Acute myeloid leukemia mitochondria hydrolyze ATP to resist chemotherapy

Abstract

Despite early optimism, therapeutics targeting oxidative phosphorylation (OxPhos) have faced clinical setbacks, stemming from their inability to distinguish healthy from cancerous mitochondria. Herein, we describe an actionable bioenergetic mechanism unique to cancerous mitochondria inside acute myeloid leukemia (AML) cells. Unlike healthy cells which couple respiration to the synthesis of ATP, AML mitochondria were discovered to support inner membrane polarization by consuming ATP. Because matrix ATP consumption allows cells to survive bioenergetic stress, we hypothesized that AML cells may resist cell death induced by OxPhos damaging chemotherapy by reversing the ATP synthase reaction. In support of this, targeted inhibition of BCL-2 with venetoclax abolished OxPhos flux without impacting mitochondrial membrane potential. In surviving AML cells, sustained polarization of the mitochondrial inner membrane was dependent on matrix ATP consumption. Mitochondrial ATP consumption was further enhanced in AML cells made refractory to venetoclax, consequential to downregulations in both the proton-pumping respiratory complexes, as well as the endogenous F₁-ATPase inhibitor ATP5IF1. In treatment-naive AML, ATP5IF1 knockdown was sufficient to drive venetoclax resistance, while ATP5IF1 overexpression impaired F₁-ATPase activity and heightened sensitivity to venetoclax. Collectively, our data identify matrix ATP consumption as a cancer-cell intrinsic bioenergetic vulnerability actionable in the context of mitochondrial damaging chemotherapy.

Figure 1.. Intrinsic lesions in respiratory complex IV limit OxPhos flux in AML cells.

All experiments were performed using whole cells and digitonin-permeabilized cells (A) Comparison of fractional OXPHOS of normal blood cells, patient AML, AML cell lines, chemoresistant AML, and mouse AML calculated as the ratio of JH⁺_OXPHOS to JH⁺_Total (n = 3–30 replicates) and represented as a percentage of total respiratory capacity. SkM Progenitors refers to myoblasts isolated from human skeletal muscle biopsy. (B) Volcano plot comparing mitochondrial and non-mitochondrial proteome of AML cells and healthy bone marrow mononuclear cells (Healthy BMC) (C) Comparison of mitochondrial content in AML cells and Healthy BMC cells (n = 3–30 replicates). (D) Comparison of whole cell respiratory capacity in Healthy BMC, MV4–11 and OCI-AML2 cells normalized to milligrams of protein (n = 3–5 replicates). (E) Comparison of the OXPHOS proteome between Healthy BMC and AML cells (n = 3 replicates). (F) Comparison of whole cell respiratory capacity in Healthy BMC, MV4–11 and OCI-AML2 cells normalized to milligrams of mitochondrial protein (n = 3–5 replicates). (G) Schematic depicting complex IV lesions and reduced respiration of individual AML mitochondria. Data are presented as mean ±SEM and analyzed by two-way ANOVA (E,F) and one-way ANOVA (A,C,D). *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001.

Figure 2.. Mitochondria in AML cells consume cellular ATP to sustain mitochondrial polarization.

All experiments were performed using whole intact cells. (A) Schematic depicting strategy to determine functional OxPhos in ΔΨ_m assays. (B) Representative image from flow cytometric analysis of intact cell ΔΨ_m in OCI_Rho0 cells. (C) Flow cytometric analysis of intact cell ΔΨ_m in OCI_Rho0 cells (n = 4 replicates). (D) Flow cytometric analysis of intact cell ΔΨ_m in Healthy BMC (n = 3 replicates). (E) Flow cytometric analysis of intact cell ΔΨ_m in myeloid and lymphoid populations sorted from PBMC (n = 4–6 replicates). (F) Flow cytometric analysis of intact cell ΔΨ_m in AML cell lines (n = 3–4 replicates). (G). Flow cytometric analysis of intact cell ΔΨ_m in patient AML (n = 3 replicates). (H) Flow cytometric analysis of intact cell ΔΨ_m in pooled MV411_WT, HL60_WT, and OCI_WT cells (n = 1 replicate per cell type). (I) Representative trace of permeabilized cell ΔΨ_m assay in OCI_WT cells. (J) Comparison of oligomycin-induced depolarization in OCI_WT cells in the presence of increasing doses of rotenone (n = 4 replicates). Data are presented as mean ±SEM and analyzed by two-way ANOVA (F), one-way ANOVA (C,D,G,H,J), or unpaired t-test (E). *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001.

Figure 3.. F₁-ATPase activity is an actionable vulnerability of AML cells.

All experiments were performed using whole intact cells, digitonin-permeabilized cells, or alamethicin-permeabilized mitochondria. (A) Effect of ΔG_ATP on maximal NADH (Complex I) or succinate/rotenone (Complex II) supported respiration in alamethicin-permeabilized MV411_WT mitochondria (n = 4 replicates). (B) Effect of oligomycin on inhibition of uncoupled respiration by ΔG_ATP in permeabilized OCI_WT cells (n = 3 replicates). (C) Effect of oligomycin on uncoupled respiration in intact OCI_WT cells (n = 3 replicates). (D) Change in TMRM fluorescence of individual mitochondria in whole mitochondrial populations isolated from mouse heart or HL60 before and after oligomycin addition during flow cytometric analysis of the ΔΨ_m. (E) Effect of ΔG_ATP on polarization by P/M/G/S/O in the presence of oligomycin, and in the presence or absence of carboxyatractyloside, in permeabilized OCI_WT cells (n = 3 replicates). (F) Fluorescent microscopy analysis of intact cell ΔΨ_m in MOLM-13_WT cells in the presence bongkrekic acid, oligomycin, or those in combination (n = 189–417 cells). (G) Schematic depicting an alternating current model of AML mitochondrial bioenergetics. (H) Schematic depicting the mitochondrial targets of IACS, antimycin A, oligomycin, and BAM15 uncoupler. (I) Effect of increasing doses of BAM15 uncoupler on OCI_WT, MV411_WT, and HL60_WT cell viability. Viability was measured using propidium iodide (n = 4–5 replicates). (J) Effect of IACS, antimycin A, or oligomycin on OCI_WT cell viability. Viability was measured using propidium iodide (n = 5 replicates). (K) Effect of IACS, antimycin A, or oligomycin on OCI_WT, MV411_WT, and HL60_WT colony formation. Colonies were manually counted using microscopy (n = 5 replicates). (L) Schematic depicting mechanism of AML cell death by oligomycin. Data are presented as mean ±SEM and analyzed by two-way ANOVA (A,C,E,F) or one-way ANOVA (A,I,J,K). *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001.

Figure 4.. Venetoclax exposure collapses mitochondrial respiration, but polarization is sustained via F₁-ATPase activity.

All experiments were performed using whole intact cells or digitonin permeabilized cells. (A) Schematic depicting characterization of treatment naïve AML cells exposed to venetoclax for 1 hour. (B) Effect of 1 hour exposure to venetoclax on AML cell viability. Viability was measured using trypan blue viable cell count (n = 9 replicates). (C) Effect of 1 hour exposure to 1μM venetoclax on respiratory capacity in permeabilized AML cells (n = 3 replicates). (D) Effect of 1 hour exposure to 100nM venetoclax on OxPhos capacity in permeabilized AML cells (n = 4 replicates). (E) Effect of 1 hour exposure to 100nM venetoclax on polarization induced by Complex I substrates (P/M/G), Complex II substrates (S/R), or Complex V substrates (ATP) in permeabilized AML cells (n = 4–6 replicates). (F) Flow cytometric analysis of intact MOLM-13_WT cell ΔΨ_m in response to single agent venetoclax, and venetoclax in combination with IACS, antimycin A, and those in combination with oligomycin (n = 3 replicates). (G) Fluorescent microscopy analysis of intact MV411_WT cell ΔΨ_m in response to single agent venetoclax, and venetoclax in combination with rotenone, antimycin A, and those in combination with oligomycin (n = 100–311 cells). (H) Effect of single agent venetoclax, venetoclax in combination with IACS and oligomycin, or IACS in combination with oligomycin on HL60_WT cell viability. Viability was measured using CellTiter-Glo (n = 4 replicates). (I) Effect of single agent venetoclax or venetoclax in combination with either oligomycin or IACS on MOLM-13_WT cell viability. Viability was measured using CellTiter-Glo (n = 4–8 replicates). (J) Schematic depicting mechanism of AML cell death in response to venetoclax and oligomycin. Data are presented as mean ±SEM and analyzed by two-way ANOVA (C,D,E,G,H), one-way ANOVA (B,F), or unpaired t-test (I). Cytochrome C effect in (C) analyzed by unpaired t-test. *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001.

Figure 5.. Chemoresistant AML mitochondria present with a phenotypic shift towards enhanced ATP consumption.

All experiments were performed using whole intact cells, permeabilized cells, or isolated mitochondria. (A) Schematic depicting generation of venetoclax resistant AML cell lines. (B) Effect of venetoclax on treatment naïve AML and chemoresistant AML cell viability. Viability was measured using propidium iodide (n = 3–6 replicates). (C) Comparison of OxPhos proteome in permeabilized treatment naïve AML cells and venetoclax resistant AML cells (n = 3 replicates). (D) Comparison of ATP5IF1 expression in mitochondria isolated from MV411_WT and MV411_Vclax cells (n = 3 replicates). (E) Comparison of respiratory capacity in permeabilized treatment naïve AML cells and venetoclax resistant AML cells (n = 3–4 replicates). (F) Comparison of OxPhos capacity in permeabilized treatment naïve AML cells and venetoclax resistant AML cells (n = 3–4 replicates). (G) Comparison of oligomycin-induced depolarization in permeabilized treatment naïve AML cells relative to permeabilized venetoclax resistant AML cells (n = 3–6 replicates). (H) Flow cytometric analysis of intact cell ΔΨ_m in venetoclax resistant AML cells (n = 4 replicates). (I) Confocal microscopy images and analysis of MV411_Vclax cells in the presence or absence of oligomycin. (J) Comparison of OxPhos capacity in treatment naïve MV411_WT cells and MV411_Vclax cells after venetoclax has been removed from culture media for 24 hours (n = 4 replicates). (K) Effect of IACS and venetoclax, or oligomycin in the presence or absence of venetoclax on MOLM-13_Vclax cell viability. Viability was measured using CellTiter-Glo (n = 4–8 replicates). Data are presented as mean ±SEM and analyzed by two-way ANOVA (B,I), one-way ANOVA (E,F,G), or unpaired t-test (C,D,H,J,K). *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001.

Figure 6.. Knockdown of ATP5IF1 confers chemoresistance

All experiments were performed using whole intact cells, digitonin-permeabilized cells, or isolated mitochondria. (A) Expression of ATP5IF1 in isolated mitochondria derived from scrambled shRNA control AML cells or ATP5IF1 knockdown AML cells (n = 3 replicates). (B) Impact of ATP5IF1 knockdown on the inhibition of respiration by ΔG_ATP in permeabilized scrambled shRNA control AML cells and permeabilized ATP5IF1 knockdown AML cells (n = 4–8 replicates). (C) Impact of ATP5IF1 knockdown on intact cell respiration of scrambled shRNA control AML cells and ATP5IF1 knockdown AML cells (n = 8–15 replicates). (D) Impact of ATP5IF1 knockdown on respiratory capacity in permeabilized scrambled shRNA control AML cells and permeabilized ATP5IF1 knockdown AML cells (n = 8–13 replicates). (E) Impact of ATP5IF1 knockdown on OxPhos capacity in permeabilized scrambled shRNA control AML cells and permeabilized ATP5IF1 knockdown AML cells (n = 8–14 replicates). (F) Impact of ATP5IF1 knockdown on Fractional OxPhos in permeabilized scrambled shRNA control AML cells and permeabilized ATP5IF1 knockdown AML cells (n = 8–14 replicates). (G) Flow cytometric analysis of intact cell ΔΨ_m in MV411_shCtrl and MV411_shIF1 cells (n = 3 replicates). (H) Effect of increasing doses of venetoclax on viability of scrambled shRNA control AML cells and ATP5IF1 knockdown AML cells after 48 hours, or the effect of venetoclax on MV411_shCtrl and MV411_shIF1 viability after 7 days of exposure. Viability measured using trypan blue viable cell count (n = 4 replicates). (I) Effect of venetoclax on colony formation of MV411_shCtrl cells and MV411_shIF1 cells. Colony formation was quantified using CellTiter-Glo (n = 3 replicates). (J) Representative image of colony formation of MV411_shCtrl cells and MV411_shIF1 cells in the presence or absence of venetoclax. Data are presented as mean ±SEM and analyzed by two-way ANOVA (B,C,D,G,H,I), one-way ANOVA (A,E), or unpaired t-test (F,H). *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001.

Figure 7.. Overexpression of ATP5IF1 enhances chemosensitivity

All experiments were performed using whole intact cells, digitonin-permeabilized cells, or isolated mitochondria. (A) Expression of ATP5IF1 in isolated mitochondria derived from lentivirus control AML cells and ATP5IF1 overexpressing AML cells (n = 3 replicates). (B) Expression of ATP5IF1 in isolated mitochondria derived from lentivirus control HCT116 cells and ATP5IF1 overexpressing HCT116 cells (n = 3 replicates). (C) Overexpression of ATP5IF1 in isolated mitochondria derived from AML and HCT116 cells infected with ATP5IF1 overexpression lentivirus represented as fold change from isolated mitochondria derived from lentivirus control AML and HCT116 cells (n = 3 replicates). (D) Impact of ATP5IF1 overexpression on the inhibition of respiration by ΔG_ATP in permeabilized lentivirus control AML cells and permeabilized ATP5IF1 overexpressing AML cells (n = 3–13 replicates). (E) Impact of ATP5IF1 overexpression on intact cell respiration in OCI_Ctrl cells and OCI_IF1 cells (n = 5 replicates). (F) Impact of ATP5IF1 overexpression on respiratory capacity in permeabilized OCI_Ctrl cells and permeabilized OCI_IF1 cells (n = 4 replicates). (G) Impact of ATP5IF1 overexpression on OxPhos capacity in permeabilized OCI_Ctrl cells and permeabilized OCI_IF1 cells. (H) Impact of ATP5IF1 overexpression on fractional OxPhos in permeabilized OCI_Ctrl cells and permeabilized OCI_IF1 cells (n = 4 replicates). (I) Effect of venetoclax on viability of OCI_Ctrl cells and OCI_IF1 cells. Viability measured using trypan blue viable cell count (n = 5 replicates). (J) Effect of venetoclax on colony formation of OCI_Ctrl cells and OCI_IF1 cells. Colony formation measured using CellTiter-Glo (n = 7–16 replicates). (K) Representative image of colony formation of OCI_Ctrl cells and OCI_IF1 cells in the presence or absence of venetoclax. (L) Schematic depicting mechanism of death of ATP5IF1 overexpressing AML cells in response to venetoclax. Data are presented as mean ±SEM and analyzed by two-way ANOVA (D,E,F,I), one-way ANOVA (C), or unpaired t-test (A,B,G,H,K) (. *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001.

Figure 8.. Dephosphorylative Oxidation or, “DephOx”, model of AML mitochondrial bioenergetics.

Schematic depicting the Dephosphorylative Oxidation or, “DephOx” model of AML mitochondrial bioenergetics.