Single-Cell DNA Sequencing Reveals Cell-Level Genetic Evolutionary Processes and Clonal Structure in B-Cell Lymphomas

Introduction: A precise depiction of intratumoral clonal diversity reveals the processes of clonal evolution and diversity within tumors. Comprehensive genetic analyses and inference of clonal structure in B-cell lymphomas, such as diffuse large B-cell lymphomas (DLBCL) and follicular lymphomas (FL), have been performed using variant allele fractions (VAFs) obtained from bulk samples. However, this method is limited in terms of predicting precise clonal structure, cell-level co-occurrence, and chronological order of genetic alterations. Moreover, the clonal relationship between the genotype and phenotype at the single-cell level has not been well characterized in B-cell lymphomas. Methods: Simultaneous single-cell DNA sequencing (scDNA-seq) and cell surface protein analyses were performed using the Tapestri platform (Mission Bio). We analyzed 32 samples from 30 patients with B-cell lymphomas, consisting of 12 DLBCLs and 18 FLs. In two cases, both lymph node and bone marrow tumor samples were analyzed. We designed 298 amplicons to cover 61 recurrently mutated genes in FL and DLBCL samples. We also selected 68 custom antibodies with oligotags to capture cell surface protein expression related to B cell differentiation (e.g., CD10, CD19, CD20, CD138, and CD83) and interactions with the tumor microenvironment (e.g., MHC PD-1, PD-L1, CTLA-4, LAG3, BTLA, and LIGHT). The sequenced data were processed using the Tapestri Pipeline for variant calling and multiple filtering processes (Miles L et al., Nature 2020). Phylogenetic trees were inferred using the COMPASS package (Sollier E et al., Nat Commun 2023). Results: We sequenced a median of 6,501 cells per sample (range: 1,039-18,099). A total of 223,336 cells were annotated using conventional B/T cell lineage markers to identify B cells (120,384), CD4-T-cells (38,863), and CD8-T-cells (21,523), with a median coverage of 78 reads per cell per amplicon. Through scDNA-seq, 97 variants were detected (median: 3.5; range: 1-10 per sample) which were used for further analysis. Notably, the average VAFs derived from scDNA-seq exhibited good concordance with the VAFs obtained from matched bulk whole-exome sequencing data (n=15), confirming the reliability of scDNA-seq (R=0.89). The most frequent mutations identified through scDNA-seq in DLBCL cases were CREBBP, TP53, B2M, and MYD88 (n=4, 33%), and CARD11 and HIST1H1E (n=3, 25%). In FL cases, CREBBP mutations were the most frequently detected (n=8, 44%), followed by EZH2, TNFRSF14, and STAT6 mutations (n=4, 23%). Notably, all CREBBP mutations detected in FL cases were found in the initiating cells of pyrogenic trees, suggesting a strong role in initiating mutations in FL. MEF2B, HIST1H1E, and GNA13 mutations were also found in the initiating cells, whereas KMT2D mutations were mainly detected in subclonal cells in FL. Consistent with the findings of previous bulk studies, all TP53 mutations were found in the subclonal cells in both DLBCL and FL cases. Strikingly, all six STAT6 mutations (D419G (n=3), D419N, N421K, and N430T) were detected in subclonal cells, suggesting a contribution to clonal expansion and progressive FL disease. Spatially separated tumors were found to share founder genetic alterations at the single-cell level, and additional alterations were acquired in lymph node (e.g., CREBBP CN-LOH and CTSS Y132D), but not in bone marrow. scDNA-seq also provided information on co-occurrence and mutual exclusiveness at the cellular level. In the case of both MYD88 (L265P) and CD79B mutations (MCD-type DLBCL), although these two gene mutations were concurrently found in most lymphoma cells, 39 of 1,523 lymphoma cells had only MYD88 mutation. Similarly, in an FL case harboring both CREBBP and EP300 mutations, these mutations diverged into two independent branches, indicating mutual exclusion at the cellular level. Conclusion: scDNA sequencing revealed the clonal structures and evolutionary processes of B-cell lymphomas at the cellular level. Specifically, some driver mutations were found to be mutually exclusive, which contrasts with the findings from previous bulk-sequencing studies that showed their co-occurrence at the patient level. Simultaneous profiling of single-cell DNA and cell surface proteins is ongoing, which will provide further insights into the genetic and phenotypic profiles of intratumor heterogeneity in B-cell lymphomas.