IJC International Journal of Cancer
Knockdown of Nrf2 suppresses glioblastoma angiogenesis by inhibiting hypoxia-induced activation of HIF-1a Xiangjun Ji1, Handong Wang1, Jianhong Zhu1, Lin Zhu1, Hao Pan1, Wei Li1, Yuan Zhou1, Zixiang Cong1, Feng Yan2 and Suihua Chen2 1
Cancer Cell Biology
2
Department of Neurosurgery, Jinling Hospital, Nanjing University School of Medicine, Nanjing Jiangsu Province, People’s Republic of China Department of Ophthalmology, Jinling Hospital, Nanjing University School of Medicine, Nanjing Jiangsu Province, People’s Republic of China
Concerns were increasingly raised that several types of cancers overexpressed the nuclear factor erythroid 2-related factor 2 (Nrf2), which contributed strikingly to cancer biological capabilities and chemoresistance. However, the role of Nrf2 in the tumor vascular biology had yet to be mechanistically determined. Here, we investigated the involvement of Nrf2 in glioblastoma (GB) angiogenesis in hypoxia. First, we detected the overexpression of Nrf2 and correlated its protein level with microvessel density (MVD) in human GB tissues. Then, we established the stable RNAi-mediated Nrf2-knockdown cells and mimicked hypoxic condition in vitro. The knockdown of Nrf2 inhibited cell proliferation in vitro and suppressed tumor growth in mouse xenografts with a concomitant reduction in VEGF expression and MVD. Similar antiangiogenic effects were documented in endothelial tube formation assays. The downregulation of Nrf2 in glioma cells led to much lower accumulation of HIF-1a protein and limited expression of VEGF and other HIF-1a target genes in mimicking hypoxia. Mechanistic investigations suggested that HIF-1a degradation during hypoxia could be attributed to reduced mitochondrial O2 consumption in Nrf2inhibited cells. It can be concluded that Nrf2, with its capacity for affecting the protein level of HIF-1a expression, has good reasons to be considered as a critical transcription factor for controlling glioma angiogenesis.
Glioblastoma (GB) is the most common and malignant form of human glioma.1 The average life expectancy of patients with GB after diagnosis is 14 months or so under the current standard of care, despite aggressive surgery, radiotherapy and chemotherapies.2 Unprecedented new knowledge has been acquired and numerous new discoveries have been made in the basic and translational research. Typically, gene-targeted therapies have found their ways into clinical practice, as evidenced by the recent approval of bevacizumab, an antibody
to vascular endothelial growth factor (VEGF), for the treatment of recurrent or progressive GB.3 But, further efforts were still required to progressively elucidate the limited administration and resistance of gene-targeted therapies. Nuclear factor erythroid 2-related factor 2 (Nrf2), a pivotal transcriptional factor for cellular responses to oxidative stress, belongs to the Keap1-Nrf2-ARE (antioxidant response element) signaling pathway and upregulates many AREcontaining genes such as aldo-keto reductase 1C1 (AKR1C1),
Key words: glioblastoma, Nrf2, HIF-1a, angiogenesis, VEGF Abbreviations: AKR1C1: aldo-keto reductase 1C1; ARE: antioxidant response element; BrdU: bromodeoxyuridine; BSA: bovine serum albumin; CCK-8: Cell-Counting kit-8; DMEM: Dulbecco’s modified Eagle’s medium; ELISA: enzyme-linked immunosorbent assay; FBS: fetal bovine serum; GB: glioblastoma; GCL: glutamylcysteine ligase; GSH: glutathione; GST: glutathione S-transferase; HIF-1a: hypoxiainducible factor-1a; HO-1: heme oxygenase-1; HUCECs: human umbilical vein endothelial cells; IRS: immunoreactivity scores; kDa: kilodalton; MVD: microvessel density; NQO1: NAD(P)H:quinone oxidoreductase-1; Nrf2: NF-E2-related factor 2; PBS: phosphate-buffered saline; PCR: polymerase chain reaction; PHDs: prolyl hydroxylase domain proteins; pVHL: Von Hippel-Lindau protein; RT: reverse transcription; VEGF: vascular endothelial growth factor. Conflict of interest: The authors declare no conflict of interest. This manuscript has been read and approved by all the authors, and not submitted or under considering for publication elsewhere. Grant sponsor: Natural Science Foundation of China; Grant number: 81271377/H0910; Grant sponsor: Science Foundation of Jinling Hospital; Grant number: 2013035 DOI: 10.1002/ijc.28699 History: Received 7 June 2013; Revised 6 Nov 2013; Accepted 17 Dec 2013; Online 28 Dec 2013 Correspondence to: H. Wang, MD, PhD, Department of Neurosurgery, Jinling Hospital, Nanjing University School of Medicine, 305 East Zhongshan Road, Nanjing 210002, Jiangsu Province, People’s Republic of China, Tel.: 186-025-51805396, Fax: 186-025-51805396, E-mail: [email protected] or S. Chen, MD, PhD, Department of Ophthalmology, Jinling Hospital, Nanjing University School of Medicine, 305 East Zhongshan Road, Nanjing 210002, Jiangsu Province, People’s Republic of China, Tel.: 186-025-80860047, E-mail: [email protected]
C 2013 UICC Int. J. Cancer: 135, 574–584 (2014) V
575
Ji et al.
What’s new? Nrf2, an important transcriptional factor in cellular responses to oxidative stress, is strongly suspected to play a pivotal role in cancer cellsurvival and tumor growth. But the role of Nrf2 in tumor vascular biology has yet to be mechanistically determined. This study demonstrated for the first time that Nrf2 played a pivotal role in glioblastoma angiogenesis. Human glioblastomatissues expressing higher Nrf2 levels showed relatively higher microvessel density. Knockdown of Nrf2 in hypoxia led to reduced angiogenesis through lower HIF-1a-VEGF signaling. The findings highlighted Nrf2 as a candidate molecular target for controlling glioblastoma angiogenesis throughblockade of HIF-1a signaling.
C 2013 UICC Int. J. Cancer: 135, 574–584 (2014) V
In our study, we investigated the role of Nrf2 in GB angiogenesis. Human GB tissues expressing higher Nrf2 levels showed relatively higher values of microvessel density (MVD). Glioma cells expressing Nrf2-specific shRNA showed slower tumor growths, lower VEGF expressions and diminishing vascular formations in mouse xenografts and angiogenesis models than they showed in the Sc-RNA control.
Material and Methods Patients and tissue samples
A consecutive series of 58 GB patients admitted at Jinling Hospital from 2008 to 2010, according to the ethical and legal standards, were included in our study. Histological diagnosis and grading of the tumors were performed in compliance with WHO criteria (World Health Organization, 2007). There were no gender or ethnicity restrictions on recruitment. None of the patients had received chemotherapy or radiotherapy before the surgery. Human normal brain tissues (all from the cortex) were obtained from five male and five female patients in the pathway during surgical removal of deep benign tumor from 2010 to 2011 in Jinling Hospital. This study was approved by the Research Ethics Committee of Jinling Hospital, School of Medicine, Nanjing University, People’s Republic of China. Written informed consent was obtained from all the patients. Totally 51 patients [35 males and 16 females, average age 53 (27–82)] completed the follow-up until death and the survival time was censored in November 2012, while seven patients were lost. There were no patients dying of diseases not directly related to their gliomas or owing to unexpected events. The adjuvant chemotherapy and radiotherapy were fully discussed with the patients before the treatment and all the patients had received the standard chemotherapy and radiotherapy. Cell culture, plasmid transfection and establishment of Nrf2-knockdown cells
Human glioma cells U251 and U87 were purchased from the Type Culture Collection of the Chinese Academy of Sciences, Shanghai, China. The cells were maintained in Dulbecco’s modified Eagle’s medium (DMEM) (Gibco, Los Angeles, CA) supplemented with 10% fetal bovine serum (FBS) (Gibco) and 1% penicillin/streptomycin (Gibco), and were incubated at 37 C in a 5% CO2 incubator. The lentiviral particles with Nrf2-specific short hairpin RNA (shRNA) were purchased
Cancer Cell Biology
heme oxygenase-1 (HO-1), NAD(P)H:quinone oxidoreductase-1 (NQO1), glutamylcysteine ligase (GCL) and glutathione S-transferase (GST). The enzymes GCL and GST are closely related to the synthesis and metabolism of the major antioxidant of glutathione (GSH) in central nervous system.4 For a long period of time, Nrf2 had been considered to help protect cells in normal tissues from harmful stimuli such as inflammation, trauma, ischemia, hemorrhage and cancer.5–9 However, the latest findings suggested that Nrf2 may have played dark and deteriorating roles in tumors. According to several of these studies, constitutively high level of Nrf2 promoted cancer formation and contributed to chemoresistance.10–13 Further investigation demonstrated that Nrf2 helped proliferate cells measurably, highly activating its target genes with various detoxification enzymes, glutathione-related enzymes and cell-cycle regulatory proteins included.14–16 Moreover, we have also observed recently that Nrf2 was involved in apoptosis, migration and invasion of glioma cell U251.17,18 All these studies supported the concept that Nrf2 actually played a pivotal role in cancer cell survival and tumor growth. In accordance with a recent finding of Kim et al., Nrf2, as a candidate molecular target, might have regulated colon tumor angiogenesis by imposing a blockade to hypoxia-inducible factor-1a (HIF-1a) signaling.19 HIF-1 is a critical mediator of cellular responses to hypoxia, containing the O2-regulated subunit HIF-1a and the constitutively expressed subunit HIF-1b.20 HIF-1a could, if in response to hypoxia, upregulate the expression of many responsible genes for the hypoxia and the adaptation. This in turn regulated multiple aspects of the endothelial behavior, including cell proliferation, chemotaxis and extracellular matrix penetration.21 Often, HIF-1a was deregulated in tumors located in hypoxic conditions and played a key role in angiogenesis, tumor progression and resistance to radiotherapy and chemotherapies.22 In normal O2 concentrations, prolyl hydroxylase domain proteins (PHDs), enzymes that consumed O2, could hydroxylate specific HIF-1a proline residues. Bound by the Von Hippel-Lindau protein (pVHL), HIF-1a was thereafter hydroxylated and degraded. In hypoxia, that PHDs were inactivated with insufficient O2, combined with the hydroxylation of HIF-1a inhibited, led to HIF-1a accumulated and the expression of target genes, including VEGF in particular, increased.21,23 Hence, HIF-1a, as the critical mediator to hypoxia, emerged as a key molecular regulator of angiogenesis.
576
from Gene-Pharma (Shanghai, China) and the target sequence was GCAGTTCAATGAAGCTCAACT. The new plasmid was named as Nrf2i. Random sequence, TTCTCC GAACGTGTCACGT, was used as the negative control, which was named as Sc. BLAST research ensured that the sequences had no significant homology with other human genes. Cell transfection was performed in six-well plates with lentiviral particles containing either Sc RNA or Nrf2i shRNA expression plasmid using Polybrene (Gene-Pharma, Shanghai, China) according to the manufacturer’s instruction. Transfection was continued for 24 hr and followed by a 24-hr recovery in the complete medium. For the selection of cells with target plasmids, cells were grown in the medium containing 1.5 lg/ml puromycin (Sigma-Aldrich, St. Louis, MO) for up to 2 weeks. Then, the positive clonal cells were selected and cultured in 96well plates with the medium containing 1.0 lg/ml puromycin.
Cancer Cell Biology
Total RNA extraction and RT-PCR analysis
Total RNA was isolated using the TRIzol reagent (Invitrogen, CA, USA) following the manufacturer’s recommendations and subjected to DNase (Promega, Madison, WI) treatment. The concentration and purity of total RNA were determined by spectrophotometer analysis (OD260/280: 1.8–2.2) and agarose gel electrophoresis. Reverse transcriptase (RT) reactions were performed by incubating 400 ng RNA with the first-strand cDNA synthesis kit (Takara, Dalian, China). Obtained cDNA was amplified immediately using the following primers: for Nrf2, 50 -TCAGCGACGGAAAGAGTATGA-30 and 50 -CCAC TGGTTTCTGACTGGATGT-30 ; for NQO1, 50 -ATGGTCGG CAGAAGAGC-30 and 50 -GGAAATGATGGGATTGAAGT-30 ; for HO-1, 50 -TCTCCGATGGGTCCTTACACTC-30 and 50 GGCATAAAGCCCTACAGCAACT-30 ; for HIF-1a, 50 -AT CCATGTGACCATGAGGAAATG-30 and 50 -CTCGGCTAGT TAGGGTACACTT-30 ; for VEGF, 50 -GAAGGAGGAGGGC AGAAT-30 and 50 -CGATTGGATGGCAGTAGC-30 and for GAPDH, 50 -GAAATCCCATCACCATCTTC-30 and 50 -GGA CTCCACGACGTACTCA-30 . The amplification and data acquisition were carried out on a real-time PCR system (Agilent) using FastStart universal SYBR green master (Roche, Mannheim, Germany). The conditions were predenaturated at 95 C for 10 min, followed by 40 circles at 95 C for 15 sec and 60 C for 1 min. All samples were analyzed in triplicates in three independent experiments. Reaction without cDNA was used as no-template control, and no-RT controls were also set up to rule out genomic DNA contamination. Relative quantification of mRNA expression was determined using the 22䉭䉭Cq method. PCR amplification efficiency of each gene was established by means of calibration curves. The level of GAPDH was used as an internal reference gene. All the real-time PCR experiments were performed in accordance to the Minimum Information for publication of quantitative real-time PCR experiments.24
Nrf2 in GB angiogenesis
Protein concentrations were estimated by Coomassie Plus Protein Assay Reagent (Pierce, Rockford, IL). Protein was separated by 8–10% sodium dodecyl sulfate polyacrylamide gels electrophoresis using the Criterion system (Bio-Rad, Hercules, CA) and electroblotted onto a nitrocellulose membrane (Millipore, Billerica, MA). For immunoblotting, membranes were blocked with 5% skim milk for 2 hr at room temperature, and the following antibodies were used: 1:500 anti-Nrf2 (Abcam, Cambridge, MA; 68 kDa), 1:1,000 antiHO-1 (Abcam; 33 kDa), 1:1,000 anti-AKR1C1 (Abcam; 37 kDa), 1:2,000 anti-b-actin (Santa Cruz Biotechnology, Santa Cruz, CA; 43 kDa) and 1:1,000 anti-HIF-1a (Novus Biologicals, LLC, Littleton, CO; 120 kDa). Each primary antibody was diluted appropriately in blocking buffer and then incubated overnight at 4 C. Then, membranes were incubated with the appropriate secondary antibodies (Cell Signaling, Danvers, MA; 1:2,000) for 2 hr at room temperature. After washing, protein bands were visualized with Chemiluminescent HRP Substrate (Millipore) at room temperature and exposed to X-ray film (Fujifilm, Tokyo, Japan). Relative changes in protein expression were estimated from the mean pixel density using Quantity One software 4.6.2 (Bio-Rad), normalized to b-actin and presented as relative density units. Cell proliferation assay
Cell proliferation was measured by Cell-Counting kit-8 (CCK-8) (Dojindo, Kumamoto, Japan) and cell proliferation enzyme-linked immunosorbent assay (ELISA) kit (Roche) according to the manufacturer’s instructions. Cells were plated in 96-well plates at a density of 2 3 103 cells per well. After the incubation for 24 and 48 hr, 10 ll CCK-8 was added and cells were further incubated for 1 hr. Then, the absorbance was measured at 450 nm using the ELISA microplate reader (Bio-Rad). DNA synthesis of cells was measured using the ELISA kit, which is a colorimetric immunoassay based on the assessment of bromodeoxyuridine (BrdU) incorporation during DNA synthesis. Cells were plated in 96-well plates at a density of 2 3 103 cells per well and grown for 24 hr. Then, cells were labeled with 10 lM BrdU for 2 hr, and incorporated BrdU was quantified at 492 nm using the ELISA microplate reader (Bio-Rad). Measurement of total GSH contents
Cells were grown in six-well plates for 24 hr, collected and lysed with protein detergent S solvent. The total GSH was determined by commercially available Total Glutathione Assay Kit (Beyotime Institute of Biotechnology, Shanghai, China). All procedures completely complied with the manufacture’s instructions. Protein concentration was estimated by Coomassie Plus Protein Assay Reagent (Pierce). Quantification of VEGF protein levels by ELISA
Western blot analysis
To obtain total protein lysate, cells were homogenized in RIPA buffer and centrifuged at 12,000g for 15 min at 4 C.
The concentration of VEGF protein was measured using Human VEGF-A Platinum ELISA kit from eBioscience (eBioscience, San Diego, CA). Sc and Nrf2i cells in FBS-free C 2013 UICC Int. J. Cancer: 135, 574–584 (2014) V
DMEM were incubated in 24-well plates, supernatants were collected and cell number of each well was counted. VEGF levels in the supernatant (100 ll) were determined and normalized to the cell number (105). A serial dilution of human recombinant VEGF was included in each assay to obtain a standard curve.
Human umbilical vein endothelial cells (HUVECs) were purchased from Lonza (Basel, Switzerland) and maintained in DMEM (Gibco) supplemented with 10% FBS (Gibco). To assess the effect of Nrf2 on HUVECs tube formation, Sc and Nrf2i cells in supplement-free DMEM were incubated in hypoxic condition for 24 hr, and supernatants (conditioned media) were collected. Then, HUVECs were plated in 96-well plates (1.8 3 104 cells per well), which were coated by Matrigel (BD Biosciences, Bedford, MA). Then, the conditioned media or fresh DMEM containing VEGF (10 ng/ml) (Sigma) were added into cells. After 10 hr, three or more random pictures were taken from each well using a digital camera system. Image J was used for the quantification of tube lengths.
eosin (HE) staining was performed to identify the histological profiles of tumor cells. The slides were scored by two independent observers. Occupied regions by tumor cells (% over trimmed tumor mass) were counted in the HE staining using Image J. The number of positive staining cells was calculated from eight representative staining fields per each specimen under 4003 magnifications. The staining intensity was estimated and stratified as 0, no staining; 1, weak staining (light yellow); 2, moderate staining (yellow brown) and 3, strong staining (brown).26 The immunohistochemistry positive results for Nrf2 were recorded as previously described.25–27 Then, we scored the percentage of immunoreactive tumor cells as 0 (0%), 1 (>0% and 10%), 2 (>10% and 50%) and 3 (>50%). Final immunoreactivity scores (IRS) of Nrf2 expression were obtained for each specimen by multiplying the percentage and the intensity score. At last we estimated the protein expression levels by classifying IRS values as negative (based on IRS value 0 and 1) and positive (based on IRS value greater than 1), combined with as low (based on IRS values as 0, 1, 2, 3 and 4) and high (based on IRS value as 6 and 9).
Tumor xenograft study
Microvessel density
All procedures in animals were approved by the Animal Care and Use Committee of Nanjing University and conformed to Guide for the Care and Use of Laboratory Animals from National Institutes of Health. Nrf2i and Sc shRNA-stabletransfected U251 or U87 cells (5.0 3 106) were suspended in 100 ll phosphate-buffered saline (PBS) and then injected subcutaneously into right side of the anterior flank of the male BALB/c athymic nude mice (Charles River Breeding Laboratories, Wilmington, MA) at 4–5 weeks of age. Tumor growth was examined every 3 days from tumor formation for at least 5 weeks. Then, tumors were stripped and weighed before paraffin embedding. Tumor volumes were determined by external measurements and calculated according to V 5 [L 3 W2 3 0.52 (V 5 volume, L 5 length and W 5 width).
MVD of tumor tissues was assessed through the immunohistochemical analysis of endothelial marker CD31 and determined according to the method of Weidner et al.28 First, immunostained sections were initially screened at low magnification (503) to identify hot spots of the neovascularization. Within hot-spot areas, the stained microvessels were counted in a single high-power (2003) field, and the average vessel counted in three hot spots was considered as the value of MVD. Any yellow-brown stained endothelial cell or endothelial cell cluster obviously isolated from the adjacent microvessels, tumor cells and other connective tissue elements was considered a single, countable microvessel. All counts were performed by three investigators in a blinded manner. Comparisons regarding microvessel counts were drawn among the observers and discrepant results were reassessed. The consensus was used as the final score for analysis.29
HUVEC culture and tube formation assay
Immunohistochemical staining
Serial sections of 3-lm thick were dewaxed and endogenous peroxidase was quenched with 3% H2O2 in methanol for 30 min.25 Before staining, nonspecific binding was blocked by incubation with 10% bovine serum albumin (BSA) in PBS at 37 C for 1 hr. Then, all incubations with 1:50 anti-Nrf2 (Abcam), 1:50 anti-Caspase-3 (Abcam) and 1:100 anti-CD31 (Abcam) antibodies in PBS containing 1% BSA were carried out at 4 C overnight. All sections were briefly washed in PBS and incubated at room temperature with the anti-rabbit antibody and avidin-biotin peroxidase (Vector Laboratories, Burlingame, CA). Color was then developed by incubation with the diaminobenzidine solution (Dako Corporation, Carpinteria, CA). Nuclei were counterstained blue with Meyer’s hematoxylin (Sigma-Aldrich). Nonneoplastic brain tissues were used as control and nonimmune IgG was used as negative control antibody for the staining. The hematoxylin and C 2013 UICC Int. J. Cancer: 135, 574–584 (2014) V
Statistical analysis
SPSS 19.0 software (SPSS, Chicago, IL) was used for statistical analysis. Statistical data were presented as mean 6 SD. One-way analysis of variance (ANOVA) followed by Tukey post hoc comparison tests was used to compare the levels of different experimental groups. p-Values of 0.05) (Fig. 5c). To conclude, the suppression effect on the proliferation rate of Nrf2i-U251 cells was far more apparent than that of Nrf2i-U87 cells. The next step was to detect the concentration of VEGF protein in the supernatant of tumor cells incubated in hypoxia. The accumulation of VEGF increased continuously throughout the incubation in both cell lines and peaked at 24–36 hr (Figs. 5d and 5e). To avoid cells dying of hunger, assessment at 24 hr was chosen for further investigation. Then, the knockdown of Nrf2 was observed to have diminished the accumulation of VEGF in hypoxia (1% O2) more greatly in U87 cells than in U251 cells (Fig. 5f). Then, the
ELISA assay on accumulated VEGF was preformed under the CoCl2-mimicking hypoxia. The administration of CoCl2 was done according to the pre-experiment and the study by Kim et al.19 Then, cells were treated with 200 lM CoCl2 for 24 hr and VEGF level was assessed. The repressive effect on VEGF level appeared to be concordant with that of 1% O2-mimicking hypoxia (Fig. 5g). To validate the role of Nrf2 in the angiogenesis of GB we performed the endothelial tube formation assay. We incubated GB cells with supplement-free DMEM medium in hypoxia and collected the supernatants. HUVECs were then incubated with DMEM or the supernatants. Cells incubated with fresh DMEM were taken as the negative control and those incubated with fresh DMEM containing VEGF (10 ng/ ml) were used as the positive control. It was observed that the hypoxia-exposed Nrf2i medium yielded notably reduced endothelial tube formation compared to the Sc group or the VEGF treatment group (Figs. 5h–5j). These findings C 2013 UICC Int. J. Cancer: 135, 574–584 (2014) V
581
Figure 5. Diminished VEGF accumulation and tube formation in Nrf2-inhibited cells. Cell proliferation was assessed using the CCK-8 assay (a, b) and BrdU incorporation assay (c) in hypoxia. (a) The CCK-8 assay on U251 cells performed at 0, 24 and 48 hr showed that the proliferation rate of Nrf2i group was suppressed significantly at 48 hr (p 5 0.0022, ANOVA). (b) U87 cells were observed to be slightly repressed in the CCK-8 assay (p 5 0.0496, ANOVA). (c) BrdU incorporation assessment confirmed the knockdown effect on the proliferation of U251 cells (p 5 0.0212), while showed no notable differences among groups in U87 (p > 0.05). Values are means 6 SD from six wells. (d–g) The ELISA assay quantified the concentration of VEGF protein in supernatants of tumor cell culture in hypoxia. (d, e) The accumulation of VEGF increased continuously throughout incubation in hypoxia (1% O2) in both cell lines and peaked at 24–36 hr compared with cells treated in normoxia. (f) Knockdown of Nrf2 diminished the 24-h accumulation of VEGF in hypoxia. (g) The ELISA assay on VEGF under CoCl2-mimicking hypoxia coincided with the Nrf2-inhibited effects. (h–j) Angiogenesis evaluated in the endothelial tube formation assay. Transfected cells from U251 (h) and U87 (i) were incubated in hypoxia (1% O2) with supplement-free DMEM medium for 24 hr and the supernatants were collected. HUVECs were then incubated with DMEM or the supernatants, and the length of formed tubes was quantified (j). VEGF treatment group was used as the positive control and supplement-free DMEM treatment group as the negative control. Bar represents mean 6 SD; *p < 0.05, **p
Knockdown of Nrf2 suppresses glioblastoma angiogenesis by inhibiting hypoxia-induced activation of HIF-1a Xiangjun Ji1, Handong Wang1, Jianhong Zhu1, Lin Zhu1, Hao Pan1, Wei Li1, Yuan Zhou1, Zixiang Cong1, Feng Yan2 and Suihua Chen2 1
Cancer Cell Biology
2
Department of Neurosurgery, Jinling Hospital, Nanjing University School of Medicine, Nanjing Jiangsu Province, People’s Republic of China Department of Ophthalmology, Jinling Hospital, Nanjing University School of Medicine, Nanjing Jiangsu Province, People’s Republic of China
Concerns were increasingly raised that several types of cancers overexpressed the nuclear factor erythroid 2-related factor 2 (Nrf2), which contributed strikingly to cancer biological capabilities and chemoresistance. However, the role of Nrf2 in the tumor vascular biology had yet to be mechanistically determined. Here, we investigated the involvement of Nrf2 in glioblastoma (GB) angiogenesis in hypoxia. First, we detected the overexpression of Nrf2 and correlated its protein level with microvessel density (MVD) in human GB tissues. Then, we established the stable RNAi-mediated Nrf2-knockdown cells and mimicked hypoxic condition in vitro. The knockdown of Nrf2 inhibited cell proliferation in vitro and suppressed tumor growth in mouse xenografts with a concomitant reduction in VEGF expression and MVD. Similar antiangiogenic effects were documented in endothelial tube formation assays. The downregulation of Nrf2 in glioma cells led to much lower accumulation of HIF-1a protein and limited expression of VEGF and other HIF-1a target genes in mimicking hypoxia. Mechanistic investigations suggested that HIF-1a degradation during hypoxia could be attributed to reduced mitochondrial O2 consumption in Nrf2inhibited cells. It can be concluded that Nrf2, with its capacity for affecting the protein level of HIF-1a expression, has good reasons to be considered as a critical transcription factor for controlling glioma angiogenesis.
Glioblastoma (GB) is the most common and malignant form of human glioma.1 The average life expectancy of patients with GB after diagnosis is 14 months or so under the current standard of care, despite aggressive surgery, radiotherapy and chemotherapies.2 Unprecedented new knowledge has been acquired and numerous new discoveries have been made in the basic and translational research. Typically, gene-targeted therapies have found their ways into clinical practice, as evidenced by the recent approval of bevacizumab, an antibody
to vascular endothelial growth factor (VEGF), for the treatment of recurrent or progressive GB.3 But, further efforts were still required to progressively elucidate the limited administration and resistance of gene-targeted therapies. Nuclear factor erythroid 2-related factor 2 (Nrf2), a pivotal transcriptional factor for cellular responses to oxidative stress, belongs to the Keap1-Nrf2-ARE (antioxidant response element) signaling pathway and upregulates many AREcontaining genes such as aldo-keto reductase 1C1 (AKR1C1),
Key words: glioblastoma, Nrf2, HIF-1a, angiogenesis, VEGF Abbreviations: AKR1C1: aldo-keto reductase 1C1; ARE: antioxidant response element; BrdU: bromodeoxyuridine; BSA: bovine serum albumin; CCK-8: Cell-Counting kit-8; DMEM: Dulbecco’s modified Eagle’s medium; ELISA: enzyme-linked immunosorbent assay; FBS: fetal bovine serum; GB: glioblastoma; GCL: glutamylcysteine ligase; GSH: glutathione; GST: glutathione S-transferase; HIF-1a: hypoxiainducible factor-1a; HO-1: heme oxygenase-1; HUCECs: human umbilical vein endothelial cells; IRS: immunoreactivity scores; kDa: kilodalton; MVD: microvessel density; NQO1: NAD(P)H:quinone oxidoreductase-1; Nrf2: NF-E2-related factor 2; PBS: phosphate-buffered saline; PCR: polymerase chain reaction; PHDs: prolyl hydroxylase domain proteins; pVHL: Von Hippel-Lindau protein; RT: reverse transcription; VEGF: vascular endothelial growth factor. Conflict of interest: The authors declare no conflict of interest. This manuscript has been read and approved by all the authors, and not submitted or under considering for publication elsewhere. Grant sponsor: Natural Science Foundation of China; Grant number: 81271377/H0910; Grant sponsor: Science Foundation of Jinling Hospital; Grant number: 2013035 DOI: 10.1002/ijc.28699 History: Received 7 June 2013; Revised 6 Nov 2013; Accepted 17 Dec 2013; Online 28 Dec 2013 Correspondence to: H. Wang, MD, PhD, Department of Neurosurgery, Jinling Hospital, Nanjing University School of Medicine, 305 East Zhongshan Road, Nanjing 210002, Jiangsu Province, People’s Republic of China, Tel.: 186-025-51805396, Fax: 186-025-51805396, E-mail: [email protected] or S. Chen, MD, PhD, Department of Ophthalmology, Jinling Hospital, Nanjing University School of Medicine, 305 East Zhongshan Road, Nanjing 210002, Jiangsu Province, People’s Republic of China, Tel.: 186-025-80860047, E-mail: [email protected]
C 2013 UICC Int. J. Cancer: 135, 574–584 (2014) V
575
Ji et al.
What’s new? Nrf2, an important transcriptional factor in cellular responses to oxidative stress, is strongly suspected to play a pivotal role in cancer cellsurvival and tumor growth. But the role of Nrf2 in tumor vascular biology has yet to be mechanistically determined. This study demonstrated for the first time that Nrf2 played a pivotal role in glioblastoma angiogenesis. Human glioblastomatissues expressing higher Nrf2 levels showed relatively higher microvessel density. Knockdown of Nrf2 in hypoxia led to reduced angiogenesis through lower HIF-1a-VEGF signaling. The findings highlighted Nrf2 as a candidate molecular target for controlling glioblastoma angiogenesis throughblockade of HIF-1a signaling.
C 2013 UICC Int. J. Cancer: 135, 574–584 (2014) V
In our study, we investigated the role of Nrf2 in GB angiogenesis. Human GB tissues expressing higher Nrf2 levels showed relatively higher values of microvessel density (MVD). Glioma cells expressing Nrf2-specific shRNA showed slower tumor growths, lower VEGF expressions and diminishing vascular formations in mouse xenografts and angiogenesis models than they showed in the Sc-RNA control.
Material and Methods Patients and tissue samples
A consecutive series of 58 GB patients admitted at Jinling Hospital from 2008 to 2010, according to the ethical and legal standards, were included in our study. Histological diagnosis and grading of the tumors were performed in compliance with WHO criteria (World Health Organization, 2007). There were no gender or ethnicity restrictions on recruitment. None of the patients had received chemotherapy or radiotherapy before the surgery. Human normal brain tissues (all from the cortex) were obtained from five male and five female patients in the pathway during surgical removal of deep benign tumor from 2010 to 2011 in Jinling Hospital. This study was approved by the Research Ethics Committee of Jinling Hospital, School of Medicine, Nanjing University, People’s Republic of China. Written informed consent was obtained from all the patients. Totally 51 patients [35 males and 16 females, average age 53 (27–82)] completed the follow-up until death and the survival time was censored in November 2012, while seven patients were lost. There were no patients dying of diseases not directly related to their gliomas or owing to unexpected events. The adjuvant chemotherapy and radiotherapy were fully discussed with the patients before the treatment and all the patients had received the standard chemotherapy and radiotherapy. Cell culture, plasmid transfection and establishment of Nrf2-knockdown cells
Human glioma cells U251 and U87 were purchased from the Type Culture Collection of the Chinese Academy of Sciences, Shanghai, China. The cells were maintained in Dulbecco’s modified Eagle’s medium (DMEM) (Gibco, Los Angeles, CA) supplemented with 10% fetal bovine serum (FBS) (Gibco) and 1% penicillin/streptomycin (Gibco), and were incubated at 37 C in a 5% CO2 incubator. The lentiviral particles with Nrf2-specific short hairpin RNA (shRNA) were purchased
Cancer Cell Biology
heme oxygenase-1 (HO-1), NAD(P)H:quinone oxidoreductase-1 (NQO1), glutamylcysteine ligase (GCL) and glutathione S-transferase (GST). The enzymes GCL and GST are closely related to the synthesis and metabolism of the major antioxidant of glutathione (GSH) in central nervous system.4 For a long period of time, Nrf2 had been considered to help protect cells in normal tissues from harmful stimuli such as inflammation, trauma, ischemia, hemorrhage and cancer.5–9 However, the latest findings suggested that Nrf2 may have played dark and deteriorating roles in tumors. According to several of these studies, constitutively high level of Nrf2 promoted cancer formation and contributed to chemoresistance.10–13 Further investigation demonstrated that Nrf2 helped proliferate cells measurably, highly activating its target genes with various detoxification enzymes, glutathione-related enzymes and cell-cycle regulatory proteins included.14–16 Moreover, we have also observed recently that Nrf2 was involved in apoptosis, migration and invasion of glioma cell U251.17,18 All these studies supported the concept that Nrf2 actually played a pivotal role in cancer cell survival and tumor growth. In accordance with a recent finding of Kim et al., Nrf2, as a candidate molecular target, might have regulated colon tumor angiogenesis by imposing a blockade to hypoxia-inducible factor-1a (HIF-1a) signaling.19 HIF-1 is a critical mediator of cellular responses to hypoxia, containing the O2-regulated subunit HIF-1a and the constitutively expressed subunit HIF-1b.20 HIF-1a could, if in response to hypoxia, upregulate the expression of many responsible genes for the hypoxia and the adaptation. This in turn regulated multiple aspects of the endothelial behavior, including cell proliferation, chemotaxis and extracellular matrix penetration.21 Often, HIF-1a was deregulated in tumors located in hypoxic conditions and played a key role in angiogenesis, tumor progression and resistance to radiotherapy and chemotherapies.22 In normal O2 concentrations, prolyl hydroxylase domain proteins (PHDs), enzymes that consumed O2, could hydroxylate specific HIF-1a proline residues. Bound by the Von Hippel-Lindau protein (pVHL), HIF-1a was thereafter hydroxylated and degraded. In hypoxia, that PHDs were inactivated with insufficient O2, combined with the hydroxylation of HIF-1a inhibited, led to HIF-1a accumulated and the expression of target genes, including VEGF in particular, increased.21,23 Hence, HIF-1a, as the critical mediator to hypoxia, emerged as a key molecular regulator of angiogenesis.
576
from Gene-Pharma (Shanghai, China) and the target sequence was GCAGTTCAATGAAGCTCAACT. The new plasmid was named as Nrf2i. Random sequence, TTCTCC GAACGTGTCACGT, was used as the negative control, which was named as Sc. BLAST research ensured that the sequences had no significant homology with other human genes. Cell transfection was performed in six-well plates with lentiviral particles containing either Sc RNA or Nrf2i shRNA expression plasmid using Polybrene (Gene-Pharma, Shanghai, China) according to the manufacturer’s instruction. Transfection was continued for 24 hr and followed by a 24-hr recovery in the complete medium. For the selection of cells with target plasmids, cells were grown in the medium containing 1.5 lg/ml puromycin (Sigma-Aldrich, St. Louis, MO) for up to 2 weeks. Then, the positive clonal cells were selected and cultured in 96well plates with the medium containing 1.0 lg/ml puromycin.
Cancer Cell Biology
Total RNA extraction and RT-PCR analysis
Total RNA was isolated using the TRIzol reagent (Invitrogen, CA, USA) following the manufacturer’s recommendations and subjected to DNase (Promega, Madison, WI) treatment. The concentration and purity of total RNA were determined by spectrophotometer analysis (OD260/280: 1.8–2.2) and agarose gel electrophoresis. Reverse transcriptase (RT) reactions were performed by incubating 400 ng RNA with the first-strand cDNA synthesis kit (Takara, Dalian, China). Obtained cDNA was amplified immediately using the following primers: for Nrf2, 50 -TCAGCGACGGAAAGAGTATGA-30 and 50 -CCAC TGGTTTCTGACTGGATGT-30 ; for NQO1, 50 -ATGGTCGG CAGAAGAGC-30 and 50 -GGAAATGATGGGATTGAAGT-30 ; for HO-1, 50 -TCTCCGATGGGTCCTTACACTC-30 and 50 GGCATAAAGCCCTACAGCAACT-30 ; for HIF-1a, 50 -AT CCATGTGACCATGAGGAAATG-30 and 50 -CTCGGCTAGT TAGGGTACACTT-30 ; for VEGF, 50 -GAAGGAGGAGGGC AGAAT-30 and 50 -CGATTGGATGGCAGTAGC-30 and for GAPDH, 50 -GAAATCCCATCACCATCTTC-30 and 50 -GGA CTCCACGACGTACTCA-30 . The amplification and data acquisition were carried out on a real-time PCR system (Agilent) using FastStart universal SYBR green master (Roche, Mannheim, Germany). The conditions were predenaturated at 95 C for 10 min, followed by 40 circles at 95 C for 15 sec and 60 C for 1 min. All samples were analyzed in triplicates in three independent experiments. Reaction without cDNA was used as no-template control, and no-RT controls were also set up to rule out genomic DNA contamination. Relative quantification of mRNA expression was determined using the 22䉭䉭Cq method. PCR amplification efficiency of each gene was established by means of calibration curves. The level of GAPDH was used as an internal reference gene. All the real-time PCR experiments were performed in accordance to the Minimum Information for publication of quantitative real-time PCR experiments.24
Nrf2 in GB angiogenesis
Protein concentrations were estimated by Coomassie Plus Protein Assay Reagent (Pierce, Rockford, IL). Protein was separated by 8–10% sodium dodecyl sulfate polyacrylamide gels electrophoresis using the Criterion system (Bio-Rad, Hercules, CA) and electroblotted onto a nitrocellulose membrane (Millipore, Billerica, MA). For immunoblotting, membranes were blocked with 5% skim milk for 2 hr at room temperature, and the following antibodies were used: 1:500 anti-Nrf2 (Abcam, Cambridge, MA; 68 kDa), 1:1,000 antiHO-1 (Abcam; 33 kDa), 1:1,000 anti-AKR1C1 (Abcam; 37 kDa), 1:2,000 anti-b-actin (Santa Cruz Biotechnology, Santa Cruz, CA; 43 kDa) and 1:1,000 anti-HIF-1a (Novus Biologicals, LLC, Littleton, CO; 120 kDa). Each primary antibody was diluted appropriately in blocking buffer and then incubated overnight at 4 C. Then, membranes were incubated with the appropriate secondary antibodies (Cell Signaling, Danvers, MA; 1:2,000) for 2 hr at room temperature. After washing, protein bands were visualized with Chemiluminescent HRP Substrate (Millipore) at room temperature and exposed to X-ray film (Fujifilm, Tokyo, Japan). Relative changes in protein expression were estimated from the mean pixel density using Quantity One software 4.6.2 (Bio-Rad), normalized to b-actin and presented as relative density units. Cell proliferation assay
Cell proliferation was measured by Cell-Counting kit-8 (CCK-8) (Dojindo, Kumamoto, Japan) and cell proliferation enzyme-linked immunosorbent assay (ELISA) kit (Roche) according to the manufacturer’s instructions. Cells were plated in 96-well plates at a density of 2 3 103 cells per well. After the incubation for 24 and 48 hr, 10 ll CCK-8 was added and cells were further incubated for 1 hr. Then, the absorbance was measured at 450 nm using the ELISA microplate reader (Bio-Rad). DNA synthesis of cells was measured using the ELISA kit, which is a colorimetric immunoassay based on the assessment of bromodeoxyuridine (BrdU) incorporation during DNA synthesis. Cells were plated in 96-well plates at a density of 2 3 103 cells per well and grown for 24 hr. Then, cells were labeled with 10 lM BrdU for 2 hr, and incorporated BrdU was quantified at 492 nm using the ELISA microplate reader (Bio-Rad). Measurement of total GSH contents
Cells were grown in six-well plates for 24 hr, collected and lysed with protein detergent S solvent. The total GSH was determined by commercially available Total Glutathione Assay Kit (Beyotime Institute of Biotechnology, Shanghai, China). All procedures completely complied with the manufacture’s instructions. Protein concentration was estimated by Coomassie Plus Protein Assay Reagent (Pierce). Quantification of VEGF protein levels by ELISA
Western blot analysis
To obtain total protein lysate, cells were homogenized in RIPA buffer and centrifuged at 12,000g for 15 min at 4 C.
The concentration of VEGF protein was measured using Human VEGF-A Platinum ELISA kit from eBioscience (eBioscience, San Diego, CA). Sc and Nrf2i cells in FBS-free C 2013 UICC Int. J. Cancer: 135, 574–584 (2014) V
DMEM were incubated in 24-well plates, supernatants were collected and cell number of each well was counted. VEGF levels in the supernatant (100 ll) were determined and normalized to the cell number (105). A serial dilution of human recombinant VEGF was included in each assay to obtain a standard curve.
Human umbilical vein endothelial cells (HUVECs) were purchased from Lonza (Basel, Switzerland) and maintained in DMEM (Gibco) supplemented with 10% FBS (Gibco). To assess the effect of Nrf2 on HUVECs tube formation, Sc and Nrf2i cells in supplement-free DMEM were incubated in hypoxic condition for 24 hr, and supernatants (conditioned media) were collected. Then, HUVECs were plated in 96-well plates (1.8 3 104 cells per well), which were coated by Matrigel (BD Biosciences, Bedford, MA). Then, the conditioned media or fresh DMEM containing VEGF (10 ng/ml) (Sigma) were added into cells. After 10 hr, three or more random pictures were taken from each well using a digital camera system. Image J was used for the quantification of tube lengths.
eosin (HE) staining was performed to identify the histological profiles of tumor cells. The slides were scored by two independent observers. Occupied regions by tumor cells (% over trimmed tumor mass) were counted in the HE staining using Image J. The number of positive staining cells was calculated from eight representative staining fields per each specimen under 4003 magnifications. The staining intensity was estimated and stratified as 0, no staining; 1, weak staining (light yellow); 2, moderate staining (yellow brown) and 3, strong staining (brown).26 The immunohistochemistry positive results for Nrf2 were recorded as previously described.25–27 Then, we scored the percentage of immunoreactive tumor cells as 0 (0%), 1 (>0% and 10%), 2 (>10% and 50%) and 3 (>50%). Final immunoreactivity scores (IRS) of Nrf2 expression were obtained for each specimen by multiplying the percentage and the intensity score. At last we estimated the protein expression levels by classifying IRS values as negative (based on IRS value 0 and 1) and positive (based on IRS value greater than 1), combined with as low (based on IRS values as 0, 1, 2, 3 and 4) and high (based on IRS value as 6 and 9).
Tumor xenograft study
Microvessel density
All procedures in animals were approved by the Animal Care and Use Committee of Nanjing University and conformed to Guide for the Care and Use of Laboratory Animals from National Institutes of Health. Nrf2i and Sc shRNA-stabletransfected U251 or U87 cells (5.0 3 106) were suspended in 100 ll phosphate-buffered saline (PBS) and then injected subcutaneously into right side of the anterior flank of the male BALB/c athymic nude mice (Charles River Breeding Laboratories, Wilmington, MA) at 4–5 weeks of age. Tumor growth was examined every 3 days from tumor formation for at least 5 weeks. Then, tumors were stripped and weighed before paraffin embedding. Tumor volumes were determined by external measurements and calculated according to V 5 [L 3 W2 3 0.52 (V 5 volume, L 5 length and W 5 width).
MVD of tumor tissues was assessed through the immunohistochemical analysis of endothelial marker CD31 and determined according to the method of Weidner et al.28 First, immunostained sections were initially screened at low magnification (503) to identify hot spots of the neovascularization. Within hot-spot areas, the stained microvessels were counted in a single high-power (2003) field, and the average vessel counted in three hot spots was considered as the value of MVD. Any yellow-brown stained endothelial cell or endothelial cell cluster obviously isolated from the adjacent microvessels, tumor cells and other connective tissue elements was considered a single, countable microvessel. All counts were performed by three investigators in a blinded manner. Comparisons regarding microvessel counts were drawn among the observers and discrepant results were reassessed. The consensus was used as the final score for analysis.29
HUVEC culture and tube formation assay
Immunohistochemical staining
Serial sections of 3-lm thick were dewaxed and endogenous peroxidase was quenched with 3% H2O2 in methanol for 30 min.25 Before staining, nonspecific binding was blocked by incubation with 10% bovine serum albumin (BSA) in PBS at 37 C for 1 hr. Then, all incubations with 1:50 anti-Nrf2 (Abcam), 1:50 anti-Caspase-3 (Abcam) and 1:100 anti-CD31 (Abcam) antibodies in PBS containing 1% BSA were carried out at 4 C overnight. All sections were briefly washed in PBS and incubated at room temperature with the anti-rabbit antibody and avidin-biotin peroxidase (Vector Laboratories, Burlingame, CA). Color was then developed by incubation with the diaminobenzidine solution (Dako Corporation, Carpinteria, CA). Nuclei were counterstained blue with Meyer’s hematoxylin (Sigma-Aldrich). Nonneoplastic brain tissues were used as control and nonimmune IgG was used as negative control antibody for the staining. The hematoxylin and C 2013 UICC Int. J. Cancer: 135, 574–584 (2014) V
Statistical analysis
SPSS 19.0 software (SPSS, Chicago, IL) was used for statistical analysis. Statistical data were presented as mean 6 SD. One-way analysis of variance (ANOVA) followed by Tukey post hoc comparison tests was used to compare the levels of different experimental groups. p-Values of 0.05) (Fig. 5c). To conclude, the suppression effect on the proliferation rate of Nrf2i-U251 cells was far more apparent than that of Nrf2i-U87 cells. The next step was to detect the concentration of VEGF protein in the supernatant of tumor cells incubated in hypoxia. The accumulation of VEGF increased continuously throughout the incubation in both cell lines and peaked at 24–36 hr (Figs. 5d and 5e). To avoid cells dying of hunger, assessment at 24 hr was chosen for further investigation. Then, the knockdown of Nrf2 was observed to have diminished the accumulation of VEGF in hypoxia (1% O2) more greatly in U87 cells than in U251 cells (Fig. 5f). Then, the
ELISA assay on accumulated VEGF was preformed under the CoCl2-mimicking hypoxia. The administration of CoCl2 was done according to the pre-experiment and the study by Kim et al.19 Then, cells were treated with 200 lM CoCl2 for 24 hr and VEGF level was assessed. The repressive effect on VEGF level appeared to be concordant with that of 1% O2-mimicking hypoxia (Fig. 5g). To validate the role of Nrf2 in the angiogenesis of GB we performed the endothelial tube formation assay. We incubated GB cells with supplement-free DMEM medium in hypoxia and collected the supernatants. HUVECs were then incubated with DMEM or the supernatants. Cells incubated with fresh DMEM were taken as the negative control and those incubated with fresh DMEM containing VEGF (10 ng/ ml) were used as the positive control. It was observed that the hypoxia-exposed Nrf2i medium yielded notably reduced endothelial tube formation compared to the Sc group or the VEGF treatment group (Figs. 5h–5j). These findings C 2013 UICC Int. J. Cancer: 135, 574–584 (2014) V
581
Figure 5. Diminished VEGF accumulation and tube formation in Nrf2-inhibited cells. Cell proliferation was assessed using the CCK-8 assay (a, b) and BrdU incorporation assay (c) in hypoxia. (a) The CCK-8 assay on U251 cells performed at 0, 24 and 48 hr showed that the proliferation rate of Nrf2i group was suppressed significantly at 48 hr (p 5 0.0022, ANOVA). (b) U87 cells were observed to be slightly repressed in the CCK-8 assay (p 5 0.0496, ANOVA). (c) BrdU incorporation assessment confirmed the knockdown effect on the proliferation of U251 cells (p 5 0.0212), while showed no notable differences among groups in U87 (p > 0.05). Values are means 6 SD from six wells. (d–g) The ELISA assay quantified the concentration of VEGF protein in supernatants of tumor cell culture in hypoxia. (d, e) The accumulation of VEGF increased continuously throughout incubation in hypoxia (1% O2) in both cell lines and peaked at 24–36 hr compared with cells treated in normoxia. (f) Knockdown of Nrf2 diminished the 24-h accumulation of VEGF in hypoxia. (g) The ELISA assay on VEGF under CoCl2-mimicking hypoxia coincided with the Nrf2-inhibited effects. (h–j) Angiogenesis evaluated in the endothelial tube formation assay. Transfected cells from U251 (h) and U87 (i) were incubated in hypoxia (1% O2) with supplement-free DMEM medium for 24 hr and the supernatants were collected. HUVECs were then incubated with DMEM or the supernatants, and the length of formed tubes was quantified (j). VEGF treatment group was used as the positive control and supplement-free DMEM treatment group as the negative control. Bar represents mean 6 SD; *p < 0.05, **p
