Identifying pathways regulating the oncogenic p53 family member ΔNp63 provides therapeutic avenues for squamous cell carcinoma

Background ΔNp63 overexpression is a common event in squamous cell carcinoma (SCC) that contributes to tumorigenesis, making ΔNp63 a potential target for therapy. Methods We created inducible TP63-shRNA cells to study the effects of p63-depletion in SCC cell lines and non-malignant HaCaT keratinocytes. DNA damaging agents, growth factors, signaling pathway inhibitors, histone deacetylase inhibitors, and metabolism-modifying drugs were also investigated for their ability to influence ΔNp63 protein and mRNA levels. Results HaCaT keratinocytes, FaDu and SCC-25 cells express high levels of ΔNp63. HaCaT and FaDu inducible TP63-shRNA cells showed reduced proliferation after p63 depletion, with greater effects on FaDu than HaCaT cells, compatible with oncogene addiction in SCC. Genotoxic insults and histone deacetylase inhibitors variably reduced ΔNp63 levels in keratinocytes and SCC cells. Growth factors that regulate proliferation/survival of squamous cells (IGF-1, EGF, amphiregulin, KGF, and HGF) and PI3K, mTOR, MAPK/ERK or EGFR inhibitors showed lesser and inconsistent effects, with dual inhibition of PI3K and mTOR or EGFR inhibition selectively reducing ΔNp63 levels in HaCaT cells. In contrast, the antihyperlipidemic drug lovastatin selectively increased ΔNp63 in HaCaT cells. Conclusions These data confirm that ΔNp63-positive SCC cells require p63 for continued growth and provide proof of concept that p63 reduction is a therapeutic option for these tumors. Investigations of ΔNp63 regulation identified agent-specific and cell-specific pathways. In particular, dual inhibition of the PI3K and mTOR pathways reduced ΔNp63 more effectively than single pathway inhibition, and broad-spectrum histone deacetylase inhibitors showed a time-dependent biphasic response, with high level downregulation at the transcriptional level within 24 h. In addition to furthering our understanding of ΔNp63 regulation in squamous cells, these data identify novel drug combinations that may be useful for p63-based therapy of SCC. Supplementary Information The online version contains supplementary material available at 10.1186/s11658-022-00323-x.

Background Squamous cell carcinoma (SCC) is one of the most prevalent forms of human cancer and is a major cause of mortality worldwide [1,2]. The most common sites are skin, head and neck, esophagus, lung and anogenital regions, and SCC is classified and clinically managed according to anatomical location. Despite advances in mechanistic understanding of SCC, improvements in patient survival have been modest, indicating that effective targeted therapeutic approaches are still lacking [3,4]. In particular, although SCCs are treated according to their site of origin, they share many biological and genetic characteristics, implying that targeting these common oncogenic pathways would be useful therapeutic approaches [1]. One common oncogenic event in SCC is overexpression of ΔNp63 [1,5,6], sometimes associated with TP63 gene amplification [7,8] although tumors without amplification also show overexpression [9]. ΔNp63 acts to maintain stem/progenitor cells, regulate differentiation and promote growth of normal squamous cells and SCCs (reviewed in [10][11][12]). In keeping with these roles, high ΔNp63 levels are associated with poor prognosis [13][14][15] and therapeutic resistance [14,[16][17][18][19]. Experimentally, ΔNp63 depletion enhances the effects of genotoxic agents on SCC cells in vitro, and ΔNp63 overexpression inhibits UV-radiation induced apoptosis of keratinocytes in vivo [16,18,20]. Thus, the ability to downregulate ΔNp63 would be therapeutically advantageous by the direct effects of p63 depletion on SCC cell proliferation, and/or by enhancing the effectiveness of conventional therapies. Achieving this aim in a clinical setting requires a fuller knowledge of the mechanisms involved in ΔNp63 regulation in tumor cells, which is known to involve multiple factors that induce positive and negative regulatory loops through cellcontext dependent pathways [11,12,21].
To explore the potential for targeting ΔNp63 in SCC treatment and uncover its regulatory pathways in these cells, we first used genetic approaches to investigate the effects of ΔNp63 depletion, providing evidence for ΔNp63 oncogene addiction in SCC. To explore whether ΔNp63 depletion can be achieved using clinically available agents, we studied the effects of known squamous cell growth factors and inhibitors of their signaling pathways, histone deacetylase inhibitors (HDACi), genotoxic agents and metabolism-modifying drugs to examine in detail their effects on ΔNp63 levels in SCC cells and non-transformed keratinocytes. These data provide a comprehensive analysis of factors involved in ΔNp63 regulation and reveal complex responses to the same treatment in different cell lines, indicating that a personalized approach will be required for optimal ΔNp63 inhibition. Nonetheless, we identify genotoxicity, dual AKT/mTOR inhibition and HDACi as major downregulators of ΔNp63, and the cholesterol-lowering agent lovastation as a selective upregulator of ΔNp63 transcription in non-malignant squamous cells. Combinations of these agents dependent on tumor characteristics will be particularly useful for ΔNp63 inhibition therapy.

Methods
All general chemicals and growth factors were obtained from Sigma-Aldrich (St Louis, MO, USA) unless stated otherwise. Drugs used in this study were of pharmaceutical grade or intended for cell culture experiments and were dissolved according to the manufacturer's instructions (details in Additional file 1). Controls were performed using the highest volume of the corresponding solute.

Cell culture
The SCC cell lines FaDu (human pharynx squamous cell carcinoma) [22] and SCC-25 (human squamous cell carcinoma of the tongue) [23] were obtained from ATCC (Manassas, VA, USA). HaCaT cells (spontaneously immortalized human keratinocytes that are non-tumorigenic and retain differentiation capacity) [24] were obtained from DKFZ (Heidelberg, Germany To determine proliferation rates after depletion, cells were seeded onto sterile 8-well slides (Ibidi Gmbh, Grafelfing, Germany) for the time required for cell adhesion. Doxycycline or control medium was then added for 4 days. Cells were fixed with cold methanol/acetone (50/50) for 10 min, dried, and immunostained with mouse monoclonal anti-Ki67 antigen (MIB-1 M7240 Dako) diluted 1:250 (0.18 µg/ml) using Envision peroxidase-polymer labeled anti-mouse Ig (Dako) for 30 min and diaminobenzidine (DAB) as the chromogen. Cells were counterstained with hematoxylin. Ki67 was quantified using QuPath image analysis [25] with default settings for hematoxylin/DAB and a detection threshold of 0.25 for all images, with 3 to 5 images (more than 1200 cells) used for each clone.
For colony-forming ability measurements, 250 single TP63-shRNA cells/well were flow-sorted into six-well plates in triplicate. After adherence, cells were cultured with or without 1 µg/ml doxycycline for 4 days before further culture without doxycycline. Colonies were stained with crystal violet (0.5% w/v in 20% methanol) and colony numbers counted in each well. After photography, colonies were destained in 1% SDS and the amount of dissolved crystal violet from each well was determined by absorbance at 570 nm. Based on previous evidence for their ability to regulate p63 [11,12,21], parental HaCaT, FaDu and SCC-25 cells were treated with insulin, insulin-like growth factor 1 (IGF-1), amphiregulin, epidermal growth factor (EGF), hepatocyte growth factor (HGF) or keratinocyte growth factor (KGF) at varying doses and varying times up to 24 h. According to the type of experiment and length of exposure, cells were grown to 30-70% confluence before treatment. Signaling inhibition employed the pan-phoshatidyl-inositol-3-kinase (PI3K) inhibitor wortmannin and the PI3K p110δ subunit inhibitor CAL-101, the mTOR1/2 inhibitor (rapamycin), the dual inhibitor of PI3K/mTOR (BEZ235), the p38 mitogen-activated protein kinase (MAPK) inhibitor (SB202190) and the EGF receptor (EGFR) inhibitor (cetuximab). In some experiments, basal signaling was reduced by incubating cells overnight in medium with reduced serum (0, 0.5 or 1% FBS) before treatment for 24 h in the same medium. These cells were also compared with those grown in medium with 10% FBS throughout to investigate the effects of serum reduction on basal ΔNp63 levels. Ultraviolet C (UVC; 254 nm), cisplatin, doxorubicin and etoposide were used as DNA damaging agents. Trichostatin A, valproic acid, sodium butyrate and SAHA (vorinostat) were used as inhibitors of class I and II HDACs, and nicotinamide was used to inhibit sirtuins (class III HDACi). Metabolism was modified using lovastatin that inhibits 3-hydroxy-3-methyl-glutaryl-coenzyme A (HMG-CoA) reductase and the mevalonate pathway, or metformin that acts through AMP-activated protein kinase (AMPK)-dependent and AMPK-independent mitochondrial pathways to regulate cellular energy metabolism. Experiments employing metformin were performed in medium with low glucose (1 g/L) for 16 h before treatment in the same medium for 24 h.

RNA isolation and reverse transcription quantitative PCR
Reverse transcription and quantitative PCR (RT-qPCR) was performed in specific situations where ΔNp63 protein levels were altered and where mechanistic data was not available previously. Total RNA was isolated using TRIzol reagent and 500 ng were reverse transcribed using High Capacity cDNA Reverse Transcription Kit (Applied Biosystems, Thermo Fisher, USA). Primers for ΔNp63 and ACTB (β-actin) (Additional file 3) were obtained from Generi Biotech (Hradec Kralove, Czech Republic). PCR was performed on a Fast Real-Time PCR System with Sybrgreen (Applied Biosystems): 95 °C for 3 min, 50 cycles of 95 °C for 5 s and 60 °C for 25 s. At least three biological replicates were performed, and each cDNA sample was analyzed in technical triplicates. Mean cycle threshold (Ct) values were transformed into relative mRNA levels [28] and ΔNp63 levels were normalized to ACTB.

Statistical analysis
Data are presented as mean ± SEM. Statistical significance (p < 0.05) was determined using unpaired 2-tailed t-tests against control values.

Preparation of inducible TP63-shRNA HaCaT and FaDu cells
Western blotting and RT-qPCR for p63 isoforms showed that HaCaT, FaDu and SCC-25 cells contain ΔNp63 mRNA and protein at high levels, with undetectable levels of TAp63 protein by western blotting and low or undetectable levels of TAp63 mRNA by RT-qPCR, in keeping with previous data [29][30][31]. Thus, we used a mono-specific mouse monoclonal antibody that recognizes only ΔNp63, showing a predominant band at 72 kDa (representing ΔNp63α, the major isoform in normal squamous cells and SCC cells [29,30]) with a minor band at approximately 68 kDa seen mainly in HaCaT cells that may represent ΔNp63β.
HaCaT, FaDu and SCC-25 cells were transduced with TRIPZ lentiviruses expressing TET-responsive TP63-shRNAs. Of the three shRNA sequences obtained, only #24246 (3' untranslated region) showed effective ΔNp63 reduction after induction by doxycycline, producing 85% to 95% reduction in ΔNp63 protein levels after 24 h in three individual primary clones of HaCaT and FaDu cells (Fig. 1A). Two HaCaT and FaDu clones were subsequently single cell re-cloned and expanded for further experiments. We were unable to produce stable SCC-25 TP63-shRNA cells, which grew poorly after single-cell cloning and during puromycin selection and maintenance.

ΔNp63 depletion selectively reduces colony formation and growth of SCC cells
To characterize the effects of depleting p63, stable HaCaT and FaDu TP63-shRNA cells were plated at cloning density and treated with doxycycline for 4 days after attachment. Clones were then allowed to grow and form colonies for 6 days (FaDu) or 10 days (HaCaT) in the absence of doxycycline. This transient p63 depletion at initial growth stages did not markedly influence the number of colonies, but significantly inhibited colony size, particularly in FaDu cells (Fig. 1B). Quantitation using crystal violet destaining (Additional file 4) confirmed that p63 depletion reduced cell  1C).

DNA damaging agents reduce ΔNp63
Previous studies have implicated DNA damage as a post-translational regulator of p63, where some forms of damage reduce ΔNp63 through a variety of proteasomal pathways [11,21] Fig. 2A-D). In addition, these treatments variably increased ΔNp63 at earlier times (4 or 8 h), most notable in doxorubicin and cisplatintreated HaCaT or FaDu cells ( Fig. 2A, C). These results are in general agreement with previous data, but also indicate cell-and agent-specific responses, despite each agent causing a similar level of DNA damage in each cell line.

Growth factor signaling regulation of ΔNp63
The effects of growth factor signaling pathways on ΔNp63 are controversial, with reports indicating both positive and negative effects for the same pathways [12]. In our experiments, addition of growth factors (insulin, IGF-1, EGF) did not alter ΔNp63 protein levels in cells under their normal growth conditions (10% FBS). Reducing the levels of endogenous growth factors by culture in medium with 1% FBS before and during growth factor treatment also did not change ΔNp63 levels compared to cells growing in 10% FBS (Fig. 3A). The addition of insulin/IGF-1 and the EGFR ligands amphiregulin and EGF to medium with 1% FBS increased the levels of ΔNp63 in SCC-25 cells but had minimal or no effects in HaCaT and FaDu cells (Fig. 3A). Cells cultured in serum-free medium before and during the 24 h treatments could not be analyzed due to extensive cell death. We also saw no effect of HGF or KGF on ΔNp63 levels under normal growth conditions or after serum depletion (see Additional file 6). Analysis of signaling activity after growth factor treatments showed minimal induction of phospho-AKT or phospho-ERK1/2 in FaDu cells, whilst phospho-ERK1/2 was increased by EGF in HaCaT cells and phospho-AKT was increased in SCC-25 cells after treatment with insulin or IGF-1 (Additional file 6).
To address the potential roles of growth factor signaling in p63 regulation further, we used small molecule or antibody inhibitors of specific pathways. Wortmannin (PI3K inhibitor) and CAL-101 (PI3K p110δ subunit inhibitor), rapamycin (mTOR1/2 inhibitor) and SB202190 (p38 MAPK inhibitor) did not cause significant changes in ΔNp63 levels, nor did they show appreciable reduction of phospho-Akt or phospho-p70 S6 kinase under the conditions used (Additional file 7). The EGFRinhibitory antibody, cetuximab, decreased ΔNp63 levels in HaCaT cells to less than 30% of the control value, with a smaller effect in SCC-25 cells (Fig. 3B). The dual PI3K/mTOR inhibitor BEZ235 selectively downregulated ΔNp63 in HaCaT cells, with more than tenfold reduction at the lowest dose tested (0.5 µM) compared to no reduction in FaDu and SCC-25 cells at this dose, and a tenfold higher dose was required to reduce ΔNp63 in SCC-25 (Fig. 3C). All doses of BEZ235 abolished phospho-Akt in all three cell lines (Additional file 7). Taken together with the growth factor supplementation data, these results indicate that the PI3K/Akt pathway is hyperactive in FaDu cells and is therefore difficult to either inhibit or activate with growth factors or inhibitors, whereas EGFR signaling and insulin/IGF signaling contribute to ΔNp63 regulation. Importantly, combined Akt/mTOR targeting effectively reduces signaling in all cells yet reduces ΔNp63 selectively in HaCaT cells.

ΔNp63 in FaDu cells
There is evidence that aspects of cellular metabolism are related to p63, with the glucose lowering agent metformin reportedly reducing ΔNp63 [32] or having no or minimal effects as a single agent in SCC cells [33]. Whether lipid metabolism is also a regulator is unknown. Treatment with lovastatin increased ΔNp63 protein levels in all three cell lines (Fig. 4). RT-qPCR was also performed to investigate whether this effect is at the protein or mRNA level. Induction by lovastatin was most pronounced in HaCaT cells, with a more than tenfold increase of mRNA and a correspondingly larger increase in ΔNp63 protein after 10 μM and 20 μM lovastatin for 24 h. This contrasts with the twofold and 3.5-fold increases in ΔNp63 mRNA in FaDu and SCC-25 cells, respectively, with correspondingly smaller increases in ΔNp63 protein levels (Fig. 4). We also investigated whether metformin influenced ΔNp63 protein levels, showing a reduction of approximately 50% in FaDu cells compared to 15-20% reduction in SCC-25 and HaCaT cells, respectively, after treatment with high doses of metformin for 24 h under low glucose conditions (Additional file 8).

HDAC inhibitors decrease ΔNp63 levels
Previous studies have investigated acetylation as a regulator of p63 activity using HDACi treatment, with both positive and negative effects reported under different conditions [34,35]. Whether these different effects are due to cell-type specific factors or differences in experimental conditions between studies is unclear. Moreover, whether HDACi regulation of ΔNp63 occurs at the transcriptional or protein level, and the effects of class-specific HDAci are unknown. Therefore, HaCaT, FaDu and SCC-25 cells were analyzed in dose response experiments after treatment with various HDACi for 24 h. All four broad spectrum HDACi (sodium butyrate, valproic acid, SAHA and trichostatin A) reduced ΔNp63 in these cells to varying extents (Fig. 5). Nicotinamide, an inhibitor of class III HDACs (sirtuins), reduced ΔNp63 only in FaDu cells (Additional file 9).
Of the range of HDACi tested, the broad spectrum agent trichostatin A (TSA) showed the greatest effect and was therefore studied in more detail at both protein and mRNA levels. Time course experiments showed notable reduction of ΔNp63 protein from 16 h onwards (Fig. 6, upper panels), while RT-qPCR showed a biphasic time response of ΔNp63 mRNA in HaCaT cells, with a transient increase at 4 h and 8 h before a reduction at 16 and 24 h (Fig. 6, lower panels). FaDu and SCC-25 cells did not show early induction of ΔNp63 mRNA and showed greater downregulation at 24 h (fourfold and 5.5-fold, respectively, compared to 2.5-fold reduction in HaCaT cells at 24 h).

Discussion
ΔNp63 overexpression is a characteristic feature of SCC, and the ability to pharmacologically reduce ΔNp63 levels would therefore be expected to have broad value for SCC patients, and for patients with other tumor types in which ΔNp63 plays an oncogenic role, such as bladder, breast and prostate cancer subtypes [11,12,[36][37][38]. We confirmed that p63 knockdown inhibits the growth of HaCaT keratinocytes and malignant SCC cells, with an increased effect in the latter, providing proof of concept that reducing ΔNp63 would have value for SCC therapy. These findings are similar to observations that enhanced ΔNp63 expression causes hyperproliferation of squamous cells and leads to malignancy [39], with p63 being required for keratinocyte proliferation and for both tumor initiation and maintenance [40,41]. Our finding that malignant FaDu cells show a more severe loss of growth potential than HaCaT keratinocytes after ΔNp63 depletion implies that SCC cells are addicted to ΔNp63 oncogenic activity, and indicates that a therapeutic window of opportunity exists for tumor-selective effects. The regulation of ΔNp63 in normal and malignant squamous cells is multifaceted, involving cell-type dependent and interactive effects of numerous growth factors, signaling pathways, metabolism and chromatin modifications (reviewed in [11,12,21]). Using this knowledge, we investigated ΔNp63 regulation in SCC cells and non-tumorigenic keratinocytes to identify factors that modulate this oncogenic protein in these cells. We aimed to identify clinically available agents that modify ΔNp63 levels and are therefore potentially useful for patient therapy. We used two distinct SCC cell lines, which represent the mesenchymal (SCC-25) and the atypical molecular subtypes (FaDu), and are derived from tongue and hypopharynx, respectively [42]. These were compared to non-malignant epidermal keratinocytes that retain squamous cell differentiation potential (HaCaT). Whilst this introduces heterogeneity in terms of tissue of origin and cancer-specific genetics, which at least in part may underlie the variable effects observed, our aim was to identify agents that consistently influence ΔNp63 in SCC and would therefore be useful across sub-types by reflecting the common features of SCC, which includes ΔNp63 upregulation [1,5]. In particular, examining three cell lines under the same conditions allowed us to make direct comparisons between cells, rather than comparing the results of previous studies that individually employed different drugs at different concentrations in different cell lines, sometimes providing different conclusions from each other. It is known that DNA damage downregulates ΔNp63 through proteasome-mediated degradation [12,21,43], and we showed downregulation of ΔNp63 by multiple DNA damaging agents. Interestingly, a transient increase was observed prior to loss, which is attributed to the initial phosphorylation of ΔNp63 prior to degradation [21,43,44]. Thus, in addition to the direct cytotoxic action of DNA damaging agents, these therapies also decrease the pro-survival effects of ΔNp63, and ΔNp63 ablation enhances DNA damage cytotoxicity in SCC cells [16,18,45], whilst forced expression of ΔNp63 protects against UV-mediated keratinocyte cell death in vivo [20]. However, we also found variability in the ΔNp63 response of SCC cells (even though DNA damage levels were relatively consistent across cells as measured by phospho-H2AX), suggesting that factors such as differing levels of endogenous DNA damage response activities in SCC [46] and/ or SCC-specific pathways of DNA repair [18] may be involved.
Favorable SCC growth conditions are mediated by the presence of tumor-or stromal-derived growth factors as well as activating mutations in signaling pathways such as PIK3CA or inactivation of the signaling inhibitor PTEN [1,5,6]. As a pro-survival factor for squamous cells, it is logical that ΔNp63 may be regulated through such pathways. Indeed, EGFR and/or PI3K activity are positively correlated with ΔNp63 [47][48][49] and dual activation of ΔNp63 and PI3K signaling is a hallmark of SCC [1,6]. Whilst initial studies indicated that EGF increases ΔNp63 in SCC cells and keratinocytes through the PI3K pathway [50][51][52], other reports indicate that PI3K activation transcriptionally reduces ΔNp63 [53,54]. Various other growth factors have also been linked to ΔNp63 regulation with similarly conflicting data reported [11,12]. In our experiments, there were no consistent effects of the growth factors tested (insulin, IGF-1, EGF, KGF, and HGF) despite serum reduction before and during treatment, as used in some studies [51]. We also did not see consistent effects of wortmannin (pan PI3K inhibitor) or rapamycin (mTOR inhibitor) at concentrations that showed cytotoxicity. These results presumably reflect differing activation of these pathways in the cells used. For examples, mutation or amplification of PIK3CA or EGFR, or PTEN inactivation are all common in SCC [1,6], but would be expected to show different sensitivities to exogenous stimulation or inhibition depending on the specific mutation and the combination of aberrations in growth factor signaling components [55]. Moreover, PI3K inhibitors may induce downstream targets due to feedback loops, again depending on the oncogenic status of the tumor cells [55]. On the other hand, our data confirm the potential role of EGF and identify insulin/IGF signaling in inducing or maintaining ΔNp63 levels in keratinocytes and SCC-25 cells. Most notably, the dual PI3K/mTOR inhibitor BEZ235 reduced ΔNp63 levels in all cells, confirming the involvement of PI3K/mTOR signaling. Thus, although growth factor signaling pathways play a role, the effects of pathway inhibitors on ΔNp63 are complex and depend on the cancer mutational status. Cellular metabolism is an emerging target for cancer treatment, and there is evidence that p63 is a regulator of metabolic activity and may itself be regulated in response to metabolic imbalance [11,12,21]. The widely used lipid lowering drugs, statins, are also being investigated for their anti-cancer effects through their action as mevalonate pathway inhibitors [56]. Importantly, statins are not universally effective, and epithelial tumors (high E-cadherin) tend to be insensitive, whilst other tumor types are insensitive due to compensatory activation of SREBP that induces mevalonate pathway genes [56]. We found that lovastatin increased ΔNp63 protein and mRNA in HaCaT cells, with lesser effects in FaDu and SCC-25, which may relate to the findings that mevalonate pathway intermediates increase ΔNp63 in non-transformed oral squamous cells [57]. In addition, lovastatin promotes survival of oral squamous cells after genotoxic insults through inhibiting ATM and/or ATR activation [58], which are themselves responsible for DNA damage-mediated destruction of ΔNp63 [12,21]. Cellular energy metabolism has also been linked to p63, and high doses of metformin reduce ΔNp63 in synergy with glycolysis inhibition or in combination with HDACi [32,33]. In our experiments, metformin reduced ΔNp63 in FaDu cells under low glucose conditions, similar to previous observations in these cells [32], whereas HaCaT and SCC-25 cells showed a minimal response to metformin alone, similar to data reported in other SCC cells [33]. Taken together, these data indicate roles for both lipid metabolism and energy balance in differential regulation of ΔNp63 in keratinocytes and SCC cells. For metformin, reduction is caused by proteasomal degradation [32], whereas we show that lovastatin acts at the transcriptional level. Acetylation of histones in gene regulatory regions is an important element of gene regulation in SCC [5,59] and influences p63 [12]. In our experiments, inhibiting class I and II HDACs reduced ΔNp63 in a time-and dose-dependent manner. Importantly, we discovered a biphasic response, with an early increase of ΔNp63 mRNA prior to its reduction, particularly noticeable in HaCaT cells (Fig. 6). These findings help to explain previous discrepancies that HDACi upregulated rather than downregulated ΔNp63 [34]-those experiments used short-term exposure compared to the longer exposures used here and elsewhere that decrease ΔNp63 [35,48]. Thus, at least two pathways are in operation, with an early induction (perhaps related to direct acetylation of ΔNp63 [34]) followed by a prolonged inhibitory response that may be transcriptional by altering histone acetylation at the promoter or enhancer regions of the ΔNp63 gene [12], or through altered regulation of other genes that influence ΔNp63 such as miRNAs [35] or PTEN [60].

Conclusion
SCC is a common malignancy that occurs at multiple anatomical locations and is often difficult to treat, with poor response and high mortality [1][2][3][4]. ΔNp63 has many oncogenic roles in SCC and its levels correlate with aggressiveness and therapy resistance. With respect to the latter, ΔNp63 protects cells from DNA damage-induced death and reducing ΔNp63 levels sensitizes to radiotherapy and chemotherapy [16, 18-20, 45, 61]. On the other hand, there is good clinical evidence that some chemotherapeutics are particularly effective in p63-positive tumors, indicating that ΔNp63 downregulation  is a major component of their anti-cancer properties in these specific cancers [62,63]. Thus, the ability to inhibit ΔNp63 is a logical approach to improve patient outcomes for these tumors, either as a single agent or as combination therapy. By screening known or suspected agents for their effects on ΔNp63, we identified selective cell-type dependent inducers and repressors (Fig. 7). Therefore, although these agents have many effects independent of ΔNp63 regulation, the data provide a valuable and practical framework for future investigations of pharmacologic p63 depletion, including indications for specific combination therapies dependent on tumor characteristics and mutational status reflected by molecular subtype. For example, lovastatin increased ΔNp63 levels predominantly in HaCaT cells and thus may be useful for p63-inhibition combination therapies by helping to maintain its levels in normal cells. That HDACi variably reduce ΔNp63 implies they may be useful for certain tumors and/or could be used in combination with EGFR inhibitor therapy or with metformin [33,48]. In addition, our data show that dual AKT/mTOR inhibition is especially effective at reducing ΔNp63, but may cause more severe side-effects on normal tissues. Taken together, we demonstrate that ΔNp63 is an amenable target for manipulation in SCC by several commonly used and well-tolerated pharmaceuticals.

Clinical implications
In summary, we have shown that ΔNp63 is required for the growth of SCC cells, confirming its' reduction as a potential therapeutic approach for these tumors. We also show that drugs with divergent mechanisms influence ΔNp63 levels in keratinocytes and/or SCC cells. The different effects seen in different cells presumably reflect the mutational characteristics of signaling pathways in individual SCC tumors, many of which regulate and/or cooperate with ΔNp63 to promote carcinogenesis. Thus, although ΔNp63 is not their sole target, many commonly used drugs influence this protein and our data indicate that inhibiting the oncogenic p53 family member ΔNp63 is a therapeutic option for SCC.