3.2 Characteristics of Studies
Of the 64 studies encompassing cell-line models identified in the search (Supplementary Table 2), 79% were compliant with the guidelines suggested by TEMTIA and 13% provided insufficient evidence for their claims of EMT [33]. The remaining 8% were marked as not applicable due to study scope and design (e.g. differential gene expression analysis). However, investigation of spatial and/or transcriptional information beyond the pure protein levels of markers as obtained by Western Blotting was rare. In addition, integration of EMT-TFs and linking of specific transcription factors to their influences on EMT markers and properties in keratinocytes is infrequent.
While many of the experimental studies met TEMTIA guidelines, the cell line models used to investigate EMT in cSCC (Table 1) misrepresent the patient population and aetiology. The most frequently used model (40%) is the vulva-derived A431 cell line. The location of the primary tumor makes UV involvement (the most frequent cause of cSCC) in the aetiology unlikely, as has been noted by others [37]. Male sex is a risk factor for developing cSCC accounting for a 1.5- to 2-fold increase in incidence and increased mortality rates [38]. The two most frequently used models, A431 and SCC-13, are both derived from female donors and account for 57% of studies reversing the clinically observed male to female ratio. This may be of concern due to hormonal differences between the sexes and the potential role of hormonal regulation of EMT in cSCC [39–42].
Table 1
Cell line models used to study EMT in cSCC (used in > 5% of studies).
Model | Frequency of studies[%] | Location of tissue of origin/Patient characteristics |
A431 | 40 | Vulva (F, 85) |
SCC-13 | 17 | Face (F, 56) |
HaCaT | 15 | Upper back/Transformed (As/ Ras) (M, 62) |
MET1/2/4 | 14 | hand/ hand/ left axillary lymph node (M, 45) |
SCL1 | 9 | Face, Caucasian (F, 74) |
SCC12 | 9 | Face (M, 60) |
HSC-5 | 6 | Location unknown/Japanese (M, 75) |
NHEK/ PHK | 6 | Normal adult skin/Variable Location(HPV/ ectopic protein expression) |
SCC-IC1 | 5 | Right temple (M, 75) |
M = male, F = female. |
There were no uniform set of markers used to assess EMT. A listing for each study specifying marker and methods is available in Supplementary Tables 1 and 2. The most frequently employed epithelial and mesenchymal markers were E-cadherin, Cytokeratins, and EpCAM as well as Vimentin, N-Cadherin, Fibronectin and α-SMA actin, respectively. Sometimes other non-canonical EMT markers were employed to assess other closely linked properties or processes such as disassembly of cellular junctions (Girdin, α-/ βcatenin, ZO1, Desmoglein 3), differentiation and stemness (KLF4, involucrin, CD133, CD44), and ECM remodelling (MMPs, uPAR, PAI-1). EMT marker analysis was primarily performed via Western blotting (84%), immunofluorescence (39%), and RT-qPCR (22%). Morphological changes towards a more spindle like phenotype are were used as starting point to prompt further investigation. Migration and invasion were frequently assessed using transwell migration assays, scratch wound assays, transwell migration assays into Matrigel coating, or organotypic assays. Cell adhesion was rarely evaluated but when done, used either a trypsinization assay or bead-based cell traction force assays. EMT property related assays were often paired with assays assessing stemness (colony/spheroid/tumor-forming assays) or proliferation (MTS/ MTT, CCK8, counting).
Of the many reports screened that included clinical samples, only 22 purely clinical investigational studies and 20 integrating both clinical investigation and cell line investigated EMT or related signalling in cSCC. A listing of each of these eligible studies specifying marker, methods and tissue comparisons is available in Supplementary 1. Again, as for experimental studies, the markers used to assess EMT were varied and included mRNA, proteins and miRNA/lncRNA expression analyses by comparison most often to normal or matched skin. Synthesis of this data in a timeline of cSCC progression is summarized in Fig. 2 and discussed below.
3.3 Synthesis of Results
3.3.1 Clinical progression of cSCC is paralleled by increased acquisition of a mesenchymal phenotype and markers.
Human epidermal keratinocytes display alterations of surface markers and transcription factors consistent with an EMT-phenotype early during the clinical progression. Bakshi et al. [43] reported reduced E-cadherin levels paralleled with increased Snail, Twist and Slug expression in sunexposed skin (SES) vs non-sunexposed skin (NSES). Others have found that miR497 levels are significantly reduced in SES vs NSES. miR-497 is a negative modulator of EMT markers Slug, N-cadherin and Vimentin, a direct repressor of SERPINE1 and inhibitor of migratory properties in cSCC cell lines [44]. In cSCC, Snail and Slug levels have been correlated negatively with E-cadherin levels [45].
The OVOL transcription factors 1 and 2 are important negative regulators of EMT [46]. Both are upregulated in AKs compared to cSCC. Loss of OVOL1/2 is associated with increased expression of EMT markers, Vimentin and Zeb1 [47]. Murata M. [47] reports an inverse association of OVOL2 and Zeb1 levels in 30 AK and 30 cSCC samples. Additionally, ZEB1 expression increased upon both OVOL1 and OVOL2 knockdown in A431 cells. Similar findings by Ito et al. [48] also suggest a role of OVOL transcription factors as the last guardian against malignancy in pre-neoplastic lesions. Conversely, OVOL1 is downregulated in Bowen’s disease vs cSCC and OVOL2 changes its location from primary nuclear to predominantly cytoplasmic upon acquisition of malignancy. Furthermore, Ito et al. [48] reported OVOL1/2 as negative modulators of the proto-oncogene c-Myc and invasiveness.
An increasing body of evidence suggests that EMT is the determining factor for the progression of AK and Bowen’s disease to cSCC. However, there are differences between the classical and differentiated pathway. Saenz-Sardà et al. [49] found significantly lower expression of EMT markers Vimentin, E-cadherin, and membranous β-catenin in the cSCC arising through the differentiated pathway when compared to the classical pathway. Additionally, the proliferation marker Ki67 was significantly increased in cSCC arising through the classical pathway [49]. Additional differences between the pathways include increased miR31 and MMP levels in the differentiated pathway [50]. Nevertheless, the expression of markers at the invasive front of cSCC and Bowen’s disease acquiring de novo invasive ability suggest a pivotal role of the EMT process in facilitating invasion in cSCC [49; 51].
The importance of EMT in cSCC malignancy is further supported by increased expression of EMT-TFs and mesenchymal markers including Snail, Slug, ZEB1, Twist, Podoplanin, and Vimentin [43; 45; 47; 52–57] with increasing progression of the disease and increasing loss of differentiation. The gain of mesenchymal markers and properties is complemented by loss of epithelial markers (Involucrin, E-cadherin, KLF4, Cytokeratin, Claudin1) and apparent disassembly of cellular junctions [43; 51; 53–55; 58–67]. For example, Girdin, an adherence junction protein closely linked to E-Cadherin, has been associated with collective migration. Furthermore, Girdin expression is correlated to well-differentiated cSCC but is lost in poorly differentiated cSCC [68]. Toll et al. [56] linked cSCC expressing Vimentin, Twist, ZEB1, nuclear βcatenin and podoplanin to lymph node metastasis in a study cohort with tumors from 146 patients. Additionally, Vimentin levels correlated with recurrence, disease specific death, tumor stage, perineural invasion (PNI), desmoplasia, and differentiation. Podoplanin, despite not being a classical EMT-marker, was also identified by another independent study as predictor of regression-Tfree survival [52]. Perineural invasion (PNI) is an established predictor of recurrence, metastasis, and poor prognosis in cSCC [69]. Brugière et al. [66] mapped cells with EMT-features to the site of PNI and increased neurotrophin signalling. This evidence links EMT to both lympho-vascular and PNI, two hallmarks of advanced disease and poor patient outcome.
Ji et al. [15] resolved the spatial architecture of 10 cSCC with matched normal skin using a combination of single cell-RNA sequencing, spatial transcriptomics, and multiplexed ion beam imaging. They identified a subpopulation of keratinocytes exclusive to tumor tissue. These tumor-specific keratinocytes (TSKs) are located towards the invasive edge of the tumor and possess an EMT-like signature. TSKs express markers of EMT (Vimentin, ITGA5) despite lacking the expression of classical EMT-transcription factors (excluding SLUG). Single-cell regulatory network interference and clustering nominated AP1 and ETS transcription factors as regulators of EMT in TSKs. Additionally, their preferential localization at the invading edge infers invasive migratory properties when compared to their basal, differentiating and cycling counterparts. Furthermore, TSKs display a broad spectrum of EMT markers reflecting the high complexity and plasticity underlying the EMT continuum.
Even further along the epithelial-mesenchymal spectrum than cSCC are spindle cell cSCC (sc-cSCC), a poorly differentiated form of cSCC with a characteristic mesenchymal phenotype [70]. Nakamura et al. [60] reported a case of cSCC mimicking an atypical fibroxanthoma staining positive for Vimentin and Snail whilst staining negative for Cytokeratin. Iwata et al. [61] reported cases of sc-cSCC with complete loss of E-Cadherin, p120catenin, and Desmoglein 3. More recently, Shimokawa et al. [59] showed increased nuclear staining of Snail, increased cytoplasmic Vimentin, and decreased levels of E-cadherin paralleled by significant reduction of COX2 levels in six sc-SCC compared to three nonsccSCC. Combined with the case reports, this warrants the consideration of sc-SCC as a tumor displaying advanced features of EMT and clinical progression of cSCC. Even more advanced, the position of cutaneous carcinosarcomas (CCS), biphasic tumors constituting of both a mesenchymal and epidermal component, in this clinical progression remains elusive [71]. The involvement of EMT in the formation of these tumors is up for debate [64; 72; 73]. Matching genotypes of both phases and EpCAM positive staining of mesenchymal and epidermal components both point towards singular epidermal origin of CCS and infer a quasi-full EMT as potential mechanism [64]. However fascinating, the rarity of sccSCC and CCS makes them less relevant in the overall picture [70; 71].
3.3.2 Genetic disorders associated with increased frequency of aggressive cSCC facilitate EMT via altered cell-matrix interactions.
Epidermolysis bullosa (EB) is a group of conditions that are associated with early onset and rapid progression of cSCC [20; 75]. Kindler syndrome is caused by mutations in the FERMT1 gene [76]. FERMT1 codes for Kindlin1, a protein co-localizing at focal adhesions and involved in the activation of their receptor functions. Lack of Kindlin1 is associated with dysregulated integrin signalling, cell adhesion and migration [77]. Immortalized patient-derived kindlin-deficient keratinocytes display reduced cell-cell adhesion, cell-matrix adhesion, epithelial markers. Conversely, induction of mesenchymal markers, ECM components and proteases point towards a more mesenchymal phenotype [78]. Conflicting data by Ji et al. [15] identifies FERMT1 as an essential transducer of integrin signalling in TSKs and amplification across many cancers. Additional studies suggest a role for FERMT1 in regulating EMT across multiple cancers including [79–81]. In cSCC, evidence points towards a contextual involvement of Kindlin1 in EMT regulation and warrants further investigation [15; 78].
RDEB patients bear loss-of-function mutations in the COL7A1 gene and produce highly aggressive and metastatic cSCC [75; 82]. Knockdown of ColVII in the non-RDEB cSCC cell lines, Met1 and SCC-IC1 increased migration and invasion by enhancing EMT and preventing differentiation [83]. In xenografts derived from ColVII knockdown SCC-IC1 cells, recombinant human ColVII reduced the effects of the ColVII deficiency. In addition to increased angiogenesis, ColVII knockdown resulted in amplified TGF-β1 signalling and upregulation of urokinase plasminogen activator (uPA), SERPINE1 and VEGFA [84]. Clinically, RDEB tumors displayed increased EMT markers and increased levels of TGFβR1. Furthermore, loss of ColVII directly correlated with decreased levels of involucrin, an epithelial differentiation marker in vivo [84]. Twaroski et al. [85] confirmed findings of increased TGF-β1 signalling and identified MEK/ERK, p38 and SMAD3 as downstream effectors and mediators of an EMT phenotype. Together these studies provide an attractive model for the high aggressiveness and high metastasis rates in RDEB cSCC.
3.3.3 Cancer-associated fibroblasts induce EMT in cSCC via paracrine growth factor, cytokine, integrin and MMP signalling.
Cancer-associated fibroblasts (CAFs) and other cancer-associated cell types in the tumor microenvironment are implicated in several cancers to induce EMT via paracrine signalling (refer to Fig. 2) [1; 15; 86]. Co-culture of fibroblast and cSCC cancer cells has been shown to be paralleled by increased CAF- and EMT marker expression [87]. In cSCC, fibroblast subtype can also play a significant role in cancer progression, with reticular fibroblasts conferring EMT and invasion more than papillary fibroblasts [87]. On the same note, Bordignon et al. [88] reported different CAF subtypes with diametric influences on EMT markers and tumor invasiveness in cSCC. A CAF population induced by TGF-β signalling significantly increased invasiveness and tumorigenic expansion in vivo compared to FGF2-induced CAFs. The increased aggressiveness was associated with increased EMT markers (Vimentin, Snail, Slug and Twist) and altered TME (e.g. COL1A1 secretion). Additionally, conditioned media derived from senescent fibroblasts induced EMT in post-senescence emergent NHEKs via MMP-PAR-signalling. Conversely, gelatinolytic activity, PAR-1 expression, as well as TWIST expression were elevated in aged human skin samples [89]. In RDEB cSCC, patient-derived fibroblasts secrete TGF-β to facilitate EMT [85].
Clinically, Sasaki et al. [90] investigated 237 specimens derived from basal cell carcinoma (BCC), malignant melanoma (MM) and cSCC for their expression of CAF and EMT-related markers. Independent clustering based on the marker expression yielded a significant correlation between clinic pathological subgroup and malignancy. However, there were BCC, MM, and cSCC represented in all subgroups. Furthermore, the subgroups also correlated with clinical parameters such as lymph node metastasis, tumor thickness and tumor size. Ji et al. [15] showed that TSKs engage in reciprocal signalling with CAFs, endothelial cells, macrophages, and myeloid-derived suppressor cells (MDSC). These interactions form a complex network of signalling molecules, including ECM components (FN1, COL1A1), cytokines (TGFB1, TGFB3, and CXCLs), growth factors (PGF, VEGFA), proteases (MMP9), and integrins (ITGA3, ITGB1). However, the influence of the individual factors on the EMT status remains subject to further investigation. For example, some evidence suggests, that stromal macrophages do not play a major role in the induction of EMT in cSCC [57].The tumor microenvironment alters the EMT status of tumor cells not only through paracrine signalling but also through the properties and composition of the ECM itself [91]. For example, HPV transformed keratinocytes (N/TERT keratinocytes) elicit an EMT response to fibronectin via α3β1integrins [92]. The influence of matrix stiffness and mechanotransduction pathways on EMT has been investigated in other cancers but remains underexplored in cSCC [93; 94].
3.3.4 βcatenin provides a mechanistic explanation for the close link between stem-like properties and EMT in cSCC.
Cancer stem cells have gained traction in recent years as they link major challenges modern cancer therapy faces including recurrence, therapy resistance and metastasis [6; 95]. Even though stemness and EMT are separate phenomena, they are closely linked [6]. Cells, staining positive for the stem cell markers CD44 and CD29, at the invasive front of tumors also displayed characteristics consistent with EMT [96]. In three kidney organ transplant recipients, EMT markers (Vimentin, Slug, and Snail) were co-expressed in CD133 expressing (stem) cells in invasive areas of skin SCCs but not concomitant AKs or normal skin [58]. A mechanistic link is provided by the downregulation of E-cadherin, which releases sequestered βcatenin from the cell membrane to the nucleus [68; 97]. Furthermore, this might work synergistically with dysregulated Wnt/ βcatenin signalling, an inducer of cancer stem cell properties [98–100]. Clinical evidence shows the co-localization of βcatenin and E-cadherin at cellular junctions [49; 68]. Loss of E-cadherin is associated with the disassembly of those junctions, decreased adherence and increased nuclear βcatenin (Figs. 2 and 4) [49; 68]. The transcriptional repression of E-cadherin is mediated by canonical EMT-TFs (e.g. Snail) as well as other transcription factors such as Grhl3 (Fig. 4) [59; 101]. Increased nuclear βcatenin has been observed in tumors with poor differentiation and correlated to lymph node metastasis [53; 56; 68].
The close link of stemness and EMT is backed by an increasing body of in vitro evidence. Biddle et al. [102] performed CD44/ EpCAM based sorting of the cell lines PM1, MET1 and MET2, derived from dysplastic skin, the primary lesion, and a recurrence at the same anatomical site, respectively [103]. They observed an increase of in the Epcam low/ CD44 high population with increasing malignancy. Further, a more prominent EMT phenotype, sphere forming ability but reduced proliferative ability distinguished the EpCAM low/ CD44 high population from the EpCAM high/ CD44 high population. CD44/ ITGB1 based sorting of the A431 cell line also identified a subset within the cancer stem cells (CSC) that display EMT characteristics. In murine xenograft models of A431 cells, the CD44 high/ ITGB1 high stem cells gave rise to significantly bigger and more aggressive tumors [96]. Additionally, the simultaneous regulation by common upstream regulators including ARMC8, ΔNp63α, p38/NFκB, transglutaminase II (TGA2) and Axl [104–109], as well as pharmaceutically active substances further tightens the link between of EMT and CSC properties [106; 110].
3.3.5 Proteases including the urokinase plasminogen activator system are underexplored markers and EMT effectors.
Proteases such as matrix metalloproteases contribute to tumor cell invasion and EMT, via multiple mechanisms [111; 112]. Proteolytic cleavage of the ECM allows for increased migration and liberates latent signalling molecules such as EGF, HGF, and TGF-β [113; 114]. In two-dimensional A431 cell culture, broad inhibition of MMPs and MMP9 knockdown reduced EMT marker expression and motility. Additionally, the EMT-TF Snail induced expression of MMP-9 [115]. MMP2 contributes to the invasiveness and migratory abilities in TGF-β-induced EMT of RDEB cells [85]. TGF-β1 and EGF treatment in Ras-transformed HaCaT cells induced an EMT phenotype and increased collagen remodelling. The broad-spectrum inhibition of MMPs using GM6001 increased cell attachment and abrogated collagen degradation [116]. An often-overlooked contribution of MMPs to EMT signalling is mediated by G-protein-coupled cell surface receptors, transducing a signal upon MMP cleavage [117]. MMPs, MMP1, MMP-2, and MMP-3, secreted by senescent fibroblasts induced EMT markers and migration via a PAR1-mediated mechanism [89].
Wilkins-Port et al. [116] identified the involvement of another protease system in TGF-β1 and EGF induced EMT of ras-transformed HaCaT cells, the urokinase plasminogen activator system (uPAS). Inhibition of the uPAS using amiloride (uPA inhibitor) or plasminogen activator inhibitor 1 (PAI-1, encoded by SERPINE1), an endogenous uPA inhibitor, reduced MMP-1 and MMP-10 levels as well as attenuated collagen. Additionally, PLAU (uPA) is part of the TSK-gene signature identified by Ji et al. [15]. TGF-β1 and EGF induce the expression of PAI1 and multiple studies report the transcriptional regulation of SERPINE1 parallels the induction of EMT [44; 118]. Mizrahi et al. [44] found progressive down-regulation of miR-497 through promotor methylation during the progression of SES to cSCC. Expression of miR497 reduced SERPINE1 levels as well as levels of other EMT related genes.
3.3.6 miRNA and lncRNA expression profile changes during EMT play a central role in its regulation.
During the progression from normal skin to cSCC, the expression of many non-coding RNAs such as miRNAs and lncRNA are altered. miRNAs regulate the transcription of proteins by regulating the stability of their respective target mRNAs [119]. Mizrahi et al. [44] investigated the differential expression of miRNAs in a clinical progression of NSES to cSCC. The changes in the miRNAs expression ranged from gradual increase/decrease to stepwise acquisition/loss creating a progression specific profile. For example, the loss of the tumor suppressor miR497 increased SMAD signalling, EMT as well as SERPINE1 expression. Other studies also found miRNAs dysregulating some of the signalling pathways central to EMT including the PI3K/Akt pathway and the Wnt-pathway. The tumor suppressive miR-451a, a suppressor of PDK1, is downregulated in metastatic vs non-metastatic tumors [120]. The suppression of PTEN by miR21 leads to increased pathway activity and Akt activation [121; 122]. The oncogenic miR-22 is gradually upregulated with increasing grade of cSCC and promotes stemness via Wnt signalling [98]. A mechanistic study by Robinson et al. [123] identified miR-211 and miR-205 as part of an iASPP/ p63 epigenetic feedback loop regulating EMT, with the latter directly targeting Zeb1 and p63.
LncRNA can act as molecular sponges for miRNAs by competing for binding with their cognate mRNAs [124]. However, other mechanism of action such as chromatin remodelling, transcriptional regulation or mRNA post-transcriptional regulation are possible [119]. For example, the lncRNA HOTAIR sponges miR326 leading to an increase of PRAF2 and a more prominent EMT phenotype [124]. MALAT1 is upregulated in cSCC compared to normal skin and promotes EMT via modulation of Wnt-signalling [125] Clinically, Li et al. [126] found a correlation between the upregulation of LINC00319 and tumor size, lymphovascular invasion, and TNM stage. In cSCC cell lines, LINC00319 favoured migration, invasion, and EMT marker expression [126]. LncRNA, H19, and miR-675 are upregulated in cSCC. Upregulation of H19 increased miR-675 levels as well as the expression of EMT markers [127].
3.3.7 Environmental factors such as arsenite induce EMT and transformation by widespread alteration of miRNA expression, mRNA expression and induction of IL-6 signalling.
After UV exposure, arsenic exposure is one of the biggest occupational hazards for developing NMSC [128]. The acute and chronic toxicity of arsenite can be replicated in vitro by exposing HaCaT cells to arsenite (0.1 µM- 1 µM) for up to 28 weeks [122; 129]. Al-Eryani et al. [130] found a significant dysregulation of EMT and cell cycle genes as early as seven weeks after exposure to arsenite. Banerjee et al. [129] reported decreases in ZO-1, a tight junction protein, and increased Slug after 19 weeks. After 28 weeks, the arsenic-exposed cells display an advanced EMT phenotype. Additionally, pathway analysis of the differentially expressed miRNAs and mRNAs showed inhibition of the ER stress pathway. Investigations into the molecular mechanism implicate NFκB, PI3K and IL-6 signalling as the major contributors to the EMT phenotype (Fig. 3). IL-6 induces miR-21 via STAT3 cytokine signalling [121]. In return, miR21 activates Akt via the inhibition of PTEN [122]. Furthermore, the increased CSC-like properties are ascribed to increase p38/ NFκB signalling as well as the induction of IL6 [105; 106]. Upregulation of the EMT-TF Snail provides a direct link to canonical EMT signalling [105; 106].
Other external factors such as UV and ROS also promote EMT in malignant keratinocytes. UV radiation induces the EMT transcription factor Snail via an AP1-dependent mechanism and UV exposure is associated with decreased E-cadherin levels clinically [43; 131]. Additionally, UV irradiation promotes cSCC via increased production of reactive oxygen species (ROS) [132]. Conversely, the tumor suppressor and negative modulator of oxidative stress, NAD(P)H dehydrogenase (NQO1), is lost in cSCC. Adenoviral expression of NQO1 reduced ROS levels and attenuated EMT [133]. On the other hand, cellular stress can induce autophagy by reducing the levels of the autophagy marker, p62 [134]. p62 directly binds and stabilizes the EMT-TF Twist1 [135]. Additionally, p62 can induce NFκB signalling further linking inhibition of autophagy to EMT [134].
3.3.8 Thyroid hormone and Estrogen signalling modulates EMT in cSCC.
Male sex is a risk factor for cSCC. Male patients are often younger, present more frequently with metastatic disease and have a worse prognosis [136–139]. The increased incidence is often attributed life-style choices. However, more recent research suggests an underlying biological cause [137; 140]. Some scarce evidence infers a role of hormonal signalling in modulating EMT in cSCC. Chen et al. [39] reported estrogen-dependent activation of the FN1-STAT3 axis in the vulva-derived A431 cell line. The subsequently induced EMT could be reversed with the inverse agonist XCT790. Nappi et al. [141] reported on a link between thyroid hormones, EMT, and tumor stage. NANOG and Deiodinase 2 (D2) were proportionately increased significantly with increased cSCC stage. D2 catalyzes the conversion of the thyroid hormone (TH) T4 to T3. TH depletion of the growth medium and inhibition of D2 with rT3 both reverted the EMT phenotype in SCC13 cells. A second study confirmed that T3 induced thyroid hormone receptor α (THR) directly binds the Zeb1 promotor and induces transcription of the EMT-TF [142]. Additionally, high D2 levels correlated with more advanced stage, a higher risk of relapse and lower overall survival in two independent datasets [142]. Together, these findings warrant a closer investigation of the hormonal influence on EMT in cSCC.
3.3.9 EMT in cSCC can be attenuated by targeting MAPK, cytokine, growth factor, and NFkB signalling.
Drug resistance and recurrence are two major challenges modern cancer therapy has to overcome. Due to its link to both these phenomena, the modulation of EMT towards an epithelial phenotype has become a desirable therapeutic avenue [6]. The development of therapy inducing MET would overcome some of these challenges. During past research efforts, some drugs have elicited the desired reversion of an EMT phenotype in cSCC cell lines and xenografts (Table 2). Successful induction of MET was mostly achieved by targeting four major signalling pathways responsible for the induction EMT in cSCC: EGFR signalling, the PI3K/ Akt/ mTOR pathway, TGFβ signalling and NFκB signalling (refer to Fig. 4 for details). The kinase, Akt, takes a central role here with several drugs, reducing its activity also favourably modulating EMT marker expression, reducing migratory and invasive properties. Direct pharmaceutical inhibition of Akt successfully induced apoptosis, reduced tumor growth in xenografts whilst inducing MET [108; 143]. Inhibition of Cyclooxygenase 2 (COX-2) or Ornithine decarboxylase (ODC) with diclofenac and difluoromethylornithine (DFMO), respectively, reduced p-Akt levels and in single agent and combination therapy [144]. Modulators of other upstream signalling of Akt like EGFR and PI3Ksignalling show promise as potential targets [122; 145–147]. Inhibition of the effectors of EGF and TGF-β induced signalling such as p38, JNK, AP1, MEK, and SMAD3 had a similar effect to inhibition of Akt [39; 85; 143; 148]. Tightly connected to p38 signalling is RelA, a component of the NFκB transcription factor. Induction of p-p38 lead to decreases in p-RelA and abrogation of migration, spheroid formation and stemness [106]. Conversely, stabilization of IκBα, a negative modulator of the NFκB pathway, increased cellular adhesion and lead to loss of migration, spheroid formation and stemness [105].
Most compounds tested targeted the afore mentioned four major signalling pathways of EMT signalling. However, there are a few exceptions involving hormonal signal transduction and cytoskeletal signalling. XTC790, a reverse agonist for the nuclear estrogen-related receptor α (ERRα), and rT3, a Dio2 inhibitor, elicited responses consistent with MET in A431 and SCC13 cell, respectively [39; 141]. Translation of the former finding to other cSCC models is questionable due to the unique origin of A431 cell line. Interestingly, inhibition of RhoA, a small GTPase involve in integrin and cell skeletal signalling, had the opposite effect of all previously discussed drugs promoting EMT [149]. Additionally, proof of concept studies using several natural compounds have report modulation EMT markers and properties in cSCC cell lines [110; 147; 150; 151]. However, these studies lack proper validation of their targets, an in-depth assessment of potential off-target activities, and consequently an understanding for the signalling pathways modulated. For example, Wogonoside modulates p-PI3K, p-β-catenin and p-Wnt levels leaving serious doubts about its selectivity especially given the lack of a specific target in the study [110]. Nevertheless, investigations into the suggested targets, EphB2 and HDAC3, using selective inhibitors might prove fruitful given their implications in other cancers [152; 153].
Table 2
Drugs and their effects on EMT markers and properties in cSCC.
Drug | Target | Markers and Signalling | Properties | Model | Ref. |
SB431542 | TGFBRI | p-p38, pERK1/2 | migration, invasion, proliferation | RDEB-cSCC | [85] |
SB203580 | p38 | Vim, FN1, MMP9, PAI-1 | migration, invasion | RDEB-cSCC | [85] |
Trametinib | MEK1/2 | Vim, FN1, MMP9 | migration, invasion | RDEB-cSCC | [85] |
PD169316 | SMAD3 | pSMAD3, PAI-1, MMP2, MMP9 | migration, invasion | RDEB-cSCC | [85] |
ARP100 | MMP2 | MMP2, PAI-1 | migration, invasion | RDEB-cSCC | [85] |
Avicularin | n.s. | E-cad, Ncad, MMP9, Vim, pMEK, p-p65 | apoptosis | SCC13 | [150] |
Aloeemodin, Kaempferitrin | EphB2 | E-cad, EphB2, MMP9, MMP2, Vim | proliferation, invasion, xenograft growth apoptosis | A431, SCL1 | [152] |
Wogonoside | n.s. | Ecad, NCad, FN1, VEGF, MMP9, MMP14, pPI3K, pWNT, pβcat, pAKT | viability, colony formation, stemness, proliferation, invasion, microtubule formation, xenograft growth, apoptosis | SCL1, SCC12 | [110] |
rT3 | Dio2 | E-Cad, N-Cad, Vim, Zeb1 | migration | SCC13 | [40] |
Ginsenoside® Rg3 | HDAC3 | E-Cad, N-Cad, Vim, Snail, HDAC3, c-Jun | migration, invasion | A431, SCC12 | [153] |
LY2109761 | TGFBRI/II | E-Cad, pSMAD2/3, Vim, FN1, Slug | migration, invasion | SCL-1 | [154] |
XTC790 | ERRα | E-Cad, p53, FN1, Vim, pSTAT3, pATR, pAMPKα | proliferation, migration, apoptosis | A431 | [39] |
Niclosamide | STAT3 | E-Cad, pSTAT3, FN1, Vim | n.s. | A431 | [39] |
Lapatinib | HER2/ EGFR | PTEN, pPTEN, ECad, pAKT, pmTOR, Wnt, βcatenin, N-cad, Vim, Slug | apoptosis, autophagy, viability | A431 | [145] |
LY294002 | PI3K | E-Cad, pAKT, Vim | migration, invasion | Astransformed HaCaT | [122] |
NC9 | TG2 | E-Cad, Twist, Snail, Slug, Vim, FN1, NCad, HIF1α | spheroid formation, migration, invasion | A431, SCC13 | [108] |
Akt inhibitor VIII | Akt | E-Cad, Vim, Slug, pAkt | adhesion, migration | PM1, MET1, MET4 | [54] |
Caffeic Acid | Fyn Kinase | E-Cad, N-Cad, Vim, Snail, pp38, pRelA | migration, stemness, spheroid formation | As-transformed HaCaT | [106] |
SB203580 | p38 | E-Cad, N-Cad, Vim, Snail, pp38, pRelA | migration, stemness, spheroid formation | As-transformed HaCaT | [106] |
BAY 11-7082 | p-IκBα | E-Cad, N-Cad, Vim, Snail | adhesion, stemness, spheroid formation, tumor formation | As-transformed HaCaT | [105] |
Y27632 | RhoA | E-Cad, Vim, NCad, Snail, Slug | n.s. | A5RT3 | [149] |
Diclofenac | COX2 | pAKT, pERK1/2, pMAPKAP2, Snail, Twist, MMP2, COX2 | tumor growth, migration, colony formation, apoptosis, | A431 (Xenograft) | [144] |
DMFO | OCD | pAKT, pERK1/2, pMAPKAP2, ODC | tumor growth, colony formation, apoptosis | A431 (Xenograft) | [144] |
Diclofenac and DMFO | COX2/ OCD | pAKT, pERK1/2, pMAPKAP2, Slug, Twist, MMP9, MMP2, COX2, ODC | tumor growth, migration, colony formation, apoptosis | A431 (Xenograft) | [144] |
API-59CJ-Ome | Akt | pAKT, Slug, NCad, FN1 | tumor growth | A431 | [144] |
Triciribine | p38 | pp38, pMAPKAP2, MMP2, MMP9, NCad, E-Cad | tumor growth, proliferation, apoptosis | CsAtreated A431 (Xenograft) | [143] |
SB203580 | Akt | pAKT, pmTOR, pp38, pMAPKAP2, MMP2, MMP9, NCad, E-Cad | tumor growth, proliferation, apoptosis | CsAtreated A431 (Xenograft) | [143] |
Triciribine/ SB203580 | p38/ Akt | pAKT, pmTOR, pp38, pMAPKAP2, MMP2, MMP9, NCad, E-Cad | tumor growth, proliferation, apoptosis | CsAtreated A431 (Xenograft) | [143] |
luteolin | n.s. | FN1, Vim, Twist, Snail, NCad, MMP9, pAkt, pGSK3β, ECad | migration, invasion | A431 | [151] |
Quercetin | n.s. | FN1, Vim, Twist, Snail, NCad, MMP9, pAkt, pGSK3β, ECad | migration, invasion | A431 | [151] |
Wortmannin | PI3K | pAkt, Vim, E-Cad | n.s. | A431 | [146] |
GSP | n.s. | EGFR, pERK1/2, NCad, FN1, Vim, ECad | migration, invasion | SCC13 | [147] |
Erlotinib | EGFR | NCad, FN1, Vim, ECad | invasion | SCC13 | [147] |
UO126 | MEK | pp38, p-ERK, Vim, ECad | n.s. | Transformed HaCaT (II3 and H375) | [148] |
SP600125 | JNK | Vim, ECad | n.s. | Transformed HaCaT (II3 and H375) | [148] |
[6]Gingerol | AP1 | Vim, ECad | n.s. | Transformed HaCaT (II3 and H375) | [148] |
italics = attenuated, bold = induced, n.s. = not specified, GSP = grape seed proanthocyanidins. |