Short-chain fatty acid receptors inhibit invasive phenotypes in breast cancer cells

Short chain fatty acids (2 to 6 carbons in length) are ubiquitous lipids that are present in human plasma at micromolar concentrations. In addition to serving as metabolic precursors for lipid and carbohydrate synthesis, they also act as cognate ligands for two known G protein-coupled receptors (GPCRs), FFAR2 and FFAR3. While there is evidence that these receptors may inhibit the progression of colorectal cancer, their roles in breast cancer cells are largely unknown. We evaluated the effects of enforced overexpression of these receptors in two phenotypically distinct breast cancer cell lines: MCF7 and MDA-MD-231. Our results demonstrate that both receptors inhibit cell invasiveness, but through different signaling processes. In invasive, mesenchymal-like MDA-MB-231 cells, FFAR2 inhibits the Hippo-Yap pathway and increases expression of adhesion protein E-cadherin, while FFAR3 inhibits MAPK signaling. Both receptors have the net effect of reducing actin polymerization and invasion of cells through a Matrigel matrix. These effects were absent in the less invasive, epithelial-like MCF7 cells. Correspondingly, there is reduced expression of both receptors in invasive breast carcinoma and in aggressive triple-negative breast tumors, relative to normal breast tissue. Cumulatively, our data suggest that the activation of cognate receptors by short chain fatty acids drives breast cancer cells toward a non-invasive phenotype and therefore may inhibit metastasis.


Introduction
Metastasis is a multi-stage process involving genetically unstable tumor cells that undergo phenotypic changes that allow them to egress from the primary tumor and colonize a distant tissue site with a favorable microenvironment [1,2]. Metastasis is responsible for 90% of cancer mortality, yet its molecular processes remain incompletely characterized [3]. One of the essential mechanisms involved in metastasis is epithelial-mesenchymal transition (EMT). EMT is a key a1111111111 a1111111111 a1111111111 a1111111111 a1111111111 antibody (ThermoFisher Scientific #R960-25). Secondary antibodies: Horseradish peroxidaseconjugated goat anti-rabbit (Cell Signaling Technologies #7074) and anti-mouse (Cell Signaling Technologies #7076), Alexa Fluor 594 (ThermoFisher Scientific #A11005) and Alexa Fluor 488 phalloidin (ThermoFisher Scientific #A12379).
To generate MCF-7 and MDA-MB-231 cell lines stably overexpressing either empty vector pcDNA3.1/V5-His (referred to as vector control) or the FFAR2-V5 and FFAR3-V5 (referred to as FFAR2-and FFAR3-overexpressing), respective plasmid constructs were linearized with restriction enzyme PvuI-HF following which they were transfected into the cells with Lipofectamine 3000 (Thermo Fisher #L3000015). The transfected cells were cultured in 1 mg/ml G418 to select for stable transfectants.

Western blot
Western analysis was performed essentially as described [35]. Cell lysate was prepared by treating the cells grown on 6-well tissue culture plates at varying time points for different experiments under starvation condition in serum-free DMEM. The cells were washed in ice cold phosphate-buffered saline (PBS) and then incubated with lysis buffer (1% Triton X-100/ 250 mM NaCL / 0.5 mM EGTA (Ethylene glycol tetraacetic acid) / 2 mM EDTA (Ethylenediaminetetraacetic acid) / 20 mM HEPES) containing protease and phosphatase inhibitors cocktail for 20 minutes on ice and harvested with a cell scraper. The cell suspension was centrifuged at 15,000 x g for 15 minutes at 4˚C. Protein concentration was quantified using a Bio-Rad protein assay dye (Bio-Rad #500006) and three parts of sample was diluted to one part of SDS-loading dye and denatured at 98˚C for 10 minutes. 20-25 μg protein was separated on a 10% SDS-polyacrylamide gel, transferred to PVDF membrane and blocked for 1 hour at room temperature in 5% nonfat milk in Tris-buffered saline containing Tween 20 (TBS-T). The blot was then incubated overnight with primary antibody at 1:2,000 dilution for anti-E-Cadherin, anti-YAP1, anti-ERK and at 1:10,000 for antiβ-actin; washed three times with TBS-T for 5 minutes each and incubated with horseradish peroxidase conjugated secondary antibody (1:10,000) for 2 hours; and visualized using the WesternBright ECL HRP substrate (Advansta, K-12045-D20). Image J software was used to quantify the images.

Immunoctochemistry
Immunocytochemistry was performed essentially as described [36]. Cells were cultured on collagen coated coverslips, fixed with 4% paraformaldehyde for 15 minutes, washed four times in 1x PBS, blocked for 1 hour in blocking solution (PBS/2.5% BSA/0.3% Triton X-100) at room temperature, incubated overnight with unconjugated primary antibodies in 10% blocking buffer, and washed with PBS. Coverslips were then incubated with secondary antibodies conjugated to Alexa Fluor dyes (Invitrogen) at 1 μg/ml in blocking solution for 1 hour with gentle rocking, then washed, incubated with 4',6-diamidino-2-phenylindole (DAPI) at 500 nM for 10 minutes, and then mounted on glass slides with Vectashield (Vector laboratories).

Invasion assay
The in vitro invasion assay was performed essentially as described [37]. Briefly, MDA-MB-231 (8 × 10 4 ) cells were seeded onto the top chamber of a Matrigel invasion chamber (Corning #354480) in serum-free media and the lower part of the chamber was in contact with 10% FBS containing culture media. The treatment where mentioned was carried out by adding 1 mM propionate diluted in plain DMEM to the upper chamber and culture media with 10% FBS was added to the lower chamber. Following a 24-hour incubation, the upper surfaces of the transwell chambers were gently wiped with cotton swabs and the invading cells were fixed in methanol, stained with crystal violet solution, washed, and air dried. The invading cell numbers were counted in five randomly selected microscope fields (20×).

Quantitative reverse transcription PCR (qPCR)
mRNA quantification was performed essentially as described [38]. Total RNA was isolated using Trizol reagent (Thermo Scientific) per manufacturer's instructions. Approximately 1 μg of each sample was primed with oligo(dT)18 and random hexamer primers prior using Maxima First Strand cDNA synthesis kit (Thermo Scientific). For quantitative real-time RT-PCR, targets were amplified with Maxima SYBR Green/ROX qPCR Master Mix (Thermo Scientific) on an Applied Biosystems ViiA 7 Real-Time PCR System (Thermo Scientific) using gene-specific primer pairs. Relative gene expression was determined using the 2 -ΔΔCT method as described [39].

TGFα shedding assay
This assay was conducted essentially as described [40,41]. Briefly, HEK293 cells were co-transfected with expression constructs for alkaline phosphatase-TGFα and the indicated receptor, along with chimeric G protein (Gq/i) using Lipofectamine 3000. Following a 24-hour incubation they were collected by trypsinization, washed with 1x PBS, and seeded into 96-well plates in Hank's balanced saline solution (HBSS). Stimulation was carried out for 1 hour following which alkaline phosphatase activity was detected in both cells and supernatant. Activity of the receptor (% shedding) was described as the alkaline phosphatase activity of the supernatant/ divided by the total alkaline phosphatase activity (cells + supernatant). Data processing and statistical analyses were performed with GraphPad Prism 6.

FFAR2 and FFAR3 differentially promote an epithelial phenotype
To understand the functions of FFAR2 and FFAR3 in breast cancer metastasis, we evaluated two breast cancer cell lines that have contrasting EMT phenotypes reflected by their EMT score: 1.0 = highly mesenchymal, -1.0 = highly epithelial [42]. MCF-7 cells are epithelial-like (EMT score = -0.412) and have high E-cadherin expression, while MDA-MB-231 cells are mesenchymal-like (EMT score = 0.454) and have low E-cadherin expression. The selected cell lines were stably transfected with V5-tagged FFAR2, FFAR3, or pcDNA3.1 vector control. The receptor over-expression was validated by immunocytochemistry and visualization with an anti-V5 antibody (S1 Fig).
As studies suggest that FFAR2 expression is reduced in certain cancers [30], and that proliferation may be inhibited by short chain fatty acids, butyrate and propionate [31,43], we speculated that short chain fatty acid receptors might play a role in regulating breast cancer progression, and possibly metastasis. To evaluate EMT in our engineered cell lines, we carried out a limited qRT-PCR screen to investigate the expression levels of the critical genes known to be regulators of EMT (Fig 1A). CDH1, the gene encoding E-cadherin, was significantly upregulated in FFAR2-expressing cells relative to control cells and FFAR3-expressing cells (Fig 1A). Furthermore, stimulation of cells with propionate, a potent and endogenous ligand for both FFAR2 and FFAR3 (S2 Fig) [44], resulted in a further increase in E-cadherin expression in FFAR2-expressing cells (Fig 1B). The increased mRNA expression of E-cadherin corresponded with increased protein expression (Fig 1C and 1D). However, FFAR2-expressing MCF-7 cells showed no change in the E-cadherin protein levels (S3 Fig), likely due to its high endogenous expression in this cell line. To confirm that this difference between the two cell lines was due to their EMT phenotypes, we repeated the experiment in two additional contrasting cell lines: BT-474 (epithelial-like, high E-cadherin expression, EMT score = -0.548) and MDA-MB-436 (mesenchymal-like, low E-cadherin expression, EMT score = 0.649). Again, FFAR2 overexpression resulted in a pronounced, ligand-mediated increase in E-cadherin expression in the mesenchymal-like cells, but not the epithelial-like cells (S4 Fig).
To determine whether these changes in E-cadherin expression correlated with changes to the invasive phenotype, cells were treated with propionate for 24 hours and evaluated for their ability to invade a Matrigel-coated matrix. FFAR2-and FFAR3-expressing cells demonstrated a ligand-dependent, significant decrease in invasiveness relative to vector control cells (Fig 2A  and 2B). Furthermore, cytoskeletal rearrangement is another notable feature of EMT [45]. Hence, we examined the morphological characteristics of the cytoskeleton by looking at the presence of actin stress fibers. We observed that both FFAR2-and FFAR3-expressing MDA-MB-231 cells showed significantly fewer cells with stress fibers as compared to the control cells (Fig 2C and 2D).

FFAR2 regulates E-cadherin by inhibition of YAP signaling
Since the related receptor FFAR1 has been shown to upregulate YAP1 signaling via Gαq and Gαi/o [46], we asked whether FFAR2 and FFAR3 similarly upregulates YAP1 signaling. Both FFAR2-and FFAR3-overexpressing cells showed significant increases in YAP1 phosphorylation (Fig 3A and 3B), consistent with downregulation of YAP1 activity, but interestingly this effect was ligand-dependent only in FFAR2-overexpressing cells. The kinase LATS1 is known to regulate YAP1 activity by phosphorylation at S127 [47]. To determine if the effect of FFAR2 and FFAR3 are mediated by LATS1, we performed a western blot analysis. We observed a significant increase in LATS1 only in FFAR2 cells, suggesting that FFAR2, but not FFAR3, regulates YAP1 via modulation of LATS1 expression (Fig 3C and 3D).
To determine whether this pathway is responsible for the FFAR2-mediated increase in Ecadherin expression, we performed siRNA knockdown of LATS1 in our engineered cell lines. This manipulation did not have a significant effect on E-cadherin protein level in control or FFAR3-overexpressing cells, but caused a significant reduction of E-cadherin in FFAR2-overexpressing cells (Fig 3E and 3F), restoring E-cadherin to control levels.
FFAR3 negatively regulates MAPK signaling FFAR2 and FFAR3 have been previously shown to regulate MAPK/ERK1/2 pathway [25] [48,49]. Therefore, we investigated whether FFAR2 and FFAR3 may be acting through MAPK/ ERK. Western blot analysis shows that FFAR3 expressing cells have a significant reduction in ERK phosphorylation, whereas FFAR2 expressing cells remained unchanged relative to control (Fig 4A and 4B).

Reduced expression of FFAR2 and FFAR3 in invasive and triple negative breast cancer tissues
Since our data suggest that FFAR2 and FFAR3 inhibit EMT, we speculated that expression of these receptors is likely to be reduced in invasive breast cancer tissues. Therefore, to determine the expression pattern of the SCFA receptors in breast tumors, we made use of Curtis breast dataset available on the Oncomine database (www.oncomine.com), which has a gene profile of 1992 breast carcinoma samples and 144 normal breast tissues [50]. Here, both FFAR2 and FFAR3 gene expression was significantly downregulated in invasive (Fig 5A and 5B) and triple negative breast cancer tissues (Fig 5C and 5D) when compared to normal breast tissues. Furthermore, to corroborate the involvement of FFAR2 in the regulation of E-cadherin expression,  we evaluated the correlation between FFAR2 and CDH1 mRNA levels in breast tumors and breast cancer cell lines (Fig 5E and 5F). Expression data for cells lines were obtained from the Cancer Cell Line Encyclopedia (https://portals.broadinstitute.org/ccle) [51]. Expression data for breast tumors were obtained from the Human Protein Atlas (www.proteinatlas.org) [52]. We observed a significant positive correlation between these two genes in both cell lines (Pearson r = 0.516) and breast tumors (Pearson r = 0.138).

Discussion
The prime goal of this study was to investigate the role of short chain fatty acid receptors (SCFA receptors) in breast cancer metastasis and the mechanism underlying this process. Since metastasis depends in part on the activation of EMT [53], we evaluated effects of FFAR2 and FFAR3 on this process. We observed that FFAR2 resulted in increased CDH1 (E-cadherin) expression specifically in mesenchymal-like MDA-MB-231 ( Fig 1A) and In contrast, E-cadherin remained unchanged by FFAR3 overexpression (Fig 1A and S3  Fig). This suggests that FFAR2, but not FFAR3, drives cells toward an epithelial phenotype. The absence of such an FFAR2-mediated effect in MCF-7 and BT-474 cells is likely due to their pre-existing epithelial phenotypes, which are characterized by high endogenous levels of E-cadherin expression [54]. That is, it is unlikely that the contribution of FFAR2 signaling would drive these epithelial cells to adopt further epithelial characteristics.
To confirm that the observed relationship between FFAR2 overexpression and E-cadherin expression was biologically relevant rather than an artefact of in vitro manipulation, we evaluated the relative expression patterns of FFAR2 and CDH1. If FFAR2 activity has a functional impact on the regulation of E-cadherin we would expect a positive correlation between these two genes. Indeed, the relationships are positive and highly significant in both cell lines and tumor samples (Fig 5E and 5F). The lower Pearson correlation coefficient observed with the primary tumors (0.138) relative to the cell lines (0.516) is likely the result of the heterogeneity of the tumor samples. Tumor samples contain varying degrees of non-tumor cells, and the net EMT status of the tumor is the result of highly multi-factorial influences. The presence of a positive, and highly significant correlation provides evidence that FFAR2 activity is one of these relevant factors.  The increased E-cadherin expression would be expected to correspond with a reduction in invasive phenotypes. Indeed, we observed a significant decrease in Matrigel invasion and stress fiber formation in FFAR2 expressing MDA-MB-231 cells. Surprisingly, similar phenotypes were also observed in FFAR3 expressing cells (Fig 2C and 2D), despite the fact that E-cadherin level was not changed. This suggested that although FFAR3 disrupts invasion, it is likely to be distinct from EMT. To elucidate alternative mechanisms, we evaluated the MAP/ERK and Hippo-Yap proliferative pathways. Previous studies demonstrated that both FFAR2 and FFAR3 may be involved in regulating the MAPK-ERK pathway [25,48,55], and possibly in the Hippo-YAP pathway [46]. We found that both FFAR2 and FFAR3 increased the phosphorylation of YAP1, but that the effect of FFAR3 was ligand-independent (Fig 3A and 3B). In contrast, we show that FFAR3, but not FFAR2, induces a significant decrease in ERK phosphorylation (Fig 4A and 4B). Since the ERK-MAPK pathway is necessary to regulate transcriptional factors crucial for EMT process [56], it is likely that the FFAR3-mediated downregulation of E-cadherin (Fig 1C) is mediated by inhibition of MAPK signaling.
Cumulatively, these results demonstrate that short-chain fatty acids can reduce the invasive potential of breast cancer cells through multiple signaling mechanisms, including the regulation of proliferative pathways, cytoskeletal organization, and expression of adhesion proteins (Fig 6). All of these processes are mediated by cognate receptors FFAR2 and FFAR3, and have the net effect of driving cells from an invasive, mesenchymal phenotype toward a quiescent, epithelial phenotype, thus potentially limiting metastasis.