Aurora Kinases as Targets in Drug-Resistant Neuroblastoma Cells

Aurora kinase inhibitors displayed activity in pre-clinical neuroblastoma models. Here, we studied the effects of the pan-aurora kinase inhibitor tozasertib (VX680, MK-0457) and the aurora kinase inhibitor alisertib (MLN8237) that shows some specificity for aurora kinase A over aurora kinase B in a panel of neuroblastoma cell lines with acquired drug resistance. Both compounds displayed anti-neuroblastoma activity in the nanomolar range. The anti-neuroblastoma mechanism included inhibition of aurora kinase signalling as indicated by decreased phosphorylation of the aurora kinase substrate histone H3, cell cycle inhibition in G2/M phase, and induction of apoptosis. The activity of alisertib but not of tozasertib was affected by ABCB1 expression. Aurora kinase inhibitors induced a p53 response and their activity was enhanced in combination with the MDM2 inhibitor and p53 activator nutlin-3 in p53 wild-type cells. In conclusion, aurora kinases are potential drug targets in therapy-refractory neuroblastoma, in particular for the vast majority of p53 wild-type cases.


Introduction
Since their discovery in 1995, the aurora kinases have gained much interest as drug targets in cancer.In humans, there are three known homologous family members, the aurora kinases A, B, and C.They are involved in the organisation of the spindle apparatus during mitosis.Various aurora kinase inhibitors are under preclinical and clinical investigation [1,2].
Resistance acquisition is a major problem in neuroblastoma [3,4] and aurora kinase inhibitors have not been investigated in neuroblastoma models of acquired resistance.Here we tested tozasertib (VX680, MK-0457), a pan aurora kinase inhibitor [16], and alisertib, a second generation aurora kinase inhibitor that inhibits aurora kinase A and B with a higher affinity to aurora kinase A [17], in a panel of drug-resistant neuroblastoma cell lines.

Cells
The MYCN-amplified neuroblastoma cell lines UKF-NB-2, UKF-NB-3, and UKF-NB-6 were established from stage 4 neuroblastoma patients [18][19][20].Parental chemosensitive cell lines were adapted to growth in the presence of anti-cancer drugs by continuous exposure of these cell lines to the increasing concentrations of these drugs as described before [18,19,21].
All cells were propagated in IMDM supplemented with 10% FBS, 100 IU/ml penicillin and 100 mg/ml streptomycin at 37uC.Cells were routinely tested for mycoplasma contamination and authenticated by short tandem repeat profiling.

Western blot
Cells were lysed in Triton X-sample buffer and separated by SDS-PAGE.Proteins were detected using specific antibodies directed against b-actin (BioVision via BioCat GmbH, Heidelberg, Germany), p21 (Cell Signaling via New England Biolabs, Frankfurt am Main, Germany), and p53 (Enzo Life Sciences, Lo ¨rrach, Germany) and were visualised by enhanced chemiluminescence using a commercially available kit (Amersham, Freiburg, Germany).

Flow cytometry
The cells were fixed and permeabilised using Cytofix/Cytoperm (BD Biosciences, Heidelberg, Germany) according to the manufacturer's protocol.To detect Bax activation, the cells were incubated with a mouse monoclonal anti-Bax antibody (clone 6A7, BD Pharmingen, Heidelberg, Germany) for 30 min at 4uC that specifically recognises a Bax binding site that is exclusively exposed upon Bax activation in the mitochondrial membrane.To detect histone H3 phosphorylation, the cells were incubated with a mouse monoclonal anti-phosphohistone H3 antibody (Cell Signaling, EMD Millipore Corporation, Billerica, MA, USA) for 30 min at 4uC.After washing and incubation with a secondary phycoerythrin (PE)-labeled anti-mouse antibody for 30 minutes at 4uC, the percentage of cells displaying activated BAX or phosphorylated histone H3 were quantified by flow cytometry.
For cell cycle analysis, cells were fixed with 70% ice-cold ethanol over night and stained with 20 mg/mL propidium iodide (Calbiochem, Merck KgaA, Darmstadt, Germany).The DNA content was determined by flow cytometry.
All experiments were performed using a FACSCanto (BD Biosciences, Franklin Lakes, NJ, USA).

Caspase 3/7 activity assay
The activity of the caspases 3 and 7 was examined using the Caspase-Glo 3/7 Assay (Promega GmbH, Mannheim, Germany) following the manufacturer's instructions.Cells were seeded in 96well 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.

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.

Effects of tozasertib and alisertib on parental neuroblastoma cell lines and their drug-resistant sub-lines
The IC 50 values of tozasertib and alisertib were determined in a panel of neuroblastoma cell lines and their drug-resistant sub-lines (Figure 1A, Table S1).The tozasertib IC 50 values displayed a much wider distribution (5.560.4 nM to 664.06257.8nM) than the alisertib IC 50 values (7.660.5 nM to 26.861.3nM).The relative resistance (IC 50 resistant sub-line/IC 50 respective parental cell line) ranged for tozasertib from 1.4 to 92.2 and for alisertib from 0.3 to 2.5.
While the average tozasertib IC 50 differed significantly between cells that show high ABCB1 expression and those that show low ABCB1 expression, there was no significant difference in the average alisertib IC 50 s (Figure 1C).In accordance, the tozasertib IC 50 was increased in UKF-NB-3 cells transduced with a lentiviral vector encoding for ABCB1 (UKF-NB-3 ABCB1 ) but the alisertib IC 50 was not (Figure 1D).The ABCB1 inhibitor zosuquidar resensitised UKF-NB-3 ABCB1 cells to tozasertib (Table S3).These data suggest that ABCB1 confers resistance to tozasertib but not to alisertib.
Some reports had suggested that the anti-cancer effects of aurora kinase inhibitors may depend on p73 activation in addition to p53 activation [39,40] or that a p73 response may even replace a p53 response in p53-deficient cells [32].In our system, however, RNAi-mediated depletion of p53 using a lentiviral vector encoding for shRNA directed against p53 abrogated the tozasertib-and alisertib-induced p53 response (Figure 3B) and reduced neuroblastoma sensitivity to tozasertib and alisertib (Figure 3C).
In order to further investigate the role of p53 in the effects of the aurora kinase inhibitors on the viability of the neuroblastoma cell lines, we combined tozasertib with the p53 activator nutlin-3 that interferes with the interaction of p53 and its endogenous inhibitor MDM2 through MDM2 binding [41].Nutlin-3 significantly increased the tozasertib-induced effects in p53 wild-type UKF-NB-3 and UKF-NB-3 r CDDP 1000 cells (Figure 4).Similar results were obtained in p53 wild-type IMR-32 neuroblastoma cells (Figure 4).However, nutlin-3 did not enhance tozasertib activity in p53-mutated UKF-NB-3 r Nutlin 10mM cells (Figure 4).Nutlin-3 also enhanced the effects of tozasertib in p53-mutated UKF-NB-3 r VCR 10 cells (Figure 4).This is most probably due to interaction of nutlin-3 with the ABCB1-mediated tozasertib efflux in the highly ABCB1-expressing UKF-NB-3 r VCR 10 cells.Nutlin-3 is known to interfere with ABCB1-mediated drug transport [42] and the ABCB1 inhibitor zosuquidar induced similar effects (Figure S2).

Effects of tozasertib on aurora kinase function, the neuroblastoma cell cycle, and neuroblastoma cell apoptosis
Histon H3 is phosphorylated by aurora kinases A and B. Therefore, determination of histon H3 phosphorylation can serve as a surrogate for determining effects on aurora kinase function [17,43].Substantially higher tozasertib concentrations were needed to suppress histone H3 phosphorylation in the ABCB1expressing cell lines UKF-NB-3 r DOX 20 and UKF-NB-3 r VCR 10 than in the low ABCB1-expressing cell lines UKF-NB-3, UKF-NB-3 r CDDP 1000 , and UKF-NB-3 r Nutlin 10mM (Figure 5A).This finding is in concert with the action of ABCB1 as efflux transporter that limits the intracellular tozasertib levels.The fact that histone H3 phosphorylation similarly affected p53 wild-type UKF-NB-3 and UKF-NB-3 r CDDP 1000 cells and p53-mutant UKF-NB-3 r Nutlin 10mM suggests that p53 inactivation reduces tozasertib sensitivity by interfering with the drug-induced signalling downstream of aurora kinase inhibition.In accordance, the effects of tozasertib on histone H3 phosphorylation were similar among the ABCB1-expressing cell lines UKF-NB-3 r DOX 20 and UKF-NB-3 r VCR 10 (Figure 5A).
Tozasertib induced a G2/M cell cycle block in p53 wild-type and p53-mutant neuroblastoma cell lines but with different kinetics and concentration-dependencies (Figure 5B) and induced more or less pronounced signs of endoreduplication (Figure 5B).
Taken together, these data suggest that the action of aurora kinase inhibitors is at least in part mediated through cell cycle inhibition and apoptosis induction as consequence of aurora kinase inhibition.

Discussion
Testing of tozasertib, a pan aurora kinase inhibitor [16], and alisertib, a second generation aurora kinase inhibitor that inhibits aurora kinase A and B with a higher affinity to aurora kinase A [17], in a panel of drug-resistant neuroblastoma cell lines revealed differing activity profiles.ABCB1 expression reduced cancer cell sensibility to tozasertib but not to alisertib.This is in concert with previous findings that had suggested tozasertib to be a substrate of ABCB1 [28,29].Moreover, we provided evidence that alisertib is not transported by ABCB1.The anti-neuroblastoma mechanism included (at least in part) aurora kinase inhibition as indicated by reduced phosporylation of the aurora kinase substrate histone H3, cell cycle inhibition, and induction of apoptosis.
Varying findings have been published on the involvement of p53 in the aurora kinase inhibitor-induced anti-cancer effects in models from various cancer entities.Various reports showed that aurora kinase inhibitors activate p53 signalling and that this p53 signalling contributed to the aurora kinase inhibitor-induced anticancer effects [10,[33][34][35].Other reports suggested that p53 may be of minor relevance for aurora kinase inhibitor activity [32,36,37] or that aurora kinase inhibitor activity may be enhanced in p53-defective cells [30,31,38].Also, the role of p53 may differ between approaches that target aurora kinase A and those that target aurora kinase B [44].Thus, the relevance of p53 in response to aurora kinase inhibition apparently depends on the cellular context.
In neuroblastoma cells, the aurora kinase A and B inhibitor CCT137690 was described to induce a p53 response [10].Our results obtained in p53 wild-type and p53-mutant cells as well as in p53-depleted cells indicated that p53 activation is of relevance for the anti-cancer effects exerted by aurora kinase inhibitors in neuroblastoma cells.The combination of the MDM2 inhibitor and p53 activator nutlin-3 with tozasertib enhanced the activity of aurora kinase inhibitors in in the presence of functional p53.This is of clinical relevance since p53 mutations were described as acquired resistance mechanism in neuroblastoma [45,46].Nevertheless, the vast majority of neuroblastomas (about 85%) harbours p53 wild-type cells [45,46].
Noteworthy, the combined effects of aurora kinase inhibitors and MDM2 inhibitors may depend on the sequence of drug administration.Previous investigations in p53 wild-type A375 melanoma cells had revealed that nutlin-3 pre-treatment had resulted in a p53-mediated cell cycle arrest that protected these cells from tozasertib-induced anti-cancer effects while tozasertib pretreatment or simultaneous combined tozasertib and nutlin-3 treatment had resulted in enhanced combined anti-cancer effects [39].Nutlin-3 pre-treatment had also protected A549 lung cancer cells, primary human keratinocytes, and HCT116p53+/+ colorectal cancer cells (but not HCT116p532/2 cells) from tozasertibinduced toxicity [39].In this context, we investigated the effects of simultaneous tozasertib and nutlin-3 treatment in primary human foreskin fibroblasts (Figure S2).The results were promising because 1) the primary fibroblasts were much less sensitive to tozasertib and nutlin-3 than p53 wild-type and p53-mutant neuroblastoma cells and 2) the combination of tozasertib and nutlin-3 resulted in contrast to the effects observed in p53 wildtype neuroblastoma cells not in enhanced toxicity compared to either single treatment (Figure S2).
Finally, it needs to be noted that although the major body of data from our study clearly demonstrated that p53 function was critically involved in the neuroblastoma cell response to aurora kinase inhibition, the p53-mutated cell line UKF-NB-6 r Nutlin 10mM was similarly sensitive to tozasertib and alisertib as the p53 wildtype neuroblastoma cell lines.The reasons for this remain unclear and emphasise that many factors may determine neuroblastoma cell sensitivity to aurora kinase inhibitors in addition to the p53 status.Possibly, aurora kinase inhibitor-induced p73 activation [32] and/or other events that need to be determined in future studies may be responsible for this.Noteworthy, kinase inhibitors may interfere with other (previously unidentified) kinases in addition to the target kinases they were designed to inhibit.For example, tozasertib was shown to interfere with additional kinases including ABL and FLT3 [16,47,48].Although the similarity of the effects exerted by two structurally different aurora kinase inhibitors suggests aurora kinases to be relevant common targets, effects on other kinases may contribute to the effects of tozasertib and/or alisertib on the viability of (certain) neuroblastoma cells.
Taken together, we provide the first data on the efficacy of aurora kinase inhibitors in neuroblastoma cells with acquired resistance to anti-cancer drugs.Our data suggest that aurora kinases represent a therapeutic target in therapy-refractory neuroblastoma cells, in particular in p53 wild-type therapyrefractory neuroblastoma cells.Table S1 Concentrations of tozasertib and alisertib that decrease neuroblastoma cell viability by 50% (IC 50 ).

Figure 5 .
Figure 5. Effects of tozasertib on histone H3 phosphorylation (indicating aurora kinase A and B activity), cell cycle distribution, and apoptosis (indicated by BAX activation) in neuroblastoma cells.A) Numbers of cells expressing phosphorylated histone H3 were determined flow cytometry; * P,0.05 relative to untreated control; B) Representative histograms indicating cell cycle distribution in neuroblastoma cells after tozasertib treatment, arrows indicate additional peaks with high DNA content, possibly indicating endoreduplication; C) Numbers of cells with activated BAX expressed as fold change relative to control as determined by flow cytometry using an antibody specific for activated BAX after 24 h of tozasertib treatment.* P,0.05 relative to untreated control.doi:10.1371/journal.pone.0108758.g005

Figure
Figure S1 Alisertib-induced expression of p53 target genes in parental UKF-NB-3 cells and their drugresistant sub-lines as indicated by qPCR.(PDF) Figure S2 Effects of tozasertib combination therapies on the viability of UKF-NB-3 r VCR 10 cells (tozasertib plus the ABCB1 inhibitor zosuquidar) or primary human foreskin fibroblasts (HFFs, tozasertib plus the MDM2 inhibitor nutlin-3) as determined by MTT assay after 5 days of incubation.(PDF)