Skip to main content
Premium Trial:

Request an Annual Quote

Harvard Team Builds Comprehensive Profile of Antibody Response to SARS-CoV-2


NEW YORK – A team led by researchers at Harvard Medical School have extensively profiled the viral epitopes recognized by antibodies present in the blood of COVID-19 patients.

Detailed in a paper published this week in Science, the profiling data demonstrates the complexity of the antibody response to SARS-CoV-2 and could be useful in the development of serology tests and vaccines as well as possibly identifying patients at risk of severe disease, said Stephen Elledge, professor of genetics at Harvard Medical School and senior author on the study.

The researchers employed the VirScan technology developed by Elledge's lab, which uses phage display of viral epitopes to profile subjects' antibody repertoires. The VirScan system presents 56-mer peptides with overlaps of 28 amino acids across the entire proteome of viruses of interest. When these libraries are exposed to patient blood samples, patient antibodies bind to these peptides. The researchers then pull down the bound antibody-phage complexes and sequence the oligonucleotide inside the bound phage to determine what epitope the antibody has bound to. In this way, they can quickly and thoroughly profile a patient's antibody repertoire.

The VirScan platform contains peptides to the full proteomes of all known pathogenic human viruses. For the SARS-CoV-2 work, the researchers also added sub-libraries including three bat coronaviruses that are closely related to SARS-CoV-2 as well as a SARS-CoV-2 proteome consisting in 20-mer peptides overlapping at intervals of five amino acids to provide additional resolution and a mutated SARS-CoV-2 library featuring triple-alanine mutants to allow for more precise mapping of epitope boundaries.

"You can program it to express hundreds of thousands of different peptides … and this way you can make many, many different derivatives and tile through different reading frames on viruses and cover their whole proteome," Elledge said.

This lets researchers look at the antibody response to viral infection in very fine detail, allowing them to identify, for instance, epitopes that are specific to a particular virus or that generate a strong immune response.

Elledge noted that in his lab's previous work using the system he and his colleagues had found that while antibody responses are diverse across populations, there are certain viral epitopes, which they termed "public epitopes," that large percentages of people all target.

"It turns out that people recurrently make the same antibodies … and they will recognize the same peptides and the same parts of those peptides over and over, no matter what continent you are from or what your MHC alleles are, it's kind of remarkable," he said. "That means that there are certain peptides that are very diagnostic for a virus, and using this system allows you to find them right away."

This, Elledge added, could aid in the development of highly sensitive and specific serology assays.

"Instead of one ELISA where you're looking for antibodies that recognize the one protein, we now have dozens of proteins that are very specific [to SARS-CoV-2]," he said.

The fact that the VirScan library contains peptides representing the full proteome of all known pathogenic human viruses allows researchers to identify potential cross-reactive peptides, as well, Elledge noted. "We can eliminate those so that it allows you to get a very high precision and highly sensitive and specific answer."

In the Science study, the researchers looked at 232 COVID-19 patients and 190 controls collected prior to the pandemic, identifying more than 800 antibody epitopes across the virus' proteome. Using machine learning, they built a model based on this data that identified individuals exposed to SARS-CoV-2 with 99 percent sensitivity and 98 percent specificity.

Elledge said that the fact that VirScan uses next-generation sequencing as its read-out makes it highly sensitive, while the fact that it can analyze antibody-epitope binding across a large library of viruses and across many individuals gives it good specificity.

He suggested that labs with access to NGS could employ the approach for serology testing, though he noted that at least in the near term it was more plausible that test developers could use the data generated by his lab to develop better SARS-CoV-2 serology assays.

"These epitopes could be used in other assays with other sorts of detection mechanisms," he said. "There are ways you could imagine using this information… and I'm sure people will be using this [epitope data] to do all kinds of creative assays for serology, which is going to become pretty important, especially once you have a lot of people who are getting vaccinated."

Elledge said that data like that provided by VirScan could also be useful for optimizing vaccines against the virus, noting that it would be important to understand which antibodies were protective and which antibodies might actually interfere with the immune response.

In some instances, viruses can evolve to generate the production of antibodies that are not actually neutralizing, which can hamper the production of neutralizing antibodies. If researchers were able to identify antibodies that appeared counterproductive, it could help with vaccine design, allowing developers to, for example, edit the epitopes to those antibodies out of their vaccine material, Elledge said.

"We're following up on that angle now," he said, noting that he and his colleagues are planning to look for correlations between antibody profiles and sensitivity to the virus.

They explored this question to an extent in the Science study, investigating how patient antibody profiles correlated with disease. They found that patients with more severe infections typically showed a stronger and broader antibody response than did those with less severe infections. Elledge said, though, that this did not suggest anti-SARS-CoV-2 antibody profiles were predictive of disease severity but rather reflected the fact that people with more severe infections typically have higher viral loads, which lead to a more intense antibody response.

He suggested that antibody profiles from previous viral infections could prove useful predictors of how COVID-19 patients will fare. In the Science study, the researchers found that weaker antibody responses to prior infections as well as past exposure to cytomegalovirus and Herpes Simplex Virus-1 were correlated with increased likelihood of hospitalization among COVID-19 patients.

"If you can look at previous infections and how well they dealt with those and how strong their immune system is, you might be able to do some predictions," he said, noting that this could also explain the strong correlation between COVID-19 outcomes and age.

"It could be that what correlates with age is a less effective immune system, and that might be the actual causal effect that age is correlated with," he said. "I think that needs to be explained in a little bit more detail."