A Mitochondrial-targeted purine-based HSP90 antagonist for leukemia therapy

Abstract

Reprogramming of mitochondrial functions sustains tumor growth and may provide therapeutic opportunities. Here, we targeted the protein folding environment in mitochondria by coupling a purine-based inhibitor of the molecular chaperone Heat Shock Protein-90 (Hsp90), PU-H71 to the mitochondrial-targeting moiety, triphenylphosphonium (TPP). Binding of PU-H71-TPP to ADP-Hsp90, Hsp90 co-chaperone complex or mitochondrial Hsp90 homolog, TRAP1 involved hydrogen bonds, π-π stacking, cation-π contacts and hydrophobic interactions with the surrounding amino acids in the active site. PU-H71-TPP selectively accumulated in mitochondria of tumor cells (17-fold increase in mitochondria/cytosol ratio), whereas unmodified PU-H71 showed minimal mitochondrial localization. Treatment of tumor cells with PU-H71-TPP dissipated mitochondrial membrane potential, inhibited oxidative phosphorylation in sensitive cell types, and reduced ATP production, resulting in apoptosis and tumor cell killing. Unmodified PU-H71 had no effect. Bioinformatics analysis identified a "mitochondrial Hsp90" signature in Acute Myeloid Leukemia (AML), which correlates with worse disease outcome. Accordingly, inhibition of mitochondrial Hsp90s killed primary and cultured AML cells, with minimal effects on normal peripheral blood mononuclear cells. These data demonstrate that directing Hsp90 inhibitors with different chemical scaffolds to mitochondria is feasible and confers improved anticancer activity. A potential "addiction" to mitochondrial Hsp90s may provide a new therapeutic target in AML.

Keywords: Hsp90; acute myeloid leukemia; chaperone; metabolism; mitochondria.

Figure 1. Chemical synthesis of mitochondrial-targeted small molecule Hsp90 inhibitor PU-H71-TPP

The individual synthesis steps and corresponding experimental conditions are indicated. The two final compounds used in this study H71-TPP-1 and H71 TPP-2 differ from the absence or presence of a iodo substituent on the methylenedioxy moiety, respectively.

Figure 2. Predicted model of PU-H71-TPP (H71-TPP-2) target binding

A, B and C. ADP-Hsp90 complex. D, E and F. Co-chaperone-Hsp90 complex. G, H and I. TRAP1. For each condition, the 3D interaction model of the ligand in the binding site showing the interacting amino acids as lines (A, D and G), and the binding mode of the ligand with hydrophobic surface representation colored by YRB highlighting scheme (B, E and H) is shown. C, F and I. 2D interaction diagram. For panels B, E and H, yellow is used for non-polar hydrocarbons; red for negatively charged oxygens of glutamate and aspartate; blue for positively charged nitrogens of lysine and arginine; blue and white for all remaining atoms including the polar backbone.

Figure 3. Ligand binding parameters

A and B. Ligand RSMD of PU-H71-TPP (H71-TPP-2) binding to ADP-Hsp90 or co-chaperone-Hsp90 complex. The average change in displacement of ligand atoms is shown with respect to the initial binding mode during the course of MD simulation. ‘Lig fit Prot’ shows the RMSD of a ligand when the protein-ligand complex is compared to the backbone of the initial binding mode complex. ‘Lig fit Lig’ shows the RMSD of the ligand when compared to its initial conformation. C and D. Ligand RSMF of PU-H71-TPP (H71-TPP-2) predicted binding to ADP-Hsp90 or co-chaperone-Hsp90 complex. Ligand contacts are the protein residues that interact with the ligand and are shown as green vertical bars. E and F. Schematic representation of the protein-ligand interactions that occur >10% of simulation time for binding to ADP-Hsp90 or co-chaperone-Hsp90 complex.

Figure 4. Mitochondria-targeted PU-H71 causes acute organelle dysfunction

A. LC/MS quantification of Gamitrinib or 17-AAG or B. PU-H71 (H71) or H71-TPP-1 (H71-TPP) in isolated subcellular fractions of treated PC3 cells. Cyto, cytosol; Mito, mitochondria; M/C, mitochondria/cytosol ratio. Data are the mean±SD (n=2). C. The indicated tumor cell lines were treated with vehicle (Veh), TPP alone, or increasing concentrations (2.5, 5, 10, 20 μM) of 17-AAG, Gamitrinib or H71-TPP-1 (H71-TPP) and analyzed by Western blotting. D. AML HL-60 cells were incubated with vehicle (Veh) or H71-TPP-1 (H71-TPP) (2.5, 5 μM), and analyzed for mitochondrial membrane potential by JC1 staining and multiparametric flow cytometry. The percentage of cells in each quadrant is indicated. The mitochondrial uncoupler CCCP was used as control. FL, fluorescence. E. The indicated AML cell lines were incubated with increasing concentrations of Gamitrinib (Gam), PU-H71 (H71), PU-H71-TPP-1 or PU-H71-TPP-2 and analyzed for cell viability by an alamarBlue assay. Data are the mean±SD (n=3).

Figure 5. Inhibition of mitochondrial Hsp90 induces metabolic defects

A. The indicated AML cell lines were incubated with vehicle (Veh) or H71-TPP-2 (H71-TPP) (2.5-5 μM) and analyzed for oxygen consumption rates (OCR) after a 360-min incubation. Data are the mean±SD (n=3). ^*, p=0.04; ^***, p<0.0001; ns, not significant. B. The indicated AML cell types were incubated with Gamitrinib (2.5-5 μM) and analyzed for OCR at increasing time intervals. Data are the mean±SD (n=3). C. The indicated AML cell types were incubated with Gamitrinib (Gam, 2.5-5 μM) and analyzed for ATP production or D. lactate generation. Data are the mean±SD (n=3). ^***, p= 0.0006-<0.0001. E. The indicated AML cell types were incubated with increasing concentrations of mitochondrial oxidative phosphorylation Complex II inhibitor, TTFA and analyzed for cell viability by an alamarBlue assay. Data are the mean±SD (n=3).

Figure 6. PU-H71-TPP induces tumor cell death

A. PC3 cells were treated with vehicle (Veh) or the indicated agents (all at 20 μM, H71-TPP-2 (H71-TPP)) and analyzed for Annexin V and propidium iodide (PI) staining by multiparametric flow cytometry. The percentage of cells in each quadrant is indicated. B. THP-1 cells were treated with the indicated agents and analyzed by Western blot for caspase cleavage. C. The indicated leukemia cell lines were treated with increasing concentrations of Gamitrinib (Gam, 2.5, 5, 10 μM) and analyzed by Western blotting. The positions of cleaved PARP or cleaved caspase-3 subunits are indicated. D. The indicated AML cell types were incubated with increasing concentrations of H71-TPP-2 (H71-TPP) and analyzed for cell viability by an alamarBlue assay. Data are the mean±SD (n=3). E. AML cell lines were incubated with increasing concentrations of Gamitrinib and analyzed for cell viability by an alamarBlue assay. Data are the mean±SD (n=3). F. The experimental conditions are as in E. except that the indicated treated cultures were quantified for cell death by direct cell counting after 36 h. Data are the mean±SD (n=4). G. The indicated AML cell lines were incubated with increasing concentrations of Ara-C, alone or in combination with 1 μM or 2 μM Gamitrinib (Gam) and analyzed for cell viability by an alamarBlue assay. Data are the mean±SD (n=2).

Figure 7. Bioinformatics analysis of TRAP1 expression in AML

A. Oncomine analysis of TRAP1 mRNA expression in normal bone marrow (n=6), AML (n=23), B-cell ALL (n=87) or T-cell ALL (n=11) [40]. TRAP1 mRNA expression in normal bone marrow versus AML: 2.39-fold induction, p=7.98×10^-8. B. Identification of 1,322 genes positively correlated with TRAP1 expression and 538 genes negatively correlated with TRAP1 expression (FDR<5%, Spearman |r|>0.3) in AML TCGA mRNA-seq data. C. Kaplan-Meier curve of differential AML survival based on the combined mRNA profile of genes associated with high or low TRAP1 expression (p=0.003, HR=2.77 by Cox regression). D. Genes positively correlated with TRAP1 expression are enriched in mitochondria-related genes (2.2-fold enrichment, p<10^-10 by hypergeometric test), and comprise 70% (52 proteins) of mitochondrial proteins that require TRAP1 for proper folding (4.9-fold enrichment, p<10^-10 by hypergeometric test). E. Combined schematic model of mitochondrial functions and cellular pathways associated with the TRAP1 protein and gene signature in AML. F. Peripheral blood normal mononuclear cells (PBMC) or leukemia cell lines were analyzed by Western blotting.

Figure 8. Inhibition of mitochondrial Hsp90s for AML therapy

A. The indicated patient-derived AML samples were treated with increasing concentrations of Gamitrinib or H71-TPP-2 (H71-TPP) and analyzed for mitochondrial function by an MTT assay. Data are the mean±SD of replicates of a representative experiment. B. The experimental conditions are as in A. except that treated AML samples were analyzed by Annexin V/PI staining and multiparametric flow cytometry. The percentage of viable cells or cells in different apoptotic stages (early, late, dead) was quantified. C. Gamitrinib-treated primary AML samples or PBMC were analyzed for cell death by direct cell counting. Data are the mean±SD of replicates of a representative experiment. D. Normal human PBMC were treated with the indicated concentrations of Gamitrinib and analyzed for Annexin V/PI staining and flow cytometry, with quantification of viable cells or cells in the various apoptotic phases (early, late, dead).