Background
p63 is the founding member of the p53 protein family, and is required for the development of limbs and epithelial structures in vertebrates [
1]. The p63 gene expresses at least 6 common transcripts by utilising two distinct promoters (TA and ΔN) and alternative splicing within the 3' end of mRNA that generates α,β and γ isoforms [
2]. TAp63 variants contain a p53-like TA1 transactivation domain. ΔNp63 variants lack a TA1 domain, but instead contain a unique 14 amino acid sequence that contributes to the formation of an alternative TA2 transactivation domain [
3]. All p63 variants contain a DNA-binding domain and a tetramerisation domain with homology to p53. However, p63 alpha isoforms encode a C-terminal extension containing a SAM protein interaction domain, a conserved functional element found in a range of developmental proteins [
4].
Initial studies identified p63 as a robust biomarker for epithelial progenitor, or stem, cells [
5]. However, the development of TA- and ΔN-isotype specific reagents revealed that ΔNp63 expression is confined to the basal layer of stratified squamous epithelium, whereas TAp63 variants predominate in suprabasal layers [
6]. Similarly,
in vitro keratinocyte differentiation induces hypoexpression of the predominant ΔNp63α isoform [
7]. TAp63 isotypes can transcriptionally activate a subset of p53 target genes involved in cell cycle checkpoint control and apoptosis [
8,
9]. In contrast, initial reports suggested that ΔNp63 variants had no intrinsic transcriptional activity, but could antagonise TAp63- and p53-dependent target gene transcription [
2]. However, recent microarray-based screening approaches have identified the transcriptional targets of distinct p63 isotypes in tumour cells and in immortalised keratinocytes [
10]. These studies have revealed that ΔNp63α can either activate or repress the transcription of many target genes involved in multiple cellular processes. The challenge now is to dissect how specific validated ΔNp63α transcriptional targets mediate ΔNp63α physiological function. For instance, loss of ΔNp63α-dependent transcriptional repression of S100A2, p21
WAF1 and 14-3-3 correlates with ΔNp63α downregulation during keratinocyte differentiation [
7,
11].
Our previous studies revealed that UV damage-induced p53 phosphorylation is restricted to the ΔNp63α-positive basal epidermal layer of UV-damaged human skin [
12], which provided an opportunity to identify novel physiological regulators of the p53 damage response. Site-specific p53 phosphorylation has already been established to play an important role in regulating the p53 response to UV-damage. For example, p53 mutation at the conserved UV-inducible CK2-site sensitizes mice to UV-induced skin cancer and attenuates the p53 transcription programme in MEFs [
13]. In this study we show that a positive association between UV-induced p53 phosphorylation in ΔNp63α-positive immortalised keratinocytes is explained by ΔNp63α-dependent transcriptional control of the ATM kinase.
Discussion
p53 is the primary mediator responsible for removing DNA damaged epidermal cells [
26], and p53 phosphorylation at the CK2-site is required to suppress UV-induced skin cancer development in mice [
13]. We previously reported the striking confinement of UV-induced p53 phosphorylation at the key damage-response CK2 and ATM sites to ΔNp63α-positive basal skin cells, despite substantial p53 stabilization throughout the epidermis [
27]. We next aimed to identify novel factors that control damage-induced p53 phosphorylation in a keratinocyte model system, and discovered that the epithelial stem cell marker ΔNp63α is a novel ATM regulator that controls p53 Serine-15 phosphorylation through transcription of the ATM kinase. Loss of ΔNp63α by RNAi or differentiation reduced ATM-dependent phosphorylation and conversely, ΔNp63α overexpression stimulated ATM signaling. A recent genome-wide screen reporting that ATM expression is reduced by 30-60% in p63 siRNA-treated epithelial cell lines supports our finding [
28].
Post-translational activation of the ATM kinase by ionizing radiation, oxidative stress, chemotherapeutic drugs [
15] and UV radiation is well-established [
29]. However, ATM transcriptional regulation has been shown to occur both
in vitro and
in vivo [
30]. E2F-1 stimulation of ATM transcription [
21] has been implicated in oncogene-mediated p53 activation [
31]. In contrast, epidermal growth factor sensitizes cells to ionizing radiation through Sp1-mediated repression of ATM transcription [
32]. We have shown that p63 ΔN-isotypes are novel positive regulators of ATM transcription that interact with the promoter CCAAT sequence. p63-dependent gene regulation has been reported to occur through interaction of the DB domain with a p53 RE [
22]. However, the lack of similarity of the CCAAT sequence to classical p53 REs suggests that p63 interaction with a CCAAT element is indirect, and requires a CCAAT-binding mediator. We show that the E2F-1 regulates ATM transcription through the same CCAAT sequence, not a canonical E2F-1 response element, suggesting that a CCAAT-binding cofactor integrates activation signals from diverse ATM transcriptional regulators. CCAAT-binding proteins include NF-1/CTF [
33], NF-Y [
34] and C/EBP [
35]. NF-Y can mediate ΔNp63α-dependent transcription in human keratinocytes [
10,
36], p53-dependent repression of cell cycle genes [
37], and transcriptional activation by p53 gain-of-function mutants [
38]. However, we found that coexpression of the NF-YA isoform inhibits ΔNp63α stimulation of the ATM reporter (data not shown), presumably by displacing the unidentified ΔNp63α coactivator from the ATM promoter CCAAT element. Ongoing studies aim to further delineate the mechanism of ΔNp63α-mediated ATM transcriptional control by identifying ΔNp63α binding partners in epithelial cells.
Based on our findings so far, cooperation of three distinct functional domains is required to mediate p63-dependent ATM transcription. We found that p63 ΔN-isotypes transcriptionally activate the ATM gene, whereas TA-isotypes do not, highlighting an essential role for the TA2 transactivation domain in mediating ΔNp63α function. Future studies will aim to determine which cofactors are recruited to this region, and whether their access is controlled by TA2 domain post-translational modification, similar to the p53 model [
39]. There was also a requirement for an intact p63 DB domain, despite the absence of a canonical p53 RE in the ATM promoter. However, in addition to providing a surface for the sequence-specific binding of DNA, the p53 DB domain modulates p53 function by providing a contact interface for regulatory proteins such as ASPP1, Mdm2, and DAPK superfamily kinases [
40,
41], and the high degree of conservation of the p63 DB domain suggests that a similar interface exists on p63. Finally, the p63 SAM domain forms a binding site for NF-Y [
36], and SAM domain disease-associated mutants have decreased transcriptional repressor and activator function [
7,
42]. We found that AEC point mutations within the SAM domain [
25] inhibited ΔNp63α-stimulated ATM transcription and ATM-dependent p53 phosphorylation, indicating that this domain may be essential for cofactor recruitment by the ΔNp63α. Interestingly, the AEC clinical phenotype predominantly involves skin defects without associated limb abnormalities [
42], consistent with a skin-specific role for ΔNp63α-ATM-p53 signaling in mediating normal ectodermal development. Therefore, the coordinated assembly of several cofactors may be required for fully functional p63 transcriptional machinery.
According to our model, elevated ΔNp63α-dependent ATM transcription primes p53 leading to damage-sensitivity in epithelial stem cells. Loss of p63-ATM-p53 pathway function will compromise epithelial stem cell function and promote premature ageing or skin carcinogenesis. Interestingly, transgenic mice with a specific p63-deficiency in the epithelium show increased senescence and an accelerated ageing phenotype [
43]. Although transgenic mice lacking the Serine-18 (equivalent to human Serine-15) ATM phosphorylation site are not cancer-prone [
44], it is now important to determine whether mutation at p53 Serine-18 enhances sensitivity to UV-induced skin tumorigenesis, similar to mutation of the CK2-site. Interestingly, p53S18A/S23A (ATM-/CHK2-sites) double mutant mice develop a spectrum of spontaneous tumours distinct from p53S23A and p53-null mice, and show accelerated skin ageing phenotypes when crossed into a repair-deficient background [
45].
Further, activation of the ATM-CHK2 pathway during early tumorigenesis has been reported to provide a selective pressure for p53 mutation [
46]. The discovery that the ΔNp63 promoter is subject to both p53-mediated activation and repression by ΔNp63α [
17], and that ATM-dependent phosphorylation mediates ΔNp63α degradation [
47] suggests that activity of the damage-response ΔNp63α-ATM-p53 pathway is finely modulated by complex feedback mechanisms. Further dissection of this pathway should provide molecular targets for combating cancer and ageing.
Materials and methods
Cell Treatments
HaCat and Saos-2 cells were maintained in DMEM supplemented with 10% FCS. H1299 cells were maintained in RPMI supplemented with 10% FCS. p63 expression plasmids were obtained from Dr Karin Nylander, and transient transfections were done using lipofectamine LF2000 (Invitrogen). Ambion silencerTM siRNA oligonucleotides were used to block ATM expression: sense 5'-gccagcaaauucuagugcctt -3' antisense: 5'-ggcacuagaauuugcuggctc-3'. Transfection of HaCaT cells with 200 pMol ATM siRNA used the siPORTTM NeoFXTM transfection reagent. Dharmacon ON-TARGETplus SMARTpool p63 siRNA was used to knockdown p63 expression using the DharmaFECT 1 transfection reagent. ON-TARGETplus siCONTROL Non-targeting pool was used for control transfections. pSUPER-p63si stable transfections were done as previously described [
17].
Site-directed mutagenesis
ΔNp63α site-directed mutagenesis used the QuikChange® Site-Directed Mutagenesis Kit (Stratagene) and the following primers:
N6H-For 5'-TTGTGAAATGGTGCCCTAACCATGAGCTGAGCCGTG-3'; N6H-Rev 5'-AATTGAGTCTGGGCATTGTGTTCCAGGTACAAC-3'; G76W-For 5'-GTACACGAACCTGTGGCTCCTGAACAGCATGG-3'; G76W-Rev 5'-CCATGCTGTTCAGGAGCCACAGGTTCGTGTAC-3'; R204W-For 5'-TTGTGAAATGGTGCCCTAACCATGAGCTGAGCCGTG-3'; R204W-Rev 5'-CACGGCTCAGCTCATGGTTAGGGCACCATTTCACAA-3'; R279H-For 5'-GCTGCGTCGGAGGAATGAACCATCGTCCAATTTTAATC-3'; R279H-Rev 5'-GATTAAAATTGGACGATGGTTCATTCCTCCGACGCAGC-3'; R298Q-For 5'-CAAGTCCTGGGCCAACGCTGCTTTG-3'; R298Q-Rev 5'-CAAAGCAGCGTTGGCCCAGGACTTG-3'; C522W-For 5'-GTTGGGCTGTTCATCATGGCTGGACTATTTCACGAC-3' C522W-Rev 5'-GTCGTGAAATAGTCCAGCCATGATGAACAGCCCAAC-3'; I537T-For 5'-GACCACCATCTATCAGACTGAGCATTACTCCATG-3'; I537T-Rev 5'-CATGGAGTAATGCTCAGTCTGATAGATGGTGGTC-3'; CUT1-For 5'-GGCCTCGAGCCACAGTACACGAACCT-3'; CUT1-Rev 5'-ACCTCTAGATCATTCTCCTTCC-3'; CUT2-For1 5'-GGCCTCGAGGACCAGCAGATTCAGAAC-3'; CUT2-For2 5'-GGCCTCGAGTTGTACCTGGAAAACAATGCCCAGACTCAATTTAGTGAGGACCAGCAGATTCAGAAC-3'; CUT2-Rev 5'-ACCTCTAGATCATTCTCCTTCC-3'; TAN-For 5'-GGCCTCGAGTGTATCCGCATGCAAGACTCAGACCTCAGTGACCCCATGTGGCCACAGTACACGAACCT-3'; TAN-Rev 5'-ACCTCTAGATCATTCTCCTTCC-3'.
Immunoblotting
Immunoblotting was done essentially as described previously [
40]. p53 protein was detected using DO-1 and DO-12 anti-p53 antibodies, specific p53 Serine-15 phosphorylation was detected using p53 phosphoSerine-15 antibodies (New England Biolabs), and all p63 isoforms were detected using the 4A4 antibody (Abcam). Anti-ATM (5C2, GeneTex) and anti-ATM phosphoSerine-1981 (clone 10H11.E12, Upstate) antibodies were used.
Reporter Assays
Wild-type and mutant human ATMpLUC reporter plasmids [
23] and the Arf exon1 βpLUC reporter plasmid [
48] were previously described. 1 μg of expression plasmid, 1 μg of reporter plasmid and 0.2 μg of pRL-CMV plasmid were cotransfected into H1299 cells using lipofectamine, and cells were harvested after 24 hours. Reporter activity was determined using the Dual-Luciferase reporter assay kit (Promega).
H1299 cells were transfected with 1 μg of p63 expression plasmid, and selected using 1 μg/ml geneticin (Invitrogen). After 14 days, colonies were stained with Giemsa and counted.
Real-Time PCR
Total mRNA was extracted using the Qiagen RNeasy Kit, and 40ng samples were analysed by real-time RT-PCR using Quantitect® SYBR® Green detection. RT-PCR conditions were: 50°C for 30 min, 95°C, 15 min, and 44 cycles of 95°C for 15 sec, 55°C for 30 sec, 72°C for 45 sec. Melting curves were recorded from 60°C to 95°C. Primers were: p63-For 5'-GGAAAACAATGCCCAGACTC-3'; p63-Rev 5'-GCTGTTCCCCTCTACTCGAA-3'; ATM-For 5'-CCAGGCAGGAATCATTCAG-3'; ATM-Rev 5'-CAATCCTTTTAAATAGACGGAAAGAA-3'; Actin-For 5'-CTACGTCGCCCTGGACTTCGAGC-3'; Actin-Rev 5'-GATGGAGCCGCCGATCCACACGG-3'.
Chromatin Immunoprecipitation Assays
4 μg of 4A4 (Abcam) antibody was used to immunoprecipitate p63-DNA complexes, and 4 μg of KH95 antibody (Santa Cruz) was used to immunoprecipitate E2F-1-DNA complexes. 2-5 μl purified DNA was analysed by real-time PCR, and input DNA dilutions are indicated in figure legends. Primers used were:
ATM For 5'-AAAACCACAGCAGGAACCAC-3'; ATM Rev 5'-TCCAAGTCTGAGGACGGAAG-3'; GAPDH For 5'-AAAAGCGGGGAGAAAGTAGG-3'; GAPDH Rev 5'-CTAGCCTCCCGGGTTTCTCT-3'. The programme used was: 95°C, 15 min, then 40 cycles of 95°C, 15 sec, 56°C, 30 sec, 72°C, 30 sec, and product melting curve was read from 60°C to 95°C at 1°C intervals.
Reporter ChIP
10 cm culture dish of H1299 cells were transiently transfected with 6.7 μg ΔNp63α or HA-E2F-1 expression plasmids 1.67 μg pGL3-basic or ATMpLUC and 1.67 μg pRL-CMV. Cells were crosslinked after 24 hrs and processed as outlined above.
Competing interests
The authors declare that they have no competing interests.
Authors' contributions
Study concept and design: ALC, JH, LE; acquisition of data: ALC, JH, LF, MN, NG, JD, GS; analysis and interpretation of data: ALC, JH, LF, MN, NG, JD, GS; drafting of the manuscript: ALC, JH, LF; critical revision of the manuscript for important intellectual content ALC, TRH, BV; statistical analysis: RH; study supervision: TRH, BV. All authors have read and approved the final manuscript.