Inhibition of translation initiation factor eIF4a inactivates heat shock factor 1 (HSF1) and exerts anti-leukemia activity in AML

Abstract

Eukaryotic initiation factor 4A (eIF4A), the enzymatic core of the eIF4F complex essential for translation initiation, plays a key role in the oncogenic reprogramming of protein synthesis, and thus is a putative therapeutic target in cancer. As important component of its anticancer activity, inhibition of translation initiation can alleviate oncogenic activation of HSF1, a stress-inducible transcription factor that enables cancer cell growth and survival. Here, we show that primary acute myeloid leukemia (AML) cells exhibit the highest transcript levels of eIF4A1 compared to other cancer types. eIF4A inhibition by the potent and specific compound rohinitib (RHT) inactivated HSF1 in these cells, and exerted pronounced in vitro and in vivo anti-leukemia effects against progenitor and leukemia-initiating cells, especially those with FLT3-internal tandem duplication (ITD). In addition to its own anti-leukemic activity, genetic knockdown of HSF1 also sensitized FLT3-mutant AML cells to clinical FLT3 inhibitors, and this synergy was conserved in FLT3 double-mutant cells carrying both ITD and tyrosine kinase domain mutations. Consistently, the combination of RHT and FLT3 inhibitors was highly synergistic in primary FLT3-mutated AML cells. Our results provide a novel therapeutic rationale for co-targeting eIF4A and FLT3 to address the clinical challenge of treating FLT3-mutant AML.

Fig. 1. High level expression of eIF4A are observed in human primary leukemias.

A Heatmap of gene expressions of eIF4A1, eIF4A2, eIF4G1, eIF4G2, eIF4G3, and eIF4E in human primary cancer specimens including leukemia in a publicly available database [34]. 1. Bladder, 2. Brain, 3. Breast, 4. Colorectal, 5. Kidney, 6. Lung, 7. Lymphoma, 8. Melanoma, 9. Ovarian, 10. Pancreatic, 11. Prostate. B Gene expression of eIF4A in various cancer types. AML acute myeloid leukemia, ALL acute lymphoblastic leukemia. C Enrichment in the gene ontology (GO) of translation initiation detected in GSEA using a publicly available microarray dataset re-analyzed as 46 intra-patient pairs of AML LICs vs non-LICs [6]. (Normalized enrichment score 2.74, P-value = 0.000, FDR q = 0.000). D Western blot for eIF4A in cord blood samples from healthy donors and in primary AML samples (left). Densitometry was performed using β-actin as a loading control. Fold-changes of eIF4A protein levels in cord blood (CB) and primary AML (ptAML) samples (right) (mean ± SEM, 1.6 ± 0.3 vs. 6.4 ± 1.0, P = 0.007).

Fig. 2. RHT inhibits growth and survival of AML cells especially cells with FLT3-ITD.

A Chemical structure of RHT. B Drug-specific apoptosis in AML cell lines treated with RHT. C Comparison of ED50 in FLT3 wild-type or FLT3-ITD-positive AML cell lines based on the results obtained in B. Mean ± SEM for ED50, 7.8 ± 1.9 vs. 61.1 ± 9.9 for FLT3-ITD and FLT3 wild-type cells respectively, P = 0.017. D Drug-specific apoptosis in Ba/F3 or OCI-AML3 cells overexpressing wild-type FLT3 or FLT3-ITD mutant protein treated with RHT. In B and D, cells were treated with RHT at indicated concentrations for 72 h, and annexin V-positive cells were counted by flow cytometry. Data represent mean ± SD. E Percent live cells (negative for annexin V or DAPI) of primary AML cells and normal BM mononuclear cells derived from heathy donors after 72 h treatment with 25 nM RHT (left). The breakdown of FLT3 mutational status in the same set of primary AML cells in the left panel (right). Data represent mean ± SD. Student t-test, **P < 0.01, ***P < 0.001, ****P < 0.0001.

Fig. 3. eIF4A inactivation exerts in vivo anti-leukemia effects in FLT3-ITD AML.

A A schematic figure of in vivo treatment of RHT in a systemic AML xenograft model using MOLM13 cells. B Luciferase activity detected by BLI on day 3 and day 8 after transplantation in each treatment group. Mean ± SD of luminous intensity is shown in the right graph. ****P < 0.0001 (C) Percentages of circulating human CD45-positive cells in PB (n = 10) and in BM (n = 3) on day 16 after transplantation. Data represent mean ± SD. (One-way ANOVA with multiple comparison: PB, %mean in untreated vs. 0.75 mg/kg groups; 1.08% vs. 0.10%, P < 0.0001, %mean in untreated vs. 1.0 mg/kg; 1.08% vs. 0.10%, P < 0.0001. BM, %mean in untreated vs. 0.75 mg/kg groups; 45.49% vs. 14.35%, P = 0.00053, %mean in untreated vs. 1.0 mg/kg; 45.49% vs. 4.12%, P = 0.00012.) D Survival curves of the mice harboring leukemia in each treatment group. Median survival of 18, 22.5, and 24 d, in the control, RHT group A, and B, respectively. Log-rank test: P < 0.0001.

Fig. 4. eIF4A inhibition diminishes engraftment potential of AML-initiating cells.

A Live cell number of CD34+CD38− primary AML cells (n = 18) or CD34+CD38− normal BM mononuclear cells (n = 8) after treatment with RHT (50 nM, for 72 h). Data represent mean ± SD. Mean: 17.06% in normal BM vs. 71.98% in primary AML. Mann–Whitney test: P < 0.0001. B A schematic figure of ex vivo treatment of the PDX AML cells (24 h treatment with 20 nM RHT) and in vivo readout in the secondary transplanted PDX AML mice. C Percentage of circulating human CD45+ cells in PB at 4 weeks after transplantation. Mean: 3.33% in vehicle group (n = 9) and 0.265% in RHT-treated group (n = 10). Data represent mean ± SD. Mann–Whitney test: P < 0.0001. D Percentage of human CD45+ cells in BM at 4 weeks after transplantation. Mean: 3.33% in vehicle group (n = 9) and 0.265% in RHT-treated group (n = 10). Data represent mean ± SD. Mann–Whitney test: P < 0.0001.

Fig. 5. Inhibition of HSF1 transcriptional activity via eIF4A potentiates FLT3 inhibitor sensitivity in FLT3-mutant AML cells.

A Immunoblot of HSF1 protein in MOLM13 cells transfected with doxycycline-inducible control shRNA (ShC) or shRNA targeting HSF1 gene (ShHSF1). Cells were treated with doxycycline at indicated concentrations for 48 h. B Live cell numbers in MOLM13 cells (ShC and ShHSF1) treated with 0.5 μg/ml doxycycline for indicated durations. Live cell numbers were counted by trypan-blue dye exclusion method. Data represent mean ± SD. C Percentage of drug-specific apoptosis in MOLM13 cells (ShC and ShHSF1) treated with indicated concentrations of doxycycline for 96 h. Annexin V-positive cells were counted by flow cytometry. Data represent mean ± SD. D Percentage of drug-specific apoptosis in MOLM13 cells treated with or without 0.5 μg/ml doxycycline for 48 h, followed by treatment of 6.25 nM ASP2215 (left) or 1.25 nM AC220 (right) for 72 h. E Percentage of drug-specific apoptosis in Ba/F3 ITD-D835Y cells treated with or without 0.5 μg/mL doxycycline (Tet) for 48 h, followed by treatment of 12.5 nM ASP2215 for 72 h. Annexin V positive cells were measured by flow-cytometry. F Immunoblotting of HSF1, phosphorylated HSF1 (S326) and β-actin in MOLM13 cells treated with or without RHT and/or sorafenib. Values indicate ratio of intensity of each protein and β-actin. Data represent mean ± SD. **P < 0.01, ***P < 0.001, ****P < 0.0001.

Fig. 6. Combinatorial RHT and FLT3 inhibition induces synergistic apoptosis in FLT3-ITD mutant AML cells.

A Percentages of drug-specific apoptosis (left) or live cell number (right) of MOLM13 cells treated with at indicated concentrations of RHT and FLT3 inhibitors (PKC412, AC220, sorafenib, E6201, or ASP2215) for 72 h. B Combination indices in each combination in A based on percent apoptosis.

Fig. 7. Combinatorial RHT and FLT3 inhibition induces synergistic apoptosis in FLT3-ITD/TKD double-mutant cells.

A, B Percentages of drug-specific apoptosis (left) or live cell number (right) of Ba/F3 cells carrying double mutations of FLT3-ITD + D835H (A) or FLT3-ITD + D835Y (B) treated with at indicated concentrations of RHT and FLT3 inhibitors (E6201 or ASP2215) for 72 h. Combination indices based on percent apoptosis are shown under each panel.