Prostaglandin E₃ metabolism and cancer

Abstract

The anticancer activity of n-3 fatty acids, especially those derived from fish, such as eicosapentaenoic acid (EPA) and docosahexaenoic acid) (DHA), has been studied for centuries. While there is a growing body of evidence that EPA and DHA may influence cancer initiation and development through targeting multiple events of tumor development, the underlying mechanisms responsible for these activities are still not fully understood. A number of studies have suggested that the anticancer activities of EPA and DHA are associated with their effects on eicosanoid metabolism by which they inhibit prostaglandin E₂ (PGE₂) production. In contrast to DHA, EPA can function as a substrate for cyclooxygenases (COXs) to synthesize unique 3-series prostaglandin compounds, especially PGE₃. With advance technology in mass spectrometry, there is renewed interest in studying the role of PGE₃ in EPA elicited anti-proliferative activity in various cancers, with some promising results. Here, we summarize the regulation of PGE₃ synthesis in cancer cells and its role in EPA elicited anticancer activity. The development of PGE₃ and its metabolites as potential biomarkers for future clinical evaluation of EPA and fish oil in cancer care is discussed.

Introduction

Cancer is a leading cause of death worldwide and a second in the United States, exceeded only by heart disease. One in every four deaths in the United States is due to cancer. Despite the advancement in various treatment strategies, such as combinations of surgical resection, radiation or chemotherapies and immune therapies, the 5-year survival rate for some cancers is still relatively low. Furthermore, the underlying cause of cancer remains unclear. Thus, there is an unmet need to develop an effective strategy for preventing the development of this devastating disease. While the results of large chemoprevention trials thus far are not encouraging, a 20-year follow up study with aspirin, a non-steriodal anti-inflammatory agent that acetylates cyclooxygenase 2 (COX-2), showed that the mortality rates from all solid cancers were 20% lower for those receiving aspirin, with adenocarcinoma being the most reduced (34%) [1], suggesting the role of anti-inflammatory agents such as COX inhibitors in cancer prevention. Further studies indicated that aspirin use has no impact on the risk of colon adenocarcinomas that do not overexpress COX-2 [2,3]. Overexpression of COX-2 has been observed in a number of malignant diseases, especially epithelial cancers. Prostaglandin E₂ (PGE₂), a metabolite of n-6 fatty acid (arachidonic acid, AA), produced by COX-2 catabolism inhibits cancer cell apoptosis, increases invasiveness and angiogenesis in the tumor through pathways such as NF-κB, MAPkinase/JNK/p38, PI3kinase/Akt [4,5] as well as epigenetic modifications [6]. Thus, there has continued interest in using selective COX-2 inhibitors, such as celecoxib and rofecoxib, in chemoprevention. However, the cardiotoxicity of these agents has become an impediment to their long-term chemopreventive usage. In contrast to synthetic COX inhibitors, n-3 fatty acids are natural modulators of COX-2, with the ability to alter COX-2 metabolites and regulate the activity of downstream receptors while reducing blood triglyceride levels [7–10].

Long chain n-3 polyunsaturated fatty acids (PUFA) or n-3 fatty acids including α-linolenic acid (C18:3, ALA), eicosapentaenoic acid (C20:5, EPA), docosapentaenoic acid (C22:5, DPA) and docosahexaenoic acid (C22:6, DHA) are a group of compounds possessing the first double bond after the third carbon atom from the methyl end of fatty acid chains. These fatty acids, especially EPA and DHA, have been shown to have anti-inflammatory and immunomodulatory properties and are believed to be beneficial to cardiac, musculoskeletal, gastrointestinal, and immune systems in humans [11]. Epidemiological and preclinical evidence support the notion that n-3 fatty acids, especially EPA and DHA, have anticancer activities. For example, n-3 fatty acids have been shown to reduce onset of different cancers and protect against late stage cancers in carcinogen induced mouse tumors, human tumor mouse xenografts and spontaneous mouse tumors induced by transgenes [12]. Additionally, human studies have demonstrated that a higher intake of n-3 fatty acid is linked to a reduced risk of skin, colorectal, lung, prostate and breast cancers [13–17]. A recent study demonstrated that intake of EPA and DHA was associated with approximate 25% reduced risk of additional breast cancer events and had a dose-dependent reduced risk of all-cause mortality [15]. These findings together provide compelling evidence that n-3 fatty acids, EPA and DHA, could not only prevent the initiation of cancer, but also delay further development of cancer possibly through different mechanisms (reviewed in [12]). Among a number of plausible mechanisms, the ability of n-3 fatty acids, especially EPA, to modulate eicosanoid metabolism, particularly reduction of cyclooxygenase derived prostaglandin metabolism, has been extensively studied. Additionally, EPA and DHA derived resolvins from acetylated COX-2 (by aspirin) or 15-lipoxygenase demonstrate anti-inflammatory, neuroprotective and anticarcinogenic activities which could be an additive benefit of EPA or DHA in cancer risk reduction [18–20]. Compelling evidence supports that EPA could function as a selective COX-2 inhibitor because it can essentially act as a competitive inhibitor of AA to COXs, resulting in reduction of the 2-series prostaglandin (PG), such as PGE₂, and concomitant generation of the 3-series PGs, i.e., PGE₃ [9,21,22]. Compared to PGE₂, PGE₃ and other 3-series prostaglandins tend to have antiproliferative and anti-inflammatory activities and could potentially antagonize the tumor promoting effect of PGE₂ in tumorigenic cells [21,23– 25]. In normal mouse colonic organoid culture, a recently report demonstrated that PGE₂ induced proliferation of Lgr5⁺ colonic stem cells and promoted growth of mouse colon organoids while PGE₃ did not support the colonic stem cell expansion in the same system [25]. The pharmacokinetics of AA derived PGE₂ and EPA derived PGE₃ are differentially regulated in normal and cancer cells, however, due to the increased expression of COX-2 in cancer cells, PGE₃ production in tumors can be much higher than the surrounding normal tissue. Given that, in general, cellular metabolism fundamentally differs in cancer cells and normal cells, an understanding of how EPA is metabolized in cancer cells becomes a critical component of research investigations focusing on anticancer activity of n-3 fatty acids.

In this review, focusing on PGE₃ metabolism, we summarize the pharmacokinetics of EPA that produces PGE₃ in normal and cancer cells, the comparisons of the 3-series to the 2-series PG receptors, the regulation of PGE₃ metabolism in cancer cells, and the association between production of PGE₃ and antitumor or chemopreventive effects of n-3 fatty acids. Additionally, we emphasize the significance of developing appropriate biomarkers for EPA, such as PGE₃ and its metabolites, to further determine the anticancer function of n-3 in future clinical applications.

Overview of AA and EPA metabolism by cyclooxygenases

In humans, n-3 and n-6 series fatty acids are ingested in the physiologically active forms of EPA and AA or as these molecules' respective precursors ALA and linoleic acid (LA, 18:2 n-6) [26]. LA, the major dietary source of n-6 fatty acids, is efficiently converted to AA. Studies have reported that because of the high oxidation rate, only a small proportion of ALA is converted to EPA, DPA or DHA [27]. Omega 3 polyunsaturated fatty acid DPA is an elongated metabolite of EPA and is an intermediary product between EPA and DHA [28]. Burdge and Wootton discovered that young men possessed the capacity to synthesize EPA and DPA from ALA, but that DHA synthesis was limited [29]. The fractional conversion of ALA to EPA varies between 0.3% and 8%, and less than 4% to DHA in males [30,31]. In women, the conversion of ALA to long chain n-3 fatty acids appears to be more efficient (up to 20% is converted to EPA and up to 9% is converted to DHA) [31]. DHA itself also serves as a substrate for metabolic retro-conversion to EPA and DPA through a β-oxidation reaction [32]. The retro-conversion rate of DHA to EPA, which can be affected by hormonal therapies in women [33], is approximately 1.4% with normal intake of DHA from food sources, while increases to 10% (9.4–12%) with DHA supplementation [32,34,35]. Similarly, DPA can also be retro-converted to EPA [36]. High ALA diets seem to only increase the rate of ALA oxidation with little conversion to EPA or DHA because diets high in ALA limit ALA accumulation in plasma [37]. With considerable variability in the conversion rates of ALA, along with modest ALA intakes and high amounts of LA in the American diet, it is reasonable to believe that ALA cannot reliably replace EPA or DHA in the diet [32,38]. However, the combined intake of EPA and DHA is estimated at only about 100 mg/d, due to limited intake of marine foods that are the major dietary source of n-3 fatty acids in the United States [39].

Fatty acid incorporation in membrane phospholipid

After being absorbed from the GI track, both n-3 and n-6 fatty acids compete for esterification at the sn-2 position of membrane phospholipids (PLs) [32]. Dietary intake of EPA and DHA, not ALA metabolic conversion, is the major contributor of EPA and DHA in cellular membrane phospholipids, because ALA is not efficiently incorporated into membrane phospholipids. In humans, short term supplementation with DPA not only significantly increases the incorporation of DPA in both plasma PL and triglycerides (TGs), but also enhances the incorporation of EPA and DHA into plasma TGs and cholesterol esters (CEs). In contrast, EPA supplementation only results in increased incorporation of EPA in plasma PLs and CEs without altering the levels of DPA and DHA incorporation into plasma lipids, suggesting that DPA and EPA incorporate into plasma phospholipids in different patterns [40]. Additionally, rodent and most recently published human data suggested that dietary DPA and EPA may be metabolized differently even though both of them exert anti-inflammatory and anticarcinogeneic activities [28,41,42]. On the other hand, high dietary intake of LA, which is efficiently incorporated into the membrane, reduces the membrane EPA phospholipid pool. Upon incorporation into the membrane, AA and EPA can be released from cellular phospholipids by phospholipases following extracellular stimulation, while free DHA released from the membrane is undetectable [43]. Cytosolic free AA or EPA in cells can further be converted to downstream metabolites that differ both in the number of their double bonds for COX products (prostaglandins, PG) and lipoxygenase (LOX) products (HETEs and leukotrienes) and in their biological actions in a wide range of physiologic and pathologic processes [44], including inflammation and immunity [45], hemostasis [46,47], atherosclerosis [48,49] and cancer [50,51]. Further, the balance in tissues eicosanoids derived from AA and EPA is regulated by complex interactions/competitions between AA and EPA at multiple levels of the eicosanoid biosynthetic pathway [52].

Eicosanoids metabolites

The term “eicosanoid” is used to denote twenty carbon PGs and lipoxins that are synthesized from n-3 and n-6 fatty acids in response to various hormones and physical stimuli [53]. The detailed prostaglandin biosynthesis including both 2 and 3 series PGs using AA and EPA has been comprehensively reviewed [54]. In this review, we will focus on PG synthesis, especially for 3-series prostaglandin, demonstrated in Fig. 1. Eicosanoids are synthesized and released from the cells rapidly (seconds) in response to extracellular stimuli [53]. Syntheses of the 2- and 3- series PG metabolites from AA or EPA, respectively, share three common steps in which particular synthases are involved (Fig. 1): firstly, AA or EPA is released from membrane phospholipids by cytosolic phospholipase A₂ (cPLA₂) or secretory sPLA₂; secondly AA/EPA is converted to prostaglandin endoperoxide (PGH₂ from AA and PGH₃ from EPA) by COX-1 or COX-2; lastly, isomerization of PGH to “2-series” or “3-series” products– PGE₂ or PGE₃, PGD₂ or PGD₃, PGF_2α or PGF_3α, PGI₂ or PGI₃, or thromboxane A₂ (TxA₂) or TxA₃ by specific synthases.

Fig. 1.

Comparison of EPA and AA cyclooxygenase metabolism. Modified from Pirman et al. [68], Wada et al. [22].

Although metabolisms of AA and EPA share the same set of enzymes, the catabolic activities of the enzymes in each of the metabolic step towards the two groups of substrates are somewhat different. cPLA₂ is found in cytosol of cells at resting state. It translocates to the plasma membrane upon hormone-induced mobilization of intracellular Ca²⁺. While both EPA- and AA-containing phospholipids are substrates for cPLA₂ with approximately equal activities [55,56], cPLA₂ hydrolyzes DHA poorly [57]. COX-1 has been reported to preferentially catalyze AA with very low activity on EPA (10% compared to AA). EPA is a reasonably good inhibitor of AA oxygenation by COX-1; at equimolar concentrations EPA inhibits AA oxygenation by 50%. In contrast to COX-1, COX-2 converts EPA to PGH₃ at about 30% of the rate of conversion of AA to PGH₂. However, studies from our laboratory and other investigators demonstrated that EPA is a better substrate for human COX-2 enzyme than COX-1 enzyme [21,58]. EPA is a poor inhibitor of AA oxygenation by COX-2 [22,58,59]. Microsomal PGE synthase-1 (mPGES-1) and PGD synthases (PGDS) were reported to be more than 3-fold less active with PGH₃ than with PGH₂ [22], but the activity of cytosolic PGES on PGH₃ has not been reported [60]. PGF synthase (PGFS) catalyzes the biosyntheses of PGF_2α and PGF_3α, however the substrate-enzyme binding/catabolic activities with PGH₂ vs. PGH₃ are unknown. PGI synthase (PGIS) and TxA synthase (TxAS) demonstrated similar activities with PGH₂ and PGH₃ [61,62]. Additionally, 15-hydroxyprostaglandin dehydrogenase (15-PGDH) catalyzes the rate-limiting step of PGE₂ catabolism [63], however whether it is responsible for PGE₃ degradation has not been reported. The ratios of the enzyme activities with EPA-vs. AA-derived reactants and products in vitro were documented [22], however the calculation needs to be tested in humans. Regardless, metabolite productivities of the entire enzymatic system closely depend on the substrate pools and levels of the various enzymes involved in the synthesis of the 2-series and the 3-series PGs.

EPA alters cellular PG metabolite profile

EPA competes with AA for incorporation into phospholipids in cell membranes, therefore partial substitution of AA by EPA in membrane phospholipids should lead to the replacement of AA-derived 2-series eicosanoid metabolites by EPA-derived 3-series metabolites. In several in vitro normal cell models, the effect of EPA on PG biosynthesis has been tested. Belury et al. first reported that PGE₃ formation in EPA treated murine epidermal cells was more than PGE₂ when these freeze–thawed cells were treated with equal molar of ¹⁴C-AA and ¹⁴C-EPA, while significant more PGF was produced in AA treated than in EPA treated cells [64]. Norris et al. showed that EPA treatment reduced the amount of AA-derived COX metabolites (20% lower) and increased EPA-derived COX metabolites PGD₃ and PGE₃ compared to control RAW264.7 macrophage cells [65]. In the same study, PGD₃ was the predominant EPA derived metabolite produced in EPA-supplemented cells and was about 3-fold higher than in control cells, which was mainly due to PGD synthase being predominant subtype prostaglandin synthases in macrophages cells [65]. In another study using cultured human mast cells, EPA treatment decreased PGD₂ generation by inhibiting the COX-2 pathway [66]. Limited in vivo data from animal studies have also shown that supplementation of dietary EPA reduces PGE₂ and increases PGE₃ in mouse colon [51,67], lung [68] or dog peripheral blood mononuclear cells (PBMCs) [69].

PGE₃ and PG receptors

Due to rapid metabolism, PGs function at or near their sites of synthesis [22]. Newly formed PGs primarily function through G-protein coupled receptors (GPCRs) as autocrine or paracrine mediators [53]. Research studying the actions of PG receptors have reported the presence of PG receptors for the D, F, I, and E types PGs and TxA, named DP, FP, IP, EP, and TP receptor, respectively (Table 1) [22,70]. Originally interactions between E type PGs and EP receptors were discovered based on PGE₂ actions [71]. There are four GPCR subtypes of EP receptors, EP₁, EP₂, EP₃, and EP₄, which exhibit differences in signal transduction, tissue localization, and regulation of expression [61]. This molecular and biochemical heterogeneity of PGE receptors leads to E type PGs being the most versatile PG [61]. Evidence for receptors for 3-series PGs is scant, with most research focused on downstream actions of PGE₃. Limited literature suggests that PGE₃ shares the same EP receptor system with PGE₂ albeit with different binding affinities and potencies (Table 1).

Table 1.

3-Series PG related EP receptors.

3-Series PGs	Receptors	EC50 ratios (PGE3 vs. PGE2)	Tissue/Cell	References
PGE3 (partial agonists for EP1, EP2, and EP3)	EP1	2	Human embryonic kidney cells	[22]
	EP2	2.55	Human pancreatic cancer cells, human embryonic kidney cells	[22,50]
	EP3	3.06	Human embryonic kidney cells	[22]
	EP4	6.03	Human CRC cells, human embryonic kidney cells	[22,24]
PGD3	DP1	0.46a	Human platelets	[22,119]
	DP2	1.14a	Human eosinophil	[22,119]
PGF3α (partial agonists)	FP	4.79	Human embryonic kidney cells	[22]
PGI3	IP	1.41	Human embryonic kidney cells, human and rabbit platelets	[22,120]
TxA3	TP	1.31	Human embryonic kidney cells	[22]

^aComparison of geometric means.

Wada et al. studied the receptor binding specificities of PGE₂/PGE₃ to EP and PGF_2α/PGF_3α to FP receptors using cell membrane from human kidney cells [22]. They determined that the relative affinities of the 2-series PGs were significantly greater than the 3-series PGs except for the EP₄ receptor. The difference between PGF_2α and PGF_3α with the FP receptor can be as high as 78-fold. In the same study, the authors reported the potencies of the 2-vs. the 3-series PGs in stimulating second messenger formation via the EP and FP receptors and found that the 2-series PGs exhibited significantly higher potency than the 3-series PGs with the exception for EP₄. Interestingly, the data further suggested that PGE₃ is a partial agonist of the EP receptors. Whether there are functional differences in downstream signaling between PGE₃ and PGE₂ was not detailed in the same publication. One recent study reported PGE₃ acts as an antagonist to EP₃ to promote the inhibitory effects of PGE₃ on platelet function, by which the cardiovascular benefits of dietary n-3 fatty acids may be conferred [72]. Another study illustrated that PGE₃ bound to EP₄ with reduced affinity and efficacy compared with PGE₂ in human colorectal cancer (CRC) cells, but in the presence of PGE₂, PGE₃ acted as an antagonist of EP₄ in cyclic AMP production [24]. A different research group found that both PGE₂ and PGE₃ induced cyclic AMP in RAW 264.7 cells, however, accumulation of intracellular cyclic AMP in PGE₃ treated cells was only half as high as PGE₂ treated cells [73]. These data suggest that although PGE₂ and PGE₃ both activate the EP receptor system, the substrate-receptor binding actions vary dramatically depending on the circumstance.

cPLA₂, COX-2, and PGE₃ metabolism in cancer

EPA functions as a substrate for COXs and results in synthesis of the 3-series PG compounds, such as PGE₃, PGD₃ and PGI₃ [9]. Even though the theory of formation of the 3-series PGs by EPA has been studied for decades, understanding of the synthetic capability of the 3-series PGs, especially PGE₃ from EPA in cancer cells is still inconclusive. This is due, in part, to the lack of specific and sensitive analytical techniques that could be used to determine the endogenous levels of PGE₃ in various biological matrices. Additionally, the amount of EPA derived PGE₃ is much lower than AA derived PGE₂ because of the low COX-2 protein expression in normal tissue [9]. Published studies reveal that COX-2 protein is overexpressed in various cancers, such as lung, colon, breast, and pancreatic tumors as well as in their relevant in vitro cancer cells [5,74]. As a result, the PGE₃ metabolism in cancer cells can be markedly different from that in the normal epithelial and smooth muscle cells mainly due to the different levels of expression and activities of the enzymes, such as cPLA₂, COX-2, and mPGES-1 regulating the release of EPA and biosynthesis of PGE₃ in the cancer cells or tumor tissues. Given that studies on the role of mPGES in PGE₃ synthesis are limited, here we will only elaborate on EPA availability and COXs regulation of the biosynthesis of PGE₃ in cancer cells or tumor tissues.

Membrane phospholipid pool

In contrast to the general consensus of PGE₃ biosynthesis being relatively low compared to that of PGE₂ in the normal cells or tissues as described earlier, the formation of PGE₃ appears to be similar to that of PGE₂ in cancer cells or tumor tissues when similar levels of EPA and AA are incorporated in the membrane phospholipid. We have found that the formation of PGE₃ was similar in the A549 non-small cell lung cancer (NSCLC) treated with EPA alone compared to that of the same cells treated with a combination of equal concentration of EPA and AA [21]. In line with this, the formation of PGE₃ and PGE₂ was generated at equivalent levels (ratio of PGE₃/PGE₂ > 1.0) when a similar amount of EPA and AA were incorporated into the membrane phospholipids of MC-26 mouse CRC liver tumors [75]. However, the actual kinetics of the formation of PGE₃ in cancer cells or tumor tissues when both AA and EPA are present has yet to be fully understood, although Hawcroft et al. have indicated that biosynthesis of PGE₃ in tumor tissues might also be associated with the ratio of EPA/AA, i.e., the ratio needs to be higher than 0.1 in order to allow biosynthesis of PGE₃ [75]. The formation of PGE₃ is relatively proportional to the amount of EPA incorporated in cell membrane of tumor tissues, but certainly the relationship is non-linear. For example, the amount of EPA incorporated in phospholipids of MC-26 mouse CRC cells liver tumor in mice fed with 5% EPA free fatty acids, was almost 3-fold higher than that from mice fed with 2.5% EPA free fatty acids. The formation of PGE₃ was only about 2-fold higher in livers from mice fed with 5% EPA free fatty acids compared to that in livers from the 2.5% group [75]. This could potentially be affected by the amount cytosolic fatty acid release from the membrane phospholipids and the status of the COXs or mPGES-2 proteins as n-3 fatty acids have the ability to inhibit the expression of COX-2 [76] and mPGES-2 [51].

cPLA₂

The release of AA or EPA from the phospholipids of cell membrane is catalyzed by phospholipases and cPLA₂ plays a critical role in regulation of eicosanoid metabolism [77,78]. For example, macrophages from ${\text{cPLA}}_2^{-/-}$ mice release much less AA than macrophages from WT mice upon inflammatory stimuli, such as phorbol mysristate acetate and lipopolysaccharides (LPS) [77]. Consequently less PGE₂ is produced in cPLA₂ deficient macrophages compared to the controls [77]. By profiling of the whole cell metabolomics using a combination of a matrix-assisted laser desorption (MALDI) mass spectrometry and LC/MS-MS, our recent study suggested that cPLA₂ status could be critical for regulation of EPA derived PG metabolism in cancer cells or tumor tissues because of its ability to affect the hydrolysis of these fatty acids from the membrane phospholipids [68]. While both A549 and H596 NSCLC cells express similar levels of COX-2 and mPGES-1, the formation of PGE₃ was substantially lower in H596 cells than in A549 cells. This appears to be due to the lower expression and activity of cPLA₂ in H596 cells than that of A549 cells. As a result, even though the abilities of EPA incorporation into membrane phospholipids were similar in both cell lines, the release of EPA from the phospholipids in H596 cells was far less than that of A549 cells because low cPLA₂ expression and activity limited the amount of free EPA presented to COXs. As described earlier, cPLA₂ hydrolyzes EPA and AA in the membrane phospholipids at similar rate, but it is less active in releasing DHA from the membrane phospholipids [9]. The differential activities on the release of EPA and DHA could potentially influence the biosynthesis of PGE₃ when both these n-3 fatty acids are incorporated into the membrane phospholipids. Based on our unpublished data, it is possible that the presence of DHA might reduce the formation of PGE₃ from EPA. This is based on the fact that the levels of PGE₃ in lung tumor tissues from mice treated with Lovaza, a fish oil supplement containing equal amounts of EPA and DHA, were only about 50% of that in the tumors from mice treated with same concentration of EPA alone. This may be caused by DHA competing with EPA for incorporation into phospholipids (Yang P unpublished data). Studies delineating the impact of DHA on EPA derived PGs and on EPA elicited anti-inflammatory and anti-proliferative activities will be interesting as the two n-3 fatty acids in general are administered simultaneously. Taken together, these studies suggested that the bioavailability of EPA, which could be regulated through the incorporation into phospholipids as well as the release from phospholipids, is important in PGE₃ metabolism in cancer cells.

Cox-2

The critical role of COX₂ in PGE₃ metabolism in cancer cells and tumors has been documented in a number of studies [21,24]. By applying a sensitive and specific LC/MS-MS method in COX-2 expressing A549 cells, Yang et al. have demonstrated that the formation of PGE₃ in these cells appears to be mainly mediated through COX-2 pathway because the level of PGE₃ was reduced by 50% when the cells were exposed to a combination of EPA (25 µM) and celecoxib, a selective COX-2 inhibitor. Compared to EPA alone, a combination of EPA and COX-1 inhibitor did not reduce the formation of PGE₃ in this particular cell line [21]. Similarly, Xia et al. and Hawcroft et al. reported that the formation of PGE₃ in EPA treated mouse melanoma and human HCA-7 CRC cells was blocked by indomethacin [79] or a selective COX-2 inhibitor, SC-236, respectively [24]. The important role of COX-2 in regulating PGE₃ metabolism was further supported by the finding that a substantially lower amount of PGE₃ was detected in the H1299 cells (COX-2 null NSCLC cells) and its xenograft tumor tissues compared to that in A549 cells (COX-2 constitutively expressed) and its xenograft [80]. The formation of PGE₃ was partially blocked in the COX-2 knockdown A549 cells compared with the control siRNA administrated A549 cells. Furthermore, while the level of PGE₂ was doubled in the colon mucosa from irradiated azoxymethane (AOM) treated rats compared to that in the colon mucosa from non-irradiated controls, there was 3-fold increase in colonic mucosal PGE₃ from irradiated AOM-rats than that of non-irradiated AOM-rats which can be due to the higher mucosal COX-2 expression in the irradiated rats than non-irradiated rats [51]. The increased PGE₃ in these studies might be linked to the anticancer activity of EPA or fish oil containing EPA. Literature on studies of different n-3 fatty acids, either EPA alone or menhaden oil in various tumors, including NSCLC, colon, breast and pancreatic cancer, consistently demonstrated that tumor cells have the ability to notably increased the ratio of PGE₃/PGE₂ in the tumor tissues of interest. Intriguingly, the ratio of PGE₃/PGE₂ was 3-fold higher in EPA free fatty acid (5%) treated MC-26 mouse CRC cell liver tumors than that in the normal liver tissues of this mouse model [75], suggesting that there should be relatively higher PGE₃ production in cancer cells as oppose to that of normal cells that potentially is due to reduced expression of COX-2 enzyme in normal tissues. Thus, these studies together provide strong support that PGE₃ metabolism is not only dependent on COX-2, but is also highly associated with cPLA₂ expression and activity. In light of COX-2 expression and activity being an important regulator of PGE₃ synthesis, it certainly challenges the concept of using combination of the n-3 fatty acids, especially EPA, and selective COX-2 inhibitor simultaneously in cancer therapy.

EPA induced changes in 2 and 3-series PGs in cancer

When EPA instead of AA is incorporated into cell membranes, not only would less AA-derived products be available but the EPA-derived substrates and products are typically less active than AA-derived substrates and products with potentially different biological activities [64]. Studies have shown that the 3-series eicosanoid metabolites are generally less pro-inflammatory than the homologous 2-series [21,22,45]. Thus, the role of n-3 fatty acids on modulation of eicosanoids has been one of the molecular mechanisms that is currently heavily studied in the context of n-3 fatty acid elicited anticancer activities [12]. The early studies conducted on n-3 fatty acids and cancer prior to the year of 2000 were mainly focused on the reduction of PGE₂ in n-3 fatty acid treated cancer cells or tissues. For examples, consumption of a diet enriched with menhaden oil significantly reduced PGE₂ metabolites in both plasma and tumor tissues in mice bearing the PG-producing HSDM1 fibrosarcoma [81]. The reduction of PGE₂ in either tumor cells or tissues has been documented in EPA, DHA or menhaden oil treated AOM-induced rat colon tissues [82–84], fish oil treated human prostate cancer DU145 xenograft model [85], 7,12-dimethylenz[a]anthracene induced rat mammary carcinoma [86], and EPA or DHA treated human lung mucoepidermoid carcinoma [87]. The reduction of PGE₂ appeared to be associated with the inhibitory effect of either fish oil (both EPA and DHA) or one of the n-3 fatty acids alone in all the aforementioned tumor models.

It was only after the year 2000 when we and other investigators established a sensitive and specific LC/MS-MS method which allowed determination of the 3-series PGs, especially PGE₃, in various biological matrices [51,67,88,89] that the role of PGE₃ in influencing the growth of cancer cells or tumor tissues began to be documented (Table 2). The anti-proliferative effect of EPA might be mediated not only through reduction of PGE₂, but also through concomitant increase of PGE₃ in A549 cells, because the COX-2 selective inhibitor, celecoxib, but not the COX-1 inhibitor SC-560, blocked formation of PGE₃ as well as antiproliferative effect of EPA in the A549 cells [21]. EPA induced inhibition of cell proliferation of A549 cells constitutively expressing COX-2 was almost 10-times stronger than that of H1299 cells (COX-2 null), and 2-fold stronger than that of COX-2 knockdown A549 cells. Additionally, dietary menhaden oil significantly inhibited the growth of A549 tumors but showed no tumor inhibitory effect in H1299 xenografts, which could be linked to the relatively higher ratio of PGE₃/PGE₂ in A549 xenografts fed with menhaden oil compared to that of H1299 tumors [80]. In line with this, Vanamala et al. reported that fish oil enriched diet protected against radiation enhanced colon cancer by reduction of PGE₂ and increased formation of PGE₃ in the colonocytes in an AOM rat colon cancer model [51]. Similar results, i.e., reduction of PGE₂ and increased PGE₃ were identified in menhaden oil treated human pancreatic cancer BxPC3 xenografts [50] and in EPA treated mouse MC-26 mouse CRC cells liver tumors [75]. These results could contribute to the mechanisms of anticancer activities of n-3 fatty acids. Furthermore, two studies using mice carrying the Fat-1 transgene, which restores the synthesis of n-3 fatty acids (EPA and DHA), inoculated with either mouse melanoma B16 cells or mouse HER2 positive breast cancer E0771 cells, showed that the development of melanoma and HER2 positive mammary gland tumors were substantially reduced after 15 days of cell inoculation while the tumors in the Fat-1 wild type (WT) mice continued to grow [79,90]. Interestingly, Xia et al. demonstrated that the reduction of PGE₂ was moderate in the stromal tissues and stronger in the tumors in the Fat-1 WT mice, while PGE₃ was markedly increased in both the stromal and the tumor tissues in the Fat-1 transgenic mice with melanoma [79]. Similarly, there was notably increase in PGE₃ production in the mammary tumors of the Fat-1 transgenic mice compared to that in the Fat-1 WT mice [90]. The significant contribution of COX-2 to the chemopreventive activity of fish oil is further strengthened by recent findings revealing that higher consumption of n-3 PUFA (EPA and DHA) lowered the risk of aggressive prostate cancer with this effect being more pronounced in men carrying a particular COX-2 variant [14]. Therefore, these studies suggested that the antiproliferative activity of n-3 fatty acids, especially EPA, could be enhanced by the formation of PGE₃ primarily through COX-2, and COX-2 could be a pivotal target for EPA or fish oil mediated anti-proliferative or chemopreventive activities.

Table 2.

Dietary n-3 fatty acids targeting 3-series PGs in cancer prevention.

Cancer type	Models	Interventionsa	Cancer outcome	PGE2	PGE3	References
Breast	Fat-1 mice injected with E0771 murine breast cancer cells	The diet contained 10% safflower oil with high LA and less than 0.1% n-3 fatty acids.	No palpable tumors in the fat-1 mice but 600 mm tumors in WT mice	No difference in the tumor	↑ From undetectable to 7.65 of ng/mg in tumor	[90]
Colon	AOM induced Sprague– Dawley rats	15% fat in the diets: a FO diet (EPA, 18.2% + DHA, 11.3%); a CO diet (LA 55.4%)	Not reported	↓ 78% in mucosa	↑ From undetectable to 1.43 pg/µg protein	[51]
Colon cancer liver metastasis	BALB/c AnN mice with intrasplenic injection of MC-26 cells	AIN-93G with 7% CO (control) vs. 4.5% CO + 2.5%EPA-FFA; 2% CO + 5%EPA-FAA	Liver metastasis was reduced; liver weight was significantly lower in 5% EPA-FFA group compared to controls	↓ 60% in the tumor in EPA-FFA fed mice (5% but not 2.5%)	↑ Dose dependently reaching 321 pg/mg in the tumors in 5% EPA-FFA treated animals	[75]
Lung	BALB/c athymic (Nu/Nu) mice injected with A549 cells	AIN-76 based: a SO diet with 15% SO; a FO diet with 10% of menhaden oil + 5% SO	Reduced tumor growth by 50–60% in A549 xenograft	↓ 52% in the tumor in A549 xenograft	↑ (0 to 0.89 ng/mg protein) in A549 xenograft	[80]
Melanoma	Fat-1 mice injected with B16 cells	A diet with 10% safflower oil	Reduced incidence of tumor formation and tumor growth rate	↓ In the tumor and surrounding tissues	↑ In the tumor	[79]
Pancreas	Nude mice injected with BxPC-3 cells	AIN-93G based diets: a CO diet (8.3% CO) vs. a FO diet (5.7% n-3 + 2.6% CO)	Tumor size reduced by 60%	↓ 35% in the tumor	↑ From undetectable to 0.5 ng/mg tumor	[50]
Colon	HCA-7 human CRC cells; MC-26 cells	0.05–0.8 µM EPA-FFA	Apoptosis induced	↓ 90% in HCA-7 cell conditioned medium (maximal synthesis at 24 h)	↑ In MC-26 cells	[24]
Lung	A549 human NSCLC cells	100 µM EPA, DHA	Not reported	↓ 64% in cells treated with EPA ↓ 48% in cells treated with DHA	↑ From 0.43 ng/5 million cells in control cells to 5.12 ng/5 million cells in treated cells	[89]
Lung	A549 human NSCLC cells	10–50 µM EPA	Reduced proliferation and induced cell death	Not reported	↑ Intracellular and extracellular production	[21]
Lung	H1299 and A549 human NSCLC cells	10 and 50 µM EPA	Reduced cell growth	Not reported	↑ Dose dependently ↑ PGE3/PGE2 ratio	[80]
Pancreas	Human pancreatic cancer cells BxPC-3	10 µM EPA	Decrease cell growth	↓ to below baseline	↑ Extracellular production to 209 pg/106 cells	[50]

^aCO: corn oil; FO: fish oil; SO: soybean oil; FFA: free fatty acid.

Potential mechanisms in PGE₃ anticancer activities

In comparison to our understanding of the biosynthesis of the n-3 series of PGs, knowledge of their biologic functions is limited. To understand the mechanism(s) underlying the effects of PGE₃ on cancer progression, researchers have been investigating how PGE₃ regulates tumor growth and its signaling pathways. It appears that the multiple signaling pathways including the three major ones AKT, ERK1/2 and PKA are involved in PGE₃ elicited anticancer activity (Fig. 2). Several reports have shown that PGE₃ suppressed tumor growth by inhibiting angiogenesis, cell invasion, cell growth and survival, which could be mediated through the EP receptors. Studies from our laboratory indicate that PGE₃ inhibits proliferation of A549 cells, whereas PGE₂ slightly stimulates the growth of these cells [21]. PGE₃ appears to inhibit the proliferation of mouse melanoma B16 cells by induction of apoptosis [79]. A similar effect was observed in human pancreatic cancer BxPC3 cells, when the cells were exposed to PGE₂ and PGE₃ [50]. In the same study, the anti-proliferative activity of PGE₃ was shown to be mediated through both the EP₂ and EP₄ receptor using EP₂ and EP₄ transfected MiaPaca cells (EP receptor null). Interestingly, the anti-proliferative effect of PGE₃ does not appear being mediated through PKA/cyclic AMP pathways. Others have shown that PGE₃ is less effective in stimulating cell proliferation and IL-6 production [45,91]. Additionally, PGE₂ eliminated the growth inhibitory effect of fish oil in hepatoma cells, while PGE₃ reduced the invasiveness of the same cells pretreated with safflower oil [92]. Moreover, PGE₃ suppressed the induction of angiopoietin-2 and resulted in inhibition of angiogenesis in human umbilical vein endothelial cells [93]. PGE₃ also modulated COX-2–mediated invasion and angiogenesis in brain-metastatic melanoma [94]. Furthermore, our studies demonstrated that PGE₃ inhibits proliferation of A549 cells and acts as an antagonist to PGE₂-mediated increases in cell proliferation [21]. Hawcroft et al. observed that PGE₃ antagonizes the protumorigenic activity of PGE₂ in EP₄ transfected HT-29 CRC cells that lack expression of COX-2 and other EP receptors, but not in HCA-7 CRC cells with higher endogenous levels of PGE₂ [24]. What was also interesting is that PGE₃ acts as an agonist when exposed alone to LoVo human CRC cells, while it becomes antagonist in the presence of natural ligand PGE₂ in these particular cells. Nevertheless, the evidence is limited and more research is necessary to address the question whether PGE₃ activates one or more of the EP receptor axis during tumor development.

Fig. 2.

Proposed schematic signaling pathways for PGE₃ induced EP receptor activation [80,121].

Recently the impact of downstream signaling by PGE₃ has been studied more rigorously regardless of how the EP receptors are involved. PGE₃ appears to downregulate PI3kinase signaling by either increasing PTEN expression [79] or directly suppressing the phopsphorylation of Akt, as opposed to increased pAkt by PGE₂ in these particular cells [21]. PGE₃ also inhibits the expression of HER3 and cMYC proteins which might contribute to the antitumor effect of n-3 fatty acids derived from de novo synthesis in Fat-1 transgenic mice. It is not clear whether or not the influence of PGE₃ on cell signaling proteins is mediated through specific EP receptors, which needs to be further investigated. Overall, these studies suggest that the 3-series metabolite of EPA, PGE₃, is associated with EPA elicited anti-proliferative activity in various types of tumors and could serve as a biomarker for EPA intake and biological response.

Prostaglandin 3-series as biomarkers for EPA status

Blood PGEs

The levels of lipids in cellular membranes reflect the net outcome of dietary intake, absorption, transport and metabolism of fats. Plasma EPA concentration increases in response to dietary EPA or DHA [32]. Plasma n-3 fatty acid levels have been used in clinical trials in testing the beneficial effects of n-3 fatty acids on a variety of debilitating conditions, including cancer [95]. However, plasma n-3 levels only provide evidence of n-3 fatty acid uptake, but not the utility of these n-3 fatty acids in relevant tissues, such as tumors, because the biosynthesis of PGE₃ in the tumor tissues or cells could be mediated by the expression and activity of other enzymes mentioned earlier that are involved in de novo prostaglandin synthesis. In terms of measuring eicosanoids in the blood, there is not sufficient evidence to extract meaningful data. Kearns et al. studied the effect of dietary n-6 and n-3 fatty acid ratios on eicosanoid production in peripheral blood mononuclear cells (PBMCs) in young and aged dogs [69]. The authors isolated PBMCs from dogs that were treated with different ratios of n-6/n-3 diets and stimulated with LPS before measuring prostaglandins. Interestingly, they found that PGE₂ production was not affected by dietary fatty acid ratios but there was a significant higher PGE₃ production in PBMCs from dogs fed with high n-3 diet. In humans, a few studies have reported blood prostaglandin production after dietary n-3 fatty acid interventions. Most data were limited to PGE₂ in ex vivo lipopolysaccharide stimulated PBMCs [96,97] and whole blood [98,99]. Dosing with 1 g/d EPA plus DHA caused the biggest decrease in PGE₂ in stimulated PBMCs [100]. One recent randomized controlled trial conducted by Dawczynski et al. showed that higher consumption of PUFAs increase EPA and PGE₃ in both plasma and red blood cells in mildly hypertriacylglycerolemic patients [101]. The available data suggested that dietary EPA consumption is correlated with reduction of PGE₂ production and increase of PGE₃ in the blood.

Tissue PGs

Compared to blood PG levels, the amount of PGE₂ and PGE₃ in tissue can be easily assayed by ELISA or LC/MS-MS, which are commonly used in research laboratories. In human studies, dietary EPA reduced PGE₂ and ornithine decarboxylase activity [83,102,103]. Several rodent studies have found that dietary EPA reduces tissue PGE₂ by 50–90% depending on models used in the studies and doses of EPA given to the animals [84,104–106]. In a human study, fish oil (MaxEPA) with 5.4 g EPA plus DHA decreased colonic PGE₂ and TxB₂ within three weeks in patients with inflammatory bowel disease [107]. One piece of information lacking in these studies was PGE₃ levels because reliable internal standard was not available for the analysis.

In carcinogenic conditions, COX-2 is commonly overexpressed which leads to induced PGE₃ formation. Several in vivo studies using rodent cancer models have reported increased PGE₃ concentrations in tumors or tumor bearing animals upon dietary EPA consumption (Table 2). Interestingly, a few studies have indicated that an increased dose of EPA in both cancer animals [75] or cell lines [68] can dose dependently increase PGE₃ production which further supports the theory that PGE₃ can be used as a biomarker for EPA intake in vivo. Our recent study also suggested that the ratio of PGE₃/PGE₂ in mouse lung tumor tissues was significantly increased by 40-fold in mice fed with EPA enriched diet compared to that in soybean-fed mice [68]. Therefore, both in vitro and in vivo data appear to support that tissue PGE₃ production is responding to dietary EPA, indicating that PGE₃ can be a useful biomarker for evaluation of EPA associated anti-proliferative activity. Although tissue production of PGE₃ may be quantifiable as a relevant biomarker linked to EPA elicited anticancer activity, obtaining tissue from human subjects would be inconvenient and prohibitory in some cases. For these reasons, development of blood or urine biomarkers is a preferable strategy.

Urinary PG metabolites

Quantification of urinary PGE metabolites [108] and the major urinary metabolite of PGI [109] can potentially be another biomarker for fatty acid status. In 1988, Kivits et al. reported that in rats, an AA diet increased urinary tetranor-prostaglandin E₁ (PGE-M) which is the major urinary metabolite of PGE₂ [110]. The authors also detected much higher conversion of PGE₃ to Δ¹⁷-tetranor-prostaglandin E₁, the major metabolite of PGE₃, upon dietary EPA feeding. There are some human studies that reported changes in PGE-M in the urine from subjects who consumed n-3 fatty acids. Ferretti and colleagues conducted a clinical trial involving a group of healthy male volunteers. They found that a low-fat diet or a high-fat diet supplemented with fish oil decreased the urinary PGE-M by 14% [111,112]. Interestingly, body weight had a significant effect on the difference in the urinary PGE-M levels in that lighter weight individuals showed a bigger reduction in urinary PGE-M levels. Murff et al. demonstrated that the dietary intake of n-3 fatty acids in women but not in men was negatively correlated to urinary PGE-M concentrations [113]. However, Young et al. performed a similar study with n-3 fatty acids in healthy postmenopausal women and did not observe changes in PGE-M in urine samples from people who consumed EPA and DHA for 8 weeks [114]. These data indicate that PGE-M as a marker for estimation of consumption of dietary fish oil may be appropriate in certain populations. More studies with larger numbers and more diverse participants are necessary to verify this observation. However, one would think that measuring urinary PGE₃ or its metabolites will provide a better approach than detecting reduction of the metabolites from the 2-series PGs upon EPA intake.

Fischer et al. reported that 10-fold increase of urinary PGE₃ was detected when the cod liver oil was ingested [115]. Additionally, two urinary PGE₃ metabolites, 7α, 11α-dihydroxy-5-ketotetranorprosta-9,13-dienoic acid and 11α-hydroxy-5-ketotetranoprosta-4(8),9, 13-trienoic acids were identified in the rat urine after the PGE₃ was administered to rat [116]. Recently data showed that two related 2,3-dinor metabolites were the major urinary metabolites of 6-keto-PGF_2α from prostacyclin (PGI₃) in mice [117], which may facilitate the development of methods to measure urinary metabolites of the 3-series PGs. Whether these metabolites could serve as urinary biomarkers for EPA status and whether this correlates to its biological responses need to be explored in cancer settings. Another important issue is that urinary metabolite levels are markers of the total body production of PGs, which may be altered in a different manner than specific tissue levels of PGs [118]. Therefore, the questions are whether systemic changes will mirror changes in a particular tissue and whether the function of these eicosanoid metabolites will correlate with changes in absolute amounts of PGs. We anticipate that systemic levels of PGs should represent dietary fatty acid consumption. Whether fatty acid derived PGs are biologically active will depend on not only the substrate availability but accessibility of the receptors that receive the signals as well.

Conclusion and challenge

Even though researchers have been studying n-3 fatty acids and cancer for decades, there is still growing interest in defining the effects of n-3 fatty acids, EPA or DHA, in preventing the development of malignant diseases. Preclinical studies continually support the notion of n-3 fatty acids being effective in preventing the initiation and progression of various cancers, however, results from human clinical evaluation on the efficacies of n-3 fatty acids in cancer management are still controversial. This certainly could be due to several factors, such as the type and dose of n-3 fatty acids used in the studies, nutritional status of patient, and lack of biomarkers to monitor bioavailability in in situ tissues. Thus, it becomes imperative to concomitantly analyze both plasma levels of EPA and DHA as well as their metabolites, such as PGE₃, in the tumor tissues. Compared to other 3-series prostaglandins, accumulating in vivo evidence supports the findings that tumor PGE₃ derived from EPA or fish oil appears to be associated with EPA elicited antiproliferative activity. However, most studies have focused thus far on determination of EPA derived eicosanoids, especially PGE₃ in cancer cells or tumor tissues. We contend that more effort needs to be taken to identify the regulation of PGE₃ de novo synthesis and its influence on downstream signaling, which will ultimately improve our understanding on the role of EPA in cancer prevention and treatment. Identifying the key factors mediating the ability of cancer cells to produce PGE₃, such as substrate availability when other n-3 (DPA and DHA) and n-6 fatty acids (AA) are present and expression of other enzymes involved in PGE synthesis, such as mPGES, are critical for understanding the biosynthesis of PGE₃ in the tumor tissues and its biological consequences. An advanced technology, such as MALDI mass spectrometry coupled with LC/MS/MS, makes it possible to measure PGE₃ metabolites in different surrogate tissues, such as erythrocytes, tissues and urine and further validate PGE₃ metabolites as potential biomarkers for the COX-2 mediated anticancer activity of EPA. This information is pivotal for clinical studies of n-3 fatty acids in cancer management. Additionally, it is critical important for delineating the toxicity of the n-3 fatty acids and for optimizing the use of the n-3 fatty acids in the prevention and treatment of cancer, which has not been fully addressed.