FAs, FA-free bovine serum albumin (BSA), and forskolin were purchased from Sigma-Aldrich (St. Louis, MO). Fura2/AM was purchased from Dojindo (Kumamoto, Japan). G1 and G15 were purchased from TOCRIS Bioscience (Ellisville, MI) and Merck (San Diego, CA), respectively. YM254890 was from Yama-nouchi Pharmaceutical Co., Ltd. (Ibaraki, Japan).

ES2 and SKOV3 cell lines were obtained from the American Type Culture Collection (ATCC). ES2 and SKOV3 cells were cultured in McCoy’s 5A medium supplemented with 10% fetal bovine serum and antibiotics. Cells were cultured in Opti-MEM (Gibco, Grand Island, NY) without fetal bovine serum. FAs were first dissolved in pure ethanol and then freshly diluted to the culture medium containing 1% BSA. The ethanol and BSA were 0.05% (v/v) and 0.1% (v/v), respectively.

For the assay of ES2 and SKOV3 cell proliferation, 5×10³ cells per well in Opti-MEM with different types and concentrations of FAs were seeded in triplicate, and cell numbers were monitored after 24, 48, and 72 hours using the Cell Counting Kit-8 (Dojindo). The cytotoxicity of various FAs at different concentrations on ES2 and SKOV3 cells was measured with the Cytotoxicity LDH Assay Kit-WST (Dojindo). All experiments were performed in triplicate wells for each condition and repeated three times independently.

A total of 0.5×10⁶ cells were cultured into 6-well culture plates and cocultured with or without 300 μM EPA for 48 hours. For cell apoptosis analysis, cells were determined with the Annexin V/PI Apoptosis Detection Kit (Dojindo) and analyzed by a FACS^Calibur (BD Bioscience, San Jose, CA).

Cells were trypsinized in 3.5-cm Petri dishes, centrifuged, and resuspended in phosphate buffered saline (PBS) with 3 mM Fura-2/AM. Cells were treated with the Ca²⁺ indicator Fura-2/AM (5 μM) to measure [Ca²⁺]i. After washed for three times to remove extracellular Fura-2/AM, the Fura-2 was detected using Felix fluorescence data acquisition as previously shown [9]. The dose of YM254890 was based on previous findings [10,11].

Cells were seeded and treated with G15 or GPR30 small interfering RNA (siRNA). The dose of G15 was based on Wang et al. [12] and Bai et al. [13], who found that the 8 μM dose would be sufficient to obtain antagonistic effects on GPER1/GPR30 without affinity for other estrogen receptors. Twenty-four hours later, the cells were seeded in Opti-MEM medium for 5 hours and then treated with BSA, palmitic acid, or EPA. Cells were washed with PBS twice and frozen and thawed three times after treatment. The cAMPs were quantified with an Enzyme Immu-noassay Kit (Cayman Chemical, Ann Arbor, MI).

The nitrocellulose membrane was spotted with EPA (1, 3, and 5 mM) and then blocked in buffer as previously reported [14]. Next, the membrane was incubated with purified GFP-GPR30 protein (100, 200, and 400 ng/mL) overnight at 4°C, then scanned by Typhoon FLA9500 (GE Healthcare, Uppsala, Sweden) to detect GFP.

Ovarian cancer xenograft models were generated in old female nude mice at 4 to 6 weeks of age by injecting 1×10⁷ ES2 cells treated with or without adenoviral GPR30 shRNA into subcutaneous fat. Animals were divided into four groups with 10 per group. For the experimental group, 10 μL EPA was added to 90 μL aqueous vehicle. For the control group, 10 μL ethanol was added to 90 μL aqueous vehicle. Mice were given EPA (4 μg) once daily. For the treatment of G15, ovarian cancer xenograft models were generated in old female nude mice at 4 to 6 weeks of age by injecting 1×10⁷ ES2 cells into subcutaneous fat. Animals were divided into four groups with 10 per group. Nude mice bearing ovarian tumors were received dimethyl sulfoxide (DMSO) in combination with MeOH as a control, EPA in combination with DMSO, MeOH in combination with G15 or EPA in combination with G15. Mice were given EPA (4 μg) once daily, while G15 (10 μg) once daily from day 5. Tumor volumes were calculated as 1/2×length×width². Ki67 and GPR30 were stained to assay the inhibitory effects of EPA and GPR30 expression, respectively.

Ovarian cancer patient (n=173) samples were collected from the Department of Pathology of Nanjing Drum Tower Hospital. The tissue microarray (TMA) was made as previously reported [15]. The immunohistochemical stains were assessed by three separate observers who did not know patient characteristics. The classification of GPR30 expression level was based on the staining intensity of GPR30. All protocols for humans were reviewed and approved (kyy2015-38). The detailed information of TMA was provided in S1-S3 Tables.

Quantitative real-time PCR was performed using SYBR Select Master Mix (Vazyme, Nanjing, China) in an ABI 7300 sequence detector. The results were normalized to the 18S rRNA gene expression level in each sample. The primers were from PrimerBank (http://pga.mgh.harvard.edu/primerbank/).

The antibodies used were cleaved caspase-3, AKT, p-AKT, ERK, p-ERK, phosphorylated protein kinase A (p-PKA) substrate (Cell Signaling Technology, Danvers, MA), actin (Sigma-Aldrich), and GPR30 (Santa Cruz Biotechnology, Santa Cruz, CA).

The siRNA against GPR30 was GCACCUGUGGCUGACGAAUUU. ES2 cells were transfected with RNA interference (RNAi) oligonucleotides using Lipofectamine RNAiMAX (Invitrogen, Carlsbad, CA).

The GPR30 shRNA sequence was GCACCTGTGGCTGACGAATTT. Adenovirus was amplified and purified using Sartorius Adenovirus Purification kits (Sartorius, Germany).

Data from at least three independent experiments were analyzed and are expressed as means±standard deviation. The nonpaired Student’s t test was used for these analyses. A difference was considered significant at p < 0.05, p < 0.01, and p < 0.001. R was used for the correlation analysis of GPR30 expression level with overall survival of ovarian cancers patients. Then packages “survival” and “survminer” were used for survival analysis using Kaplan-Meier method and generate the survival curve.

All animal procedures were carried out in accordance with the approval of the Animal Care (#CS20) and Use Committee at the Model Animal Research Center of Nanjing University in Nanjing.

FAs have an effect on cancer by altering cell membrane synthesis and saturation and via cholesterol lipid hormones that affect the function and distribution of signaling macromolecules [6]. Thus, we speculated that various FA treatments might influence different aspects of cancer development in ovarian cancer cells. Ovarian carcinoma is divided into four major histological types: clear cell, mucinous, serous, and endometrioid [16]. OCCCs makes up 25% of ovarian carcinoma in Japan, and the lethality rate of adenocarcinoma is also high [16]. Compared with other types of ovarian carcinomas, OCCCs have a poorer prognosis and are more frequently platinum resistant [17]. Here, we selected the ES2 OCCC and SKOV3 adenocarcinoma cell lines for study. ES2 and SKOV3 were first treated with several FAs at different doses for 24 hours (0-300 μM). We found that palmitic acid, a saturated FA, obviously enhanced cell growth, while EPA inhibited ES2 and SKOV3 cell proliferation in a dose-dependent manner (Fig. 1A), which is in accordance with the published anticancer effects of EPA [18]. However, EPA promoted cell proliferation in CAOV3 and A2780 cell lines, indicating the antitumor effect of EPA may be cell line specific event (S4 Fig.). We also analyzed the toxicity of palmitic acid and EPA on ES2 and SKOV3 cells by using a cytotoxicity assay. We found that cell toxicity affected ES2 cells at a 400 μM concentration, while the effect on SKOV3 cells was minor at 400 μM (Fig. 1B). Thus, 300 μM was an appropriate concentration for use in subsequent experiments. The pro-proliferation effect of palmitic acid and the inhibitory effect of EPA on ES2 and SKOV3 cells were also time-dependent. Palmitic acid treatment increased cell growth in both ovarian cancer cell lines. However, the growth of ES2 cells was significantly decreased when treated with EPA compared to that of SKOV3 cells (Fig. 1C), suggesting that EPA had a much more pronounced effect in ES2 cells than in SKOV3 cells.

As previously reported, n-3 PUFAs (docosahexaenoic acid or EPA) inhibit colorectal cancer cell proliferation by inducing cell apoptosis [19]. Therefore, we detected cell apoptosis caused by EPA treatment in ES2 cells. The flow cytometry results showed that apoptotic cells were approximately 14.4% in EPA-treated ES2 cells, more than that in control (0.03%). The protein expression analysis also confirmed that EPA led to cleaved caspase-3 activation (Fig. 1D). mRNA expression analysis confirmed that EPA-induced proapoptotic genes activation and antiapoptotic genes inhibition in ES2 cells (Fig. 1E). Importantly, the cytotoxicity and apoptosis of normal ovarian epithelial cells were not affected by 300 μM EPA treatment (Fig. 1F and G). Together, these data indicate that EPA induces ovarian cancer cell apoptosis.

Previous findings suggest the receptors of long-chain FAs are GPR40 and GPR120 [20], and these GPR agonists could induce cell proliferation [21]. There are four subfamilies of α subunits of G proteins: G_αs, G_αi, G_αq, and G_α12. GPR40 and GPR120 couple to the αq subfamily of G proteins, which can be inhibited by YM254890 [10], increasing the downstream effector Ca²⁺ [22]. To determine whether GPR40 and GPR120 mediate the effects of palmitic acid and EPA in ES2 cells, we measured changes in [Ca²⁺] in cells following palmitic acid or EPA treatments in the absence and presence of the Gq inhibitor YM254890, which can completely block the Gq pathways [10]. The modest increase in Ca²⁺ that we detected could be attributed to G_βγ, which could have activated PLCβ on the cell membrane and finally cause Ca²⁺ release from the endoplasmic reticulum. Without YM254890 treatment, palmitic acid and EPA activated [Ca²⁺] increases of 4- and 1.5-fold, respectively (Fig. 2A). Interestingly, with YM254890 treatment, the increase in [Ca²⁺] by EPA did not change, while the increase in [Ca²⁺] by palmitic acid was inhibited (Fig. 2B). These results indicate that palmitic acid functions mainly through the Gq pathway, while EPA might have its effects through other G_α pathways.

Similarly, we studied the effect of YM254890 on palmitic acid proliferation as well as the EPA antiproliferative effect in ES2 cells. As shown in Fig. 2C, YM254890 decreased palmitic acid-induced proliferation by 15% (Fig. 2C). However, YM254890 did not affect EPA-induced antiproliferation or proapoptosis effects (Fig. 2C and D). Consistent with the data above, the expression of proapoptotic genes, antiapoptotic genes, and cleaved caspase-3 as influenced by EPA was not changed by YM254890 (Fig. 2E). The results above suggest that GPR40 or GPR120, whose subunit is G_αq, is not involved in the anticancer effect of EPA in ES2 cells. Unlike the G_αq family, whose downstream effect is increasing [Ca²⁺], G_αs increases but G_αi decreases the level of cAMP [22]. To identify which G_α subunit functioned when ES2 cells were treated with EPA, we first measured the production of cAMP in ES2 cells treated with palmitic acid or EPA. The results showed that EPA but not palmitic acid increased cAMP compared to vehicle (Fig. 2F), suggesting that the EPA receptor is coupled to G_αs protein.

Based on the above results, we hypothesized that there exists some GPR whose G_αs subunit plays a dominant role in the EPA effect in ES2 cells. Thus, we detected various GPR proteins expression levels in ES2 cells. Quantitative PCR analysis showed that GPR30, coupled to G_αs protein [23], was significantly higher than other GPRs (Fig. 2G). We also detected and graded expression of GPR30 in patients with various ovarian cancer, including high-grade serous carcinomas, low-grade serous carcinomas (LGSC), mucinous carcinomas (MC), clear cell carcinomas (CCC), and endometrioid adenocarcinoma. The immunohistochemistry results showed that GPR30 was widely expressed in patients with various ovarian cancers, highly in LGSC while lowly in MC (Fig. 3A and B). Furthermore, the correlation analysis of GPR30 expression level with overall survival of ovarian cancers patients showed that patients with higher GPR30 expression level (grade 3) showed shorter survival time, while patients with lower GPR30 expression level (grade 0) showed longer survival time (Fig. 3C, S1-S3 Table).To further verify whether EPA was a ligand of GPR30 in ES2 cells, we used a modified protein lipid overlay assay [14]. As shown in Fig. 2H, the interaction between EPA and GPR30 existed and was dose-dependent. Moreover, the relative expression level of GPR30 was markedly lower in SKOV3 cells compared with ES2 cells, A2780 cells, and CAOV3 cells, indicating that GPR30 functions differently in different cell lines (Fig. 2I).

To determine whether GPR30 mediates the effect of EPA in ES2 cells, we knocked down GPR30 using RNAi and found that GPR30 expression was reduced by 80% without changes in GPR40 or GPR120 (Fig. 4A). The decrease in cell number following EPA treatment was significantly reversed in ES2 cells treated with GPR30-specific siRNA (Fig. 4B). Similarly, we used G15, the specific antagonist of GPR30, to test the EPA effect in ES2 cells. As shown in Fig. 4C, the inhibitory effect of EPA was again impaired. The siRNA and inhibitor results indicated that EPA-induced antiproliferation in ES2 cells is mediated through GPR30.

Furthermore, we knocked down GPR30 to examine its effect on the EPA proapoptotic action in ES2 cells. The number of apoptotic cells dramatically declined when ES2 cells were treated with GPR30 siRNA compared with negative-control siRNA (Fig. 4D). Additionally, the expression of apoptotic genes as influenced by EPA was reversed when GPR30 was knocked down (Fig. 4D and E). However, the proapoptotic effect of EPA on ES2 cells could not be reversed when GPR119 was knocked down (S5 Fig.), indicating GPR119, another G_αs coupled receptor, did not play an essential role in mediating the anticancer effect of EPA in ES2 cells. Altogether, the above results establish that GPR30 is essential in mediating the effect of EPA in ES2 cells.

The phosphoinositide 3-kinase (PI3K)/AKT signaling pathway is the most common pathway related to cancer proliferation [24]. In recent years, numerous reports have demonstrated that EPA is implicated in several signaling pathways, such as RAS/ERK/C/EBPβ and nuclear factor κB, contributing to tumor development and inflammation [25]. To evaluate whether EPA affects cell proliferation through these cellular response pathways, we investigated phosphorylated AKT and ERK1/2 in ES2 cells treated with or without EPA. Interestingly, palmitic acid stimulated AKT and ERK1/2 phosphorylation, while EPA inhibited their phosphorylation after 24 hours of treatment (Fig. 5A). To further determine whe-ther GPR30 was involved in the activation of AKT and ERK1/2, we knocked down GPR30 as mentioned above. The phosphorylation levels of AKT and ERK1/2 in response to palmitic acid did not obviously change, and that in response to EPA was dramatically increased when ES2 cells were treated with GPR30 siRNA (Fig. 5A and B). These findings suggest that the development of signaling pathways via GPR30 includes PI3K-AKT and ERK.

The results in Figs. 2 and 4 show that EPA inhibits ES2 cell growth through GPR30. Due to the inhibitory effect on cancer cell growth by G_αs-cAMP-PKA signaling activation [26], the G_αs-cAMP-PKA signaling pathway may be affected by EPA via GPR30 in ES2 cells. Thus, we knocked down GPR30 in ES2 cells by GPR30 siRNA and G15 as described above. The cAMP production promoted by EPA was attenuated following downregulation of GPR30 (Fig. 5C and D). Additionally, EPA increased PKA activity compared to the treatment vehicle, as detected by the phospho-(Ser/Thr) PKA substrates. Notably, G1, the specific agonist of GPR30, also increased PKA activity. Moreover, PKA activity stimulated by EPA was suppressed by GPR30 knockdown and G15. However, PKA activity stimulated by EPA, G1, and siGPR30 did not change compared to its activity stimulated by EPA and siGPR30 (Fig. 5E and F). These results indicate that GPR30-cAMP-PKA signaling is involved in the EPA effect in ES2 cells.

To confirm the above observations, we also conducted a mouse xenograft model to further verify the role of GPR30 in ES2 cells. In accordance with previous findings, EPA alone inhibited tumor growth significantly (Fig. 6A and E). To further confirm the function of GPR30, we knocked down GPR30 using adenovirus in a mouse xenograft model. As shown in Fig. 6A, EPA dramatically decreased tumor growth, while EPA was unable to inhibit tumor growth when GPR30 was knocked down. hematoxylin and eosin (H&E) staining and labeling for the cancer proliferation marker Ki67 also showed that EPA lost its inhibitory function when ES2 cells were treated with GPR30 shRNA (Fig. 6B). The change in tumor volume and weight supported the above results (Fig. 6C and D). Interestingly, we found that the deletion of GPR30 in ES2 cells decreased tumor growth in mice, suggesting that GPR30 is associated with tumor development. Therefore, we antagonized GPR30 activity by treatment with GPR30 antagonist G15 in the mouse xenograft model. Although the tumor weight had no significant change due to individual variation (Fig. 6H), the tumor size and H&E staining showed that EPA or G15 alone both inhibited cell proliferation significantly (Fig. 6E-G). The combination of EPA and G15 was unable to inhibit tumor growth, as G15 inhibited antitumor effect of EPA by impairing GPR30 activity. These in vivo results also suggest that EPA inhibits tumor growth via GPR30 in human ovarian clear cancer cells.