Abstract Despite the remarkable success of targeted therapy, the emergence of acquired resistance remains a major obstacle to cure acute myeloid leukemia (AML) patients. While existing studies have predominantly focused on genetic heterogeneity as a driver for therapy resistance, recent evidence suggests that non-genetic persister state is equally important, but such studies are lacking in AML. Here, we define the temporal transcriptome and chromatin landscape along with lineage trajectory of individual genetically identical AML persister clones to existing anti-AML therapeutics. To model persister state in AML, we utilized watermelon library (Oren et al., 2021), that expresses lineage barcode with inducible Histone-2B-mCherry to trace clonal evolution and cell division. We first evaluated whether slow-cycling, quiescent persisters emerge in AML lines after exposure to standard therapy. Our data showed the emergence of slow cycling persisters with cytarabine (AraC) in all 4 cell lines, and gilteritinib (GLT) in MOLM-13 and MV4-11 cells while BH3 mimetics (venetoclax (VEN), S63845 and navitoclax) had fast-cycling persisters. Integrating proliferative-index profiles with cell cycle profiles depicted extensive heterogeneity in AML persisters imposed by drug target rather than genetic state. Drug holiday experiments exhibited a reversible drug tolerance state in both slow and fast-cycling persisters. Longitudinal drug sensitivity and BH3 profiling assays showed that persisters emerged with adaptation to reduced mitochondrial priming that reversed after drug holiday. However, transitioning from persister to stable resistant cells resulted in a sustained loss in apoptotic priming, mirroring observations made in PDX models and primary tumors. To understand trajectory of persister emergence and underlying mechanisms, we performed single-cell multiomics on drug-naïve and persister populations (day 9) in MV4-11 (GLT), and OCI-AML2 (VEN and AraC). We expanded 10,000 unique barcoded cells and administered the drug treatment to derive persisters. Using lineage barcode analysis, we report significant reduction in the dominant lineage of day 0 and survival of multiple small lineages indicating an acquired state of persistence than selection of pre-existing resistant clone in both the lines. Further trajectory analysis defined three fates of the lineages as: expanding (log2FC1), shrinking (log2FC 1) and stable. AraC persisters showed higher expanding and stable lineages, while VEN and GLT persisters resulted in higher shrinking and stable clones, revealing heterogeneity in lineage expansion. We next asked whether persisters resulting from different therapies arised from the same parental lineage. To test this, we performed lineage tracing on VEN and AraC persisters in OCI-AML2 and compared surviving lineages at Day 9 vs Day 0 cells. This analysis identified 52 parental lineages that were able to produce both VEN and AraC persisters in AML2, providing evidence of branched evolution of persisters from the same parental clone. We finally defined transcriptional and epigenetic heterogeneity in baseline and persister cells by performing UMAP on single-cell RNAseq and ATACseq data. We identified persisters to all 3 drugs clusters distinctly from their drug naïve counterparts at both transcriptome and chromatin accessibility. Next, we investigated differentially expressed pathways in persisters and observed higher inflammatory response pathways such as TNFα signaling in GLT, enrichment of MYC targets and oxidative phosphorylation in VEN and high mTORC1 signalling in AraC persisters. We observed higher chromatin accessibility in GLT and VEN whereas condensed chromatin in AraC persisters. We next asked which motifs are highly enriched in each persister and observed the enrichment of various KLF and SP family of motifs in each persister driving multiple gene signatures in each persister type. We also observed enrichment of LSC-primed and cDC like cells in GLT persisters but with AraC and VEN persisters a uniform pattern of all states was observed, indicated stemness may not be common mechanism to persister evolution. Collectively, our integrated single-cell multiomics and lineage tracing approach reveals that AML persisters emerge through stochastic drug-specific, non-genetic adaptive mechanisms characterized by distinct transcriptional and chromatin remodeling programs rather than deterministic Darwinian selection of pre-existing resistant cells.
Yu et al. (Mon,) studied this question.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: