Functional Deconvolution of the Leukemia Stem Cell Pool Uncovers Co-Existing Alternative Stem Cell Driven Hierarchies

Leukemia stem cells (LSC) are not functionally homogenous, rather constitute a complex reservoir of multiple subtypes with distinct genetics and functional potential, that fuels disease progression in acute myeloid leukemia (AML). Understanding the intricacies within the LSC pool is critical for informed design of LSC targeting therapies. However, such in-depth studies are hampered by several technical challenges such as the scarcity of LSC and the lack of reliable cell surface markers to isolate viable LSC or distinct types of LSC. To overcome these challenges, we turned to the patient-derived OCI-AML22 model. This model includes functionally, transcriptionally, and epigenetically characterized LSC broadly representative of LSC extracted from primary AML samples. Focusing on the pool of LSC, we deconvoluted the multi-layered heterogeneity, using a panel of in vivo limiting dilution and single cell assays combined with single cell multiome analysis. We uncovered the co-existence of alternative hierarchies driven by distinct LSC subtypes that differ in quiescence depth, differentiation potential, repopulation capacities and that can be prospectively enriched. Notably, we captured the transcriptomic footprint of these newly discovered LSC subtypes within large primary AML cohorts, and even concomitantly within the same AML patient across different LSC fractions, highlighting the broad biological relevance of our findings.