Dual inhibition of Bcl-2 and Bcl-xL strikingly enhances PI3K inhibition-induced apoptosis in human myeloid leukemia cells through a GSK3- and Bim-dependent mechanism

Abstract

Effects of concomitant inhibition of the PI3K/AKT/mTOR pathway and Bcl-2/Bcl-xL (BCL2L1) were examined in human myeloid leukemia cells. Tetracycline-inducible Bcl-2 and Bcl-xL dual knockdown sharply increased PI3K/AKT/mTOR inhibitor lethality. Conversely, inducible knockdown or dominant-negative AKT increased, whereas constitutively active AKT reduced lethality of the Bcl-2/Bcl-xL inhibitor ABT-737. Furthermore, PI3K/mTOR inhibitors (e.g., BEZ235 and PI-103) synergistically increased ABT-737-mediated cell death in multiple leukemia cell lines and reduced colony formation in leukemic, but not normal, CD34+ cells. Notably, increased lethality was observed in four of six primary acute myelogenous leukemia (AML) specimens. Responding, but not nonresponding, samples exhibited basal AKT phosphorylation. PI3K/mTOR inhibitors markedly downregulated Mcl-1 but increased Bim binding to Bcl-2/Bcl-xL; the latter effect was abrogated by ABT-737. Combined treatment also markedly diminished Bax/Bak binding to Mcl-1, Bcl-2, or Bcl-xL. Bax, Bak, or Bim (BCL2L11) knockdown or Mcl-1 overexpression significantly diminished regimen-induced apoptosis. Interestingly, pharmacologic inhibition or short hairpin RNA knockdown of GSK3α/β significantly attenuated Mcl-1 downregulation and decreased apoptosis. In a systemic AML xenograft model, dual tetracycline-inducible knockdown of Bcl-2/Bcl-xL sharply increased BEZ235 antileukemic effects. In a subcutaneous xenograft model, BEZ235 and ABT-737 coadministration significantly diminished tumor growth, downregulated Mcl-1, activated caspases, and prolonged survival. Together, these findings suggest that antileukemic synergism between PI3K/AKT/mTOR inhibitors and BH3 mimetics involves multiple mechanisms, including Mcl-1 downregulation, release of Bim from Bcl-2/Bcl-xL as well as Bak and Bax from Mcl-1/Bcl-2/Bcl-xL, and GSK3α/β, culminating in Bax/Bak activation and apoptosis. They also argue that combining PI3K/AKT/mTOR inhibitors with BH3 mimetics warrants attention in AML, particularly in the setting of basal AKT activation and/or addiction.

Figure 1. Dual Bcl-2/Bcl-xL knockdown strikingly potentiates PI3K inhibitor-mediated apoptosis in U937 cells

(A, upper panels) Western blot (WB) analysis of U937 cells displaying inducible knockdown of Bcl-2 ± Bcl-xL following exposure to 1 μg/ml doxycycline (Dox). Alternatively, cells were left untreated or treated with doxycycline for 72 h, then exposed to BEZ235 (BZ; 0.5 μM), PI-103 (PI; 3 μM), or MK-2206 (MK; 3 μM) for the designated intervals, after which Annexin V/PI (A, lower panels), Bax/Bak conformational change (active Bax/Bak; B, upper panel), cytochrome c/AIF release (B, lower panel), loss of Δψ_m (C), or AKT phosphorylation and caspase activation (D) were monitored. Error Bars: S.D of 4 independent experiments.

Figure 2. PI3K/mTOR inhibitors and ABT-737 strikingly induce apoptosis in human leukemia cells

Annexin V/PI analysis of U937 cells following 18 h exposure to 0.5 μM ABT-737 and the designated concentrations of BEZ235 (A) or PI-103 (B), the designated concentrations of ABT-737 and 0.5 μM BEZ235 or 3 μM PI-103 (C). Error Bars: S.D of 3 independent experiments. (D) Annexin V/PI analysis in KG1 and MV4-11 cells following 24 h exposure to BEZ235 (500 nM and 300 nM respectively) ± ABT-737 (200 nM and 10 nM respectively). (E–F) Western Blot (WB) analysis.

Figure 3. BEZ235/ABT-737 induces apoptosis in and reduces colony formation of primary AML blasts while sparing normal CD34⁺ cells

TUNEL assay using confocal microscopy (A) and WB analysis (B) in primary AML cells (AML #2) exposed to BEZ235 (200 nM) ± ABT-737 (25 nM) for 24 h. (C) Annexin V/PI following treatment (24 h) as in (A) of six primary AML (peripheral blood: 1, 2, and 5; bone marrow: 3, 4, and 6; blast percentage of unseparated white cells were 70, 80, 74, 70, 90, and 56% respectively). Samples were further enriched for mononuclear cells by Ficoll separation. (D) Annexin V/PI analysis of two normal CD34⁺ specimens following 24 h treatment. Results are presented as the percentage of dead cells for each treatment using the formula ([treatment – control]/[100 – control]) × 100. (E) WB of six primary AML (left panel), three normal CD34⁺ specimens (N#1 and N#2 as in Figure 3D and an additional sample N#3, right panel). Colony-forming units (CFUs) for two primary AML samples AML#2 and an additional sample AML#7 exhibiting AKT phosphorylation as monitored by Western blot (F and inset) and the three normal CD34⁺ specimens shown in Figure 3E (G) treated with ABT-737 (25 nM) ± BEZ235 (100 nM) or LY294002 (LY; 3 μM) expressed as percentage relative to untreated cells. Error Bars: S.D of 3 replicates; *, p < 0.01.

Figure 4. Mcl-1 downregulation plays a functional role in BEZ235/ABT-737 lethality and involves Bax and Bak

WB of U937 cells (A), primary AML blasts (B, upper panel), and tet-inducible Bcl-2/Bcl-xL dual knockdown (B, lower panel) following exposure to BEZ235 (U937: 0.5 μM; AML: 200 nM) ± ABT-737 (U937: 0.5 μM; AML: 25 nM). (C) Annexin V/PI in U937 cells ectopically expressing Mcl-1 following treatment (18 h) with BEZ235 ± ABT-737 (0.5 μM each). Error Bars: S.D of 3 independent experiments; *, p < 0.01. (D) Bax/Bak conformational change in U937 cells following 16 h exposure to 0.5 μM ABT-737 ± 0.5 μM BEZ235 or 3 μM PI-103. WB of Bak (E) or Bax (F) immunoprecipitates or input lysates prepared from U937 cells exposed to BEZ235 ± ABT-737 (0.5 μM each). (G) Apoptosis in Bax- or Bak-knockdown U937 cells treated for 18 h as in (D). Error Bars: S.D of 3 independent experiments; *, p < 0.02 for shBak, and p < 0.01 for shBax.

Figure 5. Bim, but not Bad, plays a functional role in BEZ235/ABT-737 lethality

(A) WB of Bim immunoprecipitates or input lysates of U937 cells treated with BEZ235 ± ABT-737 (0.5 μM each). (B–C) WB and Annexin V/PI in Bim or Bad knockdown, or negative control (NC; scramble shRNA) U937 cells following 18 h exposure to ABT-737/BEZ235 (0.5 μM each) or ABT-737/PI-103 (3 μM). Error Bars: S.D of 3 independent experiments; *, p < 0.01.

Figure 6. Role of GSK3α/β in BEZ235/ABT-737 lethality

(A) WB in U937 cells following exposure (24 h) to BEZ235 ± ABT-737 (0.5 μM each). Viability assay (B) in tet-inducible Bcl-2/Bcl-xL dual knockdown U937 cells following 24 h treatment with BEZ235 ± 2 μM BIO, MeBIO, or CHIR-98014 (CHIR) in the presence or absence of doxycycline (1 μg/ml, added 72 h prior to treatment). Error Bars: S.D of 3 independent experiments; *, p < 0.01 in each case. Annexin V/PI (C) and WB (D) in U937 cells following 16 h exposure to BEZ235/ABT-737 (0.5 μM each) in the presence or absence of BIO, MeBIO, or CHIR-98014 (2 μM each). Error Bars: S.D of 3 independent experiments; * p < 0.05. WB (E) and Annexin V/PI (F) in GSK3α or GSK3β knockdown U937 cells following 16 h exposure to BEZ235/ABT-737 (0.5 μM each). Error Bars: S.D of 3 independent experiments; *, p < 0.05 for shGSK3α and p < 0.02 for shGSK3β.

Figure 7. In vivo anti-leukemic activity of combined BEZ235 and ABT-737 treatment

(A) NOD/SCID/gamma mice were tail-vein inoculated with tet-inducible Bcl-2/Bcl-xL dual knockdown U937 cells expressing luciferase, treated with BEZ235 (45 mg/Kg) ± doxycycline, and imaged using the IVIS 200 system. (B) Tumor growth in nude mice bearing subcutaneous U937 xenografts and treated with BEZ235 (35 mg/kg) ± ABT-737 (100 mg/kg) twice a day (at hour 0 and at hour 18). Error bars: SD of 3 independent experiments involving 5 mice/condition each. *, p < 0.02 versus either agent alone. (C) Photographs of 2 representative tumors for each group following 8 days of treatment. (D) Two sets of xenograft-bearing mice were treated as in (B) twice over a 24 h interval after which tumors were excised, lysed, and subjected to WB. (E) Kaplan-Meyer survival plot involving 12–13 mice/condition treated as in (B). Mice that accidentally died during the treatment procedure or were sacrificed before the maximal tumor size was reached (e.g. due to tumor bleeding) were not included in this analysis. The survival curves were significantly different for combined treatment compared to either BEZ235 or ABT-737 alone; p < 0.01; logrank test.