Mouse colon cancer cell line CT26 was kindly provided by Dr I. J. Fidler (MD Anderson Cancer Center, Houston, TX, USA) and human colon cancer cell line HT29 was purchased from Dainihon Pharmacy, Co., Ltd., (Tokyo, Japan). Cells were routinely maintained in Dulbecco's modified Eagle's medium (DMEM; Sigma-Aldrich, St. Louis, MO, USA) supplemented with 10% fetal bovine serum (FBS; Sigma-Aldrich) in 5% CO₂ at 37°C. Cell morphology was evaluated daily by performing microscopic examination. Each cell line was routinely tested for mycoplasma contamination by performing genomic PCR. Before manuscript submission, viability of each cell line was tested by performing trypan blue exclusion assay.

EA (CAS no. 112-79-8; Wako Pure Chemical Industries Ltd., Osaka, Japan), OA (CAS no. 112-80-1; Wako Pure Chemical Industries), methyl-β-cyclodextrin (MbCD; Sigma-Aldrich), human epidermal growth factor (EGF; Peprotech EC Ltd., London, UK), PP1 (SRC inhibitor; Sigma-Aldrich), Wnt inhibitor (IWP3; Miltenyi Biotec GmbH, Bergisch Gladbach, Germany) (22), ERK inhibitor (3-[2-aminoethyl]-5-[{4-ethoxyphenyl}methylene]-2,4-thiazolidinedione; Santa Cruz Biotechnology, Santa Cruz, CA, USA) and p38 inhibitor (SB 203580; Abcam, Cambridge, MA, USA) were purchased.

The cells (1×10⁴/well) were seeded in a 12-well dish. Cell growth was assessed by performing 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) assay or by counting the number of cells with autocytometer (Sysmex, Kobe, Japan) after 48 h, as previously described (23). Apoptosis and necrosis were assessed using an apoptosis/necrosis detection kit (Enzo Life Sciences, Plymouth Meeting, PA, USA). Annexin V-EnzoGold and 7-AAD present in the kit fluorescently label apoptotic and necrotic cells, respectively. Apoptosis or necrosis was determined by examining 2000 cells under a fluorescent microscope (Leica Microsystems, Tokyo, Japan). All the experiments were performed in triplicate.

The cells (1×10⁷/well) were seeded on stem cell medium II (Sigma-Aldrich) in a bacteriological 3.5-cm dish (Becton-Dickinson Labware, Bedford, MA, USA) for 1 week. Number of spheres formed was counted by performing dark-field microscopy (Nikon, Tokyo, Japan).

Mouse mRNA expression was assessed by performing reverse transcriptase-polymerase chain reaction (RT-PCR) with 0.5 µg total RNA extracted using RNeasy kit (Qiagen, Hilden, Germany). Primer sets used for amplifying mouse CD133 and nucleostemin (NS; also known as guanine nucleotide-binding protein-like 3) are as follows: CD133 (NCBI reference sequence: NM_008935) forward, 5′-GAA AAG TTG CTC TGC GAA CC-3′ and CD133 reverse 5′-TCT CAA GCT GAA AAG CAG CA-3′; NS (NCBI reference sequence: NM_153547) forward, 5′-CAG GAT GCT GAC GAT CAA GA-3′ and NS reverse 5′-TTG ATT GCT CAG GTG ACA GC-3′. These primer sets were synthesized by Sigma Genosys (Ishikari, Japan). PCR products obtained were electrophoresed on a 2% agarose gel and stained with ethidium bromide. β-actin (ACTB) mRNA (GenBank accession no. NM_001101) was amplified for use as an internal control.

Cell spheres formed were fixed in 10% buffered formalin for 12 h at 4°C and were embedded in 1% agar. The agar specimens were processed into a paraffin-embedded block and 4-µm-thick sections were cut using a microtome. Consecutive tissue sections were immunohistochemically stained by following the immunoperoxidase technique, as previously described (24). Immunoblotting analysis was performed using 0.2 µg/ml anti-MIB1 and anti-NS antibodies (Dako Corp., Carpinteria, CA, USA) after performing antibody retrieval, in which the sections were heated three times in phosphate buffer (pH 6.0) in a microwave for 15 min each. Immunoreactivity was determined using 0.2 µg/ml secondary antibodies (Medical and Biological Laboratories, Co., Ltd., Nagoya, Japan). Color development was performed using diamine benzidine hydrochloride (Dako) and counterstaining was performed using Mayer's hematoxylin (Sigma-Aldrich). The number of NS- or MIB1-positive nuclei were counted and expressed as the mean number of positive nuclei from 50 high-power fields of tumor specimens.

BALB/c male mice (4-weeks old) were purchased from Japan SLC (Shizuoka, Japan). The mice were maintained according to the institutional guidelines approved by the Committee for Animal Experimentation of Nara Medical University and the current regulations and standards established by the Ministry of Health, Labor and Welfare. Each experimental group included five mice.

A subcutaneous tumor model was established by inoculating CT26 cells (1×10⁷ cells) suspended in Hank's balanced salt solution (Sigma-Aldrich) into the scapular subcutaneous tissue of the mice, which were fed with standard diet (CE-2; CLEA Japan, Tokyo, Japan). The fatty acid composition did not contain TFAs). For analyzing cancer metastasis to the liver, lungs and peritoneum, cancer cells (1×10⁶ cells) were inoculated into the spleen, tail vain and peritoneal cavity, respectively. EA or OA (10 mg/mouse) was administered by gavage every 7 days, and metastatic status was assessed after 4 weeks. The dosage of 10 mg/mouse/week (equivalent to 0.5 g/kg body weight/week) is approximately equal to 0.53 g/kg body weight/week, which is according to the Nutrition Facts label of foods 4.6 g per day (25). Peritoneal metastasis was determined by measuring the weight of the mesenterium with tumors, which was separated from the intestinal tube. The liver and lung metastases were determined by counting the number of tumors on organ surface.

Whole-cell lysates were prepared as previously described (24). Cell fractions were extracted by processing the cells with Cell Fractionation kit (Abcam), according to the manufacturer's instruction.

Proteins (50 µg) present in cell lysates were separated by performing sodium dodecyl sulfate-polyacrylamide gel electrophoresis on 12.5% gels and were electrotransferred onto nitrocellulose membranes for immunoblotting analysis. The membranes were incubated with primary antibodies, followed by incubation with peroxidase-conjugated IgG antibodies (MBL). Tubulin antibody (Proteintech Group, Inc., Rosemont, IL, USA) was used to measure the amount of protein loaded per lane. Immune complexes were visualized using CSA system (Dako). Antibodies against GPR40 and GPR120 (Abnova Corp., Taipei City, Taiwan); EGFR, E-cadherin (ECD) and β-catenin (Transduction Laboratories, Lexington, KY, USA); phosphorylated EGFR, NS, phosphorylated ERK1/2 (pThr²⁰⁴), phosphorylated SRC (pTyr⁵³⁰), Notch1, APC, Wnt5a and CD44 (Santa Cruz Biotechnology); phosphorylated p38 (pThr¹⁸⁰/pTyr¹⁸²; Cusabio, College Park, MD, USA); CD133 and Snail1 (pSer246) (Biorbyt, Cambridge, UK); and phosphorylated β-catenin (pSer^33/37; LifeSpan Biosciences, Inc., Seattle WA, USA) were used as primary antibodies.

Stealth Select RNAi oligonucleotides (siRNAs) against human and mouse GPR40 and GPR120 and against mouse EGFR were purchased from Santa Cruz Biotechnology. AllStars Negative Control siRNA was used as a control (Qiagen, Valencia, CA, USA). Cells were transfected with 20 nM siRNA by using Lipofectamine 2000 (Invitrogen Corp., Carlsbad, CA, USA), according to the manufacturer's recommendations.

Surface expression assay was performed to determine the effect of agonist-mediated GPR internalization. Briefly, monolayer-cultured CT26 cells in a 96-well dish were preincubated with peroxidase-labeled antibody against GPR40 or GPR120, followed by treatment with EA, OA (35 µM) or vehicle. At each time-point, the cells were counted by performing the MTT assay and were fixed in 4% paraformaldehyde for 15 min at room temperature. Surface levels of GPRs were measured by performing tetramethylbenzene reaction, followed by treatment with H₂SO₄ and measurement of absorbance at 450 nm.

Statistical analyses of experimental data were performed by unpaired Student's t-test and one-way ANOVA with post hoc test. Normal distribution of the data samples was confirmed by the Kolmogorov-Smirnov normality test. A two-sided P<0.05 was considered statistically significant.

As shown in Fig. 1, CT26 mouse CRC cells were inoculated into the subcutaneous tissue (Fig. 1A), spleen (liver metastasis) (Fig. 1B), tail vein (lung metastasis) (Fig. 1C) and peritoneum (Fig. 1D) of BALB/c mice by gavage administration with EA, OA, or control vehicle. EA-treated mice showed higher growth of subcutaneous tumors and metastases to the liver, lungs and peritoneum than control mice. In contrast, OA-treated mice only showed increased peritoneal metastasis compared with the control mice. Tumors of EA-treated mice showed increased mRNA expression of CD133 and NS (Fig. 1E). Moreover, the number of NS-positive tumor cells increased in the tumors of EA-treated mice but not in the tumors of control or OA-treated mice (Fig. 1F and G). These results suggest that EA increases the stemness of CT26 cells to enhance their metastatic potential.

Next, we examined the intracellular signals of EA (Fig. 2). EA treatment induced the phosphorylation of EGFR, ERK1/2 but not p38 (Fig. 2A). The effects of EA and OA on EGFR activation were compared (Fig. 2B–D). Phosphorylation of EGFR and ERK1/2 induced by OA was less pronounced than that induced by EA in both CRC cells. In contrast, phosphorylation of p38 induced by OA was more pronounced than that induced by EA. Importantly, EA treatment induced SRC phosphorylation at higher levels than those by OA treatment in both cell types.

GPR40 and GPR120 function as receptors of long chain fatty acids (26). We observed that CRC cells expressed different levels of GPR40 and GPR120 (Fig. 3A) and that CT26 cells showed higher GPR120:GPR40 ratio (5.3) than HT29 cells (1.2).

We knocked down genes encoding GPR40 and/or GPR120 and examined the effects of gene knockdown on the number of sphere-forming cells (Fig. 3B). Knockdown of the gene encoding GPR120 decreased the number of sphere-forming cells more profoundly than the knockdown of the gene encoding GPR40. Concurrent knockdown of both GPR40- and GPR120-encoding genes blocked EA-induced increase in the number of sphere-forming cells.

Binding affinity of EA to GPR40 or GPR120 was confirmed by examining the cell surface expression of these GPRs. Cell surface expression of both GPR40 and GPR120 was down-regulated by internalization for 2 h after EA treatment, after which it recovered (Fig. 3C). Results of the surface expression assay indicate that unlike EA, OA may preferentially bind to GPR40 (Fig. 3D).

Proto-oncogene tyrosine protein kinase (SRC) plays an important role in OA-induced EGFR transactivation (27). Therefore, we examined the role of SRC in EA-induced signal transduction. We observed that SRC inhibition suppressed the phosphorylation of EGFR and ERK1/2 in CT26 cells after concurrent treatment with EA and EGF (Fig. 3E). Moreover, knockdown of the gene encoding GPR40 or GPR120 decreased the phosphorylation of EGFR and ERK1/2 (Fig. 3E). Knockdown of the gene encoding GPR120 inhibited the phosphorylation of EGFR and ERK1/2 more profoundly than the knockdown of the gene encoding GPR40. We also observed these effects in HT29 cells (Fig. 3F). These findings suggest that EA activates GPR120 and (to a lesser extent) GPR40 in CRC cells for transactivating EGFR through SRC.

Next, we examined sphere formation as a feature of cancer stem cells (Fig. 4A and B). We observed that the number of spheres and MIB1-positive proliferating cells increased after EA treatment and decreased after knocking down the genes encoding GPR40 and GPR120. Treatment of CT26 cells with EA altered stem cell marker expression, i.e., it increased NS expression and decreased CD133 expression (Fig. 4C). Moreover, EA treatment decreased Notch and Notch intracellular domain (NICD) expression and induced β-catenin nuclear translocation associated with EMT, as indicated by the upregulation of Snail expression and downregulation of E-cadherin expression. Thus, EA signaling inhibited Notch signaling and activated canonical Wnt pathway in CT26 cells. In contrast, EA treatment of HT29 cells harboring truncated APC (28), which showed constitutively activated canonical Wnt signaling, increased NS expression and decreased CD133 expression (Fig. 4D). In addition, EA treatment of HT29 cells decreased Notch and NICD expression and increased Wnt5a and CD44 expression. These results indicate that EA signaling inhibits Notch signaling and activates non-canonical Wnt signaling in HT29 cells.

To confirm that EA induced the stemness and metastatic ability of CRC cells, we examined the effects of Wnt and ERK1/2 inhibitors by performing the sphere assay (Fig. 5A). Treatment with Wnt and ERK1/2 inhibitors completely abolished EA-induced increase in sphere number and liver metastasis.

To confirm EA signaling pathway, NS expression in CT26 cells was examined at the ligand (Fig. 5B), receptor (Fig. 5C), and post-receptor signaling levels (Fig. 5D). At the ligand level, EA treatment increased NS expression to a similar level as that after EGF treatment, suggesting that EGFR transactivation induced by EA was as effective as that induced by EGF. At the receptor level, inhibition of the crosstalk linker SRC and knockdown of GPR40 and GPR120 suppressed EA-induced NS expression, which was as effective as that induced by the knockdown of EGFR. At the post-receptor signaling level, inhibition of ERK1/2, p38 and Wnt suppressed EA-induced NS expression.