Sequencing rare T-cell populations

High-throughput sequencing of T cell (TCR) receptor repertoires is a promising approach that can be used to characterize the state and dynamics of adaptive immunity and, potentially, to deduce the antigen specificity of the immune response [1]. While the number of applications of TCR profiling is constantly growing, basic protocols have several drawbacks that are making it difficult to utilize them in certain experimental settings. In particular, quantitative data interpretation is hampered by stochasticity of sampling of lymphocytes and TCR-encoding RNA/ DNA, stochasticity and biases of PCR amplification, and sequencing quality biases. These problems become more evident when profiling rare T-cell subpopulations, such as tumorinfiltrating lymphocytes (TILs) that may contain low lymphocyte counts, populations enriched for tumor antigen-specific cells, or functional T cell subests of interest, the tasks critical for studying the role of adaptive immunity in cancer progression and treatment. In this brief note we summarize what we suggest to be the most relevant issues of studying tumor-specific TCR repertoire for the minor lymphocyte counts, where derived data is more prone to noise introduced by stochastic sampling, which renders methods that operate with T-cell clonotype frequencies less useful. As a way to resolve these challenges we refer to our recently published highly-sensitive T-cell repertoire profiling protocol that enables studying rare lymphocyte populations [2]. The protocol employs molecular barcoding technique that was proven to be extremely useful for the immune repertoire profiling task [3-5]. This approach allows tracing raw sequencing reads to their initial cDNA molecules, yielding robust immune repertoire data that is normalized and corrected for PCR and sequencing errors (Figure 1). First, molecular barcoding provides control for the experimental bottlenecks. This is critical when working with limited available material, as researcher is blind in the respect of the counts of analyzed cells and molecules when using conventional approaches. For example, 10 molecules that were sufficiently amplified and analyzed with 100,000 sequencing reads may look like a highly convergent “repertoire” after all accumulated errors. Editorial