A ternary-complex of a suicide gene, a RAGE-binding peptide, and polyethylenimine as a gene delivery system with anti-tumor and anti-angiogenic dual effects in glioblastoma

Eunji Choi, Jungju Oh, Dahee Lee, Jaewon Lee, Xiaonan Tan, Minkyung Kim, Gyeungyun Kim, Chunxian Piao, Minhyung Lee

PII: S0168-3659(18)30205-0
DOI: doi:10.1016/j.jconrel.2018.04.021
Reference: COREL 9248
To appear in: Journal of Controlled Release
Received date: 19 September 2017
Revised date: 13 March 2018
Accepted date: 11 April 2018

Please cite this article as: Eunji Choi, Jungju Oh, Dahee Lee, Jaewon Lee, Xiaonan Tan, Minkyung Kim, Gyeungyun Kim, Chunxian Piao, Minhyung Lee , A ternary-complex of a suicide gene, a RAGE-binding peptide, and polyethylenimine as a gene delivery system with anti-tumor and anti-angiogenic dual effects in glioblastoma. The address for the corresponding author was captured as affiliation for all authors. Please check if appropriate. Corel(2018), doi:10.1016/j.jconrel.2018.04.021

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

A ternary-complex of a suicide gene, a RAGE-binding peptide, and polyethylenimine as a gene delivery system with anti-tumor and anti- angiogenic dual effects in glioblastoma

Eunji Choi, Jungju Oh, Dahee Lee, Jaewon Lee, Xiaonan Tan, Minkyung Kim, Gyeungyun Kim, Chunxian Piao, and Minhyung Lee*
Department of Bioengineering, College of Engineering, Hanyang University, 222 Wangsimni-ro, Seongdong-gu, Seoul 04763, Korea

*Corresponding author: Minhyung Lee, Ph.D.
E-mail: [email protected], Tel: 82-2-2220-0484, Fax: 82-2-2220-4454

Abstract
The receptor for advanced glycation end-products (RAGE) is involved in tumor angiogenesis. Inhibition of RAGE might be an effective anti-angiogenic therapy for cancer. In this study, a cationic RAGE-binding peptide (RBP) was produced as an antagonist of RAGE, and a ternary-complex consisting of RBP, polyethylenimine (2 kDa, PEI2k), and a suicide gene (pHSVtk) was developed as a gene delivery system with dual functions: the anti-tumor effect of pHSVtk and anti-angiogenic effect of RBP. As an antagonist of RAGE, RBP decreased the secretion of vascular-endothelial growth factor (VEGF) in activated macrophages and reduced the tube- formation of endothelial cells in vitro. In in vitro transfection assays, the RBP/PEI2k/plasmid DNA (pDNA) ternary-complex had higher transfection efficiency than the PEI2k/pDNA binary-complex. In an intracranial glioblastoma animal model, the RBP/PEI2k/pHSVtk ternary-complex reduced a-smooth muscle actin expression, suggesting that the complex has an anti-angiogenic effect. In addition, the ternary-complex had higher pHSVtk delivery efficiency than the PEI2k/pHSVtk and PEI25k/pHSVtk binary-complexes in an animal model. As a result, the ternary-complex induced apoptosis and reduced tumor volume more effectively than the PEI2k/pHSVtk and PEI25k/pHSVtk binary-complexes. In conclusion, due to its dual anti- tumor and anti-angiogenesis effects, the RBP/PEI2k/pHSVtk ternary-complex might be an efficient gene delivery system for the treatment of glioblastoma.

Keywords : anti-angiogenesis; gene delivery; glioblastoma; receptor for advanced glycation end- products; thymidine kinase

1. Introduction
Glioblastoma is the most common primary brain tumor. Despite progress in clinical techniques, the average lifespan of patients with glioblastoma is less than 15 months [1]. Currently, the main clinical options for the treatment of glioblastoma are surgical resection, radiotherapy, and chemotherapy [2]. Due to the critical function of the brain, aggressive surgical resection is not possible, so it is difficult to completely remove tumors by surgery. In addition, radiotherapy and chemotherapy are not effective enough to destroy all tumor cells in the brain. Thus, recurrence of glioblastoma is common. In addition, due to the aggressive nature of glioblastoma, tumor cells can migrate great distances in the brain from the original site [3]. Because of the problems with current clinical options, new treatment paradigms for glioblastoma should be developed.
The receptor for advanced glycation end-products (RAGEs) is a multi- ligand receptor of the immunoglobulin superfamily. RAGE is expressed at a high level in various tumors including glioblastoma. RAGE has various ligands, most of which are involved in the immune response. The list of RAGE ligands includes advanced glycation end-products (AGEs), S100, high mobility group box-1 (HMGB1), and amyloid-b peptides [4]. These ligands are classified as damage- associated molecular patterns (DAMPs) and pathogen-associated molecular patterns (PAMPs) and are closely related to the disease state, suggesting that RAGE is involved in the pathogenesis of various diseases [5, 6]. Especially in tumors, overexpression of RAGE stimulates tumor proliferation and migration [7]. Binding of ligands to RAGE triggers activation of cell signaling pathways such as the mitogen-activated protein (MAP) kinase pathway. Activation of RAGE eventually activates nuclear factor-kB (NF-kB) and enhances the transcription of cytokines and growth factors, including vascular endothelial growth factor (VEGF), tumor necrosis factor α (TNF-α), and interleukins (ILs) [8, 9].

Since angiogenesis plays an important role in tumor growth and metastasis, inhibition of tumor angiogenesis is an effective approach to suppress tumors [10]. Several different cytokines can influence the cellular mechanisms of angiogenesis. Cytokines such as VEGF, TNF-a, and ILs are released by various cells within solid tumors [11, 12]. Furthermore, cytokine receptors are expressed on tumor cells, and binding of cytokines to these receptors has an influence on endothelial cells and immune cells [13]. Because cytokines increase the rate of proliferation, migration, and tube formation, inhibition of cytokines can have negative effects on tumor angiogenesis. Activation of RAGE facilitates secretion of these cytokines through NF-κB activation [6].
This suggests that inhibition of ligand binding to RAGE can have therapeutic effects in cancer therapy by inhibiting cytokine secretion and tumor angiogenesis [14]. For this purpose, various RAGE inhibiting reagents have been developed. One example is the anti- S100A4 monoclonal antibody that has been used to scavenge S100A4. S100A4 is a member of the S100 calcium-binding protein family that facilitates tumorigenesis by stimulating the RAGE- mediated signal transduction pathway [15]. The anti-S100A4 antibody inhibited binding of S100A4 to RAGE and reduced its signal transduction, which effectively inhibited endothelial cell migration and blood vessel formation [15]. Similarly, an anti- RAGE antibody was used to inhibit RAGE-mediated signal transduction and subsequent angiogenesis [16]. HMGB1 is also an important ligand for RAGE and is reported to induce tumor angiogenesis [17-19]. The A-box of HMGB1 (HMGB1A) has been reported as an effective antagonist against RAGE [20, 21]. Therefore, HMGB1A can be a useful peptide for inhibition of RAGE- mediated tumor angiogenesis [22].

Gene therapy with suicide genes such as the herpes-simplex virus thymidine kinase
(HSVtk) gene has been used as a promising treatment of glioblastoma. Cerepro, an adenoviral vector containing the HSVtk gene, was developed for the treatment of glioblastoma [23]. The HSVtk gene expresses the HSVtk protein, which phosphorylates a non-toxic prodrug, ganciclovir (GCV), resulting in toxic GCV-triphosphates. Therefore, HSVtk transfected cells are under the influence of drug therapy. In addition, toxic GCV-triphosphates are transferred into neighboring cells through gap junctions, which is referred to as the bystander effect. Therefore, the HSVtk gene-GCV system has a strong anti-tumor effect even in cells that are not transfected with the HSVtk gene. Although the HSVtk gene-GCV system is effective for the treatment of glioblastoma, it cannot eradicate all tumor cells, suggesting that recurrence can occur. Clinical trials have shown that HSVtk gene therapy does not increase the overall survival of patients, and the adenoviral system can have severe adverse effects [23]. To overcome this weakness, combination therapies with genes and drugs have been developed for glioblastoma gene therapy. For example, both curcumin and the HSVtk gene were delivered to glioblastoma animal models using a micelle-based delivery system [24]. This combined delivery of curcumin and the HSVtk gene had higher therapeutic effects than either treatment alone. Similarly, combination therapy with VEGF small interfering RNA (siRNA) and doxorubicin was reported with a micelle-based delivery system [25]. However, combination therapy with a peptide and gene has not been reported.

HMGB1 is a cytokine that interacts with RAGE or toll- like receptors (TLRs). The RAGE-binding domain of HMGB1 is located at the C-terminal region of the B-box (Fig. 1A). We hypothesized that the RAGE-binding domain of HMGB1 acts as an antagonist of RAGE by interfering with its interaction with ligands when produced as an independent peptide. In this study, a RAGE-binding peptide (RBP) was designed and produced as an antagonist of RAGE based on the RAGE-binding domain of HMGB1. The RAGE-binding domain of HMGB1 has a net positive charge, suggesting that RBP can form a complex with pDNA by electrostatic interactions. RBP was expressed in bacteria and purified by affinity chromatography. The ternary- complex was developed by mixing RBP, polyethylenimine (2 kDa, PEI2k), and a suicide gene (pHSVtk). The RBP/PEI2k/pHSVtk ternary-complex was physically characterized and evaluated as a delivery system with dual functions in vitro and in vivo based on the anti-tumor effect of pHSVtk and anti-angiogenesis effect of RBP. These results suggest that the ternary-complexes can be useful for the development of an effective treatment for glioblastoma with dual anti- tumor and anti-angiogenesis therapeutic effects.

2. Materials and Methods
2.1 Materials

XhoI and NheI were purchased from Takara Bio (Shiga, Japan). pET21a was purchased from Novagen (Madison, WI). Calcein AM and isopropyl-β-D-thiogalactopyranoside (IPTG) were purchased from Sigma-Aldrich (St. Louis, MO). Phenylmethyl sulfonyl fluoride (PMSF) was purchased from AMRESCO (Solon, OH). Ni-NTA agarose was purchased from Qiagen (Valencia, CA). Spectra/Por dialysis membranes (MWCO 2,000) were purchased from Millipore (Billerica, CA). The bicinchoninic acid (BCA) assay kit, polyethylenimine (PEI, 2 kDa and 25 kDa), heparin, and 3-[4,5-dimethylthiazol-2-yl]-2,5-diphenyltetrazoliumbromide (MTT) were purchased from Pierce (Waltham, MA). C6 rat glioblastoma cells and Raw264.7 mouse macrophage cells were obtained from the Korean Cell Line Bank (Seoul, Korea).

Dulbecco’s Modified Eagle Medium (DMEM), fetal bovine serum (FBS), and trypsin- EDTA were purchased from GIBCO (Gaithersburg, MD). Human umbilical vascular endothelial cells (HUVECs), EGM-2 medium, and growth factors were purchased from Lonza (Rochester, NY). Matrigel basement membrane matrix was purchased from BD Bioscience (Franklin Lakes, NJ). Annexin V and the BD FACS Calibur™ were obtained from BD Biosciences Immunocytometry Systems (San Jose, CA). The luciferase assay kit, reporter lysis buffer, DNase I, and the DeadEndTM Colorimetric TUNEL System were purchased from Promega (Madison, WI). Lipofectamine was purchased from Invitrogen (Carlsbad, CA). Ganciclovir (GCV) was purchased from InvivoGen (San Diego, CA). Human S100B protein was purchased from Sino Biological Inc (Beijing, China). The VEGF ELISA kit was purchased from Peprotech (Rocky Hill, NJ). The TNF-a ELISA kit was purchased from eBioscience (San Diego, CA). Label IT nucleic acid labeling kits were obtained from Mirus (Madison, WI). Hamilton syringes (26-gauge) were purchased from Hamilton (Reno, NV). Anti- HSVtk antibody, anti-a-smooth muscle actin antibody, anti-EGFP antibody and horseradish peroxidase-conjugated anti-rabbit secondary antibody were obtained from Abcam (Cambridge, MA). Rabbit anti-His tag antibody and immunohistochemistry accessory kits and were purchased from Bethyl (Montgomery, AL). StaySafe nucleic acid reagent was purchased from Real Biotech (Banqiao, Taiwan). MagListo Protein G kit was obtained from Bioneer (Daejeon, Korea). Flamma 648 Sulfo-NHS ester was purchased from BioActs (Incheon, Korea). OCT Compound was purchased from Sakura Finetek (Torrance, CA).

2.2 Construction of the RBP expression vector
The HMGB1 cDNA was previously amplified by RT-PCR using total RNA from Human Embryonic K idney (HEK) 293 cells [26]. The cDNA fragment encoding amino acids 150-186 of human HMGB1 was amplified by PCR using the following primers: forward 5′- GTATGCTAGCAAGCTGAAGGAAAAATACGAAAAGG-3′ (NheI site is underlined), reverse 5′-ACCGCTCGAGACATTCCTTCTTTTTCTTGCTTTTTTC-3′ (XhoI site is underlined). The amplified cDNA was digested with NheI and XhoI and purified by agarose gel electrophoresis. The cDNA fragment was inserted into pEI21a at the NheI and XhoI sites, producing a plasmid encoding RBP (pET21a-RBP).

2.3 Expression and purification of recombinant RBP
pET21a-RBP was transformed into the E. coli BL21 (λDE3) strain. The bacteria were cultured overnight in 4 L Luria-Bertani (LB) medium containing 50 mg/ml ampicillin in a shaking incubator at 37°C and 250 rpm. RBP over-expression was induced by the addition of IPTG to a final concentration of 500 mM at an OD600 of 0.6. The bacteria were cultured for an additional 6 h in a shaking incubator at 37°C and 250 rpm. The grown bacteria were harvested by centrifugation at 6,000 × g for 15 min at 4°C. The bacterial pellets were resuspended in 50 mM NaH2PO4 (pH 8.0) containing 300 mM NaCl, 10 mM imidazole, and 1 mM PMSF. The bacterial suspension was subjected to sonication (8 × 25 sec bursts with 5 min of cooling on ice between bursts; Cell Disruptor 200, Branson Ultrasonics, Danbury, CT) for cell lysis. The lysates were subjected to centrifugation at 10,000 × g to remove cell debris. The supernatant was used for purification of RBP.

The presence of the C-terminal 6- histidine stretch of RBP allows the peptides to have high affinity for nickel ions. RBP was purified by nickel-chelate affinity chromatography. The bacterial lysate was loaded into a nickel column pre-equilibrated with an equilibration buffer containing 50 mM NaH2PO4 (pH 8.0), 300 mM NaCl, and 20 mM imidazole. After unbound proteins were removed by washing with equilibration buffer, the peptides were eluted by a step gradient of imidazole (50, 100, 150, 200, and 250 mM) in 50 mM NaH2PO4 (pH 8.0) containing300 mM NaCl at a rate of 0.5 ml/min. Each eluted fraction was assayed with the BCA assay kit and analyzed by electrophoresis in 16% SDS-PAGE. The purified protein fractions were pooled and dialyzed against distilled water using a membrane with a MW cut-off of 2,000 Da at 4°C overnight and stored at -80°C until use.

2.4 TNF-a ELISA
Raw264.7 mouse macrophage cells were cultured in Dulbecco’s Modified Eagle Medium (DMEM) containing 10% fetal bovine serum (FBS) and 1% penicillin-streptomycin (PS). Raw264.7 cells were seeded at 2×104 cells/well in 24-well plates 24 h before treatment. The cell culture medium was replaced with serum- free medium; 1.6 mg of S100B protein, RBP, and BSA were added; and the cells were incubated at 37°C for 4 h. After the incubation, the cell culture medium was replaced with fresh serum- free medium, and the cells were incubated for an additional 20 h. To measure the TNF-α level, the medium was harvested from the incubated cells, and the cytokine level was measured by ELISA.

2.5 In vitro anti-angiogenesis assay
C6 rat glioblastoma cells were cultured in DMEM containing 10% FBS and 1% PS. HUVECs were cultured in EGM-2 medium with growth factor supplements. The cells were maintained at 37°C in a humidified incubator under 5% CO2 and 95% air.
To evaluate the anti-angiogenic effect of RBP, the C6 cells were seeded in 12-well plates at a density of 1´105 cells/well 24 h before the RBP treatment. Prior to the RBP treatment, the medium was replaced with serum- free DMEM. The cells were incubated with 1 mg wtHMGB1 for 4 h. Then, various amounts of RBP were added to the cells, and they were incubated for an additional 4 h. After the incubation, the medium was replaced with fresh serum- free EGM-2 medium, and the cells were incubated for an additional 20 h. After the final incubation, the medium was harvested and used for VEGF ELISA and incubation with HUVECs.

The VEGF ELISA was performed according to the manufacturer ’s manual. For the tube formation assay, HUVECs were seeded in a 96-well plate coated with 100 ml of a Matrigel matrix and incubated in serum- free EGM-2 medium for 3 h. Then, the medium was replaced with the EGM-2 medium from the C6 cells, and the cells were incubated for an additional 4 h. Tube formation was evaluated by staining with calcein AM and imaging by microscopy. The obtained images were analyzed using the “Angiogenesis Analyzer” tool, programmed in ImageJ’s macro language. The total tube area was quantified as the mean pixel density obtained from image analysis using ImageJ software.

2.6 Plasmid DNA (pDNA) and complex formation
Two types of pDNAs were used for evaluation of gene delivery. pb-Luc and pHSVtk were previously constructed for use as a reporter gene and a therapeutic gene, respectively [27, 28]. The pDNAs were transformed to E. coli DH5a, and the bacteria were cultured overnight in LB medium containing 50 mg/ml ampicillin in a shaking incubator at 37°C and 250 rpm. The pDNAs were then extracted with the Qiagen Maxiprep kit according to the manufacturer’s manual.
For the RBP/pDNA complex, pDNA was mixed with various amounts of RBP. For the RBP/PEI2k/pDNA ternary-complex, pDNA was mixed with various amounts of RBP and PEI2k. The mixtures were incubated for 30 min at room temperature for formation of the complexes. Thecomplex formation was confirmed by gel retardation assays.

2.7 Gel retardation assays and heparin competition assays
Gel retardation assays were performed as described previously [29]. pb-Luc was used as pDNA for the assays. The amount of pb-Luc was fixed at 0.5 mg. pb-Luc was mixed with increasing amounts of RBP and PEI2k, and the mixtures were incubated for 30 min at room temperature, then analyzed by electrophoresis on a 1% agarose gel. pDNAs were visualized by a UV transilluminator.
The RBP/pb-Luc and RBP/PEI2k/pb-Luc complexes were prepared at optimal weight ratios for transfection. Increasing amounts of heparin were added to the complexes, and the mixtures were incubated for an additional 30 min at room temperature. The mixtures were analyzed on a 1% agarose gel.

2.8 DNase I protection assay
The RBP/PEI2k/pb-Luc, PEI2k/pb-Luc, and PEI25k/pb-Luc complexes were prepared with 10 mg of pDNA in 500 ml of PBS at their optimal ratio for transfection. Naked pDNA was used as a control. After complex formation, 10 units of DNase I were added to the complex solutions. The reaction mixtures were incubated at 37°C. One hundred microliters of the samples were collected at 30 or 60 min after incubation and mixed with 100 ml of a 2× stop solution (80 mM EDTA and 2% SDS). The pDNA was analyzed by 1% agarose gel electrophoresis.

2.9 Zeta-potential and complex size
The RBP/PEI2k/pb-Luc, PEI2k/pb-Luc, and PEI25k/pb-Luc complexes were prepared using the optimal conditions for transfection. The particle sizes and zeta potentials of the complexes were measured by the Zetasizer Nano ZS system (Malvern Instruments, UK).

2.10 Confirmation of ternary-complex formation: Pull-down assay
The ternary-complex of pDNA, RBP, and PEI2k was confirmed by immunoprecipitation using MagListo Protein G K it. Protein G magnetic beads and rabbit anti-His tag antibody were incubated for 10 min at room temperature. The antibody-conjugated magnetic beads were washed twice and the supernatant was removed by using a magnet. PEI2k was fluorescence-labeled with Flamma 648 Sulfo-NHS ester, according to manufacturer’s manual. The RBP/PEI2k/pb-Luc and PEI2k/pb-Luc complexes were prepared at optimal weight ratios for transfection. The complexes were mixed with the magnetic beads and the mixtures were incubated for 1 h at room temperature. The complexes and beads were washed twice and the supernatant was removed by using a magnet. The elution buffer was added and incubated at 70°C for 10 min. The supernatants were collected. StaySafe Nucleic Acid Reagent was mixed with the supernatants to detect pDNA. pDNA was measured by fluorescence at 514/537 nm (excitation/emission). PEI2k was measured by fluorescence at 648/663 nm (excitation/emission).

2.11 Cell culture and transfection
C6 cells were seeded 24 h before transfection at a density of 1× 105 cells/well in 12-well plates. To optimize the weight ratio for transfection, the RBP/PEI2k/pb-Luc ternary-complexes were prepared at various weight ratios by mixing 1 μg pβ-Luc with increasing amounts of RBP and PEI2k. The PEI25k/pb-Luc complex was prepared at 1:1 weight ratios. Prior to transfection,

the medium was replaced with serum- free DMEM. Then, the complexes were added to the cell and incubated for 4 h at 37°C in a humidified incubator under 5% CO2 and 95% air. After 4 h, the medium was replaced with fresh DMEM containing 10% FBS, and the cells were incubated for an additional 20 h at 37°C.

2.12 Luciferase assay
When the transfection was completed, the cells were washed twice with phosphate- buffered saline (PBS). After removing the PBS, 120 ml of reporter lysis buffer was added to each well, and the cells were incubated for 7 min at room temperature. The cell extracts were harvested and transferred to micro-centrifuge tubes. After vortexing for 15 s, the cells were centrifuged at 11,000 ´ g for 5 min, and the supernatant was transferred to fresh tubes. The luciferase activity of the supernatant was measured in terms of relative light units (RLU) using a 96-well plate luminometer (Berthold Detection System GmbH, Pforzheim, Germany). The protein concentration of the extract was measured by the BCA protein assay kit. The final values of luciferase activity were reported in terms of RLU/mg of total protein.

2.13 MTT assay
C6 cells were seeded in 24-well plates at a density of 5 × 104 cells/well before transfection. The amount of pDNA (pb-Luc or pHSVtk) was fixed at 0.5 mg/well, and the carrier/DNA complexes were prepared at their optimal ratios. Transfection was performed as described above. For cytotoxicity tests of prodrug therapy, GCV was added to the cells at a final concentration of 10 mg/ml. For cytotoxicity tests of carriers, GCV was not added to the cells. Then, the cells were incubated for 24 h at 37°C. After incubation, 40 ml of a 5 mg/ml MTT solution was added to each well, and the cells were incubated for 4 h at 37°C. After incubation, the MTT-containing medium was removed, and 100 ml of DMSO was added to dissolve the formazan crystals formed by the live cells. Absorbance was measured at 570 nm using a microplate reader. Cell viability (%) was calculated according to the following equation: cell viability (%) = (OD570 (sample) / OD570 (control)) × 100.

2.14 Flow cytometry analysis and confocal microscopy
For the cellular uptake test, pb-Luc was labeled with Label IT nucleic acid labeling kits according to the manufacturer’s manual. C6 cells were seeded at a density of 2.0 × 105 cells/well in 6-well plates 24 h before transfection. The RBP/PEI2k/pb-Luc, PEI2k/pb-Luc, and PEI25k/pb-Luc complexes were prepared at their optimal weight ratios. The amount of pb-Luc was fixed at 2 µg/well. Before transfection, the culture medium was replaced with serum- free DMEM, and the complexes were added to the cells. The cells were then incubated for 4 h. After the incubation, the transfection mixture was removed, and 1 ml of fresh DMEM containing 10% FBS was added. The cells were then incubated for an additional 20 h at 37°C. The cells were harvested, and Cy5-positive cells were analyzed by flow cytometry using a BD FACS CaliburTM system.
For confocal microscopy study, pb-Luc was labeled with Cy5 by using Label IT Nucleic Acid Labeling K its. C6 cells were seeded at a density of 1 × 105 cells per well in 12-well plates and incubated for 24 h at 37°C. Cy5- labeled pDNA (0.3 µg) was mixed with RBP/PEI2k, PEI2k and PEI25k, respectively. After 30 min incubation at room temperature, the Cy5- labeled pDNA/carrier complexes were added to the cells, which were incubated for an additional 24 h at 37°C. Then, the cells were washed twice with PBS and fixed with paraformaldehyde. The nuclei

were stained with DAPI in blue. The Cy5-positive cells were evaluated by confocal microscopy (Leica Microsystems, Wetzlar, Germany).
For annexin V assays, C6 cells were seeded at a density of 2 × 105 cells/well in 6-well plates 24 h before transfection. The carrier/pDNA complexes were prepared at optimized weight ratios and were transfected into the cells. After 4 h, the transfection medium was replaced with
0.5 ml of fresh DMEM containing 10% (v/v) FBS and GCV (10 mg/ml). Then, the cells were incubated for 20 h at 37°C. The cells were harvested, transferred to fresh tubes, and washed twice with cold PBS. Apoptosis was measured by annexin V staining according to the manufacturer’s instructions. Flow cytometry was performed using the BD FACS Calibur™ system.

2.15 In vivo pEGFP delivery efficiency of ternary-complex; GFP immunohistochemistry
In vivo gene delivery efficiency of ternary-complex was measured by immunohistochemistry for enhanced green fluorescent protein (EGFP). The RBP/PEI2k/pEGFP, PEI2k/pEGFP, and PEI25k/pEGFP complexes were incubated for 30 min at room temperature. The rats were anesthetized and the 10 ml of complexes were injected using a 26-gauge Hamilton syringe. After two days, the rats were sacrificed by perfusion and the brains were harvested. The harvested brains were incubated in 20% sucrose in 1× PBS at 4°C overnight. The brains were incubated in OCT Compound and frozen using liquid nitrogen. The frozen brains were cut into 10-μm-thick sections with a cryotome cryostat. The brain sections were incubated with anti- GFP antibody at 4°C overnight. The sections were washed twice in tris-buffered saline with tween (TBST) and the nuclei were stained with DAPI. The stained sections were observed by fluorescence microscopy.

2.16 In vivo experiments using an intracranial tumor rat model
All animal experimental procedures were approved by the institutional guidelines of the IACUC of Hanyang University. Intracranial tumor rat models were developed with seven-week- old male Sprague Dawley (SD) rats. The rats were anesthetized using 5% isoflurane in 70% N2O and 30% O2, and the skull was exposed to bore a 2-mm hole. The monitoring point was 2 mm lateral to bregma and carefully drilled using a gentle saline (0.89 % NaCl) drip to avoid disrupting the dura (coordinates to bregma: anteroposterior, 0 mm; lateral, 2.0 mm; ventral, 4.0 mm). Then, C6 cells at 1×105 cells/10 ml were injected into the cerebral cortex using a 26-gauge Hamilton syringe. One week after tumor implantation, the rats were randomly allocated into 4 groups: control, PEI2k/pHSVtk, PEI25k/pHSVtk, and RBP/PEI2k/pHSVtk. Each group consisted of eight rats. The carrier/pHSVtk complexes were prepared at the optimal weight ratios and injected at the same location as the tumor implantation. The pDNA amount was fixed at 1
mg/head. Ten microliters of the carrier/pDNA complexes were stereotaxically injected into the site of the tumor cell implantation. GCV (30 mg/kg rat) was injected intraperitoneally every 24 h for one week. After two weeks of tumor implantation, the rats were sacrificed by perfusion. The brains were harvested and fixed with 4% paraformaldehyde for further analysis.

2.17 Immunohistochemistry and terminal deoxynucleotidyl transferase dUTP nick end labeling (TUNEL) assay
The brains were harvested and fixed in 4% PFA, and the fixed brain samples were embedded in paraffin and cut into 5-μ m-thick sections. The sections were stained with immunohistochemistry accessory kits using a rabbit anti-HSVtk antibody or anti-a-smooth muscle actin antibody. The TUNEL assay was performed using the DeadEndTM ColorimetricTUNEL System following the manufacturer’s instructions.

2.18 Nissl staining
The brains were harvested and fixed in 4% PFA, and the fixed brain samples were embedded in paraffin and cut into 5-μm-thick sections. The brain sections were placed on slides and incubated in 0.1% cresyl violet. Sections were passed through a destaining solution (70% ethanol, 10% acetic acid), dehydrated (100 % ethanol and xylene), and cover-slipped with mounting solution. The total brain size and tumor volume were measured using Image J 1.42 software (National Institutes of Health, Bethesda, MD).

2.19 Statistical analysis
Statistical analysis of all data was performed by ANOVA followed by the Newman- Keuls test. All data are presented as the average ± standard error, and P- values less than 0.05 were considered statistically significant.

3. Results
3.1 Production and biological effects of RBP
RBP was designed as an anti- inflammatory peptide, based on the sequence of the RAGE- binding domain in wtHMGB1. As shown in Fig. 1, the RAGE-binding domain of wtHMGB1 is located at the C-terminal end of the B-box (Fig. 1A). In previous studies, the A-box of HMGB1 (HMGB1A) had anti- inflammatory effects on various inflammatory diseases including acute lung injury (ALI) and ischemic stroke [21, 30-32]. However, the RAGE-binding domain of wtHMGB1 was not contained in HMGB1A, suggesting that RBP might have a different mechanism than HMGB1A underlying its anti- inflammatory effects. In our previous study, HMGB1A bound to wtHMGB1 and inhibited the activity of wtHMGB1, which could be one of the anti- inflammatory mechanisms of HMGB1A [32]. In contrast, we hypothesized that RBP might bind to RAGE and interfere with the normal interaction between RAGE and its ligands such as wtHMGB1, S100, and LPS. As a result, RBP might induce an anti- inflammatory effect by inhibition of RAGE- mediated signal transduction [33]. In order to examine this hypothesis, we produced RBP as an anti-inflammatory peptide.

Structures of HMGB1 (A) and the RBP expression plasmid DNA (B)
The RAGE-binding domain of HMGB1 is located at the C-terminal region of the B-box. The RBP cDNA was amplified by PCR and inserted into the pET21a for construction of pET21a-RBP. The construction of pET21a-RBP was confirmed by sequencing.

To produce RBP by recombinant DNA technology, an RBP expression vector was constructed. The PCR-amplified RBP cDNA was inserted into pET21a, resulting in construction of pET21a-RBP (Fig. 1B). RBP was over-expressed in BL21 bacteria, purified by nickel affinity chromatography, and eluted from the column using 150 mM imidazole (Fig. 2A). The purification of RBP was confirmed by SDS-PAGE (Fig. 2B). In SDS-PAGE, a single band of approximately 5 kDa was detected after affinity chromatography. Contamination with endotoxins including LPS was eliminated by polymyxin B chromatography.

. Purification of RBP
(A) Nickel chelate affinity chromatography
RBP was purified by nickel chelate affinity chromatography using step gradients with increasing concentrations of imidazole. RBP was eluted at a concentration of 150 mM imidazole.
(B) SDS-PAGE
Purified RBP was analyzed by SDS-PAGE. M: molecular weight markers, 1: crude extracts after IPTG induction, 2: purified RBP.

RBP might function by binding to RAGE to inhibit RAGE-mediated induction of cytokines in cells. To examine this hypothesis, Raw264.7 cells were activated by the RAGE ligand S100B. The treatment of Raw264.7 cells with S100B increased the production of TNF-a (Fig. 3A). However, the TNF-a level was reduced by the addition of RBP in the S100B treated Raw264.7 cells (Fig. 3A). This suggests that RBP inhibited the interaction between S100B and RAGE and decreased the induction of TNF-a in the RAGE activated cells.

Biological activity of RBP
(A) TNF-a ELISA
S100B, RBP, and BSA were added to Raw264.7 cells, and the cells were incubated at 37°C for 4 h. The cell culture medium was replaced with fresh serum-free medium, and the cells were incubated for an additional 20 h. The TNF-α level was measured by ELISA. The data are presented as mean value ± standard deviation of quadruplicate experiments. *P<0.05 as compared with the other groups.
(B) VEGF ELISA
C6 cells were activated with 1 mg wtHMGB1 for 4 h. Increasing amounts of RBP were added to the cells and incubated for 4 h. Then, the medium was replaced with fresh serum-free EGM-2 medium, and the cells were incubated for an additional 20 h. After the incubation, the medium was harvested and analyzed with VEGF ELISA. The data are presented as mean value ± standard deviation of quadruplicate experiments. **, ***P<0.05 compared with the other groups.
(C) Tube formation assay
C6 cells were activated with 1 mg wtHMGB1 for 4 h. Increasing amounts of RBP were added to the cells and incubated for 4 h. The medium was replaced with fresh serum-free EGM-2 medium. After an additional incubation for 20 h, the medium was harvested and used for HUVEC culture. HUVECs were cultured in a Matrigel-coated plate in EGM-2 medium from the C6 cells for an additional 4 h. Tube formation was evaluated by staining with calcein AM, and the stained cells were visualized by microscopy.
(D) Quantitation of tube formation assay

Tube formation positive area was quantified using ImageJ software. The obtained images were analyzed using the “Angiogenesis Analyzer” tool, programmed in ImageJ’s macro language. #,##P<0.05 compared with wtHMGB1 and control.

VEGF is another important cytokine induced by the RAGE- mediated signal pathway. C6 cells were treated with wtHMGB1, which is a natural RAGE ligand. Treatment with wtHMGB1 increased the secretion of VEGF from C6 cells (Fig. 3B), confirming the effect of RAGE activation. However, treatment of the cells with RBP reduced the secretion of VEGF from the activated C6 cells in a dose-dependent manner (Fig. 3B). To confirm the anti-angiogenic effect of RBP by reducing VEGF secretion, the endothelial tube formation assays were performed. First, the C6 cells were incubated with wtHMGB1 or wtHMGB1+RBP. Then, the culture medium was harvested and added to HUVECs grown on a Matrigel layer. The medium harvested from the cells incubated with RBP inhibited the tube formation of HUVECs in a dose-dependent manner (Fig. 3C and 3D), confirming that RBP had an anti-angiogenesis effect.

3.2 Preparation and characterization of the RBP/PEI2k/pDNA ternary-complexes
The amino acid sequence of the RAGE-binding domain is KLKEKYEKDIAAYRAKGKPDAAKKGVVKAEKSKKKKE, which has a net charge of +9. The ternary-complex of RBP/PEI2k/pDNA was prepared by charge-based interactions since RBP and PEI2k have positive charges and pDNA has a negative charge (Fig. 4A).
First, RBP/pDNA complex formation was confirmed by a gel retardation assay (Fig. 4B). In the gel retardation assay, pDNA was retarded by the addition of RBP in a dose-dependent manner (Fig. 4B). pDNA was completely retarded at a weight ratio of 8:1 (RBP:pDNA). Then, the RBP/PEI2k/pDNA ternary-complex was produced. In previous reports, PEI2k had lower gene delivery efficiency than PEI25k, but did not have high toxicity [34]. Due to the low toxicity of PEI2k, PEI2k was chosen for formation of the ternary-complexes. The RBP/PEI2k/pDNA ternary-complexes were prepared at various weight ratios and analyzed by gel retardation assays. The results showed that pDNA was completely retarded at weight ratios of 8:1:1 or higher (Fig. 4B).
The stability of the ternary-complexes was measured by heparin competition assays. The ternary-complexes were prepared at an 8:20:1 weight ratio. As a control, the PEI2k/pDNA complexes were prepared at a 20:1 weight ratio. Increasing amounts of heparins were added to the complexes, and the complexes were analyzed by gel electrophoresis. The results indicated that pDNA was released from the complexes with 15 mg heparin, which was 15 times higher than the amount of pDNA in the complex (Fig. 4C). This suggests that the ternary-complex and the PEI2k/pDNA binary-complex had similar stability.
The pDNA protection effect of the delivery carriers was examined by DNase I protection assays. The complexes were prepared and incubated with DNase I for the indicated time. Naked pDNA was completely degraded after incubation for 30 min with DNase I (Fig. 4D). However, the ternary-complex protected pDNA for up to 60 min (Fig. 4D). In particular, the ternary- complex protected pDNA more efficiently than the PEI2k/pDNA complexes (Fig. 4D).
The particle size of the ternary-complex was approximately 138 nm (Table 1). The particle size of the ternary-complex was slightly larger than the particle sizes of the RBP/pDNA and PEI2k/pDNA complexes. The zeta-potential of the ternary-complex was approximately 14 mV, which was higher than the zeta-potential of the RBP/pDNA and PEI2k/pDNA complexes (Table 1).

MANUSCRIPT
The ternary-complex formation was confirmed by pull-down assay with anti-His tag antibody. Ternary-complex was prepared with pDNA, RBP, and fluorescence-labeled PEI2k. The complex was precipitated with anti- his tag antibody, because RBP contained 6´ His tag. PEI2k/pDNA complex was used as a negative control, since it was not precipitated with anti- His tag antibody. Then, the precipitated complex was analyzed to detect pDNA and PEI2k. The results showed that RBP/PEI2k/pDNA ternary-complex contained higher level of pDNA and PEI2k than negative control, PEI2k/pDNA complex (Table 2). These results confirmed the formation of ternary-complex.
Image

Physical characterization
(A) Formation of the ternary complex by electrostatic interaction (B) Gel retardation assay
The RBP/pb-Luc or RBP/PEI2k/pb-Luc complexes were analyzed with 1% agarose gel electrophoresis.
(C) Heparin competition assay
The RBP/PEI2k/pb-Luc and PEI2k/pb-Luc complexes were prepared as described in Materials and methods. The complexes were incubated with increasing amounts of heparin, and the DNA was analyzed by 1% agarose gel electrophoresis.
(D) DNase I protection assay
The PEI2k/pb-Luc and RBP/PEI2k/pb-Luc complexes were prepared as described in Materials and methods. The complexes were incubated with DNase I for 1, 30, or 60 min. Naked pDNA was used as a control. After incubation, DNA was analyzed by 1% agarose gel electrophoresis.

Table 1. Particle size and zeta-potential
Sample (Weight ratio) Particle size (nm) Zeta-potential (mV)
RBP/pDNA (8/1) 121.3 ± 19.1 11.9 ± 0.85
PEI2k/pDNA (20/1) 132.0 ± 4.13 6.77 ± 5.74
RBP/PEI2k/pDNA (8/20/1) 138.1 ± 30.5 14.09 ± 8.77

Table 2. Detection of pDNA and PEI2k in the ternary-complex immune-precipitated with anti-His tag antibody

pDNA
(excitation/emission = 514/537 nm) PEI2k
(excitation/emission = 648/663 nm)
PEI2k/pDNA 174.0 ± 3.0 11100 ± 5770
RBP/PEI2k/pDNA 211.5 ± 9.5 76010 ± 17070

3.3 Transfection efficiency and toxicity of the RBP/PEI2k/pDNA ternary-complexes
Transfection efficiency of the ternary-complexes was evaluated by delivery of the luciferase gene. First, the weight ratio of the PEI2k/pDNA complex was optimized by transfection with the complexes at various weight ratios (Fig. 5A). PEI2k had the highest transfection efficiency at a 20:1 weight ratio, so the ratio between PEI2k and pDNA was fixed at this ratio. Then, the ternary-complexes were prepared by addition of various amounts of RBP and PEI2k to pDNA. The transfection assays confirmed that the ternary-complexes had the highest transfection efficiency at a weight ratio of 8:20:1 (Fig. 5B). Therefore, the weight ratios of RBP, PEI2k, and pDNA were fixed at this ratio for the following experiments. The transfection efficiency of the ternary-complexes was compared with those of other carriers such as PEI2k and PEI25k. The transfection assays were performed in the absence of serum (Fig. 5C) or in the presence of serum (Fig. 5D). In the absence of serum, the PEI25k/pDNA complexes had the highest transfection efficiency of all tested carriers (Fig. 5C). Although the transfection efficiency of the ternary-complexes was lower than that of the PEI25k/pDNA complex, it was higher than the PEI2k/pDNA complex, which indicates that the addition of RBP to the PEI2k/pDNA complexes increased the transfection efficiency. In the presence of serum, the RBP/PEI2k/pDNA ternary-complex had the highest transfection efficiency (Fig. 5D).
The cellular uptake efficiencies of the complexes were measured by flow cytometry with Cy5-labeled pDNA. The results showed that the cellular uptake efficiency of the ternary-complex was higher than the PEI2k/pDNA complex (Fig. 6).
The toxicity of the ternary-complex was measured by the MTT assay and compared to other carriers such as PEI2k or PEI25k. The toxicity of the ternary-complex was similar to that of the PEI2k/pDNA binary-complex (Fig. 7). However, the PEI25k/pDNA complex had higher toxicity than the PEI2k/pDNA or ternary-complexes (Fig. 7).

Transfection efficiency of the RBP/PEI2k/pb-Luc complex
(A) and (B) Optimization of the weight ratios of the PEI2k/pDNA (A) and RBP/PEI2k/pb-Luc (B) complexes for transfection
The weight ratios of the PEI2k/pDNA complex and the RBP/PEI2k/pDNA ternary-complex were
optimized by a transfection assay in C6 cells. The complexes were prepared at various weight ratios and transfected into C6 cells. The transfection efficiencies were measured by a luciferase assay. The data are presented as mean value ± standard deviation of quadruplicate experiments. *P<0.05 compared with the other groups. **P<0.01 compared with the weight ratios of 0:20:1, 2:20:1, 4:20:1, and 6:20:1, but there was no statistical significance compared with 10:20:1.
(C) and (D) Comparison of the transfection efficiency of the RBP/PEI2k/pb-Luc complex with other carriers in the absence (C) or presence (D) of serum
The PEI2k/pb-Luc, RBP/PEI2k/pb-Luc, and PEI25k/pb-Luc complexes were prepared at their optimal transfection efficiency and transfected into C6 cells in the absence or presence of serum. After 24 h, the
transfection efficiencies were measured by a luciferase assay. The data are presented as mean value ±
standard deviation of quadruplicate experiments. ***, ****P<0.05 compared with the other groups.

Cellular uptake of the RBP/PEI2k/pb-Luc complex
ImageThe RBP/PEI2k/pb-Luc, PEI2k/pb-Luc, and PEI25k/pb-Luc complexes were prepared at their optimal weight ratios with Cy5 labeled pb-Luc and transfected into C6 cells. After incubation for 20 h, the cells were harvested, and Cy5-positive cells were analyzed by flow cytometry (A) and confocal microscopy (B). Nuclei were counter-stained with DAPI in blue color. Scale bar is 25 mm.

Cytotoxicity of the RBP/PEI2k/pb-Luc complex
The complexes were prepared at their optimal weight ratios and transfected into C6 cells. After 24 h, cytotoxicity was evaluated by the MTT assay. The data are presented as mean value ± standard deviation of quadruplicate experiments. **P<0.01 compared with control and PEI2k/pDNA.

3.4 Therapeutic effect of the RBP/PEI2k/pHSVtk ternary-complex
MANUSCRIPT
ACCEPTED
For the in vitro evaluation, the anti-tumor effect of the RBP/PEI2k/pHSVtk ternary- complex was measured by transfection assays in C6 cells. The PEI2k/pHSVtk, PEI25k/pHSVtk, and ternary-complexes were transfected into C6 cells, and the cells were incubated with GCV. Then, the apoptosis level and cell viability were measured 24 hr after the transfection by annexin V and MTT assays, respectively. The results showed that the RBP/PEI2k/pHSVtk ternary- complex induced higher levels of apoptosis and increased cell death compared with the PEI2k/pHSVtk complex (Figs. 8A and 8B). Considering that the PEI2k/pb-Luc and RBP/PEI2k/pb-Luc ternary-complexes had similar toxicity, as shown in Fig. 7, the higher toxicity of the RBP/PEI2k/pHSVtk complex than the PEI2k/pHSVtk complex might be due to the higher pHSVtk delivery efficiency of the ternary-complex. It was also observed that the PEI25k/pDNA complex had higher cytotoxicity to C6 cells than the ternary-complex. However, this cytotoxic effect of the PEI25k/pHSVtk complex might be due to not only higher pHSVtk delivery efficiency, but also to the higher toxicity of PEI25k itself, as shown in Fig. 7.

In vitro anti-tumor effects of delivery of the RBP/PEI2k/pHSVtk ternary-complex
The PEI2k/pHSVtk, RBP/PEI2k/pHSVtk, and PEI25k/pHSVtk complexes were prepared at their optimal weight ratios and transfected into C6 cells. Then, the cells were incubated with GCV for 24 h. (A) The apoptosis level was evaluated by an annexin V assay using flow cytometry. (B) Cell viability was measured by the MTT assay.

The pDNA delivery efficiency of the RBP/PEI2k/pDNA ternary-complex was evaluated in the rat brains after stereotaxic injection using pEGFP. The expression of the EGFP gene was measured by immunohistochemistry. The results showed that the EGFP expression level in the RBP/PEI2k/pDNA ternary-complex group was higher than the PEI25k/pDNA or PEI2k/pDNA binary complex group (Fig. 9). This result confirmed that the ternary-complex had higher gene delivery efficiency than the binary complexes in the brain in vivo.

In vivo pEGFP delivery efficiency of ternary-complex: GFP immunohistochemistry
In vivo gene delivery efficiency of ternary-complex was measured by immunohistochemistry for EGFP. The RBP/PEI2k/pEGFP, PEI2k/pEGFP, and PEI25k/pEGFP complexes were injected into the rat brains. After two days, the brains were harvested and subjected to immunohistochemistry with anti-EGFP antibody. The stained sections were observed by fluorescence microscopy. The nuclei were stained with DAPI. The scale bar is 100 mm.

The therapeutic effect of the RBP/PEI2k/pHSVtk ternary-complex was evaluated in an intracranial glioblastoma animal model. The ternary-complex was stereotaxically injected into the intracranial glioblastoma model rats, while the PEI2k/pHSVtk and PEI25k/pHSVtk complexes were injected as controls. After 1 week, the animals were sacrificed, and the brains were harvested for analysis. The expression of HSVtk was evaluated by immunostaining with an anti- HSVtk antibody. The expression level of pHSVtk was higher in the ternary-complex group compared with the control, PEI2k/pHSVtk, and PEI25k/pHSVtk complex groups (Fig. 10). These results suggest that the ternary-complex had higher efficiency in the delivery of pDNA into the glioblastoma. Interestingly, the HSVtk expression level in the PEI25k/pHSVtk complex group was not higher than the expression level in the PEI2k/pHSVtk complex in vivo (Fig. 10). PEI25k did not have higher gene delivery efficiency than PEI2k or RBP/PEI2k in the intracranial tumors in vivo. This result might be unexpected considering that PEI25k had higher transfection efficiency than PEI2k in the in vitro transfection experiment (Fig. 5C). However, this observation was not novel. In previous studies, PEI25k did not show higher gene delivery efficiency than PEI2k in brain gene delivery in vivo, although it had much higher transfection efficiency than PEI2k in vitro [35-37]. The cause of this phenomenon has not been identified, but it was speculated that a carrier with higher molecular weight and higher positive charge density can be captured more easily in the extracellular matrix of the brain and not be as efficiently internalized into cells [35].

Immunostaining of HSVtk after injection of the complexes into an intracranial glioblastoma model
The PEI2k/pHSVtk, RBP/PEI2k/pHSVtk, and PEI25k/pHSVtk complexes were prepared at their optimal
weight ratios and stereotaxically injected into tumors in the brains of rats. After 1 week, the brains were harvested, embedded in paraffin, and cut into 5-μm-thick sections. The sections were stained with immunohistochemistry accessory kits using a rabbit anti-HSVtk antibody. The scale bars represent 20 mm.

ACCEPTED
To determine the anti-angiogenesis effects of the ternary-complex, a-smooth muscle actin was stained in the tumors after delivery of the ternary-complex. The results showed that the delivery of the ternary-complex decreased the a-smooth muscle actin level compared with delivery of the control, PEI2k/pHSVtk, and PEI25k/pHSVtk complexes (Fig. 11). In addition, the apoptosis level in the tumor area was measured by the TUNEL assay. In the PEI2k/pHSVtk and PEI25k/pHSVtk complex groups, the apoptosis levels were increased compared with the control. However, the levels of apoptosis in these groups were lower than that of the ternary-complex (Fig. 12). Finally, N issl staining of the tissues was performed to determine tumor volume, which was also effectively reduced in the ternary-complex group compared with the other complex groups (Fig. 13).
ImageFig. 11. Immunostaining of a -smooth muscle actin after injection of the complexes into an intracranial glioblastoma model
The PEI2k/pHSVtk, RBP/PEI2k/pHSVtk, and PEI25k/pHSVtk complexes were prepared at their optimal
weight ratios and stereotaxically injected into tumors in the brains of rats. After 1 week, the brains were harvested, embedded in paraffin, and cut into 5-μm-thick sections. The sections were stained with

MANUSCRIPT
immunohistochemistry accessory kits using a rabbit anti-a-smooth muscle actin antibody. Left: Typical images of immunostaining. Right: Quantitation of the a-smooth muscle actin-positive area from triplicate experiments. The positive areas were measured using Image J 1.42 software (NIH, National Institutes of Health). *P<0.01 compared with the other groups. The scale bars represent 20 mm.

In vivo TUNEL staining after injection of the complexes into an intracranial glioblastoma model
The PEI2k/pHSVtk, RBP/PEI2k/pHSVtk, and PEI25k/pHSVtk complexes were prepared at their optimal
weight ratios and stereotaxically injected into tumors in the brains of rats. After 1 week, the brains were harvested, embedded in paraffin, and cut into 5-μm-thick sections. TUNEL staining was performed to identify apoptotic cells. Left: Typical images from the TUNEL assay. Right: Quantitation of TUNEL- positive cells from triplicate experiments. The positive areas were measured using Image J 1.42 software (NIH, National Institutes of Health). *P<0.01 compared with the other groups. The scale bars represent 20
mm.

Suppression of tumor growth by delivery of the RBP/PEI2k/pHSVtk ternary-complex in an intracranial glioblastoma model
The PEI2k/pHSVtk, RBP/PEI2k/pHSVtk, and PEI25k/pHSVtk complexes were prepared at their optimal weight ratios and injected into the glioblastoma rat model. After 1 week, the brains were harvested and subjected to Nissl staining. The total brain size and tumor volume were measured using Image J 1.42 software (NIH, National Institutes of Health). Left: Quantitation of tumor volume. The data are presented as mean value ± standard deviation (n=5). *P<0.01 compared with the other groups. Right: Typical
images of the Nissl staining.

4. Discussion
Clinical trials have been extensively carried out with the HSVtk genes, and their results suggest that suicide gene therapy with the HSVtk gene is a promising approach for the treatment of glioblastoma [23, 38]. However, clinical trials have shown that HSVtk gene therapy with adenoviral vector does not increase the overall survival of patients, and the adenoviral system can have severe adverse effects [23]. In this study, a ternary-complex with anti-tumor and anti- angiogenic dual effects was developed using RBP, PEI2k, and a suicide gene, pHSVtk. Despite their low transfection efficiency, polymeric gene carriers can be useful for suicide gene therapy with pHSVtk and GCV in glioblastoma. Gene delivery with polymeric carriers has three main barriers: the cellular membrane, endosomal membrane, and nuclear membrane [39]. In particular, nuclear entry of pDNA is inefficient for gene delivery with polymeric carriers. pDNA with polymeric carriers can enter nuclei in rapidly growing cells more efficiently than in quiescent cells, since the nuclear membrane disappears in the mitosis phase. In glioblastoma, the rapidly growing cells in the brain are mainly tumor cells, while most other cells are fully differentiated and quiescent. Therefore, gene delivery with polymeric carriers is more efficient in tumor cells than in normal cells in the brain, suggesting a potential method for tumor-targeting in glioblastoma gene delivery. Another consideration of HSVtk gene therapy is the ‘by-stander effect.’ pHSVtk-transfected cells convert the GCV pro-drug into active GCV-phosphate, which can be transferred to neighboring cells through gap-junctions. Therefore, GCV-HSVtk gene therapy with polymeric carriers might have relatively high anti-tumor effects due to the by- stander effect.

Recently, combination therapies with genes and drugs have been developed as a more effective treatment for glioblastoma based on delivery of a therapeutic gene and anti-cancer drug [24, 40-42]. These combination therapy studies used micelle-type carriers for combined delivery of genes and drugs. For example, polylactic acid-block-polydimethylaminoethyl methacrylate (PLA-b-PDMAEMA) micelles were used for delivery of miRNA and doxorubicin into gliomas [42]. Hydrophobic doxorubicin was loaded into the cores of the micelles, and miRNA was bound to the positive surface of the micelles. Similarly, R7L10 peptide micelles were evaluated as a carrier for combined delivery of curcumin and the HSVtk genes [24]. Hydrophobic curcumin was loaded into the cores of the micelles, and the HSVtk genes formed complexes with the R7L10 peptide by electrostatic interactions. This combination therapy was effective in glioblastoma animal models, due to the additive effect of the gene and drug. In the current study, we developed another combination therapy with the HSVtk gene and RBP. Unlike the previous combination therapy, the anti-tumor drug, RBP, is hydrophilic with positive charges. Since both therapeutic agents, pHSVtk and RBP, had opposite charges, they bound to each other when mixed together, as shown in the gel retardation assay (Fig. 4A). The RBP/pDNA complex had a relatively small size compared with the PEI2k/pDNA complex (Table 1). However, the RBP/pHSVtk complex did not have any remarkable transfection efficiency into C6 cells in vitro. These findings suggest that, although it forms a complex with pDNA, RBP does not function as a gene carrier, and an additional gene carrier is required for delivery of pDNA into a tumor. Therefore, RBP and PEI2k were combined with pDNA by electrostatic interactions, producing the RBP/PEI2k/pDNA ternary-complex.
PEI25k has been used as a standard gene carrier in polymeric gene delivery studies since it has high transfection efficiency and tolerable cytotoxicity in vitro and in vivo [43].
However, the toxicity of PEI25k is not acceptable for clinical application. PEI2k has lower toxicity than PEI25k, although the transfection efficiency of PEI2k is lower than PEI25k. In addition, polyethylene glycol (PEG) derivatives of PEI2k are approved for clinical trials in the United States [44, 45]. Due to its low toxicity, PEI2k was used as a gene carrier in this study. The in vitro transfection assays showed that the ternary-complex had the highest efficiency at an 8:20:1 weight ratio (Fig. 5B). The ternary-complex had a higher transfection efficiency than the PEI2k/pDNA binary-complex, suggesting that the addition of RBP increases the transfection efficiency of the PEI2k/pDNA complex. There are two possible explanations for this phenomenon. First, the ter
nary-complex might have higher stability than the PEI2k/pDNA complex due to the additional positive charges of RBP. Second, the ternary-complex could more efficiently interact with RAGE and enter cells by receptor- mediated endocytosis. Further study is required to identify the mechanism underlying the ability of RBP to increase the transfection efficiency of the PEI2k/pDNA complex.
The transfection efficiency of PEI25k/pDNA complex was higher than that of RBP/PEI2k/pDNA complex in vitro (Fig. 5C). However, the gene delivery efficiency of PEI25k/pDNA complex was lower than that of RBP/PEI2k/pDNA complex in the brain in vivo (Fig. 9). Similar results were reported in the previous reports [36, 46]. PEI25k had higher transfection efficiency than dexamethasone conjugated PEI2k (PEI2k-Dexa) and R3V6 in in vitro transfection experiments, while PEI2k-Dexa and R3V6 had higher transfection than PEI25k in the brain in vivo. The lower transfection efficiency of PEI25k in the brain may be due to serum proteins. For in vitro assays, transfection was performed in the absence of serum, which was different from in vivo condition. Positive proteins such as albumin may interact with PEI25k and form aggregates more easily than other carriers, due to higher charge density of PEI25k.

Indeed,the transfection efficiency of PEI25k was decreased in the presence of serum in vitro (Fig. 5D). This suggests that serum proteins may reduce the transfection efficiency of PEI25k in vivo.An in vitro tube formation assay showed that treatment with RBP reduced the production of VEGF from cells (Fig. 3B). As a result, the lower VEGF concentration in the culture medium decreased tube formation of HUVECs, suggesting an anti-angiogenic effect (Fig. 3C). The reduction of VEGF by RBP could be due to the down-regulation of RAGE- mediated signal transduction. RAGE- mediated signal transduction eventually activates nuclear factor-kB (NF-kB) [47, 48]. Activated NF-kB induces the expression of cytokines including VEGF, TNF-a, IL-1b, and IL-6 [49]. Recent studies revealed that RAGE ligands are over-expressed in most types of solid tumors, and the activation of RAGE by these ligands induced angiogenesis [17, 47, 50-52]. Therefore, the inhibition of RAGE using RBP could be an effective strategy for inhibition of angiogenesis.
The anti-tumor and anti-angiogenic effects of the ternary-complex were demonstrated in an intracranial glioblastoma animal model. Immunostaining for HSVtk and a-smooth muscle actin, the TUNEL assay, and Nissl staining indicated that the ternary-complex was more effective in gene delivery, tumor-suppression, and anti-angiogenesis than the PEI2k/pHSVtk and PEI25k/pHSVtk complexes (Figs. 11, 12, and 13).

The ternary-complex is a combination delivery system of two therapeutic agents, RBP and pHSVtk. Due to positive charge of RBP, the ternary-complex was not a simple mixture of RBP and pHSVtk. In addition to its anti-angiogenic effects, RBP participated in the complex formation with PEI2k and pHSVtk and enhanced gene delivery efficiency. Therefore, the ternary- complex of RBP, PEI2k, and pHSVtk had a synergistic effect of anti-tumor and anti-angiogenic effects.

5. Conclusions
In this study, the RBP/PEI2k/pHSVtk ternary-complex was developed and evaluated as a therapeutic gene and drug delivery system with anti-tumor and anti-angiogenic dual functions. The ternary-complex reduced the production of VEGF and decreased the proliferation of a- smooth muscle cells in the tumors. Furthermore, the ternary-complex induced HSVtk expression and activated GCV- mediated tumor cell death. Therefore, the ternary-complex might be useful for the treatment of glioblastoma due to its anti-tumor and anti-angiogenic dual effects.

Acknowledgments
This work was supported by Basic Science Research Program through the National Research Foundation of Korea (NRF) funded by the Ministry of Education (NRF- 2016R1D1A1B03934778) and the Ministry of Science and ICT (NRF-2017R1A2B4009036). It was also supported by the Bio & Medical Technology Development Program through NRF funded by the Ministry of Science and ICT (NRF-2016M3A9B4918833).

5. References
[1] D. Krex, B. Klink, C. Hartmann, A. von Deimling, T. Pietsch, M. Simon, M. Sabel, J.P. Steinbach, O. Heese, G. Reifenberger, M. Weller, G. Schackert, N. German Glioma, Long-term survival with glioblastoma multiforme, Brain, 130 (2007) 2596-2606.
[2] R. Stupp, M.E. Hegi, M.J. van den Bent, W.P. Mason, M. Weller, R.O. Mirimanoff, J.G. Cairncross, R. European Organisation for, T. Treatment of Cancer Brain, G. Radiotherapy,
G. National Cancer Institute of Canada Clinical Trials, Changing paradigms--an update on the multidisciplinary management of malignant glioma, Oncologist, 11 (2006) 165-180.
[3] S.J. Hentschel, F.F. Lang, Current surgical management of glioblastoma, Cancer J, 9 (2003) 113-125.
[4] A. Riehl, J. Nemeth, P. Angel, J. Hess, The receptor RAGE: Bridging inflammation and cancer, Cell Commun Signal, 7 (2009) 12.
[5] C.D. Logsdon, M.K. Fuentes, E.H. Huang, T. Arumugam, RAGE and RAGE ligands in cancer, Curr Mol Med, 7 (2007) 777-789.
[6] R. Ramasamy, S.J. Vannucci, S.S. Yan, K. Herold, S.F. Yan, A.M. Schmidt, Advanced glycation end products and RAGE: a common thread in aging, diabetes, neurodegeneration, and inflammation, Glycobiology, 15 (2005) 16R-28R.
[7] M.K. Fuentes, S.S. Nigavekar, T. Arumugam, C.D. Logsdon, A.M. Schmidt, J.C. Park,
E.H. Huang, RAGE activation by S100P in colon cancer stimulates growth, migration, and cell signaling pathways, Dis Colon Rectum, 50 (2007) 1230-1240.
[8] H.M. Lander, J.M. Tauras, J.S. Ogiste, O. Hori, R.A. Moss, A.M. Schmidt, Activation of the receptor for advanced glycation end products triggers a p21(ras)-dependent mitogen- activated protein kinase pathway regulated by oxidant stress, J Biol Chem, 272 (1997) 17810-17814.
[9] H.J. Huttunen, C. Fages, H. Rauvala, Receptor for advanced glycation end products (RAGE)- mediated neurite outgrowth and activation of NF-kappaB require the cytoplasmic domain of the receptor but different downstream signaling pathways, J Biol Chem, 274 (1999) 19919-19924.
[10] P. Carmeliet, VEGF as a key mediator of angiogenesis in cancer, Oncology, 69 Suppl 3 (2005) 4-10.
[11] M.P. Barr, D.J. Bouchier-Hayes, J.J. Harmey, Vascular endothelial growth factor is an autocrine survival factor for breast tumour cells under hypoxia, Int J Oncol, 32 (2008) 41- 48.
[12] R.D. Leek, A.L. Harris, C.E. Lewis, Cytokine networks in solid human tumors: regulation of angiogenesis, J Leukoc Biol, 56 (1994) 423-435.
[13] L.G. Bucciarelli, T. Wendt, L. Rong, E. Lalla, M.A. Hofmann, M.T. Goova, A. Taguchi,
S.F. Yan, S.D. Yan, D.M. Stern, A.M. Schmidt, RAGE is a multiligand receptor of the immunoglobulin superfamily: implications for homeostasis and chronic disease, Cell Mol Life Sci, 59 (2002) 1117-1128.
[14] T. Yi, Z. Yi, S.G. Cho, J. Luo, M.K. Pandey, B.B. Aggarwal, M. Liu, Gambogic acid inhibits angiogenesis and prostate tumor growth by suppressing vascular endothelial growth factor receptor 2 signaling, Cancer Res, 68 (2008) 1843-1850.
[15] J.L. Hernandez, L. Padilla, S. Dakhel, T. Coll, R. Hervas, J. Adan, M. Masa, F. Mitjans,
J.M. Martinez, S. Coma, L. Rodriguez, V. Noe, C.J. Ciudad, F. Blasco, R. Messeguer, Therapeutic targeting of tumor growth and angiogenesis with a novel anti-S100A4 monoclonal antibody, PLoS One, 8 (2013) e72480.

[16] S. Mizumoto, J. Takahashi, K. Sugahara, Receptor for advanced glycation end products (RAGE) functions as receptor for specific sulfated glycosaminoglycans, and anti-RAGE antibody or sulfated glycosaminoglycans delivered in vivo inhibit pulmonary metastasis of tumor cells, J. Biol. Chem., 287 (2012) 18985-18994.
[17] C.D. Logsdon, M.K. Fuentes, E.H. Huang, T. Arumugam, RAGE and RAGE ligands in cancer, Curr. Mol. Med., 7 (2007) 777-789.
[18] L. Campana, L. Bosurgi, P. Rovere-Querini, HMGB1: a two-headed signal regulating tumor progression and immunity, Curr. Opin. Immunol., 20 (2008) 518-523.
[19] D.L. Tang, R. Kang, H.J. Zeh, M.T. Lotze, High- mobility group box 1 and cancer, Biochim. Biophys. Acta, 1799 (2010) 131-140.
[20] H. Yang, H. Wang, C.J. Czura, K.J. Tracey, HMGB1 as a cytokine and therapeutic target, J Endotoxin Res, 8 (2002) 469-472.
[21] J.H. Song, J.Y. Kim, C. Piao, S. Lee, B. Kim, S.J. Song, J.S. Choi, M. Lee, Delivery of the high- mobility group box 1 box A peptide using heparin in the acute lung injury animal models, J. Control. Release, 234 (2016) 33-40.
[22] C.L. Zhang, M.G. Shu, H.W. Qi, L.W. Li, Inhibition of tumor angiogenesis by HMGB1 A box peptide, Med. Hypotheses, 70 (2008) 343-345.
[23] M. Westphal, S. Yla-Herttuala, J. Martin, P. Warnke, P. Menei, D. Eckland, J. Kinley, R. Kay, Z. Ram, A.S. Group, Adenovirus- mediated gene therapy with sitimagene ceradenovec followed by intravenous ganciclovir for patients with operable high- grade glioma (ASPECT): a randomised, open-label, phase 3 trial, Lancet Oncol., 14 (2013) 823- 833.
[24] J.H. Park, J. Han, M. Lee, Thymidine kinase gene delivery using curcumin loaded peptide micelles as a combination therapy for glioblastoma, Pharm. Res., 32 (2015) 528-537.
[25] H.Y. Huang, W.T. Kuo, M.J. Chou, Y.Y. Huang, Co-delivery of anti-vascular endothelial growth factor siRNA and doxorubicin by multifunctional polymeric micelle for tumor growth suppression, J. Biomed. Mater. Res. A, 97 (2011) 330-338.
[26] H. Kim, H.A. Kim, Y.M. Bae, J.S. Choi, M. Lee, Dexamethasone-conjugated polyethylenimine as an efficient gene carrier with an anti-apoptotic effect to cardiomyocytes, J. Gene Med., 11 (2009) 515-522.
[27] M. Lee, J. Rentz, M. Bikram, S. Han, D.A. Bull, S.W. Kim, Hypoxia-inducible VEGF gene delivery to ischemic myocardium using water-soluble lipopolymer, Gene Ther., 10 (2003) 1535-1542.
[28] H.A. Kim, J.H. Park, N. Yi, M. Lee, Delivery of hypoxia and glioma dual-specific suicide gene using dexamethasone conjugated polyethylenimine for glioblastoma-specific gene therapy, Mol. Pharm., 11 (2014) 938-950.
[29] M. Lee, J.W. Nah, Y. Kwon, J.J. Koh, K.S. Ko, S.W. Kim, Water-soluble and low molecular weight chitosan-based plasmid DNA delivery, Pharm. Res., 18 (2001) 427-431.
[30] Y.C. Jin, S.W. Kim, F. Cheng, J.H. Shin, J.K. Park, S. Lee, J.E. Lee, P.L. Han, M. Lee,
K.K. Kim, H. Choi, J.K. Lee, The effect of biodegradable gelatin microspheres on the neuroprotective effects of high mobility group box 1 A box in the postischemic brain, Biomaterials, 32 (2011) 899-908.
[31] B. Oh, M. Lee, Combined delivery of HMGB-1 box A peptide and S1PLyase siRNA in animal models of acute lung injury, J. Control. Release, 175 (2014) 25-35.
[32] Y.H. Hwang, M.J. Kim, Y.K. Lee, M. Lee, D.Y. Lee, HMGB1 modulation in pancreatic islets using a cell-permeable A-box fragment, J. Control. Release, 246 (2016) 155-163.
[33] S. Lee, C. Piao, G. Kim, J.Y. Kim, E. Choi, M. Lee, Production and application of

HMGB1 derived recombinant RAGE-antagonist peptide for anti- inflammatory therapy in acute lung injury, Eur. J. Pharm. Sci., 114 (2018) 275-284.
[34] W.T. Godbey, K.K. Wu, A.G. Mikos, Size matters: molecular weight affects the efficiency of poly(ethylenimine) as a gene delivery vehicle, J. Biomed. Mater. Res., 45 (1999) 268-275.
[35] H. Hyun, Y.W. Won, K.M. Kim, J. Lee, M. Lee, Y.H. Kim, Therapeutic effects of a reducible poly (oligo-D-arginine) carrier with the heme oxygenase-1 gene in the treatment of hypoxic- ischemic brain injury, Biomaterials, 31 (2010) 9128-9134.
[36] H. Hyun, J. Lee, W. Hwang do, S. Kim, D.K. Hyun, J.S. Choi, J.K. Lee, M. Lee, Combinational therapy of ischemic brain stroke by delivery of heme oxygenase-1 gene and dexamethasone, Biomaterials, 32 (2011) 306-315.
[37] P. Jeon, M. Choi, J. Oh, M. Lee, Dexamethasone-conjugated polyamidoamine dendrimer for delivery of the heme oxygenase-1 gene into the Ischemic brain, Macromol. Biosci., 15 (2015) 1021-1028.
[38] A.K. Muhammad, M. Puntel, M. Candolfi, A. Salem, K. Yagiz, C. Farrokhi, K.M. Kroeger, W. Xiong, J.F. Curtin, C. Liu, K. Lawrence, N.S. Bondale, J. Lerner, G.J. Baker,
D. Foulad, R.N. Pechnick, D. Palmer, P. Ng, P.R. Lowenstein, M.G. Castro, Study of the efficacy, biodistribution, and safety profile of therapeutic gutless adenovirus vectors as a prelude to a phase I clinical trial for glioblastoma, Clinical pharmacology and therapeutics, 88 (2010) 204-213.
[39] C.M. Wiethoff, C.R. Middaugh, Barriers to nonviral gene delivery, J. Pharm. Sci., 92 (2003) 203-217.
[40] Z.Z. Yang, J.Q. Li, Z.Z. Wang, D.W. Dong, X.R. Qi, Tumor-targeting dual peptides- modified cationic liposomes for delivery of siRNA and docetaxel to gliomas, Biomaterials, 35 (2014) 5226-5239.
[41] B. Oh, J. Han, E. Choi, X. Tan, M. Lee, Peptide micelle-mediated delivery of tissue- specific suicide gene and combined therapy with avastin in a glioblastoma model, J. Pharm. Sci., 104 (2015) 1461-1469.
[42] X. Qian, L. Long, Z. Shi, C. Liu, M. Qiu, J. Sheng, P. Pu, X. Yuan, Y. Ren, C. Kang, Star-branched amphiphilic PLA-b-PDMAEMA copolymers for co-delivery of miR-21 inhibitor and doxorubicin to treat glioma, Biomaterials, 35 (2014) 2322-2335.
[43] H.C. Kang, M. Lee, Y.H. Bae, Polymeric gene carriers, Crit. Rev. Eukaryot. Gene Expr., 15 (2005) 317-342.
[44] K. Anwer, M.N. Barnes, J. Fewell, D.H. Lewis, R.D. Alvarez, Phase-I clinical trial of IL- 12 plasmid/lipopolymer complexes for the treatment of recurrent ovarian cancer, Gene Ther., 17 (2010) 360-369.
[45] K. Anwer, F.J. Kelly, C. Chu, J.G. Fewell, D. Lewis, R.D. Alvarez, Phase I trial of a formulated IL-12 plasmid in combination with carboplatin and docetaxel chemotherapy in the treatment of platinum-sensitive recurrent ovarian cancer, Gynecol. Oncol., 131 (2013) 169-173.
[46] J. Lee, H. Hyun, J. Kim, J.H. Ryu, H.A. Kim, J.H. Park, M. Lee, Dexamethasone- loaded peptide micelles for delivery of the heme oxygenase-1 gene to ischemic brain, J. Control. Release, 158 (2012) 131-138.
[47] A. Taguchi, D.C. Blood, G. del Toro, A. Canet, D.C. Lee, W. Qu, N. Tanji, Y. Lu, E. Lalla, C. Fu, M.A. Hofmann, T. Kislinger, M. Ingram, A. Lu, H. Tanaka, O. Hori, S. Ogawa, D.M. Stern, A.M. Schmidt, Blockade of RAGE-amphoterin signalling suppresses tumour growth and metastases, Nature, 405 (2000) 354-360.

[48] B. Lv, H. Wang, Y. Tang, Z. Fan, X. Xiao, F. Chen, High- mobility group box 1 protein induces tissue factor expression in vascular endothelial cells via activation of NF-kappaB and Egr-1, Thromb. Haemost., 102 (2009) 352-359.
[49] Y. Yamamoto, A. Harashima, H. Saito, K. Tsuneyama, S. Munesue, S. Motoyoshi, D. Han, T. Watanabe, M. Asano, S. Takasawa, H. Okamoto, S. Shimura, T. Karasawa, H. Yonekura, H. Yamamoto, Septic shock is associated with receptor for advanced glycation end products ligation of LPS, J. Immunol., 186 (2011) 3248-3257.
[50] X. Chen, L. Zhang, I.Y. Zhang, J. Liang, H. Wang, M. Ouyang, S. Wu, A.C. da Fonseca,
L. Weng, Y. Yamamoto, H. Yamamoto, R. Natarajan, B. Badie, RAGE expression in tumor-associated macrophages promotes angiogenesis in glioma, Cancer Res., 74 (2014) 7285-7297.
[51] T. Okamoto, S. Yamagishi, Y. Inagaki, S. Amano, K. Koga, R. Abe, M. Takeuchi, S. Ohno, A. Yoshimura, Z. Makita, Angiogenesis induced by advanced glycation end products and its prevention by cerivastatin, FASEB J, 16 (2002) 1928-1930.
[52] C. Schlueter, H. Weber, B. Meyer, P. Rogalla, K. Roser, S. Hauke, J. Bullerdiek, Angiogenetic signaling Thymidine through hypoxia: HMGB1: an angiogenetic switch molecule, Am J Pathol, 166 (2005) 1259-1263.