P02.13.a third-generation sequencing of brain tumors on single-cell level

Abstract BACKGROUND Intratumor heterogeneity (ITH) in brain tumors has been suggested to increase tumor adaptability and thus drive tumor progression and treatment resistance. Single cell sequencing has widened the range of possibilities in detecting clonal diversity and tumor evolution. As of now, most single cell experiments rely on next-generation sequencing (NGS) platforms. However, they detect only short RNA/DNA fragments that do not show the full picture of existing transcript isoforms or genetic variants. New, long-read sequencing methods such as Oxford Nanopore Technologies (ONT) or PacBio sequencing may now enable novel approaches to single cell experiments. We here set-out to complement existing single-cell analysis workflows and data pipelines with long-read sequencing approaches. We compare and integrate the data output in order to advance the understanding of ITH in brain tumors, paradigmatically demonstrated on meningioma. MATERIAL AND METHODS We set up a pipeline for single-cell/nuclei RNA long-read sequencing on ONT devices. First, libraries as derived from the 10x genomics workflow were employed, subsequently followed by independently generated libraries. To widen the insight into ITH on multiple levels, we next assessed single-cell/nuclei DNA for small somatic variants and methylation along a modified version of this platform. RESULTS Long-read RNA-sequencing yielded results comparable to short-read sequencing. Depending on sample quality, higher reads per cell and library were required for ONT sequencing to achieve similar discrimination of clusters by expression. On the other hand, cell-type specific isoforms could be identified. Specific small somatic variants known from bulk sequencing data were identified when the region was sufficiently covered in the RNA data, which was naturally more often achieved in long read than short-read data. Thus, cell-specific variant calling was enabled in a high-throughput pipeline. Maintaining the methylation information in DNA sequencing could so far only be set-up in a plate-based approach. CONCLUSION Using these approaches, we resolved features of ITH and clonal tumor evolution, aiding us in gaining a broader understanding of brain tumor biology. Furthermore, it showed novel approaches that could possibly be implemented in routine diagnostic procedures in the future. These novel techniques could be applied to other tumor entities and possibly further omics layers such as single cell proteomics and metabolomics could be added to gain a greater understanding of different tumor types and their clonal diversity.