|Home | About | Journals | Submit | Contact Us | Français|
Increased activity of SRC family kinases promotes tumor invasion and metastasis, and overexpression of the mitotic regulator Aurora kinase A (AURKA) drives tumor aneuploidy and chromosomal instability. These actions nominate SRC and AURKA as valuable therapeutic targets for cancer, and inhibitors for SRC and Aurora kinases are now being employed in the clinic. In this study, we demonstrate potent synergy between multiple inhibitors of Aurora and SRC kinases in ovarian and colorectal cancer cell lines, but not in normal ovarian epithelial cell lines. Combination of Aurora and SRC inhibitors selectively killed cells that have undergone a preceding aberrant mitosis, and was associated with a post-mitotic reattachment defect, and selective removal of aneuploid cell populations. Combined inhibition of Aurora kinase and SRC potentiated dasatinib-dependent loss of activated (Y416-phosphorylated) SRC. SRC and AURKA share a common interaction partner, NEDD9, which serves as a scaffolding protein with activities in cell attachment and mitotic control, suggesting SRC and AURKA might interact directly. In vitro, we observed physical interaction and mutual cross-phosphorylation between SRC and AURKA that enhanced SRC kinase activity. Together, these findings suggest that combination of SRC and Aurora-targeting inhibitors in the clinic may be a productive strategy.
Substantial effort has been devoted to the creation of protein-targeted biological therapeutics intended to prevent the growth and metastasis of tumor cells (Cao and Liu, 2007; Pfeiffer et al., 2007). However, pre-clinical testing and early phase trials have generally indicated that few of these agents employed as monotherapies have produced strong gains in blocking tumor growth (Brandsma and van den Bent, 2007; Wong and Goodin, 2009). An emerging strategy in overcoming resistance is the identification of novel combination therapies involving targeted agents, as these often have enhanced clinical efficacy (Bagnyukova et al.). We now describe a novel strategy involving the combination of inhibitors of SRC and Aurora kinases.
The tyrosine kinase SRC regulates cell proliferation, survival, cytoskeleton, cell-cell contacts, and cell-matrix attachments at focal adhesions (Guarino, 2010; Yeatman, 2004). At focal adhesions, SRC forms an activated complex with focal adhesion kinase (FAK) and scaffolding proteins such as NEDD9 (Yeatman, 2004) to regulate focal adhesion turnover, stimulating downstream effector pathways that reorganize the cytoskeleton during migration. SRC also becomes hyperactive at the G2/M transition, and phosphorylates substrates that promote focal adhesion disassembly and facilitate the cell rounding phenotype characteristic of mitosis (Mustelin and Hunter, 2002). SRC is overexpressed in a large percentage of human tumors including colorectal, breast, ovarian, prostate, pancreas, lung carcinoma, glioma, melanoma, head and neck, and different types of sarcoma (Guarino, 2010; Shor et al., 2007). Interestingly, SRC is only rarely mutated in tumors, suggesting aberrant interactions with partner proteins may contribute to its increased activity. Activated SRC supports tumor growth, invasion, and metastasis (Yeatman, 2004).
Aurora serine/threonine kinases (Aurora-A/AURKA, Aurora-B/AURKB, and Aurora-C/AURKC) are best known for roles in M-phase (Gautschi et al., 2008; Keen and Taylor, 2004), although recent publications have begun to identify roles for AURKA in interphase cells (Plotnikova and Golemis, 2011; Plotnikova et al., 2010; Pugacheva et al., 2007). AURKA is best known for regulating mitotic entry and progression, influencing centrosome maturation and separation, bipolar-spindle assembly, chromosomal alignment on the metaphase plate and cytokinesis (Dutertre et al., 2002; Glover et al., 1995). As with SRC, AURKA protein overexpression often occurs in ovarian, colorectal, pancreatic, liver, bladder and gastric cancers, and is generally associated with genomic instability and aneuploidy, a higher tumor grade and a poor prognosis (Bischoff et al., 1998; Giet et al., 2005). Exogenous overexpression of AURKA either in vitro or in vivo transforms rodent fibroblast cells and induces tetraploidization, failed cytokinesis, and genomic instability. Overexpressed AURKA also affects the DNA damage-induced G2 checkpoint, and the mitotic spindle checkpoint (Anand et al., 2003). AURKB is rarely overexpressed in tumors, but plays important roles in regulating kinetochore and chromasome segregation in mitosis (Ditchfield et al., 2003). In contrast, AURKC has limited expression, and has primarily been studied in the context of meiotic cell division and early development (Fernandez-Miranda et al., 2011).
Several lines of evidence suggested the possibility of functional or direct physical interactions for SRC and AURKA. Both proteins interact with a common partner, NEDD9, with these interactions contributing to activation of each kinase (Minegishi et al., 1996; Pugacheva and Golemis, 2005; Pugacheva et al., 2007). Besides its role in mitosis, AURKA activity has recently been identified as regulating cell migration, similar to NEDD9 and SRC (Guan et al., 2007). An emerging paradigm for the design of combinatorial therapies takes as its model synthetic lethal analyses from lower organisms. These studies have established that dual inhibition of proteins operating together or in parallel pathways is often particularly effective in reducing signaling function, to potential therapeutic gain (Friedman and Perrimon, 2007). With this goal, we have here assessed the consequences of dual inhibition of Aurora kinases and SRC. We here report that multiple SRC and Aurora inhibitors synergize strongly to promote cell killing in cancer cell lines. This cell killing is associated with a post-mitotic attachment defect, clearance of aneuploid cell populations, and hypoactive SRC. We also establish that SRC and AURKA directly associate and phosphorylate each other in vitro, supporting the idea of direct interaction. These and other findings provide an initial justification for combining SRC and Aurora inhibitors in the clinic.
Dasatinib is selective for SRC family kinases (SRC, YES, LCK) and BCR-ABL, with activity at higher concentrations as an inhibitor of PDGFRβ, c-KIT, and p38 (Lombardo et al., 2004). We initially assessed the combination of dasatinib, and the Aurora kinase inhibitor PHA-680632 (Soncini et al., 2006) in limiting cell growth of a panel of human epithelial ovarian cancer (EOC) cell lines (Figure 1A, Supplemental Table 1), and normal primary human ovarian surface epithelial (HOSE) cells (Figure 1B). Chou-Talalay analysis of results indicated that at multiple different drug combination ratios, dasatinib and PHA-680632 synergized to restrict the growth of 8/8 EOC cell lines, but 0/3 HOSE cell lines. Similar synergy was obtained using either of two alternative small molecule inhibitors of Aurora kinases (MLN8237 or C1368, both preferentially inhibiting AURKA) in combination with dasatinib, or an alternative small molecule inhibitor of SRC kinases (PP2) in combination with PHA-680632 (Figure 1C). Potent synergy was also observed in 3/3 colorectal cancer cell lines, including the RAS-mutated DLD1 cell line and its RAS wild-type isogenic counterpart, DKS8 (Figure 1C).
As separate measures of efficacy of the dasatinib-PHA630632 combination, we performed clonogenic assays (Figures 2A, 2B) and assessed colony formation on Matrigel (Figures 2C, 2D). By both measures, treatment of cells with the combination resulted in significantly reduced colony size and total number of colonies versus treatment with either single agent at the same dose. Further, cells within Matrigel colonies treated with the drug combination were much more positive for cleaved caspase-3 (Figure 2E), indicating the occurrence of apoptotic cell death, rather than solely cytostasis.
To begin to explore the molecular basis for the strong synergy in growth inhibition, we investigated the consequences of combining dasatinib and PHA-680632 on cell cycle progression (Figure 3A). Treatment of thymidine-synchronized OVCAR10 cells with dasatinib caused G1 accumulation (57% ± 11% of total cells versus 45% ± 3% for vehicle treated cells) within 24 hours. Treatment with PHA-680632 caused accumulation of cells in the 4N (G2/M) compartment (38% ± 6% of total cells versus 32% ± 2% for vehicle treated cells), and also led to the appearance of a significant >4N cell population (30% ± 4% of total cells versus 5% ± 2% for vehicle treated cells). Treatment with the dasatinib/PHA-680632 combination resembled treatment with PHA-680632, but further increased the proportion of 4N cells (to 46% ± 8%) versus PHA-680632 alone. Intriguingly, the combination significantly and selectively reduced the hyperploid >4N population at 19 and 24 hours relative to PHA-680632 treatment alone (to 7% ± 2% and 11% ± 2%, respectively, versus 18% ± 2% and 30% ± 4%; P = 0.0048 and 0.0015). Similar results were obtained in a second cell line, PEO-1 (Supplementary Figure S1), but were not observed in HOSE cells treated with the same concentrations of drugs (Figure 3B). Further, although a >4N hyperploid population was induced by treatment of cells with inhibitors selective for AURKA (MLN8257) or AURKB (AZD1152), only the combination of dasatinib with MLN8257 selectively reduced this population of cells (Figure 3C), suggesting a specific interaction between dasatinib targets and AURKA.
The FACS results suggested that the drug combination-induced lethality specifically involves mitotic progression. To better address mitotic progression following drug treatment, live cell imaging was used to characterize in detail the progression through mitosis of thymidine-synchronized (S-phase arrested) OVCAR10 cells that stably expressed GFP fused to histone H2B, to allow visualization of nuclei and chromosomes. Drugs were added 3 hours after removal of thymidine. The time required for entry to mitosis, measured as chromosomal condensation, was equivalent among the treatment groups (5.28 ± 1.68 hrs for vehicle, 5.49 ± 1.52 hrs for PHA-680632, 4.63 ± 2.10 hrs for dasatinib, and 5.42 ± 1.67 hrs for the dasatinib/PHA-680632 combination) (Figure 3D). PHA-680632 treatment slightly decreased the percent of cells entering mitosis, but this effect was eliminated when cells were treated with the drug combination (Figure 3E). Treatment of cells with PHA-680632 increased the overall duration of mitosis (from 93.6 to 139.4 minutes), which was similar to the delay seen in cells treated with PHA-680632 plus dasatinib (149.2 minutes) (Figure 3F). PHA-680632 treatment induced a high frequency of cells that underwent aberrant mitosis with cleavage furrow regression to produce a binucleate single progeny (Figure 3G, Supplementary Figure S2), but this frequency was not further enhanced by addition of dasatinib. A marked increase in cells undergoing abrupt condensation and fragmentation of chromosomal material, to produce the pyknotic nuclei typically associated with apoptotic cells, was evident in live cell imaging of combination-treated cells (Figure 3H, I).
The preceding results might indicate the drug combination increased the killing of the post-mitotic, hyperploid cells, or caused death during mitosis of cells that would otherwise be destined to become hyperploid. To discriminate between these possibilities, we performed a subgroup analysis of cells that underwent cell death, as defined by nuclear fragmentation and cellular condensation, in live cell imaging experiments focusing on GFP-H2B, to visualize chromatin. We followed the fate of individual cells for up to 48 hours following release from thymidine block, and categorized cell death as 1) death without mitotic entry, 2) death in mitosis, 3) death of one or both daughter cells produced from an apparently normal mitosis, and 4) death of a multinuclear daughter cell produced after an aberrant mitosis (Supplementary Figure S2). Compared to vehicle or the individual drug treatments, the majority of cell deaths following treatment with the combination of PHA-680632 and dasatinib treatment followed an aberrant mitosis (72.3% ± 6.3 for the combination versus 46.2% ± 19.6 for PHA-680632 (P < 0.0001), 9.5% ± 8.2 for vehicle, and 5.6% ± 9.6% for dasatinib) (Figure 4A). However, death was not an immediate consequence of mitosis, but occurred typically 10–13 hours later, with no difference in timing between treatment groups (Supplementary Figure S3A).
The SRC family kinases targeted by dasatinib are important for both cell attachment and control of mitotic cell rounding (Mustelin and Hunter, 2002; Yeatman, 2004). We evaluated the bright field images from the live cell imaging experiment and stratified the cells that underwent cell death as follows: 1) attached cells that did not undergo mitotic division, 2) cells that lost attachment and did not undergo division, 3) cells that entered mitosis with normal post-mitotic reattachment, and 4) cells that entered mitosis but displayed post-mitotic reattachment failure (Supplementary Figure S2). Strikingly, a large percentage of cells treated with PHA-680632 plus dasatinib that died failed to re-spread after mitosis, a death phenotype not significantly represented with either vehicle or individual drug treated cells (Figure 4B). The latency between initial cell rounding for mitosis and cell death was similar among all the treatment groups (Supplementary Figure S3B); rather, the drug combination selectively increased the percentage of the cell population undergoing this fate.
These data implied that the interaction between PHA-680632 and dasatinib depended on disruption of cell attachment. To test this idea, we compared the efficacy of this drug combination in adherent cells, versus cells growing in detached conditions. We found that cell death was not significantly elevated in suspension cells treated with the drug combination versus either alone (Figure 4C), supporting the interpretation that disruption of adhesive signals was an important component of the drug interaction.
We next assessed whether AURKA and SRC directly or functionally interact, so that inhibition of one might depress activation of the other, contributing to the observed synergy in growth inhibition. We determined that treatment of cells with dasatinib did not inhibit AURKA activity in OVCAR10 cells (Figure 5A). Reciprocally, PHA-680632 has no effect on SRC activity at the highest concentrations used in this study (Figure 5B, C). However, PHA-680632 significantly potentiated dasatinib in blocking SRC activation, as gauged by either Western blot (Figure 5B) or immunofluorescence assays (Supplementary Figure S4). In further analysis, we found that SRC and AURKA co-immunoprecipitated following co-expression of tagged constructs (Figure 5C). This interaction was similar to the levels seen with positive control NEDD9, supporting the idea of a potential interaction; however, to date we have not observed co-immunoprecipitation of endogenous SRC and AURKA.
To further probe AURKA and SRC interactions, we next performed an in vitro kinase assay with the two kinases (Figure 5D). The auto-phosphorylation seen with recombinant SRC alone and recombinant AURKA alone is blocked by dasatinib and PHA-680632, respectively. When SRC and AURKA are combined in the same kinase reaction, we detect a very large increase in phosphorylation of both SRC and AURKA, an effect that is only partially blocked by either PHA-680632 or dasatinib treatment. Interestingly, combination of SRC and AURKA induced significant phospho-tyrosine staining on AURKA (Figure 5D), indicative of SRC substrate specificity. In contrast, combination of SRC and AURKB did not increase in auto-phophosphorylation by SRC, and SRC did not tyrosine-phosphorylate AURKB, while only weakly inducing AURKB auto-phosphorylation (Figure 5E).
To further probe the specificity of SRC and Aurora kinase interactions, we examined induction of apoptosis in cells treated with dasatinib plus siRNAs targeting AURKA versus AURKB, or with PHA-680632 plus siRNA targeting SRC (Figure 5F). Depletion of AURKA and AURKB independently increased PARP and caspase-3 cleavage in conjunction with dasatinib. Although a greater absolute magnitude of PARP induction was observed with AURKB, this was on a background in which siRNA to AURKB itself significantly induced PARP: in contrast, siRNA to AURKA only induced apoptotic signaling when combined with dasatinib. Interestingly, in the context of dasatinib treatment, siRNA depletion of AURKB led to cross-depletion of AURKA, and inhibition of AURKA cross-depleted AURKB, again suggesting dialog between the dasatinib targets and these proteins. SiRNA to SRC in combination with PHA-680632 also led to greater co-induction of PARP, although not to the same extent as with the siAurora/dasatinib combinations. The lesser effect may be due to the presence of multiple other SRC family members, such as LYN, YES, and FYN, in ovarian cancer cells, which would be inhibited by dasatinib but not an siRNA; or by inhibition of an alternative dasatinib target.
We have here described a novel in vitro synergy between dasatinib and inhibitors of Aurora kinases in ovarian and colorectal cancer cell lines, but not in normal ovarian epithelial cells, and we have shown that multiple drugs that inhibit SRC family kinases and Aurora kinases have similar phenotypes. Treatment of cells with combined AURKA inhibitors and dasatinib resulted in a specific elimination of aneuploid cells after they have undergone defective mitosis and failed to reattach to substrate. SRC and AURKA directly interacted in vitro, and administration of PHA-680632 in conjunction with treatment with dasatinib potentiated reduction of SRC kinase activity.
Both dasatinib and Aurora inhibitors are showing promise in the clinic for treatment of some malignancies. For example, in chronic myeloid leukemia (CML), dasatinib treatment has resulted in high rates of complete cytogenic response, where tumor cells are no longer detected in the blood of patients, both after failure of imatinib treatment (Hochhaus et al., 2008) and as a first line treatment (Kantarjian et al., 2010), based on the activity of this compound against BCR-ABL. Phase I data showed that treatment with the pan-Aurora inhibitor PHA-739358 resulted in clinically relevant disease stabilization in some patients, and a phase II study recently demonstrated that complete hematologic response can be reached in two of twelve patients with CML (Boss et al., 2009). Our data support the combined use of such compounds may be worth exploring in the clinic. Interestingly, Fei and colleagues (Fei et al., 2010) reported that combined treatment of Philadelphia chromosome-positive human acute lymphoblastic leukemia (ALL) cells with a combination of the pan-Aurora inhibitor VX-680 and dasatinib resulted in a significant increase in cytotoxicity compared to the individual drugs alone. Since VX-680 also has been shown to inhibit the BCR-ABL oncogene based on off-target activity (Carter et al., 2005), the authors attribute synergy to potent inhibition of this oncogenic fusion protein. Our data suggest that inhibition of Aurora kinases and SRC may also contribute, particularly as Fei et al. noted a more significant reduction in phospho-SRC when ALL cells were treated with a combination of dasatinib and VX-680 as compared to the individual agents (Fei et al., 2010), although they did not pursue this observation. It will be informative to assess whether VX-680/dasatinib synergy is observed in solid tumors in which BCR-ABL is not a driving lesion. Interestingly, in previous work, we found dual inhibition of AURKA and the receptor tyrosine kinase EGFR significantly and selectively reduced the activity of SRC and other SRC-family kinases (FGR, HCK, LYN and LCK) (Astsaturov et al., 2010), further implying close signaling interactions between AURKA, SRC, and another important therapeutic target.
The idea of dually targeting SRC and Aurora inhibitors has potentially greater impact than suggesting new applications for dasatinib. Multiple SRC inhibitors are currently undergoing clinical trials. These include not only the ATP-binding site competitive inhibitors such as dasatinib and AZD0530, but also substrate-binding site inhibitors such as KXOI (Guarino, 2010; Kopetz et al., 2007). There have been some clinical successes with these agents. However, despite their promise, SRC inhibitors used as single agents in preclinical models and phase I trials have been shown to only affect proliferation in a very limited number of cases (Johnson et al., 2005; Jones et al., 2002), and do not effectively induce tumor regression based on radiographic analysis (Kopetz et al., 2007). It would be of interest to also evaluate these for interaction with Aurora inhibitors.
Mechanistically, the death phenotype of dually treated cells indicates a failure to re-adhere after prior defects in completion of mitosis is an important component of the resulting apoptosis. There are many links between cellular attachment and cell division. Defective cellular attachment results in multinucleation, and inhibition of cytokinesis in part because of failure of cells to generate adequate traction forces to drive separation (Ben-Ze’ev and Raz, 1981; Huang et al., 2005; Orly and Sato, 1979). AURKA and SRC have each separately been shown to influence the completion of cytokinesis. SRC localized to plasma membrane subdomains at the cleavage furrow becomes hyper-phosphorylated at cytokinesis, with SRC activation supporting cleavage furrow progression (Ng et al., 2005) and abscission (Kasahara et al., 2007). Likewise, microinjection of antibodies to inhibit AURKA at metaphase was shown to prevent cytokinesis completion (Marumoto et al., 2003). NEDD9, a physical interactor of both SRC and AURKA, relocalizes from focal adhesion complexes to the spindle asters during mitosis, and misexpression of NEDD9 leads to cytokinesis failure (Dadke et al., 2006; Pugacheva and Golemis, 2005). Loss of adhesion is ultimately associated with anoikis in epithelial cells (Frisch and Screaton, 2001), due to loss of survival signals mediated through integrins. Our data are compatible with the idea that given the close interaction of AURKA, SRC, and NEDD9, inhibition of two of these proteins simultaneously may exacerbate defects in both the cell division and cell survival processes they mediate.
The OVCAR10, OVCAR5, and OVCAR3 ovarian carcinoma cell lines and the HCT116 (K-RAS mutant, TP53 wild-type) colorectal carcinoma cell line were obtained from the ATCC (USA). The DLD-1 (K-RAS mutant, TP53 mutant) and DKS-8 (isogenic to DLD-1, but with the activated K-RAS allele disrupted [hence, K-RAS wild-type], TP53 mutant) colorectal cancer cell lines were a kind gift of Dr. Robert J. Coffey (Vanderbilt University, TN). Primary HOSE cells were isolated, characterized, and cultured as previously described (Bellacosa et al., 2010; Dyck et al., 1996). All ovarian cancer cell lines were maintained in RPMI supplemented with 10% (v/v) fetal bovine serum (FBS) and insulin, L-glutamine, penicillin, and streptomycin. All colorectal cancer cell lines were maintained in DMEM supplemented with 10% (v/v) FBS, L-glutamine, penicillin, and streptomycin. Suspension cell culture was performed in 6-well Ultra-low Attachment Surface plate (Corning Life Sciences, Pittston, PA).
PP2 and C1368 were obtained from Sigma-Aldrich (St. Louis, MO, USA); dasatinib was obtained from Bristol Myers Squibb (New York, NY); AZD1152-HQPA was obtained from Selleck Chemicals (Houston, TX); PHA-680632 was obtained from Nerviano Medical Sciences (Nerviano, Italy); MLN8237 was obtained from Millenium Pharmaceuticals (Greenwood Village, CO). The Flag-Src plasmids were a kind gift of Dr. Philip Stork (Oregon Health and Science University, OR). A PCR product of mouse Red Fluorescent Protein (mRFP) was ligated into pcDNA3.1(+) (Invitrogen, Carlsbad, CA, USA) to create pcDNA3.1-mRFP. Aurora-A was expressed from pcDNA3.1-mRFP vectors; pcDNA3.1-mRFP was used as a negative control. Flag-fused HEF1/NEDD9 was expressed from the vector pCatch-FLAG (O’Neill and Golemis, 2001). On-TARGETplus SMARTpool siRNA oligos (Dharmacon, Lafayette, CO) were made against the following sequences: AURKA, J-003545 (UCGAAGAGAGUUAUUCAUA, CGGUAGGCCUGAUUGGGUU, UUCUUAGACUGUAUGGUUA, and AAUAGGAACACGUGCUCUA); AURKB, J-003326 (CCAAACUGCUCAGGCAUAA, ACGCGGCACUUCACAAUUG, GCGCAGAGAGAUCGAAAUC, and CAGAAGAGCUGCACAUUUG); SRC, J-003175 (GCAGUUGUAUGCUGUGGUU, GCAGAGAACCCGAGAGGGA, CCAAGGGCCUCAACGUGAA, and GGGAGAACCUCUAGGCACA).
For each drug tested, initial IC50 curves were established in each cell line to calibrate dose range for subsequent in vitro synergy experiments. For subsequent analysis, we used the ratio that shows the most significant drop in viability in the combined drug treatment compared to the individual drug treatment. Cells were plated at 2,000 (ovarian cancer cell lines) to 3,000 (colorectal cancer cell lines) cells/well into 96 well plates. After 24 hours, vehicle (DMSO), individual drugs or drug combinations were added, followed by 72 hours incubation. Cellular viability measurements were performed using the CellTiter Blue assay (Promega, Fitchburg, WI, USA). The coefficient of interaction (CI) between pharmacological inhibitors was established by the Chou-Talalay method (Chou and Talalay, 1984) using CalcuSyn software (Biosoft, United Kingdom).
Cells growing in 60 mm plates were synchronized overnight with 2 mM thymidine. Three hours following release from the thymidine block, cells were treated with vehicle, PHA-680632, dasatinib, or the combination of the two drugs. Attached and floating cells were harvested, fixed in 70% ethanol, and stained with propidium iodide following standard protocols provided with the Guava Cell Cycle Reagent (Millipore, Billerica, MA). Cell cycle profiles were acquired using Guava flow cytometry at time periods noted after drug treatment.
Cells stably expressing green fluorescent protein linked histone H2B (GFP-H2B) were seeded into 6 well plates and 2 mM thymidine was added for 18h, synchronizing the cells in G1/S phase. Three hours following thymidine washout, cells were treated with vehicle (DMSO), or 500 nM PHA-680632, or 156 nM dasatinib, or 500 nM PHA-680632 plus 156 nM dasatinib. Cells were then placed into a heated chamber, and bright field and fluorescent images were taken every 5 mins for up to 48 hrs using a Nikon TE2000 microscope (Nikon, Tokyo, Japan) controlled by Metamorph software (Molecular Devices, Sunnyvale, CA, USA). Stacks of individual movies were built and analyzed manually for the indicated measurements. A minimum of 90 cells was counted for each movie and experiments were conducted at least three times. For montages, selected frames representing different cell morphologies were chosen.
Cell extracts were prepared using CellLytic-M Cell Lysis Reagent (Sigma-Aldrich, St. Louis, MO, USA) supplemented with the Halt™ phosphatase inhibitor cocktail (Thermo Scientific, Rockford, IL) and the Complete Mini™ protease inhibitor cocktail (Roche Diagnostics Gmbh, Manheim, Germany). Extracts were centrifuged at 15,000 × g for 15 min at 4°C. Immunoprecipitation assays were performed as described in (Obara et al., 2004). Agarose-immobilized Aurora-A antibody (Bethyl Laboratories, Montgomery, TX) was used for immunoprecipitation. Other antibodies used were directed against phospho-tyrosine, Aurora-B/AIM-1, and Aurora-A/IAK1 (BD Transduction Laboratories, San Jose, California, USA), GST (Santa Cruz Biotechnology, Santa Cruz, CA, USA), the Flag M2 epitope tag (Sigma-Aldrich, St. Louis, MO, USA), Red fluorescent protein (RFP) (US Biological, Marblehead, MA, USA), and the PARP 214/215 cleavage site (Millipore, Billerica, MA, USA). Secondary anti-mouse or anti-rabbit horseradish peroxidase coupled antibodies were purchased from GE Healthcare (Little Chaifont, Buckinghamshire, United Kingdom). All remaining antibodies used in Western blot experiments (to phospho-Y416 Src, Src, phospho-T288 Aurora-A, pan-phospho Aurora [recognizing phosphorylation on AURKA-T288/AURKB-T232/AURKC-T198], and β-actin), cleaved caspase 3 (D175) were purchased from Cell Signaling (Danvers, MA, USA). Phospho-Y416 Src antibody (Invitrogen, Carlsbad, CA, USA) and Alexa-Fluor anti-rabbit and anti-mouse secondary antibodies (Invitrogen, Carlsbad, CA, USA) were used for immunofluorescence.
In vitro kinase assays were performed using recombinant active Aurora A (Millipore, Billerica, MA, USA), recombinant Hexa-histidine baculovirus-produced Aurora A, recombinant active Aurora B (Millipore, Billerica, MA, USA), or recombinant active Src kinase (Cell Signaling, Danvers, MA, USA). Bacterially expressed GST-fused NEDD91–363 (produced in BL21 bacteria, as described in (Pugacheva and Golemis, 2005)) and GST protein were added to reactions as indicated. For drug inhibition experiments, recombinant proteins were pretreated with drugs on ice for 20 minutes before the start of the kinase reaction. In vitro kinase reaction was performed with γ-32P-ATP (Perkin-Elmer, Waltham, MA, USA) as in (Pugacheva and Golemis, 2005). All assays were done a minimum of 3 times and a representative blot shown in experiments.
For clonogenic assays, cells were plated at concentration of 1000 cells per 35mm dish, and after 24 hours treated with vehicle (DMSO) PHA-680632, dasatinib or drug combination for 24 hours. Subsequently, cells were cultured in fresh media without drugs for 10 to 12 days. Colonies were fixed with 4% paraformaldehyde (PFA)/PBS, and stained with 0.4% crystal violet/20% methanol. Colonies larger than 10 pixels in diameter (typically ~8 cells) were counted with Metamorph software. For 3D matrigel assays, 5 × 103 OVCAR10 cells were added to 40 μl of growth factor reduced Matrigel in 8-well Lab-Tek chamber slides (Nunc, Rochester, NY). After 4 days, drugs were added in fresh medium, and cell maintained for 8 days before fixation in 4% PFA, followed by immunofluorescence or staining with crystal violet.
A Chi-square test was used to compare cell counts for different phenotypic categories among the four drug treatment types in time-lapse microscopy experiments. The Chi-square test was also used to test the type of cell division that was induced by drug treatments. The time of entry into mitosis was compared across the treatment types via the log-rank test. The duration of mitosis was compared across the treatment types via the non-parametric Wilcoxon test. Overall equality of several proportions for fraction of cell deaths and types of cell deaths over the drug treatment groups were submitted to permutation tests. FACS data was analyzed using a two-tailed Student’s t-test. Clonogenic survival and matrigel assay data were analyzed using a one-tailed Student’s t-test.
FACS analysis of synchronized PEO1 cells treated with vehicle, 250 nM PHA-680632, 625 nM dasatinib, or the drug combination. Cell cycle profiles were evaluated at times indicated. Graph below represents quantification of cells with >4N DNA from at least 2 independent repetitions of these *p = 0.0393, **p = 0.0525.
Montage of representative images of cell death phenotypes from time-lapse microscopy experiments obtained from both fluorescent (GFP-H2B) and brightfield microscopy. Time stamps represent minutes elapsed since thymidine release of cells.
A. No significant differences between the drug treatment groups in time elapsed after completion of mitosis, defined as chromosomal decondensation, until chromosomal hypercondensation and fragmentation: (11.4 ± 6.0 hrs for vehicle, 10.7 ± 8.3 hrs for PHA-680632, 13.1 ± 6.4 for dasatinib, and 12.6 ± 7.9 for the PHA-680632 and dasatinib combination). Each symbol represents an individual cell and bars represent averages and standard deviations. Data represent the compendium of 3 time lapse-microscopy experiments. B. No significant differences between the drug treatment groups in latency from cell detachment from plate surface to cell death: (11.8 ± 8.0 hrs for vehicle, 13.4 ± 7.1 hrs for PHA-680632, 11.2 ± 7.9 hrs for dasatinib, and 13.7 ± 7.3 hrs for the PHA-680632 and dasatinib combination). Each symbol represents an individual cell and bars represent averages and standard deviations. Data represent the compendium of 3 time lapse-microscopy experiments.
Immunofluorescence visualization with pY416-Src targeted antibody (red) of thymidine-synchronized OVCAR10 cells treated for 3 hours with vehicle (DMSO), 500 nM PHA-680632, 32 or 75 nM dasatinib, or the combination of PHA-680632 and dasatinib. DAPI staining is indicated in blue.
Synergy between Aurora kinase inhibitors and Src inhibitors was tested by CellTiter Blue assay in 8 different ovarian cancer cell lines at 3 different drug ratios (molar ratios) and in 3 different colorectal cancer cell lines at 2 different drug ratios. A Coefficient of Interaction (CI) value of >1 indicates antagonism; CI = 1 indicates additive effects; CI of <0.9 indicates synergy; and CI of <0.5 indicates strong synergy. Molar ratios and dose ranges for drug combination studies were determined after analyzing the individual effects of Aurora kinase inhibitor or Src inhibitor on cell viability. Inhibitory concentration (IC) values for individual Aurora kinase inhibitor or Src inhibitor as well as the IC50 for the drug combination (Combo IC50) are shown; variation in viability cutoffs for IC measurements is a result of cell line specific differences in sensitivity to the individual inhibitors.
We thank Lisa Vanderveer for assistance with maintaining EOC cell lines; Carolyn Slater for isolating and producing the primary HOSE cultures; and Christine Klimowicz for assistance with the drug synergy assays. This work was supported by R01-CA63366 and R01-CA113342; DOD W81XWH-07-1-0676 from the Army Materiel Command; and Tobacco Settlement funding from the State of Pennsylvania (to EAG); R01-CA140323 and the Ovarian Cancer Research Fund (to AKG); Ovarian SPORE P50 CA083638 (Project 4, to EAG, AKG, and DCC); R01-GM86877, an appropriation from the Commonwealth of Pennsylvania, and the Greenberg Fund (to TJY); R01-CA136596 and R01-CA151374 (to DCC); R01-CA50633 (to LMW); and by NCI CA06927 and support from the Pew Charitable Fund (to Fox Chase Cancer Center). Additional funds were provided by Fox Chase Cancer Center via institutional support of the Head and Neck Cancer Keystone Program.