Elevation of ω-3 Polyunsaturated Fatty Acids Attenuates PTEN-deficiency Induced Endometrial Cancer Development through Regulation of COX-2 and PGE2 Production

Endometrial cancer is one of the most common gynecologic malignancies. Phosphatase and tensin homologue (PTEN)-mutation is frequently identified in endometrial cancer patients. Although high dietary intake of ω-3 polyunsaturated fatty acids (PUFAs) has been associated with reduced risk of endometrial cancer, the underlying mechanisms is still unknown. To this end, we evaluated the impact of ω-3 PUFAs using several endometrial cancer cellular and animal models. While ~27% and 40% of heterozygotic PTEN mutant mice developed endometrial cancer and atypical complex hyperplasia, respectively, none of the PTEN+/− mice developed cancer when we overexpressed an mfat-1 transgene, which allowed endogenous production of ω-3 PUFAs. Fish oil-enriched diet or expression of mfat-1 transgene significantly inhibited the growth of xenograft tumor derived from RL95-2 cells bearing a PTEN null mutation. At cellular level, ω-3 PUFAs treatment decreased the viability of RL95-2 cells, AKT phosphorylation, and cyclin D1 expression. These molecular events are primarily mediated through reduction of cyclooxygenase-2 (COX-2) expression and prostaglandin E2 (PGE2) production. Exogenous PGE2 treatment completely blunted the impact of ω-3 PUFAs on endometrial cancer. Thus, we revealed the direct inhibitory effects of ω-3 PUFAs on endometrial cancer development and the underlying mechanisms involving reduction of COX-2 and PGE2.

incidence of endometrial neoplasia in mice 8,9 . Thus, targeting PTEN-deficiency initiated may represent a new therapeutic strategy for the prevention and treatment of this malignant disease.
ω -3 and ω -6 polyunsaturated fatty acids (PUFAs) are essential fatty acids necessary for human health, all of which have to be obtained through diets due to the inability of mammals synthesizing these fatty acids 10 . Epidemiological literatures on the linkage between ω -3 PUFAs and cancer incidence, including cross-sectional and migrational studies, have revealed a protective effect of ω -3 PUFAs and a promoting effect of ω -6 PUFAs on the development of cancers 11,12 . Specifically, typical modern western diets are high in ω -6 but low in ω -3 PUFAs, and are positively associated with tumorigenesis and poor prognosis of cancers 11,13 . Dietary intake of high levels of long chain ω -3 PUFAs has been shown to reduce various cancers and alleviate their complications 14,15 . Clinically, long-term high intake of diets or supplements enriched in eicosapentaenoic acid (EPA) and docosahexaenoic acid (DHA) were associated with lower risk of endometrial cancer 16,17 . Dietary ω -3 PUFAs significantly attenuated endometrial cancer cell growth in xenograft models 18 . Therefore, high circulating and tissue contents of ω -3 PUFAs may be an important tool in the prevention and treatment of cancer pathogenesis 11,19 .
We previously reported a transgenic mouse model overexpressing a C. elegans gene, mfat-1, encoding an ω -3 fatty acid desaturase 20 . This enzyme can produce ω -3 PUFAs endogenously by converting ω -6 to ω -3 PUFAs, which enables the investigation of the biological properties of ω -3 PUFAs without the need of lengthy feeding of fish oil. Furthermore, this model also makes it possible to use genetic approach by, for example, crossing the mfat-1 transgenics with the haploid PTEN-deficient mice. Such genetic approach also complements very well the xenogenic model with endometrial cancer RL95-2 cells, a PTEN-deficient cell line. With these animal models, we can interrogate the impact and underlying mechanisms of ω -3 PUFAs on spontaneously developed PTEN-deficiency-induced primary lesions. The potential positive outcomes of our studies may benefit the patients with PTEN-deficient endometrial cancer.

ω-3 PUFAs attenuates PTEN-deficiency induced primary endometrial cancer development.
To investigate the impact of ω -3 PUFAs on primary endometrial cancer development, we genetically crossed the mfat-1 transgenic mice with PTEN +/− mice to allow this enzyme to produce ω -3 PUFAs in the tissues 20,21 . The levels of PTEN mRNA and protein in the uterus of PTEN +/− mice were about half of the PTEN +/+ mice (Fig. 1a,b), confirming the haploid deficiency of PTEN expression. Analysis of fatty acid compositions confirms the activity of mfat-1 protein, with a significant decrease in arachidonic acid (AA), a concomitant increase in EPA and DHA levels, and a significantly decreased ratio of ω -6/ω -3 PUFAs compared with the mice lacking mfat-1 expression (PTEN +/+ , PTEN +/− ) ( Table 1).
Consistent with previous reports 8,9 , high incidence of endometrial complex atypical hyperplasia (CAH, 40%) and carcinoma (CA, ~27%) were observed in the uterus of PTEN +/− mice (Fig. 1c,d). In contrast, analysis of serial sections of uterus samples did not reveal any malignant lesion in any of the mfat-1;PTEN +/− animals and the CAH incidence was reduced to 30% within this group (Fig. 1c,d). Interestingly, even the rate of simple hyperplasia (SH) markedly dropped from ~33% to 20%. In the control groups (wild-type mice or mfat-1 transgenics), the uterus tissues were completely normal with no identifiable hyperplasia (Fig. 1c,d). These results suggest that elevated tissue levels of ω -3 PUFAs suppress PTEN-deficiency induced endometrial cancer development. ω-3 PUFAs inhibit endometrial cancer cell growth in xenograft models. As a further step of demonstrating the beneficial effect of ω -3 PUFAs against endometrial cancer cell development in vivo, we studied this issue in a xenograft model. We infected RL95-2 cells, a human endometrial carcinoma line containing PTEN null mutation, with lentivirus carrying the mfat-1 gene (termed "RL95-2-mfat-1" cells). As a control, we also generated a green fluorescent protein (GFP) expressing line, RL95-2-GFP. The expression of mfat-1 protein endogenously converted ω -6 PUFAs to ω -3 PUFAs in the RL95-2 cells (Table 1). Following the establishment of these cell lines, we established endometrial tumor xenografts by subcutaneously implanting these RL95-2 cell lines into immune-deficient nude mice. Although the mice implanted with the RL95-2-mfat-1 cells were maintained on regular diet, the mice implanted with the control RL95-2-GFP cells were fed either regular diet or regular diet mixed with 5% fish-oil (FO). Analysis of fatty acid composition identified significantly decreased ω -6 and elevated ω -3 PUFAs content in the tumors derived from the mice either fed 5% FO-containing diet or implanted with RL95-2-mfat-1 cells ( Table 1). Analysis of implanted human tumor xenografts revealed that elevated tissue contents of ω -3 PUFAs significantly reduced tumor growth as measured by tumor volume and weight (Fig. 2), regardless if the source of ω -3 PUFAs was coming from the mfat-1 expression or dietary intake. ω-3 PUFAs inhibit Akt phosphorylation and cyclin D1 expression in primary endometrial cancer. The constitutive activation of phosphoinositide-3 kinase (PI3K)-Akt pathway is a direct consequence of PTEN null mutation, and the primary cause of malignant cell proliferation. Through immunohistochemistry analysis, we found that PTEN deficiency-induced Akt phosphorylation and cyclin D1 expression were significantly suppressed in endometrial tissues of PTEN +/− mice when mfat-1 was transgenically expressed (Fig. 3). Meanwhile, the number of Ki-67 cells, a molecular marker of cell proliferation, was also sharply reduced in the mfat-1;PTEN +/− mice (Fig. 3). Thus, elevation of ω -3 PUFAs Scientific RepoRts | 5:14958 | DOi: 10.1038/srep14958 in the uterine tissues strongly decreased PTEN-deficiency induced Akt phosphorylation and cyclin D1 expression in malignant endometrial cancer tissues.

ω-3 PUFAs inhibit endometrial cancer cell growth in vitro. To investigate if the effect of ω -3
PUFAs is directly acted on endometrial cancer cells, we initially examined the impact of adding EPA or DHA to the cultured RL95-2 cells. DHA inhibited RL95-2 cell viability in a dose-dependent manner (Fig. 4a). At the highest concentration (100 μ M) tested, DHA caused a 5-fold reduction of cell viability in RL95-2 cells (Fig. 4a). Interestingly, addition of EPA did not significantly change the growth curve of the cultured cells (Fig. 4b). Thus, the primary effector of the ω -3 PUFAs mixture appears to be DHA in inhibiting RL95-2cell growth. Consistent with this observation, we also found mfat-1 expression significantly suppressed cell viability (Fig. 4c). Flow cytometry analysis of cell cycle further revealed that mfat-1 expression caused a G1 arrest in cell cycle in RL95-2 cell line ( Table 2). Taken together, these results indicate that ω -3 PUFA, particularly DHA, and mfat-1 expression can directly inhibit endometrial cancer cell growth.
Metabolomic analysis of eicosanoids in RL95-2 cells. The differential effects of ω -6 and ω -3 PUFAs on cancers may be mediated by their eicosanoid metabolites. As the initial step of addressing this issue, we analyzed some of the eicosanoid metabolites induced by ω -6 or ω -3 PUFA treatment. Using an LC− MS/MS-based method, we analyzed 31 AA metabolites and 15 DHA-derived products ( Table 3). As one might expect, 29 of the 31 identified AA-derived metabolites were increased in RL95-2 cells following 50 μ M AA treatment, chief among which prostaglandin E 2 (PGE 2 ), LTB 4 , 5-HETE, 12-HETE, 15-HETE, 20-HETE, 11,12-EET, and 14,15-EET, which are known to promote cancer growth 22-28 , were sharply elevated compared with the control group. Interestingly, the expression of mfat-1 and DHA treatment,   Table 1. Analysis of ω-6 and ω-3 PUFA composition. Abbreviations: ALA, α -lipoic acid; DPA, docosapentaenoic acid; LA, linoleic acid. The composition of ω -6 and ω -3 PUFAs in samples was analyzed using standard protocols (see Materials and Methods). Each species is expressed as a percentage of all fatty acid peaks, ie, the distribution areas of different ω -3 or ω -6 PUFAs peaks divided by the total peak areas of all detectable saturated and unsaturated free fatty acids (from the same sample) resolved from the gas chromatography column. a the composition of ω -6 and ω -3 PUFAs in the mice blood. All data presented are means ± SD. n = 6. * P < 0.05 and ** P < 0.01 when the mfat-1 group compared with PTEN +/+ group or the mfat-1;PTEN +/− group compared with PTEN +/− group. b the compositions of ω -6 or ω -3 PUFAs in endometrial cancer cells. c the compositions of ω -6 or ω -3 PUFAs in endometrial xenograft mice tumors. All data presented are means ± SD. n = 6. * P < 0.05 and ** P < 0.01 compared with respective controls. eicosanoids were elevated following DHA treatment. On the other hand, expression of mfat-1 transgene only increased the production of 13-HDoHE and 19,20-EDP. Despite these differences, both treatments led to reduction of AA-derived PGE 2 . Accordingly, DHA treatment and mfat-1 expression significantly decreased PGE 2 production by 3.6 fold and 3.2 fold, respectively (Table 3).
cyclooxygenase-2 (COX-2) expression and PGE 2 production were involved in ω-3 PUFAsmediated attenuation in PTEN-deficient endometrial cancer. As a bioactive lipid that elicits a wide range of biological effects associated with inflammation and cancer 29 , PGE 2 is derived from AA as a result of the sequential enzymatic reactions of COX-2 and PGE synthase. To interrogate how ω -3 PUFAs suppress cell proliferation through eicosanoid metabolites, we evaluated the expression of COX-2 following the treatment with AA or DHA. Although incubation of RL95-2 cells with AA significantly elevated COX-2 expression, DHA treatment significantly decreased COX-2 expression (Fig. 5a). The expression of mfat-1 mimicked the effect of DHA on COX-2 regulation (Fig. 5a). In the xenografts, both feeding of 5% FO-containing diet and mfat-1 expression inhibited COX-2 expression (Fig. 5b,c). Addition of PGE 2 to RL95-2 cells had a profound effect on ω -3 PUFAs-suppressed cell proliferation, Akt phosphorylation, and cyclin D1 expression. Cell proliferation assays showed that mfat-1 expression decreased cell proliferation, compared with RL95-2-GFP cells. Consistent with this observation, while exogenous PGE 2 treatment partially counteracted mfat-1-induced proliferation arrest (Fig. 5d). As shown in Fig. 5e, Akt phosphorylation and cyclin D1 expression were significantly decreased in RL95-2-mfat-1 cells compared with RL95-2-GFP cells. However, addition of PGE 2 counteracted the impact of ω -3 PUFAs on Akt phosphorylation and cyclin D1 expression, suggesting that regulation of COX-2 expression and PGE 2 production played a major role in ω -3 PUFAs-mediated growth inhibition of endometrial cancer.

Discussion
The overall aim of this study is to investigate the impact and the underlying mechanisms of ω -3 PUFAs on endometrial cancer, particularly the type carrying genetic mutation in PTEN, whose primary function is to oppose PI3K activity 7,30,31 . PTEN-deficiency is the most common genetic mutation detected thus far in endometrial cancers [3][4][5] . Mice carrying heterozygotic mutation of PTEN also have a high incidence of endometrial neoplasia 8,9 . We took advantage of a previously described transgenic model, mfat-1, that allows production of ω -3 PUFAs in all tissues including the uterus of PTEN haploid deficient mice. In contrast to the high incidence of endometrial neoplasm (~27%) and atypical hyperplasia (40%) associated with PTEN heterozygotic deficiency 8,9,21,32 , the expression of mfat-1 in the PTEN +/− mice displayed no malignant endometrial lesion and significantly reduced incidence of atypical complex hyperplasia (30%). The constitutive activation of PI3K-Akt pathway and the resulting increased expression of cyclin D1 is a critical mechanism of aggressive cell proliferation in the context of PTEN-deficiency [33][34][35][36] . Indeed, Akt1 deficiency is sufficient to inhibit the development of endometrial carcinoma in PTEN +/− mice 8 . Consistent with the pathology findings, the levels of phospho-Akt and cyclin D1 expression were greatly suppressed by elevated levels of ω -3 PUFAs as a result of mfat-1 expression. Such results derived from the primary endometrial carcinoma were also confirmed at cellular levels or in a xenograft model in which dietary addition of ω -3 PUFAs or mfat-1 expression inhibited the growth of cultured or implanted RL95-2 cells carrying a PTEN null mutation. Combined together, these data clearly demonstrate that ω -3 PUFAs can strongly suppress carcinogenesis and progression of endometrial cancer by inhibiting activation of PI3K/Akt/cyclin D1 signal pathway in the context of PTEN deficiency.
Mounting evidence has linked dietary gain of ω -3 PUFAs to the prevention or attenuation of progression of several cancers, such as colon 37 , breast 37,38 , and prostate cancers 39 . A series of recent studies have also shown that mfat-1 expression, which produces ω -3 PUFAs endogenously, also inhibited the growth of colon 40 , prostate 41 , liver 42 , and breast cancer 43 . However, these strong phenotypic results, including the aforementioned phenotypes of ω -3 PUFAs on endometrial cancer have not conclusively revealed how ω -3 PUFAs achieved these anti-tumor effects. We focused on the roles of metabolites derived from ω -3 and ω -6 PUFAs. Eicosanoids are widely believed to play important roles in carcinogenesis and aggressiveness of many cancers 23   in our assays, we found that AA treatment increased AA-derived metabolites levels including 14,15-EET, LTB 4 , 5-HETE, 12-HETE, 15-HETE, 20-HETE, 11,12-EET, and PGE 2 , all of which have been shown to promote the development of many types of cancers [22][23][24][25][26][27][28] . In contrast, treatment with ω -3 PUFAs significantly decreased PGE 2 levels while elevating the production of DHA-derived metabolites levels including 13-HDoHE and 19, 20-EDP. PGE 2 is a direct product of oxidization by COX-1 or COX-2. Previous studies have revealed elevated expression of COX-2 in various inflammatory diseases and human cancers 25,29,[46][47][48][49] , including association with poor prognosis in patients suffering endometrial cancer 50 . Importantly, selective inhibitors of COX-2 resulted in significant antitumor effects in endometrial cancer patients 51,52 . In our studies, elevation of ω -3 PUFAs caused a significant inhibition of COX-2 expression in vitro and in vivo. Treatment of RL95-2 cells with exogenous PGE 2 counteracted all of the ω -3 PUFAs-induced effects, including proliferation arrest, reduction of Akt phosphorylation and cyclin D1 expression. These data suggest that reduction of COX-2 expression and AA-derived metabolite, PGE 2, are important mechanisms underlying ω -3 PUFA-induced interruption of PI3K/Akt signal pathway.
In summary, this study provides the definitive in vivo evidence that ω -3 PUFAs strongly inhibit tumorigenesis and progression of PTEN-deficiency induced endometrial cancer through a mechanism involving reduction of COX-2 expression and PGE 2 production. Such mechanisms oppose the constitutive activation of Akt phosphorylation, cyclin D1 expression. The suppressive effect of ω -3 PUFAs in the development and progression of endometrial cancer may offer novel prevention and treatment strategy of this malignant disease in women.

Methods
Animals. All experimental protocols were approved by the Research Ethics Committee of the Nanjing Medical University. All experiments were conducted in compliance with the guidelines for the care and use of laboratory animals and approved by Institutional Animal Care and Use Committee of Nanjing Medical University. The normal diet was from Xietong Biotechnology Co.Ltd (Jiangsu, China). The fish oil was from ShangHai HOPE industry Co. Ltd (Shanghai, China). The high ω -3 PUFAs diet was made by the normal diet mixed with 5% Fish-Oil (FO). The animals were fed food and water ad libitum, housed at 22 °C and a 12-h light-dark cycle.  China). The mfat-1 transgenic mice were described in our previous studies 20 . All above strains of mice have a C57BL/6J genetic background. The PTEN gene was deleted specifically from oocytes of primary follicles by crossing Zp3-Cre mice with PTEN loxP/loxP mice as previously described 53 . The female offspring was crossed with male mfat-1 transgenic mice to generate four different genotypes for experiments (PTEN +/+ , mfat-1, PTEN +/− , mfat-1;PTEN +/− ). The PTEN +/+ and mfat-1 mice were served as littermate control.
Tumor xenograft models. Four-week-old female BALB/c nude mice were injected with 5 × 10 6 /0.1 ml of RL95-2 cells into the flanks. The mice were randomized into three groups (n = 10). Group 1 was implanted with RL95-2-GFP cells and fed with normal diet. Group 2 was implanted with RL95-2-GFP cells and fed with high ω -3 PUFAs diet (normal diet mixed with 5% FO). Group 3 was implanted with RL95-2-mfat-1 cells and fed with normal diet. Bi-dimensional tumor measurements were taken every two days. Tumor volume was measured along two major axes using calipers. Tumor volume (mm 3 ) was calculated as follows: V = 1/2 LW 2 (L: length, W: width).
Cell culture and proliferation assays. Overexpression of mfat-1 by lentiviral transfection. The description of mfat-1 cDNA has been described previously 20 . The mfat-1 cDNA was synthesized and cloned into the PLJM1 lentivirus vector (Addgene, Palo Alto, CA, USA). To produce the virus, pMD2.G and psPAX2 were co-transfected with PLJM1-mfat-1 or the vehicle plasmid into 293T cells using X-tremeGENE HP DNA transfection reagent, according to manufacturer's instructions. To obtain a stable mfat-1 overexpressing cell line, lentivirus-containing supernatant was harvested 48 hours after transfection and used to infect RL95-2 cells. Western blotting. Cells were treated with various PUFA or PGE 2 agents at 37 °C for various times, as indicated in the experiments. The cells were collected and subjected to Western blotting analysis. The immunoreactivity was detected by using a standard enhanced-chemiluminescent reaction and analyzed using Image Lab 4.0 analysis software from Bio-Rad.

Quantitative analysis of gene expression levels in cells and tissues.
Immunohistochemical staining. The tissue specimens were incubated with antibodies against phospho-Akt, Ki-67, Cyclin D1, and COX-2. The specific signals were detected using EnVision polymer technology, and visualization was performed with 3, 3N-diaminobenzidine tetrahydrochloride (DAB). The sections were photographed by Nikon microscopy and image analyses system. The expression levels of proteins were quantitated using Image Pro Plus 6.0 software (Media Cybernetics Inc., New York, USA).
Analysis of cell cycle. After RL95-2 cells were incubated in serum-free medium overnight, cells were collected and rinsed with PBS twice. For cell cycle analysis, the cells were fixed in 70% pre-cold ethanol at 4 °C overnight and stained by PI. The DNA content was analyzed using a BD FACStar flow cytometer.
Gas chromatography analysis of fatty acid compositions. Lipids extraction from cells and tissues were performed according to a previous report 41 . The gas chromatography was done on an Agilent 7890A. Identification of components was done by comparison of retention times with those of PUFA analytical standard.
Systematic metabolomic analysis of eicosanoids. The RL95-2 cells were seeded in 15 cm dish 24 hours before adding various PUFA. After digesting with 0.25% trypsin, cells were washed twice by PBS and counted. 2 × 10 7 cells were needed for eicosanoid analysis in each sample. Eicosanoid extraction was performed according to a previous report 54 . Chromatographic separation involved an ACQUITY UPLC BEH C18 column consisting of ethylene-bridged hybrid particles. The metabolites were quantified by use of a 5500 QTRAP hybrid triple quadrupole linear ion trap mass spectrometer equipped with a Turbo Ion Spray electrospray ionization (ESI) source.