Snail-Induced Down-Regulation of ΔNp63α Acquires Invasive Phenotype of Human Squamous Cell Carcinoma Free

p63 is a member of the p53 family and regulates crucial events in the formation of epithelial structures, but the role of p63 in tumor is unclear. We found that Snail-induced epithelial-to-mesenchymal transition (EMT) is accompanied by down-regulation of p63 in human squamous cell carcinomas (SCC). ΔNp63α is the predominantly expressed p63 isoform in SCC cells. ΔNp63 promoter activity required a CAAT/enhancer binding protein (C/EBP) binding element and was reduced remarkably by Snail. Down-regulation of ΔNp63α and reduction of C/EBPα were observed in EMT phenotype cells, which exhibited invasive activity in vitro. p63 knockdown in cells enhanced invasive activity in the presence of E-cadherin. Conversely, forced expression of ΔNp63α blocked invasive activity of cells with the EMT phenotype. These findings indicate that Snail down-regulates ΔNp63α, leading to acquisition of the invasive phenotype by SCC. The invasive activity caused by down-regulation of ΔNp63α does not require down-regulation of E-cadherin. [Cancer Res 2007;67(19):9207–13]

In embryonic development, epithelial-to-mesenchymal transition (EMT) is the process of disaggregating structured epithelial units to enable cell motility and morphogenesis (1, 2). Wound healing and progression of carcinomas to invasive and metastatic phenotypes also involve localized EMT (3, 4). The term “EMT” comprises a wide spectrum of changes in epithelial plasticity. Among these EMT subtypes, “complete EMT,” defined by a fibroblastoid phenotype, loss of E-cadherin, and gain of vimentin, a mesenchymal marker, was most closely correlated with local invasion (5). E-cadherin is a cell-cell adhesion molecule expressed on the cell membrane of epithelial cells. Loss of E-cadherin expression is a primal molecular event that contributes to tumor invasion and metastasis (6). Snail, a zinc finger transcription factor, triggers EMT through direct repression of E-cadherin (7, 8). The reverse correlation of Snail and E-cadherin has been reported for various human cancers, including squamous cell carcinoma (SCC; refs. 7, 9–11). Other zinc finger transcription factors, SIP1 (ZEB-2, ZFHX1B), Slug, and Twist, were also reported to repress E-cadherin and induce EMT (12–14). In this context, EMT has attracted attention in studies of tumor progression.

p63 (TP73L/TP63) is a member of the p53 gene family (15, 16) and has two different promoter usage-generating proteins containing (TA) or lacking (ΔN) an NH₂ terminus, which is homologous to the transactivation domain of p53. ΔNp63 isoforms act as transcription repressors in a dominant-negative fashion to oppose p53- or TAp63-mediated transactivation in vitro and in vivo (17). However, ΔNp63 isoforms also display transcriptional activity that is independent of the transactivating domain (18). p63^−/− mice have striking developmental defects, including complete lack of all stratified squamous epithelia, epidermal appendages, and mammary, lacrimal, and salivary glands (19, 20). Heterozygous germ-line mutations in p63 are the cause of ectrodactyly, ectodermal dysplasia, and facial clefts syndrome (EEC syndrome; 21). Thus, p63 plays essential roles during development in the formation of epithelial structures. In contrast to p53, p63 is rarely mutated in human cancers, and the role of p63 in tumors is still unclear (reviewed in ref. 22), although some links to the DNA damage response pathway have been reported (15, 23). Upstream transcriptional regulators for the p63, particularly of ΔNp63, are poorly understood in contrast to the many downstream target genes of the p63 that have recently been reported (reviewed in ref. 22). Here, we report a novel mechanism whereby down-regulation of ΔNp63α by Snail enhances invasiveness of SCC cell in parallel with down-regulation of E-cadherin.

The expression vector plasmids. ΔNp63α full-length cDNA (Genbank accession number AB042841) was amplified by reverse transcription-PCR (RT-PCR) with Pfx polymerase (Invitrogen) and cloned into the NheI and XbaI sites of pcDNA6-V5/His-tagged expression vector (Invitrogen). Primers for amplification were 5′-GCTAGCAACATGTTGTACCTGGAAAACAATGCCC-3′ and 5′-TCTAGAGGAACTCCCCCTCCTCTTTGATGC-3′. CAAT/enhancer binding protein α (C/EBPα) full-length cDNA (Genbank accession number M37197) was amplified by RT-PCR with LA Taq polymerase with GC buffer (TaKaRa BIO) and cloned into the KpnI and XhoI sites of pcDNA3.1-V5/His-tagged expression vector (Invitrogen). Primers for amplification were 5′-GGTACCATGGAGTCGGCCGACTTCTA-3′ and 5′-CTCGAGTCTCGCGCAGTTGCCCATG-3′. The sequence of these PCR products was verified by sequencing.

Cell lines and cell culture. The human vulval epidermal cell line A431 and four human oral SCC cell lines, OM-1, HOC719, HOC313, and TSU, have been described previously (9). HOC719-PE (positive E-cadherin) and HOC719-NE (negative E-cadherin) cells were isolated from HOC719 cells, which express E-cadherin heterogeneously (9). All SCC cell lines lost the functional p53 genetically (24). A431-SNA1 and OM-1-SNA1 cells were generated by transfection with pcDNA3-mm SnailHA (Genbank accession number BC034857), kindly provided by Dr. de Herreros (Universitat Pompeu Fabra, Barcelona, Spain), as described previously (25). GT-1 cells are immortalized fibroblasts derived from human gingiva by transfection with an hTERT expression vector (26). The ΔNp63α expression vector or the empty pcDNA6-V5/His vector as control was transfected into HOC313 cells, and stable cell clones were established by blasticidin selection. All cell lines were cultured at 37°C in a humidified atmosphere of 5% CO₂ in air and maintained with DMEM (Sigma) supplemented with 10% fetal bovine serum (FBS; Sigma).

RNA extraction and first-strand cDNA synthesis. Total RNAs were isolated from the cells in 70% to 80% confluence with Trizol (Invitrogen). First-strand synthesis was done with First-Strand cDNA Synthesis kit (Roche).

Semiquantitative RT-PCR. RT-PCRs (20 μL) were amplified with 30 cycles of denaturing at 95°C for 30 s, annealing for 30 s, and extension at 72°C for 1 min. For amplification of specific regions of TAp63 and ΔNp63, the primers and annealing temperatures used were described previously (15). For other amplifications, primers and annealing temperatures were follows: p63, 5′-TCCTCAGGGAGCTGTTATCC-3′ and 5′-ACATACTGGGCATGGCTGTT-3′, 56°C; ΔNp63α, 5′-ATGTTGTACCTGGAAAACAATG-3′ and 5′-ATCTGATAGATGGTGGTCAGCC-3′, 56°C; p63αβ, 5′-GGCCGTTGAGACTTATGAAATGC-3′ and 5′-GCTCAGGGATTTTCAGACTTGC-3′, 56°C; p63γ, 5′-GGCCGTTGAGACTTATGAAATGC-3′ and 5′-CTCTATGGGTACACTGATCGGTTT-3′, 56°C; C/EBPα, 5′-CAGACCACCATGCACCTG-3′ and 5′-TTGTCACTGGTCAGCTCCAG-3′, 58°C; and G3PDH, 5′-ACCACAGTCCATGCCATCAC-3′ and 5′-TCCACCACCCTGTTGCTGTA-3′, 52°C. PCR products were analyzed by 1.8% agarose gel electrophoresis and sequenced to verify their identity.

RNA interference. Small DNA fragment encoding a short hairpin RNA (shRNA) targeting all of the p63 isoforms was cloned into pRNA-U6.1 (GeneScript). The sequences of the short interfering RNAs (siRNA) were follows: p63 siRNA, GGUACCAGCACACUCUGUCUU; control siRNA, GUCGAUCCGAACACUCUCUGU. Vectors were transfected into A431 and OM-1 cells, and stable cell clones were selected with hygromycin.

Cell lysates and immunoblotting. Cells were harvested and lysates were prepared according to standard methods. Immunoblotting was also done according to a standard method. Antibodies were anti-p63, which are specific for ΔNp63 isoforms (Ab-1, Oncogene Research Products), anti-E-cadherin (H-108, Santa Cruz Biotechnology), and anti-α-tubulin (Zymed Laboratories).

Luciferase reporter assay. The ΔNp63 promoter region of nucleotide −558 to +262 (construct 1) was amplified with Pfx polymerase from genomic DNA of normal human fibroblasts with primers as described previously (27). Other fragments of the ΔNp63 promoter region were also amplified by PCR. The sense primer sequences for each fragment were follows: nucleotide −203 (construct 2), 5′-GGTACCGAAATGCCTTCTGTAAATCG-3′; nucleotide −167 (construct 3), 5′-GGTACCTGTTTGGGGAGATTTGTTTTGTTTT-3′; nucleotide −160 (construct 4), 5′-GGTACCGGAGATTTGTTTTGTTTTTAAAAGACAGTGCA-3′; nucleotide −115 (construct 5), 5′-GGTACCGAGACAGGGAAAGTTTTACC-3′; and nucleotide −44 (construct 6), 5′-GGTACCGATTGGTGATAAGGAATTC-3′.

Each PCR product was cloned into the KpnI and XhoI sites of pGL3-basic vector (Promega). The mutant C/EBP binding element clone (pGL3-ΔNp63_C/EBP mt) was also generated by changing AGATTT (italicized: nucleotide −158 to −155 on a putative C/EBP binding element) to GCTAGC in the construct 1. A431 cells were cotransfected with 4 μg of the reporter construct containing the ΔNp63 promoter sequence and 1 ng of phRL-CMV (Promega) as an inner control for transfection efficiency with LipofectAMINE 2000 (Invitrogen). Either empty pcDNA3, pcDNA3.1_C/EBPα His/V5, or pcDNA3-mm SnailHA was further transfected. At 48 h after transfection, cells were lysed with passive lysis buffer, and the promoter activity was measured with a Dual-Luciferase Reporter Assay System (Promega) according to the manufacturer's protocol. Briefly, a Firefly-derived luciferase activity of interest was measured by addition of the Firefly-specific substrate to the transfected cell extract. Subsequently, Renilla-derived luciferase activity was measured by further addition of the Renilla-specific substrate. Transfection efficiency was normalized to cytomegalovirus-driven Renilla-derived luciferase activity. The Firefly-derived luciferase activity of interest was normalized against transfection efficiency and presented in fold change toward normalized activity of the construct 1. The results correspond to the mean of at least three independent experiments.

Immunocytochemistry. The cells cultured on Lab-Tek II Chamber Slides (Nalgene Nunc) were fixed with 4% paraformaldehyde and treated with 0.1% Triton X-100 for 5 min and 1% bovine serum albumin in PBS for 30 min. To detect localization of filamentous actin (F-actin) and E-cadherin in cells, Alexa Fluor 488 phalloidin (Invitrogen) and rabbit anti-E-cadherin antibody (H-108) were used for staining. The cells were further incubated with Alexa Fluor 568–labeled goat anti-rabbit IgG (Invitrogen) for 45 min. After mounting with Vectashield (Vector Laboratories), the cells were subjected to fluorescent microscopy.

Matrigel cell invasion assay. Cell invasion activity was measured with BioCoat Matrigel Invasion Chamber (Becton Dickinson) according to the manufacturer's protocol. Briefly, cells were suspended in DMEM containing 5 × 10⁵/mL. Cell suspension (500 μL) was added to each chambers containing an 8-μm pore size PET membrane with Matrigel basement membrane and incubated for 12 to 48 h at 37°C and 5% CO₂ atmosphere. Cells on the bottom surface of the membrane were fixed with 4% paraformaldehyde and stained with trypan blue and counted as the invading cells.

In vitro three-dimensional culture. Three-dimensional cultures of epithelial cells with contracted collagen gel containing fibroblasts were prepared as described (28). GT-1 fibroblasts were suspended in a mixture of type I collagen (Koken) and DMEM containing 10% FBS and seeded in 12-well culture dishes. The collagen was allowed to solidify by incubating at 37°C for 1 h. The final concentrations of collagen and GT-1 fibroblasts were 1 mg/mL and 1 × 10⁶ cells/mL, respectively. SCC cells (1 × 10⁶) suspended in 1 mL of culture medium were seeded on the collagen gel. After incubation at 37°C for 1 h, the gels were removed from the sides and bottoms of dishes and floated in the medium. After 1 week of incubation, the contracted gel was placed on a nylon mesh, and culture medium was added until the fluid level reached the upper edge of the gel. The gels were incubated under air-liquid interface culture for 1 more week. Culture medium was changed every second day. The gel was fixed with Mildform (Wako), embedded in paraffin, and stained with H&E as described previously (29).

Down-regulation of p63 is accompanied by Snail-induced EMT. The A431 and OM-1 cell lines derived from human SCCs have an epithelial manner, which includes cuboidal cell shape, cell-cell adhesion, and strong expression of E-cadherin.

First, we examined expression of p63 and its isoforms in A431 and OM-1 cells by semiquantitative RT-PCR. Primers specific for p63 isoforms are shown in Fig. 1A. A primer set to amplify all transcripts from p63 clearly detected p63 mRNA in A431 and OM-1 cells (Fig. 1B). The series of primer sets to detect a specific isoform revealed dominant ΔN isoforms expression in these cells (Fig. 1B). Using the specific primer sets that discriminate the COOH-terminal variants, α-form was dominantly detected (Fig. 1C). Therefore, ΔNp63α is the p63 isoform specifically expressed in SCC cells with epithelial phenotype, which is consistent with previous reports (30).

As previously shown (25), exogenous expression of Snail suppressed E-cadherin completely, which was believed to confer EMT phenotype on SCC cells. We found that exogenous Snail also provoked a complete reduction of ΔNp63α in A431 and OM-1 cells (Fig. 1B and D).

Suppression of C/EBPα-dependent ΔNp63 promoter activity by Snail. To confirm that ΔNp63α expression is influenced by Snail, the ΔNp63 promoter activity in response to Snail was monitored by luciferase reporter assay. A431 cells were transiently cotransfected with the ΔNp63 promoter constructs and Snail expression vector (Snail (+)) or empty vector (Snail (−); Fig. 2A). A major basal transcriptional activity was observed in A431 cells, which were transfected with the longer reporter genes (constructs 1 to 3 in Fig. 2A). A responsible element for the clear basal activity was narrowed down between constructs 3 and 4 as indicated in Fig. 2A, although a minor transcriptional activity was still observed in reporter genes lacking the elements (constructs 4 and 5) compared with the shortest TATA-less reporter gene (construct 6). Putative RREB-1, C/EBPα, and GATA-1 binding elements were found around 5′ of the construct 3 (TFSEARCH¹

; Fig. 2B). Because the construct 4 containing the intact GATA-1 binding element exhibited only minor transcriptional activity (Fig. 2A), we determine whether the close-by C/EBPα binding element or RREB-1 binding element was required for the major basal transcriptional activity. A reporter gene (construct 1) with a mutated element only for C/EBPα binding (C/EBP mt construct in Fig. 2B) exhibited significantly lower activity than the wild-type promoter construct in A431 cells (Fig. 2C). These data identify the C/EBPα, a member of the C/EBP family of transcription factors, abundantly expressed in keratinocytes and modulates squamous differentiation (31), as a positive regulatory transcriptional factor for the ΔNp63 expression. Indeed, the C/EBPα binding to the ΔNp63 promoter was specifically observed in electrophoretic mobility shift assay² and exogenous C/EBPα expression enhanced the major transcriptional activity of the ΔNp63 reporter gene (2.13-fold) but not the C/EBP mt construct (Fig. 2C).

The coexpressed Snail consistently reduced the major transcriptional activities of the ΔNp63 longer reporter genes (constructs 1 to 3 in Fig. 2A) to the extent of the C/EBPα-independent minor activities (constructs 4 and 5 in Fig. 2A). Moreover, the minor activity of the C/EBP mt construct failed to be suppressed by coexpressed Snail (Fig. 2C). These data indicate that exogenously expressed Snail exerted its suppressive activity to the ΔNp63 promoter in control of C/EBPα.

Because Snail suppresses expression of E-cadherin by its binding to the E-cadherin promoter directly (7, 8), we next determined whether Snail directly controls the C/EBPα expression to suppress ΔNp63 expression. Despite that exogenous Snail was able to reduce the C/EBPα-dependent transcriptional activity of the ΔNp63 promoter, the abundance of C/EBPα in A431 and OM-1 cells was not altered by Snail (Fig. 2D).

The C/EBPα abundances in SCC cell lines correlated to ΔNp63α expressions and reversely correlated to EMT phenotype. We next analyzed p63 expression in cells with naturally acquired EMT phenotype by semiquantitative RT-PCR and immunoblotting (Fig. 3). The cells with EMT showed complete suppression of all transcripts encoded by p63 (Fig. 3). Intriguingly, concomitant reduction of C/EBPα was clearly observed in these cells, although cells without the EMT features obviously express both C/EBPα and ΔNp63α (Fig. 3). These data suggest that the spontaneous loss of C/EBPα in invasive cells accelerated the ΔNp63α reduction instead of Snail mediated the suppression to C/EBPα and that Snail might exert its suppressive activity toward E-cadherin.

The acquired invasiveness in p63 knockdown SCC cells was independent of EMT. To confirm the functional contribution of ΔNp63α to tumor invasion, we generated p63 knockdown SCC cell lines. A431 and OM-1 cells displayed slight changes in cell shape, cell-cell adhesion, or cell growth after p63 knockdown (Supplementary Fig. S1A and B). Expression of vimentin was not elevated remarkably by the forced depletion of ΔNp63α in these cells (Supplementary Fig. S3). Under these conditions, E-cadherin expression was observed (Fig. 4A). To confirm the functional expression of E-cadherin, immunocytochemistry was done. E-cadherin was localized properly on the cell membrane (Fig. 4B). These findings indicate that loss of p63 did not result in EMT.

The invasion assay with Matrigel basement membrane allows assessment of migration and invasive property of SCC cells. Surprisingly, p63 knockdown remarkably increased the invasiveness of A431 and OM-1 cells (9.2- and 12.9-fold, respectively; Fig. 4C). In vitro three-dimensional culture reconstructs the squamous epithelial structure by overlaying SCC cells on the collagen I gel containing immortalized fibroblasts. Intriguingly, the p63 knockdown cells formed a single-layer surface on the gel and invaded into the gel layer, although the corresponding control cells remained confined to the upper gel surface and stratified (Fig. 4D). These invasion assays provide evidence that loss of p63 leads to an acquisition of high invasiveness in vitro. It is noteworthy that the invasive activity acquired by loss of p63 does not require down-regulation of E-cadherin.

Suppression of the invasiveness by ΔNp63α. HOC313 cells exhibited the EMT features and invasiveness in vitro, which is characterized by fibroblastoid shape and scattered growth due to release of cell-cell adhesions (9). Because HOC313 did not express any transcripts despite its intact p63 allele (data not shown), we generated forced ΔNp63α-expressing HOC313 cells as a gain-of-function approach. The exogenous ΔNp63α did not affect cell morphology, growth rate, or vimentin expression in HOC313 cells (Supplementary Figs. S2A and B and S3, respectively). E-cadherin expression was not induced by the exogenous ΔNp63α (Fig. 5A). Therefore, HOC313 cells still show the EMT features even in the reexpression of ΔNp63α. However, the number of invading HOC313 cells significantly decreased 0.42-fold in response to the reexpression of ΔNp63α in the Matrigel invasion assay (Fig. 5B). Moreover, most forced ΔNp63α-expressing cells were unable to invade the gel layer and formed stratified layers on the collagen gel in in vitro three-dimensional cultures (Fig. 5C). Taken together, these findings indicate that ΔNp63α prevents the invasiveness of SCC cells with the spontaneous EMT phenotype and, conversely, loss of p63 leads to acquisition of an invasive activity regardless of functional E-cadherin expression.

In this study, we showed that Snail inhibits expression of ΔNp63α and that forced depletion of ΔNp63α enhances invasiveness of SCC cell, whereas reexpression of ΔNp63α in ΔNp63α-deficient cell suppresses its invasive activity in vitro. Furthermore, the invasive activity caused by the depletion of ΔNp63α does not require down-regulation of E-cadherin and up-regulation of vimentin and vice versa. These findings suggest that loss of ΔNp63α provides an invasive phenotype to SCC cell in parallel with complete EMT.

The molecular mechanism of tumor invasion involves altered interactions between tumor cells and their environment as well as intracellular and intercellular events, such as cell proliferation, loss of cell-cell adhesion, acquisition of cell motility, and loss of cell polarity. How does down-regulation of ΔNp63α acquire an invasive activity in SCC? In our data, ΔNp63α-deficient SCC cells formed a single layer on the collagen gel and invaded into the gel layer in in vitro three-dimensional cultures. During development of the epidermis, ΔNp63 is expressed during a late stage of the single-layered surface ectoderm and allows basal keratinocytes to commit to epidermal maturation and terminal differentiation (32). p63^−/− mice have striking developmental defects, including complete lack of all stratified squamous epithelia, epidermal appendages, and mammary, lacrimal, and salivary glands (19, 20), which suggests that p63 plays fundamental roles in formation of epithelial structures. The fact that ectoderm extends epithelial sheets to form buds into the mesoderm during development of epidermal appendages and derivative organs (33) seems to contradict our present results; however, extended epithelial sheets or buds maintain the stratified epithelial structure. The localized expression of ΔNp63α in basal keratinocytes strongly suggests its role in formation of stratified squamous epithelial structures by regulating asymmetrical division (34). Polarity of epithelial cells directs the apical basal and planar axes and plays crucial roles for cell-cell adherens and tight junctions and asymmetrical cell division (35). We speculate that down-regulation of ΔNp63α occurs in SCC cells at the invasive front, which may cause loss of cell polarity, and loss of cell polarity then promotes invasion into the adjacent connective tissue.

Because p63 is a member of the p53 family, many studies of p63 functions have been reported; nonetheless, the role of p63 in tumors is not well understood (reviewed in ref. 22). Fluorescent in situ hybridization analysis revealed frequent amplification of the p63 locus in primary SCC of the lung and head and neck cancer cell lines (36). Recently, a genome-wide microarray analysis revealed that the 3q26-29 locus encompassing p63 is frequently amplified in SCCs of the lung, suggesting that overexpression of p63 facilitates tumorigenesis (37). Barbieri et al. (38) reported that microarray analysis identified genes associated with invasion and metastasis due to loss of p63 in keratinocytes. Loss of p63 caused down-regulation of cell adhesion–associated genes, cell detachment, and anoikis in mammary epithelial cells and keratinocytes (39). Furthermore, ΔNp63α expression is directly correlated with the clinical response to cisplatin in SCCs of head and neck (40). These reports indicate that varied levels of ΔNp63α expression might decide cell fate. Our present study provides enforced evidence that loss of p63 directly involves tumor invasion. Taken together, these reports and our novel findings suggest that the overexpression of ΔNp63α might involve tumorigenesis of keratinocyte, and its suppression by Snail during progression of SCC leads to the tumor invasion.

Figure 6 here shows the schematic representation of our novel findings, indicating that Snail down-regulates ΔNp63α via suppression of C/EBPα-dependent transcription, leading to acquisition of an invasive phenotype, in parallel with down-regulation to E-cadherin. Reduction of C/EBPα itself also results in loss of ΔNp63α. During progression of SCC toward a more malignancy, Snail is expressed as an initial event and then down-regulates both E-cadherin, resulting in the EMT features, and ΔNp63α, resulting in acquisition of the invasiveness. The precise mechanism by which loss of ΔNp63α exerts the invasiveness or Snail interferes in the C/EBPα-dependent ΔNp63 expression remains defined and we currently investigate these issues. However, our results might provide a new strategy for therapeutics for progressive SCC through the modulation of ΔNp63α expression.

Grant support: Grant-in-Aid for Scientific Research (C: 17592085) from the Ministry of Education, Culture, Sports, Science and Technology, Japan.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

We thank Fumiko Higashikawa for critical discussion and encouragement.