Combined inhibition of RAC1 and Bcl-2/Bcl-xL synergistically induces glioblastoma cell death through down-regulation of the Usp9X/Mcl-1 axis

Abstract

Purpose

Anti-apoptotic and pro-migratory phenotypes are hallmarks of neoplastic diseases, including primary brain malignancies. In this work, we examined whether reprogramming of the apoptotic and migratory machineries through a multi-targeting approach would induce enhanced cell death and enhanced inhibition of the migratory capacity of glioblastoma cells.

Methods

Preclinical testing and molecular analyses of combined inhibition of Bcl-2/Bcl-xL and RAC1 were performed in established, primary cultured and stem-like glioblastoma cell systems.

Results

We found that the combined inhibition of Bcl-2/Bcl-xL and RAC1 resulted in synergistic pro-apoptotic and anti-migratory effects in a broad range of different glioblastoma cells. At the molecular level, we found that RAC1 inhibition led to a decreased expression of the deubiquitinase Usp9X, followed by a decreased stability of Mcl-1. We also found that the combined inhibition led to a significantly decreased migratory activity and that tumor formation of glioblastoma cells on chorion allantoic membranes of chicken embryos was markedly impaired following the combined inhibition.

Conclusions

Our data indicate that concomitant inhibition of RAC1 and Bcl-2/Bcl-xL induces pro-apoptotic and anti-migratory glioblastoma phenotypes as well as synergistic anti-neoplastic activities. The clinical efficacy of this inhibitory therapeutic strategy warrants further evaluation.

Electronic supplementary material

The online version of this article (10.1007/s13402-019-00425-3) contains supplementary material, which is available to authorized users.

Introduction

Anti-apoptotic and pro-migratory phenotypes are typical hallmarks of cancer cells and are critical to therapy resistance [1]. Vast genetic heterogeneities and the selection of sub-clones during therapy mark the plasticity of cancer [2]. Therefore, the simultaneous targeting of multiple features, such as cells that are primarily in a proliferation mode and cells that are primarily in a migration mode, as well as cells that are transitioning between these modes, seems a logical strategy to combat therapy resistance [3].

Glioblastoma represents one such extremely adaptive and, as a consequence, highly therapy resistant disease, which translates into a very poor prognosis, i.e., a 1–2 year survival rate after diagnosis [4]. The identification of cancer-promoting factors and the subsequent development of targeted therapies has been regarded as a promising strategy. However, in glioblastoma, this “magic bullet” approach has failed so far due to mechanisms of resistance tied to, for instance, its vast intra-tumor heterogeneity [5]. Based on this notion, multi-targeting has evolved as an approach to face this resistance.

Anti-apoptotic Bcl-2 family proteins such as Bcl-2 or Bcl-xL are important regulators of the apoptotic machinery and have been found to be overexpressed in glioblastoma [6]. These proteins bind pro-apoptotic members of the Bcl-2 family, including BAX and/or BAK, which are required to induce apoptosis [7–9]. As a result, the homeostasis of glioblastoma cells is shifted towards an anti-apoptotic phenotype. BH3 mimetics such as ABT263 are small molecules that interfere with the function of anti-apoptotic Bcl-2 family proteins by inhibiting the sequestration of the pro-apoptotic multi-effector proteins BAX and BAK [10, 11]. As a consequence, treatment with BH3 mimetics induces a pro-apoptotic cellular phenotype.

RAC1 belongs to the rho family of GTPases and is activated by tyrosine kinase or G protein-coupled receptors through the catalytical exchange of GDP for GTP to generate an active GTP-bound form [12–14]. Dysregulation of RAC1 has been observed in different malignant diseases, including cancers of the breast, lung and colon [15–17]. In addition, RAC1 has been found to modulate various cellular processes, including migration, invasion and malignant transformation, which renders it into an interesting candidate to be considered as a target for cancer therapy [18–20].

In this study, we show that dual targeting of Bcl-2/Bcl-xL and RAC1 restores pro-apoptotic and anti-migratory phenotypes in glioblastoma cells, leading to synergistic glioblastoma cell killing and impairment of migration, as well as impairment of three-dimensional tumor formation. At the molecular level, our data show that RAC1 inhibition leads to down-regulation of the deubiquitinase Usp9X and a subsequent decrease in the stability of Mcl-1.

Materials and methods

Reagents

ABT263 and ABT199 were purchased from Selleckchem (Houston, TX, USA). NSC23766 was purchased from Calbiochem (EMD Chemicals, Inc., San Diego, CA, USA). For ABT263 and ABT199 10 mM stock solutions were prepared using dimethylsulfoxide (DMSO). For NSC23766 a 50 mM stock solution was prepared using sterile water. All stock solutions were stored at −20 °C. In all experiments, the final concentration of DMSO was below 0.1% (v/v).

Cells and growth conditions

U87MG and A172 human glioblastoma-derived cell lines were obtained from LGC Standards (Wesel, Germany; 02/16/2011). T98G and U251 human glioblastoma-derived cell lines were purchased from Sigma Aldrich (St. Louis, MO, USA). T98G cells were obtained at 06/02/2016 and U251 cells were obtained at 01/19/2017. All purchased cell lines were validated for authenticity by short tandem repeat profiling and underwent a maximum of 40 passages before re-thawing from validated initial stocks. ULM-GBM-PC38 and ULM-GBM-PC40 are primary cultured human glioblastoma cells derived from tumor resections performed at our institution and were generated as previously described [20, 21]. ULM-GBM-SC35 and ULM-GBM-SC40 are primary neurosphere stem-like glioma cells that were also derived and characterized at our institute [22, 23]. All cells were cultured as previously described [24, 25]. Patient’s or next of kin’s informed consent was obtained, and procedures were used in accordance with the local ethics committee (approved ethics proposal No.162/10, University of Ulm).

Cell viability assay

In order to examine cellular viabilities, 3-[4, 5-dimethylthiazol-2-yl]-2, 5-diphenyltetrazolium bromide (MTT)-based assays were performed as described before [20].

Apoptosis and mitochondrial membrane potential assays

Apoptosis was detected using annexin V/propidium iodide (PI) or PI staining followed by flow cytometry as described before [26, 27]. Briefly, 3 × 10⁴ cells were seeded in duplicate in 12-well plates and allowed to attach overnight. After the respective treatments the supernatants and enzymatically detached cells (Trypsin/EDTA, Biochrom AG, Berlin, Germany) were collected. Next, the cells were centrifuged and washed twice with ice-cold annexin V binding buffer containing 140 mM NaCl, 2.5 mM Ca²⁺ and 1 M 4-(2-hydroxyethyl)piperazine-1-ethanesulfonic acid (pH 7.4) followed by resuspension in 100 μl binding buffer and incubation with 2.5 μl Annexin-V-FLUOS (Roche Diagnostics, Indianapolis, IN, USA) for 15 min at room temperature (RT). Subsequently, the cells were washed once with ice-cold annexin V binding buffer prior to resuspension in 300 μl buffer, after which PI was added to a final concentration of 2.5 μM right before performing each single measurement. For each flow cytometric analysis, 10,000 events were recorded using a FACSCanto™ II flow cytometer (BD Biosciences, NJ, USA). FlowJo software version 8.7.1 (Tree Star, Ashland, OR, USA) was used for further quantitative analyses.

PI staining was performed as described before [26, 27]. Briefly, cells were enzymatically detached and centrifuged for 5 min at 1800 rpm prior to washing twice with PBS. Then, 200 μl PI staining solution containing 0.05% trisodiumcitrate-dihydrate (Carl Roth, Karlsruhe, Germany), 0.05% triton-X100 and 0.05 mg/ml PI (Sigma Aldrich, St. Louis, MO, USA) was added and cells were incubated for 30 min at 4 °C prior to flow cytometric analysis.

To detect intrinsic apoptosis, staining with Tetramethylrhodamine ethyl ester (TMRE) was performed. Cell culture plates were centrifuged at 1300 rpm for 5 min prior to aspiration of the supernatant and addition of 200 nM TMRE (Thermo Fisher Scientific, Eugene, OR, USA) staining solution and incubation for 20 min at 37 °C. Next, the cells were enzymatically detached and centrifuged for 5 min at 1800 rpm, after which the supernatants were discarded and the cells were resuspended in 300 μl PBS followed by flowcytometric analysis. The data were analyzed using FlowJo software (version 7.6.5; Tree Star, Ashland, OR, USA).

Western blot analysis

Protein expression was determined by Western blot analysis as described before [28] using the following primary antibodies: anti-Mcl-1 (1:1000; #5453S, clone: D35A5, CST: Cell Signaling Technology, Danvers, MA, USA), anti-Bcl-2 (1:1000; #551098, BD Pharmingen, USA), anti-Bcl-xL (1:500; sc-1041, Santa Cruz Biotechnology, Inc., Santa Cruz, CA, USA or 1:500; #2764, clone 54H6, CST), anti-Usp9X (#5751S, 1:1000; CST), anti-Noxa (1:1000; #556585, clone 114C307.1, Enzo Life Sciences, Farmingdale, NY), anti-human caspase-3 (1:1000; #9662, CST), anti-human caspase-9 (1:1000; #9508 T, clone C9, CST), anti-RAC1 (1:1000; #1862341, Thermo Fisher Scientific), and anti-β-actin (1:2000, clone AC15; Sigma Aldrich, St. Louis, MO). Secondary HRP-linked antibodies were purchased from CST (#7076S, #7074S).

siRNA transfection

Mcl-1-siRNA (ON-TARGETplus MCL1 siRNA, J-004501-16-0002) and non-targeting siRNA (ON-TARGETplus Non-targeting siRNA, D-001810-03-05) were purchased from Dharmacon (Lafayette, CO, USA). RAC1-siRNA (Hs_RAC1_6 FlexiTube siRNA, #1027415) was purchased from Qiagen. Cells were incubated with the formed complexes of Viromer® BLUE (Lipocalyx, Halle, Germany) and the respective siRNAs in DMEM with 1.5% fetal bovine serum (FBS) for the whole duration of the experiment.

qRT-PCR

qRT-PCR was performed as described before [25] using the primers listed in Table 1.

Table 1.

qRT-PCR primers used

Gene	Forward sequence	Reverse sequence
Mcl-1	CCA AGA AAG CTG CAT CGA ACC AT	CAG CAC ATT CCT GAT GCC ACC T
USP9X	GTG TCA GTT CGT CTT GCT CAG C	GCT GTA ACG ACC CAC ATC CTG A
GAPDH	GTC TCC TCT GAC TTC AAC AGC G	ACC ACC CTG TTG CTG TAG CCA A
18S	AGT CCC TGC CCT TTG TAC ACA	GAT CCG AGG GCC TCA CTA AAC

Cell migration and motility assays

Cell migration was assessed using both transmembrane migration (Transwell®) and scratch wound healing assays. Cell motility was evaluated using time-lapse live cell imaging.

For the Transwell® assays, 3 × 10⁴ cells were seeded on Transwell® membranes (Corning Incorporated, Corning, NY, USA) with a pore size of 8 μm in DMEM containing 1.5% FBS, and allowed to migrate through the membranes towards medium containing 10% FBS. Experiments were carried out according to the manufacturer’s recommendations. After 24 h, the upper side of the membrane was wiped and washed with phosphate-buffered saline (PBS) three times. Next, the cells on the bottom side of the membrane were fixed with methanol and stained with 4′6-diamidino-2-phenylindole (DAPI) prior to mounting. The number of migrated cells was determined by counting one high-power field at 10x magnification in triplicate for each treatment condition.

Scratch wound healing assays were performed as previously reported [20]. In brief, sub-confluent cell layers in 12-well plates were scratched across the wells using a 200 μl pipet tip. Next, sequential microscopic images were taken at defined time points at 10x magnification, and the areas of the scratches were analyzed using NIH ImageJ software (http://imagej.nih.gov/ij).

For time-lapse live cell imaging, 4 × 10⁴ cells/well were seeded into 12-well plates after which microscopic images at 10x magnification were taken using a live-imaging inverted video microscope (Zeiss Observer.Z1, Göttingen, Germany) every 30 min for a total observation time of 24 h. During this period, cells were kept at standard culture conditions (37 °C, 5% CO₂, humidified atmosphere). Single-cell tracking was performed using the MtrackJ plugin [20] (www.imagescience.org/meijering/software/mtrackj/) for the NIH ImageJ software (http://imagej.nih.gov/ij). Normalized “wind-rose” plots were generated using the chemotaxis and migration tool from Integrated BioDiagnostics (Ibidi, Martinsried, Germany, www.ibidi.com).

Chorion allantoic membrane (CAM) assay

In order to assess putative effects of the respective compounds and their combination on tumor formation in a three-dimensional environment, we performed CAM assays as previously reported [20]. A 1:1 mixture of serum-free medium and Matrigel® (BD Biosciences, MA, USA) containing 1 × 10⁶ ULM-GBM-PC38 or 2 × 10⁶ U251 cells was seeded onto the CAM of one-week old fertilized chicken eggs. The experimental treatment was started after 24 h and consisted of local application of the respective agents twice a day. The tumors and its adjacent CAMs were harvested 4 days after seeding and embedded in paraffin. Sections were stained with hematoxylin and eosin and analyzed microscopically (Zeiss Imager.M1, Göttingen, Germany). Microphotographs were taken at 2.5x magnification. The tumor areas were quantified using NIH ImageJ software (http://imagej.nih.gov/ij).

Statistical analysis

Statistical significance was assessed by Student’s t test using Prism version 5.04 (GraphPad, La Jolla, CA). P < 0.05 was considered statistically significant. Combination indices and isobolograms were calculated using CompuSyn software (ComboSyn, Inc., Paramus, NJ, USA).

Results

Inhibition of RAC1 and Bcl-2/Bcl-xL synergistically decreases the viability of glioblastoma cells

Using in silico analyses, we found that both RAC1 and Bcl-xL are overexpressed in primary glioblastomas compared to normal brain tissue (Fig. 1a, b). Moreover, we found that overexpression of RAC1 correlated with a worse clinical outcome (Supplementary Fig. 1A). We also found that the combined inhibition of RAC1 by NSC23766 and of Bcl-2/Bcl-xL by ABT263 had a stronger cytotoxic effect on glioblastoma U251 and ULM-GBM-PC38 cells compared to untreated control or single agent treated cells (Fig. 1c, d and Supplementary Fig. 1B). To further explore whether concomitant inhibition of RAC1 and Bcl-xL synergistically reduces the viability of glioblastoma cells, a broad range of different cells was analyzed by MTT assay after which isobolograms were calculated (Fig. 1e, Tables 2 and 3). In the U251, U87MG, T98G and A172 glioblastoma-derived cell lines, as well as in the ULM-GBM-PC38 and ULM-GBM-PC40 primary cultured glioblastoma cells and the ULM-GBM-SC35 and ULM-GBM-SC38 stem cell-like glioma cells, a synergistic reduction in cellular viability was found when cells were treated with the combination of ABT263 and NSC23766 (Fig. 1e, Tables 2 and 3). The synergistic effect was, however, less strong in the stem cell-like glioma cells.

Fig. 1.

Combined inhibition of Bcl-2/Bcl-xL and RAC1 synergistically reduces the viability of glioblastoma cells. a, In silico analysis was performed to determine RAC1 mRNA expression levels in brain or tumor specimens. These levels were correlated with 5-year survival rates in the TCGA Brain 2 dataset using the Oncomine platform (www.oncomine.org). b, In silico analysis was performed to determine Bcl-xL mRNA expression levels in brain or tumor specimens in the TCGA Brain 2 dataset using the Oncomine platform (www.oncomine.org). c, Representative microphotographs of U251 glioblastoma cells treated with NSC23766 (NSC), ABT263 or its combination for 48 h at the indicated concentrations. Magnification: 40x. d, U251 and ULM-GBM-PC38 cells were treated for 72 h as indicated. MTT assays were performed to determine cellular viability. Columns: mean. Bars: SD. N = 3. Statistical significance was assessed by Student’s t test. e, Effects on cell viability following treatment with ABT263 (A) and NSC23766 (N) examined by MTT assay after 72 h treatment with single agents or its combination at the indicated concentrations under serum starvation (1.5% FBS). Normalized isobolograms were calculated using the CompuSyn software tool. The connecting line represents additivity. Data points located below the diagonal line indicate a synergistic drug-drug interaction and data points above the diagonal line indicate an antagonistic drug-drug interaction. Data are representative of three independent experiments

Table 2.

Combination indices (CI) for U251, U87MG, T98G and A172 glioblastoma cells treated with the indicated concentrations of ABT263 (ABT) and NSC23766 (NSC)

U251			U87MG			T98G			A172
ABT (μM)	NSC (μM)	CI	ABT (μM)	NSC (μM)	CI	ABT (μM)	NSC (μM)	CI	ABT (μM)	NSC (μM)	CI
5.0	40.0	0.98	3.0	20.0	0.99	3.0	20.0	0.68	6.25	50.0	1.07
2.5	20.0	0.68	3.0	1.25	0.28	3.0	1.25	0.61	6.25	3.125	0.41
2.5	5.0	1.22	1.5	10.0	0.96	1.5	2.5	0.86	3.125	25.0	0.77
1.25	10.0	0.89	1.5	2.5	0.41	0.75	5.0	0.57	3.125	6.25	0.91
0.625	20.0	0.89	0.75	5.0	0.91	0.375	2.5	0.54	1.56	12.5	1.19
0.625	5.0	0.79	0.375	2.5	0.73	0.1875	20.0	0.15	0.78	25.0	1.08
0.3125	40.0	0.88	0.1875	20.0	1.01	0.1875	1.25	0.18	0.39	50.0	0.83

Table 3.

Combination indices (CI) for ULM-GBM-PC38, ULM-GBM-PC40, ULM-GBM-SC35 and ULM-GBM-SC38 glioblastoma cells treated with the indicated concentrations of ABT263 (ABT) and NSC23766 (NSC)

ULM-GBM-PC38			ULM-GBM-PC40			ULM-GBM-SC35			ULM-GBM-SC38
ABT (μM)	NSC (μM)	CI	ABT (μM)	NSC (μM)	CI	ABT (μM)	NSC (μM)	CI	ABT (μM)	NSC (μM)	CI
6.25	25.0	1.10	6.25	25.0	0.73	6.25	25.0	0.89	6.25	25.0	0.78
6.25	1.56	0.56	6.25	1.56	0.43	3.125	12.5	1.61	6.25	1.56	1.15
3.125	12.5	0.79	3.125	12.5	0.85	1.56	6.25	1.47	3.125	12.5	1.19
1.56	6.25	1.09	1.56	6.25	1.11	0.78	12.5	0.95	3.125	3.125	1.61
0.78	12.5	0.85	0.78	12.5	0.85	0.78	3.125	1.26	1.56	6.25	2.75
0.39	25.0	0.88	0.39	25.0	0.68	0.39	25.0	0.61	0.78	12.5	0.98
0.39	1.56	1.24	0.39	1.56	1.46	0.39	1.56	0.84	0.39	25.0	0.13

Combined inhibition of RAC1 and Bcl-2/Bcl-xL results in enhanced apoptosis

Using light microscopy, we found that cells treated simultaneously with ABT263 and NSC23766 displayed typical morphological features of apoptosis, such as enhanced cellular fragmentation and blebbing (Fig. 1c). To further assess whether apoptosis represents a relevant part of the mechanism that is responsible for the inhibitory effect of the combination treatment on cellular viability, annexin V/PI staining and flow cytometric analyses were performed in U251, ULM-GBM-PC38, T98G and ULM-GBM-PC35 glioblastoma cells. We found that combined treatment with NSC23766 and ABT263 resulted in a significant increase in the fraction of annexin V-positive cells (apoptotic cells) compared to either treatment alone (Fig. 2a-c and Supplementary Fig. 1C, D). Concomitantly, we found that U251 cells treated with the combination therapy exhibited enhanced mitochondrial outer membrane permeability compared to cells treated with each compound alone (Fig. 2d and Supplementary Fig. 2A).

Fig. 2.

Combined inhibition of Bcl-2/Bcl-xL and RAC1 results in enhanced apoptosis in glioblastoma cells. a-c, U251 (a), ULM-GBM-PC38 (b) and T98G (c) glioblastoma cells were treated for 48 h (a and c) or 24 h (b) with NSC23766 (NSC) and/or ABT263 as indicated. Staining with annexin V/propidium iodide was performed prior to flow cytometric analysis. Representative flow plots are shown. Data are representative of three independent experiments. d, Representative histograms of U251 cells that were treated for 48 h with NSC23766 (NSC) and/or ABT263 as indicated prior to staining with TMRE and flow cytometric analysis. Data are representative of three independent experiments. e, U251 cells were treated for 24 h with NSC23766 (20 μM) and/or ABT263 (2 μM) under serum starvation (1.5% FBS). Whole cell extracts were subjected to Western blot analysis for caspase 9 (C9), cleaved caspase 9 (cC9), caspase 3 (C3) and cleaved caspase 3 (cC3) expression. Actin was used to confirm equal protein loading (arrowhead indicates 40 kDa band). Data are representative of two independent experiments. f, U251 cells were subjected for 48 h to the combination treatment in the presence or absence of zVAD.fmk prior to staining with propidium iodide and flow cytometric analysis. Representative histograms are shown. Data are representative of three independent experiments. g, U251 and ULM-GBM-PC38 cells were treated for 6 h or 24 h with increasing concentrations of NSC23766 under serum starvation (1.5% FBS). Whole cell extracts were subjected to Western blot analysis for Mcl-1, Bcl-2 and Bcl-xL expression. Actin served as a loading control. The Bcl-2 band in ULM-GBM-PC38 cells is marked by an arrowhead. Densitometric analysis was perfomed using ImageJ (NIH, Bethesda, MD; http://imagej.nih.gov/ij). Normalized data are presented on top of the respective Western blots. Data are representative of two independent experiments. h, U251 cells were treated either with non-targeting (n.t.)-siRNA or RAC1-siRNA for 72 h. Whole cell extracts were subjected to Western blot analysis for RAC1, Mcl-1, Bcl-2, Bcl-xL/S and Usp9X expression. Actin was used to confirm equal protein loading. Data are representative of two independent experiments

Combined inhibition of RAC1 and Bcl-2/Bcl-xL increases activation of initiator and effector caspases

To assess whether the pro-apoptotic effect of the combination treatment translates to enhanced cleavage (i.e., activation) of caspases at the molecular level, Western blot analyses were performed. We found that U251 cells exposed to the combination treatment exhibited enhanced cleavage of both caspase-9 and caspase-3 (Fig. 2e). Furthermore, we found that pan-caspase inhibition using zVAD.fmk resulted in abolishment of the anti-apoptotic response following combination treatment with ABT263 and NSC23766 (Fig. 2f and Supplementary Fig. 2B). In contrast, we found that treatment with Necrostatin, a necroptosis inhibitor, did not affect apoptosis driven by the combination therapy, which supports a specific caspase-mediated apoptotic response as underlying mechanism (Supplementary Fig. 2C, D).

Interference with RAC1 decreases expression of Mcl-1 and its interacting deubiquitinase Usp9X

We next addressed the question whether inhibition of RAC1 affects the expression of anti-apoptotic members of the Bcl-2 family of proteins. We found that treatment with NSC23766 decreased the protein levels of Mcl-1 in U251, ULM-GBM-PC38 and U87MG cells in a time- and dose-dependent manner (Fig. 2g and Supplementary Fig. 2E). In addition, we found that expression of the deubiquitinating enzyme Usp9X was markedly decreased (Fig. 3a). Other anti-apoptotic Bcl-2 family proteins, Bcl-2 and Bcl-xL, were also found to be down-regulated by NSC23766 treatment, however, in a less consistent manner in the cell lines tested. To verify that these effects were specifically related to interference with RAC1, knockdown experiments were performed. We found that specific siRNA-mediated down-regulation of RAC1 mirrored the effects of NSC23766 on expression of the Mcl-1, Bcl-2, Bcl-xL and Usp9X proteins (Fig. 2h).

Fig. 3.

Inhibition of RAC1 decreases Mcl-1 protein stability and enhances ABT263-mediated cell death. a, U251 and ULM-GBM-PC38 cells were treated with NSC23766 (U251: 20 μM, ULM-GBM-PC38: 10 μM), ABT263 (2 μM) or the combination as indicated. Whole cell extracts were subjected to Western blot analysis for Usp9X, Mcl-1, Bcl-2 and Bcl-xL expression. Actin was used to confirm equal protein loading. The data are representative of 2 independent experiments. b, U251 cells were treated for 48 h with non-targeting (n.t.)-siRNA or Mcl-1-siRNA, followed by treatment with ABT263 or solvent for 24 h. Staining with propidium iodide was performed prior to flow cytometric analysis. Histograms representative of three independent experiments are shown. c, Mcl-1 knockdown in cells treated as described for b confirmed by Western blot analysis. d, U251 cells were treated with non-targeting (n.t.)-siRNA or Bcl-xL-siRNA for 24 h followed by treatment with 20 μM NSC23766 or solvent for 24 h. Staining with annexin V/propidium iodide was performed prior to flow cytometric analysis. Columns: mean. Bars: SD. N = 3. Statistical significance was assessed by Student’s t test. e, Knockdown of Bcl-xL in cells treated as described under d confirmed by Western blot analysis. f, U251 cells were treated for 48 h with NSC23766, the selective Bcl-2 inhibitor ABT199 or its combination. Staining with propidium iodide was performed prior to flow cytometric analysis. Columns: mean. Bars: SD. N = 3. Statistical significance was assessed by Student’s t test. g, U251 and ULM-GBM-PC38 cells were treated for 6 h or 24 h with NSC23766, ABT263 or both prior to Mcl-1 qRT-PCR expression analysis. Columns: mean. Bars: SD. N = 3. Statistical significance was assessed by Student’s t test. h, U251 cells were treated with NSC23766 or solvent for 24 h before adding 10 μg/ml cycloheximide, after which lysates were subjected to Western blot analysis for Mcl-1 expression. Actin was used to confirm equal protein loading. Data are representative of two independent experiments. i, Graphical representation following densitometric analysis of the experiment described under h using ImageJ (NIH, Bethesda, MD; http://imagej.nih.gov/ij). j, ULM-GBM-PC38 cells were treated with NSC23766 or solvent for 24 h before adding 10 μg/ml cycloheximide after which lysates were subjected to Western blot analysis for Mcl-1 expression. Actin was used to confirm equal protein loading. Densitometric analysis was perfomed using ImageJ (NIH, Bethesda, MD; http://imagej.nih.gov/ij). k, U251 cells were treated for 5 h with 20 μM NSC23766 in the presence or absence of MG-132 (10 μM). Whole cell extracts were collected and Western blot analysis for Mcl-1 was performed. Actin was used as a loading control. Data are representative of two independent experiments

Combined inhibition of RAC1 and Bcl-2/Bcl-xL suppresses expression of Usp9X and counteracts ABT263-mediated up-regulation of Mcl-1

Up-regulation of Mcl-1 in response to treatment with ABT compounds such as ABT263 is a well described mechanism of resistance. Here, we found that treatment with ABT263 resulted in marked increases in Mcl-1 protein levels in U251 and ULM-GBM-PC38 cells (Fig. 3a). This effect was, however, abrogated by concomitant treatment with NSC23766 and was accompanied by suppression of Usp9X expression. Moreover, we found that the combination treatment led to a decreased expression of Bcl-xL.

Specific inhibition of Mcl-1 promotes an apoptotic response to BH3 mimetics

Since our data showed that Mcl-1 is suppressed by NSC23766, we next set out to determine whether Mcl-1 levels may affect the sensitivity of U251 glioblastoma cells towards a BH3-mimetic, ABT263. To this end, we knocked down the expression of Mcl-1 using a specific siRNA and found that Mcl-1 silencing strongly sensitized these glioblastoma cells to the cytotoxic effects of ABT263, which indicates that Mcl-1 acts as a key regulator of ABT263-mediated apoptosis in this setting (Fig. 3b, c and Supplementary Fig. 3A).

Selective interference with Bcl-xL acts synergistically when combined with RAC1 inhibition

ABT263 inhibits both Bcl-2 and Bcl-xL. To determine whether inhibition of Bcl-xL alone is sufficient to synergistically enhance apoptosis when combined with NSC23766, U251 cells were treated with Bcl-xL siRNA in the presence or absence of NSC23766. We found that siRNA-mediated knockdown of Bcl-xL resulted in a significant increase in the fraction of annexin V-positive (apoptotic) cells (Fig. 3d, e). To further address the question whether interference with Bcl-2 alone contributes to an enhanced pro-apoptotic response of the combination therapy, U251 cells were treated with the specific Bcl-2 inhibitor ABT199 in the presence or absence of NSC23766. We found that the combined treatment with ABT199 and NSC23766 did not result in an increase in the fraction of sub-G1 (apoptotic) cells (Fig. 3f and Supplementary Fig. 3B).

RAC1 inhibition decreases the stability of Mcl-1

We next set out to examine how Mcl-1 is downregulated by NSC23766. Using qRT-PCR, we found that a reduction in Mcl-1 transcripts was only noted in ULM-GBM-PC38 glioblastoma cells after a 6 h treatment with NSC23766 alone (Fig. 3g and Supplementary Fig. 3C). This effect was absent after a 24 h treatment with NSC23766. In addition, we found that the combination treatment resulted in an up-regulation of Mcl-1 transcripts at 6 h and 24 h, which was also noted in U251 cells after treatment with NSC23766 alone and when combined with ABT263.

Given the notion that our data do not point at a transcriptional mechanism, we decided to assess whether the NSC23766-mediated reduction in Mcl-1 protein is post-translationally driven. To this end, U251 and ULM-GBM-PC38 cells were treated with the protein synthesis inhibitor cycloheximide in the presence or absence of NSC23766. We found that the stability of the Mcl-1 protein was significantly decreased in the presence of the RAC1 inhibitor, suggesting that RAC1 controls Mcl-1 levels at least in part at a post-translational level (Fig. 3h-j and Supplementary Fig. 3D). This notion is supported by the observation that inhibition of the proteasome using MG-132 led to restoration of the Mcl-1 levels (Fig. 3k).

Combined inhibition of Bcl-2/Bcl-xL and RAC1 enhances the anti-migratory activity against glioblastoma cells

The pro-apoptotic effect of the combined treatment with ABT263 and NSC23766 on glioblastoma cells was accompanied by yet another relevant finding. Using microscopic imaging, we noted a reduction in cytoplasmic process formation and rounding up of cells as a common feature of cells treated with both agents. To test our hypothesis that the combination treatment induces an anti-migratory phenotype in a quantitative manner, we performed time-lapse live cell microscopy imaging.

Single-cell tracking revealed that while treatment with NSC23766 significantly reduced the random movement of U251, ULM-GBM-PC38 and A172 cells, ABT263 promoted the migration of U251 and ULM-GBM-PC38 cells, or at best showed a slight anti-migratory effect (A172 cells) (Fig. 4a-d and Supplementary Fig. 4A, B). Notably, we found that the combination treatment strongly impaired the migration of glioblastoma cells, which exceeded the effect of NSC23766 treatment alone.

Fig. 4.

Combined inhibition of Bcl-2/Bcl-xL and RAC1 results in enhanced anti-migratory activity. a-b, U251 (a) and ULM-GBM-PC38 (b) cells were seeded in 24-well plates followed by sequential microscopic imaging (magnification: 10x) over a time period of 24 h. Single-cell tracking was performed using the MtrackJ software tool (see Materials and methods). Wind-rose plots display the paths of 15 single cells per treatment condition during the 24 h observation period. The tracks were aligned to start from the same initial position to facilitate comparisons. c-d, Total distance covered by 45 cells within 24 h per treatment condition. Columns: mean; bars: SEM. Statistical significance was assessed by Student’s t test. Data are representative of 3 independent experiments. e, Monolayers of sub-confluent U251 cells were scratched prior to treatment with either solvent or NSC23766 (20 μM)/ABT263 (2 μM). Microscopic images were taken at time 0 h and 18 h after infliction of the scratch. Magnification: 10x. f, Graphical representation of the quantitative assessment of the scratch closures of cells treated as described under e after 6 h, 12 h, 18 h and 24 h compared to time 0 h. Data are presented as mean and SD. N = 3. Statistical significance was assessed by Student’s t test. ** signifies p < 0.01. g, 3 × 10⁴ cells were seeded onto a transwell® membrane and treated with ABT263, NSC23766 (NSC) or both agents. Medium containing 10% FBS served as a chemoattractant. After 24 h, cells on the upper side of the membrane were wiped off, and the cells on the lower side of the membrane were stained with DAPI prior to mounting. Representative microscopic images are shown. Magnification: 10x. h, Quantitative representation of trans-migrating ULM-GBM-PC38 cells treated as described under g. One representative high-power field (hpf) per transwell® membrane was counted (in total 3 hpf per treatment condition). Columns: mean. Bars: SD. N = 3. Statistical significance was assessed by Student’s t test

In order to further address the question whether the combination treatment also affects directed movement, a scratch-induced migration assay was performed. We found that U251 cells subjected to treatment with both agents showed a significantly reduced migration into the scratch after 12 h, 18 h and 24 h compared to untreated control cells and cells treated with either agent alone (Fig. 4e, f). After ~18 h of treatment, however, the cells started to display morphological changes such as a more pronounced rounding accompanied by the formation of blebs, which may reflect the initiation of cytocidal effects brought about by the combination treatment.

We further examined the capacity of the cells to traverse a collagen-coated porous membrane along a chemoattractant (FBS) gradient. As anticipated, treatment with both agents resulted in a significantly reduced transmigration of ULM-GBM-PC38 cells compared to control cells and cells treated with either agent alone (Fig. 4g, h). Overall, these results indicate that the combination treatment induces an anti-migratory phenotype in glioblastoma cells.

NSC23766 increases the inhibitory effect of ABT263 on tumor growth in chorioallantoic membranes

Next, we used a chorioallantoic membrane (CAM) assay to assess whether the combination treatment with ABT263 and NSC23766 elicits increased inhibitory effects on tumor formation and growth of U251 and ULM-GBM-PC38 cells in a three-dimensional “in vivo-near” setup compared to the effects of either agent alone. To this end, U251 and ULM-GBM-PC38 cells were seeded onto the CAMs of fertilized chicken eggs and allowed to spread for 24 h prior to treatment twice daily with ABT263, NSC23766, both drugs, or solvent for the following three days. By doing so, we found that the most striking phenotypic difference among the treatment groups was an enhanced reduction in size of those tumors that received the combination treatment, which is reflected by a significant decrease of the tumor area, compared to treatment with each agent alone (Fig. 5a-g). For U251, no tumor formation on the CAMs was noted in 50% of the cases following combination treatment.

Fig. 5.

Combined inhibition of Bcl-2/Bcl-xL and RAC1 inhibits tumor formation on CAMs. a-g, 1 × 10⁶ ULM-GBM-PC38 or 2 × 10⁶ U251 cells were seeded on CAMs of fertilized chicken eggs. After 24 h, the cells were treated twice daily with 10 μM ABT263 (U251), 5 μM ABT263 (ULM-GBM-PC38), 50 μM NSC23766 (U251 and ULM-GBM-PC38) or both compounds. On day 5, the tumors were harvested, and photographs were taken prior to fixation in 10% formalin and histological processing. a-b, Representative photographs of tumors located on CAMs at day 5. The circle has a diameter of 5 mm. c, Representative histological images of tumors on CAMs stained with hematoxylin and eosin (magnification: 2.5x; scale bar = 1000 μm). d-e, Photographs of tumors treated as indicated. Five to six replicates are shown for each treatment group. Dashed circles: tumor circumference. f-g, Quantitative representation of tumor areas as assessed by Image J (NIH, Bethesda, MD; http://imagej.nih.gov/ij). Data are presented as mean of at least 5 tumors per treatment group and SD. Statistical significance was assessed by Student’s t test

Discussion

Recent high resolution molecular analyses using single-cell RNA sequencing and next generation DNA sequencing have uncovered an immense complexity of the molecular makeup of solid tumors. As a consequence, single-targeted therapeutic strategies seem obsolete in many cases, as exemplified by the failure of different single EGFR or single VEGF-targeted approaches [29–31]. Here, we show that a combined inhibition of the anti-apoptotic Bcl-2 family protein Bcl-xL and the pro-migratory protein RAC1 leads to profound pro-apoptotic and anti-migratory effects. At the molecular level, we found that inhibition of RAC1 alone, and combined with Bcl-2/Bcl-xL inhibition, led to a decreased expression of Mcl-1. It has been well-documented that Mcl-1 up-regulation may confer resistance to BH3 mimetics such as ABT737 and ABT263 to cancer cells [11]. Therefore, we speculate that down-regulation of Mcl-1 following inhibition of RAC1, as observed in our study, likely constitutes at least part of the mechanism responsible for the synergistic anti-neoplastic activity of the combination treatment. In our setting, Mcl-1 expression seems to be modulated at the post-transcriptional level. Usp9X is a deubiquitinase that has been implicated before in post-transcriptional regulation of Mcl-1 [32]. Our data are the first to show that RAC1 inhibition decreases Usp9X expression, which is likely at least in part responsible for the reduced stability of Mcl-1 observed in our in vitro models and the subsequent synergistic anti-cancer activity of the combination treatment. This notion is supported by our previous findings [33, 34].

One of the key elements of our proposed therapeutic strategy is our finding that inhibition of RAC1 impairs the migratory activity of glioblastoma cells. Enhanced migration is one of the hallmarks of cancer and is a major driver of glioblastoma development, for instance in response to a hypoxic state within a tumor [35]. While in our study ABT263 had barely any impact on the migratory phenotype of glioblastoma cells, inhibition of RAC1 significantly impaired their migration. We have seen a similar response pattern in glioblastoma cells to RAC1 inhibition before, for instance when combined with HER1/EGFR-targeting agents [20]. These observations underline the potential utility of adding anti-migratory drugs to the treatment regimen of glioblastoma patients.

RAC1 is essential for the formation of filopodia and tunneling nanotubes, which are important for interactions of cancer cells with each other and with their microenvironment [36–38]. A missing connecting network may foster a state of stimulatory deprivation and likely facilitates anti-cancer responses. This assumption is supported by the fact that the anti-migratory activity induced by RAC1 inhibition was surpassed when this inhibition was combined with ABT263. Overall, these observations put emphasis on hitting cancer cells at different central junctions in order to ablate multiple core features at the same time.

Our data provide a proof of principle that combined inhibition of RAC1 and Bcl-2/Bcl-xL may be effective for the treatment of glioblastoma. However, from a translational perspective, our study still has a number of limitations. While we have shown that the combination therapy impairs the formation of tumors in an in vivo-like setting, including a human PDX model, we have not studied the therapeutic efficacy of our proposed strategy in an orthotopic model. Such studies would help to elucidate the prospect of this therapeutic approach and are, therefore, warranted. Another limitation of this study to be mentioned is that, while ABT263 has been used in clinical trials, RAC1 inhibition has not been applied to cancer therapy before. However, several RAC1 inhibiting drugs have been described and used to treat other medical conditions. Anti-inflammatory ketorolac and anti-hypertensive guanabenz are two drugs that are known to inhibit RAC1 and to penetrate the blood brain barrier, which could be used in a repurposing approach [39, 40]. Drug repurposing would also facilitate and accelerate the use of RAC1 inhibitors for the treatment of glioblastoma patients.

From a mechanistic point of view, it remains to be resolved how the RAC1 and Bcl-2/Bcl-xL combination therapy induces sustained down-regulation of Mcl-1. In our previous work, we showed that energy deprivation may cause an impairment of protein synthesis which, in turn, may affect expression of short-lived proteins such as Mcl-1 [25]. Future analyses, including the measurement of oxygen consumption (OXPHOS) and extracellular acidification (glycolysis) rates, are needed to decipher whether the combination therapy interferes with metabolic pathways. This could unveil additional metabolic vulnerabilities or salvage pathways that may be targeted by an extended combinatorial therapeutic approach to increase the impetus of this strategy.

Compliance with ethical standards

Conflict of interest

The authors declare that they have no conflict of interest.