Endoplasmic Reticulum Stress Contributes to Gefitinib-Induced Apoptosis in Glioma

Adequate stress on the Endoplasmic Reticulum (ER) with the Unfolded Protein Response (UPR) could maintain glioma malignancy. Uncontrolled ER stress, on the other hand, predisposes an apoptosis-dominant UPR program. We studied here the proapoptotic actions of the Epidermal Growth Factor Receptor (EGFR) inhibitor gefitinib, with the focus on ER stress. The study models were human H4 and U87 glioma cell lines. We found that the glioma cell-killing effects of gefitinib involved caspase 3 apoptotic cascades. Three branches of ER stress, namely Activating Transcription Factor-6 (ATF6), Protein Kinase R (PKR)-Like ER Kinase (PERK), and Inositol-Requiring Enzyme 1 (IRE1), were activated by gefitinib, along with the elevation of intracellular free Ca2+, Reactive Oxygen Species (ROS), and NADPH Oxidase2/4 (NOX2/4). Specifically, elevated IRE1 phosphorylation, Tumor Necrosis Factor (TNF) Receptor-Associated Factor-2 (TRAF2) expression, Apoptosis Signal-Regulating Kinase-1 (Ask1) phosphorylation, c-Jun N-Terminal Kinase (JNK) phosphorylation, and Noxa expression appeared in gefitinib-treated glioma cells. Genetic, pharmacological, and biochemical studies further indicated an active ROS/ER stress/Ask1/JNK/Noxa axis causing the glioma apoptosis induced by gefitinib. The findings suggest that ER-stress-based therapeutic targeting could be a promising option in EGFR inhibitor glioma therapy, and may ultimately achieve a better patient response.


Introduction
While cancer cells are notorious for their high proliferation and quick spread, they are surrounded by hostile microenvironments such as hypoxia, nutrient deficiency, lactic acidosis, and oxidative stress. Intrinsic events from oncogenic activation, aerobic glycolysis, and high protein demands would disrupt cellular proteostasis, leading to Endoplasmic Reticulum (ER) stress. To overcome such obstacles and turn them into survival momentum, cancer cells possess several adaptive mechanisms. One of them is the ER stress-derived Unfolded Protein Response (UPR), which is crucial for their adaptation and survival [1,2]. Conversely, in the case of unsuccessful UPR in resolving ER stress, a dominant UPR program commits the cells to death [3]. The phenomena highlight a dual role of UPR in the progression of the malignancy and provide a means for therapeutic treatment. UPR is coordinated by three different signaling branches of transmembrane proteins resident on ER: Protein Kinase R (PKR)-Like ER Kinase (PERK), Inositol-Requiring Enzyme 1 (IRE1), and Activating Transcription Factor 6 (ATF6). Under a homeostatic ER, the lumenal domains of and PERK, IRE1, and ATF6 interact with ER abundant chaperone Glucose-Regulated Protein 78 (GRP78) and remain in an inactive state. Once misfolded proteins accumulate, GRP78 titrates misfolded proteins, liberating PERK, IRE1, ATF6. The liberated PERK initiates autophosphorylation and phosphorylates Eukaryotic Translation Initiation Factor-2α (eIF2α), leading to a greater expression of ATF4 and C/EBP Homologous Protein (CHOP). Activated IRE1, which possesses kinase and RNase activities, yields a spliced form of transcription factor X-Box Binding Protein-1 (XBP1) mRNA. IRE1 also recruits Tumor Necrosis Factor (TNF) Receptor-Associated Factor-2 (TRAF2) to activate the Apoptosis Signal-Regulating Kinase-1 (Ask1)/c-Jun N-Terminal Kinase (JNK) axis. When released from GRP78, ATF6 enters the Golgi apparatus, generating a cleaved fragment that is a transcription factor. Transcription factors ATF4, CHOP, XBP1, and ATF6 have shared and individual programs, governing the expression of a variety of genes involved in cell survival, angiogenesis, metastasis, immune surveillance, inflammation, autophagy, and apoptosis [4]. Regarding malignancy, ER stress and three sensory branches are promising targets for intervention.
In the central nerve system, the most aggressive malignancy is glioma, particularly the glioblastoma multiforme. Despite conventional and novel treatment strategies, patients with malignant glioma have a poor prognosis, with high recurrence rates [5]. Genetic mutation and amplification of the Epidermal Growth Factor Receptor (EGFR) in glioma highlight an opportunity to treat patients using EGFR tyrosine kinase inhibitors. Although EGFR inhibitors have promising potential, clinical application remains challenging with monotherapy or combination therapy in combating malignant glioma. Clinical findings reveal that only 10-20% of malignant glioma patients benefit from the gefitinib and the coexpression of the EGFRvIII oncogene, and the PTEN tumor suppressor protein in patients favors a clinical response to EGFR inhibitor therapy. Besides genetic status, the combination therapy with gefitinib and temozolomide represents an alternative option [6][7][8][9][10][11]. The phenomena underscore the importance of a better understanding of anti-neoplastic mechanisms caused by EGFR inhibitors such as gefitinib.
Glioma cells over-express markers of UPR, which contribute to proliferation, angiogenesis, resistance, and stem cell-like activity, and are correlated with a poor prognosis [12][13][14]. However, anti-glioma treatments cause ER stress and UPR. With extrinsic insults and intrinsic alterations, glioma cells become sensitized to ER stress, leading to apoptosis [15][16][17][18][19]. The lung cancer cells with EGFR inhibitor gefitinib-resistance express higher levels of GRP78, and lower levels of CHOP. The induction of ER stress sensitizes these cancer cells to gefitinib therapy [20,21]. Currently, the alteration and role of ER stress in gefitinib-treated glioma cells remain unclear. In our previous studies on glioma cells, we reported that gefitinib induces apoptosis and autophagy, and the ER stress contributes to glioma apoptosis [15,16,[22][23][24]. To extend the finding on ER stress, we conducted this study to determine the role of ER stress in gefitinib-induced glioma apoptosis, and identify the molecular basis underlying the UPR-committed apoptotic program.

BAPTA-AM and N-Acetyl-Cysteine (NAC) Alleviated Gefitinib-Induced Glioma Apoptosis in H4 Cells
Gefitinib increased intracellular levels of free Ca 2+ and induced ROS generation in

BAPTA-AM and N-Acetyl-Cysteine (NAC) Alleviated Gefitinib-Induced Glioma Apoptosis in H4 Cells
Gefitinib increased intracellular levels of free Ca 2+ and induced ROS generation in H4 cells ( Figure 1C,D). To further study their roles, we applied intracellular calcium chelating agent BAPTA-AM and antioxidant NAC to these cells. BAPTA-AM ( Figure 4A-C) and NAC ( Figure 4D-F) alleviated gefitinib-induced cell viability loss ( Figure 4A,D), caspase 3 activation (Figure 4B,E) and protein hyperphosphorylation of IRE1, Ask1, and JNK ( Figure 4C,F). NAC also alleviated gefitinib-induced intracellular free Ca 2+ ( Figure 4G). Gefitinib had increased NADPH Oxidase-2 (NOX2) and NOX4 protein expression in H4 cells ( Figure 4H). These NOX family members are likely involved in IRE1-mediated ER stress [26]. Our findings suggest substantial roles played by free Ca 2+ and ROS in the gefitinib-induced ER stress and apoptosis, with NOX2/4 as candidate sources for ROS generation.

Gefitinib Caused Noxa Upregulation in Glioma Cells
BH3-only proteins are candidate downstream effectors in linking ROS/ER stress and apoptosis [27]. Gefitinib caused an elevated expression of the Noxa protein in H4 cells and the elevation was alleviated by SP600125 ( Figure 6A). Silence of endogenous Noxa ( Figure 6B) protected H4 cells against gefitinib-induced cell viability loss ( Figure 6C) and caspase 3 activation ( Figure 6D). The above-mentioned findings were duplicated in U87 cells ( Figure 6E-H). Findings were consistent with ROS/ER stress/Ask1/JNK/Noxa axis being an apoptotic cause in gefitinib-treated glioma cells.

Gefitinib Caused Noxa Upregulation in Glioma Cells
BH3-only proteins are candidate downstream effectors in linking ROS/ER stress and apoptosis [27]. Gefitinib caused an elevated expression of the Noxa protein in H4 cells and the elevation was alleviated by SP600125 ( Figure 6A). Silence of endogenous Noxa ( Figure 6B) protected H4 cells against gefitinib-induced cell viability loss ( Figure 6C) and caspase 3 activation ( Figure 6D). The above-mentioned findings were duplicated in U87 cells ( Figure 6E-H). Findings were consistent with ROS/ER stress/Ask1/JNK/Noxa axis being an apoptotic cause in gefitinib-treated glioma cells.

Discussion
Over-expressions of GRP78 and UPR components have been implicated in malignant glioma of aggressive phenotypes, while ER stress also predisposes glioma cells to

Discussion
Over-expressions of GRP78 and UPR components have been implicated in malignant glioma of aggressive phenotypes, while ER stress also predisposes glioma cells to apoptosis upon therapeutic treatments [10][11][12][13][14][15][16][17]19]. Although ER stress sensitizes lung cancer cells to gefitinib therapy [20,21], the involvement of ER stress in the actions of gefitinib on glioma cells is not clear. Continuing our earlier study on gefitinib-mediated glioma apoptosis [22], we had found here that gefitinib-induced glioma apoptosis was parallel with the following events: elevated GRP78, ATF4, and CHOP protein expressions; PERK, eIF2α, and IRE1 protein phosphorylation; ATF6 proteolytic cleavage; free Ca 2+ mobilization; and ROS generation, crucial signs of ER stress. Pharmacological and genetic studies further identified an ROS/ER stress/Ask1/JNK/Noxa axis involving NOX2/4 in mediating gefitinib-induced glioma apoptosis (Figure 7). Besides EGF/EGFR signaling, current findings are consistent with the ROS/ER stress axis likely being a target for the anti-glioma actions of gefitinib. Components of the UPR own both shared and individual biological activities. Glioma strongly expresses PERK, ATF4, ATF6, IRE1, and XBP1, and is associated with the malignant phenotype and poor prognosis [12,14,28]. Furthermore, IRE1 is also implicated in the expression of EGFR ligand epiregulin [29]. The knockdown of PERK, ATF6, or IRE1 reduces glioma cell viability with a greater sensitivity to stress-induced cell death [30][31][32]. Conversely, ER stress also leads to glioma cell death [15][16][17][18][19][20]. In an earlier study, we demonstrated that aspirin and indomethacin increased UPR component protein expression, and protein phosphorylation, resulting in glioma apoptosis. ER stress inhibitors such as salubrinal and 4-phenylbutyrate, calcium chelator BAPTA-AM, and antioxidant PDTC alleviate glioma apoptosis [15,16]. Upon gefitinib treatment, glioma cells elevated UPR components in protein expression and protein phosphorylation, along with apoptosis. Gefitinib-induced glioma apoptosis was suppressed by 4-Phenylbutyrate, BAP-TA-AM, NAC, IRE1 silencing, and Noxa silencing. Although the fine-tuning of the UPR in directing glioma cells to survival or death is not fully understood, our current findings have suggested that therapeutic or cancer cell-killing agents commit UPR to the pro-death program.
Regarding the apoptotic program, Bcl-2 family members are transcriptional targets Components of the UPR own both shared and individual biological activities. Glioma strongly expresses PERK, ATF4, ATF6, IRE1, and XBP1, and is associated with the malignant phenotype and poor prognosis [12,14,28]. Furthermore, IRE1 is also implicated in the expression of EGFR ligand epiregulin [29]. The knockdown of PERK, ATF6, or IRE1 reduces glioma cell viability with a greater sensitivity to stress-induced cell death [30][31][32]. Conversely, ER stress also leads to glioma cell death [15][16][17][18][19][20]. In an earlier study, we demonstrated that aspirin and indomethacin increased UPR component protein expression, and protein phosphorylation, resulting in glioma apoptosis. ER stress inhibitors such as salubrinal and 4-phenylbutyrate, calcium chelator BAPTA-AM, and antioxidant PDTC alleviate glioma apoptosis [15,16]. Upon gefitinib treatment, glioma cells elevated UPR components in protein expression and protein phosphorylation, along with apoptosis. Gefitinib-induced glioma apoptosis was suppressed by 4-Phenylbutyrate, BAPTA-AM, NAC, IRE1 silencing, and Noxa silencing. Although the fine-tuning of the UPR in directing glioma cells to sur-vival or death is not fully understood, our current findings have suggested that therapeutic or cancer cell-killing agents commit UPR to the pro-death program.
Regarding the apoptotic program, Bcl-2 family members are transcriptional targets of ATF4, CHOP, XBP1, and ATF6 [4]. Other evidence indicated that IRE1/TRAF2/Ask1/JNK, p38 represents an alternative mechanism to initiate apoptosis by phosphorylating and inhibiting anti-apoptotic activity of Bcl-2 members, or activating pro-apoptotic function of BH3-only proteins [33,34]. In indomethacin-treated glioma cells, activated ER stress/Ask1/p38 axis causes Akt inactivation and Mcl-1/FLIP downregulation, resulting in cell apoptosis [15]. Aspirin causes ER stress and Noxa-mediated glioma apoptosis and the death program is alleviated by silencing PERK or eIF2α [16]. Besides apoptotic induction, autophagic proteins synthesis and autophagic complexes assembly are also targets of ER stress. Gefitinib induces autophagic cell death in glioma cells involving ROS generation, and valproic acid augments ROS generation and autophagy in gefitinib-treated glioma cells [23,24]. Herein, we demonstrated an ROS/ER stress/IRE1/Ask1/JNK/Noxa axis being actively involved in the gefitinib-induced glioma apoptosis. Since JNK and Noxa have a dominant role in glioma apoptosis [35,36], the current findings further highlight a pharmacological target of JNK/Noxa in glioma therapeutic development.
ER-associated NOXs is one means of turning on the IRE1/JNK/Noxa axis through ROS generation [26,27]. Gefitinib increased NOX2/4 protein expression and ROS generation in glioma cells. Their ER stress caused by gefitinib were alleviated by NAC. Oxidative stress is typically paralleled by EGFR phosphorylation, and through the EGFR signaling, protecting against oxidative stress-induced apoptosis. Erlotinib, by inhibiting EGFR, induces metabolic oxidative stress through NOX4 in human head and neck cancer cells [37]. NOXs are overexpressed in glioma cells and support aerobic glycolysis and malignancy [38]. Although there is no direct evidence, our findings suggest that gefitinib interrupting the compensatory interplay between NOXs and EGFR signaling could lead to an overwhelmed NOX2/4 activation and ROS generation. The consequence is convergence of signals to an apoptosis dominant UPR program.
UPR components and biological consequences could vary. Their exact relationship highly and dynamically depends on the microenvironment. Apart from IRE1 and JNK, the precise involvement of NOXs, other UPR components, downstream effectors, and transcriptional events remain to be explored. Other than gefitinib, the effects of clinical relevant EGFR inhibitors centered on glioma apoptosis are of interest. Additionally, despite the encouraging apoptotic and autophagic consequences in cell studies [22][23][24], EGFR inhibitors, including gefitinib, have only marginal benefits in patients with recurrent malignant glioma. The clinical response to EGFR inhibitor monotherapy is limited to a certain genetic status such as EGFRvIII, PTEN, and IDH1. To improve clinical response, combination therapy could be an option [6][7][8][9][10][11]. These to-be-solved issues should be addressed in future studies.
In conclusion, we found that gefitinib exhibited anti-neoplastic effects against glioma H4 and U87 cells through apoptosis, involving the ROS/ER stress/Ask1/JNK/Noxa axis. Doses of gefitinib (~40 µM) inducing apoptotic cell death in glioma cells are way higher than those inhibiting EGFR signaling (~5 µM). Once gefitinib treatment is effective, ER stress overrides survival machinery, predisposing cells to the apoptosis dominant program. Therefore, ER stress might well be valuable targets in combination therapy using gefitinib to combat malignant glioma.

Cell Cultures
Human U87 glioblastoma cell line (ATCC HTB-14) and H4 neuroglioma cell line (ATCC HTB-148) purchased from American Type Culture Collection (ATCC, Manassas, VA, USA) were maintained in Dulbecco's modified Eagle medium (DMEM), containing 10% fetal bovine serum (FBS) for propagation. All experiments were conducted on cells placed in DMEM containing 2% FBS. H4 cells harbor genetic mutations in PTEN and one copy loss in TP53, but are not EGFRvIII variant. U87 cells are PTEN-deficient and TP53and EGFR-wild type. Both cells are Isocitrate Dehydrogenase (IDH)-wild type [22].

Cell Viability Assay
Cells were plated onto 96-well plates 24 h prior to treatment. Cell viability was determined with an assay kit (alamarBlue™ Cell Viability Reagent) (ThermoFisher Scientific, Waltham, MA, USA). The optical absorbance at 570 and 600 nm was measured with a spectrophotometer.

Caspase 3 Activity Assay
Cells were plated onto 6-well plates 24 h prior to treatment. Caspase 3 activity was measured within isolated protein extracts (10 µg) according to instructions of the Caspase Fluorometric Assay kit (BioVision, Mountain View, CA, USA). Fluorescent signals of released AMC moiety were measured with a fluorometer (E x 380 nm and E m 460 nm).

Measurement of ROS
Cells were plated onto 96-well plates 24 h prior to treatment. To detect ROS, cells were loaded with 2 ,7 -Dichlorofluorescein Diacetate (DCFDA, 5 µM; Molecular Probes, Eugene, OR, USA) for 30 min. Fluorescent signals were then measured in a fluorometer with excitation/emission at 495 nm/529 nm.

Measurement of Cytosolic Ca 2+
Cells were plated onto 96-well plates 24 h prior to treatment. To detect intracellular free Ca 2+ , cells were loaded with Fura-2-Acetoxymethyl Ester (Fura-2 AM, 4 µM; Molecular Probes, Eugene, OR, USA). Fluorescent signals were measured in a fluorometer with dual excitation at 340 nm and 380 nm, with emission detected at 510 nm.

Western Blot
Cells were plated onto 6-well plates 24 h prior to treatment. At the end of treatments, cells were homogenized in the Laemmli SDS buffer. Equal amounts of protein extracts were separated and analyzed in a standard SDS-PAGE, followed by reacting to horseradish peroxidase-labeled IgG, Enhanced Chemiluminescence (ECL) visualization, and densitometric measurement. Primary antibodies were used to recognize the following proteins:

Statistical Analyses
Data were represented as mean ± standard deviation, and depicted with Sigma Plot 12.3 software. Statistical comparisons were analyzed using one-way or two-way analysis of variance, followed by Tukey or Dunnett post hoc test. Statistical significance was set at p < 0.05.