Inhibition of BMI1 induces autophagy-mediated necroptosis

ABSTRACT

The clonal self-renewal property conferred by BMI1 is instrumental in maintenance of not only normal stem cells but also cancer-initiating cells from several different malignancies that represent a major challenge to chemotherapy. Realizing the immense pathological significance, PTC-209, a small molecule inhibitor of BMI1 transcription has recently been described. While targeting BMI1 in various systems significantly decreases clonal growth, the mechanisms differ, are context-dependent, and somewhat unclear. We report here that genetic or pharmacological inhibition of BMI1 significantly impacts clonal growth without altering CDKN2A/INK4/ARF or CCNG2 and induces autophagy in ovarian cancer (OvCa) cells through ATP depletion. While autophagy can promote survival or induce cell death, targeting BMI1 engages the PINK1-PARK2-dependent mitochondrial pathway and induces a novel mode of nonapoptotic, necroptosis-mediated cell death. In OvCa, necroptosis is potentiated by activation of the RIPK1-RIPK3 complex that phosphorylates its downstream substrate, MLKL. Importantly, genetic or pharmacological inhibitors of autophagy or RIPK3 rescue clonal growth in BMI1 depleted cells. Thus, we have established a novel molecular link between BMI1, clonal growth, autophagy and necroptosis. In chemoresistant OvCa where apoptotic pathways are frequently impaired, necroptotic cell death modalities provide an important alternate strategy that leverage overexpression of BMI1.

Introduction

BMI1 (BMI1 proto-oncogene, polycomb ring finger), a member of the Polycomb Repressor Complex 1 that mediates gene silencing by regulating chromatin structure, is preferentially expressed in stem cells where it supports self-renewal and clonal growth.^1,2 BMI1 is frequently upregulated and its expression correlates with poor prognosis in several types of cancer.^3-6 We have previously demonstrated that BMI1 is overexpressed in high-grade serous OvCa patient samples⁵ and targeting BMI1 decreases clonal growth and sensitizes OvCa cells to chemotherapeutics.^5,7 BMI1 thus has been implicated in the propagation of several cancers with a role in self-renewal of cancer-initiating cells in glioblastoma and clonal self-renewal in colorectal cancer.^3,4 Targeting BMI1 significantly decreases clonal growth, the mechanisms though differ and include derepression of CDKN2A/INK4/ARF in normal neural stem cells, induction of CCNG2 in leukemic cells and induction of apoptosis in colorectal cancer cells.^4,8,9

While decreased self-renewal of neural stem cells has been attributed to the derepression of the CDKN2A/INK4/ARF locus,^10,11 dual deletion of CDKN2A/INK4/ARF in the bmi1^−/− background only partially rescues neural development indicating regulation of additional pathways by BMI1.⁸ Accordingly, in a CDKN2A/INK4/ARF-independent manner, BMI1 regulates mitochondrial function as evidenced by interruption of the electron transport chain and decreased ATP levels in thymocytes from mice lacking BMI1.¹² Since metabolic stress and energy depletion are potent inducers of autophagy, we wanted to investigate if targeting BMI1 in ovarian cancer (OvCa) induced autophagy and if so what were the consequences of autophagic induction.

We propose that inhibition of BMI1 either by siRNA or PTC-209, a recently described small molecule inhibitor of BMI1 transcription, significantly impacts clonal growth and induces autophagy in OvCa cells through ATP depletion. Autophagic induction accompanies engagement of the PINK1 (PTEN induced putative kinase 1)- and PARK2 (Parkin RBR E3 ubiquitin protein ligase)-dependent mitochondrial pathway and causes nonapoptotic, necroptosis-mediated cell death through the RIPK1 (receptor interacting serine/threonine kinase 1) and RIPK3 (receptor interacting serine/threonine kinase 3), pathway. Importantly, genetic as well as pharmacological inhibitors of autophagy or necroptosis rescue clonal growth in BMI1-depleted cells. Therefore, BMI1-mediated clonal growth is linked to its mitochondrial function and autophagy in OvCa. Hence, in chemoresistant OvCa where apoptotic pathways are frequently impaired, autophagic cell death modalities provide an important alternate strategy that hinge upon depletion of BMI1.

Results

Depletion of BMI1 induces autophagy

To address a direct role for BMI1 in induction of autophagy in OvCa, high-grade serous OVCAR4 and cisplatin resistant CP20 cells were transfected with either scrambled (si-Control) or BMI1 siRNA (si-BMI1) for 48 h in 10% fetal bovine serum (FBS)-containing medium. The conversion of MAP1LC3B-I/LC3B-I (microtubule associated protein 1 light chain 3 (18 kDa), to a lipidated form (LC3B-II, 16 kDa)¹³ and levels of SQSTM1/p62, a protein that binds LC3 through a sequence-specific motif and is degraded were determined as a measure of autophagy.^14,15 Silencing BMI1 triggered significant increase in LC3B-II in both cell lines with simultaneous decrease in SQSTM1 levels (Fig. 1A). To confirm this observation, we tested with PTC-209, a novel selective transcriptional inhibitor of BMI1.⁴ In a dose-dependent manner, PTC-209 significantly increased LC3B-II and reduced SQSTM1 and BMI1 levels in OvCa cells (Fig. 1B). To further confirm, both CP20 and OVCAR4 cells were transfected with the green fluorescent protein (GFP)-fused LC3B (GFP-LC3B) and treated with 100 nM PTC-209 for 48 h. LC3 decorates the autophagic vesicles at all steps throughout the autophagosome maturation process.¹⁶ A diffuse green signal was observed throughout the cell in the vehicle-treated control groups whereas distinct puncta were observed in the PTC-209 treated cells by fluorescence microscopy, which represent formation of phagophores and autophagosomes (Fig. 1C). In order to distinguish if PTC-209 stimulated production of or alternately reduced clearance of autophagosomes, using 2 different methods we determined the autophagic flux in presence or absence of bafilomycin A₁ (BafA1), that prevents maturation of autophagic vacuoles by inhibiting fusion between autophagosomes and lysosomes¹⁷ and choloroquine (CQ), that raises the lysosomal pH, which inhibits of fusion of autophagosome with lysosome.¹⁸ In both cell lines, compared to BafA1 or CQ only, cotreatment with PTC-209 significantly increased LC3B-II and moderately decreased SQSTM1 levels (Fig. 1D). To further confirm, autophagic flux was monitored by GFP-LC3 puncta quantitation by determining the percentage of cells that had > 20 puncta. Compared to BafA1 or CQ alone the percentage of cells bearing GFP-LC3 puncta was significantly higher in the PTC-209+BafA1 or PTC-209+CQ treated cells (49% and 47.5% in CP20; 48% and 40% in OVCAR4) (Fig. S1), suggesting that PTC-209 stimulated autophagy. To establish that induction of autophagy was directly regulated by BMI1, both CP20 and OVCAR4 cells were first transfected with either si-Control or si-BMI1 for 24 h, and again transfected for another 24 h with FLAG-empty vector (FLAG-EV) or a FLAG-BMI1 construct, that is unresponsive to the siRNA. Forced expression of si-resistant BMI1 in si-BMI1 treated cells reverted LC3B-II and SQSTM1 levels to that of control cells (Fig. 1E). Interestingly, in chronic myeloid leukemia cells, treatment with PTC-209 induces CCNG2 expression and CCNG2-mediated autophagy.⁹ However, PTC-209 or BMI1 siRNA did not induce CCNG2 indicating absence of such regulation in OvCa cells (Fig. S2). Thus both genetic and pharmacological inhibition of BMI1 resulted in significant induction of autophagic flux in OvCa cells.

Figure 1.

Induction of autophagy by depletion of BMI1. (A) CP20 and OVCAR4 cells were transfected with either scrambled (si-Control) or BMI1 siRNA (si-BMI1) for 48 h in complete 10% FBS-containing medium. Expression of BMI1, SQSTM1 and MAP1LC3B was determined by western blot. ACTB/β actin was used as loading control. (B) CP20 and OVCAR4 cells were treated with the indicated concentrations of PTC-209 (PTC), along with the vehicle control for 48 h and expression of BMI1, MAP1LC3B-II, and SQSTM1 was determined by western blot. (C) CP20 and OVCAR4 cells were transfected with GFP-LC3B and treated with 100 nM PTC for 48 h to monitor autophagosome formation by fluorescence microscopy. Each image is the representation of 3 independent experiments performed in triplicate and each bar represent 20 µm. (D) CP20 and OVCAR4 cells were pretreated for 3 h with autophagy inhibitors 0.5 µM BafA1 or 5 µM CQ followed by treatment with or without 100 nm PTC for 48 h, the expression of BMI1, MAP1LC3B-II and SQSTM1 was determined by western blot. (E) CP20 and OVCAR4 cells were first transfected with either si-Control or si-BMI1 for 24 h and further transfected with FLAG-empty vector (FLAG-EV) or FLAG-BMI1 for another 24 h before determining expression of BMI1, MAP1LC3B-II, and SQSTM1 by western blot.

BMI1-mediated modulation of autophagy is ATP-dependent

Since decreased intracellular ATP might trigger autophagy, CP20 and OVCAR4 cells were treated with genetic or pharmacological inhibitors of BMI1 as above, and intracellular ATP levels determined. Significant decrease in intracellular ATP levels was observed in both cell lines either with si-BMI1 (∼50% to 60%) or with PTC-209 (∼40% to 60%) (Fig. 2A). To confirm that ATP depletion induced autophagy, siRNA-transfected cells (48 h) were supplemented with 2 µM ATP for the last 4 h. 10 µM FCCP, an uncoupling agent which dissipates the proton gradient across the mitochondrial inner membrane was used for 4 h as a positive control as it has been reported to induce autophagy in cells.¹⁹ In both cell lines, a significant decrease in LC3B-II and increase in SQSTM1 levels after ATP repletion suggested that exogenous ATP supplementation in si-BMI1 treated cells could reverse the autophagic flux while si-control remained unchanged (Fig. 2B). Similar to siRNA, ATP supplementation postpharmacological inhibition of BMI1 by PTC-209, also significantly reduced LC3B-II and increased SQSTM1 levels similar to control (Fig. 2C), thus confirming that induction of autophagy in BMI1 inhibited cells is ATP-dependent. ATP depletion can activate the energy sensor AMP-activated, protein kinase (AMPK), which then inactivates the MTOR (mechanistic target of rapamycin [serine/threonine kinase]) complex 1^20,21 Interestingly, upon treatment with PTC-209 for 48 h, phosphorylated (p)-PRKAA (protein kinase, AMP-activated, α) significantly increased (Fig. 2D) but total PRKAA levels remained unchanged in both cell lines. In corroboration, reduced phosphorylation of the 70 and 85 kDa isoforms of RPS6KB1 (ribosomal protein S6 kinase, 70 kDa, polypeptide 1; p70 RPS6KB1 and p85 RPS6KB1), which are downstream MTOR targets, was observed (Fig. 2D). These results establish that depletion of ATP is a key signal regulating autophagy in BMI1-silenced cells.

Figure 2.

BMI1 depletion-mediated autophagy is ATP-dependent. (A) CP20 and OVCAR4 cells were treated with either, si-Control, si-BMI1 or vehicle control and PTC-209 (100 nM) for 48 h and intracellular ATP levels determined and normalized with the respective number of viable cells in each group and compared with respect to control. Data represent mean ± SD of 3 independent experiments performed in triplicate. *P < 0.05 when comparing si-Control vs si-BMI1 and #, P < 0.05 when comparing control vs PTC. (B) si-RNA-transfected (48 h) CP20 and OVCAR4 cells were supplemented with 2 µM ATP for the last 4 h. 10 µM FCCP for 4 h was used as a positive control. Expression of BMI1, MAP1LC3B-II, SQSTM1 and ACTB was determined by western blot. (C) CP20 and OVCAR4 cells were treated with PTC-209 (PTC) for 48 h at 100 nM and supplemented with 2 µM ATP for the last 4 h. Expression of BMI1, MAP1LC3B-II, SQSTM1 and ACTB was determined by western blot. (D) CP20 and OVCAR4 cells were treated with 100 nM of PTC-209 (PTC) for 48 h and western blots were performed with the respective indicated antibodies.

Depletion of BMI1 induces autophagy of the mitochondria

In addition to decreased ATP, a loss in membrane potential is a hallmark of mitochondrial dysfunction that can lead to mitophagy.²² Therefore we determined the mitochondrial membrane potential (MMP) in CP20 and OVCAR4 cells treated with si-BMI1 or PTC-209 using tetramethylrhodamine ethyl ester (TMRE), a fluorescent probe sensitive to the MMP. Both genetic and pharmacological inhibition of BMI1 resulted in a significant ∼40% to 60% decrease in the MMP in both the cell lines (Fig. 3A). Recent reports suggest that the PINK1-PARK2 pathway is important in regulating mitochondrial quality control and linked to mitochondrial maintenance.²³ Upon mitochondrial damage or uncoupling due to depletion of MMP, PINK1 accumulates on the damaged mitochondria and facilitates translocation and accumulation of PARK2 to the mitochondria that mediates their autophagic elimination,²⁴ known as mitophagy.^25-27 To this end we treated both CP20 and OVCAR4 cells with PTC-209 and determined the expression of PINK1 and PARK2 in the mitochondria-isolated fraction. SOD2 (superoxide dismutase 2, mitochondrial) served as the mitochondrial fraction marker. As expected BMI1 levels were significantly decreased in the whole cell lysate from the PTC-209 treated cells (Fig. 3B). Strikingly, significant accumulation of PINK1 and PARK2 was observed in the mitochondrial fraction of the PTC-209 treated cells compared to the control (Fig. 3B). Notably, mediators of other mitophagic pathways such as the BNIP3 (BCL2/adenovirus E1B 19kDa interacting protein 3), whose levels reportedly increase in the mitophagic mitochondria,²⁸ did not increase (Fig. 3B). Similar results obtained from si-BMI1 treated cells (Fig. S3) further confirmed these observations.

Figure 3.

PTC-209 induces autophagy of the mitochondria. (A) CP20 and OVCAR4 cells were treated with either, si-Control, si-BMI1 or vehicle control and 100 nm PTC-209 (PTC) for 48 h, incubated with TMRE and fluorescence intensity determined (MMP) and normalized with the respective viable cells in each group; then expressed relative to the respective control. Data represent mean ± SD of 3 independent experiments performed in triplicate. *P < 0.05 when comparing si-Control vs si-BMI1 and # P < 0.05 when comparing control vs PTC. (B) CP20 and OVCAR4 cells were treated with 100 nM PTC for 48 h and expression of BMI1 determined in the whole cell lysate (WCL); expression of PINK1, PARK2, BNIP3 in the mitochondria isolated fraction was monitored. SOD2 served as the mitochondrial fraction marker. (C) CP20 and OVCAR4 cells were cotransfected with GFP-LC3B and (dsRed)-Mito and treated with vehicle control (Con) or 100 nM PTC for 48 h or starved for 2 h in OPTI-MEM (starvation) and analyzed by fluorescence microscopy. The green puncta represent autophagosomes (white arrow heads in starvation group) while yellow puncta represent colocalization of dsRed-Mito with GFP-LC3 (white arrow heads in the PTC-treated group). Each image is the representation of 3 independent experiments and each scale bar represents 10 µm.

At the cellular level, treatment with PTC-209 induced distinct GFP-LC3 puncta suggestive of autphagosome formation in both the cells (Fig. 1C). In order to visualize mitophagy, the cells were also transfected with (dsRed)-Mito. Frequently, yellow puncta representing colocalization of dsRed-Mito with GFP-LC3, were observed in PTC-209 treated cells compared to almost none in the control cells (Fig. 3C) or in cells that were starved to induce autophagy but not mitophagy. These results confirmed that depletion of BMI1 engaged the PINK1-PARK2-mediated mitophagic pathway in both the cells.

PTC-209 induces nonapoptotic, caspase-independent autophagic cell death by inhibiting apoptosis

To investigate whether PTC-209-induced autophagy and mitophagy led to cell death, both CP20 and OVCAR4 cells were treated with increasing doses of PTC-209 for 48 h and the ApoTox-Glo Triplex assay, that calculates cell viability and cell death simultaneously by measuring the activity profiles of 2 protease markers and CASP3-CASP7 activity through a luminogenic caspase substrate²⁹ was utilized. Interestingly, PTC-209 dose-dependently decreased cell viability, increased cell death but rather significantly, did not induce downregulation of CASP3-CASP7 activity (Fig. 4A) suggestive of a nonapoptotic mechanism responsible for cell death. The assay was further validated using Triton X-100 as a positive control for cell death and cisplatin as a positive control for apoptosis (Fig. S4A and B). To further consolidate the assessment of lack of apoptosis in PTC-209 treated cells TUNEL (terminal deoxynucleotidyl transferase dUTP nick end labeling) positivity was assessed. While cisplatin induced significant TUNEL positivity ∼44% and ∼42% in CP20 and OVCAR4 respectively, PTC-209 did not (Fig. S5A). In similarly treated samples cisplatin induced potent cleavage of PARP1 (poly[ADP-ribose] polymerase 1) but, PTC-209 did not (Fig. S5B). We next treated CP20 cells with PTC-209 or cisplatin for 24 and 48 h followed by the ANXA5 and propidium iodide (PtdIns) assay. Cisplatin significantly increased the ANXA5-only positive fraction in a time-dependent manner while PTC-209 significantly increased the PI-positive fraction with or without ANXA5 staining (Fig. S5C). Furthermore, XIAP (X-linked inhibitor of apoptosis) that reportedly inhibits CASP3-CASP7 activation³⁰ was induced by PTC-209 (Fig. 4B). Together these results indicate that PTC-209 induces a nonapoptotic, caspase-independent form of cell death.

Figure 4.

PTC-209 induces nonapoptotic, caspase-independent autophagic cell death. (A) CP20 and OVCAR4 cells were treated with increasing doses of PTC-209 (PTC) for 48 h and cell viability, cell death and CASP3-CASP7 activity was evaluated using the ApoTox-Glo Triplex assay kit. Data represent mean ± SD of 3 independent experiments performed in triplicate. *P < 0.05 when compared with the respective vehicle-treated control. (B) CP20 and OVCAR4 cells were treated with 100 nm PTC for 48 h and western blots were performed with the respective indicated antibodies. (C) CP20 cells were treated with si-Control (si-con), si-BMI1 or with vehicle control and increasing doses of PTC to evaluate the clonal growth. Colonies were stained with crystal violet and counted. the respective si-Control or vehicle-treated cells were set to 100%. Data represent mean ± SD of 3 independent experiments performed in triplicate. *P < 0.05 when compared with the respective control. (D) CP20 cells were pretreated for 3 h with either autophagy inhibitors 0.5 µM BafA1 or 5 µM CQ or the mitophagy inhibitor 0.5 µM Mdivi-1 followed by a single treatment with 100 nm PTC up to 7 d. Colonies were stained with crystal violet and counted. Vehicle-treated control cells were set to 100%. Data represent mean ± SD of 3 independent experiments performed in triplicate. *P < 0.05 when compared with the respective control and # < 0.05 when compared with only PTC-209-treated cells. (E) Following transfection with si-Con or si-ATG7, CP20 or OVCAR4 cells were treated with 100 nm PTC for 48 h and western blots were performed with the respective indicated antibodies. (F) Following transfection with si-Con or si-ATG7, CP20 cells were treated with 100 nm PTC up to 7 d and clonal growth evaluated. Colonies were stained with crystal violet and counted. The respective si-Control or vehicle-treated cells were set to 100%. Data represent mean ± SD of 3 independent experiments performed in triplicate. *P < 0.05 when compared with si-control only and # P < 0.05 when compared with the indicated PTC-treated groups.

Depending upon the cellular context autophagy can be either cytoprotective or lead to cell death. In order to determine whether inhibition of BMI1 was directly or indirectly linked to autophagy-mediated cell death, we utilized the gold-standard clonal growth assays.³¹ Treatment with si-BMI1 or with PTC-209, dose-dependently decreased clonal growth in CP20 cells (Fig. 4C). In normal neural stem cells, targeting BMI1 decreases clonal growth through derepression of CDKN2A/INK4/ARF.⁸ Therefore in si-BMI1- or PTC-209-treated cells, CDKN2A/p16 INK4A and CDKN1A/cyclin-dependent kinase inhibitor 1A/(p21, Cip1) levels, the ultimate effector of the CDKN2A/p14ARF pathway^10,32 were determined and found unchanged (Fig. S2). In PTC-209 treated cells, pretreatment with BafA1, CQ, or the mitophagy inhibitor Mdivi-1, the inhibitor of mitochondrial division,³³ significantly rescued clonal growth to near control levels (∼73%, ∼64% and ∼81% respectively compared to control) (Fig. 4D). In order to confirm that autophagy was required for PTC-209-mediated decrease in clonal growth, we employed genetic depletion of ATG7/autophagy related 7 by si-RNA. Treatment with PTC-209 moderately but consistently increased expression of ATG7 in both CP20 and OVCAR4 cells which was abolished in si-RNA (Fig. 4E) treated cells. In PTC-209 treated cells, genetic inhibition of ATG7 significantly rescued (∼81% compared to control) the decreased clonal growth (Fig. 4F). These results confirm that autophagic induction is required for PTC-209-mediated nonapoptotic, caspase-independent cell death.

PTC-209-mediated autophagy leads to necroptosis

Having established that autophagy is necessary for PTC-209-mediated cell death, we focused on understanding how the autophagic machinery is connected to cell death signaling pathways. Of the few specific cell death modalities induced by autophagy, necroptosis represents a novel method of caspase-independent cell death.^33,34 We therefore assessed if PTC-209 induced necroptosis and its relationship to autophagy and mitophagy. To this end, we treated both CP20 and OVCAR4 cells with PTC-209 for 48 h and immunoblotted for necroptosis markers, RIPK1 and RIPK3. In both the cells, upon treatment with PTC-209 while RIPK1 remained unchanged RIPK3 was significantly upregulated (Fig. 5A), a similar effect was observed in si-BMI1 treated cells. (Fig. S6A). To investigate whether autophagy and mitophagy were upstream of necroptosis, we analyzed the necroptosis markers in cells pretreated with the autophagy (BafA1) and mitophagy (Mdivi-1) inhibitors followed by PTC-209. Notably, autophagy and mitophagy inhibitors reversed the PTC-209-dependent RIPK3 induction and subsequent MLKL activation similar to that of control levels in both the cells (Fig. 5B). To further confirm that PTC-209-mediated necroptosis played a critical role in clonal growth, we pretreated CP20 cells either with a cell permeable RIPK3 inhibitor, GSK'872,³⁵ or with si-RIPK3 followed by PTC-209. Treatment with GSK'872 (Fig. 5C) or si-RIPK3 (Fig. S6B) significantly protected against PTC-209-mediated decrease in clonal growth, confirming that activation of RIPK3 is important for decreased clonal growth. In murine cells, GSK'872 has been reported to induce caspase-dependent apoptosis.³⁶ Therefore, we determined whether the PTC-209 and GSK combination used in above clonal growth assays stimulated apoptosis in the short term. PTC-209 with or without GSK'872 induced expression of XIAP (Fig. 5D, lower panel) suggesting maintenance of apoptotic inhibition by PTC-209. Compared to PTC-209 alone (Fig. 4A), cotreatment with GSK'872 significantly enhanced viability as determined by the Triplex assay (Fig. S6D). Furthermore cotreatment of GSK'872 along with PTC-209 neither stimulated CASP3-CASP7 activity (Fig. S6D) nor enhanced CASP3 cleavage as observed with the positive control cisplatin (Fig. S6E). These results thus confirm lack of any apoptotic activity of the GSK'872+PTC-209 combination in OvCa cells.

Figure 5.

PTC-209-mediated autophagy leads to necroptosis. (A) CP20 and OVCAR4 cells were treated with 100 nM PTC-209 (PTC) for 48 h and expression of necroptosis markers was determined by western blot. (B) CP20 and OVCAR4 cells were pretreated with BafA1 or Mdivi-1 followed by 100 nM PTC and necroptosis markers were determined by western blot. (C) CP20 cells were pretreated with a cell-permeable RIPK3 inhibitor GSK'872 (3 µM, GSK) followed by 100 nM PTC-209 and clonal growth was determined. Colonies were stained with crystal violet and counted. Vehicle-treated control cells were set to 100%. Data represent mean ± SD of 3 independent experiments performed in triplicate. *P < 0.05. (D) CP20 and OVCAR4 cells were pretreated with GSK at 3 µM for 3 h and further treated with 100 nM PTC for 48 h and western blots were performed with the indicated antibodies.

To further determine that the decreased clonal growth is necroptosis-dependent, we used necrostatin-1 (Nec-1), an inhibitor of RIPK1. Compared to PTC-209, significant rescue in clonal growth was observed in Nec-1+PTC-209-treated cells (Fig. S6C) indicating that RIPK1 is also important for clonal growth. This is consistent with reports indicating that the RIPK1-RIPK3 complex is required for and controls necroptosis by modulating its downstream substrate MLKL.^35-38 To investigate whether RIPK3 is required for the induction of autophagy, both CP20 and OVCAR4 cells were pretreated with the RIPK3 inhibitor GSK'872 followed by PTC-209 and LC3 conversion analyzed. Interestingly, inhibition of RIPK3 did not alter the conversion of LC3B-I to LC3B-II upon PTC-209 treatment (Fig. 5D), thus placing PTC-209-induced autophagy or mitophagy upstream of RIPK3 activation. Overall, these results confirmed that RIPK-dependent necroptosis is required for PTC-209-induced cell death.

In summary, genetic or pharmacological inhibition of BMI1 elicits a significant impairment in intracellular ATP generation due to reduction of MMP and interruption of flow in the electron transport chain (ETC)¹² (Fig. 6). Depletion of ATP triggers the stress signal that induces autophagy and simultaneously activates AMPK. Activation of AMPK downregulates the MTOR pathway^20,21 which might also potentially induce autophagy. Simultaneously, apoptosis is significantly inhibited by PTC-209 through increased expression of XIAP. At the same time RIPK3-mediated necroptosis is potentiated by phosphorylation of MLKL (Fig. 6).

Figure 6.

Schema of BMI1 depletion-mediated induction of autophagy-dependent necroptotic cell death. Genetic (si-RNA) or pharmacological (PTC-209) inhibition of BMI1 triggers a significant reduction in ETC function (reported by Liu et. al.,¹²) and MMP which leads to reduced intracellular ATP. Depletion of ATP stimulates the autophagic stress signal and also activates AMPK. AMPK activation inhibits the MTOR pathway^20,21 that might also lead to induction of autophagy. PTC-209 inhibits the apoptotoic machinery via upregulation of XIAP but facilitates induction of autophagic flux, which ultimately escalates expression of RIPK3 that mediates necroptosis through phosphorylation of its downstream substrate MLKL.

Discussion

Accumulating evidence has established BMI1 as an important therapeutic target in several different malignancies.^4,5,7,39 Accordingly, a recently developed small-molecule transcriptional inhibitor of BMI1, PTC-209 inhibits self-renewal causing long-term, irreversible impairment in primary colorectal tumor growth.⁴ However, while targeting BMI1 significantly decreases clonal growth and selfrenewal the mechanisms differ and are context-dependent. These include derepression of CDKN2A/INK4/ARF in normal neural stem cells, induction of CCNG2-mediated autophagy in leukemic cells and induction of apoptosis in colorectal cancer-initiating cells.^4,8,9 We have previously demonstrated that BMI1 is overexpressed in high-grade serous OvCa patient samples⁵ and targeting BMI1 decreases clonal growth and sensitizes OvCa cells to chemotherapeutics.^5,7 Thus, determining the mechanism of anti-BMI1 action in chemoresistant OvCa is an important step toward developing new therapeutic strategies.

Here we report for the first time that transcriptional loss of BMI1 induces autophagic flux in OvCa cells. While autophagy can promote survival or induce cell death, we show that transcriptional loss of BMI1 induces necroptosis-mediated cell death that involves the mitochondrial PINK1-PARK2 pathway and results in decreased clonal growth without altering CDKN2A/INK4/ARF or CCNG2.

The significant decrease in ATP levels observed in the BMI1 silenced OvCa cells appear to be a general phenomenon linked to the mitochondrial function of BMI1 as has been reported in thymocytes from the bmi1 knockout mice.¹² Mechanistically, we identify ATP depletion, a cause for autophagy, since supplementation with ATP in BMI1 inhibited cells significantly reverses LC3B-II and SQSTM1 levels. Moreover, reports indicate that AMPK, that is sensitive to the cellular AMP:ATP ratio is activated by metabolic stresses that either inhibit ATP production or that stimulate ATP consumption.⁴⁰ Activated AMPK then inhibits the MTOR pathway reflected by reduced phosphorylation of RPS6KB1.⁴¹ In corroboration, activation of AMPK and concomitant decrease in phosphorylation of RPS6KB1 is observed in PTC-209 treated OvCa cells.

In BMI1-silenced OvCa, in addition to decreased ATP levels, significant loss of mitochondrial membrane potential was observed which is in line with impaired electron flow reported in bmi1^−/− thymocytes.¹² Depolarization is often indicative of impaired mitochondrial function and is a prerequisite for autophagy of the mitochondria that is characterized by recruitment of PARK2 and stabilization of PINK1 at the mitochondria.^23,42,43 Accordingly the levels of PINK1 and PARK2 increased in the mitochondrial fraction of PTC-209 treated cells and colocalization of distinct GFP-LC3 puncta containing dsRed-Mito further confirmed autophagy of the mitochondria.

Importantly, while autophagy could be pro cell survival or pro cell death, inhibiting BMI1 either by siRNA or PTC-209 significantly decreased clonal growth. The tumor suppressive CDKN2A/p14 and the CDKN2A/p14 pathway were ruled out since CDKN2A/p16 levels or CDKN1A levels, the ultimate effector of the CDKN2A/p14 pathway^10,32 did not change in BMI1 siRNA or PTC-209-treated cells. Interestingly in chronic myeloid leukemia cells, PTC-209 treatment induces CCNG2 expression and CCNG2-mediated autophagy.⁹ However, PTC-209 or BMI1 siRNA did not induce CCNG2 indicating absence of such regulation in OvCa cells. In colorectal cancer cells, a 4 d treatment with PTC-209 reduces the frequency of sphere initiation from viable cells.⁴ In a separate experiment, a 4 or 5 d treatment with 500 nM PTC-209 induces apoptosis within the tumor-initiating population of colorectal cancer cells.⁴ However in OvCa cells PTC-209 dose-dependently (100 to 500 nM) significantly decreased viability at 48 h. In the same assay CASP3-CASP7 activity was also significantly downregulated, we think due to induction of XIAP which inhibits caspase activity.^30,44,45 In OvCa cells, PTC-209 did not induce apoptosis as determined by several assays such as TUNEL positivity, ANXA5-PtdIns staining and cleavage of PARP1. Therefore, PTC-209-mediated loss in viability and clonal growth was not due to apoptosis but due to autophagic cell death. Accordingly, decreased clonal growth in PTC-209-treated cells could be significantly rescued by genetic inhibition of ATG7 or by the autophagic inhibitors BafA1, CQ or the mitophagic inhibitor Mdivi-1.

We also report for the first time that PTC-209-mediated nonapoptotic autophagic cell death in OvCa is via necroptosis. Previous reports indicate that the autophagic or mitophagic machinery might induce cell death through necroptosis,^34,46,47 a novel mode of caspase-independent cell death that is distinct from apoptosis and depends upon the activation of RIPK1 and RIPK3.⁴⁸ RIPK3 then phosphorylates MLKL, the ultimate effector of necroptosis.^37,38,49 We show that treatment with PTC-209 activates RIPK3 and phosphorylates MLKL, which can be significantly reversed by either BafA1 or Mdivi-1. Furthermore RIPK1 or RIPK3 is critically required for PTC-209-mediated decrease in clonal growth as evidenced by the rescue with genetic or pharmacological inhibition, thus indicating that the RIPK1-RIPK3 complex is required for necroptosis signaling.^50,51

The mechanism of necroptosis itself or its connection to autophagy is a relatively new and evolving area of research. Regarding how necroptosis might be engaged in our system, we speculate that in the context of autophagic induction, caspase inhibition by PTC-209 might be a signal that channels autophagic cell death through RIPK-dependent necroptosis. Indeed microglia activated through Toll-like receptors (TLRs) undergo RIPK1- and RIPK3-dependent necroptosis when exposed to the pancaspase inhibitor zVAD-fmk.⁵² Also, in mouse fibrosarcoma L929 cells, zVAD induces autophagic cell death that is distinct from apoptosis.³⁴ Furthermore, according to Basit et al., Obatoclax, a pan-BCL2 family inhibitor, leads to autophagy and cell death through caspase-independent but RIPK1- and RIPK3-dependent necroptosis.⁵³ In this model, Obatoclax leads to recruitment of the necrosome on the autophagosome, RIPK1 and RIPK3 along with the adapter protein FADD are recruited to autophagosomes by interaction with ATG proteins.⁵³ Thus according to Oberst et al., autophagic membranes can recapitulate the protein complexes and cell death pathways normally activated by receptors⁵⁴ and a similar phenomenon might be envisioned in BMI1-depleted cells.

In conclusion, we have established a novel molecular link between BMI1, clonal growth, autophagy and necroptosis in OvCa. Thus, in chemoresistant OvCa where apoptotic pathways are frequently impaired, necroptotic cell death modalities provide an important alternate strategy that leverage overexpression of BMI1.

Materials and methods

Cell culture and chemicals

CP20 cell line (kind gift from Dr. Anil K. Sood, MD Anderson Cancer Center, Houston, TX, USA) and OVCAR4 (kind gift from Dr. Ronny I. Drapkin, formerly at Dana-Farber Cancer Institute, Boston, MA, USA) were routinely cultured in RPMI (Corning, 10–040-CV) and supplemented with 10% heat inactivated FBS (Gibco, 16000–044) and 100 ug penicillin-100 μg streptomycin/ml medium (Gibco, 15140–122) in a 5% CO₂ humidified atmosphere. PTC-209 was obtained from Sigma (SML1143). Gene silencing was performed using Hiperfect (Qiagen, 301707) and 10 picomoles siRNA (scrambled control Dharmacon, D-001206–13–20; BMI1 siRNA SASI-HS01-00175765, ATG7 siRNA SASI-HS01-00077648 and RIPK3 siRNA SASI-HS01-00078750 from Sigma in OPTIMEM (Gibco, 31985–070). FLAG-BMI1 construct was a kind gift from Dr. Damu Tang, McMaster University, Hamilton, ON, Canada. (GFP)-fused LC3B (GFP-LC3B) obtained from Addgene (11546, deposited by Dr. Karla Kirkegaard's Lab). (dsRed)-Mito was a kind gift from Dr. Zu-Hang Sheng, National Institutes of Health, Bethesda, MD, USA. Choloroquine (C6628) and cisplatin (P4394) were obtained from Sigma; bafilomycin A₁ (11038) from Cayman cmemicals, Mdivi-1 (BML-CM127) from Enzo Life Sciences; GSK'872 (530389) from Calbiochem and Necrostatin1 (S8037) obtained from Selleckchem. In Situ Cell Death Detection Kit, Fluorescein/TUNEL assay Kit (11684795910) and Annexin-V (ANXA5)-FLUOS Staining Kit (11858777001) were obtained from Roche.

Cell lysis, cell fractionation, SDS-PAGE, and western blotting

Total cell lysate was prepared in RIPA (Boston Bioproducts, BP115). For subcellular fractionation, a previously published protocol was followed with slight modification.^55,56 For isolation of mitochondria enriched fractions, cells were disrupted in isolation buffer (250 mM sucrose [Sigma, S1888], 1 mM EGTA, 10 mM HEPES, 10 mM Tris-HCl, pH 7.5) using 2 sets of 40 strokes in a Dounce glass homogenizer (Wheaton, 357538 with “loose” pestle). The homogenates were centrifuged at 800 g for 7 min to remove the nuclear fraction and unbroken cells. The supernatant fraction was then subjected to centrifugation at 4,000 g for 15 min and the pellet fraction was taken as the crude mitochondrial fraction. The crude mitochondrial pellet fraction was washed twice and resuspended in an EGTA-free mitochondrial buffer (250 mM sucrose, 10 mM HEPES, 10 mM Tris-HCl, pH 7.5). The mitochondrial fraction was further purified by suspending the crude mitochondrial pellet in 19% percoll (GE Healthcare, 17089101) in isolation buffer, and slowly layering on 2 layers of 30% and 60% percoll (v/v). After centrifugation at 10,000 g for 15 min, mitochondria were collected and washed (3x) with isolation buffer. Measurement of protein concentration, independent of the extraction method, was performed using a BCA assay kit (Pierce, 23225).

SDS-PAGE and western immunoblotting analysis were performed using standard protocols.⁵⁷] The cell lysates were separated on 10% or 15% glycine SDS-PAGE gels and transferred to PVDF membrane. Membranes were blocked in 5% nonfat dry milk in TBS with 0.1% Tween-20 (Fisher BioReagents, BP337) (TBST) for 1 h at room temperature followed by incubation with the indicated primary antibodies in TBST with 5% BSA (Fisher BioReagents, BP1600). Antibodies were purchased from following venders: BMI1 from Invitrogen (37–5400) and Proteintech (66161); LC3B (2775), CDKN1A (2947), phospho-RPS6KB1 (9205), PINK1 (6946), SOD2 (13194), PARP1 (9542), RIPK1 (4926), phospho-MLKL(14516) and cleaved CASP3 (9664) from Cell Signaling Technology; CCNG2 (54901), PARK2 (15954) from Abcam; phospho-PRKAA (07–681) from Upstate; PRKAA (A300–508A-T) from Bethyl Laboratories; BNIP3 (95586) from Novus Biologicals; CDKN2A (10883), from Proteintech; SQSTM1 (28359) and RIPK3 (374639) from Santa Cruz Biotechnology; XIAP (MAB822) from R&D Systems; ATG7 (04–1055) from Millipore; ACTB (A5228) and secondary antibodies conjugated with horseradish peroxidase IgG rabbit (A6154) and mouse (A4416) from Sigma. Primary antibodies were used in dilutions recommended by the manufacturer. Secondary antibodies were used at a concentration of 1:10,000.

Immunofluorescence microscopy

Cells were grown to 70% confluence on a coverslip and transfected with GFP-LC3B or cotransfected with GFP-LC3B and dsRed-Mito using FuGENE 6 transfection reagent (Promega, E2692) and treated with PTC-209. Cells were washed 3 × 3 min with PBS (Corning, 21–040-CV), and fixed with freshly prepared 3.7% formaldehyde at 37°C for 15 min and further washed with PBS. The cells were then mounted with VECTASHIELD antifade mounting medium (Vector Laboratories, H-1000). Cell images were acquired with a 63X objective using a Zeiss Axio Observer. Z1 (Göttingen, Germany).

Measurement of intracelluar ATP and mitochondrial membrane potential

Total ATP levels in BMI1 silenced cells were measured using Sigma Adenosine 5′-triphosphate (ATP) Bioluminescent Assay Kit (FLAA). Mitochondrial membrane potential was measured using a kit from Abcam (113852) as per the manufacturer's instructions. Briefly, cells were transfected as described and 48 h post-transfection, cells were replated in a 96-well plate suitable for fluorescence measurement. 250 nM positively charged TMRE dye was used to label active mitochondria that are negatively charged. FCCP, an ionophore that depolarizes mitochondria and prevents binding of the dye to the mitochondria was used as a control for the assay. Plates were read at 549 nm excitation and 575 nm emission wavelengths using CLARIOstar (BMG Labtech, Ortenberg, Germany).

Determination of apoptosis, cell viability and clonal growth

Apoptosis was determined by using the ApoTox-Glo™ triplex assay kit (G6321) from Promega. Briefly, PTC-209-treated and untreated cells were incubated simultaneously to measure 2 protease activities; one is a marker of cell viability, and the other is a marker of cell death. The live- and dead-cell proteases produced different products, AFC and R110, which had different excitation and emission spectra, allowing them to be detected concurrently. The second part of the assay utilized a luminogenic CASP3-CASP7 substrate (the tetrapeptide sequence DEVD), in a reagent optimized for caspase activity, luciferase activity and cell lysis. Luminescence was proportional to the amount of caspase activity present. For clonal growth assay, cells were seeded as single cells (200 cells/well) in 6-well plates for 24 h, treated with or without PTC-209 and other inhibitors and cells were cultured for additional 7 d before staining with crystal violet solution (0.75% crystal violet, 50% ethanol, 0.25% NaCl, 1.57% formaldehyde) and counted.

Data analysis and statistics

All the experiments unless otherwise stated were repeated independently 3 times. Data are expressed as means ± standard deviation (SD). The Student t test was used for statistical analysis. Statistical significance was set at P < 0.05, using the 2-tailed distribution.

Abbreviations

ACTB

actin, β

AMPK

AMP-activated protein kinase

BafA1

bafilomycin A₁

BMI1

BMI1 proto-oncogene, polycomb ring finger

CDKN1A/p21

cyclin-dependent kinase inhibitor 1A (p21, Cip1)

CDKN2A/p16

cyclin-dependent kinase inhibitor 2A

CQ

chloroquine

FBS

fetal bovine serum

MAP1LC3B/LC3B

microtubule associated protein 1 light chain 3 β

MLKL

mixed lineage kinase domain-like

MTOR

mechanistic target of rapamycin (serine/threonine kinase)

MTORC1

MTOR complex 1

OvCa

ovarian cancer

PARK2

parkin RBR E3 ubiquitin protein ligase

PI

propidium iodide

PINK1

PTEN induced putative kinase 1

PRKAA/AMPK

protein kinase, AMP-activated, α 2 catalytic subunit

RIPK1

receptor (TNFRSF)-interacting serine/−threonine kinase 1

RIPK3

receptor- interacting serine/−threonine kinase 3

SQSTM1/p62

sequestosome 1

TMRE

tetramethylrhodamine ethyl ester.

Disclosure of potential conflicts of interest

No potential conflicts of interest were disclosed.

Funding

This study was supported by the National Institutes of Health (NIH) CA 157481 to RB. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.