Blockade of the forward Na+/Ca2+ exchanger suppresses the growth of glioblastoma cells through Ca2+‐mediated cell death

Background and Purpose The Na+/Ca2+ exchanger (NCX) working in either forward or reverse mode participates in maintaining intracellular Ca2+ ([Ca2+]i) homeostasis, which is essential for determining cell fate. Previously, numerous blockers targeting reverse or forward NCX have been developed and studied in ischaemic tissue injury but barely examined in glioblastoma for the purpose of anti‐tumour therapy. We assessed the effect of NCX blockers on glioblastoma growth and whether NCX can become a therapeutic target. Experimental Approach Patch‐clamp recording, Ca2+ imaging, flow cytometry, and Western blot were used to study the effects of specific and non‐specific NCX blockers on cultured glioblastoma cells. In vivo bioluminescent imaging was used to measure effects on grafted glioblastoma. Key Results Selectively blocking the reverse NCX with SEA0400, SN‐6, and YM‐244769 did not affect tumour cell viability. Blocking the forward NCX with bepridil, CB‐DMB, or KB‐R7943 elevated [Ca2+]i and killed glioblastoma cells. Bepridil and CB‐DMB caused Ca2+‐dependent cell cycle arrest together with apoptosis, which were all attenuated by a Ca2+ chelator BAPTA‐AM. Systemic administration of bepridil inhibited growth of brain‐grafted glioblastoma. Bepridil did not appear to have a cytotoxic effect on human astrocytes, which have higher functional expression of NCX than glioblastoma cells. Conclusions and Implications Low expression of the NCX makes glioblastoma cells sensitive to disturbance of [Ca2+]i. Interventions designed to block the forward NCX can cause Ca2+‐mediated injury to glioblastoma thus having therapeutic potential. Bepridil could be a lead compound for developing new anti‐tumour drugs.


| INTRODUCTION
Intracellular Ca 2+ signalling plays an important role in cell fate determination such as quiescence or proliferation, survival, or death and, therefore, is receiving increasing attention in the study of glioblastoma evolution (Leclerc et al., 2016;Morrone, Gehring, & Nicoletti, 2016).
Tumour biology studies document that glioblastoma cells release toxic concentrations of glutamate to surrounding neurons (Buckingham et al., 2011;Ye & Sontheimer, 1999). Glutamate constantly activates glutamate receptors in glioblastoma cells and causes Ca 2+ influx. If Ca 2+ extrusion mechanisms are inhibited, the intracellular Ca 2+ ([Ca 2+ ] i ) will build up in glioblastoma cells. The accumulation of [Ca 2+ ] i is a well-known step that causes ischaemic/hypoxic cell death (Dong, Saikumar, Weinberg, & Venkatachalam, 2006;Trump & Berezesky, 1995). An elevation of [Ca 2+ ] i may also be lethal to tumour cells and become a new approach for combating glioblastoma.
However, the data regarding how to induce [Ca 2+ ] i overload and Ca 2+ -dependent injury in glioblastoma are very limited (Hsu et al., 2016;Liang & Lu, 2012).
The Na + /Ca 2+ exchanger (NCX) is an anti-porter located in the plasma membrane. NCX participates in maintaining intracellular Ca 2+ homeostasis by working either in the forward mode (Ca 2+ extrusion) or in the reverse mode (Ca 2+ entry) depending on the membrane potential and the transmembrane Ca 2+ and Na + concentrations Nagy et al., 2004). The mammalian NCX (SLC8) family contains three isoforms, NCX1, NCX2, and NCX3. NCX1 predominately exists in neurons, glial cells, cardiomyocytes, and renal tissue, while NCX2-NCX3 barely appear in tissues other than brain and skeletal muscle (Annunziato, Pignataro, & Di Renzo, 2004). NCX and the plasma membrane Ca 2+ -ATPase are the main Ca 2+ transporters that are in charge of extruding [Ca 2+ ] i in neurons and glial cells (Blaustein, Juhaszova, Golovina, Church, & Stanley, 2002). The NCX has 20-fold to 60-fold higher Ca 2+ -transporting capacity than plasma membrane Ca 2+ -ATPase (Herchuelz, Kamagate, Ximenes, & Van Eylen, 2007), so NCX is the primary means of Ca 2+ extrusion when [Ca 2+ ] i accumulates above the basal level. Inhibition of the NCX occludes the main route of Ca 2+ extrusion and thus will result in [Ca 2+ ] i overload. To validate this speculation, we examined whether NCX blockers can elevate [Ca 2+ ] i in glioblastoma cells and induce Ca 2+ -dependent cell death.
KB-R7943 is a blocker that inhibits the NCX in both operation modes (Watano, Harada, Harada, & Nishimura, 1999). Inhibition of forward NCX by bepridil aggravates ischaemic brain damage, whereas inhibitors that selectively target the reverse NCX, such as SEA0400, SN-6, and YM-244769, produce protection in various hypoxic/ ischaemic cell injury models (Hu & Song, 2017;. Thus, pharmacological effects of NCX blockers rely on what mode of NCX operation is blocked at a given time. Theoretically, inhibition of the reverse NCX impedes NCX-mediated Ca 2+ entry and reduces Ca 2+ -mediated injury. In contrast, blockade of the forward NCX (Ca 2+ efflux) will lead to Ca 2+ accumulation and cytotoxicity.
Given that NCX blockers have been well developed and studied, it is timely and of great interest to know what action these NCX blockers will exert on glioblastoma. Particularly, involvement of the NCX in glioblastoma growth has not been well studied using either pharmacological or genetic approaches (Amoroso et al., 1997;Hsu, Chou, & Chueh, 1995).
Here, we showed that blocking reverse NCX with its selective inhibitors does not affect tumour growth; the glioblastoma is only suppressed when the forward NCX is blocked. Blocking the forward NCX is an effective strategy to beat glioblastoma. Therefore, the plasma membrane NCX may represent a new potential therapeutic target.

| Animals and ethical statement
Male athymic BALB/c nude mice were provided by Shanghai Laboratory Animal Center (Chinese Academy of Sciences, Shanghai, China).

BALB/c nude mouse was developed through crosses and backcrosses
What is already known • A host of NCX blockers has been developed and studied in tissues other than glioblastoma.

What this study adds
• Blocking the forward but not reverse NCX can suppress glioblastoma growth via Ca 2+ -mediated injury.

What is the clinical significance
• The NCX is a potential therapeutic target, and bepridil could be a leading compound.
• Summarized in 50 words, the significance of the work in language can be understood by a first-year university science student, including what this paper adds to existing literature.
• Existing literature shows that the Na + /Ca 2+ exchanger (NCX) and its blockers have been studied in tissues other than glioblastoma. We here found that blocking the forward NCX with bepridil can kill glioblastoma. We suggest that the NCX is a potential therapeutic target and bepridil could be a lead compound for this strategy.
between BALB/cABom-nu and BALB/cAnNCrj-nu at Charles River Laboratories Japan, Inc.; this mouse is inbred, and genetic monitoring results confirm it to be a BALB/c nude. The animal lacks a thymus, is unable to produce T cells, and is therefore immunodeficient. Mice were housed in the specific pathogen-free animal facility with free access to food and water and a 12-hr light to dark cycle. Mice at 8 weeks of age were chosen for experiments. Animal studies are reported in compliance with the ARRIVE guidelines (Kilkenny, Browne, Cuthill, Emerson, & Altman, 2010)   purchased from Nanjing COBIOER biotechnology company (Nanjing, China) and cultured in DMEM + 10% FBS + 1 μg·ml −1 puromycin.

| Culture of glioblastoma cells and astrocytes
They were used at less than 30 passages. Human astrocytes (Cat# 1800) were purchased from ScienCell Research Laboratories (Carlsbad, CA, USA) and cultured in human astrocyte medium supplemented with 10% FBS and astrocyte growth supplement. After selecting for astrocytes in passage 1, passages 2-6 of human astrocytes were used for experiments. All cells were maintained at 37°C in an incubator with a humidified atmosphere containing 5% CO 2 and 95% air.

| Cell viability assay
Cell viability was determined by using a Cell Counting Kit-8 (Dojindo, Japan). Briefly, cells were harvested from flasks and plated in 96-well cell culture plates at 5 × 10 3 per well and cultured overnight. Cells were treated with test reagents for 72 hr. Control cells were treated with cell culture medium and measured at the same time point as treatment groups. CCK-8 solution (10 μl) was added to each well of the 96-well plates, and the cultures were incubated for 30 min at 37°C. Absorbance at 450 nm was measured using an automatic microplate reader (Molecular Devices, San Jose, CA, USA).

| Whole-cell patch-clamp recording
Whole-cell patch-clamp recording was performed using a Multiclamp to +60 mV, and then a voltage ramp starting from +60 to −120 mV was applied for 1,000 ms before returning back to −50 mV. The reverse and forward NCX currents were measured at +50 and −110 mV holding potentials respectively (Molinaro et al., 2008;Rahman, Inman, Kiss, & Janssen, 2012;Song, Chen, & Yu, 2014). Data were filtered at 3 KHz and digitized at sampling rates of 20 KHz.
For drug application, cells were perfused with a VC-6 Six Channel Perfusion Valve Control Systems (Warner Instruments, Greenwich, CT, USA) connected to a gravity-fed perfusion systems.

| Ca 2+ and Na + imaging
Cells were cultured in 96-well plates at 8 × 10 3 per well and loaded with the cell membrane-permeable Ca 2+ dye Fluo-4 AM (5 μM) in 100-μl HEPES-buffered solution for 50 min. The solution contained (mM): 135 NaCl, 3 KCl, 2 CaCl 2 , 1 MgCl 2 , 10 glucose, and 10 HEPES, supplemented with 1% FBS at pH 7.4. To prepare the Ca 2+ -free solution, 2-mM CaCl 2 was replaced with 1-mM EGTA in the HEPESbuffered solution. The 96-well plate containing the Fluo-4 AM-loaded cells was mounted in a Leica TCS SP8 confocal system (Leica Microsystems, Wetzlar, Germany). The confocal microscope armed with 488-nm laser light was employed to excite the dynamic fluorescent signals in time and synchronously, while test regents were added to cells. To do Na + imaging, Fluo-4 AM was replaced with the cell membrane-permeable Na + dye Asante NaTrium Green-2 AM (ANG-2 AM) at 5 μM, which was excited by 514-nm laser light. Visual inspection and fluorescent imaging were carried out at room temperature.
The imaging data were analysed with the software LAS-AF-Lite 2.5 (Leica). The imaging traces shown are representative of at least three separate experiments; 12-18 cells were imaged in each experiment.
2.6 | Cell cycle and apoptosis assay U87 cells were seeded in six-well plates (Corning, Corning, NY, USA) with 1 × 10 6 cells per well and treated with NCX blockers, BAPTA-AM, or MAPK inhibitors. Drug treatment was performed by the person who is blinded to the experimental groups. After treatment, cells were harvested in 15-ml tubes and washed with cold sterile PBS.
After being centrifuged, cells were fixed with 70% ethanol for 30 min at 4°C for at least 1 hr. For propidium iodide (PI) staining, cells were washed one time with PBS and then were treated with ribonuclease A (RNase; 100 μg·ml −1 ) for 30 min. PI (10 μg·ml −1 ) was added, and stained cells were kept in the dark at 4°C until analysis. The number of cells analysed for each sample was 10,000 events. FACS analysis was performed using the Coulter CytoFlexS flow cytometer (Beckman Coulter, CA, USA), and the percentage of cells in the G 0 /G 1 , S, and G 2 /M phases was determined using cell cycle analysis software, ModFit LT 5.0 (Verity Software House, Topsham, ME, USA). For the apoptosis assay, cells were harvested and washed twice with cold PBS and resuspended at a density of 1 × 10 6 cells·ml −1 in 100 μl of binding buffer containing 5 μl of Annexin V-FITC and 5 μl of PI working solution (100 μg·ml −1 ). After incubation at room temperature for 15 min in the dark, 400 μl of binding buffer was added to each sample. Apoptosis was analysed by Coulter CytoFlexS flow cytometer (Beckman Coulter) for at least 10,000 events, and data were analysed with the software, CyExpert 2.0 (Beckman Coulter).

| Nucleoprotein extraction
Adherent cells were washed twice with 10 ml Dulbecco's PBS. Then 1-mM PMSF, 1-mM DTT, and 10-μl protease inhibitor cocktail (without EDTA) per 1 ml of hypotonic buffer, mix hypotonic buffer, were added and samples placed on ice for a few minutes. After which 0.6-ml mixed hypotonic buffer was added per 10,000,000 cells, on ice, then cells scraped off, and samples transferred to 1.5-ml EP tube, centrifuged for 5 min (4°C, 835× g), and the supernatant removed. The cell lysates were resuspended by adding 0.6-ml hypotonic buffer, centrifuged for 5 min (4°C, 835× g), and the supernatant removed. The pellet was resuspended in 0.3 ml of lysis buffer (lysis buffer contained 1-mM PMSF, 1-mM DTT, and 10-μl protease inhibitor cocktail per 1 ml of lysis buffer), incubated on ice for 20 min, and centrifuged for 10 min (4°C, 13,362× g). The supernatant was nuclear protein extraction.

| U87-luciferase xenograft model and treatment
The athymic BALB/c nude mice were anaesthetized by subjecting them to 3% isoflurane (RWD Life Science, Shenzhen, China) in a mixture of 30% O 2 and 70% N 2 O. After induction of anaesthesia, 1.5% isoflurane was maintained, and body temperature was kept at 37 ± 0.5°C by a heating pad. The mouse head was maintained within a stereotactic frame (RWD Life Science), allowing a precise and reproducible injection. The coordinate of the injection site was 0.5 mm anterior and 2.5 mm right from the bregma and at a depth of 3 mm.

U87-Luc cells (COBIOER) were assessed by the trypan blue exclusion
assay. Only cell suspensions with greater than 95% viability were used and implanted into the right striatum of nude mice. U87-Luc cells (5 × 10 5 in 5-μl PBS) were inoculated into the right striatum of each mouse at a controlled flow rate of 0.5 μl·min −1 , and the total injection volume was about 5 μl. Sixteen days after implantation, the mice were randomly divided into two groups: treated with vehicle (n = 5) and treated with bepridil (n = 5). Bepridil was first dissolved in ethanol (final volume 5%) and then added to a mixture of polyethylene glycol (50%) and 0.9% NaCl (45%). Mice received bepridil (50 mg·kg −1 body weight) or vehicle treatment once a day via oral gavage for 14 days.
Drug treatment was performed by a person who was blinded to the experimental groups.

| MS analysis
Athymic BALB/c nude mice were administered bepridil by oral gavage at doses of 50 mg·kg −1 as described above. Two hours later, mice were killed, and the brain sample was homogenized using the tissue homogenizers (Bertin Instruments, Paris, France) in ice-cold deionized water at a concentration of 3 ml·g −1 tissue. Then, methanol with 0.1% formic acid was added to the brain homogenate in a 4:1 proportion (vol/vol) and centrifuged at 18,188× g and 4°C for 10 min to precipitate protein. The supernatant was transferred to a vial to be injected directly into the LC-MS. The chromatographic system consisted of Shimadzu quaternary pump, vacuum degasser, and autosampler (Shimadzu, LC-20ADXR, Tokyo, Japan) coupled to triple quad LC-MS systems, API-4000 mass spectrometer (SCIEX, Redwood City, CA, USA). HPLC separation was performed on a Thermo Syncronis C8 column 100 × 2.1 mm, 5.0 μm. The mobile phase consisted of methanol (A) and water with 0.1% formic acid (B), measure time was 9 min per run. A flow rate of 0.3 ml·min −1 was used using a gradient elution of 70% B at 1 min and 5-70% B between 1 and 5 min and maintained for 2 min at 5% B and back to 70% B at 7-7.1 min. An API-4000 mass spectrometer equipped with ion spray source was employed for obtaining mass spectra. Data acquisition was carried out by analysis software. Ion spray voltage was set at 5,500 V. Curtain gas was kept at 35 psi. Ion source temperature was 550°C. Nebulizing gas and drying gas were at 50 psi. Multiple reaction monitoring mode was utilized to detect the compound of interest. Collision energy (CE) is an instrument parameter that is frequently optimized to increase fragment ion intensity. An alternative to empirically optimizing the CE for bepridil is to predict the best CE value based on the precursor mass-to-charge ratio of bepridil (m/z). The CE for bepridil was optimized to 30 V. The precursor-to-product ions (Q1 → Q3) selected for bepridil during quantitative optimization were (m/z) 367.1 → 84.3/184.3.

| In vivo bioluminescent imaging
Tumours were measured by bioluminescence using an IVIS Spectrum CT Imaging System (PerkinElmer, Waltham, MA, USA), which captured the luminescence signal emitted from the engrafted tumour. Prior to imaging, the mice were anaesthetized with inhalation of isoflurane gas (RWD Life Science) and injected i.p. with 150 mg·kg −1 D-luciferin potassium (J&K Scientific) aqueous solution. The isoflurane was balanced with oxygen and dialled to 2.0% for the induction of anaesthesia and 1.0% for maintenance. Images were subsequently captured 10 min following injection. Signal intensity was quantified within a region of interested over the head, as defined by Living Image software. All images represent 2-min exposure time, and the average number of photons·s −1 ·cm −2 per steradian were recorded. The data were analysed using Living Image 4.4.5 software (PerkinElmer, RRID: SCR_014247).

| Data and statistical analysis
Þ Þ, where I 0 and I are current amplitudes measured in control and in the presence of NCX blockers, C is the logarithm of concentration, and n is the Hill coefficient. The statistical difference between two independent groups was analysed by Student's parametric unpaired t test. And the original data of more than two groups were assessed by the parametric one-way ANOVA followed by a Tukey's post hoc test. The data for cell viability and
(d-f) The currents recorded before and after application of the reverse NCX blockers SEA0400 (0.1-100 μM), SN-6 (0.1-30 μM), and YM244769 (0.1-30 μM). The reverse NCX current was measured at +50 mV, and forward NCX current was measured at −110 mV Western blot were usually normalized to the control group, and were thus analysed by the non-parametric Kruskal-Wallis test followed by a Dunn's post hoc test. Differences were considered to be significant when P < .05. Post hoc tests were conducted when the F value achieved the necessary level (P < .05) and there was no significant variance inhomogeneity.

| Inhibition of the Na + /Ca 2+ exchanger by its blockers in glioblastoma cells
We characterized the membrane currents associated with NCX activity in glioblastoma cells using whole-cell patch-clamp recording. As illustrated in Figure 1a, cell membrane potential was held at −50 mV, and the NCX currents were elicited by a voltage ramp from +60 to −120 mV. The NCX current in its reverse mode (Ca 2+ entry) was measured at +50, and the forward-mode current (Ca 2+ exit) was measured at −110 mV. The Ni 2+ (10 mM) was employed to inhibit and identify the NCX currents in its either reverse or forward mode as previously reported (Maack, Ganesan, Sidor, & O'Rourke, 2005). The reversal potential of NCX current is around −60 mV determined by intersection of current-voltage relation (I/V) in the absence and presence of

BJP
Ni 2+ . Bepridil has been reported to block forward-mode NCX and also partially inhibits its reverse mode . We found bepridil blocked both direction of NCX currents in U87 cells, with halfmaximal inhibition (IC 50 ) values of 19.1 ± 1.2 and 15.6 ± 1.1 μM ( Figure 1b and Table 1). KB-R7943 preferentially inhibits the reverse NCX, and it also blocks the forward NCX .

| Blocking the forward NCX operation generates cytotoxicity in glioblastoma cells
We next examined impact of NCX blockers on viability of glioblastoma cells. U87 and U251 cell lines were treated with bepridil or KB-R7943 BAPTA-AM significantly suppressed bepridil-induced apoptosis. n = 5 independent experiments; *P < .05, apoptotic cell fraction in BAPTA-AM + bepridil versus bepridil alone, with the parametric one-way ANOVA followed by Tukey's post hoc test for 72 hr. We found that viability of U87 and U251 cells was reduced to 30-40%, 6-7%, and 3-4% of control level by 20-, 50-, and 100-μM bepridil (Figure 2a,b). Similarly, U87 and U251 cells were regressed to 50-60%, 7-18%, and 2-3% of control level by 20-, 50-, and 100-μM KB-R7943. Figure 2d lists IC 50 values of bepridil and KB-R7943 that cause cytotoxicity. Since bepridil and KB-R7943 inhibit both forward and reverse NCX, we need to clarify whether this cytotoxicity resulted from blocking the reverse or forward NCX operation.
We also applied NCX blockers to a paediatric glioblastoma cell line SF188. These cells were suppressed by bepridil and KB-R7943 in a dose-dependent manner but slightly affected by the compounds that selectively block the reverse NCX (Figure 2c). CB-DMB is a more specific blocker for the forward NCX than bepridil and KB-R7943. It is very exciting to find that CB-DMB is more potent than bepridil and KB-R7943 in suppressing glioblastoma cells ( Figure S1a). The IC 50 values of CB-DMB that cause cytotoxicity in U87, U251, and SF188 are 3.1 ± 0.6, 2.7 ± 0.1, and 2.6 ± 0.1 μM. These data confirm that blocking the forward but not reverse mode of NCX can suppress the growth of glioblastoma cells.    (Brustovetsky et al., 2011).

| Bepridil, CB-DMB, and KB-R7943 cause cell cycle arrest and apoptosis
To do cell cycle analysis, U87 cell cultures were treated with 25-μM bepridil for 6, 12, 18, and 24 hr and stained with PI. We show that bepridil reduced the cell fraction in S phase and increased cells in G 0 /G 1 phase in a time-dependent manner (Figure 4a,b). To determine the role of [Ca 2+ ] i in bepridil-induced G 0 /G 1 arrest, BAPTA-AM was used to prevent [Ca 2+ ] i accumulation in U87 cells. Figure 4c shows that BAPTA-AM (10 μM) alone for 18 hr did not incur significant cell fraction changes in G 0 /G 1 , S and G 2 /M phases. Bepridil (25 μM, 18 hr) markedly increased cell fraction in G 0 /G 1 phase. This G 0 /G 1 arresting effect was significantly attenuated by 10-μM BAPTA-AM (Figure 4c). To detect any apoptotic cell death involved, U87 cells were treated with bepridil (25 μM) for 6, 12, 24, and 48 hr and then analysed with Annexin V-FITC/PI apoptosis assay. Figure 4 d,e shows that bepridil incurred a time-dependent increase of apoptotic (Annexin V + ) cells. BAPTA-AM (10 μM) alone for 18 hr did not increase apoptosis in U87 cells but significantly attenuated bepridil-induced apoptosis (Figure 4f). Application of CB-DMB (3.5 μM) to U87 cells also induced G 0 /G 1 arrest and activated apoptosis in time-dependent manner ( Figure S3a,b,d-e), and these cell death processes were significantly attenuated by BAPTA-AM (Figure 3c,f).
These data indicate that bepridil and CB-DMB generate cytotoxicity through Ca 2+ -mediated cell cycle arrest and apoptosis in glioblastoma cells.

| Involvement of MAPK signalling in bepridil-induced cell death
The above data indicate that bepridil killed glioblastoma cells via blockade of the forward NCX and elevation of [Ca 2+ ] i . It is well known that [Ca 2+ ] i regulates the MAPK signalling pathway, which plays essential role in cell fate determination (White & Sacks, 2010). We here measured the activities of ERK, p38-MAPK, and JNK in U87 cells at the time points 3.6 | Functional expression of NCX isoforms and impact of bepridil on human astrocytes Glioblastoma arises from glial cells in the brain (Ostrom, Gittleman, Stetson, Virk, & Barnholtz-Sloan, 2018). Bepridil was previously applied by Johnson & Johnson Pharmaceutical Research for the treatment of angina and hypertension. We here assessed viability of human astrocytes after exposed to bepridil (25 μM) for 48 hr and compared it with glioblastoma cells. Bepridil did not generate apparent cytotoxicity in human astrocytes, meanwhile markedly suppressed growth of U87, U251, and SF188 cells (Figure 8a). To explore why human astrocyte is not sensitive to NCX blocker bepridil, we measured expression of three NCX isoforms in astrocytes and glioblastoma cells. Western blot analysis shows that NCX1 expression is significantly higher in human astrocytes than glioblastoma cells (Figure 8b). But the level of NCX2 and NCX3 is . n = 5 independent tests in each group not high; especially, NCX2 is even lower than that in glioblastoma cells. This observation is identical to the literature, documenting that in astrocytes, NCX1 is the most highly expressed among the three NCX isoforms (Boscia et al., 2016;Pappalardo, Samad, Black, & Waxman, 2014). We then decided to compare the activity of NCX among them by using whole-cell patch-clamp recording. The amplitude of the NCX currents in human astrocytes is much larger than glioblastoma cells (Figure 8c). After performing a calculation, we show current density of the NCX in human astrocytes is several-fold higher than that in glioblastoma cells (Table S1). This high NCX activity and astrocyte's resistance to bepridil's toxicity most likely result from high expression of NCX1 isoform. On the other, KB-R7943 is deleterious to astrocytes probably due to its multiple actions besides blocking NCX (Figure 8d). images of U87-luciferase glioblastoma cells implanted into the right cerebral striatum of nude mice. Images were taken by an IVIS Spectrum CT Imaging System before and after the period of treatment with bepridil or vehicle as described in methods. (e) Tumour size determined by bioluminescence and caliper measurement in each group before and after treatment. *P < .05, comparison between the two groups; n = 5 mice per group, analysed by Student's parametric unpaired t test 3.7 | Bepridil inhibits growth of brain-grafted tumour cells in vivo Bepridil has been shown to able to penetrate the blood-brain barrier (BBB) and reach the brain after oral administration or intraperitoneal injection at 50 mg·kg −1 body weight in two separate animal studies (Lipsanen et al., 2013;Mitterreiter et al., 2010). Our MS experiment shows that bepridil can cross the BBB and get into brain tissue (Figure 9a,b)  however, there is increasing evidence that KB-R7943 also inhibits many other ionic channels (Barrientos, Bose, Feng, Padilla, & Pessah, 2009;Brustovetsky et al., 2011;Pezier, Bobkov, & Ache, 2009). In our study, KB-R7943 is also toxic to normal astrocytes at the concentration range for it suppresses glioblastoma cells. We did not conduct further study to reveal intracellular signalling that mediates toxicity of KB-R7943.
Our test of bepridil generates more encouraging results. Bepridil can penetrate the BBB and reach the brain. Animal experiment demonstrated that bepridil reduces in vivo growth of brain-grafted glioblastoma. In addition, bepridil does not produce apparent cytotoxicity to human astrocytes when it markedly inhibits glioblastoma cells in vitro. This resistance to bepridil's cytotoxicity may result from high NCX1 expression and high NCX activity in astrocytes. The NCX1 expression is much lower in glioblastoma cells than in astrocytes. From the perspective of pharmacodynamics, the potency of a drug is proportional to the percentage of targets being occupied (Pugsley, Authier, & Curtis, 2008). When most NCX1s on glioblastoma cells are occupied by bepridil at an appropriate concentration, only a small fraction of NCX1s are blocked in astrocytes. Therefore, viability of astrocytes is not affected by bepridil when it already hurts glioblastoma cells.
The present study cannot completely exclude the possibility that other mechanisms also contribute to the sensitivity difference between astrocytes and glioblastoma cells. For example, cytosolic Na + inactivates operation of the NCX1 and NCX3 (not NCX2), while increase of cytosolic Ca 2+ activates all NCXs and also alleviates the Na + -induced inactivation (Tal, Kozlovsky, Brisker, Giladi, & Khananshvili, 2016;Verkhratsky, Trebak, Perocchi, Khananshvili, & Sekler, 2018). In glioblastoma cells, the elevated [Ca 2+ ] i cannot activate NCXs because most NCXs were occupied and blocked by bepridil. In human astrocytes, NCX1 is the most highly expressed isoform, and only a small portion of NCX1s were blocked by bepridil. The remaining NCX1s can be readily activated by [Ca 2+ ] i and operate in the forward mode, which potentially underlies astrocyte's resistance to bepridil.
Second, the reverse NCX operation accounts for [Ca 2+ ] i overload and ensuing injury in astrocytes when these glial cells are in resting, stimulating, or ischaemic conditions, especially when Na + -driven uptake of neurotransmitters occurs (Reyes, Verkhratsky, & Parpura, 2012;Verkhratsky et al., 2018). It is possible that the reverse NCXs were blocked by bepridil in human astrocytes, and thereafter, the cytotoxicity was regressed, whereas this situation is different in glioblastoma and the cytotoxicity was augmented because the forward NCXs were inhibited.
The role of NCX has been well described in glial cells by these articles (Boscia et al., 2016;Parpura, Sekler, & Fern, 2016;Rose & Verkhratsky, 2016). But there are very few studies that report NCXs in the tissue of glioblastoma; particularly, it lacks data that compare NCXs between glioblastoma and glial cells. We here got a new observation that the NCX activity is higher in normal astrocytes than that in glioblastoma cells. The direction of NCX operation is controlled by change of membrane potential and gradients of Na + and Ca 2+ , which constantly coordinate with neuronal activity. Conversely, the NCX operation tightly regulates Na + and Ca 2+ signalling, which manipulates neuronal excitability and neurotransmission. Down-regulated NCX activity in glioblastoma could disrupt the crosstalk between glial and neuronal cells, leading to a detrimental ionic environment in the brain. So far, our materials verify our hypothesis and suggest that the plasma membrane NCX represents a new potential therapeutic target for glioblastoma.
Glioblastoma is the most common type of glioma that arises from glial cells in the brain or spinal cord. The annual incidence rate of glioblastoma in adults is between 0.6 and 3.7 per 100,000 persons (Ostrom et al., 2018). Patients with glioblastoma usually have a mean survival rate of 3-4 months if without treatment (Krex et al., 2007;Rao, 2003). Under combined treatment of radiotherapy and chemotherapy, the median survival time with grade IV glioblastoma can be prolonged to 15 months (Wen & Reardon, 2016 Bepridil was previously approved by the Food and Drug Administration to treat angina but withdrawn later because it can induce new arrhythmias. We recommend that a structural modification is required to reduce the cardiac side effect of bepridil before it is re-purposed to treat human glioblastoma.