LC21 Agenda

Human papillomavirus (HPV) is a necessary but insufficient driver of cervical cancer. We used nanopore sequencing to generate genome, methylome, and transcriptome data for cervical cancers from Uganda (HTMCP) and the USA (TCGA). HPV integration events underlie complex intra- and inter-chromosomal structural variants that sometimes resulted in highly expressed gene fusions. Haplotype-resolved DNA methylation showed human enhancer dysregulation affecting the HPV-containing allele and allowed us to determine the methylation status of individual viral integration events. Long-read profiling will thus reveal new cancer features that were previously cryptic, and such features will clearly advance cervical cancer biology and perhaps cancer genomic medicine.

Mutations in RNA splicing factors are commonly observed in cancer. For example, splicing factor 3B subunit 1 (SF3B1) mutations are the most common genetic alterations in myelodysplastic syndrome (MDS) patients and are also associated with a high-risk of leukemic transformation in clonal hematopoiesis (CH) individuals. SF3B1 mutations are associated with aberrant RNA splicing, resulting in increased cryptic 3’ splice site usage, and often lead to refractory anemia with ringed sideroblasts clinical phenotype. However, the full impact of SF3B1 mutations on cellular fitness leading to clonal outgrowth in humans remains largely unknown. The study of SF3B1-mediated splicing aberrations has been hampered by the inability to distinguish mutated and wildtype cells in human MDS and CH samples on the one hand, and the limitation of short-read single-cell RNA sequencing to adequately cover splice junctions on the other hand. To overcome these limitations, we developed GoT-Splice by integrating Genotyping of Transcriptomes (GoT) with long-read full-length cDNA profiles at single-cell resolution, using Oxford Nanopore sequencing with the PromethION. This allowed for the simultaneous profiling of gene expression, somatic mutation status, and alternative splicing in single cells. Furthermore, we significantly improved read recovery previously reported with Oxford Nanopore flow cells. Using a biotin enrichment protocol, GoT-Splice allowed for the selective amplification of full-length reads with the accurate structure, thus maximizing reads that can be used for confident alternative splice junction calling at the single-cell level. We applied GoT-Splice to CD34+ cells from MDS patients with SF3B1mut to study how SF3B1 mutations corrupt the complex process of human haematopoiesis. High-resolution mapping of malignant vs. normal hematopoietic progenitors revealed an increasing fitness advantage of SF3B1mut cells with myeloid differentiation, resulting in a build-up of SF3B1mut cells in the erythroid progenitor state. GoT-Splice uniquely allowed us to jointly capture SF3B1 mutations and genome-wide splicing information to chart the variation of splicing alterations in SF3B1mut cells within individual hematopoietic cell types. We not only showed increased detection of cryptic 3’ splice signal over previous bulk RNA-sequencing methods but also revealed that MDS patients exhibit distinct cryptic 3’ splice site usage in SF3B1mut cells as a function of hematopoietic progenitor cell identity. Specifically, we observed the largest increase in cryptic 3’ splice site usage in the erythroid progenitors, consistent with the ineffective erythropoiesis phenotype associated with this mutation. Similar to MDS, the application of GoT-splice to CD34+ progenitor cells from individuals with SF3B1mut CH revealed increased mutant cell frequency in EPs and cell-type specific cryptic 3’ splice site usage in SF3B1mut cells. In summary, we developed a novel multi-omics single-cell toolkit to examine the impact of somatic splicing factor mutations on cellular fitness and its molecular determinants from the earliest phases of clonal blood disorders (CH) to overt MDS, providing a widely applicable methodology and related analytics to study the effect of somatic mutations on isoform-level gene expression patterns in single cells directly in human samples.