5-azacytidine treatment reshapes clonal structure and alters mutation-dependent splicing in MDS patients Free

Among molecular and cellular hallmarks of aging, changes in genetics, the epigenome and inflammation are now recognized as key mediators of decreased regenerative potential. Such molecular changes compound the impact of Myelodysplastic Syndrome (MDS) mutations (e.g. SRSF2, RUNX1), and underlie selection of mutant clones. To understand the mechanistic basis of MDS, leukemic transformation and therapy resistance we applied a multimodal single cell profiling framework to serial MDS patient bone marrow CD34+ hematopoietic stem and progenitors (HSPCs). This integrated approach simultaneously captured transcriptome profiles and surface protein expression through CITE-seq, while leveraging PacBio long-read sequencing to resolve alternative splicing patterns and perform single-cell genotyping with unprecedented resolution. Our analysis spanned young and aged normal donors to complement MDS at diagnosis, HMA treatment and secondary AML (sAML); mapping mutation specific gene regulation and alternative isoforms throughout aging, MDS progression, and therapeutic response.

Aged donor analysis revealed myeloid skewing without affecting the numbers of the most primitive HSC population, but decline in mature ErP and MkP populations, indicating impaired differentiation efficiency. Transcriptome analysis revealed age-dependent molecular divergence within specific HSPC populations, most notably in primitive HSCs where we identified significant dysregulation of inflammatory and ubiquitin pathway gene expression. These cluster-specific transcriptional changes were coupled with altered splicing patterns affecting critical regulators of cell survival, clonal hematopoiesis, and inflammation.

Comparison of aged normal marrow versus diagnostic MDS biopsies revealed a coordinated gene program in MDS HSPCs characterized by downregulation of inflammatory, glucocorticoid, and cytokine signaling pathways, coupled with widespread splicing dysregulation. While no consensus gene programs were altered by AZA treatment, we observed that AZA therapy consistently depleted HSC numbers while driving differentiation toward MEP and MkP lineages, We note that EMP3 and CD9 are differential markers of erythroid and megakaryocytic response. Single-cell genotyping revealed the double-edged consequences of AZA treatment on clonal dynamics: high-risk RUNX1-W279-nonsense mutations became enriched in HSPCs post-treatment, whereas founding mutations like SRSF2-P95 remained stable throughout disease progression, highlighting mutation-specific therapeutic vulnerabilities and resistance mechanisms.

Clonal analysis revealed that different genetic lesions confer distinct cellular advantages through either lineage reprogramming or survival enhancement. Within the LMPP-2 population, which represents a key cellular reservoir for leukemic transformation, RUNX1-mutated cells exhibited aberrant upregulation of cell activation markers and ectopic expression of B-cell differentiation genes, accompanied by surface overexpression of lymphoid markers including CD24, indicating mutation-driven lineage infidelity. Conversely, PHF6 mutations, acquired during secondary AML relapse following Venetoclax/AZA combination therapy, conferred resistance through a distinct mechanism: downregulation of BCL2 and cell proliferation genes within the MPP-MEP cells, suggesting therapeutic evasion via target elimination and enforced quiescence.

AZA treatment induced genotype-specific splicing alterations, including changes in some bona fide SRSF2-P95 target genes. Importantly, we validated this finding in both an Srsf2-P95 Runx1-/- mouse model and human SRSF2-P95 cell lines. However, the clonal stability of SRSF2-P95 mutations throughout disease evolution, combined with a subset of consistently perturbed splicing targets, presents a novel therapeutic opportunity to selectively target alternatively spliced protein isoforms as a mutation-specific treatment strategy.

Overall, using a new bioinformatics workflow called AltAnalyze-LR, our multimodal single-cell framework reveals the interplay between aging-related hematopoietic remodeling, mutation-specific cellular reprogramming, and treatment-induced clonal selection, identifying new biomarkers and therapeutic targets for MDS management.