YB-1 dependent oncolytic adenovirus efficiently inhibits tumor growth of glioma cancer stem like cells

U87-MG (ATCC), U373-MG and LN-18 cells (kindly provided by Dr. N. de Tribolet, Zurich, Switzerland) were maintained in DMEM with glutamine (Biochrom, Berlin, Germany) containing 10% FCS (PAN-Biotech, Aidenbach, Germany). Brain CSC lines R11, R28, R40, and R49 were obtained from patients with primary GBM as previously described [31] and were maintained as tumorspheres in stem cell-permissive DMEM-F12 medium supplemented with 20 ng/ml of each human recombinant epidermal growth factor (EGF; BD Biosciences, Heidelberg, Germany), human recombinant basic fibroblast growth factor (bFGF; R&D Systems, Wiesbaden, Germany), human leukemia inhibitory factor (LIF; Millipore, Billerica, MA, USA), and 2% B27 (Life Technologies, Carlsbad, CA, USA) for preservation of the tumors’ original molecular characteristics and for minor differentiation. SV-GA cells (a human astrocytic subclone of human fetal glial cells transduced with an origin-defective mutant of simian virus 40) have been previously described [32] and were maintained in MEM medium with 2 mM L-glutamine, 10% fetal bovine serum, and antibiotic solution.

The following viruses were used: (i) wild type adenovirus of serotype 5 (Ad-WT), (ii) Ad-Delo3-RGD, an oncolytic adenovirus (OAV) that combines the dl520 genotype (CR3 deletion of E1A restricting viral amplification and anti-tumor activity to drug-resistant cells displaying nuclear YB-1 expression) with an E1B19K deletion and a RGD motif in the fiber knob [27], (iii) dl703 [33] that contains expanded deletions in early region 1 (3180 bp) and (iv) the E1A-deleted adenovirus dl312 described in detail in [34]. All viruses were produced in HEK293 cells and purified by two consecutive standard cesium chloride gradient centrifugations and size-exclusion chromatography (PD-10 Desalting Columns, GE Healthcare, Freiburg, Germany). Viral titers were determined by plaque assay using HEK293 cells. Multiplicity of infection (MOI) is therefore indicated as plaque forming units (pfu) per cell. Virus dose was optimized for each in vitro experiment. The absence of replication competent adenovirus (RCA) in Ad-Delo3-RGD preparations was excluded by PCR using specific primers for the E1A-CR3 region (for primer sequences proceed to “DNA isolation and PCR”). In general, particle (determined by OD-measurement) to PFU ratio in virus preparation were between 30 and 50.

To calculate the EC₅₀ of TMZ (Schering-Plough, Kenilworth, NJ, USA) in brain CSC lines, the cells were separated by trituation and viability was assessed by trypan blue staining. 10.000 viable brain CSC were treated with increasing concentrations of TMZ (1–2000 μM) for 24 h followed by a medium change. After further 72 h, cell viability was assessed using the MTT assay.

Cells were lysed using ProteoJET Mammalian Cell Lysis Reagent (Fermentas, St. Lon-Rot, Germany) supplemented with complete protease inhibitor cocktail (Roche Diagnostics, Filderstadt, Germany) and incubated at room temperature for 10 min. The lysates were clarified by centrifugation and protein concentration was measured using the Bradford assay. 40 μg protein was separated on SDS-polyacrylamide gels and transferred onto poly-vinylidine-diluloride (PVDF) membranes (Millipore, Schwalbach, Germany). For detection, the following antibodies were used: rabbit anti-phospho-YB1 (Ser102), rabbit PathScan^® Multiplex Western Cocktail I (anti-phospho-p90RSK, anti-phospho-AKT, anti-phospho-p44/42 MAPK (Erk1/2), anti-phospho-S6 Ribosomal Protein Detection Kit), rabbit anti-phospho-AKT (all antibodies were purchase from Cell Signaling/Millipore), goat anti-MGMT (R&D Systems) or rabbit anti-YB-1 [27]. Immunoreactive proteins were detected using the Amersham enhanced chemiluminescence (ECL) or ECL plus western blot detection system (GE Healthcare).

For the assessment of viral replication, total DNA from infected cells (50 MOI) was isolated using digestion buffer (100 mM NaCl, 10 mM TrisHCl pH 8.0, 25 mM EDTA pH 8.0, 0.5% SDS), Proteinase K and phenol-chloroform. After precipitation with ethanol, DNA was solubilized in 10 mM TrisHCl pH 8.0. Quantitative real-time PCR was performed using the ABI Prism 7900HT sequence detection system (Applied Biosystems) using 100 ng of total DNA per reaction and SYBR green fluorescent dye (Agilent Technologies, Waldbronn, Germany). The specific primers (Eurofins, Hamburg, Germany) used for real-time PCR analyses were: fiber-fw: 5′-AAGCTAGCCCTGCAAACATCA, fiber-rev: 5′-CCCAAGCTACCAGTGGCAGTA, ß-actin-fw: 5′-TAAGTAGGTGCACAGTAGGTCTGA, and ß-actin-rev: 5′-AAAGTGCAAAGAACACGGCTAAG. For identity testing following primer were used: 12Sfw: 5’-AATGGCCGCCAGTCTTTT, 12Srev: 5’-GCCATGCAAGTTAAACATTATC, 13Sfw: 5’-GGCATGTTTGTCTACAGTAAG, 13Srev: 5’-GCCATGCAAGTTAAACATTATC, E1b19kfw: 5’-CGTGAGAGTTGGTGGGCGT, E1b19krev: 5’-CTTCGCTCCATTTATCCT, E3ADPfw: 5’-ATGTCAGCATCTGACTTTGGCC, E3ADPrev: 5’-CTCGAGGAATCATGTCTC, E3fw: 5’-GTTAATGTCAGGTCGCCTAAGTCG, E3rev: 5’-GTGTGTTGCCCGCGACCATT RGDfw: 5’-CTGCCGCGGAGACTGTTTC, RGDrev: 5’-CTGCAATTGAAAAATAAACACG. Cycling conditions started with initial enzyme activation at 95°C for 15 min, followed by 40 cycles of 15 sec denaturation at 95°C, 15 sec, annealing at 60°C, and 15 sec elongation at 72°C. Homogeneity of the amplification product was confirmed by melting curve analysis (T_m-fiber: 85°C, T_m-ß-actin: 83°C). Detection of amplification of the adenoviral sequences E1A12S, E1A13S, E1B19, E3, ADP and RGD (35 cycles, annealing at 55°C) was done by agarose gel electrophoresis.

Brain CSC were grown on slides and fixed for 20 min in a methanol/acetone (1:1) mixture at −20°C. Immunocytochemical reactions were performed using rabbit anti-YB-1 antibody followed by a FITC-conjugated swine anti-rabbit secondary antibody (1:20; Dako, Hamburg, Germany). Slides were mounted with Vectashield (Vector Laboratories/Axxora, Lörrach, Germany) and images were taken with the AxioImagerZ1 with ApoTome (Zeiss Opticals, Jena, Germany). Immunohistochemical reactions were conducted using polyclonal rabbit antibodies directed against YB-1 as previously described [35].

R11, R28 or R40 cells were seeded in 12-well plates (1 × 10⁵ cells in 0.5 ml stem cell medium per well) and infected with the indicated viruses the next day. Cells were cultivated for further 24–36 h under normoxic or hypoxic (<0.66% O₂) conditions [36]. After this treatment, cells from 12-well plates were diluted for the clonogenic dilution assay into 24-well plates containing 1ml of stem cell medium in 1:10 dilution steps and incubated for approximately 10 doubling times (4–6 weeks).

1.5 x 10⁶ R28 cells were transfected with 1000 pmol annealed double-strand control-siRNA (Qiagen, Hilden, Germany; sense 5’-UUCUCCGAACGUGUCACG UdTdT-3’, antisense 5’-ACGUGACACGUUCGGAGAAdTdT-3’)’ or a siRNA specific for YB-1 (Qiagen; sense 5’-GGCGAAGGUUCCCACCUUATT-3’, antisense 5’-UAAGGUGGGAACCUUCGCCTG-3), using Lipofectamine™ 2000 (Life Technologies). After 24 h, cells were divided into three aliquots and infected with 50 MOI dl703, Ad-Delo3-RGD or Ad-WT, respectively. Cells were harvested after 4 h and after 48 h. DNA was isolated as described above and solubilized in 10 mM TrisHCl pH 8.0. Expression of YB-1 was confirmed in cells harvested 48 h after siRNA transfection using immunoblot.

NMRI nude mice (Janvier, Le Genest Saint Isle, France) were anesthetized and placed into a stereotactic fixation device (Stoelting, Wood Dale, USA). 10⁵ viable R28 cells were injected into the right striatum. At day 7 post implantation, the mice were randomly separated in 4 groups (n=7 to 8 mice per group), followed by an intratumoral injection of either PBS (mock) or 3 x 10⁸ plaques forming units( pfu) Ad-Delo3-RGD. Half of the mice were treated intraperitoneally with TMZ (5 mg/kg) at day 10 and 17. The mice were sacrificed when developing neurological symptoms. Brains were isolated and fixed with 4% paraformaldehyde for further analysis. All animal research was carried out in accordance with the German Animal Welfare Act and was approved by local authorities.

Tumors of mice were dissected, fixed in 4% formaldehyde, and embedded into paraffin. Serial 5 μm sections were cut and stained with hematoxylin and eosin (H&E). Histopathological evaluations were done on a light microscope (Eclipse E200, Nikon Instruments, Düsseldorf, Germany).

It is established that AKT, MAPK/ERK and ribosomal S6 kinase (RSK) can interact with and phosphorylate YB-1 [37‐39]. To explore the presence of activated AKT, MAPK/ERK and RSK in brain CSC, we examined protein phosphorylation by immunoblot analysis (Figure 1A). All four brain CSC lines showed phospho-AKT and phospho-ERK expression to a different degree. Especially striking was the phosphorylation of RSK in R11, R28 and R40 CSCs in comparison to established GBM cell lines (U87-MG, U373-MG, LN-18). In LN-18 cells which are characterized by high O⁶-Methylguanin-DNA-Methyltransferase (MGMT) expression and TMZ resistance [40], phospho-AKT and phospho-ERK_1/2 are equally expressed as it is the case for the brain CSC lines R28 and R40. In contrast to established GBM cell lines, brain CSC lines showed no expression of phospho-S6.

TMZ resistance was reported in brain CSC [8], and different responsiveness to treatment with TMZ according to the MGMT status of these cells was described [9]. Since our set of brain CSCs showed different grades of TMZ resistance (high in R28, low in R11, intermediate in R40 and R49; being EC₅₀ in all CSC higher than the clinical therapeutic TMZ concentration of ~50 μM; data not shown), we tested MGMT expression in these cells using immunoblot analysis. We also included the GBM cell line LN-18 known to express MGMT and being highly resistant to TMZ [40] as well as U87-MG and U373-MG cells in this analysis. As shown in Figure 1B, MGMT was detectable in LN-18, R28, R40 and R49 cell lines, but not in U87-MG, U373-MG or R11 cells.

Considering YB-1 as a downstream target of PI3-K/AKT and MAPK/RSK signaling, we next evaluated YB-1 expression in normal brain tissue, GBM cell as well as brain CSC lines. Western blot analysis demonstrates elevated levels of both total and phosphorylated YB-1 in brain CSC lines and established GBM cell lines. In contrast, virtually no YB-1 expression was detectable in normal CNS tissue (Figure 2A right panel and Figure 2C, upper left micrograph). Localization of YB-1 varied in the examined brain CSC lines: In the TMZ-resistant brain CSC line R28, YB-1 is highly phosphorylated and located in the nucleus, whereas in R11 cells, showing the lowest IC₅₀ value for TMZ, YB-1 is predominately located in the cytoplasm (Figure 2B), corresponding to the weak phosphorylation status of YB-1 these cells. As shown in Figure 2C, immunohistochemical analysis of non-neoplastic brain tissue demonstrated no expression of YB-1. In glioma specimen, diversifying YB-1 expression patterns could be observed. A few glioma showed no YB-1 expression whereas in the majority of glioma biopsies, YB-1 could be easily detected. Some glioma exhibited cytoplasmic YB-1 expression whereas in other GBM specimen nuclear YB-1-specific staining could be demonstrated.

We have shown that YB-1 plays an important role in the adenoviral life cycle [41]. In addition, we have proven that drug-resistant cells displaying nuclear YB-1 expression facilitate adenovirus replication independent of the adenoviral E1A protein [42]. Thus, it seems reasonable to assess the replication capacity of an YB-1 dependent adenoviral vector in brain CSC. To exclude contamination of Ad-Delo3-RGD preparation with an YB-1-independent replication competent adenovirus (RCA), we analyzed all Ad-Delo3-RGD preparations for absence of E1A13S sequences which are essential for YB-1 independent virus replication [27]. As demonstrated in Figure 3A, no amplification of adenoviral DNA was measured using primers specific for E1A13S. To analyze whether Ad-Delo3-RGD was able to replicate in brain CSC, we infected these cells with a E1-deleted, replication deficient adenovirus (dl703), Ad-WT or Ad-Delo3-RGD and measured viral DNA replication by real-time PCR 4 h and 72 h post infection. As shown in Figure 3B, Ad-Delo3-RGD replicates around 100–600 fold better than dl703, and in R40 cells even 4.3 fold better than Ad-WT, which shows slightly better replication in the two other CSC lines (R11, R28). Under hypoxic conditions Ad-Delo3-RGD retains its ability to replicate (data not shown). Taken together, these data demonstrate that Ad-Delo3-RGD replicates efficiently in brain CSC.

To investigate whether Ad-Delo3-RGD induces a virus based cytopathic effect (CPE) in infected brain CSC lines, CPE assays were performed. As shown in Figure 3C, Ad-Delo3-RGD infection resulted in nearly complete cytolysis of infected brain CSC lines within 7 days, which was confirmed by MTT assay (data not shown). In contrast to Ad-Delo3-RGD infected brain CSC, in R28 and R49 cells expressing high levels of MGMT morphology was not altered by treatment with TMZ. In addition, infection of all CSC lines with the replication deficient adenovirus dl703 did not show any lytic effect. To confirm our results we additionally performed cell clonogenic dilution assays with infected brain CSC lines. As shown in Figure 4 (left panel), treatment of brain CSC lines with Ad-Delo3-RGD considerably inhibited growth in a sustained manner (indicated by the red color of the medium six weeks post infection). In contrast, TMZ treated or dl703 infected brain CSC, even under hypoxic conditions (Figure 4, right panel), retain their capacity to grow and of being metabolically active (leading to a yellow color of the medium six weeks post infection).

The impact of YB-1 in adenoviral replication was verified by siRNA-mediated knockdown of YB-1 in R28 cells. Down regulation of YB-1 was demonstrated by immunoblot 48 h post transfection. YB-1 was down regulated to 58% in YB-1 siRNA transfected R28 cells (Figure 5A,B). Knockdown of YB-1 in R28 cells resulted in reduced adenoviral replication compared to control siRNA transfected R28 cells (Figure 5C). 48 h after infection, copy numbers of adenoviral vectors in YB-1 knockdown R28 cells decreased to 76% (Ad-WT) and 42% (Ad-Delo3-RGD) in comparison to copy numbers in control siRNA transfected cells.

Based on the observation that nuclear YB-1 expression is a prerequisite of Ad-Delo3-RGD replication we suggested that Ad-Delo3-RGD will be unable to replicate in human non-dividing non-neoplastic brain cells. To confirm this hypothesis, we infected human immortalized SV-GA astrocytic cells with 10 MOI of Ad-Delo3-RGD or Ad-WT and compared particle formation in these cells in comparison to U87MG glioma cells. Ad-WT replication was even better in SV-GA cells compared to U87MG glioma cells (33-fold increase), but replication of the oncolytic adenovirus Ad-Delo3-RGD was markedly and significantly reduced in SV-GA cells compared to U87MG cells (Figure 5D).

To evaluate the potential of Ad-Delo3-RGD in affecting GBM growth, we established a mouse orthotopic brain tumor model using TMZ-resistant R28 brain CSC cells. These cells express high levels of MGMT (Figure 1B) as well as of aldehyde dehydrogenase (ALDH) 1A1 [43], both proteins procuring resistance to TMZ. R28-derived GBM-bearing mice were treated with either TMZ, were intratumorally injected with Ad-Delo3-RGD or achieved a combination therapy. The two groups treated with Ad-Delo3-RGD (either with or without additional TMZ treatment) survived significantly longer than mice which achieved solely TMZ or were mock-(PBS)-treated (p < 0.001). Median survival time increased from 125 days in mock treated mice up to 176 days in Ad-Delo3-RGD treated mice. 167 days post treatment none of the mice receiving an intratumoral PBS (mock) injection or were treated solely with TMZ (0/15) survived, whereas 50% (7/14) of the mice receiving an intratumoral Ad-Delo3-RGD injection were still alive. TMZ treatment of mice did not show any significant effect in none of the treatment groups (Figure 6A). Histopathological evaluation of mock-treated tumors (Figure 6B, upper left) as well as of tumors of TMZ treated mice (Figure 6B, upper middle) showed a similar histopathological appearance. In these mice, polymorphic, highly mitotic tumor tissue was detectable. In contrast, tumors of mice achieving intratumoral Ad-Delo3-RGD injections were smaller, presenting high numbers of apoptotic cells and less vital tumor cells (Figure 6B, upper right). However, in virus treated animals no toxicity or inflammation could be observed in any non-neoplastic part of the brain adjacent to the tumor including the subventricular zone (Figure 6B, lower left), cerebellum (Figure 6B, lower middle) and cortex (Figure 6B, lower right).