Endoplasmic reticulum-Golgi intermediate compartment protein 3 knockdown suppresses lung cancer through endoplasmic reticulum stress-induced autophagy

Trafficking from the endoplasmic reticulum (ER) to the Golgi apparatus is elevated in cancer cells. Therefore, proteins of the ER-Golgi intermediate compartment (ERGIC) attract significant attention as targets for cancer treatment. Enhanced cancer cell growth and epithelial-mesenchymal transition by ERGICs correlates with poor-prognosis of lung cancer. This prompted us to assess whether knockdown of ERGIC3 may decrease lung cancer growth. To test the hypothesis, the effects of ERGIC3 short hairpin RNA (shERGIC3) on ER stress-induced cell death and lung tumorigenesis were investigated both in vitro and in vivo. Knockdown of ERGIC3 led to ER stress-induced autophagic cell death and suppression of proliferation in the A549 human lung cancer cell-line. Moreover, non-invasive aerosol-delivery of shERGIC3 using the biocompatible carrier glycerol propoxylate triacrylate and spermine (GPT-SPE) inhibited lung tumorigenesis in the K-rasLA1 murine model of lung cancer. Our data suggest that suppression of ERGIC3 could provide a framework for the development of effective lung cancer therapies.


INTRODUCTION
The endoplasmic reticulum-Golgi intermediate compartment (ERGIC) is a dynamic and mobile early secretory pathway located between the endoplasmic reticulum (ER) and the Golgi apparatus in mammalian cells [1]. ER-Golgi trafficking is elevated in cancer cells; therefore, ER-and Golgi-related proteins could be good targets for cancer therapy [2].
ERGIC3 is a human ER-related 43-kDa protein (ERp43). Over-expression of ERGIC3 promotes cell growth and significantly reduces ER stress-mediated cell death [3].
Secretory proteins are properly folded by various enzymes called chaperones, which function in protein folding and promote degradation of misfolded polypeptides [7]. Calcium depletion in the ER lumen, glucose deprivation, inhibition of asparagine (N)-linked glycosylation, and disulfide bond reduction can induce the accumulation of unfolded proteins and disrupt ER function, a phenomenon known as ER stress [8]. Under ER stress conditions, attenuation of translation as well as degradation responses and transcriptional induction of chaperone proteins are activated to prevent unfolded protein accumulation. To increase ER folding capacity, the intracellular signaling pathway known as the unfolded protein response (UPR) is activated, resulting in transcriptional up-regulation of ER resident proteins [9].
ER stress can trigger autophagy, which is considered a downstream mediator in ER stress-induced cell death [10,11]. Autophagy, known as cellular self-digestion, includes the degradation of cytoplasmic proteins and organelles. Autophagy is activated in cells for protein clearance in conditions such as starvation, irradiation, and hypoxia or in diseases such as cancer [12]. When ER stress or damage cannot be repaired by the UPR pathway, cell death may be induced [13,14].
ERGIC3 is over-expressed in cancer, and also correlates with oncogenic lung diseases. Moreover, the relation between increased cancer cell growth and epithelial-mesenchymal transition with ERGIC3 led us to investigate the therapeutic effect of ERGIC3 downregulation in lung cancer. In this study, we show that delivery of a small hairpin RNA targeting ERGIC3 (shERGIC3) suppresses lung cancer through ER stressinduced autophagic cell death in both in vitro and in vivo experiments.

Expression of ERGIC3 in human normal lung and lung adenocarcinoma tissues
ERGIC3 was over-expressed in human lung adenocarcinoma (Grade I, II and III) compared to normal lung tissues. Twenty human tissue samples (five samples per group) were analyzed using Western blot and densitometry ( Figure 1A and Supplementary Figure S1).

Effect of ERGIC3 knockdown on cell proliferation and Akt1 phosphorylation
Over-expression of ERGIC3 in lung cancer tissue led us to study the effect of ERGIC3 knockdown for lung cancer therapy. Sequence number 58 small hairpin RNA (shRNA) targeting ERGIC3 (shERGIC3-58) showed the best silencing efficacy compared to other target sequences (Supplementary Figure S2). Therefore, this target sequence was chosen for further experiments. Stable cell lines expressing scrambled vector (shScr) or shERGIC3 were generated in A549 cells. Significant silencing of ERGIC3 downregulated stable cell line was confirmed by quantitative PCR (qPCR) ( Figure 1B) and Western blot (Figure 2A). Using the real-time xCELLigence proliferation detection system, ERGIC3 downregulated stable cells showed a remarkable suppression of cell proliferation compared to control (*P < 0.05) and shScr ( ## P < 0.01) groups ( Figure 1C). Moreover, proliferating cell nuclear antigen (PCNA) level was also decreased in the shERGIC3 stable cell line ( Figure 1D). In addition, Akt1 (protein kinase B) phosphorylation at Thr308 and Ser473 residues was significantly decreased in ERGIC3 downregulated cells ( Figure 1E). Therefore, we confirmed that suppression of ERGIC3 decreased Akt1 activation and lung cancer cell proliferation.

Knockdown of ERGIC3 triggers ER stressinduced autophagy in lung cancer cell lines
Expression of ER stress-related proteins was assessed using Western blot. In the shERGIC3 stable cell line, expression of C/EBP homology protein (CHOP), Calnexin, and inositol-requiring ER to nucleus signal kinase-1α (IRE1α) was up-regulated compared to other control groups. Moreover, induction of autophagy by ERGIC3 downregulation was observed with the increase of the light chain (LC) 3-II to LC3-I ratio. Degradation of p62/ sequestosome-1 (SQSTM1) was induced by an increase in autophagy ( Figure 2A). These results were also confirmed by both using another shRNA targeting ERGIC3 (shERGIC3-59, GTGGAACACAACCTGTTCAAGCAACGACT) and another human lung cancer cell line H460 (Supplementary Figure S3). Moreover, after treatment of chloroquine, p62/SQSTM1 was more degraded in shERGIC3 stable cell line compared to shScr (Supplementary Figure S5A).
Dilated ER, swelling of the Golgi apparatus, vacuoles in mitochondria and distorted nuclei were observed in shERGIC3 stable cells in transmission electron microscopy (TEM) images. Engulfed organelles in autophagosomes and autolysosomes were also found in ERGIC3 downregulated stable cell line ( Figure 2B, Supplementary Figure S4).
Oncotarget 65338 www.impactjournals.com/oncotarget localized to the ER in normal condition (control) and partially localized to the ER in ER stress conditions ( Figure 3A). Furthermore, ERGIC3 was partially colocalized to the cis-Golgi matrix protein 130 (GM130) ( Figure 3B). To examine the expression of ERGIC3 in ER stress conditions, A549 cells were treated with various concentrations (0.1-5 μg/mL) of tunicamycin. Expression of ERGIC3 was decreased in ER stress condition induced by tunicamycin ( Figure 3C). However, co-treatment with tunicamycin and tauroursodeoxycholic acid (TUDCA; an ER stress inhibitor) did not decrease ERGIC3 expression ( Figure 3D). Moreover, treatment of TUDCA inhibits changes of ER stress and autophagy-related proteins in ERGIC3 downregulated cell line compared to TUDCA non-treated cell line (Supplementary Figure S5B).

ERGIC3 overexpression alleviates ER stress and promotes cell proliferation
Overexpression of ERGIC3 alleviated ER stress induced by tunicamycin treatment. After treatment with tunicamycin, expression of ER stress proteins including CHOP, Calnexin, IRE1α, p62/SQSTM1 and LC3 in A549 control, shScr and shERGIC3 groups. Cells were cultured for 2 days and then collected for Western blot analysis. Significant differences are indicated by *P < 0.05, **P < 0.01 and ***P < 0.001 (*: compared to control, #: compared to shScr, n = 4). (B) Subcellular morphological changes in shERGIC3 stable cells were revealed by transmission electron microscopy (TEM) analysis. Cells were cultured for 2 days then were fixed and observed using TEM. In each group of figures, calibration bars in the left figure of each group = 2 μm, while those in the right figure of each group = 0.5 μm. Arrow indicates distorted nucleus. N, nucleus; G, Golgi apparatus; ER, endoplasmic reticulum; M, mitochondria; A, autophagosome; AL, Autolysosomes; shScr, small hairpin scramble; shERGIC3, small hairpin ERGIC3.

Aerosol delivery of shERGIC3 suppresses lung cancer by affecting ER stress-induced autophagy, proliferation and angiogenesis in K-ras LA1 mice
To test the silencing efficacy of four targets of mouse ERGIC3 shRNA, Western blot was performed after transfection of shRNAs into LA-4 (mouse lung adenoma) cells. Target sequence A showed the best silencing effect compared to other targets (Supplementary Figure S7); therefore, it was chosen for gene delivery to mice.
Repeated aerosol delivery of shERGIC3 using glycerol propoxylate triacrylate spermine (GPT-SPE) complexes significantly reduced the total number of tumors as well as tumors larger than 1 mm in K-ras LA1 mice compared to control and shScr-treated groups ( Figure  4A, 4B). Significant suppression of ERGIC3 expression in the lungs of mice treated with shERGIC3 was confirmed using qPCR and Western blot ( Figure 5). Histological assessment revealed a clear reduction in tumor size, and no adenocarcinoma was observed in the shERGIC3-treated group. Reduced number of adenoma and hyperplasia analysis of ERGIC3 (Green) and an ER marker (Red). After transfection of the ER marker, tunicamycin (1 μg/mL) was applied for 24 h to A549 cells. After fixation, incubation with an anti-ERGIC3 antibody was performed for 24 h. Calibration bars = 10 μm. Green, ERGIC3; Red, ER marker; Blue, nucleus. (B) ERGIC3-tGFP was transfected into A549 cells for 24 h, followed by incubation with anti-GM130 antibody for 24 h, and nuclei staining (DAPI). Calibration bars = 10 μm. Green, ERGIC3; Red, GM130; Blue: nucleus (C) Western blot analysis of ERGIC3 and Calnexin expression depending on tunicamycin concentration (0.1 to 5 μg/mL) for 24 h. (D) Pre-treatment with 5 mg/mL tauroursodeoxycholic acid (TUDCA) was performed for 2 h, followed by 5 μg/mL tunicamycin for 24 h. Western blot assays of ERGIC3, Calnexin and GAPDH were performed after sampling cell lysates. Con, control; Tuni only, tunicamycin treated; TUDCA, TUDCA pretreated and tunicamycin treated).
Oncotarget 65340 www.impactjournals.com/oncotarget foci was also observed in the shERGIC3-treated group compared to other control groups ( Figure 4C, Table 1). In TEM images of the lungs of shERGIC3-treated mice, dilated ER and ingested organelles in autophagosomes were observed (Supplementary Figure S8). To assess whether aerosol delivery of shERGIC3 stimulated ER stress-induced autophagy in K-ras LA1 mice, ER stressand autophagy-related proteins were analyzed. CHOP and IRE1α expression levels increased as a result of ERGIC3 suppression. LC3-II expression also increased, while p62 level decreased because of autophagy in the lung tissues of mice treated with shERGIC3 ( Figure 5B). The expression of matrix metalloproteinase-9 (MMP-9), proliferating cell nuclear antigen (PCNA), vascular endothelial growth factor (VEGF) and cyclin B1 was also significantly decreased in the shERGIC3-treated group compared to other control groups ( Figure 6A). Decreased Akt1 phosphorylation (residues Thr308 and Ser473) was also observed in the lungs of shERGIC3-treated mice, similarly to the in vitro studies ( Figure 6B).

DISCUSSION
Cancer statistics show that lung cancer causes the highest number of estimated deaths in the United States [15]. Gene therapy is a promising and efficient approach for the suppression of tumorigenesis by directly influencing the defective genes [16]. Moreover, this therapy has been Oncotarget 65341 www.impactjournals.com/oncotarget applied to various diseases, and the efficiency of delivery is essential for the success of this therapy [17,18]. Aerosol delivery enables non-invasive and efficient delivery of macromolecules to the lung [19,20]. Moreover, aerosol delivery using various cationic carriers and viral vectors has been successfully used in lung cancer therapy [21]. We previously developed a GPT-SPE cationic polymer which showed biocompatibility and high transfection efficiency [22]. Therefore in this study, GPT-SPE cationic carrier was used for aerosol delivery of shERGIC3 to K-ras LA1 mice.
In cancer cell, ER to Golgi trafficking is elevated, therefore, the ER to Golgi network has emerged as a potential target for cancer treatment [2,23]. Our data shows that ERGIC3 is localized to ER and cis-Golgi ( Figure 3A, 3B). Some binding partner proteins of ERGIC3 are also localized to ER and Golgi apparatus (Supplementary  Table S1) Moreover, knockdown of ERGIC3 suppressed proliferation and tumor growth (Figures 1, 4). Although the role of ERGIC3 in membrane trafficking still needs to be proven, our results may indicate the relationship between ERGIC3 and ER to Golgi transport.
In our study, we focused on the anti-cancer effect of the ERGIC3 downregulation. Suppression of ERGIC3 increased ER stress-related proteins and LC3-II levels ( Figure 2A). Furthermore, treatment of tunicamycin induced suppression of ERGIC3 and increase of Calnexin expressions. Increase of ER stress proteins may be due to the decrease of the ERGIC3 expression under ER stress condition ( Figure 3C). Moreover, ER dilation, Golgi swelling, autophagosomes and autolysosomes were found in the ERGIC3 downregulated stable cell line ( Figure 2B and Supplementary Figure S3). These results show that downregulation of ERGIC3 induces autophagy and ER stress in lung cancer cells.
Autophagy is a major intracellular process that degrades defective cytoplasmic organelles or protein aggregates via lysosomal machinery, and is also associated with various pathological processes [24,25]. Autophagy not only facilitates cell survival, it can also suppress tumor development. This dual function of autophagy is still controversial; however, if cellular stress is too high, an increase in autophagic flux can induce cell death [26]. Autophagic cell death is also involved in cancer cell death. Whereas reduced autophagy function is related to tumorigenesis [27] Autophagy can act as a tumor suppressor by limiting cell growth or inducing cell death [28]. Although the complex roles of autophagy need to be elucidated, we can consider activation of autophagy as a therapeutic target in cancer treatment. The p62 protein (also known as SQSTM1), is an autophagy substrate, and an increase in its expression has been demonstrated in various cancers [26,29]. The p62/SQSTM1 protein accumulates because of defective autophagy, and this is directly linked to tumorigenesis. Therefore, control of p62/SQSTM1 is crucial for cancer therapy [30,31]. The absence of p62/SQSTM1 inhibits ras-induced lung adenocarcinomas and increased cell death [32]. In this study, suppression of ERGIC3 increased LC3-II levels, and reduced p62 expression both in vitro and in vivo (Figures 2A, 5B).
UPR is activated under ER stress, and is the subcellular response for the reestablishment of ER function and adaptation to a changed environment [33]. In human lung cancer lesions, expression of ER stress proteins is linked to autophagy-related proteins such as LC3 and Beclin-1 [34]. ER stress is also an essential inducer of preautophagosomal structures, LC3 conversion and autophagy [35,36]. Moreover, ER stress-induced autophagy is activated through IRE1α mediation [14]. In our data, stimulation of ER stress in ERGIC3 knockdown status was confirmed by an increase in related proteins (i.e., IRE1α and CHOP) in vitro and in vivo. Moreover, shERGIC3 triggered autophagy and autophagosome formation (Figures 2, 5B). Therefore, we can conclude that knockdown of ERGIC3 stimulates ER stress-induced autophagy. (qPCR) analysis of ERGIC3 in lung tissue. Aerosol delivery of shERGIC3 significantly decreased ERGIC3 expression in the lungs compared to control (***P < 0.001) and shScr ( ### P < 0.001) (n = 4). (B) Western blot and densitometric analyses of ERGIC3, CHOP, IRE1α, LC3 and p62/SQSTM1 in the lungs of control, shScr and shERGIC3-treated mice. Significant differences are indicated by *P < 0.05, **P < 0.01 and ***P < 0.001 (*: compared to control, #: compared to shScr, n = 4).
Oncotarget 65343 www.impactjournals.com/oncotarget Prolonged and severe ER stress can induce cell death [33]. Additionally, ER stress can induce autophagic cell death [37,38]. ER stress-inducing reagents including tunicamycin and dithiothreitol (DTT) also aggravate autophagy and cell death [39]. Our data showed that aerosol delivery of shERGIC3 inhibited lung tumorigenesis in K-ras LA1 mice through ER stress-induced autophagy (Figures 4, 5). The number of tumors was also decreased and no adenocarcinoma was observed in shERGIC3-treated mice ( Table 1).
Activation of Akt1 is associated with the proliferation, angiogenesis, growth and survival of cancer cells. Akt1 activation is induced by phosphorylation at Thr308 and Ser473 residues [40]. Whereas inhibition of Akt1 can promote the induction of fatal ER stress and autophagy mediated cell death [41][42][43][44]. Moreover, ER stress-induced autophagy is attributed to the suppression of AKT/TSC/mTOR pathway [39]. Our data show that suppression of ERGIC3 inhibited Akt1 phosphorylation at Thr308 and Ser473 residues (Figures 1E, 6B). Increased ER stress-induced autophagy was also observed in ERGIC3 downregulated stable cell line (Figures 2, 5B). Therefore, depletion of ERGIC3 suppresses Akt1 activation as downstream of the ER stress induction. In addition, according to our analysis of proteomics, downregulation of ERGIC3 induced suppression of the cell cycle, cytokinesis and cell proliferation (Supplementary Figure S9).
In conclusion, our data showed that knockdown of ERGIC3 triggers ER stress-induced autophagic cancer cell death and the suppression of Akt1 activation (Supplementary Figure S10). Moreover, suppression of ERGIC3 is associated with the expression of cell cycle-and mitosis-related proteins. In addition to lung cancer, ERGIC3 is also associated with hepatocellular carcinomas, and its expression is enhanced in colorectal tumors [4,6]. Therefore, ERGIC3 could be a novel target and marker for various cancers and cancer gene therapy. (A) Western blot and densitometry analyses of PCNA, VEGF, MMP-9 and Cyclin B1 in lungs of control, shScr and shERGIC3-treated mice (n = 4). (B) Western blot and densitometry analyses of p-Akt1 (Thr308 and Ser473). Statistical significance is indicated by *P < 0.05, **P < 0.01 and ***P < 0.001 (*: compared to control, #: compared to shScr, n = 3).

Western blot analysis
Human lung tissue samples were obtained from the Korea Lung Tissue Bank (KLTB, Seoul, Korea). All experiments involving human tissues were authorized by the Seoul National University Institutional Review Board (SNUIRB-E1201/001-001) and KLTB (KU Guro Gene Bank 2012-004). Lung lobes were homogenized using 300 μL of a 2.5× Passive Lysis Buffer (Promega, Madison, WI, USA), while A549 cells were lysed using 1× Cell Culture Lysis Buffer (Promega). Immunoblotting was performed by incubating with the primary antibody (1:2500) at 4°C overnight, and then with a horseradish peroxidase (HRP)-conjugated secondary antibody (1:2000; Invitrogen, Carlsbad, CA, USA) for 3 h at room temperature (RT). Bands were detected using a luminescent image detector (Ez-Capture MG), and analysis was performed using the CS Analyzer program (ATTO, Tokyo, Japan).

In vivo Aerosol delivery of GPT-SPE/shERGIC3 complex
Animals were maintained under the animal guidelines of the Seoul National University, and all animal experiments were reviewed and approved by the Institutional Animal Care and Use Committee of Seoul National University (SNU-130710-4). K-ras LA1 (non-small cell murine lung cancer model) mice were obtained from the Human Cancer Consortium-National Cancer Institute (Frederick, MD, USA). Ten-week-old male K-ras LA1 mice were divided into the following three groups (6 mice per group, total 18 mice); control, small hairpin scramble vector control and shERGIC3-treated groups. Mice in the control (Con) group were untreated, while mice in the other two groups were treated with aerosol containing 8 mg GPT-SPE and 0.8 mg shRNA (small hairpin ERGIC3 or small hairpin scramble) twice a week for 4 weeks. GPT-SPE is a spermine-based biocompatible DNA carrier which can be used in shRNA-based lung cancer therapy [22]. Mice were treated with aerosol containing GPT-SPE/ shERGIC3 or GPT-SPE/shScr complex in a nose-only exposure chamber. Aerosol was generated by a nebulizer (Korean patent #20304964). Tumor lesions on the entire lung surface were carefully counted, and the diameter of the tumors was also measured using a digital caliper. Left lung lobes were collected for histopathological analysis. After necropsy, remaining lung lobes were stored at −70°C for Western blot and RNA analysis.

Histopathology of lung tissues
Left lung lobes were fixed in 10% neutral-buffered formalin, and embedded in paraffin blocks. Blocks were sectioned at 5 μm thickness, and sections were stained with hematoxylin and eosin (H&E, Sigma) for histological analysis.

Transmission electron microscope (TEM)
Mice lung tissues, A549 and stable cell lines were fixed in 2.5% glutaraldehyde (EMS, Hatfield, PA, USA) overnight. Samples were carefully washed and stained using 1% osmium tetroxide solution (EMS) for 2 h at 4°C. Samples were dehydrated with a gradient of ethanol and infiltrated in a 1:1 propylene oxide Epon resin (EMS) Oncotarget 65345 www.impactjournals.com/oncotarget mixture. Then, samples were embedded in Epon resin and polymerized for 24 h at 70°C. Ultrathin sections (40-70 nm thick) were obtained on an ultramicrotome (Leica, Nussloch, Germany), and mounted on copper grids. Finally, sectioned samples were counterstained with lead citrate and uranyl acetate. Samples were observed using a transmission electron microscope (JEOL, Tokyo, Japan).
For immunostaining, slides were washed in phosphate buffered-saline (PBS), fixed in 3% paraformaldehyde (PFA) solution for 5 min, and post fixed in a 4% PFA solution for 10 min. Samples were incubated in 3% BSA and 0.1% saponin (Sigma) in PBS for 1h at RT, for the blocking of non-specific binding sites. Primary antibodies (diluted 1:250) were incubated overnight at 4°C. After washing, cells were incubated with secondary antibodies (diluted 1:500) conjugated to Alexa Fluor 488 or 555 (Invitrogen) for 1h at RT. Slides were washed in PBS and cover slipped using Fluoroshield ™ (Sigma) for DAPI staining. Slides were observed using a confocal laser scanning microscope (LSM710, Carl Zeiss GmbH, Jena, Germany).

Cell proliferation analysis
Real-time cell proliferation analysis was performed using xCELLigence RTCA DP system (Roche Applied Science, Indianapolis, IN, USA). This device analyzes real-time cell proliferation through measurement of the electrical impedance of microelectrodes integrated in the bottom of plates. For analysis, 1.5 × 10 3 A549 cells were seeded in 16-well E-plates and incubated for 72 h.

Statistical analyses
Statistical significances of difference were analyzed using Microcal Origin student's t-test two populations (Microcal Software, Northampton, MA, USA). The significances were set by probability values (*P < 0.05, ** P < 0.01 and ***P < 0.001) compared to the corresponding values. Each bar in the graphs indicates the mean ± standard error of mean (SEM).