Poor chemosensitivity and the development of chemoresistance remain major obstacles to successful chemotherapy of malignant gliomas. GRP78 is a key regulator of the unfolded protein response (UPR). As a Ca2+-binding molecular chaperone in the endoplasmic reticulum (ER), GRP78 maintains ER homeostasis, suppresses stress-induced apoptosis, and controls UPR signaling. We report here that GRP78 is expressed at low levels in normal adult brain, but is significantly elevated in malignant glioma specimens and human malignant glioma cell lines, correlating with their rate of proliferation. Down-regulation of GRP78 by small interfering RNA leads to a slowdown in glioma cell growth. Our studies further reveal that temozolomide, the chemotherapeutic agent of choice for treatment of malignant gliomas, leads to induction of CHOP, a major proapoptotic arm of the UPR. Knockdown of GRP78 in glioblastoma cell lines induces CHOP and activates caspase-7 in temozolomide-treated cells. Colony survival assays further establish that knockdown of GRP78 lowers resistance of glioma cells to temozolomide, and, conversely, overexpression of GRP78 confers higher resistance. Knockdown of GRP78 also sensitizes glioma cells to 5-fluorouracil and CPT-11. Treatment of glioma cells with (−)-epigallocatechin gallate, which targets the ATP-binding domain of GRP78 and blocks its protective function, sensitizes glioma cells to temozolomide. These results identify a novel chemoresistance mechanism in malignant gliomas and show that combination of drugs capable of suppressing GRP78 with conventional agents such as temozolomide might represent a novel approach to eliminate residual tumor cells after surgery and increase the effectiveness of malignant glioma chemotherapy. [Cancer Res 2007;67(20):9809–16]
- malignant glioma
Malignant gliomas are the most commonly diagnosed malignant adult primary brain tumors, including anaplastic astrocytoma (grade 3) and glioblastoma multiforme (GBM; grade 4). Median survival for GBM is ∼12 to 15 months. Surgical resection or diagnostic biopsy is usually the first step in therapy, followed by adjuvant radiation and chemotherapy ( 1). Despite significant improvements in the surgical treatment of malignant gliomas, a complete resection is impossible; after diffuse brain invasion, cells were often found at great distances from the site of the major tumor ( 2).
Because a cure remains elusive, it is important to identify new treatment modalities, as well as to modify existing therapies, to possibly change malignant gliomas from a deadly disease into a chronic one. A promising approach to achieving these goals is to enhance glioma chemosensitivity to existing treatment regimens. Temozolomide (Temodar) is currently the standard of care for treatment of newly diagnosed and recurrent malignant gliomas. This well-tolerated drug has been shown to delay tumor progression and prolong patient survival ( 3). The action of temozolomide is associated with its ability to damage DNA through DNA methylation and subsequent substitution of cytosine by thymine. This leads to the activation of the mismatch repair mechanism, which recognizes a recurring error and triggers apoptosis ( 4). In patient samples, resistance to temozolomide in GBMs has been linked to expression of O6-methylguanine DNA methyltransferase ( 5).
The unfolded protein response (UPR) consists of a set of adaptive pathways that are triggered by disparate perturbations in normal function of the endoplasmic reticulum (ER) that lead to the production of misfolded proteins. The UPR alleviates ER stress by arresting general translation, up-regulation of chaperones and folding enzymes, and degradation of misfolded proteins; however, in the case of intense or persistent ER stress, the UPR will trigger apoptosis resulting in cell death ( 6). A major UPR target is the induction of the glucose-regulated protein 78 (GRP78/BiP), which has important roles in protein folding and assembly, targeting misfolded proteins for degradation, ER Ca2+ binding, and controlling the activation of transmembrane ER stress sensors ( 7). Thus, GRP78 represents a prosurvival arm of the UPR. On the other hand, the CCAAT/enhancer binding protein homologous transcription factor (CHOP/GADD153) is one of the critical executioners of the proapoptotic arm of the UPR ( 8). Induction of CHOP signals that the cells are experiencing ER stress and hence initiating the cell death process.
As malignant gliomas represent tumors that are highly resistant to chemotherapy and elimination of residual tumor cells remains a major challenge, we initiated this study to determine whether GRP78 plays a role in the chemoresistance of such cancers. Here, we provide evidence that malignant glioma cells overexpress GRP78. Our results reveal that knockdown of GRP78 suppresses glioma cell proliferation, induces CHOP, and activates caspase-7 in temozolomide-treated cells. Through colony survival assays, we have established that manipulation of GRP78 levels alters the sensitivity of glioma cells to temozolomide and other chemotherapeutic agents. Further, combination treatment of glioma cells with temozolomide and (−)-epigallocatechin gallate (EGCG), a known inhibitor of the protective function of GRP78 ( 9), sensitizes glioma cells to temozolomide. These results show that GRP78 status is a potential prognostic marker for malignant gliomas and that GRP78 is a novel chemosensitizing target for malignant brain tumors.
Materials and Methods
Cell lines, culture, and drug treatment conditions. U87 cell lines were obtained from the American Tissue Culture Collection. All other malignant glioma cell lines were provided by Frank B. Furnari and Webster K. Cavenee (Ludwig Institute of Cancer Research, La Jolla, CA). The cells were propagated in DMEM supplemented with 10% fetal bovine serum, 100 units/mL penicillin, 0.1 mg/mL streptomycin at 37°C, and 5% CO2. Human astrocytes and the astrocyte medium, in which astrocytes were grown, were purchased from ScienCell Research Laboratories. Temozolomide (Temodar) was purchased from Schering Plough and dissolved in DMSO as a 25 mmol/L stock solution. 5-Fluorouracil (5-FU) and irinotecan (CPT-11) were obtained from Sigma-Aldrich.
Immunoblots and antibodies. Fifty micrograms of total cell lysate prepared in radioimmunoprecipitation assay buffer were processed for Western blot analysis as described ( 10). The antibodies against GRP78, CHOP, actin (Santa Cruz Biotechnology, Inc.), and caspase-7 (BD PharMingen) were used according to the manufacturer's recommendations. The secondary antibodies were coupled to horseradish peroxidase, and were detected by chemiluminescence using the SuperSignal West substrate (Pierce). Each immunoblot was done at least twice to confirm the results. The blots were quantified using ImageJ software ( 11).
Immunohistochemistry. Malignant glioma tissues were obtained from glioblastoma patients undergoing primary resection or biopsy of their tumor. No exposure to radiation or chemotherapy had been experienced by this group of patients. The peritumoral tissue was obtained from patients undergoing a partial or full lobectomy for their GBM. Consent for use of tissue specimens was obtained from the University of Southern California Institutional Review Board. Immunohistochemical analysis of GRP78 expression in tumor tissues and cell lines was done with anti-GRP78 antibody (Santa Cruz Biotechnology) diluted 1:100 in 2% goat blocking serum as described ( 12). Specificity of the antibody was tested as previously described ( 12).
Transfections. For knockdown of GRP78, the glioblastoma cells were transfected with small interfering RNA (siRNA) against GRP78: sense 5′-GGAGCGCAUUGAUACUAGATT-3′ and antisense 5′-UCUAGUAUCAAUGCGCUCCTT-3′ ( 13). The siRNA targeted against green fluorescent protein (GFP) purchased from Qiagen served as negative control. A final concentration of 80 nmol/L was used for both siRNAs. Transfections were done in six-well plates following Qiagen's instructions. Ten microliters of each siRNA were used with 5 μL of Lipofectamine 2000 per well.
Lentivirus infection. Lentiviruses expressing siRNA against GRP78 as described above or full-length human GRP78 were constructed as described ( 14). All lentiviral constructs expressed GFP. For infection, 104 cells were plated in six-well plates and infected with the lentivirus at titers of 5 × 106 units/mL. Infected cells were monitored for GFP under the fluorescence microscope.
Colony formation assays. Twenty-four hours after transfection with siRNA, the cells were seeded in six-well plates at 200 cells per well (when higher concentrations of temozolomide were used, appropriately increased numbers of cells were seeded). The next day, the medium was changed to include 20 to 160 μmol/L temozolomide. After a change of fresh medium 48 h later, the cells were allowed to form colonies for 14 days in the absence of drug. The colonies were stained with 1% methylene blue (in methanol) for 4 h and then counted. The experiments were done in three or six replicates and each experiment was done at least twice. Statistical analysis was done using Student's t test, two-tailed distribution, assuming equal variance for the samples.
Elevation of GRP78 levels in human glioma tissue specimens. Tumor specimens from patients undergoing craniotomy before adjunctive therapy, including radiation and chemotherapy, were examined using both Western blot analysis and immunohistochemistry to determine whether GRP78 is up-regulated in human glioma tissue. The Western blot results show that GRP78 level was generally elevated with varying degrees in the glioma tissues ( Fig. 1A ). For example, compared with normal brain tissue where GRP78 was below detection limit, specimens 4 and 6 showed high GRP78 level, specimens 2 and 3 showed intermediate level, specimen 1 was at low level, and specimen 5 was at very low level. GRP78 overexpression was confirmed by immunohistochemical staining of tissue sections from human glioma specimens that were different from those used in the Western blots. All four sections were positive for GRP78 staining ( Fig. 1B). In agreement with the immunoblot result, peritumoral brain tissue showed very little GRP78 staining ( Fig. 1B). These results reveal that GRP78 is highly elevated in this sample of human malignant gliomas.
GRP78 expression level correlates with proliferation rates and is required for glioma cell growth. A panel of malignant glioma cell lines was tested for GRP78 expression in Western blot analysis to address whether human glioma cells constitutively express this protein and at what level. The results show that compared with human astrocytes grown under normal culture conditions, five of the six glioma cell lines constitutively expressed higher level of GRP78, with the apparent exception of the U138 cell line ( Fig. 2A ). The level of GRP78 overexpression ranged from 2- to 3-fold, with the upper range comparable with the level of GRP78 induction observed in human astrocytes treated with thapsigargin, an inhibitor of the ER Ca2+-ATPase, and an established ER stress inducer. Interestingly, the level of GRP78 overexpression in glioma cell lines seems to correlate with their rate of proliferation. For example, highly replicating lines (U251, LN229, T98G, U87, and A172) with relatively shorter doubling times (19, 22, 22, 24, and 37.5 h, respectively) showed higher level of GRP78 (3-, 2-, 2-, 2-, and 1.5-fold increase) compared with the slower replicating U138 cells, which have a doubling time of >80 h, for which the GRP78 level was similar to the baseline control. Doubling times of U251, T98G, A172, and U87 cells had previously been reported and were confirmed in our laboratory ( 15– 17). Doubling times of U138 and LN229 cells were obtained in our laboratory.
To confirm the immunoblot results, immunohistochemical staining was done. Human astrocytes grown under normal conditions and treated with thapsigargin were used as negative and positive controls, respectively ( Fig. 2B). As shown in Fig. 2C, the majority (>90%) of U251 cells stained positively for GRP78; the majority of T98G, LN229, and U-87 cells (>70%) also stained positively for GRP78. A172 cells showed intermediate staining, with ∼25% of the cells stained. By contrast, U138 cell staining was comparable with the negative control with only an occasional positive cell, consistent with the immunoblot results. The cellular staining pattern of GRP78 was perinuclear, within the cytosol, demonstrating a staining pattern consistent with ER localization.
GRP78 was down-regulated using specific GRP78 siRNA (siGRP78) to determine whether it was required for glioma cell growth. Two glioma cell lines (U87 and U251) were examined ( Fig. 2D). Twenty-four hours posttransfection, equal numbers of cells transfected with siGRP78 or an siRNA directed against GFP (siCtrl) were plated in six-well plates. Subsequently, one-well per group was trypsinized and counted each day for 3 days and the numbers of cells were plotted against time. Down-regulation of GRP78 resulted in a slower growth rate for both cell lines, with the effect being more severe in U251 than in U87 cells, correlating with near-complete versus partial knockdown of GRP78 in U251 and U87 cells, respectively ( Fig. 2D). Taken together, these data show that the majority of malignant glioma cell lines express elevated levels of GRP78 protein, with the fastest proliferating cells exhibiting the highest level of overexpression. In addition, down-regulation of GRP78 protein levels results in a significant slowing down of the glioma cell growth.
Temozolomide treatment of glioma cells induces CHOP/GADD153 and GRP78. Temozolomide is widely used as an antiglioma drug. The effect of temozolomide on GRP78 (i.e., the protective arm of the UPR) as well as on CHOP (i.e., the proapoptotic component of the UPR) were examined to determine whether this drug would impinge on ER stress or UPR levels. First, U87 cells were treated with a range (30–500 μmol/L) of temozolomide concentrations for 72 h, and then subjected to Western blot analysis. As shown in Fig. 3A , CHOP induction was dramatically stimulated by temozolomide in a dosage-dependent manner, whereas a moderate induction of GRP78 was detected in parallel. We then tested whether this observation is applicable to other glioma cells and to determine the kinetics of CHOP and GRP78 induction. U251 cells were incubated with 250 μmol/L of temozolomide from 4 to 72 h. The results showed that temozolomide could induce GRP78 and CHOP in the U251 cells; the induction became evident after 36 h, and intensified and persisted at least through 72 h of treatment ( Fig. 3B). Thus, the induction of the representative UPR markers GRP78 and CHOP indicates that temozolomide is a novel inducer of the ER stress response.
Interestingly, the suppression of GRP78 expression with the use of siRNA resulted in a substantial increase in CHOP levels in U251 cells, even in the absence of any drug treatment ( Fig. 3C). This result was consistently obtained in U87 and LN229 cells as well (data not shown). Therefore, GRP78 suppresses CHOP induction. GRP78 is known to interact with procaspase-7, suppressing its activation ( 9, 18, 19). U251 cells infected with lentivirus expressing siRNA against GRP78 (L-siGRP78) or the GFP alone (L-GFP) were treated with temozolomide. Whole cellular lysates were analyzed by Western blot with antibodies to GRP78 and caspase-7. The cells that detached and were floating in the medium were included in this analysis. We observed significant caspase-7 cleavage in temozolomide-treated U251 cells in which GRP78 level was down-regulated by siRNA ( Fig. 3D). These results imply that elevated level of GRP78 in gliomas may contribute to increased resistance to cell death of these tumor cells.
Knockdown of GRP78 or treatment with EGCG sensitizes gliomas to temozolomide. GRP78 knockdown experiments were done to test directly whether gliomas depend on GRP78 for protection against temozolomide-mediated cell death. U251, U87, and LN229 glioma cells were transfected with siRNA targeted against human GRP78 (siGRP78) or control siRNA (siCtrl) targeted at GFP. Colony survival assays were done using a range of temozolomide concentrations to evaluate the drug sensitivity of the transfected cells ( Fig. 4A–C ). Western blot assays verified that siRNA against GRP78 was highly effective in suppressing GRP78 expression in all three cell lines ( Fig. 4A–C). The results showed that suppression of GRP78 expression caused a statistically significant increase in drug sensitivity in all three glioma cell lines tested, indicating a critical role of GRP78 as a protector against the cytotoxic effects of temozolomide. To determine whether GRP78 protection extends to drugs other than temozolomide, the same cells were treated with 5-FU or CPT-11. Down-regulation of GRP78 resulted in increased sensitivity of LN229 cell to both drugs ( Fig. 4C). Treatment of cells with these drugs did not result in induction of ER stress markers (data not shown).
As a direct test that GRP78 overexpression protects glioma cells against temozolomide-induced cell death, U251 cells were infected with a lentivirus expressing human GRP78 (L-GRP78) or a control vector expressing GFP alone (L-GFP). Colony survival assays were done using a range of temozolomide concentrations to evaluate the drug sensitivity of the infected cells ( Fig. 5 ). Western blot assay verified that infection with L-GRP78 increased the level of GRP78 protein by ∼2-fold. The results showed that elevation of GRP78 level caused a statistically significant protection from temozolomide-induced cell death in U251 cells.
Our results predict that down-regulation of GRP78 by compounds directed against GRP78 expression or activity could lead to increased chemosensitivity of glioma cells to temozolomide. One such compound is EGCG, which directly interacts with the ATP-binding domain of intracellular GRP78, blocking its interaction with procaspase-7 and suppressing the protective function of GRP78 ( 9). In colony survival assays, concomitant treatment of U251 cells with temozolomide and EGCG led to increases in temozolomide chemosensitivity ( Fig. 6A and B ). Importantly, EGCG at the effective concentrations did not cause cell death in U251 cells by itself. The combination of EGCG and temozolomide, however, caused significantly more cell death than temozolomide alone. These results imply that combination of drugs capable of GRP78 suppression could sensitize glioma cells to temozolomide.
Gliomas, and GBM in particular, are among the most chemoresistant tumors ( 20). In this study, we have made several new observations, which may have important prognostic and therapeutic applications. In our survey of several malignant glioma cell lines, GRP78 overexpression was generally observed and this overexpression was independent of their p53 or PTEN status. Our results showed that glioma cell lines with rapid proliferation rates had the highest levels of GRP78 expression. In addition, down-regulation of GRP78 led to a significant decrease in glioma cell growth. Varying levels of elevated GRP78 expression were also detected in patient biopsies before adjuvant treatment. Taken together, our results raise the possibility that the level of GRP78 in patient biopsies could be used as a predictive factor of aggressiveness of the glioblastoma, independent of their p53 and PTEN status. Our findings predict that tumors with low GRP78 expression represent slow growing, less aggressive tumors, whereas those with high GRP78 are rapidly proliferating, aggressive tumors. Furthermore, elevated GRP78 expression in highly proliferating tumors may provide malignant gliomas with a possible protective mechanism against cytotoxic drugs such as temozolomide. Cytotoxic drugs that induce DNA mismatch work best in rapidly proliferating cells. Rapidly proliferating malignant glioma cells are expected to be more chemosensitive to temozolomide but they may also be protected from temozolomide by higher GRP78 levels.
An important question arises as to how GRP78, an ER molecular chaperone, protects gliomas against temozolomide, a DNA-damaging agent. Our studies suggest that several novel mechanisms may be involved. Evidence is emerging that some therapeutic agents recognized as working through DNA damage can also elicit cytosolic stress independent of the nucleus. For example, cisplatin can induce apoptosis in the absence of DNA damage, at least in part through activation of ER stress as evidenced by GRP78 induction and activation of the ER-associated caspase-12 ( 21). Our discovery that temozolomide is an inducer of both GRP78 and CHOP suggests that it also causes ER stress. The ER could be a direct target site of temozolomide or temozolomide may indirectly elicit ER stress acting through other cellular pathways. Recently, Das et al. ( 22) reported that temozolomide treatment at 100 μmol/L for 6 h leads to increased intracellular Ca2+ levels and higher expression of proapoptotic markers. Elevated cytoplasmic Ca2+ levels have been reported to activate the ER stress response, as exemplified by ATF6 cleavage and GRP78 induction ( 23). GRP78 is known to function as a Ca2+-binding protein in the ER, thereby preventing Ca2+ efflux into the cytosol. In such capacity, GRP78 may provide a buffer against temozolomide-induced increases in cytosolic Ca2+ levels, thereby suppressing apoptosis. In addition, ER stress causes protein misfolding in the ER. GRP78, in its capacity as a major ER molecular chaperone, offers protection against ER stress and suppresses apoptosis.
The mechanism of cell death and resistance to temozolomide is just emerging and is likely to be complex, involving apoptotic as well as nonapoptotic pathways ( 5, 22, 24). In the glioma cell lines, temozolomide induces CHOP, which as a transcription factor, sensitizes cells to ER stress-induced apoptosis through down-regulation of BCL-2 and activation of GADD34 and ER01α, an ER oxidase ( 25, 26), and this may represent a new antitumoral mechanism of temozolomide. Our studies reveal that CHOP is constitutively suppressed by GRP78 in glioblastoma. Further, temozolomide treatment leads to activation of caspase-7 when GRP78 level is down-regulated in glioma cells. In view that GRP78 also protects glioma cells against drugs such as 5-FU and CPT-11, which do not induce ER stress, pathways independent of ER stress are likely to be affected by GRP78. These may include its role in cell proliferation and other yet unknown mechanisms ( 27– 30).
Because GRP78 provides protection to tumor cells and endows them with increased chemoresistance, down-regulation of GRP78, therefore, may become an important adjunct in future treatments of malignant gliomas. Our results with siGRP78 show that lowering GRP78 levels significantly increases chemosensitivity to temozolomide. Recent developments of small molecules that can specifically block GRP78 expression and/or its activity may make this approach clinically feasible in the near future ( 9, 30– 34). Recently, Ermakova et al. ( 9) showed that EGCG, a major component of green tea, can bind to intracellular GRP78, inhibit its protective function, and increase chemosensitivity to etoposide in breast and bladder carcinoma cell lines. In our study, we showed that treatment with EGCG had similar effects as down-regulation of GRP78 by siRNA such that cells treated with the combination of EGCG and temozolomide exhibited more cell death than cells treated with temozolomide alone. EGCG alone did not cause any cell death in glioma cells. EGCG is a water-soluble molecule that may potentially be formulated for i.v. infusions. As a small molecule, it should cross the blood-brain barrier readily. Further in vivo studies, including toxicity studies on systemic EGCG administration, are pending. Thus, in our study, we present a target for increasing glioma cell sensitivity to temozolomide (GRP78), and a clinically feasible way to affect this target (treatment with EGCG) in gliomas. Other agents, such as ABT 828, recombinant human plasminogen Kringle 5, has been recently used in phase I/II studies in various carcinomas. Preclinical studies showed that GRP78 is the receptor for Kringle 5, which induces apoptosis of fibrosarcoma and growth-stimulated endothelial cells ( 18). Its mechanism of action has been hypothesized to be secondary to binding of surface GRP78, internalization of ABT 828 into the cytosol, and blocking of GRP78 binding to the proapoptotic molecule procaspase-7. Combining these and other drugs that can suppress GRP78 with conventional agents, such as temozolomide, may present a novel approach to the elimination of residual tumor cells after surgery, and thereby increase the effectiveness of malignant glioma chemotherapy.
Grant support: NIH grants CA027607 and CA111700 (Amy S. Lee), Accelerate Brain Cancer Cure (Axel H. Schönthal and Thomas C. Chen), and American Medical Association Seed Grant (Peter Pyrko).
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.
We thank Frank B. Furnari and Webster K. Cavenee (Ludwig Institute for Cancer Research) for the various glioblastoma cell lines; Suzan Su for help with some immunohistochemical staining; members of the Lee laboratory, Dezheng Dong and Min Ni, and Caryn Stiles and John B. Patterson (MannKind Corp.), for lentiviral construction and infection; and Peter Baumeister and Yong Fu for helpful discussions.
- Received February 14, 2007.
- Revision received July 31, 2007.
- Accepted August 14, 2007.
- ©2007 American Association for Cancer Research.