ATM kinase is a master switch for the ΔNp63α phosphorylation/degradation in human head and neck squamous cell carcinoma cells upon DNA damage

Abstract

We previously found that the pro-apoptotic DNA damaging agent, cisplatin, mediated the proteasome-dependent degradation of ΔNp63α associated with its increased phosphorylated status. Since ΔNp63α usually plays an opposite role to p53 and TAp63 in human cancers, we tested the notion that phosphorylation events induced by DNA damage would affect the protein degradation of ΔNp63α in HNSCC cells upon cisplatin exposure. We found that ΔNp63α is phosphorylated in the time-dependent fashion at the following positions: S385, T397 and S466, which were surrounded by recognition motifs for ATM, CDK2 and p70s6K kinases, respectively. We showed that chemical agents or siRNA inhibiting the activity of ATM, CDK2 and p70s6K kinases blocked degradation of ΔNp63α in HNSCC cells after cisplatin exposure. Site-specific mutagenesis of ΔNp63α residues targeted for phosphorylation by ATM, CDK2 or p70s6k led to dramatic modulation of ΔNp63α degradation. Finally, we demonstrated that the ΔNp63α protein is a target for direct in vitro phosphorylation by ATM, CDK2 or p70s6K. Our results implicate specific kinases, and target phosphorylation sites in the degradation of ΔNp63α following DNA damage.

Introduction

Although p63 displays a similar modular structure and extensive homology to that of p53, it encodes at least six different protein isotypes resulting from alternative splicing.^64,76 P63 isotypes consist of two groups, including those with the transactivation (TA)-domain at the amino terminus (TA-) or those without it (ΔN-). Similarly to p53, TAp63 isotypes bind specific DNA sequences activating transcription from p53-responsive genes that induce cell cycle arrest or apoptosis.^{13,22,23,26,27,38,45,46,53,67,73} ΔNp63α is the most abundant p63 isotype in proliferating basal layers of many epithelial tissues and is highly expressed in various human cancers (e.g., head and neck squamous cell carcinomas, HNSCC), supporting its role in cell proliferation and neoplastic development.^{42,61,74,76,79,82,83} Since ΔNp63 proteins retain the core DNA-binding domain, simple competition for DNA sites might prevent p53 or TAp63 from binding target gene promoters.^42,71,74,76 ΔNp63α was also shown to play a more proactive role as a transcriptional regulator by activating or downregulating several critical downstream gene targets.^42,61,74 Moreover, ΔNp63α was found to induce accumulation and signaling of β-catenin supporting an oncogenic role for ΔNp63 in HNSCC cells.⁶¹

Previous studies showed that genotoxic stress agents including ultraviolet (UV)-irradiation, actinomycin D, bleomycin and etoposide led to accumulation of TAp63α and TAp63gamma proteins, and TAp63 proteins were degraded via a proteasome complex under normal cellular circumstances.^40,56 In addition, TAp63 was shown to activate p53-responsive genes involved in regulation of cell cycle arrest and apoptosis in response to UV-irradiation, or actinomycin D.^40,56 However, the effect of genotoxic stress on ΔNp63 isotypes in most cancers, including head and neck cancers has not been well studied. A few reports showed that UV-irradiation decreased levels of ΔNp63 in normal keratinocyte cultures.^46,79,80

Apoptosis is a common feature of the cytotoxicity caused by DNA damaging anti-cancer agents.^{1,2,4,7,18,29,34,43,58,59,63,70} The overexpression of ΔNp63α was shown to an increase in cell proliferation, and enhanced tumor growth in vitro and in vivo and also to block UV-induced p53-dependent apoptosis in vivo.^42,46 However, siRNA silencing of ΔNp63α inhibited proliferation and mediated the programmed cell death of HNSCC cells supporting the notion that ΔNp63α functions as an anti-apoptotic factor.^13,45,67

We previously found that cisplatin dramatically decreased the survival of HNSCC cells and also led to a proteasome-dependent degradation of ΔNp63α^25,37,85 We further found that cisplatin exposure decreased protein levels of ΔNp63α in HNSCC cells, while increasing overall phosphorylation levels of ΔNp63α based on in vivo cell labeling and treatment with a serine/threonine kinase inhibitor.^25,37 This degradation of ΔNp63α was also enhanced by the physical interaction of ΔNp63α and stratifin, which preferentially binds to phosphorylated protein targets, such as ΔNp63α^25,80

It is well known that p53 is phosphorylated upon DNA damage at the multiple sites, in both the N-termini and C-termini.^1,2,18,72 Although stress-related phosphorylation of the p53 tumor suppressor at the N-terminus leads to its stabilization, there are a few reports showing that p53 phosphorylation occurring at the C-terminus might lead to its subsequent degradation.¹⁴ However, little is known about p63 phosphorylation events, especially after genotoxic stress. Since ΔNp63α usually plays an opposite role in human cancers to p53 and TAp63, we hypothesized that this protein might react differently to DNA damage. In this scenario, the cisplatin-induced degradation of ΔNp63α in HNSCC cells might occur through its phosphorylation rather than dephosphorylation. Earlier reports suggested that p63 is phosphorylated under unstressed conditions and after exposure of epithelial cells to cisplatin or UV-irradiation.^25,37,79,80 Therefore, we examined the type and consequences of phosphorylation events induced by DNA damage on ΔNp63α levels in HNSCC cells after cisplatin exposure.

Results

To examine whether ΔNp63α, the predominant p63 isotype in HNSCC cells (reviewed in refs. ^{42, 60, 61} and ⁶⁵) is modified after exposure to cisplatin and to track down changes in the posttranslational modification status of ΔNp63α, we employed a combination of immunoprecipitation with Ab-1 antibody against ΔNp63 and subsequent 2D-gel separation of target proteins followed by mass spectrometry analysis.^25,37

We previously found that endogenous ΔNp63α produced multiple modified p63 proteins in HaCaT cells shown by subsequent immunoprecipitation (with the Ab-1 antibody) and 2D-gel isolectrofocusing/SDS-electrophoresis.³⁷ This observation suggested that ΔNp63α is phosphorylated and degraded in HNSCC cell lines exposed to genotoxic stress agents (cisplatin, UV) supporting our previous findings and those of other groups.^{25,37,79 ,80} The initial bioinformatics searches performed by us (www.cbs.dtu.dk/services/NetPhosK/, see, Table 1) and others (reviewed in refs. ^{20, 21, 31, 41} and ⁵⁷) suggested that the ΔNp63α protein might be phosphorylated at the multiple motifs by a variety of protein kinases (e.g., ATM, CDK, p70s6K, MAPK and protein C kinase).

Table 1.

Predicted phosphorylation sites in ΔNp63α

(A) Potentially phosphorylated serine residues
Position	Sequence	Score	Predictions
36	QNGSSSTSP	0.928	*S*
53	VTAPSPYAQ	0.994	*S*
60	AQPSSTFDA	0.958	*S*
73	PAIPSNTDY	0.911	*S*
95	STAKSATWT	0.981	*S*
156	NHELSREFN	0.987	*S*
191	TGRQSVLVP	0.878	*S*
264	ADEDSIRKQ	0.920	*S*
273	QVSDSTKNG	0.997	*S*
295	IQMTSIKKR	0.826	*S*
301	KKRRSPDDE	0.998	*S*
364	QSPSSYGNS	0.996	*S*
369	YGNSSPPLN	0.800	*S*
385	NKLPSVSQL	0.866	*S*
466	LGCSSCLDY	0.925	*S*
492	DDLASLKIP	0.871	*S*
533	ASTVSVGSS	0.961	*S*
536	VSVGSSETR	0.865	*S*
537	SVGSSETRG	0.979	*S*

(B) Potentially phosphorylated threonine residues
Position	Sequence	Score	Prediction
187	VEDPITGRQ	0.987	*T*
207	EFTTYLYNF	0.977	*T*
316	RGRETYEML	0.862	*T*
397	RNALTPTTI	0.927	*T*
428	GLSPTQALP	0.913	*T*
551	AVRFTLRQT	0.952	*T*

(C) Potentially phosphorylated tyrosine residues
Position	Sequence	Score	Prediction
17	SEPQYTNLG	0.993	*Y*
77	SNTDYPGPH	0.935	*Y*
100	ATWTYSTEL	0.977	*Y*
181	SHAQYVEDP	0.850	*Y*
365	SPSSYGNSS	0.872	*Y*

www.cbs.dtu.dk/services/NetPhosK/

To further pursue these observations, we used HNSCC 028 cells (10⁸) infected with Ad5-ΔNp63α-myc for 18 h and then treated with 10 μM cisplatin or control medium for an additional 18 h. ΔNp63α-myc was then immunoprecipitated overnight from cell lysates (500–2000 μg) using the ProFound Mammalian C-myc Tag IP/Co-IP kit and purified by fast-performance liquid chromatography. The purified ΔNp63α-myc was in-gel digested with chymotrypsin, and the peptide mixture was subjected directly to peptide profiling by MALDI-TOF-MS. Among a few phosphorylation sites in ΔNp63α that were affected by cisplatin exposure (Table 2), we found ATM, CDK2 and p70s6K kinase recognition motifs [S385, ATM motif, PSVSQL), T397, CDK2 motif, NALTPTT), and S466, p70s6K, GCSSCLD].

Table 2.

ΔNp63α phosphorylation sites identified by mass spectrometry

Observed proteolytic	Residue	Kinase	Mascot	Xcorr
phosphopeptide1			Score2	Value3
Chymotrypsin: VSQLINPQQRNAL (384–396)	S385	ATM	41	2.21
Chymotrypsin: LTPTTIPDGM (396–405)	T397	CDK2	43	2.37
Plus Proteinase K (396 397 399 400 401)
Chymotrypsin: LGCSSCL (462–468)	S466	p70s6K	44	2.53
Plus NTCB: GCS (463–466)	S466	p70s6K	49	3.23

¹All phosphopeptides were confirmed by tandem mass spectrometry. NTCB (2-nitro-5-thiocyanobenzoic acid).

²Mascot scores of individual peptides were determined as described elsewhere.⁶²

³SEQUEST Xcorr value of individual peptides were determined as described elsewhere.⁴⁹

To examine protein levels and phosphorylation levels of ΔNp63α possibly affected by cisplatin exposure, we used HNSCC 028 cells (without endogenous p63 expression) ectopically expressing Ad5-ΔNp63α-myc, HaCaT immortalized keratinocytes (with endogenous p63 expression) and Flip-In HaCaT stable clones that expressed wild type or mutated ΔNp63α (L514F mutation derived from Hay-Wells Syndrome), introduced into a genomic DNA (reviewed in ref. 37).

Cells were treated with 10 μM cisplatin or control media for 24 h. First, we examined levels of ΔNp63α using immunoblotting with a custom antibody generated against the phosphorylated S385 (NKLPSVSQLINPQQ, residues 379–392, designated as the anti-pATM motif). We found evidence of an increased ΔNp63α phosphorylation at the ATM motif upon exposure of HNSCC cells to cisplatin (Fig. 1A and B).

Figure 1.

Cisplatin increased phosphorylation levels and decreased protein levels of ΔNp63α. (A). HNSCC 028 cells (10⁵) were infected with an empty vector Ad5 or Ad5-ΔNp63α myc for 18 h and then treated with mock medium (−) or 10 μg/ml cisplatin (+) for an additional 18 h. The vertical line in the upper panel divides two independent gel experiments. (B). Flp-In HaCaT cells (10⁵) expressing wild type or L514F mutant ΔNp63 were treated with mock medium (−) or 10 μg/ml cisplatin (+) for 18 h. Immunoblotting was performed with the indicated antibodies. Protein levels were normalized with antibody to β-actin.

We further used HNSCC 029 cells endogenously expressing an abundant amount of ΔNp63α (reviewed in ref. 32) and treated the cells with 10 μM cisplatin for 0, 12 and 24 h. Initially, we tested the effect of cisplatin exposure on cell survival using the MTT assay, as previously described.³² Using both immunoblotting and combined immunoprecipitation/in vitro protein kinase assay, we also tested whether the protein and activity levels of ATM, CDK2 or p70s6K were affected upon exposure of HNSCC 029 cells to cisplatin. We observed that cisplatin exposure decreased the survival of HNSCC 029 cells (by 2-fold in 0–48 h time frame), associated with clear induction of protein levels, and most importantly enzymatic activity, of all the kinases tested (data not shown). To evaluate the effect of cisplatin on phosphorylation levels of ΔNp63α, we employed immunoblotting of total lysates obtained from HNSCC 029 cells with custom antibodies against a phosphorylated ATM motif, CDK2 motif or p70s6K motif in the ΔNp63α polypeptide. We found that phosphorylated protein levels at the indicated positions/motifs dramatically increased upon exposure of HNSCC 029 cells to cisplatin in a time-dependent fashion relative to total ΔNp63α protein levels (Fig. 2).

Figure 2.

Custom antibodies to phosphorylated motifs of ATM (S385), CDK2 (T397) and p70s6K (S466) kinase recognize site-specific phosphorylation of ΔNp63α in HNSCC 029 cells upon cisplatin exposure. HNSCC 029 cells (10⁵) were treated with 10 μg/ml cisplatin for the indicated time periods in the presence of 20 μM MG-132 (26S proteasome inhibitor). Total lysates resolved by SDS-PAGE were immunoblotted with the indicated antibodies. Protein levels were normalized with antibody to β-actin.

Next, we grew HNSCC 029 cells in the presence or absence of 10 μM cisplatin for 0, 6, 12 or 24 h and cell-permeable inhibitors for ATM (5 μM KU-55933, reviewed in refs. ^{34, 39, 44} and ⁶³), CDK2 (20 μM roscovitine, reviewed in refs. ^{14, 17, 18, 24, 52} and ⁸¹) or p70s6K (20 nM rapamycin, reviewed in refs. ¹⁶ and ³⁰) kinases (added for the last 60 min before cell harvesting). We first showed that these protein kinase inhibitors dramatically blocked the enzymatic activities of ATM, CDK2 and p70s6K, while increasing the cell survival compared to the inhibitory effect of cisplatin exposure (data not shown). Then, the protein levels for ΔNp63α were tested by immunoblotting with Ab-1 antibody. We observed that all protein kinase inhibitors used in this set of experiments clearly blocked degradation of ΔNp63α in HNSCC cells induced by cisplatin exposure (Fig. 3).

Figure 3.

Inhibitors of protein kinases blocked the protein degradation of ΔNp63α in HNSCC 029 cells. HNSCC 029 cells (10⁵) were treated with 10 μg/ml cisplatin for the indicated time periods in the presence or absence (control, DMSO) of inhibitors of ATM (5 μM KU-55933), Cdk2 (20 μM roscovitine) or mTOR/p70s6K (20 nM rapamycin), which were added for the last 60 min before cell harvesting. Total lysates resolved by SDS-PAGE were immunoblotted with Ab-1 antibody against ΔNp63. Protein levels were normalized with antibody to β-actin.

We then transiently introduced siRNA against ATM, CDK2 or p70s6K (300 nM each) into HNSCC 029 cells versus scrambled siRNA to examine their effectiveness in silencing expression of their respective target (Fig. 4). We observed that siRNA dramatically silenced the expression of the protein kinases tested (Fig. 4A), and in turn decreased the cisplatin-induced degradation of ΔNp63α in HNSCC 029 cells, while scrambled siRNA failed to do so (Fig. 4B).

Figure 4.

Small interference RNA (siRNA) to protein kinases inhibited the protein degradation of ΔNp63α in HNSCC 029 cells. (A). In contrast to a negative control (scrambled siRNA), siRNA against ATM, CDK2 or S6K dramatically inhibited the protein levels of ATM, CDK2 or p70S6K in 029 cells (10⁵, 48 h post-transfection, 300 nM of siRNA), respectively, as shown by immunoblotting with the indicated antibodies. (B). HNSCC 029 cells were transiently transfected with siRNA for 48 h and then subsequently treated with 10 μg/ml cisplatin for indicated additional time periods. SiRNA against ATM, CDK2 or p70S6K dramatically inhibited the protein degradation of ΔNp63α upon cisplatin exposure. Protein levels for ΔNp63α were analyzed by immunoblotting with Ab-1 antibody, and normalized with antibody to β-actin.

We further examined whether genetic changes in putative phosphorylation sites in the ΔNp63α protein expressed in HNSCC 029 cells would affect its degradation upon cisplatin exposure. We established HNSCC 029 stable cell lines expressing the wild type ΔNp63α or mutated ΔNp63α (ATM phosphorylation site, CDK2 phosphorylation site and p70s6K phosphorylation site) using the Flp-In technology, as previously described.³⁷ This approach allowed us to introduce p63 mutations (S385, T397 and S466) into the genomic DNA of target 029 cells resembling a knock-in technique. The cells harboring p63 mutations exclusively expressed RNA transcripts with desired mutations, as confirmed by sequencing of individual clones derived from RT-PCR of mRNA obtained from mutated cells (data not shown). Interestingly, cells expressing mutated ΔNp63α grew and proliferated at a similar rate to cells expressing the wild type ΔNp63α, while protein and activity levels for ATM, CDK2 or p70s6K were the same (data not shown). These cells were then treated with 10 mM cisplatin and protein levels of ΔNp63α were examined by immunoblotting with Ab-1 antibody against ΔNp63. We observed that the absence of serine or threonine residues necessary for kinase phosphorylation [due to mutations in Ser (S385 or S466), or Thr (T397 for Ala] completely blocked the cisplatin-induced degradation of ΔNp63α suggesting a critical role of all three of these residues in the phosphorylation-mediated degradation of ΔNp63α in HNSCC cells (Fig. 5).

Figure 5.

Mutations of serine (S/A-385, S/A-466) or threonine (T/A-397) in the ΔNp63α protein kinase motifs prevented the protein degradation of ΔNp63α in HNSCC 029 cells. Stable clones of HNSCC 029 cells (10⁵) harboring plasmids with mutations in ΔNp63α for the putative phosphorylation motifs for ATM, CDK2 or p70s6K protein kinases were generated. HNSCC 029 cells were then treated with 10 μg/ml cisplatin for the indicated time periods. Protein levels for ΔNp63α were analyzed by immunoblotting with Ab-1 antibody, and normalized with antibody to β-actin.

Next, we further explored whether the cisplatin exposure of HNSCC cells induced a possible sequence of events leading to phosphorylation of ΔNp63α. Stable clones of HNSCC 029 cells harboring plasmids with wild type ΔNp63α or ΔNp63α mutated in the putative phosphorylation motifs (ATM, CDK2 or p70s6K) were grown up in the presence of 10 μg/ml cisplatin for 0, 6, 12, 18, 24 h (Fig. 6). Levels for phospho-ΔNp63α were analyzed by immunoblotting with the indicated custom antibodies against the pATM, pCDK2 or pp70s6k motifs, respectively (Fig. 6). We found that in cells expressing the wild type ΔNp63α cisplatin-inducedATM-mediated phosphorylation of ΔNp63α started to reach maximal levels at ~6 h, while CDK2 phosphorylation occurred at 12 h and p70s6K phosphorylation at 18 h (Fig. 6A). We further observed that when testing cells with ATM-mutated phosphorylation site, neither of our custom antibodies recognized the phospho-ΔNp63α protein (Fig. 6B). Moreover, the antibody to the pATM-motif (in cells harboring mutated CDK2 site) or antibodies to pATM-motif or pCDK2 motif (in cells harboring mutated p70s6K site) recognized the phospho-ΔNp63α protein (Fig. 6C and D). Therefore, in HNSCC cells treated with cisplatin, the ΔNp63α protein is phosphorylated first by ATM, then by CDK2 and finally by p70s6K kinases.

Figure 6.

Time-course of the ΔNp63α phosphorylation by protein kinases in HNSCC cells upon cisplatin exposure. Stable clones of HNSCC 029 cells (10⁵) harboring the wild type ΔNp63α (A. ΔNp63α-wt) or mutations in the certain ΔNp63α putative phosphorylation motifs (B, ΔNp63α-ATM), (C, ΔNp63α-CDK2) or (D, ΔNp63α-s6K) were grown-up in the presence of 10 μg/ml cisplatin for the indicated time periods (0, 6, 12, 18, 24 h). Protein levels for phospho-ΔNp63α were analyzed by immunoblotting with indicated custom antibodies against pATM motif, pCDK2 motif or p70s6k motif, and normalized with antibody to β-actin.

Finally, we tested the ability of purified protein kinases (ATM, CDK2 or p70s6K) to phosphorylate the purified ΔNp63α protein in vitro.^11,25 We showed that all tested kinases phosphorylated ΔNp63α in vitro suggesting that ΔNp63α is a direct target of these protein kinases (Fig. 7).

Figure 7.

In vitro phosphorylation of ΔNp63α by ATM, CDK2 or p70s6K. Kinase reactions contained 25 mM Tris-HCl, pH 8, 50 mM KCl, 5% glycerol, 0.5 mM DTT, 5 μCi of [gamma³²P]-ATP, 10 μM cold ATP, 10 mM MnCl₂, 0.25–0.5 μg of PHAS-I (Stratagene) 2 ng of purified ATM (CDK2 or p70s6K or nothing) and 2 ng of ΔNp63α. ΔNp63α was precipitated with Ab-1 antibody, analyzed by 10% SDS-PAGE, and subjected to autoradiography. Blots were immunoblotted with 4A4 antibody to p63, with custom antibodies against ATM motif, CDK2 motif or p70s6K motif. The cDNA for ΔNp63α was mutated at the following positions [S-385 (S-A), T-397 (T-A), or S-466 (S-A)] for alanine residue (A) using the QuickStep site-directed mutagenesis assay. In vitro translated unlabeled products representing wild type (S or T) or mutated (A) ΔNp63α polypeptides were synthesized using the Promega TNT assay as previously described (reviewed in refs. ²⁵, ³⁷ and ⁶¹), and mixed with 1 ng of purified ATM, CDK2 or p70s6K for kinase reaction as described above. Resulting mixes were precipitated with Ab-1 antibody, analyzed by 10% SDS-PAGE, and subjected to autoradiography.

Discussion

We previously found that genotoxic stress agents (e.g., cisplatin, doxorubicin, etc.,) induced phosphorylation of ΔNp63α leading to a dramatic decrease of the ΔNp63α protein levels in HNSCC cells.^25,85 By using MALDI-TOF-MS, we have identified phosphorylated sites in ΔNp63α in HNSCC cells upon cisplatin exposure. They included the S385, T397 and S466 sites surrounded by recognition motifs for the ATM, CDK2 and p70s6K kinases, respectively. We also showed that the cisplatin-exposed HNSCC cells harbored phosphorylated ΔNp63α at the ATM, CDK2 and p70s6K kinase recognition sites in a time-dependent fashion. We further showed that chemicals or siRNA inhibiting the activity of ATM, CDK2 and p70s6K kinases blocked degradation of ΔNp63α in HNSCC cells usually induced upon cisplatin exposure. We also found that engineered genetic changes in ΔNp63α that replace specific residues targeted for phosphorylation by ATM, CDK2 or p70s6k led to a dramatic modulation of ΔNp63α degradation. We also showed that the ΔNp63α protein in HNSCC cells exposed to cisplatin is probably phosphorylated first by ATM, then by CDK2 and finally by p70s6K kinases ultimately targeting ΔNp63α to a proteasome-dependent degradation pathway.^12,25 Finally, we demonstrated that the ΔNp63α protein is a target for direct in vitro phosphorylation by ATM, CDK2 or p70s6K. Overall, we suggest that ATM kinase is a master switch for the ΔNp63α phosphorylation/degradation in human HNSCC cells upon DNA damage.

Accumulating evidence points to a role for multiple posttranslational modifications (e.g., phosphorylation) in response of cells to DNA damage induced by various extracellular stimuli mediating these events.^{1,2,6,7,20,21,27,43,50,72} Recent studies suggest that N-terminal phosphorylations are important for stabilizing p53 in cells exposed to genotoxic stresses.⁶ However, the modifications to the C-terminus of p53 inhibit its ability to negatively regulate sequence-specific DNA-binding, and also modulate protein stability, the oligomerization state, the nuclear import/export process, and the degree of p53 ubiquitination.¹⁴

From other hand, ΔNp63 isotypes were shown to manifest cellular response to DNA damage in an opposite fashion via protein degradation, because they lack the N-terminal region of TA-isotypes found in all p53 family members.²⁵ Keeping this in mind, we now have evidence that cisplatin-induced degradation of ΔNp63α in HNSCC cells occurs after its phosphorylation at specific sites. Since ΔNp63α usually plays an opposite role in human cancers compared to p53 and TAp63, we now add additional molecular evidence that this protein reacts differently to cisplatin ands possibly to other DNA damaging agents.

Although both exogenous ΔNp63α protein expressed in transfected cells under unstressed conditions (reviewed in ref. 79) and endogenous ΔNp63α expressed in human keratinocytes and HNSCC cells under stressed conditions^25,37,80 were reported to be phosphorylated, a comprehensive and direct analysis of the p63 sites phosphorylated by specific kinases was not previously performed. The relationship between p63 and key protein kinases is mutually intertwined. A few protein kinases were found to be involved in the upstream regulation of p63 transcription and several other protein kinases were shown to be p63 downstream transcription targets.^{5,48,54,64,78,82}

Westfall and coworkers have shown that certain amino acid residues are phosphorylated in ΔNp63α in human keratinocytes upon UV-irradiation.^79,80 All the modified species of ΔNp63α (12—in UV-treated cells, and 7—in untreated cells) were reduced to one by treatment of ΔNp63α precipitates with calf intestine phosphatase supporting the notion that ΔNp63α is indeed a phosphoprotein.^79,80 Using antibodies to certain phosphorylated residues of ΔNp63α, Westfall and coworkers have also found an increase in phosphorylation of S66/S68 and S361 in ΔNp63α upon paclitaxel or UV exposure.⁸⁰ The latter was also shown to induce the rapid phosphorylation of ΔNp63α by the p38 stress MAPK leading to inhibition of ΔNp63α transcriptional activity and its degradation.^35,59 Phosphorylated ΔNp63 displayed a decreased ability to bind certain cell cycle arrest and apoptotic promoters, thus allowing rapid activation of the p53-dependent transcriptional apoptotic program.⁵⁹ Interestingly, the Abl inhibitor, Gleevec was shown to reduce TAp63 expression in a dose-dependent manner in HNSCC cells thereby overriding its induction by DNA damaging agents.^58,77

Since p63 proteins share a strong homology with p53, it is likely that many protein kinases implicated in phosphorylation of p53 might play a similar role in p63 (reviewed in refs. ²⁰ and ²¹). Using a bioinformatics approach (GPS kinase prediction module), Finlan and Hupp uncovered several putative protein kinase motifs (Abl, PKB, ATM, S6K, CDK, MAPK and EGFR) for potential TAp63α phosphorylation (Fig. 8A and reviewed in ref. 21). A fewer motifs were predicted in the protein sequence of ΔNp63α that lacks the N-terminal domain (Fig. 8B).

Figure 8.

Schematic representation of phosphorylation sites in ΔNp63α. (A) “The sequence of the TAp63α protein with highlighted motifs depicting potential phosphorylation sites within TAp63α. Four clusters for potential kinases have been identified using the GPS phosphorylation prediction module. Of note, cluster I and II reside within the N-terminal transactivation domain (TAp63 isoforms, and devoid in ΔNp63 isoforms).” ((A) is courtesy of Lee Finlan and Ted Hupp. P63. The Phantom of the Tumor Suppressor. Cell Cycle 2007; 6:1062–71. Permission has been received from Dr. Ted Hupp and Cell Cycle Editorial. Copyright of Landes Biosciences Press). (B) Modular structure of ΔNp63α (reviewed in ref. 64) with putative phosphorylation sites predicted for TAp63α (reviewed in ref. 21) and adjusted according to differences between TAp63α and ΔNp63α. ΔN-transactivation domain ΔN-TAD, residues 1–26), proline-rich domain (PRD, residues 37–97), DNA-binding domain (DBD, residues 98–322), oligomerization domain (ODD, residues 323–357), C-terminal TAD (CTAD, residues 380–470), sterile α-motif (SAM, residues 472–537), and transcription inhibitory domain (TID, last 71 residue). The arrows indicate the newly identified phosphorylation sites for ATM (S385), CDK2 (T397) and p70s6K (S466) kinases found in p63 (the current study) and also reported by Westfall et al.⁸⁰

Multiple signaling pathways exist to stabilize p53 in response to different forms of stress through phosphorylation of p53 by a variety of protein kinases.^1,6,20,50 They include the following kinases serving as DNA damage checkpoints and sensor proteins: ATM, ATR, AURKA, DNA-PK, CHK1 and 2, CDK2 and 5, HIPK2, PKC, PKR, FACT-CK2, ERK, p38 stress MAPK, JNK, GSK3β, etc.^1,6 It is not surprising that at least two of these kinases (ATM and CDK2) shown in the current study play a role in modification of ΔNp63α after DNA damage induced by cisplatin exposure.

Mammalian cells respond to DNA damage by activating signal transduction pathways that arrest cell cycle progression and initiate DNA repair.^{1,27,29,50,55} A key regulator of the cell response to DNA damage is the ATM protein kinase, whose activation/homodimerization leads to association of ATM with its protein targets and their phosphorylation with subsequent effects on cell cycle checkpoints, apoptosis and DNA repair.^1,43,50,55 Abnormal regulation of progression from G₁ to S phase of the cell cycle by altered activity of CDKs was shown to be a hallmark of cancer.^18,55,87 There is also evidence implicating the mTOR/p70s6K signaling pathway in regulating cell proliferation and the resistance of cancer cells to cisplatin.^47,77,84

Our study shows that the above-mentioned protein kinases phosphorylate ΔNp63α in HNSCC cells upon cisplatin exposure. Additional studies are necessary to perform a comprehensive mapping of phosphorylation events and protein kinases involved in the regulation of p63 in a variety of cell types. Moreover, it is still unclear how phosphorylation and/or dephosphorylation events affect the transcriptional function of p63. As in the case of p53 posttranslational modifications, these modifications are likely to play a critical role in regulating p63 function as a transcriptional regulator of tumorigenesis, cell proliferation/differentiation or DNA replication stress.^28,55 Since ΔNp63 isotypes play an opposite role towards p53, we suggest that the long-term DNA damage induces a dramatic decrease of the ΔNp63α protein levels decrease leading to an increase in the ability of the wild type p53 to function as tumor suppressor, thereby modulating cell proliferation (by cell cycle arrest or apoptosis) and directing cells to senesce.⁸⁵

Experimental Procedures

Cells, reagents and antibodies

Head and neck squamous cell carcinoma (HNSCC) cell line 029 (expressing wild type p53 and p63), and 028 (no p63 expression, while expressing wild type p53) were initially isolated from primary tissues and maintained at the Department of Otolaryngology/Head and Neck Surgery of the Johns Hopkins University School of Medicine.³² Cells were maintained in RPMI medium 1640, 10% fetal bovine serum (FBS). Cells were incubated with cisplatin (10 μg/ml, Sigma) and a 26S proteasome inhibitor (MG-132, 20 μM, American Peptide Company). We also treated cells for 60 min with the cell-permeable inhibitors for ATM kinase, 5 μM KU-55933 (2-Morpholin-4-yl-6-thianthren-1-yl-pyran-4-one (#118500, EMD Chemicals Inc.,), for CDK2, 20 μM roscovitine 2-(R)-(1-ethyl-2-hydroxyethylamino)-6-benzylamino-9-isopropylpurine (EMD Chemicals), and for mTOR/p70s6K kinase, 20 nM rapamycin (EMD Chemicals).

We used a rabbit polyclonal antibody Ab-1 directed against human ΔNp63 (EMD Chemicals), a mouse monoclonal antibody against all human p63 isotypes (4A4, Santa Cruz Biotechnology), a monoclonal antibody against human β-actin (Sigma), a mouse anti-human Cdk1/Cdk2 monoclonal antibody (clone AN21.2, ab6434, Abcam) and a rabbit anti-human s6K polyclonal antibody (ab36864, Abcam), a mouse anti-human ATM monoclonal antibody (clone 10H11.E12, ab36810, Abcam), a mouse anti-human p70s6k monoclonal antibody (clone 16, #611260, BD Biosciences/Pharmingen). Custom rabbit polyclonal antibodies against phosphorylated peptides encompassing the ΔNp63α protein sequences (ATM motif, NKLPSV-pS-QLINPQQ, residues 379-392; CDK2 motif, QQRNAL-pT-PTTIPDG, residues 391–404; p70s6K motif, LARLGC-pS-CLDYFT, residues 459–472) were prepared and purified against the phosphorylated peptide vs. non-phosphorylated peptide with the aid of Sigma Genosys. Ad-ΔNp63α-myc was prepared in our laboratory, as previously described.⁶¹

The following recombinant proteins: ATM (#HZ-2029-10), Cdk2 (HZ-2018-10), p70s6K (HZ-2053-10) were purchased from Humanzyme, Inc., Recombinant GST-ΔNp63α was purified, as previously described.²⁵ About ~1.0–2.0 μg of fusion protein was purified from 3–4 g of bacterial cells. To purify intact ΔNp63α, GST-taq was cleaved by thrombin and separated by second round of glutathione-agarose column chromatography (Sigma) according to the manufacturer’s protocol.

Immunoblotting and immunoprecipitation

Cells were lysed in buffer (50 mM Tris, pH 7.5, 100 mM NaCl, 2 mM EDTA, 0.5% Triton X-100, 0.5% Brij-50, 1 mM PMSF, 0.5 mM NaF, 0.1 mM Na₃VO₄, 2X complete protease inhibitor cocktail), sonicated five times for 10 sec time intervals, and clarified for 30 min at 15,000 x g.^61,65 Supernatants (designated as total lysates) were resolved by 4–10% SDS-PAGE and then analyzed by immunoblotting or immunoprecipitation, as previously described.⁶¹

Generation of mutated p63 stable clones

The stable HNSCC 029 clones expressing wild type and mutated ΔNp63α were generated using the Flp-In technology (Invitrogen), as previously described.³⁷ Flp-In HNSCC 029 host cells were generated by stable insertion of pFRT/lacZeo plasmid and selected on Zeocin.³⁷ To establish the isogeneic HNSCC 029 clones expressing wild type and mutant ΔNp63α, we subcloned the wild type or mutated cDNA for ΔNp63α (prepared by site-directed mutagenesis using the Quick-Step mutagenesis kit (Stratagene) with the following primers:

ΔNp63S385G_F (a1153g):

5′-CAAGCTGCCTTCTGTGGGCCAGCTTATCAACCC-3′ ΔNp63S385G_R:

5′-GGGTTGATAAGCTGGCCCACAGAAGGCAGCTTG-3′; ΔNp63T397A_F (a1189g):

5′-CAGCGCAACGCCCTCGCTCCTACAACCATTCC-3′ ΔNp63T397A_R:

5′-GGAATGGTTGTAGGAGCGAGGGCGTTGCGCTG-3′; ΔNp63S466A_F (t1396g):

5′-CGAGGTTGGGCTGTTCAGCATGTCTGGACTATTTC-3′ ΔNp63S466A_R,

5′-GAAATAGTCCAGACATGCTGAACAGCCCAACCTCG-3′ and then confirmed by sequencing) into the pcDNA5/FRT/V5-His-TOPO vector.

Flp-In HNSCC 029 cells were transfected with an empty pcDNA5/FRT/V5-His-T

OPO vector or expression cassettes for ΔNp63α (wild type or mutant) along with

pOG44 plasmid, bearing the Flp recombinase.³⁷ Resulting Flp-In HNSCC 029 clones were selected on hygromycin B and verified by RT-PCR and sequencing. Flp-In 029 cells expressing wild type or mutated p63 grew and proliferate under unstressed conditions. The resulting cells were treated with 10 μg/ml cisplatin in a time-dependent fashion.

Small-interfering RNA (siRNA)

We used the human ATM siRNA, CDK2 siRNA, p70s6K siRNA and validated negative control scrambled siRNA (SMARTpool, Dharmacon (Lafayette, CO, USA). SiRNA were combined with DharmaFECT transfection reagent (Dharmacon), and the HNSCC 029 cells were transfected according to the recommended protocol with siRNA (100 nM final concentration). After 48 h of transfection, cells were starved in RPMI medium 1640 containing 0.5% FBS before treatment. After 48 h incubation, cells were treated with cisplatin for indicated time periods and then total cell lysates were used for immunoblotting analysis, as previously described.^25,37

Mass-spectrometry analysis of phosphorylation sites in ΔNp63α

028 cells were infected with the recombinant Ad5-ΔNp63α-myc for 18 h (reviewed in refs. ^{25, 61, 82} and ⁸³). Resulting cells were treated with 10 μM cisplatin or with control medium for an additional 18 h, and ΔNp63α-myc was immunoprecipitated overnight from cell lysates (500–2000 μg) using the ProFound Mammalian C-myc Tag IP/Co-IP kit (Pierce) following the manufacturer’s protocol #2. The protein then was purified by fast-performance liquid chromatography allowing to achieve ~90% purity of ΔNp63α and yielding ~10 μg/10⁸ cells. Purified ΔNp63α-myc proteins from cells treated and untreated with cisplatin were subjected to the enzyme in-gel digestion, and the peptide mixture was subjected directly to peptide profiling by matrix-assisted laser desorption/ionization time-of-flight mass spectrometry (MALDI-TOF-MS) to pinpoint naturally occurring phosphopeptides, and to sequence and identify each peptide within the mixture.^{3,9,19,36,49,68,86}

MALDI-TOF-MS was performed using the Voyager DE ProTM mass spectrometer (Applied Biosystems). Aliquots of digested peptides (1 pmol) in 1 μl of H₂O and 0.1% trifluoroacetic acid were mixed with 1 μl of cyano-4-hydroxycinnamic acid, spotted onto a sample target, dried and loaded into the mass spectrometer. The MS-digest computer programs (MS-Fit, Protein Prospector, version 3.4.1. and ProFound, version 4.10.5) were provided by the Johns Hopkins University Mass Spectrometry Core and used to calculate the average masses of all possible peptide and phosphopeptide fragments of ΔNp63α, and the m/z value of the mass spectral peaks for the corresponding MH⁺ ions.^8,28 The peptide masses were entered into the MASCOT Search Engine-2 and the National Center for Biotechnology Information database was searched to match the tryptic peptide fingerprint with a parent polypeptide.^49,62

In vitro protein kinase assay

Kinase reactions contained 25 mM Tris-HCl, pH 8, 50 mM KCl, 5% glycerol, 0.5 mM DTT, 5 μCi of [gamma³²P]-ATP, 10 μM cold ATP, 10 mM MnCl₂, 0.25–0.5 μg of PHAS-I (Stratagene), 2–4 ng of purified recombinant ATM (CDK2 or p70s6K or nothing, all purified kinases were purchased from Humanzyme) and 2–4 ng of purified ΔNp63α as previously described.^11,25 The reactions were incubated at 30°C for 30 min. ΔNp63α was precipitated with Ab-1 antibody, and analyzed by 10% SDS-PAGE. Gels were stained with Coomassie Blue, destained, dried and exposed to X-ray film with intensifying screen at 80°C.