Trifluoperazine, a Well-Known Antipsychotic, Inhibits Glioblastoma Invasion by Binding to Calmodulin and Disinhibiting Calcium Release Channel IP₃R Free

Glioblastoma is the most common and aggressive malignant primary brain tumor, accounting for more than 50% of all brain tumor cases. Moreover, glioblastoma has a very poor prognosis, with a less than 5% 5-year survival rate and a median survival time of 1 year (1–4). Standard therapy of glioblastoma consists of a surgical resection of glioblastoma and adjuvant therapy with anticancer drug, such as the alkylating agent temozolomide. However, a complete surgical resection is almost impossible due to aggressive invasion of glioblastoma over time throughout the brain (5). In addition, temozolomide provides only modest benefits and increases the patients' life expectancy merely by about 2.5 months, with tumor relapse frequently observed (6, 7). Therefore, there is a pressing need for finding more effective anticancer therapy for glioblastoma.

Interestingly, some commonly used antipsychotic drugs show antiproliferative effect (8) and have been recently proposed as anticancer drugs. A recent study showed that patients with schizophrenia are not only less susceptible to cancer (1.93%) than non-schizophrenic controls (2.97%) during a 9-year follow-up period (9), but also some psychotropic agents, including antipsychotics, antidepressants, and mood stabilizers, possess a significant in vitro antiproliferative activity (10, 11). For example, trifluoperazine (TFP), which is an FDA-approved antipsychotic and antiemetic drug (12) used for treating schizophrenia (13), has been shown to inhibit cell proliferation and invasion and to induce cell death in several types of cancer cell lines and animal models (14–21). However, whether TFP has any anticancer effect in glioblastoma has not been investigated.

One of the proposed modes of action of TFP is its ability to bind to a well-known Ca²⁺-binding protein, calmodulin (CaM). It has been reported that TFP exerts an inhibitory action on the function of CaM by directly binding to CaM (22). In particular, CaM is known to interact with inositol 1,4,5-trisphosphate receptor (IP₃R), which is a Ca²⁺ release channel located on intracellular Ca²⁺ stores, such as endoplasmic reticulum (23). IP₃R is a tetrameric protein, and each of its subunits consists of an N-terminal ligand-binding domain, a C-terminal transmembrane pore domain, and an intervening modulatory domain (24). It is well known that the opening of IP₃R is regulated by both IP₃ and Ca²⁺ (24). When Ca²⁺ binds to CaM, the Ca²⁺/CaM complex can interact with N-terminal ligand-binding domain of IP₃R, causing it to inhibit IP₃R (24, 25). Therefore, it is possible that TFP binding to CaM might have a profound effect on IP₃R channel function and IP₃R-mediated Ca²⁺ release. However, this possibility has not been tested yet.

Ca²⁺ is one of the crucial molecules involved in intracellular signaling, which is important for cell proliferation, and survival (5, 26). When it comes to glioblastoma, Ca²⁺ signaling is critical for proliferation, migration, and invasion (27). We previously reported that caffeine blocks an increase in intracellular Ca²⁺ concentration, which is caused by the signaling of several G-protein–coupled receptors, in glioblastoma cells by inhibition of Ca²⁺ release channel IP₃R subtype 3 (5). As a result, caffeine suppresses glioblastoma cells proliferation, migration, and invasion and extends survival in animal models (5). From that study, we proposed that targeting Ca²⁺ signaling can be a valuable therapeutic target for treating glioblastoma.

In the current study, we investigated whether TFP could inhibit glioblastoma proliferation and invasion by targeting Ca²⁺ signaling. We found that, unlike caffeine, TFP binding to CaM2 increased intracellular Ca²⁺ release through opening of IP₃R subtype 1 and 2 by causing a dissociation of CaM2 from IP₃R. Through this mechanism, TFP suppressed the proliferation, motility, and invasion of glioblastoma cells in in vitro and in vivo xenograft model.

U87MG human glioblastoma cells were obtained from the ATCC in 2006. Human primary glioblastoma (GBL) cell lines (GBL12, GBL13, GBL14, GBL15, GBL28, GBL30, GBL227, and GBL232) and neural stem cells (NSC) were generated from the Seoul National University Hospital (Seoul, Korea, IRB nr.; H-0507-509-153). All cell lines originated from the human neuronal system. All cells were cultured in 5% CO₂ and air humidified in a 37°C incubator. Culture media were DMEM (Corning Costar) containing 10% FBS, 100 U/mL penicillin, and 100 μg/mL streptomycin (Gibco Invitrogen). Each cell line at early passages was stocked, and cultures were maintained until passage 20 (within 2 months).

Cells were seeded into 96-well plates at a density of 1.5 × 10³ cells per well and allowed to attach for 24 hours. Cells were treated with vehicle or the indicated concentrations of TFP (Sigma) diluted in complete media for each 24, 48, and 72 hours. Cell proliferation was determined by 3-[4, 5-dimethylthiazol-2-yl]-2, 5-diphenyltetrazolium bromide (MTT) assay. At the end of the TFP treatment, 100 μL of 1× MTT (Amresco) labeling reagent was added to each well and incubated for another 2 hours. The proliferative activities were expressed as the relative percentage of cell numbers at initial time.

Cells per well (5 × 10⁴) were seeded in 0.3% cell agar layer, which was on top of 0.6% base agar layer in 6-well culture plates. The solidified cell layer was covered with the medium containing TFP, which was replaced every 4 days. Cells were then incubated for a further 2 weeks at 37°C and 5% CO₂. Afterward, colonies were stained with 0.05% cresyl violet and photographed.

Cell invasion was assayed as described previously (5). The mean number of untreated cells was considered as 100% invasion. Each condition was triplicated, and five fields were randomly selected and counted for each assay. TFP was added at the time of cell plating.

In the skin xenograft model, 5-week-old athymic mice (BALB/c nu/nu) were obtained from Central Laboratory Animal, Inc. For the xenograft tumor growth assay, U87MG cells (150 μL; 3 × 10⁶ cells) were subcutaneously injected into the right flanks of the mice. On the third day after injection, TFP was given as an intraperitoneal injection (5 mg/kg/day). The control animals were given 0.9% saline. A total of 21 days treatments were given, with a 24-hour rest period. Athymic mice (BALB/c nu/nu) bearing U87MG cells were randomized into two groups (control and TFP; n = 5 per group) when the volume of tumor reached 40 mm³. Tumor volume and body weight were measured daily for 21 days with a caliper and electronic balance. Tumor volumes were calculated by the following formula: volume = length × width²/2. After 21 days, all animals were euthanized and autopsied. Tumors and lung organs were collected and fixed by 4% paraformaldehyde in PBS.

In the brain xenograft model, an orthotopic implantation model was established with same cell line. U87MG cells (5 μL; 2.5 × 10⁵ cells) were implanted by intracranial injections in the left frontal lobe at coordinates 2 mm lateral from the bregma, 0.5 mm anterior, and 3.5 mm intraparenchymal. TFP (10 mg/kg/day) was given as an intraperitoneal injection 3 days after implantation and drug administration schedule was same as flank xenograft model. Mice were monitored daily for general appearance, behavioral changes, and neurologic deficits. Mice were sacrificed when moribund. All protocols were approved by the Gyeongsang National University Institutional Animal Care and Use Committee.

Mouse lung and brain tissues were stained with hematoxylin and eosin (H&E). For H&E staining, the fixed tumors were embedded in paraffin, cut into 5 μm sections, and stained with H&E. Slides were photographed using an optical or confocal microscope (BX61VS, Olympus).

Before replating onto a cover glass for calcium imaging and Western blotting, lentivirus carrying scrambled shRNA or IP₃R subtype 1, 2, 3 shRNA or CaM shRNA was infected to cultured U87MG cells or GBL28 cells seeded on 35- or 60-mm dishes. Each shRNA sequence for each IP₃R subtype and CaM is described in Supplementary Table S1.

Ca²⁺ imaging was performed as described previously (5). Intensity images of 510 nm wavelength were taken at 340 and 380 nm excitation wavelengths using either iXon EMCCD (DV887 DCS-BV, ANDOR Technology). The two resulting images were used for ratio calculations in Axon Imaging Workbench version 6.2 (Axon Instruments).

qRT-PCR was done by using SYBR Green PCR Master Mix. Reactions were performed in duplicates in a total volume of 10 μL containing 10 pmol/L primer, 4 μL cDNA, and 5 μL SYBR Green PCR Master Mix (Applied Biosystems). The mRNA level of CaM subtypes was normalized to that of GAPDH mRNA. Fold induction was calculated using the 2^−ΔΔC_t method. The following sequences of primers were used for real-time RT-PCR. CaM primer sequences were as follows: GAPDH_Fwd 5′-AGCTGAACGGGAAGCTCA-3′; GAPDH_Rev 5′-TGCTGTAGCCAAATTCGT-3′; calmodulin1_Fwd: 5′-ACTGGGTCAGAACCCAACAG-3′; calmodulin1_ Rev: 5′-TGCCTCACGGATTTCTTCTT-3′; calmodulin2_Fwd: 5′- ATGCTGATGGTAATGGCACA-3′; calmodulin2_Rev: 5′–TGTCATCACATGGCGAAGTT-3′; calmodulin3_Fwd: 5′–GATGAGATCCCTGGGACAGA-3′; calmodulin3_Rev: 5′–CTCGGATCTCCTCCTCACTG-3′.

Cell Observer (Carl Zeiss) was used to observe real-time glioblastoma behavior pattern in TFP-treated medium. Cells were seeded at 1 × 10⁴ cells per well on 12-well plates and cultured overnight. Cells were treated with TFP in medium under standard cell culture conditions. Cultured cells in the medium without adding TFP were taken as a control. After 4 hours, culture plates were transferred to the Cell Observer and maintained at 37°C in 5% CO₂. Images were captured at 5-minute intervals for 25 hours. The recorded images were analyzed by supported software (Axio Vision, Carl Zeiss) to determine motility for each sample.

U87MG cells were lysed with RIPA buffer (Thermo Fisher Scientific). Total protein concentrations were determined using BCA assay (Thermo Fisher Scientific). An equal volume of 4× SDS sample buffer was added, and the samples were boiled for 5 minutes. Equivalent amounts of total protein (20–30 μg) were separated by SDS-PAGE on 4% to 10% polyacrylamide gel and then transferred to nitrocellulose membrane. The membrane was blocked with 5% skimmed milk or 3% BSA in 0.1% Tween-20/Tris-buffered saline (TBS-T). The membrane was incubated with IP₃R1 (1:1,000, ab5804, Abcam), IP₃R2 (1:1,000, ab55981, Abcam), IP₃R3 (1:1,000, ab55983, Abcam), and β-actin (1:10,000, Thermo Fisher Scientific) antibodies overnight at 4°C.

When two groups were being compared, the significance of data was assessed by the two-tailed Student unpaired t test using Microsoft Excel software. Other statistical analyses were done using GraphPad Prism for Windows (Version 5.0, GraphPad Software). Differences between 3 or more means were determined by one-way ANOVA with Dunnett multiple comparison test. Linear mixed effects regression models were used to estimate and compare the group-specific change in tumor growth curves. Differences in survival curves were determined by log-rank test. All statistical analyses were performed at the P < 0.05 level of significance.

First, we investigated the effect of TFP on the cell viability of U87MG glioblastoma cells by MTT assay. We found that U87MG cell viability decreased time dependently (24, 48, and 72-hour TFP treatment) as well as dose dependently (1, 2, 5, 10, and 20 μmol/L of TFP; Fig. 1A), suggesting that TFP has a cytotoxic effect on glioblastoma cells.

To test the effect of TFP on anchorage-independent growth of U87MG glioblastoma cells, we performed colony formation assay. We observed that TFP significantly suppressed anchorage-independent growth at a concentration higher than 2 μmol/L (Fig. 1B). Because of the high invasive activity of glioblastoma cells in the brain tissue (6), we investigated the effect of TFP on invasion of U87MG cells by performing Matrigel transwell invasion assay (Fig. 1B). We observed that TFP significantly restricted the invasion of glioblastoma cells at a concentration higher than 1 μmol/L, with a half-maximal effective concentration at around 10 μmol/L (Fig. 1B). These results suggested that TFP potently inhibits colony formation and invasion of glioblastoma cells.

To investigate the effect on TFP on cellular morphology of U87MG cells, we performed live cell imaging and analysis (Supplementary movie S1). We observed dynamic changes in cellular morphology in U87MG cells during TFP treatment in a dose-dependent manner (5, 10, 15, 20, and 25 μmol/L; Supplementary Fig. S1A). The cellular morphology changed quickly in response to TFP, and eventually, most cells were detached from the plate and died after 25 hours (Supplementary Fig. S1A). Furthermore, the ratio of proliferated glioblastoma cells was significantly decreased in TFP-treated cells compared with 0.9% saline-treated cells as control (Supplementary Fig. S1B). These results were consistent with the results of MTT assay, suggesting that TFP potently induces glioblastoma cell death.

To examine the in vivo effect of TFP, we transplanted U87MG cells subcutaneously into the athymic mice (BALB/c nu/nu) and checked tumor volume, weight, and lung tumor incidence during 21 days (Fig. 2A). We found that TFP treatment (5 mg/kg/day) significantly suppressed the glioblastoma growth of the xenograft mice as evidenced by a marked decrease of glioblastoma tumor size in TFP-treated mice at day 21 (Fig. 2A). Next, we monitored daily progression of the tumor volume during 21 days from the day of intraperitoneal injection of U87MG cells. We found that tumor volume of TFP-treated mice was significantly decreased compared with control mice without TFP starting from day 17 (Fig. 2B). Furthermore, tumor weight of TFP-treated mice was about 50% of control at day 21 (Fig. 2C). However, there was no difference between the mean bodyweights of the two groups over the same period (Supplementary Fig. S2). The lung is known to contain many blood and lymphatic vessels originating from various other organs and is prone to invasion of tumors from several other organs. Therefore, we measured whether TFP can inhibit lung metastasis in the skin xenograft model. The results from the autopsy showed that the lung metastasis was observed in the control group (71.4%, 4/7 mice) but not in the TFP-treated group (0%, 0/7 mice; Fig. 2D). Consistent with this result, H&E staining also showed that there was no lung metastasis in TFP-treated group compared with control in the skin xenograft model (Fig. 2D).

Intracellular Ca²⁺ is an important signal for gene expression, motility, differentiation, and survival of glioblastoma cells. In our previous study, we demonstrated that caffeine blocks the Ca²⁺ release from the IP₃R subtype 3, decreases the invasion of glioblastoma cells, and finally increases the survival of the mouse implanted with glioblastoma tumor (5). Therefore, we examined whether TFP is associated with the intracellular Ca²⁺ in glioblastoma cells using Ca²⁺ indicator dye, Fura-2-AM. Unexpectedly, TFP application to cultured U87MG cells induced an increase in cytoplasmic Ca²⁺ in the U87MG cells concentration dependently (Fig. 3A). This effect of Ca²⁺ increase in U87MG cells by TFP is opposite to the effect of Ca²⁺ inhibition by caffeine, yet both TFP and caffeine showed inhibitory effect on invasion and proliferation, suggesting that not only inhibiting intracellular Ca²⁺ increase but also increasing intracellular Ca²⁺ is critical for blocking glioblastoma cell growth and invasion.

To determine whether TFP-induced Ca²⁺ increase is caused by Ca²⁺ entry from extracellular space or Ca²⁺ release from intracellular stores, such as endoplasmic reticulum (ER), we performed Ca²⁺ imaging experiments under three conditions: normal 2 mmol/L Ca²⁺–containing bath condition, Ca²⁺-free bath condition; and thapsigargin (TG) treatment in Ca²⁺-free bath condition. We firstly treated with 100 μmol/L TFP in 2 mmol/L Ca²⁺ bath solution and found a robust Ca²⁺ increase in U87MG cells like in Fig. 3A. Next, we removed Ca²⁺ entry from outside by using Ca²⁺-free bath solution and found that 100 μmol/L TFP also induced Ca²⁺ increase in U87MG cells, although there was about 32% decrease of Ca²⁺ peak ratio compared with 2 mmol/L calcium–containing bath condition (Fig. 3B). Next, we depleted Ca²⁺ stores by treating with TG in Ca²⁺-free bath solution and found that there was almost no calcium increase by 100 μmol/L TFP (Fig. 3B). These results indicate that ER Ca²⁺ release is the major initiating contributor to TFP-induced Ca²⁺ increase in glioblastoma cells.

It has been reported that Ca²⁺-bound CaM is known to inhibit the opening of Ca²⁺ release channel, IP₃R (23). In addition, TFP binding to CaM is reported to cause conformational change and functional loss of CaM (22). Therefore, we hypothesized that the molecular mechanism of TFP-induced Ca²⁺ response might involve IP₃R and CaM. We first examined whether TFP-induced Ca²⁺ response is affected by gene silencing of IP₃R in U87MG cells. We confirmed the knockdown efficiency of each IP₃R subtype 1, 2, and 3 shRNA by Western blot analysis (Supplementary Fig. S3A). Gene silencing of IP₃R subtype 1 or 2 by lentivirus carrying IP₃R1 or IP₃R2 shRNA resulted in a significant decrease in 100 μmol/L TFP-induced Ca²⁺ response compared with that of the control scrambled shRNA in U87MG cells (Fig. 4A). However, Ca²⁺ response by 100 μmol/L TFP in IP₃R3 shRNA-infected U87MG cells was not different with that in scrambled shRNA-infected U87MG cells (Fig. 4A). These results showed that TFP-induced Ca²⁺ increase in glioblastoma cells is mediated by opening of IP₃R subtype 1 and 2, but not subtype 3.

Next, we checked whether the IP₃R1 and 2 can affect glioblastoma cell viability by silencing the IP₃R1 or 2 gene by shRNA for IP₃R1 or 2 in cell counting assay (Fig. 4B). We observed that surviving U87MG cell number in IP₃R1 or IP₃R2 shRNA-infected U87MG cells was significantly more than that in scrambled shRNA-infected U87MG cells after 5 μmol/L TFP treatment (Fig. 4B), indicating that IP₃R1 and 2 contribute to the TFP-induced glioblastoma cell death. However, cell number in IP₃R3 shRNA-infected U87MG cells did not differ with that in scrambled shRNA-infected U87MG cells after 5 μmol/L TFP treatment (Fig. 4B), implying that IP₃R3 did not contribute to the antiglioblastoma effect by TFP. This was also confirmed with GBL28 by IP₃R1 shRNA, which showed most effect on U87MG cell viability (Supplementary Fig. S3B). Result of cell counting assay using GBL28 cells (Supplementary Fig. S3B) was consistent with the result in U87MG cells (Fig. 4B). On the basis of these results, we conclude that antiglioblastoma effect by TFP is mainly mediated by IP₃R subtype 1.

We next investigated whether CaM is associated with TFP-induced Ca²⁺ increase in U87MG cells. First, the expression level of CaM subtypes was measured by real-time PCR in GBL28 cells and NSCs as control (Fig. 4C). We found that CaM2 is expressed at a much higher level than other CaM subtypes in GBL28 cells (Fig. 4C). Moreover, the expression level of CaM2 in GBL28 cells was up to 2-fold higher than that in NSCs (Fig. 4C). Then, we tested the involvement of CaM2 in TFP-induced Ca²⁺ response by CaM2 shRNA-infected U87MG cells. We found that CaM2 shRNA decreased approximately 70% of the Ca²⁺ release by 100 μmol/L TFP compared with scrambled shRNA (Fig. 4D). To test whether the reduction in TFP-induced Ca²⁺ response in U87MG cells is caused by the specific gene silencing of CaM2, we coinfected wild-type CaM2 (sh-insensitive), which has mutations in CaM2 shRNA target sequence, so that it is no longer sensitive to CaM2 shRNA, along with CaM2 shRNA. Under this condition, Ca²⁺ response was rescued to the control scrambled shRNA condition (Fig. 4D). In addition, we tested whether the physical interaction between TFP and CaM is critical for TFP-induced Ca²⁺ increase by introducing a Q to A mutation at 120 residue in CaM2 (CaM2-Q120A), which is known to be the critical amino acid residue for TFP binding to CaM (22). The Q120A mutation was introduced in the CaM2 gene that contains CaM2 shRNA-insensitive sequence so that when coinfected with CaM2 shRNA, CaM2-Q120A would not be able to interact with TFP in the absence of the endogenous CaM2. We observed that the TFP-induced Ca²⁺ response was not rescued by CaM2-Q120A after infected with CaM2 shRNA and CaM2-Q120A, which has TFP-binding mutant CaM2 and CaM2 shRNA-insensitive sequence (Fig. 4D). On the basis of these results, we concluded that TFP opens the IP₃R subtype 1 and 2 indirectly by binding to CaM2 and subsequently causing a conformational change of CaM2 that results in disinhibition (or relief of inhibition) of IP₃R, leading to opening of IP₃R.

Next, we examined whether the effect of TFP in cell viability is cell-type specific. We performed MTT assay and tested the inhibitory effect of TFP in GBL cells and NSC viability. As shown in Fig. 5A, concentrations higher than 5 μmol/L TFP showed a significant toxicity for NSCs, whereas concentrations higher than 2 μmol/L TFP showed a significant toxicity for GBL28 cells (Fig. 5A). Furthermore, cell viability of NSCs decreased only about 10% at 10 μmol/L TFP, whereas GBL28 cells showed 40% decrease in cell viability at the same concentration (Fig. 5A). In addition, we found that cell viability of other GBL12, GBL13, GBL15, GBL30, GBL227, and GBL232 was significantly decreased compared with NSCs (Supplementary Fig. S4). These results indicated that TFP can affect both types of cell, but glioblastoma cells show much higher sensitivity to TFP.

To see whether the heightened sensitivity to TFP in GBL cells is due to an enhanced Ca²⁺ increase, we compared the TFP-induced Ca²⁺ increase in NSCs, U87MG cells, and several GBL cells (GBL12, GBL14, GBL15, GBL28, and GBL30). Consistent with MTT assay results, TFP-induced Ca²⁺ response in GBL cells (GBL12, GBL14, GBL15, GBL28, and GBL30) and U87MG cells was significantly higher than NSCs Ca²⁺ response (Fig. 5B). On the basis of these results, we concluded that TFP is more effective and potent in glioblastoma cells than in normal cells.

To further test the potential therapeutic effect of TFP on glioblastoma, we examined the inhibitory effects of TFP on glioblastoma growth in orthotopic xenograft mice. We implanted U87MG cells into the left frontal lobe of athymic mice (BALB/c nu/nu) by intracranial injections. TFP (10 mg/kg/day) was given as an intraperitoneal injection 3 days after implantation, and drug administration schedule was same as the skin xenograft model. Mice were monitored daily for general appearance, behavioral changes, and neurologic deficits. Mice were sacrificed when moribund. Mouse brain was stained with H&E to visualize and measure the tumor size in the brain (Fig. 6A). We found that TFP treatment reduced over 75% of the tumor volume compared with control mice (Fig. 6B). However, in the survival test in orthotopic xenograft mice, TFP only modestly extended the survival time compared with the control group (Fig. 6C). From these results, we concluded that TFP shows a strong inhibitory effect on glioblastoma growth in vivo but, for unknown reasons, only modestly extends the survival time in mice.

We investigated here the therapeutic potential of TFP on glioblastoma and underlying detailed molecular and cellular mechanisms of TFP action by using both in vitro and in vivo assays. In summary, TFP suppressed proliferation (Fig. 1A), migration (Fig. 1B), and invasion (Fig. 1B) of glioblastoma cells in several in vitro assays. Furthermore, tumor size and weight in in vivo skin xenograft model (Fig. 2) and in orthotopic brain xenograft model (Fig. 6) of glioblastoma were markedly decreased by TFP treatment. We found that the glioblastoma-specific cytotoxic effect of TFP is linked to intracellular Ca²⁺ increase by release of Ca²⁺ from ER through IP₃Rs. Among the three types of IP₃R, by using IP₃R1, 2, and 3 shRNA, we identified that IP₃R subtype 1 and 2 are the major contributors of TFP-induced Ca²⁺ increase in glioblastoma (Fig. 4A). Furthermore, we identified that activation of IP₃R1 by TFP can be critical for the cell viability of GBL cells (Fig. 4B). TFP-induced Ca²⁺ increase is also associated with CaM, which is involved in modulation of Ca²⁺ signaling through interaction with Ca²⁺ (28, 29). We found that CaM2 is the major type among three types of CaM and is overexpressed in GBL cell lines compared with NSCs (Fig. 4C). Moreover, the gene silencing of CaM2 by shRNA reduced TFP-induced Ca²⁺ increase significantly (Fig. 4D).

Regulation of IP₃R opening has been investigated in previous studies: one of the well-known mechanisms is the regulation of IP₃R by CaM and Ca²⁺ (24, 25). According to this mechanism, IP₃R is blocked by Ca²⁺/CaM complex, which binds to the N-terminal domain of IP₃R under normal condition (24, 25). TFP, also known as a CaM antagonist, binds to CaM and induces conformational change and functional loss of CaM (22). On the basis of these well-known mechanisms, we hypothesized that TFP inhibits CaM function, which normally inhibits IP₃R opening, and subsequently induces robust Ca²⁺ release from ER by disinhibiting (or relieving the inhibition of) IP₃R (Fig. 6D). In fact, when we used CaM2 shRNA and CaM2-Q120A, which has a mutation in TFP-binding site with CaM2 shRNA-insensitive sequence, TFP-induced Ca²⁺ increase was impaired (Fig. 4D), supporting our hypothesis.

In our previous study, we demonstrated that caffeine inhibits intracellular Ca²⁺ increase by IP₃R3, whose expression level is increased in glioblastoma cells compared with normal glial cells (5). We further demonstrated that caffeine blocks glioblastoma migration, invasion in vitro, and extends survival in the xenograft mouse model of glioblastoma by inhibiting IP₃R3-mediated Ca²⁺ release from ER (5). These results suggested that inhibiting the Ca²⁺ signaling could suppress the aggressive motility and invasiveness of glioblastoma cells (5). On the other hand, TFP increased intracellular Ca²⁺ in glioblastoma cells by inducing opening of IP₃R1 and 2. Consequently, TFP inhibited glioblastoma migration and invasion in vitro and in vivo. These results indicated that aberrant increase in the Ca²⁺ signaling pathway could also inhibit the motile and invasive nature of glioblastoma cells. Intracellular Ca²⁺ is important for the regulation of various intracellular events, such as metabolism, gene expression, and survival, and therefore, it is highly regulated (30). If Ca²⁺ homeostasis is broken, regardless of overload or lack of intracellular Ca²⁺, cells experience adverse effects, such as cell death. Therefore, both caffeine and TFP could have a similar inhibitory effect on glioblastoma growth and invasion by either inhibiting or increasing intracellular Ca²⁺, respectively.

The practical dose of TFP for mental disorder therapy is usually 10 mg/day for adult (bodyweight, 60 kg) and is adjusted according to the degree of the side effects (13). A dose of 100 to 150 mg/day is also used in a particular case of patients (13). In the current study, a dose of TFP at 5 mg/kg was used for skin xenograft model (Fig. 2), and a dose of 10 mg/kg was used for orthotopic brain xenograft model (Fig. 6). The dose of TFP used for animal experiments was much higher than that for patients with a mental disorder. Therefore, we were concerned about potential side effects of TFP experimental animals, such as Parkinsonism and extrapyramidal side effect (EPS) as described previously (12). However, at a dose of 5 mg or 10 mg/kg/day in our animal experiments, very little signs of side effects were observed, and even if EPS was observed in a few cases, it disappeared within a few minutes. Therefore, we concluded that at the current doses, there might be very few side effects by TFP.

Recently, numerous studies have focused on drug repositioning, namely discovering new indications for existing drugs for cancer therapy (31–33). Repositioning of currently used medications can be time saving and money saving in terms of research funding. It has been recently proposed that new glioblastoma treatment approach using nine drugs that are currently marketed for cytotoxic chemotherapy along with temozolomide could have a therapeutic effect in recurrent glioblastoma. The nine adjuvant drug regimen, so called coordinated undermining of survival paths (CUSP9), includes aprepitant, artesunate, auranofin, captopril, celecoxib, disulfiram, itraconazole, ritonavir, and sertraline, augmenting continuous low-dose temozolomide (34, 35). It has been reported that glioblastoma cell viability was markedly decreased in CUSP9-treated cells compared with temozolomide alone (35). In addition to CUSP9, here, we propose another antiglioblastoma drug TFP that satisfies the concept of drug repositioning. It has been reported that TFP, a well-known antipsychotic drug, has shown anticancer effect in several cancer cell lines, such as lung cancer, hepatocellular carcinoma, and T-cell lymphoma (14–21). Moreover, TFP was viewed as a potential reagent for glioblastoma treatment due to its safety and high permeability of blood–brain barrier (36). In this study, we identified a new use of TFP for glioblastoma therapy and delineated detailed molecular and cellular mechanism of TFP action on glioblastoma. Our study proposes TFP as a valuable therapeutic drug for glioblastoma and also calls for future development of more effective TFP derivatives.

No potential conflicts of interest were disclosed.

Conception and design: S. Kang, J.M. Lee, S.H. Paek, E.J. Roh, S.S. Kang

Development of methodology: J. Hong, J. Choi, S.S. Kang

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): S. Kang, H.E. Moon, S.H. Paek, C.J. Lee, S.S. Kang

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): J. Hong, J.M. Lee, J. Choi, S.H. Paek, C.J. Lee, S.S. Kang

Writing, review, and/or revision of the manuscript: S. Kang, J.M. Lee, N.A. Yoon, C.J. Lee, S.S. Kang

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): S. Kang, J.M. Lee, N.A. Yoon, E.J. Roh, C.J. Lee, S.S. Kang

Study supervision: S.H. Paek, S.S. Kang

Other (synthesis of TFP derivatives): B. Jeon

We wish to thank Richard Eric Kast (Department of Psychiatry, University of Vermont, Burlington, VT), who provided original ideas about TFP and glioblastoma.

This study was supported by the National Research Foundation of Korea (NRF) and by a grant funded by the Korean Government (MEST) nr. 2013R1A2A2A01068964 (to S.S. Kang), the Creative Research Initiative Program, the Korean National Research Foundation 2015R1A3A2066619 (to C.J. Lee), the KU-KIST Graduate School of Science and Technology program R1435281 (to C.J. Lee), Brain Research Program through the NRF funded by Ministry of Science, ICT & Future Planning NRF-2012M3C7A1055412 (to C.J. Lee), KIST Institutional Grant 2E22662 (to C.J. Lee), the Creative Fusion Research Program through the Creative Allied Project funded by the National Research Council of Science & Technology CAP-12-1-KIST (to E.J. Roh), the Technology Innovation Program 10050154 (Business Model Development for Personalized Medicine Based on Integrated Genome and Clinical Information) funded by the Ministry of Trade, Industry & Energy (MI, Korea; to S.H. Paek), and the Bio & Medical Technology Development Program of the NRF funded by the Korean government, MSIP 2015M3C7A1028926 (to S.H. Paek).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.