AUTHORS
Haiming Xu, PhD, Wenhao Zhang, PhD2, Lanpeng Chen, PhD3, Lindsey Rosas Togo4, Tony Cheung5, Sungmi Park-Chouinard, PhD6, Lisa Drew, PhD7, Omid Tavana, PhD8
AFFILIATIONS
1 AstraZeneca, Waltham, Massachusetts, United States
2 AstraZeneca, Waltham, Massachusetts, United States
3 Acerta Pharma AstraZeneca, Oss, Netherlands
4 AstraZeneca, Waltham, Massachusetts, United States
5 AstraZeneca, Waltham, Massachusetts, United States
6 AstraZeneca, Waltham, Massachusetts, United States
7 AstraZeneca, Waltham, Massachusetts, United States
8 AstraZeneca, Waltham, Massachusetts, United States
ABSTRACT
Leukemic stem cells (LSCs) are responsible for leukemia initiation, disease propagation, therapeutic resistance, and relapses in acute myeloid leukemia (AML). Functional heterogeneity within the LSC compartment likely drives differential therapeutic responses. Defining the molecular and epigenetic basis of this heterogeneity will facilitate the development of more effective therapies capable of eradicating the LSC populations and reducing the risk of relapse. However, progress has been limited by the lack of models that faithfully retain distinct LSC populations. The patient-derived AML models OCI-AML22 (Boutzen et al 2022) and OCI-AML8227 (Lechman et al., 2016), developed by John Dick's group, retain biologically and clinically relevant LSC properties, providing powerful platforms for mechanistic and functional studies of leukemic stem cell biology in a physiologically relevant cellular context.
To define the molecular and epigenetic heterogeneity of AML LSCs, we utilized these two physiologically relevant human AML LSC models (OCI-AML22 and OCI-AML8227), together with the DFAM-33766 (Olesinski et al., 2024) PDX model, which is enriched for phenotypic CD34⁺CD38⁻ human LSCs. We performed Single-cell Combinatorial Chromatin Accessibility and mRNA Expression Sequencing (SHARE-seq) to simultaneously profile chromatin accessibility and gene expression at single-cell resolution.
AML heterogeneity was captured by both scRNA-seq and differential chromatin accessibility across distinct leukemic cell subpopulations, including LSC-like, granulocyte-monocyte progenitor (GMP)-like, and monocyte-like cells. Additionally, by analyzing the single-cell ATAC-seq data from SHARE-seq, we identified hyperactivity of RUNX and GATA transcription factors across these AML PDX models, supporting their lineage-biased differentiation states. Projection of the scRNA-seq data onto a reference bone marrow lineage differentiation map (Zeng et al., 2025) uncovered that these AML models recapitulate the cellular plasticity, differentiation arrest, and lineage bias observed in primary AML patient samples. Specifically, DFAM-33766 model aligned with a primitive LSC-like state, OCI-AML8227 model displayed a monocytic-biased AML-like phenotype, and OCI-AML22 was enriched for an Erythroid/Megakaryocyte progenitor (MEPs)-biased AML program.
Venetoclax-based therapies have significantly improved clinical outcomes for patients with AML; however, the transcriptional and epigenetic mechanisms governing treatment response and resistance remain poorly understood. Erythroid and megakaryocytic differentiation has been associated with increased BCL-XL dependency and venetoclax resistance (Kuusanmäki et al., 2023). Consistent with these findings, our trajectory analysis suggested that venetoclax-insensitive cells arise from an MEP-biased cellular program enriched in OCI-AML22 model. The OCI-AML22 model also exhibited a venetoclax-insensitive phenotype ex vivo and significantly higher BCL-XL protein expression compared with the other two AML models. Further supporting these findings, preliminary analysis of samples from the Tempus AML cohort revealed enrichment of erythroid/megakaryocytic lineage programs and increased BCL-XL expression in relapsed/refractory (R/R) AML following venetoclax-based therapy.
Collectively, our integrated SHARE-seq analysis provides a comprehensive multiomic atlas of AML leukemic stem cell heterogeneity. These findings demonstrate the power of SHARE-seq to uncover the transcriptional and epigenetic mechanisms underlying therapeutic resistance and identify novel regulatory pathways that may be co-targeted to overcome venetoclax resistance and improve the durability of AML treatment responses.
DISCLOSURES
Haiming Xu, PhD: AstraZeneca - Current equity holder in publicly-traded company; AstraZeneca - Current equity holder in publicly-traded company; AstraZeneca - Stock (publicly traded); AstraZeneca - Stock (publicly traded); AstraZeneca - Current Employment; AstraZeneca - Current Employment