The MEK inhibitor PD98059 was purchased from Cell Signaling Technology (Beverly, MA, USA). The PI3K inhibitor LY294002 was purchased from Sigma-Aldrich (St. Louis, MO, USA). PD98059 and LY294002 were dissolved in dimethyl sulfoxide (DMSO) and stored at −20°C. In all experiments, cells were treated with or without PD98059 (50 μM) and LY292004 (10 μM). Untreated cells were treated identically as the treated cells, except for PD98059 and LY292004.

The SHG44, U87 and U251 human glioblastoma cell lines were obtained from the American Type Culture Collection (ATCC, Manassas, VA, USA) and cultured in Dulbecco’s modified Eagle’s medium (DMEM) (HyClone, Rockford, IL, USA) containing 10% fetal bovine serum (Sijiqing, Hangzhou, China) and 1% penicillin/streptomycin (HyClone) in a humidified incubator with 5% CO₂ at 37°C.

The method for construction of the recombinant plasmids and transfection efficiency has been described in our previous study (30). Human Nrf2 cDNA was cloned into the pEGFP-N1 vector (pEGFP-Nrf2), and an empty vector was used as the negative control (pEGFP). Nrf2 shRNA was cloned into the pGPH1/GFP/Neo vector (sh-Nrf2), and a random sequence was used as the negative control (sh-control). Cells (10⁶/well) were seeded into 6-well plates and allowed to attach for 24 h. The cells were then transfected with the various plasmids (pEGFP, pEGFP-Nrf2, sh-control or sh-Nrf2) using Lipofectamine 2000 (Invitrogen, Carlsbad, CA, USA) according to the manufacturer’s protocol. Transfected cells were cultured for at least 24 h after transfection. Transfection efficiency, monitored by GFP fluorescence, was ~80%.

Sixteen patients with primary glioblastoma who underwent maximal safe surgical resection at Jinling Hospital (Nanjing, China) were selected for the present study. All the patients received no chemotherapy or radiotherapy before surgery. The study was approved by the local ethics committee. Informed consent was obtained for the use of brain tissue. Sixteen tumor tissues were collected from the main bulk of the tumor tissues. Eleven peritumoral normal tissues were collected from the border zones between the tumors and the normal brain tissues. All specimens were fixed in 10% neutral formalin and were then embedded in paraffin. All cases were reviewed independently by two pathologists. The diagnosis of glioblastoma was confirmed according to the current World Health Organization (WHO) classification of central nervous system tumors.

For immunohistochemistry, sections (4-μm) from 10% buffered formalin-fixed, paraffin-embedded tumor tissues were cut and placed onto slides. All the sections were deparaffinized in xylene and rehydrated through a series of graded alcohol. Endogenous peroxidase activity was blocked with 3% hydrogen peroxidase and methanol. The sections were autoclaved in 10 mmol/l citrate buffer (pH 6.0) at 120°C for 2 min for antigen retrieval, and then cooled at room temperature. The sections were blocked with 10% normal calf serum in PBS for 30 min, and incubated with the primary antibody against Nrf2 (1:50; Abcam, Cambridge, MA, USA) overnight at 4°C. The sections were then incubated with the secondary antibody (Dako, UK) for 30 min at room temperature. The peroxidase reaction was developed using 3,3′-diaminobenzidine as chromogen. Nuclei were counterstained with hematoxylin. Serial sections of the samples were also subjected to hematoxylin and eosin staining. The specificity of the immunohistochemical reaction was evaluated by replacement of the primary antibody with non-specific rabbit IgG. Serial sections of the individual samples were also subjected to hematoxylin and eosin staining.

The scoring of nuclear Nrf2 expression was carried out semi-quantitatively by assessing the staining intensity and the percentage of nuclear-positive cells. The staining intensity was scored as: 0 for no staining, 1 for weak staining, 2 for moderate staining and 3 for strong staining. The percentage of nuclear-positive cells was scored as: 0 when 0–1% of the tumor cells were positive, 1 when 2–10% of the tumor cells were positive, 2 when 11–50% of the tumor cells were positive and 3 when 51–100% of the tumor cells were positive. The percentage and intensity scores were added to obtain the total score (range, 0–6). The Nrf2 staining was divided into four levels: negative (score 0), weak (score 1–2), moderate (score 3–4) and strong (score 5–6). The nuclear-positive cells were identified, counted and analyzed by two pathologists blinded to the experimental design.

To prepare the total proteins, cells were lysed in cold RIPA lysis buffer (Beyotime, Jiangsu, China) containing 1 mM phenylmethylsulphonyl fluoride and centrifuged at 13,000 × g for 5 min at 4°C to remove debris. To prepare the cytoplasmic and nuclear proteins, cells were lysed using a nuclear and cytoplasmic protein extraction kit (Beyotime) according to the manufacturer’s instructions. Protein concentrations were determined using the BCA assay, and the protein extracts were heat denatured in Laemmli sample loading buffer. Equal amounts (50 μg/lane) of protein were separated by 10% SDS-polyacrylamide gel electrophoresis, and electroblotted onto polyvinylidene-difluoride membranes. The membranes were blocked for 2 h at room temperature with 5% nonfat milk in Tris-buffered saline containing 0.1% Tween-20 (TBST), and incubated overnight at 4°C with the primary antibodies diluted in blocking buffer: anti-Nrf2 (1:500; Abcam), anti-β-actin (1:1,000; Cell Signaling Technology) and anti-Lamin B1 (1:1,000; Abcam). The membranes were washed with TBST (3 × 10 min), and incubated with goat anti-rabbit horseradish peroxidase (HRP)-conjugated IgG (1:1,000; Cell Signaling Technology) for 2 h at room temperature. The blotted protein bands were developed using an ECL detection reagent (Amersham Biosciences, Buckinghamshire, UK). Lamin B1 was used as a loading control for nuclear extracts. β-actin was used as a loading control for cytoplasmic and whole cell extracts.

Cells growing on glass coverslips were fixed in 4% paraformaldehyde, permeabilized in 0.1% Triton X-100 and then blocked with Immunol Staining Blocking Buffer (Beyotime). The coverslips were incubated with the primary antibody against Nrf2 (1:100; Abcam) overnight at 4°C. Following washes with PBS, the coverslips were incubated with an FITC-labeled secondary antibody (Beijing Zhongshan Jinqiao Biotechnology Co., Ltd., Beijing, China) for 1 h at 37°C. Nuclei were then counterstained with DAPI. The specificity of the immunofluorescence reaction was evaluated by replacement of the primary antibody with non-specific rabbit IgG.

Total RNA was isolated from cells with RNAiso Plus (Takara, Dalian, China). The concentration of total RNA was determined by a spectrophotometer (OD260/280 = 1.8–2.2). Single-stranded cDNA was synthesized from 2 μg of total RNA using the PrimeScript™ RT Reagent kit (Takara). Quantitative real-time PCR analysis was performed using the iQ5 Real-Time PCR Detection System (Bio-Rad Laboratories, Hercules, CA, USA), applying real-time SYBR-Green PCR technology. The reaction mixtures contained 10 μl of Real-Time PCR Master Mix (Toyobo, Osaka, Japan), 1 μl of cDNA, 1 μl of each forward and reverse primer (10 μM) and 7 μl of DEPC-treated water. The PCR thermal cycling was carried out with a 5-min initial denaturation step at 95°C, followed by 40 three-step cycles: a denaturation step (95°C for 15 sec), an annealing step (60°C for 30 sec) and an extension step (72°C for 30 sec). The experimental results were analyzed by the 2^−ΔΔCt method.

Primer sequences used in this study were as follows: HO-1 forward and reverse primers were 5′-TCTCCGATGGGTCCT TACACTC-3′ and 5′-GGCATAAAGCCCTACAGCAACT-3′. NQO1 forward and reverse primers were 5′-ATGGTCGG CAGAAGAGC-3′ and 5′-GGAAATGATGGGATTGAAGT-3′. β-actin forward and reverse primers were 5′-GCAGAAGGA GATCACTGCCCT-3′ and 5′-GCTGATCCACATCTGCTG GAA-3′.

In brief, 24 h after transfection, the transfected and untransfected cells (4,000/well) were treated with or without PD98059 (50 μM) plus LY294002 (10 μM) for another 48 h. Afterward, MTT (5 mg/ml, 20 μl) was added to each well and the plates were incubated for 4 h. The blue formazan product was then dissolved with DMSO, and the optical density (OD) value was measured at 490 nm using an enzyme-labeling instrument.

The wells containing only cells and complete medium were used as the negative control. The wells containing only complete medium were used as the blank control.

All experiments were repeated at least three times, and the results were analyzed by SPSS 16.0 (SPSS Inc., Chicago, IL, USA). Data are presented as means ± SEM, and evaluated by the Student’s t-test or one-way ANOVA followed by LSD test. P-value <0.05 was considered to indicate a statistically significant result.

To evaluate the expression and distribution of Nrf2, we performed immunohistochemistry on tumor and peritumoral normal tissue sections. As shown in Fig. 1A, the expression and nuclear localization of Nrf2 in the main bulk of tumor tissues were markedly increased when compared with the expression and the nuclear localization in the peritumoral normal tissues which were located at a distance from the tumor tissues. Fig. 1B shows Nrf2 staining in a sample of tumor and peritumoral normal tissues. Peritumoral normal tissues exhibited weak Nrf2 staining which was predominant in the cytoplasm. In the adjacent tumor tissues, strong Nrf2 staining was found in most tumor cells and was located in both the cytoplasm and nucleus. Semi-quantitative analysis of the unclear Nrf2 protein is shown in Fig. 1C. In the peritumoral normal tissues, 0 (0/11), 72.72 (8/11), 27.28 (3/11) and 0% (0/11) of samples showed negative, weak, moderate and strong staining, respectively. In the tumor tissues, 0 (0/16), 18.75 (3/16), 56.25 (9/16) and 25% (4/16) of samples showed negative, weak, moderate and strong staining, respectively. These results suggest that expression and nuclear localization of Nrf2 are increased in human glioblastoma tissues.

It is well known that nuclear accumulation of Nrf2 proteins indicates activation of the Nrf2-ARE pathway. In the present study, we aimed to ascertain whether inhibition of ERK and PI3K decreases nuclear distribution of Nrf2. To this end, SHG44, U87 and U251 cell lines were treated with or without the MEK inhibitor PD08059 plus PI3K inhibitor LY292004 for 3 h. We then performed immunofluorescence to analyze the Nrf2 distribution. Representative images of immunofluorescence staining are shown in Fig. 2. Three hours after treatment with PD98059 plus LY292004, the nuclear staining of Nrf2 was decreased in all three cell lines. In addition, the cytoplasmic staining of Nrf2 was also decreased in the U251 and U87 cells. We found that the U251 cells were more sensitive to the inhibitors than the other cell lines. Thus, we chose the U251 cells to further verify these findings and carry out the subsequent studies.

Next, to further determine whether combined inhibition of ERK and PI3K decreases the expression and activation of Nrf2, we analyzed Nrf2 protein in the nuclear and cytoplasmic fractions of U251 cells by western blot analysis. In addition, we further investigated the effect of each inhibitor on the expression and activation of Nrf2. The U251 cells were harvested after different treatments (untreated, PD98059, LY292004 or PD98059+LY292004) for 3 h and the cytoplasmic and nuclear fractions were isolated from the cells. Representative immunoblots are shown in Fig. 3. Three hours after treatment with PD98059 plus LY292004, there was a significant decrease in the protein level of nuclear Nrf2 when compared with the protein level in the untreated group. Consistent with the immunofluorescence results, there was also a decrease in the protein level of cytoplasmic Nrf2 following combined inhibition of ERK and PI3K. As expected, inhibiting either ERK or PI3K moderately suppressed nuclear accumulation of Nrf2 protein. However, there was no significant decrease in the expression of cytoplasmic Nrf2 following treatment with either inhibitor alone. In addition, there was no statistical change in the mRNA level of Nrf2 after treatment with both inhibitors (data not shown). Therefore, combined inhibition of ERK and PI3K signaling not only reduced nuclear accumulation of Nrf2 but also suppressed the protein expression of Nrf2 in the U251 cells.

Nrf2 activation plays a cytoprotective role through upregulating the expression of Nrf2-ARE downstream target genes, therefore, we examined whether inhibition of ERK and PI3K suppresses the expression of Nrf2 target genes. The U251 cells were treated with PD98059 plus LY292004 for 3 h, and then real-time PCR analysis was performed. As shown in Fig. 4, the mRNA levels of Nrf2-ARE downstream target genes (HO-1 and NQO-1) were decreased in the PD98059+LY292004 group in comparison with the DMSO group. Together, these results imply that the ERK and PI3K signaling cascades are closely associated with the expression and activation of Nrf2 in human glioblastoma cells.

It is well known that the ERK and PI3K signaling cascades play central roles in controlling cancer cell viability. There are numerous downstream proteins involved in cell viability. Thus, we evaluated whether the ERK and PI3K signaling cascades regulate cell viability through Nrf2. Here, first we assessed the effect of Nrf2 on U251 cell viability. The U251 cells were transfected with different plasmids (sh-control, sh-Nrf2, pEGFP and pEGFP-Nrf2). Following transfection, cells were incubated for 48 h, and cell viability was detected by MTT analysis. As shown in Fig. 5A, downregulation of Nrf2 protein suppressed U251 cell viability; the viability rate decreased from 96.02 (the sh-control group) to 86.99% (the sh-Nrf2 group). On the contrary, upregulation of Nrf2 protein promoted U251 cell growth; the relative growth rate increased from 96.90 (the pEGFP group) to 105.90% (the pEGFP-Nrf2 group). There was no difference between the two control groups (the sh-control group compared with the pEGFP group). In addition, the transfection effect of the four plasmids was further verified by western blot analysis (Fig. 5B).

Next, we investigated whether the overexpression of Nrf2 protein could partly reverse the MEK and PI3K inhibitor-induced inhibition of cancer cell viability. The U251 cells were treated with PD98059 plus LY294002 for 48 h. The morphological changes in the U251 cells in the different groups were observed by microscopy. As shown in Fig. 6A, we observed that many cells lost normal shape and detached from the culture plates (the PD98059+LY292004 group and the PD98059+LY292004+pEGFP group). However, fewer abnormal cells were noted in the PD98059+LY292004+pEGFP-Nrf2 group than that in the PD98059+LY292004+pEGFP group. Cell viability was further measured by MTT assay. As shown in Fig. 6B, combined inhibition of ERK and PI3K markedly reduced U251 cell viability; the viability rate decreased from 98.20 (the untreated group) to 65.60% (the PD98059+LY292004 group). Upregulation of Nrf2 protein partly reversed the ERK and PI3K inhibitor-induced U251 cell viability; the viability rate increased from 64.57 (the PD98059+LY292004+pEGFP group) to 69.41% (the PD98059+LY292004+pEGFP-Nrf2 group). Thus, overexpression of Nrf2 reversed nearly one-sixth of the total inhibition. There was no difference between the PD98059+LY292004 group and the PD98059+LY292004+pEGFP group. These results suggest that the ERK and PI3K signaling cascades regulate cell viability partly through Nrf2 in the U251 cells.