This story has been updated to include comments from an outside researcher.
SAN FRANCISCO (GenomeWeb) – By sequencing the immune repertoires of more than 600 individuals, Adaptive Biotechnologies was able to identify a set of T cell receptor sequences associated with cytomegalovirus exposure. Those T cell receptor sequences were present at elevated levels in individuals exposed to the virus and absent in those who were not exposed.
The company published the results of the proof-of-principal study in Nature Genetics this week and is now applying the same approach to look for immune repertoire signatures in other diseases, according to Harlan Robins, head of innovation at Adaptive Biotechnologies.
Importantly, Robins said, the team found that there is "enough overlap in the adaptive immune response to be able to find enough shared T cell receptors to be able to diagnose the disease," he said.
Healthy humans express approximately 10,000,000 unique T cell receptor beta chains. While specific T cells multiply in response to exposure to an antigen, it has been unclear whether T cells that respond to a specific antigen in one individual would be the same in another or whether that response is unique. In the Nature Genetics study, the Adaptive group adds to evidence suggesting that individuals do share T cell receptor (TCR) sequences in response to exposure to the same antigen.
One goal of the study, Robins said, was to answer the question of whether there is a shared adaptive immune response to a specific pathogen among individuals. "Reading the adaptive immune system will be hard if there's no shared response," Robins said. "If the T cells you make in response to a pathogen are different than the ones I make, it would be nearly impossible to read," he said.
Robins said that the team decided to look at CMV infection as a test case because up to 50 percent of individuals have been exposed to CMV, so it would be relatively easy to get enough individuals who were CMV positive. The researchers sequenced the TCRs of 666 individuals with known CMV serostatus.
In the study, the Adaptive team first sequenced the T cell repertoires of 666 healthy bone marrow donors. CMV serotyping confirmed that 289 were positive for CMV and 352 were negative for CMV.
Sequencing identified more than 89 million unique TCR beta sequences, or an average of around 192,000 per individual. Next, the researchers analyzed the sequences to look for those that were enriched in CMV-positive individuals compared to the CMV-negative group and identified 164 TCR beta sequences that were associated with CMV.
They then came up with statistical methods to predict CMV status. Essentially, individuals with elevated levels of a handful of the TCR clonotypes would have CMV, Robins said.
To test this, the team sequenced the TCRs of a second unrelated cohort of 120 individuals and also performed CMV serotyping, demonstrating that they could accurately predict CMV infection with 90 percent sensitivity, 88 percent specificity, and a diagnostic odds ratio of 70.
"We were able to basically diagnose CMV as the underlying pathogen" using the set of shared TCR beta sequences, Robins said.
Jian Han, a faculty investigator at the HudsonAlpha Institute for Biotechnology, who was not involved with the study, said that the study was "very good." Han has used a similar strategy for identifying disease signatures as part of the R10K project to sequence the immune repertoires of 10,000 individuals. He added that one important next step would be to validate that the Adaptive team's results were in fact a real signal and not due to either analytical or biological noise. Nonetheless, he said the idea of identifying disease signatures in the adaptive immune system is one his team has also been pursuing and has found promising results.
Aside from demonstrating the proof of principal that immune sequencing can identify disease-associated TCRs, Robins said that the study also illustrated that some of the traditional immunological techniques are not very sensitive. For instance, he said, the researchers identified in the literature around 900 TCR sequences that have previously been reported as being related to CMV. Those 900 TCR sequences had been identified by a method called tetramer sorting. Of those, 900, the Adaptive team observed about half in their cohort, but only nine were enriched in the CMV-positive cohort.
Robins said that although the tetramer sorting technique does identify the correct TCR sequences, it also results in a lot of false positives, making it difficult to separate out the real signal from the noise.
Robins said that the firm plans to move beyond this proof of principal to further refine the techniques and apply them to other diseases. He said researchers plan to use similar approaches on more complex and rare diseases, including other infectious diseases, autoimmune disorders, and even cancer. In addition, he said, the firm is also developing new methods and techniques that will enable it to match TCRs with their antigens as well as techniques to help tease out the function of various TCRs. Robins declined to elaborate further on these techniques, but said the company was looking to commercialize them.
This latest research from Adaptive, while seemingly a new direction for the company, is nonetheless a natural extension of the firm's immune repertoire sequencing business, Robins said. The company markets a TCR repertoire profiling research-use-only assay dubbed ImmunoSeq and the clinical version, ClonoSeq, which is used for minimal residual disease analysis in blood-based cancers. Through its collaborations, it has generated a wealth of immune repertoire sequencing data, which can now be leveraged to identify these pathogen-specific TCRs, like in the Nature Genetics study.
"It's all about the data," Robins said. "If you can get data that contains massive amounts of information and start to understand that, the power is incredible."
Robins said the firm now has ongoing research collaborations in the infectious disease, autoimmune, and cancer spaces. He said that so far, preliminary results are promising that it will be able to identify disease-specific TCRs. However, he said, cancer is more complicated, "because while some of the immune response is to a shared antigen, there are also unique neoantigens," that elicit unique immune responses. That's in contrast to say, an infectious disease, he said, where "everyone more or less responds to the same antigen."