(P 390) NOVEL ISOFORM DISCOVERY IN SARCOMA USING FULL-LENGTH PACBIO ISO-SEQ

Objective: Sarcomas can be driven either by genomic translocation resulting in fusion genes, or genomic instability. Both classes of sarcoma have poor outcomes, highlighting the need to better understand these diseases and develop novel therapeutic strategies. One aspect of tumour biology with a high potential to yield new insights into tumour-specific targets is transcriptional diversity. However, commonly used short-read sequencing approaches are insufficiently powered for high-confidence transcript isoform inference. Here, we employ the high-fidelity (HiFi), long-read sequencing PacBio Iso-Seq platform to directly sequence full-length (FL) transcripts and discover novel isoforms. We apply this approach to a sarcoma cohort including (1) alveolar soft part sarcoma (ASPS) and synovial sarcoma (SYNS), both fusion-driven tumour types, and (2) angiosarcoma (ANGS) and undifferentiated pleomorphic sarcoma (UPS), both genetically complex.

Methods: Twelve patient samples were profiled with the PacBio Iso-Seq platform from total RNA, including 2 SYNS, 3 ASPS (primary & metastasis), 4 ANGS (biopsy, primary, recurrence), and 3 UPS (untreated, primary/neoadjuvant treated) cases. HiFi reads were aligned to the reference genome and cleaned FL reads were clustered and classified into structural isoform categories based on the PacBio IsoSeq3 and Pigeon pipelines. We explored the distribution of isoform categories across the cohort and within individual tumour types, specifically focusing on novel isoforms. Functional enrichment analysis of genes with novel isoforms was carried out using g:Profiler.

Results: We identified a total of 24,130 unique genes with FL read support among all libraries, with an average of 13,400 unique genes/library. Isoforms were classified into 9 different categories (e.g., full- or incomplete-splice match, novel isoforms involving known and/or new splice sites, etc.) and further divided into 14 sub-categories (e.g., alternative 3/5end, multi-exon, intron retention). We focused on genes with novel isoforms that met the following conditions: (i) novel isoforms each had a minimum of 5 FL read counts, and (ii) at least 10% of the FL reads in the gene were supported novel isoforms. This strategy shortlisted 6,774 genes with novel isoforms across the whole cohort. Functional enrichment analysis revealed that novel isoforms are primarily overrepresented in genes associated with cell-substrate and cadherin binding, focal adhesion and junction assembly, cytoskeleton reorganization and cell migration. A further distinction was observed within each tumour subtype, including signalling by Rho GTPases and ECM reorganization in UPS and ANGS, and mRNA splicing terms in SYNS. Additionally, novel isoforms were found for genes in currently ongoing clinical trials in sarcomas (i.e., MAGEA4, CD276) in SYNS samples, and driver fusions were detected in both ASPS and SYNS samples.

Conclusion: We used PacBio Iso-Seq data to comprehensively profile isoform diversity in both fusion-driven and genetically complex soft-tissue sarcomas. Genes with novel isoforms were highly enriched in functions related to cell motility and cell-cell/cell-matrix interactions, indicating that this aspect of tumour biology harbours a high capacity for fine-tuning via transcriptional diversity.