Targeting Notch Signaling and Autophagy Increases Cytotoxicity in Glioblastoma Neurospheres

Abstract

Glioblastomas are highly aggressive tumors that contain treatment resistant stem‐like cells. Therapies targeting developmental pathways such as Notch eliminate many neoplastic glioma cells, including those with stem cell features, but their efficacy can be limited by various mechanisms. One potential avenue for chemotherapeutic resistance is the induction of autophagy, but little is known how it might modulate the response to Notch inhibitors. We used the γ‐secretase inhibitor MRK003 to block Notch pathway activity in glioblastoma neurospheres and assessed its effects on autophagy. A dramatic, several fold increase of LC3B‐II/LC3B‐I autophagy marker was noted on western blots, along with the emergence of punctate LC3B immunostaining in cultured cells. By combining the late stage autophagy inhibitor chloroquine (CQ) with MRK003, a significant induction in apoptosis and reduction in growth was noted as compared to Notch inhibition alone. A similar beneficial effect on inhibition of cloogenicity in soft agar was seen using the combination treatment. These results demonstrated that pharmacological Notch blockade can induce protective autophagy in glioma neurospheres, resulting in chemoresistance, which can be abrogated by combination treatment with autophagy inhibitors.

INTRODUCTION

Glioblastomas are the most common primary adult brain malignancy, comprising more than 50% of all gliomas 23. Despite some advances in therapeutic options, outcomes remain dismal with a median overall survival of 14.6 months and a 2‐year survival rate of 26.5% 33. Treatment resistant stem‐like glioma cells have been proposed as one reason for treatment failure 4, 8, 10, 25, 36, 38. We 3, 9 and others 30, 34 have demonstrated the efficacy of γ‐secretase inhibitors (GSI), which block Notch pathway activity, using in vitro and in vivo glioma models. However, success has been limited in clinical trials using GSIs as a single agent in gliomas 17, and in other solid tumors alone 17, 35 or in combination with other drugs 5, 26, 29. We, therefore, examined if induction of autophagy might be limiting the therapeutic efficacy of Notch inhibitors.

Autophagy is a dynamic process by which cellular organelles and proteins are sequestered, delivered to lysosomes and digested 37. Induction of autophagy has been identified as a potential mechanism for resistance to chemotherapy 16 including temozolomide 14, 18, 19, 22, 42 as well as radiation treatment 13, 41, which represent the standard therapies for glioblastoma. Manipulation of autophagy has increasingly been explored as a treatment strategy in many cancers 21. However, there is minimal data on the effects of the Notch pathway on autophagy in cancer 12. We examined if Notch pathway blockade using the GSI MRK003 would induce autophagy in glioma neurosphere lines, and found that it does. We next explored a therapeutic regimen using the late stage autophagy inhibitor, chloroquine (CQ), in conjunction with MRK003, and showed that the combination induced apoptosis and more effectively decreased cell growth, proliferation and clonogenicity.

MATERIALS AND METHODS

Cell culture conditions and drug preparation

HSR‐GBM1 and JHH520 glioblastoma neurosphere lines were used, and their identity was authenticated at the Johns Hopkins Core Laboratory through short tandem repeat (STR) analysis. HSR‐GBM1 was a kind gift from Dr. Angelo Vescovi, JHH520 from Dr. Gregory Riggins. The HSR‐GBM1 is completely methylated at the MGMT locus but resistant to temozolomide in vitro, lacks mutation at the IDH1 and alterations in p53. For treatment studies, neurospheres were disassociated into single cell suspensions using Accutase (Sigma Aldrich, St Louis, MO, USA), counted, plated and allowed to grow overnight in Neural Stem Cell medium supplemented with 20 ng/mL of human epidermal growth factor and 10 ng/mL of human fibroblast growth factor (Peprotech, Rocky Hill, NJ, USA). All cells were grown in a humidified atmosphere containing 5% CO₂ at 37°C. The next morning, GSI MRK003, generously provided by Merck Research Laboratories (Boston, MA, USA), dissolved in dimethyl sulfoxide (DMSO) (Sigma Aldrich, St Louis, MO, USA) at final concentrations of 0.5 to 5 μM, chloroquine diphosphate (CQ) (Sigma Aldrich, St Louis, MO, USA), dissolved in phosphate buffered saline (PBS) at a final concentration of 10 μM, or DMSO (1:1000 dilution) was added to the media. Bafilomycin A1 (BafA1) (Sigma‐Aldrich, St Louis, MO, USA) was dissolved in DMSO at final concentrations of 0.5–5 nM. 3‐methyladenine (3‐MA) (Selleckchem, Houston, TX, USA) was diluted in DMEM at a final concentration of 10 mM.

Quantitative real‐time PCR analyses

Cells were collected after treatment with 0–5 µL of MRK003 for 48 h. RNA extracted was performed using the RNeasy Mini Kit (Quiagen) with on‐column DNA digestion to prevent amplification of genomic DNA. Reverse trancription into complementary DNA (cDNA) was performed. cDNA levels were analyzed by quantitative real‐time PCR analysis with SsoAdvanced Universal SYBR Green Supermix (Bio‐Rad, Hercules, CA, USA) on an iQ5 Multicolor real‐time PCR detection system. Standard curves were used to determine expression levels and all values were normalized to β‐actin. Statistical comparisons were between at least three biological replicates, each with triplicate technical replicates. Primer sequences were as follows: HES1 forward F: 5′‐GTG‐AAG‐CAC‐CTC‐CGG‐AAC‐3′, reverse R: 5′‐CGT‐TCA‐TGC‐ACT‐CGC‐TGA‐3′; HES5 forward F: 5′‐GTG‐CCT‐CCA‐CTA‐TGA‐TCC‐TTA‐AA‐3′, reverse R: 5′‐AGT‐ACA‐AAG‐TCG‐TGC‐CCA‐CA‐3′; HEY1 forward F: 5′‐TCT‐GAG‐CTG‐AGA‐AGG‐CTG‐GT‐3′, reverse R: 5′‐CGA‐AAT‐CCC‐AAA‐CTC‐CGA‐TA‐3′; β‐actin forward F: 5′‐CCC‐AGC‐ACA‐ATG‐AAG‐ATC‐AA‐3′, reverse R: 5′‐CGA‐TCC‐ACA‐CGG‐AGT‐ACT‐TG‐ 3′.

Western blot analysis

Proteins were extracted from HSR‐GBM1 and JHH520 after treatment with MRK003 and/or CQ for 48–72 h, unless otherwise indicated. The cells were lysed in RIPA buffer containing protease inhibitor cocktail (Sigma‐Aldrich, St Louis, MO, USA) and phosphatase inhibitors on ice for 10 minutes then centrifuged at 4°C and 14,000 rpm for 5 minutes to remove cellular debris. Protein quantification was done using the Bio‐Rad protein assay (Bio‐Rad, Hercules, CA, USA). Subsequently, protein separation was performed on 12% (for LC3B protein analysis) or 4%–12% (for analysis of other proteins) SDS‐polyacrylamide gel, and transferred onto nitrocellulose membranes (Invitrogen, Carlsbad, CA, USA). Membranes were then probed with primary antibodies at 4°C overnight. The following primary antibodies were used: LC3B, P62/SQSTM1, phospho‐AKT (Ser 473), AKT (C6E7), cleaved PARP, cleaved NOTCH1 (Val1744)(D3B8) (1:1000 dilution, Cell Signaling, Danvers, MA, USA), HES1 (1:800 dilution, Aviva Systems Biology, San Diego, CA, USA), β‐actin (1:5000 dilution, Sigma‐Aldrich, St Louis, MO, USA). Secondary antibodies conjugated to horseradish peroxidase (KPL, Gaitersberg, MD, USA) were incubated for 1 h at room temperature and detected with a Western Lightning Plus ECL chemiluminescent kit (PerkinElmer, Waltham, MA). All blots were repeated at least three times and representative blots are shown. Densitometry was performed using Image J Ver. 1.440 software (http://rsb.info.nih.gov/ij/) 31.

Immunofluorescence and immunohistochemical analyses

HSR‐GBM1 cells were treated with DMSO or 5 µM MRK003 for 48 h, or starved for 4 h in Earle's Basic Salt Solution (EBSS). The cells were subsequently pelleted, fixed in 10% formalin and embedded in paraffin. Antigen retrieval was performed by heating in a pressure cooker for 30 minutes in Target Retrieval Solution (Dako, Glostrup, Denmark). Sections were stained using anti‐LC3B rabbit polyclonal antibodies (1:10,000 dilution, ab51520, Abcam, Cambridge, UK and 1:500 dilution, Cell Signaling, Danvers, MA, USA) at 4°C overnight. The sections were stained by avidin‐biotin‐peroxidase complex method (Vector, Burlingame, CA, USA) with secondary antibody incubation for 30 minutes at a dilution of 1:5000, diaminobenzidine as the chromogen and counterstained with hematoxylin.

For immunofluorescence staining, HSR‐GBM1 and JHH520 cells were grown in 10% FBS DMEM in 12 well plates on coverslips. From the next day, the cells were treated with DMSO or 5 µM MRK003 for 48 h. The cells were subsequently fixed in 4% paraformaldehyde, permeabilized in ice cold (−20°C) methanol for 15 minutes, blocked for 30 minutes with 5% normal goat serum (NGS) with 0.3% Triton X solution. Sections were stained using anti‐LC3B rabbit monoclonal antibody (1:2000 dilution, ab51520, Abcam, Cambridge, UK), which was added in 5% NGS with 0.3% Triton X solution at 4°C overnight. After washing with PBS, cells were incubated for 90 minutes at a dilution of 1:400 in Cy‐3 (Jackson Immunoresearch Laboratories Inc., West Grove, PA, USA) conjugated secondary antibody. After washing with PBS, nuclei were counterstained with DAPI 4′, 6‐diamindino‐2‐phenylindole and coverslips were mounted on slides with antifade (Vectastain Elite, Vector Laboratories, Burlingame, CA, USA). Confocal imaging was done using the Fluoview 1000 (Olympus America, Melville, NY, USA).

Determination of cell growth

3‐(4,5‐dimethylthiazol‐2‐yl)−5‐(3‐carboxymethoxyphenyl)−2‐(4‐sulfophenyl)‐2H‐tetrazolium (MTS) assays were performed to determine growth in viable cell mass. Cells were dissociated and seeded into 96‐well plates at a density of 5000 per well in 200 µL of medium and treated with DMSO, 5 µM MRK003 and/or 10 µM CQ and incubated in 5% CO₂ at 37°C. For the plate readings, 20 µL of MTS solution was added to each well after 0–96 h postplating and incubated for 1 h, protected from light. The optical density at 490 nm was subsequently measured by spectrophotometer. The experiments were repeated at least three times for each cell line.

Cell proliferation assay

Cells were treated with DMSO, 5 µM MRK003 and/or 10 µM CQ for 72 h. Cells were subsequently fixed and stained for Ki67 per manufacturer's instructions using the Muse Ki67 Proliferation Assay (Millipore, Billerica, MA, USA). Cells incubated with IgG1 control in parallel were used to set the gates. The percentage of Ki67 positive cells was determined using the Muse Cell Analyzer. Analyses were repeated at least three separate times for each cell line.

Cell cycle analyses

Cells were treated with DMSO, 5 µM MRK003 and/or 10 µM CQ for 72 h. Cells were then fixed with 70% ethanol for at least 3 h and stained with Muse Cell Cycle Reagent. Cell cycle analyses were performed using the Muse Cell Analyzer and percentages of sub‐G1, G0/G1, S and G2/M populations were determined using FlowJo software. Analyses were repeated at least three separate times for each cell line.

Apoptosis assay

Apoptosis was measured using the Muse Annexin V and Dead Cell Assay per the manufacturer's protocol. Neurospheres were treated with DMSO, 5 µM MRK003 and/or 10 µM CQ for 72 h. Cells were resuspended in DMEM/F12 without phenol red with 1% Bovine Serum Albumin (BSA) before analysis. Percentage of apoptotic and dead cells were analyzed using the Muse Cell Analyzer. Analyses were repeated at least three separate times for each cell line.

Clonogenic assays

Neurosphere media was mixed 3:1 with 4% agarose (Invitrogen, Carlsbad, CA, USA) to make the bottom layer, which was used to coat each well of a six‐well plate. Neurospheres treated with DMSO, 2 µM MRK003 or a combination of 2 µM MRK003 and 10 µM CQ for 72 h were dissociated and added to 7:1 media and agarose mixture at a concentration of 5000 cells per 2 mL to make the top layer. After the agarose gelled, 2 mL of fresh media and DMSO or drug were added to the top of each well. Media and drug were changed every 1–4 days during the growth phase of the clonogenic assay. The cells were allowed to form colonies for 19–23 days. Colonies were visualized by staining with nitroblue tetrazolium (NBT) in a tissue culture incubator overnight at 37°C after which they were imaged. Colonies greater than 50 µm in diameter were scanned and counted using MCID Elite software (Cambridge, England, UK). For each well, at least three images were scanned and counted. Triplicates of each condition were plated and the experiment was repeated three times.

Statistical analyses

Differences between three or more groups were assessed using one way ANOVA test with post hoc Tukey's multiple comparison test. Error bars represent standard error of means. All statistical tests were performed using the GraphPad Prism 6 software (GraphPad Software, La Jolla, CA, USA). P values less than 0.05 were considered statistically significant.

RESULTS

MRK003 inhibits Notch pathway targets

Notch pathway activity was assessed using quantitative RT PCR to measure levels of the canonical Notch pathway transcriptional targets HES1, HES5 and HEY1 after 48‐h treatment of glioblastoma neurospheres with MRK003. A statistically significant, dose dependent, up to 50% reduction of HES1 was seen after treatment with 5 µM MRK003. Likewise, an approximately 75% decrease of HES5 and approximately 80% decrease of HEY1 was seen after 2 µM MRK003 treatment in both cell lines (Figure 1A). Interestingly, the treatment with MRK003 at a high dose (5 µM) resulted in a parodoxical increase in the mRNA levels of HES5 and HEY1. We have previously observed this paradoxical increase when blocking Notch activity in brain tumors with relatively high concentrations of GSI or arsenic trioxide, and speculated a feedback or resistance mechanism of cells surviving maximal therapy [6 and unpublished data]. We next assessed the effects of MRK003 on Notch pathway activity at the protein level. Treatment with 0.5 µM MRK003 almost completely inhibited cleaved (active) Notch1 expression in JHH520 and HSR‐GBM1; inhibition of HES1 was also observed after treatment with 5 µM MRK003 in both neurosphere lines (Figure 1B).

Figure 1.

MRK003 inhibits Notch pathway. A. Treatment with MRK003 inhibited downstream Notch pathway transcription factors HES1, HES5, HEY1 in JHH520 and HSR‐GBM1 lines at the mRNA level. A statistically significant up to 50% reduction of HES1 was seen after treatment with 5 μM MRK003. Likewise, an approximately 75% decrease of HES5 and approximately 80% decrease of HEY1 compared to DMSO control was seen after 2 μM MRK003 treatment in both lines. Stars denote P values vs. DMSO. B. Treatment with 0.5 μM MRK003 almost completely inhibited cleaved (active) Notch1 expression in JHH520 and HSR‐GBM1 lines. Inhibition of HES1 was observed after treatment with 5 μM MRK003 in both cell lines.

Notch inhibition induces autophagy in glioblastoma

Induction of autophagy, evidenced by a dose‐dependent increase in LC3B‐II/LC3B‐I ratio, was seen after 48‐h treatment with 0.5–5 µM of MRK003 in both glioblastoma neurosphere lines (Figure 2A,B). In mature autophagosomes, LC3B‐I undergoes a post‐translational modification to LC3B‐II increasing its hydrophobic nature resulting in further migration on electrophoresis. A significant (P < 0.05) increase in LC3B‐II/LC3B‐I ratio averaging over threefold was seen after 5 µM treatment with MRK003 in JHH520; likewise, over fourfold (P < 0.05) and sixfold (P = 0.0008) increases were seen after 2 and 5 µM treatment in HSR‐GBM1 (Figure 2B).

Figure 2.

Autophagy is induced after MRK003 treatment in vitro. A. Western blot showed induction of autophagy as seen by the increase of the LC3B‐II band after GSI treatment in JHH520 and HSR‐GBM1 lines in a dose‐dependent manner. B. Quantification of LC3B‐II/LC3B‐I from three separate blots. C. HSR‐GBM1 cells were treated with DMSO, 5 μM MRK003, or EBSS, then pelleted and stained for LC3B. Note the punctate staining of the cytoplasm of glioma cells after treatment with MRK003 and EBSS, but not DMSO (original magnification all ×400). D. HSR‐GBM1 and JHH520 cells were treated with DMSO or 5 μM MRK003 and stained for LC3B. Note the markedly increased punctate staining in the cytoplasm of glioma cells after treatment with MRK003 compared to DMSO by immunofluorescence analysis of LC3B (original magnification all ×640). E. Accumulation (loss of clearance) of LC3B‐II after treatment with chloroquine (CQ). A decrease in p62 after MRK003 treatment alone and increase of p62 was observed after combined treatment with MRK003 and CQ, suggesting inhibition of autophagy by CQ.

Having observed an induction of autophagy after treatment with MRK003 by western blotting, we next sought to determine if induction of autophagy could be confirmed by immunohistochemical analysis in HSR‐GBM1 cells and by immunoflourescence analysis in both cell lines. Increased punctate staining of LC3B in the cytoplasm of HSR‐GBM1 cells, suggesting mature autophagosomes, was observed after 48 h treatment with 5 µM MRK003 and 4 h starvation in EBSS media, but not in DMSO control (Figure 2C). Immunoflourescence analysis showed markedly increased punctate staining for LC3B in the cytoplasm of tumor cells compared to DMSO control in both cell lines (Figure 2D).

Given the induction of potentially protective autophagy after treatment with MRK003, a combinatorial regimen using the late stage autophagy inhibitor CQ in conjunction with MRK003 was explored. CQ blocks LC3B clearance via lysosomal inhibition, causing its accumulation in tumor cells 1. Consistent with this mechanism, we saw a very strong induction of the LC3B‐II band after CQ treatment in both lines, indicating in this context a blockage of late autophagy (Figure 2E). We chose a dosage of 5 µM MRK003 for combination experiments, because maximal suppression of HES1 protein and induction of autophagy was seen at this level. An increase in LC3B‐II/LC3B‐I ratio was also seen after treatment with the two drugs in combination (Figure 2E). To help determine if this reflected an ongoing increase in autophagy or a late blockade of the process, we assessed expression of p62, which recognizes autophagic cargo and mediates formation of selective autophagy 27. p62 is known to be upregulated when autophagy is inhibited, resulting in intracellular waste accumulation. A 40% decrease in p62 was seen after induction of autophagy by MRK003 and was subsequently increased 2.4 fold after treatment with both 5 µM MRK003 and 10 µM CQ in HSR‐GBM1, suggesting that the dominant process when both agents are used is inhibition of autophagy (Figure 2E). Interestingly, p62 was decreased by only 20% in JHH520 after treatment with 5 µM MRK003, and the increase in p62 after combination MRK003 and CQ treatment was a modest 1.8‐fold compared to DMSO control (Figure 2E). Likewise, a similar lack of p62 decrease after MRK003 treatment and only modest (1.6‐fold compared to 3.9‐fold for HSR‐GBM1) increase of p62 was seen after combination MRK003 and BafA1 treatment in JHH520 (Supporting Information Figure 1B), suggesting a cell line dependent variability in effects.

Combination treatment with MRK003 and chloroquine inhibits growth and proliferation

The effects of combination treatment with MRK003 and CQ on growth of the two glioblastoma neurosphere lines were examined using the MTS assay. In both lines, 10 µM CQ did not have a significant effect on the viable cell mass, while Notch inhibition caused a moderate slowing (P < 0.05) in the glioma growth (Figure 3A). In both cell lines, culture growth was significantly reduced compared to DMSO controls (JHH520; P = 0.0004, HSR‐GBM1; P = 0.0002) and MRK003 alone (JHH520; P < 0.01, HSR‐GBM1; P < 0.05) when MRK003 treatment was paired with inhibition of autophagy (Figure 3A).

Figure 3.

Combination treatment with MRK003 and chloroquine more effectively inhibits glioma neurosphere growth and proliferation. A. 5,000 JHH520 and HSR‐GBM1 cells per well were plated in a 96‐well plate and cell viability was assessed by MTS assay after treatment with DMSO, 5 μM MRK003, 10 μM CQ, and a combination of 5 μM MRK003 and 10 μM CQ at days 0–5 of treatment. Combination treatment markedly reduced growth compared to DMSO (JHH520; P = 0.0004, HSR‐GBM1; P = 0.0002) and MRK003 alone (JHH520; P < 0.01, HSR‐GBM1; P < 0.05) in both neurosphere lines. Stars denote P values vs. DMSO unless otherwise specified. B. Ki67 MUSE assays were performed to assess effects on proliferation. A modest but significant decrease in the percentage of Ki67‐positive cells was observed after combination treatment compared to DMSO control in the HSR‐GBM1 line.

We sought to determine whether combination treatment of MRK003 with another late stage autophagy inhibitor, BafA1, would exhibit increased cytotoxicity. Combination treatment with 2 µM of MRK003 and 0.5 µM of BafA1 showed a drastic reduction of cell growth compared to DMSO (JHH520; P < 0.0001, HSR‐GBM1; P < 0.001) (Supporting Information Figure 1A) after 5 days of treatment. However, while the addition of BafA1 to 2 µM MRK003 resulted in fewer viable cells, this increase over the GSI alone was not statistically significant, perhaps because the antigrowth effects of MRK003 were particularly prominent in these studies. Similar effects were seen with combination of 2 µM MRK003 and 5 µm of BafA1 (data not shown). However, no reduction in p62 was noted in JHH520 cells after MRK003 treatment. Increases in LC3B‐II/LCB‐I ratio and p62 were seen after combination treatment with MRK003 and BafA1, suggesting inhibitory effects of BafA1 at the late stage of autophagy similar to CQ (Supporting Information Figure 1B).

A combination of 2 µM MRK003 and 10 mM of the early stage autophagy inhibitor, 3‐MA, showed significant reduction of cell growth compared to DMSO control (JHH520; P = 0.002, HSR‐GBM1; P = 0.002) and MRK003 alone (JHH520; P = 0.03, HSR‐GBM1; P = 0.0009) after 5 days of treatment (data not shown). Western blots showed a 50% decrease in p62 expression compared to DMSO after MRK003 treatment and a 3.2‐fold increase in p62 after combination MRK003 and 3‐MA treatment in HSR‐GBM1 (Supporting Information Figure 2). A 40% reduction in LC3B‐II/I ratio was seen after 10 mM 3‐MA treatment compared to DMSO control and 30% decrease in LC3BII/I ratio was seen after combination MRK003 and 3‐MA compared to MRK003 alone, suggesting the inhibition of early stage autophagy.

To determine if combination treatment altered the proliferation of glioblastoma neurospheres, the percentage of Ki67‐positive cells was assessed using the MUSE Ki67 assay. A modest but significant reduction in the percentage of Ki67‐positive cells was observed after combination treatment with 5 µM MRK003 and 10 µM CQ compared to DMSO control in the HSR‐GBM1 line (P < 0.05) (Figure 3B). However, this reduction in proliferation following combination therapy was not significantly greater than noted after Notch inhibition alone, suggesting that it was not the only mechanism by which overall growth was affected.

Combination treatment with MRK003 and chloroquine induces apoptosis

Given the limited effect of combination treatment on proliferation, we examined changes in apoptosis. An approximately fourfold induction in the apoptotic mediator cleaved PARP was noted after 5 µM MRK003 treatment in both lines, with a lesser increase seen after phamacological inhibition of autophagy (Figure 4A). In contrast, a dramatic 12‐fold to 21‐fold induction was seen after combination treatment. Only a minor, nonsignificant increase was observed in the percentage of sub‐G1 population by cell cycle analysis after treatment with 5 µM of MRK003 compared to DMSO. In contrast, an 8.9‐fold to 10.6‐fold increase in percentage of sub‐G1 population was observed after combination treatment (Figure 4B, Supporting Information Figure 2A,B).

Figure 4.

Induction of apoptosis is observed after combination treatment with MRK003 and chloroquine. A. Western blot showed approximately fourfold increase of cleaved PARP after 5 μM MRK003 treatment compared to DMSO control in both cell lines and 12.7‐fold to 21.1‐fold increase after combination treatment compared to DMSO control. B. Cell cycle analysis and C. Annexin V MUSE assays were performed to assess effects on apoptosis. An approximately 10‐fold increase in percentage of sub‐G1 population and modest but significant increase in percentage of Annexin V‐positive cells were seen after combination treatment compared to DMSO control.

With respect to other cell cycle alterations, no significant change in percentage of G0/G1, S, or G2/M population was noted after 72‐h combination treatment of JHH520; a modest reduction of percentage of G0/G1 population was noted after 72‐h combination treatment of HSR‐GBM1 (Supporting Information Figure 2C). Although cell cycle arrest was not noted after 72‐h combination treatment, G2/M arrest was seen after 48‐h combination treatment in line (data not shown), suggesting that cells in G2/M arrest eventually undergo apoptosis. Finally, a significant but less pronounced increase in the percentage of total apoptotic cells was observed after combination treatment compared to DMSO control in both lines using the Annexin V assay (JHH520; P < 0.05, HSR‐GBM1; P < 0.01) (Figure 4C). These data support the concept that induction of autophagy prevents apoptosis following Notch blockade in glioma cells, and that CQ can suppress this and increase the cytotoxicity of GSI.

Clonogenic capacity is reduced after combination treatment with MRK003 and chloroquine

Having confirmed the increase of apoptosis caused by inhibition of autophagy by CQ, we next looked at clonogenic capacity after monotherapy and combination treatment using the soft agar assay. These were performed using 2 µM rather than 5 µM MRK003, as the higher GSI dose resulted in too few colonies for an additive effect of the autophagy inhibitor to be examined. A representative experiment shows an approximately 40%–50% decrease in the number of colonies after treatment with 2 µM MRK003 alone compared to DMSO in both cell lines (Figure 5A,B). However, with combination therapy a 75%–85% reduction in clonogenicity was observed, and this was significantly more pronounced than the effect with 2 µM MRK003 alone (both P < 0.0001). The clonogenic assays were repeated three times for each cell line, and all of these experiments showed a significantly lower number of colonies for combination treatment than the GSI alone.

Figure 5.

Combination treatment with MRK003 and chloroquine reduces clonogenicity in glioblastoma neurosphere lines. A. Representative images of nitroblue tetrazolium‐stained colonies formed by JHH520 and HSR‐GBM1 cells after treatment with DMSO, 2 μM MRK003, 10 μM CQ and combination MRK003 and CQ are shown. B. Quantification of number of colonies depicted in A, showing markedly reduced colony formation after combination therapy.

Inhibition of AKT following Notch inhibition

AKT activity is known to suppress autophagy 39, and is also modulated by Notch in cancers other than glioma 7, 24. We have previously shown that GSIs other than MRK003 can inhibit phosphorylation and activation of AKT 9 in glioblastoma. Using MRK003, we found that phospho‐AKT levels decreased dramatically 48 h after Notch inhibition at concentrations as low as 0.5 uM in HSR‐GBM1 neurospheres (Figure 6A,B). JHH520 cells showed modest inhibition of AKT activity with greatest inhibition after 24 h treatment with 5 µM MRK003 (Figure 6C). However, further investigation is needed to elucidate whether autophagy after MRK003 treatment is mediated by mTORC2 inhibition.

Figure 6.

Phospho‐AKT is inhibited by MRK003 and MRK003 induces autophagy in vivo. A, B. Phospho‐AKT (Ser 473) was significantly inhibited by MRK003 after treatment with 0.5–5 μM MRK003 in HSR‐GBM1 for 48 h (P ≤ 0.0001). However, inhibition of phospho‐AKT was not seen in JHH520 after 5 μM MRK003 treatment for 48 h. C. Phospho‐AKT was most inhibited by 5 μM MRK003 treatment for 24 h, but subsequently upregulated after 48–72 h treatment in JHH520 line.

DISCUSSION

Despite some recent advances in treatment, the prognosis for glioblastoma patients remains dismal. Induction of autophagy has been implicated as a potential resistance mechanism to general treatments in gliomas, and its manipulation has actively been investigated as a therapeutic strategy 15. However, little is known about the role of Notch in regulating autophagy in gliomas, or the potential for autophagy to mediate resistance to pathway inhibitors such as GSIs. Increasing evidence suggests that the Notch pathway affects autophagosomal trafficking 2, and that inhibition of the Notch pathway induces autophagy in some types of cancer 12 and during differentiation of stem cells 32.

Our data provide initial evidence that pharmacological Notch pathway blockade can induce autophagy in gliomas. When neurosphere lines were treated with GSI, a dramatic several fold induction of the LC3B‐II protein compared to LC3B‐I was noted, as well as the emergence of punctate immunohistochemical LC3B staining in fixed cells, consistent with an increase in autophagic flux. However, a robust decrease in p62 following Notch blockade was only seen in one line, thus, clear induction of autophagy may be restricted to a subset of cases. Autophagy may also be present in a more complex in vivo treatment milieu, as analysis of orthotopic glioblastoma xenografts treated with the GSI from a prior study 3 has revealed focal punctate LC3B staining, but too little material was present for definitive conclusions to be drawn. Our findings are consistent with a very recent report demonstrating induction of autophagy after NOTCH1 knockdown by shRNA in a single glioma cell line 40.

The induction of autophagy is of potential clinical relevance, as combining the late stage autophagy inhibitor CQ with MRK003 resulted in a significant induction in apoptosis and reduction in growth and clonogenicity as compared to Notch inhibition alone. Similar growth changes were noted when combining Notch inhibition with the early stage autophagy inhibitor 3‐MA. A few autophagy inhibitors have been clinically tested in gliomas. CQ is an antimalarial drug with lysosomotropic properties that inhibit lysosomal function. It is known to have central nervous system penetration and has been shown to effectively inhibit autophagy in vitro 28. A previous trial using radiation and carmustine with or without CQ in glioblastoma patients showed a trend toward increased median overall survival that was not significant 18. Hydroxychloroquine (HCQ), a derivative of CQ, shows equal potency to inhibit autophagy in vitro and has demonstrated less retinal toxicity. A trial of HCQ in conjunction with radiation therapy and temozolomide showed biological response caused by HCQ, but failed to improve overall survival in glioblastoma 28. New and more potent inhibitors of autophagy such as Lys05 20 and VATG‐027 11 have been tested preclinically and await clinical validation.

We have demonstrated that pharmacological Notch blockade can induce protective autophagy in glioma neurospheres, which can be abrogated by combination treatment with CQ. Together, the two drugs efficiently induce apoptosis and decrease cell growth, proliferation and clonogenicity. A range of options for pharmacological inhibition of both Notch and autophagy exist, and combinatorial blockade of these two pathways represents a promising new treatment strategy for glioblastoma.

Supporting information

Additional Supporting Information may be found in the online version of this article at the publisher's web‐site:

Figure 1. A Combination treatment with 2 µM MRK003 and Bafilomycin A1 (BafA1) significantly decreased cell growth compared to DMSO in both cell lines. B Increase in LC3B‐II/LC3B‐I ratio expression and marked elevation of p62 is noted after treatment with BafA1, suggesting late stage inhibition of autophagy. Conversely, LC3B‐II/LC3B‐I ratio decreases and p62 increases after treatment with 3‐MA, suggesting early stage autophagy inhibition.

Figure 2. A, B Representative cell cycle analysis data depicting marked increases in percentages of sub‐G1 population in JHH520 (A) and HSR‐GBM1 (B) lines after combination treatment with MRK003 and CQ. C Cell cycle arrest was not observed after 72‐hour treatment with MRK003 and CQ.

Supporting Information