Wnt/CTNNB1 Signal Transduction Pathway Inhibits the Expression of ZFP36 in Squamous Cell Carcinoma, by Inducing Transcriptional Repressors SNAI1, SLUG and TWIST

The Wnt/CTNNB1 pathway is often deregulated in epithelial tumors. The ZFP36 gene, encoding the mRNA binding protein Tristetraprolin (TTP), is downregulated in several cancers, where it has been described to behave as a tumor suppressor. By this report, we show that Wnt/CTNNB1 pathway is constitutively activated, and ZFP36 expression is downregulated in Squamous Cell Carcinoma (SCC) cell lines compared to normal keratinocytes. Moreover, we suggest that the decrease of ZFP36 expression might depend on the activity of transcriptional repressors SNAI1, SLUG and TWIST, whose expression is induced by Wnt/CTNNB1, highlighting a potential regulatory mechanism underlying ZFP36 downregulation in epithelial cancers.


Introduction
ZFP36 gene encodes the Zinc-finger RNA-binding protein Tristetraprolin (TTP). TTP expression is induced by different stimuli, such as mitogenic agents [1] and inflammation [2,3]. TTP localizes in the nucleus, but stimulation leads to its translocation in the cytosol where it recognizes and binds to specific AU motifs at the 3 UTR (AREs) of its target mRNAs, thereby triggering their degradation [4,5]. Owing to this activity, TTP is involved in the anti-inflammatory response, by targeting and inducing degradation of several pro-inflammatory cytokines mRNAs [6], thereby preventing chronical inflammation [7]. With respect to TTP mediated RNA degradation activity, this protein works most efficiently at a specific concentration, and it is able to self-regulate its own mRNA's stability in order to remain at optimal levels [8]. TTP's role in cancer is still controversial. It is recognized as a protein that inhibits tumor progression although, for instance, it also acts on CD8+ lymphocytes by blocking cell-mediated antitumor immune response [9]. ZFP36 gene is frequently downregulated in different cancers [9], where, although specific mutations affecting it are not described, it seems to act as a tumor Figure 1. Immunofluorescence highlighting CTNNB1 localization and immunoblotting of ZFP36 expression. Panel A shows confocal microscope images of healthy keratinocytes (panel A1) and SCC12, SCC13 and SCC15 (Panels A2, A3 and A4 respectively). Scale bars 30 µm, 50 µm and 100 µm. Panel B shows the western blot analysis of TTP expression in the three SCC cell lines compared to normal keratinocytes. The plot below indicates the normalized densitometric analysis of immunoblot. Figure 2 depicts a western blot analysis of the transcriptional repressors SLUG, SNAI1 and TWIST ( Figure 2, panel A, B and C, respectively) in three SCC cell lines compared to normal keratinocytes. The three repressors are expressed at different levels in the three cell lines. SLUG protein is upregulated predominantly in SCC13 and to a lesser extent in SCC15 and SCC12 cells. SNAI1 protein is induced especially in SCC13 and SCC15 cells and is almost completely absent in SCC12. TWIST protein is only upregulated in SCC15 cells, compared to normal keratinocytes.   (Figure 2A, B and C, respectively) in three SCC cell lines compared to normal keratinocytes. The three repressors are expressed at different levels in the three cell lines. SLUG protein is upregulated predominantly in SCC13 and to a lesser extent in SCC15 and SCC12 cells. SNAI1 protein is induced especially in SCC13 and SCC15 cells and is almost completely absent in SCC12. TWIST protein is only upregulated in SCC15 cells, compared to normal keratinocytes. Overall, at least one of the three repressors is induced in every SCC cell line compared to healthy keratinocytes.

Treatment with the Wnt/CTNNB1 Inhibitor FH535 Induces ZFP36 Up-Regulation and the Downregulation of Transcriptional Repressors SNAI1, SLUG and TWIST
Western blot analysis was performed 48 and 72 h upon treatment with FH535 in the cell lines SCC12, SCC13 and SCC15. Following FH535 treatment, the expression of CTNNB1 tends to decrease, while ZFP36 is upregulated in the three cell lines ( Figure 3, panels A, B and C). Figure 4, panels A, B and C show a western blot analyzing the expression of repressors SNAI1, SLUG and TWIST in SCC12, SCC13 and SCC15 cell lines (panels A, B, C, respectively). SCC12 cells express neither SNAI1 nor TWIST. They only show SLUG expression, which is downregulated following treatment with the FH535 inhibitor as represented in Figure 4A. SCC13 cells show SNAI1 and SLUG expression. These two repressors are both down-regulated following treatment with FH535 ( Figure 4B). SCC13 do not show TWIST expression ( Figure 4B). Figure 4C shows the expression of the three transcriptional repressors in the SCC15 cell line. As in the previous tumor lines, SLUG and SNAI1 are downregulated by the treatment with FH535. Moreover, in this cell line TWIST follows the same path and it shows complete downregulation at 72 h after treatment.   Figure 4A-C show a western blot analyzing the expression of repressors SNAI1, SLUG and TWIST in SCC12, SCC13 and SCC15 cell lines (panels A, B, C, respectively). SCC12 cells express neither SNAI1 nor TWIST. They only show SLUG expression, which is downregulated following treatment with the FH535 inhibitor as represented in Figure 4A. SCC13 cells show SNAI1 and SLUG expression. These two repressors are both down-regulated following treatment with FH535 ( Figure 4B). SCC13 do not show TWIST expression ( Figure 4B). Figure 4C shows the expression of the three transcriptional repressors in the SCC15 cell line. As in the previous tumor lines, SLUG and SNAI1 are downregulated by the treatment with FH535. Moreover, in this cell line TWIST follows the same path and it shows complete downregulation at 72 h after treatment.

SNAI1 and SLUG Proteins Bind the ZFP36 Promoter Region Both in Healthy Keratinocytes and in SCC15 Cell Line
To characterize the ZFP36 promoter, and to verify a possible regulation of its expression by the above mentioned transcriptional repressors, we performed an in silico analysis of 2994 base-pairs' (bp) genomic region upstream of the human ZFP36 gene. To this purpose E-box sequences (5′-CANNTG-3′), known binding sites for SNAI1, SLUG and TWIST proteins, were searched using MatInspector software (Genomatix). We discovered a nearly canonical E-Box element, composed by two distinct and adjacent 5′-CANNTG-3′ sequences, located 2416 bp upstream from the most common annotated transcriptional start site (TSS1) of ZFP36. Moreover, a TATA-box regulatory sequences was annotated 24 nt upstream from the TSS1. A second, alternative TSS (TSS2) has been mapped 16 nt downstream from TSS1 ( Figure 5 panel A).

SNAI1 and SLUG Proteins Bind the ZFP36 Promoter Region Both in Healthy Keratinocytes and in SCC15 Cell Line
To characterize the ZFP36 promoter, and to verify a possible regulation of its expression by the above mentioned transcriptional repressors, we performed an in silico analysis of 2994 base-pairs' (bp) genomic region upstream of the human ZFP36 gene. To this purpose E-box sequences (5 -CANNTG-3 ), known binding sites for SNAI1, SLUG and TWIST proteins, were searched using MatInspector software (Genomatix). We discovered a nearly canonical E-Box element, composed by two distinct and adjacent 5 -CANNTG-3 sequences, located 2416 bp upstream from the most common annotated transcriptional start site (TSS1) of ZFP36. Moreover, a TATA-box regulatory sequences was annotated 24 nt upstream from the TSS1. A second, alternative TSS (TSS2) has been mapped 16 nt downstream from TSS1 ( Figure 5A). mapped 16 nt downstream from TSS1 ( Figure 5 panel A).
To test whether this E-Box could indeed bind to these transcriptional repressors, we performed chromatin immunoprecipitation assay in healthy keratinocytes and SCC15, unique among the previously exploited squamous cancer cell lines that simultaneously expresses all the three mentioned repressors (as shown in Figure 2). To date, both SNAI1 and SLUG proteins were found to bind significatively to the E-Box within the ZFP36 promoter when compared to a negative control region located downstream to the ZFP36 gene both in human keratinocytes and in SCC15 cells ( Figure  5 panel B-C). In addition, the binding of SLUG to the ZFP36 promoter in SCC15 cells is significantly higher than in healthy keratinocytes although no obvious alteration in its expression was observed ( Figure 2). No significant differences were instead observed between the binding capability of SNAI1 to ZFP36 promoter in healthy keratinocytes and SCC15 cell line ( Figure 5 panel C).
Together, these data hence demonstrate that SNAI1 and SLUG repressors actually bind to ZFP36 proximal putative promoter region and hence might be involved in its transcriptional repression. To test whether this E-Box could indeed bind to these transcriptional repressors, we performed chromatin immunoprecipitation assay in healthy keratinocytes and SCC15, unique among the previously exploited squamous cancer cell lines that simultaneously expresses all the three mentioned repressors (as shown in Figure 2). To date, both SNAI1 and SLUG proteins were found to bind significatively to the E-Box within the ZFP36 promoter when compared to a negative control region located downstream to the ZFP36 gene both in human keratinocytes and in SCC15 cells ( Figure 5B,C). In addition, the binding of SLUG to the ZFP36 promoter in SCC15 cells is significantly higher than in healthy keratinocytes although no obvious alteration in its expression was observed ( Figure 2). No significant differences were instead observed between the binding capability of SNAI1 to ZFP36 promoter in healthy keratinocytes and SCC15 cell line ( Figure 5C).
Together, these data hence demonstrate that SNAI1 and SLUG repressors actually bind to ZFP36 proximal putative promoter region and hence might be involved in its transcriptional repression.

Discussion
Although ZFP36 expression is lost in several malignancies, the mechanisms underlying this event are still poorly understood. Specific mutations in the promoter region of ZFP36 are not been described so far. Epigenetic causes have been proposed [24]. However, we hypothesize that, since, in general, loss of ZFP36 fosters tumor progression, in different cell contexts ZFP36 down-regulation might result from the modulation of pathways that are crucial to a specific tumor. To this regard, we observed in a previous work that ZFP36 expression in colorectal carcinoma cell lines is inversely correlated to Wnt/CTNNB1 pathway activity [6]. Recently, it has also been suggested that miRNA-423 is capable of promoting tumor cell proliferation and migration by enhancing Wnt/CTNNB1 activity, thereby inducing ZFP36 down-regulation [31]. We also analyzed the activity toward ZFP36 promoter of specific microRNAs induced by Wnt/CTNNB1, among others (microRNA29b1/a), but never get conclusive results (data not shown). Moreover, we were able to exclude the possibility that TCF7L2/CTNNB1 transcriptional complex might directly bind ZFP36 promoter, thereby repressing its expression (data not shown). The correlation between Wnt/CTNNB1 and ZFP36 is noteworthy and the molecular process that links the signal transduction pathway to ZFP36 modulation deserves to be clarified. The aim of this work is to demonstrate that the correlation existing between ZFP36 and the Wnt/CTNNB1 signal transduction pathway is mediated by three transcriptional repressors (SNAI1, SLUG and TWIST) that are induced by CTNNB1 pathway and that modulate ZFP36 by binding to its promoter.
The protein product of ZFP36 (TTP) is known to directly target the mRNAs encoding SNAI1, SLUG and TWIST [6]. When Wnt/CTNNB1 signaling is deregulated, the sustained expression of SNAI1, SLUG and TWIST is involved in the epithelial to mesenchymal transition. Since it is known that ZFP36 is downregulated in epithelial cancer, that its down-regulation stabilizes the expression of a set of genes involved in EMT and that restoration of its expression contributes to the rescue of a more epithelial phenotype [6], the existence of this functional inverse correlation between TTP and the repressors made it plausible to hypothesize that the repressors might be the link allowing Wnt/CTNNB1 signaling to negatively regulate ZFP36 gene.
We elected to perform this study on a model of epithelial cancer and chose SCC since, to date, there were no data on this disease and ZFP36. As we observed by immunofluorescence, Wnt/CTNNB1 is constitutively active in the cell lines SCC12, SCC13 and SCC15. Moreover, we observed that ZFP36 is down-regulated in these cells compared to healthy keratinocytes, as opposed to transcriptional repressors SNAI1, SLUG and TWIST, that are upregulated in these SCC lines as a consequence of Wnt/CTNNB1 constitutive activity. We observed that inhibition of Wnt/CTNNB1 via the specific molecule FH535 determines a decrease of SNAI1, SLUG and TWIST expression and the rescue of ZFP36 expression. Western blots throughout the paper show the presence of multiple differently migrating bands for TTP. In our opinion these bands are specific and their different height represents a different phosphorylation level. In fact, it is well known that TTP can be phosphorylated by p38 MAP kinase and that this event results in protein stabilization and in the loss of mRNA destabilizing activity [32]. Moreover, as mentioned in the introduction, TTP mRNA destabilizing capability is active at a specific protein concentration [8]. At different concentrations the protein undergoes inactivation by phosphorylation. In our hands this makes over-expression experiments particularly tricky, since the achievement of very high expression levels often leads to the accumulation of an inactive form of TTP.
In order to further verify our hypothesis, we analyzed by chromatin immunoprecipitation the ability of the transcriptional repressors SNAI1 and SLUG to bind ZFP36 promoter at specific E-boxes. Results showed that both repressors are indeed capable of binding ZFP36 promoter and, in particular, the binding of SLUG to the promoter is significantly higher in SCC15 cells than in healthy keratinocytes. Results obtained on SNAI1, although statistically not significant, suggest that also this repressor preferentially binds ZFP36 promoter in SCC15 cells compared to healthy keratinocytes. A wider set of chromatin immunoprecipitation data is needed to confirm statistical significance. Chromatin immunoprecipitation assays on TWIST binding are currently ongoing.
In general, we showed that in SCC, and possibly in other tumors of epithelial origin characterized by EMT, where ZFP36 is down-regulated and the alteration of Wnt/CTNNB1 signaling plays a major role, the link between the signal transduction pathway and ZFP36 is represented by the transcriptional repressors described in this paper. In tumors of different origin, where Wnt/CTNNB1 signaling is not involved and ZFP36 is down-regulated, in the absence of specific mutations or epigenetic events (that are not to be excluded), the tumor advantage-conferring TTP silencing can be achieved as a consequence of the deregulated activity of other specific pathways. This idea seems also acceptable considering the fact that TTP expression is often induced physiologically in order to switch off a biological process, be it the inflammatory response, proliferation or other. Conversely, different deregulated pathways underlying transformation and the origin of different tumors might converge, among many other consequences, to the inactivation of the expression of the ZFP36 gene.
The tumor lines were treated with FH535 (Sigma, St. Louis, Missouri, USA) dissolved in DMSO and diluted in culture medium at the concentration of 30 µM. Control cells received the same volume of DMSO [6].

Western Blot Analysis
Total protein extracts were obtained by lysing the cells in RIPA buffer (50 mM TRIS-HCl pH 7.4, 150 mM NaCl, 1% NP40, 1 mM sodium deoxycholate, 1 mM sodium orthovanadate, 1 mM EDTA) with addition of Protease Inhibitor Cocktail 1x (Roche Applied Science, Penzberg, Germany). Cells were resuspended and centrifuged in the buffer and subsequently quantified by Bradford assay. A fixed amount of protein for each sample was loaded on SDS-Polyacrylamide gels for each sample and transferred to a nitrocellulose membrane. The membranes were blocked with 5% non-fat milk in TBS + 0.1% Tween, and then immunoblotted overnight at 4 • C with the following primary antibodies:

Immunofluorescence
Untreated cells were grown on slides coated with 1 mL of collagen IV and fixed with 4% paraformaldehyde for 20 min at room temperature, then washed with PBS. Cells were permeated with 1% Triton X100 for 5 min on ice, and then incubated with Blocking Solution (Goat Serum 10% and BSA 1% in PBS) for 20 min. The slides were incubated with the primary antibody anti-CTNNB1 (BD Transduction Laboratories 610154) diluted 1:100 in blocking solution, overnight at 4 • C in humidified chamber. Subsequently the slides were incubated for 45 min at room temperature with anti-mouse Alexa Flour 488 antibody (Invitrogen, Carlsbad, California, USA), diluted 1:300. Fluorescent samples were analyzed by confocal scanning laser microscope (Leica TCS SP2, Wetzlar, Germany).

Conflicts of Interest:
The authors declare no conflict of interest.