Anti-inflammatory/anti-amyloidogenic effects of plasmalogens in lipopolysaccharide-induced neuroinflammation in adult mice

Male C57/6J mice weighing 32 to 37 g (10 months old) were used in all experiments. The animals were housed five per cage at a temperature of 22 ± 2°C with 12 hour light/12 hour dark cycles (lights on at 8:00) with free access to laboratory food and water. The mice were randomly divided into four groups: control (Con), LPS, LPS + Pls and Pls. LPS (Sigma-Aldrich, St. Louis, MO, USA) was dissolved in saline, while the Pls were dissolved in corn oil then sonicated to ensure complete solubilization. The LPS group received an i.p. injection of LPS (250 μg/kg) followed by corn oil in the morning (9:00 to 10:00) daily for seven days (days 1 to 7). The LPS + Pls group was treated with LPS followed by Pls (20 mg/kg) for seven days, while the Pls group was injected with saline and Pls. The Con group was given saline and corn oil for seven days. All animals were sacrificed on day 8. The body weights were measured in the morning before injection on day 1 to day 8.

The Pls used in the present study were prepared from chicken breast muscle using a previously reported method [21]. A HPLC used for phospholipid separation [22] indicated that the purified Pls consisted of 47.6% PlsEtn, 49.3% PlsCho, 2.4% sphingomyelin (SM) and 0.5% other phospholipids. The composition of fatty acids of PlsEtn and PlsCho was analyzed using the previously described HPLC method [21], as shown in Table 1.

The mice were deeply anesthetized with pentobarbital (50 mg/kg) and transcardially perfused with PBS followed by 4% paraformaldehyde. For each animal, the brain was removed, post-fixed for 24 hours and then transferred successively to 20% and 30% sucrose solutions. Subsequently, the brains were frozen on a cold stage and sliced into 30 μm thicknesses using a cryostat. The sections were permeabilized with 0.3% Triton-X 100 (Sigma-Aldrich) in PBS for 15 minutes and blocked in PBS containing 1% BSA and 5% normal donkey serum (Jackson ImmunoResearch Lab., West Grove, PA, USA) for 60 minutes at room temperature. The sections were incubated in blocking solution (Block Ace, Dainippon Pharmaceutical, Osaka, Japan) for 30 minutes at room temperature and then incubated with rabbit polyclonal antibodies against Iba-1 (1:10000; Wako Pure Chem. Indus., Osaka, Japan), which are known to have specific affinity for microglial Ca²⁺-binding proteins and are highly expressed by activated microglia, and anti-glial fibrillary acidic proteins (GFAPs, 1:3000; Sigma-Aldrich) for astrocytes in 10% Block Ace in PBS at 4°C overnight. Other sections were incubated with polyclonal antibodies against Aβ_3–16 (1:1000; ab14220, Abcam, Cambridge, UK) and NeuN (1:1000; Millipore, Billerica, MA, USA). According to the manufacturer’s instructions, ab14220s react with all isoforms of mouse and rat Aβ. The rinsed sections were incubated for six hours with Alexa Fluor 488 goat anti-rabbit IgG or Alexa Flour 568 goat anti-mouse immunoglobulin G (IgG) (1:1000; Invitrogen, Eugene, OR, USA) at room temperature. Every treatment was followed by washing three times for five minutes with PBS. The sections were then mounted in the perma fluor aqueous mounting medium (Thermo Fisher Scientific, Waltham. MA, USA).

All samples were analyzed with a confocal laser-scanning microscope (LSM510 Meta; Carl Zeiss, Jena, Germany). The number of glial cells in 90 to 100 areas of 200 μm x 200 μm in four slices per brain was counted and the average number/4 x 10⁴ μm² was obtained for each brain.

Mice were deeply anaesthetized with pentobarbital and perfused transcardially with phosphate buffered saline, then the PFC and hippocampus were removed immediately. Total RNA was isolated from the samples using magnetic beads (MagExtractor system; Toyobo, Tokyo, Japan) after homogenizing the tissues. Primer pairs were chosen to flank at least one intron. The amount of total RNA was quantified by measuring the optical density 250 using a Nanodrop spectrophotometer (Nanodrop, Wilmington, DE, USA). For reverse transcription, 100 ng of total RNA was transferred to the reaction using an RNA PCR kit (AMV) (Takara Bio Inc., Ootsu, Japan) and 9-mer random primers. SYBR-Green real-time PCR (Applied Biosystems, Foster City, CA, USA) was performed on cDNA prepared from each sample using the THUNDERBIRD SYBR qPCR Mix, ROX reference dye (Toyobo) and 0.5 mM of each primer. The data were analyzed using 7500 System software v2.0 (Applied Biosystems). All values of cytokines were normalized to the corresponding β-actin concentration obtained using the same method. The sequences of primers were follows: IL-1β, sense; 5′- CTCCATGAGCTTTGTACAAGG -3′, antisense; 5′- TGCTGATGTACCAGTTGGGG -3′; TNF-α, sense; 5′- CCACCACGCTCTTCTGTCTAC -3′, antisense; 5′- TGGGCTACAGGCTTGTCACT -3′ β-actin, sense; 5′- TTGCTGACAGGATGCAGAAGGAG -3′, antisense; 5′- GTGGACAGTGAGGCCAGGAT -3′. The predicted sizes of the PCR products were 245 bp for IL-1β, 196 bp for TNF-α, and 127 bp for β-actin mRNA.

Mice were deeply anesthetized with pentobarbital (50 mg/kg) and transcardially perfused with sterile PBS. For each animal, the brain was removed and the PFC and hippocampus were dissected in a dish filled with ice-cold PBS. The samples (300 mg to 500 mg) were stored at −80°C until Pl measurement. Total lipids were extracted using the method of Folch and coworkers [23], and the relative composition of phospholipid classes, including Pls, was measured as previously reported [22].

The results are expressed as the mean ± SEM. The body weights (BWs), numbers of Iba-1⁺ and GFAP⁺ cells and amounts of mRNAs were compared using one-way analysis of variance (ANOVA) followed by post hoc (Scheffe’s) test. Changes in the PlsEtn levels and the ratio of PlsEtn/Phosphatidyl Etn (PEtn) determined after LPS and Pl injection were evaluated using the non-parametric Kruskal-Wallis test followed by the Steel test for multiple comparisons. Values of P <0.05 were considered to be statistically significant.

The BWs of the mice in the LPS group started to decrease on day 2 and showed significant differences between groups on day 4 that lasted until day 8 (day 4: F(3,28) = 7.1, day 5: F(3,28) = 8.1, day 6: F(3,28) = 6.0, day 7: F(3,28) = 9.0 and day 8: F(3,28) = 9.4, P <0.01, respectively, each group, n = 8). The post hoc test indicated that the BWs of the LPS group were different from those of the control (Con) group (day 4, 5, 7 and 8, P <0.05) and the Pls group (from day 4 to 8, P <0.01). However, the LPS + Pls group showed no significant differences between either the Con or Pls group in terms of BW at any point (Figure 1).

As shown in Figure 2Aa, the Con group that received saline and corn oil for seven days showed typical features of Iba-1-positive (green) resting microglia with small and compact soma bearing ramified processes (a’) in the PFC. GFAP was immunostained with weak fluorescence (red) in astrocytes (b). However, the i.p. administration of LPS (250 μg/kg/day) for seven days (LPS group, second row) resulted in neuroinflammation showing increased numbers of Iba-1-positive microglia and intense immunoreactivity (d) with activated phenotypes of marked cellular hypertrophy and retraction of cytoplasmic processes (d’). GFAP-positive astrocytes also increased in number and intensity (e). As shown in Figure 2Ag and h, the increases in the number of activated microglia and astrocytes in the PFC were suppressed by i.p. administration of LPS and Pls (20 mg/kg) (LPS + Pls group). Iba-1-positive microglia and GFAP-positive astrocytes did not merge with each other in all groups (c, f, i, and l). Figure 2B shows a summary of the LPS-induced increases in the number of glial cells and the suppression of this increase by Pls (each bar, n = 8). The number of microglia (left) and astrocytes (right) significantly increased following LPS injection (F(3,28) = 38.4, P <0.01; F(3,28) = 45.8, P <0.01, respectively). The multiple-range test indicated that the numbers of microglia and astrocytes in the LPS group were different from those in the Con, Pls, and LPS + Pls groups (Scheffe’s test, P <0.01, respectively), while the LPS + Pls group did not differ from the Con or Pls groups for either microglia or astrocytes.

In the CA1 region of the hippocampus, both Iba-1-positive microglia and GFAP-positive astrocytes increased in number in the LPS group (Figure 3Ad and e) compared with that observed in the control group (a and b). Similar to that observed in the PFC, the increases in the number of activated glial cells were attenuated following the administration of Pls (g and h). As shown in Figure 3B, the statistical analysis indicated significant differences in the numbers of microglia (left: F(3,28) = 10.7, P <0.01) and astrocytes (right: F(3,28) = 11.2, P <0.01) between the groups (each bar, n = 8). LPS injection increased the numbers of microglia (left: P <0.01) and astrocytes (right: P <0.01). The increases were significantly suppressed by Pls to the levels of microglia and astrocytes observed in the Con and Pls groups (LPS + Pls group, P <0.05, respectively).

As shown in Figure 4A, the dentate gyrus (DG) of the hippocampus also demonstrated LPS-induced increases in the number and intensity of immunostaining for Iba-1 in microglia (d) and GFAP in astrocytes (e) compared with that observed in the Con group (a and b). These increases were suppressed by simultaneous injection of Pls (g and f, respectively). One-way ANOVA indicated significant differences in the numbers of microglia (Figure 4B, left: F(3,28) = 41.4, P <0.01) and astrocytes (right: F(3,28) = 11.5, P <0.01) between the groups (each bar, n = 8). The differences in the numbers of microglia and astrocytes in the Con and Pls groups (P <0.01) and the LPS + Pls group were significant (P <0.01 and P <0.05, respectively).

As shown in Figure 5A, the relative amounts of mRNAs for IL-1β (left) and TNF-α (right) in the PFC were significantly different (ANOVA test, F(3,16) = 19.3, P <0.01, and F(3,16) = 14.9, P <0.01, respectively) between the groups (each bar, n = 5). The post hoc test indicated that the levels of both IL-1β and TNF-α mRNAs in the LPS group were significantly higher than those observed in the other groups (Scheffe’s test, P <0.01, respectively), while the levels in the LPS + Pls group were not different from those observed in the Con and Pls groups, suggesting that Pls have a suppressive effect on LPS-induced cytokine expression (Figure 5A).

Similar to the findings observed in the PFC, both cytokine mRNAs increased in the hippocampus in the LPS group (left: IL-1β, F(3,16) = 10.1, P <0.01; right: TNF-α, F(3,16) = 17.7, P <0.01) in comparison to that observed in the other three groups (Figure 5 B, P <0.01, respectively; each bar, n = 5).

The presence of LPS-induced amyloidogenesis in the PFC and hippocampus following LPS injection was investigated. As shown in Figure 6, a weak fluorescence for Aβ_3–16 immunoreactivity (green) in the PFC of the control group (Con, b) was apparently increased in the LPS group (e). The increase in Aβ_3–16 immunoreactivity was attenuated in the LPS + Pls group (h). Neurons in the cortex were stained with NeuN (red, a, d and g), and most of the Aβ_3–16 fluorescence merged with the NeuN immunoreactivity, indicating intracellular localization of Aβ (yellow in f).

Accumulations of Aβ were also observed in the CA1 region of the hippocampus in the LPS group (Figure 7, e). Similar to that observed in the PFC, enhanced Aβ_3–16 immunoreactivity was markedly attenuated by Pl administration (LPS + Pls group, h). Again, the presence of intracellular accumulation of Aβ in the pyramidal neurons was detected using double staining for Aβ_3–16 and NeuN (f and i).

As shown in Table 2, the levels of PlsEtn were much higher than those of PlsCho both in the PFC and the hippocampus, although more than 60% of the phospholipids were diacylgylycereophospholipids (PEtn, PCho, PS and PI). The relative levels of PlsEtn in the PFC significantly changed following LPS injection (Table 2, upper, Kruskal-Wallis test, χ ²(3) = 10.7, P <0.05). A multiple comparison analysis using the Steel test revealed that the amount of PlsEtn significantly decreased in the LPS group in comparison to that observed in the control group (P <0.05); however, the levels in the LPS + Pls group were not different from those observed in the control group (each group: n = 5). Therefore, the ratio of PlsEtn/PEtn significantly decreased following LPS treatment (χ ²(3) = 11.5, P <0.01 and Steel test, P <0.05) and recovered after the i.p. administration of Pls.

As shown in the lower part of Table 2, the relative levels of PlsEtn in the hippocampus were greater than those in the PFC in any group (for example, 16.40 ± 0.59 versus 24.18 ± 0.64 in the Con group). Similar to that observed in the PFC, both the relative levels of PlsEtn and the PlsEtn/PEtn ratio were significantly reduced by LPS treatment (Kruskal-Wallis test, χ ²(3) = 11.3 and χ ²(3) = 11.4, P <0.01, and Steel test, P <0.05, respectively). The reduction of the PlsEtn levels in the hippocampus was also reversed by Pl administration. The levels of the other phospholipids did not show any significant changes following all of the treatments.

It is well known that the activation of microglia and astrocytes plays an important role in neuroinflammation induced by systemic LPS by enhancing the secretion of cytokines, prostanoids, ROS/RNS and related substances. In the present study, i.p. injection of LPS for seven days induced morphological activation and increased the number of glial cells in the PFC and hippocampus (Figures 2 4). The amounts of IL-1β and TNF-α mRNAs, which are considered to be derived primarily from glial cells, also increased following LPS injection (Figure 5). Furthermore, LPS injection resulted in the intracellular accumulation of Aβ proteins in both regions (Figures 6 and 7). It has been reported that i.p. injection of LPS induces deficits in spatial learning in mice [24, 25] that may be due to the enhancement of Aβ generation in the hippocampus [1].

It is known that β-secretase is involved in amyloidogenic processing of amyloid precursor proteins at the first step, while γ-secretase yields Aβ isoforms such as the more prevalent Aβ₄₀ and aggregation-prone Aβ₄₂ at the last step [26]. It has been recently shown that the activities of β- and γ-secretase are increased in the cortex and hippocampus following systemic injection of LPS [1]. It is possible that microglia play important roles in this phenomenon since proinflammatory cytokines, as well as ROS/RNS, released from activated microglia augment Aβ formation by upregulating β-secretase mRNA and enzymatic activity [5, 27]. Microglia are activated further through receptors for advanced glycation end products (RAGE), which bind to Aβ and induce phagocytosis of Aβ, thereby amplifying the generation of ROS/RNS and cytokines [26].

In addition to observing the activation of glial cells and Aβ accumulation, we found that the Pl levels in the PFC and hippocampus decreased following LPS administration (Table 2). It is possible that the decreases in the amount of Pls during neuroinflammation are due to the anti-oxidant properties of Pls. It has been shown that the Pl-specific vinyl ether bond at the sn-1 of the glycerol backbone is targeted by a variety of oxidants, including ROS/RNS [10, 11, 13], and oxidative stress preferentially oxidizes PlsEtn over phosphatidyl ethanolamine (PEtn) [28, 29], resulting in the disruption of vesicular fusion in the synaptosomes and the decrease in acetylcholine release [30]. This may at least partly explain why AD patients show decreases in Pl levels in the brain [14‐16]. It has been suggested that abnormal membrane lipid compositions, namely decreases in the ratio of Pl to non-Pl ethanolamine glycerophospholipids, cause membrane instability in AD, which may contribute to amyloidogensis by cooperatively acting with amyloid cascade mechanisms [14]. Furthermore, since PlsEtn are major endogenous lipid constituents that facilitate membrane fusion of synaptic vesicles associated with neurotransmitter release [31, 32], pathological and/or age-related alterations in the Pl levels may be attributed to neurological disorders including AD [7]. In accordance with this, it has been reported that decreases in the amount of Pls are closely correlated with the severity of dementia in humans [18, 19].

There seems to be a causes/consequences loop involving neuroinflammation that includes cytokine and ROS/RNS production, Aβ accumulation and decreases in the amount of Pls. LPS-induced activation of β-secretase [1], which is predominantly localized in cholesterol-rich lipid rafts [33, 34], causes accumulation of Aβ proteins. Aβ-induced production of ROS/RNS that enhance lipid peroxidation [35, 36] may decrease Pl levels, as mentioned above. In addition, increases in Aβ, cytokines and ROS/RNS reduce the expression of alkyl-dihydroxyacetone phosphate-synthase, a rate-limiting enzyme for Pl de novo synthesis, by inducing the dysfunction of peroxisomes, where Pls are biosynthesized, resulting in decreases in the Pl levels [37]. It has also been reported that TNF-α down regulates another key enzyme in Pl biosynthesis in peroxisomes, glycerol-3-phosphate-O-acyltransferase [38], and up regulates myeloperoxidase, which generates one of the reactive species, hypochlorous acid (HOCl), in the brain, targeting Pls to be oxidized [39]. Finally, Pls-PLA₂, which degrades Pls to release DHA or ARA from the sn-2 position of the glycerol moiety, is possibly activated by ceramide produced under inflammatory conditions, and contributes to the loss of Pls in the brain [9, 40].

It is well known that the generation and clearance of Aβ are affected by cholesterol metabolism, as evidenced by the identification of a variant gene of apolipoprotein E, a cholesterol transporter, as a major genetic risk factor for AD [26, 41, 42]. It has been shown that decreases in the amount of Pls induce a decreased rate of intracellular cholesterol transport from cell membranes to the endoplasmic reticulum, which increases the cholesterol levels in cell membranes [43]. Mankidy et al. further indicated that esterification of cholesterol, an obligate step that occurs prior to efflux from cells, is dependent upon the amount of polyunsaturated fatty acid (PUFA)-containing PlsEtn present in the membrane with increasing levels of the membrane-bound cholesterol-processing enzyme, sterol-O-acyltransferase-1 [44]. Increases in the cholesterol levels promote the secretion of Aβ [41, 42, 45], while depletion of cholesterol inhibits the generation of Aβ [46, 47]. Furthermore, it has been shown that membrane Pls block cholesterol-mediated increases in β-secretase activity and directly increase the activity of α-secretase, which is known to promote non-amyloidogenic processing of amyloid precursor proteins [48]. Therefore, a vicious circle in which LPS-induced Aβ accumulation decreases the Pl levels, which leads to increased cholesterol levels, which further enhances the generation of Aβ may be involved in the pathological conditions of neuroinflammation.

In the present study, we showed that LPS-induced activation of glial cells (Figures 2 4), expression of IL-1β and TNF-α mRNAs (Figure 5), accumulation of Aβ proteins (Figures 6 and 7) and decreases in the PlsEtn levels (Table 2) in the PFC and hippocampus are all prevented by co-administration of Pls. Although the precise mechanisms underlying the effects of Pls in this study are not known, supplementation with Pls could improve pathological disorders. The most important question may be whether peripheral Pls can enter into the brain. So far, there are no reports indicating that Pls directly cross the blood–brain barrier (BBB). Therefore, it is not excluded that the anti-oxidative effects of Pls are exerted outside the brain in order to suppress primary inflammation induced by peripheral LPS. However, it has been shown that the Pls levels in sera are decreased in parallel with or even at earlier times than decreases in the brain Pl levels in AD patients [18, 19]. Furthermore, our results showed that LPS-induced decreases in the Pls levels in the PFC and hippocampus are corrected with peripheral administration of Pls (Table 2). Therefore, it is possible that peripheral supplementation with Pls would have effects on the CNS by changing the Pl levels in the brain.

Another question is whether the effective molecules in our experiment are PUFAs, not Pls, which Pls must carry at the sn-2 position. Several lines of evidence show that n-3 PUFAs, such as eicosapentaenoic acid, DHA, and its derivative, neuroprotectin D1, have anti-inflammatory and neuroprotective effects [49‐52]. Furthermore, DHA has been reported to suppress the production of Aβ proteins through multiple mechanisms, including inhibition of β-/γ-secretase activities and alteration of membrane cholesterol distribution [53‐55]. Since the purified Pls used in the present study contained DHA and its precursor, α-linolenic acid, especially in PlsEtn (Table 1), it cannot be excluded that DHA derived from PlsEtn plays a significant role in the CNS effects of Pls. Indeed, it has been shown that DHA is synthesized from α-linolenic acid and incorporated into phospholipids in the liver then transported to the brain through the peripheral circulation [56]. On the other hand, it has also been shown that lyso-type phospholipids, which contain DHA at the sn-2, show preferential transfer over DHA in in vitro models of the BBB [57]. Furthermore, it has been suggested that specific transport mechanisms to import Pls and their synthetic precursors exist in brain capillary epithelial cells [58, 59]. These findings suggest that Pls containing DHA exert more effective actions in the CNS than DHA alone.

The present study suggests that co-administration of Pls suppresses systemic LPS-induced neuroinflammation in the brain. Although further studies on the mechanisms underlying these CNS effects, including the metabolism of the administered Pls and the pathways used to enter the brain, are needed, the present results indicate that Pls may possibly be used in new preventive and therapeutic strategies for treating AD.