Aurora B kinase phosphorylates and instigates degradation of p53

Abstract

Aurora B is a mitotic checkpoint kinase that plays a pivotal role in the cell cycle, ensuring correct chromosome segregation and normal progression through mitosis. Aurora B is overexpressed in many types of human cancers, which has made it an attractive target for cancer therapies. Tumor suppressor p53 is a genome guardian and important negative regulator of the cell cycle. Whether Aurora B and p53 are coordinately regulated during the cell cycle is not known. We report that Aurora B directly interacts with p53 at different subcellular localizations and during different phases of the cell cycle (for instance, at the nucleus in interphase and the centromeres in prometaphase of mitosis). We show that Aurora B phosphorylates p53 at S183, T211, and S215 to accelerate the degradation of p53 through the polyubiquitination–proteasome pathway, thus functionally suppressing the expression of p53 target genes involved in cell cycle inhibition and apoptosis (e.g., p21 and PUMA). Pharmacologic inhibition of Aurora B in cancer cells with WT p53 increased p53 protein level and expression of p53 target genes to inhibit tumor growth. Together, these results define a mechanism of p53 inactivation during the cell cycle and imply that oncogenic hyperactivation or overexpression of Aurora B may compromise the tumor suppressor function of p53. We have elucidated the antineoplastic mechanism for Aurora B kinase inhibitors in cancer cells with WT p53.

Aurora kinases are serine/threonine kinases essential for cell cycle control and mitosis (1–7). Mammals have three Aurora kinase family members (A, B and C), and these kinases are expressed at maximum levels during mitosis. Although all three kinases regulate mitosis, Aurora A and Aurora B differ in subcellular localization, and each kinase performs a distinct task (1). Aurora A is located at the centrosomes at prophase, and as mitosis progresses, it is located at the spindle poles during prometaphase and metaphase. In contrast, Aurora B, part of the chromosome passenger complex (CPC), is located on the chromosome arms during prophase and at the centromeres during prometaphase and metaphase (1). Aurora B subsequently localizes to the midbody during cytokinesis. A recent report showed that Aurora C behaves like Aurora B and is also a chromosome passenger protein (8). These different subcellular localizations suggest that Aurora kinases may recruit different substrates to regulate different processes during mitosis. Whereas many Aurora A substrates have been characterized (2), few substrates for Aurora B have been identified.

Tumor suppressor gene p53 is a guardian of the genome and an important negative regulator of the cell cycle; p53 delays or arrests cell cycling at DNA damage checkpoints preceding DNA replication (the G1/S checkpoint) as well as inhibits damaged cells from entering mitosis (the G2/M checkpoint) (9). However, it is not clear whether and when p53 has function during mitosis. A recent report shows that cells lacking p53 function lose spindle assembly checkpoint control (10), which suggests a role for p53 in mitosis. How the information on the status of the spindle assembly is relayed to p53 is, however, not known. Importantly, activity of tumor suppressor p53 is lost in 50% of human cancers by mutation, deletion of the p53 gene, or loss of cell signaling upstream or downstream of p53 (11)_.

Aurora B has been shown to be overexpressed in many types of cancers, including multiple myeloma, colorectal, prostate, and pancreatic cancers (12). This overexpression has made Aurora B an attractive target for therapeutic cancer drugs (13). Although high levels of Aurora B are associated with advanced clinical stage and poor prognosis in several cancers, the functional significance of aberrant overexpression of Aurora B remains unclear. Many small-molecule inhibitors of Aurora kinases are being investigated as potential drugs for cancer therapy (13) in translational studies and early-phase clinical trials (12). These small molecules can inhibit autophosphorylation of Aurora B and phosphorylation of histone H3 on Ser10 (7), but it is not clear how these small molecules can affect the uncharacterized functions of Aurora B. It is also not known whether Aurora B and p53 are coordinately regulated during the cell cycle. Elucidating the mechanisms that coordinate Aurora B and p53 would help us to better understand the regulation of p53 during mitosis and may lead to the development of more effective or more individualized cancer therapies targeting Aurora B. Thus, we sought to investigate these mechanisms by investigating the interaction and any mutual regulation of Aurora B and p53 using various cell cycle synchronizations, biochemical interaction techniques, and animal models. Our findings show that p53 is a substrate of Aurora B and that Aurora B directly interacts with p53 at the CPC during mitosis. We also indicate that Aurora B negatively regulates p53 and that a specific Aurora B inhibitor can antagonize this impact. Our studies provide a mechanism-based application for effective cancer therapy when using Aurora B kinase inhibitors in cancers with WT p53 status.

Results

Aurora B Associates with p53.

Previously, we discovered that the specific Aurora B kinase inhibitor AZD1152 (13–15) reduced the steady-state protein level of Aurora B (16). We also reported that AZD1152 reduced the protein level of Aurora B and concurrently elevated the protein level of p53 in cancer cells. From this finding, we hypothesized that Aurora B and p53 have an interactive and/or regulatory relationship. Indeed, coimmunoprecipitation (co-IP) experiments showed their association in intact cells and in vitro (Fig. 1 A and B). This association was shown to be independent of the presence of Novel inhibitor of histone acetyltransferase (INHAT) Repressor (NIR), which was previously shown to associate with both Aurora B and p53 (Fig. S1A) (17). Next, we mapped the structural regions of p53 required for the association with Aurora B. The results showed that Aurora B was bound to the C terminus of p53 (amino acids 160–393) but not the N terminus (amino acids 1–160) (Fig. 1C and Fig. S1 B and C). To examine whether the interaction was cell cycle regulated, we collected cell lysates from synchronized HCT-116 cells at various time points after release from double thymidine block (Fig. 1D). Aurora B protein was detected in every cell cycle phase, but it was reduced after mitosis (Fig. 1E). Co-IP of Aurora B and p53 showed that they interacted during most of the cell cycle but not in the late M phase when Aurora B is degraded (Fig. 1E). These results suggest that Aurora B may have an uncharacterized function in interphase that involves interaction with p53. Phosphorylation of H3 (HH3), a substrate of Aurora B (18, 19), was elevated during mitosis and then diminished after the down-regulation of Aurora B after mitosis. The interaction pattern between p53 and Aurora B did not follow the temporal pattern of HH3 phosphorylation, which suggests that p53 was not involved in the regulation of phosphorylation of HH3 by Aurora B (Fig. 1E). Noticeably, the expression levels of p53 and Aurora B seemed inversely correlated during the cell cycle (Fig. 1E).

Fig. 1.

Aurora B associates with p53 at various phases of the cell cycle. (A) Lysates of U2OS cells were IP with either anti-Aurora B (AB2254; Abcam) or anti-p53 antibodies (AB-2) or preimmune IgG (negative control) followed by immunoblotting (IB) with antibodies as labeled to show co-IP of Aurora B and p53. (B) GST pull-down assay was performed with combinations of in vitro translated Aurora B, GST, and GST-tagged p53 as labeled. Aurora B that was bound to GST-p53 was detected by IB. Coomassie staining of GST and GST-p53 inputs are in C. In vitro translated Aurora B was detected by immunoblot (Lower). (C) As presented in B, results of in vitro GST pull-down assay of immunopurified Flag-Aurora B with GST or GST-tagged p53 deletion constructs are shown. The asterisks indicate the stained bands of GST, GST-p53, and GST-p53 deletion mutants. (D) Hct116 cells were synchronized to S phase by double thymidine block. Cell samples at labeled time points after release of thymidine block were stained with PI and analyzed by FACS for DNA content. DNA content histograms are shown for the time points as labeled. (E) Lysates of synchronized Hct116 cells from D were analyzed by immunoblot with indicated antibodies. Aurora B–p53 interaction at various phases of the cell cycle (as labeled above) was detected by IP with anti-Aurora B antibody followed by IB for p53 and Aurora B (IP:AurB and IB:AurB).

Aurora B Interacts with p53 During both Interphase and Mitosis.

To investigate whether the Aurora B–p53 interaction could occur subcellularly during mitosis, we used immunofluorescence microscopy to visualize colocalization of endogenous p53 and Aurora B. The images indicated that these two proteins colocalized at the midzone in anaphase and telophase (Fig. 2A). To prove direct molecular interaction between Aurora B and p53 in whole cells, we used the method of bimolecular fluorescent complementation (BiFC) (Fig. 2B). Cells were cotransfected with a plasmid expressing the chimera of Aurora B and the C-terminal portion of Venus (an enhanced YFP) and a plasmid expressing the chimera of p53 and the N-terminal portion of Venus (20) (Fig. 2B). Consistent with the co-IP results from synchronized cells (Fig. 1E), fluorescent microscopy of MCF7 cells transfected with such plasmids showed direct intermolecular interaction between Aurora B and p53 (pseudocolored green), whereas the control plasmids (Venus-N-Term and Venus-C-Term) did not produce fluorescence (Fig. 2C). The fluorescence can be observed in prometaphase and interphase (Fig. 2C). Synchronization study revealed that p53 and Aurora B colocalized during different phases of mitosis (prometaphase, metaphase, anaphase, and telophase) (Fig. S1D). Close examination of mitotic cells revealed that the Aurora B–p53 interaction (BiFC signal) was located in the middle part of each chromosome at prometaphase (Fig. 2D, arrows). Given that Aurora B is located at the centromeres (21–23), it is likely at this location that the p53–Aurora B interaction occurs. To verify this location, we cotransfected cells with Venus C-term-AurB and Venus N-Term-p53 and then immunostained for the CPC member Survivin (24), which is dynamically localized at the centromeres to activate Aurora B (22, 25). We observed that Aurora B, p53, and Survivin all colocalized to the DNA of prometaphase cells, suggesting that Aurora B and p53 interact at the centromeres (Fig. 2E). Furthermore, we cotransfected Venus plasmids (as previously described) into synchronized cells and immunostained for the centromere marker CENP-A. We observed that the Venus interaction of Aurora B and p53 occurred at distinct points that also stained clearly for CENP-A (Fig. 2F). Similar results were obtained with other cell lines: BiFC was observed in interphase U2OS cells (Fig. S1E) and also, at a distinct foci on each chromosome in prometaphase 293T cells (Fig. S2A). These data indicate that the interaction between Aurora B and p53 occurs at different subcellular locations (e.g., interphase nuclei and the centromeres) and during different phases of the cell cycle.

Fig. 2.

Aurora B colocalizes and directly interacts with p53 in interphase and mitosis. (A) MCF7-Her18 cells were stained with DAPI, mouse anti-p53, and rabbit anti-Aurora B antibodies. Confocal immunofluorescence images of cells in anaphase and telophase are shown. (B) Schematic diagram to explain the use of Venus fusion proteins for BiFC. (C) Plasmids containing the N terminus of Venus fluorescent protein fused to p53 (Venus N-term-p53) and the C terminus of Venus fused to Aurora B (Venus C-term-AurB) were cotransfected in MCF7. Cells were synchronized to prophase by thymidine–nocodazole block. Venus fluorescence is pseudocolored green. (D) Deconvolved fluorescent micrograph of a synchronized MCF7 nucleus from C is shown at a high magnification. Arrows indicate p53 and Aurora B direct intermolecular interaction (BiFC) at the centromeres. (E) 293T cells transfected with Venus C-term-AurB and Venus N-Term-p53 and immunostained for the CPC member Survivin (red). The merged image shows that BiFC (pseudocolored green) colocalizes with at least some of the Survivin immunofluorescence (red). (F) MCF7 cells were transfected with Venus C-term-AurB and Venus N-term-p53 and then treated with thymidine–nocodazole for synchronization. Deconvolved fluorescent micrograph of a synchronized MCF7 nucleus showing colocalization of Venus (Aurora B–p53 interaction) and the centromere marker CENP-A (red).

Aurora B Phosphorylates p53.

After establishing that Aurora B and p53 interact directly, we investigated whether p53 is a kinase substrate for Aurora B. Using an in vitro kinase assay (Fig. 3A), we found that purified recombinant GST-p53 was phosphorylated by Aurora B. As negative controls, Flag-tagged Aurora B K106R (kinase dead) or Aurora B (WT) were immunopurified from transfected cell lysates and assayed for their ability to phosphorylate recombinant GST-p53. Aurora B WT, but not Aurora B K106R, efficiently phosphorylated p53 (Fig. 3B), confirming that phosphorylation of p53 in these reactions required the presence of the functioning kinase domain of Aurora B. The specific Aurora B kinase inhibitor AZD1152-hydroxyquinazoline pyrazole anilide (HQPA) (active form) blocked Aurora B-mediated p53 phosphorylation in a dose-dependent manner in vitro (Fig. 3C). Taken together, these results suggest that Aurora B has kinase activity to p53.

Fig. 3.

Aurora B phosphorylates p53 at multiple serine/threonine residues in the DNA binding domain. (A) Flag-Aurora B was immunoprecipitated with anti-Flag antibody from lysates of 293T cells transfected with Flag-Aurora B or vector (negative control). The anti-Aurora B immunoblot is shown in A Middle. Coomassie blue-stained SDS/PAGE gel of GST and GST-p53 is shown (A Bottom). Phosphorylation resulting from Flag-Aurora B catalyzed in vitro kinase reactions using GST or GST-tagged p53 as the substrate was detected by autoradiography (A Top). (B) Flag-Aurora B or Flag-Aurora B K106R was expressed in 293T cells and immunoprecipitated to catalyze in vitro kinase reactions as in A. Phosphorylation results are shown in a similar manner to A. (C) Recombinant GST-tagged Aurora B was used in an in vitro kinase assay with GST or GST-p53 as the substrate in the presence of an increasing dose of the specific Aurora B inhibitor AZD1152-HQPA. Phosphorylation results are shown in a similar manner to A. (D) Recombinant GST-Aurora B was used in an in vitro kinase assay as before with GST-tagged deletion mutants of p53. Phosphorylation results are shown in a similar manner to A. *Phosphorylated GST-p53 (amino acids 160–393) fragment. (E) Consensus phosphorylation sequence and Netphos 2.0 scores for potential Aurora B phosphorylation sites in human p53. (F) In vitro kinase assay with recombinant GST-Aurora B and various GST-p53 DNA binding domain mutant substrates. Ratio of phosphorylation relative to control is indicated above each lane.

To further investigate Aurora B-mediated phosphorylation of p53, we used different p53 fragments in another series of in vitro kinase assays. We found that Aurora B specifically phosphorylated the fragment of p53 containing the DNA binding domain (amino acids 160–393) but not the fragment containing the transactivating domain (amino acids 1–160) or the C-terminal fragment (amino acids 320–393) (Fig. 3D). To determine the p53 phosphorylation sites, we scanned the p53 sequence using the NetPhos algorithm (http://www.cbs.dtu.dk/services/NetPhos/) and identified five potential Aurora B phosphorylation sites (S183, T211, S215, S269, and T284) that were conserved across different species and shared the Aurora B phosphorylation site consensus motif (Fig. 3E and Fig. S2B). In vitro kinase assays using alanine mutants of potential phosphorylation sites indicated that several sites (S183, T211, and S215) were phosphorylated by Aurora B (Fig. 3F).

Aurora B Enhances p53 Ubiquitination.

Because the stability of p53 is regulated by its phosphorylation (11), we hypothesized that p53 phosphorylation by Aurora B accelerated p53 protein turnover by polyubiquitination and subsequent proteasomal degradation. We found that expression of Aurora B reduced the steady-state protein level of p53 in a dose-dependent manner (Fig. 4A). Furthermore, in the presence of the de novo protein synthesis inhibitor cycloheximide, Aurora B overexpression accelerated the turnover rate of p53 (Fig. 4B). Consistently, Aurora B knockdown using siRNA reduced the turnover rate of p53 (Fig. 4C and Fig. S2C). Real-time quantitative PCR (qPCR) analysis showed that p53 mRNA levels were not affected by Aurora B overexpression (Fig. S3B), suggesting that Aurora B down-regulated p53 at the posttranscriptional level. We also found that Aurora B-mediated p53 down-regulation was antagonized by the proteasome inhibitor MG341 (bortezomib), indicating the involvement of ubiquitin–proteasome degradation in this process (Fig. 4D). The murine double minute 2 (MDM2) is the E3 ubiquitin ligase for p53, and we found that Aurora B kinase potentiated MDM2-mediated p53 ubiquitination (Fig. 4E). Importantly, AZD1152-HQPA, which specifically inhibits the kinase activity of Aurora B, was able to block p53 ubiquitination in a dose-dependent manner (Fig. 4F). To investigate the role of Aurora B-mediated p53 phosphorylation in regulating p53 stability, we constructed a p53 triple phosphorylation site mutant, S183A/T211A/S215A → (AAA), and we found that, compared with the turnover rate of WT p53, the turnover rate of this p53 AAA mutant was not influenced by Aurora B (Fig. 4G). Together, these results indicate that Aurora B kinase plays a role in enhancing the degradation of p53 through the ubiquitination–proteasome pathway, therefore reducing p53 steady state in the cell.

Fig. 4.

Aurora B phosphorylates p53 at multiple sites resulting in polyubiquitination and degradation by the proteasome. (A) H1299 cells were transfected with a fixed amount of plasmid expressing GFP-p53 and an increasing amount of plasmid expressing Flag-Aurora B. Immunoblots for GFP-p53, Flag-Aurora B, and Actin are shown. (B) U2OS cells were transfected with Flag-Aurora B or control vector and then exposed to cycloheximide followed by immunoblot for p53 turnover. (C) 293T cells were infected with lentivirus expressing shRNA 468 AurB or control lentivirus (shRNA luc). Infected cells were exposed to cycloheximide and then analyzed by immunoblot for p53 turnover. The asterisk (*) represents nonspecific band. (D) H1299 cells were cotransfected with GFP-p53 and Flag-Aurora B in the presence and absence of the proteasome inhibitor MG341. (E) 293T cells were transfected as indicated and immunoprecipitated with anti-HA antibody followed by immunoblot with anti-GFP antibody to detect ubiquitinated p53. (F) 293T cells were transfected as indicated, treated for 24 h before harvest with AZD1152-HQPA, and immunoprecipitated with anti-HA antibody followed by immunoblot with anti-p53 antibody to detect polyubiquitinated p53. (G) H1299 cells were transfected with plasmids as indicated and treated with cycloheximide to evaluate the turnover of GFP-tagged p53 or GFP-p53 AAA. Relative remaining GFP-p53 expression value is indicated in a line graph.

Aurora B Affects p53 Transcriptional Activity.

Because Aurora B was shown to down-regulate p53 protein level, we examined the impact of Aurora B on the transcriptional activity of p53 using a p53-responsive luciferase reporter gene assay. We observed that Aurora B expression impaired p53 transcriptional activity (Fig. 5A), whereas Aurora B knockdown enhanced p53 transcriptional activity (Fig. 5B). A control experiment using a p53-independent promoter is shown in Fig. S3A. Next, we examined the expression of p53 target genes in Aurora B-overexpressing cells. Our results showed that expressions of p53 transcriptional targets MDM2 and p21 were lower in Aurora B-overexpressing cells than control cells (Fig. 5C and Fig. S3B). Because p53 is important in inducing apoptosis genes to regulate cell death under genotoxic stress, we used detection of sub-G1 DNA content in flow cytometry to analyze the impact of Aurora B on genotoxic stress-induced apoptosis. Compared with findings in control cells that did not overexpress Aurora B, Aurora B overexpression decreased apoptosis caused by the DNA-damaging agent cisplatin (CDDP) (Fig. 5D and Figs. S3C and S4A). Additional analysis of the functional effect of Aurora B-mediated p53 phosphorylation on transcriptional activity was carried out using p53 phosphorylation mutants S183A, T211A, S215A, S183A/T211A, or S183A/T211A/S215A (the AAA mutant) in a p53-responsive luciferase reporter gene assay. T211A and AAA mutants were resistant to Aurora B-mediated repression of p53 transcription, whereas S183A and S215A as well as S183A/S215A mutants remained sensitive to the presence of Aurora B (Fig. 5E). A Western blot of GFP-p53 construct expression levels is shown in Fig. S4B. In agreement, we found that, compared with WT p53, the p53 AAA mutant was resistant to Aurora B-mediated repression of p21, a p53 target gene (Fig. 5F). Therefore, T211 seems to be the most influential site phosphorylated by Aurora B in terms of p53 transcriptional activity. Because the p53 AAA mutant was resistant to the Aurora B-mediated increase of turnover (Fig. 4G), this mutant was, thus, resistant to Aurora B-mediated repression of its transcriptional activity. Taken together, these results suggest that overexpression of Aurora B led to phosphorylation of p53 at multiple sites, which increased p53 turnover, resulting in the repression of p53-mediated transcriptional activity.

Fig. 5.

Phosphorylation of p53 by Aurora B inhibits p53 transcriptional activity. (A) 293T cells were transfected with fixed amounts of plasmids expressing GFP-p53, p53-Luciferase reporter plasmid (BDS2-3x-luc containing a p53-responsive element), and increasing doses of plasmid expressing GFP-Aurora B. p53 transcriptional activity was measured by luciferase activity. All of the error bars in the bar charts of this figure represent 95% confidence intervals. (B) p53-Luciferase reporter assay (BDS2-3x-luc) in 293T cells transfected with increasing dose of shRNA 468-AurB plasmid. (C) qRT-PCR analysis was performed to measure p53 and p53 target gene mRNAs in U2OS transfected with vector or Flag-Aurora B. (D) Hct116 cells were transfected with Flag-Aurora B, treated with cisplatin (CDDP), and analyzed by FACS after PI staining. Percentage of apoptotic cells (sub-G1 fraction) was plotted. *P < 0.05 compared with CDDP alone by one-way ANOVA posthoc intergroup comparison by Tukey test. Analysis of cleaved PARP and Caspase 3 are shown in Fig. S3C. (E) H1299 cells were transfected with indicated GFP-p53 phosphorylation site mutants and p53-luciferase reporter plasmid (MDM2-luc containing a p53-responsive element) in the absence (vector) or presence of Flag-Aurora B. Relative luciferase activity of each p53 construct in the presence or absence of Aurora B was measured. Error bars represent 95% confidence intervals. (F) qRT-PCR analysis was used to measure p21 mRNA in Hct116 p53^−/− cells transfected with Aurora B and either GFP-p53 or GFP-p53 AAA mutant. *P < 0.05 by one-way ANOVA posthoc intergroup comparison by Tukey test.

Aurora B Kinase Inhibitor Potentiates p53 Stabilization and p53 Transcriptional Activity in Vivo.

To evaluate the in vivo relationship between Aurora B and p53, we used Aurora B-specific inhibitor AZD1152-HQPA against aggressive human breast cancer MCF7-Her18 cells (WT p53). Our results showed that, after AZD1152-HQPA exposure, p53 levels in these cells increased in a dose-responsive fashion concurrent with the increase of p53 target genes (MDM2 and p21) (Fig. S4C), whereas phosphorylation of H3 (substrate of Aurora B) showed a decrease (Fig. 6A). In agreement with these findings, immunoblots for p53 in tumor samples from xenografted MCF7-Her18 cells (16) showed that, after a dose of AZD1152, p53 levels increased compared with levels in vehicle-treated mice (Fig. 6B). Accordingly, MDM2, Bax, and p53 up-regulated modulator of apoptosis (PUMA) levels in these tumor samples increased compared with levels in controls. Immunohistochemical staining for p53 in these tumor samples indicated that both low- and high-dose AZD1152 treatments increased the number of p53-positive cells and increased p53 staining intensity compared with the respective values in controls (Fig. 6C). The p53 expression level determined on the basis of staining intensity was quantitated by ACIS III and image analysis software (Dako) and presented as a bar graph (Fig. 6D). In addition, qRT-PCR indicated that AZD1152-treated tumors had higher mRNA levels of p53 target genes, including PUMA and p21, than control tumors (Fig. 6E). These data suggest that inhibition of Aurora B in vivo increased p53 protein level, which in turn, increased target gene expression to inhibit tumor growth (∼50% reduction in tumor weight) (16).

Fig. 6.

Inhibition of Aurora B induces p53 in a breast cancer xenograft model. (A) MCF7-Her18 cells were treated with increasing concentrations of AZD1152-HQPA as labeled for 48 h. Antigens for the immunoblots shown are labeled on the left. Mdm2 is indicated as full length (90 kDa) and cleavage product (60 kDa). (B) MCF7-Her18 xenografts from nude mice (indicated by mouse number) treated with AZD1152 were immunoblotted for p53 and p53 target genes. The protein expression of Bax, Puma, and Actin was quantified from Western blot films using Image J program. Bax/Actin and Puma/Actin ratios were calculated. Bar graphs and one-way ANOVA statistic analyses with Turkey test were done with GraphPad Prism 5.0c. (C) Representative photomicrographs are shown for immunohistochemical staining of p53 in MCF7-Her18 xenograft tumors treated with AZD1152. (D) Percentage p53-positive (Upper) and average p53 immunostaining intensity (Lower) from automated quantitative image analysis of immunohistochemical staining of MCF7-Her18 nude mouse xenografts treated with AZD1152. *P < 0.05 compared with vehicle control by one-way ANOVA posthoc comparison by Tukey test. Error bars represent 95% confidence intervals. (E) qRT-PCR analysis was used to measure p53 and p53 target gene mRNA in MCF7-Her18 cells treated with 20 nM AZD1152-HQPA for 48 h. Error bars represent 95% confidence intervals. (F) Model of p53 regulation by Aurora B.

Discussion

As an important tumor suppressor, p53 needs to keep many oncogenic signals or gene mutation events at bay. For example, DNA damage will activate ataxia telangiectasia-mutated (26, 27) or Chk (28) kinase to modify and/or strengthen the stability and activity of p53 for initiating DNA repair or causing apoptosis. However, during tumorigenesis, p53 is also a target of many oncogenic signals, such as Akt (29, 30), MDM2 (31, 32), COP1 (33–35), or CSN6 (36), and these oncogenic signals will decrease the protein level of p53. Aurora B is a protein overexpressed in many types of cancer. Here, we report a critical role for Aurora B in controlling p53 homeostasis by regulating its phosphorylation and subsequent ubiquitin–proteasome degradation. Our results provide a mechanism to explain the role of Aurora B expression in carcinogenesis and cancer progression. This layer of p53 posttranslational regulation also provides an example of an interphase function for mitotic kinase Aurora B. Our data indicate that p53 and Aurora B directly interact during interphase (Fig. 1E). This finding suggests a functional role for Aurora B outside mitosis. It is important to point out that Aurora A, another mitotic kinase, was shown by other investigators to phosphorylate p53 at Ser315 (37), which is not a consensus site for Aurora B and can also cause p53 destabilization (37). However, the phase of the cell cycle on which the regulation of p53 by Aurora A occurs remains to be determined. Our data show that Aurora B was able to regulate p53 stability during interphase, thereby decreasing p53 transcriptional activity, facilitating cell cycle progression, and antagonizing apoptosis.

The significance of Aurora B-regulated p53 destabilization during interphase could serve to drive the cell cycle through the following mechanisms by (i) blocking p53-mediated expression of p21 (38, 39), an important cyclin-dependent kinase inhibitor that blocks G1/S phase progression, and (ii) alleviating p53-mediated suppression of Cyclin B/Cdc2 and Survivin (40–44). These effects will aid cell cycle progression through interphase to enter mitosis.

As for the role of Aurora B–p53 interaction during mitosis, it has been shown that general suppression of transcription is observed during mitosis (45, 46); therefore, Aurora B-mediated p53 transcriptional suppression will not play a role during mitosis. The work by Cross et al. (10) showed that p53 is involved in facilitating chromosome segregation to ensure the maintenance of diploid cells, because p53 deficiency leads to tetraploidy in vivo. It is possible that Aurora B coordinates with p53 to mediate the spindle checkpoint (10) and aid progression through mitosis. Given that Aurora B deregulation also results in polyploidy, the interplay between p53 and Aurora B could conceivably be important for spindle checkpoint; however, this issue remains to be investigated. Nevertheless, our data fill an important gap in the knowledge regarding the localization of p53 during different stages of mitosis—p53 associates with the CPC (Fig. 2E) and is located at the centromeres (Fig. 2F) at prometaphase (Fig. 2D) or the mid-zone/cleavage furrow at anaphase/telophase (Fig. 2A). The functional significance of Aurora B–p53 interaction during different stages of mitosis also remains to be investigated, but p53 is definitely involved in the spindle checkpoint (10), and our data serve to confirm the presence of this tumor suppressor in the spindle checkpoint machinery. It is important to point out that the binding between Aurora B and p53 decreases after the end of mitosis because of the down-regulation of Aurora B by anaphase promoting complex (47). After mitosis, Aurora B is recovered from degradation, and binds again to p53 (Fig. 1E) and thus, potentially preventing p53 from arresting the cell cycle at G1 (48) or causing cell death by maintaining negative phosphoregulatory control of p53. Future experiments will focus on dissecting the role of p53 within the CPC complex and the role of Aurora B-mediated p53 phosphorylation in this context. Given that overexpression or depletion of p53 will disturb the cell cycle, investigating this regulation during the mitosis window has major technical challenges. Nevertheless, our observations open a research avenue for additional studies in this area.

On the basis of our biochemical studies, we propose the model that Aurora B phosphorylates p53, leading to p53 ubiquitination/degradation and loss of p53 transcriptional activity as well as p53-mediated cell cycle suppression and apoptosis (Fig. 6F). Conversely, inhibition of Aurora B can reverse this effect and is sufficient to inhibit tumor growth. These findings indicate that Aurora B negatively regulates p53 and that Aurora B inhibitors may have value as effective therapeutic agents against cancers with WT p53 status.

Materials and Methods

Tissue Culture.

MCF-Her18 cells, a stably transfected subline of MCF7 that overexpresses Her2, have been described previously (49). These cells were cultured in DME/F12 (Sigma) supplemented with either 5% or 10% (vol/vol) FBS (Gemini). U2OS, 293T, H1299, MCF7, and HeLa cells were obtained from the ATCC and maintained in DME/F12 media (in-house supplier) supplemented with 5% or 10% (vol/vol) FBS. Hct116 p53^+/+ and Hct116 p53^−/− (gifts from Bert Vogelstein, The Johns Hopkins University, Baltimore, MD) (39) were cultured in McCoy’s 5A media (HyClone) supplemented with 5% or 10% (vol/vol) FBS, 2 mM l-glutamine (Cellgro), and 1% antibiotic–antimycotic solution (Invitrogen). All cells were incubated in a humidified incubator at 37 °C with 5% (vol/vol) CO₂.

Drugs and Reagents.

AZD1152 (orally bioavailable prodrug) and AZD1152-HQPA (active specific inhibitor of Aurora B kinase) were provided by Kirsten Mundt (Astra Zeneca, Cheshire, United Kingdom) (50). AZD1152-HQPA was dissolved in 100% DMSO at a 10 mM concentration as a stock solution before further dilution in appropriate aqueous solutions/media at final concentrations as indicated. De novo protein synthesis inhibitor cycloheximide and proteasome inhibitor MG132 were obtained from Sigma and used at 100 and 5 μg/mL, respectively. MG341 (bortezomib) was purchased from Millenium Laboratories and used at 70 ng/mL overnight (8–12 h). Cisplatin was purchased from Bedford and used as indicated.

Immunofluorescence.

Indicated cell lines were grown on chamber slides, tissue culture dishes, or cover glasses to 50–75% confluence. Cells were treated with vehicle (20 nM AZD1152-HQPA), treated with indicated transfection, or not treated for 48 h. Cells were rinsed two times with cold PBS and then fixed with 3% (vol/vol) paraformaldehyde (Electron Microscopy Sciences) solution for 15 min at room temperature. Cells were then rinsed three times with cold PBS for 10 min, and each rinse was followed by permeabilization with 0.2% Triton X-100 (Sigma). Plates were blocked in either 5% BSA (Sigma) or 5% (vol/vol) normal goat serum (a gift from Elsa Flores, The University of Texas M.D. Anderson Cancer Center, Houston, TX) diluted in PBS for 1 h. Fixed and blocked cells were stained with antibodies diluted appropriately in PBS and applied for a period of 1 h to overnight. Cells were stained with DAPI and mounted on microscope slides using Fluoromount G (Southern Biotech) or Prolong Gold (Invitrogen) that contained DAPI. Immunostained cells were visualized with an Olympus IX81 confocal microscope, an Olympus IX70 fluorescent microscope, a PerkinElmer Ultraview ERS spinning disk confocal microscope, or a Nikon Ti with a Photometric CoolSnap HQ2 camera driven by Nikon Elements software. Deconvolved images were processed using AutoQuant ×2 software (Media Cybernetics).

Immunoblotting and Immunoprecipitation.

Cells for Western blot or immunoprecipitation were collected from tissue culture dishes after two rinses with cold PBS. Cells were centrifuged at low speed for 10 min, and supernatants were discarded. Pellets were then either frozen at −80 °C for additional processing later or lysed with 100–300 μL 1× lysis buffer [0.5-L batch: 7.5 g 1 M Tris (Fisher), 15 mL 5 M NaCl (Fisher), 0.5 mL Nonidet P-40 (USB Corp.), 0.5 mL Triton X-100 (Sigma), 1 mL 0.5 M EDTA (Fisher)] for 20 min at 4 °C. Lysis buffer also contained a mixture of protease/phosphatase inhibitors: 5 mM NaV, 1 mM NaF, 1 μM DTT, 0.1 mg/mL Pepstatin A, 1 mM PMSF, and 1,000× Complete Mixture Protease Inhibitor (Roche). Lysates were centrifuged at high speed for 10 min, and cell debris was discarded. Protein concentration was measured using the Bradford method with protein assay reagent (Biorad) and read on a Powerwave XS (Biotek) spectrophotometer at 595 nm. Protein samples were standardized and mixed with 5× loading dye [100-mL batch: 3.78 g Tris base, 5 g SDS, 25 g sucrose (Sigma), 0.04 g bromophenol blue (Sigma), pH adjusted to 6.8], and they were boiled 5 min before SDS gel analysis. SDS/PAGE was performed according to standard procedures. All gels were 10% (vol/vol) polyacrilamide except MDM2 (8%), p-HH3 (15%), Bax (15%), oligomerization of p53 (6%), p53 ubiquitination (6%), and p21 (15%). Transfer of proteins was performed using 1× transfer buffer [1-L batch: 30.3 g Tris base, 144 g glycine (Fisher), 10 g SDS, pH adjusted to 8.3] to PVDF membrane (Millipore). For immunoprecipitation, cell lysates were prepared and standardized as before, and 1 mg protein was immunoprecipitated with appropriately diluted antibody in lysis buffer overnight. Antibody was pulled down with 50 μL either Protein A or G beads (Santa Cruz Biotechnology) for 1 h. Beads were centrifuged at low speed for 10 min, and the supernatant was discarded. Dried beads were mixed with 2× loading dye and boiled for 5 min. Lysate samples were loaded onto gels, and SDS/PAGE was performed as before.

Fluorescence Cell Sorting.

Cells to be analyzed were plated in six-well tissue culture dishes and grown to log phase. Appropriate treatments by either transfection of plasmids or treatment with cisplatin (indicated dose) were performed for 24–48 h. Monolayers were rinsed two times with PBS, and the cells were scraped into microcentrifuge tubes. Cells were centrifuged at low speed and rinsed one time with PBS. Pellets were resuspended in 0.5 mL hypotonic propidium iodide (PI) solution [0.85 mg/mL sodium citrate (Sigma), 0.1 mg/mL RNase A (Qiagen), 0.1% Triton X-100, 20 mg/mL PI (Roche)] and incubated in the dark for 30 min. Cell cycle/PI analysis was performed using a FACScalibur flow cytometer (Becton Dickinson).

Cell Cycle Synchronization.

Hct116, MCF7, U2OS, HeLa, or 293T cells were plated in appropriate complete media and grown to 25–30% confluence; 2 mM thymidine (Sigma) was added to the media for 18 h followed by block release by washing two times with warm PBS and refeeding with fresh complete media. Cells were allowed to grow for 9 h and then retreated with 2 mM thymidine for 17 h (second block). After the second block, cells were released to cycle as before.

Thymidine–nocodazole block.

Hct116, MCF7, U2OS, HeLa, or 293T cells were synchronized by plating to 40% confluence in normal complete media; 2 mM thymidine was added for 24 h followed by cell release by washing two times with warm PBS and refeeding with fresh complete media. Cells were released for 3 h, and then, 100 ng/mL nocodazole (Sigma) was added to the media for 12 h. Cells were then released again by washing two times with PBS and changing to fresh complete media.

Dual Luciferase Reporter Assays.

Analysis of p53 transcriptional activity was performed by transfecting log-phase cells cultured in 12-well dishes with either a p53 luciferase reporter plasmid containing the three copies of the p53 binding sites from the 14-3-3σ or MDM2 promoters. Cells were also cotransfected with the Renilla luciferase reporter plasmid. After transfection, cells were treated with either AZD1152-HQPA or vehicle and incubated for 24 h. At harvest, cells were collected using passive lysis buffer (Promega) and analyzed according to the manufacturer’s protocol for the Dual Luciferase Reporter Assay kit (Promega).

Bimolecular Fluorescence Complementation Assays.

C- and N-terminal Venus plasmids were provided by Gordon Mills (51) and modified to contain Aurora B or p53 by subcloning. Primers for Aurora B and p53 were designed to contain AscI and EcoRV restriction sites (Table S1). Cells were transfected by the liposome method with Venus plasmids containing Aurora B, p53, or empty Venus plasmids as appropriate controls. After 24 h, cells were imaged by fluorescent microscopy as described above.

Construction of Mutants.

Aurora B and p53 mutants were constructed using a site-directed mutagenesis technique. Forward and reverse primers (complementary) that were ∼30 bases long were used in a PCR with PFU Turbo polymerase (Stratagene) to amplify plasmids in their entirety. The primers were designed to change one or two bases to effect the change in the amino acid sequence. Table S1 shows the sequences of the primers used for this purpose. After amplification, plasmids were treated with restriction enzyme DpnI (New England Biolabs) to digest any remaining template. Plasmids were then transformed into DH5α Escherichia coli competent cells and selected with appropriate antibiotics.

In Vitro Kinase/Binding Assays.

Immunopurified Aurora B (IP as described previously) or recombinant Aurora B (Cell Signaling) was incubated in 1× kinase buffer [80 mM Mops (Sigma), 7.5 mM MgCl₂ (Fisher), pH 7.0] with GST-purified p53 substrates, cold ATP, and γ³² ATP (Perkin-Elmer) at 30 °C for 15 min. Kinase reactions were mixed with loading dye and analyzed by SDS/PAGE as described before. SDS/PAGE gels were dried and imaged using a phosphoimager cassette (Molecular Dynamics) and a Typhoon Trio variable mode imager. Images were processed using Image Quant 5.1 software. Recombinant p53 substrates were produced by growing BL-21 E. coli bacteria transformed with the GST-p53 plasmid of interest in 250 mL LB for 1 h followed by induction of expression with 1 mM isopropyl β-D-1 thiogalactopyranoside (IPTG) (Fisher). Cells were grown for 4 h and harvested by centrifugation. Cells were lysed with NaCl, EDTA, Tris, NP40 buffer (NETN) buffer (20 mM Tris⋅HCl, pH 8.0, 150 mM NaCl, 1 mM EDTA, 1% (vol/vol) Nonidet P-40) plus protease–phosphatase inhibitor mixture and sonicated for 5 min. Cell debris was removed by centrifugation (10,000 × g), and then, 200 μL GST beads (GE Healthcare) were added. Lysates were incubated with beads overnight at 4 °C. The next day, the beads were washed three times in NETN plus inhibitors followed by one wash in kinase buffer. Recombinant p53 substrates for in vitro binding assays were prepared as for kinase assays. Substrates were incubated with IP-purified Flag-Aurora B overnight in 1 mL lysis buffer. GST-tagged substrates were pulled down using GST beads followed by SDS/PAGE and Western blotting with anti-Flag antibody.

Real-Time qPCR.

Total RNA was extracted from 293T, MCF7-Her18, or Hct116 p53^−/− cells with TRIzol (Invitrogen) according to the manufacturer’s instructions. RT was performed according to the manufacturer’s instructions using the iScript cDNA synthesis kit (BioRad); 1 µL per reaction of cDNA product was used in real-time qPCR according to the manufacturer’s instructions with the iQ SYBR Green Supermix (BioRad) and iCycler (BioRad) thermocycler. The following cycle was used: 95 °C for 10 min (1 cycle), 95 °C for 15 s, 60 °C for 1 min, 95 °C for 15 s for 40 cycles and then, 95 °C for 15 s and 60 °C for 1 min. Nucleotide sequences of forward and reverse primers are listed in Table S2.

Antibodies.

The following antibodies were used in this study: 14-3-3σ (1433S01; RDI), Actin (A2066; Sigma), Annexin V-FITC (556419; BD Biosciences), Aurora B (Ab2254; Abcam), Bad (B36420; BD Biosciences), Bax (B73520; BD Transduction), Cyclin A (SC751; Santa Cruz), Cyclin B1 (SC245; Santa Cruz), Cyclin D (MS-2110; Neomarkers), Cyclin E (SC-247; Santa Cruz), Flag (A804-200; Sigma), GFP (SC-9996; Santa Cruz), HA (12CA5; Roche), His (SC-803; Santa Cruz), HH3 (p-S10, 05–817; Upstate), MDM2 (S3813; Santa Cruz), Mouse IgG (488; Alexa; A11029; Molecular Probes), Mouse IgG (568; Alexa; A11031; Molecular Probes), p53 for IP (AB-1, PAB1801; Oncogene Science), p53 for IF (SC-6243; Santa Cruz), p53 (610183; BD Biosciences), p53 (p-S315, 2528S0; Cell Signaling), P21 (610233; Transduction Labs), PUMA (SC-28226; Santa Cruz), Rabbit IgG (488; Alexa; A1103; Molecular Probes), rabbit IgG (568; Alexa; A11011; Molecular Probes), and Survivin (2808; Cell Signaling).