Parallel Sequencing of Transposons and Transcriptomes in Single Cells for Genome-Wide Cancer Gene Discovery

• Published in: Blood 2024-11, Vol.144, p.1351-1351
• Main Authors: Holz, Maximilian, Klinger, Bertram, Stuckenbrok, Felix, Bamopoulos, Stefanos A., Gosavi, Sanket, Bräuning, Caroline, Altmüller, Janine, Rad, Roland, Haferlach, Torsten, Blüthgen, Nils, Keller, Ulrich, Conrad, Thomas, Baumgartner, Francis
• Format: Article
• Language: English
• ISSN: 0006-4971
• DOI: 10.1182/blood-2024-206939

Introduction Genome-wide transposon mutagenesis screens in mice are powerful tools for identifying genomically, transcriptionally, and epigenetically deregulated cancer genes in a single experimental approach. A major limitation, however, is the need for large cohort sizes and substantial sequencing and bioinformatic resources to differentiate causal genetic changes from silent transposon insertions. We address these challenges with single cell PiggyBac Transposon and Transcriptome Sequencing (scPB&T-seq), a novel dual-assay that enables the identification of disrupted molecular networks at the single cell level, thus improving precision and efficacy of cancer gene studies. Methods A murine PiggyBac (PB) transposition screen was performed to generate Asxl1-mutated (Asxl1mut) acute myeloid leukemias (AMLs), a subgroup with particularly poor prognosis. For scPB&T-seq, single cells (sc) from three AMLs were isolated, and genomic DNA was physically separated from mRNA. Sc transposon insertions were amplified via multiple displacement followed by a custom PCR amplification and sequenced alongside sc transcriptomes derived by Smart-Seq2. The data was then integrated with high-throughput droplet-based scRNAseq from the same samples. Results Activating transposon mutagenesis in an Asxl1mut background led to the identification of a large candidate gene catalogue for Asxl1mut AML pathogenesis. Using scPB&T-seq we identified an average of 10-25 transposons per cell, aligning with expected frequencies. All insertions were confirmed in concurrent bulk insertion analyses. About 60% of the insertions impacted gene expression and were retained for further analysis. By integrating scRNAseq data, the insertions were assigned to specific cellular clusters, allowing the reconstruction of clonal evolution and identification of intratumoral heterogeneity. This approach confirmed known AML drivers co-occurring with ASXL1 mutations, such as PRDM16 and MECOM, and revealed novel genomic loci linked to AML pathogenesis. These findings were validated in whole genome and whole transcriptome sequencing data from 774 human AML patients generated within the Munich Leukemia Laboratory 5K Genomes Project. Interestingly, 45 (5.9%) patients from this dataset exhibited a transcriptional profile similar to the main clone of an Asxl1mut PB-AML, showcasing an upregulation of e.g., DOCK5, TCF4, and CSF1. Furthermore, this subgroup of patients displayed a significant enrichment for ASXL1 (31.1% vs