Targeting Aurora Kinase with MK-0457 Inhibits Ovarian Cancer Growth

The Aurora family of serine/threonine kinases is essential for many cellular functions including high-fidelity progression through mitosis (1–4). Aurora-A is located on chromosome 20q13.2-q13.3 and is required for centrosome separation and maturation as well as proper mitotic spindle formation and function (1, 4, 5). Although the Aurora-A kinase activity is cell cycle dependent and highest during G₂-M, phosphorylation of human Aurora-A kinase, particularly at the Thr²⁸⁸ residue, appears to be requisite for enhanced kinase activity (3). Following the G₂-M phase of the cell cycle, degradation of Aurora-A kinase is mediated by the ubiquitin-proteasome pathway (5). The ability of Aurora-A kinase to induce multipolar spindles by overriding the mitotic spindle checkpoints and transforming fibroblasts into aneuploid cells supports its role as a potential oncogene (1, 5). Aurora-B kinase, located on chromosome 17p13.1, is a “chromosomal passenger” protein that plays an important role in regulating mitosis, particularly cytokinesis. Aurora-C kinase is not as well understood but appears to have functions during mitosis that overlap with Aurora-B kinase (3, 5).

We reported recently that Aurora-A is overexpressed in 83% of human epithelial ovarian carcinomas and predicts poor clinical outcome (6). Furthermore, the chromosome 20 amplicon corresponding to the Aurora-A gene location has been reported in not only ovarian cancer cell lines but also in 54% to 100% of sporadic and hereditary human ovarian carcinomas (7, 8). Although Aurora-A kinase overexpression has also been correlated with centrosome amplification (6, 8), distinct polymorphisms within the Aurora-A kinase gene locus are also associated with ∼20% increased risk of invasive ovarian cancer (9), further implicating Aurora-A kinase in tumor development. Through mechanisms including Akt activation and checkpoint dysregulation, Aurora-A kinase has also been implicated in protecting cells from apoptosis induced by traditional chemotherapy agents, including mainstay cytotoxic agents against ovarian cancer such as cisplatin and paclitaxel (10, 11). Moreover, Sun et al. have recently shown that inhibition of Aurora kinase can sensitize SKOV3 cells to traditional chemotherapeutic agents via NF-κB down-regulation, further supporting the therapeutic role of Aurora kinase targeting in oncology (12). Recent reports have emerged highlighting the role of Aurora-B in maintaining the spindle assembly checkpoint and supporting it as a valid and individual therapeutic target (13–15).

Given the high prevalence of Aurora kinase overexpression in ovarian cancer and its diverse protumorigenic roles, inhibiting the Aurora kinase family appears to be an attractive therapeutic goal, particularly as ovarian cancer remains the leading cause of death from gynecologic cancer (16). Based on the role Aurora kinase plays during the cell cycle, we examined the effects of pan-Aurora kinase inhibition using a highly selective small-molecule inhibitor, MK-0457 (formerly known as VX-680), on ovarian cancer growth in preclinical orthotopic models of metastatic ovarian carcinoma using both chemotherapy-sensitive and resistant cell lines.

Cell lines. To study the effects of Aurora kinase inhibition in ovarian carcinoma, we used two highly metastatic chemosensitive human ovarian cancer cell lines, HeyA8 and SKOV3ip1. Because most patients with recurrent ovarian cancer develop chemotherapy-resistant disease, we also used the taxane- and platinum-resistant human ovarian cancer cell lines, HeyA8-MDR and A2780-CP20, respectively. The derivation and source of these cell lines have been reported elsewhere (17–19). HeyA8, SKOV3ip1, and A2780-CP20 cells were maintained in monolayer cultures at 37°C in RPMI 1640 supplemented with 15% fetal bovine serum and 0.1% gentamicin sulfate (Gemini Bioproducts). The HeyA8-MDR cell line, a generous gift from Dr. Isaiah J. Fidler (The University of Texas M. D. Anderson Cancer Center), was generated by sequential exposure to increasing sublethal doses of paclitaxel and grown in the same medium as the parental cells supplemented with 300 μg/mL paclitaxel (Bristol-Myers Squibb). All cell lines were routinely screened for Mycoplasma using MycoAlert (Cambrex Bio Science) as described by the manufacturer. In vitro and in vivo experiments were conducted with cell lines at 70% to 80% confluence.

MK-0457. Inhibition of Aurora kinase was achieved using the small-molecule pan-Aurora kinase inhibitor, MK-0457, obtained from Merck & Co. The kinase specificity for this compound has been reported previously with reported activity against wild-type and mutated bcr-abl, including the T313I mutation, as well as JAK2 and FLT3 (1). In vitro experiments were conducted using dilutions from a 2 mmol/L stock of MK-0457 dissolved in DMSO. In vivo studies were conducted using MK-0457 dissolved in 1:1 PEG300/PBS for i.p. administration.

Functional assays of Aurora kinase inhibition. Because Aurora-A kinase autophosphorylation on Thr²⁸⁸ as well as phosphorylation of histone H3 on Ser¹⁰ and the centromeric histone variant, Cenp-A on Ser⁷, are important indicators of Aurora kinase activity during mitosis, we assayed for these known mitosis-specific functions. Twenty-thousand HeyA8 and SKOV3ip1 ovarian carcinoma cells were plated in 6 cm plates and allowed to adhere overnight. All plates were then treated with the G₂-M blocker, demecolcine solution (Sigma-Aldrich), for 8 h. Mitotic cells were collected by mitotic shake-off, washed in fresh medium, and then released into new medium containing MK-0457 (10, 100, 500, or 1,000 nmol/L) for 1 h at 37°C and 5% CO₂. One plate maintained in demecolcine served as a control. Cells were collected and washed in PBS. Cell pellets were lysed directly in NP-40 sample disruption buffer and separated on a 4% to 12% gradient polyacrylamide gel electrophoresis (Invitrogen). Proteins were transferred onto Immobilon-P membranes (Millipore) for 1 h using standard Western blot methods. Immobilized proteins were blocked in 5% milk/PBS with 0.1% Tween 20 and incubated overnight at 4°C with antibodies against phospho-Aurora-A (Thr²⁸⁸; 1:1,000; Abcam), phospho-histone H3 (Ser¹⁰; 1:1,000; Upstate), and phospho-Cenp-A (Ser⁷; 1:1,000; Upstate). After washing and incubating with the corresponding secondary antibodies, blots were developed using enhanced chemiluminescence (Amersham) reagents.

Western blot. Whole-cell lysates were used for Western blot analysis to characterize the in vitro kinetics of Aurora kinase inhibition by MK-0457. One million HeyA8 cells were plated onto 10 cm plates and allowed to adhere overnight. Cells were then treated with MK-0457 (10 nmol/L) for 5, 10, and 30 min and 1, 2, 4, 6, and 12 h. Cell lysates were prepared by incubating plates on ice for 20 min with 1× modified radioimmunoprecipitation assay lysis buffer with 1× protease inhibitor (Roche) supplemented with sodium orthovanadate. After centrifuging at 13,000 rpm for 20 min at 4°C, the supernatant was collected and stored at −80°C until ready for use.

Western blotting for phospho-Aurora-A (Thr²⁸⁸) and total Aurora-A was done using 20 μg total protein as determined by BCA Protein Assay Kit (Pierce Biotechnology). After separation by 12% SDS-PAGE with wet transfer onto a nitrocellulose membrane, probing was done using an anti-phospho-Aurora-A (Thr²⁸⁸) antibody (1:1,000) and anti-total Aurora-A antibody (1:1,000, Bethyl Laboratories). Visualization was achieved using a horseradish peroxidase–conjugated anti-rabbit (1:2,000; Amersham) antibody and enhanced chemiluminescence. Equal loading was verified using β-actin.

Cytotoxicity assay. The cytotoxic effects of Aurora kinase inhibition on tumor cells were determined using the 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT; Sigma-Aldrich) uptake method as described previously (20). Briefly, 1,000 HeyA8 or 2,000 SKOV3ip1 cells in RPMI 1640 + 15% fetal bovine serum were seeded into each well of a 96-well plate and allowed to adhere overnight. Treatment conditions were conducted in replicates of 5. Cells were then treated once with increasing concentrations of MK-0457 at 37°C for 96 h before 50 μL/well of 0.15% MTT solution were added. After incubation for 2 h at 37°C, the medium/MTT solution was replaced with 100 μL/well DMSO (Sigma-Aldrich), and the absorbance was measured at 570 nm using a 96-well multiscanner (Dynex Technologies, MRX Revelation). The IC₅₀ was determined by calculating the mean absorbance at 570 nm and then identifying the corresponding MK-0457 concentration on the dose-response curve using regression analysis.

To characterize effects of combining MK-0457 with docetaxel on tumor cells, MTT assays were done. One thousand HeyA8 or 3,000 SKOV3ip1 cells per well were seeded into a 96-well plate and allowed to adhere overnight. Cells were then treated with either 1 or 0 nmol/L (fresh medium) of MK-0457 for 24 h. Sequential doses of docetaxel mixed with medium and MK-0457 (0 or 1 nmol/L) were then administered to the cells for 72 h. MTT assay was then done as above, and IC₅₀ levels were determined based on A₅₇₀ readings.

Cell cycle and apoptosis analysis by flow cytometry. Because of the role of Aurora kinase in cell cycle integrity, the ability of MK-0457 to modulate the cell cycle and affect apoptosis in HeyA8 and SKOV3ip1 in vitro was evaluated using flow cytometry. Experimental conditions were done in replicates of 5. For each cell line, 1 × 10⁶ cells were seeded into 10 cm dishes and allowed to adhere overnight. Cell cultures were washed with PBS and then treated with RPMI 1640 (negative control) or medium containing MK-0457 or MK-0457 plus docetaxel. Then, 12, 24, and 48 h after treatment, cells were collected by trypsinization and pooled with floating cells, which consisted of detached mitotic, apoptotic, and/or dead cells. After trypsin neutralization with fetal bovine serum–containing medium, cell suspensions were centrifuged for 5 min at 1,500 rpm at room temperature and then washed with PBS twice before being fixed in 70% ethanol. Cells were stored at −20°C for at least 18 h after fixation. Immediately before analysis, fixed samples were washed with PBS and then resuspended in propidium iodide (50 μg/mL) and RNase A (20 μg/mL; Phoenix Flow System) for at least 30 min at room temperature protected from light. Stained cells were analyzed on an EPICS XL flow cytometer (Beckman-Coulter) within 2 h of staining. The low-level gate was set at the base of the G₁ peak and the percentages of cells within the G₁, S, and G₂-M phases of the cell cycle were determined by analysis with Multicycle (Phoenix Flow System).

Immunohistochemistry. Phospho-histone H3 (Ser¹⁰) and proliferating cell nuclear antigen (PCNA) immunohistochemistry was done on 5-μm-thick, formalin-fixed, paraffin-embedded slides. Deparaffinization was achieved with xylene followed by descending grades of ethanol. Antigen retrieval was done by microwave-heated citrate buffer (pH 6.0) for 20 min. Endogenous peroxidases were blocked with 3% H₂O₂/methanol for 12 min at room temperature. Nonspecific epitopes were blocked with 10% normal goat serum (for phospho-histone H3) or 5% normal horse serum/1% normal goat serum (for PCNA) for 30 min at room temperature. Slides were then incubated with anti-phospho-histone H3 (Ser¹⁰) antibody (1:300) or PCNA clone PC10 (1:50; DAKOCytomation) at 4°C overnight. Slides were then developed with either biotinylated goat anti-rabbit (BioCare Medical) for phospho-histone H3 detection followed by streptavidin-horseradish peroxidase (1:300; DAKOCytomation) or rat anti-mouse IgG2a-horseradish peroxidase (1:200; Serotec) for PCNA. Visualization was achieved with 3,3′-diaminobenzidine (Open Biosystems), and counterstaining was done with Gill's hematoxylin (Sigma-Aldrich). Phospho-histone H3 status was determined as the number of phospho-histone H3-positive cells averaged over five “hotspot” high-power fields at ×100 per specimen. Proliferative index was calculated as the proportion of PCNA-positive cells over five high-power fields at ×200 per specimen.

To quantify apoptosis, terminal deoxynucleotidyl transferase–mediated dUTP nick end labeling assay was done on 5-μm thick, paraffin-embedded tumor slides as described previously (20, 21). Briefly, after deparaffinization, all slides were treated with proteinase K (1:500). One DNase-treated specimen served as a positive control. Then, 3% H₂O₂/methanol was applied to all specimens to block endogenous peroxidases. After a terminal deoxynucleotidyl transferase buffer (30 mmol/L Trizma, 140 mmol/L sodium cacodylate, 1 mmol/L CoCl₂) rinse, all slides were incubated with terminal transferase (1:400; Roche Diagnostics) and biotin-16-dUTP (1:200; Roche) and then blocked with 2% bovine serum albumin. Samples were then incubated in streptavidin-horseradish peroxidase (1:400; DAKO) for 40 min at 37°C and rinsed with PBS. Visualization was done with 3,3′-diaminobenzidine and counterstained with Gill's hematoxylin. The apoptotic index was quantified as the number of apoptotic tumor cells in five randomly selected ×100 high-power fields exclusive of necrotic areas.

Animals. For all in vivo studies, female athymic mice (NCr-nu/nu) were purchased from the National Cancer Institute-Frederick Cancer Research and Development Center. Mice were housed and maintained under specific pathogen-free conditions in accordance with guidelines from the American Association for Accreditation of Laboratory Animal Care and the NIH. All studies were approved and overseen by The University of Texas M. D. Anderson Cancer Center Institutional Animal Care and Use Committee.

Orthotopic inoculation of tumor cells and necropsy. At ∼75% confluence, HeyA8, SKOV3ip1, HeyA8-MDR, and A2780-CP20 cells were collected from cultures using either 0.25% trypsin-EDTA (Life Technologies) or 0.1% EDTA depending on the cell line. Cells lifted with trypsin underwent trypsin-neutralization with fetal bovine serum–containing medium before being centrifuged and then resuspended in the appropriate volume of serum-free HBSS (Invitrogen) for animal inoculation. Cell lines not requiring trypsin-neutralization were directly centrifuged at 1,000 rpm for 7 min at 4°C, washed with PBS, and then resuspended in serum-free HBSS at the appropriate concentrations for inoculation. HeyA8 cells were injected i.p. at 2.5 × 10⁵ per 200 μL HBSS. SKOV3ip1, HeyA8-MDR, and A2780-CP20 cells were injected i.p. at 1 × 10⁶ per 200 μL HBSS.

Long-term therapy experiments were done using all four cell lines. Mice were sacrificed when the control group appeared near moribund, ∼3 to 5 weeks after commencing therapy, depending on the cell line. Tumors were harvested from the peritoneal cavities of mice, tumor nodules were quantified, and total tumor weight was determined. Malignant ascites was aspirated and the volume was measured. Additional tumor tissue for H&E staining and immunohistochemistry was formalin fixed at the time of tumor collection and then paraffin embedded. Paraffin sections were uniformly cut at 5 μm thickness.

Therapy experiments using MK-0457 in orthotopic murine models. Dose-finding experiments were done by injecting HeyA8 tumor cells i.p. (2.5 × 10⁵) into athymic female mice. Nineteen days after tumor cell injection when i.p. tumors were palpable, the mice were randomized into three dosage groups: 0 mg (vehicle alone), 25 mg/kg, and 50 mg/kg. Twice daily doses of inhibitor or vehicle were administered by i.p. injections for 2 days (total four doses, 12 h apart). Mice were sacrificed at 24, 48, and 72 h after the final i.p. injection. Immunohistochemistry for phospho-histone H3 was done on the tumors as described earlier.

To determine the antitumor effects of Aurora kinase inhibition, we initiated treatment with MK-0457 and/or cytotoxic chemotherapy injections 1 week after tumor cell inoculation using a minimal residual disease model (20, 22–25). Docetaxel (Sanofi-Aventis; 35 μg/mouse for SKOV3ip1 tumor-bearing mice or 50 μg/mouse for HeyA8 and HeyA8-MDR), cisplatin (160 μg/mouse for A2780-CP20; Bristol-Myers Squibb), or vehicle was injected i.p. once weekly. Docetaxel was the chosen taxane given its favorable side-effect profile over paclitaxel in human studies (26–28). MK-0457 was administered twice daily for 2 days, starting 1 day before treatment with docetaxel or cisplatin. Mice were monitored daily for adverse effects and drug tolerance. All animals were sacrificed and tumors were harvested at necropsy when the control mice began to appear moribund, ∼3 to 4 weeks after the initiation of therapy, depending on the cell line used (19). Mouse weight, tumor weight, tumor distribution, and ascites volume were recorded.

To explore the therapeutic effect of the timing at which Aurora kinase inhibition occurred relative to cytotoxic chemotherapy treatment, we employed the in vivo HeyA8 tumor model and initiated MK-0457 treatment either 2 days before, 1 day before and with, concurrently and 1 day after, and 1 and 2 days after weekly docetaxel. Treatment continued until the vehicle-treated animals showed significant tumor burden and/or were moribund at which point all animals were sacrificed simultaneously. All tumor nodules were collected, counted, and weighed at necropsy.

To compare the biological activity of i.v. versus i.p. aurora kinase inhibition, we used the in vivo HeyA8 tumor model and initiated twice weekly either vehicle alone, i.v. MK-0457 therapy, or i.p. MK-0457. Dosages between the two treatment groups were matched and animals were followed until animals in any group became moribund at which time all animals were sacrificed and tumors were harvested, weighed, and recorded.

Microarray analysis of tumors following MK-0457 treatment. Five vehicle-treated control mice and four MK-0457-treated experimental mice bearing orthotopic HeyA8 tumors were sacrificed 24 h after i.p. treatment (50 mg/kg). Tumors were immediately removed and preserved in RNAlater solution (Ambion) for subsequent RNA extraction with RNeasy kit (Qiagen). The quality and purity were assessed by agarose gel electrophoresis and absorbance measurement at A₂₆₀/A₂₈₀. Commercially available high-density oligonucleotide microarrays (Human Genome U133 Plus 2.0; Affymetrix) were used for expression analysis. Preparation of cRNA, hybridization, scanning, and image analysis of the arrays were done according to the manufacturer's protocols (Affymetrix) as described previously (29). Microarray data were processed with dChip software (30) and differentially expressed genes were identified using SAM analysis (31).

Real-time PCR. cDNA was synthesized from total RNA using the High-Capacity cDNA Reverse Transcription kit (Applied Biosystems). Quantitative real-time PCR was done in a MX4000 multiplex quantitative PCR system (Stratagene) using predesigned TaqMan primers and probe sets (Applied Biosystems) and the Brilliant QPCR kit (Stratagene). The conditions for the reaction were as follows: 1 cycle at 95°C for 10 min and 40 to 50 cycles at 95°C for 15 s and 60°C for 1 min. Quantitative real-time PCR for each primer and probe set was done either in duplicate or triplicate, and the means are reported. Samples were normalized to cyclophilin A and the relative expression level was calculated by the comparative C_T method using an untreated sample as the calibrator.

Statistical analysis.In vivo therapy experiments were powered to detect a 50% reduction in tumor weight (β error = 0.2). The Mann-Whitney rank-sum test was used to analyze nonparametric and nonnormally distributed data sets. Comparison of dose-response curves in the MTT analysis for single-agent and combination treatments was made by logistic regression analysis (Prism; GraphPad Software). Regression and statistical analyses were done using SPSS 12.0 for Windows (SPSS). A two-tailed P ≤ 0.05 was deemed statistically significant.

In vitro inhibition of Aurora kinase activity. Before initiating therapy experiments, we first did functional kinase assays to interrogate the phosphorylation status of Aurora-A (Thr²⁸⁸), required for kinase activity, as well as additional downstream targets, phospho-histone H3 (Ser¹⁰) and phospho-Cenp-A (Ser⁷) following treatment with MK-0457. In both HeyA8 and SKOV3ip1 mitotic cells, inhibition of autophosphorylation of Aurora-A kinase at Thr²⁸⁸ was observed within 1 h of treatment with the inhibitor (10 nmol/L). Treatment with MK-0457 also resulted in a gradual decline in phosphorylation status of histone H3 (Ser¹⁰) and a dramatic down-regulation of Cenp-A (Ser⁷) phosphorylation (Fig. 1A).

To characterize the onset of Aurora kinase inhibition by MK-0457, we examined Aurora-A phosphorylation in plated HeyA8 cells by Western blot. Levels of phospho-Aurora-A relative to total-Aurora-A began to decrease 10 min after exposure to the inhibitor. Inhibition of phospho-Aurora-A persisted through 6 h after exposure. At 12 h after exposure to the inhibitor, relative levels of phospho-Aurora-A began to increase toward baseline (Fig. 1B).

In vitro cytotoxic effects of inhibiting Aurora kinase on ovarian carcinoma. Before in vivo experiments, we examined the in vitro effects of targeting Aurora kinase on ovarian carcinoma cytotoxicity. The IC₅₀ level of the Aurora kinase inhibitor was ∼50 nmol/L for the HeyA8 cell line and 20 to 70 nmol/L for the SKOV3ip1 cell line (Fig. 2A). Treating HeyA8 cells with the traditional cytotoxic agent, docetaxel, resulted in an IC₅₀ of ∼10 nmol/L; however, treating HeyA8 cells with varying concentrations of docetaxel after 24 h of pretreatment with MK-0457 resulted in at least a 10-fold improvement in the IC₅₀ of docetaxel to 0.5 to 1 nmol/L (Fig. 2A). This enhanced effect was similar in the SKOV3ip1 cell line in which treating SKOV3ip1 cells with increasing concentrations of docetaxel after 24 h of MK-0457 pretreatment also resulted in a significant improvement compared with treatment with docetaxel alone (IC₅₀ = 0.2-0.5 nmol/L with pretreatment versus 5-10 nmol/L without pretreatment; P = 0.001; Fig. 2A).

Based on the role of Aurora kinases in cell cycle progression, we did flow cytometry to determine the effect on cell cycle after inhibition with the pan-Aurora kinase inhibitor. In the HeyA8 and SKOV3ip1 cell lines, a one-time exposure of cells to MK-0457 alone resulted in >3-fold increase in G₂-M arrest within 12 h of treatment (P = 0.02). A trend of persistent G₂-M arrest was evidenced in the HeyA8 cell line through 48 h after exposure to the inhibitor (P = 0.08). However, in the SKOV3ip1 cell line, this 3-fold increase in G₂-M arrest was present through 48 h after exposure to the Aurora kinase inhibitor (P = 0.02; Fig. 2B). Endoreduplication, a phenotype of Aurora-B inhibition, has been identified as a hallmark of aberrant cytokinesis (32); therefore, we did flow cytometry to examine cell ploidy. Twenty-four hours after treatment with the inhibitor, 71% of the HeyA8 cells showed aneuploidy or ≥4N (Fig. 2C). Because an important consequence of G₂-M arrest is apoptosis, we used flow cytometry to determine the apoptotic fraction of cells treated with the Aurora kinase inhibitor as represented by the sub-G₁ cell population. Within 48 h after Aurora kinase inhibition, a 30-fold increase in apoptotic HeyA8 cells was seen (P = 0.02) compared with controls (Fig. 2D). In the SKOV3ip1 cell line, treatment with the inhibitor elicited a 3.5-7-fold increase in apoptosis by 48 h after exposure (P = 0.02) compared with controls (Fig. 2D). Based on the induction of G₂-M arrest by MK-0457, we next asked whether docetaxel-induced apoptosis would be further enhanced by this inhibitor. Combining MK-0457 with docetaxel in the SKOV3ip1 cell line resulted in a rapid and sustained 25- to 40-fold increase in apoptosis beginning 12 h after treatment and lasting through 48 h compared with controls (P = 0.05; data not shown).

In vivo effects of Aurora kinase inhibition on ovarian carcinoma. To establish the optimal dose and frequency of dosing to effectively inhibit Aurora kinase in vivo, we initiated dose-finding experiments using phospho-histone H3 status as a biological indicator of Aurora kinase activity. Four twice-daily doses of MK-0457 (25 or 50 mg/kg) or vehicle alone were administered by i.p. injection to athymic female mice bearing HeyA8 i.p. tumors 19 days after tumor cell inoculation when the tumors were palpable. Animals were sacrificed 24, 48, and 72 h after the last dose, and tumors were harvested. Examination of the tumors by immunohistochemistry revealed 40% to 50% lower levels of phospho-histone H3 in the 25 and 50 mg/kg groups, respectively, within 24 h after the last dose of inhibitor (P values < 0.001, compared with controls; Fig. 3A). Although some degree of reduced phospho-histone H3 levels was seen at 48 h after the last dose of MK-0457, the most consistently observed response was at 24 h post-treatment; therefore, subsequent in vivo therapy experiments used MK-0457 dosed at 50 mg/kg starting 24 h before taxane-based chemotherapy.

In vivo experiments with diverse cell lines in an orthotopic murine model for metastatic ovarian cancer were employed to characterize the antitumor effects of Aurora kinase inhibition. Aurora kinase inhibition using MK-0457 was initiated 1 week after tumor cell inoculation to model the clinical scenario of minimal residual disease as described previously (22). The four treatment groups consisted of (a) vehicle alone, (b) MK-0457 twice daily for 2 days weekly, (c) docetaxel (or cisplatin for the A2780-CP20 tumor model) i.p. once weekly, and (d) MK-0457 twice daily for 2 days weekly starting 1 day before weekly docetaxel or cisplatin (A2780-CP20). Compared with therapy with vehicle alone, treatment with the inhibitor alone resulted in highly significant 80% and 90% reductions in tumor weight (P values = 0.002 versus controls) in the HeyA8 and SKOV3ip1 tumor models, respectively (Fig. 3B). As expected in these chemosensitive tumor models, docetaxel effectively reduced tumor growth. Combining MK-0457 and docetaxel resulted in the greatest efficacy in reducing tumor burden, eliciting a >90% reduction in tumor weight in both HeyA8 and SKOV3ip1 tumor models (P values = 0.002 versus controls). Furthermore, MK-0457 plus docetaxel exhibited significantly improved efficacy in decreasing tumor burden compared with docetaxel monotherapy in both HeyA8 (68% reduction; P < 0.01) and SKOV3ip1 (85% reduction; P = 0.01) tumor models.

Because recurrent and advanced ovarian cancer is typically refractory to traditional cytotoxic agents, particularly taxanes and platinum agents, we studied the effects of Aurora kinase inhibition in the taxane-resistant, HeyA8-MDR, and platinum-resistant, A2780-CP20, tumor models. Consistent with their resistance profiles, docetaxel monotherapy in the HeyA8-MDR model and cisplatin in the A2780-CP20 model did not alter tumor growth compared with vehicle treatment (P values ≥ 0.9). MK-0457 monotherapy and combination therapy with docetaxel in the HeyA8-MDR tumor model resulted in significant reductions in tumor burden (70% and 90%, respectively; P values < 0.01) compared with controls (Fig. 3B). Similarly, in the A2780-CP20 tumor model, treatment with either MK-0457 monotherapy or combined with cisplatin produced 78% and 92% reductions in tumor weight compared with vehicle (P values < 0.05) and 80% and 91% reductions compared with cisplatin-treated animals (P values = 0.001), respectively.

To further characterize the effects of Aurora kinase inhibition on tumor growth inhibition, we examined tumor nodule formation (Table 1). In the chemosensitive (HeyA8 and SKOV3ip1) and chemoresistant (HeyA8-MDR and A2780-CP20) tumor models, MK-0457 treatment alone resulted in significantly fewer tumor nodules compared with controls (65%, 77%, 60%, and 93% reduction, respectively; all P values ≤ 0.01). Combining MK-0457 with either docetaxel or cisplatin also elicited marked reductions in tumor nodule formation compared with controls (76%, 91%, 55%, and 96%, respectively; all P values ≤ 0.01). In the two chemoresistant tumor models, both MK-0457 monotherapy and combination therapy resulted in significant reductions in tumor nodule formation compared with treatment with docetaxel or cisplatin alone (P values ≤ 0.01).

To examine the therapeutic effect of altering the timing of Aurora kinase inhibition, we varied the time at which MK-0457 was administered relative to the time of docetaxel treatment in the HeyA8 tumor model. No statistically significant difference was observed among the four treatment groups relative to each other (data not shown). We also asked whether the route of MK-0457 delivery could alter efficacy. No significant differences were noted in tumor weight between animals treated via i.v. versus i.p. routes (0.22 versus 0.27 g; P = nonsignificant). Both routes of administration (i.v. and i.p.), however, yielded robust reductions in tumor weight compared with animals treated with vehicle alone (85% and 82%; P < 0.001; Supplementary Fig. S1).

Daily observation of all mice in long-term therapy experiments revealed no overt signs of toxicity such as alterations in bowel or feeding habits, posture, or mobility. No significant differences in body weight were observed among treatment groups in the chemosensitive HeyA8 and SKOV3ip1 models as well as the taxane-resistant, HeyA8-MDR tumor model. Within the A2780-CP20 experiment, mice treated with cisplatin alone weighed 13% and 10% less than mice receiving vehicle alone (P = 0.01) or MK-0457 alone (P = 0.01), respectively. Mice receiving combination MK-0457/cisplatin displayed 21% and 12% reduced body weights compared with both MK-0457 monotherapy and single-agent cisplatin groups (P ≤ 0.001), respectively. Of note, mice receiving MK-0457 alone and vehicle alone showed no difference in body weight in our A2780-CP20 tumor model, further implicating cisplatin as the agent primarily responsible for this side effect. These observations are also consistent with observed cachexia and weight loss seen with cisplatin in the clinical setting (33, 34).

Effects of MK-0457 on proliferation and apoptosis. To explore possible mechanisms underlying the tumor growth inhibition evidenced in the MK-0457 therapy experiments, we first examined its effects on tumor cell proliferation by calculating the proliferative index after PCNA immunohistochemistry on tumors collected at necropsy from all therapy experiments. In the HeyA8 model, the proliferation index for animals treated with vehicle alone was 71%; however, treatment with MK-0457 alone and in combination with docetaxel resulted in significantly lower proliferation indices (23% and 36% reductions, respectively, P values < 0.001; Fig. 4A). Notably, treatment with combination MK-0457 and docetaxel was superior to treatment with docetaxel alone in decreasing tumor cell proliferation (27% reduction; P = 0.004). The SKOV3ip1 tumor model also showed significant 15% and 34% decreases in proliferation indices compared with controls after treatment with MK-0457 both as a single agent and combined with docetaxel, respectively (P values < 0.001; data not shown). Similar to the HeyA8 tumor model, the antiproliferative effect of combining MK-0457 with docetaxel was significantly better than the effect of docetaxel alone (P < 0.05; data not shown).

In the taxane-resistant model, treatment with docetaxel did not decrease the proliferative index (data not shown); however, treatment with either MK-0457 alone or in combination with docetaxel significantly reduced the proliferative indices more than 20% from controls, respectively (P values < 0.001). In the cisplatin-resistant model, treatment with MK-0457 alone and combined with cisplatin resulted in similarly significant reductions in tumor cell proliferation (P values < 0.001) compared with controls. Furthermore, treatment with combination MK-0457 and cisplatin offered a significant 14% reduction in proliferation index beyond that offered by MK-0457 monotherapy (P = 0.02).

Due to the proposed role of Aurora kinases in apoptosis, we examined tumor cell apoptosis using terminal deoxynucleotidyl transferase–mediated dUTP nick end labeling staining in animals from therapy experiments. In the HeyA8 tumor model, inhibition of Aurora kinases using MK-0457 both as a single-agent and combined with docetaxel yielded significant 3-fold increases in the number of apoptotic tumor cells (P < 0.01) compared with untreated animals (Fig. 4A). In the taxane-resistant tumor model (HeyA8-MDR), MK-0457 alone and with docetaxel also resulted in statistically significant >77% increases in the number of apoptotic tumor cells compared with tumors from vehicle-treated animals (P values < 0.03; data not shown).

To gain additional insight into downstream effects of MK-0457, we did expression profile studies on HeyA8 tumors harvested from either vehicle- or MK-0457-treated animals. The expression profiles of tumors from MK-0457-treated animals differed from the vehicle controls as revealed by unsupervised clustering analysis (Supplementary Fig. S2). Tumors from four treated mice formed a single cluster, whereas the five vehicle controls formed another cluster. Using SAM analysis with permutation to estimate the false discovery rate, we identified 174 significant probe sets (false discovery rate < 0.25%) that were up-regulated ≥2-fold in tumors from MK-0457-treated mice (Supplementary Table S1). Several protease gene family members including chymotrypsin, elastase, and carboxypeptidase were found to be highly induced. To validate the up-regulation of these protease genes, real-time reverse transcription-PCR was done using both human- and mouse-specific TaqMan probes. The results indicated that the up-regulation of CPB1 (carboxypeptidase B1), CTRB1 (chymotrypsinogen B1), and ELA2A (elastase 2A) are mouse specific (Fig. 4B), suggesting stromal effects of therapy.

Notable findings from this current study are that inhibiting Aurora kinases using an engineered small-molecule inhibitor shows marked antitumor efficacy in ovarian carcinoma models. These effects were mediated, in part, by significantly decreasing tumor cell proliferation and increasing apoptosis. Furthermore, these mechanisms were consistently shown in both chemosensitive tumor models and two additional models resistant to taxane- and platinum-based chemotherapies.

To date, several studies support the important role Aurora kinases play in cell cycle regulation and high-fidelity mitosis (35). Aurora-A kinase is necessary for mitotic spindle assembly and balanced chromosome segregation between daughter cells. Overexpression results in a tumorigenic phenotype (3); however, the specific mechanisms of regulating Aurora-A expression are still being elucidated. Recently, Kiat et al. identified Aurora-A kinase interacting protein, an endogenous negative regulator of Aurora-A kinase, from a yeast dosage suppressor screen (36). Although they were able to show specificity and efficacy in Aurora-A down-regulation via proteasome-dependent pathways, the clinical potential of degrading Aurora-A kinase was not exploited. Manfredi et al. reported using a small molecule to exogenously inhibit Aurora-A kinase to elicit tumor cell apoptosis and tumor growth inhibition in colorectal and prostate nonorthotopic xenograft models (37). Although they were able to show reductions in tumor growth after long-term treatment with this inhibitor, the use of a nonorthotopic in vivo system may not consider the influence of the relevant tumor microenvironment, an important factor in tumor growth and metastasis (38). This current study extends the available body of knowledge by demonstrating antitumor effects and mechanisms of activity of MK-0457, a highly potent pan-Aurora kinase inhibitor, in an orthotopic in vivo model of metastatic ovarian cancer.

In addition to the fundamental role Aurora kinases play in cell cycle regulation, increasing interest exists in examining its potential role in chemoresistance. In ovarian cancer, chemoresistant recurrence is a significant clinical problem and second-line therapies have limited efficacy; therefore, the potential clinical role for Aurora kinase manipulation in reversing drug resistance may be useful clinically. In vitro, HeLa cells stably overexpressing Aurora-A kinase were shown to be more resistant to taxane-induced apoptosis (10). Similarly, Noguchi showed that patients with breast tumors with high Aurora-A mRNA levels exhibited a lower response rate to docetaxel treatment than patients with low Aurora-A mRNA breast tumors (41% versus 71%; ref. 39). Hata et al. showed that down-regulation of Aurora-A kinase in pancreatic cancer cell lines using small interfering RNA–based targeting resulted in increased sensitivity to paclitaxel (40). Although the specific mechanism for taxane sensitization is not thoroughly elucidated and is likely multifactorial, evidence suggests that apoptosis inhibition plays an important role (11). Our study shows that therapeutic inhibition of Aurora kinases in our taxane-resistant tumor model results in decreased tumor growth with a concomitant increase in apoptosis, further emphasizing apoptosis as an important antitumor mechanism of Aurora kinase inhibition. Remarkably, we discovered and validated that several protease-related genes (CPB1, CTRB1, and ELA2A) were highly up-regulated in the stroma (mouse-derived tissue within the HeyA8 tumor). Expression of these degradative genes within the stroma may be related to the decrease in tumor growth. Further work to gain mechanistic insights regarding stromal effects following Aurora kinase inhibition is being actively pursued.

Based on their important roles in the cell cycle, Aurora kinases represent an intriguing therapeutic target. In fact, several groups have identified small-molecule inhibitors of Aurora kinases, each with different degrees of selectivity for Aurora-A or B. Although other pathways such as the JAK/STAT have recently been implicated in increased aggressiveness and drug sensitivity of ovarian cancer (41, 42), the specificity of MK-0457 for Aurora kinases is significantly greater for Aurora kinases (43), thereby supporting Aurora kinase inhibition as the predominant mechanism for the therapeutic effects observed in these experiments. Here, we describe the antitumor activity of MK-0457 in orthotopic advanced ovarian cancer models that implicates the Aurora kinase family as an important therapeutic target in ovarian cancer. In conclusion, our findings support pan-Aurora kinase targeting using the potent small-molecule inhibitor, MK-0457, alone or in combination with traditional cytotoxic agents, for the treatment of ovarian cancer. Although a recent industry newswire by manufacturers Merck/Vertex (November 20, 2007) reporting QTc prolongation has placed ongoing clinical trials with MK-0457 on hold, the antitumorigenic and therapeutic benefits of targeting the Aurora kinase family in ovarian cancer remain the fundamental findings from our investigations and support additional development of Aurora kinases as a therapeutic target. Our study extends previous work by showing potent antitumor activity in both taxane-resistant and platinum-resistant orthotopic tumor models of metastatic ovarian carcinoma. The robust antitumor effects, including cell cycle disruption and apoptosis induction, seen in our studies provide preclinical rationale for upcoming clinical trials targeting Aurora kinase in ovarian cancer.

No potential conflicts of interest were disclosed.

We thank Derek J. Fiterman, Nicholas Jennings, Karen Ramirez, Donna Reynolds, Diana Urbauer, and Drs. Robert Langley and Yun-Fang Wang for insightful discussions and expertise.