Combination of Ad-SGE-REIC and bevacizumab modulates glioma progression by suppressing invasion and angiogenesis

Reduced expression in immortalized cells/Dickkopf-3 (REIC/Dkk-3) is a tumor suppressor and its overexpression has been shown to exert anti-tumor effects as a therapeutic target gene in many human cancers. Recently, we demonstrated the anti-glioma effects of an adenoviral vector carrying REIC/Dkk-3 with the super gene expression system (Ad-SGE-REIC). Anti-vascular endothelial growth factor treatments such as bevacizumab have demonstrated convincing therapeutic advantage in patients with glioblastoma. However, bevacizumab could not improve overall survival in patients with newly diagnosed glioblastoma. In this study, we examined the effects of Ad-SGE-REIC on glioma treated with bevacizumab. Treatment of Ad-SGE-REIC resulted in significant reduced numbers of invasion cells treated with bevacizumab.. Western blot analyses revealed increased expression of several endoplasmic reticulum stress markers in cells treated with both bevacizumab and Ad-SGE-REIC and decreased β -catenin protein levels. Expressions of apoptosis markers were also increased in cells with combination therapy. In malignant glioma mouse models, overall survival was extended in the combination therapy group. These results suggest that the combination therapy of Ad-SGE-REIC and bevacizumab exerts anti-glioma effects by suppressing angiogenesis and invasion of tumors. Combined Ad-SGE-REIC and bevacizumab might indicate a promising strategy for the treatment of malignant glioma.

Glioma is the most common type of primary brain tumor, and malignant gliomas are aggressive intracranial neoplasms in humans 1,2 . The characters are proliferation, angiogenesis, and invasion. Effective chemotherapeutic or molecular-targeted agents for malignant glioma have yet to be developed. Owing to the high resistance of malignant glioma to current therapies, new therapeutic agents are needed immediately.
Several clinical trials have investigated the efficacy of molecular targeted drugs in glioma 3,4 .
Bevacizumab, an anti-vascular endothelial growth factor (VEGF) antibody, showed promising effects for the treatment of recurrent glioblastoma patients in trials [5][6][7] . However, the phase III AVAglio and RTOG 0825 studies did not show a statistically significant prolongation in overall survival in glioblastoma patients treated with bevacizumab 8,9 . In addition, Piao et al. found that anti-VEGF therapy made gliomas more resistant, invasive and aggressive and enhanced epithelial-to-mesenchymal transition 10 . In our previous research, we demonstrated that increased glioma cell invasion due to anti-VEGF therapy was associated with extracellular matrix changes, and an integrin inhibitor (cilengitide) reduced anti-VEGF therapy-induced glioma cell invasion 11 . We also showed that fibroblast growth factor 13 and delta-catenin regulate glioma cell invasion and are important for bevacizumab-induced glioma invasion 12,13 .
Expression of the gene encoding reduced expression in immortalized cells/Dickkopf-3 (REIC/Dkk-3) was shown to be decreased in various human tumors [14][15][16][17]  In this study, we evaluated the combination treatment of Ad-SGE-REIC and bevacizumab on glioma. Bevacizumab suppressed tumor angiogenesis, and Ad-SGE-REIC reduced bevacizumab-induced glioma invasion. We also observed decreased activation of the β-catenin signaling pathways in glioma cells treated with combined Ad-SGE-REIC and bevacizumab.

Anti-angiogenic effect of bevacizumab in combination treatment
To investigate the anti-angiogenic effect of bevacizumab, tube formation assays were performed using HUVECs. When HUVECs were cultured in the presence of VEGF, efficient tube formation was observed (Figure 2A, B). Treatment with bevacizumab efficiently inhibited tube formation compared with controls. In contrast, Ad-SGE-REIC conditioned medium (CM) increased the tube length in HUVECs. Notably, co-treatment of bevacizumab with Ad-SGE-REIC (CM) significantly decreased tube length compared with HUVECs only treated with Ad-SGE-REIC (CM) (Ad-SGE-REIC and bevacizumab combination: 8.4 ×10 3 ± 2.1 × 10 2 pixels vs.

Ad-SGE-REIC-infected glioma cells showed inhibited invasion activity in vitro
We next examined the effects of bevacizumab and Ad-SGE-REIC on glioma cell invasion. The CM of malignant glioma cells infected by Ad-SGE-REIC was centrifuged and filtrated to eliminate virus and cell debris. The expression of REIC was detected by western blotting in Ad-SGE-REIC-infected glioma cells but not in Ad-SGE-REIC-infected glioma cells. Glioma cells treated with bevacizumab and/or Ad-SGE-REIC were seeded into upper chambers, and then cells invading through the membrane were counted after 24 h ( Figure 2C). REIC significantly reduced the number of invading cells of U87ΔEGFR, U251MG, and U87MG cell lines compared with controls ( Figure 2D). Furthermore, the numbers of invading cells treated with bevacizumab were reduced by co-treatment with REIC.

ER stress and β-catenin degradation by combination of REIC/Dkk-3 and bevacizumab in glioma cells
Western blot analysis revealed increased expressions of ER stress marker molecules Bip and phosphorylated IRE1α in U87ΔEGFR, U251 and MGG23 cells treated with the combination treatment compared with cells treated with either Ad-SGE-REIC or bevacizumab and control cells (Fig. 3A).
The Wnt signaling pathway regulates cell invasion by inhibition of proteasome-dependent proteolysis of β-catenin. Therefore, we evaluated the impact of the Ad-SGE-REIC and bevacizumab combination treatment on β-catenin expression in glioma cells (Fig. 3B). The results showed that β-catenin protein levels were more potently reduced by the combination treatment compared with levels in cells treated with Ad-SGE-REIC or bevacizumab as well as controls.

Anti-tumor effect of combination therapy with bevacizumab and Ad-SGE-REIC in xenograft mice
We evaluated the potential antitumor effect of the combination therapy of bevacizumab and Ad-SGE-REIC on mice harboring intracerebral U87ΔEGFR glioma cells (Fig. 4A). We compared mice bearing U87ΔEGFR glioma cells treated with saline, bevacizumab at 10 mg/kg, Ad-SGE-REIC at 3.6×10 7 pfu, and bevacizumab at 10 mg/kg, and Ad-SGE-REIC at 3.6×10 7 pfu and bevacizumab at 10 mg/kg. Control mice treated with PBS had a median survival of 14 days after tumor cell implantation, and mice treated with Ad-SGE-REIC similarly had a median survival of 14 days after tumor cell implantation (Fig. 4B). Mice treated with bevacizumab had a median survival of 19 days after tumor cell implantation. However, mice treated with the bevacizumab and Ad-SGE-REIC combination had a median survival of 29 days, which was significantly longer than mice treated with PBS, Ad-SGE-REIC alone, or bevacizumab alone (log-rank test: p=0.004, p=0.005, and p=0.003, respectively).
We next evaluated the potential antitumor effect with single cell invasive model (supplementary data). We compared mice bearing MGG23 glioma cells treated with PBS, bevacizumab at 10 mg/kg, Ad-SGE-REIC at 3.6×10 7 pfu, and bevacizumab at 10 mg/kg, and Ad-SGE-REIC at 3.6×10 7 pfu and bevacizumab at 10 mg/kg. Control mice treated with PBS had a median survival of 70 days after tumor cell implantation, and mice treated with bevacizumab similarly had a median survival of 75 days after tumor cell implantation (Fig. 4B). Mice treated with Ad-SGE-REIC prolonged a median survival of 88 days after tumor cell implantation.
However, mice treated with the bevacizumab and Ad-SGE-REIC combination had a median survival of 89 days, which was not significantly longer than mice treated with Ad-SGE-REIC alone.

Effect of combination Ad-SGE-REIC with bevacizumab on angiogenesis in vivo
Athymic mice harboring U87ΔEGFR cell-derived brain tumors were sacrificed at 18 days after tumor implantation and immunohistochemical staining using anti-human CD31 was performed ( Fig. 5A). Treatment with bevacizumab significantly decreased vessel density compared with controls (p<0.005) (Fig. 5B). In the group treated with Ad-SGE-REIC, vessel density was significantly increased compared with controls. However, treatment of bevacizumab combined with Ad-SGE-REIC significantly decreased the vessel density compared with levels in Ad-SGE-REIC-treated mice (p<0.005).

Effect of Ad-SGE-REIC on bevacizumab-induced invasion in vivo
In another experiment, immunodeficient mice harboring U87ΔEGFR glioma cells were sacrificed at 18 days after tumor implantation, and immunohistochemical staining with antihuman leukocyte antigen was performed. U87ΔEGFR cells treated with bevacizumab alone showed greater invasion to the ipsilateral cerebral cortex adjacent to the injection site compared with saline controls or the combined bevacizumab and Ad-SGE-REIC treated group ( Figure   5C). Invasion activity was assessed by the number of tumor cells in the ipsilateral cerebral cortex. We observed a significant increase of glioma cells invading into the cerebral cortex in the bevacizumab-treated U87ΔEGFR cell group compared with saline controls (p<0.05).
However, combination therapy with bevacizumab and Ad-SGE-REIC significantly decreased the depth of glioma invasion induced by bevacizumab (ipsilateral cortex: p<0.05, Figure 5D).
These results demonstrated that Ad-SGE-REIC reduced bevacizumab-induced invasion of glioma cells.

Microarray analysis of the effect of combination treatment on the U87ΔEGFR orthotopic mouse model
To more closely examine the tumor microenvironment response to the combination therapy, we analyzed the changes in gene expression in tumor tissues from the U87ΔEGFR orthotopic mouse model treated with Ad-SGE-REIC and bevacizumab combination therapy compared with tissues from mice treated with Ad-SGE-REIC monotherapy. We identified 937 upregulated genes and 2565 downregulated genes in Ad-SGE-REIC-treated U87ΔEGFR glioma brain tissue compared with control glioma tissue. We also identified 934 upregulated genes and 397 downregulated genes in the combination treatment group compared with the Ad-SGE-REIC-treated U87ΔEGFR glioma tissue.
We then characterized the functional significance of these dysregulated genes using pathway analysis. For the upregulated genes in Ad-SGE-REIC-treated U87ΔEGFR glioma brain tissue compared with controls, 14 significantly enriched pathways were identified, including 'TNF-alpha NF-kB Signaling Pathway,' 'Wnt Signaling Pathway NetPathand,' and 'IL-2 Signaling Pathway' (Table 1A). For the downregulated genes in Ad-SGE-REIC-treated U87ΔEGFR glioma brain tissue compared with controls, 24 significantly enriched pathways were identified, such as 'Notch Signaling Pathway,' 'TNF-alpha NF-kB Signaling Pathway,' and 'Mitochondrial Gene Expression' (Table 1B). For the upregulated genes in the combination treated tissue compared with Ad-SGE-REIC-treated U87ΔEGFR glioma tissue, 12 significantly enriched pathways were identified, such as 'TNF-alpha NF-kB Signaling Pathway' and 'Delta-Notch Signaling Pathway' (Table 2A). For the downregulated genes in the combination treated tissue compared with Ad-SGE-REIC-treated U87ΔEGFR glioma tissue, 7 significantly enriched pathways were identified, including 'Robo4 and VEGF Signaling Pathways Crosstalk' and 'TNF-alpha NF-kB Signaling Pathway' (Table 2B).

DISCUSSION
Our results indicate that combination therapy of bevacizumab with Ad-SGE-REIC had additional therapeutic effects on glioma cells compared with monotherapy using bevacizumab or Ad-SGE-REIC. After treatment with Ad-SGE-REIC and bevacizumab, the number of surviving malignant glioma cells was significantly decreased in a time-dependent manner. Western blot analyses also showed increased expression of several ER stress markers in cells treated with both bevacizumab and Ad-SGE-REIC. Additionally, β-catenin protein levels were potently decreased by combination therapy. In malignant glioma mouse models, overall survival was extended in the combination therapy group. Our results showed that invasive activity increased by bevacizumab counteracted the effectiveness of bevacizumab. However, our experiments using two different mouse glioma models indicated that Ad-SGE-REIC inhibited glioma cell invasion induced by bevacizumab, resulting in a synergistic effect.

Ad-SGE-REIC
The Ad-SGE-REIC adenovirus vector was developed to increase REIC/Dkk-3 expression and showed enhanced therapeutic effects. We previously showed that Ad-SGE-REIC exhibited timedependent and significant effects on reducing the number of viable malignant glioma cells in cytotoxicity assays 18 . Xenograft and syngeneic mouse intracranial glioma models treated with Ad-SGE-REIC had significantly longer survival than those treated with the control Ad-LacZ vector or Ad-CAG-REIC 18 . Moreover, MGG 23 human primary culture xenograft mouse intracranial glioma models treated with Ad-SGE-REIC had significantly longer survival than controls.

Combination effect of Ad-SGE-REIC with bevacizumab
In our previous study, we observed a decreased number of vessels in the tumor xenograft model treated with bevacizumab 11 , and we demonstrated prolonged survival of mouse treated with bevacizumab. The VEGF autocrine signaling loop is suppressed, the Akt and Erk pathways are activated, and tumor growth and invasion are stimulated by anti-VEGF therapy 19 . Molecules within the extracellular matrix microenvironment such as proteoglycans and collagens may influence tumor invasion during anti-VEGF therapy 20 . Our results showed that REIC could inhibit glioma invasion induced by bevacizumab in vitro and in vivo.
ER stress markers and β-catenin in the nucleus were downregulated in response to the combination treatment. These results correlated with anti-invasive effects, which were associated with IRE1α endoribonuclease activity 21,22 . In microarray data, the significantly enriched pathways including 'TNF-alpha NF-kB Signaling Pathway' were identified in the combination treatment sample compared with Ad-SGE-REIC alone sample. 'TNF-alpha NF-kB Signaling Pathway' is upstream of ER stress 23 . We need investigate more information for this upstream pathway.
The Wnt/β-catenin signaling plays an essential role in cellular proliferation, migration, invasion, and angiogenesis, therefore contributing to glioma progression 24 . Notch1 promotes glioma cell migration and invasion by stimulating β-catenin 25 . In microarray data, the significantly enriched pathways including the 'Notch Signaling Pathway' and 'Delta-Notch Signaling Pathway' were identified in the combination treatment sample compared with Ad-SGE-REIC alone sample. The Notch signaling pathway is an evolutionarily conserved intercellular signaling mechanism essential for proper embryonic development in all metazoan organisms in the Animal kingdom (https://www.wikipathways.org/index.php/Pathway:WP29).

Anti-angiogenic effect of bevacizumab on angiogenesis after REIC treatment
REIC was reported to be induced angiogenesis. Untergasser reported that Dkk-3 is expressed in tumor endothelial cells and supports capillary formation 26

Future directions
Ad-REIC is being evaluated in clinical studies. The first in-human, phase I/IIa study of in situ Ad-REIC gene therapy for prostate cancer and a phase I/II clinical trial of Ad-SGE-REIC for malignant mesothelioma was performed in Japan 28,29 . Recently, the safety and efficacy of Ad-SGE-REIC to liver tumors in patients are evaluated in a Phase I/Ib study 30

Cell lines, drugs, and adenovirus vector
The human glioma cell lines U87ΔEGFR, U87MG, U251MG, A172, and LNZ308 were prepared and/or maintained as described previously 33 . The human glioblastoma-derived cell lines MGG8, MGG18, and MGG23 were provided by Dr. Hiroaki Wakimoto at Massachusetts General Hospital and cultured as previously described 34,35 . The human GBM cell lines were The cells were infected with Ad-SGE-REIC at an MOI of 30 or treated with 1 mg/mL bevacizumab. Spheres that formed after 10 days were counted.

Western blot analysis
Total cell protein was extracted from cells in ice-cold lysis buffer (20 mM Tris pH 7.5, 150 mM NaCl, 1 mM EDTA, 1 mM EGTA, 1.0% Triton X-100, one tablet per 10cc buffer of PhosSTOP (Roche Applied Science, Mannheim, Germany), and protease inhibitor cocktail) and quantified using the Bradford method 38 . Equal amounts of protein (20 µg) were separated by SDS-PAGE.
Then, we performed Western blotting as described previously 13 with conditioned medium obtained by cultivating U251 at cytotoxicity assay. After 24 h, the cells were viewed under a microscope (BZ-8000; Keyence, Osaka, Japan). The total tube length in each well was calculated under high magnification (x20), and the average length of four wells in each treatment group was taken as the tube formation per view field.

Transwell migration assay
In vitro migration assays were performed using a 24-well plate and ThinCert (8 μm-pore, 24well format, Greiner Bio-one; Kremsmunster, Austria) according to the manufacturer's instructions. We prepared U87ΔEGFR cells as described previously 11

Ethics and animal use statement
This study was conducted in strict accordance to the recommendations in the Guide for the Care and Use of Laboratory Animals in Japan. All procedures and animal protocols were in accordance with arrive guidelines and approved by the Committee on the Ethics of Animal Experimentation at Okayama University (Permit No. OKU-2016554).

Brain xenografts
Athymic mice (BALB/c-nu/nu) were obtained from CLEA Japan, Inc. (Tokyo, Japan). We prepared 2.0 × 10 5 U87ΔEGFR or, MGG23 cells/μL as described previously 11 . They were anesthetized with an intraperitoneal (ip) injection of ketamine (2.7-3.0mg/100g) and xylazine (0.36-4.0mg/100g). Cells (2 µL) were injected into the right frontal lobe of mice (3 mm lateral and 1 mm anterior to the bregma at a depth of 3 mm), as described previously. PBS or bevacizumab (10 mg/kg) was intraperitoneally administered two times per week, starting on day 5 after tumor cell implantation. At 7 days after tumor inoculation, all mice bearing brain tumors were reanesthetized and stereotactically injected with Ad-SGE-REIC or PBS at the tumor inoculation site using the same coordinates. We assessed the survival time of the U87ΔEGFR mouse glioma model using Kaplan-Meier survival analysis.
Animals were sacrificed at 28 days after tumor implantation, following six administrations of PBS or bevacizumab. The maximum transverse diameter of tumors was measured. Hematoxylin and eosin staining was performed as described previously 40 .

Immunohistochemistry
Immunohistochemistry was performed using the avidin-biotin-peroxidase complex method (Ultrasensitive; MaiXin, Fuzhou, China). CD31 mouse monoclonal antibody (1:300 dilution; Abcam, Inc., Cambridge, UK) and mouse immunoglobulin (a negative control) were used as previously described 34 . Hematoxylin was used for counterstaining. We evaluated the positivity of cytoplasmic immunostaining in tumor cells.

Microarray analysis
Orthotopic U87ΔEGFR xenograft mouse models treated with control, Ad-SGE-REIC or the combination of bevacizumab and Ad-SGE-REIC were killed at 18 days after tumor implantation Inc.). The data were extracted using the following criteria: Z score > 0 and p value < 0.05 11,42 .

Statistical analysis
The Student's t-test, Mann-Whitney U test, and ANOVA were used to test for statistical significance. Data are presented as the mean ± standard deviation and standard error.
Differences were considered to denote statistical significance when p<0.05. All statistical analyses were performed using SPSS statistical software, version 20 (SPSS, Inc., Chicago, IL, USA).    Athymic nude mice bearing intracranial U87ΔEGFR gliomas were treated with 3.6×10 7 pfu Ad-SGE-REIC, and bevacizumab was administered intraperitoneally at 10 µg/g. Statistical significance was calculated by the log-rank test.