Distinguishing Health Benefits of Eicosapentaenoic and Docosahexaenoic Acids

Long chain omega-3 polyunsaturated fatty acids (LC n-3 PUFAs) are recommended for management of patients with wide-ranging chronic diseases, including coronary heart disease, rheumatoid arthritis, dementia, and depression. Increased consumption of eicosapentaenoic acid (EPA) and docosahexaenoic acid (DHA) is recommended by many health authorities to prevent (up to 0.5 g/day) or treat chronic disease (1.0 g/day for coronary heart disease; 1.2–4 g/day for elevated triglyceride levels). Recommendations for dietary intake of LC n-3 PUFAs are often provided for α-linolenic acid, and for the combination of EPA and DHA. However, many studies have also reported differential effects of EPA, DHA and their metabolites in the clinic and at the laboratory bench. The aim of this article is to review studies that have identified divergent responses to EPA and DHA, and to explore reasons for these differences. In particular, we review potential contributing factors such as differential membrane incorporation, modulation of gene expression, activation of signaling pathways and metabolite formation. We suggest that there may be future opportunity to refine recommendations for intake of individual LC n-3 PUFAs.

health and government organizations. Most of these organizations have issued recommendations for combined intake of EPA and DHA, but do not provide advice for the ratio of EPA to DHA. The French Agency for Food, Environmental and Occupational Health & Safety [20] is a notable exception, providing separate recommendations for dietary intake of EPA and DHA (Table 1). Following the American Psychiatric Association recommendation of 1.0 g/day DHA+EPA for the treatment of affective disorders ( [23], Table 1), McNamara [24] proposed a 2:1 ratio of EPA:DHA for optimal patient management.
Differential associations for cell membrane EPA and DHA levels and resting state electroencephalographic (EEG) activity have been reported. DHA was associated with fast frequency EEG activity while EPA was associated with slow frequency EEG activity in 46 adolescent boys with attention deficit hyperactivity disorder [25]. Not surprisingly, the health benefits of the LC-n-3 PUFAs have often been ascribed to particular types of LC-n-3 PUFAs. For example, studies have reported blood pressure lowering effects of DHA but not EPA [26,27], while EPA has been reported to be more efficacious than DHA in reducing platelet activation [28,29]. The differential response to EPA and DHA suggests that an opportunity might exist to customize advice for the types or ratios of LC n-3 PUFAs to be used in the treatment of particular conditions or diseases. The implementation of such a strategy is certainly achievable, with a plethora of commercially available dietary marine oil supplements of defined LC n-3 PUFA composition [30]. For example, EPAX 5510 TG/N is an EPA-rich oil (EPA/DHA ratio of 5:1), while EPAX 1050 TG/N is a DHA-rich oil (EPA/DHA ratio of 1:5) [31]. Salunkhe et al. [32] recently identified a marine bacterial isolate that produces high concentrations of EPA (60% of total fatty acid content) at 30 °C , with no detectable production of DHA. The strategy may also be achievable by considering the dietary intake of fish and meat sources. For example, of fish high in LC n-3 PUFAs content, the ratio of EPA:DHA can range from ~1:2 for Atlantic Salmon and Ocean Trout to ~1:30 for Snook [33].
The purpose of this review is to examine studies that have reported heterogeneity with respect to effects of the LC n-3 PUFAs, chiefly EPA and DHA, and to examine some of the reasons for these differential responses. In particular, we review the impact of differential metabolism, membrane incorporation, modulation of gene expression, and activation of signaling pathways for EPA and DHA. Table 1. Recommendations for Long chain omega-3 polyunsaturated fatty acids (LC n-3 PUFA) intake obtained from selected national and international health and government organizations.

Health Organization
Country Recommendation Ref.
National Heart Foundation Australia 0.5 g/day EPA + DHA plus 2 g/day ALA to lower the risk of coronary heart disease; 1.0 g/day EPA + DHA plus 2 g/day ALA for patients with documented coronary heart disease; 1.2-4.0 g/day EPA + DHA for patients with elevated serum triglyceride levels. [22] American Heart Association USA ≥2 fish meals/week plus oils rich in ALA in subjects without coronary heart disease; 1.0 g/day EPA + DHA for patients with documented coronary heart disease; 2.0-4.0 g/day EPA + DHA for patients with elevated serum triglyceride levels. [34] World Health Organization International 0.2-0.5 g/day EPA+DHA to prevent coronary heart disease and ischemic stroke. [35] American

Evidence for Differential Responses to EPA and DHA
Although EPA and DHA are both long chain polyunsaturated fatty acids (EPA, C20:5n3; DHA, C22:6n3), the molecules are often reported to produce biochemical and physiological responses that are qualitatively and quantitatively different to each other. Table 2 provides a non-exhaustive overview of studies that report differential effects of EPA and DHA, and their metabolites. Each individual study represented in Table 2 was internally controlled, thus allowing direct comparisons of responses to the fatty acids under identical experimental conditions. This is an important point, since differences in experimental design between studies have sometimes led to inconsistent conclusions in regard to the effects of the LC n-3 PUFAs. When considering the blood pressure lowering effects of LC n-3 PUFAs for example, the reader needs to be cognisant of the effects of dose and duration of treatment, baseline characteristics of the subject for dietary intake of LC n-3 PUFAs and blood pressure, subject age, co-morbidities, and use of anti-hypertensive medications. It is nonetheless noteworthy that several studies in Table 2 reported response, and/or lack of response that was based on the examination of a single concentration of the LC n-3 PUFAs. In those studies it is possible that information obtained from construction of full dose-response curves to EPA and DHA might have shown the same qualitative response, albeit with different potencies and/or efficacies. For example, Yusufi et al. [39] reported an inhibitory effect of low concentration (10 μM) DHA, but not EPA on EGF-stimulated mitogenesis in mesangial cells. However, an inhibitory effect of EPA, quantitatively similar to that of DHA, was revealed when the LC n-3 PUFAs were used at a 10-fold higher concentration [39]. Similarly, studies that rely on single time-points for measurement of response may not capture changes that occur after short-term-, or more typically after long-term treatment regimens. This point is exemplified by the early incorporation (2-14 days) of EPA, but not DHA into erythrocyte membranes, with DHA incorporation occurring later (14-28 days) [40].
Interestingly, there are instances where EPA and DHA produce opposite responses, for example in the regulation of c-abl proto-oncogene, glutathione S-transferase A1, and breast cancer type 2 susceptibility protein gene expression [41], modulation of pulse pressure in a rat model of hypertension [27], and changes to resting heart rate in healthy human subjects [42] (Table 2). These examples suggest fundamental differences in the mechanism of action for the different LC n-3 PUFAs. Discussion of factors likely to contribute to the divergent responses to individual LC n-3 PUFAs is reported in Section 3. Table 2. Summary of selected studies identifying differential effects of docosahexaenoic acid (DHA), eicosapentaenoic acid (EPA) and their metabolites. c-abl proto-oncogene: DHA ↑, EPA ↓.
* Result likely linked to differential incorporation of the LC n-3 PUFAs into diacylglycerol and not different affinities of the phospholipids for RasGRP.

Regulation of Transcription Factors
LC n-3 PUFAs regulate gene transcription by binding to a variety of nuclear receptors, including the retinoid-activated nuclear receptor, RXRβ [64] (via association with the brain fatty acid-binding protein-7 transporter [65]), and peroxisome proliferator-activated receptor (PPAR) [66,67]. In 3T3-L1 adipocytes, 125 µM DHA upregulated PPARγ mRNA expression levels while 125 µM EPA was without effect [46]. PPARγ is a transcription factor for the adiponectin gene, and activation of PPARγ leads to increased synthesis of adiponectin. DHA stimulated secretion of greater amounts of adiponectin from adipocytes than EPA, and only the DHA response was blocked by an antagonist of PPARγ [46]. The authors concluded that there may be independent actions of EPA and DHA on PPARγ/adiponectin signaling in adipocytes [46]. The selectivity of DHA over EPA for PPARγ activation may not apply to all cell types, with evidence for stimulation of PPARγ activity by both EPA and DHA in human kidney-2 (HK-2) cells [68], and by EPA in human umbilical vein endothelial cells [47]. It is possible that differential effects of the LC n-3 PUFAs may be caused by metabolism of the LC n-3 PUFA to its metabolites. A number of oxidation products of DHA were found to be more potent activators of PPARγ than DHA [69,70], although metabolites resolvin D1 and resolvin E1 were without effect [71]. It is not yet known whether EPA and metabolites of EPA have differential activity at PPAR. The contribution of LC n-3 PUFA metabolites to differential responses observed for EPA and DHA are discussed in Section 3.4.
In addition to the direct effects of EPA and DHA on transcription factors, these LC n-3 PUFAs also indirectly modulate transcription factors. For example, LC n-3 PUFAs inhibit activity of the transcription factor, nuclear factor kappa B (NFκB) by attenuating phosphorylation and degradation of the inhibitory factor, IκB-α (reported for EPA [72]), and by inhibiting the recruitment of toll-like receptor 4 to lipid rafts (reported for DHA [73]). Although 100 µM DHA was more effective than 100 µM EPA in retaining IκB-α in the cytosol of lipopolysaccharide (LPS)-stimulated macrophages, EPA and DHA were equally effective at inhibiting LPS-stimulated NFκB/DNA binding activity [43].

Receptor-Mediated Effects of EPA and DHA
Free fatty acids were identified as endogenous ligands for the orphan G-protein coupled receptor, GPR120 following extensive ligand-binding screening [74]. GPR120 is expressed in gastrointestinal epithelial cells [74,75], and in macrophages [76], and its activation by LC n-3 PUFAs leads to secretion of glucagon-like peptide-1 from epithelial cells [74,77] and anti-inflammatory effects in macrophages [76]. Recent findings revealed short and long isoforms of GPR120, both of which were responsive to free fatty acids [78]. Although differential maximal efficacies for phosphorylation of these isoforms was reported in HEK293 cells exposed to α-linolenic acid and DHA, a comparison of EPA and DHA-mediated receptor phosphorylation was not made [78]. Nonetheless, it was speculated that the length and degree of saturation of the fatty acids may affect efficacy [78], raising the possibility that differences in efficacy might also be observed for EPA (20 carbons, 5 double bonds) and DHA (22 carbons, 6 double bonds). Whilst further studies are required to test this hypothesis, differences in amplitude of response to EPA and DHA have been reported previously. For example, serum response element-luciferase (SRE-luc) activity in HEK293 cells expressing GPR120 and the SRE-luc promoter appeared to be greater for cells treated with DHA than EPA [76]. Furthermore, secretion of glucagon-like peptide-1 from mouse colon was significantly elevated after intracolonic administration of DHA, but not EPA [77]. It is thus possible that the above interactions of EPA and DHA with GPR120 could in part explain the greater efficacy of DHA compared to EPA for the inhibition of release of pro-inflammatory cytokines from human macrophages [43] ( Table 2). In that study, a single, high concentration of EPA and DHA was used (100 μM), and this was therefore likely to produce a maximal response (see [76]). Interestingly, potency does not appear to be different for EPA and DHA at GPR120. Potency of EPA and DHA for stimulation of calcium mobilization in HEK293 cells expressing GPR120, and promoter activity in HEK293 cells expressing GPR120 and SRE-luc promoter, were similar [74,76].

Incorporation of EPA and DHA into Phospholipids
Glycerophospholipids (phospholipids) are the main component of the cell membrane, and include structures such as phosphatidylcholine and phosphatidylserine. Dietary intake of EPA and DHA increases the LC n-3 PUFA content of phospholipids with an associated reduction in arachidonic acid levels [49,79,80]. The source of LC n-3 PUFAs may affect the extent to which EPA and DHA are incorporated into phospholipids. A recent study recruited healthy, young (20-50 years) men and compared incorporation of fish oil (EPA and DHA as re-esterified triglycerides or ethyl esters, with no free fatty acids) and krill oil-derived EPA and DHA (mainly bound in phospholipids, with 21%-22% free fatty acids) into plasma phospholipids [81]. A non-significant trend for higher incorporation of EPA and DHA from krill oil was reported [81]. Although the duration of the experiment was short (24 h), the investigators were careful to match the amount of EPA and DHA that was administered for the three preparations.
To investigate possible differential incorporation of EPA and DHA into membrane phospholipids, Judé et al. [79] fed dogs a diet containing a greater amount of DHA than EPA, for 8 weeks. As expected, the plasma concentration of DHA was greater than EPA. However, despite the intake of a DHA-rich diet, EPA was preferentially incorporated into erythrocyte and cardiac membrane phospholipids. The incorporation of EPA and DHA into phospholipids was also investigated in humans receiving dietary supplementation with LC n-3 PUFAs [14,82,83]. In these studies, EPA was also more efficiently incorporated into cholesteryl esters than DHA [14], and this was ascribed to a higher efficiency of lecithin-cholesterol acyltransferase (LCAT) activity for transfer of EPA from phosphatidylcholine to cholesteryl esters [82,84]. In contrast, DHA was preferentially incorporated into triglycerides [14], with DHA serving as a preferential substrate for diacylglycerol acyltransferase [52]. Interestingly, not all studies concur with this pattern of fatty acid metabolism. The supplementation of guinea-pigs with marine oils rich in EPA led to preferential incorporation of DHA into cardiac muscle total phospholipids [85]. Whilst species differences may be responsible for the different findings, this is unlikely to occur at the level of LCAT activity since LCAT activity in guinea-pig is intermediate between that reported in dog (lower activity than in guinea pig) and human (higher activity than in guinea pig) [86].
Both EPA and DHA decrease the liberation of arachidonic acid from phospholipids by inhibiting phospholipase A 2 activity [87]. Arachidonic acid is a substrate for cyclooxygenase and lipoxygenase enzymes, and the competitive reduction in arachidonic acid by EPA and DHA inhibits the generation of the 2-series prostaglandins, and 4-series leukotrienes [88,89]. As described above, EPA is a better substrate than DHA for LCAT activity [82]. The displacement of arachidonic acid by EPA in phospholipids increases cyclooxygenase-2-mediated production of PGE 3 , and lipoxygenase-mediated production of LTB 5 , from EPA, with a concomitant reduction in levels of PGE 2 and LTB 4 [90][91][92]. The PGE 2 -to-PGE 3 and LTB 4 -to-LTB 5 switching described by these investigators is likely to be of clinical importance in the in vivo modulation of disease since the EPA-derived products will tend to favour anti-inflammatory, anti-mitotic and anti-allergic activities [90][91][92].
Diacylglycerol is one of the lipid molecules into which EPA and DHA can be incorporated. This molecule has a key role in cell signaling, being a potent endogenous activator of conventional and novel subclasses of protein kinase C (PKC). Diets containing LC n-3 PUFAs lead to enrichment of these fatty acids in diacylglycerol [93], primarily in the sn-2 position [94]. Judé et al. [79] showed a preferential enrichment of cardiac membrane diacylglycerol with EPA compared to DHA in dogs that were fed a fish oil diet, while Madani et al. [56] showed preferential enrichment of Jurkat T cell diacylglycerol with DHA compared to EPA. In mice fed diets rich in EPA or DHA for 10 days, isolated splenic cells showed an apparent (data for n = 2 experiments only) differential enrichment of diacylglcerol with 18:1-22:6(n-3) and 18:1-20:1(n-9) (DHA fed mice) or 18:0-18:2(n-6) (EPA fed mice) [95]. The findings show that diets containing high levels of EPA or DHA can lead to differential incorporation of fatty acids, including LC n-3 PUFAs, into diacylglycerol. The question arises as to whether this type of response might translate to differential effects on PKC signaling. This was examined in an in vitro study that compared the effects of diacylglycerol containing EPA or DHA in the sn-2 position, on PKC activity [96]. Both EPA and DHA-containing forms of diacylglycerol stimulated concentration-dependent PKCα, βI, γ, δ and ε activity, however differential responses were not observed [96]. The implication of this finding is that whilst differential incorporation of EPA and DHA into membrane phospholipids may contribute to differences in health benefits of the LC n-3 PUFAs, this is unlikely to occur at the level of diacylglycerol/ PKC signaling.

LC n-3 PUFA Metabolites
LC n-3 PUFAs undergo enzyme-independent auto-oxidation under cell culture conditions, increasing with time and with the level of unsaturation of the fatty acid (DHA > EPA; [97]. Since oxidation products can have different activity to the parent LC n-3 PUFA, differential stability of EPA and DHA may impact on the pharmacological effect of these fatty acids. Multiple enzyme systems also convert LC n-3 PUFAs to metabolites that have biological activity [98,99] (Figure 1). The family of cytochrome P450 (CYP450) enzymes convert EPA to primary metabolites epoxyeicosatetraenoic acid (17, and hydroxyeicosapentaenoic acid , and DHA to epoxydocosapentaenoic acid (19, and hydroxydocosahexaenoic acid   [100]. Treatment of human subjects with 1.86 g EPA and 1.5 g DHA per day for 4 weeks caused a 4.7-and 2.1-fold increases in plasma levels of 17,18-EEQ and 19,20-EDP, respectively [101]. EPA is also converted to 18S-and 18R-resolvins E1 and E2 by aspirin-acetylated cyclooxygenase-2 (COX-2) and 5-lipoxygenase activity [102]. DHA is converted by aspirin-acetylated COX-2 to 17R-resolvins D1-D4 and 17R-protectin D1, and by 5-and 15-lipoxygenase to the 17S-series of these compounds [103][104][105].
Several studies have examined divergent activity EPA and DHA by studying the responses to their metabolites. Multiple isoforms of CYP450 contribute to the metabolism of EPA and DHA, and these can perform their catalytic function with different preference for the EPA and DHA substrates. For example, DHA is metabolized at a greater rate than EPA by CYP4F2, CYP4F3B, and CYP4F3A, while CYP4A11 and CYP2J2 preferentially metabolize EPA [99]. The response to the LC n-3 PUFAs will therefore be affected by expression levels of CYP450 isoforms in the tissues, as this will influence the degree to which EPA and DHA are metabolized. Importantly, the metabolites have different biological activities to each other, and to their parent LC n-3 PUFA (Figure 1). This point is exemplified by the markedly different potencies of EPA, DHA, and CYP450 metabolites of EPA and DHA for eliciting vasorelaxation. The potency of DHA for dilation of pre-constricted porcine coronary arterioles (apparent pEC 50 , 5.1) was ~336,000-fold lower than for its metabolite, 19,20-EDP (pEC 50 , 10.6) [106]. In contrast to the high potency of the DHA metabolite, the EPA metabolite 17,18-EEQ relaxed pre-constricted mouse mesenteric arteries with comparatively low potency (~pEC 50 , 7.0) [107]. Polymorphisms in the gene encoding for the CYP450 isoforms may contribute to variation in metabolism of EPA and DHA between individuals. Several polymorphisms have been identified in the CYP1A1 gene, including a CYP1A1.2 variant (Ile462Val) that has allelic frequency up to 10% in Caucasians, and up to 33% in Asians (see [124]). This variant form of CYP1A1 metabolizes EPA to 17,18-EEQ and 19-HEPE with a 2.1 and 5.2-fold higher efficiency than the wild type enzyme [124]. The ratio of efficiency of epoxygenation to hydroxylation is also different for the variant and wild-type enzymes [124]. Since the biological activity of the metabolites can differ (see above), polymorphisms may be expected to contribute to variability in differential effects of the LC n-3 PUFAs and their metabolites between individuals.
Differential responses have also been noted for the resolvins and lipoxins; families of autacoids with anti-inflammatory and pro-resolving activities. The resolvins mediate their effects, at least in part, through the activation of G protein coupled receptors; ALX/FPR2 and GPR32 for resolvin D1 and lipoxin A4 [71,108] and ChemR23 and BLT1 for resolvin E1 [108,125]. Some similarities exist in the functional response to the autacoids and their precursor molecules. For example, both resolvin D1 and DHA induced polarization of adipose tissue-infiltrating macrophages from an M1 to an M2 phenotype, as evidenced by the attenuation of M1 markers TNF-α and IL-6 and induction of the M2 marker Arg1 [126]. However, despite these similarities, differences between the LC n-3 PUFAs and their metabolites have also been noted. Whereas resolvin D1 potently induced macrophage phagocytic activity, DHA had the opposite effect, and whereas 1 nM resolvin D1 stimulated ROS production in peritoneal macrophages, DHA was without effect even when used at a 10,000-fold higher concentration [126]. It is thus possible that some of the differential responses reported for EPA and DHA might be attributable to the extent to which they are metabolized by the different enzyme systems to their endogenous autacoid metabolites. In addition to the differences observed between the metabolites and their parent LC n-3 PUFA, differences have also been reported for anti-inflammatory and pro-resolving activity between the various metabolites. For example, 17R-resolvin D1, formed by aspirin-acetylated COX-2, was more potent than lipoxygenase-generated 17S-resolvin D1 at inhibiting leukocyte infiltration in a mouse-model of peritonitis [105]. Differential kinetics for onset of anti-inflammatory and pro-resolving activity for the metabolites of arachidonic acid, EPA and DHA may contribute to some of the differential responses reported for these mediators. A stable analogue of 15-epi-lipoxin A4 inhibited pro-inflammatory cytokine production in a mouse-model of peritonitis, 4 h after administration [127]. At this time-point, resolvin E1 was without effect, however when assessed 12 h post-administration, it inhibited pro-inflammatory cytokine production with magnitude equal to, or greater than that observed for the lipoxin [127]. Few studies have examined the biological activities of the CYP450 metabolites, and further research into this area is warranted.

Conclusions
Long chain omega-3 polyunsaturated fatty acids (LC n-3 PUFAs), eicosapentaenoic acid (EPA) and docosahexaenoic acid (DHA), have been recommended for management of patients with wide-ranging chronic diseases, including coronary heart disease, rheumatoid arthritis, dementia, and depression. Although there is much evidence supporting uniformity of response to the individual LC n-3 PUFAs, there are also a plethora of studies that have shown qualitative and quantitative differences in response to EPA and DHA. In this review, we have explored some of the reasons for these differences. We propose that there are multiple factors that contribute to the differential effects of EPA and DHA, including differences in direct and indirect activation of transcription factors, impact of length, degree of saturation and stability of the fatty acid on efficacy, and differential efficiency for incorporation of the fatty acids into phospholipids. In addition, potency of the metabolites of EPA and DHA are often markedly different to the parent LC n-3 PUFA, and divergence in efficiency of enzymes to metabolize EPA and DHA can contribute to observed diversity in cellular response. We suggest that with improved understanding of the similarities and differences for EPA and DHA, an opportunity exists to customize recommendations for intake of EPA or DHA that can be tailored to the patient's condition.