Tumor‐infiltrating lymphocytes (TILs) are direct effectors of tumor immunity, and their characterization is important for further development of immunotherapy. Recent advances in high‐throughput sequencing technologies have enabled a comprehensive analysis of T‐cell receptor (TCR) complementarity‐determining region 3 (CDR3) sequences, which may provide information of therapeutic importance. We developed a high‐fidelity target sequencing method with the ability for absolute quantitation, and performed large‐scale sequencing of TCR beta chain (TCRB) CDR3 regions in TILs and peripheral blood lymphocytes (PBLs). The estimated TCRB repertoire sizes of PBLs from four healthy individuals and TILs from four colorectal cancer tissue samples were 608,664–1,003,098 and 90,228–223,757, respectively. The usage of J‐ and V‐regions was similar in PBLs and TILs. Proportions of CDR3 amino acid (aa) sequences occupying more than 0.01% of the total molecular population were 0.33–0.43% in PBLs and 1.3–3.6% in TILs. Additional low coverage sequencing of 15 samples identified five CDR3 aa sequences that were shared by nine patients, one sequence shared by 10 patients, and one sequence shared by 12 patients. The estimated size of the TCRB repertoire in TILs was significantly smaller than that in PBLs. The proportion of abundant species (>0.01%) in TILs was larger than that in PBLs. Shared CDR3 aa sequences represent a response to common antigens, and the identification of such CDR3 sequences may be beneficial in developing clinical biomarkers.

BackgroundGenotyping of the human leucocyte antigen (HLA) is indispensable for various medical treatments. However, unambiguous genotyping is technically challenging due to high polymorphism of the corresponding genomic region. Next-generation sequencing is changing the landscape of genotyping. In addition to high throughput of data, its additional advantage is that DNA templates are derived from single molecules, which is a strong merit for the phasing problem. Although most currently developed technologies use genomic DNA, use of cDNA could enable genotyping with reduced costs in data production and analysis. We thus developed an HLA genotyping system based on next-generation sequencing of cDNA.MethodsEach HLA gene was divided into 3 or 4 target regions subjected to PCR amplification and subsequent sequencing with Ion Torrent PGM. The sequence data were then subjected to an automated analysis. The principle of the analysis was to construct candidate sequences generated from all possible combinations of variable bases and arrange them in decreasing order of the number of reads. Upon collecting candidate sequences from all target regions, 2 haplotypes were usually assigned. Cases not assigned 2 haplotypes were forwarded to 4 additional processes: selection of candidate sequences applying more stringent criteria, removal of artificial haplotypes, selection of candidate sequences with a relaxed threshold for sequence matching, and countermeasure for incomplete sequences in the HLA database.ResultsThe genotyping system was evaluated using 30 samples; the overall accuracy was 97.0% at the field 3 level and 98.3% at the G group level. With one sample, genotyping of DPB1 was not completed due to short read size. We then developed a method for complete sequencing of individual molecules of the DPB1 gene, using the molecular barcode technology.ConclusionThe performance of the automatic genotyping system was comparable to that of systems developed in previous studies. Thus, next-generation sequencing of cDNA is a viable option for HLA genotyping.Electronic supplementary materialThe online version of this article (10.1186/s12864-017-4300-7) contains supplementary material, which is available to authorized users.