Reduced prostaglandin F2 alpha release from blood mononuclear leukocytes after oral supplementation of omega3 fatty acids: the OmegAD study.

Omega-3 fatty acids, e.g., dokosahexaenoic acid (DHA) and eikosapentaenoic acid (EPA), ameliorate inflammatory reactions by various mechanisms, but the role of prostaglandins remains unclear. Our aim was to determine if dietary supplementation with a DHA-rich fish oil influenced the release of PGF2α from peripheral blood mononuclear cells (PBMC). In the OmegAD study, 174 Alzheimer disease patients received either 1.7 g DHA plus 0.6 g EPA or a placebo daily for six months. PBMCs from the 21 (9 on fish oil and 12 on placebo) first-randomized patients were stimulated with either lipopolysaccharide (LPS) or phytohemagglutinin (PHA) before and after 6 months. Our results showed that plasma concentrations of DHA and EPA increased significantly at 6 months in the omega-3 group. PGF2α release from LPS- (but not from PHA-) stimulated PBMC was significantly diminished in this group; no change was noted in the placebo group. PGF2α changes correlated inversely with changes in plasma DHA and EPA. Decreased IL-6 and IL-1β levels correlated with decreased PGF2α levels. The stimulus-specific PGF2α release from PBMC after 6 months of oral supplementation with the DHA-rich fish oil might be one event related to reduced inflammatory reactions associated with omega-3 fatty acid intake.

of the remaining patients received the 3 FA preparation and 12 (58-79 y; median 71 y; 4 women) the placebo capsules.
No change in peripheral blood neutrophil, monocyte, and lymphocyte cell counts were recorded after 6 months of 3 FAs supplementation. Patients were not given specifi c advice on food intake or time points for -3 capsule intake during the study. Food intake in the AD subjects will be reported separately. The two groups did not differ with regard to age, Mini Mental Test Scores (i.e., degree of cognitive deterioration), serum C-reactive protein levels, plasma DHA or EPA levels, blood pressure, body weight, or intake of aspirin.
The study was approved by the ethical committee of the Karolinska Institutet ( 7 ). The 3 FA treatment was safe and well tolerated.

Blood sampling
PBMCs were isolated form EDTA anticoagulated venous blood by means of Lymphoprep (Nycomed Pharma, Oslo, Norway) gradient centrifugation. The cell preparations obtained before and after treatment with 3 FAs, contained on average, 15 ± 5% monocytes and 85 ± 5% lymphocytes on both occasions (means and SD values). Corresponding fi gures for the placebo group were 15 ± 5% and 85 ± 5%, respectively. The cell viability in both groups was 96%, as assessed by trypan blue staining.
PGF2 ␣ release was measured using an enzyme immunoassay kit (Correlate-EIA, Assay Designs, Inc., Ann Arbor, MI) and is expressed in ng/ml. The lower limit for detection of PGF2 ␣ was annotated to be 3 pg/ml.

Plasma fatty acid analyses
Plasma fatty acids were analyzed by gas chromatography (THERMO TR-Fame column (30 m × 0.32 mm ID × 0,25 µm fi lm; Thermo Electron Corp., Waltham, MA) and results are given as the relative abundance of individual fatty acids ( 29 ). Data for all 174 patients in the OmegAD study have been given previously ( 7 ). Likewise, data for the present 21 patients have been given ( 28 ).

Statistical analyses
We used the Wilcoxon signed rank test for analyses of dependent data. For comparison of differences in responses between groups over time, we used a Mann-Whitney U test for independent data. For correlation analyses, the Spearman's rank correlation test was applied. P < 0.05 were considered signifi cant. We used median values surrounded by the values for the 25 th and 75 th percentiles.

Plasma fatty acids
As reported previously ( 28 ), at study entry DHA and EPA concentration in plasma were not signifi cantly different between the 3 FA and the placebo group. In the 3 FA group, give rise to the highly anti-infl ammatory metabolites resolvins and protectins, which interacts with prostaglandin synthesis ( 11 ).
The aim of this study was to evaluate the effects of oral supplementation for 6 months of 3 FA on PGE 2 release from peripheral blood mononuclear cells (PBMC) by measuring its stable metabolite PGF 2 ␣ . We used two different stimuli, one with effects mainly on monocytes [lipopolysaccharide (LPS)], and the second on T-lymphocytes [phytohemagglutinin (PHA)]. The study is a part of a trial, the OmegAD study, where a product rich in DHA was given to patients with mild to moderate AD. The goal of the OmegAD study was, among other things, to see if this 3 FA preparation would reduce the cognitive deterioration in AD ( 7 ).

SUBJECTS AND METHODS
This study included 25 patients. They were the fi rst to be randomized in the OmegAD study, described in detail in Freund-Levi, et al. ( 7 ). In summary, the double-blind, placebo controlled OmegAD study included a total of 204 patients (73 +/ Ϫ 9 y; 52% woman) with mild to moderated AD. Patients were randomized to 6 months of nutritional supplementation with a 3 fi sh oil rich in DHA or to placebo. Patients were treated daily with either 1.7 g DHA plus 0.6 g EPA (EPAX 1050TG (Pronova Biocare A/S Lysaker, Norway) or with an isocaloric placebo oil (1 g corn oil, including 0.6 g linoleic acid). EPAX 1050TG is a 60% 3 FA concentrate in triacylglycerol form, produced according to good manufacturing practices. DHA and EPA comprise appr. 67% of the fatty acid content. Four milligrams of vitamin E (tocopherol) was added to each EPAX 1050TG and placebo capsule. A total of 174 patients concluded the OmegAD study. Plasma fatty acid profi les and cognition and behavioral data have been published ( 7,26,27 ). Based on pretrial power calculation concerning cytokine profi les with a statistical signifi cance level of P < 0.05 and 80% power, a minimum of 20 patients was required to detect a difference of 30% between the 3 FAs and placebo groups through use of cytokine assays.
Blood samples for preparation of PBMCs or plasma for the present study were obtained from 23 patients before and after 6 months of treatment (2 of the 25 patients did not complete the OmegAD trial). Samples from 2 patients had to be excluded because of technical laboratory failure. Thus, 9 (57-82 y; median 75 y; 3 women)

Correlation analyses
When relating values for the LPS-induced release of PGF 2 ␣ to plasma concentrations of DHA and EPA, we found that changes in DHA and EPA for all 21 subjects correlated signifi cantly to changes in PGF 2 ␣ release ( r = Ϫ 0.6271; P = 0.003 for DHA, and r = Ϫ 0.6662; P = 0.001 for EPA) ( Fig. 2 ). Thus, the more DHA or EPA increased, the lower was the PGF 2 ␣ release.

DISCUSSION
This study has shown that oral supplementation with a DHA-enriched 3 marine FAs preparation reduced the release of the stable arachidonate metabolite PGF 2 ␣ from ex vivo LPS-stimulated PBMC. This reduction correlated signifi cantly with the rise of plasma EPA and DHA, as well as to reductions of IL-6 and IL-1 ␤ , simultaneously released from these same LPS-stimulated PBMC. These results point to interactions between the eikosanoid, FA, and cytokine systems and may be part of the anti-infl ammatory reactions associated with 3 FA treatment.
In the presently employed assay for PGF 2 ␣ , the manufacturer (Invitrogen, Inc.) declares that the cross-reactivity with the 3 based PGF 3 ␣ is 21%. Thus, we cannot determine how much of the assayed PGF 2 ␣ actually originated from AA or from EPA (hence being PGF 3 ␣ ). Nonetheless, even if one assumes that the proportion of PGF 3 ␣ increased after 6 months of 3 supplementation, the total outcome of all isoforms of PGF was a reduction. It is also reasonable to assume that the biological activity of the PGF mixture might be lower than if all PGF originated from AA. The same reasoning is valid for PGE 2 and PGE 3 . plasma values for DHA as well as for EPA were signifi cantly higher at 6 months compared with pretrial values ( Table 1 ). The placebo group displayed no signifi cant changes of DHA or EPA in plasma compared with pretrial values ( Table 1 ). The rise of DHA levels was larger than that of EPA in the 3 FA group (+3.7 percentage units and +2.7 percentage units, respectively), suggesting that some conversion of DHA to EPA had taken place as discussed in ( 28 ).

PGF 2 ␣ synthesis in cell supernatants
Quiescent PBMCs released only minute amounts of PGF 2 ␣ . At baseline, the two groups did not differ significantly as to PGF 2 ␣ release from stimulated PBMCs induced by LPS or PHA.
LPS . LPS conferred a 100-fold rise of the PGF 2 ␣ concentration in supernatants. At 6 months of treatment, mean values for PGF 2 ␣ release from LPS-stimulated PBMCs from the 9 AD patients given the 3 FAs preparation were signifi cantly lower than baseline ( P = 0.0076) ( Fig. 1 ). In contrast, mean values for PGF 2 ␣ releases for the placebo-treated AD patients were not signifi cantly lower at 6 months compared with pretrial values ( P > 0.05) ( Fig. 1 ). The reduction of PGF 2 ␣ in the 3 FA group for LPS values between baseline and 6 months was trendwise signifi cant for the difference from the corresponding values for the placebo group ( P = 0.06).
PHA . PHA conferred a 30-fold rise of the PGF 2 ␣ concentration in supernatants. PGF 2 ␣ release from PBMCs stimulated with 10 µg PHA /ml from 3 FA treated patients (n = 9) or the placebo oil (n = 10, due to insuffi cient amounts of donor cells from 2 patients) was not changed after 6 months of treatment ( Table 1 ).

TNF-␣ , IL-1 ␤ , IL-6, and G-CSF release
As described previously ( 28 ), supplementation with 3 FAs was associated with signifi cant reductions of the release of IL-1 ␤ , IL-6, and G-CSF after 6 months compared with pretrial values when PBMCs were stimulated with 10 ng LPS/ml. The placebo group displayed no or minor changes ( Table 1 ). In contrast, the TNF-␣ release did not change in any of the treatment groups ( 28 ). th percentile values are given in parentheses. P values are given for changes between baseline and 6-mo values. The decrease in IL-6 was signifi cantly larger after 6 mo in the 3 FA group than in the placebo group ( P = 0.039). DHA, dokosahexaenoic acid; EPA, eikosapentaenoic acid; G-CSF, granulocyte colony-stimulating factor; IL, interleukin; LPS, lipopolysaccharide; NS, not signifi cant; PGF 2 ␣ , prostaglandin F 2 ␣ ; PHA, phytohemagglutinin; TNF, tumor necrosis factor. a phenomenon involved in neutrophil recruitment and migration ( 24 ). As well, PGE 2 and PGF 2 ␣ may promote vascular infl ammation ( 16,25 ). Consequently, one might ask about the mechanisms for the simultaneous reductions of PGF 2 ␣ and several cytokines, as well as the statistically signifi cant correlation between changes in PGF 2 ␣ , IL-6, IL-1 observed in our study. Do all reactions depend on the effects of 3 FA on common mechanisms for generation of these molecules? Or do reduced release of prostaglandins directly infl uence generation and release of IL-6, IL-1, G-CSF (but not TNF)? Our study and current literature cannot resolve these issues but might serve as a starting point for research on proand anti-infl ammatory effects of prostaglandins. Moreover, whether the effect of 3 FA is on CD14, TLR4 or other parts of the receptor system for LPS or downstream remains to be settled ( 40 ).
The PGF 2 ␣ results obtained here after stimulation with the alternate stimulus PHA strongly suggest that the structures and cells targeted by PHA (lectins on most T-lymphocytes) were not affected in this capacity by the increase of 3 FAs. This is in accordance with studies by Wasserman et al. showing that PGF 2 ␣ had no effect on PHA-mediated lymphocyte proliferation ( 18 ). In contrast, T-lymphocytes might be infl uenced by 3 FAs, as shown by Trebble et al. ( 41 ) when using concanavalin A, a different T-cell mitogen, as the stimulus and cell proliferation as the read-out system. These considerations point to a rather specifi c effect of 3 FAs on certain but not all signaling pathways. However, we have no data here to further dissect the various signaling pathways for LPS and PHA.
Our PGF 2 ␣ data compare well with previous results for PGE 2 generation from LPS-stimulated PBMC obtained from healthy subjects after oral supplementation with 3 FAs. Thus, two studies using varying doses of EPArich fi sh oil preparations reported decreases of PGE 2 in a dose-dependent way ( 41,42 ). Trebble et al. ( 41 ) also found a negative correlation between generated PGE 2 and plasma EPA but did not report if there also was a relation to DHA. Rees et al. ( 42 ) observed no changes of released TNF-␣ , IL-1 ␤ , or IL-6, which differs from our results. As discussed previously ( 28 ), effects of EPA and DHA differ in a number of respects, such as binding to PPAR ␥ or to the RX receptor, and for membrane fl uidity ( 9 ). Our PGF 2 ␣ data also compare well with previous in vitro results for PGE 2 and PGF 2 ␣ generation from various cells after 3 FA treatment ( 43 ).
The relation of prostaglandins to the AD brain pathology has attracted attention over the years. There are recent data suggesting that PGE 2 stimulates production of amyloid-␤ peptides (which can result in the AD typical plaques) ( 44 ) and that these peptides can further stimulate PGE 2 generation ( 45 ). Hence, reducing PGE 2 (and thus PGF 2 ␣ ) might be of signifi cance for progression of AD. However, therapy with blockers of cyclooxygenase activity (impairing generation of prostaglandins) have not been successful in reducing cognitive decline in AD. PGF 2 ␣ is considered to be a major and stable metabolite of prostaglandin E 2 (PGE 2 ) ( 12,13 ). The generation of PGE and PGF proceeds from AA and the common precursor prostaglandin H. By means of an enzyme, 9KPGR, PGE can be converted to PGF ( 12 ). Various factors can alter the generation of either eikosanoid, such as mutations in generating enzymes, stimuli, and environment (30)(31)(32)(33). Thus, interactions of EPA, DHA, LPS, and PHA might be of signifi cance for generation of one or the other eikosanoid. Moreover, PGE and PGF can interact on the receptor level ( 34 ). Therefore, we decided to measure a common end product, PGF, which would refl ect turnover of more than one single eikosanoid.
Given the effects of PGE and PGF to infl uence cytokine production ( 20-22, 24, 25 ), one might hypothesize that PGE 2 (as well as PGE 3 ), generated and released rapidly in and from the LPS-stimulated PBMC, might have interacted with the relatively slower signaling systems for generation of the aforementioned cytokines. In our study, the release of PGF 2 ␣ into the culture medium corresponds to a concentration of approx. 1.5 nM (being similar to previous data on PGE 2 by Harizi et al. ( 35 ). The question then arises if this concentration is of biological signifi cance. Previous studies have shown that much higher concentrations of PGE 2 are needed for blockade of, for example, PHA-mediated release of proinfl ammatory cytokines and lymphocyte proliferation ( 18,36 ). Although previous studies usually have focused on inhibitory effects of prostaglandins on generation of cytokines such as TNF-␣ and IL-1 ␤ , recent data have emphasized that PGE 2 might also enhance immune and infl ammatory reactions ( 37, 38 , Vedin unpublished data). Thus, PGE 2 induced pro-stimulatory molecules of the TNF/TNF receptor superfamily ( 39 ) enhanced hematopoietic stem cell homing, survival, and proliferation ( 23 ) and increased IL-23-induced IL-17 production, In conclusion, this study shows that release from PBMC of PGF 2 ␣ , generated directly from AA or via PGE 2 , decreased in the DHA-enriched 3 FA supplemented group compared with the placebo group in a stimulus-specifi c way. This novel fi nding agrees with and adds to previous data on effects of EPA supplementation, suggesting that EPA and DHA effects are rather similar in this particular respect, although differences are noted for other effector variables. In this context, it may be speculated that DHA (and EPA) gives rise to anti-infl ammatory and neuroprotective lipid mediators, which appears to be part of the resolution phase of infl ammation ( 33,46 ).
The authors thank Bengt Vessby and Siv Tengblad for fatty acid analyses; A-C Tysén-Bäckström and Andreas Svensson for patient management; and Lilian Walther Jallow, for support with the Bioplex 100 Systemreader. The authors thank Jesper Haeggström, Karolinska Institutet, for valuable review of prostaglandin metabolism.