Protein kinase C α enhances migration of breast cancer cells through FOXC2-mediated repression of p120-catenin

All cells were maintained in a humidified incubator with 5% CO₂ at 37 °C. MCF7 cells were originally obtained from the Michigan Cancer Foundation (Detroit, MI) in 1992 and T47D cells were originally obtained from ATCC in 1996; both cell lines were stored at early passage. T47D:A18, a hormone-responsive clone, has been described previously [33]. T47D:A18 and MCF7 cells were cultured in RPMI with 10% FBS. MCF7:TAM1 [34], MCF7/PKCα [35, 36], MCF7:5C [37] and T47D:C42 [33] are hormone-independent and endocrine-resistant clones that were previously described. MCF7:TAM1 and MCF7/PKCα were cultured in RPMI with 10% FBS supplemented with 4-hydroxytamoxifen (4-OHT, 10⁻⁷ M) and G418 (100 μg/mL) respectively¹. MCF7:5C and T47D:C42 were cultured in phenol red-free RPMI with 10% charcoal stripped FBS [33, 37]. Before experiments, estrogen-dependent cell lines were stripped in phenol red free media for 3 days. TNBC cell lines HCC1937 (CRL 2336™) and HCC1143 (CRL 2321™) were obtained from ATCC (Manassas, VA, USA). They were cultured and passaged in RPMI with 10% FBS according to the ATCC’s instruction. The TNBC cell line MDA-MB-231 (CL#10A) was cultured in MEM supplemented with 10% FBS. All cell culture reagents were obtained from Life Technologies (Carlsbad, CA, USA). Cell lines were tested negative for Mycoplasma contamination (MycoAlertTM Mycoplasm Detection Kit, Lonza Ltd., Walkersville, MD, USA), and were authenticated using Short Tandem Repeat (STR) method by the Research Resource Center core at the University of Illinois at Chicago (Chicago, IL, USA) in 2016. For TPA treatment, cells were treated with 100 nM for 2 h before mRNA was collected and analyzed.

Whole cell extracts of cultured cells were prepared in lysis buffer (Cell Signaling Technology, Danvers, MA, USA) supplemented with the protease inhibitor phenylmethane sulfonyl fluoride (PMSF). Protein concentration was determined by the bicinchoninic acid assay (BCA) (Thermo Fisher Scientific, Waltham, MA, US) and separated on SDS-PAGE gel. The following antibodies and dilution factors were used: PKCα (1:200, Santa Cruz Biotechnology, Santa Cruz, CA, USA), E-cadherin (1:1000, Cell Signaling Technology, Danvers, MA, USA), p120-catenin (1:200, Santa Cruz Biotechnology, Santa Cruz, CA, USA), FOXC2 (1:1000, Abcam, Cambridge, MA, USA). β-actin (1:1000, Sigma-Aldrich, St.​ Louis, MO, USA) was used as loading control. Blocking agents were either 5% non-fat dry milk or 5% bovine serum albumin (BSA) depending on the specific antibody. Mouse and rabbit horseradish peroxidase-conjugated secondary antibodies were purchased from GE Healthcare Life Sciences (Pittsburgh, PA, USA) and used at a 1:2000 dilution factor. Images of blots were acquired on a Bio-Rad ChemiDoc System following incubation with SuperSignal West Dura luminol solution (Thermo Fisher Scientific, Waltham, MA, USA). Protein bands were quantified using densitometry measured in Quantity One (Bio-Rad, Hercules, CA, USA). When necessary, membrane was stripped using Restore Western Blot Stripping Buffer (Thermo Fisher Scientific, Waltham, MA, USA).

Corning ^® transwell inserts (Corning Inc., Corning, NY, USA) were used for the migration and invasion assays following the manufacturer’s instruction. For invasion assay, inserts were coated with reconstituted Corning ^® Matrigel ^® Growth Factor Reduced (GFR) Basement Membrane (Corning Inc., Corning, NY, USA) and incubated for 2 h at 37 °C. Cells (1 × 10⁵) were plated in the upper chamber and FBS was used as the chemoattractant in the bottom chamber. For the experiments that involved MCF7/PKCα, fibroblast-conditioned media was used as the chemoattractant instead because we found FBS to be inhibitory to their migration and invasion (data not shown). After overnight incubation, inserts were fixed in ice cold 100% methanol and stained with a 0.2% crystal violet/ 2% ethanol solution. Following staining, inserts were rinsed with water and allowed to air dry before imaging. Total number of migrated and invasive cells/well was counted with 100X total magnification light microscopy. At least four areas per well were counted and averaged for analysis. Graph represents the fold change of number of migrating or invading cells relative to the control as explained in the legend.

mRNA was extracted by Trizol^® reagent (Thermo Fisher Scientific, Waltham, MA, USA) and purified following the manufacturer’s instruction. mRNA was reverse transcribed using the High Capacity cDNA Reverse Transcription kit (Applied Biosystems, Foster City, CA, USA). Detection of transcripts was done using a SYBR green reaction mixture in the StepOne Plus Real Time PCR Machine (Applied Biosystems, Foster City, CA, USA) using the standard amplification and detection protocol. Primer sequences are shown in Table 1.

Cells were transfected with 50 nM (C_f) siRNA targeting PKCα or FOXC2 following the manufacturer’s instruction. PKCα siRNA was purchased from Dharmacon (Lafayette, CO) (ON-TARGET plus SMARTpool) and Sigma Aldrich (predesigned, lab-validated siRNA). FOXC2 siRNA was purchased from Dharmacon (Lafayette, CO) (ON-TARGET plus SMARTpool) and IDT (San Jose, CA, USA) (Dicer-substrate, lab-validated siRNAs). Media was changed 24 h following transfection and every 3–4 days for the duration of the experiment. Efficiency of siRNA knockdown was confirmed with either qRT-PCR or Western blot. siRNA sequences are shown in Table 2. Specificity of PKCα siRNA is shown in Additional file 1: Figure S1.

The p120-catenin short luciferase reporter construct was kindly provided by Dr. Fariborz Mortazavi (Department of Medicine, David Geffen School of Medicine, University of California, Los Angeles, CA). To assess p120-catenin promoter activation, cells were co-transfected with p120-catenin reporter construct and β-galactosidase using Lipofectamine ^® 2000 (Thermo Fisher Scientific, Waltham, MA, USA). Luciferase activity was measured using the Dual Luciferase Reporter Assay (Applied Biosystems, Foster City, CA) and normalized against the activity of β-galactosidase following the manufacturer’s instructions.

Cells (2–4 × 10⁵) were seeded on coverslips in 6 well plates to reach 80% confluence in 2 days. Cells were fixed by incubating with 4% paraformaldehyde in PBS, pH 7.4 for 10 min at room temperature, and washed three times with ice-cold PBS. Permeabilization was achieved with 0.1% Triton-100X in PBS for 1 min. After three PBS washes, cells were incubated with blocking buffer (10% normal goat serum (Cell Signaling Technology, Danvers, MA, USA) in 1X PBS) for 1 h at room temperature, followed by overnight incubation with primary antibody in a humidified chamber at 4 °C. On the next day, coverslips were rinsed three times with wash buffer (0.1% BSA in 1X PBS), followed by 1 h incubation with secondary antibody for 1 h at room temperature. The following antibodies were used: mouse E-cadherin (Cell Signaling Technology, Danvers, MA, USA), rabbit p120-catenin, rabbit PKCα (Santa Cruz Biotechnology, Santa Cruz, CA, USA), anti-rabbit IgG (H + L), F(ab’)₂ Fragment (Alexa Fluor^® 488 Conjugate) (Cell Signaling Technology, Danvers, MA USA), anti-mouse IgG (H + L), F(ab’)₂ Fragment (Alexa Fluor^® 555 Conjugate) (Cell Signaling, Danvers, MA, USA). Following the manufacturer, all antibodies were used at 1:100 dilution factor: primary antibodies were diluted in blocking buffer (10% normal goat serum/PBS) and secondary antibodies were diluted in dilution buffer (1% normal goat serum/PBS). Coverslips were then incubated with Prolong® Gold Antifade Reagent with DAPI (Cell Signaling, Danvers, MA, USA) overnight. Images were obtained by the Zeiss Laser Scanning Microscope (LSM) 710 at the Core Imaging Facility at the University of Illinois at Chicago (Chicago, IL, USA). Intensity quantification was done using ImageJ.

Protocol was optimized from a protocol previously described by Carey et al. [38]. Specifically, 80-100μg of chromatin was incubated with either FOXC2 (ChIP grade, Abcam, Cambridge, MA, USA) or the negative control IgG (Cell Signaling Technology, Danvers, MA, USA) overnight at 4 °C. The antibody-DNA complex was captured by Protein G Agarose/ Salmon Sperm DNA bead (Millipore, Billerica, MA, USA). DNA was purified and analyzed by qRT-PCR using the previously reported primers that recognize p120-catenin promoter region ((20) as shown below. Primers that recognize the upstream and downstream region from the reported binding site (+127 to +309) of FOXC2 on p120-catenin were used as negative controls Table 3.

Oncomine™ (Compendium Bioscience, Ann Arbor, MI, USA) overexpression analysis was used to probe for PRKCA (PKCα) and FOXC2 expression in mRNA datasets. P values less than 0.05 were considered significant.

For the generation of PRKCA, FOXC2, and CTNND1 heat map, the TCGA data, analyzed using the AgilentG4502A_07_3 array platform, were obtained from the UC Santa Cruz platform (https://​genome-cancer.​ucsc.​edu). All samples were intrinsically classified by PAM50 assay and the expression of ER, PR, and HER2. They were then stratified based on the relative transcripts expression of the selected gene (PRKCA, FOXC2, and CTNND1).

To investigate the relationship between PKCα, FOXC2 and p120-catenin, we first examined the Oncomine™ database for relative transcript levels of PRKCA (encoding for PKCα) and FOXC2. In four independent reports examining TNBC samples, both PRKCA and FOXC2 rank among the top 10% of genes associated with the TNBC subtype (Additional file 2: Figure S2). These results are in agreement with previous reports that TNBC tumors express high PKCα [29, 30] and FOXC2 protein expression [21, 22]. Whereas FOXC2 was previously demonstrated to be a repressor of p120-catenin expression in lung cancer [20], it is not known whether this inverse relationship holds true in breast cancer. Kaplan-Meier analyses on two independent datasets, GSE22219 [39] and GSE42568 [40], support the hypothesis that in patients whose tumors lack ER expression, high FOXC2/CTNND1 (p120-catenin) ratio (high FOXC2, low CTNND1) was associated with shorter relapse free survival (RFS) (P < 0.001 and P = 0.08 for GSE22219 and GSE42568 respectively) (Fig. 1a). Using microarray data provided by The Cancer Genome Atlas (TCGA) dataset, we were able to evaluate the expression levels of PRKCA (PKCα), FOXC2, and CTNND1 in breast cancer patients (http://​genome.​ucsc.​edu/). Patients were classified molecularly as normal-like, luminal B, luminal A, HER2-enriched, and basal-like [41], as well as by ER, PR, and HER2 expression. Relative expression of PRKCA, FOXC2, and CTNND1 in all tumor samples, using the AgilentG4502A_07_3 array platform, were computed and visualized by a heat map. Overall, patients whose tumors are classified as basal-like and/or as TNBC express higher levels of PRKCA and FOXC2 along with lower levels of CTNND1 when compared to tumors of other subtypes (Fig. 1b). Interestingly, in the same GSE22219 and GSE42568 datasets, high FOXC2/CTNND1 ratio also indicated a trend for reduced RFS for ER⁺ patients although the association is weaker than that in ER⁻ patients (P = 0.3 and 0.13 for GSE22219 and GSE42568 respectively) (Fig. 1c). Together, these data prompted us to further examine the functional consequence of the PKCα, FOXC2, and p120-catenin relationship in breast cancer at the molecular level.

We examined the expression pattern of PKCα and FOXC2 in ER⁺ and TNBC breast cancer cell lines. Among ER⁺ cell lines, T47D:A18 and MCF7 cell lines are sensitive to endocrine treatment (such as tamoxifen) whereas T47D:C42, MCF7/PKCα, MCF7:TAM1 and MCF7:5C are all resistant to endocrine treatment as previously described [33, 34]. TNBC cell lines HCC1143 and HCC1937 (basal A) and MDA-MB-231 (basal B) were chosen based on molecular profiling [31, 32]. The basal B subgroup is reported to be highly enriched with EMT and stem cell signatures whereas basal A cell lines are characterized by upregulation of ETS- and BRCA-related pathways [31, 32]. Compared to basal B, the basal A subgroup is reported to better reflect the biology of the clinical basal-like breast cancer [31]. PKCα and FOXC2 are expressed in all endocrine resistant and basal TNBC (A and B) cell lines and among ER⁺ breast cancer cell lines, the endocrine resistant cells (T47D:C42, MCF7/PKCα, MCF7:TAM1, and MCF7:5C) have higher expression of PKCα compared to their endocrine sensitive counterparts T47D:A18 and MCF7 (Fig. 2a). When compared to their endocrine sensitive parental cell lines MCF7 and T47D:A18, MCF7/PKCα and T47D:C42 are more migratory and MCF7/PKCα cells are more invasive compared to MCF7 (Fig. 2b and Additional file 3: Figure S3). Interestingly, characteristics of enhanced migration and invasion observed in MCF7/PKCα partially correlate with markers consistent with an EMT (Fig. 2c). Specifically, MC7/PKCα cells show down-regulation of epithelial markers ZO-1 and E-cadherin compared to MCF7, however, elevated expression of mesenchymal markers Vimentin, N-cadherin, and P-cadherin is not observed (Fig. 2c). This result suggests that MCF7/PKCα cells have not undergone a complete EMT and perhaps this is not necessary for cancer cells to acquire a migratory and invasive phenotype.

Interestingly, either PKCα or FOXC2 knockdown was sufficient to reduce the migratory and invasive capabilities of MCF7/PKCα cells (Fig. 2d). Similarly, PKCα or FOXC2 knockdown in basal A cell lines HCC1937 and HCC1143 resulted in significantly lower migration and invasion capabilities (Fig. 2d). Therefore, we concluded that the positive contribution of PKCα and FOXC2 on migration and invasion can be extended beyond the scope of basal B TNBC [21, 22].

To assess a possible relationship between PKCα and FOXC2, all cell lines were treated with 12-O-tetradecanoylphorbol-13-acetate (TPA), an activator of several PKC family members, including PKCα. TPA treatment results in relocation of PKC isoforms from the cytoplasm to the cell membrane, indicative of activation. Indeed, upon TPA treatment, we observed a clear translocation of PKCα from the cytoplasm to the cell membrane in two representative cell lines (MCF7 and MCF7/PKCα) (Additional file 4: Figure S4). Following TPA treatment of endocrine resistant ER⁺ (T47D:C42, MCF7/PKCα) and basal A TNBC cell lines (HCC1143, HCC1937) we observed a significant induction of FOXC2 expression as measured by qRT-PCR (Fig. 3a). Correspondingly, PKCα knockdown using siRNA was sufficient to reduce FOXC2 expression at both the transcript and protein level in these cell lines (Fig. 3b). In contrast, TPA treatment did not have any effect on FOXC2 expression in either endocrine sensitive (T47D:A18, MCF7) (Fig. 3a) or basal B TNBC cell line (MDA-MB-231) (Additional file 5: Figure S5), suggesting a relationship between PKCα and FOXC2 in these two subtypes is unlikely. Altogether, our findings suggest that PKCα is a positive regulator of FOXC2 expression in endocrine resistant and basal A TNBC subgroups.

Loss of E-cadherin has been recognized as a characteristic of the transition from benign lesions to invasive, metastatic cancer [42]. At the molecular level, loss or reduction of E-cadherin expression precedes and is often causal to the dissociation of other members of the AJ, signifying the dissolution of intercellular adhesion [42, 43]. In agreement with the observation that PKCα enhances breast cancer cell motility (Fig. 2d), we examined the effect PKCα has on AJ components. PKCα knockdown resulted in a significant increase in E-cadherin and p120-catenin protein expression (Fig. 4a), suggesting that PKCα is a repressor of the two proteins. The increase of p120-catenin protein upon PKCα knockdown correlated with an increase in p120-catenin transcripts (Fig. 4b). However, no changes in E-cadherin transcripts were observed (Fig. 4b). This result suggests that E-cadherin repression by PKCα is not a transcriptional event and more likely a result from reduced protein stability. As loss of p120-catenin was previously reported to result in a transcription-independent reduction of E-cadherin [13, 20], we reasoned that PKCα-mediated repression of p120-catenin may be the underlying mechanism for E-cadherin loss. To address this hypothesis we examined p120-catenin and E-cadherin protein expression by immunofluorescent staining following PKCα knockdown. We determined that p120-catenin was recovered and localized at the cell membrane at 72 h after siRNA transfection, followed by a recovery of E-cadherin at approximately 24 h later (Fig. 4c). Quantitatively, we show that p120-catenin significantly recovered at an earlier time point than E-cadherin, supporting the notion that E-cadherin recovery is a downstream effect of p120-catenin recovery.

FOXC2 was reported to be a transcriptional repressor of p120-catenin in NSCLC cell lines [20]. We sought to determine if the inverse relationship between FOXC2 and p120-catenin is also true in our breast cancer cell lines. FOXC2 knockdown in two representative cell lines, MCF7/PKCα and HCC1937, efficiently rescued p120-catenin expression at both the transcript and protein level (Fig. 5a). Furthermore, FOXC2 knockdown significantly increased the p120-catenin promoter activity as determined using a luciferase reporter construct (Fig. 5b). These data suggest that FOXC2 suppresses p120-catenin expression by repressing its transcription. It is noteworthy that PKCα knockdown did not result in recovery of p120-catenin expression in either MCF7 (Fig. 5c) or MDA-MB-231 (Additional file 5: Figure S5) even though both cell lines co-express PKCα and FOXC2 (Fig. 2a). Accordingly, we detected a significant enrichment of FOXC2 occupancy on the p120-catenin promoter in all cell lines representing either endocrine resistant or basal A breast cancer subtypes, but not in endocrine sensitive MCF7 (Fig. 5d). The interaction between FOXC2 and p120-catenin seems to take place within the +127 to +309 region of the p120-catenin promoter, as we were not able to detect FOXC2 binding either downstream or upstream from this region (Additional file 6: Figure S6). Finally, we found that FOXC2 binding to p120-catenin likely depends on PKCα expression because PKCα knockdown significantly reduced FOXC2 enrichment on p120-catenin (Fig. 5e). These findings cumulatively support the hypothesis that PKCα is a novel regulator of FOXC2-mediated repression of p120-catenin in breast cancer. Specifically, PKCα down-regulates p120-catenin by sustaining the expression and activity of FOXC2, a p120-catenin repressor. By repressing p120-catenin, PKCα promotes E-cadherin down-regulation and dissolution of the AJ, which impairs intercellular adhesion and promotes cellular migration. Specifically we found that this signaling axis is relevant in two breast cancer subtypes: endocrine resistant ER⁺ and basal A TNBC. Endocrine sensitive ER⁺ and basal B TNBC, despite being positive for PKCα and/or FOXC2, do not rely on PKCα for the repression of p120-catenin: PKCα knockdown in either MCF7 (endocrine sensitive) or MDA-MB-231 (basal B TNBC) was not sufficient to recover p120-catenin expression (Fig. 5c and Additional file 5: Figure S5).