Aurora A mediates cross-talk between N- and C-terminal post-translational modifications of p53

Abstract

The serine/threonine protein kinase Aurora A is known to interact with and phosphorylate tumor suppressor p53 at Serine 215 (S215), inhibiting the transcriptional activity of p53. We show that Aurora A positively regulates human p53 protein levels and, using isogenic p53 wild-type and p53-null colorectal carcinoma cells, further show that p53 regulates human Aurora A protein expression. S215 is located in the DNA-binding core of p53 and at the center of the cryptic epitope for PAb240 antibody, which is used to detect mutant and denatured p53. Following denaturing SDS PAGE, the PAb240 epitope was detectable by immunoblotting in only two out of eight cell lines. The efficacy of novel p53-targeted anticancer therapies may be influenced by the conformational state of p53, therefore, the initial determination of p53 status may be relevant. We found no correlation between phosphorylation of p53 at S215 and PAb240 antibody recognition. However, phosphorylation at S37 was positively associated with PAb240 reactivity. More importantly, we provide the first evidence of Aurora A-mediated cross-talk between N- and C-terminal p53 post-translational modifications. As p53 and Aurora A are targets for anticancer therapy the impact of their reciprocal relationship and Aurora A-induced post-translational modification of p53 should be considered.

Introduction

The tumor suppressor p53 is expressed in normal tissues at extremely low levels, principally due to its rapid ubiquitination and degradation driven by Mdm2.¹ In the event of cellular and genotoxic stress, wild-type p53 is transiently stabilized and transported to the nucleus where it binds to DNA and regulates the transcription of p53-dependent genes mediating autophagy and cellular metabolism and stress responses including apoptosis, DNA repair, senescence and cell cycle arrest.^2–5 Post-translational modifications stabilize and activate p53 in order to elicit the appropriate biological response pathway (Fig. 1A).^3,6–8 Mutations in p53 are found in over 50% of all human cancers and in up to 60% of colorectal cancers, therefore the p53 protein is a major target for anticancer therapy.⁹

Figure 1.

(A) Schematic indicating the positions of DO-11, PAb240 and DO-12 antibody epitopes, the site-specific N- and C-terminal post-translational modifications of the human p53 protein investigated in this study and recently identified sites of acetylation and phosphorylation (K120, K164, S215 and K357). (B) Panel a, protein gel blot analyses of p53 and Aurora A proteins endogenously expressed from p53 isogenic HCT116^−/−, HCT116^−/+ and HCT116^+/+ cell lines. Panel b, schematic demonstrating cross-talk between p53 and Aurora A proteins mediated by Fbxw7.

Aurora A is required for control of centrosome duplication, separation and maturation, bipolar spindle assembly, chromosome alignment, entry into and exit from mitosis and is described as an oncogene as it is amplified in many human cancers.^10,11 Expression of Aurora A is cell cycle dependent, being initiated at late S phase and maximum at G₂-M phase.¹² In murine embryonic fibroblasts (MEFs), the inhibition of Aurora A can alter the rate of cell growth depending on the level of expression of p53.¹³ De-regulation of Aurora A expression and kinase activity has been associated with the development of malignancy, hence Aurora A is also a critical target for cancer therapeutics.^11,14–16

Aurora A kinase has been shown to phosphorylate p53 at serine 215, the first phosphorylation site to be described located within the p53 DNA-binding domain. This post-translational modification plays a major role in cell cycle control, apoptosis and tumor development as Aurora A-mediated p53 S215 phosphorylation is reported to directly inhibit p53 function, abrogating p53 DNA binding and transactivation activity.¹⁷

Here, we show that Aurora A regulates human p53 protein levels and additional post-translational modifications of p53 and demonstrate a reciprocal role for p53 in the regulation of Aurora A protein in human colon carcinoma epithelial cells, thereby indicating important cross-talk between human p53 and Aurora A.

Results

Cross-talk between Aurora A and p53 protein levels.

Figure 1B, panel a, demonstrates that increased expression of endogenous human wild type p53 protein in p53 isogenic HCT116 colon carcinoma epithelial cell lines (p53^−/−, p53^+/− and p53^+/+) corresponds with a reduction in the level of Aurora A protein. In murine cells, loss of p53 leads to upregulation of Aurora A through reduced expression of Fbxw7, a p53-dependent tumor suppressor gene which controls Aurora A protein expression as depicted schematically in Figure 1B, panel b.^13,18

Following treatment with Aurora A siRNA which efficiently and specifically silenced Aurora A mRNA and protein in SW620 cells (Fig. 2A and B) and HCT116^+/+ cell line (Fig. 2B, mRNA results not shown), we observed a large increase in the expression of mutant and wild type p53 protein respectively, as detected by the antibody DO-1 (Fig. 2B). In the case of SW620 cell line (but not HCT116^+/+ cell line), there was an accompanying increase in mutant p53 mRNA levels, suggesting a possible increase in transcription (Fig. 2A). However, mRNA levels of p21 and HDM2 in SW620 cells were unaltered and p21 and HDM2 proteins remained undetectable (Fig. 2A and B).

Figure 2.

(A) Results from real-time RT-PCR demonstrating changes in mRNA levels following treatment of SW620 cells with Aurora A siRNA. Note the large increase in p53 mRNA without the corresponding increase in mRNA of p53-dependent genes p21 and HDM2. (B) Protein gel blot analyses of endogenously expressed mutant and wild-type p53 proteins in SW620 and HCT116^+/+ cell lines, respectively, following treatment with Aurora A siRNA.

S215 phosphorylation is not associated with PAb240 epitope recognition.

Aurora A is thought to exert an effect at a number of stages in the p53 pathway indicative of their potential involvement in an integrated functional network. Serine 215 lies within the epitope RHSVV of the antibody PAb240 which is routinely used to distinguish between wild-type and mutant p53 proteins (Fig. 1A).^19–22 We have observed that following protein gel analysis of a series of human colon carcinoma epithelial cell lines (Table 1) the PAb240 epitope was detectable by immunoblotting in only two out of eight cell lines investigated in our study (Fig. 3Aa). We have previously demonstrated that the availability of the PAb421 antibody epitope is dependent upon p53 C-terminal acetylation at K382.²³ Therefore, we were interested in investigating the conditions under which the PAb240 epitope is available. In particular, we wished to establish whether phosphorylation at serine 215 could influence the ability of p53 to recognize the PAb240 antibody epitope.

Table 1.

Summary of the cell lines investigated in this study indicating their p53 status and position and class of p53 mutations

Cell	Origi	P53	Mutation	Mutation
ARPE-	Normal retinal	Wild	N	N
HCT116-	Colon carcinoma	Nul	N	N
HCT116+/	Colon carcinoma	Wild	N	N
RK	Colon carcinoma	Wild	N	N
LoV	Colon carcinoma	Wild	N	N
SW1116	Colon carcinoma	Mutan	A159D	Structura
HT2	Colon carcinoma	Mutan	R273H	Contac
SW480	Colon carcinmoa	Mutan	R273H	Contac
SW620	Colon carcinoma	Mutan	R273H	Contac
TOV-	Ovarian adenocarcinoma	Mutan	R175H	Structura

Figure 3.

(A) Panel a, protein gel blot analyses of p53 proteins from human colon carcinoma epithelial cell lines. Total protein loading results described in Figure 3A (panels a and b) for wild type proteins represent a 4-fold loading of protein (4 x actin) compared with mutant p53 proteins. The PAb240 epitope is only detected in twp out of eight cell lines. Panel b, protein gel analyses of human colon carcinoma epithelial cell lines demonstrating the post-translational modification status of p53 proteins in the absence of applied stress. Due to low endogenous p53 levels in cell lines expressing wild-type p53 total protein loading results described in Figure 3A for wild-type proteins represent a 4-fold loading of protein (4 x actin) compared with mutant p53 proteins. (B) Protein gel analyses of wild-type and phosphorylation mutant proteins p53S215A and p53S215D exogenously expressed in HCT116^−/− cells. Expression of p53S215D abrogates the expression of p53-dependent proteins p21 and HDM2. (C) Protein gel blot analyses of post-translational modification status of p53 wild-type and phosphorylation mutant proteins p53S215A and p53S215D exogenously expressed in HCT116^−/− cells. The result from hwtp53 DO-1 reactivity is from the same blot and film as the other results, but has been cut and pasted to provide the order of samples required for the figure. (D) Schematic highlighting the similar effects of the exogenous expression of human p53 phosphorylation mutant proteins p53S215A in HCT116^−/− cells and siRNA silencing of Aurora A in SW620 cells on site-specific N- and C-terminal phosphorylation and acetylation events. Open circles depict non-phosphorylated sites. (P) Demonstrates phosphorylated sites.

Actin was used as a reference control for equivalent total protein loading between the cell lines.^24,25 The expression of both wild-type and mutant p53 in cell lines, in the absence of applied stress, was detected with antibody DO-1 (Fig. 3Aa). Due to low endogenous p53 levels in cell lines expressing wild-type p53 total protein loading results described in Figure 3A (panels a and b) for wild-type proteins represent a 4-fold loading of protein (4 x actin) compared with mutant p53 proteins.¹ In cell lines expressing wild-type p53, the PAb240 epitope remained undetectable despite the increase in total protein loading, suggesting this is not a threshold effect (Fig. 3Aa).

Mutations of the p53 gene are predominantly single point mutations within the DNA-binding region such as in the cell lines incorporated into this study (Table 1).^9,26 Interestingly, though the p53 contact mutant cell lines HT29, SW480 and SW620 shared the same point mutation (R273H), the PAb240 epitope was detected in HT29 and SW620 but not SW480 cells (Fig. 3Aa). To exclude the possibility that further mutations of p53 may influence reactivity with the PAb240 epitope we cloned and sequenced the p53 core domain DNA in our cell lines and were unable to detect further p53 mutations or splicing events (data not shown).²⁷

The epitopes for antibodies DO-11 and DO-12 occupy constrained loops at the S5 domain and the junction of S9/S10 domains of the p53 protein respectively, adjacent to but discrete from the PAb240 epitope (Fig. 1A).^28–30 These antibodies detected mutant p53 proteins in HT29, SW480 and SW620 cell lines (Fig. 3Aa). Thus, the constraints upon the availability of epitopes for DO-11 and DO-12 are uncoupled from the detection of PAb240.

As there is no antibody available which recognizes phosphorylation at S215 we studied the specific effect of phosphorylation at this residue via exogenous expression of p53 constructs; human wild-type p53 and phosphorylation mutant constructs in which S215 was mutated to either alanine (S215A) or aspartic acid (S215D), mimicking the non-phosphorylated and phosphorylated forms of S215 respectively were expressed in HCT116 p53^−/− cells.¹⁷ Following protein gel blotting, p53S215A and p53S215D proteins were both reactive with the PAb240 antibody, indicating that the phosphorylation status of p53 at S215 does not influence PAb240 antibody reactivity (Fig. 3B). Here we demonstrate that exogenously expressed p53S215D (which mimics phosphorylated p53S215) but not p53S215A, inhibits the expression of p21 and HDM2 proteins in HCT116 cells (Fig. 3B).

To further substantiate our findings we performed the converse experiment treating SW620 (mutant p53, reactive with PAb240) and HCT116^+/+ (wild-type p53, non-reactive with PAb240) cells with Aurora A siRNA in order to deplete Aurora A and by inference, phosphorylation at S215. Following treatment with Aurora A siRNA p53 proteins expressed in SW620 cells recognized the PAb240 epitope whereas proteins expressed in HCT116^+/+ were unable to do so. Therefore, we found no change in PAb240 reactivity in the either cell line (Fig. 2B), consistent with our previous findings that phosphorylation at S215 does not influence the availability of the PAb240 epitope. In the HCT116^+/+ cell line expressing endogenous wild-type p53, the increase in p53 protein levels were associated with an increase in expression of the p53 target gene HDM2 protein whereas increased levels of mutant p53 protein expressed in SW620 cells was not accompanied by an increase in either p21 of HDM2 (Fig. 2B).

Mediation of p53 post-translational modifications by Aurora A.

Aurora A siRNA treatment in HCT116^+/+ cells expressing very low levels of wild-type p53 had no effect on p53 post-translational modifications, despite the elevated levels of p53 as demonstrated by antibody DO-1. Conversely, increased levels of specific p53 post-translational modifications were seen in SW620 cells. In order to obtain meaningful protein gel results, the amount of protein loaded in Figure 2B was lower than that in Figure 3A, thereby accounting for the discrepancy in post-translational modifications seen in untreated SW620 cells in the two figures. Following siRNA silencing of Aurora A in SW620 cells we report a specific increase in phosphorylation at the N-terminus of p53 at S33 and loss of phosphorylation at the C-terminus at S392 (Fig. 2B) correlating with post-translational modifications observed following the expression of p53S215A and p53S215D, namely loss of phosphorylation at S392 and loss of phosphorylation at S33 respectively (Fig. 3C). Figure 3D schematically depicts the post-translational modifications of p53 that arise following inhibition of phosphorylation of p53 at S215 by the expression of p53S215A in HCT116^−/− cells and Aurora A silencing in SW620 cells.

S37 phosphorylation correlates with PAb240 epitope recognition.

Using antibodies specific for modified residues of p53, we further explored the relationship between specific individual post-translational modifications (Fig. 1A) and PAb240 epitope recognition in human colon carcinoma epithelial cell lines in the absence of stress. Cell lines expressing mutant p53 were phosphorylated at S15, whereas phosphorylation at S33 was barely detectable (Fig. 3Ab). This may be relevant to mutant p53 function as S33 phosphorylation is involved in UV-induced apoptosis and activation of p53-dependent genes.³¹ Phosphorylation at S46 was confined to the cell line SW620 (Fig. 3Ab) which is important since phosphorylation at this site disrupts the interaction between p53 and HDM2 inhibiting HDM2-mediated p53 nuclear export and ubiquitination.^1,32 Here, S37 phosphorylation was detected in cell lines HT29 and SW620 alone, therefore correlating with the recognition of the PAb240 epitope (Fig. 3Ab).

Acetylation stabilizes p53 by excluding ubiquitination at the same site. The major sites for acetylation of p53 are lysines K373 and K382 at the C-terminus.^33–35 Phosphorylation of p53 at S392 following genotoxic stress regulates the oligomerisation of p53 and sequence-specific DNA-binding and impairs p53 nucleolar localization.³⁶ Both acetylation at K382 and phosphorylation at S392 were detected in HT29, SW480 and SW620 mutant p53 cell lines and did not correlate with the detection of the PAb240 epitope (Fig. 3Ab).

We further investigated the observed correlation between phosphorylation at p53 S37 and PAb240 epitope recognition by expressing p53 constructs; wild-type p53, p53S37A and p53S37D in HCT116 p53^−/− cells.³⁷ Previous studies describe transfection-mediated activation of wild-type p53 and in this study, exogenously expressed wild-type p53 was reactive with PAb240 antibody and constitutively phosphorylated at S15, S33, S37, S46, S392 and acetylated at K382 (Fig. 4A).³⁸ Expression of p53S37A and p53S37D reduced phosphorylation at S33, illustrating that alterations to residue serine 37 influence phosphorylation events at S33.²⁴ Importantly, PAb240 reactivity was detectable for p53S37D but not p53S37A enhancing evidence that recognition of the PAb240 epitope correlates with phosphorylation of p53 at S37 (Fig. 4A). p53S215A and p53S215D proteins expressed in HCT116^−/− cells were also phosphorylated at S37 and reactive with PAb240 lending further support for the relationship between PAb240 antibody reactivity and phosphorylation at S37 (Fig. 3B).

Figure 4.

(A) Protein gel analyses of the post-translational modification status of p53 and reactivity with PAb240 antibody in wild-type and phosphorylation mutant proteins p53S37A and p53S37D exogenously expressed in HCT116^−/− cells. (B) Protein gel blot analyses of alterations in the S37 phosphorylation status of p53 in exogenously expressed wild-type and phosphorylation mutant proteins p53S37A and p53S37D in the presence of alkaline phosphatase (AP). (C) Schematic highlighting the effects of human p53 phosphorylation mutations on site-specific N- and C-terminal phosphorylation and acetylation events. Phosphorylation mutation sites S37A and S37D are shown in bold. Open circles and squares depict non-phosphorylated and non-acetylation sites respectively.

We investigated the effect of phosphorylation on immunoreactivity with PAb240 by treating identical lysates from wild type, p53S37A and p53S37D proteins exogenously expressed in HCT116^−/− cells with alkaline phosphatase (AP, see Materials and Methods).³⁹ Wild-type p53 protein was de-phosphorylated at S37 and non-reactive with PAb240 following treatment with AP (Fig. 4B). p53S37A protein served as a negative control, non-phosphorylated at S37, and non-reactive with PAb240. In contrast, p53S37D remained reactive with both anti-phospho S37P antibody and PAb240 antibody (Fig. 4B) further supporting the relationship between phosphorylation at p53 S37 and PAb240 epitope recognition.

Cross-talk between N- and C-terminal post-translational modifications of p53.

Multisite modification of proteins constitutes a complex regulatory program. We have previously reported evidence of interdependency between N- and C-terminal post-translational p53 modifications on human p53.²⁴ In this study we show site-interdependency between N-terminal phosphorylation at S37 and modifications at both the N- and C-terminus of p53 since the expression of p53S37A (but not p53S37D) in HCT116^−/− cells reduced phosphorylation at S15 and S46 and acetylation at K382 (Fig. 4A). This cross-talk is depicted schematically in Figure 4C.

The expression of p53S215A and p53S215D proteins resulted in specific loss of phosphorylation at S392 and S33 respectively (Fig. 3C) and is summarized schematically in Figure 3D. This is consistent with p53 post-translational modifications observed following treatment of SW620 cells with Aurora A siRNA, namely, increased phosphorylation of p53 at S33 and loss of phosphorylation at S392 (Figs. 2B and 3D). Expression of p53S215 phosphorylation mutant proteins correlated with a large increase in acetylation at K382 compared with wild-type p53 (Fig. 3C). However, we were unable to detect upregulation of p53 target genes p21 and HDM2 upon expression of p53S215A and were unable to detect p21 or HDM2 following expression of p53S215D (Fig. 3B) which may be important given the role that acetylation plays in the activation of p53.^33,40

Discussion

It has been established that the tumor suppressor p53 and Aurora A kinase proteins interact, p53 binding to the Aurora box of Aurora-A.¹⁶ Significant correlations have been described between Aurora A and p53 protein levels in human breast cancer cell lines and primary human tumors.^11,13,41 What is not clearly established is the extent of the cross-talk between these two proteins and the functional implications of specific interactions, though it has been suggested that alterations in the balance between p53 and Aurora A could result in checkpoint abnormalities, chromosome instability and carcinogenesis.¹²

In this study, we demonstrate that increased expression of endogenous human wild-type p53 protein in p53 isogenic HCT116 colon carcinoma epithelial cells is associated with a reduction in the level of Aurora A protein. Our findings support data from murine studies in which Aurora A protein levels in normal thymus and other tissues from untreated p53^−/− mice are consistently higher than in corresponding p53^+/− mice, possibly contributing to the genomic instability and aneuploidy seen in cells from these mice.^11,13 Loss of p53 has been associated with the upregulation of Aurora A through reduced expression of the p53-dependent gene Fbxw7, which controls Aurora A protein expression^13,18 At present this is an important area of research, the Breakthrough Research Center is currently investigating whether Aurora-A overexpression and p53/Fbxw7 co-operate in breast tumorigenesis.

Following treatment with Aurora A siRNA we observed a large increase in the expression of mutant and wild-type p53 protein in human colon carcinoma cell lines providing further evidence of cross-talk between human p53 and Aurora A proteins. The corresponding mRNA and protein levels of p21 and HDM2 in SW620 cells were unaltered, consistent with the inability of mutant p53 proteins to upregulate p53-dependent genes.²

As previously stated, Aurora A-mediated p53 S215 phosphorylation is thought to inhibit p53 function, abrogating p53 DNA binding and transactivation activity.¹⁷ These findings are corroborated in our study as we demonstrate that exogenous expression of p53S215D in HCT116 cells inhibits the expression of p21 and HDM2 proteins.

Here we provide the first report of site-interdependent post-translational modification of p53 involving Aurora A and phosphorylation at S215 in the DNA-binding domain of p53, since the expression of p53S215A and p53S215D proteins resulted in the specific loss of phosphorylation at S392 and S33 respectively. In addition, treatment of SW620 cells with Aurora A siRNA increased phosphorylation of p53 at S33 and abrogated phosphorylation at S392. In view of the effect of Aurora A-mediated phosphorylation of p53 at serine 215 on cell cycle control, apoptosis and tumor development, the reciprocal phosphorylation status of p53 at S33 and S392 is potentially very important.^{31,32,38,40,42,43} Loss of phosphorylation at S33 may be relevant as S33 phosphorylation regulates p53 stability and UV-induced apoptosis and may be required for phosphorylation at S37.³¹ Phosphorylation of S392 is known to occur in response to UV radiation and has been implicated in alterations in p53 oligomerization, nucleolar localization and sequence-specific DNA-binding.^31,36 Changes in the phosphorylation status at both S33 and S392 have also been shown to influence the activation of p53-dependent genes.^36,44 It remains unclear whether the cross-talk between Aurora A phosphorylation at S215 and post-translational modifications at S33 and S392 are dependent upon the prior exposure of the PAb240 epitope, the level of expression of p53 or upon p53 being in the active or mutant conformation.

Aurora A has been shown to regulate the phosphorylation of Histone H3 specifically at S10.⁴⁵ The histone code suggests that an increase in histone acetylation is a signal to transcribe open chromatin.⁴⁶ We have previously demonstrated the ability of p53 to regulate phosphorylation and acetylation of Histone H3 at S10 and K14 respectively.^47,48 More recently, we have identified clear knock-on effects on Histone H3 at S10, K4, K9 and K14 following post-translational modifications of p53.²⁴ We and others have proposed the integration of a histone code and p53 code. This histone-p53 code may be an important epigenetic means of maintaining the balance between DNA repair, apoptosis and cell cycle arrest that is required to prevent adverse cellular transformation.^24,49

In this study, we demonstrate that following protein gel blot analysis the PAb240 epitope was detectable by immunoblotting in only two out of eight human colon carcinoma cell lines. Many tumors with a p53 mutation have been shown to be resistant to anticancer therapy whereas, other tumors with mutant p53 demonstrate an improved therapeutic response.^26,30 Conflicting results may be due, in part, to incorrect assessment of p53 status as highlighted in this study. p53 status would also be relevant to efficacy of treatment where the choice of anticancer therapy may be influenced by the conformation state of p53 (wild-type or mutant). Therefore, we suggest that antibodies such as DO-11 and DO-12 may be used in conjunction with PAb240 to verify the detection of wild-type or mutant p53 proteins.

Our experimental findings suggest that there is no direct correlation between mutational status of p53 and PAb240 epitope recognition. In a previous study, we have shown that the detection of the PAb421 epitope is dependent upon C-terminal acetylation at K382.²³ In this study, we found no association between S215 phosphorylation and PAb240 epitope recognition. Importantly, using phospho-specific antibodies and expression of phosphorylation mutant p53 proteins we identify a relationship between S37 phosphorylation and detection of the PAb240 epitope. Phosphorylation events at the N-terminus of p53, particularly at serines 15 and 37 are crucial in the transactivation of p53-responsive genes.^38,50 N-terminal phosphorylation may act as a switch to promote p53-dependent transactivation while simultaneously preventing targeting of p53 for degradation by MDM2.¹

A cascade of post-translational modifications tightly controls p53 activity, stability and conformation, thereby invoking an appropriate cellular response to intra- and extracellular stresses.^3,8,32 We have previously reported evidence of interdependency between N- and C-terminal post-translational p53 modifications on human p53.²⁴ Earlier, we discussed the involvement of S215 phosphorylation on p53 site-interdependent post-translational modifications. We extend this evidence to include site-interdependency between N-terminal phosphorylation at S37 and modifications at both the N- and C-terminus of p53 since the expression of p53S37A reduced phosphorylation at S15 and S46 and acetylation at K382. Phosphorylation at S15 blocks the binding of p53 and MDM2 leading to an increase in p53 stability as seen with mutant p53 proteins.³² Phosphorylation at S46 disrupts the interaction between p53 and HDM2 inhibiting HDM2-mediated p53 nuclear export and ubiquitination.^1,32 Phosphorylation at S46 is also crucial for the induction of p53-dependent pro-apoptotic genes and enhances the protein-protein interaction of p53 at TAD2 with proteins such as p62.^43,51,52 S33 and S46 phosphorylation following UV treatment regulate p53 stability and are a prerequisite for phosphorylation at S37 for wild-type p53.⁵³ Given the importance of phosphorylation of p53 at S15 and S46 in the activation and function of p53, our discovery that S37 phosphorylation may also regulate S46 phosphorylation in human cells is highly significant. Acetylation of p53 activates sequence-specific DNA-binding regulating the transcription of pro-apoptotic target genes. Mutation of multiple acetylation sites results in the expression of p53 proteins unable to activate many p53 target genes.³³ The importance of cross-talk between p53 post-translational modifications is highly relevant following the resolution of a novel structure for full-length p53 protein which predicts that the N- and C-termini interact directly in tetrameric p53.^25,54

The development of novel anticancer therapies has involved the identification of small molecules which stabilize p53 and activate apoptosis in tumor cells expressing wild-type or mutant p53.² Small-molecule inhibitors of Aurora kinases have also been used in targeted anticancer therapy to block tumor growth and induce tumor regression.^{11,15,55–57} Potentially, inhibitors of Aurora A may induce post-translational modifications of p53 influencing p53 stability and/or the interaction with other cellular components such as Histone H3 and ultimately the mediation of an appropriate cellular response to stress. Complex interactions therefore may exist between anticancer therapies which stabilize p53 and inhibit Aurora kinases.^5,11,15,24

Materials and Methods

Tissue culture.

Colon carcinoma epithelial cell lines SW480 and SW620 cells were cultured in Leibovitz's L-15 medium at 37°C in the absence of 5% CO₂ in air. All other cell lines were grown in DMEM at 37°C in the presence of 5% CO₂. All media were supplemented with 10% FCS and 2 mM glutamine. Control cell lines included in the study were the isogenic cell clone HCT116 p53^−/−, null for p53 and the ARPE-19 normal human epithelial cell line which expresses wild-type p53.^58,59

Transfection with p53 phosphorylation mutant constructs and Aurora A siRNA.

Human wild-type p53 and site-specific mutant S37A, S37D, S215A and S215D cDNAs were cloned into vector pcDNA3, and constructs were verified by sequencing.^17,37 HCT116 p53^−/− cells were seeded into 6-well plates and transfected with Lipofectamine 2000 (Invitrogen Corporation) in accordance with manufacturer's instructions. Cells were harvested 24 h post-transfection and cell lysates prepared as previously described.²⁴ Each experiment was repeated three times. HCT116^+/+ cells were seeded into 6-well plates and transfected with 10 or 20 µM siRNA and 3 µl Oligofectamine (Invitrogen Corporation) in a final volume of 200 µl DMEM antibiotic-free medium. SW620 cells were transfected with 10 or 20 µM siRNA in the absence of Oligofectamine in a final volume of 200 µl Leibovitz's L-15 antibiotic-free medium. Lamin A/C siRNA was used in the experiment as an internal positive control for siRNA knockdown. The Lamin A/C and Aurora A siRNA sequences used (Perbio Dharmacon) have been previously described.^60,61 Cells were harvested 48 h post-transfection and prepared for protein and mRNA analysis as previously described.²⁴

Protein gel blotting.

Protein gel blot analysis was performed as previously described using actin as a control for total protein loading.²⁴ Total human p53 was detected using DO-1 (Santa Cruz, sc-126). Antibodies PAb240, DO-11 and DO-12 were produced in-house from hybridoma cell lines.^28,29 Phosphorylation of p53 protein at S15, S33, S37, S46 and S392 were detected by specific anti-phospho-p53 antibodies (Cell Signaling Technology, 9284, 2526, 9289, 2521 and 9281). Acetylation at K382 was detected using anti-acetyl-p53 (K382) (Epitomics, 2485-1). Aurora A was detected using antibody 4718, New England Biosciences. Each experiment was repeated three times.

Quantitative real time RT-PCR.

Total RNA was isolated using an RNeasy kit (Qiagen) in accordance with manufacturer's instructions. mRNA levels were measured in quadruplicate by quantitative real-time RT-PCR performed with a SYBER green Gigakit (Qiagen) using an Opticon II Engine (BioRad) as previously described.²⁴ The GAPDH, Lamin A/C and Aurora A primer sequences have been published previously.^24,60,61

Alkaline phosphatase dephosphorylation of p53.

Cell lysates were prepared as described above then clarified by centrifugation at 14,000 rpm for 30 min at 4°C and divided into two aliquots. One aliquot was treated with 10 units of Shrimp Alkaline Phosphatase (Promega, M8201) for 30 min at 37°C and one aliquot was mock-treated in accordance with manufacturer's instructions. The resultant lysates were then used in proteingel blotting experiments as described earlier.²⁴

Disclosure of Potential Conflicts of Interest

No potential conflicts of interest were disclosed.