DeltaNp63alpha-Mediated Induction of Epidermal Growth Factor Receptor Promotes Pancreatic Cancer Cell Growth and Chemoresistance

Pancreatic ductal adenocarcinoma (PDAC) is highly resistant to current chemotherapy regimens, in part due to alterations in the p53 tumor suppressor pathway. p53 homolog p63 is a transcription factor essential for the development and differentiation of epithelial surfaces. However its function in cancer is controversial and its role in PDAC is not known. We discovered that ΔNp63α was the predominantly expressed p63 variant in pancreatic cancer cell lines. ΔNp63α protein and mRNA levels were high in T3M4, BxPC3 and COLO-357 pancreatic cancer cells and low in ASPC-1 and PANC-1 cells. Overexpression of ΔNp63α in PANC-1 cells and shRNA-mediated knockdown in T3M4 cells indicated that ΔNp63α promoted anchorage-dependent and -independent growth, motility and invasion, and enhanced resistance to cisplatin-induced apoptosis. Epidermal growth factor receptor (EGFR) signaling pathways contribute to the biological aggressiveness of PDAC, and we found that the motogenic effects of ΔNp63α were augmented in presence of EGF. Ectopic expression of ΔNp63α resulted in upregulation of EGFR and β1-integrin in PANC-1 cells. Conversely, ΔNp63α knockdown had an opposite effect in T3M4 cells. ΔNp63α potentiated EGF-mediated activation of ERK, Akt and JNK signaling. Chromatin immunoprecipitation and functional reporter assays demonstrated that ΔNp63α activated EGFR transcription. 14-3-3σ transcription was also positively regulated by ΔNp63α and we have previously shown that 14-3-3σ contributes to chemoresistance in pancreatic cancer cell lines. Conversely, shRNA-mediated knockdown of 14-3-3σ led to abrogation of the ΔNp63α effects on cell proliferation and invasion. Thus, p53 homolog ΔNp63α enhances the oncogenic potential of pancreatic cancer cells through trans-activation of EGFR and 14-3-3σ.


Introduction
Studies in human disease demonstrate that most established tumors carry more than one genetic defect. Pancreatic ductal adenocarcinoma (PDAC) results from the successive accumulation of gene mutations [1]. Activating mutations in K-Ras oncogene and inactivation of tumor suppressors CDKN2A, p53 and SMAD4 are implicated in PDAC development and progression [2]. Genetically engineered mouse models have supported the ''double-hit'' hypothesis where introduction of either mutant p53 allele or biallelic deletion of ink4a/Arf in mice results in progression of pancreatic intraepithelial neoplastic lesions with local invasion and metastases [3,4]. p53 mutations are found in 60 to 70% of PDAC. By contrast, p63, an ancestral member of p53 family, is rarely mutated in cancer.
Six variants of p63 are generated through transcription from two distinct promoters and alternative splicing. Isoforms transcribed from P1 contain a full-length trans-activation domain (TAp63a, b and c). Transcription from P2 generates amino-terminally truncated variants (DNp63a, b and c), whose exact role in cancer is not clear. The DNp63 variant is overexpressed in a variety of human cancers, including tumors of squamous cell origin (head and neck, lung), breast and bladder [5]. In head and neck squamous cell carcinoma and ''triple-negative'' breast cancer cells, DNp63 suppresses p73-dependent apoptosis and thus promotes tumor survival [6,7]. By contrast, downregulation of DNp63a in urothelial carcinoma cell lines promotes cancer invasiveness [8], suggesting that the DN variant may function in a cell type-specific manner.
The role of p63 in PDAC is poorly understood. Here we demonstrate that the DNp63a variant is expressed at variable levels in PDAC cell lines, and provide evidence that DNp63a promotes pancreatic cancer cell growth, migration, invasion and chemoresistance. Via direct transcriptional activation, DNp63a leads to the up-regulation of epidermal growth factor receptor (EGFR) and 14-3-3s, sensitizing cancer cells to EGF and enhancing their oncogenic potential.

Cell lines
Human pancreatic cancer cell lines ASPC-1, BxPC3 and PANC-1; HEK293 human embryonic kidney cells and H1299 human lung carcinoma cells were purchased from American Type Culture Collection (Manassas, VA). COLO-357 and T3M4 human pancreatic cancer cell lines were a gift from Dr.R.Metzgar (Duke University, Durham, NC). Cell lines were grown in RPMI or DMEM (HEK293) supplemented with 10% fetal bovine serum, 100 U/ml penicillin, and 100 mg/ml streptomycin (complete medium).

Reverse transcription polymerase chain reaction (RT-PCR)
Total RNA was isolated using RNeasy Mini Kit (Qiagen, Valencia, CA). cDNA was synthesized from 1 mg RNA by random hexamer priming using the Superscript III RT Kit (Invitrogen, Carlsbad, CA). Semi-quantitative RT-PCR was carried out using forward and reverse primers specific for p63 isoforms as previously published [12]. The following PCR cycling conditions were used: 94uC for 3 min, 35 cycles of 94uC for 40 sec, 55uC for 40 sec, and 72uC for 1.5 min; and 72uC for 4 min.
For the analysis of DNp63 and TAp63 transcript expression quantitative real time PCR (Q-PCR) was performed in a 7300 Sequence Detector using Universal PCR Master Mix according to manufacturer's instructions (Applied Biosystems, Foster City, CA), template cDNA and gene specific probes (Hs00978339_m1 [DNp63]; Hs00978349_m1 [TAp63]; Applied Biosystems). All samples were analyzed in duplicates.

Transient transfections, adenoviral infection and luciferase assays
PANC-1 and T3M4 cells were transiently transfected with ExGen 500 in vitro Transfection Reagent (Fermentas Life Sciences, Glen Burnie, MD), using equal amounts of experimental or control plasmid. Cells were incubated for 48 hours and protein expression was verified by immunoblotting. The control and DNp63a-expressing adenovirus were used as described [13]. ASPC-1 and PANC-1 cells were infected with adenovirus at MOI of ,5 in complete medium for 24 hours.
For luciferase assays, PANC-1 cells were co-transfected with experimental plasmid or control and the luciferase reporter construct (pBV-14-3-3s promoter BDS 2 36 (p53 binding site)luc; or pGL3-EGFR promoter-luc) along with pCMVb vector (Clontech Laboratories, Mountain View, CA). The amount of DNA per transfection was kept constant by using empty pcDNA3.1 vector. Cells were harvested 48 hours post-transfection and luciferase assays were performed using the Dual-Luciferase Reporter Assay System (Promega). Relative light units were determined using a luminometer (LMaxII 384 , Molecular Devices, Sunnyvale, CA) for firefly luciferase. b-galactosidase activity was determined using a colorimetric method to normalize transfection efficiency [14].

3-(4,5-Dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) assay
Cells were plated in 96-well plates (3000 cell per well, 6 wells per sample) and cultured in the absence or presence of 10 mg/mL cisplatin (Sigma-Aldrich, St. Louis, MO) for 48 hours. MTT (Sigma-Aldrich) was added at a final concentration 0.55 mg/mL. After additional 4 hours incubation, absorbance at 570 nm was determined using an Emax precision microplate reader (Molecular Devices). We have previously observed that in pancreatic cancer cell lines the MTT assay correlates with cell growth as determined in a doubling and [ 3 H]thymidine incorporation assays [16].

Soft agar assays
Soft agar assays were set up in twelve-well plates, each well containing a bottom layer of 0.5% Difco agar noble (BD Biosciences), a middle layer of 0.3% agar including 1500 cells, and a top layer of 0.3% agar. The plates were incubated for 14 days. Next, 150 mL of 5 mg/mL MTT solution were added to each well. After incubation at 37uC for 4 hours plates were photographed and colonies were counted.

Cell migration and invasion assays
Cell motility was assessed in in vitro wound healing and Transwell migration assays. For wound healing assays, cells were grown to confluency in 6-well tissue culture plates, and serumstarved for 12 hours. The resulting cell monolayer was scratched with a 10 mL pipette tip generating two parallel wounds and incubated for 18 hours in serum-free medium (SF; 0.1% bovine serum albumin) in the absence or presence 1 nM EGF (Millipore). Photographs were taken of each well at four marked locations under 406 magnification at zero and 18 hours after wounding. The wound area of matched pictures was measured using the ImageJ software (National Institutes of Health). For Transwell migration assays, cell were suspended in 100 mL SF medium and p63 in Pancreatic Cancer placed onto the upper compartment of Transwell chambers (8 mm pore size, Corning Incorporated, Corning, NY). For invasion assays, cells were suspended in 500 mL SF medium and placed onto upper compartment of Matrigel-coated Transwell chambers (8 mm pore size, BioCoat Matrigel Invasion Chambers; BD Biosciences; Franklin Lakes, NJ). In both Transwell assays, the lower compartment was filled with 750 mL SF medium in the absence or presence of 1 nM EGF. After 18-20 h, membranes were fixed in methanol, stained with toluidine blue (Fisher Scientific, Pittsburgh, PA) and counted using a light microscope.

Formaldehyde crosslinking, chromatin immunoprecipitation (ChIP) and PCR amplification
For DNA-histone crosslinking, 2610 6 cells were incubated with 1% formaldehyde in complete medium for 10 minutes. Cells were washed twice in ice-cold PBS and lysed in SDS Lysis Buffer (Millipore) complete with protease inhibitor cocktail (Roche). Protein lysates were sonicated to yield chromatin fragments of ,500 bp as assessed by agarose gel electrophoresis. After preclearing, protein lysates were incubated at 4uC overnight with 2 mg p63a antibodies (H-129, Santa Cruz) or with rabbit IgG as isotype-specific control antibodies (Santa Cruz). Immunocomplexes were washed and DNA was purified using the ChIP Assay Kit (Millipore) and QiaQuick PCR Purification Kit (Qiagen) according to the manufacturer's instructions. PCR conditions and primers for 14-3-3s consensus binding sites 1 and 2 were previously reported [17]. The following PCR primers and conditions were used for the EGFR promoter p53 binding site: forward primer 59 GGCCGCTGGCCTTGGGTC 39; reverse primer 59 GCCGTGCGCGGTGGTTG 39; 1 cycle of 94uC for 5 min, 43 cycles of 95uC for 30 sec, 55uC for 30 sec, 72uC for 50 sec, followed by 1 cycle of 72uC for 10 min. PCR was performed using HotStarTaq polymerase Kit (Qiagen), with use of Q-Solution in the EGFR PCR assay.

Cisplatin-induced apoptosis
Pancreatic cancer cells were incubated for 24 h in complete medium in the absence or presence of 5 or 10 mg/mL cisplatin for the initial 2 h, 6 h or the entire 24 h. Both floating and adherent cells were collected and subjected to immunoblotting.

DNp63a expression in pancreatic cancer cell lines
We found that p63 was differentially expressed in pancreatic cancer cell lines. p63 protein levels were highest in T3M4 and BxPC3 cells, intermediate in COLO-357 and below the detection level in ASPC-1 and PANC-1 cells (Fig. 1A). Since the p63 antibody (clone 4A4) recognized all six known isoforms of p63, we compared p63 mRNA transcript levels in the above cell lines. All the cell lines expressed low levels of TAp63 mRNA (Fig. 1B). By contrast, DNp63 mRNA transcripts were relatively elevated in BxPC3, COLO-357, and T3M4 cells (Fig. 1B).
To further elucidate whether a particular splicing variant of p63 is expressed in pancreatic cancer cell lines, we employed semiquantitative RT-PCR with p63 splicing variant-specific primers (Table S1). We confirmed that TAp63a mRNA transcripts were expressed at relatively low and similar levels in all five cell lines. By contrast, DNp63a mRNA transcripts were only expressed in BxPC3, COLO-357, and T3M4 cells, whereas DNp63c mRNA was barely detectable in BxPC3 and T3M4 cells (Fig. 1C). Although p63a can be detected in a specific PCR assay, isolated detection of p63b mRNA transcripts is unreliable since its Cterminus is not unique (Fig. S1). In our experiments, expression of DNp63b mRNA transcript correlated with DNp63a expression, probably due to non-specific amplification of DNp63a. Since we detected p63 protein migration at ,68 kDa, and DNp63b protein migrates at ,52 kDa [18][19][20], we concluded that DNp63a was the predominantly expressed p63 variant in pancreatic cancer cells.

Oncogenic effects of DNp63a in pancreatic cancer cells
To study the biologic effects of DNp63a in PDAC, we manipulated its expression in PANC-1 and T3M4 cells, which had low and high endogenous levels of DNp63a, respectively ( Fig. 1). PANC-1 cells were transiently transfected with a DNp63aexpressing vector (PANC-1DN), a TAp63a-expressing vector (PANC-1TA), or control plasmid. Lentiviral-mediated shRNA was used to down-regulate p63 gene isoforms in T3M4 cells. Two different shRNA sequences, one complementary to a sequence within a p63 DNA-binding domain (sh DBD) and thus targeting all isoforms of p63, and another complementary to a sequence within sterile alpha motif domain (SAM) and thus targeting TAp63a and DNp63a (sh p63a) were determined to reduce the DNp63a protein and transcript levels in T3M4 cells ( Fig. 2A). While shRNA targeting trans-activation domain (sh TAp63) resulted in reduced TAp63 transcript levels, this did not translate into an appreciable change in the total p63 protein levels ( Fig. 2A), confirming our observation that DNp63a variant predominates in pancreatic cancer cell lines.
Since anchorage-independent growth is a hallmark of malignant cells, we studied whether DNp63a has an effect on anchorageindependent growth and cell proliferation in PDAC. PANC-1DN, but not PANC-1TA cells exhibited increased colony formation in soft agar assays and increased cell proliferation in MTT assays (Fig. 2B). Since T3M4 cells are incapable of forming colonies in soft agar, proliferation was studied in a clonogenic assay. As compared with control cells, proliferation of T3M4 cells was reduced in response to shRNAs that target the DBD or the aspecific C-terminus and unchanged in response to sh TAp63 (Fig. 2C). Transient overexpression of mouse DNp63a cDNA, which carries a silent mutation in the region targeted by the aspecific shRNA, resulted in a partial rescue of the proliferative ability of sh p63a T3M4 cells, as compared with cells transfected with control vector (Fig. 2C).
DNp63a is known to regulate the adhesion program in mammary epithelial cells and keratinocytes [15]. Given the importance of cell adhesion in tumor invasion and progression, we studied the effects of DNp63a on the motility and invasive capabilities of the pancreatic cancer cells. In wound-healing assays, PANC-1DN cells demonstrated enhanced motility in SF conditions, while overexpression of TAp63a had no effect (Fig. 2D). Since activation of the EGFR pathway is associated with increased tumor aggressiveness and decreased survival in PDAC [21], we sought to determine the effect of EGF on motility. EGF stimulation enhanced migration in PANC-1DN and control cells (Fig. 2D). Similarly, T3M4 cells expressing sh p63a, but not sh TAp63, exhibited reduced motility both at baseline and in presence of EGF (Fig. 2E). Transient overexpression of mouse DNp63a in T3M4 cells expressing sh p63a resulted in a partial rescue of the motility phenotype (Fig. 2F). Manipulation of DNp63a expression levels had no effect on the invasion of pancreatic cancer cells in Matrigel chambers in SF conditions (Fig. 2G, H). However, stimulation with EGF led to a dramatic increase in the invasive capability of PANC1-DN cells (Fig. 2G). Consistent with the above findings, shRNA-mediated suppression of DNp63a, but not TAp63, significantly reduced the ability of T3M4 cells to invade through Matrigel in response to EGF stimulation (Fig. 2H).
Collectively, our results indicate that DNp63a enhances the colony-forming, proliferative and invasive capacity of the pancreatic cancer cells.

DNp63a upregulates EGFR and sensitizes pancreatic cancer cells to the effects of EGF
Compared with the normal pancreas, EGFR protein and mRNA are expressed at high levels in pancreatic cancer [22]. EGFR mRNA expression predicts decreased survival and poor response to chemotherapy in patients with PDAC [23]. In our work, DNp63a dramatically potentiated pancreatic cancer cell motility and invasive abilities in the presence of EGF. To explain this phenomenon, we studied the effect of DNp63a on EGFR levels. We determined that engineered expression of DNp63a in PANC-1 cells enhanced EGFR expression, while ectopic TAp63a had no effect (Fig. 3A). Consistent with this observation, we found that EGFR levels were decreased in T3M4 cells in response to shRNA-mediated suppression of DNp63a (Fig. 3B), thus documenting a direct correlation between DNp63a and EGFR expression in pancreatic cancer cell lines. To study the effects of DNp63a overexpression on EGFR signaling, we assessed activation of downstream kinases upon EGF stimulation. Treatment of PANC1-DN cells with EGF resulted in enhanced phosphorylation of ERK, Akt and JNK (Fig. 3C). In those cells, all three kinases exhibited increased level of activation at as early as 10 minutes after cell stimulation, and ERK activity persisted for 60 minutes.
Next, we employed a tyrosine kinase inhibitor to assert that the effects of DNp63a were mediated through the EGFR pathway. Erlotinib is a reversible inhibitor of the EGF signaling, which associates with the ATP-binding site of the receptor. Incubation of PANC-1 cells with 1 mM erlotinib attenuated the DNp63amediated enhancement of colony formation, motility and invasion (Fig. S2), thus excluding a possibility of a non-specific effect. By contrast, p63 protein levels were not affected when cells were incubated with either EGF or erlotinib (data not shown).
Previous reports suggested that DNp63a is able to regulate cell adhesion proteins in mammary epithelial cells [15]. Since integrinmediated tumor cell interactions with the extracellular matrix play PLoS ONE | www.plosone.org a critical role in determining the invasive phenotype in PDAC, we investigated whether DNp63a had an effect on integrin expression in pancreatic cancer cells. Ectopic expression of DNp63a in PANC-1 cells increased expression of b1-integrin, whereas the expression of a3and a5-integrins was unchanged (Fig. 3A). Conversely, shRNA-mediated suppression of DNp63a led to a decrease in b1-integrin in T3M4 cells (Fig. 3B).
Thus, DNp63a contributes to the oncogenic potential of pancreatic cancer cells, at least in part, through upregulation of EGFR and b1-integrin.

DNp63a contributes to chemotherapy resistance in pancreatic cancer cells
Since PDAC is notoriously resistant to chemotherapy agents, we studied the role of DNp63a in chemotherapy resistance. It has been previously shown that cisplatin induces degradation of DNp63a via stimulation of IkB kinase [24]. Consistent with those data, incubation of pancreatic cancer cell lines with 10 mg/ml of cisplatin resulted in degradation of endogenous DNp63a. This effect of cisplatin was evident after 6 hours of incubation and was accompanied by increased apoptosis in all tested cell lines (Fig. 4A). We then investigated whether ectopic expression of DNp63a would contribute to resistance to cisplatin-induced apoptosis in PDAC. PANC-1 cells were incubated for 24 hours in complete medium supplemented with 5 mg/mL cisplatin for 2, 6, or 24 hours. PANC-1DN cells exhibited a mild attenuation of apoptosis compared with control cells, as manifested by decreased PARP and caspase cleavage (Fig. 4B). This decrease in cisplatin-induced apoptosis correlated with an increase in viability of PANC-1DN cells as determined in an MTT assay, while PANC-1TA cells exhibited a slightly reduced survival (Fig. 4C). By contrast, T3M4 cells expressing sh p63a were sensitized to cisplatin-induced apoptosis, as demonstrated by an increased cleavage of PARP and caspase (Fig. 4D). Thus, DNp63a attenuates the sensitivity of pancreatic cancer cells to cisplatin.
To determine if DNp63a modulates 14-3-3s expression in pancreatic cancer cells, we introduced DNp63a cDNA via adenoviral infection in ASPC-1 and PANC-1 cells, which express low levels of 14-3-3s and DNp63a. We observed an increase in 14-3-3s protein levels in the DNp63a overexpressing cells compared to the control-infected cells (Fig. 5A). This DNp63amediated increase in 14-3-3s was not associated with alterations of protein levels of the other 14-3-3 isoforms (Fig. 5A). By contrast, manipulation of 14-3-3s levels in PANC-1 and T3M4 cells did not have an effect on DNp63a (Fig. 5B and 5C).

DNp63a is a transcriptional activator of EGFR and 14-3-3s in pancreatic cancer
DNp63a is a regulator of the transcriptional activities of the p53 family member proteins. As such, DNp63a may inhibit transcription via its dominant negative effect on p53 and TAp63c [27]. In addition, the DN variant is able to trans-activate p53 target genes via binding to p53 responsive elements within the corresponding gene promoters. While TAp63c is a potent gene trans-activator due to the absence of an auto-inhibitory domain [27], recent work demonstrated that it has a very low ability to bind the p53 responsive element within the EGFR promoter, and in fact represses promoter activity [28]. Since pancreatic cancer cell lines expressed very low levels of the TAp63 isoform, and since we found no correlation between DNp63a and TAp63c expression in those cell lines (Fig. 1A), the dominant negative effect of DNp63a Figure 2. Effect of DNp63a on anchorage-independent growth, motility, and invasion in pancreatic cancer cells. A, T3M4 cells were infected with GFP-expressing virus, or p63-specific shRNA complementary to DBD, TA and a-specific domains of p63. Whole-cell protein lysates were subjected to immunoblotting (top panel). Total RNA was isolated, reverse-transcribed and subjected to real-time PCR with probes specific for DN and TA isoforms of p63. Results were normalized to 18S levels (bottom panel). Knockdown of p63 isoforms was routinely monitored during the subsequent experiments. B, DNp63a stimulates anchorage-independent growth of PANC-1 cells in soft agar assay (left and bottom panels) and cell proliferation in MTT assay (right panel). PANC-1 cells were transiently transfected with DNp63a-expressing vector, TAp63a or control vector. Cell were plated in soft agar at a density of 1500/well of a 12-well plate, four wells per sample. Colonies were counted after 14 days of incubation. For MTT assay cells were plated in 96-well plates, six wells per sample. MTT was added after incubation for 48 h. Data are the mean 6 SE of four independent experiments. *, p,0.001 compared with control. C, Downregulation of DNp63a slows proliferation of T3M4 cells in a clonogenic assay. Reconstitution of DNp63a partially restores the proliferative ability. Cells were plated on 6 well plates at a density of 500 cells/well. 14 days later, plates were fixed in 3:1 methanol:glacial acetic acid and stained with 2% crystal violet. Data are the mean 6 SE of three independent experiments done in triplicates. *, p,0.01 compared with control (sh ctrl). D-F, Effect of p63 on cell motility measured in wound-healing assays in PANC-1 (D) and T3M4 cells (E, F). Cells were incubated in serum-free (SF) conditions in the absence or presence of 1 nM EGF for 18 h after making a scratch. Quantitative analysis of the images was performed. Ectopic expression of mouse DNp63a in sh p63a T3M4 cells resulted in a partial restoration of cell motility (F); corresponding DNp63a protein levels shown below. Data are the m 6 SE of at least three independent experiments. Representative pictures shown (magnification 640). *, p,0.001 compared with control (sh ctrl). G and H, Effect of DNp63a on the invasion in Matrigel chambers. PANC-1 (G) or T3M4 (H) cells were plated in Matrigel chambers (5610 4 /ml) and incubated in absence (SF) or presence of 1 nM EGF for 18 hours. Effect was normalized to invasion of control in SF conditions. *, p,0.001 compared with control (sh ctrl). doi:10.1371/journal.pone.0026815.g002 was an unlikely explanation of the changes in EGFR and 14-3-3s expression. Both EGFR and 14-3-3s promoters contain wellconserved p53 binding sites, suggesting they are transcriptional targets of DNp63a in PDAC. The 14-3-3s promoter is known to be unmethylated in pancreatic cancer cell lines [29].
DNp63a requires an intact DNA-binding domain for its transcriptional activity [18]. We generated a mutant vector where we introduced a missense mutation into human DNp63a cDNA sequence resulting in a single amino acid substitution at position 202 (DNp63a DBDmut ; Fig.S3). Enforced expression of DNp63a DBDmut in PANC-1 cells resulted in a detectable DNp63a protein, but failed to upregulate 14-3-3s (Fig. 6A) indicating that the ability of DNp63a to upregulate 14-3-3s is dependent upon its ability to bind to DNA. We tested if DNp63a was able to activate EGFR promoter in an in vitro luciferase assay. The details of the p53 binding sites within  the EGFR promoter and the corresponding luciferase vector were published previously [10]. Ectopic expression of DNp63a resulted in a dramatic increase of the activity of EGFR reporter plasmid in PANC-1 cells (Fig. 6B). The effect of DNp63a significantly exceeded that of wild-type p53 and TAp63a. Next, we introduced a single nucleotide mutation within the p53 BS 1 of the EGFR promoter. Specifically, cytosine was replaced with thymidine at position 2241 (EGFR C(2241)T ). As a further confirmation of our findings, co-transfection of a mutant EGFR reporter plasmid with wild-type DNp63a, or wild-type EGFR reporter plasmid and mutant DNp63a DBDmut resulted in attenuated luciferase activity (Fig. 6B). Consistent with this observation, ectopic expression of Figure 5. DNp63a upregulates 14-3-3s in pancreatic cancer cells. A, ASPC-1 and PANC-1 cells were infected with adenovirus containing either empty vector or full length DNp63a, and whole cell protein lysate was prepared 48 hours after infection and probed with anti-DNp63a and 14-3-3s and other 14-3-3 isoforms. B, Overexpression of 14-3-3s had no effect on DNp63a levels. PANC-1 cells were transfected with an empty vector (Sham) or with the full-length human 14-3-3s cDNA that was tagged with HA. C, Silencing 14-3-3s does not affect DNp63a. T3M4 cells were infected with lentivirus containing control or 14-3-3s specific shRNAs (sh 14-3-3s 1 and 2), whole cell lysates were prepared 72 hrs post infection and probed for DNp63a and 14-3-3s. D-F. 14-3-3s is required for the oncogenic effects of DNp63a in PDAC. PANC-1 cells were infected with control sh RNA or sh 14-3-3s 2. Subsequent to that, cells were transiently transfected with DNp63a-expressing vector or vector control. D, Anchorage-independent growth. Cells were incubated in soft agar as described in the methods. Data are the mean 6 SE of three independent experiments. E, Cells (5610 4 /well) were subjected to a migration assay in Transwell chambers in the presence of EGF (1 nmol/L) for 18 hours in duplicates. Data are the mean 6 SE of three independent experiments. F, Cells were subjected to invasion in Matrigel chambers in the presence of EGF (1 nmol/L) for 20 h in duplicates. Data are the mean 6 SE of three independent experiments. doi:10.1371/journal.pone.0026815.g005 Figure 6. DNp63a is a transcriptional activator of EGFR and 14-3-3s in pancreatic cancer cells. A, DNp63a DBDmut does not upregulate 14-3-3s. B and C, DNp63a upregulates EGFR and 14-3-3s promoter in a luciferase assay. PANC-1 cells were transfected with wildtype EGFR, mutant EGFR or 14-3-3s promoter luciferase construct as indicated, along with pCMVb plasmid, with or without wildtype p53, DNp63a, DNp63a DBDmut , or TAp63a. At 48-72 h after transfection, the luciferase activity was determined. The transfection efficiency was standardized against b-galactosidase activity. Results are indicative of four independent experiments. *, p,0.0001. D, p63 binds EGFR and 14-3-3s promoters in T3M4 cells. Binding is decreased upon serum-starvation. T3M4 cells were grown in medium containing 10% FBS (rapidly growing cells) or in serum-free medium (predominantly resting cells). Cells were pelleted and ChIP was performed as described in materials and methods. Uncrosslinked protein lysate, protein lysate immunoprecipitated with an irrelevant antibody, and a protein lysate immunoprecipitated with p63 antibody, but where irrelevant DNA sequences located ,3000 bp downstream of the 14-3-3s or EGFR promoter regions were amplified served as negative controls (the latter not shown). doi:10.1371/journal.pone.0026815.g006 p63 in Pancreatic Cancer PLoS ONE | www.plosone.org DNp63a resulted in a significant increase of the 14-3-3s reporter plasmid activity in PANC-1 cells, which was attenuated when a DNp63a DBDmut expression vector was co-transfected in place of wild type DNp63a (Fig. 6C).
Next, we tested whether DNp63a is capable of binding EGFR and 14-3-3s promoters at their corresponding p53 binding sites. Our ChIP experiments revealed that DNp63a exhibited significant binding to p53 binding sites in the endogenous EGFR and 14-3-3s promoters in T3M4 cells (Fig. 6D). Chromatin binding depended on the growth phase of T3M4 cells as serum starvation resulted in a significant decrease in promoter occupancy by DNp63a (Fig. 6D), while DNp63a protein expression was unchanged (data not shown).
We next sought to determine whether the ability of DNp63a to trans-activate EGFR and 14-3-3s promoters was unique to pancreatic cancer. We studied HEK293 cells, a transformed nonmalignant cell line, and H1299 cells, a non-small cell lung cancer cell line [30]. Similar to our pancreatic cancer cells, H1299 cells carry a mutant p53. While DNp63a is known to be expressed in non-small cell lung cancer [31], H1299 cells predominantly express the TAp63 variant [30].
Both HEK293 and H1299 cells expressed lower levels of p63 and 14-3-3s compared to T3M4 pancreatic cancer cells (Fig. 7A). Engineered expression of DNp63a in these cells did not lead to an increase in either EGFR or 14-3-3s protein levels (Fig. 7A). Similarly, DNp63a did not activate the EGFR promoter in either cell line in an in vitro luciferase reporter assay (Fig. 7B, left panels). By contrast, DNp63a transactivated the 14-3-3s promoter in H1299 cells, albeit to a lesser degree than in pancreatic cancer cells (Fig. 7B, top right panel). In HEK293 cells, DNp63a repressed the 14-3-3s promoter, similar to what was previously reported in human keratinocytes [25], whereas in H1299 cells enforced expression of DNp63a led to an increase in 14-3-3s mRNA transcript levels (Fig. 7C).
DNp63 has been shown to enhance proliferative capacity in normal and malignant cells of epithelial origin [32][33][34]. While DNp63a stimulated anchorage-independent growth, invasion and migration of pancreatic cancer cells and sensitized them to EGF, such effects were not observed in H1299 cells (Fig. 7D). Collectively, these data indicate that DNp63a binds and transactivates the EGFR and 14-3-3s promoters in pancreatic cancer cells and that these actions of DNp63a do not necessarily occur in other cell types.

Discussion
Genetic defects which result in inactivation of tumor suppression genes and accumulation of oncogenic alleles promote development and progression of PDAC. p53 is a prototypic tumor suppressor that is vital in cell growth control. However, normal p53 function is commonly lost in PDAC. p53 homolog p63 possesses an N-terminal trans-activation domain (TA), a DNAbinding domain (DBD) and a C-terminal oligomerization domain (OD). Studies in knockout mice demonstrated that despite structural similarity with p53, p63 has distinct functional properties [35]. Moreover, variability of expression of six known p63 isoforms among normal and malignant tissues suggests that the functional role of p63 is dependent on the cellular context. TAp63 isoforms may carry an anti-oncogenic potential as they mediate Ras-induced cellular senescence, antagonize tumorigenesis in vivo and suppress development of metastases through regulation of microRNA network [36,37]. The role of DNp63 in cancer is less clear. Recent studies have indicated that DNp63a is necessary and sufficient to by-pass oncogene-induced senescence suggesting that it plays a critical role in the very early steps of cancer initiation [38]. Overexpression of DNp63 is found in cancers of epithelial origin (lung, head and neck), where it promotes tumor survival [7]. DNp63 has been shown to enhance the proliferative capacity of both epithelial stem cells and cancer cells, and loss of p63 reduced the proliferative rate of MCF-7 breast cancer cells [32,33]. However, the role of p63 in PDAC is not known. Here we report that DNp63a is the predominantly expressed p63 isoform in pancreatic cancer cell lines. We demonstrate that ectopic expression of DNp63a in PANC-1 cells, which have low p63 mRNA transcript and protein levels, resulted in enhancement of anchorage-independent growth, cell motility and invasion. Conversely, shRNA-mediated suppression of DNp63a in T3M4 cells resulted in the opposite effect, i.e. decreased proliferation, promigratory and pro-invasive responses. In earlier reports, downregulation of endogenous p63 led to enhanced apoptosis in head and neck squamous cell cancer cell lines irrespective of the p53 gene status [7]. By contrast, exogenous DNp63a induced cell cycle arrest and apoptosis in p53-null non-small cell lung cancer cells [39]. In our studies, neither engineered expression of DNp63a, nor downregulation of endogenous DNp63a induced apoptosis in p53mutant pancreatic cancer cells (data not shown). While migration and invasion are both paramount to embryonic development and wound healing in normal tissues, in cancer those processes are involved in local tumor invasion and metastasis. Hence, these data underscore the importance of DNp63a in tumor progression in PDAC.
Tyrosine kinase and serine/threonine kinase pathways regulate multiple cellular processes and are major effectors in PDAC development [40]. EGFR is a 170 kDa transmembrane glycoprotein of the ErbB family of tyrosine kinase growth factor receptors. EGFR activates Ras/Raf/MAPK-ERK and PI-3 kinase/Akt pathways leading to increased cell proliferation, reduced apoptosis, increased angiogenesis, enhanced motility, invasion and metastasis [41]. EGFR signaling blockade via a dominant negative mechanism leads to reduction in mitogenic activity in pancreatic cancer cells through decreased activation of MAPK pathway [42]. In our experiments, DNp63a enhanced EGF-mediated motogenic effects and potentiated EGFR signaling as evidenced by enhanced phosphorylation of ERK and Akt and increased activation of JNK in PANC-1DN cells. We and others previously reported that EGF is capable of promoting cancer cell growth through JNK activation [42,43]. Suppression of DNp63a in T3M4 cells resulted in a remarkable attenuation of EGF-mediated invasion. Collectively, these observations suggest that PDAC requires DNp63a to fulfill its pro-invasive potential in response to EGF stimulation.
Clinical trials of EGFR-targeting agents reported modest effects on patient survival in PDAC [44,45]. The therapeutic efficacy of these agents could be improved if regulation of EGFR signaling was better understood. Here we provide evidence of a functional interplay between DNp63a and EGFR in pancreatic cancer. Ectopic expression of DNp63a, but not TAp63a, resulted in increased EGFR protein levels in PANC-1 cells. Conversely, downregulation of DNp63a in T3M4 cells led to a decreased expression of EGFR. It was initially felt that whereas TAp63 is a strong gene trans-activator, DNp63 functions as a dominant negative isoform [27]. However, recent findings point out that DNp63 is able to trans-activate p53 target genes as well as distinct targets [46,47]. TAp63c has been reported to repress EGFR promoter activity in H1299 lung cancer cells [28]. However, we determined that DNp63a is a transcriptional activator of EGFR in pancreatic cancer cells, but not in HEK293 or H1299 cells. Moreover, compared to p53, DNp63a was a strong trans-activator of EGFR in a functional reporter assay, and ChIP experiments Figure 7. A distinct role of DNp63a in HEK293 and H1299 cells. A, DNp63a does not affect EGFR or 14-3-3s protein levels in HEK293 or H1299 cells. Representative blot of three independent experiments is shown. B, DNp63a fails to upregulate EGFR promoter but modulates 14-3-3s promoter in a luciferase reporter assay. Cells were transfected with wildtype EGFR or 14-3-3s promoter luciferase construct as indicated, along with pCMVb plasmid, with or without wildtype p53, DNp63a, DNp63a DBDmut , or TAp63a. At 48-72 h after transfection, the luciferase activity was determined. The transfection efficiency was standardized against b-galactosidase activity. Results are indicative of three independent experiments performed in duplicates. C, DNp63a increases 14-3-3s mRNA transcript levels in H1299 cells. H1299 cells were transfected with DNp63a, DNp63a DBDmut or vector confirmed that DNp63a binds the EGFR promoter at the p53binding site. By contrast, TAp63a was a weak transcriptional activator of EGFR in a functional reporter assay. However, downregulation of TAp63 isoforms also decreased EGFR expression in T3M4 cells, suggesting that in those cells TAp63 isoforms could still be contributing to EGFR expression.
PDAC is notoriously resistant to conventional chemotherapy agents. This is partly explained by frequent inactivation of p53 in pancreatic tumors and thus unresponsiveness to genotoxic stress. Here we demonstrate that DNp63a further contributes to chemotherapy resistance in PDAC. While overexpression of DNp63a in PANC-1 cells resulted in a marginal increase in apoptotic death upon exposure to cisplatin, its downregulation in T3M4 cells led to a markedly increased sensitivity of those cells to cisplatin-induced apoptosis. Similar to our observations in pancreatic cancer cells, cisplatin induces proteosomal degradation of DNp63a in other cell types [24]. Additional mechanisms that contribute to cisplatin resistance in pancreatic cancer include overexpression of cyclin D1 [48] and multidrug resistanceassociated proteins [49]. Moreover, DNp63a itself may promote chemoresistance through several mechanisms. For example, it can have a dominant negative effect on the proapoptotic partners within the p53 family, as suggested by earlier experiments performed in head and neck and breast cancer [6]. Although DNp63a contributed to chemoresistance through regulation of Akt1 expression in ovarian and head and neck cancer [50], we did not observe such effect in pancreatic cancer cells (Fig. 3A). However, we highlight a new potential mechanism of DNp63amediated chemoresistance in PDAC. In our earlier experiments, 14-3-3s dramatically increased chemoresistance and enhanced the pro-invasive potential of pancreatic cancer cells [11]. Here we demonstrate that expression of 14-3-3s protein correlates with DNp63a levels in those cells. While others have identified 14-3-3s as a DNp63 repression target in human embryonic keratinocytes [25], we established that DNp63a activates the 14-3-3s promoter and upregulates 14-3-3s protein expression in pancreatic cancer cells (and in H1299 lung cancer cells), thus decreasing sensitivity of pancreatic cancer cells to cisplatin-induced apoptosis. This discrepancy provides additional evidence that DNp63a actions are tissue-specific, and its function in cancer is distinct from its role in embryonic cells. We also found that 14-3-3s promoter binding by DNp63 was more pronounced in rapidly cycling as compared with resting T3M4 cells. This suggests that at least some DNp63a target sites may be poorly accessible in quiescent cells and require chromatin remodeling to occur prior to binding, further implicating cellular context in modulation of p63 function.
In our immunoprecipitation experiments we found no interaction between p63 and 14-3-3s proteins (data not shown), however 14-3-3s contributed to the mitogenic and motogenic effects of DNp63a in PDAC. Lentiviral knockdown of 14-3-3s resulted in a dramatic attenuation of the effects of DNp63a on anchorageindependent growth and EGF-stimulated invasion of PANC-1 cells. Silencing of 14-3-3s had no consequences on the effects of DNp63a on cell motility, leading to the conclusion that DNp63a affects migration through potentiation of the other signaling pathways, i.e. EGFR and integrins.
p63 controls the adhesion program in MCF-10A mammary epithelial tissues through transcriptional regulation of integrins [15]. A number of integrin subunits have been shown to be upregulated in PDAC [51]. We demonstrate that DNp63a modulates expression of b1-integrins, which are involved in regulation of cell adhesion and migration of tumor on stroma proteins, thus further implicating DNp63a in pancreatic carcinogenesis. Importantly, integrin and EGFR pathways are closely intertwined and orchestrate cancer growth and invasion. Integrin-mediated adhesion of cells to extracellular matrix induces EGFR activation in a ligand-independent manner, while EGFR regulates integrin signaling and is necessary for adhesion-induced activation of ERK and other signaling molecules [52]. Cross-talk between integrins and EGFR family members affects multiple aspects of tumor progression, including proliferation, migration and invasion [53]. For example, blockade of integrin a v b 5 reverses the EGF-stimulated invasion and metastasis in pancreatic cancer cells [54]. On the other hand, overexpression of b1-integrin has been shown to correlate with acquired resistance to EGFR inhibitors in lung cancer [55]. The unique ability of DNp63a to regulate both EGFR and b1-integrin in PDAC renders it a particularly promising therapeutic target.
In conclusion, DNp63a is the predominantly expressed p63 variant in PDAC cell lines with oncogenic properties. In PANC-1 and T3M4 cells DNp63a enhanced anchorage independent growth, cell proliferation, and basal and EGF-stimulated motility and invasion and conferred chemoresistance. The oncogenic effects of DNp63a were mediated via transcriptional activation of EGFR and 14-3-3s. Our observations indicate that targeting DNp63a and 14-3-3s may present a strategy to potentiate the efficacy of EGFR-targeting therapies in PDAC.

Supporting Information
Table S1 p63 primer sequence and expected size of PCR products.  Figure S2 Erlotinib attenuates DNp63a-mediated enhancement of migration, invasion and anchorage-independent growth in PANC-1 cells. A, Erlotinib attenuates DNp63a-mediated anchorage-independent growth of PANC-1 cells in soft agar assay. Soft agar assay was performed as described previously in presence of 1 mM erlotinib or vehicle control. B, PANC-1 cells (5610 4 /well) were subjected to a migration assay in Transwell chambers in presence of EGF (1 nmol/L) as described above, with or without 1 mM erlotinib. C, PANC-1 cells (1610 4 / well) were subjected to invasion assay in Matrigel chambers in presence of EGF (1 nmol/L) as described above, with or without 1 mM erlotinib. *, p,0.05 compared with control. (TIF) Figure S3 DNp63a DBDmut cDNA-expressing vector. Partial sequence of the DNp63a cDNA shown (nucleotides control. At 36 hours total RNA was isolated, reverse-transcribed and subjected to real-time PCR with probe specific for 14-3-3s. Results were normalized to 18S levels. Data are the mean of two independent experiments done in duplicates in which similar results were obtained. D, DNp63a does not enhance anchorage-independent growth, migration or invasion in H1299 cells. H1299 cells were transfected with DNp63a or vector control. Soft agar, transwell migration and invasion assays were performed as described above. *, p,0.05 compared with control. doi:10.1371/journal.pone.0026815.g007 p63 in Pancreatic Cancer PLoS ONE | www.plosone.org 601 to 612, within DBD). Substituted nucleotide is shown in capital. (TIF)