Next Generation Sequence Immunoprofiling of the T-Cell Repertoire in Chronic Lymphocytic Leukemia Supports Selection By Shared Antigenic Elements

Chronic lymphocytic leukemia (CLL) is characterized by remarkable skewing of the B-cell receptor immunoglobulin (BcR IG) repertoire, culminating in the existence of subsets of patients with stereotyped BcR IGs. This implies antigen selection in the natural history of CLL, ultimately affecting clonal behavior. Currently, limited information is available regarding the role of antigens in the selection and activation of cognate T cells, although this is relevant in light of B and T cell interactions inducing T cell tolerance. Our preliminary next-generation sequencing (NGS) studies in 11 patients assigned to stereotyped subset #4, a clinically indolent disease subgroup, indicated T-cell receptor beta chain (TRB) gene repertoire restriction and oligoclonality. Prompted by these observations, here we sought to obtain a comprehensive view of the T-cell repertoire in CLL by extending our analysis to 36 untreated CLL patients, either assigned to major stereotyped subsets [subset #4 (n=11), subset #1 (n=10), subset #2 (n=4), subset #16 (n=1)] or non-subset cases [with mutated-M (n=5) or unmutated-UM clonotypic BcR IGs (n=5)]. Starting material was PB mononuclear cells (n=27), purified CD4⁺ and CD8⁺ T-cell subpopulations (n=10), bone marrow (n=2) or lymph node (LN) tissue (n=1). Three patients were studied overtime. Multiple sample and PCR replicates, as well as 3 age-matched healthy controls were also included. TRBV-TRBD-TRBJ gene rearrangements were amplified on cDNA (BIOMED2 protocol) and subjected to paired-end NGS, designed to cover the complementarity determining region 3 (CDR3) twice/sequence. In order to further increase the accuracy of results, raw NGS reads were subjected to a purpose-built, bioinformatics algorithm, performing: (i) length and quality filtering of raw reads; (ii) merging of filtered-in paired reads via local alignment; (iii) length and quality filtering of stitched sequences. No base calls of Q-score<30 were allowed in the 75 nucleotide stretch ahead of the FGXG motif, thus further increasing the CDR3 sequencing reliability. Filtered-in sequences were submitted to IMGT/HighV-QUEST, and metadata was processed by an in-house bioinformatics pipeline designed for clonotype computation and repertoire analysis. Overall, 66 samples were analyzed, producing 23,238,779 filtered-in sequences. Only productive, in-frame TRBV-TRBD-TRBJ rearrangements were included in the analysis (85.8% of filtered-in sequences, median 354,972/sample). For repertoire analysis, clonotypes (i.e. TRB rearrangements with identical TRBV gene usage and amino acid CDR3 sequence) were considered (median 37,550 distinct clonotypes/sample; 13,134 expanded versus 22,550 singletons). The distribution of TRBV genes was asymmetric with only 5 genes [TRBV12-3/12-4 (8.4%), TRBV19 (6.8%), TRBV29-1 (6.6%), TRBV5-1 (6.1%), TRBV6-5 (5.2%)] collectively accounting for one-third of the repertoire, indicating functional constraints. No significant TRBV gene repertoire difference was identified across the different immunogenetic CLL subgroups analyzed. In contrast, significant (p<0.001) differences were seen between CD8⁺ versus CD4⁺ cells regarding certain genes (e.g. over-representation of TRBV7-2 in CD4⁺ and under-representation of TRBV5-1 in CD8⁺ cells), alluding to distinct immune selective processes. Notably, the TRB repertoire was significantly more oligoclonal in CLL compared to the healthy controls (median frequency of the predominant clonotype: 5.7% versus 1.27%, respectively, p<0.05), albeit with differences between cases with distinct BcR IG, perhaps reflecting their different immunopathogenesis. Overtime analysis showed persistence of most major clonotypes in 2/3 cases studied, likely in a context of persistent antigenic stimulation. Cluster analysis of all CLL cases identified 17,592 different clonotypes (excluding singletons) shared by different patients and not present in the healthy controls, arguing in favor of a disease-biased immunoprofile; formal proof of this claim will require analysis of a larger number of healthy controls, currently underway by our group. In conclusion, massive parallel sequencing documents the restricted nature of the TRB repertoire in CLL, supporting selection by shared antigenic elements. Whether these are the same antigens interacting with the malignant clone remains to be elucidated.