Purpose: To investigate the regulation of epithelial-to-mesenchymal transition (EMT) in head and neck squamous cell carcinoma (HNSCC) and its importance in tumor invasion.

Experimental Design: We use a three-dimensional invasive organotypic raft culture model of human foreskin keratinocytes expressing the E6/E7 genes of the human papilloma virus-16, coupled with bioinformatic and IHC analysis of patient samples to investigate the role played by EMT in invasion and identify effectors and upstream regulatory pathways.

Results: We identify SNAI2 (Slug) as a critical effector of EMT-activated downstream of TP63 overexpression in HNSCC. Splice-form–specific depletion and rescue experiments further identify the ΔNp63γ isoform as both necessary and sufficient to activate the SRC signaling axis and SNAI2-mediated EMT and invasion. Moreover, elevated SRC levels are associated with poor outcome in patients with HNSCC in The Cancer Genome Atlas dataset. Importantly, the effects on EMT and invasions and SNAI2 expression can be reversed by genetic or pharmacologic inhibition of SRC.

Conclusions: Overexpression of ΔNp63γ modulates cell invasion by inducing targetable SRC-Slug–evoked EMT in HNSCC, which can be reversed by inhibitors of the SRC signaling. Clin Cancer Res; 24(16); 3917–27. ©2018 AACR.

This article is featured in Highlights of This Issue, p. 3785

Translational Relevance

We used The Cancer Genome Atlas datasets from patients with head and neck squamous cell carcinoma (HNSCC) and patient-derived tumor sections from human papillomavirus (HPV)-positive/-negative oropharyngeal HNSCC to clinically validate in vitro data generated using primary human foreskin keratinocytes (HFK), expressing the HPV16 E6/E7 oncogenes. The novel and translational aspects of this study are as follows:

  1. Slug/SNAI2 is the main epithelial-to-mesenchymal transition (EMT)-activating transcription factor in HNSCC and E6/E7-HFK.

  2. Activation of SRC and downstream targets mediates the Slug/SNAI2-evoked EMT.

  3. We show for the first time that a particular p63 isoform, namely ΔNp63γ, is necessary and sufficient to activate SRC signaling axis, induce EMT, and invasion.

This manuscript is relevant to those investigating (i) the oncogenic significance of p63 transcription factors, (ii) the role of upstream pathways in the activation of Slug, and (iii) the therapeutic potential of SRC inhibitors in clinical trials in patients with EGFR-resistant HNSCC.

Head and neck squamous cell carcinoma (HNSCC) represents the sixth most commonly diagnosed cancer worldwide (>500,000 new cases yearly; ref. 1). The frequency of HNSCC is increasing, particularly in a younger age group associated with human papilloma virus (HPV) infection, which accounts for 23% to 60% of cases (2). Currently, there are limited options for biomarker-guided, molecular-targeted therapies in HNSCC because of a poor understanding of the disease at the molecular level.

Amplification and overexpression of TP63 occurs independently of TP53 mutation or HPV infection in the majority of squamous cell carcinomas (SCC; refs. 3, 4), the clinical importance of which remains unclear. The TP63 locus is complex and encodes at least six well-described isoforms that play overlapping but distinct roles in activating or repressing target gene expression. This occurs as a result of utilization of two alternative promotors resulting in TA (transactivation) and ΔN (lacks the TA domain) N-terminal isoforms, each of which can be alternatively spliced with C-terminal variants α, β, and γ (5). Traditionally, elevated expression of the predominant ΔNp63α isoform has been assumed to promote SCC growth by opposing activation of canonical TP53/TP73-regulated cell-cycle repressive and proapoptotic targets (6, 7). Importantly, recent data from our group and others indicate that TP63 plays essential TP53-independent roles in promoting and maintaining squamous transformation stimulating invasion and migration and is paradoxically, necessary to induce differentiation in normal cells (8–10). Our integrative genome-wide analyses of TP53–TP63 function highlighted the extent of the network of genes potentially affected by the TP63–TP53 axis, and the importance of upregulation of TP63 target genes in HNSCC (11, 12).

Previously, we and others have shown that TP63 is involved in supporting an epithelial-to-mesenchymal transition (EMT) phenotype in normal breast epithelial cells (13, 14). E-cadherin (CDH1), an adherens junction protein and an epithelial marker, is essential for knitting the epithelial cells together, and the loss (through suppression of CDH1 expression and/or its relocalization away from cell–cell contacts) is critical for the acquisition of EMT (15). This can be mediated through the activities of EMT-inducing transcription factors including Twist (TWIST1), Snail (SNAI1), and Slug (SNAI2), which are known to directly repress transcription from the CDH1 promoter, thereby promoting the disassembly of the cell–cell contacts (16, 17). Vimentin, a hallmark of EMT encoded by the VIM gene, is also overexpressed in malignant epithelial breast and vulvar cancers and correlates with poor prognosis (18, 19). Hence, in invasive cancers, several molecular pathways are altered to support the upregulation and protein stabilization of EMT-promoting genes.

We recently used a three-dimensional (3D)–organotypic model of human foreskin keratinocytes (HFK), stably expressing the high-risk HPV-16 E6 and E7 oncoproteins (E6/E7-HFK) to identify the non-receptor tyrosine kinase Src, as a TP63 target gene in oro-pharyngeal cell carcinoma, a subset of HNSCC (8, 20, 21). Src is a critical regulator of cell migration, the upregulation of which has been observed in several cancers including HNSCC, where this has a direct correlation with disease progression (18, 19). Our study elucidated the importance of TP63 in transcriptionally regulating a Src–MMP axis that is required for migration and invasion, which could be inhibited by TP63 or Src depletion or Src activity inhibition.

In this study, we show for the first time that the ΔNp63γ isoform is an important factor in regulating EMT through Slug/SNAI2 and SRC and that the EMT phenotype and invasion can be reversed by inhibition of SRC activity.

Cell culture

Primary neonatal HFK expressing an empty vector (pBabe-HFK) or E6/E7-HFK were cultured as monolayer to subconfluence in Epilife medium on collagen-I–coated plates before harvesting mRNA and proteins. A fraction of these cells were seeded on Rb-depleted human foreskin fibroblast embedded in collagen-I plugs to establish 3D-organotypic rafts for studying invasion (8, 22). After 14 days of incubation in E-medium, the rafts were sliced, embedded in paraffin, sectioned, and used for immunofluorescence and H&E stains. The invasive incidents were quantified using H&E-stained sections and represented as number of invasions/cm recorded from three independent experiments.

Retroviral constructs and stable knockdown

The stable knockdowns of nonspecific (scram) or p63 isoforms (UTR, DBD) in E6/E7-HFK were established by shRNA molecules (9) ligated in the pSuper-retro-neo constructs before their transfection in the 293T cells. The retroviruses generated using the phoenix system were used to infect HFK and GFP-tagged constructs acted as positive controls for measuring infection efficiency (22).

Adenovirus-mediated overexpression

To generate shRNA resistance ΔNp63α/β/γ isoforms, 5 μg of entry vector pENTR11 (Life Technologies) carrying the gene of interest (GFP or ΔNp63α/β/γ) were subjected to site-directed mutagenesis for four-point mutations as previously described (13), and were sequence verified before recombination with the adenoviral pAd/CMV/V5-DEST Gateway vector (Life Technologies) using Gateway-LR Clonase-II enzyme mix (Life Technologies) according to the manufacturer's guidelines. Similarly, the constitutively active Src (Src-531) construct was generated by site-directed mutagenesis (8) before recombination with adenoviral vector. The recombined vector was transfected in 293T cells before generation, purification, and titration of adenovirus as previously described (23).

Transient knockdown

Transient gene/protein knockdowns in E6/E7-HFK were established by transfection with 50 nmol/L of nonspecific (scram) or specific siRNA molecules targeting Slug (mol-1:5′-CAAACGACTTTGCAACTCC-3′, mol-2:5′-CCTCTTGGCATACTCCTCT-3′; ref. 24), Src, and p63 (UTR, Pan) as previously described (8).

RNA extraction and qRT-PCR

The total RNA was harvested from HFK samples using Trizol reagent (Roche), according to the manufacturer's instructions. After measuring the purity, 1 μg RNA was used to synthesize cDNA by single-strand cDNA Synthesis Kit (Roche) followed by PCR amplification to study the fold difference in mRNA levels after normalization against reference gene RPLP0 as previously described (12).

Western blotting

Fifty micrograms of whole cell lysates were resolved on 10% SDS-PAGE gels before transfer of proteins onto nitrocellulose membrane, which was followed by blocking in 5% milk and incubation with primary antibodies against human E-cadherin, N-cadherin, fibronectin (BD biosciences), Slug, Snail, vimentin, Src, Src-pY416, AKT-pS473, AKT (Cell Signaling Technology), p63 (Abcam), Twist, ZEB1, ZEB2 (Santa Cruz Biotechnology) GAPDH, and β-actin (Sigma Aldrich). Using HRP-tagged species-specific secondary antibodies, the differential protein expressions were visualized by chemiluminescence detection.

Immunofluorescence

The cellular localization of proteins was studied by fixing monolayer of HFK, which was followed by permeabilization and exposure to primary and species-specific Alexa Fluor-488/-594-tagged secondary antibodies (Life Technologies). The proteins were visualized through 20× and 60× magnification by confocal microscopy. The cellular localization of proteins in HNSCC sections or on 3D-organotypic rafts were studied by heat-induced antigen retrieval methods (tris buffer and citrate buffer) followed by immunofluorescence detection. The intensities of staining (Q-score) were quantified as previous described (8).

The Cancer Genome Atlas dataset analysis

HNSCC samples processed for the The Cancer Genome Atlas (TCGA) resource (4) were utilized for in silico analyses/support of laboratory findings. Processed (level three) gene expression data for 277 patients with HNSCC, 277 tumor, and 44 matched normal tissue, was downloaded from Gene Expression Omnibus, GEO accession number GSE62944 (25) and supporting clinical data from University of California Santa Cruz Cancer Browser (https://genome-cancer.ucsc.edu/proj/site/hgHeatmap/). HPV status was previously defined by the TCGA (5) in these samples, as the presence of >1,000 mapped RNA sequencing reads aligning to HPV viral genes E6 and E7 (5) and was obtained through cbioportal (http://www.cbioportal.org/). Gene expression data were log2(x + 1) transformed before merging with clinical data. The resulting data matrix was used to plot expression of genes of interest between normal, HPV-positive, and HPV-negative. Patients were dichotomized into high-/low-expressing groups for survival analyses by receiver–operating characteristic analysis of the gene of interest against survival as previously described (26). Survival analyses were performed using the Kaplan–Meier estimate on a subcohort of 241 HPV-negative patients with survival data and the log-rank test used to calculate univariate associations between genes of interest and survival. Only 5-year survival was considered and defined as the time, in months, from sample collection until death by any cause, with right censoring applied to patients lost to follow-up or with a survival time of greater than 60 months. These analyses were performed using R v.3.3.1.

ChIP-seq and analysis of public ChIP-seq datasets

TP63 ChIP-seq data were generated from HFK-E6E7 expressing cells as previously described (12). Raw FASTQ data and those from our previously published ChIP-seq for p63 in primary HFKs (12) were reanalyzed as follows: adapter sequences were removed and FASTQC conducted with trimgalore and resulting reads aligned to hg19 with Bowtie 2 default settings (27). Reads filtered for blacklist regions with samtools were used as inputs for peak calling with MACS2 (28), comparing ChIP with input control and resultant SPMR normalized bedgraphs converted to bigwig format for visualization using UCSC bedGraphToBigWig script. Relevant bigwig files from encode (29) were downloaded and visualized alongside p63. Integrative analysis of narrowpeak calls was conducted using custom workflows in Cistrome (30).

Statistical analysis

The statistics for lab experiments were performed by comparing the mean values by Student t test and one-way ANOVA followed by Dunnet's post hoc analysis using IBM SPSS 20.0 software. The results were presented as mean ± SEM from five independent experiments and P < 0.05 was considered to be significant in all the experiments. For TCGA gene expression analysis, results were presented as mean ± SEM and statistical significance defined as ***P < 0.001, **P < 0.01, and *P < 0.05 when calculated by either Welch t test (A) or ANOVA (C, D) when compared with the normal/controls.

Slug/SNAI2 is the predominant EMT-promoting gene expressed in HNSCC

To investigate deregulation of EMT-promoting transcription factors in HNSCC, we first examined expression of EMT master regulatory transcription factors Twist/TWIST1, Snail/SNAI1, and Slug/SNAI2 in patient with HNSCC samples in TCGA [HPV-positive (n = 36) and HPV-negative (n = 241); ref. 4]. This revealed that expression of SNAI1, SNAI2, and TWIST1 were significantly increased in both HPV-positive and HPV-negative tumors compared with normal (Fig. 1A). Importantly, although significant increases of both SNAI1 and TWIST1 were observed, the absolute expression levels of SNAI1/Snail and TWIST1 were 5- to 10-fold lower than Slug/SNAI2 as measured by fragments per kilobase of exon per million fragments mapped (FPKM; Fig. 1A), with SNAI1 and TWIST exhibiting more modest fold change in expression compared with normal than SNAI2/Slug (SNAI2 2.02/3.70; SNAI1 1.40/1.59, TWIST1 1.51/1.89 in HPV-positive and HPV-negative tumors, respectively), suggesting that Slug/SNAI2 is the predominantly expressed EMT-regulating transcription factor activated in HNSCC. Further examination of expression levels and correlation of SNAI1, SNAI2, TWIST1, and other EMT regulators ZEB1 and ZEB2 and markers Vimentin (VIM), E-Cadherin (CDH1), N-Cadherin (CDH2), and fibronectin (FN1) revealed similarly modest expression of both ZEB1 and ZEB2, with only ZEB2 demonstrating significant upregulation (ZEB1 1.09/1.16 and ZEB2 1.36/1.33 in HPV-positive and HPV-negative tumors, respectively).

Figure 1.

A, Comparison of mRNA expression of SNAI1, SNAI2, and TWIST1 mRNA levels in normal tissue (n = 44) and HPV-positive (n = 36) and HPV-negative (n = 241) HNSCC in the TCGA RNA-seq cohort. B and C, Immunofluorescence detection and quantification (Q-score) of Slug protein in normal epithelium and tumor areas from HPV-positive and HPV-negative oropharyngeal HNSCC. N = 5 HPV(positive) and 6 HPV(negative) tumor sections. D, Fold difference in mRNA expression of TWIST1, SNAI1, SNAI2, VIM, and CDH1 genes; and (E) protein expression of E-cadherin, Slug, vimentin, and β-actin in control (pBabe-HFK) and E6/E7-HFK, respectively. F, Phase contrast images indicating normal cobblestone (green asterisk) and elongated spindle-like EMT (red arrows) morphology. G, Immunofluorescence visualization of E-cadherin (white arrows) and nuclei (DAPI). Scale bars, 20 and 100 μm. Data represented as mean ± SEM and statistical significance defined as: ***, P < 0.001; **, P < 0.01; and *, P < 0.05, when calculated by either Welch t test (A) or a one-way ANOVA (C and D) when compared with the normal/controls.

Figure 1.

A, Comparison of mRNA expression of SNAI1, SNAI2, and TWIST1 mRNA levels in normal tissue (n = 44) and HPV-positive (n = 36) and HPV-negative (n = 241) HNSCC in the TCGA RNA-seq cohort. B and C, Immunofluorescence detection and quantification (Q-score) of Slug protein in normal epithelium and tumor areas from HPV-positive and HPV-negative oropharyngeal HNSCC. N = 5 HPV(positive) and 6 HPV(negative) tumor sections. D, Fold difference in mRNA expression of TWIST1, SNAI1, SNAI2, VIM, and CDH1 genes; and (E) protein expression of E-cadherin, Slug, vimentin, and β-actin in control (pBabe-HFK) and E6/E7-HFK, respectively. F, Phase contrast images indicating normal cobblestone (green asterisk) and elongated spindle-like EMT (red arrows) morphology. G, Immunofluorescence visualization of E-cadherin (white arrows) and nuclei (DAPI). Scale bars, 20 and 100 μm. Data represented as mean ± SEM and statistical significance defined as: ***, P < 0.001; **, P < 0.01; and *, P < 0.05, when calculated by either Welch t test (A) or a one-way ANOVA (C and D) when compared with the normal/controls.

Close modal

In support of the prediction that Slug/SNAI2 is important for EMT, we observed significant increases in Slug/SNAI2 protein levels (as measured by indirect immunfluorescent staining) when comparing tumor versus adjacent normal in full face sections of both HPV-positive (n = 5) and HPV-negative (n = 6) oropharyngeal HNSCC (Fig. 1B and C; Supplementary Fig. S2A). Interestingly, in agreement with recent studies in normal skin (31), only a few basal cells in normal regions stained positive for Slug/SNAI2, whereas, in contrast, intense nuclear localization of Slug/SNAI2 was observed throughout both HPV-positive and HPV-negative tumors (Fig. 1B and C; Supplementary Fig. S2A). In addition, the faint staining of Snail/SNAI1 and TWIST1 proteins in epithelial cells from normal and tumor sections suggested low protein expression in these cells compared with the adjacent stromal cells (Supplementary Fig. S2A).

Further evaluation of SNAI1, SNAI2, TWIST1, ZEB1, and ZEB2 protein and mRNA levels in our invasive HPV-E6/E7 expression model system (8, 22, 32) also indicates that Slug/SNAI2 is the predominant EMT-activating transcription factor expressed in E6/E7-HFK and normal matched keratinocyte controls, and is significantly upregulated at both the mRNA and protein level in E6/E7-HFK (Fig. 1D and E; Supplementary Fig. S2B). Similar to TCGA patient analysis, although relative increases in both Twist/TWIST1 and Snail/SNAI1 mRNA expressions in E6/E7-HFK were also observed (Fig. 1D), their levels were at the limit of detection by both RT-PCR and Western blot analysis (Fig. 1D; Supplementary Fig. S2B). Importantly, elevated Slug/SNAI2 coincided with a switch to spindle-like morphology (Fig. 1F), significant decrease in expression of both E-cadherin/CDH1 mRNA and protein and concomitant increase in vimentin/VIM and fibronectin/FN1 (Fig. 1D and E; Supplementary Fig. S2B) as well a loss of junctional E-Cadherin staining (Fig. 1G), suggesting that E6/E7 expressing cells exhibit an EMT phenotpye.

Depletion of Slug attenuates EMT and mitigates invasion

To determine the functional contribution of Slug/SNAI2 in mediating EMT and invasion, we used two independent siRNA molecules (24) to knockdown Slug/SNAI2 expression in E6/E7-HFK (Fig. 2A–C; Supplementary S3A). Depletion of Slug/SNAI2 resulted in a reversal of EMT as measured by increase in mRNA and protein expression of E-cadherin, concomitant reduction in the expression of vimentin (Fig. 2A and C), reversion of spindle-like to cobble-stone morphology, and concurrent increase in E-cadherin/CDH1 localization at cell junctions (Fig. 2C; Supplementary Fig. S3A). Importantly, qualitative and quantitative analysis of 3D-organotypic rafts established using E6/E7-HFK with confirmed knockdown of Slug expression by the more effective siRNA (mol-2) showed a significant reduction in number of invasions (Fig. 2D and E) and that these effects were independent of cell proliferation as measured by BrdUrd uptake (Supplementary Fig. S3B and S3C), indicating that Slug/SNAI2 is necessary for both EMT and invasion in this model.

Figure 2.

Depletion of Slug expression restores normal phenotype and attenuates cell invasion. A, Fold difference in mRNA expression of SNAI2, VIM, and CDH1 genes in Scram siRNA (control) and Slug siRNA mol-1– and mol-2–transfected cells. B, Phase contrast images indicating EMT (red arrows) and normal (green asterisk) morphology and immunofluorescence visualization of E-cadherin (white arrows), Slug and nuclei (DAPI) in control (scram) and Slug (mol-2) depleted E6/E7-HFK, respectively. C, Protein expressions of Slug, vimentin, E-cadherin, and β-actin in the same cells as (A). H&E staining showing invasions (arrows) on the 3D-organotypic rafts established with control and Slug (mol-2) depleted E6/E7-HFK (D); and relative number of invasive incidents across these rafts per centimeter (E). Scale bars, 20 and 100 μm. N = 5 independent experiments. Data represented as mean ± SEM and statistical significance calculated by Student t test with: *, P < 0.05 compared with the control; **, P < 0.01 compared with the control.

Figure 2.

Depletion of Slug expression restores normal phenotype and attenuates cell invasion. A, Fold difference in mRNA expression of SNAI2, VIM, and CDH1 genes in Scram siRNA (control) and Slug siRNA mol-1– and mol-2–transfected cells. B, Phase contrast images indicating EMT (red arrows) and normal (green asterisk) morphology and immunofluorescence visualization of E-cadherin (white arrows), Slug and nuclei (DAPI) in control (scram) and Slug (mol-2) depleted E6/E7-HFK, respectively. C, Protein expressions of Slug, vimentin, E-cadherin, and β-actin in the same cells as (A). H&E staining showing invasions (arrows) on the 3D-organotypic rafts established with control and Slug (mol-2) depleted E6/E7-HFK (D); and relative number of invasive incidents across these rafts per centimeter (E). Scale bars, 20 and 100 μm. N = 5 independent experiments. Data represented as mean ± SEM and statistical significance calculated by Student t test with: *, P < 0.05 compared with the control; **, P < 0.01 compared with the control.

Close modal

SRC signaling axis modulates the expression of EMT markers

Western blot analysis revealed that Slug/SNAI2 depletion (Supplementary Fig. S4A) did not affect SRC levels and activatory phosphorylation (Y416), which we have previously shown to be necessary and sufficient for invasion. We next evaluated whether Slug/SNAI2 was activated downstream of SRC and required for EMT and invasion. In support of this hypothesis both Slug/SNAI2 mRNA and protein expression were significantly decreased in 2D-culture upon transient siRNA-mediated depletion of SRC expression with two independent siRNAs (Fig. 3A and B). Moreover, a similar reduction of Slug/SNAI2 protein was observed upon inhibition of Src activity using the kinase inhibitor dasatanib, both in 2D-culture (Fig. 3C) and in treated organotypic raft cultures (Fig. 3D), where we previously demonstrated the ability of dasatanib or SRC depletion to block invasion (8). In support of a mechanistic role for EMT, both SRC depletion or inhibition were accompanied by a cognate decreased vimentin and increased E-cadherin expression (Fig. 3A–C) and restoration of gross cellular morphology suggestive of a reversion of EMT (Supplementary Fig. S4B). Moreover, exogenous adenoviral expression of constitutively active Src (Src-531), which is sufficient to promote cell invasion (8), also upregulated the protein expression of Slug and vimentin, while suppressing E-cadherin and inducing EMT morphology in noninvasive cells (Supplementary Fig. S4C and S4D).

Figure 3.

Depletion of Src expression alleviates EMT. Fold difference in mRNA levels (A) and/or protein expression of Src, Slug, vimentin, E-cadherin, and β-actin in Src (siRNA Src 1 and 2) depleted cells (B). C, Protein expression of Src-pY416, total Src, Slug, vimentin, E-cadherin, and β-actin in E6/E7-HFK treated in the absence and presence of specific Src inhibitor (dasatinib; 10, 50 nmol/L). D, Immunofluorescence visualization of Slug (red) and nuclei (DAPI) on rafts established with E6/E7-HFK and treated in the absence and presence of dasatinib (50 nmol/L). Arrows indicate the invasive incidents in vehicle (control) rafts. High SRC expression correlates with poor survival. E, Comparison of mRNA expression of SRC from normal tissue (n = 44), HPV-positive (n = 36) and HPV-negative (n = 241) HNSCC in the TCGA RNA-seq cohort. F, Kaplan–Meier plots of 5-year OS ROC outcome-based stratification of low and high mRNA levels of SRC in the aforementioned dataset. Scale bars, 100 μm. N = 3 independent experiments. Data represented as mean ± SEM and statistical significance defined as: ***, P < 0.001; **P < 0.01; and *, P < 0.05, when calculated by either a one-way ANOVA (A) or Welch t test (E) when compared with the control/normal.

Figure 3.

Depletion of Src expression alleviates EMT. Fold difference in mRNA levels (A) and/or protein expression of Src, Slug, vimentin, E-cadherin, and β-actin in Src (siRNA Src 1 and 2) depleted cells (B). C, Protein expression of Src-pY416, total Src, Slug, vimentin, E-cadherin, and β-actin in E6/E7-HFK treated in the absence and presence of specific Src inhibitor (dasatinib; 10, 50 nmol/L). D, Immunofluorescence visualization of Slug (red) and nuclei (DAPI) on rafts established with E6/E7-HFK and treated in the absence and presence of dasatinib (50 nmol/L). Arrows indicate the invasive incidents in vehicle (control) rafts. High SRC expression correlates with poor survival. E, Comparison of mRNA expression of SRC from normal tissue (n = 44), HPV-positive (n = 36) and HPV-negative (n = 241) HNSCC in the TCGA RNA-seq cohort. F, Kaplan–Meier plots of 5-year OS ROC outcome-based stratification of low and high mRNA levels of SRC in the aforementioned dataset. Scale bars, 100 μm. N = 3 independent experiments. Data represented as mean ± SEM and statistical significance defined as: ***, P < 0.001; **P < 0.01; and *, P < 0.05, when calculated by either a one-way ANOVA (A) or Welch t test (E) when compared with the control/normal.

Close modal

Analysis of TCGA patient RNA-seq data revealed that SRC mRNA expression is significantly elevated in both HPV-positive and HPV-negative HNSCC tumors compared with normal (Fig. 3E). Importantly, dichotomizing patients into high and low expressing groups (based on ROC outcome based dichomization of 5-year overall survival; ref. 26) identified a significant association between high SRC expression (high = 48, low = 192) and worse outcome in HPV-negative patients (Fig. 3F).

Ascertaining impact of intratumoral expression of other EMT transcription factors (SNAI1/Snail, TWIST1, ZEB1, and ZEB2) and markers (CDH1, CDH2, VIM, FN1) on survival is challenging owing to the high-level expression of EMT factors in stromal cells (Fig. 1B; Supplementary Fig. S2A). Further analysis revealed high levels of correlation between all of these stromally derived markers (data not shown) and no coherent pattern of impact on outcome (data not shown).

TP63 modulates the expression of SNAI2

Interestingly, a similar trend was observed for total TP63 expression (Fig. 4A), where high TP63 levels were associated with worse outcome. Because we have previously shown that TP63 is necessary to effect invasion in this model by regulating SRC–AKT–AP1 signaling (8), we next compared TP63-ChIP-seq in E6/E7-HFK with our previous TP63 ChIP-seq in primary HFKs (Supplementary Fig. S5A; ref. 12). Similar TP63 binding patterns were observed globally and specifically at the previously described SRC- (Fig. 4B) and MMP14 (not shown)-associated enhancer regions (8). Furthermore, we also identified direct TP63 binding to upstream enhancer and promoter of SNAI2 and upstream and within CDH1 loci (E-cadherin; Fig. 4B; Supplementary Fig. S5B). Importantly, siRNA-mediated depletion of all TP63 isoforms resulted in significant decreased Slug/SNAI2 mRNA and protein levels (Fig. 4C and D). Significantly, depletion of Slug did not affect TP63 mRNA or mRNA splice-form or protein isoform expression (Supplementary Fig. S6A and S6B), indicating that TP63 is required for SRC-dependent transcription of Slug/SNAI2 and EMT, which is necessary for invasion.

Figure 4.

TP63 modulates SNAI2 expression. A, Kaplan–Meier plots of 5-year OS ROC outcome-based stratification of low and high mRNA levels of TP63 in the TCGA HNSCC HPV-negative cohort of 241 patients. B, Visualization of E6/E7 and normal HFK TP63 ChIP-seq tracks around SRC and SNAI2 loci annotated with encode histone modification data from normal human epidermal keratinocytes (NHEK; ref. 29). C and D, Relative mRNA and protein expressions of TP63 and SNAI2 after transient knockdown of total TP63 in E6/E7-HFK. Cells transfected with scram siRNA molecule were treated as the control. N = 3 independent experiments. Data represented as mean ± SEM and statistical significance calculated by Student t test with: *, P < 0.05 and **, P < 0.01 compared with the controls.

Figure 4.

TP63 modulates SNAI2 expression. A, Kaplan–Meier plots of 5-year OS ROC outcome-based stratification of low and high mRNA levels of TP63 in the TCGA HNSCC HPV-negative cohort of 241 patients. B, Visualization of E6/E7 and normal HFK TP63 ChIP-seq tracks around SRC and SNAI2 loci annotated with encode histone modification data from normal human epidermal keratinocytes (NHEK; ref. 29). C and D, Relative mRNA and protein expressions of TP63 and SNAI2 after transient knockdown of total TP63 in E6/E7-HFK. Cells transfected with scram siRNA molecule were treated as the control. N = 3 independent experiments. Data represented as mean ± SEM and statistical significance calculated by Student t test with: *, P < 0.05 and **, P < 0.01 compared with the controls.

Close modal

TP63γ isoform is necessary for the Src–Slug signaling, EMT, and invasion

Because we and others have consistently observed differential and often opposing effects of expressing different TP63 isoforms on target gene expression and resulting phenotypes (11, 33–35), we speculated that TP63 mediated activation of SRC and Slug/SNAI2 may be differentially affected by TP63 isoforms similar to our previous observations with regard to the role of ΔNp63γ isoform in affecting EMT in breast epithelial cells (13). To investigate this possibility, we first compared the effects of specifically depleting the α and β isoforms through targeting their shared 3′UTR (UTR; with a validated shRNA molecule that does not affect γ-isoforms; ref. 9) with an shRNA targeting the DNA-binding domain (DBD) and therefore depleting all TP63 isoforms (DBD), which we had previously reported to attenuate invasion in this model (8). This analysis revealed that depletion of the α and β isoforms had little or no effect on number of invasions (Supplementary Fig. S7A–S7C), suggesting that residual γ-isoforms are sufficient to maintain invasive capacity. These results were further supported by similar analysis with independent siRNAs, where total p63 depletion (Pan molecule) was sufficient to significantly reduce the number of invasions (Fig. 5A and B). In support of these results, mRNA or protein expression of Src, Slug/SNAI2, vimentin, and Src activity were unaffected in UTR-depleted cells grown in 2D on collagen I (Fig. 5C–E), whereas, in contrast, total TP63 knockdown depleted mRNA and protein expression of both Src and Slug (Fig. 5D and E) with a concomitant increase in the protein levels of E-cadherin while suppressing vimentin (Fig. 5D and E). Cumulatively, this suggests that the residual endogenous TP63γ isoform sufficient to maintain the expression/activity of Src–Slug signaling axis, EMT, and invasive phenotype.

Figure 5.

Depletion of different isoforms of TP63 attenuates EMT. A, H&E staining showing invasive incidents (arrows) on 3D-organotypic rafts established with E6/E7-HFK with transient TP63 knockdown by UTR (TP63α, β) and Pan (total TP63) siRNA molecules; and quantification of the number of invasive incidents across the rafts per centimeter established with these cells (B). C and D, Relative mRNA expressions of TP63α, β, γ isoforms, SNAI2, and SRC; and protein expressions of total p63, total Src, Src-pY416, Slug, E-cadherin, vimentin, and β-actin after transient knockdown with control (Scram), UTR p63, and Pan p63 molecules in E6/E7-HFK (E). Scale bars, 100 μm. N = 3 independent experiments. Data represented as mean ± SEM and statistical significance calculated by one-way ANOVA with: *, P < 0.05 and **, P < 0.01 compared with the Scram controls.

Figure 5.

Depletion of different isoforms of TP63 attenuates EMT. A, H&E staining showing invasive incidents (arrows) on 3D-organotypic rafts established with E6/E7-HFK with transient TP63 knockdown by UTR (TP63α, β) and Pan (total TP63) siRNA molecules; and quantification of the number of invasive incidents across the rafts per centimeter established with these cells (B). C and D, Relative mRNA expressions of TP63α, β, γ isoforms, SNAI2, and SRC; and protein expressions of total p63, total Src, Src-pY416, Slug, E-cadherin, vimentin, and β-actin after transient knockdown with control (Scram), UTR p63, and Pan p63 molecules in E6/E7-HFK (E). Scale bars, 100 μm. N = 3 independent experiments. Data represented as mean ± SEM and statistical significance calculated by one-way ANOVA with: *, P < 0.05 and **, P < 0.01 compared with the Scram controls.

Close modal

ΔNp63γ is sufficient to induce Src–Slug axis, EMT, and invasion

We have previously shown that ΔN N-terminal variants are the dominant forms in both normal (9) and E6/E7-expressing HFKs (36) with endogenous TA isoforms below the limit of Western blotting and not normally detectable by RT-PCR (data not shown). Therefore, we next tested if expression of ΔNp63γ or indeed ΔNp63α or ΔNp63β alone is sufficient to rescue EMT and invasion in E6/E7-HFK rendered noninvasive through stable shRNA depletion of total TP63. To achieve this, we compared the effects of adenoviral-mediated reintroduction of individual TP63 isoforms rendered resistant to the shRNA into E6/E7-HFK cells (13). Growth of these reconstituted cells in organotypic cultures revealed that only reexpression of the ΔNp63γ isoform was capable of significantly inducing invasions (Fig. 6A and B), whereas ΔNp63α had a modest repressive effect on residual invasion and ΔNp63β no significant impact. Importantly, ectopic expression of ΔNp63γ isoform in keratinocytes was sufficient to significantly upregulate mRNA levels of SRC and SNAI2, while suppressing CDH1 (E-cadherin) compared with its controls (GFP; Fig. 6C), coincident with upregulation of protein expression of Src, activated Src, Slug, activated AKT and vimentin, and suppression of E-cadherin (Fig. 6D; Supplementary Fig. S7D). Together these results identify a potentially important novel oncogenic role of ΔNp63γ isoform in E6/E7-HFK, which is sufficient to induced Src-activated EMT and invasion in squamous cells.

Figure 6.

TP63γ upregulation is sufficient for EMT. A, H&E staining showing invasion (arrows) on 3D-organotypic rafts established after adenovirus-mediated introduction of GFP (control), ΔNp63α, β, and γ isoforms in the E6/E7-HFK with stable total TP63 (DBD) depletion; and quantification of number of invasive incidents across the rafts per centimeter established with these cells (B). C, Relative mRNA levels of TP63γ, SRC, SNAI2, and CDH1 gene; and (D) protein expression of HA (measure of transfected ΔNp63γ), total p63, total Src, Src-pY416, Slug, E-cadherin, vimentin, AKT-pS473, total AKT, and β-actin in keratinocytes after adenovirus-mediated expression of GFP and ΔNp63γ isoform. Scale bar, 100 μm. N = 3 independent experiments. Data represented as mean ± SEM and statistical significance calculated by one-way ANOVA and Student t test with: *, P < 0.05; **, P < 0.01; and ***, P < 0.001 compared with GFP.

Figure 6.

TP63γ upregulation is sufficient for EMT. A, H&E staining showing invasion (arrows) on 3D-organotypic rafts established after adenovirus-mediated introduction of GFP (control), ΔNp63α, β, and γ isoforms in the E6/E7-HFK with stable total TP63 (DBD) depletion; and quantification of number of invasive incidents across the rafts per centimeter established with these cells (B). C, Relative mRNA levels of TP63γ, SRC, SNAI2, and CDH1 gene; and (D) protein expression of HA (measure of transfected ΔNp63γ), total p63, total Src, Src-pY416, Slug, E-cadherin, vimentin, AKT-pS473, total AKT, and β-actin in keratinocytes after adenovirus-mediated expression of GFP and ΔNp63γ isoform. Scale bar, 100 μm. N = 3 independent experiments. Data represented as mean ± SEM and statistical significance calculated by one-way ANOVA and Student t test with: *, P < 0.05; **, P < 0.01; and ***, P < 0.001 compared with GFP.

Close modal

Depletion of cellular epithelial markers with concurrent appearance of those of mesenchymal cells is an important switch required to enable morphological and mechanical changes necessary to underpin the invasive capacity of cancer cells. Here, we identify that expression of ΔNp63γ isoform as both necessary and sufficient to induce such a switch in E6/E7 transformed primary human keratinocytes to promote an EMT-like phenotype and invasion. This is mediated through direct transcriptional modulation and activation of Slug/SNAI2 and Src, the increased expression of which is observed along with TP63 in both patients with HPV-positive and -negative HNSCC. In particular, increased SRC levels correlates with poor prognosis and represents a potential clinically actionable event through inhibition of Src or downstream AKT activity.

Our previous studies demonstrated that E6/E7-HFK exhibited increased cell migration and invasion compared with normal HFK (8). In this study, we focused on determining the importance of master transcriptional regulators of EMT namely SNAI1/Snail, SNAI2/Slug, TWIST1/Twist, given their importance for EMT required for cancer cell migration, invasion, and ultimately metastasis. Our analyses revealed that Slug is the dominant EMT regulator increased in both our invasive E6/E7-HFK model and in patient samples, wherein both Snail and Twist are expressed at much lower levels, a finding that is strongly supported by a recent landmark single-cell study of patient with HNSCC samples (37), which indicates that the majority of signal for EMT markers with the exception of SNAI2/Slug are stromally derived in HNSCC and is associated with a partial EMT. Importantly, depletion of Slug is sufficient to reverse the increases in vimentin, decrease E-Cadherin, and change cells from a fibroblast-like to an epithelial-like morphology and prevent invasion in 3D-organotypic cultures, suggesting Slug is necessary for the invasive phenotype. There is significant evidence in the literature to support the broader applicability of our findings with regard to the importance of SNAI2/Slug for HNSCC cell survival, migration, invasion, stemness, and radioresistance (38–41). Consistent with our hypothesis, the overexpression of Slug/SNAI2 and its increased transcriptional activity in normal human keratinocytes has been shown to induce (i) transcriptional silencing of differentiation genes, (ii) dedifferentiation, (iii) EMT, (iv) wound healing, and (v) enhanced cell migration (31, 42, 43).

Moreover, our IHC studies in HPV-positive and -negative oropharyngeal HNSCC are supported by findings by Wang and colleagues, which indicated that elevated Slug/SNAI2 expression in HNSCC tumor cells in a cohort of 129 HNSCC correlated with poor prognosis (38). Because the expression or activities of components from TP63/Src signaling axis remained unaffected by Slug/SNAI2 depletion, it suggests that it is a downstream target of this signaling cascade. This is consistent with a recently published study, which also identified Slug/SNAI2 as a direct p63 target in lung SCC and breast cancer cell lines (44).

Transcriptional and protein levels of TP63 isoforms are elevated in E6/E7-HFK and depletion of total TP63 isoforms has been previously shown to attenuate the cell invasion (8). However, in this study, stable knockdown of the TP63 α and β isoforms (UTR) in the invasive population had no significant impact in reducing the number of invasions. However, because total TP63 ablation depleted the expression of Src and reversed expression of EMT genes, it indicates that either TAp63γ or ΔNp63γ isoforms are responsible for promoting EMT and invasion. Our previous studies on normal breast cells have shown the role ΔNp63γ in promoting EMT phenotype and development of epithelial cancers (13). Because the ΔNp63 isoforms are more stable compared with the TAp63 isoforms due to the absence of structural features like the FWL motif which accelerates the TAp63 protein degradation (45, 46), and because TAp63 levels are almost undetectable in HFK, we expressed ΔNp63 isoforms (α, β, and γ) in a background where all TP63 isoforms were depleted by stable knockdown. Ectopic expression of the ΔNp63γ isoform only induced invasion concomitant with activation of both SRC/AKT signaling and Slug-mediated EMT.

In conclusion, the ΔNp63γ isoform appears to be essential for mediating invasion by modulating the expression of Slug and by activating Src signaling. Specific inhibition of Src activity has a therapeutic potential alone or as recently suggested in combination with epidermal growth factor receptor and other receptor tyrosine kinase inhibitors to protect against progression and relapse of HNSCC, which are problems in treating these cancers (47–49).

No potential conflicts of interest were disclosed.

Conception and design: K. Srivastava, S.S. McDade, D.J. McCance

Development of methodology: K. Srivastava, S.S. McDade, D.J. McCance

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): K. Srivastava, A. Pickard, G.P. Quinn, J.A. James, S.S. McDade

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): K. Srivastava, A. Pickard, S.G. Craig, G.P. Quinn, S.M. Lambe, S.S. McDade, D.J. McCance

Writing, review, and/or revision of the manuscript: K. Srivastava, A. Pickard, S.G. Craig, J.A. James, S.S. McDade, D.J. McCance

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): K. Srivastava, J.A. James, D.J. McCance

Study supervision: D.J. McCance

The authors thank Medical Research Council, UK grant G1001692, for funding this study. The patient samples used in this research were received from the Northern Ireland Biobank (NIB13-001), which has received funds from HSC Research and Development Division of the Public Health Agency in Northern Ireland and the Friends of the Cancer Centre. ChIP-seq data generation in this study was supported by the FMHLS, Genomics Core Technology Unit, Queen's University Belfast.

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.

1.
Ferlay
J
,
Shin
H-R
,
Bray
F
,
Forman
D
,
Mathers
C
,
Parkin
DM
. 
Estimates of worldwide burden of cancer in 2008: GLOBOCAN 2008
.
Int J Cancer
2010
;
127
:
2893
917
.
2.
Goon
PKC
,
Stanley
MA
,
Ebmeyer
J
,
Steinstraesser
L
,
Upile
T
,
Jerjes
W
, et al
HPV & head and neck cancer: a descriptive update
.
Head Neck Oncol
2009
;
1
:
36
.
3.
Hoadley
KA
,
Yau
C
,
Wolf
DM
,
Cherniack
AD
,
Tamborero
D
,
Ng
S
, et al
Multiplatform analysis of 12 cancer types reveals molecular classification within and across tissues of origin
.
Cell
2014
;
158
:
929
44
.
4.
Lawrence
MS
,
Sougnez
C
,
Lichtenstein
L
,
Cibulskisl
K
,
Lander
E
,
Gabriel
SB
, et al
Comprehensive genomic characterization of head and neck squamous cell carcinomas
.
Nature
2015
;
517
:
576
82
.
5.
Trink
B
,
Okami
K
,
Wu
L
,
Sriuranpong
V
,
Jen
J
,
Sidransky
D
. 
A new human p53 homologue
.
Nat Med
1998
;
4
:
747
8
.
6.
McDade
SS
,
McCance
DJ
. 
The role of p63 in epidermal morphogenesis and neoplasia
.
Biochem Soc Trans
2010
;
38
:
223
8
.
7.
Rocco
JW
,
Leong
CO
,
Kuperwasser
N
,
DeYoung
MP
,
Ellisen
LW
. 
p63 mediates survival in squamous cell carcinoma by suppression of p73-dependent apoptosis
.
Cancer Cell
2006
;
9
:
45
56
.
8.
Srivastava
K
,
Pickard
A
,
McDade
S
,
McCance
DJ
. 
p63 drives invasion in keratinocytes expressing HPV16 E6/E7 genes through regulation of Src-FAK signalling
.
Oncotarget
2015
;
8
:
16202
19
.
9.
McDade
SS
,
Patel
D
,
McCance
DJ
. 
p63 maintains keratinocyte proliferative capacity through regulation of Skp2-p130 levels
.
J Cell Sci
2011
;
124
:
1635
43
.
10.
Ramsey
MR
,
Wilson
C
,
Ory
B
,
Rothenberg
SM
,
Faquin
W
,
Mills
AA
, et al
FGFR2 signaling underlies p63 oncogenic function in squamous cell carcinoma
.
J Clin Invest
2013
;
123
:
3525
38
.
11.
McDade
SS
,
Henry
AE
,
Pivato
GP
,
Kozarewa
I
,
Mitsopoulos
C
,
Fenwick
K
, et al
Genome-wide analysis of p63 binding sites identifies AP-2 factors as co-regulators of epidermal differentiation
.
Nucleic Acids Res
2012
;
40
:
7190
206
.
12.
McDade
SS
,
Patel
D
,
Moran
M
,
Campbell
J
,
Fenwick
K
,
Kozarewa
I
, et al
Genome-wide characterization reveals complex interplay between TP53 and TP63 in response to genotoxic stress
.
Nucleic Acids Res
2014
;
42
:
6270
85
.
13.
Lindsay
J
,
McDade
SS
,
Pickard
A
,
McCloskey
KD
,
McCance
DJ
. 
Role of delta Np63 gamma in epithelial to mesenchymal transition
.
J Biol Chem
2011
;
286
:
3915
24
.
14.
Zhang
Y
,
Yan
W
,
Chen
X
. 
P63 regulates tubular formation via epithelial-to-mesenchymal transition
.
Oncogene
2014
;
33
:
1548
57
.
15.
Onder
TT
,
Gupta
PB
,
Mani
SA
,
Yang
J
,
Lander
ES
,
Weinberg
RA
. 
Loss of E-cadherin promotes metastasis via multiple downstream transcriptional pathways
.
Cancer Res
2008
;
68
:
3645
54
.
16.
Hajra
KM
,
Chen
DYS
,
Fearon
ER
. 
The SLUG zinc-finger protein represses E-cadherin in breast cancer
.
Cancer Res
2002
;
62
:
1613
8
.
17.
Ye
Y
,
Xiao
Y
,
Wang
W
,
Yearsley
K
,
Gao
JX
,
Shetuni
B
, et al
ER alpha signaling through slug regulates E-cadherin and EMT
.
Oncogene
2010
;
29
:
1451
62
.
18.
Rodrigues
IS
,
Lavorato-Rocha
AM
,
Maia
BdM
,
Stiepcich
MMA
,
de Carvalho
FM
,
Baiocchi
G
, et al
Epithelial-mesenchymal transition-like events in vulvar cancer and its relation with HPV
.
Br J Cancer
2013
;
109
:
184
94
.
19.
Satelli
A
,
Li
S
. 
Vimentin in cancer and its potential as a molecular target for cancer therapy
.
Cell Mol Life Sci
2011
;
68
:
3033
46
.
20.
Fincham
VJ
,
Frame
MC
. 
The catalytic activity of Src is dispensable for translocation to focal adhesions but controls the turnover of these structures during cell motility
.
EMBO J
1998
;
17
:
81
92
.
21.
Bianchi-Smiraglia
A
,
Paesante
S
,
Bakin
AV
. 
Integrin beta 5 contributes to the tumorigenic potential of breast cancer cells through the Src-FAK and MEK-ERK signaling pathways
.
Oncogene
2013
;
32
:
3049
58
.
22.
Pickard
A
,
Cichon
A-C
,
Barry
A
,
Kieran
D
,
Patel
D
,
Hamilton
P
, et al
Inactivation of Rb in stromal fibroblasts promotes epithelial cell invasion
.
EMBO J
2012
;
31
:
3092
103
.
23.
Lowenstein
PR
,
Enquist
LW
. 
Protocols for gene transfer in neuroscience: towards gene therapy of neurological disorders
.
Chichester, NY
:
Wiley and Sons
; 
1996
.
24.
Mittal
MK
,
Singh
K
,
Misra
S
,
Chaudhuri
G
. 
SLUG-induced Elevation of D1 cyclin in breast cancer cells through the inhibition of its ubiquitination
.
J Biol Chem
2011
;
286
:
469
79
.
25.
Rahman
M
,
Jackson
LK
,
Johnson
WE
,
Li
DY
,
Bild
AH
,
Piccolo
SR
. 
Alternative preprocessing of RNA-Sequencing data in The Cancer Genome Atlas leads to improved analysis results
.
Bioinformatics
2015
;
31
:
3666
72
.
26.
Heagerty
PJ
,
Lumley
T
,
Pepe
MS
. 
Time-dependent ROC curves for censored survival data and a diagnostic marker
.
Biometrics
2000
;
56
:
337
44
.
27.
Langmead
B
,
Salzberg
SL
. 
Fast gapped-read alignment with Bowtie 2
.
Nat Methods
2012
;
9
:
357
U54
.
28.
Zhang
Y
,
Liu
T
,
Meyer
CA
,
Eeckhoute
J
,
Johnson
DS
,
Bernstein
BE
, et al
Model-based analysis of ChIP-Seq (MACS)
.
Genome Biol
2008
;
9
:
R137
.
29.
Myers
RM
,
Stamatoyannopoulos
J
,
Snyder
M
,
Dunham
I
,
Hardison
RC
,
Bernstein
BE
, et al
A user's guide to the encyclopedia of DNA elements (ENCODE)
.
PLoS Biol
2011
;
9
.
30.
Liu
T
,
Ortiz
JA
,
Taing
L
,
Meyer
CA
,
Lee
B
,
Zhang
Y
, et al
Cistrome: an integrative platform for transcriptional regulation studies
.
Genome Biol
2011
;
12
:
R83
.
31.
Mistry
DS
,
Chen
Y
,
Wang
Y
,
Zhang
K
,
Sen
GL
. 
SNAI2 controls the undifferentiated state of human epidermal progenitor cells
.
Stem Cells
2014
;
32
:
3209
18
.
32.
Cichon
A-C
,
Pickard
A
,
McDade
SS
,
Sharpe
DJ
,
Moran
M
,
James
JA
, et al
AKT in stromal fibroblasts controls invasion of epithelial cells
.
Oncotarget
2013
;
4
:
1103
16
.
33.
Gu
X
,
Coates
PJ
,
Boldrup
L
,
Nylander
K
. 
p63 contributes to cell invasion and migration in squamous cell carcinoma of the head and neck
.
Cancer Lett
2008
;
263
:
26
34
.
34.
Candi
E
,
Rufini
A
,
Terrinoni
A
,
Dinsdale
D
,
Ranalli
M
,
Paradisi
A
, et al
Differential roles of p63 isoforms in epidermal development: selective genetic complementation in p63 null mice
.
Cell Death Differ
2006
;
13
:
1037
47
.
35.
Boldrup
L
,
Coates
PJ
,
Gu
XL
,
Nylander
K
. 
Delta Np63 isoforms differentially regulate gene expression in squamous cell carcinoma: identification of Cox-2 as a novel p63 target
.
J Pathol
2009
;
218
:
428
36
.
36.
McKenna
DJ
,
McDade
SS
,
Patel
D
,
McCance
DJ
. 
MicroRNA 203 expression in keratinocytes is dependent on regulation of p53 levels by E6
.
J Virol
2010
;
84
:
10644
52
.
37.
Puram
SV
,
Tirosh
I
,
Parikh
AS
,
Patel
AP
,
Yizhak
K
,
Gillespie
S
, et al
Single-cell transcriptomic analysis of primary and metastatic tumor ecosystems in head and neck cancer
.
Cell
2017
;
171
:
1611
24
.
e24
.
38.
Wang
C
,
Liu
XQ
,
Huang
HZ
,
Ma
HB
,
Cai
WX
,
Hou
JS
, et al
Deregulation of Snai2 is associated with metastasis and poor prognosis in tongue squamous cell carcinoma
.
Int J Cancer
2012
;
130
:
2249
58
.
39.
Zhao
TT
,
He
QT
,
Liu
ZH
,
Ding
XQ
,
Zhou
XF
,
Wang
AX
. 
Angiotensin II type 2 receptor-interacting protein 3a suppresses proliferation, migration and invasion in tongue squamous cell carcinoma via the extracellular signal-regulated kinase-Snai2 pathway
.
Oncol Lett
2016
;
11
:
340
4
.
40.
Jiang
FF
,
Zhou
LJ
,
Wei
CB
,
Zhao
W
,
Yu
DS
. 
Slug inhibition increases radiosensitivity of oral squamous cell carcinoma cells by upregulating PUMA
.
Int J Oncol
2016
;
49
:
709
19
.
41.
Chung
MK
,
Jung
YH
,
Lee
JK
,
Cho
SY
,
Murillo-Sauca
O
,
Uppaluri
R
, et al
CD271 confers an invasive and metastatic phenotype of head and neck squamous cell carcinoma through the upregulation of slug
.
Clin Cancer Res
2018
;
24
:
674
83
.
42.
Savagner
P
,
Kusewitt
DF
,
Carver
EA
,
Magnino
F
,
Choi
C
,
Gridley
T
, et al
Developmental transcription factor slug is required for effective re-epithelialization by adult keratinocytes
.
J Cell Physiol
2005
;
202
:
858
66
.
43.
Hudson
LG
,
Newkirk
KM
,
Chandler
HL
,
Choi
C
,
Fossey
SL
,
Parent
AE
, et al
Cutaneous wound reepithelialization is compromised in mice lacking functional Slug (Snai2)
.
J Dermatol Sci
2009
;
56
:
19
26
.
44.
Dang
TT
,
Westcott
JM
,
Maine
EA
,
Kanchwala
M
,
Xing
C
,
Pearson
GW
. 
Delta Np63 alpha induces the expression of FAT2 and Slug to promote tumor invasion
.
Oncotarget
2016
;
7
:
28592
611
.
45.
Serber
Z
,
Lai
HC
,
Yang
A
,
Ou
HD
,
Sigal
MS
,
Kelly
AE
, et al
A C-terminal inhibitory domain controls the activity of p63 by an intramolecular mechanism
.
Mol Cell Biol
2002
;
22
:
8601
11
.
46.
Okada
Y
,
Osada
M
,
Kurata
S
,
Sato
S
,
Aisaki
K
,
Kageyama
Y
, et al
p53 gene family p51 (p63)-encoded, secondary transactivator p51 B(TAp63 alpha) occurs without forming an immunoprecipitable complex with MDM2, but responds to genotoxic stress by accumulation
.
Exp Cell Res
2002
;
276
:
194
200
.
47.
Stabile
LP
,
He
GQ
,
Lui
VWY
,
Henry
C
,
Gubish
CT
,
Joyce
S
, et al
c-Src activation mediates erlotinib resistance in head and neck cancer by stimulating c-Met
.
Clin Cancer Res
2013
;
19
:
380
92
.
48.
Egloff
AM
,
Grandis
JR
. 
Targeting epidermal growth factor receptor and Src pathways in head and neck cancer
.
Semin Oncol
2008
;
35
:
286
97
.
49.
Koppikar
P
,
Choi
SH
,
Egloff
AM
,
Cai
Q
,
Suzuki
S
,
Freilino
M
, et al
Combined inhibition of c-Src and epidermal growth factor receptor abrogates growth and invasion of head and neck squamous cell carcinoma
.
Clin Cancer Res
2008
;
14
:
4284
91
.

Supplementary data