Identification of flubendazole as potential anti-neuroblastoma compound in a large cell line screen.

Flubendazole was shown to exert anti-leukaemia and anti-myeloma activity through inhibition of microtubule function. Here, flubendazole was tested for its effects on the viability of in total 461 cancer cell lines. Neuroblastoma was identified as highly flubendazole-sensitive cancer entity in a screen of 321 cell lines from 26 cancer entities. Flubendazole also reduced the viability of five primary neuroblastoma samples in nanomolar concentrations thought to be achievable in humans and inhibited vessel formation and neuroblastoma tumour growth in the chick chorioallantoic membrane assay. Resistance acquisition is a major problem in high-risk neuroblastoma. 119 cell lines from a panel of 140 neuroblastoma cell lines with acquired resistance to various anti-cancer drugs were sensitive to flubendazole in nanomolar concentrations. Tubulin-binding agent-resistant cell lines displayed the highest flubendazole IC50 and IC90 values but differences between drug classes did not reach statistical significance. Flubendazole induced p53-mediated apoptosis. The siRNA-mediated depletion of the p53 targets p21, BAX, or PUMA reduced the neuroblastoma cell sensitivity to flubendazole with PUMA depletion resulting in the most pronounced effects. The MDM2 inhibitor and p53 activator nutlin-3 increased flubendazole efficacy while RNAi-mediated p53-depletion reduced its activity. In conclusion, flubendazole represents a potential treatment option for neuroblastoma including therapy-refractory cells.

Flubendazole was shown to exert anti-leukaemia and anti-myeloma activity through inhibition of microtubule function. Here, flubendazole was tested for its effects on the viability of in total 461 cancer cell lines. Neuroblastoma was identified as highly flubendazole-sensitive cancer entity in a screen of 321 cell lines from 26 cancer entities. Flubendazole also reduced the viability of five primary neuroblastoma samples in nanomolar concentrations thought to be achievable in humans and inhibited vessel formation and neuroblastoma tumour growth in the chick chorioallantoic membrane assay. Resistance acquisition is a major problem in high-risk neuroblastoma. 119 cell lines from a panel of 140 neuroblastoma cell lines with acquired resistance to various anti-cancer drugs were sensitive to flubendazole in nanomolar concentrations. Tubulin-binding agent-resistant cell lines displayed the highest flubendazole IC 50 and IC 90 values but differences between drug classes did not reach statistical significance. Flubendazole induced p53-mediated apoptosis. The siRNA-mediated depletion of the p53 targets p21, BAX, or PUMA reduced the neuroblastoma cell sensitivity to flubendazole with PUMA depletion resulting in the most pronounced effects. The MDM2 inhibitor and p53 activator nutlin-3 increased flubendazole efficacy while RNAi-mediated p53-depletion reduced its activity. In conclusion, flubendazole represents a potential treatment option for neuroblastoma including therapy-refractory cells.
Mebendazole affects the viability of cancer cells in experimental systems from a broad spectrum of cancer entities including lung cancer, breast cancer, ovary cancer, adrenocortical carcinoma, osteosarcoma, melanoma, glioblastoma, and colorectal carcinoma 1,3-7 . More recently, flubendazole was shown to affect the viability of leukaemia and myeloma cells in nanomolar concentrations 2 .
Here, we performed a screen of flubendazole in a panel of 321 cell lines including cell lines from 26 cancer entities. Among leukaemia and multiple myeloma, neuroblastoma was identified as a highly flubendazolesensitive cancer entity. Flubendazole showed broad activity in primary neuroblastoma cells and a panel of 140 neuroblastoma cell lines with acquired drug resistance. The anti-neuroblastoma activity of flubendazole involved p53-mediated apoptosis and the MDM2 inhibitor and p53 activator nutlin-3 strongly enhanced the flubendazole  effects. A water-soluble flubendazole-(2-hydroxypropyl)-b-cyclodextrin preparation inhibited vessel formation and tumour growth in the chick chorioallantoic membrane (CAM) model in vivo.

Results
Effects of flubendazole on cancer cell viability in a panel of 321 cancer cell lines from 26 cancer entities. Flubendazole was screened in a panel of 321 cancer cell lines from 26 cancer entities. Concentrations of 1 mM appear pharmacologically achievable based on prior studies in mice that demonstrated a dose of 5 mg/kg to produce a C max of 3.6 mM 8 . The maximum concentration tested was 5 mM. Multiple myeloma, neuroblastoma, and leukaemia/lymphoma consistently belonged to the cancer entities that displayed the highest sensitivity to flubendazole (Fig. 1a, Suppl. Fig. S1, Suppl. Table S1). This confirmed previous investigations that had suggested multiple myeloma and leukaemia to be flubendazole-sensitive cancer types 2 and identified neuroblastoma as an additional flubendazole-sensitive entity. Statistical testing using the Wilcoxon rank sum test 9 with subsequent Benjamini-Hochberg correction 10 indicated the flubendazole IC 90 values in neuroblastoma cells to be significantly lower than those from 21 out of the 25 other investigated cancer cell types (Fig. 1c).
117 (36%) of the 321 cell lines displayed IC 90 s , 1 mM. 31 cell lines (10%) displayed IC 90 s . 1 mM and ,5 mM, and 173 cell lines (54%) displayed IC 90 s . 5 mM. There were clear differences between the individual cancer entities. In leukaemia/lymphoma 40 (82%) out of 49 cell lines, in multiple myeloma 10 (100%) out of 10 cell lines, and in neuroblastoma 28 (88%) out of 32 cell lines displayed an IC 90 , 1 mM. Together, these three entities accounted for 78 (67%) of the 117 cell lines that displayed IC 90 s , 1 mM among the 26 cancer entities. In all other entities, except Ewing's sarcoma (4 (57%) out of 7 cell lines with an IC 90 , 1 mM) and head and neck cancer (3 (60%) out of 5 cell lines with an IC 90 , 1 mM), the majority of the cell lines displayed IC 90 s . 1 mM. None of the 9 gastric cancer, the 13 melanoma, the 6 oesophageal cancer, the 10 ovarian cancer, the 10 pancreas cancer, the 5 prostate cancer, or the 3 retinoblastoma cell lines displayed an IC 90 , 1 mM ( Fig. 1; Suppl. Table S1).
The majority (87, 62%) of the 140 drug-resistant neuroblastoma cell lines was similarly sensitive to flubendazole like the corresponding parental cell lines as indicated by fold changes IC 90 drug-resistant sub-line/IC 90 respective parental cell line between 0.5 and 2.0. In 39 (28%) of the drug-resistant cell lines the fold change was .2.0. In 14 drug-resistant cell lines it was ,0.5 (Fig. 2, Suppl. Table S2).
In concordance with its anticipated effects on microtubule function 2 , flubendazole induced a G2/M arrest in neuroblastoma cells (Suppl. Fig. S3). The tubulin-binding agents can be divided into destabilising agents that inhibit tubulin polymerisation and stabilising agents that inhibit microtubule degradation in super-therapeutic concentrations 12,13 . The taxanes docetaxel, paclitaxel, and cabazitaxel are stabilising agents that target the taxoid domain. The Vinca alkaloids vinblastine, vincristine, and vinorelbine are destabilising agents that target the vinca domain. Flubendazole and 2-methoxyestradiol are destabilising agents that target the colchicine domain 2,12,13 . The neuroblastoma cell lines with acquired resistance to tubulin-binding agents displayed the highest average IC 90 and IC 50 values for flubendazole (Fig. 2 Table S2).  Table S3).
Effects of ABCB1 or ABCG2 expression on cancer cell sensitivity to flubendazole. The anti-cancer effects of many anti-cancer drugs are affected by the expression of ATP-binding cassette (ABC) transporters including ABCB1 and ABCG2 14,15 . UKF-NB-3 cells were transduced with lentiviral vectors encoding for ABCB1 or ABCG2, respectively, as described previously 16,17 . ABCB1 expression enhanced UKF-NB-3 cell resistance to the ABCB1 substrate vincristine but not to flubendazole (Suppl. Table S4). The ABCB1 inhibitor verapamil sensitised ABCB1-expressing neuroblastoma cells to vincristine but not to flubendazole. Similarly, ABCG2 expression increased UKF-NB-3 cell resistance to the ABCG2 substrate mitoxantrone but not to flubendazole. The ABCG2 inhibitor WK-X-34 18 sensitised ABCG2expressing UKF-NB-3 cells to mitoxantrone but not to flubendazole (Suppl. Table S4).
The MDM2 inhibitor nutlin-3 20 enhanced the effects of flubendazole in p53 wild-type UKF-NB-3 cells ( Fig. 5b; Suppl. Table S6) but did not affect the effects of flubendazole in p53-mutated UKF-NB-3 r VCR 10 cells and hardly influenced the effect of flubendazole in p53depleted UKF-NB-3 p53shRNA cells (Suppl . Table S6). Some combined effects may be observed in UKF-NB-3 p53shRNA cells treated with the combination of flubendazole 80 nM and nutlin-3 1.25 mM. This is most likely caused by some remaining p53 activity due to incomplete p53 depletion.  Table S7). The preparation was stable during storage for eight weeks as indicated by HPLC and assessment of biological activity (Suppl . Table S7).
For the investigation of vessel formation, the CAM assay was performed using drug-loaded gelatin sponges as described previously 21 . Vessel formation was scored from 0 (complete suppression of vessel formation) to 5 (vessel formation using non-treated control sponges). The analysis of vessel formation surrounding 10 vehicletreated control sponges resulted in a vessel formation score of 4.5 6 0.7. The analysis of vessel formation surrounding 10 flubendazoletreated sponges resulted in a vessel formation score of 0.7 6 0.8 (P 5 1 3 10 26 relative to vehicle-treated control sponges). Representative photographs are presented in Fig. 6a.
To analyse the effects of flubendazole on neuroblastoma xenograft growth in the CAM, three cell lines were used. UKF-NB-3 (flubendazole IC 90 Fig. 6b and Suppl. Fig. S6. To quantify the drug effects, the fractions of necrotic cells were determined and cancer cell invasion was analysed. The flubendazoleinduced necrosis levels were higher in UKF-NB-3 and UKF-NB-3 r CDDP 1000 tumours than in UKF-NB-3 r VCR 10 tumours although differences did not reach statistical significance (Fig. 6c). To measure cancer cell invasion, the minimum distance of the cancer cell invasion front to the inner CAM border was determined. Higher values indicate a greater distance to the inner membrane and, therefore, lower penetration. Again, flubendazole exerted stronger effects on UKF-NB-3 and UKF-NB-3 r CDDP 1000 tumours than on UKF-NB-3 r VCR 10 tumours (Fig. 6c).

Discussion
Flubendazole had been demonstrated to exert activity against leukaemia and myeloma cells 2 . Here, a screen of flubendazole in 321 cancer cell lines from 26 entities confirmed the flubendazole-sensitivity of leukaemia and multiple myeloma cells and identified neuroblastoma as potential additional flubendazole-sensitive cancer entity. 140 neuroblastoma cell lines and primary neuroblastoma samples from five different patients were similarly sensitive to flubendazole. 119 (85%) out of 140 neuroblastoma cell lines and all five primary neuroblastoma isolates displayed IC 90 values ,1000 nM. Only in two further cancer entities (Ewing's sarcoma, head and neck cancer), a substantial fraction of the investigated cell lines was sensitive to flubendazole in concentrations ,1 mM that appear to be pharmacologically achievable 8 . Notably, a number of the entities that were in general regarded to be insensitive to flubendazole included flubendazole-sensitive cell lines emphasising the need for markers indicating cancer cell sensitivity to flubendazole.
Neuroblastoma is the most frequent extracranial solid childhood tumour. About half of patients suffer from high-risk disease associated with overall survival rates below 50% despite intensive therapy 22 . Resistance acquisition is a major problem in neuroblastoma 11 . Among 140 neuroblastoma cell lines with acquired resistance to a range of anti-cancer drugs, 119 cell lines (85%) displayed IC 90 values below 1 mM. 87 drug-resistant neuroblastoma cell lines (62%) were similar sensitive to flubendazole like the corresponding parental cells. 39 drug-resistant cell lines (28%) showed cross-resistance to flubendazole, 14 drug-resistant cell lines (10%) were more sensitive to flubendazole than their parental counterparts.
Flubendazole has been described to interfere with microtubule function 2 . There are different classes of tubulin-binding agents. In super-therapeutic concentrations, destabilising agents inhibit tubulin polymerisation, stabilising agents microtubule degradation 12,13 . Our panel of resistant neuroblastoma cell lines contained cell lines resistant to the taxanes docetaxel, paclitaxel, and cabazitaxel (stabilising agents, target the taxoid domain), to the Vinca alkaloids vinblastine, vincristine, and vinorelbine (destabilising agents, target the vinca domain), and to 2-methoxyestradiol that is like flubendazole a destabilising agents that targets the colchicine domain 2,12,13 . The neuroblastoma cell lines with acquired resistance to tubulin-binding agents displayed the highest average flubendazole IC 90 and IC 50 values but the differences did not reach statistical significance. These findings suggest that acquired resistance to a tubulin-binding agent may not necessarily be associated with a substantially decreased sensitivity to flubendazole. In concordance, albendazole, another benzimidazole anthelminthic, had been shown to be active in cancer cell lines resistant to the stabilising tubulin-binding agents paclitaxel and epothilone B 23,24 . Also, albendazole and flubendazole had been shown to increase the anti-cancer effects of paclitaxel and vinblastine 2,25 .
In concordance with previous results 2 , the anti-cancer effects of flubendazole were not impaired by ABCB1 expression. Moreover, ABCG2, another major ABC transporter 14,15 , did not affect flubendazole efficacy. However, flubendazole appears to exert its anti-neuroblastoma effects in part via p53 activation. Flubendazole induced p53 signalling, displayed a strongly enhanced potency in combination with the MDM2 inhibitor and p53 activator nutlin-3, and the flubendazole activity was substantially reduced in the absence of functional p53. RNAi-mediated depletion experiments suggested PUMA to be a critical mediator of the flubendazole-induced anti-neuroblastoma effects. Since inactivation of p53 signalling has been suggested as an acquired resistance mechanism in neuroblastoma 11 , this may be of clinical relevance. While p53 was found mutated in only 2% of neuroblastomas at diagnosis, p53 mutations were detected in about 15% of neuroblastomas at relapse 11,26,27 . In concordance, different cell line-based studies pointed towards a role of p53 inactivation as (acquired) resistance mechanism in neuroblastoma 11,28-33 . Therefore, flubendazole appears to be a promising treatment option in particular for the majority of p53 wild-type neuroblastomas.
The poorly water-soluble flubendazole was administered in 0.9% NaCl and 0.01% Tween-80 in a previous study 2 . We were not successful in preparing a suitable flubendazole preparation using this method. Acidic b-CD complexes of flubendazole for oral use that had been described before 8,34 not be suited for parenteral application or the CAM assay. Therefore, we developed a (2-hydroxypropyl)-b-CD preparation of flubendazole that was equally effective as DMSOdissolved flubendazole in cell culture. This flubendazole (2-hydroxypropyl)-b-CD preparation inhibited vessel and tumour formation in the chick chorioallantoic membrane. These findings are in concert with reports that demonstrated anti-leukaemia and anti-myeloma activity of flubendazole in vivo 2 . In addition, we provide evidence that flubendazole also exerts anti-angiogenic effects.
Mebendazole and albendazole are further benzimidazole anthelminthic agents that have been shown to exert anti-cancer effects 1,3-7,35-39 . Recent results suggested that mebendazole and albendazole differ significantly in their anti-cancer mechanisms of action 7 . Based on a comparison of flubendazole and mebendazole in a panel of 39 cancer cell lines from 13 cancer entities (Suppl. Table 8) and in 25 drug-resistant neuroblastoma cell lines (Suppl. Table 9) these two compounds appear to exert similar anti-cancer effects.
In conclusion, we show that the well-tolerated anthelminthic flubendazole [40][41][42] , represents a potential treatment option for neuroblastoma, in particular for the majority of neuroblastomas with functional p53.
Parental chemosensitive cell lines were adapted to growth in the presence of anticancer drugs by continuous exposure of these cells to increasing drug concentrations as described previously 28,29,43 . The drug-resistant cell lines were derived from the resistant cancer cell line (RCCL) collection (www.kent.ac.uk/stms/cmp/RCCL/ RCCLabout.html) (Suppl . Table S10).
Cells were routinely tested for mycoplasma contamination and authenticated by short tandem repeat or variable number tandem repeat profiling. p53-depleted cells or cells showing high expression of ABCB1 (also known as MDR1, gene product also known as P-glycopotein) or ABCG2 (also known as BCRP) were established as described previously 17 using the Lentiviral Gene Ontology (LeGO) vector technology 44,45 ; www.lentigo-vectors.de).
Fresh neuroblastoma cells (MYCN amplified) were isolated from the bone marrow aspirate of five patients with metastasised INSS stage 4 neuroblastoma following informed consent. Primary cells were cultivated in IMDM supplemented with 10% FCS, 100 IU/ml penicillin, and 100 mg/ml streptomycin at 37uC.
Cell viability was tested in 96-well plates using 1 in 4 dilution steps. The maximum flubendazole concentration tested was 5000 nM. IC 50 and IC 90 values were determined using the CalcuSyn software (Biosoft, Cambridge, UK). Differences in the IC 50 and IC 90 values between the cancer entities were tested using the Wilcoxon rank sum test 9 with subsequent correction for multiple testing by the Benjamini-Hochberg method 10 .
Caspase 3/7 activity assay. The activity of the caspases 3 and 7 was examined using the Caspase-GloH 3/7 Assay (Promega GmbH, Mannheim, Germany) following the manufacturer's instructions. Cells were seeded in 96-well cell culture plates and allowed to adhere overnight. After drug treatment, the culture plates were adjusted to room temperature. Then, the cells were incubated for 5 min with the pre-mixed substrate and the luminescent signal was measured with a plate reader (Tecan, Crailsheim, Germany) for 30 cycles.
Development of water-soluble flubendazole preparations. 5.0 mg flubendazole were suspended in 10.0 mL aqueous solutions of the cyclodextrin (CD) derivatives (2-hydroxypropyl)-b-CD (0.2-30% m/V, Cargill, Krefeld, Germany) or c-CD (1-20% m/V, Wacker-Chemie, Munich, Germany). After vortexing and ultrasonication (15 min), the samples were stirred at room temperature for 5 days. After 48 h and 120 h samples were taken, filtered through pre-saturated cellulose acetate membrane filters (0.45 mm), diluted with purified water and analysed by HPLC for the flubendazole concentration using an isocratic mixture of water and methanol (31569) containing 0.1% trifluoracetic acid at a temperature of 30uC and a reverse phase column (Kinetex PFP, 2.6 mm, 100 3 4.6 mm) in combination with a KrudKatcher TM Ultra In-line filter (Phenomenex, Aschaffenburg, Germany) in an Agilent 1200 HPLC system (Agilent, Böblingen, Germany) with DAD detection (flow rate 1.0 mL/min). Flubendazole was detected at 234 nm (retention time 2.1 min).
Validation of the HPLC analysis was performed according to the ICH harmonised tripartite guideline ''Validation of analytical procedures: Text and Methodology Q2(R1)'' (www.ich.org). An acceptable degree of linearity, accuracy and precision was confirmed for concentrations between 0.25 and 1 mg/mL. The limit of detection was ,0.01 mg/mL; the limit of quantification was .0.2 mg/mL.
Chorioallantoic membrane (CAM) assay. The effects of flubendazole on vessel formation and in vivo tumour growth were investigated in the chorioallantoic membrane (CAM) assay. For the investigation of vessel formation the CAM assay was performed using drug-loaded gelatin sponges as described previously 21 . Sponges were placed onto the CAM at day 8. Vessel formation was determined at day 12. 20% Luconyl Black in phosphate-buffered saline was injected into a vitelline vein. Vessel formation was scored from 0 (complete suppression of vessel formation) to 5 (vessel formation using non-treated control sponges).
Tumour growth in the CAM was examined following described methods 46 . Briefly, 5 3 10 6 cells were suspended in 30 ml extracellular matrix (Growth-factor reduced Matrigel, BD Biosciences, Heidelberg, Germany) and implanted on the CAM of fertilised chicken eggs on day 11 of embryo development. Eggs were incubated for another 3 days to allow formation of a distinct tumour mass. On day 14, a small silicone ring was placed around the tumour mass and drugs were administered. On day 18, the tumours were sampled with the surrounding CAM, fixed using 4% paraformaldehyde, and embedded in paraffin. Sections (4 mM) were haematoxylin/ eosin-stained.
Statistics. Results are expressed as mean 6 S.D. of at least three experiments. Comparisons between two groups were performed using Student's t-test. Three and more groups were compared by ANOVA followed by the Student-Newman-Keuls test. P values lower than 0.05 were considered to be significant.