Central role of Snail1 in the regulation of EMT and resistance in cancer: a target for therapeutic intervention

The Notch intracellular domain, LOXL2, NF-κB, HIF-1α, IKKα, SMAD, HMGA2, Egr-1, PARP-1, STAT3, MTA3, and Gli1 all interact directly with the Snail1 promoter to regulate Snail1 at the transcriptional level [17][29]. Hypoxic stress, caused by insufficient oxygen, prompts a transcriptional response mediated by hypoxia-inducible factors (HIFs) [17]. Notch increases HIF-1α recruitment to the LOX promoter, and LOXL2 oxidizes K98 and/or K127 on the Snail1 promoter, leading to a conformational change in shape [18]. Under hypoxic conditions, HIF-1α binds to HRE2, contained within -750 to -643 bp of the Snail1 promoter, and increases Snail1 transcription. Knockdown of HIF-1α results in the repression of both Snail1 and EMT [19]. NF-κB also binds to the Snail1 promoter, between -194 and -78 bp, and increases its transcription [20]. SMAD2 and IKKα bind concurrently to the Snail1 promoter between -631 and -506 bp, resulting in Snail1’s upregulation [21]. HMGA2 cooperates in this complex as well, as the binding of HMGA2 to the Snail1 promoter increases SMAD binding [22].

In addition, ILK promotes PARP-1 binding, and STAT3 binds as a final result of an IL-6/JAK/STAT pathway [23],[24]. In mice, a pathway beginning with HB-EGF and progressing through the MEK/ERK pathway has also induced STAT3 binding to the Snail1 promoter [25]. Gli1 and Snail1 interact through a positive feedback loop: Shh and Wnt crosstalk results in the upregulation of both [26]. MTA3, a subunit of the Mi-2/NuRD complex, transcriptionally represses Snail1 in an ER-dependent manner. Snail1, in turn, binds to the ER promoter to complete the negative feedback loop [27],[28]. In a similar fashion, Egr-1 and Snail1 relate via a negative feedback loop. Egr-1, another zinc-finger transcription factor, binds to the Snail1 promoter at four sites between -450 and -50 bp. This process necessitates the presence of HGF and is mediated by the MAPK pathway, and it ultimately results in Snail1 upregulation. Snail1, in turn, represses Egr-1 [29].

YY1 and Snail1 itself are two special instances of transcriptional Snail1 regulation. YY1 binds to the3’ enhancer, rather than the promoter, and knockdown of YY1 has been shown to decrease Snail1 expression [30]. Furthermore, Snail1 is capable of binding to its own promoter and upregulating itself [31]. Snail1 binds to the E box region within the Snail ILK Responsive Element (SIRE); PARP-1 also binds to the SIRE, which is located between -134 and -69 bp, when induced by ILK [23] (Figure 2).

Experiments conducted to elucidate the relationship between p53, a tumor suppressor protein, and Snail1 have shown that p53 acts via miR-34a, -34b, and -34c to repress Snail1 at a 3’ untranslated region (UTR). Consequently, when p53 is repressed, the repression of Snail1 is lifted, and the expression of Snail1 rises [32].

Two instances of phosphorylation are crucial to Snail1's post-transcriptional regulation. GSK-3β phosphorylates Snail1 at two consensus motifs in serine-rich regions. The first phosphorylation, at motif 2 (S¹⁰⁷, S¹¹¹, S¹¹⁵, S¹¹⁹), results in Snail1's being exported to the cytoplasm. The second instance of phosphorylation (S⁹⁶, S¹⁰⁰, S¹⁰⁴) leads to its ubiquitination by β-Trcp, which recognizes the destruction motif D⁹⁵SGxxS¹⁰⁰ and ubiquitinates Lys98, 137, and 146. Consequential proteasomal degradation follows [33],[34]. In conditions that prevent GSK-β from phosphorylating Snail1, the F-box E3 ubiquitin ligase FBXL14 appears to cause proteasomal degradation by ubiquitinating the same lysine residues as β-Trcp [35]. P21-activated kinase 1 (PAK1) also phosphorylates Snail1 at S²⁴⁶[36]. Phosphorylation determines Snail1’s subcellular location, as GSK-3β -mediated phosphorylation induces Snail1's export to the cytoplasm through exportins such as chromosome region maintenance 1 (CRM1) [33],[37]. By contrast, PAK1 phosphorylation promotes Snail1’s presence in the nucleus and, therefore, increases its activity [36]. In the cytoplasm, Snail1 is quickly degraded; it has a half-life of only twenty-five minutes [33]. To protect from this degradation, Snail1 has nuclear localization signals (NLS): one monopartite from amino acids 151-152 and one bipartite overlapping the SNAG domain between amino acids 8 and 16 [38]. These signals are responsible for the nuclear transport of Snail1, which in turn is required for proper expression. β-catenin, Lef-1, and IκB employ similar systems [38] (Figure 3, Table 1).

TNFα, NF-κB, FGF, Wnt, and microRNA signals also influence the regulation of GSK-3β-mediated phosphorylation. The TNFα/NF-κB pathway induces CSN2, which protects Snail1 from degradation by interfering with the binding of GSK-3β and β-Trcp. Thus, Snail1 is neither phosphorylated nor ubiquitylated [39]. FGF operates through the PI3K/Akt pathway to downregulate GSK-3β, and receptor tyrosine kinase induces EGF suppression of GSK-3β [34],[40]. Wnt can also suppress GSK-3β and, thus, the phosphorylation of Snail1 [41]. Additionally, miR-148a causes the phosphorylation of AKT and GSK-3β, which results in less Snail1 localized in the nucleus. This, in turn, inhibited EMT in hepatocellular carcinoma [42].

Phosphorylation of upstream targets also influences the regulation of Snail1. For example, RANKL, in association with IκB, activates the NF-κB p65 subunit, and Akt influences the nuclear localization of NF-κB through its phosphorylation of IKKα and IκB in turn [43],[44]. TGF-β1 induces the phosphorylation of SMAD2 and SMAD3, which is necessary for their binding to Snail1 and the consequential upregulation of Snail1’s activities [45]. However, the cooperation of Ras signals is required for this pathway, since TGF-β1-mediated induction of Snail1 ceases with the silencing of Ras [46].

Other mechanisms of regulation contribute to the expression levels of Snail1, too. The small C-terminal domain phosphatase (SCP) induces dephosphorylation of both GSK-3β and the affected Snail1 motifs, thereby stabilizing Snail1 [47]. Additionally, histone deacetylase inhibitors promote the acetylation, likely of lysines, and increase Snail 1’s nuclear localization by inhibiting ubiquitination [48].

E-cadherin is a transmembrane glycoprotein responsible for calcium-dependent cell-to-cell adhesion [49]. E-cadherin is a type I cadherin encoded by the gene CDH1, which is located on human chromosome 16q22.1 [50]. The founding member of the cadherin superfamily, E-cadherin plays a pivotal role in cadherin-catenin-cytoskeleton complexes, and it grants anti-invasive and anti-migratory properties to epithelial cells [51]. E-cadherin expression naturally decreases during gastrulation in order to properly form the mesoderm, and its expression increases once more for kidney organogenesis [52],[53]. The CDH1 promoter contains multiple E-boxes, and Snail1, Slug, ZEB1, ZEB2, and Twist, among others, have been shown to directly repress E-cadherin [54]. Total E-cadherin knockout in mice resulted in immediate death at implantation [55]. Decreases in E-cadherin expression correlate with epithelial-mesenchymal transition, metastasis, and lower patient survival rates [10].

Four Snail1 complexes have been identified as mechanisms of E-cadherin repression. (1) Snail1 interacts with G9a, which concurrently recruits DNA methyltransferases (DNMTs) to the E-cadherin promoter. Snail1's zinc fingers are thought to interact with the G9a ankyrin repeats, SET domain, or both. The complex has been shown to increase H3K9me2 and decrease H3K9 acetylation [56]. (2) The Snail1-Ajuba-PRMT5 complex promotes the methylation of H4R3. This, too, operates at the E-cadherin promoter [57]. The demethylation of H3K4 by Co-REST, CtBP, and HDAC complexes also factors into the last two mechanisms [58]. (3) Snail1 works in conjunction with Sin3A and HDAC1/2 to deacetylate H3 and H4, which suppress E-cadherin [59]. (4) In perhaps the most elucidated case, the Snail1 SNAG domain interacts with the LSD1 AO domain to form a Snail1-LSD1-CoREST complex. Snail1 residues Pro2, Arg3, Ser4, Phe5, Arg8, and Lys9 have been shown to be particularly crucial to this union, since mutants could not interact with LSD1. Likewise, LSD1 requires functional Asp375 and Glu379, Glu553, Glu555 and Glu556 to cooperate with Snail1. LSD1 inhibitors, histone H3, and SNAG peptides also hamper the activity of the complex. The formation of the Snail1-LSD1-CoREST complex results in the demethylation of H3K4me2 and consequential suppression of E-cadherin, while also increasing the stability of each of the components of the complex [60]. In a proposed second step to this mechanism, Snail1 recruits Suv39H1 to the E-cadherin promoter. Similar to prior cases, the Snail1 SNAG domain interacts with the Suv39H1 SET domain to suppress E-cadherin. Knockdown of Suv39H1 restored E-cadherin expression by inhibiting H3K9me3 [61].

Raf kinase inhibitor protein (RKIP), a member of the phosphatidylethanolamine-binding protein (PEBP) group, suppresses metastasis by inhibiting the Raf-MEK-ERK and NF-κB pathways [62] [65]. In prostate, breast, and colorectal cancers, among others, RKIP expression is downregulated [64],[66]. Furthermore, elevated RKIP expression is a positive prognostic indicator for survival [66],[67]. Expression levels of RKIP correlate with those of E-cadherin, another Snail1 target, as they are both repressed by means of the E-boxes in their promoters [68].

Phosphatase and tensin homolog deleted in chromosome 10 (PTEN) dephosphorylates phosphoinositide-3,4,5-triphosphate (PIP3) and, thus, inhibits the PI3K pathway [69]. In this way, PTEN functions as a tumor suppressor. Snail1 binds to the PTEN promoter, which contains two E-boxes, and represses PTEN [70]. The specificity of this interaction is emphasized by the fact that neither Slug nor ZEB1 expression significantly alters PTEN levels [70]. Snail1's association with the PTEN promoter inhibits the binding of p53, which activates PTEN during apoptosis, and it consequently increases resistance to gamma radiation-induced apoptosis [70],[71]. A positive feedback loop has been established around this interaction as well, since the repression of PTEN increases the expression of Akt [72]. Akt, operating through NF-κB, increases the expression of Snail1 [44]. Through this pathway, Snail1 may contribute to raising its own expression levels [70].

Occludin, an integral membrane protein crucial to the integrity of tight junctions, was first identified in 1993. The transmembrane protein has four hydrophobic domains within its 522 amino acid sequence and a molecular weight of 65 kDa [73],[74]. Though it is considered similar to connexins in gap junctions, occludin is found exclusively at tight junctions in epithelial and endothelial cells [73]. Snail1 functions as a transcriptional repressor of occludin, just as it does E-cadherin in adherens junctions. By binding to the E-box in the occludin promoter sequence, Snail1 can completely repress the promoter activity [75]. Immunoblot analysis and immunocytochemistry confirm the considerable reduction of occludin expression in the presence of Snail1 [13]. This repression, along with that of E-cadherin and claudins, is critical to the loss of cell-to-cell adhesion observed in EMT.

The claudin family contains more than twenty members, all of which are integral proteins spanning the membrane four times. Family members range from 20-27 kDa, but they all share PDZ binding motifs, which allow them to interact with ZO-1, ZO-2, and MUPP-1, among others [76]. Claudins are components of tight junctions, and claudin-1 binds with occludin [76],[77]. The expression of claudins is frequently low or nonexistent in breast cancer cell lines, and it shares an inverse relationship with Snail1 expression levels in invasive breast tumors [77].

Specifically, claudin-1, -3, -4, and -7 are all susceptible to repression by Snail1. The promoter sequence of each of these proteins contains multiple E-box binding motifs: claudin-1 has two E-boxes, claudin-3 has six, claudin-4 has 8, and claudin-7 has eight. As such, Snail1 can completely inhibit their transcription [75]. The destruction of tight junctions that accompanies the repression of claudins and occludin leads to epithelial cells' loss of apical polarity and increases proliferation [78]. This mechanism helps drive Snail1-induced EMT.

Mucin-1, a transmembrane glycoprotein encoded by MUC1, is an epithelial marker expressed at the apical surface of epithelial cells in the reproductive tract, digestive tract, lungs, kidney, liver, eyes, and other tissues [79] [81]. Additionally, it is expressed in hematopoietic and T cells [80]. Mucin-1's functions include lubrication and protection from pathogens, and its association with β-catenin has implicated Mucin-1 in cell signaling [80]. O-linked glycosylation affects the protein significantly, as the core protein ranges from 120-225 kDa and the glycosylated form can reach up to 500 kDa [82]. In epithelial tumors, Mucin-1 is upregulated, and disparities in splice variants and glycosylation become apparent [79],[80]. Splice variants differ greatly-the protein can vary from 4-7 kb [82]. Perhaps most importantly, Mucin-1 also loses its apical restriction in malignant cases [80].

The 2872 bp promoter facilitates much of Mucin-1's regulation, and it notably includes five sites for YY1 binding [79]. Snail1 interacts with the two E-boxes that begin -84 bp from the start of transcription. Like E-cadherin, Mucin-1 is an epithelial marker repressed by Snail1 during the induction of EMT [83].

ZEB-1, like Snail1, is a zinc-finger transcription factor that assists in the induction of EMT. Using E-boxes and co-repressors such as CtBP and BRG1, ZEB-1 represses E-cadherin and Mucin-1 [83],[84]. However, ZEB-1 is at least ten times less potent a repressor of both E-cadherin and Mucin-1 than Snail1 [83]. Interference with the interaction between ZEB-1 and BRG1 results in the upregulation of E-cadherin and simultaneous downregulation of vimentin, so an abundance of functional ZEB-1 is associated with a mesenchymal phenotype [84]. In contrast to the lethal effects of Snail1 knockout, ZEB-1 knockout does not prevent development to term and, thus, is not as critical for gastrulation [83].

The presence of Snail1 increases both RNA and protein levels of ZEB-1 during EMT. Snail1 expression in MDCK clones causes a 2.5-fold increase in ZEB-1 promoter activity compared to control cells. The abilities of Snail1 and ZEB-1 to repress E-cadherin are additive, and the two transcription factors work together to achieve a complete EMT [83].

Vimentin is 57 kDa intermediate filament generally restricted to mesenchymal cells [85]. Vimentin regulation is a complex interplay of epigenetic and post-translational modifications in addition to transcriptional regulation. Of note, the human vimentin promoter contains an NF-κB binding site as well as a TGF-β1 response element [86],[87]. Akt1 protects vimentin from caspase proteolysis via phosphorylation of Ser39 [88]. During EMT, epithelial cells, which normally express keratin intermediate filaments, begin to express vimentin. Overexpression of vimentin is evident in breast and prostate cancers, among many other types, and overexpression generally correlates with invasiveness, migration, and poor prognosis [89] [91]. Snail1 upregulates vimentin during EMT [54].

Fibronectin is a glycoprotein involved in cell adhesion, differentiation, and migration [92],[93]. A dimer with two 250 kDa components, fibronectin is greatly affected by splicing, and at least twenty variants of the human form have been identified [94]. Fibronectin interacts with many integrins in addition to heparin, collagen, and fibrin [95] [99]. Inactivation of fibronectin is lethal in mice [100]. Snail1 upregulates fibronectin, a mesenchymal marker indicative of EMT [54].

Cytokeratins exist in two types, and each cytokeratin works with a complementary partner to form keratin filaments [101]. Cytokeratin 18 is the first type, acidic, and interacts with the basic cytokeratin 8 [101]. The cytokeratin 18 protein is encoded by the CK18 gene, which is located on chromosome 12q13. Cytokeratin 18 is an intermediate filament protein involved in cell structure, cell signaling, and the cell cycle [101] [104]. Cytokeratin 18 serves as an epithelial marker, and it localizes in epithelial organs, such as the kidney, liver, gastrointestinal tract, and mammary glands [105]. Snail1 represses cytokeratin 18 during the induction of EMT [83]. Unlike other targets, though, cytokeratin 18 expression is not completely subdued by Snail1's presence [75].

Matrix metalloproteinases (MMP) cleave extracellular matrix substrates and, thereby, alter cell-matrix adhesions [106]. MMP-2 and -9 are a subcategory within the MMP group because they specifically act on gelatin, collagen, elastin, and fibronectin [107] [111]. The genes that encode MMP-2 and -9 both contain fibronectin type II domains and are consequently three exons longer than the other MMP genes [107]. MMP-2 is a 72 kDa protein while MMP-9 is 92 kDa, and the main difference between them is the MMP-9's 54 amino acid hinge region [107],[112]. Additionally, MMP-2 localizes in the nucleus and MMP-9 in the cytoplasm [113]. Overexpression of MMP-2 and MMP-9 is frequently associated with invasive, metastatic tumors [114] [117].

Snail1's presence increases the mRNA levels of both MMP-2 and -9 [118]. One suggested interaction includes the upregulation of MMP-2 and -9 by Snail1 to trigger EMT and, then, the coordinated effort of Snail1 and Slug to sustain EMT by continually stimulating MMP-9 [113].

Lymphoid enhancer-binding factor 1 (LEF-1) is a T-cell factor commonly detected in tumors [119],[120]. The transcription factor represses E-cadherin by forming complexes with β-catenin, which, like Snail1, is degraded as a result of GSK-3β-mediated phosphorylation [11],[121] [123]. LEF-1 interacts with Snail1 via Wnt, PI3K and TGF-β1 pathways, and both Snail1 and LEF-1 are necessary for a complete EMT [124]. LEF-1 is considered a mesenchymal marker, and Snail1 induces its expression and continues to upregulate it [82],[125].

Invasive breast carcinomas, including infiltrating ductal (IDC) and infiltrating lobular carcinomas (ILC), spread to surrounding breast tissues, lymph nodes and the pleural cavity. Assigned histological grades, with three being the highest, correlate with prognosis [126]. Breast carcinomas can give rise to malignant pleural effusions, and typical survival rates at that point are a matter of months [127].

Snail1 is not present in normal breast epithelium, nor is it present in ILCs (n = 21). Of 17 patients, Snail1 was expressed in 47% of IDCs, and its expression correlated with lymph node metastases and high histologic grades [128]. E-cadherin and Snail1 expression levels are inversely related, and high expression levels of Snail1 correlate with shorter effusion-free, disease-free, and overall survival rates (n = 16) [129]. As such, Snail1 has prognostic significance as a marker of IDC malignancy [128].

Snail1 mRNA and protein levels are inversely correlated with E-cadherin in hepatocellular carcinoma (HCC) [130]. Snail1 overexpression, which in one study included 23% of cases (n = 47), is associated with portal vein invasion, metastasis, and poor differentiation. Furthermore, Snail1 expression correlates with a poor prognosis in recurrence-free survival and, thus, is considered a potential risk factor for early recurrence [131],[132].

Overall, Snail1 expression is lower in ovarian carcinoma than in breast carcinoma, though its expression is still associated with distant metastases [129]. Expression is higher among primary tumors and metastases than effusions, and effusions show complete cytoplasmic localization of Snail1 [133]. Snail1 represses E-cadherin and upregulates MMPs, and E-cadherin expression correlates with disease-free survival while MMP-2 is considered a marker of poor prognosis [129].

E-cadherin expression is drastically reduced in gastric carcinoma, and Snail1 expression levels once again share an inverse relationship with E-cadherin expression levels [129]. Snail1 expression levels are more comparable to breast than ovarian carcinomas, and Snail1 expression is still higher in diffuse rather than intestinal varieties of gastric carcinomas [129],[134]. Elevated Snail1 expression increases cells' capacities for migration and invasion. Overexpression correlates with tumor size, depth of invasion, and lymph node metastasis. Shortened survival rates are also directly related to Snail1 overexpression, and Snail1 is considered a predictor of poor prognosis [135].

Oral squamous carcinoma is another case of E-cadherin/Snail1 expression inversion, and the higher the Snail1 expression, the more invasive the cancer. E-cadherin positive cells maintain their cuboidal shape while E-cadherin negative cells turn spindle-shaped. This is a typical sign of EMT, and it shows Snail1's repression of E-cadherin [136].

Pancreatic carcinoma tissues show significantly reduced E-cadherin levels and relatively high Snail1 expression [129]. In one study, 78% (n = 36) of ductal adenocarcinoma tissues expressed Snail1, and Snail1 expression is higher in undifferentiated cell lines than in differentiated ones [137].

Colorectal cancer (CRC) begins in gland cells that line the colon and rectum, and it is one of the most commonly newly diagnosed cancers and a leading cause of cancer-related deaths [138]. Snail1 expression is again inversely correlated to E-cadherin expression in CRC, and the expression level of Snail1 is quite high in CRC (78%, n = 59) [130],[139]. Interestingly, the mean age of the Snail1-positive group was nine years older than the Snail1-negative group in one study, with a standard deviation of 12.7 years (58.9 years vs. 49.8 years, n = 59) [139]. In another study, Snail1 expression was detected by Western blot in all tested CRC lines, and its expression increased both migratory and invasive properties. Additionally, Snail1 expression led to a stem-cell like phenotype and spindle shape, as usually accompanies the loss of E-cadherin [140]. Snail1 expression also increased with the stage of the tumors, with 15/23 stage III expressing Snail1 and 6/6 of stage IV. The significantly higher rate of metastasis associated with Snail1 expression suggests that Snail1's presence indicates a high risk of distant metastases [139],[140].

Though the expression level of Snail1 is lower in bladder carcinoma than in other types of cancer, its presence still has a significant impact on the cancer's progression. In one study, only 16% of the 120 tested tissues expressed Snail1, indicating that Slug and Twist, whose expression levels were 63% and 44% respectively, play larger roles. However, Snail1 expression increased in node-positive compared to node-negative tumors, and Snail1's presence lowered the three-year progression free survival rate to only 15% [141]. Since Snail1 expression is closely linked with tumor recurrence, its elevation is considered a significant prognostic factor [141],[142].

In melanoma, there is increased Snail1 mRNA and low E-cadherin in the presence of Snail1 expression. By contrast, no Snail1 mRNA was detected in primary melanocytes [143]. Snail1 expression confers both invasive and immunosuppressive properties in melanoma [144].

Saito et al. reported that Snail1 mRNA was found in all cases tested of synovial sarcoma (n = 20) and E-cadherin mRNA was detected by RT-PCR in 14/20 cases. This does not show the same strong inverse correlation that has come to be expected of Snail1 and E-cadherin. In this case, mutations of the CDH1 gene, which encodes E-cadherin, seem to be more influential than the presence of Snail1 [145].

Prostate cancer is the second most commonly diagnosed cancer in men worldwide, with estimates of over 900,000 new cases per year [146]. A Gleason grade, which describes the two most important histopathological patterns of that patient's cancer, accompanies a diagnosis. The grade ranges from 2-10 with a higher score meaning less differentiated [147]. Significant losses of E-cadherin and syndecan 1, two proteins involved in cellular adhesion, have been observed in malignant prostate cancer [148],[149]. Both promoters contain E-boxes, so Snail1 can directly bind and repress them [150],[151]. The presence of E-boxes may explain the inverse correlation between E-cadherin/syndecan 1 and Snail1 expression levels. Poblete et al. found that high Snail1 expression correlated with a high Gleason grade and increased malignancy. Furthermore, in more malignant cell lines, like PC3, Snail1 had exclusively nuclear localization. By contrast, Snail1 had both cytoplasmic and nuclear localization in less malignant cell lines [152].

Cervical cancer is one of the most common malignancies in women worldwide [138]. Chen et al. found Snail1 expressed in 94% of samples (n = 70), and the elevated expression of Snail1 correlated with late FIGO stage, lymph node metastasis, and poor differentiation [153].