Abstract
To address the increased need to understand the similarities and differences in targeting Aurora A or Aurora B for the treatment of cancer, we systematically evaluated the relative importance of Aurora A and/or Aurora B as molecular targets using antisense oligonucleotides. It was found that perturbations in Aurora A and Aurora B signaling result in growth arrest and apoptosis preferentially in cancer cells. The biological fingerprints of Aurora A and Aurora B inhibition were compared and contrasted in efforts to identify the superior therapeutic target. Due to the different biological responses, we conclude that each Aurora kinase should be treated as autonomous drug targets, which can be targeted independently or in combination. We observed no advantages to targeting both kinases simultaneously and feel that an Aurora A–targeted therapy may have some beneficial consequences over an Aurora B–targeted therapy, such as mitotic arrest and the rapid induction of apoptosis. [Mol Cancer Ther 2006;5(10):2450–8]
- antisense oligonucleotides
- Aurora kinases
- target validation
- pancreatic cancer
Introduction
Several groups have published recently work describing a role for Aurora A kinase in the progression of pancreatic cancer and its potential use as a therapeutic target for the treatment of this disease (1–3). Additionally, several groups have simultaneously developed small-molecule inhibitors for Aurora A; however, due to amino acid sequence similarities and other undetermined factors, the biological consequences of treatment with these compounds is more consistent with Aurora B inhibition rather than Aurora A (4–6). Such findings have fueled a debate among scientists as to which Aurora kinase is a more rational target.
There are three members in the mammalian Aurora kinase family, Aurora A, Aurora B, and Aurora C. Aurora A (STK15/BTAK/Aurora-2) is an oncogenic serine/threonine kinase that plays a role in centrosome separation and in the formation of the mitotic bipolar spindle (2, 7). Aurora B (STK12/Aurora-1) is required for chromosome alignment, kinetochore-microtubule biorientation, activation of the spindle assembly checkpoint and cytokinesis (8, 9). The cellular functions for Aurora C (Aurora-3) have not been elucidated (10), yet recent work points toward a complimentary role to Aurora B function (11–13). Aurora A and Aurora B are up-regulated in various cancers (14, 15). The Aurora A gene is amplified in primary breast and colorectal tumors and in cell lines derived from many cancer types (16, 17). Ectopic expression of Aurora A in fibroblast and near-diploid human breast epithelial cells leads to centrosome amplification, genomic instability, and transformation in vitro and in vivo (14, 18). To the extent of Aurora A, a role for Aurora B in tumorigenesis has not been characterized, although up-regulation of Aurora B has been noted in various cell lines (19) and in some primary tumors, such as high-grade gliomas (20).
Although both Aurora A and Aurora B are considered to be validated anticancer targets for many tumor types, a parallel comparison of Aurora A and Aurora B as molecular therapeutic targets has not been reported. Using antisense oligonucleotides (ASO), we systematically evaluated Aurora A and Aurora B as drug targets in an effort to gain insight into the biological consequences and potential clinical outcomes of inhibiting these kinases independently and in combination.
Materials and Methods
Cell Lines and Tissue Culture
Pancreatic cancer cell lines MIA PaCa-2 and PANC-1, the cervical cancer cell line HeLa S3, the normal lung fibroblast cell line IMR-90, and the normal mammary ductal epithelial cell line MCF10A were all purchased from the American Type Culture Collection (Manassas, VA). The cancer cell lines were maintained in RPMI 1640 supplemented with 10% fetal bovine serum, 100 units/mL penicillin, and 100 μg/mL streptomycin. IMR-90 cells were cultured in EMEM with the same supplements added to the RPMI 1640. MCF10A cells were cultured in MEGM according to American Type Culture Collection guidelines. The cell line HPDE6 (an immortalized but not transformed human pancreatic ductal epithelial cell line) was obtained from Dr. Ming-Sound Tsao (University of Toronto, Toronto, Canada) and maintained in keratinocyte serum-free medium supplemented by epidermal growth factor and bovine pituitary extract (Invitrogen, Carlsbad, CA). The HCT-116 p53 isogenic cell lines were obtained from Dr. Burt Vogelstein (Johns Hopkins University, Baltimore, MD) and cultured in supplemented RPMI 1640. All cells were grown in a humidified incubator at 37°C and 5% CO2. Cells were harvested with trypsin at 80% to 90% confluency. Cell counting was done using trypan blue staining on a hemocytometer.
Quantitative Reverse Transcription-PCR
Total RNA from cell pellets was isolated using the NucleoSpin RNA II isolation kit (BD Biosciences, Palo Alto, CA). Total RNA (1 μg) was used for reverse transcriptase reactions (20 μL total volume), which was carried out using the iScript cDNA Synthesis kit (Bio-Rad, Hercules, CA). An iCycler (Bio-Rad) was used to do real-time fluorescence detection PCR. Primer sequences for Aurora A were 5′-GAATGCTGTGTGTCTGTCCG-3′ and 5′-GCCTCTTCTGTATCCCAAGC-3′ and for Aurora B were 5′-GATGACTTTGAGATTGGGCG-3′ and 5′-GGGACTTGAAGAGGACCTTG-3′. Reactions were carried out in a 16 μL reactions with 200 nmol/L of each primer, iQ SYBRGreen Supermix (Bio-Rad), and 1 μL cDNA. Two-step amplification (95°C for 15 seconds and 56°C for 15 seconds) was repeated for 40 cycles. Following the PCR, a melting curve analysis was done to determine PCR efficiency and purity of the amplified product. Data were provided as a threshold cycle value (Ct) for each sample, indicating the cycle at which a statistically significant increase in fluorescence was first detected. These data were then normalized to β-actin, which served as a reference gene, and then compared with a data from cell treated with the scrambled ASO to determine a relative expression ratio using the Pfaffl (21) or the ΔΔCt methods (22). Primer sequences used for β-actin were 5′-CTGGAACGGTGAAGGTGACA-3′ and 5′-AAGGGACTTCCTGTAACAACGCA-3′.
Western Blot Analysis
Western blots were done as described previously (23) using 50 μg protein per sample and 4% to 12% NuPAGE Bis-Tris gels (Invitrogen). Membranes were probed with a mouse monoclonal antibody against Aurora B (Abcam, Cambridge, MA) at a 1:500 dilution, a mouse monoclonal antibody against Aurora A (BD Biosciences) at a 1:250 dilution, or a mouse monoclonal antibody against β-actin (Sigma, St. Louis, MO) at a 1:20,000 dilution. The membrane was washed and probed with an anti-mouse horseradish peroxidase–linked secondary antibody (Bio-Rad) and visualized with a chemiluminescence kit (Cell Signaling, Beverly, MA) and X-ray film.
ASO Synthesis and Transfection
2′-Methoxyethyl-modified antisense phosphorothioate oligonucleotides were synthesized as described previously (24). The ASOs used were 20 mers and contained 2′-methoxyethyl modifications at the five terminal residues at both the 5′- and 3′-ends of the molecule with 10 contiguous oligodeoxy residues in the center. Oligonucleotides were analyzed by capillary gel electrophoresis and judged to be at least 85% full-length material. The sequences for the ASOs are the following: ASO AurA (Isis 127697), 5′-GTTCTAGATTGAGGGCAGCA-3′; ASO AurB (Isis 173848), 5′-GCCTGGATTTCGATCTCTCT-3′; and scrambled ASO (ASO Scr; Isis 129694), 5′-GTACAGTTATGCGCGGTAGA-3′.
As described previously (1, 23), cells were grown to 40% to 50% confluency and washed with Dulbecco's PBS (Cellgro, Herdon, VA). Opti-MEM (Invitrogen) containing 3 μL Lipofectin reagent (Invitrogen) per milliliter of medium for each 100 nmol/L ASOs was added to the cell culture. ASOs were added dropwise to the final concentrations of 200 nmol/L. Cells were incubated in transfection medium for 6 hours and then given normal growth medium. Cells were harvested using trypsin at the time points indicated.
Flow Cytometry
Cells were treated and harvested as described and stored at −80°C until day of assay. Cell pellets were resuspended in 1 mL Krishan's buffer (0.1% sodium citrate, 0.02 mg/mL RNase A, 0.3% NP40) containing 0.05 mg/mL propidium iodide and incubated for 4 hours. DNA content analysis was then done using a Becton Dickinson FACScan (Palo Alto, CA), modeling 10,000 events per sample.
Apoptosis Assays
Cells grown to 25% to 50% confluency were treated and harvested as described above, counted, and stored at −80°C until day of assay. One million cells for each treatment were resuspended in a cell lysis buffer and incubated for 10 minutes on ice. The cell lysates were centrifuged at 10,000 rpm for 10 minutes at 4°C and the supernatant was transferred to new tubes for detection of caspase-3 activity using the ApoAlert Caspase-3 Fluorescent Assay kit (BD Biosciences) by following the manufacturer's protocol. The samples were read using a Wallac Victor2 multilabel fluorometer (Perkin-Elmer, Boston, MA) with 405 nm excitation filter and 500 nm emission filter.
Cell Morphology Studies
MIA PaCa-2 cells were cultured and transfected with ASOs as described above. After 48 hours of treatment, cells were inspected using a Nikon Eclipse TS100 inverted microscope (Melville, NY) and images were captured using a SPOT Insight digital camera (Diagnostic Instruments, Sterling Heights, MI).
Colony Formation Assay in Soft Agar
Cells were treated with ASOs for 6 hours, trypsinized, mixed with Difco agar (final concentration of 0.26%; BD Biosciences) and RPMI 1640 containing 10% fetal bovine serum, and overlaid onto an underlayer of 0.45% Difco agar containing the same medium in a 35-mm grid Petri dish. Cells (3,000) per Petri dish were seeded and allowed to grow for 19 days at 37°C before counting the number of colonies (≥50 cells) under a light microscope.
Phosphorylated Histone H3 Assay
Approximately 10,000 cells were grown and transfected with ASOs on chambered microscope slides (Nalge Nunc International, Rochester, NY). After treatment, the cells were washed with PBS and fixed in 4% paraformaldehyde solution for 20 minutes at room temperature. The fixed cells were washed with 0.1 mol/L phosphate buffer (pH 7.2) and incubated with blocking buffer containing 0.1 mol/L phosphate buffer (pH 7.2), 0.2% Tween 20, and 2% bovine serum albumin for 1 hour. Anti-phosphorylated histone H3 (pHH3) antibody (Cell Signaling) in blocking buffer was added to the fixed cells at a dilution of 1:50 for 1 hour. The cells were washed again and incubated with an Alexa Fluor 488 goat anti-mouse antibody (Invitrogen) at a dilution of 1:1,000 for 1 hour. The cells were washed and mounted by removing the chamber, adding Aquamount (Lerner Laboratories, Pittsburgh, PA), and placing a coverslip over the cells. Positive staining was quantified by counting stained cells under a fluorescence microscope and dividing by the number of total cells visible under a light microscope.
Gene Expression Profiling
The Agilent Human 1A (v2) expression array (Palo Alto, CA) containing 18,411 unique discovery probes was used to analyze the knockdown potential of ASOs in MIA PaCa-2 cells. Furthermore, the Affymetrix (Santa Clara, CA) U133A Plus 2 platform was used for additional ASO experiments and for gene expression profiling of the panel of pancreatic cancer cell lines.
For the ASO experiments, RNA isolates were generated after 48 hours of incubation with ASOs as described above. The RNA preparation, labeling, hybridization, and scanning were according to the manufacturer's recommendations. The Agilent data consisted of two channels of intensity values, control versus antisense. The ratios of the two channels were loaded into GeneSpring software, version 7.2 (Agilent), and analyzed using a cutoff of 2-fold to determine genes, which changed expression due to ASO AurA or ASO AurB treatments.
The experimental design for the Affymetrix platform used a reference array containing the ASO Scr and additional arrays using the ASO AurA, ASO AurB, and ASO Both treatments. GC-biased Robust Multichip Averaging (25) and MAS5 (Affymetrix GeneChip Operating software, version 5.0) were used to compare and accommodate the associated varying type I and type II error rates for each normalization scheme.
Reverse transcription-PCR experiments were done as described above to confirm changes in expression levels identified from arrays. Primer sequences used were as follows: 5′-GCCTGAGCCTATTTTGGTTG-3′ and 5′-GGATCAGCTCCATCTTCTGC-3′ (cyclin B1); 5′-GCGTACTCAAATCCGAGAGC-3′ and 5′-ATAATGGTGGCCACAGCTTC-3′ (PRC-1); 5′-TGTACCTGTGGAGTGCAAGC-3′ and 5′-GATCCAGGCCACAGAGGATA-3′ (CDC20); 5′-GATCAAGTGTGACCCGGACT-3′ and 5′-TCCTCCTCTTCCTCCTCCTC-3′ (cyclin D1); 5′-GTGGCCAAACAGGAAAGTGT-3′ and 5′-TCAGGAGCCTCTCCAGGTAA-3′ (INCENP); 5′-GGCAGTTGACCAACACAATG-3′ and 5′-CTTCTAGCATGGCCTTTTGC-3′ (Eg5/KSP); 5′-TACTGTCGGGGAAGTTCCAG-3′ and 5′-GAGACCTGTCCTCTTCACGC-3′ (CDC25C); 5′-TGGCCGAGTTCTTCTCATTC-3′ and 5′-AAACCTCCTGGTTCCTTTTGA-3′ (Mad2L1); 5′-AAGAGATCCCGGAGGTCCTA-3′ and 5′-TCATTCAGGAAAAGGTTGCC-3′ (Plk-1); 5′-CTGGCAGCCCTTTCTCAAGGACCA-3′ and 5′-CCAGCCTTCCAGCTCCTTGAAGCA-3′ (survivin); 5′-CACCATGGCACAAGTTAAAAG-3′ and 5′-ATTCTCCAAATTGGCCTTCTC-3′ (TPX2); and 5′-TTGTCAAGCTGCTGGATGTC-3′ and 5′-TGAGTCCAAATAGCCCAAGG-3′ (CDK2).
Statistical Analyses
For all quantitative reverse transcription-PCR experiments, model I (fixed effects) one-way ANOVA analyses were done to determine whether there was a significant difference among the measured groups. A strong between-group significance resulted in all cases. Pair-wise t tests were then done to identify significance compared with the proper control for each experiment. Ps reported result from t tests. A similar approach was used to analyze the data from pHH3 experiments.
Results
Inhibition of Aurora Kinase Expression by ASO
Using ASO, the expression of Aurora A and Aurora B were effectively reduced at both transcript and protein levels (Fig. 1A and B ). Target reduction was observed in all four cell lines used throughout this study (MIA PaCa-2, PANC-1, HeLa, and HPDE6) at time intervals between 24 and 72 hours after treatment. Using quantitative reverse transcription-PCR, it was determined that the reduction in expression of Aurora A and Aurora B ranged between 85% to 97% in the cell lines (Fig. 1A). The same samples were also evaluated for target reduction by Western blot, which showed similar levels of reduction (Fig. 1B for MIA PaCa-2 cells; data not shown for other cell lines).
Inhibition of Aurora A and Aurora B expression by ASO. A, quantitative reverse transcription-PCR measurements of mRNA levels in MIA PaCa-2 cells treated with 200 nmol/L ASOs for 24, 48, and 72 h. *, P < 0.05; **, P = 0.002, comparing the expression of Aurora B in ASO Scr to ASO AurA. B, Western blot analysis of Aurora A and Aurora B protein levels in the same ASO-treated samples. Similar experiments were done for PANC-1, HPEDE6, and HeLa (data not shown).
Cell Cycle Analysis on ASO-Treated Cancer Cells
Approximately 5% of untreated MIA PaCa-2 cells or cells treated with a ASO Scr were in the G2-M fraction of the cell cycle (Fig. 2A and B ) as determined by FACScan. After 24 hours of treatment with ASO AurA or ASO AurB, this fraction increased to 13% and 31%, respectively. At 48 hours, 41% (ASO AurA) and 59% (ASO AurB) of the cells were arrested and after 72 hours 26% (ASO AurA) and 48% (ASO AurB) were in G2-M. These results for ASO AurA are consistent with those previously published (1).
A, FACScan profiles of MIA PaCa-2 cells treated with the indicated ASO for 48 and 72 h. Although cells treated with ASO AurB showed a marked increase in cells with 4N DNA, they do not undergo a cell cycle arrest. Rather, these cells proceed through the cell cycle without dividing resulting in cells with 8N DNA. These results are also seen in ASO Both-treated MIA PaCa-2 cells; however, the accumulation of 8N cells was slower. B, percentage of cells with 4N DNA following Aurora ASO treatment as described. C, caspase-3 activity in Aurora ASO-treated cells. Note that HPDE6 cells did not share the same increase in caspase-3 activity following ASO treatment as MIA PaCa-2 and PANC-1 cells.
To determine the effects of targeting Aurora A and Aurora B simultaneously, MIA PaCa-2 cells were treated with ASO AurA and ASO AurB in combination (ASO Both). These cells were also arrested where 26% (24 hours), 48% (48 hours), and 40% (72 hours) of the cells were in the G2-M phase of the cell cycle. These percentages were significantly higher than those for ASO AurA and slightly lower than those for ASO AurB, thus indicating no additive or synergistic effects in cell cycle distribution when both kinases are targeted.
Cell cycle arrest profiles with ASO-treated PANC-1 and HPDE6 lines were similar to those of MIA PaCa-2, although the percentage of arrested cells was less. This is likely due to the combination of two reasons: (a) slower doubling time of these cell lines compared with MIA PaCa-2 and (b) target reduction is slightly lower in these cell lines following ASO treatment compared with MIA PaCa-2 cells (85-90% versus 95-97%). Interestingly, PANC-1 and HPDE6 cells had greater cell cycle arrest when treated with ASO Both. This was not observed in MIA PaCa-2 where simultaneous treatment with ASO Both resulted in a G2-M fraction between those obtained for single treatments.
ASO AurB-Induced Polyploidy
On further analysis of the FACScan profiles, it became evident that some cell lines treated with ASO AurB did not arrest in mitosis. Beyond 24 hours after treatment, ASO AurB-treated MIA PaCa-2 cells began to show cells with >4N DNA content (Fig. 2A). This observation was expected considering the biological function of Aurora B and has been reported in other cell types (26, 27). ASO AurB-treated MIA PaCa-2 cells entered and exited mitosis; however, cytokinesis failed. Therefore, these cells were able to continue through the cell cycle yet did not divide and thereby produced the phenotype of polyploidy and multinucleation as observed.
MIA PaCa-2 cells treated with both ASO AurA and ASO AurB showed a phenotype most similar to ASO AurB treatment alone. Both treatments resulted in a population of cells with 8N DNA content. These findings were somewhat consistent with a recent study where Aurora A and Aurora B were knocked down individually and in combination with small interfering RNAs (28). From that study, the authors concluded that Aurora A function was not required in mitosis when Aurora B was inactivated. In other words, knocking down both kinases produced the exact same result as knocking down Aurora B alone. In contrast to this previously reported study, we observed a small difference between ASO Both treatments and ASO AurB treatments in that the polyploidy cells accumulated slower in the ASO Both treatments.
ASO AurB- or ASO Both-induced polyploidy was also observed in PANC-1 and HPDE6 cells but, interestingly, not when used in HeLa cells (data not shown). To determine if this differential response was solely due to p53 status, the isogenic cell lines HCT-116 (p53+/+) and HCT-116 (p53−/−) were treated with ASO AurB. Cell cycle analysis of these samples by flow cytometry did not indicate any differences (data not shown).
Apoptosis Analysis on ASO-Treated Cancer Cells
Apoptosis was monitored in ASO-treated cells by measuring caspase-3 activation (Fig. 2C). No significant amount of caspase-3 activity was detected after 24 hours of treatment with any of the ASOs. However, at 48 hours, there was a noteworthy increase in caspase-3 activity in some of the cell lines treated with ASOs. MIA PaCa-2 cells showed a 6- to 7-fold increase in caspase-3 activity following treatment with ASO AurA at both 48 and 72 hours. At 48 hours, ASO AurB- and ASO Both-treated MIA PaCa-2 cells increased their caspase-3 activity by 2.6- and 2.5-fold, respectively. This activity increased to 3.3- and 4.7-fold at 72 hours for the same treatments. PANC-1 cells exhibited a similar apoptotic response, however, to a much lesser degree compared with results obtained in MIA PaCa-2 cells. However, HPDE6 cells showed no significant (P > 0.05) increase in caspase-3 activity in any of the treatments at these three time points. To see if the lack of caspase-3 activity carried over to other normal cell lines, we evaluated the effects of the ASOs on apoptosis in IMR-90 and MCF10A cells. Similar to HPDE6, we did not observe any apoptosis in these cell lines in any of the ASO treatments over 72 hours (data not shown). However, it should be noted that transfection efficiency and expression knockdown was less effective in these two cell lines compared with other cell lines used in this study.
Cell Morphologic Changes in ASO-Treated Pancreatic Cancer Cells
Cells treated with the ASOs underwent morphologic changes consistent with the findings from FACScan and caspase-3 analyses. Visual inspections of cells treated with ASO AurA revealed that they had rounded up, indicating mitotic arrest and apoptosis (Fig. 3 ). At times beyond 48 hours, the cells began to die and detach from the surface of the flask (data not shown). Alternatively, cells treated with ASO AurB and ASO Both become oversized and multinucleated (MIA PaCa-2 and PANC-1), but the majority of the cells remained firmly attached to the flask and maintained viability.
Morphologic changes induced by Aurora ASO treatment in MIA PaCa-2 cells. Cells treated with ASO AurA were rounded up and detached from growth surface, indicating mitotic arrest and apoptosis. Alternatively, cells treated with ASO AurB and ASO Both became oversized and multinucleated, but the majority of the cells remained firmly attached to the flask and maintained viability. Black bar, distance of 200 μm.
Effects of ASO Treatment on Growth in Soft Agar
To expound on these observations, we evaluated the roles of Aurora A and Aurora B in maintaining tumorigenicity in established pancreatic cancer cell lines. ASO-treated MIA PaCa-2 and PANC-1 cells were grown in an anchorage-independent manner in soft agar. MIA PaCa-2 cell exhibited a 69%, 81%, and 78% reduction in colony-forming units for ASO AurA, ASO AurB, and ASO Both, respectively (Fig. 4 ). Similarly, PANC-1 cells showed reductions of 65%, 87%, and 91% for the same treatments, respectively.
Colony formation in soft agar by cells (MIA PaCa-2 and PANC-1) treated with Aurora ASOs.
The reduction in colony formation in both cell lines was statistically significant (P < 0.05) for all three treatments when compared with the ASO Scr-treated control. Hata et al. (29) reported recently that knockdown of Aurora A with small interfering RNA resulted in suppression of colony formation in soft agar. Our results agree with theirs. Correlating with results from other types of assays (cell cycle arrest, apoptosis, etc.), reductions in colony formation for cells treated with ASO AurB and ASO Both were very similar. Generally, the effects of an ASO AurB treatment were greater compared with the ASO AurA treatment in the soft agar assay. This observation was more obvious in the PANC-1 cells than it was in the MIA PaCa-2 cells (Fig. 4).
Treatment with ASO AurA Induces an Increase in Aurora B Expression and Activity
We observed from target reduction validation studies (Fig. 1) that Aurora B was up-regulated in cells treated with ASO AurA. At the mRNA level, this increase in Aurora B expression was subtle yet statistically significant (Fig. 1A). Additionally, the up-regulation of Aurora B protein as determined by Western blot (Fig. 1B) was more remarkable. The reasons for the up-regulation of Aurora B in the absence of Aurora A are not known; however, it could be due to a compensatory role of Aurora B for the loss of Aurora A function. It could also simply be due to the fact that cells treated with ASO AurA are arrested in mitosis, which is the portion of the cell cycle when Aurora B is typically expressed. Therefore, it is plausible that the up-regulation of Aurora B is simply due to a larger fraction of cells in the ASO AurA-treated population expressing Aurora B compared with an asynchronous population.
Regardless of the reasons for this up-regulation, we determined if Aurora B activity was higher in ASO AurA-treated cells by detecting pHH3 levels, a cellular substrate for Aurora B. In ASO Scr-treated MIA PaCa-2 cells, only 1.45% of the cells stained positive for pHH3, which is similar to that of untreated cells (1.31%; Fig. 5 ). However, treatment with ASO AurA resulted in an increase to 2.99%, indicating an ∼2-fold increase in Aurora B activity. In cells treated with ASO AurB, the percentage of pHH3-stained cells fell to 0.59% (Fig. 5).
pHH3 levels in ASO-treated MIA PaCa-2 cells. A, ASO AurA resulted in an increase in pHH3 levels and ASO AurB caused a decrease. Black bar, distance of 200 μm. B, quantitation of pHH3-positive cells shown in A. *, P = 0.005; **, P = 0.001 when compared with ASO Scr.
Gene Expression Profile Changes in ASO-Treated Pancreatic Cancer Cells
To gain insight into additional genes whose expression level changed following perturbations in Aurora A signaling, microarray experiments were done on MIA PaCa-2 cells treated with ASOs for 48 hours. The microarray analysis on the ASO-treated cells confirmed the efficacy of the ASOs at reducing expression of their target (Fig. 6 ). Additionally, the analysis also confirmed the up-regulation of Aurora B in ASO AurA-treated cells. As indicated above, this could be either indicative of a compensatory role for Aurora B in the loss of Aurora A function or simply due to ASO AurA-induced mitotic arrest. We felt that if the up-regulation of Aurora B were due to cell growth arrest at the mitotic phase, other mitotic genes would also be up-regulated in the ASO AurA-treated sample. Conversely, if the up-regulation was due to a compensatory role, then it should be unique compared with other mitotic genes. The microarray analysis revealed several mitotic genes likewise up-regulated in the ASO AurA-treated cells. These genes included a 2-fold or greater induction of cyclin B1, survivin, Plk-1, PRC-1, CDC20, and CDC25C, to name a few (Fig. 6). The expression levels of several genes were confirmed by quantitative PCR. Considering these data, we conclude that the up-regulation of Aurora B in ASO AurA-treated cells is likely due to a mitotic arrest of the cells and not a function of Aurora B compensating for the loss of Aurora A expression.
A, gene expression changes on the treatment of ASO determined by DNA microarrays. Changes in Aurora A and Aurora B expression validated the down-regulation of these genes by ASO treatment. Treatment with ASO AurA resulted in an increase in Aurora B expression, which was also seen with other mitotic genes. B, quantitative reverse transcription-PCRs were done to confirm the expression of other mitotic genes change in parallel with Aurora B on treatment with ASO AurA. In contrast, other genes, such as CDK2 and cyclin D1, which are involved in cell cycle control but not mitosis, were unchanged.
Discussion
The Aurora kinases have emerged as attractive therapeutic targets for the treatment of various tumor types. Unexpected results from drug development efforts have led many to ask which Aurora kinase should be targeted to produce the most beneficial effect. To study some of the differential biological consequences between Aurora A- and Aurora B-specific therapies, we used ASOs to down-regulate these kinases independently and in combination.
Effects on Cell Cycle Distribution
As described, ASO treatment resulted in accumulation of 4N cells over 72 hours. The lower percentage of cells in the G2-M population at 72 hours compared with 48 hours may suggest the effects of ASO treatment (target reduction) wane with time and that cells can return to a normal cell cycle distribution following transient perturbations in Aurora kinase signaling; however, this explanation is not supported by target reduction validation studies (Fig. 1A and B). Another reason for a reduction in the percentage of cells in G2-M at 72 hours compared with 48 hours may be due to arrested cells having undergone apoptosis.
Although we saw similar results to those reported by Yang et al. the accumulation of 8N cells was slower in ASO Both-treated cells compared with ASO AurB. The reason(s) for the lower accumulation of 8N cells in the ASO Both treatment is not clear; however, at least two possibilities exist. First, treating cells with ASO Both is not as effective at target reduction as treating with ASO AurB. If this were the case, a lower accumulation of 8N cells would be expected in the ASO Both-treated cells because a higher percentage of cells would have some remaining amounts of functional Aurora B.
However, this potential explanation is not supported by target reduction studies (Fig. 1A and B). Secondly, as suggested by Yang et al., Aurora A activity may not be required for mitotic progression in cells without active Aurora B; however, the absence or inhibition of Aurora A in combination with the inhibition of Aurora B may have resulted in a mitotic delay that is not sustained but eventually circumvented. Therefore, cells without functioning Aurora A and Aurora B enzymes become polyploid; however, this happens at a slower rate than cells with functional Aurora A but no Aurora B. This conclusion is consistent with the observations made in this study; however, Yang et al. did not claim to have seen a slower accumulation of 8N cells when treated with both Aurora A and Aurora B small interfering RNA.
It has been noted that the inhibition of the Aurora kinases by small molecules can result in two fates: (a) a cell enters additional replicative cycles followed by the continued failure of cell division resulting in polyploid cells and (b) tetraploid cells exit from mitosis and undergo a sustained G1-like arrest (4, 30). MIA PaCa-2, PANC-1, and HPDE6 cells treated with ASO AurB experience the first of these fates and HeLa cells the second. The literature suggests that an important factor in determining which fate is reached depends on the status of a p53-dependent postmitotic checkpoint (31). It has been proposed that, aside from its role in responding to DNA damage, p53 can respond to cytokinetic failures by inducing a G1-like arrest. In this manner, p53 may compliment other checkpoints in averting the propagation of cells with an aberrant genome. Consistent with this notion, MIA PaCa-2 and PANC-1 cells are reported to harbor p53 mutations (32, 33) and HPDE6 cells have been immortalized by transduction of the human papillomavirus 16-E6E7, which leads to the inactivation of the p53 pathway (34). However, in contrast, HeLa cells express low levels of p53 protein (35), but it has been inactivated by human papillomavirus 18. Furthermore, HCT-116 p53 isogenic cell lines responded no differently to ASO AurB treatment. Taken together, these data suggest that p53 status alone probably does not account for the fate reached on Aurora B inhibition.
Effects on Apoptosis
Caspase-3 activity experiments suggest that apoptosis is rapidly induced in cancer cells treated with ASO AurA, which peaks at 48 hours. Apoptosis was also detected in ASO AurB-treated cancer cells; however, its induction was more gradual compared with that in ASO AurA-treated cells (Fig. 2C). It is likely that caspase-3 activity in ASO AurB-treated cells would reach similar levels as ASO AurA-treated cells at time points beyond 72 hours. Interestingly, the levels of caspase-3 activity in ASO AurB-treated MIA PaCa-2 cells and HeLa cells were very similar, although this same treatment resulted in different morphologic responses. Therefore, the postmitotic checkpoint potentially responsible for the G1-like arrest seen in ASO AurB-treated HeLa cells does not likely determine the viability of cells exposed to Aurora B kinase inhibition.
The lack of apoptosis in HPDE6, IMR-90, and MCF10A cells treated with ASOs suggests that these cells possess some resistance to the inhibition of the Aurora kinases. Apparently, rapidly dividing cells are more sensitive to the inhibition of both Aurora A and Aurora B. Although this is likely true, it does not entirely account for this observation because PANC-1 cells have a similar doubling time as HPDE6 cells and apoptosis was detected in these cells. Another contributing factor to the lack of apoptosis in normal cells treated with ASO may be that these cells have intact spindle assembly checkpoints that allow the cells to manage Aurora A or Aurora B inhibition without undergoing apoptosis. In other words, on inhibition of Aurora A or Aurora B expression, normal cells arrest and remain arrested due to proper cellular checkpoints. In cancer cells with compromised checkpoints, cells respond to cell cycle arrest induced by Aurora A or Aurora B inhibition by undergoing apoptosis.
Effects on Growth in Soft Agar
It is clear that both Aurora A and Aurora B contribute to the tumorigenicity of these pancreatic cancer cell lines and to their ability for self-renewal. Surprisingly, Aurora B inhibition consistently had a greater effect in inhibiting pancreatic cancer cell growth in soft agar. This finding is not in line with result from the caspase-3 activity analysis. One would reason that increased apoptosis (seen with ASO AurA) would lead to a larger decrease in growth in soft agar (seen with ASO AurB). Mechanisms behind this difference were not further explored; however, a one possible explanation is that the caspase-3 activity assays were carried out over a period of just 72 hours and soft agar experiments ran for up to 21 days. As mentioned above, we speculate that due to the gradual increasing levels of caspase-3 activity over the 72 hours following ASO AurB treatment that apoptosis levels would eventually match if not exceed those seen in ASO AurA treatments if time points beyond 72 hours were considered. Therefore, the differences between the gradual induction of caspase-3 activity seen in ASO AurB treatments versus the rapid activity seen in ASO AurA treatments would be less important in an assay that spanned several weeks. A second potential explanation that we have no supporting evidence for is that effects of ASO AurB treatment may differ in cells growing anchorage dependently versus anchorage independently. For example, Aurora B may play a more important role in a cellular response that may be more important for survival when under stressful conditions. This is, however, highly speculative.
Conclusion
In summary, Aurora A and Aurora B kinases have emerged as important potential therapeutic targets in recent years (36). Initially, due to the well-established role of Aurora A in tumorigenesis, it was considered the primary useful drug target of the two kinases. Surprisingly, as inhibitors of the Aurora kinases have advanced into preclinical and clinical development, it was noted that treatment with these compounds resulted in a cellular phenotype entirely consistent with Aurora B inhibition and not Aurora A. This finding has fueled a healthy debate as to which Aurora kinase is the most appropriate target, which encouraged us to embark on a systematic analysis of comparing expression patterns and the biological consequences of inhibiting Aurora A, Aurora B, or both in combination by using ASOs.
To further evaluate Aurora A and Aurora B as therapeutic targets, we subjected ASO-treated cells to a series of assays to determine potential consequences of perturbing the function of the Aurora kinases. In summary, inhibiting either Aurora A or Aurora B results in effects that may be therapeutically useful, such as the inhibition of cell proliferation, cell cycle arrest, and apoptosis. Interestingly, the induction of apoptosis was more rapid in ASO AurA-treated cells, probably due to the mitotic arrest observed in these cells. In contrast, ASO AurB-treated cells showed a more gradual induction of apoptosis. The precise benefit of one response over the other is not clear; there does not seem to be an advantage to targeting both Aurora A and Aurora B in combination because such an approach would result in consequences very similar if not identical to targeting Aurora B alone. Finally, because Aurora A and Aurora B kinases are often not distinguished as distinctive therapeutic targets, we feel one important conclusion from this study is that the biological consequences and potential clinical outcomes of inhibiting Aurora A or Aurora B are very dissimilar and that they should be distinguished as two distinctive therapeutic targets that can be targeted independently.
Acknowledgments
We thank the Flow Cytometry shared services at the Arizona Cancer Center for their assistance, Dr. Burt Vogelstein for the HCT-116 isogenic cell lines, and Dr. Ming-Sound Tsao for providing us with the HPDE6 cell line.
Footnotes
Grant support: American Chemical Society Division of Medicinal Chemistry and Wyeth predoctoral fellowship (S.L. Warner) and NIH grant CA 95031 (D.D. Von Hoff).
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
- Accepted August 16, 2006.
- Received April 11, 2006.
- Revision received July 28, 2006.
- American Association for Cancer Research
References
Volume 5, Issue 10
