Targeted lipidomics reveals mPGES-1-PGE2 as a therapeutic target for multiple sclerosis

Abstract

The arachidonic acid (AA) cascade produces eicosanoids, such as prostaglandins (PGs), that regulate physiological and pathological functions. Although various nonsteroidal anti-inflammatory drugs have been developed, blocking upstream components (cyclooxygenase-1 and -2) of the AA cascade leads to severe side effects, including gastrointestinal ulcers and cardiovascular events, respectively, due to the complexity of the AA cascade. Here, using an AA cascade-targeted lipidomics approach, we report that microsomal PGE synthase 1 (mPGES-1) plays a key role in experimental autoimmune encephalomyelitis (EAE). Eicosanoids (mainly PGD₂) are produced constitutively in the spinal cord of naive mice. However, in EAE lesions, the PGE₂ pathway is favored and the PGD₂, PGI₂, and 5-lipoxygenase pathways are attenuated. Furthermore, mPGES-1^−/− mice showed less severe symptoms of EAE and lower production of IL-17 and IFN-γ than mPGES-1^+/+ mice. Expression of PGE₂ receptors (EP1, EP2, and EP4) was elevated in EAE lesions and correlated with clinical symptoms. Immunohistochemistry on central nervous systems of EAE mice and multiple sclerosis (MS) patients revealed overt expression of mPGES-1 protein in microglia/macrophages. Thus, the mPGES-1-PGE₂-EPs axis of the AA cascade may exacerbate EAE pathology. Our findings have important implications for the design of therapies for MS.

Multiple sclerosis (MS) is the most prevalent autoimmune disorder of the central nervous system (CNS), with neurological symptoms caused by inflammation and demyelination (1). Studies of experimental autoimmune encephalomyelitis (EAE), an animal model for MS, have shown that autoreactive T cells secreting IL-17 (T_H17 cells) and IFN-γ (T_H1 cells) are involved in EAE/MS pathogenesis (2–4).

In an effort to understand and treat this complex disease, lipids have emerged as one of the targets for developing drugs. Membrane lipids have been identified as autoantigens in MS and EAE pathologies (5). In addition to membrane components, lipids play roles in cell–cell interaction as autacoids. Arachidonic acid (AA) is released from membrane glycerophospholipids by the action of cytosolic phospholipase A₂α (cPLA₂α) (6, 7). Released AA is further converted into prostaglandins (PGs), leukotrienes (LTs), lipoxins, and hydroxy-eicosatetraenoic acids (HETEs), collectively termed eicosanoids, by cyclooxygenases (COXs), lipoxygenases (LOs), and terminal enzymes (8). Involvement of eicosanoids and related lipid mediators has been reported in EAE, collagen-induced arthritis, and other immunological disorders (6, 7, 9). Previously, we and others clearly demonstrated the importance of cPLA₂α in EAE pathology by genetically and pharmacologically ablated mouse studies (10–12). Furthermore, cPLA₂α expression and activities were up-regulated in the spinal cords (SCs) of EAE mice (13). However, it remains totally elusive which eicosanoids downstream of cPLA₂α are involved in the EAE induction and exacerbation. Miyamoto et al. (14) demonstrated that COX-2^−/− mice develop EAE with comparable severity to wild-type control mice, while indomethacin, a nonsteroidal anti-inflammatory drug (NSAID), prevents EAE. However, all mice treated with effective doses of indomethacin died of gastrointestinal bleeding (14). On the other hand, 5-LO^−/− or 12/15-LO^−/− mice developed more severe EAE than control wild-type mice (15), while pharmacological studies showed different results (12). Previous sketchy studies to focus on individual eicosanoids in EAE or MS patients have obvious limitations, as >20 different eicosanoids are produced by the concerted actions of cPLA₂α and downstream enzymes (Fig. S1) (16–18). To overcome such complexity and some controversy, we evaluated the cascade in the SCs of naive and EAE mice by the nonbiased top-down approach (AA cascade-targeted lipidomics and transcriptomics) and confirmed the results from knockout studies.

Results

AA Cascade-Targeted Transcriptomics Imply That AA Cascade Plays Important Roles in EAE Pathology.

C57BL/6 mice were immunized with a myelin oligodendrocyte glycoprotein 35–55 (MOG_35–55) peptide to induce EAE and monitored daily (Fig. 1A). As described in refs. 13 and 19, the course of the disease is divided into induction, acute, and chronic phases in accordance with the clinical symptoms (Fig. 1A). We collected the SCs, spleen, and plasma of naive mice and EAE mice in these three phases.

Fig. 1.

AA cascade-targeted transcriptomics analysis. (A) C57BL/6 female mice were immunized with the MOG_35–55 peptide. Mice were monitored and weighed daily. Data are the mean clinical score and body weight ± SEM of eight animals. (B) Expression of AA cascade related transcripts was estimated using the comparative C_T method with the SCs of naive mice and EAE mice in the induction, acute, and chronic phases (n = 6, 5, 6, and 5 animals, respectively). The relative abundance of mRNA levels in EAE mice compared with naive mice is shown. Data represent means ± SEM. #, P < 0.001, **, P < 0.01, *, P < 0.05 compared with naive mice using the Kruskal-Wallis test with Dunn's post-hoc test.

To determine which enzymes and receptors in the AA cascade are involved in EAE pathology, AA cascade-targeted transcriptomics of the SCs was performed by quantitative RT-PCR (Fig. 1B). Components immediately downstream of cPLA₂α are COX-1/2, 5-LO/FLAP, and 12/15-LO, whose expression levels were highly up-regulated (Fig. 1B). Large part of terminal enzymes and receptors, such as microsomal PGE synthase 1 (mPGES-1), were substantially up-regulated in the SCs of EAE mice (Fig. 1B). However, expression levels of PGI synthase (PGIS) and lipocalin-type PGDS (L-PGDS) were down-regulated in the induction and acute phases of EAE, respectively, and then returned to the basal levels in the chronic phase (Fig. 1B). The PGI₂, PGE₂, and LT receptor (IP, EP1/2/4, BLT1, and CysLT1) gene expression was up-regulated in the acute phase of EAE (Fig. 1B). A correlation between the gene expression and the clinical score was observed for COX-1, H-PGDS, EP1/2/4, BLT1, CysLT1, etc. (Fig. S2). These results imply that AA cascade profoundly affects the pathogenesis of EAE.

AA Cascade-Targeted Lipidomics Reveals That the PGE₂ Pathway Is Favored in EAE Lesions.

The lipidomics approaches disclosed the constitutive production of eicosanoids in the SCs of naive mice and the AA cascade dynamics during EAE (Fig. 2). Approximately 90% of the eicosanoids in the SCs of naive mice were derived from the COX pathway (Fig. S3). The rank order of eicosanoids in the SCs of naive mice was the following: PGD₂ and metabolites > PGE₂ and metabolites >6-keto-PGF_1α > PGF_2α and metabolites > cysteinyl LTs >11-HETE >12-HETE >15-HETE > thromboxane B₂ (TXB₂) > 5-HETE >8-HETE > LTB₄ (Fig. 2A). Cluster analysis demonstrated pathway-dependent fluctuations in the AA cascade, such as COX (clusters I and III) and LO (cluster II), during the course of EAE (Fig. 2B). Eicosanoids belonging to cluster I, such as PGD₂ and 6-keto-PGF_1α, were markedly suppressed in the acute phase and returned to the basal levels in the chronic phase (Fig. 2 A and B). Suppression of these metabolites was probably caused by the fluctuation of L-PGDS and PGIS, respectively (Fig. 1B). 5-LO metabolites (LTB₄, LTC₄, LTD₄, and 5-HETE) belonging to cluster II considerably diminished during the disease course (Fig. 2 A and B), although the up-regulation of 5-LO pathway components were remarkable (Fig. 1B). In the cluster III, the levels of PGE₂ and 13, 14-dihydro-15-keto-PGE₂ (DHK-PGE₂, a major tissue metabolite of PGE₂) were increased, thereby making them the major eicosanoids in acute phase SCs instead of PGD₂ and its metabolites (Fig. 2A). In the chronic phase of EAE, the level of PGE₂ remained higher than in naive mice (Fig. 2A). Among PGESs, only mPGES-1 was up-regulated in the SCs of EAE mice (Fig. 1B), suggesting that this is the key enzyme for the PGE₂ accumulation in EAE lesions. Cluster IV consists of only platelet-activating factor (PAF), which is structurally distinct from the eicosanoids. We previously reported that PAF exacerbates inflammation in the chronic phase of EAE by the enhancement of phagocytosis in microglia/macrophages and subsequent production of TNF-α (19). Furthermore, there was a strong correlation between PAF levels and clinical scores of EAE (13, 19). However, in the current study, no eicosanoid was correlated with the clinical scores of EAE. Although we also measured eicosanoid levels in the spleen and plasma, there were only a few fluctuations in these samples (Fig. S4). These results suggest that the AA cascade was strictly regulated in the SCs of naive mice and this regulation was disrupted by EAE pathology.

Fig. 2.

AA cascade-targeted metabolomics analysis. (A) Eicosanoid levels in the SCs of naive mice and EAE mice in the induction, acute, and chronic phases (n = 8–10 animals) were determined. Data represent means ± SEM. #, P < 0.001, **, P < 0.01, *, P < 0.05 compared with naive mice using the Kruskal-Wallis test with Dunn's post-hoc test. (B) The means of eicosanoid levels were normalized with those of naive mice for each eicosanoid. Then, a cluster analysis was performed by the Ward method. The normalized eicosanoid levels were divided into seven levels according to the indicated color scale. (C) Correlations of COX/5-LO in naive mice (light blue) and EAE mice in the induction (pink), acute (red), and chronic (orange) phases are shown (n = 8–10 animals). Each data point represents the results from a single animal. Data were analyzed statistically by Pearson's correlation.

Next, correlations between the pathways were analyzed to understand the selectivity of the downstream pathways in the AA cascade (Fig. 2C, Fig. S5, and Table S1), because the cluster analysis revealed the pathway-dependent fluctuations. The COX pathway (PGs, TXs, and 11-HETE) correlated with the 5-LO pathway (LTs and 5-HETE) in naive mice (Fig. 2C). Of note, AA metabolism was shifted into the COX pathway, rather than the 5-LO pathway, in the acute phase of EAE (Fig. 2C). Within the COX pathway, the PGE₂ pathway was correlated with both the PGD₂ and PGI₂ pathways in naive mice (Fig. S5 and Table S1). After the induction phase, the PGE₂ pathway was correlated with both the PGD₂ and PGI₂ pathways, while PGE₂ production was facilitated, as shown by the small slope (Fig. S5). Correlations between other pathways (Table S1), such as the PGD₂ and PGI₂ pathways (Fig. S5), were largely conserved throughout the disease course, as indicated by the stable slopes. Taken together, the COX-mPGES-1-PGE₂-EP axis may aggravate EAE pathology.

mPGES-1 Exacerbates EAE Pathology Through T_H1 and T_H17 Cytokine Production.

To further elucidate the roles of PGE₂ pathway in EAE, we particularly focused on mPGES-1, a key enzyme responsible for PGE₂ production in inflammation (20–22). In EAE lesions, mPGES-1 was colocalized with F4/80, a marker for macrophages, but not with CD4 (Fig. 3A). PGE₂ production in SCs of EAE mice was completely blocked by disruption of the mPGES-1 gene (Fig. 3B), suggesting that PGE₂ production in EAE lesions depends on mPGES-1 expressed in macrophages/microglia. The clinical course of EAE in mPGES-1^−/− mice was less severe than in mPGES-1^+/+ mice (Fig. 3C), consistent with the lower cumulative scores in mPGES-1^−/− mice (Table S2). To understand the mechanisms underlying the attenuated symptoms of EAE in mPGES-1^−/− mice, proliferation and cytokine production of T cells in response to the MOG_35–55 peptide were investigated (Fig. 3 D–F). When stimulated with the MOG_35–55 peptide, cells isolated from lymph nodes (LNs) of immunized mPGES-1^+/+ and mPGES-1^−/− mice proliferated similarly (Fig. 3D). However, LN cells from mPGES-1^−/− mice produced significantly lower levels of cytokines (TNF-α, IFN-γ, IL-6, and IL-17) than mPGES-1^+/+ mice (Fig. 3E). Flow cytometry of MOG_35–55-treated CD4⁺ T cells also revealed that IFN-γ and IL-17 production in mPGES-1^−/− mice was reduced (Fig. 3F). Taken together, PGE₂ derived from mPGES-1 appears to support T_H1 and T_H17 cytokine production in EAE lesions.

Fig. 3.

Suppression of EAE pathology and T_H1/T_H17 responses in the absence of mPGES-1. (A) SCs of EAE mice in the acute phase were stained with anti-mPGES-1 (green), F4/80 (red), and CD4 (red) Abs. (Scale bar, 50 μm.) These pictures are representative of two different experiments. (B) PGE₂ levels were measured in the SCs of mPGES-1^{+/+, +/−} and mPGES-1^−/− mice on day 26 (n = 13 and 7 animals, respectively). *, P < 0.05 versus mPGES-1^{+/+, +/−} mice by Mann–Whitney U test. Data represent means ± SEM. (C) mPGES-1^+/+ (filled circles) and mPGES-1^−/− (open circles) mice were immunized with the MOG_35–55 peptide and clinical scores were assessed daily for 26 days. Data represent means ± SEM from two independent experiments with a total of 14 and 9 animals for mPGES-1^+/+ and mPGES-1^−/− mice, respectively. *, P < 0.0001 versus mPGES-1^+/+ mice determined by two-way repeated measures ANOVA. (D) LN cells from EAE-induced mPGES-1^+/+ and mPGES-1^−/− mice on day 11 were stimulated in vitro with various concentrations of MOG_35–55 peptide, and the proliferative responses were measured (n = 5 animals). Data represent means ± SEM. (E) Supernatants of LN cell cultures were used to measure the concentrations of IFN-γ, IL-17, TNF-α, and IL-6 (n = 5 animals). Data represent means ± SEM. *, P < 0.05, **, P < 0.01 versus mPGES-1^+/+ mice by two-tailed Student's t test. (F) Two days after activation, cells were restimulated with PMA/ionomycin and subjected to intracellular cytokine staining for IFN-γ and IL-17 (n = 5 animals). Each data point represents the results from a single animal and the horizontal bars designate the mean values of individual groups. *, P < 0.05, by two-tailed Student's t test.

mPGES-1 Is Expressed in Human MS Lesions.

To investigate whether the mPGES-1 is a therapeutic target in MS patients, we examined autopsy brain tissues obtained from MS patients (Fig. 4). In accordance with murine EAE (Fig. 3A), immunohistochemistry on MS lesions revealed the overt expression of mPGES-1 protein (Fig. 4 A, B, D, and E) in CD68⁺ macrophages (Fig. 4 G–Q). Immunoreactivity of mPGES-1 was not detected with antigen-absorbed Ab (Fig. 4 C and F). These data suggest that not only murine EAE, but also human MS pathology seems to be influenced by the mPGES-1-PGE₂ axis of the AA cascade.

Fig. 4.

mPGES-1 expression in infiltrated macrophages of MS lesions. Brain tissues of MS patients' autopsied materials displaying periventricular demyelinating lesions were stained with anti-mPGES-1. (Scale bar, 50 μm A, B, D, and E.) The high magnification images of A and D are shown in B and E, respectively. (Scale bar, 10 μm.) The antigen-peptide blocks the signals of anti-mPGES-1 Ab staining. (Scale bar, 10 μm C and F.) Both MS #1 (subacute stage, A–C) and MS #2 (acute stage, D–F) showed extensive infiltration of macrophages characterized by foamy appearance. The mPGES-1 is colocalized with CD68-immunoreactive macrophage in the MS lesions. (Scale bar, 10 μm G–Q.)

Discussion

In the present study, we provided a comprehensive overview of the AA cascade dynamics in EAE lesions and determined the roles of mPGES-1 and PGE₂ in EAE. By the AA cascade-targeted lipidomics approach, we found that the PGE₂ pathway is favored and the PGD₂, PGI₂, and 5-LO pathways are attenuated. Correlation analysis imply that the AA producing enzyme cPLA₂α possibly passes more AA to the COX than the 5-LO pathway in EAE, because the sequential actions of eicosanoid-synthesizing enzymes are regulated spatially and temporally in the cells, the so-called functional coupling (16–18). Likewise, PGH₂, a common precursor of PGs, appears to be selectively consumed by the PGE₂ pathway rather than the PGI₂ and PGD₂ pathways, thereby producing PGE₂ in the SCs of EAE mice.

The symptoms of EAE in COX-2^−/− mice were comparable to those of wild-type controls and administration of indomethacin, a nonselective COX inhibitor, prevented development of the disease (14). We found that COX-2 expression was substantially up-regulated in the SCs of EAE mice from the induction phase (Fig. 1B). The absolute number of microglia/macrophages increased before the onset and throughout the disease course (23), and COX-2 is generally induced by the inflammatory mediators in these cells (24), implying that the up-regulation of COX-2 depended on these cells. In human MS, Rose et al. reported the COX-2 expression in microglia/macrophages of MS lesions (25). We demonstrated here that COX-1 mRNA expression is up-regulated in the SCs of EAE mice (Fig. 1B) and is correlated with clinical symptoms (Fig. S2). Deininger and Schluesener showed the constitutive expression of COX-1 in the microglia/macrophages of rat brain and an elevation of its expression levels in the cells with the progression of EAE (26). Further studies are needed to disclose the functions of COX-1 and COX-2 in EAE pathology.

Among the eicosanoids examined, only the PGE₂ level was significantly elevated in the SCs of EAE mice (Fig. 2). This up-regulation seems to depended on the mPGES-1 expressed in the microglia/macrophages of EAE lesions, because the PGE₂ production was almost completely suppressed in mPGES-1^−/− mice (Fig. 3B). The expression of EP1, EP2, and EP4 was altered depending on the disease severity (Fig. 1B and Fig. S2). Therefore, these EPs are candidates for the downstream effectors of PGE₂. Indeed, the exogenous and endogenous PGE₂ activates EP2 and EP4 on antigen presenting cells to stimulate expression of IL-23 and IL-6, resulting in a shift toward T_H17 responses (27–30). PGE₂ synergized with IL-23 in expanding human T_H17 cell (31) and directly promoted differentiation of human and murine T_H17 cells through EP2/EP4 signaling (32). In addition, Yao et al. (30) recently reported that PGE₂ acting on EP4 on T cells amplifies IL-23-mediated T_H17 expansion in vitro and an EP4-selective antagonist suppressed the EAE pathology. In the case of T_H1 immune responses, differentiation of T_H1 cells from naive T cells in vitro was facilitated by EP1 (33), EP2, or EP4 agonists (30). T cells appear to encounter the autoantigens mainly in the peripheral lymphoid organs in the induction phase, and then in inflammatory foci of CNS in the acute and chronic phases. The levels of PGE₂ in the spleen were unchanged during the disease course (Fig. S4), whereas those in the SCs were drastically elevated in wild-type mice (Fig. 2). Furthermore, the EAE pathology of mPGES-1^−/− mice was attenuated in the chronic phase. These results imply that EP4-expressing T cells are affected by PGE₂ in the CNS rather than in peripheral lymphoid organs. We showed that the mPGES-1 deficiency abrogated the T_H1 and T_H17 cytokine production in EAE (Fig. 3 E and F), suggesting that PGE₂ produced by microglia/macrophages in inflammatory foci may aggravate EAE by promoting the differentiation of T_H1 and the expansion of T_H17 cells through these EPs (Fig. 5).

Fig. 5.

Conceptual model for the role of AA cascade in EAE pathology. PGE₂ is produced by mPGES-1 in activated microglia (MG), infiltrating macrophages (MΦ) and dendritic cells (DC), and then activates these cells in an autocrine/paracrine manner through EP2/EP4. T cells differentiate into T_H17 cells by stimulation with IL-6 and IL-23 secreted from activated MG/MΦ/DC. T_H1 differentiation and expansion of T_H17 cells are accelerated by the direct effects of PGE₂ on EP1 and EP2/EP4.

We found that the PGIS expression was essentially diminished in EAE-induced C57BL/6 mice, resulting in a significant decrease of 6-keto-PGF_1α levels in the acute phase of EAE (Fig. 2). Because PGI₂ enhances endothelial barrier function (34), PGI₂ seems to protect mice from EAE by suppressing the inflammatory cell infiltration. Several reports showed that loss of mPGES-1 results in the rediversion of a substrate PGH₂ to other prostanoids, like PGI₂, which may affect the physiological and pathophysiological conditions of mPGES-1^−/− mice (35–37). Therefore, it is possible that the amelioration of EAE in mPGES-1^−/− mice is due to the up-regulation of PGI₂ levels (Fig. S6). On the other hand, the collagen-induced arthritis was aggravated by the PGI₂-IP axis (38), raising another possibility that the PGI₂ is an exacerbating factor in T_H1 and T_H17 immune responses and EAE pathology. The roles of PGI₂ in EAE pathology remain controversial, and further studies are needed.

Reduction in PGD₂ levels in the acute phase and its recovery in the chronic phase could be explained by the expression changes of L-PGDS (Figs. 1B and 2). One of the features of EAE/MS is demyelination, which coincides with oligodendrocyte cell death. L-PGDS is thought to be an anti-apoptotic molecule that protects oligodendrocytes, because the oligodendrocyte apoptosis was enhanced in the brain of L-PGDS-deficient twitcher mice (39). In human MS, L-PGDS expression was increased especially in the remyelinated lesions (40). Thus, the decrease and the subsequent recovery of L-PGDS may be associated with the demyelination and remyelination in EAE/MS, respectively. Another PGD synthase, H-PGDS, showed the pattern opposite to L-PGDS expression (Fig. 1B). Although H-PGDS is highly expressed in hematopoietic cells and microglia/macrophages (41), PGD₂ levels were not elevated in the SCs of EAE mice in the acute phase. Because these cells also express mPGES-1, COX-1/2 seem to prefer to functionally couple with mPGES-1 rather than H-PGDS in the cells of EAE lesions.

Contrary to the up-regulation of 5-LO/FLAP transcripts (Fig. 1B), we demonstrated here the significant suppression of 5-LO metabolites during EAE pathology (Fig. 2). The discrepancy between transcripts and metabolites is explained by posttranscriptional/translational (such as phosphorylation) regulation of enzymes (42–44) and/or functional coupling with cPLA₂α. Furthermore, COX enzymes were up-regulated in the SC around the same time (Fig. 1B), which results in the reduction of the substrate availability for 5-LO. Emerson and LeVine reported that 5-LO^−/− mice developed more severe EAE than control wild-type mice (15), suggesting a possibility of increased PGE₂ production in the SCs of 5-LO^−/− mice. Meanwhile, the report raises another possibility that lipid mediators downstream of 5-LO may play protective roles in EAE pathology. Because Gladue et al. (45) showed that a BLT1 antagonist protects mice from EAE, LTB₄ is not such a candidate for the protective lipid mediators. 5-LO generates not only LTs but also lipoxins and resolvins (46, 47), which are anti-inflammatory/proresolving lipid mediators derived from AA and eicosapentaenoic acid, respectively, although we did not measure these lipid mediators in the current study. 12/15-LO deficiency conferred more severe EAE pathology (15), possibly because 12/15-LO also mediates lipoxin production. Resolvin E1 reduces dendritic cell migration and IL-12 production, suggesting a protective role in EAE pathology (48).

As a therapeutic strategy, blocking the hub of the AA cascade, like COX-1/2, seems to be highly effective in preventing EAE pathology. Although treatment of indomethacin actually ameliorates EAE, all mice died of gastrointestinal ulcers and bleeding (14), a well-known side effect of NSAIDs (49). In 2004, rofecoxib, a selective COX-2 inhibitor, was withdrawn because of an increased cardiovascular risk in patients taking the drug for >18 months (50, 51). A possible reason why NSAIDs are currently only used for treatment of flu-like symptoms associated with IFN therapy for MS is that inhibition of COX-1/2 may put the patients at high risk for adverse effects. Therefore, it seems that targeting AA cascade downstream of COX-1/2 is highly effective in treating MS without side effects. As discussed above, components of AA cascade contribute to EAE pathology via intricate mechanisms (Fig. 5), while our AA cascade-targeted lipidomics approach and knockout study identified the mPGES-1-PGE₂-EPs axis as the critical pathway of AA cascade in EAE pathology. Given the function of EPs in T_H1 and T_H17 immune responses (27–29, 38) and our findings of mPGES-1 expression in MS lesions (Fig. 4), inhibition of mPGES-1 has a much better chance of blocking individual EPs. Although mPGES-1 inhibitors which effectively suppress PGE₂ production in rodents (22) have not been published yet, MF63, a selective human mPGES-1 inhibitor, suppressed PGE₂ production, pyresis and inflammatory pain in knock-in mouse expressing human mPGES-1 (52). We believe that mPGES-1 inhibitors provide a significant treatment option for MS.

Materials and Methods

Mice.

Female C57BL/6 mice were purchased from Charles River for lipidomics analysis. The mPGES-1^−/− mice were generated as described in ref. 53. See SI Materials and Methods for more information.

Induction of EAE.

Mice were s.c. immunized with MOG_35–55 peptide. Pertussis toxin was i.p. injected. See SI Materials and Methods for more information.

Quantification of Eicosanoids and mRNA Levels in SCs.

Eicosanoid levels were estimated simultaneously as described in refs. 13, 19, 54, and 55. See SI Materials and Methods for more information.

Ex Vivo Experiments.

The details of recall responses to the MOG_35–55 peptide antigen, ELISA, and flow cytometry are provided in SI Materials and Methods.

Immunohistochemistry.

For murine EAE, SCs obtained from EAE-induced mice were stained with either anti-F4/80 (Serotec) or CD4 (BD Biosciences) Ab and mPGES-1 (Affinity BioReagents) Ab as primary Abs. For human MS, cerebral specimens were stained with rabbit anti-human mPGES-1 polyclonal Ab (Cayman) or mouse anti-human CD68 monoclonal Ab (DAKO, clone KP1) as a first Ab. See SI Materials and Methods for more information.

Statistical Analysis.

As appropriate, data were analyzed statistically by means of a Student's t test, Mann–Whitney U test, Fisher's exact test, two-way repeated measures ANOVA, or Kruskal-Wallis test with Dunn's post-hoc test. Statistical, clustering, and correlation analyses were performed using JMP6 (SAS Institute) or Prism 4 (GraphPad). P values <0.05 were considered to be statistically significant.