually evolve into an affordable assay for viral exposures in general.
"Right now, many people are walking around with some diseases — for example, hepatitis C virus — and they're unaware that they have them because they've never been tested," co-first author Tomasz Kula, a graduate student in senior author Stephen Elledge's genetics laboratory at Harvard and Brigham and Women's Hospital, told GenomeWeb.
"Something like this, which you could do on a regular basis, would pick up diseases or ongoing infections that maybe you weren't even aware that you had," Kula added.
VirScan is an extension of the so-called phage immunoprecipitation sequencing, or PhIP-seq, method that members of the same team used to profile autoantibody patterns in individuals' spinal fluid for a Nature Biotechnology study published in 2011.
For autoimmune conditions tested by PhIP-seq in the past, Kula explained, the team established a library of bacteriophages that displayed peptide particles coinciding with human proteome components, providing a platform to test each individual's antibody reactions to human proteins.
In the VirScan method, the researchers tweaked this strategy to produce a library of phages presenting viral epitope particles instead of human proteins. After capturing anti-viral antibodies from individuals' blood samples, the researchers sequence the oligonucleotide taken up by the phage to reveal the identity of the virus that produced the bound protein.
Using viral sequences from RefSeq, Kula and his colleagues designed oligonucleotides encoding 56 amino acids apiece from more than 1,000 viral strains, synthesizing the oligos on a programmable microarray before introducing them to the viral epitope-producing phage library.
Using a 96-well liquid handling system, the team can simultaneously test all of the interactions between these viral proteins and antibodies found in a microliter or less of blood per person, with pooled, multiplexed oligo sequencing providing the readout for these interactions.
At the moment, it costs around $25 per person for VirScan when researchers run 96 PCR products in parallel on a Illumina HiSeq 2000 or 2500 lane, once the phage library has been established, Kula said, with a turnaround time of between two and three days.
In contrast to conventional ELISA-based approaches for identifying past viral exposures, which typically target antibodies against one or a few viruses at a time, the approach makes it possible to screen for exposure to thousands of viruses simultaneously.
But unlike sequencing-based methods that detect viral nucleic acids in the blood, VirScan appears adept at uncovering cleared infections and exposures to viruses that don't circulate in the blood.
For the current study, the team tested samples from 569 individuals from four continents, testing their blood for antibodies against peptides produced by more than 1,000 viral strains representing 206 viral species.
To get a sense of how well various viral epitopes were represented in their phage library, the researchers also sequenced DNA in this library prior to immunoprecipitation with participant blood samples.
Results from those experiments indicated that the vast majority of intended viral strains were found in the viral library, Kula noted. "We were able to detect more than 99 percent of the sequences that we were hoping would be there … that's one of the advantages of using the DNA synthesis approach: it's good at getting a uniform representation across a large number of different fragments," he said.
In samples sequenced after immunoprecipitation, meanwhile, the researchers found that each individual's antibodies interacted with roughly 10 viruses, on average, though two of the individuals carried antibodies coinciding with 84 different viruses.
The method successfully detected known HIV-1 infections around 95 percent of the time with 100 percent specificity, though VirScan's sensitivity was slightly lower for finding known infections with some other viruses such as hepatitis C virus and herpes simplex virus 2.
As anticipated, the team saw evidence of exposure to far more viruses in adults than they did in children under 10 years. On the other hand, blood samples from the children seemed to display stronger responses to some viruses, including viruses coinciding with common childhood infections, Epstein-Barr virus, and cytomegalovirus.
The VirScan sequences also offered a look at the varied viral exposure histories found in individuals with or without HIV or HCV infections. These included a jump in antibodies against a wide range of viruses in the HIV-positive participants.
Within the HIV-negative adults, the researchers reported regional differences in viral exposures. A larger proportion of participants from South Africa, Peru, and Thailand had antibodies recognizing cytomegalovirus than did individuals in the US, for example, while American individuals generally showed signs of lower viral exposure.
Although antibody responses can last a lifetime, Kula cautioned that some responses do wane with time if individuals aren't re-exposed to the same pathogen.
"There are some things that we miss," he explained. "We think that some of the things that we miss are things that you may have been vaccinated against — in which case, you were exposed to the antigen only once and over a long period of time your response is waning."
Nevertheless, the analysis uncovered antibodies from viral epitope exposures stretching back many years, including antibodies reacting to poliovirus. Roughly one-third of HIV-negative individuals carried antibodies against the poliovirus — albeit at relatively low levels — apparently due to childhood vaccinations.
More unexpectedly, the researchers found that antibodies from many individuals tended to interact with shared amino acid sequences for each virus, suggesting these epitopes are especially well-recognized by the human immune system.
The team has already made some improvements to its VirScan analyses by giving more computational heft to antibody interactions involving these common epitopes, Kula noted.
And he and his co-authors argued that taking a closer look at the main epitopes circulating for a given virus may help in coming up with new vaccine targets, getting a more refined look at how our immune systems interact with different viral strains, or developing leaner and more targeted phage libraries for VirScan.
"One improvement that we can make in future generations [of VirScan] is we can make a library that's focused on the peptides against which we already know people have antibodies," Kula said. "That will make the library much smaller, which could mean that you don't need as much sequencing depth to read out the results of the immunoprecipitation."
Although the researchers used Illumina instruments to sequence immunoprecipitation products from VirScan, they suspect the methodology will be platform agnostic and compatible with other sequencing technologies as well.
At the moment, the team generates around one million reads per sample, though there appears to be some flexibility in the depth needed to identify the viruses present.
Because the epitope library is produced in phages, VirScan will not pick up on the presence of host antigens that recognize viral epitopes carrying post-translational modifications, meaning the approach may miss some epitopes from a particular virus.
The approach is also limited to finding known viruses, though Kula said it should be relatively straightforward to add epitopes from newly discovered viruses once sequences are available for them.
Similarly, he noted that it might be useful to add epitopes from some viruses that are currently circulating in animals, but are not believed to have crossed over into humans as a way of monitoring such exposures.
Those involved in the study are also considering the possibility of expanding the assay to interrogate individuals' bacterial or fungal exposures, though the much larger genome sizes for these organisms is expected to make phage library production far trickier than that required for the virome assay.
In the nearer term, the researchers hope to apply VirScan to studies of complex human diseases suspected of having viral triggers such as type 1 diabetes. They are also interested in investigating pathogen interactions with the immune system that may inform studies of immunotherapy in cancer.
In particular, Kula pointed to a New England Journal of Medicine paper published a few months ago, which hinted that the success of treating cancer with immunotherapy-based approaches might depend, in part, on the presence or absence of amino acid motifs shared between tumors and common pathogens.
If that's the case, he explained, it's possible that a pre-existing immune response to these pathogens could boost the immune system's tumor-fighting potential — a hypothesis that the researchers hope to explore with VirScan.
"This is just an intriguing possibility, but up until now it's been hard to test unless you know ahead of time which virus might be involved," he said. "So we'd be very interested in getting samples from patients who did or did not respond to tumor immunotherapy and seeing whether there are any viral correlates."