Docosahexaenoic Acid–Derived Fatty Acid Esters of Hydroxy Fatty Acids (FAHFAs) With Anti-inflammatory Properties

White adipose tissue (WAT) is an extremely plastic organ with important roles in energy balance, whole-body glucose homeostasis, and the immune system (1–4). The systemic effects of WAT largely reflect its role in the control of blood lipid levels, as well as in the secretion of numerous bioactive peptides (e.g., adiponectin, leptin) from WAT cells (5,6). Adipocytes also release lipid-based mediators such as branched fatty acid esters of hydroxy fatty acids (FAHFAs) (7) and palmitoleate (8), which could improve local and whole-body glucose metabolism (9,10). Moreover, FAHFA administration also stimulated glucagon-like peptide 1 and insulin secretion, and reduced obesity-associated WAT inflammation in mice through cell surface G-protein–coupled receptor 120–dependent signaling (7). Both FAHFAs and palmitoleate are produced in adipocytes via the fatty acid (FA) synthesis pathway (de novo lipogenesis). The activity of this pathway is decreased in response to obesity-associated hyperinsulinemia and WAT inflammation (8,11,12), resulting in reduced levels of the beneficial lipid mediators. Although the biosynthetic enzymes of FAHFAs are unknown, two FAHFA-specific hydrolases, AIG1 and ADTRP, were recently identified (13).

These newly described FAHFAs are members of a lipid class called estolides (intermolecular esters of hydroxy FAs) serving mainly as biodegradable lubricants (14). Recently, the discovery of FA estolides in humans and mice (7) and triacylglycerol estolides in the brushtail possum (15) brought these lipids to mammalian physiology. The FAHFA nomenclature introduced by Yore et al. (7) combines abbreviations of esterified FAs and hydroxy FAs (e.g., the combination of esterified palmitic acid and hydroxy stearic acid [HSA] was abbreviated as PAHSA). Although a recently published in silico library of all potential FAHFAs uses a nomenclature based on chemical structure (16), for practical reasons, here we use the shorter abbreviations (e.g., PA for palmitic acid, LA for linoleic acid, and DHA for docosahexaenoic acid, with the “H” prefix for hydroxy FAs [HFAs] and the position of branching).

The obesity-associated WAT inflammation imposes adverse local as well as whole-body metabolic effects (1,2,11,17–19). This inflammatory response is accompanied by macrophage repolarization to the proinflammatory (classically activated) M1 state, which negatively affects WAT functions (20) and could be counteracted using both dietary and pharmacological interventions (21,22). Regarding the dietary influence, namely n-3 polyunsaturated FAs (PUFAs) of marine origin, which are considered to be healthy dietary constituents in individuals with diabetes (23) (see discussion), play an important role in the resolution of WAT inflammation (24–28) and probably potentiate beneficial functions of immune cells in WAT like efferocytosis or autophagy (17).

n-3 PUFAs exhibit their anti-inflammatory effects through G-protein–coupled receptor 120, as well as through several other signaling pathways, resulting in improved insulin sensitivity in obese mice (29–31). In addition, DHA-derived lipid mediators such as resolvin D1 have been reported to decrease WAT inflammation, shifting macrophage polarization toward the M2 form and improving insulin sensitivity in obese mice (24–27). A wide range of lipid mediators, including resolvins D1 and D2, protectin D1, lipoxin A4, 17-hydroxydocosahexaenoic acid (HDHA), 18-HEPE, and 14-HDHA were identified in human subcutaneous WAT, whereas the levels of protectin D1 and 17-HDHA decreased in the subcutaneous WAT of patients with peripheral vascular disease (32). Protectin DX, also known to be produced in WAT, alleviated insulin resistance in db/db mice, but did not resolve WAT inflammation (33).

Given the beneficial effects of n-3 PUFAs on WAT inflammatory status, we hypothesized that novel FAHFA structures derived from n-3 PUFAs, with possible anti-inflammatory properties, could be found. To test this hypothesis, we performed lipidomic analysis using human and murine serum and WAT samples collected from subjects supplemented or not supplemented with n-3 PUFAs.

All chemicals were purchased from Sigma-Aldrich (Prague, Czech Republic), unless otherwise stated. FAHFA standards (5-, 9-, 12-, 13-PAHSA, respectively, and 5-PAHSA-²H₃₁ and 9-PAHSA-¹³C₄) were purchased from Cayman Pharma (Neratovice, Czech Republic).

Serum samples were acquired within the framework of a clinical trial focused on the combined effects of the antidiabetic drug pioglitazone and n-3 PUFAs (34). Briefly, overweight/obese patients 40–70 years of age, in whom type 2 diabetes had been diagnosed and who had already been treated with metformin, were given either 5 g/day corn oil (Placebo) or 5 g/day eicosapentaenoic acid (EPA) plus DHA concentrate (EPAX 1050TG, EPAX AS, containing about 15% EPA, 40% DHA, wt/wt [i.e., ∼2.8 g of EPA plus DHA]) for 24 weeks. The serum samples and biopsy samples of abdominal subcutaneous WAT collected during the final visit after an overnight fast were stored at −80°C until liquid chromatography (LC)–tandem mass spectrometry (MS/MS) analysis.

Male mice (C57BL/6J; The Jackson Laboratory, ME) were maintained in a controlled environment (22°C, 12-h light/dark cycle, light from 6.00 a.m.) and were fed a corn oil–based high-fat (HF) diet (lipid content 35%, wt/wt) or an HF diet with EPA plus DHA concentrate (HFF) (EPAX 1050TG) for 8 weeks, as previously described (21). Epididymal WAT, subcutaneous WAT, liver and interscapular brown adipose tissue, and serum samples were collected after an overnight fast, and were stored in liquid nitrogen. All animal experiments were approved by the Animal Care and Use Committee of the Institute of Physiology of the Czech Academy of Sciences (Approval Number 172/2009) and followed the guidelines.

Murine adipocyte cell line 3T3-L1 and human multipotent adipose-derived stem (hMADS) cells (7) were grown according to standard protocols. Differentiated adipocytes were incubated with 100 μmol/L LA and 100 μmol/L DHA complexed to BSA 3:1 for 24 h and extracted for FAHFA analysis. Adipocytes and stromal-vascular cells (SVCs) were prepared as before (26). RAW 264.7 cells and murine bone marrow–derived macrophages (BMDMs) were grown and stimulated with lipopolysaccharide (LPS) as before (35,36). Human peripheral blood mononuclear cells (PBMCs) were isolated from buffy coats and treated with lectin, as described previously (37).

FAHFA extraction was performed based on the published method (7). Murine tissue (∼300 mg) or cells were homogenized using a MM400 bead mill (Retsch GmbH, Haan, Germany) chilled to −20°C in a mixture of citric acid buffer and methanol, and further extracted with dichloromethane (1:1:2 final ratio). Internal standards of 5-PAHSA-²H₃₁ and 9-PAHSA-¹³C₄ were added to the homogenate (100 pg/sample). Serum samples were extracted according to the same protocol, apart from the homogenization. The organic phase was collected, dried in a Speed-vac (Savant SPD121P; ThermoFisher Scientific), resuspended in dichloromethane, and applied on Strata SI-1 Silica SPE columns (55 µm, 70 Å; Sigma-Aldrich). FAHFAs were eluted from the SPE columns with ethylacetate, concentrated in the Speed-vac, resuspended in methanol, and immediately measured using LC-MS as follows.

Chromatographic separation was performed in an ultra-performance LC UltiMate 3000 Rapid Separation LC System (Thermo Scientific) equipped with a Kinetex C18 1.7 µm 2.1 × 150 mm column (Phenomenex). The flow rate was 200 µL/min at 50°C. The gradient program used to separate FAHFA was as follows: solvent A (70% water, 30% acetonitrile, 0.01% acetic acid, pH 4), solvent B (50% acetonitrile, 50% isopropanol), for 1 min (100% solvent A), 5 min (20% solvent A), 18 min (10% solvent A), 20 min (100% solvent A), and 25 min (100% solvent A), while a linear gradient was maintained between the steps. Isocratic elution (20% solvent A, 80% solvent B, for 60 min) was used for structural studies. Ultra-performance LC was coupled to a QTRAP 5500/SelexION, a hybrid, triple-quadrupole, linear ion trap mass spectrometer equipped with an ion mobility cell (Sciex). FAHFAs were detected in negative electrospray ionization mode with the following parameters: declustering potential, −130; collision energy, −35; collision cell exit potential, −15; curtain gas, 25; collision gas, high; ionspray voltage, −4500; temperature, 400; ionsource gas 1, 40; and ionsource gas 2, 50. For ion mobility experiments, differential mobility spectrometry (DMS) settings were as follows: DMS temperature, high; DMS modifier, isopropanol/high; separation voltage, 3,800; DMS offset, −3; DMS resolution enhancement, low; and compensation voltage, −5.0 for 13-PAHSA and −1.2 for 12-PAHSA. Multiple reaction monitoring (MRM) mode with one quantifier and two qualifier transitions per FAHFA was used for quantitation (7) (Table 1). Quantifier ion MRM was used as a survey scan for information-dependent acquisition in the linear ion trap for enhanced-resolution MS/MS and 2nd generation (MS/MS/MS) product ion spectra (scan rate 1,000 Da/s, scan mode profile, step size 0.05 Da, linear ion trap fill time 200 ms). Pure standards of PAHSAs were used for quantitation. For DHAHLA and DHAHDHA compounds, the 5-PAHSA calibration curve was used as a surrogate.

Organic synthesis of 13-DHAHLA was performed according to Steglich esterification from DHA and 13-HODE (38,39). Details are provided in the Supplementary Data.

The anti-inflammatory properties of DHAHLA were assessed according to published methods (7,36,37,40). Briefly, murine BMDMs or RAW 264.7 macrophages were incubated in the presence of LPS (100 ng/mL; Escherichia coli 0111:B4; Sigma-Aldrich) or 13-DHAHLA (10 μmol/L) alone or in combination with 9-PAHSA (10 μmol/L) or 13-DHAHLA (10 μmol/L) for 18 h. Control cells were incubated with the vehicle alone. Murine IL-6 ELISA (Cayman Chemicals) and quantitative PCR (21) were used to measure the markers of macrophage activation; details are provided in the Supplementary Data. BMDMs were incubated in the presence of LPS alone or in combination with DHA (10 μmol/L), interferon-γ (IFN-γ) (50 ng/mL) or 13-DHAHLA (10 μmol/L) for 18 h. LC-MS/MS metabolipidomics (26,36,37,41) was used to measure macrophage metabolic activation and the levels of lipid mediators. PBMCs were pretreated with 1 µmol/L 13-DHAHLA for 30 min and stimulated with lectin (phytohemagglutitin 10 µg/mL) for 24 h, and the levels of tryptophan and kynurenine were measured in the media. RAW 264.7 macrophages were stimulated with LPS (10 ng/mL) and incubated in the presence of various 13-DHAHLA concentrations for 18 h to explore the dose-dependent inhibition of macrophage activation. The phagocytosis of fluorescein-labeled zymosan (Life Technologies) by BMDMs was measured using a Victor X4 plate reader (PerkinElmer) (40).

Statistical analysis was performed with SigmaStat and P < 0.05 was considered significant.

In view of the beneficial effects of PAHSAs (7), we developed a targeted lipidomic methodology using LC coupled to hybrid tandem mass linear ion trap spectrometry to identify and quantify FAHFAs in human and murine samples. We took advantage of the ability of MS to switch from sensitive triple-quadrupole scan modes to highly sensitive full-scan ion trap mode within one analysis to obtain both quantitative and qualitative (structural) information. This approach enabled us to precisely quantify FAHFA levels using MRM and to identify the branching position on the backbone HFA using MS/MS/MS. For instance, 9-PAHSA can be ionized in negative mode to [M-H]⁻ ion 537.488 mass-to-charge ratio (m/z). Fragmentation by collision-induced dissociation results in the following three daughter ions: 299.3, 281.2, and 255.2 m/z, identified as fragments of hydroxystearic, octadecenoic, and palmitic acid, respectively (Fig. 1A and B) (7). Further fragmentation of ion 299.3 m/z (hydroxystearic acid) in the linear ion trap gave rise to the ions 155.144 and 127.113 m/z, which are specific to the position of the hydroxyl group on the hydroxystearic acid backbone (Fig. 1A), thus enabling us to identify the branching carbon.

The quantities and structures of FAHFA isomers were analyzed in human serum samples, and with the PAHSA isomers, we were able to identify nine positional isomers (13-, 12-, 11-, 10-, 9-, 8-, 7-, 6-, and 5-PAHSA) (Fig. 1C) according to their MS/MS/MS spectra (Fig. 1D). Given the practical impossibility of separating the positional isomers of 12- and 13-PAHSA in human samples (7) (Supplementary Fig. 1), the 12- and 13-PAHSA levels are reported together.

First, we investigated the possible effect of n-3 PUFA supplementation on the levels of the known FAHFAs (see Introduction) in humans with diabetes, and therefore serum samples of metformin-treated patients supplemented with either corn oil capsules (Placebo) or n-3 PUFA capsules (n-3 PUFA) for 24 weeks (34) were analyzed. Serum levels of 5-, 9-, and 12/13-PAHSA isomers were not altered by n-3 PUFA supplementation (Fig. 2A). This observation was also supported by animal experiments on dietary obese mice that were fed an n-3 PUFA diet for 8 weeks (21), where no differences in serum PAHSA levels were detected (Fig. 2B).

Using the same MS approach as in the identification of PAHSA isomers, any combination of FAs and the branching position theoretically could be detected. Therefore, we focused on alternative combinations of FAs besides PAHSAs, and were able to identify novel members of the FAHFA family derived from LA and DHA, specifically DHAHLA, LAHDHA, and DHAHDHA in human serum (Supplementary Fig. 2, structures).

With DHAHLA, DHA esterified to a hydroxy LA, the following two positional isomers of the hydroxy FA backbone were detected: 9- and 13-HLA (also known as HODE); and therefore 9- and 13-DHAHLA. This is in agreement with the high concentrations of 9(S)- and 13(S)-HODE, enzymatic products of 15-lipoxygenase in the organism. As shown in the 13-DHAHLA fragmentation scheme (Fig. 3A), the ion 605.457 m/z gave rise to the fragment 295.228 m/z (13-HODE), which was further fragmented to characteristic ions 179.144 and 195.139 m/z (for details, see Supplementary Fig. 3).

Chromatographic separation of the murine serum sample revealed additional complexity of DHAHLA isomers when four separated DHAHLA peaks were detected (Fig. 3C). Structural characterization in the linear ion trap revealed that two major peaks were 13-DHAHLA and two minor peaks were 9-DHAHLA cis-trans isomers of double bonds in HLA acyl chains. Identity of the backbone fragmentation was confirmed using synthetic standards for 9(S)-HODE and 13(S)-HODE (Fig. 3C and Supplementary Fig. 2). Of note, only two peaks corresponding to physiologically relevant (9Z,11E,13S)-13-hydroxy-9,11-octadecadienoic acid- and (9S,10E,12Z)-9-hydroxy-10,12-octadecadienoic acid-derivatives, 9(S)-HODE and 13(S)-HODE, respectively, were observed in human WAT and serum samples, as well as in cultured cells (hMADS cells; data not shown); therefore, only these were considered for further analyses.

Similar to PAHSA tissue distribution (7), 13-DHAHLA was detected in murine adipose tissue depots and was upregulated after n-3 PUFA supplementation (Fig. 4A). Levels of FAHFA-specific hydrolases (13), Aig1 and Adtrp, were downregulated in murine adipose tissue and upregulated in the liver after n-3 PUFA supplementation. Interestingly, although Adtrp was exclusively associated with adipocytes, Aig1 was present also in SVC (Supplementary Fig. 4). Very low levels of 13-DHAHLA were detected in human subcutaneous fat biopsy samples after n-3 PUFA supplementation (0.38 ± 0.06 pmol/g), but no DHAHLA was detected in placebo-treated patients. Cultured 3T3-L1 and hMADS adipocytes, when supplemented with DHA and LA, were able to synthesize DHAHLA isomers (Fig. 4B). Although no DHAHLA was detected in macrophages treated similarly, macrophages were able to synthesize DHAHLA, when exposed to high concentrations of DHA and LA (Supplementary Fig. 5).

Although several positional isomers of HDHA exist (42), only one peak for DHAHDHA was detected in human serum. MS/MS/MS analysis revealed 14-HDHA–specific fragments (205.2 and 161.2 m/z) (39,42) in the corresponding peak. Importantly, oxidized phospholipids containing predominantly the 14-HDHA positional isomer were identified in human platelets (39). Besides the 14-HDHA backbone, smaller peaks of DHAHDHA were detected in murine samples. Because of very low intensities, we were unable to identify their branching positions, and thus we cannot rule out that other isomers were also present in murine samples. Overall, DHAHLA and DHAHDHA levels were elevated in the experimental groups supplemented with n-3 PUFAs (Fig. 5A and B), and a positive correlation between the omega-3 index in serum phospholipids (34) and their levels of 13-DHAHLA was found (Fig. 5C).

Isomers of PAHSA were shown to be able to decrease macrophage activation and the release of proinflammatory cytokines after LPS stimulation (7). Also, oxidized derivatives of DHA, especially resolvins, protectins, and maresins (31,43), possess potent anti-inflammatory and proresolving activities. Therefore, RAW macrophages were stimulated with LPS, and the effects of 9-PAHSA (7) and 13-DHAHLA on macrophage activation were analyzed. Both PAHSA and DHAHLA prevented the increase in proinflammatory interleukin (IL)-6 concentrations in media and also decreased mRNA levels of IL-6, tumor necrosis factor-α (TNF-α), IL-1β, and Ptgs2 in cells (Fig. 6A and Supplementary Fig. 6). Next, the model of mitogen-stimulated human PBMCs was used to test the immunomodulatory potential of 13-DHAHLA. The activation of indoleamine 2,3-dioxygenase (IDO), the rate-limiting enzyme of the essential amino acid tryptophan catabolism toward kynurenine, causes tryptophan depletion, reduced the growth of microbes (37), and serves as an immune checkpoint (44). Freshly isolated PBMCs were preincubated with 1 μmol/L 13-DHAHLA for 30 min and then exposed to lectin phytohemagglutinin to activate T-helper type 1 lymphocytes and also the IDO pathway in macrophages (37). Activation of the IDO enzyme was indicated by a strong increase in the kynurenine/tryptophan ratio, which was partially prevented by 13-DHAHLA preincubation (Fig. 6B).

FAHFA might also serve as a pool of FA, and the anti-inflammatory properties of 13-DHAHLA could be partially mediated by DHA released via DHAHLA hydrolysis. To test this hypothesis, murine BMDMs were stimulated with LPS, and the effects of 10 μmol/L 13-DHAHLA and 10 μmol/L DHA on macrophage activation were analyzed using metabolipidomics (26,36). The ratio of the intracellular concentrations of (citrulline plus ornithine)/(arginine plus aspartate), which summarizes the intermediates of the related metabolic pathways of nitric oxide and reactive oxygen species production, was the most sensitive early marker of macrophage activation (36), and the proinflammatory effect of LPS was significantly reduced by both 13-DHAHLA and DHA (Fig. 6C). However, a detailed analysis of proresolving lipid mediators revealed that the macrophages converted DHA into 17-HDHA, and further to protectin D1 or resolvin D1, with anti-inflammatory and proresolving properties; whereas, with 13-DHAHLA, only a small amount of DHA was converted to 17-HDHA and protectin D1, and no resolvin D1 was detected (Fig. 6D). To further explore the ability of 13-DHAHLA to reduce macrophage activation by LPS, a dose-dependent experiment was performed. The lowest effective concentration of 13-DHAHLA was 10 nmol/L (Fig. 6E). Because the resolution of inflammation also involves phagocytosis, 13-DHAHLA was tested for its ability to enhance phagocytic activity toward zymosan particles. BMDMs pretreated with 13-DHAHLA increased the phagocytosis of fluorescein-labeled zymosan in a dose-dependent manner with a peak value at 100 nmol/L (Fig. 6F). Finally, an effect of 13-DHAHLA on the stimulation by proinflammatory cytokine IFN-γ (50 ng/mL) was tested. 13-DHAHLA prevented macrophage activation measured as levels of citrulline, the byproduct of nitric oxide synthase (Fig. 6G). These results document that 13-DHAHLA itself exerts anti-inflammatory and proresolving properties.

We identified novel members of FAHFA lipid class derived from DHA and LA (i.e., DHAHLA and DHAHDHA) with anti-inflammatory properties in the serum and WAT of both mice and patients with diabetes supplemented with n-3 PUFAs. In addition to DHAHLA and DHAHDHA isomers, which were elevated after the supplementation, other members of the FAHFA family not described before (7,16) were also found using our novel LC-MS/MS/MS screening for all theoretical FA combinations (e.g., LAHDHA, DHAHSA, OAHDHA). However, these lipids were not altered by n-3 PUFAs supplementation in humans, and therefore we did not explore them in detail.

The levels of DHAHLA and DHAHDHA were comparable to the concentrations of DHA-derived docosanoids (protectins, resolvins) in human serum. Hydroperoxy/hydroxy DHA serves as an intermediate in enzymatic reactions, which produce dihydroxylated and trihydroxylated docosanoids (43). FAHFAs are also produced by enzymatic reaction (7), probably using monohydroxylated DHA or HLA as a substrate, and therefore the concentrations of DHAHLA and DHAHDHA were within the expected range. The levels of 13-DHAHLA also correlated with the omega-3 index in the serum of patients. Compared with 5-PAHSA, 13-DHAHLA concentrations were ∼100-fold lower in serum. Similar to PAHSA, the novel lipids were also synthesized in WAT and adipocytes, expanding the family of adipose-derived lipid mediators. In addition to 13- and 9-DHAHLA derived from 13(S)- and 9(S)-HODE, two related compounds were identified as DHAHLA cis/trans-isomers where the backbone was conjugated LA (e.g., rumenic acid). Because these compounds were detected only in mice fed a custom-made diet, and were found in neither human samples nor in cell cultures, casein in the diet was most likely the source of conjugated LA (21). Other derivatives of HDHA were detected, but only 14-DHAHDHA was upregulated in humans in response to n-3 PUFA supplementation. The exceptional role of DHA hydroxylated at position 14 was also observed in oxidized phospholipids (39). Due to the demanding organic synthesis of the 14-HDHA backbone, HDHA derivatives were not explored further.

The chemical stability of DHAHLA is similar to that of resolvins and protectins (45). The pure compound had to be stored in amber glass vials under an argon atmosphere at −80°C. Although the ester bond between DHA and HLA can be hydrolyzed, yielding 13-HODE and DHA, we did not observe any extensive DHAHLA decomposition during short-term cell preincubations. However, sample acquisition and storage were critical for the successful analysis of DHAHLA. Murine samples had to be processed immediately or stored in liquid nitrogen. The storage of human serum or WAT samples at −80°C for periods longer than ∼6 months resulted in FAHFA degradation. Therefore, low levels of 13-DHAHLA were detected in WAT from patients supplemented with n-3 PUFAs, whereas no DHAHLA was detected in the placebo group, as these samples were collected 1–3 years before the analysis (data not shown).

Reflecting the anti-inflammatory effects of PAHSAs, the already known members of the FAHFA lipid class, on adipose tissue macrophages from obese mice (7) and the beneficial effects of DHA and its metabolites on adipose tissue inflammation (24–26), we hypothesized that DHAHLA could also have immunomodulatory properties. First, we tested 13-DHAHLA for the ability to alter macrophage activation triggered by LPS. Our results demonstrated that the activation of macrophages by LPS and IFN-γ, measured as a production of IL-6, expression of IL-6, TNF-α and IL-1β, and as an alteration in metabolic pathways (metabolomic marker) (36) was alleviated by both 9-PAHSA and 13-DHAHLA. When compared with stimulation with pure DHA, 13-DHAHLA exerted anti-inflammatory properties, whereas only a limited production of proresolving mediators derived from DHA was detected, probably because of nonspecific hydrolysis in the media. This suggests that 13-DHAHLA does not serve as a temporary storage location of DHA for the production of docosanoids. Also, 13-DHAHLA was able to partially prevent IDO activation in PBMCs stimulated with lectin, to limit immunosuppression caused by tryptophan depletion (37,44), and stimulated the clearance of zymosan particles by BMDMs in a higher nanomolar range. Our results document that 13-DHAHLA has the ability to affect immune cells, alleviate macrophage activation, and stimulate the proresolving processes. It is conceivable that DHAHLA may contribute to the anti-inflammatory and proresolving effects attributed to the DHA in human and murine WAT (lower density of crown-like structures, lower local concentration of proinflammatory cytokines), as was already documented for the other DHA metabolites, namely, docosanoids and related endocannabinoids (24–30,32). The expression of FAHFA hydrolases in WAT, liver, isolated adipocytes, and SVCs revealed that FAHFA degradation is probably regulated on both local and systemic levels.

Importantly, our results also documented that serum PAHSA levels, either in mice or in humans, were not affected by n-3 PUFA supplementation, and, therefore, that PAHSA was not involved in the anti-inflammatory effects of DHA.

The detection of DHAHLA and DHAHDHA in both mice and humans, and the emerging concept that the immune and metabolic systems are interconnected (17,26), strongly suggest the involvement of the novel lipids in the broad beneficial effects of n-3 PUFAs on health. Thus, in humans, in addition to their anti-inflammatory effects in WAT (see above), n-3 PUFAs attenuate systemic inflammatory processes (46), help to prevent cardiovascular disease (47), ameliorate nonalcoholic fatty liver disease (48), lower hypertriacylglycerolemia (49), and increase circulating adiponectin levels (50). Although n-3 PUFAs could not reverse type 2 diabetes and their impact on insulin sensitivity is controversial (23,28,34), n-3 PUFAs exert anti-inflammatory effects in the WAT of human subjects with insulin resistance (28). As demonstrated by our clinical trial (34), which also served as a source of the samples for this study, n-3 PUFAs could improve postprandial lipid metabolism in overweight/obese patients with type 2 diabetes, even in the face of a combined pharmacotherapy.

The demanding organic synthesis of pure compounds in higher amounts will be needed to further explore the proresolving properties of DHAHLA and related molecules in vivo, their ability to interact with G-protein–coupled receptors, and the potential involvement of these novel lipids in various biological effects of DHA. Future studies focused on these metabolites of DHA may provide important insights into the anti-inflammatory properties of n-3 PUFAs and provide new tools for the treatment of diseases linked with inflammation.

Acknowledgments. The authors thank Professor Charles N. Serhan (Harvard Medical School, Boston, MA) for the PD1 standard. hMADS were provided by the laboratory of Christian Dani (University of Nice Sophia Antipolis) and Amri Ez-Zoubir (University of Nice Sophia Antipolis) under the conditions of a material transfer agreement (Centre National de la Recherche Scientifique).

Funding. This work was supported by the Czech Science Foundation (grant 14-3684G) and the Ministry of Education, Youth and Sports of the Czech Republic (grant LH14040).

Duality of Interest. No potential conflicts of interest relevant to this article were reported.

Author Contributions. O.K. conceived of, designed, performed, and interpreted the LC-MS, animal, and cell culture studies; performed the biological studies; and wrote the manuscript. M.B. and M.R. performed the LC-MS and biological studies. B.S. and E.K. designed the organic synthesis of FAHFA standards. M.P. and P.B. performed the synthesis of FAHFA standards. P.J., J.V., and J.K. Jr. collected human samples. T.P. contributed reagents and designed the human study. J.K. contributed reagents, designed the human study, and contributed to the data discussion and the writing of the manuscript. J.K. and O.K. are the guarantors of this work and, as such, had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.