Clonal heterogeneity of acute myeloid leukemia under nutrient and chemotherapy stress Free

Background and Objective: Acute myeloid leukemia (AML) is a genetically heterogeneous disease, but even isogenic leukemic populations exhibit phenotypic variability. To determine how non-genetic, biological heterogeneity influences growth, therapy response, and adaptation to metabolic stress, we generated and deeply characterized isogenic human and mouse AML clones. By longitudinally tracking single-cell clones, we found that metabolic plasticity, not newly acquired mutations or changes in transcriptional programs, dictates cell growth and therapeutic response to venetoclax. These data highlight the central importance of biological heterogeneity in AML that is not dictated by genetic changes.

Methods: We isolated 29 single-cell clones from the MLL-rearranged AML cell line MOLM13 and assessed longitudinal differences in baseline growth, adaptation to nutrient stress, and drug response. Clones were tested under glutamine or serum restriction and treated with cytarabine, daunorubicin, venetoclax, and azacitidine. Mitochondrial respiration was measured using Seahorse assays. To explore the genetic basis for phenotypic divergence, we performed RNA-seq and whole-exome sequencing on the most venetoclax-sensitive and -resistant clones. We validated findings in a primary mouse MLL-AF9 AML using both a barcoding approach to tracking clonal dynamics by NGS and with individually isolated and cultured clones.

Results: Even without exogenous stress, baseline growth rates varied >3-fold among MOLM13 clones and remained stable over time. Nutrient stress further amplified heterogeneity: glutamine deprivation suppressed growth by 40–70% across clones, whereas 0.1% FBS had a more uniform suppressive effect. Glutamine restriction reduced basal oxygen consumption rates (OCR) in all clones but preserved maximal respiration after FCCP uncoupling, indicating metabolic adaptation via alternative substrates. Faster-growing clones displayed higher basal OCR and relative resistance to venetoclax, while slower-growing clones were more venetoclax-sensitive. Cytarabine responses showed an opposite trend, suggesting that drug sensitivity is growth rate-dependent. The IC₅₀ of venetoclax varied by ~12-fold, compared to ~2-fold for daunorubicin, indicating a context-dependent influence of clonal heterogeneity on therapy response. Next, we compared the genetic and transcriptional profile of clones that displayed the most sensitivity and resistance to venetoclax. RNA-seq showed no differential expression of BCL2-family genes between sensitive and resistant clones, and whole-exome sequencing revealed hundreds of unique, non-overlapping non-synonymous variants per clone, but no recurrent driver mutations were linked to the phenotype. These results suggest non-genetic mechanisms underlie observed phenotypic diversity. Finally, we extended our findings using the primary mouse MLL-AF9 AML. Using a lentiviral barcoding approach, we again observed interclonal heterogeneity and changes in clonal representation under glutamine restriction and venetoclax exposure. We then isolated and grew individual MLL-AF9 clones, following a similar approach to our work with the MOLM13 cells, and observed interclonal variability in OCR and drug sensitivity, confirming our findings in an independent, primary AML.

Conclusions: Our study reveals substantial biological heterogeneity among isogenic AML clones, with stable differences in growth, metabolic plasticity, and drug response. These phenotypes were not attributable to known driver mutations, which indicate that non-genetic factors may be involved in clonal adaptation. Clonal traits were context-dependent: venetoclax and glutamine restriction revealed the most heterogeneity, suggesting that selective pressures shape phenotypic diversity. These findings highlight the need to consider isogenic clonal variability when interpreting experimental results or designing therapies and highlight that functional heterogeneity may be an underappreciated driver of therapeutic response in AML.