Introduction

In healthy humans, neuroinflammation serves as a vital protective mechanism which play a defensive role against pathogens. However, when it becomes chronic and unregulated, it can become detrimental due to the sustained release of cytokines and other proinflammatory molecules [1]. Chronic low-grade neuroinflammation is a characteristic feature of aging and plays a key role in the onset of age-related cognitive decline and various health issues [2]. It induces the activation of brain immune cells called microglia, which regulate inflammation. In a healthy brain, these cells express a homeostatic signature and are involved in environmental monitoring and neurogenesis. During aging, microglial cells become more reactive and produce more proinflammatory cytokines and chemokines, such as TNF-α, IL-1β, IL-6, CCL2 and CXCL10 [3,4,5]. They accumulate and may also polarize towards an MGnD phenotype (microglial neurodegenerative phenotype), which is characteristic of aging, thereby disrupting the inflammatory response [6]. The chronic low-grade inflammation associated with aging induces neuronal alterations leading to a decline in cognitive function, with short- and long-term memory deficits. Indeed, high levels of plasma IL-6 are correlated with impaired cognitive function in elderly people [7, 8]. Conversely, IL-6 deficiency in mice has been shown to protect against age-related cognitive decline [9]. Inflammatory processes are also intricately regulated by specific lipid mediators, called oxylipins, which, while triggering inflammation on one hand, preferably facilitate self-resolution on the other [10, 11]. Oxylipins are produced on demand and are not stocked, they can be found throughout the body in all tissues, urine and blood. Brain is equipped with enzymes allowing their synthesis [12]. There is a growing literature on the role of oxylipins in inflammation during normal or pathological aging [13, 14]. In fact, older healthy people are reported to have a different plasma oxylipin profile than healthy young people [15]. Moreover, LxA4, a proresolving mediator, was found to be reduced in the cerebrospinal fluid and hippocampus of Alzheimer’s patients and to be positively correlated with better scores on a memory test [16]. Therefore, restricting overly brain inflammatory reactions and enhancing the process of resolution are crucial for the return to homeostasis and to prevent alterations linked to an uncontrolled inflammatory response.

Strong evidence suggests that nutrition plays a significant role in influencing inflammation. Several nutrients, such as polyphenols, carotenoids and omega-3 polyunsaturated fatty acids (omega-3), have immunomodulatory properties and are good candidates to prevent impairments related to neuroinflammation. In fact, polyphenols exhibit anti-inflammatory effects through their ability to reduce LPS-induced microglial activation [17]. Notably, it has been shown that a blueberry-mulberry extract, rich in polyphenols, decreases the expression of proinflammatory markers in the brain and intestine of aged mice [18]. Furthermore, resveratrol, a stilbene present in grapes, decreases LPS-induced protein and gene expressions of IL-1β in the plasma and hippocampus of aged mice and reduces the expression of proinflammatory markers such as IL-1β and COX-2 in LPS-stimulated microglial cells [19, 20]. Studies reported that saffron carotenoids also have anti-inflammatory properties, notably by decreasing the levels of IL-6 and TNF-α in the plasma of patients with type 2 diabetes compared to placebo [21]. In rats, saffron used as a treatment for 7 days in a model of chronic neuropathy, attenuates the production of proinflammatory factors such as IL-6, IL-1β and TNF-α in the spinal cord [22]. Moreover, it has been demonstrated in LPS-stimulated microglial cells from rats that saffron carotenoids, crocin and crocetin, decrease the expression of the proinflammatory cytokines TNF-α and IL-1β [23]. Safranal, another constituent of saffron, also decreases neuroinflammation by downregulating NF-κB signaling pathway and proinflammatory cytokines expression in mice model of epilepsy and in rat model of Alzheimer’s disease [24, 25]. In addition, omega-3 have been widely described for their immunomodulatory properties. Indeed, DHA supplementation reduces the serum levels of inflammatory markers in humans [26]. In aged mice, 2-month EPA and DHA supplementation prevents the increase of proinflammatory cytokines IL-6, IL-1β and TNF-α expression in the hippocampus and plasma [27]. Oxylipin profile can be modulated by some nutrients such as polyphenols and omega-3. Indeed, despite few studies on the effects of polyphenols, Hartung et al.., reported that genistein, a flavonoid, resveratrol and its derivatives modulate the oxylipin profile and inhibit 5-LOX, an enzyme implicated in oxylipin biosynthesis, in human neutrophils [28]. Moreover, 18 days of blueberry supplementation increases plasma levels of proresolutive oxylipins and decreases those of proinflammatory derivates in humans [29]. DHA has also been shown, both in humans and mice, to increase the synthesis of proresolving oxylipins, which play a key role in resolving inflammation [12, 30,31,32,33]. To our knowledge, no studies have investigated the effects of saffron carotenoids on brain oxylipin profile.

Combining these nutrients could be pertinent for preventing neuroinflammation by having additional effects. Hence, our study explored the impact of a mix of plant extracts that were chosen according to previous results [34,35,36,37,38,39]. This mix contained polyphenols sourced from grapes and blueberries, as well as saffron carotenoids, and omega-3 fatty acids from fish, on LPS-induced neuroinflammation in mice and compare it to the effects of plant extracts or omega-3 alone.

Materials and methods

Animals and treatments

All tests were conducted with 5-month-old male C57Bl/6J mice from Janvier Labs (Le Genest-Saint-Isle, France). The mice were kept under standard housing conditions with cellulose litter in an environment controlled for temperature (21 ± 2 °C) and humidity (40%), following a 12-hour light/dark cycle (7:00–19:00), and were provided ad libitum access to water and food (A04, Safe, Augy, France). A total of 149 animals were used. Over an 18-day period, the mice were subjected to force-feeding through a gastric tube with 150 µL of supplements, including a plant extract containing polyphenols from grape and blueberry (3.69 mg/day, equivalent to 600 mg/day in humans) (Memophenol™ Activ’Inside, Beychac-et-Caillau, France) and carotenoids and safranal from saffron extract (0.1845 mg/day, equivalent to 30 mg/day in humans) (patent pending WO2021209455A1 Activ’Inside, Beychac-et-Caillau, France) (n = 30), omega-3 (om-3) in triglyceride form (containing 6.15 mg/day of DHA and 1.41 mg/day of EPA, equivalent to 1000 and 229.27 mg/day and in humans) (n = 30), or a combination of both (n = 29) (Table 1). Control mice received water and peanut oil in the same way (n = 60). Plant extracts and omega-3 were diluted in water or peanut oil, respectively. At the end of the experiment, mice were intraperitoneally injected with a saline solution for the control group (n = 30) or a solution of 250 µg/kg of LPS (lipopolysaccharide; Escherichia coli, 0127: B8; Sigma-Aldrich, Lyon, France, n = 119) to induce acute inflammation. To examine the impact of inflammation and supplementation over time, mice were euthanized at 30 min (n = 49), 2 h (n = 50), or 6 h (n = 50) after the injection. Hippocampus and spleen were isolated and stored at -80 °C until analysis.

Table 1 Composition of the supplements
Full size table

Quantitative real time PCR

Total RNA from hippocampus was extracted using the TRIzol extraction protocol (Invitrogen, Life Technologies, France). The quantity and purity of the RNA were assessed using a Nanodrop photospectrometer (Nanodrop One, Life Technologies, France). Two micrograms of RNA were subjected to reverse transcription to generate complementary DNA (cDNA) using Superscript IV (Invitrogen, Life Technologies, France). Subsequently, the cDNAs were amplified through PCR with TaqMan® primers specific to the target genes studied, as described in previous studies [12, 40]. We focused on IL-6 (Mm00446190_m1), IL-1β (Mm00434228_m1), TNFα (Mm00443258_m1), TGF-β1 (Mm01178820_m1), CCL2 (Mm00441242_m1), CXCL9 (Mm00434946_m1), CXCL10 (Mm00445235_m1), IFNγ (Mm01168134_m1), NF-κB (Mm00476361_m1), TLR4 (Mm00445273_m1), CD11b (Mm00434455_m1), iNOS (Mm00440502_m1), CD68 (Mm03047343_m1), CD86 (Mm00444540_m1), Tspo (Mm00437828_m1), SOCS3 (Mm00545913_s1), CD36 (Mm00432403_m1), Arg1 (Mm00475988_m1), Glut5 (Mm00600311_m1), P2ry12 (Mm00446026_m1), COX-2 (Mm00478374_m1), CD206 (Mm00485148_m1), CD11c (Mm00498698_m1), Trem2 (Mm04209424_g1), ApoE (Mm01307193_g1), Clec7a (Mm01183349_m1), Bax (Mm00432051_m1), Bcl2 (Mm00477631_m1), Caspase-3 (Mm01195085_m1), RAGE (Mm00545815_m1), Catalase (Mm00437992_m1), GPx1 (Mm00656767_g1), SOD1 (Mm01344233_g1), SOD2 (Mm01313000_m1), NRF2 (Mm00477784_m1), GSS (Mm00515065_m1), GSTp1 (Mm04213618_gH), BDNF (Mm04230607_s1), TrkB (Mm00435422_m1), NGF (Mm00443039_m1), TrkA (Mm01219406_m1), DCX (Mm00438400_m1), b3-tubulin (Mm00727586_s1), CaMKII (Mm00437967_m1), 5-LOX (Mm01182747_m1), 12-LOX (Mm00545833_m1) and 15-LOX (Mm00507789_m1) in hippocampus. GAPDH (Mm99999915_g1) was used as reference gene. Fluorescence levels were assessed using a LightCycler® 480 instrument II (Roche, La Rochelle, France), and the final quantification was analyzed through the comparative threshold cycle (Ct) method. The outcomes are presented as relative fold changes, calculated in reference to the control target mRNA expression, as outlined in previous studies [12, 40].

Measurement of IL-6 in the spleen

The measurement of IL-6 level in spleen tissue was conducted using the Mouse IL-6 ELISA kit (Sigma-Aldrich, Lyon, France) according to the manufacturer protocol. The concentration of IL-6 in each sample, expressed in pg/mL/mg of protein, was determined by spectrophotometry (Victor3V, PerkinElmer, France) based on the standard range provided by the kit supplier.

Lipids analysis

Lipids were extracted from the hippocampus at CSGA lab (CSGA, INRAE UMR 1324, Dijon, France), and the fatty acids were methylated following established procedures outlined in a previous study [41]. The resulting fatty acid methyl esters (FAMEs) were analyzed through gas chromatography using a Hewlett Packard Model 5890 series II gas chromatograph (Palo Alto, CA, USA). This system was equipped with a split/spitless injector, a flame ionization detector (Palo Alto, CA, USA), and a CPSIL-88 column (100 m×0.25 mm internal diameter; film thickness, 0.20 μm; Varian, Les Ulis, France). Hydrogen served as the carrier gas at an inlet pressure of 210 kPa. The procedure involved maintaining the oven temperature at 60 °C for 5 min, followed by an increase to 165 °C at a rate of 15 °C/min, holding for 1 min, and then raising it to 225 °C at 2 °C/min. The final temperature was maintained at 225 °C for 17 min. The injector and detector temperatures were kept at 250 °C and 280 °C, respectively. Identification of FAMEs was accomplished by comparing them with commercially available standards. The fatty acid composition was expressed as a percentage of the total detected fatty acids.

Oxylipin quantification

Metabolites derived from linoleic acid (LA), arachidonic acid (AA), docosahexaenoic acid (DHA), and eicosapentaenoic acid (EPA) were isolated from the hippocampus and subjected to analysis through mass spectrometry (LC-MS/MS) at the METATOUL platform (MetaboHUB, INSERM UMR 1048, I2MC, Toulouse, France). The analytical procedures closely followed the methods outlined by Le Faouder et al. [42]. The obtained results were expressed as pg/mg of protein.

Statistical analyses

Principal component analysis (PCA) was performed to differentiate the main profiles of biomarkers using R Software (R version 4.1.1, RStudio environment). Indeed, this statistical approach is suitable to large-scale data: PCA aims to summarize the variability of the data while maximizing their variance in order to identify profiles, called components. The components were calculated as a weighted linear combination of the initial variables, with higher weights for the most contributory variables. We conducted independent PCAs, between Control and LPS groups and between all LPS groups with or without supplementation, at 3 times (30 min, 2 h and 6 h after LPS injection), each time on the 90 available biomarkers (gene expression and oxylipin concentration in hippocampus, protein expression in spleen). The number of selected components was based on the bar plot (Cattell criteria) and on the percentage of the information contained in the selected components. For each component, biomarkers were considered contributors when their absolute correlation with the component was ≥ 0.50. Two-dimensional plots of individuals were drawn to illustrate two components at the same time (on the x-axis and y-axis).

The group variable was not considered in the construction of PCAs but was added as an additional variable to describe the components. In this way, a one-factor analysis of variance was performed between each component and group variable (coordinates of the individuals) to identify whether the group was characteristic of the component, retained for a critical probability of the overall Fisher test < 0.05. In the significant case, Student t-tests were achieved to compare the average of the category with the general mean, in order to identify potential characteristic category of each component (i.e. potential distinction according to the LPS, LPS PE, LPS om-3 or LPS PE + om-3). The distinction of a group in a component was considered when the associated critical probabilities p-value < 0.05 (Student t-test). To visualize group distinction, barycenter of individuals in each category and their confidence ellipses were added on the plots.

Statistical analyses of oxylipins and fatty acid composition were performed using GraphPad Prism software (version 10.1.0, GraphPad Software, Boston, MA, USA). For fatty acid analysis, the 5 groups were compared using a 2-way ANOVA (with time and treatment as factors), followed by Dunnett’s post hoc test when appropriate. For oxylipin analysis, following PCA, the 4 groups treated with LPS were compared at each time point using a 1-way ANOVA followed by Tukey’s post hoc test when appropriate or a Kruskal-Wallis test followed by Dunn’s post hoc test when appropriate, when data were non-normal. To complete the oxylipin analysis, a heatmap based on the z-score calculated for each data point compared with the mean of the data group was generated using R software (R version 4.4.1, corrplot package version 0.92) for the three time points studied in the 5 groups.

Results

Principal component analysis (PCA) was performed at each time point of euthanasia on the 90 available biomarkers. At each time point, the effect of LPS treatment was evaluated comparing Control vs. LPS groups and the effect of supplementations was evaluated comparing non-supplemented vs. supplemented LPS groups. For each analysis, we will only describe the number of components that explain at least 50% of the total variance.

30 Min post-LPS: multivariate analysis

LPS-treated mice did not significantly differ from control mice after 30 min of exposure

Fig. 1
figure 1

Multivariate analysis of control and LPS groups at 30 min: individual map of PCA

Full size image

The first 3 components from the PCA analysis explained 54.2% of the total variance (first component, ‘dim 1’: 26.6%; second component, ‘dim 2’: 16.1%; third component, ‘dim 3’: 11.5%) (Fig. 1). No significant differences between LPS and Control group were revealed on these dimensions meaning that the profiles between Control and LPS groups cannot be distinguished.

Supplementations increased inflammatory and antioxidant responses in 30 min LPS-treated animals

Fig. 2
figure 2

Multivariate analysis of supplemented and non-supplemented LPS groups at 30 min: individual map of PCA

Full size image

The first 4 components derived from the PCA analysis explained 53.9% of the total variance (first component, ‘dim 1’: 27.2%; second component, ‘dim 2’: 10.4%; third component, ‘dim 3’: 9%; fourth component, ‘dim 4’: 7.3%) (Fig. 2). Table 2 shows the correlations between biomarkers and components. PE supplemented group was significantly different from the other groups on dim 2, for which it showed positive scores (p = 0.001). Specifically, dim 2 revealed a positive correlation with proinflammatory markers CCL2, TNFα, IL-1β, NFκB, CXCL10, CXCL9, microglial markers Clec7a, SOCS3, CD11c, CD206, ApoE and antioxidant markers GPx1 and Catalase. PE + om-3 supplemented group was significantly different from the other groups on dim 4, for which it showed negative scores (p = 0.001). Dim 4 showed a positive correlation with antioxidant marker GSS and C16-2OH, a hydroxy fatty acid and a negative correlation with PGD2, LTB4, PGE2, derived from AA and 10-HODE, derived from LA. No distinction between the supplemented groups were found on dimensions 1 and 3.

Table 2 Multivariate analysis of supplemented and non-supplemented LPS groups at 30 min
Full size table

2 H post-LPS: multivariate analysis

LPS modulated the oxylipin profile and increased proinflammatory and antioxidant markers after 2 h of exposure

Fig. 3
figure 3

Multivariate analysis of control and LPS groups at 2 h: individual map of PCA

Full size image

The first 4 components from the PCA analysis explained 56.9% of the total variance (first component, ‘dim 1’: 25.6%; second component, ‘dim 2’: 13.7%; third component, ‘dim 3’: 10%; fourth component, ‘dim 4’: 7.6%) (Fig. 3). Control and LPS groups significantly differed from each other on dim 1 and dim 2. In fact, LPS group exhibited negative scores on dim 1 and positive scores on dim 2 whereas it was the opposite for Control group (dim 1: p < 0.001 and dim 2: p = 0.016). Table 3 shows the correlation between biomarkers and components. Dim 1 was positively correlated with various LA- and AA-derived oxylipins and hydroxy fatty acids, as well as 12-LOX, an oxylipin biosynthesis enzyme and Clec7a, a microglial marker. Dim 1 was also negatively correlated with PGE2 and LTB4, derived from AA, COX-2, involved in oxylipin biosynthesis, microglial marker CD86 and proinflammatory cytokines and chemokines TNFα, IL-1β, CCL2, IL-6 (both in spleen and hippocampus), CXCL10 and CXCL9. Dim 2 revealed only positive correlation with oxylipins derived from AA, DHA and LA, pro-inflammatory factors NFκB, IL-6 (both in spleen and hippocampus), TNFα and IL-1β, microglial marker SOCS3, antioxidant markers Sod2, Catalase, GPx1, Sod1 and NRF2, and 15-LOX, an oxylipin biosynthesis enzyme. No significant differences between LPS group and Control group were found on dimensions 3 and 4.

Table 3 Multivariate analysis of control and LPS groups at 2 h
Full size table

PE + om-3 supplementation modulated oxylipin composition in 2 h LPS-treated animals

Fig. 4
figure 4

Multivariate analysis of supplemented and non-supplemented LPS groups at 2 h: individual map of PCA

Full size image

The first 5 components derived from the PCA analysis accounted for 55.1% of the total variance, with the breakdown as follows: the first component (dim 1) contributing 23.7%, the second component (dim 2) 10.9%, the third component (dim 3) 8%, the fourth component (dim 4) 6.7% and the fifth component (dim 5) 5.8% (Fig. 4). PE + om-3 group was significantly different from LPS group on dim 4, for which it showed positive scores (p < 0.001) whereas LPS group showed negative scores (p < 0.001). Table 4 shows the correlation between biomarkers and components. Dim 4 was positively correlated with LxA4, derived from AA and C10-3OH, a hydroxy fatty acid, and negatively correlated with LTB4, also derived from AA. No distinction between the LPS groups were found on dimensions 1, 2, 3 and 5.

Table 4 Multivariate analysis of supplemented and non-supplemented LPS groups at 2 h
Full size table

6 H post-LPS: multivariate analysis

LPS increased the proinflammatory response, decreased neuroprotection and modulated oxylipin composition after 6 h of exposure

Fig. 5
figure 5

Multivariate analysis of control and LPS groups at 6 h: individual map of PCA

Full size image

The first 2 components from the PCA analysis explained 54.2% of the total variance (first component, ‘dim 1’: 36.9%; second component, ‘dim 2’: 17.3%) (Fig. 5). Control group was significantly different from LPS group on dim 1 and dim 2. In fact, LPS group exhibited negative scores on dim 1 and positive scores on dim 2 whereas it was the opposite for Control group (dim 1: p = 0.039 and dim 2: p < 0.001). Table 5 shows the correlation between biomarkers and components. Dim 1 was positively correlated with various LA-, AA- and DHA-derived oxylipins and hydroxy fatty acids as well as neurotrophic factor BDNF and negatively correlated with LTB4, derived from AA, and IDO1. Dim 2 was positively correlated with PGE2 and PGD2, derived from AA, 3 hydroxy fatty acids, proinflammatory markers IL-1β, NFκB, IL-6 (both in spleen and hippocampus), CCL2, TNFα, IDO1, TLR4, CXCL9, CXCL10 and iNOS, microglial markers SOCS3, Arg1, CD86, CD11b and ApoE, oxidative markers NRF2, RAGE and GSS, oxylipin biosynthesis enzymes 15-LOX, 5-LOX, COX-2, and apoptosis marker Bax. Dim 2 was negatively correlated with microglial markers Trem2 and Glut5, oxylipin biosynthesis enzyme 12-LOX and neurogenesis marker DCX.

Table 5 Multivariate analysis of control and LPS groups at 6 h
Full size table

PE + om-3 supplementation induced a change in oxylipin profile in 6 h LPS-treated animals

Fig. 6
figure 6

Multivariate analysis of supplemented and non-supplemented LPS groups at 6 h: individual map of PCA

Full size image

The first 5 components derived from the PCA analysis explained 53.8% of the total variance. Specifically, the first component (dim 1) contributed 29.4%, the second component (dim 2) 10.2%, the third component (dim 3) 8.2% and the fourth component (dim 4) 6.2% (Fig. 6). Table 6 shows the correlation between biomarkers and components. LPS group showed negative scores on dim 1 (p = 0.001) whereas PE + om-3 supplemented group showed positive scores (p = 0.040). Dim 1 was positively correlated with oxylipins derived from LA, AA, EPA and DHA and hydroxy fatty acids as well as microglial markers Tspo and ApoE, neuroprotective receptor TrkA and antioxidant markers GPx1, Sod2 and Sod1. PE + om-3 group was significantly different from the other groups on dim 3, for which it revealed negative scores (p < 0.001). Dim 3 was positively correlated with PGD2, derived from AA, microglial markers CD68, Glut5, CD11b and ApoE and C18-3OH. No distinctions of supplemented groups were found on dim 2 and 4.

Table 6 Multivariate analysis of supplemented and non-supplemented LPS groups at 6 h
Full size table

Oxylipin analysis

Fig. 7
figure 7

Heatmap representation of oxylipin concentrations in LPS groups. Red indicates higher concentration and blue indicates lower concentration (n = 9–10/group)

Full size image

As the effect of supplementations seemed to be driven by a change in oxylipin profile, we investigated the impact of supplementations on the oxylipin concentrations derived from LA, AA, EPA and DHA at each time point. The generated heatmap demonstrated that gene regulation and oxylipin production were dependent on the time point and the group considered (Fig. 7). Interestingly, heatmap clearly highlighted the underexpression of oxylipins at 30 min post-LPS in the PE + om-3 group compared with the other groups and the overexpression of oxylipins at 6 h in the supplemented groups compared with the LPS and control groups.

After 30 min of exposure (Table 7), supplementations affected the amount of prostaglandins derived from AA: 8-iso-PGA2 (F(3,35) = 3.457, p = 0.027), PGD2 (F(3,35) = 3.463, p = 0.027), PGE2 (H(3) = 13.030, p = 0.005), PGF2α (H(3) = 13.870, p = 0.003), 15dPGJ2 (F(3,35) = 6.589, p = 0.001) and that of 18-HEPE (F(3,35) = 29.160, p < 0.001), derived from EPA. Indeed, PGF2α concentration was significantly higher in PE group than in LPS group (p = 0.006). PE + om-3 supplementation significantly increased PGD2, PGE2, PGF2α and decreased 15dPGJ2 compared to LPS group (PGD2: p = 0.049, PGE2: p = 0.040, PGF2α: p = 0.010, 15dPGJ2: p = 0.040). Moreover, PE + om-3 supplementation significantly increased PGD2 (p = 0.031) and PGE2 (p = 0.011) compared to om-3 group and decreased 8-iso-PGA2 compared to PE supplementation (p = 0.015) and 15dPGJ2 compared to all other groups (p = 0.001 vs. PE, p = 0.006 vs. om-3). Finally, om-3 and PE + om-3 supplementations significantly increased 18-HEPE compared to LPS and PE groups (p < 0.001 for all).

Table 7 Effect of supplementations on oxylipins concentrations in hippocampus 30 min post-LPS treatment
Full size table

After 2 h of exposure (Table 8), supplementations modulated the concentrations of LA-derived 9-oxo-ODE (F(3,35) = 6.195, p = 0.002) and 13-oxo-ODE (H(3) = 12.110, p = 0.007), as well as AA-derived 8,9-EET (F(3,35) = 4.129, p = 0.013), 5-HETE (F(3,34) = 6.513, p = 0.001), 8-HETE (F(3,35) = 5.740, p = 0.003), 12-HETE (H(3) = 11.560, p = 0.009), 15-HETE (F(3,35) = 4.026, p = 0.015), 5-oxo-ETE (F(3,35) = 8.448, p < 0.001), LTB4 (F(3,28) = 3.715, p = 0.023), PGD2 (F(3,35) = 3.380, p = 0.029) and PGE2 (H(3) = 8.090., p = 0.044). 18-HEPE (F(3,35) = 23.99, p < 0.001) and 14-HDHA (F(3,35) = 3.021, p = 0.043), derived respectively from EPA and DHA, were also affected. In fact, PE supplementation significantly decreased 14-HDHA (p = 0.038), 9-oxo-ODE (p = 0.016), 5-oxo-ETE (p = 0.002), 5-HETE (p = 0.003), 8-HETE (p = 0.009), 15-HETE (p = 0.040) and LTB4 (p = 0.033) compared to LPS group. Om-3 supplementation significantly decreased 5-oxo-ETE compared to LPS group (p = 0.052). PE + om-3 supplementation significantly decreased 9-oxo-ODE (p = 0.001), 13-oxo-ODE (p = 0.005), 8,9-EET (p = 0.008), 5-oxo-ETE (p < 0.001) 5-HETE (p = 0.003), 8-HETE (p = 0.004), 12-HETE (0.006), 15-HETE (p = 0.016) and LTB4 (p = 0.053) and tented to increase PGD2 (p = 0.061) compared to LPS group. Moreover, PE supplementation significantly increased PGE2 and PE + om-3 supplementation significantly increased PGD2 compared to om-3 group (p = 0.049 and p = 0.034, respectively). Finally, om-3 and PE + om-3 supplementations significantly increased 18-HEPE compared to LPS and PE groups (p < 0.001 for all).

Table 8 Effect of supplementations on oxylipins concentrations in hippocampus 2 h post-LPS treatment
Full size table

After 6 h of exposure (Table 9), supplementations significantly impacted the concentration of most oxylipins derived from LA: 9-HODE (F(3,36) = 5.133, p = 0.005), 13-HODE (F(3,36) = 4.213, p = 0.012), 9,10-DiHOME (F(3,36) = 4.511, p = 0.009), 12,13-DiHOME (F(3,36) = 9.159, p < 0.001), 9-oxo-ODE (F(3,36) = 6.698, p = 0.001), 13-oxo-ODE (F(3,36) = 5.257, p = 0.004), 9,10,13-TriHOME (F(3,36) = 4.767, p = 0.007), 9,12,13-TriHOME (H(3) = 15.570, p = 0.001). Most AA-derived oxylipins were also modulated by the supplementations: 5,6-EET (F(3,36) = 3.368, p = 0.029), 8,9-EET (F(3,36) = 10.74, p < 0.001), 11,12-EET (F(3,35) = 6.812, p = 0.001), 14,15-EET (H(3) = 12.73, p = 0.005), 5-HETE (F(3,36) = 6.736, p = 0.001), 8-HETE (F(3,36) = 6.363, p = 0.001), 15-HETE (F(3,36) = 4.274, p = 0.011), 5-oxo-ETE (F(3,36) = 9.646, p < 0.001), LTB4 (F(3,26) = 4.729, p = 0.009), LxA4 (H(3) = 11.20, p = 0.011), 8-iso-PGA2 (H(3) = 21.67, p < 0.001), PGD2 (F(3,36) = 6.050, p = 0.002), PGF2α (F(3,36) = 2.877, p = 0.049), 15dPGJ2 (F(3,36) = 10.21, p < 0.001), TXB2 (F(3,36) = 7.746, p < 0.001). The same results were found for EPA-derived 18-HEPE (H(3) = 27.41, p < 0.001) and DHA-derived 14-HDHA (F(3,36) = 5.365, p = 0.004) and 17-HDHA (F(3,36) = 6.470, p = 0.001). Indeed, compared to LPS group, all supplementations significantly increased 9,10-DiHOME (trend at p = 0.090 vs. PE, p = 0.007 vs. om-3, p = 0.048 vs. PE + om-3), 12,13-DiHOME (p = 0.003 vs. PE, p < 0.001 vs. om-3 and vs. PE + om-3), 9-oxo-ODE (p = 0.040 vs. PE, p = 0.012 vs. om-3, p < 0.001 vs. PE + om-3), 9,10,13-TriHOME (p = 0.028 vs. PE, p = 0.007 vs. om-3, trend at p = 0.065 vs. PE + om-3), 15dPGJ2 (p = 0.002 vs. PE, p = 0.012 vs. om-3, p < 0.001 vs. PE + om-3), TXB2 (p = 0.004 vs. PE, p < 0.001 vs. om-3, p = 0.005 vs. PE + om-3) and 8,9-EET (p = 0.050 vs. PE, p = 0.010 vs. om-3, p < 0.001 vs. PE + om-3). Compared to LPS group, PE supplementation induced a higher level of 9,12,13-TriHOME (p = 0.011), 5-HETE (p = 0.010), LTB4 (p = 0.006), LxA4 (p = 0.014) and 8-iso-PGA2 (p < 0.001). Om-3 supplementation increased the level of 9-HODE (p = 0.055), 13-HODE (p = 0.046), 13-oxo-ODE (p = 0.028), 9,12,13-TriHOME (p = 0.002), LxA4 (p = 0.040), 8-iso-PGA2 (p = 0.017) and tended to increase 11,12-EET (p = 0.060), 14,15-EET (p = 0.091) and PGF2α (p = 0.055) levels compared to LPS group. PE + om-3 supplementation increased the level of 9-HODE (p = 0.003), 13-HODE (p = 0.010), 13-oxo-ODE (p = 0.003), 11,12-EET (p < 0.001), 14,15-EET (p = 0.003), 5-HETE (p < 0.001), 8-HETE (p < 0.001), 15-HETE (p = 0.006), 8-iso-PGA2 (p < 0.001), 14-HDHA (p = 0.004) and 17-HDHA (p < 0.001) and decreased that of PGD2 (p = 0.005), compared to LPS group. Moreover, PE + om-3 supplementation increased the concentrations of 8,9-EET, 14-HDHA and 17-HDHA compared to PE group (p = 0.028, p = 0.030, p = 0.050, respectively) and decreased that of PGD2 compared to om-3 group (p = 0.004). In addition, PE + om-3 supplementation increased the concentrations of 5-oxo-ETE compared to the other groups (p < 0.001 vs. LPS, p = 0.012 vs. PE, p = 0.017 vs. om-3). Finally, 18-HEPE was also significantly higher in om-3 and PE + om-3 groups than in LPS and PE groups (p < 0.001 vs. LPS, om-3 vs. PE p = 0.006, PE + om-3 vs. PE p = 0.003).

Table 9 Effect of supplementations on oxylipins concentrations in hippocampus 6 h post-LPS treatment
Full size table

Fatty acid analysis

Fig. 8
figure 8

Effect of LPS and supplementations on fatty acid concentrations in hippocampus. (A) Total SFAs concentration. (B) Total MUFAs concentration. (CC) Total omega-6 + omega-3 concentration. (*p < 0.05, **p < 0.01, ****p < 0.0001 vs. LPS by 2-WAY ANOVA and Dunnett’s post-hoc test, n = 9–10/group). Data are presented as mean ± SEM

Full size image

We then analyzed the hippocampal composition of fatty acids that are precursors of oxylipin compounds (Fig. 8A-C). Time of exposure to LPS and treatment significantly impacted the proportions of total SFAs (time: F(2,128) = 3.522, p = 0.032; treatment: F(4,128) = 8.757, p < 0.001) and total omega-6 + omega-3 (om6 + om-3) (F(2,128) = 13.51, p < 0.001; treatment: F(4,128) = 19.97, p < 0.001). At each time point, the omega-6 and omega-3 proportions followed the same evolution as total om-6 + om-3 (data not shown). Moreover, an interaction between time and treatment was observed for the proportions of total SFAs (F(8,128) = 3.931, p < 0.001), total MUFAs (F(8,128) = 2.984, p = 0.004) and total om-6 + om-3 (F(8,128) = 9.801, p < 0.001). Indeed, compared to control group, LPS significantly decreased the proportion of total SFAs (p < 0.001) and increased that of total om-6 + om-3 (p < 0.001), 30 min after exposure but not later. Concerning the effect of the supplementations, after 30 min of exposure, all of them significantly increased total SFAs proportion (p < 0.001 for all) and decreased total om-6 + om-3 proportions compared to LPS (p < 0.001 for all). After 2 h of exposure, no effects of the supplementations were observed. After 6 h of exposure, all the supplementation significantly increased total MUFAs (PE: p = 0.011, om-3: p = 0.021, PE + om-3: p = 0.007) and decreased total om-6 + om-3 proportion (p < 0.001 for all) compared to LPS group.

Discussion

Neuroinflammation, when it becomes excessive, unregulated and prolonged, can switch from a protective to a detrimental mechanism [1]. Several mechanisms contribute to the resolution of inflammation, including the inhibition of proinflammatory markers, the production of anti-inflammatory cytokines, and the generation of oxylipins [10, 43]. This study provides evidence that a combination of polyphenols and carotenoids from plant extracts (Memophenol™ and a patented saffron extract) and omega-3 from fish induced a different resolution of inflammation through a switch in hippocampal oxylipin profile after LPS injection.

Acute LPS injection is a model widely used to study neuroinflammation, as it induced the release of pro- and anti-inflammatory factors [44]. LPS is associated with memory deficits and neurodegenerative disorders such as Alzheimer’s or Parkinson’s diseases in humans [45,46,47] As expected, we observed, at 2 h and 6 h, an increase in the expression of proinflammatory molecules such as NFκB, IL-6, IL-1β, TNFα, CCL2, CXCL9 and CXCL10, as well as activated microglial markers, of proinflammatory and proresolutive phenotypes, in LPS-treated animals, as previously reported by Chataigner et al.., and Hasegawa-Ishii et al. [30, 48]. 2 h after injection, an increase in antioxidant enzymes expression in the hippocampus was observed in LPS-treated animals, as already shown by Essadek et al.., suggesting an increase in the body’s defense system against cellular free radicals [49]. This was accompanied by an increase in the levels of proinflammatory oxylipins PGE2, LTB4, 12-HETE and 15-HETE, as well as proresolutive DHA derivates 14-HDHA and 17-HDHA. Same changes in oxylipin profile 2 h after LPS treatment were previously reported in mice hippocampus [30]. 6 h after treatment, LPS group showed a lower expression of Glut5, a marker of microglial homeostasis, and Trem2, which was positively linked to the inhibition of neuroinflammation in BV2 microglia [50, 51]. These changes were associated with higher concentrations of proinflammatory mediators PGE2, PGD2 and LTB4 as well as higher gene expression of their biosynthesis enzymes COX-2, 5-LOX and 15-LOX, but not for 12-LOX, which was less expressed in LPS group than in control group. Prostaglandins and leukotrienes production begins within a few minutes, at the start of the inflammatory response, and peaks after a few hours [52]. This production is followed by a switch in oxylipins class to lipoxins and other proresolving mediators production, lasting from a few hours to a few days [53]. In our study, the higher levels of proinflammatory mediators in LPS group were observed at 6 h. A study quantifying the plasma level of some oxylipins between the basal condition and 96 h post-LPS already described a peak at 3–6 h, depending on the derivate [54]. The expression of the neuroprotective and neurogenesis markers BDNF and DCX also decreased, as already described in the literature but after 2 days of LPS injections [55]. LPS treatment was also associated with changes in fatty acid composition. Specifically, total SFAs levels decreased after 30 min of treatment but not later, and this decrease was offset by the increase in total om-6 + om-3 levels at the same time. Similarly, a previous study showed that 24 h of LPS treatment on BV2 microglial murine cells induced an increase in total om-6 and total om-3, associated with an increase in total SFAs and a decrease in total MUFAs [56]. We hypothesized that in our study, the decrease in SFAs was due to their oxidation to yield energy to fight against inflammation, with the increase in om-6 + om-3 as a consequence since the results are expressed in % of total fatty acids.

In this study, the mix of plant extracts (Memophenol™ and a patented saffron extract) and omega-3 had anti-inflammatory and proresolutive effects, mostly driven by changes in oxylipins. Indeed, at 30 min post-LPS, the mix increased the amount of proresolutive EPA-derived 18-HEPE. 18-HEPE has been demonstrated to suppress LPS-induced TNFα expression in macrophage cell culture [57]. Surprisingly, while most of the studies described lower AA-oxylipins levels after omega-3 supplementation, PE + om-3 also increased some prostaglandins, such as PGD2, PGE2 and PGF2α, and decreased others, such as 15dPGJ2 and 8-iso-PGA2, which is produced by the oxidation of AA [12, 58, 59]. As the proinflammatory mediator production begins earlier with the mix supplementation, this may suggest that the time course of inflammation resolution is different than that in the LPS group. Indeed, at 2 h post-LPS, PE + om-3 supplementation induced lower concentrations of 9-oxo-ODE, 13-oxo-ODE, 5-HETE, 8-HETE, 12-HETE, 15-HETE, 5-oxo-ETE and LTB4, mainly described as proinflammatory [60]. Higher concentrations of the proresolving mediators LxA4, reduced in AD patients, and 18-HEPE, were also observed, indicating a switch to a proresolutive profile [16, 53]. Finally, at 6 h post-LPS, mice supplemented with the mix showed a completely different oxylipin profile, characterized by an increase in almost all oxylipins detected and a decrease in PGD2. The proresolutive properties of both 14-HDHA and 17-HDHA have been extensively documented; notably they suppress the inflammatory response within the central nervous system (CNS) and in microglial cells [12, 61, 62]. Moreover, oxidized linoleic acid metabolites, such as 13-HODE, 9-HODE, 9,10-DiHOME, 9,10,13-TriHOME and 12,13-DiHOME, which were highly overexpressed by the mix, regulate neuronal morphogenesis in vitro [63]. 13-HODE and 18-HEPE are largely described as potent anti-inflammatory molecules. Indeed, 13-HODE is a ligand for PPARα and PPARγ, whose activation reduces the expression of TNFα, IL-1β and IL-6 [64]. 18-HEPE is a metabolic precursor of resolvin E involved in the resolution of inflammation and is significantly increased in the plasma of subjects supplemented with blueberries [29]. These changes were accompanied by an increase in the gene expression of antioxidant enzymes and a decrease in the gene expression of microglial markers CD68 and CD11b, both described as proinflammatory [65, 66]. All these modifications of oxylipin concentrations could be interpreted as a modification of inflammation resolution kinetic. Furthermore, all supplementations counteracted the LPS effect on fatty acid composition. Indeed, they increased total SFAs and decreased total om-6 + om-3 levels at 30 min. At 6 h post-LPS, all supplementations also increased total MUFAs levels and decreased total om-6 + om-3 levels. The lower proportion of om-6 + om-3 may be explained by their oxidation to produce oxylipins or by the increase in SFA level.

Moreover, the combination of plant extracts and omega-3 exhibited a greater effect than each nutrient alone. In fact, PE supplementation only showed antioxidant effect 30 min after LPS treatment, associated with a proinflammatory effect. After 2 h and 6 h of treatment, PE supplementation also induced changes in the oxylipin profile, but to a lesser extent. Indeed, it decreased the levels of some LA- and AA-derived oxylipins at 2 h and increased the levels of others at 6 h. Similarly, om-3 supplementation increased only the 18-HEPE concentration after 30 min of LPS. This effect was also observed after 2 h and 6 h, accompanied by other changes in the oxylipin profile, mostly by an increase in LA-derived oxylipins at 6 h. Om-3 supplementation did increase DHA-derived oxylipin levels compared to LPS group, in accordance with previous studies [12, 30]. As PE supplementation had no effect on either 14- or 17-HDHA, metabolite precursors of maresin and resolvin D that are pro-resolving molecules, but PE + om-3 supplementation did, these results suggest that plant extracts and omega-3 exerted a potentiating effect.

To our knowledge, this study is the first to provide interesting initial evidence on the potential beneficial effects of a mix containing grapes and blueberries polyphenols, saffron carotenoids and fish omega-3 on neuroinflammation and oxylipin profile. However, it has some limitations. Indeed, this study focused on a relatively short-term period with limited time points (6 h maximum post-LPS injection), further studies on chronic inflammatory models, are necessary to conclude on the long-term effects of the nutrients. Behavioral assessments could also be added to understand how these changes impacted functional changes in mice. Moreover, even if many biomarkers have been measured, the exact mechanisms by which the supplementations exert their effects remain unclear. Further analysis of the pathways involved could provide a more complete and detailed overview of key players in this biological process. In addition, the use of only male mice limits the generalizability of our findings, especially given sex-related differences in inflammatory responses [67]. Further studies have to be conducted in female mice.

Conclusion

This study presented evidence of the additional effect of a mix containing PE (Memophenol™ and a patented saffron extract) and omega-3 on LPS-induced neuroinflammation, as it showed modulations of some genes and oxylipins not found with the two other supplementations. The mix supplementation, for only 18 consecutive days, induced a shift in the oxylipin profile, demonstrating a different resolution of inflammation than without supplementation or with nutrients alone. Thus, our results confirmed the interest of a multinutrient approach to limit alterations induced by age, such as chronic low-grade inflammation.