Inhibitory Mechanism of FAT4 Gene Expression in Response to Actin Dynamics during Src-Induced Carcinogenesis

Oncogenic transformation is characterized by morphological changes resulting from alterations in actin dynamics and adhesive activities. Emerging evidence suggests that the protocadherin FAT4 acts as a tumor suppressor in humans, and reduced FAT4 gene expression has been reported in breast and lung cancers and melanoma. However, the mechanism controlling FAT4 gene expression is poorly understood. In this study, we show that transient activation of the Src oncoprotein represses FAT4 mRNA expression through actin depolymerization in the immortalized normal human mammary epithelial cell line MCF-10A. Src activation causes actin depolymerization via the MEK/Erk/Cofilin cascade. The MEK inhibitor U0126 blocks the inhibitory effect of Src on FAT4 mRNA expression and Src-induced actin depolymerization. To determine whether actin dynamics act on the regulation of FAT4 mRNA expression, we treated MCF-10A cells with the ROCK inhibitor Y-27632. Y-27632 treatment decreased FAT4 mRNA expression. This suppressive effect was blocked by siRNA-mediated knockdown of Cofilin1. Furthermore, simultaneous administration of Latrunculin A (an actin depolymerizing agent), Y-27632, and Cofilin1 siRNA to the cells resulted in a marked reduction of FAT4 mRNA expression. Intriguingly, we also found that FAT4 mRNA expression was reduced under both low cell density and low stiffness conditions, which suggests that mechanotransduction affects FAT4 mRNA expression. Additionally, we show that siRNA-mediated FAT4 knockdown induced the activity of the Hippo effector YAP/TAZ in MCF-10A cells. Taken together, our results reveal a novel inhibitory mechanism of FAT4 gene expression through actin depolymerization during Src-induced carcinogenesis in human breast cells.


Introduction
Oncogenic cell transformation results from the summation of changes in cell growth, cell viability, cell motility and cell morphology. The v-Src oncogene, a product of the Rous sarcoma virus, is the constitutively active form of c-Src. Src has the ability to regulate various signal transduction pathways, including the Ras/MEK/Erk, PI3K/Akt, STAT3, and Rho/ROCK pathways [1][2][3][4]. More specifically, Src has been reported to induce alterations in cell morphology through actin dynamics and to depolymerize the actin cytoskeleton via the MEK/Erk/Cofilin cascade [5]. Additionally, members of the cadherin superfamily have been implicated in Srcinduced tumor transformation. Src downregulates E-cadherin expression and triggers morphological changes in multiple cancers [6][7][8]. These findings suggest the importance of both actin dynamics and the loss of cadherin-mediated cell-cell adhesion in Src-induced tumorigenesis.
FAT4, a protocadherin, is the human ortholog of Drosophila Fat [9,10]. Recent studies indicate that Drosophila Fat suppresses tumorigenesis through activation of the Hippo pathway. In support of this finding, PrognoScan, a new microarray database [11], and other recent studies [12][13][14] have demonstrated that human FAT4 gene expression is repressed in breast and lung cancers and in melanoma, which suggests that reduced FAT4 gene expression can trigger carcinogenesis. However, the molecular mechanisms underlying the down-regulation of FAT4 gene expression in human cancers remain unknown.
The Hippo pathway is involved in tumor suppressor signaling and regulates organ size, cell proliferation, apoptosis and stemness [9,10,15,16]. In mammals, the core components of the Hippo pathway are primarily composed of Mst1/2 and Lats1/2, which inactivate the Hippo effector Yap/Taz via phosphorylation [17][18][19][20][21][22]. Yap/Taz acts as a transcriptional co-activator by binding the transcription factors Tead1-4 to induce the expression of genes such as the connective tissue growth factor CTGF [18,23,24]. Fat4 has been shown to repress Yap activity and thereby negatively regulates neural progenitor cell proliferation during mammalian neurogenesis [25,26]. Thus, the FAT4-Hippo axis is believed to be functionally conserved in humans; however, the functional significance and the molecular mechanisms that confer this axis are poorly understood.
Here, we show that transient Src activation represses FAT4 mRNA expression and leads to tumor transformation in MCF-10A cells, which are immortalized normal human mammary epithelial cells. We further demonstrate that the inhibitory effect of Src on FAT4 mRNA expression depends on actin depolymerization that is induced by the MEK/Erk/Cofilin cascade. Finally, we demonstrate that FAT4 represses YAP/TAZ activity in MCF-10A cells.

FAT4 gene expression is repressed during human breast cell transformation by the Src oncogene
To analyze the mechanisms that repress FAT4 gene expression in human breast cancer, we generated a model of carcinogenesis using MCF-10A cells, an immortalized normal human mammary epithelial cell line [27][28][29][30][31]. In brief, we used retroviruses to create MCF-10A cells that stably express the Src kinase oncoprotein (v-Src) fused to the ligand binding domain of the estrogen receptor (ER). In MCF-10A v-Src:ER cells, the addition of 4-Hydroxytamoxifen (TAM) induces Src activation and their subsequent cell transformation. TAM treatment resulted in morphological transformation (Fig. 1A), enhanced cell proliferation (Fig. 1B), and promoted anchorage-independent growth (Fig. 1C). In this model, we found that FAT4 mRNA expression was repressed by Src activation in MCF-10A cells (Fig. 1D). On the other hand, TWIST1 mRNA expression was not altered under similar experimental conditions (S1 Fig.). This points to the specificity of the inhibitory effect of Src on FAT4 mRNA expression. Thus, these results indicate that Src induces tumorigenesis and suppresses FAT4 mRNA expression in MCF-10A cells.
Src activation induces both FAT4 gene repression and Cofilin-mediated actin depolymerization through the MEK/Erk pathway Src has been shown to elicit Cofilin-mediated disruption of the actin cytoskeleton through the MEK/Erk pathway [5]. The function of Cofilin is activated via its dephosphorylation [32]. Thus, we examined whether Src can disrupt the actin cytoskeleton through dephosphorylation of Cofilin in MCF-10A v-Src:ER cells. Indeed, phalloidin staining and Western blot analysis demonstrated that Src can depolymerize the actin cytoskeleton and dephosphorylate Cofilin (Fig. 2). We further examined the involvement of the MEK/Erk pathway in Src-mediated modulation of the actin cytoskeleton. Treatment of MCF-10A v-Src:ER cells with the MEK inhibitor U0126 completely inhibited Src-mediated phosphorylation of Erk1/2, a major target of MEK1/2 (Fig. 3A). Under this condition, U0126 significantly inhibited both Src-induced actin dynamics (Fig. 3B) and the dephosphorylation of Cofilin (Fig. 3C). Therefore, these results indicate that Src induces Cofilin-mediated actin depolymerization through the MEK/Erk pathway in MCF-10A cells.
We also investigated whether the MEK/Erk pathway is involved in Src-reduced FAT4 mRNA expression by RT-qPCR analysis; the results suggest that Src-mediated inhibition of FAT4 mRNA expression is repressed in MCF-10A cells by U0126 or PD0325901, a MEK inhibitor ( Fig. 3D and S2 Fig.). In addition, to emphasize the importance of the MEK/Erk pathway in FAT4 mRNA repression we evaluated the effect of U0126 on FAT4 mRNA expression in MDA-MB-231 cells, a malignant breast cancer cell line. Treatment of MDA-MB-231 cells with U0126 also up-regulated FAT4 mRNA expression (Fig. 3E). Thus, these results indicate that Src represses FAT4 mRNA expression through the MEK/Erk pathway. Intriguingly, siRNA-    (Fig. 3F), perhaps because the MEK/Erk pathway regulates various targets that include FAT4 gene. Collectively, these results suggest that Src activation induces not only Cofilinmediated actin depolymerization but also FAT4 mRNA repression through the MEK/ Erk pathway.

Src represses FAT4 gene expression through Cofilin1-mediated actin depolymerization
Next, we examined the causal relationship between FAT4 mRNA repression and Cofilinmediated actin depolymerization, both of which are downstream of the Src/MEK/Erk pathway. The ROCK pathway is a major pathway that regulates Cofilin-mediated actin depolymerization [33,34]. Thus, to determine whether Src downregulates FAT4 mRNA expression by altering actin structures, we examined the effect of the ROCK inhibitor Y-27632 on FAT4 mRNA expression. Treatment of MCF-10A cells with Y-27632 reduced FAT4 mRNA expression in a dose-dependent manner (Fig. 4A). The inhibitory effect of Y-27632 on FAT4 mRNA expression was abrogated by siRNA-mediated knockdown of Cofilin1 (Fig. 4B). Cofilin1 is a nonmuscle type of Cofilin [32]. Effective knockdown of Cofilin1 was supported by RT-qPCR ( Fig. 4B) and Western blot analysis (Fig. 4C). Moreover, simultaneous administration of the actin depolymerizing agent Latrunculin A, Y-27632, and siCofilin1 to the cells resulted in a striking decrease of FAT4 mRNA expression equal to that of the Y-27632 treatment alone ( Fig. 4B and S3 Fig.). Phalloidin staining showed that Latrunculin A sufficiently depolymerizes the actin cytoskeleton (Fig. 4D). Therefore, on one hand, these results suggest that ROCK1/2 regulates FAT4 mRNA expression via Cofilin1-mediated actin depolymerization. On the other hand, they suggest that inhibition of Cofilin1 significantly attenuates FAT4 mRNA repression by Src (Fig. 4E). Therefore, we propose that Src suppresses FAT4 mRNA expression through actin depolymerization by Cofilin1.

Mechanotransduction affects FAT4 gene expression
Recent studies have reported that mechanical stress on cells affects cellular behavior via the modulation of actin dynamics [35][36][37]. Based on our findings that FAT4 mRNA expression is regulated by actin depolmerization through Cofilin1 (Fig. 4), we examined whether mechanical stress also affects FAT4 mRNA expression. To test this hypothesis, we performed a Cell Density Assay and a Softwell Assay. In the Cell Density Assay, MCF-10A cells were seeded at high and low cell densities to evaluate the effect of cell-cell contact on FAT4 mRNA expression. The Softwell Assay used collagen type I-coated Softwell plates of varying degrees of stiffness (50 kPa and 0.5 kPa) to assess the effect of anchorage stiffness on FAT4 mRNA expression. FAT4 mRNA expression was reduced under both low cell density (Fig. 5A) and low anchorage stiffness conditions (Fig. 5B). These findings suggest that the mechanical stress that stems from the extracellular environment is intimately linked with FAT4 mRNA expression. Moreover, we found no significant difference in phoshorylated-Cofilin expression between high and low density cells (S4 Fig.). Thus, it may be assumed that the observed alteration in FAT4 mRNA expression under low density conditions is not mediated through the MEK/ERK/Cofilin cascade.

FAT4 suppresses the activity of the Hippo effector YAP/TAZ
Accumulating evidence suggests that the FAT4-Hippo axis is also conserved in humans [12][13][14]25,26]. We next examined the effects of FAT4 knockdown on the function of the Hippo effector YAP/TAZ. YAP/TAZ activity is suppressed by cytoplasmic retention and proteasomal degradation due to LATS1/2-mediated phosphorylation of YAP/TAZ [38][39][40][41][42]. FAT4 siRNA increased mRNA expression levels of CTGF, a YAP/TAZ target gene (Fig. 6A) [18][19][20][21]23,24], by promoting the nuclear accumulation of YAP/TAZ (Fig. 6B). Furthermore, we found that the oncoprotein Src promoted YAP/TAZ activity in MCF-10A cells (Figs. 6C and 6D). These results are consistent with those of a recent study that demonstrated Drosophila Src64B activates Yorkie, the mammalian Yap/Taz homologue, by repressing the core components of the Hippo pathway (Hippo/Warts) [43]. Thus, our results suggest that FAT4 represses CTGF mRNA expression through the inactivation of YAP/TAZ. We next examined the molecular basis of YAP/TAZ activation by FAT4 knockdown. Because YAP/TAZ activity is regulated by mechanotransduction via actin dynamics [35,36,38], we investigated the effects of FAT4 siRNA on actin dynamics. Phalloidin staining revealed that FAT4 knockdown induced abnormal spiny actin structures that radiated in all directions from the cells (Fig. 6E and S5 Fig.). However, this structural modulation of actin was independent of Cofilin because we observed no increase in phosphorylated Cofilin (Fig. 6F). Taken together, our results may suggest that FAT4 represses YAP/TAZ activity by regulating actin structures.

FAT4 knockdown does not induce cell transformation
Finally, we investigated whether FAT4 knockdown alone can transform MCF-10A cells because FAT4 knockdown activates YAP/TAZ, which elicits cell transformation (Figs. 6A and 6B) [23,24]. However, FAT4 knockdown alone did not cause increased cell proliferation nor anchorage-independent growth in MCF-10A cells (S6 Fig.). Therefore, combinatorial effects of reduced FAT4 gene expression and other factors could be necessary for tumor transformation in MCF-10A cells.

Discussion
Reduced FAT4 gene expression in human breast cancers has been reported in several studies [11,12]. Nevertheless, the molecular mechanism that governs this down-regulation in gene expression remains unknown. Here, we show that FAT4 mRNA expression is repressed in response to actin dynamics during Src-induced tumor transformation of MCF-10A cells (Fig. 7). Src-mediated actin depolymerization is regulated by the MEK/Erk/Cofilin1 cascade and results in the repression of FAT4 mRNA expression. Finally, we demonstrate that reduced expression of FAT4 induces the activity of the Hippo effector YAP/TAZ in MCF-10A cells. Altogether, these results propose a novel mechanism by which actin dynamics regulate cancer-related traits through the loss of FAT4-mediated cell-cell adhesion. In this study, we demonstrate that the Src/MEK/Erk/Cofilin1 cascade, which mediates actin depolymerization, represses FAT4 gene expression in MCF-10A cells (Fig. 7). The mechanism by which the MEK/Erk pathway represses FAT4 gene expression is conserved in even the malignant breast cancer cell line MDA-MB-231 cells as well as the normal breast cell line MCF-10A cells (Fig. 3D and 3E). Src activates Cofilin via dephosphorylation by components of the MEK/Erk pathway (Fig. 3C) and thereby induces actin depolymerization (Fig. 3B). Notably, the MEK/Erk pathway plays an important role in activating Cofilin in a variety of human cancer cell lines [5,[44][45][46][47]. Moreover, simultaneous administration of the ROCK inhibitor Y-27632 and siCofilin1 lead to restore FAT4 mRNA expression repressed by Y-27632 in the cells (Fig. 4B). Thus, these results suggest that Src-Cofilin1 axis plays an important role in the regulation of FAT4 mRNA expression in breast cancer cells. Unexpectedly, the siRNAmediated Cofilin knockdown did not completely block the reduction in FAT4 mRNA expression by Src, although it abolished the inhibitory effect of Y-27632 on FAT4 mRNA expression ( Fig. 4B and 4E). Generally, Src regulates a wider array of signaling pathways than ROCK. As such, this may result in the significant differences presented in these data. Taken together, these results clearly show the inhibitory effect of Src on FAT4 mRNA expression in response to actin dynamics during carcinogenesis.
The molecular mechanisms by which Src-induced actin depolymerization represses FAT4 mRNA expression have not yet been elucidated. The FAT4 gene is silenced by promoter hypermethylation in breast cancer cells [12]. Interestingly, Src elicits the overexpression of DNA methyltransferase (DNMT) 1 and induces DNA hypermethylation in rat 3Y1 fibroblasts [48]. Additionally, the MEK/Erk pathway induces DNA hypermethylation via the regulation of DNMT1 [49][50][51]. Therefore, Src-induced actin depolymerization has been speculated to repress FAT4 mRNA expression via DNMT1-induced promoter hypermethylation during carcinogenesis, but further studies are needed for clarification.
Our study demonstrates that siRNA-mediated knockdown of FAT4 induces YAP/TAZ activity in MCF-10A cells (Fig. 6A and 6B). In line with this view, recent studies utilizing in vivo electroporation have demonstrated that Fat4 downregulates Yap activity leading to the inhibition of neural progenitor cell proliferation during murine and avian neurogenesis [25,26]. In addition, it has been demonstrated that deletion of Fat4 in mice causes primarily the nuclear accumulation of Yap in the nephron progenitor cells [52]. As such, these in vivo findings suggest that the Fat4-Hippo axis is highly conserved in mammals. However, the molecular underpinnings of Fat4-mediated nuclear translocation of Yap/Taz in mammals still remain unknown. Because there is no mammalian counterpart of Drosophila Dachs, which is regulated by Fat to activate the Hippo pathway [9,10,18], Fat4 should modulate Yap/Taz activity differently from Fat. In this regard, our research demonstrates that YAP phosphorylation levels (at least at Ser127) in FAT4 siRNA-treated cells are consistent with those found in control siRNAtreated MCF-10A cells (S7 Fig.). Nevertheless, MST1 kinase (mammalian Hippo ortholog) expression is reduced in FAT4 siRNA-treated cells (S7 Fig.). Accordingly, it is possible that nuclear localization of YAP is regulated by FAT4 through a Ser127 phosphorylationindependent mechanism.
It is also known that FAT4 is involved in the maintenance of planar cell polarity (PCP) [9]. Indeed, Fat4 mutant mice and PCP gene mutant mice show similar defects of the inner ear such as significant disruptions in orientation of hair cells and elongation of the cochlea [53]. Moreover, PCP genes have also been shown to regulate actin dynamics [54]. Collectively, these findings suggest that Fat4 regulates actin dynamics as well as vertebrate PCP. Interestingly, YAP/TAZ activity can be modulated by cytoskeletal tension induced by cell spreading in a Hippo pathway-independent manner [10,21,35]. Therefore, we can surmise that FAT4 regulates the YAP/TAZ activity through modulating actin cytoskeleton. In fact, we show that FAT4 knockdown induces abnormal spiny actin protrusions, which may enhance cell spreading (Fig. 6E). Thus, FAT4 may negatively regulate YAP/TAZ function by modulating actin cytoskeletal tension. Altogether, our current study presents a novel model in which FAT4 represses YAP/TAZ activity by regulating actin structures, at least in breast cancer cells. Further examination is needed to elucidate the molecular basis of YAP/TAZ inactivation by FAT4.
Accumulating evidences suggest the pathophysiological relevance between FAT4 reduction and carcinogenesis. Qi et al. have clearly demonstrated that FAT4 mRNA is not expressed at all in many breast cancer cell lines (BT20, ZR75-1, and BT474 cells) [12]. We also found that MCF-7 breast cancer cells express far too low a level of FAT4 mRNA compared with the MCF-10A normal mammary cell line (data not shown). Moreover, FAT4 knockdown has been shown to induce the malignant phenotype of human gastric cancer cell lines [55]. However, we show that FAT4 knockdown alone is not sufficient to induce cell proliferation or anchorage-independent growth in MCF-10A cells despite promoting YAP/TAZ activity (S6 Fig.; Fig. 6A and 6B); nevertheless, overexpression of YAP/TAZ can transform the cells [23,24]. Additionally, FAT4 knockdown alone does not induce cell proliferation in malignant breast cancer MDA-MB-231 cells (Fig. 3F). These data strongly demonstrate that cooperative effects between FAT4 reduction and other factors induce tumor transformation in MCF-10A cells. We propose that the MEK/Erk pathway involves in the Src-induced tumorigenesis as well as FAT4 mRNA repression (Fig. 3A-D), because numerous studies have already established the importance of the MEK/Erk pathway for cell transformation [2,5,44,56,57]. Further comprehensive analyses are required to solve molecular mechanisms of Src-mediated carcinogenesis through repressing FAT4 gene expression.

siRNA knockdown experiments
Cells were cultured in Growth Medium without antibiotics at 37°C for 24 h and were then transfected with siRNA using both Lipofectamine RNAiMAX (Invitrogen) and Opti-MEM I Reduced Serum Medium (31985-070; Gibco) according to the manufacturer's protocol. The sequences of the siRNAs employed in this study are listed in S1 Table. The siRNA concentrations are described in the Figure Legends.

RNA extraction and real-time quantitative PCR (RT-qPCR)
Total RNA was extracted from cells with ISOGENII (311-07361; Wako) or the RNeasy Mini Kit (Qiagen) and was subjected to cDNA synthesis with random hexamer primers and Moloney murine leukemia virus (M-MLV) reverse transcriptase (Invitrogen) according to the manufacturer's protocols. RT-qPCR was performed with FastStart Universal SYBR (Roche) and a Thermal Cycler Dice Real Time System II (TP900; Takara Bio). The PCR primers that were used are listed in S2 Table. 18S rRNA was used for normalization.

Fluorescent staining
The cells were fixed with 4% formaldehyde solution in PBS on ice for 15 min, washed 3 times, and were permeabilized with 0.5% Triton X-100 in PBS for 10 min at room temperature (RT). After blocking cells with PBS containing 1% skim milk for 20 min at RT, the cells were incubated overnight at 4°C with anti-YAP antibody diluted in PBS containing 1% skim milk. After washing the cells 3 times with PBS, the cells were incubated with Alexa Fluor 546-conjugated secondary antibody (A-11030; Invitrogen) diluted in PBS containing 1% skim milk for 1 h at RT. To stain actin filaments (F-actin), the cells were blocked with PBS containing 5% BSA for 30 min at RT and were incubated with CF594 Phalloidin (00045; Biotium) in PBS containing 1% BSA for 20 min at RT, followed by PBS washes. The nuclei were stained with 4',6'-diamidino-2-phenylindole (DAPI) (100 ng/ml) for 10 min at RT. Finally, the cells were sealed with fluorescence mounting medium (S3023; Dako). Fluorescent images were captured with an Olympus LX71 fluorescence microscope.

Soft Agar Colony Formation Assay
The cells were seeded at 4.6×10 5 cells (6 cm dish), cultured in Assay Medium at 37°C for 24 h and were then treated with TAM at 37°C for 72 h before the colony formation assay. For siRNA-mediated FAT4 knockdown experiments, the cells were seeded at 1.8×10 5 cells (6 cm dish) and cultured in Growth Medium without antibiotics at 37°C for 24 h. The cells were then transfected twice with FAT4 or control siRNA (at 24 h and 48 h after plating) and cultured for a further 24 h after the last transfection. Subsequently, the cells were seeded at 7.0×10 2 cells per well (96-well plate) in Growth Medium containing 0.4% agarose (100 μl) onto a lower layer of Growth Medium containing 0.5% agar (100 μl). Cells were fed 2 times per a week. The number of colonies was counted after 14 days.

Cell Density Assay
The cells were seeded at 4.2×10 5 cells (High density) or 7.0×10 4 (Low density) cells per well (6-well plate), cultured in Assay Medium at 37°C for 24 h and subjected to RT-qPCR analysis.

Softwell Assay
Twenty-four well Softwell plates of two different qualities of stiffness (50 kPa: stiff or 0.5 kPa: soft) (Matrigen) were coated by Collagen I, rat tail (354236, BD Biosciences) (10°g/ml) for 1 h at RT. The cells were seeded at 8.1×10 4 cells per well in Softwell plates. The cells were cultured in Assay Medium for 48 h and subjected to RT-qPCR analysis.

Statistical analyses
Statistical analyses were performed using Student's t-test; a P < 0.05 was considered significant.