Inhibition of Amino Acid Metabolism Selectively Targets Human Leukemia Stem Cells

Outcomes for AML patients remain poor because of the inability to fully eliminate leukemia stem cells (LSCs). We have previously shown that primary human LSCs reside in a unique metabolic condition characterized by a relatively low oxidative state (termed "ROS-low") and increased levels of glutathione (Lagadinou et al. Cell Stem Cell, 2013). Cells in this condition are highly dependent on oxidative phosphorylation (OXPHOS) for survival, in striking contrast to many tumor cells which often rely on glycolysis, suggesting that LSCs are governed by distinct metabolic properties. Therefore, the goal of this project was to identify and target metabolic vulnerabilities of LSCs.

To achieve this objective, we used mass spectroscopy to interrogate the metabolome of leukemia stem cells (LSCs) isolated from primary human AML specimens. We observed significant increases in the levels, uptake, and metabolism of amino acids in LSCs compared to bulk AML cells. These data suggest that LSCs may preferentially rely on amino acids for survival. To investigate this hypothesis, we cultured LSCs and bulk leukemia cells isolated from primary leukemia specimens in media lacking amino acids and measured cell viability and colony forming potential. We found that LSCs were uniquely sensitive to amino acid loss. In addition, LSCs formed significantly fewer colonies upon amino acid depletion compared to LSCs cultured in media containing amino acids. To confirm that amino acid depletion was targeting functionally-defined LSCs, we employed engraftment assays in immune incompetent mice. Culturing primary AML cells without amino acids for 24 hours resulted in significantly reduced levels of leukemia cell engraftment. Next, we interrogated whether amino acid depletion impaired normal HSC survival and function by culturing mobilized peripheral blood without amino acids and measuring the frequency of CD34+ cells, colony forming ability, and engraftment into immune deficient mice. HSC frequency, colony forming ability, and engraftment potential were not changed by amino acid depletion. Altogether, these data demonstrate the LSCs are selectively dependent on amino acids for survival.

We next determined how amino acids modulate LSC biology by measuring the consequences of amino acid loss on LSC metabolism. We observed that amino acid depletion decreased OXPHOS specifically in LSCs and not in bulk leukemia cells. We have previously shown that BCL-2 inhibition decreases OXPHOS in LSCs (Lagadinou et al. Cell Stem Cell, 2013). Importantly, recent studies have shown that inhibition of BCL-2 using the BCL-2 inhibitor venetoclax in combination with azacitidine has resulted in superior outcomes for AML patients (Dinardo et al. Lancet Oncology, 2018). Furthermore, our preliminary data demonstrates that venetoclax with azacitidine targets LSCs in AML patients. Therefore, we hypothesized that venetoclax with azacitidine may be targeting LSCs by modulating OXPHOS via amino acid metabolism.

To test this hypothesis, we isolated LSCs from AML patients undergoing treatment with venetoclax and azacitidine. LSC specimens obtained pre and 24 hours after initiation of therapy were analyzed for changes in OXPHOS, gene expression, and the metabolome. We observed that venetoclax with azacitidine treatment decreased OXPHOS and reduced amino acid levels. In addition, expression of amino acid transporters was down-regulated. Finally, we sought to determine if culturing LSCs in high levels of amino acids before venetoclax and azacitidine treatment could rescue LSC viability and OXHPOS. We found that culturing LSCs with increased levels of amino acids rescued LSCs survival and OXPHOS, demonstrating that venetoclax with azacitidine targets LSCs by decreasing amino acid levels.

Taken together, our data indicate that LSCs are selectively reliant on amino acid metabolism to fuel OXPHOS. Furthermore, amino acid metabolism can be targeted in AML patients by venetoclax with azacitidine treatment. These studies are the first to characterize metabolic targeting of LSCs in AML patients.