Profiling the B/T cell receptor repertoire of lymphocyte derived cell lines

Clonal V(D)J [variable (V), diversity (D) and joining (J)] rearrangement which occurs during development of B/T lymphocytes has been used as a marker to track the clonality of B/T cell populations [1, 2]. This approach is feasible because lymphoid neoplasm/lymphoproliferative cells originate and expand from a single cell; and the progeny cells share the same VDJ rearrangement. A pattern of a monoclonal/oligoclonal population (manifested as the over-representation of either one or a few uniquely rearranged sequences) suggests the presence of a lymphoid neoplasm [or non-malignant clonal lymphoproliferative disorder, such as monoclonal B cell lymphocytosis [3] or monoclonal gammopathy of undetermined significance (MGUS)].

In this study, we systematically profiled the B/T cell receptor repertoire of 936 cancer cell lines across a variety of cancer types, as well as 462 Epstein-Barr Virus (EBV) transformed normal B lymphocyte lines, using RNA sequencing data from the Cancer Cell Line Encyclopedia (CCLE) [4] and Geuvadis RNA sequencing project of 1000 Genomes samples [5]. This study cohort contains cell lines from a variety of solid tumors and 164 blood cancer cell lines (annotated as haematopoietic and lymphoid tissue in CCLE), as well as immortalized “normal” B-lymphocyte cell lines. Cancer cell lines are typically deemed to be “pure”, due to the lack of normal stroma cells and infiltrating T/B cells which are frequently presented in primary tumor samples; thus, this cell line collection provides a unique opportunity to profile faithfully and comprehensively the immunoglobulin/TCR gene rearrangement events in different types of blood cancers.

Transcriptome sequencing data were downloaded from the CCLE and Geuvadis RNA sequencing databases; and the B/T cell receptor repertoire of each cell line was analyzed using MiXCR [6]. The 936 CCLE cancer cell lines were authenticated before uniformly processed RNA sequencing (paired-end 100 × 2 bp) [4]. The growth and uniformly processed RNA sequencing (paired-end 75 × 2 bp) of 462 Epstein-Barr Virus (EBV) transformed normal B lymphocyte lines were described in ref [5]. The B cell and T cell gene expression signature and expression of lineage specific markers (e.g., CD4/CD8 for T cell and CD19/CD20 for B cells) were analyzed from the cell line microarray expression data. The relative clonal CPM value (count per million RNA sequencing reads) was calculated by dividing the clonal read counts by the total RNA sequencing read counts. Each clonotypes were determined based on a unique nucleotide sequence of the VDJ junction (which codes for the CDR3 region) and supported by ≥30 sequencing reads. Two clonotypes have the same rearrangement pattern but have ≥1 bp difference in their nucleotide sequence VDJ junction were considered as different clonotypes. As we cannot exclude the possibility of expression of biallelic rearrangements (where the second allele is usually non-productive or has markedly decreased expression), we refer to bi/oligoclonality as the observation of more than two/four clonotypes within the same type of rearrangement within a cell line [7] (either ≥3 IGH or ≥ 3 IGL or ≥ 3 TRAV, or > 4 IGKV or > 4 TRBV [7]). For example, a cell line with three different IGH rearrangements (e.g., 60% of IGHV6–1-IGHD1–20-IGHJ4, 20% IGHV3–23-IGHD2–8-IGHJ6 and 18% of IGHV3–20-IGHD2–8-IGHJ5) will be regarded as potential bi/oligoclonality. Nonproductive rearrangement was referred as out of frame rearrangement (denoted with “_” in the CDR sequence) or rearrangement with stop codon inside the CDR3 region (denoted with “*” in the CDR3 sequence).

One of the potential pitfalls of profiling the BCR/TCR repertoire using RNA sequencing is that non-transcribed rearrangements may not be detected, and the clonality analysis may be biased by the expression level of BCR/TCR loci. Nevertheless, in comparison to traditional PCR based method which is labor intensive and requires well trained staffs, NGS/RNA sequencing based analysis is more standardized and can be simply outsourced to big sequencing centers. In contrast to PCR based analysis which is limited by the number of loci that can be feasibly examined, NGS based profiling can comprehensively detect all expressed TCR/BCR rearrangements, including some rearrangements which are not usually included in PCR based analysis (e.g., TCRA alpha rearrangement). The BCR/TCR repertoire identified using RNA sequencing were consistent with previous studies using BIOMED-2 PCR Sanger sequencing [29, 30] (The TCR gene names in their paper utilized old aliases and need to be converted to standard gene name using Genecards https://​www.​genecards.​org). In addition, the feasibility of profiling BCR repertoires from RNA sequencing data have recently been explored in CLL [44]. In their study comparing two approaches, BCR repertoire profiling using RNA sequencing showed equal or superior results as compared with traditional PCR and Sanger sequencing (clinical technique) [44]. In addition, with a few exceptions which are likely caused by accidental mix-up, a consistent BCR/TCR repertoire could be obtained when comparing the RNA sequencing data from CCLE with sequencing results for blood cancer cell lines generated by different researchers (for example, cancer cell lines RNA sequencing data in SRA database, https://​www.​ncbi.​nlm.​nih.​gov/​sra). This suggests that the BCR/TCR repertoire pattern may be used as an alternative/complement authorization method for lymphocyte derived cell lines. The advantage of this approach is that when RNA sequencing data is available (which is already widely applied in many kinds of research), the cell lines (need to be lymphocyte derived cells) used in these studies can be simultaneously validated based on the BCR/TCR profile. As such, the RNA sequencing data deposited in the online public database (GEO, SRA) can be further checked and authenticated independently by any researchers. Crucially, we noticed that poly A selection before sequencing is important for examining the BCR/TCR repertoire, as poly A enriched samples generated 100-fold more BCR/TCR sequencing reads as compared to samples which were not subjected to  poly A selection but were sequenced at similar sequencing depth. Nonetheless, RNA sequencing of samples without poly A selection can still discover the majority of BCR/TCR rearrangements. However, accuracy of clonotype fraction of subclonal rearrangements may be significantly affected.

Currently, multiplex PCR based T/B cell receptor rearrangement testing has been used as a clinical approach to detect suspected lymphoproliferative disease [45]. This has occasionally been hindered by a deletion/translocation (e.g., t(11;14)) event in the Ig loci and different set of primer panels may be tested before clonality can be inferred. Recently, the cost of next-generation sequencing (NGS) has quickly decreased and gradually become comparable with traditional PCR-Sanger analysis. RNA sequencing is straightforward with standardized procedures, eliminating the need of patient based personalized BCR/TCR primer sets selection and optimization. Furthermore, RNA sequencing can be scaled up to a large number of samples easily, allowing simultaneous examination of gene expression, SNP and somatic mutations, in addition to the B/TCR rearrangement repertoire. Our analysis highlights the potential of using RNA sequencing as a diagnostic test to examine the BCR/TCR clonal rearrangement in lymphoid malignancy.

The observation of subclonality or potential biclonality in a number of blood cancer cell lines is interesting. In most of these cases, the subclones appear to have been derived from the major clones through somatic hypermutation (e.g., Burkitt lymphoma cell lines NAMALWA, GA-10, EB1, CA46 and Hodgkin lymphoma cell line HS611.T etc.). However, in the B cell lymphoma cell lines NU-DUL-1, JM1 and the Multiple myeloma cell line AMO-1, more than three different IGL (cell lines NU-DUL-1, JM1) or IGH (cell lines AMO-1) rearrangements were detected, suggesting independent biclonality in these cell lines. As cancer cell lines are generally regarded as of monoclonal origin, the potential biclonality at these cell lines is interesting and may require further detailed study. On the other hand, our observation of high clonality and potential clonal selection/evolution in EBV transformed normal B lymphocytes suggests that careful experimental design and interpretation of the result may be required when using EBV transformed lines as a model to study normal B cell population, B cell gene expression or quantitative trait loci (QTL) [40, 46].

In summary, we comprehensively profiled the B/T cell receptor repertoire in 936 cancer cell lines and 462 samples of EBV transformed normal B lymphocytes. The relative “pure” feature of cancer cell lines circumvents the problem of tumor infiltrating T/B and stroma cells in primary tumor samples. Our analysis provides unique insights into the BCR/TCR rearrangement repertoire and clonality of cell lines derived from lymphocyte cells.