High Glucose Shifts the Oxylipin Profiles in the Astrocytes towards Pro-Inflammatory States

Hyperglycemia is associated with several complications in the brain, which are also characterized by inflammatory conditions. Astrocytes are responsible for glucose metabolism in the brain and are also important participants of inflammatory responses. Oxylipins are lipid mediators, derived from the metabolism of polyunsaturated fatty acids (PUFAs) and are generally considered to be a link between metabolic and inflammatory processes. High glucose exposure causes astrocyte dysregulation, but its effects on the metabolism of oxylipins are relatively unknown and therefore, constituted the focus of our work. We used normal glucose (NG, 5.5 mM) vs. high glucose (HG, 25 mM) feeding media in primary rat astrocytes-enriched cultures and measured the extracellular release of oxylipins (UPLC-MS/MS) in response to lipopolysaccharide (LPS). The sensitivity of HG and NG growing astrocytes in oxylipin synthesis for various serum concentrations was also tested. Our data reveal shifts towards pro-inflammatory states in HG non-stimulated cells: an increase in the amounts of free PUFAs, including arachidonic (AA), docosahexaenoic (DHA) and eicosapentaenoic (EPA) acids, and cyclooxygenase (COX) mediated metabolites. Astrocytes cultivated in HG showed a tolerance to the LPS, and an imbalance between inflammatory cytokine (IL-6) and oxylipins release. These results suggest a regulation of COX-mediated oxylipin synthesis in astrocytes as a potential new target in treating brain impairment associated with hyperglycemia.


Introduction
Oxylipins represent a superfamily of bioactive lipid mediators, derived from the metabolism of polyunsaturated fatty acids (PUFAs), through a complex network of biochemical reactions [1][2][3]. Oxylipins have fundamental roles in a diverse set of homeostatic and inflammatory processes, and therefore, their synthesis is under intensive consideration. The conversion of PUFAs into oxylipins occurs via four major pathways, involving cyclooxygenases (COX), lipoxygenases (LOX), and cytochrome P450 monooxygenases (CYP450), and non-enzymatically [1, 4,5]. When cells are stimulated, it is important that different PUFAs, and even some of their derivatives, are released from membrane phospholipids and metabolized into oxylipins simultaneously [6,7]. Therefore, there is a competition between pathways for their substrates [1,4]. Essential for oxylipin synthesis, PUFAs are subdivided into omega-3 (n-3) and omega-6 (n-6) groups; docosahexaenoic acid (DHA) and eicosapentaenoic acid (EPA) belong to n-3, and linoleic acid (LA) and arachidonic acid (AA) belong to the n-6 group. An imbalance between n-3 and n-6 PUFAs is generally attributed to chronic inflammation [8,9], although the system of oxylipins synthesis is currently viewed as being more complicated [3,10]. For example, the oxylipins of the n-6 group include both pro-inflammatory compounds and resolution substances, which are responsible for restoring the system after the pro-inflammatory stimulus has been applied [2,3,10]. The of HG on the oxylipin release in rat primary astrocytes and their ability to respond to a lipopolysaccharide (LPS) stimulation.

Changes in the Oxylipin Profiles in Astrocytes, Cultured in Normal and High Glucose
To compare oxylipin profiles, primary rat astrocytes were cultured in medium with normal (5.5 mM) or high (25 mM) glucose concentrations. In the experiment, cells were stimulated with lipopolysaccharide (LPS, 100 ng/mL) for 4 h. Oxylipins in supernatants were extracted and measured using ultra-performance liquid chromatography-tandem mass spectrometry (UPLC-MS/MS). We obtained oxylipin profiles for four groups: (1) normal (NG) naive cells; (2) high (HG) glucose naive cells; (3) NG cells, stimulated by LPS; (4) HG cells, stimulated by LPS (Figure 1). For details of data, see Supplementary Table S1.
Detected oxylipins can be subdivided into three groups: (1) free PUFAs ( Figure 1A); (2) derivatives of LOX-pathway ( Figure 1B); and (3) derivatives of COX-pathway ( Figure 1C). The PUFAs group contains three acids: EPA, DHA, and AA. Although cell cultivation in HG increases the total content of PUFAs nearly twofold, LPS stimulation does not lead to an increase in the content of PUFAs in the medium ( Figure 1A). The ratio between substances is maintained (Table S1).
Based on the previous data [3,4], we classified the following metabolites as the LOX metabolic pathway: 12-and 5-hydroxyeicosatetraenoic acids (12-HETE, 5-HETE), 9-hydroxyoctadecadienoic acid (9-HODE), 8-, 14-, and 17-hydroxydocosahexaenoic acids (8-HDoHE, 14-HDoHE, and 17-HDoHE). Cell cultivation in HG or LPS stimulation did not modulate the content of these pathway substances ( Figure 1B). We detected the following derivatives of AA: 12-hydroxyheptadecatrenoic acid (12-HHT), prostaglandins (PGs) 6-keto-PGF2a, PGA2, PGE2, PGD2, PGF2a, thromboxane B2 (TXB2), 11-HETE, and also 13-HDoHE, which is a derivative of DHA. All of these oxylipins are attributed to the COX pathway of oxylipin synthesis [4,39,40]. Cell cultivation in HG increases the total content of COX derivatives ( Figure 1C). The sensitivity of COX-mediated synthesis for LPS stimulation decreases for the HG group ( Figure 1C). Indeed, we observed a nearly twofold increase in the concentration of COX derivatives in LPS versus naive groups for HG cells and near fourfold for NG cells ( Figure 1C). Thus, although many more metabolites of the COX pathway are released at high glucose, the response to the stimulation of the pro-inflammatory stimulus became less pronounced. COX-mediated derivatives contained both pro-inflammatory and anti-inflammatory or resolution relative substances [2,3]. A detailed analysis of this group of substances reveals variations in their sensitivity for HG and LPS stimulations ( Figure 1D).
For the characterization of oxylipin profiles, we compared pie plots, calculated for the COX-mediated metabolites, released from naive or LPS-stimulated astrocytes ( Figure 2). The plot considers compounds regarding their concentration (%), where total COX-mediated substances concentration is taken as 100%. There is a significant change in the profile during incubation in HG, which is also notable following LPS stimulation ( Figure 2). In comparison to NG/HG naive cells, there is a decrease in the proportion of a substance among others for 12-HHT, 6-keto-PGF2a, PGA2, PGD2, PGF2a, TXB2, 11-HETE, and 13-HDoHE ( Figure 2). Only the proportion of PGE2 significantly increased ( Figure 2). Since PGE2 is usually referred to as a marker of the inflammatory process [39], it can be assumed that there is a shift in HG towards the inflammatory process even in naive cells. For LPS-stimulated oxylipin profiles, the increase in PGE2 proportion is also notable ( Figure 2). Note that a proportion of 6-keto-PGF2a decreased in LPS-stimulated oxylipin profiles for HG compared to NG ( Figure 2). normal (5.5 mM) or high (25 mM) glucose concentrations. In the experiment, cells were stimulated with lipopolysaccharide (LPS, 100 ng/mL) for 4 h. Oxylipins in supernatants were extracted and measured using ultra-performance liquid chromatography-tandem mass spectrometry (UPLC-MS/MS). We obtained oxylipin profiles for four groups: (1) normal (NG) naive cells; (2) high (HG) glucose naive cells; (3) NG cells, stimulated by LPS; (4) HG cells, stimulated by LPS (Figure 1). For details of data, see Supplementary Table S1.

Effect of Serum on Oxylipin Profiles in Astrocytes, Cultured at Normal and High Glucose
To further characterize oxylipin synthesis in NG and HG astrocytes, we decreased the amount of serum in the experimental culture medium and tested 10% (normal concentration for cell cultivation), 5%, and 1% of the serum. This was carried out for several reasons. Firstly, oxylipins bind with extracellular proteins [11,41], and these processes may influence oxylipin profiles and cellular responses [42][43][44]. Secondly, in our experimental protocol, HG as a stressor has a long-term application, therefore, we added a set of experiments with an acute stressor. Serum as a stressor for astrocytes was reported previously [45].
We obtained that in naive cells, in both the NG and HG groups, the serum affects the medium/cell oxylipins equilibrium (the less serum, the fewer oxylipins) ( Figure S1). This effect was observed for TXB2, 11-HETE, 13-HdoHE, 12-HHT, 6-keto-PGF1a, PGA2, and PGE2. In the case of PGD2, PGE2, and PGF2a, the concentration of the compound, detected in the extracellular medium, does not depend on the serum concentration ( Figure S1). Therefore, for most oxylipins, the simple assumption of the dependence of oxylipins' concentrations on their binding with the serum proteins, works well. However, the nondependence for three PGs reveals certain additional effects of low serum concentration. To test these assumptions, we tested how serum affects the LPS-mediated cellular response in astrocytes.
We obtained unexpected results when estimating the influence of serum concentration on the LPS-stimulated release of metabolites ( Figure 3). During the first stage, we compared the total concentrations of metabolites, divided into PUFA, LOX-mediated, and COXmediated metabolites. We obtained that the summarized level of PUFAs depends on the serum (in 10% FBS more than in 1% FBS) and the glucose (more in HG than in NG, with the greatest difference observed in 10% serum). Simultaneously, the stimulation of LPS cells does not significantly affect the concentration of the summarized PUFAs ( Figure 3A). Cell cultivation in HG or LPS stimulation, with different serum concentrations, did not modulate the content of LOX-derived metabolites ( Figure 3B). Different serum concentrations do not affect the total amount of COX metabolites, while HG increases the total level of these metabolites in all tested serum concentrations ( Figure 3C).
A detailed analysis of the COX-mediated group of substances reveals variations in their sensitivity for serum, HG, and LPS stimulations. We presented the data as a "fold to a control", which means that the synthesis of a compound in the naive cells at an appropriate serum concentration was taken as 1. A decrease in serum concentration leads to an increase in TXB2, 11-HETE, 13-HDoHE, PGA2, and PGD2. The release of 12-HHT, 6-keto-PGF2a, PGE2, and PGF2a was not dependent on serum concentration. Therefore, upon stimulation of the oxylipins synthesis by LPS, the serum acts as an additional stimulus, and a change in serum concentration affects the amount of emitted oxylipins and their ratio.

Effect of Serum Level and Different Glucose Concentrations in the Culture Medium on the IL-6 Release
For further characterization of the inflammatory response in astrocytes cultivated in NG and HG, we compared the release of IL-6 in LPS-stimulated cells in the presence of 10% and 1% of serum. It is known that the synthesis and release of IL-6 were at a detectable level only in the presence of LPS, not in the naive cells [46,47], therefore it can be used as a marker of astrocyte activation upon stimulation.
There is no difference in IL-6 release between cells, cultivated in NG and HG ( Figure 4A). A decrease in serum concentration resulted in a nearly twofold increase in LPS-stimulated IL-6 release, although there was no difference between the NG and HG groups ( Figure 4A). Thus, culturing cells in HG does not change the release of the pro-inflammatory cytokine, i.e., the work of the ensemble of the cell response to the pro-inflammatory stimulus is impaired (the ratio between the released cytokines and oxylipins).
Glial fibrillary acidic protein (GFAP) formed part of the cytoskeleton of mature astrocytes and other glial cells and was used as a marker of injured astrocytes [14]. To characterize the differences between the NG and HG groups, we measured the level of GFAP expression ( Figure 4B). Expression levels of GFAP decreased in the HG group compared to the NG group ( Figure 4B). Both groups did not respond to 4 h LPS stimulation by altering the level of GFAP expression ( Figure 4B).  There is no difference in IL-6 release between cells, cultivated in NG and HG ( Figure  4A). A decrease in serum concentration resulted in a nearly twofold increase in LPS-stimulated IL-6 release, although there was no difference between the NG and HG groups ( Figure 4A). Thus, culturing cells in HG does not change the release of the pro-inflammatory cytokine, i.e., the work of the ensemble of the cell response to the pro-inflammatory stimulus is impaired (the ratio between the released cytokines and oxylipins). Glial fibrillary acidic protein (GFAP) formed part of the cytoskeleton of mature astrocytes and other glial cells and was used as a marker of injured astrocytes [14]. To characterize the differences between the NG and HG groups, we measured the level of GFAP expression ( Figure 4B). Expression levels of GFAP decreased in the HG group compared to the NG group ( Figure 4B). Both groups did not respond to 4 h LPS stimulation by altering the level of GFAP expression ( Figure 4B).

Discussion
Our data reveal that astrocyte cultivation in the high concentration of glucose (25 mM) demonstrated impairment in the innate immune system statements. Compared with astrocytes cultivated in a normal concentration of glucose (5.5 mM), cells from an HG cultivation medium exhibit the following properties: (1) an inflamed state of the naive cells measured by oxylipins levels; (2) an imbalance between the pro-inflammatory cytokine

Discussion
Our data reveal that astrocyte cultivation in the high concentration of glucose (25 mM) demonstrated impairment in the innate immune system statements. Compared with astrocytes cultivated in a normal concentration of glucose (5.5 mM), cells from an HG cultivation medium exhibit the following properties: (1) an inflamed state of the naive cells measured by oxylipins levels; (2) an imbalance between the pro-inflammatory cytokine (IL-6) and oxylipins' release; and (3) a tolerance in the oxylipins' synthesis response to the pro-inflammatory stimulus, LPS.
It has been known for more than a decade that metabolism and immunity are closely linked, and lipids play an important role in this link [48]; however, molecular mechanisms are still largely unknown. Primary cell cultures are extensively used to facilitate the study of the molecular mechanisms involved in various physiological and pathophysiological processes. Astrocyte-enriched cultures are usually used for glial cell response studies [49]. Although Dulbecco's Modified Eagle's Standard Medium (DMEM) contains 25 mM or 5.5 mM glucose, glucose concentration in the culture medium used is often not indicated. Supplementation with 5.5 mM glucose in culture media approximates normal blood glucose levels in vivo, glucose levels approaching 10 mM are close to pre-diabetic levels, whereas 25 mM is considered to be a diabetic-like condition. Relatively recently, it became clear that the glucose concentration differences can affect intracellular signaling responses in various cell types (for example, [50]), including glial cells [51,52].
Various glucose applications have been tested for astrocytes: short-term (study of the glucose effect for several hours) [36,37], medium-term (48 h) [34,37], and long-term (two or more weeks of cultivation from cellular preparation) [35,51,52]. The data in the short-term glucose stimulation demonstrate the involvement of innate immune signaling pathways in the response, such as the activation of NF-kB and p38 MAPK, and proinflammatory cytokines' release [36][37][38]. Such glucose effects may induce cellular responses of adaptation. The data relating to experiments with a middle-term application of high glucose concentration are in accordance with this assumption. Indeed, HG enhanced the LPS-stimulated pro-inflammatory cytokine release, increased p38, decreased JNK MAPK activities, and increased PGE2 release [34]. The long-term cultivation of cells in an HG medium resulted in the impairment of gap junctional communication [35], metabolic efficiency and capacity [53], activated the AMP-activated protein kinase (AMPK) signaling pathway, increased glycogen storage and ATP, and decreased proliferation [51]. One study claims that long-term HG cultivation reduces the ability of astrocytes to respond to LPS, but this phenomenon has not been studied in detail [52]. Interestingly, a decrease in glucose concentration below the normal level (2 mM) also induces the pro-inflammatory responses on astrocytes, manifested by increased pro-inflammatory cytokines' release [54]. It is known that changes in the glucose transporter's expression and functions correlate with diseases in the brain [55], which are also attributed to diseases with neuroinflammatory complications. Our data and those of others indicate that the relationship between glucose metabolism and inflammatory responses is related to maintaining the homeostasis system in the brain, and its changes are manifested in various pathologies.
It is important to note that it is currently difficult to assess the intracellular mechanisms of glucose modulation of the cellular response to inflammatory stimuli, firstly because of the differences in the experimental protocols, and secondly because of the multiple reactivities of cells to extracellular glucose concentrations. Differing glucose addition times and media change protocols can affect the readings. For example, we found that GFAP levels depend only on glucose concentration and not on LPS stimulation. The absence of changes after stimulation can be explained by the short time of action of LPS (4 h); the data are consistent with others [56]. A change in GFAP level was found 24 h or more after stimulation with LPS [37,57]. From another point of view, it should be borne in mind that the protocol in Wang's work with colleges [37] included the stage of processing cells without glucose in the medium, which may be the reason for the difference in the obtained data.
The synthesis of oxylipins at the cellular level and the mechanism of their action via special receptors is a complex system. Oxylipins are generated by complex enzymatic reactions that begin with PUFA release from membrane phospholipids by phospholipase A2 (PLA2) [7,40]. It is known that various PLA2s are responsible for the release of AA and DHA in astrocytes [58]. Some of the oxylipins can be included in the composition of phospholipids and released by them [6]. Therefore, it is difficult to directly relate solely the oxylipins released into the extracellular medium, with their enzymatic synthesis. The scarcity of data concerning cellular oxylipin synthesis does not identify the connection between PUFAs and oxylipin profiles in the naive cells, cultivated in HG and NG with intracellular processes of oxylipin synthesis. For COX-mediated branches of metabolism, we detected, besides PGs and Tx, the following derivatives of AA: 12-HHT, 11-HETE, and also 13-HDoHE, which is a derivative of DHA. Note that DHA works as an inhibitor for COX-1 [59] and a substrate for COX-2 [60]. In addition, PGF2a may be the result of several enzyme activities: PGF synthase (from PGH2), PGD211 keto-reductase (from PGD2), and PGE2 9 keto-reductase (from PGE2) [61]. 12-HHT may result from the Tx synthase pathway or non-enzymatically from PGH2. Therefore, its synthesis is solely dependent on COX activities [62]. An elucidation of the reasons for changes in oxylipin profiles at the level of intracellular oxylipin metabolism is the subject of further research. Our data reveal the tune regulation of oxylipin synthesis and its significant dependence on glucose metabolism in astrocytes.
The observed increase in the synthesis of prostaglandins at high glucose levels is in good agreement with the data relating to a decrease in the proliferation of astrocytes [51] and the data on the summation of the effects of prostaglandins in suppressing proliferation [63]. Our data support the view of astrocytes as key cells at the crossroad of metabolic and inflammatory pathways.

Primary Astrocyte Cell Culture
Astrocyte-enriched cultures were obtained from newborn rats of both sexes as previously reported [34,58]. More than 95% of the cells were positive for the astrocyte marker glial fibrillary acidic protein in these cultures, and only <2% were positive for a microgliaspecific marker. In brief, brains from newborn rats were triturated by the use of nylon meshes of 250 and 136 mL pore width, in consecutive order. Dissociated cells were plated into culture flasks. The cells were subsequently cultured in DMEM (1 g/L D-glucose or 4.5 g/L D-glucose, 10% bovine fetal serum (FBS), 50 units/mL streptomycin, 50 µg/mL penicillin) at 37 • C, with 10% CO 2 . After 5 days of cultivation in DMEM, the culture medium was replaced with a fresh medium, and flasks were placed on a shaker at 200 RPM for 4 h to dissociate microglial cells. The culture medium was changed to a fresh one. After 2 days, the monolayer of astrocytes was trypsinized and plated into six-well plates and maintained for 2 days in DMEM. After this, the medium was changed, and the cells were used for experiments. At each change of medium, the cells were cultured only with the glucose concentration upon their preparation. This was performed to eliminate the possible reaction of cells to changes in glucose concentration. After the cell experiments, the supernatant was collected and stored with 0.005% BHT at −80 • C for further analysis.

UPLC-MS/MS Conditions and Sample Preparation
After the cell experiments, the supernatant was collected and stored at −80 • C for further analysis. The cell-free culture media were taken for the solid-phase lipid extraction (Oasis ® PRIME HLB cartridge (60 mg, 3 cc)). The cartridge was washed with 2 mL of 15% methanol containing 0.1% formic acid, and the lipids were sequentially eluted with 500 µL of anhydrous methanol and 500 µL of acetonitrile. The resulting samples were mixed, concentrated by the evaporation of the solvent under a gentle stream of nitrogen, and stored at −80 • C. For the identification of lipid mediators, the respective lipid extracts were analyzed using an 8040 series UPLC-MS/MS mass spectrometer (Shimadzu, Kyoto, Japan) in multiple-reaction monitoring mode at a unit mass resolution for both the precursor and product ions as described previously [63]. Briefly, the lipid compounds were separated by reverse-phase UPLC (injection volume 20 µL) using Phenomenex C8 column (2.1 mm × 150 mm × 2.6 µm) with the flow rate of 0.4 mL/min. The elution was performed using an acetonitrile gradient in 0.1% (v/v) formic acid. The target PUFAs and oxylipins were identified and quantified by comparing their MS, MS/MS, and UPLC (retention times, peak areas) data with the data obtained for deuterated internal standard compounds using Lipid Mediator Version 2 software (Shimadzu, Kyoto, Japan).

Western Blot Analysis
Western blot analysis was performed as described earlier [47]. In brief, cells were lysed in a radio immunoprecipitation assay (RIPA) buffer. The protein concentration was determined by the standard Bradford assay. After blocking, the PVDF membranes were subsequently subjected to a Phosphate-Buffered Saline with Tween 20 0.05%, with a respective primary antibody at 4 • C overnight with constant rotation. Secondary species-specific antibodies were applied at the concentration of 1:10,000 for 1 h at room temperature. The signals were detected using the Pierce ECL Plus Western Blotting Substrate (Thermo Scientific, Waltham, MA, USA). Mild stripping buffer (Abcam, Cambridge, UK) was used to re-probe western blot membranes. Densitometry was carried out on three different experiments. The band intensity was measured using a ChemiDoc™ XRS+ gel documentation system (Bio-Rad, Hercules, CA, USA). Densitometry was performed with ImageJ (1.51 s) software (National Institute of Health, Bethesda, MD, USA) using β-tubulin as the loading control.

Determination of IL-6 Enzyme-Linked Immunoassay
ELISA for IL-6 was performed using the BD OptEIA, (cat No.:550319; BD Biosciences, San Jose, CA, USA) antibody kits following the manufacturer's instructions. Optical density was measured using a microplate reader Synergy H4 (BioTek, Winooski, VT, USA).

Data Analysis and Statistics
The data are expressed as mean ± SEM. The normality of data sets was assessed using the Shapiro-Wilk test. The data were subjected to a two-way ANOVA, followed by Bonferroni's post hoc test, in order to determine the statistical significance, p < 0.05 was considered statistically significant. All of the experiments were repeated at least three times.

Conclusions
Our data reveal shifts towards pro-inflammatory states in HG, non-stimulated, naive cells and also a more pro-inflammatory mixture of LPS-stimulated oxylipin synthesis in HG in comparison to NG. Astrocytes cultivated in HG showed a tolerance to LPS. The data suppose a regulation of COX-mediated oxylipin synthesis as a potential new target in the treatment of brain impairment associated with hyperglycemia.