Glioblastoma Proto-oncogene SEC61γ Is Required for Tumor Cell Survival and Response to Endoplasmic Reticulum Stress

In tumors, hypoxia and nutrient deprivation of the solid-tumor microenvironment inevitably subject tumor cells to ongoing cell stress (1–3). Under conditions of reduced oxygen and/or glucose availability, tumors exploit cellular stress response pathways to support tumor cell proliferation (4). The unfolded protein response (UPR) is a set of cytoprotective signaling pathways that expands the endoplasmic reticulum (ER) processing capacity for misfolded proteins and activates adaptive, antiapoptotic pathways (5, 6). However, if the survival mechanisms are exhausted, the UPR can commit cells to apoptosis (7). These findings may be of particular relevance to glioblastoma multiforme, which, because they are highly proliferative tumors, display substantial microheterogeneity in glucose and oxygen delivery. Recent studies show that UPR genes are upregulated in gliomas (8), suggesting that the genetic alteration of the ER-based, protein folding machinery, a prominent metabolic feature of glioblastoma multiforme, confers a growth advantage.

SEC61γ is a member of the SEC61 translocon, a heterotrimeric protein channel comprising three subunits, SEC61 α, β, and γ (9). The SEC61 complex forms a transmembrane pore for the translocation of nascent polypeptides into the ER lumen as well as the integration of transmembrane proteins into the ER bilayer (10). In addition to the SEC61 complex, other proteins, such as ERj1, SEC62, and SEC63, form the complete protein translocase involved in protein folding, modification, and translocation (11). Mutations of ER-resident chaperones and the translocon have been identified in cancer cells, indicating that ER proteins play an important role in tumor pathogenesis. Of particular note, SEC62 and SEC63 are among the most frequently mutated and/or overexpressed genes in prostate, gastric, and colorectal cancers (12).

In the study described here, we found that SEC61γ is not only always coamplified with epidermal growth factor receptor (EGFR) in 47% of glioblastoma multiforme but also overexpressed in 77% of glioblastoma multiforme. In addition, SEC61γ is required for tumor cell survival and response to ER stress reagents.

Glioblastoma multiforme cell lines and frozen xenograft tumor samples were obtained from the tissue bank of the Preston Robert Tisch Brain Tumor Center at Duke. Acquisition of tissue specimens was approved by the Duke University Health System Institutional Review Board.

Digital karyotyping library construction and data analysis based on 25 glioblastoma multiforme samples were done as described previously (13, 14). Digital karyotyping protocols and software for extraction and analysis of genomic tags are available online.⁵

Single nucleotide polymorphism genotyping on genomic DNA from 32 glioblastoma multiforme samples, including 7 pediatric glioblastoma multiforme, was done using the Illumina HumanHap550 Genotyping BeadChip array. Raw data from the single nucleotide polymorphism chips were collected and subjected to copy-number analysis using Nexus Copy Number Professional software (BioDiscovery).

Differences between glioblastoma multiforme and normal brain cells in genomic DNA content of EGFR or SEC61γ were determined by quantitative real-time PCR (Q-PCR) as described previously (15). Relative gene expression levels of ATF4, Bip, CHOP, ATF6, Xbp1, EGFR, cyclin D1, and SEC61 α, β, and γ were measured before and after treating the cells with 1 μg/mL of the ER stress inducer tunicamycin, and cDNA content was normalized to that of glyceraldehyde-3-phosphate dehydrogenase.

Human U133A GeneChips (Affymetrix) were used for analysis of gene expression in 43 primary glioblastoma multiforme samples and 4 normal brain tissues. The experimental protocols and data processing were described in a previous report (16). A hierarchical clustering tree and heat map were generated with dChip. For both cancer and normal tissues, an average difference value was normalized against the average difference of β-actin to obtain the internally normalized expression value. The data were inputted into dChip, normalized against a normal tissue sample, and divided by the SD for each gene.

SEC61γ small interfering RNA (siRNA) sequences 5′-GCCAAGUCGGCAGUUUGUAAAGGAC-3′ and 5′-GUCCUUUACAAACUGCCGACUUGGC-3′ and a control scrambled GC siRNA were purchased from Invitrogen.

Cell viability assays were done as described previously (14). Briefly, assays were done consecutively from days 1 to 6 after siRNA transfection. A 10% solution of 5 mg/mL MTT diluted in PBS was added to the cells and incubated for 30 min, and 600 μL isopropanol was then added to the plate and incubated for 5 min. Next, 200 μL of the mixed solution were transferred to a 96-well plate, and absorbance was measured at 570 nm. All experiments were done in triplicate.

Twenty-four hours after siRNA transfection, H80 and HeLa were seeded at 500 cells in 10 cm plates and cultured for 14 days. The colonies were identified by crystal violet staining for 4 h and then counted. The experiments were done at least twice. Statistical analysis was done using Student's t test.

Rabbit anti-SEC61γ polyclonal antibody (pAb) was purchased from Proteintech Group; anti–glyceraldehyde-3-phosphate dehydrogenase monoclonal antibody was from Santa Cruz Biotechnology; anti-AKT pAb, anti–phospho-Ser⁴⁷³-AKT pAb, anti-extracellular signal-regulated kinase 1/2 pAb, anti–phospho-Thr²⁰²/Tyr²⁰⁴ extracellular signal-regulated kinase 1/2 pAb, anti-EGFR pAb, and anti–phospho-EGFR were from Cell Signaling Technology; the antibody for detection of the active caspase-3 was from Abcam; and anti-GRP78/Bip pAb was from Cell Signaling Technology.

To identify cancer-specific genetic changes in glioblastoma multiforme, we applied two genomic approaches, digital karyotyping and the Illumina HumanHap550 Genotyping BeadChip array, to interrogate the glioblastoma multiforme genome for gene copy-number variations in 59 human patient glioblastoma multiforme samples. Digital karyotyping allows precise, quantitative delineation of chromosomal amplifications at a resolution of 100 kb (13). Of the 16 cases of primary tumors and xenografts examined by digital karyotyping, 10 samples displayed dramatic amplification (8- to 103-fold) at chromosome 7p11.2 (Fig. 1A), which is the most frequently amplified region in the samples analyzed. However, none of the 10 glioblastoma multiforme cell lines contained chromosome 7p11.2 amplification. In all the samples that displayed chromosome 7p11.2 amplification, two genes, SEC61γ and EGFR, were encompassed in the minimal, overlapped regions of amplification (Fig. 1A). In an additional 33 primary tumors and xenografts, the Illumina HumanHap550 single nucleotide polymorphism data revealed chromosome 7p11.2 focal, high gain of copy number in 12 cases and gain of copy number in 14 cases. In agreement with the digital karyotyping results, the Illumina data also displayed minimal overlapped region containing only SEC61γ and EGFR (Supplementary Fig. S1).

EGFR amplification is a well-known genetic event occurring in 40% to 60% of primary glioblastoma multiforme (17, 18). In addition, amplification of the EGFR is often associated with structural alterations in the gene, with the most common being variant III (EGFR vIII), a 801-bp in-frame deletion of exons 2 to 7 of EGFR present in 20% to 50% of glioblastoma multiforme with EGFR amplification (19). Here, the high resolution of genomic approaches revealed both focal amplification and intragenetic deletions of EGFR (Fig. 1A; Supplementary Fig. S1).

It is of note that the genomic analyses revealed that SEC61γ, adjacent to EGFR, is always coamplified with EGFR in glioblastoma multiforme. To verify the results obtained from genome-wide studies, we performed Q-PCR analysis of EGFR and SEC61γ copy numbers on genomic DNA extracted from an additional 43 tumors of primary glioblastoma multiforme patients (age >50 years). Of the 43 samples, 20 samples (47%) displayed SEC61γ high copy-number gain (>4-fold) and 17 samples (40%) displayed coincident EGFR high copy-number gain (>4-fold; Fig. 1B). Interestingly, in no case was EGFR high copy-number gain present without a corresponding SEC61γ high copy-number gain in the Q-PCR analysis (Fig. 1B; Supplementary Table S1).

A previous population-based study indicated that the presence of EGFR amplification does not affect survival of glioblastoma multiforme patients at any age (20). We performed a survival analysis of the 43 glioblastoma multiforme cases and did not observe an association between SEC61γ amplification and a patient's survival (Supplementary Fig. S2), which indicates that chromosome 7p11.2 amplification alone may not have a prognostic value to predict glioblastoma multiforme patients' survival.

We also examined the expression levels of genes within the chromosome 7p11 amplicon using data derived from the human GeneChip array U133A (Affymetrix) of 4 normal adult brain tissues and the 43 primary glioblastoma multiforme samples that we analyzed by Q-PCR for amplification. Relative to its neighboring genes, SEC61γ displayed more prominent overexpression. Among the 43 glioblastoma multiforme, 33 (77%) tumors expressed SEC61γ at significantly higher levels than normal brain tissues, and genes in the vicinity of SEC61γ, that is, EGFR, LANCL2, and ECOP, displayed significant overexpression in 25 (58%), 10 (23%), and 27 (63%) of the 43 glioblastoma multiforme, respectively (Fig. 2A). Although in most cases the level of EGFR expression correlated with that of SEC61γ, in two glioblastoma multiforme samples, SEC61γ was expressed at a relatively higher level, whereas EGFR was expressed at the same level as the control sample (Fig. 2A, samples labeled with asterisk). Additionally, 7 samples in clade I (Fig. 2A) have overexpression of SEC61γ and EGFR but not LANCL2 and ECOP. Furthermore, SEC61γ is overexpressed in every sample with SEC61γ amplification (Supplementary Table S1). Consistent with the RNA data, immunoblotting analysis with anti-SEC61γ antibody detected high levels of SEC61γ expression in 6 of 8 (75%) randomly selected glioblastoma multiforme xenografts (Supplementary Fig. S3). Furthermore, we examined 3 secondary glioblastoma multiforme but did not detect SEC61γ overexpression (Supplementary Fig. S4).

We also compared the differential expression profiles of the three ER translocon subunit genes, SEC61α, SEC61β, and SEC61γ, in glioblastoma multiforme samples versus control samples. We found that although the expression of all three gene subunits tends to be elevated in glioblastoma multiforme samples, the expression of SEC61γ is often at greater levels than the α and β subunits (Supplementary Fig. S5).

To determine whether SEC61γ overexpression is associated with stage of astrocytoma progression, we analyzed the SEC61γ gene expression in different grades of astrocytomas using the Serial Analysis of Gene Expression data from the National Cancer Institute Cancer Genome Anatomy Project Web site.⁶

Based on our genetic studies, our hypothesis is that overexpression of SEC61γ is needed for glioblastoma multiforme tumor cell progression. To assess the requirement for SEC61γ in malignant cells, we investigated the effect of siRNA-mediated knockdown of SEC61γ on tumor cell growth. First, SEC61γ siRNA-transfected tumor cells were examined for viability by a MTT assay. In glioblastoma multiforme cell line H80, significantly reduced cell viability was observable 48 h post-transfection with siRNA against SEC61γ (Fig. 3A). The growth-inhibitory effects of SEC61γ knockdown were observed under complete medium conditions and more prominently under serum starvation. The siRNA also markedly repressed anchorage-independent tumor cell growth in soft agar (Fig. 3B). To determine if the SEC61γ knockdown-induced decrease in cell viability was due to increased apoptosis, we treated H80 cells with SEC61γ siRNA and assayed them for cleaved caspase-3, a protein that plays a key role in the execution of the late-stage apoptotic program. SEC61γ siRNA-mediated knockdown of SEC61γ led to significantly elevated caspase-3 activity (Fig. 3C). To verify the results, we repeated the above experiments using the highly proliferative cervical cancer cell line HeLa. We found that siRNA-mediated knockdown of SEC61γ also resulted in deceased cell growth and increased caspase-3 activity in HeLa cells (Fig. 3A-C).

Aggressive solid tumors like glioblastoma multiforme typically outgrow their blood supply, which leads to chronic hypoxia and nutrient deprivation. This physiologic stress usually results in disruption of ER homeostasis and leads to ER stress, a condition in which the protein-folding capacity of the ER is overwhelmed and misfolded proteins accumulate in the ER lumen. To survive in this condition, solid tumors cope with ER stress through the UPR, a cytoprotective signaling network that enables cells to process misfolded proteins and activate adaptive, antiapoptotic pathways. Given that the SEC61 complex forms the core of the ER protein translocation apparatus, we postulated that SEC61γ may play a role in the ER stress response in glioblastoma multiforme cells.

To assess the involvement of SEC61γ in the cellular response to ER stress, we exposed H80 and HeLa cells to the pharmacologic ER stress inducer tunicamycin and examined the expression of SEC61γ and GRP78, an ER chaperone protein and central regulator of ER homeostasis, the upregulation of which is widely used as a sentinel marker for ER stress under pathologic conditions (21). We observed upregulation of both SEC61γ and GRP78 proteins 3 h after tunicamycin treatment, with peak levels at 24 h (Fig. 4). To further evaluate expression of the SEC61 translocon subunits under ER stress, the expression of SEC61 α, β, and γ and of five UPR activation markers was examined by Q-PCR after tunicamycin treatment in H80 cells. EGFR and cyclin D were included as controls. Consistent with previous reports, cyclin D expression was downregulated in response to tunicamycin (22), whereas EGFR transcription was not upregulated in the cells. We found significant upregulation of the ER stress response genes GRP78, Xbp1, ATF4, ATF6, and CHOP by tunicamycin treatment (Supplementary Fig. S6). Following a short exposure to tunicamycin, the expression of CHOP mRNA was first observed to increase in H80 cultures and started to decrease 3 h post-treatment. At later intervals, ATF4 and Xbp1 were induced and reached their peak levels at 6 h post-treatment. Transcription levels of SEC61 α, β, and γ, ATF6, and GRP78 reached a peak at 12 h after tunicamycin treatment (Supplementary Fig. S6).

Chromosome 7p11.2 amplification has long been identified as the most prominent genetic lesion for the carcinogenesis of glioblastoma multiforme. Using high-resolution digital karyotyping and single nucleotide polymorphism arrays, we found that the minimal amplicon of chromosome 7p11.2 contains two genes, EGFR and SEC61γ. SEC61γ is not only always coamplified with EGFR in 47% of glioblastoma multiforme but also overexpressed in 77% of glioblastoma multiforme. In addition, we found that SEC61γ is required for tumor cell survival and for the cellular response to ER stress. Our findings suggest that SEC61γ exists as a glioblastoma multiforme–specific proto-oncogene, the product of which may facilitate tumor cells in coping with cellular stress in the tumor microenvironment to support glioblastoma multiforme cell proliferation.

Chromosome 7p11 is the most frequently amplified genomic region in glioblastoma multiforme. Among the genes within the chromosome 7p11 amplicon, our results indicate that, in addition to EGFR, SEC61γ is frequently amplified and overexpressed. Previous studies have shown that the size of a specific amplicon varies among tumors, ranging from a few hundred to a few thousand kilobases (23–28). In many tumor samples, an amplicon may harbor a cancer-specific gene and its adjacent bystander genes (29–32). However, it is also possible that more than one gene can exist as a tumor-specific gene in the same amplicon. For example, PDGFRA has been found to be coamplified with KIT, along with the vascular endothelial growth factor receptor gene KDR, at 4q12, and DDX1 and N-Myc are coamplified at 2p24 (33). The coamplification of multiple genes within a genomic region may have synergistic effects on neoplastic pathogenesis. Within the chromosome 7p11.2 amplicon, EGFR is amplified and overexpressed in 46% of glioblastoma multiforme (34), and EGFR truncated mutations have been identified to be oncogenic in glioblastoma multiforme (34, 35). Given that receptor tyrosine kinases and the signaling pathways they control constitute potential therapeutic targets, EGFR has been regarded as a major focus of research in glioblastoma multiforme. Two genes adjacent to EGFR, LANCL2 (36) and ECOP (37–39), have been reported to coamplify with EGFR in 50% and 33% of glioblastoma multiforme, respectively. However, the role of SEC61γ in glioblastoma multiforme pathogenesis has not been fully characterized.

Due to inadequate perfusion, the microenvironment of highly proliferative solid tumors is characterized by chronic hypoxia and a lack of nutrients. These conditions induce ER stress and activate both cytoprotective and cytodestructive branches of the UPR (40). Recent work has linked various types of cancer to genetic alterations in ER-resident chaperones (1–3). The ER translocon proteins SEC62 and SEC63 have been identified as the most frequently mutated and overexpressed genes in prostate, gastric, and colorectal cancers (12). An exacerbated proliferation is a hallmark of glioblastoma multiforme cells as further indicated by the significant overexpression of the UPR target genes, including GRP78, in glioblastoma multiforme (8).

Recently, SEC61β has been implicated in EGFR trafficking and EGFR-mediated activation of the phosphoinositide 3-kinase/AKT pathway (41–43), which is known for promoting cell survival and inhibiting apoptosis in most cell types (44, 45), including glioblastoma multiforme (46, 47). The phosphoinositide 3-kinase/AKT pathway may counteract ER stress–induced apoptotic signaling (48, 49). To determine if the growth-inhibitory effects of SEC61γ knockdown might be mediated through the EGFR/AKT signaling pathway, we evaluated the consequence of SEC61γ knockdown on the EGFR/AKT pathway activity and found that EGF-mediated activation of EGFR and AKT was significantly inhibited in SEC61γ siRNA-transfected cells (Supplementary Fig. S7). These results suggest that inhibition of SEC61γ could be an alternative way to inhibit various arms of the AKT signaling network and reverse the fate of tumor cells from survival to cell death. However, the molecular mechanism underlying the observation needs to be illuminated by further investigation of the role of SEC61γ in the EGFR-AKT signaling pathway.

In the present study, we show that the expression of SEC61γ is positively correlated with astrocytoma grade and the primary glioblastoma multiforme subtype. Moreover, we show that SEC61γ is upregulated in glioblastoma multiforme cell lines in response to the pharmacologic stress agent tunicamycin. The SEC61γ amplification and increased expression in glioblastoma multiforme in proliferative situations probably reflects increased synthesis and translocation of proteins to the lumen of the ER and increased activity of the quality-control process of proteins destined for membranes and for secretion. Knocking down SEC61γ expression resulted in apoptosis and abrogation of EGFR/AKT survival signaling. These results suggest that SEC61γ confers a selective growth advantage under physiologic conditions by facilitating a cytoprotective response to ER stress.

Further studies are needed to investigate the molecular mechanisms through which ER stress stimulates the expression of SEC61γ and mediates its prosurvival effects. A detailed understanding of the signaling networks underlying SEC61γ involvement in ER stress response, along with those underlying its high overexpression in malignant cells and near absence in normal cells, will make SEC61γ as an attractive viable therapeutic target for pharmaceutical intervention.

No potential conflicts of interest were disclosed.

Grant support: The Pediatric Brain Tumor Foundation Institute at Duke and NIH grants: National Cancer Institute grant R01CA118822, National Institute of Neurological Disorders and Stroke grant 5P50 NS20023, National Cancer Institute Specialized Program of Research Excellence grant 5P50 CA108786, and National Cancer Institute merit award R37 CA 011898.

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.