TP-0903 inhibits neuroblastoma cell growth and enhances the sensitivity to conventional chemotherapy

European Journal of Pharmacology 818 (2018) 435–448

Contents lists available at ScienceDirect

European Journal of Pharmacology journal homepage: www.elsevier.com/locate/ejphar

Molecular and cellular pharmacology

TP-0903 inhibits neuroblastoma cell growth and enhances the sensitivity to conventional chemotherapy

T

⁎

Sanja Aveica, , Diana Coralloa, Elena Porcùb, Marcella Pantilea, Daniele Bosob, Carlo Zanona, Giampietro Violab, Viktoryia Sidarovichc, Elena Mariottob, Alessandro Quattronec, Giuseppe Bassob, Gian Paolo Toninia Pediatric Research Institute – Città della Speranza, Neuroblastoma Laboratory, Padua, Italy University of Padua, Laboratory of Oncohematology, SDB Department, Padua, Italy c University of Trento, Centre for integrative biology - CIBIO, Trento, Italy a

b

A R T I C L E I N F O

A B S T R A C T

Keywords: Neuroblastoma ATRA TP-0903 Spheroids Zebrafish Drug combination

Neuroblastoma (NB) is an embryonal tumor with low cure rate for patients classified as high-risk. This class of NB tumors shows a very complex genomic background and requires aggressive treatment strategies. In this work we evaluated the efficacy of the novel multi-kinase inhibitor TP-0903 in impairing NB cells’ growth, proliferation and motility. In vitro studies were performed using cell lines with different molecular background, and in vivo studies were done using the zebrafish experimental model. Our results confirmed a strong cytotoxicity of TP-0903 already at the sub-micro molar concentrations. The observed cytotoxicity of TP-0903 was irreversible and the resulting apoptosis was caspase dependent. In addition, TP-0903 impaired colony formation and neurosphere creation. Depending on the molecular background of the selected NB cell lines, TP-0903 influenced either their capacity to migrate, to complete their cell cycle or both. Likewise, TP-0903 reduced NB cells intravasation in vitro and in vivo. Importantly, TP-0903 showed remarkable pharmacological efficacy not only as a mono-treatment, but also in combination with conventional chemotherapy drugs (ATRA, cisplatin, and VP16) in different types of NB cells. In conclusion, the multi-kinase activity of TP-0903 allowed the impairment of several biological processes required for expansion of NB cells, making them more vulnerable to the conventional chemotherapeutics. Altogether, our results support the eligibility of TP-0903 for further (pre)clinical assessments in NB.

1. Introduction Neuroblastoma (NB) is a tumor of the sympathetic nervous system that occurs predominantly in infants, toddlers and children at prescholar age (Luksch et al., 2016; Maris et al., 2007). Although the effectiveness of therapy settings for pediatric cancers is constantly improving, a portfolio of drugs regarding patients with high-risk (HR) NB has not been significantly amended yet. Therefore, the introduction of novel compounds is necessary (Lonergan et al., 2003). One of the reasons for poor survival of HR-NB patients is the elevated biological variability among the NB tumors (Brodeur, 2003). Analysis of the entire genome of patients with NB, by the next-generation sequencing, revealed a very large intra-tumoral heterogeneity (Esposito et al., 2017; Pugh et al., 2013). Beside well known prognostic factors for NB, such as ALK, MYCN and PHOX2B (Raabe et al., 2008), the functional analyses that followed the whole-genome sequencing confirmed the relevance of

⁎

additional molecules that might determine future clinical improvements (Peifer et al., 2015; Zage et al., 2012) or unravel new insights about already known targets (Bellini et al., 2015). Protein kinases are well known molecular markers in a variety of malignancies, and their targeting represents a very promising treatment strategy. The TAM protein family of receptor tyrosine kinases, particularly AXL, was proposed to be of particular relevance in conferring survival advantages to cancer cells (Graham et al., 2014). For these reasons, a raising interest of biologists and oncologists is directed toward AXL targeting (Verma et al., 2011). AXL is an important tumor propagator that impacts cell migration, adhesion and response to drugs through several signaling pathways: the extra-cellular signal-regulated kinase (ERK), the phosphatidylinositol 3-kinase/AKT and GTP-ases from Rho protein family (Korshunov, 2012). Also, the involvement of AXL oncoprotein in NB pathology has been reported recently (Debruyne et al., 2015). Considering that all beforehand mentioned AXL-

Corresponding author. E-mail address: [email protected] (S. Aveic).

https://doi.org/10.1016/j.ejphar.2017.11.016 Received 7 August 2017; Received in revised form 10 November 2017; Accepted 13 November 2017 Available online 14 November 2017 0014-2999/ © 2017 Elsevier B.V. All rights reserved.

European Journal of Pharmacology 818 (2018) 435–448

S. Aveic et al.

2.3. Cell growth recovery measurement

dependent pathways were found deregulated in NB (Brodeur, 2003), the inhibition of AXL, as their governor, would theoretically abolish the survival advantages that they assure to NB cells. Another clinically relevant kinase for NB tumors is Aurora A (AURKA). While the precise role of AXL in pathology of NB tumors is currently under investigation, AURKA represents a clear negative prognostic marker for NB patients (Ramani et al., 2015). Beside the regulation of the cell cycle and progression of mitosis, AURKA also protects MYCN from degradation (Otto et al., 2009). These attributes made AURKA an attractive target for decades (Romain et al., 2014), and a search for a more effective AURKA inhibitor is still ongoing. Lately, a new multi-kinase drug called TP-0903, able to inhibit either AXL or AURKA, has been proposed (Myers et al., 2016). TP-0903 was highly efficient in the treatment of chronic lymphocytic leukemia and invasive breast cancer (Sinha et al., 2015; Soh et al., 2016). Its successful activity against other types of cancer prompted us to evaluate whether TP-0903 could confirm the same anti-tumoral effects in NB cells as well. We established the tumor cell directed cytotoxicity which caused an irreversible impairment of NB cells growth, sustaining the TP-0903 for further (pre)clinical evaluations in this cancer.

The capacity of cells to recover from TP-0903 treatment was measured after drug wash-out. Cells were plated as described in detail elsewhere (Aveic et al., 2016), and treated 24 h with increasing concentrations of TP-0903 (10, 20, 40, 80 and 100 nM) and DMSO. Afterwards, cell medium was removed and substituted with fresh drugfree RPMI. Cell viability was controlled after 24 h (24 h with TP-0903 + 24 h without TP-0903) and 48 h (24 h with TP-0903 + 48 h without TP-0903), and changes in cell growth were calculated as percent growth relative to the starting conditions applying the MTT assay. 2.4. Determination of cells' proliferation rate The 5-ethynyl-2'-deoxyuridine (EDU)–Click kit (BCK-Edu488, Sigma-Aldrich) was used for the evaluation of DNA synthesis as suggested by manufacturer. Shortly, 1 × 105 NB cells were seeded in the chamber slides one day before the treatment. Then TP-0903 was left for 24 h adding the EdU solution (10 μM) for 8 h before fixation with 3.7% formaldehyde (Sigma-Aldrich). After permeabilization with 0.5% Triton-X100, the reaction mix was added and the experiment carried out as recommended by the protocol. EdU-positive cells (green) were counted under the inverted fluorescence microscope (Vico, Eclipse Ti80, Nikon, Tokyo) and normalized to the number of total nuclei (stained with DAPI; blue).

2. Materials and methods 2.1. Cell culturing and treatments The human NB tumor cell lines NB3, kindly provided by Dr Luca Longo (IRCCS AOU San Martino-IST, Genoa), SH-SY5Y (from DSMZ) and IMR32 (from ATCC), were cultured in RPMI medium (SigmaAldrich, Milan, Italy). The selectivity of TP-0903 towards transformed cells and not normal ones was tested using Hek293T and BJ fibroblast cells. In the lack of healthy human neuroblasts as non-tumoral control for our study, we relied on the findings of Shaw et al. (Shaw et al., 2002) that attributed certain neuro-specific features to Hek293T cell line, which was kindly provided by Dr Stefano Indraccolo (IOV, Padua). The BJ cell line was kindly provided by Dr Michela Pozzobon (IRP CDS, Padua). Both non-cancer, control cell lines were grown in DMEM medium enriched with antibiotics and glutamine (1%), fetal bovine serum (FBS; 10%; all from Gibco, Life Technologies, Monza, Italy). Cell lines authentication was done at the BMR Genomics, Padua. Cell culture media were tested by PCR to exclude mycoplasma infections. The primary human umbilical vein endothelial cells (HUVEC) were purchased from Life Technology and maintained in M200 medium with addition of Low Serum Growth Supplement (LSGS; Life technology) until the sixth passage. All treatments were done using less than 0.1% of DMSO (SigmaAldrich) as control and the following chemical agents: all-trans retinoic acid (ATRA), VP16, Cisplatin, SB203580, Z-VAD, 5-aza-2′deoxycytidine (AZA; all from Sigma-Aldrich) and TP-0903 (Selleck Chemicals, Munich, Germany). Concentrations used for single or combined experiments are mentioned within the text.

2.5. RNA isolation, cDNA synthesis and quantitative RT-PCR analysis (qPCR) Total RNA was extracted with TRIzol reagent (Invitrogen, Life Technologies) and cDNA synthesized using Super Script II (Invitrogen), according to the manufacturer's recommendations. One-twentieth of produced cDNA was then used to determine gene expression by the ABI prism 7900 (Applied Biosystems, Forest City, CA) using the SYBR Green PCR Master Mix (Applied Biosystems) as described before (Aveic et al., 2016). The primer sequences are available upon request. 2.6. DNA isolation, bisulfite conversion and sequencing Genomic DNA from SH-SY5Y and IMR32 was extracted using columns (Invisorb Spin Tissue Mini Kit, Stratec), and 1 μg was treated with sodium bisulfite (EpiTect Bisulfite kit; Qiagen) according to the manufacturer's instructions. Bisulfite converted DNA was subjected to PCR reaction using High fidelity Taq polymerase (Qiagen) to amplify zone of CpG island found within the AXL gene promoter. Primers for the reaction were designed with the MethPrimer program: AXLbisuF 5’GTTTGAGTGTGTTTGTGGGTTAGTA-3’, AXLbisuR 5’-CAAACCTCCTTA ACCCTTCATT-3’ (Li and Dahiya, 2002); the PCR amplicons were cloned in TOP10 competent bacteria (Invitrogen) following manufacturer instructions. Ten colonies per sample were sequenced using BigDye Terminator Cycle Sequencing Kit v3.1 on the ABI PRISM 3500DX Genetic Analyzer sequencer (Applied Biosystems). The same sequencing procedure was used to check for the presence of exon 10 in AXL mRNA. DNA fragmentation was validated on 1% agarose gel (Invitrogen), after being processed as described elsewhere (Kasibhatla et al., 2006).

2.2. Viability assay and percent growth calculation Cells viability was determined 24 h, 48 h and 72 h post-treatment by measuring the amount of the resulting formazan, using Multilabel Plate Reader VICTOR (PerkinElmer, Waltham, MA) at 486 nm, and the number of cells used for the analyses was determined as described in detail before (Aveic et al., 2016). Results were calculated with respect to the starting conditions or to the DMSO control treatments. Growth rate percentage was used to quantify the change over time in treated and control samples respect to the initial conditions (0 h). Similarly, the MTT assay served for the Inhibitory Concentration 50% (IC50) determination.

2.7. Wound healing (scratch) and Boyden chamber assays Details for wound healing assay (IBIDI, Milano, Italy) settings were described previously (Aveic et al., 2016). Wound healing was analyzed at 0 h, 24 h and 48 h by measuring the percentage of scratched area (ImageJ software; NIH, Bethesda, MD). To avoid that the cell proliferation influence on the final assay results, the cells were kept under low serum (3% FBS) conditions. For cell migration measurements, the cells in DMSO or TP-0903 were plated on 8 µm porous membrane (BD

436

European Journal of Pharmacology 818 (2018) 435–448

S. Aveic et al.

supplemented with calcein-AM dye. The images of TP-0903-treated spheroids were captured in bright field and green fluorescent channel using the Operetta (PerkinElmer). Viability of spheroids was accessed by CellTiter-Glo 3D assay (Promega; Milan, Italy) following the manufacturer’s instructions.

Bioscience; San Jose, CA) with 3% of serum, whereas 20% of FBS was added to the lower chamber. Cells were left to migrate for 24 h. The following day, a fixation with 3.7% of formaldehyde was done and cells were permeabilized with methanol. After crystal violet staining, and removal of the cells from the upper part of the chamber using cotton swabs, the images were obtained using light microscope.

2.13. Colony formation assay 2.8. Western blot analysis Two thousand cells pre-treated with TP-0903 or DMSO were plated in MethoCult medium (Stemcell Technologies, Milan, Italy). Pre-treatments were done for 6 h and 48 h. Cells were then left for 10–14 days to form colonies. Colonies were visualized as described before (Aveic et al., 2016).

Treated cells were lysed with commercially available buffer (Biosource International; Camarillo, CA) and 20 μg of total proteins were analyzed as described before (Aveic et al., 2016). Double protein loading for each experimental condition was done on the same gel in order to evaluate the expression of more proteins in parallel, and to limit the number of membrane stripping. The following antibodies were tested: anti-PARP, cleaved anti-PARP, anti-ERK1/2, anti-pERK1/2, antiCaspase-3, anti-AXL, anti-H2AX and anti-γH2AX (Cell Signaling, Danvers, MA), anti-β TUBULIN (Novus Biologicals, Littleton, CO), antiAURKA (Millipore, Darmstadt, Germany), anti-PCNA (SCBT, Dallas,TX), diluted as suggested by manufacturer.

2.14. Flow cytometry Cell cycle profile was defined by flow cytometry (Cytomics FC500; Beckman Coulter, Brea, CA) after treatment with increasing concentrations of TP-0903 for 24 h. The procedure was described previously in detail (Aveic et al., 2016). Mitosis-specific marker phosphohistone H3 (pH3) was analyzed by flow cytometry after staining with primary anti-pH3 antibody (BioLegend, San Diego, CA). The cells were fixed using the same procedure described for cell cycle analysis. After centrifugation the cells were left in 1xPBS/1% BSA/0.25% Triton-X100. Afterwards, cell pellets were incubated with anti-pH3 antibody at dark. After the wash, pellets were processed as described for the cell cycle procedure. The expression of CD133 and CD44 cell surface markers was validated after incubation with anti-CD133-PE and anti-CD44-PE (BD Bioscience). Eventual apoptosis activation was validated using Annexin V/(PI) staining (BD Bioscience). For each sample minimum 10.000 events were recorded.

2.9. Rock activity assay Activation of Rho family of proteins was evaluated indirectly by measuring the activity of the downstream Rho effector - Rho-associated kinase (ROCK) using the 96-well ROCK activity immunoassay (Cell Biolabs. Inc.; San Diego, CA). Equal amounts of protein lysates from each experimental setting were loaded in triplicate as suggested by the manufacturer. The absorbance was measured using the VICTOR Multilabel Plate Reader (450 nm). 2.10. Human phospho-kinase array

2.15. Trans-endothelial migration assay The cells treated with TP-0903 or DMSO for 24 h were used for protein extraction according to the R&D Systems protocol (R&D Systems, Minneapolis, MN). A total of 500 μg of the proteins were used for the analyses, and the validation of the single spot intensity was done using ImageJ - Array analysis program (Carpentier, 2008). A simultaneous screening of 43 different kinases was done. The resulting data were then normalized for internal controls spotted on each membrane, and results presented as mean ± S.D.

Cell intravasation was assessed within the CytoSelect™ Trans-endothelial Migration Assay provided by Cell Biolabs, Inc. According to manufacturer protocol. Pre-treated cells (TP-0903 or DMSO) were labeled with the fluorescent dye CytoTracker™; then the cell suspension was laid down on a HUVEC-based endothelial monolayer pre-activated with TNFα (10 ng/µl). After 24 h from the co-culture assessment, the fluorescence intensity of NB cells migrated through the endothelium was measured using VICTOR Multilabel Plate Reader (530 nm).

2.11. Immunostaining 2.16. Zebrafish maintenance For immunostaining, the cells were washed, fixed in 3.7% formaldehyde and permeabilized with 0.5% Triton-X100. Samples were then blocked with 3% BSA and incubated with anti-Phalloidin antibody (1:400; Sigma-Aldrich) at 4 °C overnight for F-actin visualization. After washing, the incubation with Alexa Fluor 594-conjugated goat antirabbit antibody (red; 1:2000, Life technologies) was done. Cell nuclei were marked with DAPI (1:10.000; Sigma-Aldrich). The images were examined under a fluorescence microscope (Nikon) under 20X and 60X objectives.

Wild-type (AB/TU) and the Tg(fli1: EGFP) zebrafish line (Stoletov et al., 2007) were raised, staged and maintained as described by Kimmel et al. (Kimmel et al., 1995). The project, with protocol number 86/2016-PR, was examined and approved by the Italian Ethical Committee OPBA. 2.17. Fluorescent cell labeling, embryo preparation and tumor cell implantation

2.12. Matrigel matrix for spheroid study

Dechorionized (2 dpf) zebrafish embryos were anaesthetized with 0.003% tricaine (Sigma-Aldrich) and positioned on a 10 cm Petridish coated with 3% agarose. Non-fluorescent NB cells were labeled with the Vybrant® DiI Cell-Labeling Solution (Invitrogen) according to the manufacturer’s instructions. Fluorescent NB cells were re-suspended in 1xPBS and implanted within 3 h using borosilicate glass capillary needles (OD/ID: 1.0/0.75 mm, WPI), Pneumatic Picopump and a micromanipulator (WPI). Approximately 100 cells were injected within the duct of Cuvier of anesthetized embryos. After implantation, zebrafish embryos were maintained at 33 °C. Embryos showing less than 30–40 cells after 4 h post-injection were discarded from the analysis. For each cell line, more than 50 embryos per group were analyzed from at least

For this study, 2 × 105 cells were re-suspended in 1 ml of medium with DMSO or TP-0903 (10 or 20 nM) and left to grow for 6 days over matrix gel (Corning, Inc.; Corning, NY). The cells were checked for sphere formation and their sizes were measured using ImageJ software. For each experimental replicate at least 10 spheres were considered. To evaluate the viability of the cells grown as three-dimensional (3D) multi-cellular spheroids, 2.000–4.000 cells were plated in a round bottom ultra-low attachment 96-well plate (Corning) for 2–5 days. The formed spheroids were treated with increasing concentrations of TP0903 (2.5–100 nM) for 48 h. Afterwards, the growing media was 437

European Journal of Pharmacology 818 (2018) 435–448

S. Aveic et al.

et al., 1994), as confirmed by cDNA sequencing (Supplementary Fig. S1A; right image). In order to explain the reason of undetectable AXL expression in IMR32 cell line, we looked for the methylation status of the CpG island of the AXL gene promoter. CpG methylation was confirmed only for IMR32 cells and not for SH-SY5Y, by sequencing the PCR products of bisulfite converted DNA (data not shown). Additionally, treatment with the de-methylation agent AZA strengthened these observations since a 5-fold increase in the expression of AXL in IMR32, but not in SH-SY5Y cell line, was detected (Supplementary Fig. S1B). Together, these results proved that AXL and AURKA mRNA, along with their protein levels, differ among NB cells and suggested a cell dependent epigenetic regulation of AXL gene transcription.

three independent experiments. Embryos at 1 and 4 dpi were photographed live using a Nikon C2 H600L confocal microscope (20X water dipping objective). 2.18. Drug synergism calculation For the calculation of the Combination Index (CI), the Chou-Talalay test was used (Chou, 2007). The cells were treated for 24 h by the serial dilutions of each drug (TP-0903: 1–100 nM; ATRA: 1-10 μM; VP16: 1–10 μM; Cisplatin: 1–10 μM) whereas combined treatment was done applying several constant molar ratios (1:1; 1:10; 1:100, first position refers to TP-0903). Cell viability was calculated using MTT and data were processed for the determination of the interaction between drugs as synergistic (CI < 1), additive (CI=1) or antoagonistic (CI > 1) using CalcuSyn software 2.0 (Biosoft, Cambridge, UK).

3.2. TP-0903 works at nano-molar range in NB cells To determine the concentration range within which TP-0903 worked best, we measured the viability of NB cells after addition of increasing concentrations of the compound. Our data demonstrated a strong sensitivity of NB cells to TP-0903 action already at nM concentrations. This effect was time-dependent and particularly evident for NB3 and IMR32 cell lines (Supplementary Table S1). Notably, the viability of Hek293T cells did not yield important changes upon administration of TP-0903 (1–200 nM) for 24, 48 h or 72 h (MTT assay; data not shown), and neither it caused a cell death activation in any of control cell lines that have been used, Hek293T and BJ human fibroblasts (Supplementary Fig. S2). The concentrations of TP-0903 used for further studies were then selected to correspond to the pre-determined IC20 or IC40 (24 h post-treatment) for each cell in order to examine the effects of a grading scale of cytotoxicity on NB cells behavior.

2.19. Statistical analyses The experiments were done at least as triplicates and results expressed as the mean ± S.D. Statistical analyses were done on GraphPad Software 7.00 (La Jolla, CA) and based on Student's t-test (unpaired), one-way ANOVA (Dunnett’s multiple comparison test) or two-way ANOVA (for more than one experimental variable). For the in vivo studies, P-value was obtained by Fisher's exact test. The P-values minor or equal to 0.05 were considered statistically significant. 3. Results 3.1. AXL and AURKA oncoproteins are differentially expressed in NB cell lines

3.3. TP-0903 is efficient in reducing NB cells viability and proliferation The levels of AXL and AURKA transcripts and the corresponding proteins were examined in the SH-SY5Y, NB3 and IMR32 cell lines. These cell lines were chosen in order to include the most prevalent molecular profiles associated with NB (see Supplementary Table S1). We confirmed the expression of AXL at mRNA and protein levels in SHSY5Y, its less marked expression in NB3 and no quantifiable amount of AXL in IMR32 cells; AURKA showed the most evident levels in NB3 cell line and smaller amounts in SH-SY5Y and IMR32 cells (Fig. 1A and B). Interestingly, the PCR analysis confirmed no AXL transcript in IMR32 cells and showed products of different molecular weights between NB cells (SH-SY5Y and NB3) and Hek293T control (Supplementary Fig. S1A; left image). This difference was a result of the exclusion of exon 10 in NB, but not in Hek293T cells, due to its altered splicing (Neubauer

To understand how TP-0903 affects the NB cells number, we calculated the changes in cell growth after administration of increasing TP-0903 concentrations. As shown in Fig. 2A, the percent growth as a function of time was concentration dependent. Moreover, viability of NB3 and IMR32 cells was completely abolished 72 h upon addition of 40–80–100 nM of TP-0903 and 100 nM in the case of SH-SY5Y cell line (Fig. 2A; Supplementary Table S2). In order to assess whether the observed reduction in cell viability was a result of impaired cell proliferation, we evaluated the expression of the Ki-67 and PCNA, proliferation-specific markers, upon treatment with TP-0903. We confirmed a marked down-regulation of Ki-67 (Fig. 2B; Supplementary Table S3) and PCNA (Fig. 2C) expressions in treated versus control Fig. 1. The expression of AXL and AURKA mRNAs and oncoproteins in NB cell lines. A) Relative mRNA quantities (RQ) of AXL and AURKA in SH-SY5Y, NB3 and IMR32 cell lines were measured by qPCR. Bars represent the relative transcript expression as mean ± S.D. of triplicate measures for each gene after normalization for GAPDH as an internal control. B) Protein level for AXL and AURKA was validated by western blot (WB). The level of beta TUBULIN was considered as a control for proper protein loading. The differences in molecular weight (kDa) of AXL observed between SH-SY5Y and NB3 cells were the result of complete, and partial AXL glycosylation (140 kDa and 120 kDa protein bands), respectively.

438

European Journal of Pharmacology 818 (2018) 435–448

S. Aveic et al.

Fig. 2. Cells viability and proliferation after treatment with TP-0903. A) Cell viability was measured after addition of increasing concentrations of TP-0903 (10–100 nM) and the changes in the percent growth were calculated with respect to the initial condition (time point zero – 0 h). Three time points were considered (24 h, 48 h and 72 h). B) The RQ for Ki-67 mRNA was measured 24 h post-treatment with TP-0903. Results of the triplicate measures are presented as mean ± S.D. fold change with respect to DMSO control sample (RQ=1). Pvalues under or equal to 0.05 were considered statistically significant and were indicated with asterisks: *** for P-values ≤ 0.001, ** for P-values ≤ 0.01, and * for P-values ≤ 0.05 for each experiment. C) The level of PCNA protein was validated by WB, with beta TUBULIN as a loading control. Increasing concentrations of TP-0903 adapted for cell type were used for 48 h. D) Cell proliferation was assessed by measuring the percentage of cells with green signal (sign for EdU incorporation into the DNA of replicating cells) over the total cell number (blue signal). The images are representative for each experimental setting (10× objective lens), and corresponding quantitative data are presented below. Graph bars show the percentage of proliferating cells as mean ± S.D. All experiments were done at least in triplicate. The 4’,6-diamidino-2-phenylindole (DAPI; blue) served for nuclear counter-staining. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

death was caspase dependent, we pre-treated the cells with pan-caspase (Z-VAD) or a p38 (SB203580) inhibitor before adding TP-0903. We confirmed the inhibition of PARP and Caspase-3 cleavage only in the presence of Z-VAD, but not SB203580 (Supplementary Fig. S3A). Finally, we established that the TP-0903 induced apoptosis was not an early event, since no cleavage of PARP was observed 6 h post-treatment (Supplementary Fig. S3B). Together, these data uncover a role of TP0903 in the successful impairment of both proliferation and survival of NB cell lines.

samples, particularly in NB3 and SH-SY5Y cell lines. The expression of the same markers was not significantly impaired in IMR32 cell line (data not shown). Moreover, we performed an ethynyldeoxyuridine (EdU) assay and confirmed a concentration dependent reduction of DNA synthesis in the cells exposed to TP-0903 (Fig. 2D; Supplementary Table S4). Together, these results suggest that TP-0903 can successfully impair the proliferative capacities of AXL and AURKA expressing NB cells.

3.4. Treatment with TP-0903 blocks cell cycle and induces NB cell death 3.5. TP-0903 reduces colony formation and provokes an irreversible growth arrest of NB cells

In addition to attenuated cell proliferation, the cell cycle profiles underwent a significant alteration in the presence of TP-0903. We detected an important G2/M phase arrest in the TP-0903 treated samples (Fig. 3A and B; Supplementary Table S5A). Further on, in order to better define the effects of TP-0903 on mitotic entry, we measured the fraction of phospho-histone 3 (phospho-H3) signal in treated cells. Flow cytometry validation showed an evident fading out of the H3-phosphorylation in these cells, that was a clear sign of mitotic delay (Fig. 3C; Supplementary Table S5B). Moreover, to assess if TP-0903 induced cell death as well, we assessed the status of the apoptotic hallmarks, Caspase-3 and PARP (Chaitanya et al., 2010). We detected a marked cleavage of both proteins particularly in NB3 and IMR32 cells (Fig. 3D) which proved the activation of apoptosis. To verify whether the cell

To determine whether TP-0903 influenced the clonogenic capacities of NB cells, we looked for the differences in colony formation capacity between DMSO and TP-0903 pre-treated (for 6 h and 48 h) cells. The exposure to TP-0903 significantly reduced the number of colonies in each cell line (Fig. 4A and B; Supplementary Table S6). At this point we inquired whether the observed modulation of clonogenic potential was a result of an irreversible negative impact that TP-0903 had on cell viability. For that purpose, the cells were treated with several concentrations of TP-0903 and the drug was then washed-out (Fig. 4C). Results confirmed a dose-dependent response to TP-0903 (Supplementary Table S7). Interestingly, NB3 and IMR32 cells treated 439

European Journal of Pharmacology 818 (2018) 435–448

S. Aveic et al.

Fig. 3. Cell cycle and cell death analyses in TP-0903 treated samples. A) Percentage of cells in each phase of the cell cycle (G1, S and G2/M) is presented for a number of concentrations of TP-0903 and DMSO controls (mean ± S.D.). B) A representative cell cycle profile is shown for each cell line and two experimental settings: DMSO control and IC40 of TP0903 adapted for each cell type. 2n and 4n indicate DNA content during G1 and G2/M phase, respectively. C) Percent of cells positive for phospho-H3 signal is shown as mean ± S.D. of triplicate measures. ***P ≤ 0.001, **P ≤ 0.01 and *P ≤ 0.05. D). Activation of the principal apoptotic molecules, PARP and Caspase-3 were validated by WB for two time points, and two different TP-0903 concentrations. Level of beta TUBULIN was used as a loading control.

smaller in size (Fig. 4D; Supplementary Fig. S4). In the case of control SH-SY5Y and NB3 cells, the shape of the spheres was less regular when compared to IMR32, appearing smaller in size and with more pronounced sprouting. Instead, the spheres of TP-0903 treated cells did not boost the sprouting features and appeared much smaller (Supplementary Table S8). Finally, the IC50 values calculated in the 3D cell cultures (Supplementary Fig. S5A and S5B) were comparable to the values obtained in 2D systems (Supplementary Table S1), thus

with 80 nM and 100 nM of TP-0903 maintained a negative growth rate even after TP-0903 removal (Fig. 4C). Additionally, since the effectiveness of some anti-cancer agents diminishes in the presence of extracellular matrix - ECM (Weigelt et al., 2010), we tested whether matrigel presence might affect long-term treatment with TP-0903. We observed that the DMSO treated IMR32 cells grown on the matrigel scaffold created very regular spheres that showed a discrete sprouting effect, whereas the spheres derived from TP-0903 treated cells remained 440

European Journal of Pharmacology 818 (2018) 435–448

S. Aveic et al.

Fig. 4. Colony formation and growth impediment due to TP-0903 treatment. A) Capacity of NB cells to form the colonies after being pre-treated with TP-0903 for different time points (6 h, 48 h) is shown. Representative images for each cell line and experimental setting can be seen. B) A quantification of colony formation assay is presented as a mean number of colonies ± S.D. formed after 10–14 days of growth in semi-solid medium. C) Changes in the percentage of growth are presented after wash-out of TP-0903. Pre-treatment with TP-0903 was done with the increasing concentration of the drug for 24 h. The black arrow indicates the moment of drug wash-out, and fresh medium addition. D) Spheres dimension (mm) in the presence of ECM in addition to TP-0903 or DMSO is presented. A least 10 single spheres have been considered for dimension validation in each experimental replicate. P-values under or equal to 0.05 were considered statistically significant (***P ≤ 0.001, **P ≤ 0.01 and *P ≤ 0.05).

3.6. TP-0903 impairs NB cells migration

confirming the efficiency of TP-0903 also in a multi-cellular tumor sphere model (Santini and Rainaldi, 1999). These results are of particular interest since they indicate that NB cells grown in either 2D or 3D systems are highly sensitive to TP-0903, and that NB cells cannot recover their proliferating capacities after treatment, confirming a strong anti-tumor effect of the compound.

Suppression of sprout formation observed in TP-0903 treated cells suggested that the migration of these cells was probably impacted. Even though AXL is preferentially correlated with cell migration, there are evidences that indicate the importance of AURKA in the regulation of 441

European Journal of Pharmacology 818 (2018) 435–448

S. Aveic et al.

Fig. 5. Cell migration and F-actin distribution after TP-0903 addition. A) Wound healing assay was done for all three cell lines treated with sub-lethal concentrations of TP-0903 (left panel). The quantification of scratched area was done by means of ImageJ for the starting conditions (0 h) and following 24 h and 48 h, under 10× objectives. Results show mean ± S.D. of triplicate measures (right panel). ***P ≤ 0.001, **P ≤ 0.01 and *P ≤ 0.05. B) Rho-associated kinase (ROCK) is measured indirectly by evaluating the phosphorylation of MYPT1 at T696 by ROCK. The results of this immunoassay were obtained by measuring the absorbance for TP-0903 treated samples and comparing it to the co-respective DMSO control. C) Increasing concentrations of TP-0903 were used for 24 h and phalloidin staining (indicative for the F-actin) was done afterwards. DAPI served for nuclear counter-staining. Scale bars, 100 µm.

demonstrated a significant reduction (Fig. 5A, left panel) in the rate of scratch closure upon addition of TP-0903 with respect to the controls (Fig. 5A, graph bars; Supplementary Table S9). In accordance, Boyden

the same process (Chien et al., 2014). To examine how the migratory feature of NB cells with different levels of AXL and AURKA changed upon TP-0903 treatment we performed a wound-healing assay. We 442

European Journal of Pharmacology 818 (2018) 435–448

S. Aveic et al.

concentrations of TP-0903 used in our study reduced both, cell vitality and extravasation capabilities of NB cells.

chamber assay demonstrated that the migration of NB cells was depleted in the presence of TP-0903 compared to their DMSO treated counterparts (Supplementary Fig. S6A). However, since the observed phenomenon could occur due to an impaired cell cycle profile as well (Kubens et al., 2001), we included an additional test in order to unveil to which extent reduced migration was either AXL-dependent or G2 checkpoint block-dependent. For that purpose, we examined the activation status of the AXL downstream target Rho-associated protein kinase (ROCK). This protein belongs to Rho family of GTP-ases and is required for proper cell movement and adequate cytoskeletal dynamics (Bustelo et al., 2007). In treated SH-SY5Y cells we detected a significant decrease of ROCK activity (from 100% AU to 31 ± 15% AU; P = 0.002; Fig. 5B), which was accompanied by important rearrangements of the F-actin cytoskeleton (Koorstra et al., 2009). The phalloidin staining demonstrated less dramatic changes in F-actin distribution in NB3 (Fig. 5C), whereas no particular differences were observed in treated IMR32 cells (Supplementary Fig. S6B). Taken together, these data imply that TP-0903 is very effective in hampering tumor cells motility and migration by inhibiting AXL (SH-SY5Y and NB3), whereas impairment of these processes is likely a consequence of increased cell death and cell cycle block in cell lines with less or no AXL (NB3 and IMR32, respectively).

3.8. TP-0903 treatment leads to genotoxic stress activation To understand which intra-cellular pathways were preferentially activated upon addition of TP-0903, we carried out the phospho-kinase array. We identified 21 out of 43 (49%) kinase phosphorylation sites to be changing more than 20% (corresponding to 18 different proteins; Supplementary Table S11) in TP-0903 treated sample (Fig. 7A). Among them, 6 kinases showed a change greater than 50%: P53 S392 (119%), CHK-2 T68 (72%), PLC-γ1 Y783 (68%), WNK1 T60 (57%), PRAS40 T246 (58%), and PYK2 Y402 (76%; Fig. 7B). Additional pathway network analysis showed how these proteins were related to each other (Fig. 7C). The observed decrease in the expression of PLC-γ1, PYK2 and WNK1 kinases is in line with the detected motility decline and with dynamic changes of F-actin profile (Dbouk et al., 2014; Ostergaard and Lysechko, 2005; Piccolo et al., 2002). On the contrary, the observed activation of P53 and CHK-2 proteins were indicative of genotoxic stress activation (Cox and Meek, 2010; Xu et al., 2002). In order to extend the observation that TP-0903 stimulated DNA damage in SHSY5Y cells, we evaluated the level of histone H2AX phosphorylation at S139 (γH2AX), and detected an increase of γH2AX expression in TP0903 treated samples versus controls (Fig. 7D). The same observations were made for NB3, while IMR32 cell line showed a less dramatic change at this residue (Supplementary Fig. S9A). Since γH2AX is a dual marker, activated during G2/M block and DNA fragmentation, we also analyzed and confirmed, the presence of double strand brakes (DSB) in the cells exposed to TP-0903 (Fig. 7E). Finally, we looked for one of the AXL downstream targets, ERK1/2, and verified its decrease after TP0903 treatment by both phospho-kinase array and WB analysis (Fig. 7F). Similar results were found for NB3 and IMR32 cell lines (Supplementary Fig. S9B). Altogether, protein array data indicated that the main TP-0903 effects were exerted on proteins involved in cell migration, proliferation and response to genotoxic stress, sustaining previously obtained in vitro and in vivo data.

3.7. TP-0903 occludes NB cells intravasation in vitro and in vivo To verify if TP-0903 can hinder cells intravasation as one of the critical steps involved in cancer cell dissemination (Nguyen et al., 2009), we adopted two experimental approaches. Firstly, we used HUVEC as a barrier for cell migration to test the potential of NB cells to invade it. By using 0–10% serum gradient, we compared the differences in the intensity of the fluorescent signal (sign of migrated cells) in the TP-0903 versus DMSO pre-treated samples, and observed its clear decline (Supplementary Fig. S7A). Even though statistical significance was obtained only for NB3 cell line, a trend in declined signaling was observed for the remaining two cell lines, SH-SY5Y in particular (Supplementary Table S10), indicating that NB cells invasiveness was reduced by TP-0903 although to a different extent. Secondly, to investigate whether TP-0903 affects NB cells migration in vivo, the zebrafish xenograft model was exploited. The NB3 and SH-SY5Y cells were pre-treated for 2 days with TP-0903 (IC40), subsequently labeled with fluorescent DiI and injected into the Tg(fli1: EGFP) embryos (Supplementary Fig. S7B, upper image). A direct injection into the blood circulation allows the cells to be distributed throughout the organism, mimicking the advanced stages of metastasis spreading (He et al., 2012). Upon transplantation, tumor cells immediately disseminated through the blood circulation and arrested within the head, heart and venous plexus (Supplementary Fig. S7B, lower image). Some cells penetrated into the smaller optic veins and the inter-segmental vessels (Supplementary Fig. S7C). Both NB3 and SH-SY5Y lines began the extravasation 1 day post implantation (dpi), but failed to invade the surrounding tissues and to form new sites of tumor formation. From 1 to 4 dpi we observed a progressive depletion of the fluorescent signal of TP-0903 treated NB3 cells (Fig. 6A), which was indicative of cell number regression. Thus, the percentage of embryos maintaining treated NB3 cells into the blood flow decreased significantly (90% at 1 dpi and 33% at 4 dpi) compared to the control group (100% at 1 dpi and 88% at 4 dpi; Fig. 6B; P < 0.0001). In addition, control NB3 cells progressively extravasated, whereas TP-0903 treated cells displayed unchanged levels of extravasation events (Fig. 6C; P = 0.007). The percentage of embryos retaining SH-SY5Y cells into the blood circulation diminished after treatment with TP-0903 although the results did not reach a statistically significant level (Supplementary Fig. S8A and S8B; P=0.09), and the extravasation was less dramatically modified compared to NB3 cell line (Supplementary Fig. S8C; P=0.15). Taken together, from these results we could conclude that the sub-toxic

3.9. TP-0903 works in synergy with chemotherapy drugs After having confirmed the cytotoxic effects of TP-0903, we performed a combo-drug screening in order to understand how TP-0903 behaved when used together with standard chemotherapy drugs. We simultaneously administered TP-0903 and all-trans retinoic acid (ATRA), cisplatin or VP16 and found that TP-0903 did not have negative effects on cytotoxicity with any of them. More precisely, we observed that TP-0903 worked in synergy with VP16 and cisplatin particularly when 1:100 molar combination was used (CI < 1; Supplementary Table S12A). In the case of TP-0903 and ATRA combination, the results were very encouraging, showing a synergistic action between the two drugs under each molar ratio (CI < 1 for 1:1; 1:10 and 1:100; Supplementary Table S12B). These analyses sustained the possible use of TP-0903 along with other chemotherapy drugs. 3.10. TP-0903/ATRA combination is effective against CD133+ cells We then repeated the TP-0903/ATRA combination in NB3 and IMR32 cell lines, confirming the synergy for NB3, but not for IMR32 (Supplementary Table S13). We speculated that the observed discrepancy in the combo-treatment results was due to the existence of more than one NB cell population, such as CD133, showing a distinct grade of differentiation (Hämmerle et al., 2013). As shown by others (Khalil et al., 2016), and confirmed by ourselves (data not shown), the analysis of the CD133 expression separated the cell lines in two groups: CD133 positive (CD133+) (SH-SY5Y and NB3) and cells with an almost undetectable CD133+ sub-population of transformed neuroblasts

443

European Journal of Pharmacology 818 (2018) 435–448

S. Aveic et al.

Fig. 6. Extravasation of NB3 in a zebrafish xenotransplantation model after TP-0903 treatment. A) Fluorescent microscopy images of the trunk region of Tg(fli1:GFP) zebrafish injected with NB3 cells pre-treated as indicated. Each group contained the DMSO treated cells as a control. The percentage of animals (n = 104 embryos) maintaining TP-0903-treated NB3 cells decreased over time (c, d) whereas extravasation of control NB3 cells (a, b) remained detectable into the zebrafish host (n = 85 embryos). Extravasating cells were still in contact with the external wall of the endothelium, as reported in the transverse view in b’’. White boxes indicate position of the higher magnification in a’, b’, c’ and d’. White arrowheads in a’ and c’ indicate extravasated NB3 cells. Scale bars, 100 µm. B) Graphical representation of the percentage of embryonic zebrafish displaying visible NB3 cells as a function of time (1–4 dpi; P < 0.0001). C) Graphical representation of the percentage of embryonic zebrafish displaying extravasating NB3 cells. Treatment of NB3 with TP-0903 showed a significant reduction of extravasating cells at 4 dpi (**P < 0.01). Graph represent the mean ± S.D. dpi - days post implantation.

whereas the expression of BNIP3, used as a non stem-control marker, did not change significantly (Supplementary Table S14). These findings suggested that TP-0903 had a stronger toxic impact on the population of cells that did not express stem-specific markers. Since ATRA was proven to be an effective agent against CD133+cells (Takenobu et al., 2011), we combined it with TP-0903 and re-evaluated the percentage of CD133+ cells. The TP-0903/ATRA combination resulted to be more effective against CD133+ cells than the TP-0903 mono-treatment, since it caused a marked decrease of detected CD133+ signal (Fig. 8B). These findings gave an additional explanation to the previously observed synergistic action between TP-0903/ATRA combo-treatment, sustaining their efficacy in the elimination of different NB cells sub-populations.

(IMR32). As a cancer stem cell marker, CD133 expression was related to decreased drug efficacy and impaired differentiating potential in several types of tumors including NB (Walton et al., 2004). At first, to investigate how the population of CD133+cells changed after treatment with toxic doses of TP-0903 (IC60, IC70 and IC80), we performed flow cytometry on the SH-SY5Y cell line. The adopted concentration range of TP-0903 caused a decrease of total living cells (sum of CD133- and CD133+) in a dose dependent way (Fig. 8A; solid line). Furthermore, increasing fractions of dead cells were observed (DMSO = 16.3 ± 8.6%; IC40 = 51 ± 7%; IC60 = 62.3 ± 14.5%; IC80 = 68.3 ± 15.1%; P < 0.01; dashed line). When the remaining still living cells were considered separately, the percentage of CD133+ cells increased as the alive cells decreased due to the higher toxicity of TP0903 (Fig. 8A; pie chart). Other markers of stemness in NB cells, including CD44, Nestin and BMI-1 (Hämmerle et al., 2013; Kamijo, 2012), also increased significantly upon treatment with IC80 of TP-0903,

4. Discussion NB is a tumor of embryonal origin that affects the sympathetic 444

European Journal of Pharmacology 818 (2018) 435–448

S. Aveic et al.

Fig. 7. Phospho-kinase proteins profile. A) Long and short exposure of membranes composed of two parts (part A and B) is presented. Proteins extracted from SH-SY5Y cells from two experimental conditions are reported: DMSO and TP-0903 (100 nM; TP100), 24 h post-treatment. Numbers indicate corresponding proteins lined in Fig. 7B. Colors refer to increase (red) and decrease (blue) of evaluated intensity acquired after treatment. B) Percent of change (arbitrary units - AU) was calculated for TP-0903 samples with respect to DMSO. Results from Fig. 7A are shown here as mean ± S.D. of experimental duplicates. C) Pathway network analysis of two TP-0903 targets (AXL and AURKA) and related up- and down-regulated downstream molecules from Fig. 7B is shown. Red/blue color intensities are proportional to the amount of up/down regulation; arrows width is proportional to the mean value of disregulation of the connected proteins. Connecting arrows include different types of interactions between proteins: neighbor of-; controls transport of-; in complex with-; controls phosphorylation of-; interacts with-; controls expression of- (data source: www.pathwaycommons.org). D) Levels of H2AX protein were validated by WB 24 h post-treatment with IC20 and IC40 of TP-0903. Levels of beta TUBULIN were used as loading controls. E) Validation of DSB due to TP-0903 treatment was performed by checking for the DNA smear appearance on agarose gel. Ladder (L) of 50 base pairs (bp) served for validation of DNA fragments length. F) The WB analysis was done to test ERK1/2 expression in addition to TP-0903 (IC20 and IC40). The level of beta TUBULIN was used as loading control. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

2008). Hence, AXL targeting has become a challenge in several cancer types (Leconet et al., 2014). Lately, a highly efficient anti-AXL compound TP-0903, that has been proposed in pre-clinical studies involving different types of cancers, gave very promising anti-tumor effects (Sinha et al., 2015). These results prompted for a phase I clinical trial involving the patients with advanced solid tumors (https:// clinicaltrials.gov/ct2/show/NCT02729298). In the case of NB, this drug has not been considered yet, neither at pre-clinical nor at the clinical level. However, considering that TP-0903 impacts the

nervous system. The majority of cases are diagnosed during childhood and at diagnosis these patients can be classified as low, intermediate or high-risk (HR), depending on the clinical and biological features of the tumor. Protein kinases are frequently deregulated in solid tumors, including NB. The association between AXL, a receptor tyrosine kinase (RTK) that belongs to the TAM family of proteins, and NB has been described recently (Debruyne et al., 2015). This RTK gains increasing attention since it regulates crucial biological pathways involved in tumor cell survival, proliferation, motility and invasion (Linger et al.,

445

European Journal of Pharmacology 818 (2018) 435–448

S. Aveic et al.

Fig. 8. Combined therapeutic strategy of TP-0903 with ATRA. A) Flow cytometry was used to measure the percentage of alive (solid line) and dead cells (dashed line) within the total cell fraction (sum of CD133− and CD133+) 72 h post-treatment for 3 concentrations of TP-0903 (IC60, IC70, IC80). Pie chart shows the percentage of CD133+ cells within the fraction of alive cells. B) The percentage of CD133+ cells is shown for DMSO, TP-0903, ATRA and TP0903/ATRA combination. A decrease in CD133+ signal in the portion of alive cells can be observed in combined versus ATRA mono-treatment. Percentage of CD133+ cells was presented as mean ± S.D. of triplicates, *P = 0.05.

One of the processes that emerged from the studies of the mechanisms responsible for the resistance to administered RTK-inhibitors is autophagy. A cytoprotective feature of autophagy is frequently used by cancer cells in order to avoid stress stimuli and survive toxic drug inputs (Aveic et al., 2016; Mokarram et al., 2017). However, we did not notice significant autophagy activation in NB cells upon TP-0903 treatment (data not shown), which additionally sustains its cytotoxic potential against NB cells. We are aware that cancer cells are very efficient in finding alternatives to death. This is particularly evident in the tumors that are composed of heterogeneous cell populations, among which NB, where certain sub-populations are less sensitive to the drug cytotoxicity. One of such populations include CD133+ cells that define a stem compartment within the tumor and are frequently associated with drug resistant phenotypes (Hanahan and Weinberg, 2011). Therefore, it is very important to know whether an anti-tumor drug is effective against this cell population or not. Sometimes it is essential to combine different therapeutics in order to assure the eradication of all tumor cells, including those with self-renewal (stem) characteristics (Likus et al., 2016). Otherwise, the remaining, even minimal, number of tumor stem cells could cause a tumor re-growth or the appearance of the metastatic sites. Retinoic acid is effective against cells with more immature phenotype stimulating their differentiation and is used during post-consolidation therapy in HR-NB patients (Peinemann et al., 2015). Our results clarify why ATRA still remains a drug of election for recurrent disease and why many in vitro studies still seek for the possibility of improving ATRA efficiency in cancer treatments (Hämmerle et al., 2013). The ability of TP-0903 to work in synergy with ATRA is encouraging and sustains this combination for further testing. Moreover, the synergy might be explained by the pro-differentiating effects of ATRA which renders more immature cells susceptible to TP-0903 treatment. This result is partially explained by the recent report in which TP-0903 was used in zebrafish model to study migration of neural crest cells (Jimenez et al., 2016), where the authors described the increase of the retinoic acid biosynthesis upon treatment. Hence, the use of TP-0903 together with other chemotherapeutic agents is promising, but suggests that the ratio of the administered drugs can be crucial in obtaining their synergistic action, as suggested previously (Chou, 2007). In conclusion, our results show that TP-0903 works through several targets/mechanisms favored by NB cells, leading to a significant impairment of their fitness. Although the use of TP-0903 as an addendum to already adopted therapies for patients with NB and poor prognosis is still to be defined, our results clearly show that it is a very interesting compound to be considered for additional (pre)clinical testing in NB.

biological processes strongly related to NB development, we thought that this drug might be efficient against NB cells as well. Moreover, the knowledge that TP-0903 can also target AURKA, one of the strongest negative prognostic factors in NB (Ramani et al., 2015), strengthen our choice for its pre-clinical testing. Our results confirmed the impressive efficacy of TP-0903 in impairing the beforehand mentioned biological processes already at nMconcentrations. This efficacy was observed in both 2D and 3D cell cultures. Additionally, we proved that TP-0903 was responsible for an increased cell death and for their reduced clonogenic potential, independently from their mutational background. Crucially, the effects of TP-0903 were irreversible. Moreover, we observed mitotic arrest in treated cells and apoptosis triggering which were accompanied by P53 and CHK-2 proteins and genotoxic stress activation (Hirao, 2000). On the other side, reported changes in the expression of kinases, such as PRAS40 and ERK1/2, supported the observed decrease of cell growth (Wang et al., 2012). In addition, TP-0903 significantly influenced the formation of neurospheres even in the presence of extra-cellular matrix support. This finding has significant relevance for TP-0903’s effectiveness since not all drugs work well under more complex 3D environmental conditions (Hurst et al., 2005). We confirmed both, in vitro and in vivo, that TP-0903 pre-treatment reduced NB cells’ invasion. Taking the advantages of zebrafish embryos with fluorescently marked vessels, we were able to follow spatio-temporally the migratory capacities of pre-treated NB cells. These approaches allowed us to observe that not all biological processes were affected in equal proportions in cell lines with various mutational make-ups. Accordingly, in IMR32, we reported strong negative effects on cell viability and survival caused by TP-0903, but a less striking impact was done on their drifting capacities. On the other side, migrating and invasive properties of SH-SY5Y and NB3 treated cells were highly dependent on the undisturbed activity of AXL/Rho-associated protein kinase (ROCK) axis. In fact, TP-0903 influenced both features in these cells, and provoked an important variation in actin filaments remodeling. The observed dissimilarities in the action of TP-0903 on different NB cell types could be explained by the fact that TP-0903 is a drug with more than one molecular target. Therefore it is plausible that in different cell types it can act through diverse (kinase) targets (Mollard et al., 2011). As we summarized in the Supplementary Fig. S10, it is likely that TP-0903 affects several kinases that are heterogeneously expressed in different NB cell types. The aforementioned concept supports the idea that “dirty drugs” might be useful and more effective as an approach for targeted therapy in NB (Fojo, 2008). According to this view, targeting of multiple kinases might turn off the parallel inputs driven by several RTK simultaneously, which stand behind the decreased efficiency, and/or emerging resistance to RTK-inhibitors (Aveic and Tonini, 2016). This can be additionally strengthen by the fact that multi-signal targeting could guarantee the impairment of both, growth and survival of tumor cells, making the administered therapy more effective (Ham et al., 2016).

Acknowledgements This work was supported by funds from Fondazione Italiana per la Lotta al Neuroblastoma. EP has a post-doc fellowship grant from the Italian Association for Cancer Research (AIRC). 446

European Journal of Pharmacology 818 (2018) 435–448

S. Aveic et al.

Author contributions

Hanahan, D., Weinberg, R.A., 2011. Hallmarks of cancer: the next generation. Cell. http://dx.doi.org/10.1016/j.cell.2011.02.013. He, S., Lamers, G.E.M., Beenakker, J.W.M., Cui, C., Ghotra, V.P.S., Danen, E.H.J., Meijer, A.H., Spaink, H.P., Snaar-Jagalska, B.E., 2012. Neutrophil-mediated experimental metastasis is enhanced by VEGFR inhibition in a zebrafish xenograft model. J. Pathol. 227, 431–445. http://dx.doi.org/10.1002/path.4013. Hirao, a., 2000. DNA damage-induced activation of p53 by the checkpoint kinase Chk2. Science 80–287, 1824–1827. http://dx.doi.org/10.1126/science.287.5459.1824. Hurst, R.E., Kamat, C.D., Kyker, K.D., Green, D.E., Ihnat, M.A., 2005. A novel multidrug resistance phenotype of bladder tumor cells grown on Matrigel or SIS gel. Cancer Lett. 217, 171–180. http://dx.doi.org/10.1016/j.canlet.2004.07.043. Jimenez, L., Wang, J., Morrison, M.A., Whatcott, C., Soh, K.K., Warner, S., Bearss, D., Jette, C.A., Stewart, R.A., 2016. Phenotypic chemical screening using a zebrafish neural crest EMT reporter identifies retinoic acid as an inhibitor of epithelial morphogenesis. Dis. Models Mech. 9, 389–400. http://dx.doi.org/10.1242/dmm.021790. Kamijo, T., 2012. Role of stemness-related molecules in neuroblastoma. Pediatr. Res. 71, 511–515. http://dx.doi.org/10.1038/pr.2011.54. Kasibhatla, S., Amarante-Mendes, G.P., Finucane, D., Brunner, T., Bossy-Wetzel, E., Green, D.R., 2006. Analysis of DNA fragmentation using agarose gel electrophoresis. CSH Protoc. 2006http://dx.doi.org/10.1101/pdb.prot4429. (pdb.prot4429-). Khalil, M., Hraběta, J., Groh, T., Procházka, P., Doktorová, H., Al, E., 2016. Valproic acid increases CD133 positive cells that show low sensitivity to cytostatics in neuroblastoma. PLoS One 11, e0162916. Kimmel, C.B.C.B., Ballard, W.W., Kimmel, S.R., Ullmann, B., Schilling, T.F., 1995. Stages of embryonic development of the zebrafish. Dev. Dyn. 10, 253–310. http://dx.doi. org/10.1002/aja.1002030302. Koorstra, J.B.M., Karikari, C.A., Feldmann, G., Bisht, S., Rojas, P.L., Offerhaus, G.J.A., Alvarez, H., Maitra, A., 2009. The Axl receptor tyrosine kinase confers an adverse prognostic influence in pancreatic cancer and represents a new therapeutic target. Cancer Biol. Ther. 8. http://dx.doi.org/10.4161/cbt.8.7.7923. Korshunov, V.A., 2012. Axl-dependent signaling: a clinical update. Clin. Sci. (Lond.). 122, 361–368. http://dx.doi.org/10.1042/CS20110411. Kubens, B.S., Niggemann, B., Zänker, K.S., 2001. Prevention of entrance into G2 cell cycle phase by mimosine decreases locomotion of cells from the tumor cell line SW480. Cancer Lett. 162. http://dx.doi.org/10.1016/S0304-3835(00)00651-0. Leconet, W., Larbouret, C., Chardès, T., Thomas, G., Neiveyans, M., Busson, M., Jarlier, M., Radosevic-Robin, N., Pugnière, M., Bernex, F., Penault-Llorca, F., Pasquet, J.-M., Pèlegrin, A., Robert, B., 2014. Preclinical validation of AXL receptor as a target for antibody-based pancreatic cancer immunotherapy. Oncogene 33, 5405–5414. http:// dx.doi.org/10.1038/onc.2013.487. Li, L.-C., Dahiya, R., 2002. MethPrimer: designing primers for methylation PCRs. Bioinformatics 18, 1427–1431. http://dx.doi.org/10.1016/0014-5793(85)80078-8. Likus, W., Siemianowicz, K., Bieńk, K., Pakuła, M., Pathak, H., Dutta, C., Wang, Q., Shojaei, S., Assaraf, Y.G., Ghavami, S., Cies̈lar-Pobuda, A., Łos, M.J., 2016. Could drugs inhibiting the mevalonate pathway also target cancer stem cells? Drug Resist. Updat. 25, 13–25. http://dx.doi.org/10.1016/j.drup.2016.02.001. Linger, R.M.A., Keating, A.K., Earp, H.S., Graham, D.K., 2008. TAM receptor tyrosine kinases: biologic functions, signaling, and potential therapeutic targeting in human cancer. Adv. Cancer Res. http://dx.doi.org/10.1016/S0065-230X(08)00002-X. Lonergan, G.J., Schwab, C.M., Suarez, E.S., Carlson, C.L., 2003. Neuroblastoma, ganglioneuroblastoma, and ganglioneuroma: radiologic-pathologic correlation. Radiographics 22, 911–934. http://dx.doi.org/10.1148/radiographics.22.4. g02jl15911. Luksch, R., Castellani, M.R., Collini, P., De Bernardi, B., Conte, M., Gambini, C., Gandola, L., Garaventa, A., Biasoni, D., Podda, M., Sementa, A.R., Gatta, G., Tonini, G.P., 2016. Neuroblastoma (Peripheral neuroblastic tumours). Crit. Rev. Oncol. Hematol. 107, 163–181. http://dx.doi.org/10.1016/j.critrevonc.2016.10.001. Maris, J.M., Hogarty, M.D., Bagatell, R., Cohn, S.L., 2007. Neuroblastoma. Lancet. http:// dx.doi.org/10.1016/S0140-6736(07)60983-0. Mokarram, P., Albokashy, M., Zarghooni, M., Moosavi, M.A., Sepehri, Z., Chen, Q.M., Hudecki, A., Sargazi, A., Alizadeh, J., Moghadam, A.R., Hashemi, M., Movassagh, H., Klonisch, T., Owji, A.A., Łos, M.J., Ghavami, S., 2017. New frontiers in the treatment of colorectal cancer: autophagy and the unfolded protein response as promising targets. Autophagy. http://dx.doi.org/10.1080/15548627.2017.1290751. Mollard, A., Warner, S.L., Call, L.T., Wade, M.L., Bearss, J.J., Verma, A., Sharma, S., Vankayalapati, H., Bearss, D.J., 2011. Design, synthesis, and biological evaluation of a series of novel AXL kinase inhibitors. ACS Med. Chem. Lett. 2, 907–912. http://dx. doi.org/10.1021/ml200198x. Myers, S.H., Brunton, V.G., Unciti-Broceta, A., 2016. AXL inhibitors in cancer: a medicinal chemistry perspective. J. Med. Chem. http://dx.doi.org/10.1021/acs.jmedchem. 5b01273. Neubauer, A., Fiebeler, A., Graham, D.K., O’Bryan, J.P., Schmidt, C.A., Barckow, P., Serke, S., Siegert, W., Snodgrass, H.R., Huhn, D., 1994. Expression of axl, a transforming receptor tyrosine kinase, in normal and malignant hematopoiesis. Blood 84, 1931–1941. Nguyen, D.X., Bos, P.D., Massagué, J., 2009. Metastasis: from dissemination to organspecific colonization. Nat. Rev. Cancer 9, 274–284. http://dx.doi.org/10.1038/ nrc2622. Ostergaard, H.L., Lysechko, T.L., 2005. Focal adhesion kinase-related protein tyrosine kinase Pyk2 in T-cell activation and function. Immunol. Res. 31, 267–282. http://dx. doi.org/10.1385/IR:31:3:267. Otto, T., Horn, S., Brockmann, M., Eilers, U., Schüttrumpf, L., Popov, N., Kenney, A.M., Schulte, J.H., Beijersbergen, R., Christiansen, H., Berwanger, B., Eilers, M., 2009. Stabilization of N-Myc is a critical function of aurora a in human neuroblastoma. Cancer Cell 15, 67–78. http://dx.doi.org/10.1016/j.ccr.2008.12.005. Peifer, M., Hertwig, F., Roels, F., Dreidax, D., Gartlgruber, M., Menon, R., Krämer, A.,

Participated in research design: SA, DC, EP; Conducted experiments: SA, DB, DC, MP, VS, EM, GV, EP; Contributed new reagents or analytic tools: AQ, GB, GPT; Performed data analysis: SA, CZ, DC; Wrote or contribute in the writing of the manuscript: SA, DC, EP, MP, DB, CZ, GV; VS, EM, AQ, GB, GPT. All the authors have read, revised and approved the manuscript. Conflict of interest statement No potential conflicts of interest were disclosed by the authors. Appendix A. Supporting information Supplementary data associated with this article can be found in the online version at http://dx.doi.org/10.1016/j.ejphar.2017.11.016. References Aveic, S., Pantile, M., Seydel, A., Esposito, M.R., Zanon, C., Li, G., Tonini, G.P., 2016. Combating autophagy is a strategy to increase cytotoxic effects of novel ALK inhibitor entrectinib in neuroblastoma cells. Oncotarget 7. http://dx.doi.org/10.18632/ oncotarget.6778. Aveic, S., Tonini, G.P., 2016. Resistance to receptor tyrosine kinase inhibitors in solid tumors: can we improve the cancer fighting strategy by blocking autophagy? Cancer Cell Int. 16. http://dx.doi.org/10.1186/s12935-016-0341-2. Bellini, A., Bernard, V., Leroy, Q., Frio, T.R., Pierron, G., Combaret, V., Lapouble, E., Clement, N., Rubie, H., Thebaud, E., Chastagner, P., Defachelles, A.S., Bergeron, C., Buchbinder, N., Taque, S., Auvrignon, A., Valteau-Couanet, D., Michon, J., JanoueixLerosey, I., Delattre, O., Schleiermacher, G., 2015. Deep sequencing reveals occurrence of subclonal ALK mutations in neuroblastoma at diagnosis. Clin. Cancer Res. 21, 4913–4921. http://dx.doi.org/10.1158/1078-0432.CCR-15-0423. Brodeur, G.M., 2003. Neuroblastoma: biological insights into a clinical enigma. Nat. Rev. Cancer 3, 203–216. http://dx.doi.org/10.1038/nrc1014. Bustelo, X.R., Sauzeau, V., Berenjeno, I.M., 2007. GTP-binding proteins of the Rho/Rac family: regulation, effectors and functions in vivo. BioEssays. http://dx.doi.org/10. 1002/bies.20558. Carpentier, G., 2008. Dot Blot Analyzer: Software development using the macro language of ImageJ. In: Proceedings of the ImageJ User and Developer Conference, p. ISBN 2919941-06-2. Chaitanya, G.V., Steven, A.J., Babu, P.P., 2010. PARP-1 cleavage fragments: signatures of cell-death proteases in neurodegeneration. Cell Commun. Signal 8, 31. http://dx.doi. org/10.1186/1478-811X-8-31. Chien, C.-Y., Tsai, H.-T., Su, L.-J., Chuang, H.-C., Shiu, L.-Y., Huang, C.-C., Fang, F.-M., Yu, C.-C., Su, H.-T., Chen, C.-H., 2014. Aurora-A signaling is activated in advanced stage of squamous cell carcinoma of head and neck cancer and requires osteopontin to stimulate invasive behavior. Oncotarget 5, 2243–2262. Chou, T.-C., 2007. Correction to “Theoretical basis, experimental design, and computerized simulation of synergism and antagonism in drug-combination studies. Pharmacol. Rev. 59, 124 (LP-124). Cox, M.L., Meek, D.W., 2010. Phosphorylation of serine 392 in p53 is a common and integral event during p53 induction by diverse stimuli. Cell. Signal. 22, 564–571. http://dx.doi.org/10.1016/j.cellsig.2009.11.014. Dbouk, H.A., Weil, L.M., Perera, G.K.S., Dellinger, M.T., Pearson, G., Brekken, R.A., Cobb, M.H., 2014. Actions of the protein kinase WNK1 on endothelial cells are differentially mediated by its substrate kinases OSR1 and SPAK. Proc. Natl. Acad. Sci. USA 111, 15999–16004. http://dx.doi.org/10.1073/pnas.1419057111. Debruyne, D.N., Bhatnagar, N., Sharma, B., Luther, W., Moore, N.F., Cheung, N.-K., Gray, N.S., George, R.E., 2015. ALK inhibitor resistance in ALKF1174L-driven neuroblastoma is associated with AXL activation and induction of EMT. Oncogene 35, 1–11. http://dx.doi.org/10.1038/onc.2015.434. Esposito, M.R., Aveic, S., Seydel, A., Tonini, G.P., 2017. Neuroblastoma treatment in the post-genomic era. J. Biomed. Sci. 24. http://dx.doi.org/10.1186/s12929-017-0319-y. Fojo, T., 2008. Novel therapies for cancer: why dirty might be better? Oncologist 13, 277–283. Graham, D.K., DeRyckere, D., Davies, K.D., Earp, H.S., 2014. The TAM family: phosphatidylserine-sensing receptor tyrosine kinases gone awry in cancer. Nat. Rev. Cancer 14, 769–785. http://dx.doi.org/10.1038/nrc3847. Ham, J., Costa, C., Sano, R., Lochmann, T.L., Sennott, E.M., Patel, N.U., Dastur, A., Gomez-Caraballo, M., Krytska, K., Hata, A.N., Floros, K.V., Hughes, M.T., Jakubik, C.T., Heisey, D.A.R., Ferrell, J.T., Bristol, M.L., March, R.J., Yates, C., Hicks, M.A., Nakajima, W., Gowda, M., Windle, B.E., Dozmorov, M.G., Garnett, M.J., McDermott, U., Harada, H., Taylor, S.M., Morgan, I.M., Benes, C.H., Engelman, J.A., Mossé, Y.P., Faber, A.C., 2016. Exploitation of the apoptosis-primed state of MYCN-amplified neuroblastoma to develop a potent and specific targeted therapy combination. Cancer Cell 29, 159–172. http://dx.doi.org/10.1016/j.ccell.2016.01.002. Hämmerle, B., Yañez, Y., Palanca, S., Cañete, A., Burks, D.J., Castel, V., Font de Mora, J., 2013. Targeting neuroblastoma stem cells with retinoic acid and proteasome inhibitor. PLoS One 8. http://dx.doi.org/10.1371/journal.pone.0076761.

447

European Journal of Pharmacology 818 (2018) 435–448

S. Aveic et al.

neuronal cells by human adenoviruses and the origin of HEK 293 cells. FASEB J. 16, 869–871. http://dx.doi.org/10.1096/fj.01-0995fje/r01-0995fje. (pii). Sinha, S., Boysen, J., Nelson, M., Secreto, C., Warner, S.L., Bearss, D.J., Lesnick, C., Shanafelt, T.D., Kay, N.E., Ghosh, A.K., 2015. Targeted Axl inhibition primes chronic lymphocytic leukemia B cells to apoptosis and shows synergistic/additive effects in combination with BTK inhibitors. Clin. Cancer Res. 21, 2115–2126. http://dx.doi. org/10.1158/1078-0432.CCR-14-1892. Soh, K.K., Kim, W., Lee, Y.S., Peterson, P., Siddiqui-Jain, A., Warner, S.L., Bearss, D.J., Whatcott, C.J., 2016. AXL inhibition leads to a reversal of a mesenchymal phenotype sensitizing cancer cells to targeted agents and .immuno-oncology therapies, In: 10 Proceedings of the 7th Annual Meeting of the American Association for Cancer Research. p. AACR; Cancer Res., 2016, 76. 〈http://dx.doi.org/10.1158/1538-7445. AM2016-235〉. Stoletov, K., Montel, V., Lester, R.D., Gonias, S.L., Klemke, R., 2007. High-resolution imaging of the dynamic tumor cell vascular interface in transparent zebrafish. Proc. Natl. Acad. Sci. USA 104, 17406–17411. http://dx.doi.org/10.1073/pnas. 0703446104. Takenobu, H., Shimozato, O., Nakamura, T., Ochiai, H., Yamaguchi, Y., Ohira, M., Nakagawara, a., Kamijo, T., 2011. CD133 suppresses neuroblastoma cell differentiation via signal pathway modification. Oncogene 30, 97–105. http://dx.doi.org/ 10.1038/onc.2010.383. Verma, A., Warner, S.L., Vankayalapati, H., Bearss, D.J., Sharma, S., 2011. Targeting Axl and Mer kinases in cancer. Mol. Cancer Ther. 10, 1763–1773. http://dx.doi.org/10. 1158/1535-7163.MCT-11-0116. Walton, J.D., Kattan, D.R., Thomas, S.K., Spengler, B.A., Guo, H.-F., Biedler, J.L., Cheung, N.-K.V., Ross, R.A., 2004. Characteristics of stem cells from human neuroblastoma cell lines and in tumors. Neoplasia 6, 838–845. http://dx.doi.org/10.1593/neo. 04310. Wang, H., Zhang, Q., Wen, Q., Zheng, Y., Philip, L., Jiang, H., Lin, J., Zheng, W.H., 2012. Proline-rich Akt substrate of 40kDa (PRAS40): a novel downstream target of PI3k/ Akt signaling pathway. Cell. Signal. http://dx.doi.org/10.1016/j.cellsig.2011.08. 010. Weigelt, B., Lo, A.T., Park, C.C., Gray, J.W., Bissell, M.J., 2010. HER2 signaling pathway activation and response of breast cancer cells to HER2-targeting agents is dependent strongly on the 3D microenvironment. Breast Cancer Res. Treat. 122, 35–43. http:// dx.doi.org/10.1007/s10549-009-0502-2. Xu, X., Tsvetkov, L.M., Stern, D.F., 2002. Chk2 activation and phosphorylation-dependent oligomerization. Mol. Cell. Biol. 22, 4419–4432. http://dx.doi.org/10.1128/MCB.22. 12.4419-4432.2002. Zage, P.E., Nolo, R., Fang, W., Stewart, J., Garcia-Manero, G., Zweidler-McKay, P.A., 2012. Notch pathway activation induces neuroblastoma tumor cell growth arrest. Pediatr. Blood Cancer 58, 682–689. http://dx.doi.org/10.1002/pbc.23202.

Roncaioli, J.L., Sand, F., Heuckmann, J.M., Ikram, F., Schmidt, R., Ackermann, S., Engesser, A., Kahlert, Y., Vogel, W., Altmüller, J., Nürnberg, P., Thierry-Mieg, J., Thierry-Mieg, D., Mariappan, A., Heynck, S., Mariotti, E., Henrich, K.-O., Gloeckner, C., Bosco, G., Leuschner, I., Schweiger, M.R., Savelyeva, L., Watkins, S.C., Shao, C., Bell, E., Höfer, T., Achter, V., Lang, U., Theissen, J., Volland, R., Saadati, M., Eggert, A., de Wilde, B., Berthold, F., Peng, Z., Zhao, C., Shi, L., Ortmann, M., Büttner, R., Perner, S., Hero, B., Schramm, A., Schulte, J.H., Herrmann, C., O’Sullivan, R.J., Westermann, F., Thomas, R.K., Fischer, M., 2015. Telomerase activation by genomic rearrangements in high-risk neuroblastoma. Nature 526, 700–704. http://dx.doi.org/ 10.1038/nature14980. Peinemann, F., van Dalen, E.C., Tushabe, D.A., Berthold, F., 2015. Retinoic acid post consolidation therapy for high-risk neuroblastoma patients treated with autologous hematopoietic stem cell transplantation. Cochrane Database Syst. Rev. 1, CD010685. http://dx.doi.org/10.1002/14651858.CD010685.pub2. Piccolo, E., Innominato, P.F., Mariggio, M.A., Maffucci, T., Iacobelli, S., Falasca, M., 2002. The mechanism involved in the regulation of phospholipase Cgamma1 activity in cell migration. Oncogene 21, 6520–6529. http://dx.doi.org/10.1038/sj.onc.1205821/ n1205821. (pii). Pugh, T.J., Morozova, O., Attiyeh, E.F., Asgharzadeh, S., Wei, J.S., Auclair, D., Carter, S.L., Cibulskis, K., Hanna, M., Kiezun, A., Kim, J., Lawrence, M.S., Lichenstein, L., McKenna, A., Pedamallu, C.S., Ramos, A.H., Shefler, E., Sivachenko, A., Sougnez, C., Stewart, C., Ally, A., Birol, I., Chiu, R., Corbett, R.D., Hirst, M., Jackman, S.D., Kamoh, B., Khodabakshi, A.H., Krzywinski, M., Lo, A., Moore, R.A., Mungall, K.L., Qian, J., Tam, A., Thiessen, N., Zhao, Y., Cole, K.A., Diamond, M., Diskin, S.J., Mosse, Y.P., Wood, A.C., Ji, L., Sposto, R., Badgett, T., London, W.B., Moyer, Y., GastierFoster, J.M., Smith, M.A., Auvil, J.M.G., Gerhard, D.S., Hogarty, M.D., Jones, S.J.M., Lander, E.S., Gabriel, S.B., Getz, G., Seeger, R.C., Khan, J., Marra, M.A., Meyerson, M., Maris, J.M., 2013. The genetic landscape of high-risk neuroblastoma. Nat. Genet. 45, 279–284. http://dx.doi.org/10.1038/ng.2529. Raabe, E.H., Laudenslager, M., Winter, C., Wasserman, N., Cole, K., LaQuaglia, M., Maris, D.J., Mosse, Y.P., Maris, J.M., 2008. Prevalence and functional consequence of PHOX2B mutations in neuroblastoma. Oncogene 27, 469–476. http://dx.doi.org/10. 1038/sj.onc.1210659. Ramani, P., Nash, R., Rogers, C.A., 2015. Aurora kinase A is superior to Ki67 as a prognostic indicator of survival in neuroblastoma. Histopathology 66, 370–379. http://dx.doi.org/10.1111/his.12520. Romain, C., Paul, P., Kim, K.W., Lee, S., Qiao, J., Chung, D.H., 2014. Targeting Aurora kinase-A downregulates cell proliferation and angiogenesis in neuroblastoma. J. Pediatr. Surg. 49, 159–165. http://dx.doi.org/10.1016/j.jpedsurg.2013.09.051. Santini, M.T., Rainaldi, G., 1999. Three-dimensional spheroid model in tumor biology. Pathobiology. http://dx.doi.org/10.1159/000028065. Shaw, G., Morse, S., Ararat, M., Graham, F.L., 2002. Preferential transformation of human

448