INTEGRATED SINGLE-CELL MUTATION, GENE EXPRESSION AND ISOFORM ANALYSIS TO DECONVOLVE ACUTE MYELOID LEUKEMIA HETEROGENEITY

Acute myeloid leukemia (AML) is an aggressive cancer arising from the hematopoietic stem cell (HSC). As other tumor types, AMLs are characterized by multiple and interconnected levels of intra-tumor heterogeneity, including genetic (DNA mutations), phenotypic (transcriptional patterns) and ecological (interactions with host immune-cells) diversity. Emerging evidence suggest that intra-tumor heterogeneity impacts directly on leukemogenesis, disease prognosis and sensitivity/resistance to available treatments. How the different layers of intra-tumoral heterogeneity interact with each other and shape the different leukemia phenotypes at single-cell level, however, is still missing. One major limit is the lack of technologies allowing ecosystem-wide characterization of tumor samples, including the simultaneous multiomic analyses of both malignant and immune populations at single-cell level. In this work, we have developed a novel high-throughput multiomics approach to integrate gene mutation, expression and isoform information at single-cell resolution. SCM-seq (Single Cell and Molecule sequencing) combines high- throughput droplet-based scRNA-seq to Nanopore single-molecule sequencing of full- length whole transcriptome and enriched mutated transcripts. This technology allows the integration at single cell-level of expression profiles and lineage-imputation (from scRNA-seq data) with mutation burden and transcript isoform diversity (from Nanopore data). We have applied this methodology to the analysis of three AML samples sharing a mutation in a spliceosome factor, with the aim to investigate how phenotypic heterogeneity is related to genetic complexity in both the malignant and immune compartments of a coherent AML subgroup. Results showed that SCM-seq allows multiomic characterization at single-cell level with sufficiently high throughput to represent sample complexity. We identified mutations at cell-level with high sensitivity and were able to stratify groups of cells based on their genetic complexity and mutations co-occurrences. For selected variants, we were also able to genotype both mutant and wild-type cells, which is the premise to investigate genotype-phenotype interactions. HSC/progenitor-like AML cells accumulated higher numbers of mutations and shared specific transcriptional features, including leukemia stem cell properties, thus enabling the identification of the putative malignant compartment of the AML samples. We found, however, that mutant cells were also represented in all remaining hematopoietic lineages, including differentiated myeloid cells and lymphocytes, recapitulating the genetic hierarchy observed in HSCs. Increasing genetic complexity in HSC/progenitor-like AML cells was associated to increasing transcriptional heterogeneity and correlated with the expression of genes and signatures related to cell cycle control, proliferation, stress response, RNA splicing regulation, MTORC1 signaling and MYC targets. Moreover, HSC/ progenitor-like AML cells with high mutation burden displayed limited isoform abundance, as related to the number of expressed genes, indicating a progressively restricted repertoire of isoforms in the presence of increasing genetic complexity. In all lineages, the presence of the SRSF2 mutation was associated to increased isoforms diversity, with mutated cells carrying significantly higher proportions of genes expressed with more than one isoform or expressing novel or alternative transcripts, as compared to AML SRSF2-wild-type cells. Together, these preliminary data show the capability of our method to integrate different sources of AML heterogeneity and their relevance within the tumor ecosystem.