Background
Although multiple mechanisms underlie the pathophysiology of ischemic stroke, there is increasing evidence that inflammation accounts for its progression [
1,
2]. A robust inflammatory reaction characterized by peripheral leukocyte influx into the cerebral parenchyma and the activation of endogenous microglia follows focal cerebral ischemia [
3‐
7]. Reactive oxygen species (ROS) and reactive nitrogen species (RNS) play pivotal roles in neuronal damage. ROS and/or RNS are, at least in part, produced by cerebral parenchymal cells, including neurons and glial cells, as well as peripheral inflammatory cells such as leukocytes. The origin of ROS appears to be from activated NADPH oxidase (NOX), and ROS, in turn, reacts with nitric oxide (NO) produced by activated inducible nitric oxide synthase (iNOS), leading to peroxynitrite generation, which is a highly toxic reactive species that can induce lipid peroxidation and protein oxidative damage [
8]. Thus, the intrinsic protective mechanism that eliminates ROS may play an important role in countering post-stroke inflammation.
Toll-like receptors (TLRs) mediate innate immune response, and recent findings suggest that they also mediate post-stroke inflammatory responses [
9]. There are at least 11 TLR proteins in mammals [
10], and TLR4 is implicated in the pathogenesis of stroke-related inflammation. TLR4 is predominantly expressed in both microglia and astroglia, and also in neurons [
11]. Although lipopolysaccharide (LPS), a component of bacterial cell walls, is a well-known classical ligand of TLR4, extracellular matrix and dysfunctional mitochondria released from damaged cells have been reported to be endogenous ligands of TLR4 [
12]. TLR4 stimulation of microglia plays a harmful role in stroke through downstream signaling pathways: myeloid differentiation protein 88–nuclear factor κB (NFκB), mitogen-activated protein kinase (MAPK), also called extracellular signal-regulated kinase 1/2, and Janus kinases, which are signal transducers and activators of transcription. In particular, the activation of NFκB induces NOX and iNOS, which enhance ROS and NO production by microglia.
Astroglia express TLR4 even in the human brain, and the stimulation of TLR4 by LPS in cultured astroglia induces pro-inflammatory responses [
13]. In contrast, LPS may also induce an astroglial anti-inflammatory, protective function via the pentose phosphate pathway (PPP) [
14]. PPP is a shunt pathway of glycolysis and produces NADPH, which recycles the oxidized from of glutathione (GSH) back to the reduced form, leading to the removal of ROS by GSH peroxidase. Thus, PPP activity is an index of endogenous neuroprotective function. In fact, astroglial PPP activity is much higher than that of neurons [
15], and PPP activation plays an important role in countering oxygen stress. The activation of astroglial TLR4 by LPS activates astroglial PPP through NO [
14,
16,
17]. The exact mechanism by which NO activates PPP remains to be elucidated.
The rate-limiting enzyme of PPP, glucose 6-phosphate dehydrogenase (G6PDH), is regulated both allosterically and transcriptionally. Under stress conditions, the master regulator of the anti-stress response, the Kelch-like ECH-associated protein 1 (Keap1)/nuclear factor-erythroid-2-related factor 2 (Nrf2) system induces various enzymes via transcriptional regulation mediated by Nrf2. In fact, G6PDH also harbors an antioxidant response element (ARE) to which Nrf2 binds and initiates the transcription of G6PDH, leading to PPP activation [
18]. Nrf2 is bound to its adaptor protein Keap1 and is constantly degraded by the ubiquitin–proteasome system in the cytosol. However, several signals can affect the cysteine residues of Keap1 to induce a conformational change for releasing Nrf2. Free Nrf2 translocates into the nucleus and initiates transcription. Numerous reports show that both ROS and/or NO can induce Keap1 conformational change and Nrf2 activation [
19‐
22]. Thus, TLR4-mediated production of ROS, NO, or both by astroglia is a potential mediator.
Another important issue that should be clarified is the cellular origin of ROS or NO. There is no doubt that LPS induces ROS and NO production in microglia. However, whether astroglial ROS and NO production are induced by LPS remains to be established. Most studies describing the astroglial response to LPS have been based on cell culture experiments [
23]. Usually, an astroglial culture contains a small (<10 %) population of microglia. If LPS acts on microglia, even on <10 % of the total cells, microglia can produce sufficient ROS and NO in response to LPS.
The present study used a conventional secondary astroglial culture prepared from pups of Sprague–Dawley rats. To eliminate microglial cells, the conventional astroglial culture was treated with cytosine β-
d-arabinofuranoside hydrochloride (Ara-C), followed by
l-leucine methyl ester (LME) [
24,
25]. This procedure completely eliminates Iba1-positive cells. We used both microglia-present and microglia-absent astroglial cultures and evaluated PPP activity, as well as ROS and NO production in response to LPS.
Methods
Animals
Timed-pregnant Sprague–Dawley rats were obtained from Japan SLC, Inc. (Hamamatsu, Japan). Male C57BL/6 (wild-type (WT)) mice (Japan SLC, Inc.) and Nrf2 gene knockout (KO) mice (on C57BL/6 background, originally generated by Dr. Masayuki Yamamoto) were used in this study. Nrf2-KO mice were genotyped by PCR amplification of genomic DNA extracted from tail snips. All animal procedures were performed in accordance with The Animal Experimentation Guidelines of Keio University School of Medicine and were approved by the Committee on Animal Care and Use, Keio University (08097-[15]).
Chemicals
Chemicals and materials were obtained from the following sources: Insta-Fluor Plus and Hyamine hydroxide 10-X from Perkin-Elmer Life Sciences (Boston, MA, USA); d-[1-14C]glucose (specific activity, 2.035 GBq/mmol) and d-[6-14C]glucose (specific activity, 2.035 GBq/mmol) from American Radiolabeled Chemicals, Inc. (St. Louis, MO, USA); normal goat serum from Jackson ImmunoResearch (West Grove, PA, USA); rabbit Iba1 antibody from Wako Pure Chemical Industries (Osaka, Japan); mouse monoclonal anti-glial fibrillary acidic protein (GFAP) from Sigma (St. Louis, MO, USA); goat polyclonal antibody against TLR4 from Santa Cruz Biochemistry (Delaware, CA, USA); 1,4-diamino-2,3-dicyano-1,4-bis[2-aminophenylthio] butadiene (U0126), a selective inhibitor of MAPK/ERK kinase1/2 (MEK1/2) from Cell Signaling Technology (Danvers, MA, USA); mouse monoclonal antibody against anti-β actin from Abcam (Cambridge, UK); rabbit anti-Nrf2 (H-300) polyclonal antibody, rhodamine-conjugated goat anti-mouse (for GFAP), anti-rabbit (for Nrf2) IgG antibody, and fluorescein-isothiocyanate (FITC)-conjugated goat anti-rabbit (for Iba1) antibody from Santa Cruz Biochemistry; Dulbecco’s modified Eagle media (DMEM) with or without glucose, penicillin, and streptomycin from Life Technologies (Grand Island, NY, USA); defined fetal bovine from HyClone Laboratories (Logan, UT, USA); 2′,7′-dichlorodihydrofluorescein diacetate (H2DCFDA) and monochlorobimane (MCB) from Molecular Probe Inc. (Eugene, OR, USA); diaminofluorescein-FM diacetate (DAF-FM DA) from Sekisui Medical (Chuo-ku, Tokyo, Japan); and all other chemicals were obtained from Sigma (St. Louis, MO, USA).
Preparation of cells
Primary astroglial cultures were prepared from the rat or mouse cerebral cortices 24–48 h after birth [
26]. Dissociated cells from the cerebral cortices (2.5 × 10
5 cells/mL for rat and 5.0 × 10
5 cells/mL for mouse) were plated (15 mL/flask) in uncoated 75-cm
2 culture flasks (Sumitomo Bakelite, Tokyo, Japan) and cultured in a glucose-containing medium (final concentration, 12 mmol/L of
d-glucose) comprising DMEM with 10 % (
v/
v) fetal bovine serum, penicillin (100 U/mL), and streptomycin (100 μg/mL) at 37 °C in humidified air containing 21 % O
2 and 7 % CO
2 (day 0). The culture medium was changed every 2 days until the cultures reached confluence. On day 11, the adherent cells were treated with trypsin–EDTA solution, suspended in a fresh high (12 mmol/L)-glucose medium, and placed in uncoated 12-well culture plates (0.8 mL/well, respectively; Nalge Nunc, Rochester, NY, USA) or in 25-cm
2 culture flasks (5 mL/flask; Nalge Nunc). For immunohistochemical staining, the cells were plated (2 mL/dish) on 35-mm glass-bottomed dishes (Matsunami Glass, Industry, Osaka, Japan) precoated with poly-
l-lysine. Some cells were grown on glass-bottomed 24-well culture plates (EZView culture plate LB, Iwaki, Tokyo, Japan) for fluorometric assay. From the day after subculturing, the cells were cultured in DMEM, fetal bovine serum, penicillin, and streptomycin and a final
d-glucose concentration of 12 mmol/L for 10 days. The culture medium was changed twice a week, and the cells were used once they reached confluence (on day 21 in vitro). Some secondary astroglial cells were treated with Ara-C followed by LME to eliminate microglia.
Primary neuronal cultures were prepared from the striatum of fetal rats on embryonic day 16, as previously described [
26]. Mechanically dissociated cells were placed in 12-well culture plates (0.8 mL/well, respectively) or 25-cm
2 culture flasks (5 mL/flask) coated with poly-
l-lysine (5 μg/mL). For the neuronal cultures, viable cells (1.5 × 10
6 cells/mL) that excluded trypan blue were placed in the cultures, and Ara-C (10 μmol/L) was added 72 h later to induce mitotic arrest of the astroglia. The cells were cultured in a glucose medium (final concentration, 12 mmol/L
d-glucose) at 37 °C in humidified air containing 21 % O
2 and 7 % CO
2. In vitro assays were performed using cultures that were 7 or 8 days old. The nutrient medium remained untouched until the experiments were initiated.
Microglial cells were obtained by collecting floating or loosely adherent cells in primary astroglial culture and were placed in 12-well culture plates or glass-bottomed dishes precoated with poly-
l-lysine [
27].
Depletion of microglia from astroglial cultures
Usually, secondary astroglial cultures contain only small amounts (<10 % of total cells) of microglia, as previously described [
27,
28]. To eliminate all possible metabolic effects of microglial contamination, the treatment with a combination of Ara-C and LME was applied according to the method described by Pont-Lezica et al. [
25]. Briefly, 10 μmol/L (final concentration in nutrient medium) of Ara-C was added to the astroglial culture 3 days prior to the assay. After 48 h, cells were further incubated with 50 mmol/L of LME (pH adjusted to ~7.4) for 60 min. Cells were then washed with phosphate-buffered saline (PBS) without Ca
2+ and Mg
2+ and further cultured with a regular nutrient medium for 24 h. This treatment destroyed microglia by osmotic effects and yielded almost 100 % pure astroglial cell culture. The exact mechanism by which LME eliminates microglia remains to be elucidated. It has been speculated that LME is internalized by microglia, wherein it causes lysosomal disruption and subsequent apoptosis. In fact, macrophages and microglia are characteristically rich in lysosomes, causing them to be particularly vulnerable upon exposure to LME treatment [
29,
30].
Experimental protocol
To assess the effects of LPS on astroglial PPP activation, ROS/NO production, and immunohistochemistry, the nutrient medium (12 mmol/L) was removed and fresh nutrient medium containing LPS (0.01, 0.1, or 1 μg/mL) was added and kept in a CO
2 incubator for 15 h. For neuronal cultures, a medium change per se damages cells via glutamate toxicity [
31], LPS was directly added to the culture media of cells to obtain final concentrations of 0.01, 0.1, or 1 μg/mL, as described above. When astroglial cells were exposed to 1-isothiocyanato-(4R,S)-(methylsulfinyl)butane (sulforaphane), an Nrf2 activator [
32‐
34] or S-nitroso-
N-acetyl-
dl-penicillamine (SNAP), an NO donor [
35], the nutrient medium was replaced with a fresh medium containing the drug or vehicle (dimethyl sulfoxide (DMSO)), and cells were incubated for 15 h. When the effects of sulforaphane (10 μmol/L), U0126 (10 μmol/L),
l-buthionine-sulfoximine (BSO; 50 μmol/L)—an inhibitor of GSH synthesis [
36]—on LPS-induced ROS/NO production were assessed, cells were pre-incubated with these drugs for 12 h before LPS treatment and further incubated for 12 h with LPS, followed by the measurement of ROS and NO.
We next assessed the effects of hypoxia because it is also well recognized that hypoxic conditions per se shift the cellular metabolism from oxidative to glycolytic in various cells. However, whether hypoxia enhances PPP flux has not been addressed. An early study revealed that chronic hypoxia induced PPP activity in C6 glioma cells [
37], but the mechanism that conveys the hypoxic signal for PPP activation is undetermined. In a recent study [
38], hypoxia and reperfusion induced PPP activation in astroglia. Although the underlying mechanism was not described, the reperfusion-induced massive production of ROS in association with NO might exert a strong influence on PPP activation. Moreover, the effect of LPS or LPS in combination with hypoxia on astroglial PPP activity has not been reported. In the present study, we examined the effect of LPS stimulation in combination with hypoxia on PPP flux in astroglia. To assess the effect of hypoxia on PPP activation, we placed cells for 12 or 24 h in a hypoxic chamber (multi-gas incubator APM-30DR, ASTEC, Fukuoka-ken, Japan) at 37 °C in humidified air containing 1 % O
2 and 7 % CO
2. We assessed the combined effects of LPS (0.01 μg/mL) and hypoxia (1 % O
2) for 15 h in some neuronal and astroglial cells.
Measurement of the rates of d-[1-14C]glucose and d-[6-14C]glucose oxidation to 14CO2
The rate of [
14C]glucose oxidation to
14CO
2 was measured using a modification of a previously described method [
39] after cells cultured in 25-cm
2 culture flasks were washed twice with PBS containing no glucose. The assay solutions consisted of Dulbecco’s balanced salt solution (DBSS) containing (mmol/L) 110 NaCl, 5.4 KCl, 1.8 CaCl
2, 0.8 MgSO
4, 0.9 NaH
2PO
4, and 44 NaHCO
3, in addition to 2 mmol/L of
d-glucose labeled with 1.0 μL/mL of
d-[1-
14C]glucose or
d-[6-
14C]glucose (original concentrations, 3.7 MBq/mL), were added, and the flasks were capped with rubber stoppers containing a center well and incubated at 37 °C for 60 min. The
14CO
2 produced was trapped by a cotton ball placed in the center well containing 100 μL of Hyamine hydroxide 10-X. The reactions were terminated by the injection of 250 μL of perchloric acid (2 mol/L) through the rubber stopper, and the flasks were kept at 4 °C overnight to trap the
14CO
2. The center wells were then transferred to 20-mL glass scintillation counter vials, and 500 μL of ethanol and 10 mL of Insta-Fluor Plus were added. The
14C contents of the vials were evaluated in a liquid scintillation counter. The assay solutions consisted of 2.5 mL of DBSS containing 2 mmol/L
d-glucose; these solutions were labeled by the addition of 1.0 μL/mL of
d-[1-
14C]glucose or
d-[6-
14C]glucose (original concentrations, 3.7 MBq/mL).
Waniewski and Martin [
40] reported that substantial
14C counts were obtained from a flask without cells. Accordingly, the
14C counts obtained from a flask without cells in which the reaction had been stopped at 60 min were treated as background values and were subtracted from the measured counts. The cell carpets remaining in the incubation flasks after the removal of the reaction mixtures were then digested with 5 mL of 0.1 M NaOH, and their protein contents were determined using the bicinchoninic acid (BCA) assay kit (Thermo Scientific Pierce, Rockford, IL, USA). The rates of total glucose oxidation (pmol glucose/μg protein/60 min) based on the conversion from [
14C]glucose to
14CO
2 over 60 min were measured.
Measurement of PPP activity in cultured astroglia and neurons
PPP activity was measured using a modification of the method described by Hothersall et al. [
41]. Briefly, cells were incubated with glucose with tracer doses of [1-
14C]glucose or [6-
14C]glucose for 60 min, and the PPP rate calculated as the difference between the
14CO
2 derived from [1-
14C]glucose (metabolized by both PPP and the tricarboxylic acid [TCA] cycle,) and that derived from [6-
14C]glucose (metabolized by only the TCA cycle) was taken as an indicator of PPP activity.
Measurement of ROS and NO production
The production of ROS, mainly H
2O
2, and NO in cells was assessed by H
2DCFDA [
42] and DAF-FM DA [
43] using semiquantitative fluorometric measurement. Immediately prior to the assay, the nutrient medium was removed and the cells were washed twice with PBS without glucose. Then, DBSS containing 10 μmol/L of H
2DCFDA or DAF-FM DA dissolved in DMSO (final volume of DMSO 0.1 %) supplemented with 2 mmol/L glucose were added, and the cells were incubated at 37 °C in humidified air with 7 % CO
2 for 30 min. After the H
2DCFDA or DAF-FM DA were loaded for 30 min, the cells were washed twice again with PBS without glucose and DBSS containing 2 mmol/L
d-glucose. The cells were further incubated for 60 min, and the fluorescence level indicating intracellular ROS or NO production was measured at 0 and 60 min using a fluorescent microplate reader (Infinite 200 PRO; Tecan Japan, Kanagawa, Japan) with excitation at 485 nm and emission at 530 nm. As the fluorescent signals linearly increased for up to 60 min (data not shown), the results were expressed as the percent increase in fluorescent signal at 60 min compared with that at 0 min.
Measurement of total glutathione content using a fluorometric method
The intracellular content of total glutathione in astroglia was assessed using MCB which forms an adduct with glutathione via an enzymatic reaction catalyzed by glutathione transferases [
44]. Cells were incubated with MCB (50 μmol/l) for 30 min. The intracellular formation of the GSH–MCB adduct was assessed using a microplate reader with excitation at 360 nm and emission at 465 nm. The results were expressed as the fluorescence signal standardized by cellular protein content.
Immunohistochemistry
The translocation of Nrf2 from the cytosol to the nucleus was assessed using immunohistochemistry. Astroglial cells grown in a glass-bottomed 35-mm dish were fixed with 4 % paraformaldehyde for 10 min on ice. The cells were permeabilized with 0.5 % Triton X-100 for 10 min at room temperature, followed by post-fixation with 4 % paraformaldehyde for 5 min on ice. For immunostaining, the cells were washed twice with PBS containing MgCl2 and CaCl2, and nonspecific IgG binding sites were blocked by the incubation of cells in PBS containing 3 % bovine serum albumin and 3 % normal goat serum for 30 min at room temperature. Cells were incubated with primary antibodies (anti-GFAP [1:200], anti-Iba1 [1:200], and anti-Nrf2 polyclonal antibody [1:100]) for 2 h at room temperature. They were then incubated with the secondary antibodies (rhodamine-conjugated goat anti-rabbit IgG antibody or FITC-conjugated goat anti-mouse IgG antibody) with 4′,6-diamino-2-phenylindole (DAPI) for nuclear staining for 1 h. The cells were examined using a laser confocal microscopy system (Leica TCS SP5; Leica Microsystems, Exton, PA, USA).
Western blotting
Cells were homogenized in an ice-cold lysis buffer; radioimmunoprecipitation assay buffer, 0.5 % sodium dodecyl sulfate, ~0.1 M phenylmethanesulfonyl fluoride, and protease inhibitor cocktail. The concentration of total protein in the supernatant was determined using the BCA protein assay kit (Thermo Scientific Pierce, Rockford, IL, USA). The proteins were separated using 10 % sodium dodecyl sulfate polyacrylamide gel electrophoresis and electrophoretically transferred to polyvinylidene difluoride membranes (Millipore Corporation, Bedford, MA, USA). After treatment for 1 h with a blocking solution containing 5 % nonfat dry milk, the membrane was incubated with anti-TLR4 antibody (1:200), and anti-β actin antibody (1:5000). After washing, the membranes were reacted with the second bodies, the membrane was incubated with horse-radish peroxidase-conjugated anti-goat (Santa Cruz), or anti-mouse antibody (Jackson ImmunoResearch) and the immunoreactive bands were visualized by enhanced chemiluminescence and detected with a lumino-image analyzer (Las-4000; Fujifilm, Tokyo, Japan).
Evaluation of Nrf2 activation by quantitative reverse transcription PCR (qRT-PCR) analysis for mRNA of heme oxygenase-1 (HO-1)
Total RNA from cultured cells was isolated using RNeasy mini kit (Qiagen, Valencia, CA) with a DNase (Qiagen, Valencia, CA) treatment. First-stranded cDNA was synthesized using PrimeScript RT reagent Kit (TaKaRa) according to the manufacturer’s instructions. qRT-PCR was performed using predesigned gene-specific TaqMan Gene Expression Assays (Applied Biosystems) with the StepOnePlus real-time PCR system (Applied Biosystems). Rat HO-1 TaqMan probe and primers (cat# Rn00561387_m1) and rat GAPDH TaqMan probe and primers (cat# Rn01775763_g1) were from Applied Biosystems. The housekeeping gene glyceraldehyde 3-phosphate dehydrogenase (GAPDH) was used for normalization. The fold difference between levels of transcripts was calculated using the ΔΔCT method.
The PPP flux and other metabolic pathway of glucose were analyzed using CE/ESI MS (Additional file
1).
Statistical analyses
Statistical comparisons among the values obtained for each group were performed using grouped t tests or one-way analysis of variance followed by the Dunnett test for multiple comparisons. A P value of <0.05 was considered statistically significant.
Discussion
Inflammatory responses by neural cells (neurons and glial cells) as well as blood cells have been implicated in the underlying mechanisms of ischemic cell damage [
46‐
53]. TLRs, in particular TLR4, play important roles in the initiation and progression of inflammatory response by intrinsic brain cells, given that brain ischemia induces neuronal release or the increased production of endogenous ligands of TLR4 (damage-associated molecular patterns such as high-mobility group box 1 and peroxiredoxin) in a relatively early stage of an ischemic event, i.e., within 24 h [
9,
54,
55]. There is accumulating evidence that not only immune cells in the brain parenchyma, such as microglia, but also astroglia and neurons harbor TLR4 [
56‐
58]. It is likely that microglia respond to these endogenous TLR4 ligands by producing inflammatory cytokines and ROS/NO, leading to further neuronal damage [
59]. However, the full role of TLR4 in non-immune cells has not been elucidated. In the present study, we confirmed the presence of TLR4 in astroglia by Western blotting using an astroglial culture devoid of microglia. However, the role of astroglial TLR4 per se does not appear relevant.
The roles of astroglia in ischemic stroke are disputed. Astroglial TLR4 may play either a detrimental or beneficial role or both [
60]. Gorina et al. reported that TLR4 stimulation by LPS even at a low concentration (0.01 μg/mL) induced the release of inflammatory cytokines from astroglia, suggestive of a harmful role [
13]. For identifying the role of TLR4 in astroglia, cultured cells are useful tools [
61‐
64]. However, usually microglial cells are also present and may affect the results obtained from astroglial cultures [
65‐
68]. In the present study, we obtained a pure astroglial culture using Ara-C/LME-treatment. Although we did not detect cytokine release using this culture, ROS/NO production was not altered by the addition of LPS to the pure astroglial culture, whereas it showed drastic increases in the microglia-containing astroglial culture.
In contrast, Bolaños et al. reported the astroglial activation of PPP flux as a cytoprotective response induced by LPS (1 μg/mL) through NO [
16]. PPP is a cytoprotective metabolic pathway of glucose, given that PPP activation induces NADPH production, which is necessary to recycle the oxidized from of GSH back to the reduced form, leading to removal of ROS by GSH peroxidase [
69‐
71]. Importantly, in ischemic stroke, ischemia–reperfusion enhances ROS production in the mitochondria of neurons, leading to further neuronal damage [
72,
73]. Thus, PPP activation after ischemia–reperfusion could be an important player in preventing cellular damage during ischemic stroke. Furthermore, they found that NO has a key role in activating PPP in astroglia in response to LPS stimulation [
16]. The present study partly supported these beneficial effects of LPS on astroglia. We identified two important findings using two different cultures of astroglia with or without microglia. First, the origin of NO and ROS production in response to LPS stimulation was coexisting microglia. NO elimination by MAPK inhibition eliminated increases in PPP in astroglia, supporting the PPP-activating role of NO. Second, this NO-induced astroglial PPP activation was induced via the Keap1/Nrf2 system in astroglia. Interestingly, MAPK does not seem to be involved in the basal production of NO in microglia. In contrast, basal production of ROS in microglia decreased by MAPK inhibition, indicating different roles of MAPK in microglial production of NO and ROS.
Hypoxia per se induced PPP activation in astroglia by a completely different mechanism. Hypoxia could potentially activate PPP flux, given that the PPP is a branch pathway of glycolysis that is well known to be activated under hypoxic conditions [
74,
75]. However, the effects of hypoxia on PPP flux of astroglial cells have been extensively examined [
75,
76]. Instead, inflammatory cells such as macrophages alter their metabolic character to increase glycolysis in response to inflammation, which is associated with hypoxic environments. With respect to neurons, oxidative metabolism of glucose is required to produce sufficient ATP to maintain neuronal action potentials [
77,
78]. In contrast, astroglia energy metabolism is more dependent on glycolysis [
78‐
80]. We have reported that astroglial glycolysis and its branch pathway, the PPP, are important for protection of neurons against oxygen stress [
45]. In the present study, we found that hypoxia-induced activation of the PPP in astroglia is mediated by a different pathway from that mediating NO-induced activation of the PPP and that these two pathways act in an additive manner.
Astroglial cultures always contain a small but appreciable number of microglia [
81‐
83]. The present study showed that a normal astroglial culture contains as much as 10 % of Iba1-positive cells. Elimination of microglia from astroglia is important for distinguishing the astroglial response to TLR4 stimulation. Ara-C combined with LME eliminates microglia effectively [
24,
25]. The mechanism by which LME destroys microglia has not been fully established. The lysosomal enzyme, which is abundant in microglia, is involved [
29,
30]. In the present study, we confirmed that this treatment succeeded in completely eliminating microglia and that LPS-induced but not hypoxia-induced PPP activation occurred only in the presence of microglia. Interestingly, however, TLR4 expression in pure astroglia devoid of microglia was unaltered, indicating that astroglia per se express TLR4. Thus, LPS stimulation could possibly induce astroglial response through TLR4 activation, whether pro- or anti-inflammatory. We speculated that astroglial PPP activity is high enough to mask ROS production in response to LPS. In fact, LPS did induce increases in ROS production in astroglia in the absence of microglia after GSH depletion. We do not know why LPS did not induce NO production in astroglia.
PPP is a cytoprotective metabolic pathway of glucose, given that PPP activation induces the NADPH production that is necessary to recycle the oxidized form of GSH back to the reduced form, leading to the removal of ROS by GSH peroxidase [
84]. The rate-limiting enzyme of PPP, G6PDH, is transcriptionally regulated: the transcriptional factor Nrf2 binds to the ARE located upstream of the G6PDH gene [
18,
85]. Nrf2 is released from its adaptor protein Keap1 under stress imposed by ROS and RNS [
19]. Thus, ROS or NO can elicit Nrf2 activation, leading to PPP activation. Bolaños et al. indicated that NO induces astroglial PPP activation via mitochondrial inhibition and resulting activation of AMP-activated protein kinase (AMPK) [
16]. The results of the present study supported it, since the TCA flux as assessed by [6-
14C]glucose oxidation decreased by NO. However, increases in PPP flux did not seem to be explained merely by a decreased flux to the TCA cycle. In the present study, we confirmed that NO per se induced Nrf2 activation and increases in PPP flux, and after the inhibition of NO production from microglia by U0126, LPS failed to induce PPP activation, even though ROS production was maintained. NO reportedly activates the Keap1/Nrf2 system by S-nitrosylation of the Keap1 protein, facilitating the release of Nrf2 [
21]. ROS may also act as an activator of the Keap1/Nrf2 system through the modification of thiol residues of Keap1 [
86‐
89]. The present study, however, did not support a major role of ROS in activating the Keap1/Nrf2 system.
Another possibility is astroglial TLR4-mediated response via MAPK-associated phosphorylation of Nrf2. The mechanisms by which Nrf2 is activated are multiple and the phosphorylation of Nrf2 at serine 40 is one of them [
90‐
93]. Various kinases may phosphorylate Nrf2 [
20]. However, our results showed that nuclear translocation of Nrf2 was induced only when microglia were also present. Thus, the direct involvement of MAPK-induced phosphorylation of Nrf2 in astroglia is unlikely.
TLR4 activation is an important mechanism in ischemic stroke, given that ischemia-induced neuronal damage triggers the release of endogenous ligands for TLR4 [
9,
53,
94‐
96]. In the present study, we also examined the effects of hypoxia per se, given that hypoxia induces marked glycolysis. If glycolysis is activated, PPP, a branch pathway of glycolysis, is simultaneously activated. In fact, PPP in astroglia was activated irrespective of the presence of microglia. LPS and hypoxia showed additive effects on PPP activation in astroglia containing microglia, indicating that hypoxia and TLR4 activation act independently. Hypoxia induced the activation of both glycolysis and PPP. Similarly, LPS induced PPP activation as well as glycolysis. In a recent study, hypoxia induced PPP activation coupled with glycolytic activation [
97]. In this context, PPP activation is merely a consequence of glycolytic activation induced by hypoxia. Whether or not LPS-induced activation of glycolysis is dependent on the Nrf2 system is unclear from the results of the present study. These pathways are probably independent, given that LPS-induced increases in PPP flux were preserved under hypoxia. However, in brain ischemia, primary hypoxia causes cellular damage, and endogenous TLR4 ligands are released from the damaged cells. It is reasonable to speculate that hypoxia and TLR4 activation work independently to protect the living body against oxygen stress, given that ROS exerts deleterious effects in the progression of neuronal damage, especially when reperfusion of the occluded artery occurs. The protective mechanism of astroglia may play important roles, and its enhancement may lead to a novel therapeutic strategy for acute ischemic stroke.
Finally, we treated astroglia with sulforaphane to induce Nrf2 activation as evidenced by HO-1 expression. By the pretreatment of astroglia with sulforaphane, which had been confirmed to activate the PPP, ROS and NO production were completely blocked. Thus, the activation of PPP by the enhancement of Nrf2 in astroglia may lead to a novel therapeutic strategy for stroke.
Competing interests
The authors declare that they have no competing interests.
Authors’ contributions
ST designed experiments. ST, TI, KM, KM, YI, TH, and MK performed experiments. ST, TA, MS, and NS analyzed data. ST and TI wrote the manuscript. All authors read and approved the final manuscript.