2.1 Antibodies and chemicals

The following antibodies were used: mouse anti-β-catenin monoclonal antibody (15B8, #C7207), mouse anti-E-cadherin monoclonal antibody (36, #610181; BD Biosciences); rabbit anti-COX-2 monoclonal antibody (D5H5, #12282; Cell Signaling Technology); and mouse anti-GAPDH monoclonal antibody (6C5, #AM4300; Thermo Fisher Scientific). The secondary antibodies goat anti-rabbit (#NA934) and anti-mouse (#NA931) IgG horseradish peroxidase-conjugated were obtained from GE Healthcare. Alexa 488-conjugated goat anti-rabbit IgG and Alexa 546-conjugated goat anti-mouse IgG were purchased from Molecular Probes. Transforming growth factor-β (TGF-β; #PHG9204)) was purchased from Life Technologies and diluted in phosphate-buffered saline (PBS) + 0.1% bovine serum albumin (BSA) to obtain a stock concentration of 10 µM. Prostaglandin E₂ (#14520) was purchased from Cayman Chemical, and diluted in dimethyl sulfoxide (DMSO) to obtain a stock concentration of 100 mM.

2.2 Cell culture and treatments

HT-29 human colon adenocarcinoma cell lines (ATCC; #HTB-38) obtained from American Type Culture Collection were passaged weekly using a solution of 0.05% trypsin/0.02% EDTA in PBS solution. HT-29 cells are moderately differentiated, with PIK3CA, TP53, SMAD4, and APC mutations (Bastos et al., 2014). IEC-6 (ATCC; #CRL-1592) cells had growth medium supplemented with 0.1 unit of insulin. All cells were grown in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% fetal bovine serum (FBS), penicillin G (60 mg/L), and streptomycin (100 mg/L) at 37°C in a humidified atmosphere of 5% CO₂/air. Cell cultures were switched to serum-free medium for 24 h before PGE₂ (10 nM) and TGF-β (5 ng/ml) treatment. IEC-6 pBabe and IEC-6 H-RasV12 cells were also used in the present study and obtained as previously reported by Accioly et al. (2008).

2.3 Cell extraction and Western blot analysis

Total cell lysates were obtained by incubating the cells in lysis buffer (150 mM NaCl, 10 mM Hepes, pH 7.3, 0.2% sodium dodecyl sulfate, 1% TX-100, 0.5% deoxycholate, and 2 mM EDTA) containing 2 mM orthovanadate, 20 mM NaF, and protease inhibitor cocktail (1:100; Sigma-Aldrich). Cells were then scratched from the plates, homogenized, and centrifuged at 11,000g for 10 min at 4°C. The supernatant was collected and stored at –20°C.

For Western blot analysis, equal amounts of protein (30 μg/lane of cell lysate) were electrophoretically separated by sodium dodecyl sulfate-polyacrylamide gel electrophoresis on 10% or 12% polyacrylamide gel. Protein was then transferred to nitrocellulose membranes using a semidry transfer cell. After transfer onto nitrocellulose membranes, nonspecific binding sites were blocked with 5% nonfat milk in TBS-Tween (TBST; 50 mmol/L Tris–HCl [pH 7.4], 150 mmol/L NaCl, 0.05% Tween 20). Membranes were probed overnight in primary antibodies against COX-2 (1:1000), and GAPDH (1:10,000) in TBS-T. Proteins of interest were then identified by incubating the membrane with horseradish peroxidase–conjugated secondary antibodies in TBST. Membranes were developed with an enhanced chemiluminescence reagent (Amersham Biosciences), and luminescence was visualized by the ChemiDoc MP Imaging System from Bio-Rad.

2.4 Immunofluorescence

To analyze the subcellular localization of proteins, cell monolayers were grown on sterile glass coverslips and then manually wounded by scraping with a pipet specifically to analyze the front of the monolayer. After treatment, cells were washed in PBS and fixed in 4% paraformaldehyde for 10 min and permeabilized/blocked with 0.1% Triton X-100 3% BSA in PBS for 2 h. Coverslips were incubated overnight at 4°C in primary antibodies against E-cadherin (1:500) and β-catenin (1:200), followed by 1 h in the respective Alexa 488-conjugated secondary antibodies (1:500). Afterward, the coverslips were incubated with 4′,6-diamidino-2-phenylindole (DAPI; 1:1000) for 1 min, washed, and mounted using Prolong Gold antifade reagent (Invitrogen). Cell staining was detected using a Fluoview FV10i (Olympus Co.) laser scanning confocal microscope with a 543 nm excitation laser. Individual images were collected of regions whereof the monolayers were scrapped. The images shown are representative of at least three independent experiments.

2.5 Cell proliferation assay

HT-29 cells (2 × 10⁴ cells/ml) were cultured in 96-well plates and treated with PGE₂ (10 nM) and TGF-β (5 ng/ml) for 24 or 48 h. After this time, they were fixed with ethanol for 10 min and incubated with a crystal violet solution (0.05% crystal violet and 20% ethanol) for an additional 10 min. Wells were washed twice with water and then solubilized with 100% methanol. The absorbance at 595 nm was measured with a Spectra Max 190 spectrophotometer (Molecular Devices).

2.6 Anchorage-dependent and independent colony formation assays

Anchorage-dependent and anchorage-independent colonies constitute an important parameter for measuring tumorigenic potential in vitro. First, for anchorage-dependent colony formation assays, HT-29 cells were seeded at low density (500 cells/well) in 12-well plates. Cells were monitored for 10 days to check the ability of a single cell to form a colony. Treatment with PGE₂ (10 nM) and TGF-β (5 ng/ml) was continuous over 10 days. At the end of the experiment, cells were fixed with 100% ethanol for 10 min and subsequently incubated with a solution containing 0.05% crystal violet and 20% ethanol for 10 min. Then, wells were washed with distilled water and the crystal violet eluted with 100% methanol. The absorbance at 595 nm was quantified using a SpectraMax 190 spectrophotometer (Molecular Devices). Second, for anchorage-independent colony formation assays, the cells (500 cells/well) were seeded into a 12-well plate previously coated with 1 ml of a semi-solid medium containing 0.3% agarose. After 10 days of treatment, cell colonies were observed and counted with an Axio Observer Z1 microscope (Carl Zeiss, Inc.).

2.7 Migration assay

Cell migration is an important parameter to evaluate the tumorigenic potential of cancer cells. To do that, HT-29 cells were seeded into a six-well plate and allowed to grow until they reached confluency. Cellular monolayers were then manually wounded by scraping with a pipette tip to carry out the wound healing assay. Wells were washed with PBS and incubated in complete growth medium in the absence or presence of PGE₂ and TGF-β for 12 and 24 h. For each well, five sites of a unique regular wound were analyzed under an Axio Observer Z1 microscope equipped with an Axio Cam HRc and Axio Vision Release 8.2 Image Analyzer. The sites were then selected and marked. After treatment, cells were allowed to migrate into the clearing area for 24 h. Immediately after wounding (0 h) and at the end of the experiment (24 h), the entire wound of the five previously selected sites of each scrap/well were photographed.

The lesion area was manually quantified using ImageJ software version 1.51r (http://imagej.nih.gov/ij). Values were represented by the percentages of cell migration in bar graphs, and the data are presented as the means ± SEM of triplicate assays for each cell line. At least three independent experiments were performed.

2.8 Luciferase assay for TCF activity

Top-flash luciferase reporter assay is fundamental to analyze activation of the Wnt/β-catenin pathway. Cells were seeded in 24-well plates and transiently transfected with 2 μg SUPER-8 Top-flash reporter plasmid and 3 μl FUGENE HD Transfection Reagent (Roche Applied Science). For transfection efficiency control, 0.2 μg of a Renilla luciferase construct (pRL-Tk) was included in each transfection. Twenty-four hours after transfection, the cells were washed twice with PBS and then treated with 5 ng/ml TGF-β, 10 nM PGE₂, or 10 mM LiCl as a positive control. After 24 h, cells were harvested, and the extracts were prepared with 50 µl reporter lysis buffer (Promega). Renilla and luciferase activity were assayed according to the manufacturer's protocol with a Kit Dual-Luciferase Reporter Assay System (Promega). The luciferase activity in each well was normalized to the Renilla activity.

2.9 In vitro invasion assay

To analyze cell invasion, HT-29, IEC-6 pBabe, and IEC-6 H-RasV12 cells (3 × 10⁴) were seeded in the upper surface of an 8 μm pore size Polycaronate Membrane Transwell® insert (Corning Costar). This was previously coated with 20 µl Matrigel® (BD Biosciences) diluted 1:10 in DMEM. Cells were seeded in 200 µl of serum-free medium and incubated at 37°C for 30 min. DMEM with 20% FBS was added as a chemoattractant in the lower chamber. After 24 h of incubation, the upper surface of the membrane was scrubbed with a cotton swab. Invasive cells in the lower membrane were fixed with ethanol for 10 min, stained with 0.05% crystal violet, and analyzed under the Axio Observer Z1 microscope (Carl Zeiss, Inc.). The number of invaded cells was expressed as the average of five random fields under the microscope. The cell invasion values are represented as fold increases of cell invasion via bar graphs, and data are presented as the mean ± SEM of triplicate assays of three independent experiments.

2.10 Statistical analysis

Statistical analysis was performed using GraphPad Prism 5.0 (GraphPad Software). Data from at least three independent experiments are expressed as mean ± SEM and were interpreted using one-way analysis of variance (ANOVA) followed by the Bonferroni or Dunnet's posttest where relevant. All comparisons were determined using a one-way ANOVA between the experimental group and the control group. Differences were considered statistically significant when *p < .05; **p < .01 and ***p <.001.

3.1 Combination of TGF-β plus PGE₂ treatment hinders the increased cell migration and invasion induced by PGE₂ alone

Initially, to elucidate the role that PGE₂ and TGF-β play on the aggressive features of HT-29 cells, we investigated the cell migration at 12 and 24 h and the invasive potential of these cells after 24 h of treatment with PGE₂, TGF-β, or a combination of these compounds. We observed that after 24 h, all treatments caused a significant increase of wound healing, specially in the treatment with PGE₂ alone. Interestingly, combined treatment with these compounds induced cellular migration, but it inhibited the effect caused by treatment with TGF-β alone (Figure 1a). Additionally, we observed that in shortened time points (12 h), when the impact on cell proliferation could be less pronounced the treatments did not alter cell migration (Figure S1). Furthermore, we found that only PGE₂ alone significantly increased invasiveness compared to untreated cells after 24 h of treatment. Even though, TGF-β alone had no significant effect on cell invasion, the association of it with PGE₂ hindered its sole effect in invasion (Figure 1b).

3.2 Joint treatment of TGF-β and PGE₂ prevents increase in colony formation and proliferation induced by PGE₂ alone

Tumor cells can form anchorage-dependent and anchorage-independent colonies, which constitutes an important parameter for measuring tumorigenic potential in vitro. Therefore, we evaluated the effects of PGE₂ and TGF-β on the clonogenic properties of HT-29 cells. Figure 2a shows that HT-29 cells exposed to PGE₂ had significantly increased numbers of anchorage-dependent colonies when compared to the control group. TGF-β treatment did not induce changes, but when PGE₂ and TGF-β were applied in combination, the increased number of colonies caused by PGE₂ was completely inhibited. Additionally, we assessed the effects of PGE₂ and TGF-β treatment on anchorage-independent growth using a soft agar colony formation assay. Our results show that PGE₂ significantly increased colony size, while TGF-β alone and TGF-β in combination with PGE₂ impaired this increase (Figure 2b).

We then decided to verify whether these treatments alter cellular viability over time using the crystal violet assay. We observed after 24 h of PGE₂ and TGF-β treatment, an increase in cell viability indicating that HT-29 cells had significantly increased cell growth. However, after 48 h, only PGE₂ showed this effect. PGE₂ and TGF-β joint treatment reduced cell viability after 48 h of treatment, but not at 24 h (Figure 3).

3.3 TGF-β increases COX-2 protein levels and invasiveness in a H-Ras dependent fashion

Our results showed that HT-29 cells treated with PGE₂ had increased growth, migration, and invasion, but that the combination with TGF-β inhibited these effects. This might indicate that TGF-β works as a tumor suppressor in the presence of PGE₂. There is evidence that for TGF-β to induce invasion and tumoral progression, Ras signaling must be active (Liu et al., 2018). Thus, to evaluate the contribution of the Ras oncogene to TGF-β-associated tumor promotion, we used rat intestinal cells (IEC-6) transfected with H-Ras oncogene (IEC-6 H-RasV12). This cellular model was previously established and characterized by increased COX-2 expression and therefore, increased PGE₂ production and release in the culture medium (Accioly et al., 2008). Matrigel-coated transwell invasion assay showed that TGF-β treatment alone did not cause a significant increase in invasion in IEC-6 pBabe. However, when associated with IEC-6 H-RasV12 cells, that are already more invasive, TGF-β lead to more invasion (Figure 4a). Next, we used Western blot analysis to confirm whether TGF-β treatment-induced changes in COX-2 expression in IEC-6 pBabe and IEC-6 H-RasV12 cells. Figure 4b shows that, in the presence of TGF-β, IEC-6 H-RasV12 cells have higher COX-2 protein levels. These results suggest that H-Ras is necessary for TGF-β to enhance the invasive potential and when combined, both lead to COX-2 upregulation.

3.4 TGF-β and oncogenic H-Ras induce Wnt/β-catenin pathway activation

Previous studies have demonstrated that H-Ras and TGF-β can increase COX-2 expression in human CRC cell lines (Du et al., 2005; Roman et al., 2002). However, the mechanisms through which this occurs remain unclear. It is known that COX-2 is a transcriptional target of Wnt/β-catenin signaling (Araki et al., 2003). Therefore, Wnt/β-catenin activation could be the regulator of COX-2 expression in our model. To test this hypothesis, we investigated Wnt/β-catenin activation in IEC-6 pBabe and IEC-6 H-RasV12 cells, with or without TGF-β treatment. This was done using a Top-flash luciferase reporter assay. Our results show that IEC-6 H-RasV12 cells have higher luciferase activity (Top activation) when compared to IEC-6 pBabe and that after 24 h, TGF-β treatment significantly increased luciferase activation of IEC-6 pBabe and IEC-6 H-RasV12 cells compared to the respective control groups. In this experiment, lithium treatment was used as a positive control of TCF activation (Figure 5a). Wnt/β-catenin pathway activation is characterized by the translocation of β-catenin into the nucleus before TCF activation. This has been implicated in tumorigenesis and increased cell invasion (Haase et al., 2016). For this reason, we analyzed β-catenin localization in response to TGF-β via immunofluorescence assay of IEC-6 pBabe and IEC-6 H-RasV12 cells. As shown in Figure 6b, β-catenin was observed at cell-cell contact points and in the cytoplasm of IEC-6 pBabe cells. When treated with TGF-β for 24 h, we observed the translocation of β-catenin from the membrane to the nucleus. H-Ras oncogene expression alone was enough to determine β-catenin localization in the nucleus. Associated with TGF-β, IEC-6 H-RasV12 treated cells displayed intense β-catenin staining in the nucleus (Figure 5b). These findings support the notion that the H-Ras oncogene and TGF-β could increase COX-2 expression via Wnt/β-catenin signaling.

3.5 Independent treatment of HT-29 cells with PGE₂ and TGF-β causes subcellular E-cadherin and β-catenin redistribution, but not when combined

The loss of cell-cell contacts is a feature that potentially causes tumor cells to migrate and invade tissues (Zhou et al., 2017). Thus, we raised the question whether the subcellular localization of E-cadherin and β-catenin would be affected after treatments with PGE₂, TGF-β, or their combination. To address this question, we used immunofluorescence to analyze the subcellular redistribution of E-cadherin and β-catenin, specifically in the front of the monolayer, focusing on cells that may migrate after these treatments. As shown in Figure 6a,b, we observed that 24 h after PGE₂ and TGF-β treatment, E-cadherin and β-catenin lost their predominant membrane staining and showed a more diffuse pattern with some punctate in the cytoplasm. This indicates a disorganization of cellular contacts when compared to the control group. Cells in the combined treatment showed majority labeling of both proteins at the cell-cell contact points, similar to the control group. Considerable β-catenin labeling was detected at the nucleus after TGF-β treatment, unlike cells that received the combined treatment. Taken together, these data indicate that PGE₂ and TGF-β both individually promote the loss of cell-cell contacts. Interestingly, the presence of both agents seems to hinder their individual effects. As PGE₂ treatment triggers cell-cell disorganization, as observed by redistribution of E-Cadherin and β-catenin, and this can be a clue of epithelial–mesenchymal transition (EMT), we need to analyze by reverse-transcription polymerase chain reaction the mRNA levels of E-cadherin, a classic EMT marker. Our result showed that the treatments with PGE₂ and TGF-β did not alter the RNA levels of this proteins, but the treatment combined increased these levels (Figure S2).

3.6 Wnt/β-catenin signaling activation is a differential regulator of cell invasiveness in the presence of TGF-β

Our prior results using HT-29 cells showed that PGE₂ treatment induced disorganization of the cellular contacts through E-cadherin and β-catenin relocalization, increased migration, proliferation, colony formation, and invasiveness. However, when cells were treated with PGE₂ and TGF-β in combination, these effects were absent. Conversely, in IEC-6 H-RasV12 cells, which showed increased COX-2 expression and Wnt/β-catenin activation, TGF-β treatment increased invasiveness. We, therefore, decided to analyze whether COX-2 expression and Wnt/β-catenin signaling activation occurs in HT-29 cells treated with PGE₂ and TGF-β. As shown in Figure 7a, PGE₂, TGF-β, and PGE₂ in combination with TGF-β treatments did not induce significant β-catenin transcriptional activity in comparison with the control group. In this assay, lithium chloride (10 mM) was used as positive control. To further confirm this result, we stimulated HT-29 cells with the same treatments for 24 h. Interestingly only TGF-β increased COX-2 protein levels. PGE₂ left COX-2 levels unaltered and so did the treatment with PGE₂ in combination with TGF-β (Figure 7b). These results confirm that Wnt/β-catenin pathway activation did not occur when HT-29 cells are stimulated with PGE₂, suggesting that this is the differential regulator of cell invasiveness. It was interesting to see the suppressive effect of TGF-β on increased invasiveness caused by PGE₂, and we decided to verify what mechanism will be underlying this event. It is widely known that invasion is associated with the dynamic and regulation of actin cytoskeleton, and that non-Smad signaling of TGF-β induces changes in actin dynamics via rapid activation of RhoA GTPase (Souza-Squiavinato et al., 2019). Therefore, we analyzed the activity of this protein by using a specific G-LISATM RhoA Activation Assay Biochem KitTM, which showed that RhoA activity was increased in cells exposed to early time to PGE₂ and to TGF-β at 15 min. However, in the combined treatment with these two compounds, the RhoA activity was abolished (Figure S3).