Adding Splicing to the Cake: Enhancing Single-cell Multi-omics With Detection of Splicing Changes

RNA splicing is the process by which the non-coding regions of RNA are removed and the coding regions are sewn back together. Alternative splicing results in the production of different mRNA transcripts from the same mRNA precursor, enabling protein diversity among cell types and tissues and the fine-tuning of cell function via the generation of alternative isoforms under specific physiological conditions.

Splicing factors are often mutated in hematologic cancers, especially myelodysplastic syndromes (MDS), chronic lymphocytic leukemia (CLL), and myeloproliferative neoplasms (MPNs), with SF3B1 being one of the most commonly affected genes.¹  SF3B1-mutated MDS is recognized as a distinct disease entity, characterized by the presence of ring sideroblasts, ineffective erythropoiesis, and a relatively indolent clinical course.²  SF3B1-mutated MDS is particularly responsive to luspatercept — a transforming growth factor beta (TGFβ) superfamily ligand trap that abolishes transfusion requirements in up to 40% of patients.³  This is thought to be due to a reduction in TGFβ/SMAD family member 2/3 (SMAD2/3) signaling that enables cells to overcome a block in late erythroid maturation. However, why changes in SF3B1 cause ineffective erythropoiesis in the first place has been poorly understood.

A major challenge in unraveling the impact of splicing factor mutations on cell fate and phenotype has been our inability to examine spliceosome changes at the single-cell level while simultaneously genotyping to distinguish between wild-type and mutant cells. In myeloid blood cancers, wild-type and mutant cells are typically indistinguishable, and stem cells and progenitors are difficult to disentangle even using rigorous immunophenotyping definitions. Therefore, it has not been possible to distinguish the specific changes caused by splicing factor mutations from differentiation state-specific changes in gene expression. Several years ago, Anna S. Nam, MD, Dan A. Landau, MD, PhD, and colleagues developed the Genotyping of Transcriptomes (GoT) method, integrating targeted single-cell genotyping with high-throughput single-cell RNA-sequencing (scRNAseq) and allowing them to identify the transcriptional impact of oncogenic perturbations.^4,5  Like the majority of scRNAseq methods, GoT is based on short-read sequencing and therefore does not sensitively detect splicing alterations. In their recent study,⁶  the Landau lab added splicing analysis to the platform to also include long-read transcriptomics with proteogenomics and developed a novel computational pipeline, allowing measurement of the cell type-specific impact of splicing factor mutations for the first time.

Their study, which primarily focused on SF3B1-mutant MDS, applied GoT-Splice to analyze cells from six patients with MDS who had SF3B1 K700E mutations. Genotyping was possible for 64% of the approximately 24,000 cells analyzed, revealing that wild-type and mutant cells were mixed throughout the hematopoietic hierarchy and not distinguishable based on transcriptomic information alone. The authors found that the SF3B1 mutations arose in hematopoietic stem cells (HSCs) but that mutant cells were significantly enriched in erythroid progenitors.

More than half of the mis-splicing events in mutant cells involved alternative 3’ splice sites in the SF3B1^mut+ MDS samples. In contrast, GoT-Splice applied to a CLL dataset revealed exon skipping as the most common event. A high frequency of exon skipping was also observed in mutant cells from AML samples with U2AF1 mutations, demonstrating that the GoT-Splice platform can be applied in a range of contexts.

Examining progenitor-specific patterns of mis-splicing showed that the majority of the cryptic 3’ splice events in the MDS samples occurred in erythroid progenitors and involved genes involved in the cell cycle, oxygen homeostasis, and erythroid differentiation. Aberrant splicing increased expression of some genes, while an inverse correlation with gene expression was observed in other cases, in which mis-splicing induced nonsense-mediated decay (NMD) of aberrant transcripts. NMD-inducing events affected genes key for erythroid development, disrupting erythroid differentiation. Compounding the phenotype, altered splicing led to overproduction of an isoform of BAX, which encodes a protein known to protect cells from apoptosis. The authors went on to functionally validate the ability of this BAX isoform to protect cells from apoptosis using a genetically engineered erythroleukemia cell line. Collectively, these data showed that the SF3B1 mutation led to downregulation of genes required for erythroid maturation while increasing resistance to apoptosis.

The presence of SF3B1 in precursor neoplasms (clonal hematopoiesis [CH] and clonal cytopenias of uncertain significance [CCUS]) is invariably associated with progression to overt hematological neoplasms.¹  Mariela Cortés-Lopéz, PhD, and colleagues next applied GoT-Splice to samples from adult donors with normal blood counts and SF3B1 CH. They found that, even prior to the onset of overt malignancy, SF3B1-mutant cells were enriched in erythroid progenitors and that the mutation induced a bias towards erythroid differentiation, with increased usage of cryptic 3’ splice sites and similar dynamics of BAX isoforms.