Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

The Aurora-A inhibitor MLN8237 affects multiple mitotic processes and induces dose-dependent mitotic abnormalities and aneuploidy

The Aurora-A inhibitor MLN8237 affects multiple mitotic processes and induces dose-dependent... www.impactjournals.com/oncotarget/ Oncotarget, Vol. 5, No. 15 The Aurora-A inhibitor MLN8237 affects multiple mitotic processes and induces dose-dependent mitotic abnormalities and aneuploidy 1 1 1 2 Italia Anna Asteriti , Erica Di Cesare , Fabiola De Mattia , Volker Hilsenstein , 2 1 1 1 Beate Neumann , Enrico Cundari , Patrizia Lavia and Giulia Guarguaglini Institute of Biology, Molecular Medicine and Nanobiotechnology (formerly Institute of Molecular Biology and Pathology), CNR National Research Council, c/o Department of Biology and Biotechnology, Sapienza University of Rome, Rome, Italy Advanced Light Microscopy Facility, EMBL, Heidelberg, Germany Correspondence to: Giulia Guarguaglini, email: giulia.guarguaglini@uniroma1.it Keywords: Aurora kinases, MLN8237, mitotic spindle, aneuploidy, time-lapse microscopy Received: March 7, 2014 Accepted: July 8, 2014 Published: July 9, 2014 This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. ABSTRACT Inhibition of Aurora kinase activity by small molecules is being actively investigated as a potential anti-cancer strategy. A successful therapeutic use of Aurora inhibitors relies on a comprehensive understanding of the effects of inactivating Aurora kinases on cell division, a challenging aim given the pleiotropic roles of those kinases during mitosis. Here we have used the Aurora-A inhibitor MLN8237, currently under phase-I/III clinical trials, in dose-response assays in U2OS human cancer cells synchronously proceeding towards mitosis. By following the behaviour and fate of single Aurora-inhibited cells in mitosis by live microscopy, we show that MLN8237 treatment affects multiple processes that are differentially sensitive to the loss of Aurora-A function. A role of Aurora-A in controlling the orientation of cell division emerges. MLN8237 treatment, even in high doses, fails to induce efficient elimination of dividing cells, or of their progeny, while inducing significant aneuploidy in daughter cells. The results of single-cell analyses show a complex cellular response to MLN8237 and evidence that its effects are strongly dose-dependent: these issues deserve consideration in the light of the design of strategies to kill cancer cells via inhibition of Aurora kinases. often abnormally expressed in tumor cells and are being INTRODUCTION investigated as targets of anti-mitotic compounds for cancer therapy [9, 10]. Many efforts have converged in the The Aurora-A kinase is a major regulator of cell last years to develop Aurora inhibitors: molecules acting division and operates in distinct processes required for as ATP-competitors have been identified and some of them spindle assembly: in human cells it regulates separation are currently in clinical trials [11]. Only a few of those and maturation of centrosomes at mitotic entry, mitotic molecules discriminate Aurora-A vs Aurora-B and may microtubule (MT) nucleation [1- 3] and the integrity of thus prove useful both in clinical studies for comparing spindle poles [2, 4, 5]. Recent data also indicate a role the efficacy of anti-tumor responses and for dissecting the of Aurora-A in central spindle assembly at telophase [6, functions of Aurora kinases in mammalian cells. 7]. The highly homologous Aurora-B kinase also operates MLN8237 (Alisertib) is a second generation Aurora in control of the fidelity of chromosome segregation, inhibitor currently undergoing Phase-I/III clinical trials by regulating chromosome condensation, correction of [11-16; www.clinicaltrials.gov]. Thus far, it is one of the improper attachments between MTs and kinetochores, molecules displaying highest specificity for Aurora-A spindle checkpoint function, cytokinesis and abscission over Aurora-B (300-fold in in vitro assays and 200-fold [8]. in HCT116 colorectal carcinoma cells [17]). Most pre- As other mitotic regulators, Aurora kinases are www.impactjournals.com/oncotarget 6229 Oncotarget clinical studies based on whole cell population analyses suitable for high-resolution single-cell microscopy in tumor cell lines showed cell growth inhibition, analysis, and employed in our previous studies of RNA accumulation of polyploid cells over time, as well as interference-mediated Aurora-A inactivation [4, 5, 23]. induction of cell death [17-19]. Anti-tumor activity was We set up a protocol by pre-synchronizing U2OS also demonstrated in xenograft mouse models [17, 20, 21]. cells at the G1/S transition by thymidine treatment, then Available data on MLN8237-treated cells were releasing from arrest into G2 and mitosis (Figure 1A). mostly obtained from asynchronous cultures analyzed in MLN8237 was added 6 hours after thymidine release (late bulk populations. This approach reveals the predominant S-phase/early G2) and cells were harvested after further 4 cellular behaviour after long exposure to Aurora-A hours. Aurora-A activity was measured at the single cell inhibition (24 to 96 hours) but can miss out transient level by anti-Aur-A-phospho-Thr288 immunofluorescence phenomena and so mask the unfolding of relevant (IF) staining in dose-response assays (Figure 1B, left processes. In addition, inhibition of as pleiotropic a kinase panels). Aurora-A auto-phosphorylation was significantly as is Aurora-A, yields multiple phenotypes over time, inhibited at concentrations ranging from 5 nM to 250 nM. making it difficult to dissect distinct functional roles With concentrations higher than 20 nM the residual signal within a bulk population. Microscopy-based single cell at spindle poles was below 15% compared to controls. In analyses are proving of critical importance to visualize Western blot analysis, no phospho-Thr288-Aurora-A was the array of possible cell responses to anti-mitotic drugs detectable in mitotic extracts from cultures treated with [22]. Here we have coupled high resolution microscopy 20 and 50 nM MLN8237 for 4 hours, while some residual and high-throughput analysis of single cells treated with amount was present after 1 hour (Figure 1C). increasing doses of Aurora-A inhibitor to investigate the Previous reports indicated that MLN8237 above possible fates of cells with inactive Aurora-A. 100 nM also inhibits Aurora-B activity in other cell lines A protocol was set up for treating pre-synchronized [19, 25-27]. We therefore assessed the specificity of cultures when they reach G2 and analyze progression MLN8237 by measuring Aurora-B activity using anti-Aur- through G2 and mitosis as soon as Aurora-A inhibition B-phospho-Thr232 antibody (Figure 1B, right panels). is achieved. Because MLN8237 induces spindle pole Surprisingly, we noticed that Aurora-B activity is already abnormalities [23], we assessed the occurrence of significantly compromised by 50 nM MLN8237; that was chromosome mis-segregation events and aneuploidy not evident when using anti-phospho-Histone-H3 (Ser10) induction, which would represent undesirable effects of as a reporter of Aurora-B activity (Supplementary Figure the treatment in anti-cancer therapy. Our results highlight S1), possibly reflecting kinase redundancy or delay in a partial specificity of MLN8237 in the U2OS cell line, detecting modulation of phosphorylation of downstream with multiple cellular responses in a dose-dependent targets vs auto-phosphorylation. manner. The single cell analysis enabled us to depict Our single-cell analysis in U2OS cultures delimits a fraction of cells with defective spindle orientation, therefore a narrow MLN8237 concentration window (20- a defect that was not appreciated in previous studies of 50 nM) yielding effective and specific Aurora-A inhibition. Aurora-A inhibition in human cells. In addition, we find that low and high MLN8237 concentrations yield mild MLN8237 delays mitotic entry and prolongs and massive aneuploidy, respectively, representing a mitotic duration in a dose-dependent manner tumor-inducing or a tumor-suppressing condition [24]. Collectively, these results draw attention to the variability We investigated the influence of MLN8237 on and the nature of cellular responses to the loss of Aurora mitotic entry: after 4 hours of treatment, we found kinase function, which may represent potential caveats a significantly lower percentage of mitotic cells in deserving consideration when designing and interpreting MLN8237-treated cultures compared to controls (Figure clinical trials. 2A). This effect is dose-dependent, appearing at ≥ 20 nM, and is stronger at 250 nM MLN8237. To clarify whether RESULTS cells were arrested in the G2 phase or rather delayed in progression through the G2/M transition, we recorded cultures in time-lapse experiments from the treatment start MLN8237 displays dose-dependent target up to 16 hours later (Figure 2B). The peak of interphases selectivity on Aurora kinases entering mitosis in control cultures (DMSO) was between 4 and 8 hours from the treatment start (Figure 2B) and was not significantly affected by partial Aurora-A Prior to analyzing mitotic division in cells with inhibition (5-10 nM MLN8237). Entry into mitosis was inhibited Aurora-A, we sought to precisely define the instead delayed above 20 nM MLN8237. In the 16 hours specificity of MLN8237 inhibition in dose-response of time-lapse recording, 70-80% of the interphase cells assays. We used the U2OS osteosarcoma cell line for its entered mitosis in all treated cultures (about 60% with ease of cytological analysis, which renders it especially www.impactjournals.com/oncotarget 6230 Oncotarget 250 nM MLN8237): thus, the majority of cells exposed an accumulation of prometaphase figures over all mitoses to MLN8237 are delayed in G2, yet mitotic onset is not in fixed samples (Supplementary Figure S2). The MT prevented. nucleation defect was strongly dose-dependent and Extending the time-lapse recording to 30 hours appeared in a relevant fraction of mitotic cells treated with indicated that MLN8237 prolonged the duration of mitosis 50 nM MLN8237 or above. in a dose-dependent manner (about 300 minutes with 50 In cells in which MT nucleation was not visibly nM, compared to about 80 minutes in control cells; Figure affected, spindles were highly disorganized (affecting 30% 3). Importantly, although slowed down, cells eventually to 60% of all prometaphases with 10 nM or above; Figure exited mitosis. Together, the results indicate that both the 4); a fraction of these prometaphases displayed spindles G2-to-mitosis transition and the overall duration of the with multiple poles, consistent with previous observations mitotic process are strongly dependent on Aurora-A. [17, 23]. A non-significant fraction of monopolar spindles was present at 10-20 nM MLN8237. The influence of these defects on the global Inhibition of Aurora kinases yields impaired MT execution of mitosis was examined in depth in time- nucleation, disorganized spindles and multipolar lapse movies of MLN8237-treated cells. Figure 5, first or failed cell division row, and Supplementary Movie S1 show examples of normal mitosis. We observed that a significant fraction We next analyzed spindle structure in cells that of cells with inhibited Aurora-A (about 20% with 50 entered mitosis with different degrees of Aurora-A and nM MLN8237, and a smaller fraction with 250 nM) Aurora-B inhibition. underwent multipolar ana-telophase (Figure 5, second With the highest MLN8237 concentration (250 row; Supplementary Movie S2). Multipolar mitoses nM) a strong impairment of MT nucleation was evident, took a longer time to reach the stage of chromosome with 70% of prometaphases displaying no MTs (Figure segregation (about 125 minutes average, compared to 35 4); this was associated with a prolonged prometaphase minutes in control cells). In some cases two of the multiple duration in time-lapse recording experiments, yielding groups of segregating chromosomes eventually re-joined, Figure 1: Dose-dependent inhibition of Aurora-A and Aurora-B by MLN8237. A. Protocol for MLN8237 treatment in cells progressing towards mitosis after thymidine (Thym) arrest and release. B. Quantification of IF signals for active pThr288-Aurora-A (left, mean intensity at poles) or active pThr232-Aurora-B (right, sum intensity at chromosomes) in control (DMSO) or MLN8237-treated prometaphases is shown in the box-plots (center lines show the medians; box limits indicate the 25th and 75th percentiles as determined by R software; whiskers extend 1.5 times the interquartile range from the 25th and 75th percentiles, outliers are represented by dots). Fluorescence intensity is shown in arbitrary units (a.u.). **: p<0.0001, unpaired t test or Mann-Whitney test. n=90 spindle poles (p-Aurora-A) or 50 prometaphases (p-Aurora-B) from 3 experiments. Representative IF images are shown. Scale bars: 10 µm. C. p-Aurora-A (active) levels decrease in mitotic extracts (shake-off) from MLN8237-treated (1 or 4 hours before harvesting) compared to DMSO- or nocodazole (NOC)-treated (4 hours) U2OS cultures. Total Aurora-A levels are also shown; actin is used as loading control. p-Aurora-B was not assessed in Western blot due to the lack of a suitable antibody for this application. www.impactjournals.com/oncotarget 6231 Oncotarget originating two asymmetric daughter cells (Figure 5, third activity assays in Figure 1 indicate dual inhibition of both row). Lower MLN8237 doses (10-20 nM), which yielded Aurora-A and -B under high doses, it was important to disorganized spindles in fixed prometaphases (Figure 4), establish the contribution of each individual kinase to the do not yield multipolar divisions, suggesting that in those no-division phenotype appearing above 50 nM. Small- cultures a bipolar or pseudo-bipolar spindle is eventually scale time-lapse recording of cell cultures subjected to assembled before anaphase. individual (Aurora-A or Aurora-B) or combined (Aurora- In cultures treated with high MLN8237 doses we A+Aurora-B) RNA interference were set up to clarify observed prolonged prometaphase (average duration this issue (Supplementary Figure S3): the no-division 150-200 minutes) followed by a complete lack of phenotype was not recorded in mitoses with selective chromosome segregation and cell division (Figure 5, inactivation of Aurora-A alone, yet appeared in a small fourth row): cells eventually re-adhered to form a single fraction of Aurora-B defective cells (15,4%) and was large or multinucleated interphase, often preceded by amplified by the concomitant inactivation of both kinases repeated “blebbing” movements (Figure 5, fifth row; (41,9%). Supplementary Movie S3). The “lack of division” phenotype appeared in a small fraction of mitoses treated with 50 nM MLN8237 and became predominant (about 90% of mitoses) with 250 nM. Similar phenotypes were previously observed in other human cell lines treated with high MLN8237 concentrations [26, 27]. Since the Figure 2: MLN8237 delays entry into mitosis. A. Mitotic index (MI) from control (DMSO) and MLN8237-treated Figure 3: Dose-dependent lengthening of mitosis in cultures (protocol as in Figure 1A), as assessed by DAPI (DNA) MLN8237-treated cultures. The protocol for time-lapse and alpha-tubulin (spindle) staining. p values (χ2 test) relative to recording of MLN8237-treated cultures progressing towards control cultures, number of scored cells (n) and of independent experiments (exp) are indicated. B. Control (DMSO) and mitosis from thymidine (Thym) arrest and release is depicted on MLN8237-treated cultures were recorded by time-lapse imaging top. Phase-contrast microscopy images were acquired with a 10x objective every 30 minutes. Duration of mitosis is calculated from the treatment start for the following 16 hours. The graph from round-up to visualization of 2 distinct daughter cells; each shows the percentage of interphases entering mitosis during the bar represents a mitotic cell. At least 90 cells per condition are recording period; results are grouped in 4-hours intervals. 250 recorded interphases in 3 experiments for each condition. displayed from 3 experiments. www.impactjournals.com/oncotarget 6232 Oncotarget In summary, these results show that spindle from mitotic round-up (on average 65 minutes with organization is the most sensitive process affected 5-10 nM and 110 with 20 nM MLN8237) compared to by MLN8237 and is readily altered by even a partial control cells (35 minutes). The time from the onset of reduction of Aurora-A activity, associated with unbalanced chromosome segregation to re-formation of daughter chromosome segregation. Stronger inhibition of the kinase interphase cells was instead unaltered, indicating that induces impairment in MT nucleation, associated with the process of chromosome segregation per se was not prolonged prometaphase duration. The highest MLN8237 disrupted. However, we often recorded a delay between dose leads to a complete failure of cell division, largely re-adhesion of the lower and the upper cell (see the ascribable to the concomitant inhibition of Aurora-B. example in Figure 6A and Supplementary Movie S4). No significant induction of mis-oriented division was observed above 50 nM MLN8237 (Figure 6B). In the case Aurora-A inactivation induces defects in the orientation of cell division The time-lapse recording experiments also revealed that a fraction of cells did not divide parallel to the growing surface: that was already evident under conditions of partial Aurora-A inhibiton, occurring in about 15% of mitoses in U2OS cultures treated with 5, 10 or 20 nM MLN8237 (Figure 6A-B and Supplementary Movie S4). These cells took a longer time to reach the stage of chromosome segregation Figure 4: Spindle defects in MLN8237- treated mitoses. Cultures harvested 4 hours after MLN8237 treatment (protocol as in Figure 1A) were stained for DNA and alpha- tubulin. Histograms represent the percentage of prometaphases displaying normal or defective spindles (IF panels on top). 250- 300 counted cells per condition from 3 experiments; s.d. are shown. Scale bar: 10 µm. Figure 5: Time-lapse analysis of MLN8237-treated mitoses reveals multipolar and “no division” phenotypes. Cultures treated as in the protocol in Figure 3 were recorded by time- lapse from treatment start for the following 24 hours. DIC images were acquired with a 40x objective every 5 minutes; representative single photograms are shown; time from round-up is indicated. First row: normal mitosis; second and third rows: multipolar mitoses (a, b and c indicate daughter cells); fourth and fifth rows: mitoses passing directly from prometaphase to defective interphase, with (lower) or without (upper) a “blebbing” phase. Defects are quantified (%) in the table below; number of recorded mitoses (n) and independent experiments per condition are indicated. *: 0.01<p<0.02; **: p< 0.001, χ2 test. Scale bar: 10 µm. www.impactjournals.com/oncotarget 6233 Oncotarget of 250 nM MLN8237 treatment, this is consistent with the and that spindle mis-orientation drives the abnormally predominance of the no-division phenotype (see Figure 5). oriented divisions. The absence of mis-oriented divisions with 50 nM, which To gain better resolution we used an U2OS cell induced highly disorganized spindles and/or defective MT line derivative stably expressing fluorescently labelled nucleation (see Figure 4), suggests that MTs are required alpha-tubulin and histone H2B in video recoding assays (Supplementary Figure S4). MLN8237-treated mitoses that did not divide parallel to the culture dish displayed spindle rotation during prometaphase, such that often only one of the two poles was visible (Figure 6A, lower panels). To define the spindle orientation axis, we analyzed fixed samples and measured the angle formed between the centrosome-centrosome axis and the growing surface (Figure 6C). This analysis was performed in Figure 6: Aurora-A inhibition induces mis-oriented cell division. A. An example of normally dividing cell is shown on top. Mis-oriented division in the presence of MLN8237 in U2OS Figure 7: Outcome of mitoses treated with the cells with fluorescently labeled H2B (green) and alpha-tubulin MLN8237 inhibitor. A. Schematization of the protocol for (red) is shown below. Minutes from round-up are indicated. Scale the analysis of the progeny of MLN8237-treated mitoses. B. bars: 10 µm. B. Quantification (%) of mis-oriented cell division The IF panels show different scored categories. Markers and recorded from MLN8237 treatment start for the following 20 color codes for DNA, MTs, nuclear envelope and centrosomes hours (80-200 mitoses per condition from 3-4 independent are indicated on the left. In the upper panels, the inset in the experiments). *: 0.01<p<0.02; **: p< 0.01, χ2 test. C. The angle DAPI channel shows an enlargement of a micronucleus, while between the growth surface and the centrosome-centrosome axis insets in the merged images show enlargements of centrosomes. in MLN8237-treated mitoses is calculated as schematized on top Scale bars: 10 μm. C. Histograms represent the occurrence (%) (centrosomes are in red; see Methods for details); histograms of defects in B. At least 1000 cells per condition were counted in represent the distribution of prometaphases in 3 classes (50-60 3 experiments; s.d. are shown. cells from 3 independent experiments). www.impactjournals.com/oncotarget 6234 Oncotarget prometaphases from cultures treated with 5, 10 or 20 nM (Figure 7). Cells were stained with combinations of DAPI MLN8237 compared to controls. The average angle in and antibodies against alpha-tubulin and pericentrin control prometaphases was 11° and almost doubled (19°, (examples in Figure 7B, upper panels) or lamin-B1 and mean value) in MLN8237-treated cells (p<0,01), reaching alpha- or gamma-tubulin (middle and lower panels) a >30° distortion in about 20% of prometaphases. Thus, and categorized: cells with loss or gain of 1 or few the inhibition of Aurora-A under conditions that do not chromosomes (1-2 micronuclei), polyploid (1 large impair spindle formation altogether influences the proper nucleus with 4 pericentrin or gamma-tubulin signals), orientation of the spindle axis and hence of cell division. binucleated, or multinucleated. Indeed, following treatment with 250 nM MLN8237, most interphases appeared to have undergone Induction of aneuploidy in the progeny of chromosome mis-segregation (>65% multinucleated MLN8237-treated mitoses cells); a smaller fraction (about 10%) became polyploid. Only less than 10% of interphases were apparently normal Some of the defects observed in MLN8237-treated (Figure 7C). Abnormalities were also observed in cells mitoses suggest the possibility that genetically imbalanced generated during treatment with 50 nM MLN8237: some daughter cells are generated. 20% were multinucleated or polyploid and about 15% To address this issue we treated pre-synchronized were binucleated. Interestingly, in about one third of the cells with MLN8237 in G2 as described, fixed the cells binucleated cells the nuclei were not equivalent in size, after 24 hours and screened defects in interphase cells suggesting that they represent aberrant products of the presumably representing the progeny of treated mitoses multipolar/asymmetric divisions (see Figure 5). Time- lapse imaging of fluorescently labelled U2OS cells showed that some interphases remained very close after division with a connecting alpha-tubulin bridge (Supplementary Figure S4), possibly representing intermediate figures before binucleation. 20 nM and 50 nM MLN8237 also yielded a significant induction (14-15%) of cells with micronuclei, indicative of mild aneuploidy (Figure 7C). To assess whether micronuclei reflected genuine chromosome loss events, and hence aneuploidy, we assessed whether they contained whole chromosomes by staining with CREST antibodies to human centromeres (Figure 8A). We observed a 6- and 17-fold increase of CREST-positive micronuclei in 20 and 50 nM MLN8237-treated cultures, respectively, compared to controls. Consistent with this, recording of GFP-labelled chromosomes dynamically visualized chromosome bridges and micronuclei formation; interestingly, these defects were always present in mitoses with multipolar spindles (Supplementary Figure S4). These results indicate different extents of ploidy alterations in MLN8237-treated cultures, which we decided to investigate directly by employing FISH (Fluorescence In Situ Hybridization) analysis. We counted hybridization signals produced by chromosome-specific centromeric probes (chromosome 7 and 11) in interphases originated from MLN8237-treated mitoses (protocol in Figure 7). Most (about 85%) control interphases displayed Figure 8: Ploidy alteration in MLN8237-treated cells. 3-4 signals for both chromosome 7 and 11 (Figure 8B), A. IF images of CREST-positive (left panels) or -negative confirming that U2OS cells are hypertriploid. 20 nM (right panels) micronuclei. The insets show enlargements of MLN8237 induced a slight shift in the frequency of cells micronuclei. Histograms represent interphases with CREST- positive or -negative micronuclei in MLN8237-treated cultures with fewer signals (2-fold increase in interphases with 1-2 (fold-increase relative to controls). 400 cells per condition were signals compared to controls), likely corresponding to the counted from 4 experiments. s.d. are shown. Scale bar: 10 µm. chromosome loss events evidenced by the micronuclei B. Distribution of interphases with different numbers of signals analysis. With 50 nM MLN8237, cells displaying 3-4 for chromosomes 7 and 11 (FISH hybridization). 800-1000 cells signals decreased to about 70% and cells with > 4 signals per condition were counted in 2 experiments. www.impactjournals.com/oncotarget 6235 Oncotarget correspondingly increased; a minor fraction of truly duration) or from the generation of abnormal interphase polyploid cells (≥6 signals for both chromosomes within cells. The kinetics of appearance of mitotic cells depicted a single large nuclei) was also identified, consistent with two waves of division in control cultures (Figure 9C). In the IF results in Figure 7. With 250 nM MLN8237 > 50% MLN8237-treated cultures the first mitotic peaks were interphases had ≥ 6 signals, indicative of gain of complete shifted in time and appeared broader. Both effects were sets of chromosomes, consistent with the lack of division dose-dependent, consistent with our data on mitotic entry observed in time-lapse experiments (Figure 5) and with and duration (Figures 2 and 3). At 250 nM inhibitor, no the frequent multinucleated state depicted in Figure 7. second wave of division was observed. Concomitantly, FACS analysis confirmed the increase in ploidy in about we observed a strong increase of multinucleated cells at 45% of the viable population (data not shown) after 48 and 250 nM and a milder effect at 50 nM MLN8237. Some 96 hours of treatment with 250 nM MLN8237. polyploid cells appeared under these conditions (below Together these analyses indicate that MLN8237 3%). Detection of multinucleated and polyploid cells was treatment yields variable levels of aneuploidy in daughter therefore consistent with the results from fixed samples cells, in a dose-dependent manner. (Figures 7 and 8). Importantly, the induction of cell death remained below 3% throughout the recording time (Figure 9C, bottom right panel), and remained at a similar level Long term fate of MLN8237-treated cells in one time-lapse experiment extended to 65 hours (not shown). The time-lapse analysis thus far indicates that Together, the high-throughput data indicate that MLN8237 (5-250 nM range) induces no significant cell MLN8237 induces a dose-dependent lengthening of death in U2OS mitotic cells. We investigated the long- mitotic progression and the generation of abnormal term outcome of the treatment in high-throughput time- daughter cells that are impaired in re-dividing, with a lapse analyses for a length of time (48 hours) roughly cytostatic effect, while low cytotoxicity is observed within corresponding to 2 division cycles. Using the fluorescently the first 48 hours of treatment. labelled U2OS cell line, we analyzed the data with an We also performed proliferation/cell death analyses automated method, the CellCognition software [28], trained to classify cells as interphasic, mitotic, dead, multinucleated or polyploid (examples in Figure 9A). In the 48 hours of the recording time, the cell number increased threefold in control cultures, yet dose-dependent growth inhibition was observed in cultures treated with MLN8237, with almost no increase with 250 nM MLN8237 (Figure 9B, left panel). Consistent with this, the number of normal interphase cells was dramatically reduced by the treatment (Figure 9B, right panel): this effect could result either from an increase in mitotic cells (due to prolonged mitotic Figure 9: Long-term high-throughput analysis of MLN8237-treated cultures. Cultures treated as in Fig. 3 were recorded for 48 hours. Automated segmentation and classification on the images was performed using the CellCognition software. A. Classes defined for training the classifier are shown with representative examples for each class. B. Histograms on the left represent the increase in the number of viable cells from the first to the last frame of the acquisition. Average values and s.d. of 10 replicates from 3 independent experiments are shown. The analyzed sample size at time 0 was at least 500 cells per replicate. Histograms on the right represent the increase in normal interphases only under the same conditions. C. The line charts represent the percentage of cells per class during the recording time. www.impactjournals.com/oncotarget 6236 Oncotarget after several days (Figure 10) by cell counting and FACS genuinely inhibit Aurora-A does not efficiently promote measurement of the DNA content. After 48 hours of cell death in U2OS cells; a toxic effects is only observed treatment, corresponding to the end of the high-throughput after very long exposure times to high doses that video-recording, a slight decrease in the total cell number simultaneously target Aurora-B. was observed with 5-10-20 nM MLN8237 compared to control cultures (Figure 10A); the overall proliferation DISCUSSION trend remained comparable to controls over 96 hours. Cell growth inhibition was instead appreciated with 50 Here we have used the MLN8237 small molecule nM MLN8237 and even more effectively with 250 nM inhibitor to investigate mitotic roles of the Aurora-A MLN8237 (Figure 10A). kinase in osteosarcoma U2OS cells; this molecule is To assess cell death we measured cells with a under clinical trial in several cancer types including sub-G1 DNA content by FACS analysis (Figure 10B). That osteosarcoma. revealed a generally higher level of cell death compared We first determined the extent and specificity of to the microscopy analysis, possibly reflecting technical kinase inhibition by assessing Aurora-A and Aurora-B specificities in the methodology: indeed, detached self-phosphorylation. MLN8237 was effective over dead cells are counted by cytofluorimetry, while being Aurora-A in the 20-50 nM range, but also inhibited preferentially lost in microscopy analysis. Nevertheless, Aurora-B above 50 nM. The loss of selectivity with high by FACS analysis, only 250 nM MLN8237 induced doses was previously reported [25-27]. The inhibition of remarkable cell death (about 30% of sub-G1 cells after 48 Aurora-B identified here with 50 nM MLN8237 (a very hours of treatment, increasing to about 50% after 96 hours, close condition to that required for complete Aurora-A compared to about 10% in control cultures; Figure 10B). inhibition) was instead not noticed before using histone H3 Thus, MLN8237 treatment in concentrations that phosphorylation as a reporter, indicating that the selection of activity reporters is critical to ascertain the selectivity of kinase inhibitors. It also evidences that the window at which MLN8237 fully inhibits Aurora-A, without concomitantly affecting Aurora-B, can be very narrow in some cell lines, indicating that selective inhibition of Aurora-A vs Aurora-B remains a critical issue, even with the best performing ATP-competitors. MLN8237 treatment prolongs the G2 phase duration. The G2 delay was under-appreciated in previous studies using MLN8237 in asynchronous cultures, yet was observed when Aurora-A was inhibited by either antibody microinjection or RNA interference [1, 29], or by conditional ablation in mouse embryonic fibroblasts (MEFs; [30]). No permanent arrest is however induced, suggesting that Aurora-A functions in G2 are important, but can be taken over by other kinases. Indeed, the G2 delay is more severe with 250 nM, under which condition Aurora-B is also inhibited. These findings raise the possibility that centrosome- and/or MT-associated defects induced by Aurora-A inactivation in G2 evoke a checkpoint response that delays the transition towards mitosis onset. Dose-response analyses of MLN8237-treated cells that entered mitosis revealed processes that are differentially sensitive to Aurora-A inactivation, as depicted in the schematics in Figure 11. Complete Aurora-A inhibition impaired MT nucleation; consistent with this, previous data showed that partial vs complete Figure 10: Dose-dependent effects of MLN8237 on cell RNA interference-mediated depletion of Aurora-A growth and viability. A. Cells were counted after 48 and differentially affects maturation of centrosomes, required 96 hours from MLN8237 treatment start. Values represent the for mitotic MT nucleation [1, 2]. Partial Aurora-A increase in the number of cells respect to t=0h (3 independent inhibition (10 nM MLN8237) instead yielded multipolar experiments). B. The percentage of sub-G1 cells in MLN8237- or disorganized spindles, the frequency of which treated cultures, detected by FACS analysis, is shown in the increased in a dose-dependent manner up to 50 nM. histograms (3 experiments). s.d. are shown. www.impactjournals.com/oncotarget 6237 Oncotarget Defective spindle assembly was associated with longer is implicated in spindle orientation in asymmetric prometaphase duration compared to untreated cells. The cell divisions in Drosophila [34] via phosphorylation highest frequency of induction of spindle organization of Pins [35]. Excess Aurora-A can also influence the defects (50 nM) was associated with multipolar divisions. spindle orientation and the cell fate in human mammary 250 nM MLN8237 also prolonged prometaphase epithelium stem/progenitor cells [36]. In our time-lapse duration, after which cells re-adhered and eventually assays using a U2OS cell line with fluorescent MTs, we exited mitosis without division. The failure of actually visualized spindle rotation movements: thus, chromosome segregation could be appreciated in time- Aurora-A activity is required for pathways that determine lapse experiments and was consistent with previous results the mitotic spindle orientation, raising the possibility that obtained from concomitant inactivation of Aurora-A and Aurora-A inhibition influences the fate of asymmetrically B kinases by high MLN8237 doses in Hep3B or HeLa dividing cells and/or tissue architecture. cells or, to a lesser extent, following Aurora-B inhibition Time-lapse experiments also revealed that in HeLa cells [26, 27]. Our own recording experiments of MLN8237 generates chromosome mis-segregation. At interfered mitoses for Aurora-A, or -B, or both, confirm a high concentrations, when both Aurora-A and Aurora-B mild effect of Aurora-B inactivation alone and a synergic are inhibited, massive aneuploidy is observed, with effect of the inactivation of both kinases. Complementary the generation of multinucleated daughter cells. Lower functions of Aurora-A and B in chromosome segregation doses that inhibit Aurora-A alone yield mild aneuploidy, KO also emerged in chicken DT40 Aurora-A cells treated with increased number of CREST-positive micronuclei with an Aurora-B specific inhibitor [31]. None of the (schematics in Figure 11). FISH analysis depicted the approaches used to inactivate Aurora-A alone in human differential generation of mild and massive aneuploidy cultured cells yielded chromosome segregation failure [2, by low and high MLN8237 doses, respectively. This is 27, 32]. The latter was instead described in Aurora-A-null relevant, given the pro- or anti-tumorigenic effects of MEFs [30, 33], suggesting specific modes of action of the aneuploidy depending on the extent of chromosome mis- two kinases in this cellular context, which may reflect a segregation [24]. Consistent with our observations, Yang different stoichiometry between Aurora-A and B and/or and colleagues [37] described the generation of an 8N their substrates. population following simultaneous RNA interference- An interesting finding from this study is the mediated inactivation of Aurora-A and Aurora-B, but not induction of mis-oriented divisions by low MLN8237 of Aurora-A alone, in U2OS cells. Segregation defects concentrations (5-20 nM); the inhibitor is selective for (lagging chromosomes and chromatin bridges) were Aurora-A at these doses and, as recalled above, does observed in Aurora-A-null MEFs, with increased ploidy not impair MT nucleation: this directly implicates the over time [30, 33], again evidencing a crucial contribution lack of Aurora-A in the mis-orientation phenotype and of Aurora-A to chromosome segregation in this system. suggests that MTs are required. Interestingly, Aurora-A We also observed binucleated cells in the progeny of MLN8237-treated mitoses; in time-lapse analyses, telophase cells often remained in closer proximity to one another compared to controls, and tubulin bridges persisted, a phenotype that may possibly evolve into a binucleated cell. The observation of binucleated cells with 20 nM MLN8237 would support a recently proposed direct role of Aurora-A in late steps of cell division [6, 7]. Importantly, MLN8237 failed to induce cell death in mitosis nor did it cause a highly effective elimination of the cellular offspring within the first 48 hours of treatment. Recent time-lapse studies with MLN8237 reported variable results regarding mitotic cell death: no mitotic toxicity was observed in Hep3B cells [26], whereas some death from mitosis was recorded in HeLa cells [27]. Aurora-A inactivation therefore appears to require a set of concomitant conditions, as yet elusive, Figure 11: Dose-dependent effects of MLN8237 on for mitotic cell death activation. The highly aneuploid cell division and aneuploidy induction: a schematic progeny generated in our assays at 50 nM (and, to a higher overview. Schematic representation of the phenotype distribution (Up: mitotic defects. Low: abnormal daughter extent, 250 nM) MLN8237 originates cells impaired in interphases) along a MLN8237 concentration gradient in further cell division, hence unviable in the long term. The U2OS cells. Gradual inhibition of Aurora-A (blue) and mechanisms underlying the cytostatic effects of Aurora-A Aurora-B (orange) is indicated by the gradients within shapes inactivation are controversial: a dose- and time-dependent (dose intervals are not in scale); note that at 20 nM MLN8237 induction of apoptosis was described in different cell Aurora-A inhibition is virtually complete. www.impactjournals.com/oncotarget 6238 Oncotarget lines treated with MLN8237 [18, 19, 38, 39], while in FS and FL-2, in a log scale. Cell aggregates were gated other cases the MLN8237-induced cytostatic effect is out on the bi-parametric graph FL-3lin/Ratio. Apoptosis attributed to senescence [40] consistent with results was determined as the proportion of cells exhibiting a described in Aurora-A-null MEFs [33], or to induction DNA content lower than that of G1 cells after gating out of differentiation pathways [41]. These observations cell debris on the bi-parametric graph FS/SS, using the suggest that both the treatment parameters and the WinMDI software. cellular background contribute to determine the long-term outcome of MLN8237-treated cultures. IF In conclusion, the broad variability in the U2OS cell response to MLN8237 highlighted in this study is an Cells grown on coverslips were fixed as follows: important issue in the light of the use of this compound (a) –20°C methanol, 6 minutes; or, for phospho- in anti-cancer therapy. In the human organism undergoing Aurora-B (Thr232) staining, (b) 3.7% PFA in PBS, 10 treatment, the compound dose cannot be constant and is minutes at room temperature followed by 5 minutes in expected to rise and fall over time. It will be important 0.1% TritonX-100 in PBS. Blocking and all antibody to extend these studies and shed light on the pathways incubations were performed at room temperature in driving the response towards one or another cell fate, with PBS/0.05% Tween 20/3% BSA. Cells were counterstained the perspective to modulate such choice and drive cells with 4,6-diamidino-2-phenylindole (DAPI, 0.1 µg/ml) and towards death pathways. mounted using Vectashield (Vector Laboratories). Primary antibodies were: mouse anti-alpha-tubulin (1:2000, B-5-1- METHODS 2, Sigma-Aldrich), mouse anti-Aurora-A (0.5 µg/ml, BD Transduction Laboratories), rabbit anti-phospho-Aurora-A (Thr288) (1:250, C39D8; Cell Signaling Technology), Cell cultures, synchronization protocols and rabbit anti-pericentrin (2 µg/ml, ab4448; Abcam), rabbit treatments anti-phospho-Aurora-B (Thr232) (1:50, Poly6361, BioLegend), mouse anti-phospho-Histone-H3 (Ser 10) (0.25 µg/ml, clone 3H10, Millipore), mouse anti-γ-tubulin The human U2OS osteosarcoma cell line (ATCC: 0 (1:1000, GTU-88, Sigma-Aldrich), rabbit anti-lamin B1 HTB-96) was grown at 37 C in a 5% CO atmosphere in (1 µg/ml, ab16048; Abcam), human anti-centromere DMEM, supplemented with 10% faetal bovine serum. (CREST; 1:20, Antibodies Incorporated). Samples were U2OS cells stably expressing H2B-GFP and RFP-alpha- analyzed using a Nikon Eclipse 90i microscope equipped tubulin (kind gift of L. Lanzetti, Institute for Cancer with a Qicam Fast 1394 CCD camera (QImaging). Image Research at Candiolo, Italy) were cultured under the same acquisition, deconvolution and Extended Depth of Focus conditions and used in time-lapse imaging experiments. on Z-serial optical sections were performed using Nis- For synchronization, cells were subjected to a 24 hours Elements AR 4.2 (Nikon); images were further processed block in 2 mM thymidine. Cultures were then released with Adobe Photoshop CS 8.0. from the G1/S arrest by washing away the thymidine and adding fresh medium containing 30 µM deoxycytidine; 6 Quantitative analysis of IF signals. hours post-release cultures were treated with MLN8237 (5-250 nM as indicated; Selleck Chemicals) or 0.1 µg/ml nocodazole (Sigma Aldrich). Mock-treated cultures were Signals were measured using Nis Elements AR incubated with DMSO. 4.2 (nd2 file format). Analysis of endogenous Aurora-A and Aurora-B activity in mitotic cells was performed Cell counting and FACS analysis as follows: a) p-Thr288-Aurora-A staining: average pixel intensity at spindle poles, corrected for external background; b) p-Thr232-Aurora-B staining: sum intensity Cells were harvested after 48 and 96 hours from at chromosomes, corrected for external background. treatment. One sample was harvested when cultures were Images for quantification of mitotic signals were released in thymidine–free medium (t=0h) as reference for Maximum Intensity Projections from z-stacks (0.6 µm, the initial number of cells and for verifying the efficacy of ranging over a 5-10 µm). Box-plots were generated using the thymidine arrest. For counting the number of cells a the web-tool BoxPlotR. Z1 Coulter Particle Counter (Beckman Coulter) was used. For measuring the angle between the centrosome- For FACS analysis samples were permeabilized with centrosome axis and the culture surface z-stacks serial 0.1% TritonX-100 in PBS. Cell cycle phase distribution images were used. The “arctan(xy/z)” formula for was analyzed after incubation with propidium iodide (PI, calculating the angle of a right triangle was applied with Sigma P4170, 0.04 mg/ml) using a flow cytofluorimeter “xy” being the distance between centrosomes in xy in the Epics XL apparatus (Beckman Coulter). Parameters SS maximum intensity projection, and “z” being the distance and FL-3 were acquired in a linear amplification scale, www.impactjournals.com/oncotarget 6239 Oncotarget between centrosomes along the z-axis. A schematization conditions, time-lapse acquisition, starting 6 hours after is shown in Figure 6C. Values were statistically analyzed the treatment, was extended for 8 hours only to avoid using the InStat3 software, using either (i) the unpaired phototoxic effects. t test (for Gaussian distributions), applying the Welch For high-throughput experiments, cells were seeded correction when required, or (ii) the Mann-Whitney in 96-well plates. Images were acquired with a ScanR test, when the populations did not follow a Gaussian microscope (Olympus) using a 10x objective, an Olympus distribution. DBH1 camera and the ScanR acquisition software. Temperature and CO were kept constant by an incubation system on the microscope. Acquisition was performed Fluorescent in situ hybridization (FISH) every 30 minutes using Phase contrast and fluorescence imaging, for a total duration of 48-65 hours. Cells grown on coverslips were fixed at room temperature with 3:1 methanol:acetic acid, air dried and Automated analysis of high-throughput data stored at −20°C for at least one day. Before hybridization, coverslips were again incubated in 3:1 methanol:acetic acid, 1 hour at room temperature, then heated 2 hours Data from high-throughput experiments were at 65 C and dehydrated in 70%-90%-100% cold analyzed by the CellCognition software (v 1.3.3-28, ethanol. Denaturation of probes in the hybridization mix [28]). The software was used to segment the cells, extract (Fluorescein-labeled Chromosome 7 Satellite Probe, features and classify them using a support vector machine. cat: PSAT0007-G; rhodamine-labeled Chromosome 11 The classes defined were: interphases, prometaphases, Satellite Probe, cat: PSAT001-R; Hybridization Buffer metaphases, ana-telophases, multinucleated interphases, QB007, all from Q-BIOgene) was performed at 96 C for poliployd cells, dead cells; an additional class of debris or 10 minutes. When the mix was applied on coverslips a of cells with non homogeneous morphology which could co-denaturation step was performed (2 minutes, 72 °C), not be included in any of the other categories was created followed by the hybridization incubation, performed at and was then excluded from subsequent data elaboration to 37 C, overnight. Coverslips were then washed in SSC, at avoid artifacts. The classifier was trained with images from 0 0 37 C and 60 C, and counterstained with DAPI (0.2 µg/ independent experiments with a training set comprising a ml) in SSC, 10 minutes at room temperature. Coverslips minimum of 100 cells per class (with the exception of the were mounted with Vectashield and analyzed with a Nikon polyploid class, of which only 33 examples were found). Eclipse 90i microscope, using a 20x objective (Plan Fluor, A confusion matrix and a classification test were used to 0.5 N.A.). Signals per nucleus were counted using the assess the quality of classification before the analysis. The Object count function of Nis Elements AR 4.2 (nd2 file output data were further analyzed using Microsoft Excel; format), setting the parameters using the Spot detection a single class for Mitoses was generated by pooling the function. prometaphase, metaphase and ana-telophase classes. Values for each well were pooled from 6 imaging fields; average values from 4 replicates within one experiment Time-lapse recording are shown in Figure 9. A similar trend was observed in 2 independent experiments (6 additional replicates). Cells seeded in 35 mm dishes (ibiTreat, cod. 81156, or glass bottom, cod. 81158, both from Ibidi) or 8-well Immunoblotting micro-slides (ibiTreat, cod. 80826, Ibidi) were observed under an Eclipse Ti inverted microscope (Nikon), using 60x [Plan Apo, 1.4 N.A. Differential Interference Contrast Mitotic cells collected by shake off were lysed in (DIC), oil immersion] or 40x (Plan Fluor, 0.60 N.A. DIC) RIPA buffer (50 mM Tris-HCl pH 8.0, 150 mM NaCl, objectives (Nikon); during the whole observation cells 1% NP40, 1 mM EGTA, 0.25% sodium deoxycholate) were kept in a microscope stage incubator (Basic WJ, supplemented with protease and phosphatase inhibitors. Okolab), at 37°C and 5% CO . DIC images were acquired Proteins were resolved by electrophoresis on 10% every 5 minutes over 24 hours using a DS-Qi1Mc camera Laemmli gel and transferred on a nitrocellulose membrane and the NIS-Elements AR 3.22 software (Nikon). Image (Protran BA83, Whatman) using a semi-dry system (BIO- and movie processing were performed with NIS-Elements RAD). 40 µg of extract per lane were loaded. Blocking AR 4.2. and antibody incubations were performed at room Imaging of the U2OS cell line stably expressing temperature in PBS/0.1% Tween 20/5% low fat milk, H2B-GFP and RFP-alpha-tubulin was performed with the or in PBS/0.1% Tween 20/5% BSA (for anti-phospho- 60x objective: images were acquired every 7 minutes in Aurora-A hybridization). Antibodies were: mouse anti- the 3 channels. z-stacks of the fluorescent channels were Aurora-A (1 µg/ml; BD Transduction Laboratories), rabbit acquired every 1 µm over a range of 8 µm, attenuating anti-phospho-Aurora-A (Thr288) (1:1000; C39D8; Cell the fluorescence lamp intensity to 1/32. Under these Signaling Technology), goat anti-actin (0.5 µg/ml, I-19; www.impactjournals.com/oncotarget 6240 Oncotarget SantaCruz Biotechnology). Signals were visualized by chromosomal passenger complex (CPC): from easy rider to enhanced chemiluminescence detection (ECL plus, GE the godfather of mitosis. Nat Rev Mol Cell Biol. 2012; 13: Healthcare, and Protein Detection System, GeneSpin). 789-803. 9. Lapenna S, Giordano A. Cell cycle kinases as therapeutic targets for cancer. Nat Rev Drug Discov. 2009; 8: 547-566. ACKNOWLEDGEMENTS 10. Lens SM, Voest EE, Medema RH. Shared and separate We are grateful to Dr. Rainer Pepperkok and the functions of polo-like kinases and aurora kinases in cancer. EuroBioImaging project for access to the Advanced Light Nat Rev Cancer. 2010; 10: 825-841. Microscopy Facility of the EMBL (Heidelberg) and to the 11. Kollareddy M, Zheleva D, Dzubak P, Brahmkshatriya PS, team at the Facility for advice and technical help. We thank Lepsik M, Hajduch M. Aurora kinase inhibitors: progress Dr. Letizia Lanzetti for providing the U2OS cell line stably towards the clinic. Invest New Drugs. 2012; 30: 2411-2432. expressing fluorescent GFP-H2B and RFP-alpha-tubulin 12. Cervantes A, Elez E, Roda D, Ecsedy J, Macarulla T, and Dr. Francesca Degrassi for reagents and advice with Venkatakrishnan K, Roselló S, Andreu J, Jung J, Sanchis- FISH analysis. We are grateful to Dr. Pietro Cirigliano and Garcia JM, Piera A, Blasco I, Maños L, et al. Phase I to staff of Nikon Instruments S.p.A. (Italy) for continuous pharmacokinetic/pharmacodynamic study of MLN8237, collaborative support for imaging experiments performed an investigational, oral, selective aurora a kinase inhibitor, at the CNR-IBMN Nikon Reference Center. This work in patients with advanced solid tumors. Clin Cancer Res. was supported by grants from AIRC-Italian Association 2012; 18: 4764-4774. for Cancer Research (MFAG-13350 to G.G. and IG14534 13. Dees EC, Cohen RB, von Mehren M, Stinchcombe TE, to P.L.) and Fondazione Monte dei Paschi di Siena (to Liu H, Venkatakrishnan K, Manfredi M, Fingert H, Burris G.G. and P.L.). Visits of IAA to the ALMF were supported HA 3rd, Infante JR. Phase I study of aurora A kinase by an EMBO Short-term fellowship and a SIBBM travel inhibitor MLN8237 in advanced solid tumors: safety, grant. pharmacokinetics, pharmacodynamics, and bioavailability of two oral formulations. Clin Cancer Res. 2012; 18: 4775- REFERENCES 14. Matulonis UA, Sharma S, Ghamande S, Gordon MS, Del 1. Hirota T, Kunitoku N, Sasayama T, Marumoto T, Zhang Prete SA, Ray-Coquard I, Kutarska E, Liu H, Fingert D, Nitta M Hatakeyama K, Saya H. Aurora-A and an H, Zhou X, Danaee H, Schilder RJ. Phase II study of interacting activator, the LIM protein Ajuba, are required MLN8237 (alisertib), an investigational Aurora A kinase for mitotic commitment in human cells. Cell. 2003; 114: inhibitor, in patients with platinum-resistant or -refractory 585-598. epithelial ovarian, fallopian tube, or primary peritoneal 2. Marumoto T, Honda S, Hara T, Nitta M, Hirota T, Kohmura carcinoma. Gynecol Oncol. 2012; 127: 63-69. E, Saya H. Aurora-A kinase maintains the fidelity of early 15. Doménech E, Malumbres M. Mitosis-targeting therapies: and late mitotic events in HeLa cells. J Biol Chem. 2003; a troubleshooting guide. Curr Opin Pharmacol. 2013; 13: 278: 51786-51795. 519-528. 3. Katayama H, Sasai K, Kloc M, Brinkley BR, Sen S. Aurora 16. Friedberg JW, Mahadevan D, Cebula E, Persky D, Lossos kinase-A regulates kinetochore/chromatin associated I, Agarwal AB, Jung J, Burack R, Zhou X, Leonard EJ, microtubule assembly in human cells. Cell Cycle. 2008; 7: Fingert H, Danaee H, Bernstein SH. Phase II study of 2691-2704. alisertib, a selective Aurora A kinase inhibitor, in relapsed 4. De Luca M, Lavia P, Guarguaglini G. A functional interplay and refractory aggressive B- and T-cell non-Hodgkin between Aurora-A, Plk1 and TPX2 at spindle poles: Plk1 lymphomas. J Clin Oncol. 2014; 32: 44-50. controls centrosomal localization of Aurora-A and TPX2 17. Manfredi MG, Ecsedy JA, Chakravarty A, Silverman spindle association. Cell Cycle. 2006; 5: 296-303. L, Zhang M, Hoar KM, Stroud SG, Chen W, Shinde V, 5. De Luca M, Brunetto L, Asteriti IA, Giubettini M, Lavia P, Huck JJ, Wysong DR, Janowick DA, Hyer ML et al. Guarguaglini G. Aurora-A and ch-TOG act in a common Characterization of Alisertib (MLN8237), an investigational pathway in control of spindle pole integrity. Oncogene. small-molecule inhibitor of aurora A kinase using novel in 2008; 27: 6539-6549. vivo pharmacodynamic assays. Clin Cancer Res. 2011; 17: 6. Lioutas A, Vernos I. Aurora A kinase and its substrate 7614-7624. TACC3 are required for central spindle assembly. EMBO 18. Kelly KR, Nawrocki ST, Espitia CM, Zhang M, Yang JJ, Rep. 2013; 14: 829-836. Padmanabhan S, Ecsedy J, Giles FJ, Carew JS. Targeting 7. Reboutier D, Troadec MB, Cremet JY, Chauvin L, Guen V, Aurora A kinase activity with the investigational agent Salaun P, Prigent C. Aurora A is involved in central spindle alisertib increases the efficacy of cytarabine through a assembly through phosphorylation of Ser 19 in P150Glued. FOXO-dependent mechanism. Int J Cancer. 2012; 131: J Cell Biol. 2013; 201: 65-79. 2693-2703. 8. Carmena M, Wheelock M, Funabiki H, Earnshaw WC. The www.impactjournals.com/oncotarget 6241 Oncotarget 19. Qi W, Spier C, Liu X, Agarwal A, Cooke LS, Persky DO, 31. Hégarat N, Smith E, Nayak G, Takeda S, Eyers PA, Chen D, Miller TP, Mahadevan D. Alisertib (MLN8237) an Hochegger H. Aurora A and Aurora B jointly coordinate investigational agent suppresses Aurora A and B activity, chromosome segregation and anaphase microtubule inhibits proliferation, promotes endo-reduplication and dynamics. J Cell Biol. 2011; 195: 1103-1113. induces apoptosis in T-NHL cell lines supporting its 32. Du J, Hannon GJ. Suppression of p160ROCK bypasses cell importance in PTCL treatment. Leuk Res. 2013; 37: 434- cycle arrest after Aurora-A/STK15 depletion. Proc Natl Acad Sci U S A. 2004; 101: 8975-8980. 20. Maris JM, Morton CL, Gorlick R, Kolb EA, Lock R, Carol 33. Pérez de Castro I, Aguirre-Portolés C, Fernández-Miranda H, Keir ST, Reynolds CP, Kang MH, Wu J, Smith MA, G, Cañamero M, Cowley DO, Van Dyke T, Malumbres Houghton PJ. Initial testing of the aurora kinase A inhibitor M. Requirements for Aurora-A in tissue regeneration and MLN8237 by the Pediatric Preclinical Testing Program tumor development in adult mammals. Cancer Res. 2013; (PPTP). Pediatr Blood Cancer. 2010; 55: 26-34. 73: 6804-6815. 21. Zhou N, Singh K, Mir MC, Parker Y, Lindner DJ, Dreicer 34. Lee CY, Andersen RO, Cabernard C, Manning L, Tran KD, R, Ecsedy JA, The BT, Zhang Z, Almasan A, Hansel DE. Lanskey MJ, Bashirullah A, Doe CQ. Drosophila Aurora-A The investigational Aurora kinase A inhibitor MLN8237 kinase inhibits neuroblast self-renewal by regulating aPKC/ induces defects in cell viability and cell-cycle progression Numb cortical polarity and spindle orientation. Genes Dev. in malignant bladder cancer cells in vitro and in vivo. Clin 2006; 20: 3464-3474. Cancer Res. 2013; 19: 1717-1728. 35. Johnston CA, Hirono K, Prehoda KE, Doe CQ. 22. Gascoigne KE, Taylor SS. Cancer cells display profound Identification of an Aurora-A/PinsLINKER/Dlg spindle intra- and interline variation following prolonged exposure orientation pathway using induced cell polarity in S2 cells. to antimitotic drugs. Cancer Cell. 2008; 14: 111-122. Cell. 2009; 138: 1150-1163. 23. Asteriti IA, Giubettini M, Lavia P, Guarguaglini G. 36. Regan JL, Sourisseau T, Soady K, Kendrick H, McCarthy Aurora-A inactivation causes mitotic spindle pole A, Tang C, Brennan K, Linardopoulos S, White DE, fragmentation by unbalancing microtubule-generated Smalley MJ. Aurora A kinase regulates mammary epithelial forces. Mol Cancer. 2011; 10: 131. cell fate by determining mitotic spindle orientation in a 24. Weaver BA, Cleveland DW. The aneuploidy paradox in cell Notch-dependent manner. Cell Rep. 2013; 4: 110-123. growth and tumorigenesis. Cancer Cell. 2008; 14: 431-433. 37. Yang H, Burke T, Dempsey J, Diaz B, Collins E, Toth 25. Qi W, Cooke LS, Liu X, Rimsza L, Roe DJ, Manziolli A, J, Beckmann R, Ye X. Mitotic requirement for aurora A Persky DO, Miller TP, Mahadevan D. Aurora inhibitor kinase is bypassed in the absence of aurora B kinase. FEBS MLN8237 in combination with docetaxel enhances Lett. 2005; 579: 3385-3391. apoptosis and anti-tumor activity in mantle cell lymphoma. 38. Kelly KR, Ecsedy J, Medina E, Mahalingam D, Biochem Pharmacol. 2011; 81: 881-890. Padmanabhan S, Nawrocki ST, Giles FJ, Carew JS. The 26. Marxer M, Foucar CE, Man WY, Chen Y, Ma HT, Poon novel Aurora A kinase inhibitor MLN8237 is active in RY. Tetraploidization increases sensitivity to Aurora B resistant chronic myeloid leukaemia and significantly kinase inhibition. Cell Cycle. 2012; 11: 2567-2577. increases the efficacy of nilotinib. J Cell Mol Med. 2011; 15: 2057-2070. 27. Marxer M, Ma HT, Man WY, Poon RY. p53 deficiency enhances mitotic arrest and slippage induced by 39. Sehdev V, Peng D, Soutto M, Washington MK, Revetta F, pharmacological inhibition of Aurora kinases. Oncogene. Ecsedy J, Zaika A, Rau TT, Schneider-Stock R, Belkhiri 2014; 33: 3550-3560. A, El-Rifai W. The aurora kinase A inhibitor MLN8237 enhances cisplatin-induced cell death in esophageal 28. Held M, Schmitz MH, Fischer B, Walter T, Neumann adenocarcinoma cells. Mol Cancer Ther. 2012; 11: 763-774. B, Olma MH, Peter M, Ellenberg J, Gerlich DW. CellCognition: time-resolved phenotype annotation in high- 40. Liu Y, Hawkins OE, Su Y, Vilgelm AE, Sobolik T, Thu throughput live cell imaging. Nat Methods. 2010; 7: 747- YM, Kantrow S, Splittgerber RC, Short S, Amiri KI, Ecsedy 754. JA, Sosman JA, Kelley MC, Richmond A. Targeting aurora kinases limits tumour growth through DNA damage- 29. Marumoto T, Hirota T, Morisaki T, Kunitoku N, Zhang D, mediated senescence and blockade of NF-κB impairs this Ichikawa Y, Sasayama T, Kuninaka S, Mimori T, Tamaki drug-induced senescence. EMBO Mol Med. 2013; 5: 149- N, Kimura M, Okano Y, Saya H. Roles of aurora-A kinase in mitotic entry and G2 checkpoint in mammalian cells. Genes Cells. 2002; 7: 1173-1182. 41. Mohan P, Castellsague J, Jiang J, Allen K, Chen H, Nemirovsky O, Spyra M, Hu K, Kluwe L, Pujana MA, 30. Cowley DO, Rivera-Pérez JA, Schliekelman M, He YJ, Villanueva A, Mautner VF, Keats JJ, et al. Genomic Oliver TG, Lu L, O’Quinn R, Salmon ED, Magnuson imbalance of HMMR/RHAMM regulates the sensitivity and T, Van Dyke T. Aurora-A kinase is essential for bipolar response of malignant peripheral nerve sheath tumour cells spindle formation and early development. Mol Cell Biol. to aurora kinase inhibition. Oncotarget. 2013; 4: 80-93. 2009; 29: 1059-1071. www.impactjournals.com/oncotarget 6242 Oncotarget http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Oncotarget Unpaywall

The Aurora-A inhibitor MLN8237 affects multiple mitotic processes and induces dose-dependent mitotic abnormalities and aneuploidy

Loading next page...
 
/lp/unpaywall/the-aurora-a-inhibitor-mln8237-affects-multiple-mitotic-processes-and-vkyIIK5R0N

References

References for this paper are not available at this time. We will be adding them shortly, thank you for your patience.

Publisher
Unpaywall
ISSN
1949-2553
DOI
10.18632/oncotarget.2190
Publisher site
See Article on Publisher Site

Abstract

www.impactjournals.com/oncotarget/ Oncotarget, Vol. 5, No. 15 The Aurora-A inhibitor MLN8237 affects multiple mitotic processes and induces dose-dependent mitotic abnormalities and aneuploidy 1 1 1 2 Italia Anna Asteriti , Erica Di Cesare , Fabiola De Mattia , Volker Hilsenstein , 2 1 1 1 Beate Neumann , Enrico Cundari , Patrizia Lavia and Giulia Guarguaglini Institute of Biology, Molecular Medicine and Nanobiotechnology (formerly Institute of Molecular Biology and Pathology), CNR National Research Council, c/o Department of Biology and Biotechnology, Sapienza University of Rome, Rome, Italy Advanced Light Microscopy Facility, EMBL, Heidelberg, Germany Correspondence to: Giulia Guarguaglini, email: giulia.guarguaglini@uniroma1.it Keywords: Aurora kinases, MLN8237, mitotic spindle, aneuploidy, time-lapse microscopy Received: March 7, 2014 Accepted: July 8, 2014 Published: July 9, 2014 This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. ABSTRACT Inhibition of Aurora kinase activity by small molecules is being actively investigated as a potential anti-cancer strategy. A successful therapeutic use of Aurora inhibitors relies on a comprehensive understanding of the effects of inactivating Aurora kinases on cell division, a challenging aim given the pleiotropic roles of those kinases during mitosis. Here we have used the Aurora-A inhibitor MLN8237, currently under phase-I/III clinical trials, in dose-response assays in U2OS human cancer cells synchronously proceeding towards mitosis. By following the behaviour and fate of single Aurora-inhibited cells in mitosis by live microscopy, we show that MLN8237 treatment affects multiple processes that are differentially sensitive to the loss of Aurora-A function. A role of Aurora-A in controlling the orientation of cell division emerges. MLN8237 treatment, even in high doses, fails to induce efficient elimination of dividing cells, or of their progeny, while inducing significant aneuploidy in daughter cells. The results of single-cell analyses show a complex cellular response to MLN8237 and evidence that its effects are strongly dose-dependent: these issues deserve consideration in the light of the design of strategies to kill cancer cells via inhibition of Aurora kinases. often abnormally expressed in tumor cells and are being INTRODUCTION investigated as targets of anti-mitotic compounds for cancer therapy [9, 10]. Many efforts have converged in the The Aurora-A kinase is a major regulator of cell last years to develop Aurora inhibitors: molecules acting division and operates in distinct processes required for as ATP-competitors have been identified and some of them spindle assembly: in human cells it regulates separation are currently in clinical trials [11]. Only a few of those and maturation of centrosomes at mitotic entry, mitotic molecules discriminate Aurora-A vs Aurora-B and may microtubule (MT) nucleation [1- 3] and the integrity of thus prove useful both in clinical studies for comparing spindle poles [2, 4, 5]. Recent data also indicate a role the efficacy of anti-tumor responses and for dissecting the of Aurora-A in central spindle assembly at telophase [6, functions of Aurora kinases in mammalian cells. 7]. The highly homologous Aurora-B kinase also operates MLN8237 (Alisertib) is a second generation Aurora in control of the fidelity of chromosome segregation, inhibitor currently undergoing Phase-I/III clinical trials by regulating chromosome condensation, correction of [11-16; www.clinicaltrials.gov]. Thus far, it is one of the improper attachments between MTs and kinetochores, molecules displaying highest specificity for Aurora-A spindle checkpoint function, cytokinesis and abscission over Aurora-B (300-fold in in vitro assays and 200-fold [8]. in HCT116 colorectal carcinoma cells [17]). Most pre- As other mitotic regulators, Aurora kinases are www.impactjournals.com/oncotarget 6229 Oncotarget clinical studies based on whole cell population analyses suitable for high-resolution single-cell microscopy in tumor cell lines showed cell growth inhibition, analysis, and employed in our previous studies of RNA accumulation of polyploid cells over time, as well as interference-mediated Aurora-A inactivation [4, 5, 23]. induction of cell death [17-19]. Anti-tumor activity was We set up a protocol by pre-synchronizing U2OS also demonstrated in xenograft mouse models [17, 20, 21]. cells at the G1/S transition by thymidine treatment, then Available data on MLN8237-treated cells were releasing from arrest into G2 and mitosis (Figure 1A). mostly obtained from asynchronous cultures analyzed in MLN8237 was added 6 hours after thymidine release (late bulk populations. This approach reveals the predominant S-phase/early G2) and cells were harvested after further 4 cellular behaviour after long exposure to Aurora-A hours. Aurora-A activity was measured at the single cell inhibition (24 to 96 hours) but can miss out transient level by anti-Aur-A-phospho-Thr288 immunofluorescence phenomena and so mask the unfolding of relevant (IF) staining in dose-response assays (Figure 1B, left processes. In addition, inhibition of as pleiotropic a kinase panels). Aurora-A auto-phosphorylation was significantly as is Aurora-A, yields multiple phenotypes over time, inhibited at concentrations ranging from 5 nM to 250 nM. making it difficult to dissect distinct functional roles With concentrations higher than 20 nM the residual signal within a bulk population. Microscopy-based single cell at spindle poles was below 15% compared to controls. In analyses are proving of critical importance to visualize Western blot analysis, no phospho-Thr288-Aurora-A was the array of possible cell responses to anti-mitotic drugs detectable in mitotic extracts from cultures treated with [22]. Here we have coupled high resolution microscopy 20 and 50 nM MLN8237 for 4 hours, while some residual and high-throughput analysis of single cells treated with amount was present after 1 hour (Figure 1C). increasing doses of Aurora-A inhibitor to investigate the Previous reports indicated that MLN8237 above possible fates of cells with inactive Aurora-A. 100 nM also inhibits Aurora-B activity in other cell lines A protocol was set up for treating pre-synchronized [19, 25-27]. We therefore assessed the specificity of cultures when they reach G2 and analyze progression MLN8237 by measuring Aurora-B activity using anti-Aur- through G2 and mitosis as soon as Aurora-A inhibition B-phospho-Thr232 antibody (Figure 1B, right panels). is achieved. Because MLN8237 induces spindle pole Surprisingly, we noticed that Aurora-B activity is already abnormalities [23], we assessed the occurrence of significantly compromised by 50 nM MLN8237; that was chromosome mis-segregation events and aneuploidy not evident when using anti-phospho-Histone-H3 (Ser10) induction, which would represent undesirable effects of as a reporter of Aurora-B activity (Supplementary Figure the treatment in anti-cancer therapy. Our results highlight S1), possibly reflecting kinase redundancy or delay in a partial specificity of MLN8237 in the U2OS cell line, detecting modulation of phosphorylation of downstream with multiple cellular responses in a dose-dependent targets vs auto-phosphorylation. manner. The single cell analysis enabled us to depict Our single-cell analysis in U2OS cultures delimits a fraction of cells with defective spindle orientation, therefore a narrow MLN8237 concentration window (20- a defect that was not appreciated in previous studies of 50 nM) yielding effective and specific Aurora-A inhibition. Aurora-A inhibition in human cells. In addition, we find that low and high MLN8237 concentrations yield mild MLN8237 delays mitotic entry and prolongs and massive aneuploidy, respectively, representing a mitotic duration in a dose-dependent manner tumor-inducing or a tumor-suppressing condition [24]. Collectively, these results draw attention to the variability We investigated the influence of MLN8237 on and the nature of cellular responses to the loss of Aurora mitotic entry: after 4 hours of treatment, we found kinase function, which may represent potential caveats a significantly lower percentage of mitotic cells in deserving consideration when designing and interpreting MLN8237-treated cultures compared to controls (Figure clinical trials. 2A). This effect is dose-dependent, appearing at ≥ 20 nM, and is stronger at 250 nM MLN8237. To clarify whether RESULTS cells were arrested in the G2 phase or rather delayed in progression through the G2/M transition, we recorded cultures in time-lapse experiments from the treatment start MLN8237 displays dose-dependent target up to 16 hours later (Figure 2B). The peak of interphases selectivity on Aurora kinases entering mitosis in control cultures (DMSO) was between 4 and 8 hours from the treatment start (Figure 2B) and was not significantly affected by partial Aurora-A Prior to analyzing mitotic division in cells with inhibition (5-10 nM MLN8237). Entry into mitosis was inhibited Aurora-A, we sought to precisely define the instead delayed above 20 nM MLN8237. In the 16 hours specificity of MLN8237 inhibition in dose-response of time-lapse recording, 70-80% of the interphase cells assays. We used the U2OS osteosarcoma cell line for its entered mitosis in all treated cultures (about 60% with ease of cytological analysis, which renders it especially www.impactjournals.com/oncotarget 6230 Oncotarget 250 nM MLN8237): thus, the majority of cells exposed an accumulation of prometaphase figures over all mitoses to MLN8237 are delayed in G2, yet mitotic onset is not in fixed samples (Supplementary Figure S2). The MT prevented. nucleation defect was strongly dose-dependent and Extending the time-lapse recording to 30 hours appeared in a relevant fraction of mitotic cells treated with indicated that MLN8237 prolonged the duration of mitosis 50 nM MLN8237 or above. in a dose-dependent manner (about 300 minutes with 50 In cells in which MT nucleation was not visibly nM, compared to about 80 minutes in control cells; Figure affected, spindles were highly disorganized (affecting 30% 3). Importantly, although slowed down, cells eventually to 60% of all prometaphases with 10 nM or above; Figure exited mitosis. Together, the results indicate that both the 4); a fraction of these prometaphases displayed spindles G2-to-mitosis transition and the overall duration of the with multiple poles, consistent with previous observations mitotic process are strongly dependent on Aurora-A. [17, 23]. A non-significant fraction of monopolar spindles was present at 10-20 nM MLN8237. The influence of these defects on the global Inhibition of Aurora kinases yields impaired MT execution of mitosis was examined in depth in time- nucleation, disorganized spindles and multipolar lapse movies of MLN8237-treated cells. Figure 5, first or failed cell division row, and Supplementary Movie S1 show examples of normal mitosis. We observed that a significant fraction We next analyzed spindle structure in cells that of cells with inhibited Aurora-A (about 20% with 50 entered mitosis with different degrees of Aurora-A and nM MLN8237, and a smaller fraction with 250 nM) Aurora-B inhibition. underwent multipolar ana-telophase (Figure 5, second With the highest MLN8237 concentration (250 row; Supplementary Movie S2). Multipolar mitoses nM) a strong impairment of MT nucleation was evident, took a longer time to reach the stage of chromosome with 70% of prometaphases displaying no MTs (Figure segregation (about 125 minutes average, compared to 35 4); this was associated with a prolonged prometaphase minutes in control cells). In some cases two of the multiple duration in time-lapse recording experiments, yielding groups of segregating chromosomes eventually re-joined, Figure 1: Dose-dependent inhibition of Aurora-A and Aurora-B by MLN8237. A. Protocol for MLN8237 treatment in cells progressing towards mitosis after thymidine (Thym) arrest and release. B. Quantification of IF signals for active pThr288-Aurora-A (left, mean intensity at poles) or active pThr232-Aurora-B (right, sum intensity at chromosomes) in control (DMSO) or MLN8237-treated prometaphases is shown in the box-plots (center lines show the medians; box limits indicate the 25th and 75th percentiles as determined by R software; whiskers extend 1.5 times the interquartile range from the 25th and 75th percentiles, outliers are represented by dots). Fluorescence intensity is shown in arbitrary units (a.u.). **: p<0.0001, unpaired t test or Mann-Whitney test. n=90 spindle poles (p-Aurora-A) or 50 prometaphases (p-Aurora-B) from 3 experiments. Representative IF images are shown. Scale bars: 10 µm. C. p-Aurora-A (active) levels decrease in mitotic extracts (shake-off) from MLN8237-treated (1 or 4 hours before harvesting) compared to DMSO- or nocodazole (NOC)-treated (4 hours) U2OS cultures. Total Aurora-A levels are also shown; actin is used as loading control. p-Aurora-B was not assessed in Western blot due to the lack of a suitable antibody for this application. www.impactjournals.com/oncotarget 6231 Oncotarget originating two asymmetric daughter cells (Figure 5, third activity assays in Figure 1 indicate dual inhibition of both row). Lower MLN8237 doses (10-20 nM), which yielded Aurora-A and -B under high doses, it was important to disorganized spindles in fixed prometaphases (Figure 4), establish the contribution of each individual kinase to the do not yield multipolar divisions, suggesting that in those no-division phenotype appearing above 50 nM. Small- cultures a bipolar or pseudo-bipolar spindle is eventually scale time-lapse recording of cell cultures subjected to assembled before anaphase. individual (Aurora-A or Aurora-B) or combined (Aurora- In cultures treated with high MLN8237 doses we A+Aurora-B) RNA interference were set up to clarify observed prolonged prometaphase (average duration this issue (Supplementary Figure S3): the no-division 150-200 minutes) followed by a complete lack of phenotype was not recorded in mitoses with selective chromosome segregation and cell division (Figure 5, inactivation of Aurora-A alone, yet appeared in a small fourth row): cells eventually re-adhered to form a single fraction of Aurora-B defective cells (15,4%) and was large or multinucleated interphase, often preceded by amplified by the concomitant inactivation of both kinases repeated “blebbing” movements (Figure 5, fifth row; (41,9%). Supplementary Movie S3). The “lack of division” phenotype appeared in a small fraction of mitoses treated with 50 nM MLN8237 and became predominant (about 90% of mitoses) with 250 nM. Similar phenotypes were previously observed in other human cell lines treated with high MLN8237 concentrations [26, 27]. Since the Figure 2: MLN8237 delays entry into mitosis. A. Mitotic index (MI) from control (DMSO) and MLN8237-treated Figure 3: Dose-dependent lengthening of mitosis in cultures (protocol as in Figure 1A), as assessed by DAPI (DNA) MLN8237-treated cultures. The protocol for time-lapse and alpha-tubulin (spindle) staining. p values (χ2 test) relative to recording of MLN8237-treated cultures progressing towards control cultures, number of scored cells (n) and of independent experiments (exp) are indicated. B. Control (DMSO) and mitosis from thymidine (Thym) arrest and release is depicted on MLN8237-treated cultures were recorded by time-lapse imaging top. Phase-contrast microscopy images were acquired with a 10x objective every 30 minutes. Duration of mitosis is calculated from the treatment start for the following 16 hours. The graph from round-up to visualization of 2 distinct daughter cells; each shows the percentage of interphases entering mitosis during the bar represents a mitotic cell. At least 90 cells per condition are recording period; results are grouped in 4-hours intervals. 250 recorded interphases in 3 experiments for each condition. displayed from 3 experiments. www.impactjournals.com/oncotarget 6232 Oncotarget In summary, these results show that spindle from mitotic round-up (on average 65 minutes with organization is the most sensitive process affected 5-10 nM and 110 with 20 nM MLN8237) compared to by MLN8237 and is readily altered by even a partial control cells (35 minutes). The time from the onset of reduction of Aurora-A activity, associated with unbalanced chromosome segregation to re-formation of daughter chromosome segregation. Stronger inhibition of the kinase interphase cells was instead unaltered, indicating that induces impairment in MT nucleation, associated with the process of chromosome segregation per se was not prolonged prometaphase duration. The highest MLN8237 disrupted. However, we often recorded a delay between dose leads to a complete failure of cell division, largely re-adhesion of the lower and the upper cell (see the ascribable to the concomitant inhibition of Aurora-B. example in Figure 6A and Supplementary Movie S4). No significant induction of mis-oriented division was observed above 50 nM MLN8237 (Figure 6B). In the case Aurora-A inactivation induces defects in the orientation of cell division The time-lapse recording experiments also revealed that a fraction of cells did not divide parallel to the growing surface: that was already evident under conditions of partial Aurora-A inhibiton, occurring in about 15% of mitoses in U2OS cultures treated with 5, 10 or 20 nM MLN8237 (Figure 6A-B and Supplementary Movie S4). These cells took a longer time to reach the stage of chromosome segregation Figure 4: Spindle defects in MLN8237- treated mitoses. Cultures harvested 4 hours after MLN8237 treatment (protocol as in Figure 1A) were stained for DNA and alpha- tubulin. Histograms represent the percentage of prometaphases displaying normal or defective spindles (IF panels on top). 250- 300 counted cells per condition from 3 experiments; s.d. are shown. Scale bar: 10 µm. Figure 5: Time-lapse analysis of MLN8237-treated mitoses reveals multipolar and “no division” phenotypes. Cultures treated as in the protocol in Figure 3 were recorded by time- lapse from treatment start for the following 24 hours. DIC images were acquired with a 40x objective every 5 minutes; representative single photograms are shown; time from round-up is indicated. First row: normal mitosis; second and third rows: multipolar mitoses (a, b and c indicate daughter cells); fourth and fifth rows: mitoses passing directly from prometaphase to defective interphase, with (lower) or without (upper) a “blebbing” phase. Defects are quantified (%) in the table below; number of recorded mitoses (n) and independent experiments per condition are indicated. *: 0.01<p<0.02; **: p< 0.001, χ2 test. Scale bar: 10 µm. www.impactjournals.com/oncotarget 6233 Oncotarget of 250 nM MLN8237 treatment, this is consistent with the and that spindle mis-orientation drives the abnormally predominance of the no-division phenotype (see Figure 5). oriented divisions. The absence of mis-oriented divisions with 50 nM, which To gain better resolution we used an U2OS cell induced highly disorganized spindles and/or defective MT line derivative stably expressing fluorescently labelled nucleation (see Figure 4), suggests that MTs are required alpha-tubulin and histone H2B in video recoding assays (Supplementary Figure S4). MLN8237-treated mitoses that did not divide parallel to the culture dish displayed spindle rotation during prometaphase, such that often only one of the two poles was visible (Figure 6A, lower panels). To define the spindle orientation axis, we analyzed fixed samples and measured the angle formed between the centrosome-centrosome axis and the growing surface (Figure 6C). This analysis was performed in Figure 6: Aurora-A inhibition induces mis-oriented cell division. A. An example of normally dividing cell is shown on top. Mis-oriented division in the presence of MLN8237 in U2OS Figure 7: Outcome of mitoses treated with the cells with fluorescently labeled H2B (green) and alpha-tubulin MLN8237 inhibitor. A. Schematization of the protocol for (red) is shown below. Minutes from round-up are indicated. Scale the analysis of the progeny of MLN8237-treated mitoses. B. bars: 10 µm. B. Quantification (%) of mis-oriented cell division The IF panels show different scored categories. Markers and recorded from MLN8237 treatment start for the following 20 color codes for DNA, MTs, nuclear envelope and centrosomes hours (80-200 mitoses per condition from 3-4 independent are indicated on the left. In the upper panels, the inset in the experiments). *: 0.01<p<0.02; **: p< 0.01, χ2 test. C. The angle DAPI channel shows an enlargement of a micronucleus, while between the growth surface and the centrosome-centrosome axis insets in the merged images show enlargements of centrosomes. in MLN8237-treated mitoses is calculated as schematized on top Scale bars: 10 μm. C. Histograms represent the occurrence (%) (centrosomes are in red; see Methods for details); histograms of defects in B. At least 1000 cells per condition were counted in represent the distribution of prometaphases in 3 classes (50-60 3 experiments; s.d. are shown. cells from 3 independent experiments). www.impactjournals.com/oncotarget 6234 Oncotarget prometaphases from cultures treated with 5, 10 or 20 nM (Figure 7). Cells were stained with combinations of DAPI MLN8237 compared to controls. The average angle in and antibodies against alpha-tubulin and pericentrin control prometaphases was 11° and almost doubled (19°, (examples in Figure 7B, upper panels) or lamin-B1 and mean value) in MLN8237-treated cells (p<0,01), reaching alpha- or gamma-tubulin (middle and lower panels) a >30° distortion in about 20% of prometaphases. Thus, and categorized: cells with loss or gain of 1 or few the inhibition of Aurora-A under conditions that do not chromosomes (1-2 micronuclei), polyploid (1 large impair spindle formation altogether influences the proper nucleus with 4 pericentrin or gamma-tubulin signals), orientation of the spindle axis and hence of cell division. binucleated, or multinucleated. Indeed, following treatment with 250 nM MLN8237, most interphases appeared to have undergone Induction of aneuploidy in the progeny of chromosome mis-segregation (>65% multinucleated MLN8237-treated mitoses cells); a smaller fraction (about 10%) became polyploid. Only less than 10% of interphases were apparently normal Some of the defects observed in MLN8237-treated (Figure 7C). Abnormalities were also observed in cells mitoses suggest the possibility that genetically imbalanced generated during treatment with 50 nM MLN8237: some daughter cells are generated. 20% were multinucleated or polyploid and about 15% To address this issue we treated pre-synchronized were binucleated. Interestingly, in about one third of the cells with MLN8237 in G2 as described, fixed the cells binucleated cells the nuclei were not equivalent in size, after 24 hours and screened defects in interphase cells suggesting that they represent aberrant products of the presumably representing the progeny of treated mitoses multipolar/asymmetric divisions (see Figure 5). Time- lapse imaging of fluorescently labelled U2OS cells showed that some interphases remained very close after division with a connecting alpha-tubulin bridge (Supplementary Figure S4), possibly representing intermediate figures before binucleation. 20 nM and 50 nM MLN8237 also yielded a significant induction (14-15%) of cells with micronuclei, indicative of mild aneuploidy (Figure 7C). To assess whether micronuclei reflected genuine chromosome loss events, and hence aneuploidy, we assessed whether they contained whole chromosomes by staining with CREST antibodies to human centromeres (Figure 8A). We observed a 6- and 17-fold increase of CREST-positive micronuclei in 20 and 50 nM MLN8237-treated cultures, respectively, compared to controls. Consistent with this, recording of GFP-labelled chromosomes dynamically visualized chromosome bridges and micronuclei formation; interestingly, these defects were always present in mitoses with multipolar spindles (Supplementary Figure S4). These results indicate different extents of ploidy alterations in MLN8237-treated cultures, which we decided to investigate directly by employing FISH (Fluorescence In Situ Hybridization) analysis. We counted hybridization signals produced by chromosome-specific centromeric probes (chromosome 7 and 11) in interphases originated from MLN8237-treated mitoses (protocol in Figure 7). Most (about 85%) control interphases displayed Figure 8: Ploidy alteration in MLN8237-treated cells. 3-4 signals for both chromosome 7 and 11 (Figure 8B), A. IF images of CREST-positive (left panels) or -negative confirming that U2OS cells are hypertriploid. 20 nM (right panels) micronuclei. The insets show enlargements of MLN8237 induced a slight shift in the frequency of cells micronuclei. Histograms represent interphases with CREST- positive or -negative micronuclei in MLN8237-treated cultures with fewer signals (2-fold increase in interphases with 1-2 (fold-increase relative to controls). 400 cells per condition were signals compared to controls), likely corresponding to the counted from 4 experiments. s.d. are shown. Scale bar: 10 µm. chromosome loss events evidenced by the micronuclei B. Distribution of interphases with different numbers of signals analysis. With 50 nM MLN8237, cells displaying 3-4 for chromosomes 7 and 11 (FISH hybridization). 800-1000 cells signals decreased to about 70% and cells with > 4 signals per condition were counted in 2 experiments. www.impactjournals.com/oncotarget 6235 Oncotarget correspondingly increased; a minor fraction of truly duration) or from the generation of abnormal interphase polyploid cells (≥6 signals for both chromosomes within cells. The kinetics of appearance of mitotic cells depicted a single large nuclei) was also identified, consistent with two waves of division in control cultures (Figure 9C). In the IF results in Figure 7. With 250 nM MLN8237 > 50% MLN8237-treated cultures the first mitotic peaks were interphases had ≥ 6 signals, indicative of gain of complete shifted in time and appeared broader. Both effects were sets of chromosomes, consistent with the lack of division dose-dependent, consistent with our data on mitotic entry observed in time-lapse experiments (Figure 5) and with and duration (Figures 2 and 3). At 250 nM inhibitor, no the frequent multinucleated state depicted in Figure 7. second wave of division was observed. Concomitantly, FACS analysis confirmed the increase in ploidy in about we observed a strong increase of multinucleated cells at 45% of the viable population (data not shown) after 48 and 250 nM and a milder effect at 50 nM MLN8237. Some 96 hours of treatment with 250 nM MLN8237. polyploid cells appeared under these conditions (below Together these analyses indicate that MLN8237 3%). Detection of multinucleated and polyploid cells was treatment yields variable levels of aneuploidy in daughter therefore consistent with the results from fixed samples cells, in a dose-dependent manner. (Figures 7 and 8). Importantly, the induction of cell death remained below 3% throughout the recording time (Figure 9C, bottom right panel), and remained at a similar level Long term fate of MLN8237-treated cells in one time-lapse experiment extended to 65 hours (not shown). The time-lapse analysis thus far indicates that Together, the high-throughput data indicate that MLN8237 (5-250 nM range) induces no significant cell MLN8237 induces a dose-dependent lengthening of death in U2OS mitotic cells. We investigated the long- mitotic progression and the generation of abnormal term outcome of the treatment in high-throughput time- daughter cells that are impaired in re-dividing, with a lapse analyses for a length of time (48 hours) roughly cytostatic effect, while low cytotoxicity is observed within corresponding to 2 division cycles. Using the fluorescently the first 48 hours of treatment. labelled U2OS cell line, we analyzed the data with an We also performed proliferation/cell death analyses automated method, the CellCognition software [28], trained to classify cells as interphasic, mitotic, dead, multinucleated or polyploid (examples in Figure 9A). In the 48 hours of the recording time, the cell number increased threefold in control cultures, yet dose-dependent growth inhibition was observed in cultures treated with MLN8237, with almost no increase with 250 nM MLN8237 (Figure 9B, left panel). Consistent with this, the number of normal interphase cells was dramatically reduced by the treatment (Figure 9B, right panel): this effect could result either from an increase in mitotic cells (due to prolonged mitotic Figure 9: Long-term high-throughput analysis of MLN8237-treated cultures. Cultures treated as in Fig. 3 were recorded for 48 hours. Automated segmentation and classification on the images was performed using the CellCognition software. A. Classes defined for training the classifier are shown with representative examples for each class. B. Histograms on the left represent the increase in the number of viable cells from the first to the last frame of the acquisition. Average values and s.d. of 10 replicates from 3 independent experiments are shown. The analyzed sample size at time 0 was at least 500 cells per replicate. Histograms on the right represent the increase in normal interphases only under the same conditions. C. The line charts represent the percentage of cells per class during the recording time. www.impactjournals.com/oncotarget 6236 Oncotarget after several days (Figure 10) by cell counting and FACS genuinely inhibit Aurora-A does not efficiently promote measurement of the DNA content. After 48 hours of cell death in U2OS cells; a toxic effects is only observed treatment, corresponding to the end of the high-throughput after very long exposure times to high doses that video-recording, a slight decrease in the total cell number simultaneously target Aurora-B. was observed with 5-10-20 nM MLN8237 compared to control cultures (Figure 10A); the overall proliferation DISCUSSION trend remained comparable to controls over 96 hours. Cell growth inhibition was instead appreciated with 50 Here we have used the MLN8237 small molecule nM MLN8237 and even more effectively with 250 nM inhibitor to investigate mitotic roles of the Aurora-A MLN8237 (Figure 10A). kinase in osteosarcoma U2OS cells; this molecule is To assess cell death we measured cells with a under clinical trial in several cancer types including sub-G1 DNA content by FACS analysis (Figure 10B). That osteosarcoma. revealed a generally higher level of cell death compared We first determined the extent and specificity of to the microscopy analysis, possibly reflecting technical kinase inhibition by assessing Aurora-A and Aurora-B specificities in the methodology: indeed, detached self-phosphorylation. MLN8237 was effective over dead cells are counted by cytofluorimetry, while being Aurora-A in the 20-50 nM range, but also inhibited preferentially lost in microscopy analysis. Nevertheless, Aurora-B above 50 nM. The loss of selectivity with high by FACS analysis, only 250 nM MLN8237 induced doses was previously reported [25-27]. The inhibition of remarkable cell death (about 30% of sub-G1 cells after 48 Aurora-B identified here with 50 nM MLN8237 (a very hours of treatment, increasing to about 50% after 96 hours, close condition to that required for complete Aurora-A compared to about 10% in control cultures; Figure 10B). inhibition) was instead not noticed before using histone H3 Thus, MLN8237 treatment in concentrations that phosphorylation as a reporter, indicating that the selection of activity reporters is critical to ascertain the selectivity of kinase inhibitors. It also evidences that the window at which MLN8237 fully inhibits Aurora-A, without concomitantly affecting Aurora-B, can be very narrow in some cell lines, indicating that selective inhibition of Aurora-A vs Aurora-B remains a critical issue, even with the best performing ATP-competitors. MLN8237 treatment prolongs the G2 phase duration. The G2 delay was under-appreciated in previous studies using MLN8237 in asynchronous cultures, yet was observed when Aurora-A was inhibited by either antibody microinjection or RNA interference [1, 29], or by conditional ablation in mouse embryonic fibroblasts (MEFs; [30]). No permanent arrest is however induced, suggesting that Aurora-A functions in G2 are important, but can be taken over by other kinases. Indeed, the G2 delay is more severe with 250 nM, under which condition Aurora-B is also inhibited. These findings raise the possibility that centrosome- and/or MT-associated defects induced by Aurora-A inactivation in G2 evoke a checkpoint response that delays the transition towards mitosis onset. Dose-response analyses of MLN8237-treated cells that entered mitosis revealed processes that are differentially sensitive to Aurora-A inactivation, as depicted in the schematics in Figure 11. Complete Aurora-A inhibition impaired MT nucleation; consistent with this, previous data showed that partial vs complete Figure 10: Dose-dependent effects of MLN8237 on cell RNA interference-mediated depletion of Aurora-A growth and viability. A. Cells were counted after 48 and differentially affects maturation of centrosomes, required 96 hours from MLN8237 treatment start. Values represent the for mitotic MT nucleation [1, 2]. Partial Aurora-A increase in the number of cells respect to t=0h (3 independent inhibition (10 nM MLN8237) instead yielded multipolar experiments). B. The percentage of sub-G1 cells in MLN8237- or disorganized spindles, the frequency of which treated cultures, detected by FACS analysis, is shown in the increased in a dose-dependent manner up to 50 nM. histograms (3 experiments). s.d. are shown. www.impactjournals.com/oncotarget 6237 Oncotarget Defective spindle assembly was associated with longer is implicated in spindle orientation in asymmetric prometaphase duration compared to untreated cells. The cell divisions in Drosophila [34] via phosphorylation highest frequency of induction of spindle organization of Pins [35]. Excess Aurora-A can also influence the defects (50 nM) was associated with multipolar divisions. spindle orientation and the cell fate in human mammary 250 nM MLN8237 also prolonged prometaphase epithelium stem/progenitor cells [36]. In our time-lapse duration, after which cells re-adhered and eventually assays using a U2OS cell line with fluorescent MTs, we exited mitosis without division. The failure of actually visualized spindle rotation movements: thus, chromosome segregation could be appreciated in time- Aurora-A activity is required for pathways that determine lapse experiments and was consistent with previous results the mitotic spindle orientation, raising the possibility that obtained from concomitant inactivation of Aurora-A and Aurora-A inhibition influences the fate of asymmetrically B kinases by high MLN8237 doses in Hep3B or HeLa dividing cells and/or tissue architecture. cells or, to a lesser extent, following Aurora-B inhibition Time-lapse experiments also revealed that in HeLa cells [26, 27]. Our own recording experiments of MLN8237 generates chromosome mis-segregation. At interfered mitoses for Aurora-A, or -B, or both, confirm a high concentrations, when both Aurora-A and Aurora-B mild effect of Aurora-B inactivation alone and a synergic are inhibited, massive aneuploidy is observed, with effect of the inactivation of both kinases. Complementary the generation of multinucleated daughter cells. Lower functions of Aurora-A and B in chromosome segregation doses that inhibit Aurora-A alone yield mild aneuploidy, KO also emerged in chicken DT40 Aurora-A cells treated with increased number of CREST-positive micronuclei with an Aurora-B specific inhibitor [31]. None of the (schematics in Figure 11). FISH analysis depicted the approaches used to inactivate Aurora-A alone in human differential generation of mild and massive aneuploidy cultured cells yielded chromosome segregation failure [2, by low and high MLN8237 doses, respectively. This is 27, 32]. The latter was instead described in Aurora-A-null relevant, given the pro- or anti-tumorigenic effects of MEFs [30, 33], suggesting specific modes of action of the aneuploidy depending on the extent of chromosome mis- two kinases in this cellular context, which may reflect a segregation [24]. Consistent with our observations, Yang different stoichiometry between Aurora-A and B and/or and colleagues [37] described the generation of an 8N their substrates. population following simultaneous RNA interference- An interesting finding from this study is the mediated inactivation of Aurora-A and Aurora-B, but not induction of mis-oriented divisions by low MLN8237 of Aurora-A alone, in U2OS cells. Segregation defects concentrations (5-20 nM); the inhibitor is selective for (lagging chromosomes and chromatin bridges) were Aurora-A at these doses and, as recalled above, does observed in Aurora-A-null MEFs, with increased ploidy not impair MT nucleation: this directly implicates the over time [30, 33], again evidencing a crucial contribution lack of Aurora-A in the mis-orientation phenotype and of Aurora-A to chromosome segregation in this system. suggests that MTs are required. Interestingly, Aurora-A We also observed binucleated cells in the progeny of MLN8237-treated mitoses; in time-lapse analyses, telophase cells often remained in closer proximity to one another compared to controls, and tubulin bridges persisted, a phenotype that may possibly evolve into a binucleated cell. The observation of binucleated cells with 20 nM MLN8237 would support a recently proposed direct role of Aurora-A in late steps of cell division [6, 7]. Importantly, MLN8237 failed to induce cell death in mitosis nor did it cause a highly effective elimination of the cellular offspring within the first 48 hours of treatment. Recent time-lapse studies with MLN8237 reported variable results regarding mitotic cell death: no mitotic toxicity was observed in Hep3B cells [26], whereas some death from mitosis was recorded in HeLa cells [27]. Aurora-A inactivation therefore appears to require a set of concomitant conditions, as yet elusive, Figure 11: Dose-dependent effects of MLN8237 on for mitotic cell death activation. The highly aneuploid cell division and aneuploidy induction: a schematic progeny generated in our assays at 50 nM (and, to a higher overview. Schematic representation of the phenotype distribution (Up: mitotic defects. Low: abnormal daughter extent, 250 nM) MLN8237 originates cells impaired in interphases) along a MLN8237 concentration gradient in further cell division, hence unviable in the long term. The U2OS cells. Gradual inhibition of Aurora-A (blue) and mechanisms underlying the cytostatic effects of Aurora-A Aurora-B (orange) is indicated by the gradients within shapes inactivation are controversial: a dose- and time-dependent (dose intervals are not in scale); note that at 20 nM MLN8237 induction of apoptosis was described in different cell Aurora-A inhibition is virtually complete. www.impactjournals.com/oncotarget 6238 Oncotarget lines treated with MLN8237 [18, 19, 38, 39], while in FS and FL-2, in a log scale. Cell aggregates were gated other cases the MLN8237-induced cytostatic effect is out on the bi-parametric graph FL-3lin/Ratio. Apoptosis attributed to senescence [40] consistent with results was determined as the proportion of cells exhibiting a described in Aurora-A-null MEFs [33], or to induction DNA content lower than that of G1 cells after gating out of differentiation pathways [41]. These observations cell debris on the bi-parametric graph FS/SS, using the suggest that both the treatment parameters and the WinMDI software. cellular background contribute to determine the long-term outcome of MLN8237-treated cultures. IF In conclusion, the broad variability in the U2OS cell response to MLN8237 highlighted in this study is an Cells grown on coverslips were fixed as follows: important issue in the light of the use of this compound (a) –20°C methanol, 6 minutes; or, for phospho- in anti-cancer therapy. In the human organism undergoing Aurora-B (Thr232) staining, (b) 3.7% PFA in PBS, 10 treatment, the compound dose cannot be constant and is minutes at room temperature followed by 5 minutes in expected to rise and fall over time. It will be important 0.1% TritonX-100 in PBS. Blocking and all antibody to extend these studies and shed light on the pathways incubations were performed at room temperature in driving the response towards one or another cell fate, with PBS/0.05% Tween 20/3% BSA. Cells were counterstained the perspective to modulate such choice and drive cells with 4,6-diamidino-2-phenylindole (DAPI, 0.1 µg/ml) and towards death pathways. mounted using Vectashield (Vector Laboratories). Primary antibodies were: mouse anti-alpha-tubulin (1:2000, B-5-1- METHODS 2, Sigma-Aldrich), mouse anti-Aurora-A (0.5 µg/ml, BD Transduction Laboratories), rabbit anti-phospho-Aurora-A (Thr288) (1:250, C39D8; Cell Signaling Technology), Cell cultures, synchronization protocols and rabbit anti-pericentrin (2 µg/ml, ab4448; Abcam), rabbit treatments anti-phospho-Aurora-B (Thr232) (1:50, Poly6361, BioLegend), mouse anti-phospho-Histone-H3 (Ser 10) (0.25 µg/ml, clone 3H10, Millipore), mouse anti-γ-tubulin The human U2OS osteosarcoma cell line (ATCC: 0 (1:1000, GTU-88, Sigma-Aldrich), rabbit anti-lamin B1 HTB-96) was grown at 37 C in a 5% CO atmosphere in (1 µg/ml, ab16048; Abcam), human anti-centromere DMEM, supplemented with 10% faetal bovine serum. (CREST; 1:20, Antibodies Incorporated). Samples were U2OS cells stably expressing H2B-GFP and RFP-alpha- analyzed using a Nikon Eclipse 90i microscope equipped tubulin (kind gift of L. Lanzetti, Institute for Cancer with a Qicam Fast 1394 CCD camera (QImaging). Image Research at Candiolo, Italy) were cultured under the same acquisition, deconvolution and Extended Depth of Focus conditions and used in time-lapse imaging experiments. on Z-serial optical sections were performed using Nis- For synchronization, cells were subjected to a 24 hours Elements AR 4.2 (Nikon); images were further processed block in 2 mM thymidine. Cultures were then released with Adobe Photoshop CS 8.0. from the G1/S arrest by washing away the thymidine and adding fresh medium containing 30 µM deoxycytidine; 6 Quantitative analysis of IF signals. hours post-release cultures were treated with MLN8237 (5-250 nM as indicated; Selleck Chemicals) or 0.1 µg/ml nocodazole (Sigma Aldrich). Mock-treated cultures were Signals were measured using Nis Elements AR incubated with DMSO. 4.2 (nd2 file format). Analysis of endogenous Aurora-A and Aurora-B activity in mitotic cells was performed Cell counting and FACS analysis as follows: a) p-Thr288-Aurora-A staining: average pixel intensity at spindle poles, corrected for external background; b) p-Thr232-Aurora-B staining: sum intensity Cells were harvested after 48 and 96 hours from at chromosomes, corrected for external background. treatment. One sample was harvested when cultures were Images for quantification of mitotic signals were released in thymidine–free medium (t=0h) as reference for Maximum Intensity Projections from z-stacks (0.6 µm, the initial number of cells and for verifying the efficacy of ranging over a 5-10 µm). Box-plots were generated using the thymidine arrest. For counting the number of cells a the web-tool BoxPlotR. Z1 Coulter Particle Counter (Beckman Coulter) was used. For measuring the angle between the centrosome- For FACS analysis samples were permeabilized with centrosome axis and the culture surface z-stacks serial 0.1% TritonX-100 in PBS. Cell cycle phase distribution images were used. The “arctan(xy/z)” formula for was analyzed after incubation with propidium iodide (PI, calculating the angle of a right triangle was applied with Sigma P4170, 0.04 mg/ml) using a flow cytofluorimeter “xy” being the distance between centrosomes in xy in the Epics XL apparatus (Beckman Coulter). Parameters SS maximum intensity projection, and “z” being the distance and FL-3 were acquired in a linear amplification scale, www.impactjournals.com/oncotarget 6239 Oncotarget between centrosomes along the z-axis. A schematization conditions, time-lapse acquisition, starting 6 hours after is shown in Figure 6C. Values were statistically analyzed the treatment, was extended for 8 hours only to avoid using the InStat3 software, using either (i) the unpaired phototoxic effects. t test (for Gaussian distributions), applying the Welch For high-throughput experiments, cells were seeded correction when required, or (ii) the Mann-Whitney in 96-well plates. Images were acquired with a ScanR test, when the populations did not follow a Gaussian microscope (Olympus) using a 10x objective, an Olympus distribution. DBH1 camera and the ScanR acquisition software. Temperature and CO were kept constant by an incubation system on the microscope. Acquisition was performed Fluorescent in situ hybridization (FISH) every 30 minutes using Phase contrast and fluorescence imaging, for a total duration of 48-65 hours. Cells grown on coverslips were fixed at room temperature with 3:1 methanol:acetic acid, air dried and Automated analysis of high-throughput data stored at −20°C for at least one day. Before hybridization, coverslips were again incubated in 3:1 methanol:acetic acid, 1 hour at room temperature, then heated 2 hours Data from high-throughput experiments were at 65 C and dehydrated in 70%-90%-100% cold analyzed by the CellCognition software (v 1.3.3-28, ethanol. Denaturation of probes in the hybridization mix [28]). The software was used to segment the cells, extract (Fluorescein-labeled Chromosome 7 Satellite Probe, features and classify them using a support vector machine. cat: PSAT0007-G; rhodamine-labeled Chromosome 11 The classes defined were: interphases, prometaphases, Satellite Probe, cat: PSAT001-R; Hybridization Buffer metaphases, ana-telophases, multinucleated interphases, QB007, all from Q-BIOgene) was performed at 96 C for poliployd cells, dead cells; an additional class of debris or 10 minutes. When the mix was applied on coverslips a of cells with non homogeneous morphology which could co-denaturation step was performed (2 minutes, 72 °C), not be included in any of the other categories was created followed by the hybridization incubation, performed at and was then excluded from subsequent data elaboration to 37 C, overnight. Coverslips were then washed in SSC, at avoid artifacts. The classifier was trained with images from 0 0 37 C and 60 C, and counterstained with DAPI (0.2 µg/ independent experiments with a training set comprising a ml) in SSC, 10 minutes at room temperature. Coverslips minimum of 100 cells per class (with the exception of the were mounted with Vectashield and analyzed with a Nikon polyploid class, of which only 33 examples were found). Eclipse 90i microscope, using a 20x objective (Plan Fluor, A confusion matrix and a classification test were used to 0.5 N.A.). Signals per nucleus were counted using the assess the quality of classification before the analysis. The Object count function of Nis Elements AR 4.2 (nd2 file output data were further analyzed using Microsoft Excel; format), setting the parameters using the Spot detection a single class for Mitoses was generated by pooling the function. prometaphase, metaphase and ana-telophase classes. Values for each well were pooled from 6 imaging fields; average values from 4 replicates within one experiment Time-lapse recording are shown in Figure 9. A similar trend was observed in 2 independent experiments (6 additional replicates). Cells seeded in 35 mm dishes (ibiTreat, cod. 81156, or glass bottom, cod. 81158, both from Ibidi) or 8-well Immunoblotting micro-slides (ibiTreat, cod. 80826, Ibidi) were observed under an Eclipse Ti inverted microscope (Nikon), using 60x [Plan Apo, 1.4 N.A. Differential Interference Contrast Mitotic cells collected by shake off were lysed in (DIC), oil immersion] or 40x (Plan Fluor, 0.60 N.A. DIC) RIPA buffer (50 mM Tris-HCl pH 8.0, 150 mM NaCl, objectives (Nikon); during the whole observation cells 1% NP40, 1 mM EGTA, 0.25% sodium deoxycholate) were kept in a microscope stage incubator (Basic WJ, supplemented with protease and phosphatase inhibitors. Okolab), at 37°C and 5% CO . DIC images were acquired Proteins were resolved by electrophoresis on 10% every 5 minutes over 24 hours using a DS-Qi1Mc camera Laemmli gel and transferred on a nitrocellulose membrane and the NIS-Elements AR 3.22 software (Nikon). Image (Protran BA83, Whatman) using a semi-dry system (BIO- and movie processing were performed with NIS-Elements RAD). 40 µg of extract per lane were loaded. Blocking AR 4.2. and antibody incubations were performed at room Imaging of the U2OS cell line stably expressing temperature in PBS/0.1% Tween 20/5% low fat milk, H2B-GFP and RFP-alpha-tubulin was performed with the or in PBS/0.1% Tween 20/5% BSA (for anti-phospho- 60x objective: images were acquired every 7 minutes in Aurora-A hybridization). Antibodies were: mouse anti- the 3 channels. z-stacks of the fluorescent channels were Aurora-A (1 µg/ml; BD Transduction Laboratories), rabbit acquired every 1 µm over a range of 8 µm, attenuating anti-phospho-Aurora-A (Thr288) (1:1000; C39D8; Cell the fluorescence lamp intensity to 1/32. Under these Signaling Technology), goat anti-actin (0.5 µg/ml, I-19; www.impactjournals.com/oncotarget 6240 Oncotarget SantaCruz Biotechnology). Signals were visualized by chromosomal passenger complex (CPC): from easy rider to enhanced chemiluminescence detection (ECL plus, GE the godfather of mitosis. Nat Rev Mol Cell Biol. 2012; 13: Healthcare, and Protein Detection System, GeneSpin). 789-803. 9. Lapenna S, Giordano A. Cell cycle kinases as therapeutic targets for cancer. Nat Rev Drug Discov. 2009; 8: 547-566. ACKNOWLEDGEMENTS 10. Lens SM, Voest EE, Medema RH. Shared and separate We are grateful to Dr. Rainer Pepperkok and the functions of polo-like kinases and aurora kinases in cancer. EuroBioImaging project for access to the Advanced Light Nat Rev Cancer. 2010; 10: 825-841. Microscopy Facility of the EMBL (Heidelberg) and to the 11. Kollareddy M, Zheleva D, Dzubak P, Brahmkshatriya PS, team at the Facility for advice and technical help. We thank Lepsik M, Hajduch M. Aurora kinase inhibitors: progress Dr. Letizia Lanzetti for providing the U2OS cell line stably towards the clinic. Invest New Drugs. 2012; 30: 2411-2432. expressing fluorescent GFP-H2B and RFP-alpha-tubulin 12. Cervantes A, Elez E, Roda D, Ecsedy J, Macarulla T, and Dr. Francesca Degrassi for reagents and advice with Venkatakrishnan K, Roselló S, Andreu J, Jung J, Sanchis- FISH analysis. We are grateful to Dr. Pietro Cirigliano and Garcia JM, Piera A, Blasco I, Maños L, et al. Phase I to staff of Nikon Instruments S.p.A. (Italy) for continuous pharmacokinetic/pharmacodynamic study of MLN8237, collaborative support for imaging experiments performed an investigational, oral, selective aurora a kinase inhibitor, at the CNR-IBMN Nikon Reference Center. This work in patients with advanced solid tumors. Clin Cancer Res. was supported by grants from AIRC-Italian Association 2012; 18: 4764-4774. for Cancer Research (MFAG-13350 to G.G. and IG14534 13. Dees EC, Cohen RB, von Mehren M, Stinchcombe TE, to P.L.) and Fondazione Monte dei Paschi di Siena (to Liu H, Venkatakrishnan K, Manfredi M, Fingert H, Burris G.G. and P.L.). Visits of IAA to the ALMF were supported HA 3rd, Infante JR. Phase I study of aurora A kinase by an EMBO Short-term fellowship and a SIBBM travel inhibitor MLN8237 in advanced solid tumors: safety, grant. pharmacokinetics, pharmacodynamics, and bioavailability of two oral formulations. Clin Cancer Res. 2012; 18: 4775- REFERENCES 14. Matulonis UA, Sharma S, Ghamande S, Gordon MS, Del 1. Hirota T, Kunitoku N, Sasayama T, Marumoto T, Zhang Prete SA, Ray-Coquard I, Kutarska E, Liu H, Fingert D, Nitta M Hatakeyama K, Saya H. Aurora-A and an H, Zhou X, Danaee H, Schilder RJ. Phase II study of interacting activator, the LIM protein Ajuba, are required MLN8237 (alisertib), an investigational Aurora A kinase for mitotic commitment in human cells. Cell. 2003; 114: inhibitor, in patients with platinum-resistant or -refractory 585-598. epithelial ovarian, fallopian tube, or primary peritoneal 2. Marumoto T, Honda S, Hara T, Nitta M, Hirota T, Kohmura carcinoma. Gynecol Oncol. 2012; 127: 63-69. E, Saya H. Aurora-A kinase maintains the fidelity of early 15. Doménech E, Malumbres M. Mitosis-targeting therapies: and late mitotic events in HeLa cells. J Biol Chem. 2003; a troubleshooting guide. Curr Opin Pharmacol. 2013; 13: 278: 51786-51795. 519-528. 3. Katayama H, Sasai K, Kloc M, Brinkley BR, Sen S. Aurora 16. Friedberg JW, Mahadevan D, Cebula E, Persky D, Lossos kinase-A regulates kinetochore/chromatin associated I, Agarwal AB, Jung J, Burack R, Zhou X, Leonard EJ, microtubule assembly in human cells. Cell Cycle. 2008; 7: Fingert H, Danaee H, Bernstein SH. Phase II study of 2691-2704. alisertib, a selective Aurora A kinase inhibitor, in relapsed 4. De Luca M, Lavia P, Guarguaglini G. A functional interplay and refractory aggressive B- and T-cell non-Hodgkin between Aurora-A, Plk1 and TPX2 at spindle poles: Plk1 lymphomas. J Clin Oncol. 2014; 32: 44-50. controls centrosomal localization of Aurora-A and TPX2 17. Manfredi MG, Ecsedy JA, Chakravarty A, Silverman spindle association. Cell Cycle. 2006; 5: 296-303. L, Zhang M, Hoar KM, Stroud SG, Chen W, Shinde V, 5. De Luca M, Brunetto L, Asteriti IA, Giubettini M, Lavia P, Huck JJ, Wysong DR, Janowick DA, Hyer ML et al. Guarguaglini G. Aurora-A and ch-TOG act in a common Characterization of Alisertib (MLN8237), an investigational pathway in control of spindle pole integrity. Oncogene. small-molecule inhibitor of aurora A kinase using novel in 2008; 27: 6539-6549. vivo pharmacodynamic assays. Clin Cancer Res. 2011; 17: 6. Lioutas A, Vernos I. Aurora A kinase and its substrate 7614-7624. TACC3 are required for central spindle assembly. EMBO 18. Kelly KR, Nawrocki ST, Espitia CM, Zhang M, Yang JJ, Rep. 2013; 14: 829-836. Padmanabhan S, Ecsedy J, Giles FJ, Carew JS. Targeting 7. Reboutier D, Troadec MB, Cremet JY, Chauvin L, Guen V, Aurora A kinase activity with the investigational agent Salaun P, Prigent C. Aurora A is involved in central spindle alisertib increases the efficacy of cytarabine through a assembly through phosphorylation of Ser 19 in P150Glued. FOXO-dependent mechanism. Int J Cancer. 2012; 131: J Cell Biol. 2013; 201: 65-79. 2693-2703. 8. Carmena M, Wheelock M, Funabiki H, Earnshaw WC. The www.impactjournals.com/oncotarget 6241 Oncotarget 19. Qi W, Spier C, Liu X, Agarwal A, Cooke LS, Persky DO, 31. Hégarat N, Smith E, Nayak G, Takeda S, Eyers PA, Chen D, Miller TP, Mahadevan D. Alisertib (MLN8237) an Hochegger H. Aurora A and Aurora B jointly coordinate investigational agent suppresses Aurora A and B activity, chromosome segregation and anaphase microtubule inhibits proliferation, promotes endo-reduplication and dynamics. J Cell Biol. 2011; 195: 1103-1113. induces apoptosis in T-NHL cell lines supporting its 32. Du J, Hannon GJ. Suppression of p160ROCK bypasses cell importance in PTCL treatment. Leuk Res. 2013; 37: 434- cycle arrest after Aurora-A/STK15 depletion. Proc Natl Acad Sci U S A. 2004; 101: 8975-8980. 20. Maris JM, Morton CL, Gorlick R, Kolb EA, Lock R, Carol 33. Pérez de Castro I, Aguirre-Portolés C, Fernández-Miranda H, Keir ST, Reynolds CP, Kang MH, Wu J, Smith MA, G, Cañamero M, Cowley DO, Van Dyke T, Malumbres Houghton PJ. Initial testing of the aurora kinase A inhibitor M. Requirements for Aurora-A in tissue regeneration and MLN8237 by the Pediatric Preclinical Testing Program tumor development in adult mammals. Cancer Res. 2013; (PPTP). Pediatr Blood Cancer. 2010; 55: 26-34. 73: 6804-6815. 21. Zhou N, Singh K, Mir MC, Parker Y, Lindner DJ, Dreicer 34. Lee CY, Andersen RO, Cabernard C, Manning L, Tran KD, R, Ecsedy JA, The BT, Zhang Z, Almasan A, Hansel DE. Lanskey MJ, Bashirullah A, Doe CQ. Drosophila Aurora-A The investigational Aurora kinase A inhibitor MLN8237 kinase inhibits neuroblast self-renewal by regulating aPKC/ induces defects in cell viability and cell-cycle progression Numb cortical polarity and spindle orientation. Genes Dev. in malignant bladder cancer cells in vitro and in vivo. Clin 2006; 20: 3464-3474. Cancer Res. 2013; 19: 1717-1728. 35. Johnston CA, Hirono K, Prehoda KE, Doe CQ. 22. Gascoigne KE, Taylor SS. Cancer cells display profound Identification of an Aurora-A/PinsLINKER/Dlg spindle intra- and interline variation following prolonged exposure orientation pathway using induced cell polarity in S2 cells. to antimitotic drugs. Cancer Cell. 2008; 14: 111-122. Cell. 2009; 138: 1150-1163. 23. Asteriti IA, Giubettini M, Lavia P, Guarguaglini G. 36. Regan JL, Sourisseau T, Soady K, Kendrick H, McCarthy Aurora-A inactivation causes mitotic spindle pole A, Tang C, Brennan K, Linardopoulos S, White DE, fragmentation by unbalancing microtubule-generated Smalley MJ. Aurora A kinase regulates mammary epithelial forces. Mol Cancer. 2011; 10: 131. cell fate by determining mitotic spindle orientation in a 24. Weaver BA, Cleveland DW. The aneuploidy paradox in cell Notch-dependent manner. Cell Rep. 2013; 4: 110-123. growth and tumorigenesis. Cancer Cell. 2008; 14: 431-433. 37. Yang H, Burke T, Dempsey J, Diaz B, Collins E, Toth 25. Qi W, Cooke LS, Liu X, Rimsza L, Roe DJ, Manziolli A, J, Beckmann R, Ye X. Mitotic requirement for aurora A Persky DO, Miller TP, Mahadevan D. Aurora inhibitor kinase is bypassed in the absence of aurora B kinase. FEBS MLN8237 in combination with docetaxel enhances Lett. 2005; 579: 3385-3391. apoptosis and anti-tumor activity in mantle cell lymphoma. 38. Kelly KR, Ecsedy J, Medina E, Mahalingam D, Biochem Pharmacol. 2011; 81: 881-890. Padmanabhan S, Nawrocki ST, Giles FJ, Carew JS. The 26. Marxer M, Foucar CE, Man WY, Chen Y, Ma HT, Poon novel Aurora A kinase inhibitor MLN8237 is active in RY. Tetraploidization increases sensitivity to Aurora B resistant chronic myeloid leukaemia and significantly kinase inhibition. Cell Cycle. 2012; 11: 2567-2577. increases the efficacy of nilotinib. J Cell Mol Med. 2011; 15: 2057-2070. 27. Marxer M, Ma HT, Man WY, Poon RY. p53 deficiency enhances mitotic arrest and slippage induced by 39. Sehdev V, Peng D, Soutto M, Washington MK, Revetta F, pharmacological inhibition of Aurora kinases. Oncogene. Ecsedy J, Zaika A, Rau TT, Schneider-Stock R, Belkhiri 2014; 33: 3550-3560. A, El-Rifai W. The aurora kinase A inhibitor MLN8237 enhances cisplatin-induced cell death in esophageal 28. Held M, Schmitz MH, Fischer B, Walter T, Neumann adenocarcinoma cells. Mol Cancer Ther. 2012; 11: 763-774. B, Olma MH, Peter M, Ellenberg J, Gerlich DW. CellCognition: time-resolved phenotype annotation in high- 40. Liu Y, Hawkins OE, Su Y, Vilgelm AE, Sobolik T, Thu throughput live cell imaging. Nat Methods. 2010; 7: 747- YM, Kantrow S, Splittgerber RC, Short S, Amiri KI, Ecsedy 754. JA, Sosman JA, Kelley MC, Richmond A. Targeting aurora kinases limits tumour growth through DNA damage- 29. Marumoto T, Hirota T, Morisaki T, Kunitoku N, Zhang D, mediated senescence and blockade of NF-κB impairs this Ichikawa Y, Sasayama T, Kuninaka S, Mimori T, Tamaki drug-induced senescence. EMBO Mol Med. 2013; 5: 149- N, Kimura M, Okano Y, Saya H. Roles of aurora-A kinase in mitotic entry and G2 checkpoint in mammalian cells. Genes Cells. 2002; 7: 1173-1182. 41. Mohan P, Castellsague J, Jiang J, Allen K, Chen H, Nemirovsky O, Spyra M, Hu K, Kluwe L, Pujana MA, 30. Cowley DO, Rivera-Pérez JA, Schliekelman M, He YJ, Villanueva A, Mautner VF, Keats JJ, et al. Genomic Oliver TG, Lu L, O’Quinn R, Salmon ED, Magnuson imbalance of HMMR/RHAMM regulates the sensitivity and T, Van Dyke T. Aurora-A kinase is essential for bipolar response of malignant peripheral nerve sheath tumour cells spindle formation and early development. Mol Cell Biol. to aurora kinase inhibition. Oncotarget. 2013; 4: 80-93. 2009; 29: 1059-1071. www.impactjournals.com/oncotarget 6242 Oncotarget

Journal

OncotargetUnpaywall

Published: Jul 9, 2014

There are no references for this article.