Identification of ANKRD11 as a p53 coactivator

The p53 protein is the most important tumor suppressor in the cell and is described as `the guardian of the genome' (Lane, 1992). The function of p53 is predominantly dependent on its activity as a sequence-specific transcription factor, controlling the expression of various target genes that mediate biological functions such as cell-cycle arrest, apoptosis, senescence, differentiation, DNA repair and inhibition of angiogenesis and metastasis (Liu and Chen, 2006). Approximately 50% of all cancers contain a mutation in the TP53 gene (Vogelstein et al., 2000). The six most frequently occurring tumorigenic mutations of p53 cluster within the DNA-binding surface, involving residues that either directly contact DNA (`DNA-contact' mutations – residues R248 and R273) or perturb the structural integrity of the DNA-binding surface of p53 (`structural' mutations – residues R175, G245, R249 and R282) (Bullock and Fersht, 2001). DNA-contact mutants generally retain a similar structural integrity and thermodynamic stability to that displayed by wild-type p53; however, the mutation might significantly impair or even completely abrogate its ability to interact with its cognate DNA-response element (Bullock et al., 2000). Rescue of DNA-contact mutants has been demonstrated through restoration of their DNA-binding affinities following treatment with a p53-derived C-terminal peptide (Selivanova et al., 1997), small p53-activating molecules (CP-257042, CP-31398 and PRIMA-1) (Bykov et al., 2002; Foster et al., 1999) or the introduction of second site suppressor mutations (Brachmann et al., 1998). Despite these recent advances, the development of successful pharmacological therapeutics designed to rescue p53 DNA-contact mutants has proven to be a challenging task because of the complex biomechanistics that occur during the restoration of mutant p53 function.

The p53 transcription factor is stabilized and activated in response to various forms of intracellular or extracellular stress. Stabilization and activation of p53 occurs through a complex pattern of post-translational modifications, including phosphorylation and acetylation. The acetylation of specific lysine residues of p53 is mediated by several acetyltransferases, including p300/CBP (Lys164, Lys370, Lys372, Lys373, Lys381, Lys382 and Lys386), Tip60/hMOF (Lys120) and P/CAF (Lys320) (Gu and Roeder, 1997; Sakaguchi et al., 1998; Sykes et al., 2006; Tang et al., 2008). Earlier studies have shown that acetylation of p53 promotes its ability to bind its cognate DNA-response element in vitro (Gu and Roeder, 1997; Sakaguchi et al., 1998); however, these findings are not consistent with recent in vivo data (Liu and Chen, 2006; Tang et al., 2008; Toledo and Wahl, 2006). It has since been shown that the critical function of acetylation in vivo is not to enhance the DNA-binding affinity of p53, but rather to enhance its interaction with coactivators or histone acetyltransferases (HATs) (Barlev et al., 2001), or perturb its interaction with Mdm2 (Tang et al., 2008) at the promoters of p53-responsive genes. Recent evidence also suggests that acetylation of specific lysine residues in p53 may influence the selection of specific p53 target genes, resulting in induction of either growth arrest or apoptotic pathways (Di Stefano et al., 2005). hADA3 (alteration/deficiency in activation) has been identified as a novel p53-interacting protein capable of recruiting p300/CBP and P/CAF to p53 (Wang et al., 2001). Ectopic expression of hADA3 increased the stability and transcriptional activity of p53, and was shown to have an essential role in p53 acetylation (Wang et al., 2001). Nevertheless, we currently do not have a comprehensive understanding of the complex molecular mechanisms involved in the regulation of p53 acetylation.

ANKRD11 was initially selected as a candidate breast cancer tumor suppressor gene because of its location within the 16q24.3 breast cancer LOH (loss of heterozygosity) region (Powell et al., 2002). LOH of human chromosome 16q occurs in at least half of all breast tumors (Miller et al., 2003) and cytogenetic studies have implicated this to be an early event in breast carcinogenesis (Gong et al., 2001). ANKRD11 (also termed ANCO1; official protein symbol ANR11) was also reported to interact with and suppress the function of the p160 coactivator family, including the oncogene AIB1 (amplified in breast cancer-1) (Zhang et al., 2004). AIB1 enhances ligand-dependent transactivation of steroid nuclear receptors, including estrogen receptor (ER), and is frequently amplified and overexpressed in breast and ovarian cancers (Anzick et al., 1997). ANKRD11 was reported to recruit histone deacetylases (HDACs) through its C-terminus to the AIB1/nuclear receptor complex, resulting in the inhibition of ligand-dependent transactivation (Zhang et al., 2004). Based on these findings, we investigated whether ANKRD11 has properties similar to that of a breast cancer tumor suppressor.

In this study we report the functional characterization of ANKRD11 as a novel p53-interacting protein. We find that ANKRD11 increases the acetylation of p53, potentiating the DNA-binding property and transcriptional activity of p53. Findings from this study suggest that ANKRD11 might have a crucial role in p53-mediated tumor suppression and might also be implicated in the restoration of mutant p53^R273H function. We also show that ANKRD11 itself is a novel p53 target gene.

To gain insight into the functional role of ANKRD11, we investigated its subcellular localization in breast epithelial cells. GFP-ANKRD11 was retrovirally expressed in MCF-10A, a nonmalignant human breast epithelial cell. Fluorescent and phase-contrast microscopy showed that the GFP-ANKRD11 protein forms distinct foci within the nuclei of MCF-10A cells (Fig. 1A). Subsequent counterstaining with the nucleolar marker nucleophosmin revealed that these domains were distinct from nucleoli (Fig. 1A).

Recent studies refer to the above-mentioned domains as `extranucleolar inclusions' (Kashuba et al., 2003). These extranucleolar domains have been reported to house several crucial cell cycle proteins, including p53, p14^ARF and the promyelocytic leukemia (PML) protein. To investigate if ANKRD11 associates with these proteins within extranucleolar inclusions, we determined the localization of endogenous ANKRD11 and p53 in MCF-10A cells in the presence or absence of mitomycin C, a DNA-damaging agent that induces elevated levels of p53 protein. Our findings show that endogenous ANKRD11 and p53 colocalize within nuclear foci, and this interaction is enhanced in the presence of elevated p53 protein levels in response to mitomycin C treatment (Fig. 1B). Similar findings were observed using an alternative anti-p53 antibody (supplementary material Fig. S1). It has also been previously reported that exogenous p14^ARF localizes to both extranucleolar inclusions and the nucleolus (Kashuba et al., 2003). ANKRD11-myc and GFP-p14^ARF were ectopically expressed in HEK293T cells and imaged using confocal microscopy (Fig. 1C). Coimmunolocalization results showed that ANKRD11-myc localized exclusively to extranucleolar p14^ARF foci (filled arrowheads), but not to the larger nucleolar p14^ARF foci (open arrowheads). Endogenous PML counterstaining showed that PML bodies were also recruited to ANKRD11/p14^ARF-positive extranucleolar inclusions (Fig. 1D) as previously reported (Kashuba et al., 2003). In the absence of exogenous p14^ARF, ANKRD11 nuclear foci were observed to exhibit only partial colocalization with PML bodies during late G2 phase of the cell cycle (data not shown). This observation suggests that the association of ANKRD11-PML complex is enhanced in the presence of p14^ARF. The location of nucleoli in these cells was determined through detection of nucleophosmin, and nuclei were counterstained with DAPI. Further studies were undertaken using glycerol gradient centrifugation of lysates of MCF-7 cells stably expressing GFP-ANKRD11-myc. Both p53 and PML cofractionated with ANKRD11-myc (supplementary material Fig. S2), providing preliminary evidence that these proteins may coexist as a complex. Detection of p14^ARF in these fractions was not possible because the MCF-7 cell line does not express endogenous p14^ARF as a result of deletion of the CDKN2A locus (Stott et al., 1998).

Coimmunoprecipitation experiments were used to determine whether ANKRD11 interacts with p53 in vivo. Endogenous p53 was detected in protein complexes immunoprecipitated using an anti-ANKRD11 antibody from protein lysates of MCF-7 cells transduced with recombinant retroviruses expressing GFP-ANKRD11 and treated with mitomycin C for 6 hours (Fig. 2A, upper panel). Similar results were observed after immunoprecipitation of endogenous ANKRD11 from HEK293T cell lysates (Fig. 2A, lower panel). This interaction was specific, because p53 was not detected in complexes immunoprecipitated with preimmune serum. To further map the region of ANKRD11 that interacts with p53, HEK293T cells were transfected with constructs expressing Flag-tagged ANKRD11 regions. Anti-Flag immunoprecipitation of these fragments showed that endogenous p53 interacts with the 817 N-terminal residues of the ANKRD11 protein (Fig. 2B).

Intriguingly, this region of ANKRD11 contains an ankyrin-repeat domain. In silico analysis of the predicted ANKRD11 amino acid sequence revealed that the five ankyrin repeats (residues 133-296) constituting the ankyrin-repeat domain in ANKRD11 were highly homologous with the ankyrin-repeat domains of BARD1 (45% identical; 72% similar), IκBα (27% identical; 68% similar) and 53bp2 (24% identical; 57% similar). Since it has been previously reported that these proteins interact with p53 through their ankyrin domain (Dreyfus et al., 2005; Feki et al., 2005; Gorina and Pavletich, 1996), we investigated whether the homologous ankyrin-repeat domain of ANKRD11 was indeed the minimal region of interaction with p53. For this purpose, an N-terminal myc-tagged construct expressing a region of ANKRD11 encompassing the homologous ankyrin domain (residues 144-313) was used. HCT116 cells were transfected with either myc-ANKRD11^144-313aa or empty vector and treated with mitomycin C for 6 hours. Anti-myc immunoprecipitation of HCT116 cell lysates showed that ANKRD11 interacts with p53 through this 144-313 residue region (Fig. 2C, upper panel). Similar results were obtained after anti-Flag immunoprecipitation of lysates from HEK293T cells expressing Flag-ANKRD11^144-313aa (Fig. 2C, lower panel). Subsequently, an in vitro MBP pull-down experiment was used to determine whether ANKRD11 interacted directly with p53. GST-ANKRD11^144-313aa protein was precipitated after incubation with MBP-p53 amylose beads (Fig. 2D). This interaction was shown to be specific, because GST-ANKRD11^144-313aa protein did not bind MBP amylose beads alone, and GST alone did not interact with the MBP-p53 amylose beads. A schematic diagram depicts the ankyrin repeat (ANK), repressor (RD) and activator (AD) domains of ANKRD11 and summarizes the in vivo and in vitro ANKRD11-p53 interaction results (supplementary material Fig. S3), illustrating the minimal region of ANKRD11 required for p53 interaction (ANKRD11^144-313aa).

Since ANKRD11 directly interacts with p53, we investigated whether ANKRD11 was able to modulate the transcriptional activity of p53. Results from dual luciferase assays show that expression of ANKRD11 in Saos-2 or HeLa cells led to a significant (2.0-fold and 1.8-fold, respectively; P<0.05) dose-dependent increase in p53-mediated transactivation of the pGL2-Promoter-p53-RE-Luc reporter construct containing 17 tandem p53-response elements (p53-REs) (Fig. 3). A similar effect was observed when endogenous p53-REs from CDKN1A (p21^waf1/CIP1) and MDM2 promoter sequences were used in reporter assays, because expression of ANKRD11 caused a significant (maximum 3.3-fold and 4.1-fold; P<0.05) dose-dependent increase in p53-mediated transactivation of the pGL3-Basic-CDKN1A-pro-Luc and pGL2-Basic-MDM2-pro-Luc reporter constructs, respectively (Fig. 3).

Since ANKRD11 is located within the 16q24.3 breast cancer LOH region and is able to enhance p53 activity, it was thought possible that ANKRD11 imparts a biological function as a tumor suppressor. To investigate this possibility further, the relative expression of ANKRD11 in a panel of breast cell lines was determined by real-time RT-PCR. Results showed that the average ANKRD11 expression in the six breast cancer cell lines was downregulated (89% average reduction; P<0.01) when compared with the average ANKRD11 expression in finite life-span HMECs and nonmalignant immortalized breast epithelial cells (Fig. 4A, dark shading).

Our findings suggest that increased ANKRD11 expression enhances p53-mediated transcription of a reporter gene driven by the CDKN1A promoter region (Fig. 3) and ANKRD11 expression is down-regulated in breast cancer cell lines (Fig. 4A). Therefore, we hypothesized that restoring the expression of ANKRD11 in breast cancer cell lines that retain an intact p53 pathway may reduce their oncogenic phenotype by enhancing the transcriptional activity of p53. To test this hypothesis, exogenous ANKRD11 was re-introduced into the MCF-7 (wild-type p53), MB-468 (mutant p53^R273H) and MB-231 (mutant p53^R280K) cell lines. In contrast to the p53^R273H mutant expressed in the MB-468 cell line that possesses the ability to bind its cognate DNA-response element and transactivate target genes, the MB-231 cell line expresses the p53^R280K mutation, completely abrogating its DNA-binding affinity and transactivation potential (Park et al., 1994; Prasad and Church, 1997). Cultures stably expressing GFP-ANKRD11 (MCF-7-ANK-1 to MCF-7-ANK-3; MB-468-ANK-1; MB-231-ANK-1 to MB-231-ANK-3) or negative control cell lines (MCF-7; MB-468; MB-231) (see Materials and Methods) were established from single colonies after retroviral transduction and geneticin selection. The level of ANKRD11 re-expression in these stable cell isolates (Fig. 4A, light shading) was determined and was shown to be comparable to the level of endogenous ANKRD11 expressed by finite life span or nonmalignant immortalized breast epithelial cells (Fig. 4A, dark shading).

The relative CDKN1A expression levels in these MCF-7, MB-468 or MB-231 cultures stably expressing ANKRD11 were determined by real-time RT-PCR (Fig. 4B). MCF-7 and MB-468 cultures stably expressing ANKRD11 showed a 3.3-fold or 2.4-fold increase in CDKN1A expression levels, respectively. These observations were also consistent with p21^waf1 protein levels (Fig. 4B, insets). Restoration of ANKRD11 expression was also associated with increased expression of p53-responsive apoptotic genes, including FAS and NOXA. MCF-7 cultures stably expressing ANKRD11 showed a 2.6-fold increase in FAS expression, whereas a 2.1-fold increase in NOXA expression was observed in MB-468 cultures stably expressing ANKRD11 (supplementary material Fig. S4). Restoration of ANKRD11 expression in the MB-231 cell line caused no significant change in the expression levels of p21^waf1, FAS or NOXA, suggesting that the presence of an intact p53 pathway is required for ANKRD11 to enhance the transcription of downstream p53 target genes (Fig. 4B).

The p21^waf1 expression levels were subsequently determined in MCF-7 cultures stably expressing ANKRD11 (MCF-7-ANK-1) or negative control cells (MCF-7) in response to knockdown of p53 expression. Significant reduction of p21^waf1 protein levels was observed in both the MCF-7-ANK-1 derivative and MCF-7 parental cell line following transfection with p53-specific siRNA (Fig. 4C). Furthermore, ectopic expression of ANKRD11 in a p53-null Saos-2 cell line (Saos-2-ANKRD11) was not associated with increased p21^waf1 expression (supplementary material Fig. S5). Restoration of p53 expression in the Saos-2-ANKRD11 derivative led to a 3.6-fold increase in p21^waf1 expression, while re-expression of similar levels of p53 in the parental Saos-2 cell line was only associated with 1.5-fold induction of p21^waf1 expression (supplementary material Fig. S5). Taken together, these findings confirm that ANKRD11-mediated modulation of p21^waf1 levels is indeed p53 dependent.

The clonogenic properties of the stably expressing MCF-7, MB-468 or MB-231 cultures were subsequently determined to investigate the effect of restoring physiologically relevant levels of ANKRD11 expression on the growth characteristics of breast cancer cells (supplementary material Fig. S6). The average number of colonies initiated on plastic by MCF-7 or MB-468 cultures stably expressing ANKRD11 was reduced by 76% or 92% respectively, when compared with the negative control cultures (MCF-7 or MB-468). Restoration of ANKRD11 expression in MB-231 cells resulted in a minimal inhibition (11% average reduction) of the clonogenic properties of this breast cancer cell line, supporting the notion that ANKRD11 suppresses the oncogenic phenotypes of these breast cancer cell lines by a p53-dependent mechanism. In addition, the effect of restoration of ANKRD11 expression in these breast cancer cell lines showed an average reduction in proliferation by 36%, 47% or 30% respectively (after 72 hours), when compared with the proliferation of negative control cultures (supplementary material Fig. S6). The findings that ANKRD11 expression is downregulated in breast cancer cell lines and that restoration of ANKRD11 expression suppresses the oncogenic characteristics of breast cancer cells suggest that ANKRD11 possesses characteristics consistent with those of a tumor suppressor.

To further investigate a role for ANKRD11 as a p53 coactivator, we silenced the expression of endogenous ANKRD11 using a previously validated short hairpin RNA (shRNA) (Zhang et al., 2007a). Stable expression of this shRNA in the MCF-10A breast epithelial cell line (MCF-10A-shANK) resulted in reduced levels of endogenous ANKRD11 mRNA and protein when compared with MCF-10A cells stably expressing scramble shRNA (MCF-10A-SCR) (Fig. 4D). Results show that shRNA-mediated silencing of ANKRD11 reduced the ability of p53 to activate p21^waf1 expression in response to DNA damage. These findings provide additional evidence to support a functional role for ANKRD11 as a p53 coactivator.

To gain further mechanistic insight into the role of ANKRD11 as a p53 coactivator, we endeavored to identify other known p53-regulatory proteins that may be recruited by ANKRD11. hADA3 was identified from a yeast two-hybrid screen as an ANKRD11-interacting protein, as determined by the β-galactosidase reporter assay (Fig. 5A). To confirm this interaction, a GST pull-down assay was conducted using in vitro translated hADA3. Results show that hADA3 interacted specifically with GST-ANKRD11^2369-2663aa, but not with GST alone (Fig. 5A). Recent studies have shown that hADA3 plays a key role in the regulation of p53 acetylation, stability and activity through the recruitment of acetyltransferase complexes such as p300 and P/CAF to p53 (Nag et al., 2007; Wang et al., 2001). The localization of P/CAF was subsequently determined in relation to ANKRD11 nuclear foci. Coimmunolocalization studies demonstrated that Flag-P/CAF protein is recruited to HA-ANKRD11-positive nuclear foci (Fig. 5A).

Expression of ANKRD11 was associated with increased acetylation of p53 at Lys320 in both MCF-7 and MB-468 stably expressing cell isolates (Fig. 5B). ANKRD11-mediated increase in Lys320-acetylated p53 levels was shown to occur solely through increased post-translation modification of p53, because exogenous ANKRD11 had no effect on total p53 protein levels in these cell lines (Fig. 5B). Exogenous expression of ANKRD11 in the MB-231 cell line had no effect on Lys320-acetylated p53 levels, which is consistent with our observation that ANKRD11 did not affect p21^waf1 expression in this cell line (Fig. 4B). Results from dual luciferase assays in H1299 (p53-null) cells show that the ANKRD11-mediated increase in p53-mediated transactivation of the CDKN1A promoter was attenuated upon mutation of p53 at Lys320 (p53-K320R) (Fig. 5C). Furthermore, expression of ANKRD11 had no effect on acetylation of p53 at Lys373, a lysine residue that is modified by the p300/CBP acetyltransferase (supplementary material Fig. S7). Taken together, these findings suggest that ANKRD11 enhances p53 transcriptional activity through increased P/CAF-mediated acetylation of p53.

The finding that restored ANKRD11 expression can enhance the transcriptional activity of the p53 through increased Lys320 acetylation prompted us to further investigate whether ANKRD11 influences the DNA-binding properties of p53. Results from promoter-binding assays show that both wild-type p53 (MCF-7 lysates) and mutant p53^R273H (MB-468 lysates) bind the p53-responsive CDKN1A promoter region and that this binding is enhanced following restoration of ANKRD11 expression in these cell lines (MCF-7-ANK-1 and MB-468-ANK-1) (Fig. 5D). Furthermore, the enhanced DNA-binding activity of p53 in MCF-7-ANK-1 cells was also associated with increased levels of Lys320 acetylation. Therefore, these findings suggest that ANKRD11 enhances the DNA-binding activity of p53 and implicates a possible role for ANKRD11 in the functional rescue of the p53^R273H mutant.

A putative p53-RE in the first intron of ANKRD11 was recently identified through two independent genome-wide chromatin immunoprecipitation (ChIP)-based studies (Hearnes et al., 2005; Wei et al., 2006). This region of ANKRD11 intronic sequence was cloned upstream of a luciferase reporter (pGL3-Basic-ANKRD11-p53-RE-Luc), and the dual-luciferase reporter system was used to investigate the transcriptional activity of this putative p53-RE located within the ANKRD11 intronic region. Increasing expression of myc-p53 led to a dose-dependent increase in luciferase expression from this reporter construct (Fig. 6A). Furthermore, p53-mediated activation of this region of ANKRD11 intronic sequence was ablated when the putative p53-RE was mutated (pGL3-Basic-ANKRD11-p53-RE-mut-Luc) (Fig. 6A).

To further investigate ANKRD11 as a p53 target gene, we monitored endogenous ANKRD11 expression levels during the p53 response evoked through treatment with a DNA-damaging agent. p53-dependent ANKRD11 transcription showed a concordant increase with p53 protein levels, increasing 8.2-fold 48 hours post treatment (Fig. 6B). In addition, endogenous ANKRD11 mRNA levels increased in a dose-dependent manner (P<0.005) in response to re-introduction of physiologically relevant levels of exogenous p53 in the HCT116 (TP53^–/–) derivative (Fig. 6C). These data, together with the previous ChIP-verified studies, suggest that ANKRD11 is a novel p53 target gene.

In this study, we have characterized ANKRD11 as a novel p53 coactivator. We found that ANKRD11 can enhance the transactivation of p53 target genes such as CDKN1A (p21^waf1) and that this transcriptional activation is dependent on p53 status. We have also shown that silencing of endogenous ANKRD11 expression reduces the transcriptional activity of p53. We show that ANKRD11-mediated enhancement of p53 activity occurs through increased p53 acetylation and DNA-binding activity.

Immunofluorescence studies have demonstrated that both endogenous and exogenous ANKRD11 localizes to nuclear foci, and both colocalization and cofractionation data indicate that ANKRD11 associates with p53, p14^ARF and PML within these nuclear domains (Fig. 1). Previous studies have also shown that these latter three proteins associate within similar domains, which were referred to as `extranucleolar inclusions' (Kashuba et al., 2003). Interestingly, the association between endogenous ANKRD11 and p53 in MCF-10A cells was enhanced in the presence of a DNA-damaging agent, suggesting that ANKRD11 may be involved in post-translational regulation of p53 activity during the p53 response. Our data suggest that ANKRD11 might have a role in the acetylation of p53 acetylation within these extranucleolar inclusions upon cellular stress. This notion is supported by previous findings whereby PML, another component protein recruited to extranucleolar inclusions, was also implicated in the regulation of p53 transcriptional activity (Guo et al., 2000). PML has been shown to interact directly with the DNA-binding domain of p53 and enhance p300/CBP-mediated acetylation of p53 within PML nuclear bodies (Guo et al., 2000).

This study uses the MCF-7, MB-468 and MB-231 breast cancer cell lines as models to demonstrate the biological function of ANKRD11 as a p53 coactivator. Although both MB-468 and MB-231 cell lines express mutated p53, the DNA-binding properties and transactivation potential of these mutants differ substantially. The p53^R280K `DNA-contact' mutant expressed in MB-231 cells is unable to bind DNA or transactivate p53 target genes (Park et al., 1994), making this an ideal negative control cell line for these studies. The observation that exogenous ANKRD11 had no significant effects on the expression of p21^waf1, FAS or NOXA in the MB-231 cell line provides evidence to suggest that the coactivator function of ANKRD11 was indeed p53 dependent (Fig. 4; supplementary material Fig. S6). The ANKRD11-mediated modulation of p21^waf1 was further confirmed to be p53 dependent through the use of studies involving p53 knockdown (Fig. 4C) and the expression of ANKRD11 in a p53-null cell line (supplementary material Fig. S5).

The p53^R273H `DNA-contact' mutant expressed in MB-468 cells retains the ability to interact with its cognate p53-DNA-response element in vivo (Park et al., 1994; Prasad and Church, 1997). This was in contradiction to an in vitro study suggesting that p53^R273H possessed no DNA binding affinity (Bullock et al., 2000). These discrepancies between in vivo and in vitro studies probably result from the use of recombinant, purified mutant p53 protein throughout the in vitro binding assays, suggesting that cellular post-translation modifications of p53 may influence its affinity for DNA. Our findings suggest that the restoration of mutant p53^R273H function may be possible through ANKRD11-mediated enhancement of Lys320 acetylation and subsequent increased DNA binding activity (Fig. 5).

p53^R273H is a potential target for restoration of mutant p53 function in cancer as it is the second most frequent cancer-associated mutation of p53, accounting for approximately 7.5% of entries in the IARC TP53 mutation database (www.iarc.fr/p53/index.html). Crystallographic studies have demonstrated that mutation of the R273 DNA contact residue significantly impairs the affinity of p53 for DNA, but has no effect on the thermodynamic stability or structural integrity of the DNA-binding surface of p53 (Ang et al., 2006). Therefore, it is suggested that only the enhancement of the DNA-binding properties of this mutant is required to restore the normal function of the p53^R273H mutant (Bullock and Fersht, 2001). However, recent in vitro data suggests that the DNA-binding affinity of p53^R273H is ∼700 to 1000 times weaker than that of wild-type p53, making the therapeutic revival of this mutant a challenging task (Ang et al., 2006).

The transcriptional regulatory domains of ANKRD11, including two intrinsic repressor domains (RD1: residues 318-611 and RD2: residues 2369-2663) and an activator domain (AD1: 2076-2145), have recently been defined (Zhang et al., 2007a). It is likely that P/CAF or other HATs are recruited to transcription factors by ANKRD11 through this AD1 domain, although this is yet to be formally demonstrated. It has previously been reported that ANKRD11 can function as a co-repressor of steroid nuclear receptor transactivation through the recruitment of HDACs via the RD2 domain of ANKRD11 to the p160 and nuclear receptor complex (Zhang et al., 2007a; Zhang et al., 2004). Furthermore, ANKRD11 subnuclear localization is regulated by both nuclear import and export signals, and is essential for ANKRD11 to negatively regulate transcription of steroid nuclear receptors (Zhang et al., 2007b). It is apparent that ANKRD11 can function as either a coactivator or corepressor of transcription, although the underlying mechanisms that govern this dual functionality are yet to be elucidated.

ANKRD11 was initially identified as a putative breast cancer tumor suppressor gene because of its location within the 16q24.3 breast cancer LOH region (Powell et al., 2002). A number of other tumor suppressors have been identified within this region, including CBFA2T3 (Kochetkova et al., 2002) and FBXO31 (Kumar et al., 2005). It is hypothesized that LOH of chromosome 16q during the onset of breast cancer results in the simultaneous loss or reduction in activity of several tumor suppressor genes, and this contributes to the early stages of tumorigenesis. A role for ANKRD11 as a breast cancer tumor suppressor was supported by the findings that ANKRD11 expression is downregulated in breast cancer cell lines, and restoration of ANKRD11 expression in MCF-7 and MB-468 cells suppressed their oncogenic growth characteristics (supplementary material Fig. S6). A concordant increase in p21^waf1 expression was also observed in response to restored ANKRD11 expression, suggesting that the anti-oncogenic effects of ANKRD11 occur through activation of downstream p53 pathways.

ANKRD11 itself is a novel p53 target gene (Fig. 6). A putative p53-RE was identified 50 kb from the 5′ end of the first intron of ANKRD11 by two independent ChIP-based genome wide screens mapping functional p53-REs (Hearnes et al., 2005; Wei et al., 2006). Our results have validated this p53-RE and shown that p53 does indeed modulate endogenous ANKRD11 transcript levels. Furthermore, endogenous ANKRD11 expression directly correlates with p53 status in breast cell lines (P<0.05) (Fig. 4A). Taken together, these results further support a role for ANKRD11 as a p53 target gene. Interestingly, PCAF itself was also recently identified as a p53 target gene in breast epithelial cell lines (Watts et al., 2004). The findings of both ANKRD11 and PCAF as p53 target genes suggest the existence of a positive feedback loop in the p53 acetylation pathway, although this remains to be investigated. Detailed investigations into the regulation of p53 activity by ANKRD11 or its interacting partners will significantly improve our understanding of the molecular pathways surrounding the modulation of p53 and will provide deeper insight into the development of improved pharmacological approaches to restore the function of mutant p53 in tumors.

HCT116 and its TP53^–/– derivative were supplied by B. Vogelstein (Bunz et al., 1998). All other cell lines were as previously reported (Kumar et al., 2005). Antibodies used were mouse α-p53 (Ab-2 and Ab-5), mouse α-p21^waf1 (Neomarkers, Fremont, CA); rabbit α-acetyl-p53 Lys320 (Upstate, Temecula, CA); rabbit α-acetyl-p53 Lys373 (Abcam, Cambridge, UK); mouse α-nucleophosmin, mouse α-Flag (Sigma-Aldrich, Saint Louis, Missouri); goat α-GST (Amersham Biosciences, Uppsala, Sweden); mouse α-GFP (Roche, Indianapolis, IN); rabbit α-PML and rabbit α-HA (Santa Cruz Biotechnology, Santa Cruz, CA). A rabbit α-ANKRD11 polyclonal antibody was raised against a synthetic peptide representing residues 326-341 of the ANKRD11 sequence, conjugated to KLH and affinity-purified on Sepharose beads coupled with the same peptide. Other antibodies were as previously reported (Kumar et al., 2006; Kumar et al., 2005).

C-terminal myc-tagged ANKRD11 open reading frame (ORF) was cloned into pLNCX2 retroviral vector (Clontech Laboratories, Mountain View, CA) to generate pLNCX2-ANKRD11-myc using standard procedures. PCR amplified EGFP ORF from pEGFP-C1 (Clontech Laboratories) was inserted in-frame within pLNCX2-ANKRD11-myc to generate a construct expressing the GFP-ANKRD11-myc fusion protein. To generate epitope-tagged ANKRD11 truncated constructs, appropriate regions of ANKRD11 were PCR amplified from pLNCX2-ANKRD11-myc and cloned in-frame into the mammalian expression vector pCMV-Tag2 (FLAG-ANKRD11^1-817aa, ANKRD11^144–313aa, ANKRD11^816–1802aa, ANKRD11^{1803–2203aa} and ANKRD11^2352-2663aa), pCMV-myc (myc-ANKRD11^144–313aa) or pGEX-5X-1 (GST-ANKRD11^144–313aa). The pGBT-ANKRD11^2369-2663aa construct was constructed by subcloning the ANKRD11 C-terminal coding region from pACT2-ANKRD11^2369-2663aa plasmid (Zhang et al., 2004) into pGBT9 vector at BamHI and XhoI sites. Protein coding ORFs for p14^ARF or p53 were PCR amplified from breast cDNA and cloned in-frame with the indicated tags into the mammalian expression vectors pEGFP-C1 (GFP-p14^ARF), pLNCX2 (myc-p53) or pMAL-c2x (MBP-p53). The p53-K320R mutant was generated from the p53-K320/373/381/382R construct supplied by Shelly Berger (Barlev et al., 2001). The ANKRD11 shRNA expression vector (pMSCV-shANK) was generated through cloning of a previously reported ANKRD11-specific oligonucleotide sequence (Zhang et al., 2007a) or scrambled oligonucleotide into a modified pMSCV retroviral vector (also expressing EGFP and puromycin resistance) downstream of a H1 promoter. Reporter constructs pGL2-Promoter-p53-RE-Luc [17 tandem p53 response elements (REs)], pGL3-Basic-CDKN1A-pro-Luc and pGL2-Basic-MDM2-pro-Luc [CDKN1A and MDM2 promoter regions containing multiple endogenous p53-RE (el-Deiry et al., 1993; Juven et al., 1993)] were kind gifts from Moshe Oren (Weisman Institute of Science, Rehovot, Israel). The CDKN1A promoter construct region was generated through PCR amplification of the CDKN1A promoter region from pGL3-Basic-CDKN1A-pro-Luc construct using biotinylated primers. To generate the pGL3-Basic-ANKRD11-p53-RE-Luc reporter construct, a region from the first intron of the ANKRD11 sequence (accession no. NM_013275) carrying the putative p53-binding site was PCR amplified from human genomic DNA using the forward: 5′-CAAAGCCACCAGACCTCCGTTC-3′ and reverse: 5′-CAGCAGAACCTTGCTGTGCGTGTC-3′ primers, and cloned into the pGL3-Basic-Luc reporter vector (Promega, Madison, WI). The putative p53-RE (GAACATGCCAGGTCATGTCT) within this ANKRD11 promoter region was mutated to GCTTGCAAAAGCTTGCAAAA using overlap PCR amplification to generate the pGL3-Basic-ANKRD11-p53-RE-mut-Luc reporter construct. pACT-hADA3 was isolated from a human placental cDNA library in the yeast two-hybrid screen, and it encodes full-length hADA3. pCMXHA-hADA3f was generated by subcloning full-length hADA3 coding region (residues 1-432) into pCMXHA vector at Asp718 and NheI sites. pGBT-RAC3-N (1-408) (Wu et al., 2001), GST-ANKRD11^2369-2663aa and HA-ANKRD11 have been described (Zhang et al., 2004). pAB-FLAG-hP/CAF was kindly provided by Bert O'Malley, Baylor College of Medicine. The sequences of all constructs were confirmed by DNA sequencing.

In silico analysis of the predicted ANKRD11 amino acid sequence was performed using blastp (NCBI) and ClustalW (EBI) programs, with ANKRD11 (NM_013275), BARD1 (NM_000465), IκBα (NM_020529) and 53bp2 (NM_001031685) sequences derived from GenBank, NCBI.

Cells were seeded in Lab-Tek II Chamber Slides (Nalge Nunc, Naperville, IL), transiently transfected with the indicated plasmids using Lipofectamine 2000 (Invitrogen, Carlsbad, CA) or treated with mitomycin C (20 μg/ml). Twenty four hours post transfection, cells were fixed in 2% paraformaldehyde for 15 minutes at room temperature (RT) and permeabilized in 0.4% Triton X-100 (15 minutes, RT). Cells were incubated with the indicated primary antibody in 5% donkey serum (overnight, 4°C), followed by incubation with the appropriate Alexa-Fluor-488-, Alexa-Fluor-594-, Rhodamine- or Fluorescein-conjugated antibodies (Molecular Probes, OR) in 5% donkey serum (1 hour, RT) and mounted in Vectashield mounting medium with DAPI (Vector Laboratories, Burlingame, CA). Cells were imaged using either Bio-Rad Radiance 2100 confocal or Olympus IX70 inverted microscopes.

MCF-7 cells (1×10⁸) stably expressing the GFP-ANKRD11-myc protein were lysed in buffer containing 50 mM Tris-HCl pH 7.5, 250 mM NaCl, 0.1% Triton-X-100, 1 mM EDTA, 50 mM NaF, 1 mM Na₃VO₄ with 1× complete protease inhibitor cocktail (Roche, Indianapolis, IN), sonicated and centrifuged. The clarified protein lysate was loaded on a 10-40% continuous glycerol gradient prepared in 20 mM Tris-HCl pH 8.0, 5 mM MgCl₂, 100 mM KCl, 0.1% NP-40 and centrifuged overnight at 50,000 r.p.m. in an MLS-50 swing-out rotor (Beckman Coulter, Fullerton, CA). Protein complexes from 24 gradient fractions were analyzed by western blot analysis using α-myc, α-p53 and α-PML antibodies.

Nuclei were harvested from 5×10⁷ MCF-7 or HEK293T cells as described previously (Wysocka et al., 2001), resuspended in 50 mM Tris-HCl, pH 8.0, 150 mM NaCl, 1% Triton-X-100 and 1× complete protease inhibitor cocktail, sonicated and centrifuged. Clarified lysates were pre-cleared with pre-immune serum from the rabbit used to generate the α-ANKRD11 polyclonal antibody (1 hour, 4°C), incubated with Protein-A-conjugated Sepharose beads (Amersham Biosciences, Uppsala, Sweden) (1 hour, 4°C) and immunoprecipitated using the α-ANKRD11 antibody. Inputs and coimmunoprecipitated protein complexes were subjected to western blot analysis as previously described (Kumar et al., 2005). Coimmunoprecipitation of endogenous p53 with protein fragments of ANKRD11 was performed as previously described (Kumar et al., 2005).

For in vitro binding assays, MBP (maltose-binding protein), MBP-p53, GST (glutathione S-transferase), GST-ANKRD11^144-313aa or GST-ANKRD11^2369-2663aa proteins were induced in BL21pLysS bacteria. GST, GST-ANKRD11^144-313aa or GST-ANKRD11^2369-2663aa fusion proteins were purified using glutathione Sepharose 4B beads (Amersham Pharmacia Biotech) using a standard protocol. MBP or MBP-p53 fusion proteins were associated with the amylose resin for use as bait in the in vitro binding assay. ³⁵S-labeled hADA3 was synthesized from pCMXHA-hADA3f by in vitro transcription/translation reactions using T7-Quick reticulocyte lysate (Promega). To confirm that proteins were intact, they were resolved by SDS-PAGE and visualized by colloidal Coomassie Blue staining. For MBP pull-down experiments, purified GST or GST-ANKRD11^144-313aa fusion proteins were incubated with either MBP- or MBP-p53-amylose beads in TNME buffer (20 mM Tris-HCl pH 8.0, 100 mM NaCl, 5 mM MgCl₂, 1 mM EDTA, 0.1% NP-40, 20% glycerol at 4°C), washed in TNME buffer, eluted with 10 mM maltose in 20 mM Tris-HCl pH 7.5, 200 mM NaCl, 1 mM EDTA, 5% glycerol and detected by western blot analysis. GST pull-down assays were performed as previously described (Zhang et al., 2004).

For promoter binding assays, a biotinylated p53-responsive CDKN1A promoter region was immobilized to streptavidin-M280 magnetic beads (Invitrogen, Oslo, Norway) following the manufacturer's protocol. Promoter binding assays were performed as previously described (Goardon et al., 2006). MDA-MB-468 (5×10⁶) cells were lysed in binding buffer (20 mM Tris-HCl pH 8.0, 6 mM MgCl₂, 5 mM DTT, 0.1 mM EDTA, 0.01% NP-40, 10% glycerol), sonicated and centrifuged. Clarified lysates were incubated (1 hour, RT) with either CDKN1A-charged or uncharged beads, washed in binding buffer containing 250 mM NaCl and eluted in 1× SDS loading buffer.

HeLa or Saos-2 cells (2×10⁵) were seeded in 24-well plates and transfected with 100 ng reporter plasmid with varying amounts of myc-p53 or ANKRD11-myc and 25 ng pRL-TK plasmid (Promega) as a transfection control. Empty vector was added to compensate for amounts of plasmid used in various treatments. Dual-reporter assays were performed as described previously (Kumar et al., 2006).

ANKRD11 (forward: 5′-AAGGAGCTGTTCAGGCAGCAGGAG-3′ and reverse: 5′-AGTCGTCGTTGACGTCGACCATG-3′ primers), CDKN1A (p21^waf1) (forward: 5′-TGGACCTGGAGACTCTCAGGGTCG-3′ and reverse: 5′-TTAGGGCTTCCTCTTGGAGAAGATC-3′ primers), FAS (forward: 5′-ATGCTGGGCATCTGGACCCT-3′ and reverse: 5′-GCCATGTCCTTCATCACACAA-3′ primers) and NOXA (forward: 5′-AGAGCTGGAAGTCGAGTGT and reverse: 5′-GCACCTTCACATTCCTCTC-3′ primers) expression were determined by real-time RT-PCR as previously described (Kumar et al., 2005). The significance of the downregulation of average ANKRD11 expression in breast cancer cell lines when compared with the average ANKRD11 expression in finite life-span HMECs and nonmalignant immortalized breast epithelial cells was determined using analysis of variance (ANOVA).

Generation of amphitrophic recombinant retroviruses was performed as previously described (Kumar et al., 2005). Stable cell isolates were established by transduction of MCF-7, MDA-MB-468 (MB-468), MDA-MB-231 (MB-231) or Saos-2 cell lines with retroviral particles derived from either pLNCX2-GFP-ANKRD11 or pLNCX2 vector (negative control). The clonogenicity of these stable cell isolates was determined through scoring the number of colonies initiated following the growth of low-density (1000 cells per well) cultures for a further 10-14 days in the presence of geneticin (600 μg/ml). Cell proliferation of these stable clones was performed by plating cells at 10-20% confluence in 96-well plates and was assayed using the CellTitre-Glo Luminescent Cell Viability Assay (Promega). Knockdown of p53 expression in MCF-7 stable cell isolate was achieved through transfection with p53-specific siRNA (CGGCAUGAACCGGAGGCCCAU) or scrambled siRNA at 100 nM using lipitoid transfection reagent according to the manufacturer's protocol (Utku et al., 2006). The shRNA-mediated silencing of ANKRD11 was performed through generation of stable cell isolates established by transduction of MCF-10A cells with retroviral particles derived from either pMSCV-shANK of pMSCV-SCR (scrambled control).

The pGBT-ANKRD11^2369-2663aa plasmid encoding a GAL4 DBD-ANKRD11 C-terminal domain (residues 2369-2663) fusion protein was used as bait to screen a human placental yeast two-hybrid cDNA library. Yeast Y190 strain was sequentially transformed with the bait and library plasmids following the manufacturer's protocols (Clontech Laboratories). Yeast colonies that survived on synthetic conditional plates supplemented with 50 mM 3-aminotriazole (Sigma-Aldrich) were then tested for β-galactosidase expression (Zhang et al., 2004). Library plasmids from double reporter positive colonies were rescued and reconfirmed by co-transformation with the bait plasmid.

We wish to thank Moshe Oren, Shelley Berger and Bert O'Malley for providing constructs; Martha Stampfer and Bert Vogelstein for providing cell lines; and Ghafar Sarvestani for assistance with the confocal microscopy. National Health and Medical Research Council of Australia grant 207703 (D.C.) and Faculty of Health Sciences and Department of Medicine, University of Adelaide and the Royal Adelaide Hospital. We also thank the Hanson Institute for generous financial support.