Regulation of Superoxide by BAP31 through Its Effect on p22phox and Keap1/Nrf2/HO-1 Signaling Pathway in Microglia

Reactive oxygen species (ROS) production by activation of microglia is considered to be a major cause of neuronal dysfunction, which can lead to damage and death through direct oxidative damage to neuronal macromolecules or derangement of neuronal redox signaling circuits. BAP31, an integral ER membrane protein, has been defined as a regulatory molecule in the CNS. Our latest studies have found that BAP31 deficiency leads to activation of microglia. In this study, we discovered that BAP31 deficiency upregulated LPS-induced superoxide anion production in BV2 cells and mice by upregulating the expression level of p22phox and by inhibiting the activation of Nrf2-HO-1 signaling. Knockdown of p22phox/keap1 or use of an NADPH oxidase inhibitor (apocynin) reversed the production of superoxide anion and inflammatory cytokines, which then reduced neuronal damage and death in vitro and in vivo. These results suggest that BAP31 deficiency contributes to microglia-related superoxide anion production and neuroinflammation through p22phox and keap1. Furthermore, the excess superoxide anion cooperated with inflammatory cytokines to induce the damage and death of neurons. Thus, we determined that BAP31 is an important regulator in superoxide anion production and neuroinflammation, and the downstream regulators or agonists of BAP31 could therefore be considered as potential therapeutic targets in microglial-related superoxide anion production and neuroinflammation.


Introduction
Neurodegenerative diseases including Alzheimer's disease (AD), Parkinson's disease (PD), and multiple sclerosis (MS) are characterized by oxidative damage, chronic neuroinflammation, neuronal degeneration, and death in specific regions of the central nervous system (CNS) [1][2][3]. Reactive oxygen species (ROS), including the superoxide anion (O 2 •-), hydrogen peroxide (H 2 O 2 ), and hydroxyl radical (OH -), are the natural by-products of aerobic metabolism. O 2 •-, which is the precursor for most other ROS, can be catalyzed into H 2 O 2 by superoxide dismutases (SODs) and further reduced to the hydroxyl radical or water by peroxidases. ROS are wellknown stress signaling molecules in cells, and can be increased dramatically by environmental stress and disease [4]. The formation of superoxide anion and hydrogen peroxide is one of several common mechanisms that are described as triggers of neurodegeneration, with persistently unquenched levels of superoxide anion and hydrogen peroxide ultimately leading to oxidative damage to the lipids, proteins, and DNA essential for biological homeostatic states. Since the effects of oxidative stress are widespread and the brain has low antioxidant capacities, neurons are particularly vulnerable to oxidative damage induced by excess superoxide anion. In particular, superoxide anion production by microglia is considered to be a major cause of neuronal dysfunction, damage, and death through direct oxidative damage to neuronal macromolecules or derangement of neuronal redox signaling circuits, and the ROS-amplified proinflammatory response in microglia also drives neuropathology [5,6].
The generation of superoxide anion and hydrogen peroxide is regulated by specific enzymes, such as NADPH oxidase (NOX). NOX is a multisubunit enzyme complex; under resting conditions, the different subunits of this complex are localized in the cytosol (p40 phox , p47 phox , and p67 phox ) and cellular membrane (p22 phox and gp91 phox ), while upon stimulation, the catalytically active complex is assembled in the plasma membrane [7,8], resulting in superoxide anion and hydrogen peroxide production.
Nuclear factor erythroid 2-related factor 2 (Nrf2), a key regulator of endogenous defensive systems against oxidative stress, is produced by active microglia induced by oxidative stress in the brain [9]. Kelch-like ECH-associated protein 1 (keap1) is responsible for the cytosolic sequestration of Nrf2 under physiological conditions. Nrf2 is constitutively expressed in the cytoplasm and translocated into the nucleus under oxidative injury condition [10]. Heme oxygenase-1 (HO-1), a 32 kDa cytoprotective enzyme with strong immunomodulatory and anti-inflammatory properties, is positively regulated by Nrf2 [11]. Recent studies indicate that activation of Nrf2 signaling could be a popular strategy to prevent superoxide anion and inflammation-mediated neuronal toxicity [12,13].
Superoxide anion and hydrogen peroxide produced by either microglia or the surrounding environment are currently considered not only to impact neurons but also to modulate microglial activity. They act as both a signaling molecule and a mediator of inflammation. LPS induces microglia to produce superoxide anion and hydrogen peroxide and several proinflammatory molecules such as TNFα, IL-1β, inducible nitric-oxide synthase, prostaglandin E2, and monocyte chemotactic protein-1 (MCP-1) via the NOX pathway.
B cell receptor-associated protein 31 (BAP31) is an endoplasmic reticulum (ER) membrane protein. Recently, BAP31 was found to be a regulatory molecule for immunity in the central nervous system [14]. BAP31 deficiency accelerates the formation of amyloid-β plaques in APP/PS1 mice [15] and exacerbates the activation of microglia and the death of neurons induced by LPS [16].
In this study, we investigated whether BAP31 influenced the superoxide anion production. Our findings implied that BAP31 deficiency accentuated LPS-induced superoxide anion and hydrogen peroxide production by both increasing the superoxide anion production by upregulating p22 phox expression and inhibiting superoxide anion scavenging through inhibiting the keap1/Nrf2/HO-1 signaling pathways. Knockdown of p22 phox /keap1 or use of an inhibitor of NOX reversed these functions and protected neurons, indicating that BAP31 may be a key regulatory molecule in the nervous system by alleviating superoxide anion production in microglial cells.

Primary Microglia Cell
Culture. Primary microglial cells were prepared from 1-to 3-day-old newborn pups (BAP31 fl/fl and LysM-Cre-BAP31 fl/fl mice) of either sex as described [16]. Briefly, brains were dissected, and the meninges were removed. Then, the cortices were digested using trypsin and through a 70 μm cell strainer; then, cell suspensions were incubated in 25 cm 2 flasks pretreated with poly-L-lysine. After 4-7 days, astrocytes were recovered and microglia were generated by the addition of DMEM medium containing 25% of L929 conditioned medium. Three to four days later, the confluent mixed glia cells were subjected to shaking at 37°C at 100 rpm for an hour. Microglial cells were resuspended in DMEM/F12 medium containing 25% L929 conditioned medium for experiments. The microglial cells 2 Oxidative Medicine and Cellular Longevity were incubated with LPS (100 ng/ml) for 12 h after having been pretreated with or without apocynin (1 mM) [17] for 1 h, and then superoxide anion production was measured by dying with NBT and DHE. (1) Nitroblue Tetrazolium (NBT) Reduction Assay. The NBT assay is simple, sensitive, and quantitative and can be used to determine the amounts of intracellular O 2 •produced by a wide variety of cells [18]. The quantitation of O 2 •production in BV2 cells and primary microglia cells was performed using a nitroblue tetrazolium reduction assay. Cells were seeded at a density of 2 × 10 4 cells/well in 12-well culture plates for 24 h. Cells were treated with LPS (100 ng/ml) for 12 h; then, cells were incubated with nitroblue tetrazolium chloride for 2 h at 37°C. Cells were then washed with PBS and fixed in 2% paraformaldehyde (PFA). Images of cells were taken, and then reduced formazan particles, contained inside cells, were dissolved with 2 M KOH in DMSO and the absorbance was read at 630 nm on a multimode microplate reader (Bio-Tek, USA).
(2) Measurement of H 2 O 2 . The levels of H 2 O 2 were measured by using a H 2 O 2 assay kit from Beyotime Biotechnology. Briefly, scramble and shBAP31 cells were treated with LPS for 12 h; then, cells were washed with cold PBS and lysed using the assay buffer provided by the kit. After homogenization, the cell samples were centrifuged at 12000 × g for 5 min and the supernatant was collected. After deproteinization, the working solution containing an OxiRed probe and horseradish peroxidase (HRP) was added to samples according to the manufacturer's instructions. In the presence of HRP, the OxiRed probe reacts with H 2 O 2 to produce a product with color (λ max = 570 nm). The absorbance at 570 nm was read with a multimode microplate reader (Bio-Tek, USA). The amount of H 2 O 2 was calculated according to the standard curve (y = 0:0069x + 0:064, R 2 = 0:9985) and expressed as fold of scramble control cells.
2.2.7. Western Blot Analysis. Cells were collected and lysed with RIPA lysis buffer (1 mol/l Tris-HCl, pH 7.4; 1% Triton X-100; 1% sodium deoxycholate; and 150 mM NaCl 0.1% SDS) with protease and phosphatase inhibitor. The lysates were centrifuged at 12000 × g for 15 min at 4°C to produce whole-cell extracts. Protein concentrations were measured by the micro-BCA protein assay kit (Thermo Fisher Scientific, Waltham, MA, USA). The same amount of total protein lysates was separated by 12% SDS-PAGE and transferred to Immobilon polyvinylidene difluoride (PVDF) membranes (Millipore). After blocking with 5% nonfat milk in PBS, immunoblots were incubated with primary antibodies overnight at 4°C, followed by treatment with HRP-linked secondary antibodies. The intensity of immune-reactive bands was quantified using Image Lab Software.
2.2.8. Flow Cytometry Analysis. Scramble and shBAP31 microglial cells were treated with LPS (100 ng/ml) for 24 h, and the cells were harvested; then, 1 × 10 6 cells were fixed with 2% PFA for 30 min, punched with saponin for 20 min, and then blocked with BSA for 1 h. Subsequently, cells were suspended in 100 μl primary antibody (CD86 and CD206; 1 : 1000 in BSA) for 50 min on ice. Flow cytometry analyses were performed using BD Accuri™ C6 Flow Cytometer.
2.2.9. Enzyme-Linked Immunosorbent Assay. The levels of interleukin-1β (IL-1β) and TNFα in the culture supernatants were measured by ELISA kits according to the manufacturer's instructions (R&D Systems, Minneapolis, MN). Briefly, scramble and shBAP31 microglial cells were treated with LPS (100 ng/ml) for 24 h and culture supernatants were harvested; the medium was collected and centrifuged for 20 min at 1000 × g to remove the pellet. A volume of each 20 μl of the supernatant was sampled for measuring IL-1β and TNFα according to the manufacturer's protocol.
2.2.10. Nitrite Assay. Accumulation of nitrite (NO 2-) in culture supernatant fluids was measured by the Griess assay. Microglial cells (5 × 10 4 cells/well) were plated into 96-well plates, then treated with LPS (100 ng/ml) for 24 h. Then, 50 μl culture supernatant fluids were mixed with 50 μl Griess reagent at 37°C. Fifteen minutes later, the absorbance was determined at 540 nm using a multimode microplate reader. Quantifications of NO 2after LPS treatment were calculated as fold of untreated scramble cells.

BV2 Microglia Conditioned Medium-Induced SHSY5Y
Neurotoxicity. The effect of microglia conditioned medium (MCM) on the viability of SHSY5Y cells was measured using the MTT assay [21]. Scramble and shBAP31 cells were stimulated with LPS (100 ng/ml) for 24 h. Stimulation was terminated by collecting conditioned medium from the cells and centrifuged and stored at -80°C. SHSY5Y neurons cells were seeded at a density of 2 × 10 5 cells/ml in 96-well cell culture plates and incubated at 37°C. When cells reach confluence, the culture medium was removed and replaced with 100 μl of conditioned medium and DMEM medium containing LPS and/or apocynin and further incubated for 48 h. Stimulation was terminated by adding 20 μl of MTT (5 mg/ml in PBS) solution and incubating for 4 h at 37°C; then, the supernatant was removed, and 150 μl dimethyl sulfoxide (DMSO) was added to solubilize the formazan crystals. The absorbance was measured at 490 nm using a multimode microplate reader.

Animal and Surgical Procedures.
Mice with BAP31 deficiency in microglia were generated by crossing transgenic mice expressing Cre-recombinase under the lysozyme M (LysM) promoter with mice carrying a BAP31 gene flanked by LoxP sites [16]. The mice (BAP31 fl/fl and LysM-Cre-BAP31 fl/fl mice) were injected intraperitoneally (i.p.) with LPS (250 μg/kg/day) or saline for 1 week (8 mice per group) to make a model of oxidative stress in the brain according to a previous study [22][23][24]. Apocynin (30 mg/kg) was orally administrated once daily before LPS injection [25,26] to verify BAP31 functions through NADPH oxidase. The mice were divided into six groups: BAP31 fl/fl ; BAP31 fl/fl + LPS; LysM-Cre-BAP31 fl/fl ; LysM-Cre-BAP31 fl/fl + LPS; LysM-Cre-BAP31 fl/fl + apo; and LysM-Cre-BAP31 fl/fl + LPS + apo for the study. Mice were sacrificed, and brains were used for superoxide anion and morphological analysis. All animal experiments were carried out as per the National Institutes of Health Guide for the Care and Use of Laboratory Animals, as well as according to the guidelines of animal handling of our university authority.

Measurement of Reactive Oxygen Species Levels in
Brain. To assess superoxide anion production in the brain, the mice (BAP31 fl/fl and LysM-Cre-BAP31 fl/fl mice) were deeply anesthetized and perfused with saline and 4% paraformaldehyde; the brains were fixed in 4% paraformaldehyde in PBS overnight at 4°C; and the brains were first dehydrated by placing the tissue for 30 min in each of 70%, 95%, and 100% ethanol solutions, and then embedded (n = 8 per group for each experiment). The brains were cut at 10 μm thickness using a microtome blade (Leica, Wetzlar, HE, Germany), and sections (10 μm) were incubated with 5 μmol/l fluorescent dye dihydroethidium (DHE, Molecular Probes) at 37°C for 30 min in a humidified chamber and protected from light. Digital images were captured by a Leica (Wetzlar, HE, Germany) scanning confocal microscope, and the red fluorescence intensity was quantified by using NIH ImageJ 4 Oxidative Medicine and Cellular Longevity software. The fluorescence intensity was expressed relative to that of BAP31 fl/fl .

Immunofluorescence Staining.
Brain sections from all groups were prepared as described in Section 2.3.2; then, endogenous peroxidase activity was blocked with 0.3% H 2 O 2 for 10 min, washed in PBS, blocked with 5% BSA for 1 h, then species were incubated with the primary antibody (mouse 8-OHdG 1 : 200 in 10% goat serum PBS buffer) overnight at 4°C. The slices were then incubated with red fluorescent-conjugated secondary antibody (Thermo Fisher Scientific Inc.) for 1 h. After washing, the slices were incubated with rabbit anti-NeuN (1 : 1000 in 10% goat serum PBS buffer) followed by a green fluorescent-conjugated secondary antibody (Thermo Fisher Scientific Inc.) for another 1 h. Sections were then washed in PBS, and DAPI was used as the blue nuclear stain. Sections were viewed and processed in a Leica scanning confocal microscope. NeuN-positive cells were counted using ImageJ software (NIH) with a DAPI counterstain. The relative intensity of fluorescence in NeuN-positive cells/field of view was used for statistical analysis.

Statistical Analysis.
Statistical analyses were conducted with GraphPad Prism 7.0 Software (GraphPad, La Jolla, CA, USA) according to previous studies [27,28]. Data are expressed as the mean ± SEM. Student's t-test was used to compare two groups and was always used as two-tailed in Figure 1(e). For other Figures, analyses were performed using two-way ANOVA followed by Tukey's multiple-comparison analyses, p values < 0.05 were considered significant. * p < 0:05, * * p < 0:01, and * * * p < 0:001, and ns (no significant difference) denotes the significance thresholds.
Then, we verified the above results by flow cytometry. The primary microglial cells from BAP31 fl/fl and LysM-Cre-BAP31 fl/fl mice were specifically stained with CD86 and CD206, respectively. As shown in Figure 3 Fizz1 (h), and Ym1 (i) were analyzed by RT-PCR. Flow cytometry shows that the expression of CD86 (j) and CD206 (k) in primary microglial cells were treated with LPS (100 ng/ml) for 24 h. All the data are indicated as mean ± SEM of three independent experiments. * p < 0:05, * * p < 0:01, and * * * p < 0:001 versus the control group. 12 Oxidative Medicine and Cellular Longevity  13 Oxidative Medicine and Cellular Longevity were downregulated in BAP31-deficient cells; after LPS stimulation, CD206-positive cells were reduced further, which was consistent with the RT-PCR results.
Subsequently, we examined whether apocynin treatment influenced superoxide anion production in BAP31-deficient microglial cells. Primary microglial cells from BAP31 fl/fl and LysM-Cre-BAP31 fl/fl mice were treated with apocynin for 1 h, cells were stimulated with LPS, and then superoxide production was detected by the NBT and DHE assay. As shown in the NBT assay (Figures 5(c) and 5(d)), apocynin treatment decreased LPS-induced superoxide production caused by BAP31 deficiency from 2:24 ± 0:12-fold to 1:61 ± 0:01-fold, F ð2,12Þ = 152:5, p < 0:001. As shown in the DHE assay (supplementary Figure 4), apocynin treatment decreased LPS-induced superoxide production caused by BAP31 deficiency from 3:98 ± 0:24-fold to 1:97 ± 0:07-fold, F ð2,12Þ = 143:3, p < 0:001. Primary microglial cells were transfected with p22 phox siRNA for 60 h, followed by treatment with LPS for 12 h. The relative superoxide anion levels were measured by staining with NBT and quantified using a multimode microplate reader. (c, d, e) Primary microglial cells were transfected with keap1 siRNA for 60 h, followed by treatment with LPS for 24 h. The cytosolic and nuclear fractions were analyzed by Western blotting with antibodies against Nrf2, histone, and β-actin. (f) Primary microglial cells were transfected with keap1 siRNA for 60 h, followed by treatment with LPS for 12 h. The relative superoxide anion was measured by staining with NBT and quantified using a multimode microplate reader. All the data are indicated as mean ± SEM of three independent experiments. * p < 0:05, * * p < 0:01, and * * * p < 0:001 versus the control group.
To investigate whether BAP31 influenced the survival of neurons through proinflammatory cytokines, SHSY5Y cells were incubated with microglial conditioned medium from  (a, b, and c) Scramble and shBAP31 BV2 cells were treated with LPS (100 ng/ml) for 24 h, and the secreted protein levels of the cytokines IL-1β (a) and TNFα (b) in the supernatant were analyzed using ELISA kits. NO production (c) was measured by the Griess assay. (d, e) Visualization of SHSY5Y cells after coculture with microglial conditional medium (MCM) from scramble and shBAP31 BV2 cells exposed to LPS for 24 h after treatment with apocynin for 1 h. Cell viability was measured by the MTT assay. Scale bars = 200 μm. All the data are indicated as mean ± SEM of three independent experiments. * p < 0:05, * * p < 0:01, and * * * p < 0:001 versus the control group. 18 Oxidative Medicine and Cellular Longevity   [23]. BAP31 fl/fl and LysM-Cre-BAP31 fl/fl mice were intraperitoneally injected with LPS to assess the superoxide anion production in the brain. Consistently, we observed that systemic LPS administration enhanced the production of superoxide anion in both the BAP31 fl/fl and LysM-Cre-BAP31 fl/fl groups, but BAP31 deficiency significantly increased the accumulation of superoxide anion. As shown in Figure 8, BAP31 deficiency resulted in excessive superoxide production in the hippocampus (from 120:5 ± 3:39 to 177:1 ± 5:01, To verify the function of BAP31 on inflammatory cytokines in vivo, we examined the production of the proinflammatory cytokines IL-1β and TNFα in the hippocampus of the mice described above by RT-PCR. LysM-Cre-BAP31 fl/fl mice administered LPS showed higher expression (Figures 9(c) and 9(d)) of the two cytokines compared with the BAP31 fl/fl mice (IL-1β: 275 ± 54:07-fold versus 700:8 ± 114:8-fold, F ð1,32Þ = 58:39, p < 0:001; TNFα: 59:2 ± 4:81-fold versus 247:3 ± 51:8 -fold). Consistent with the in vitro results, BAP31 deficiency significantly exacerbated cytokine production.
As BAP31 regulated superoxide anion and neuroinflammation, which are harmful to neurons, we examined the neuronal integrity using a NeuN antibody that detects intact neurons. Similar sections of the hippocampus were compared between the six groups of mice. As shown in Figure 10(a), NeuN-positive cells in the BAP31 fl/fl groups were densely packed in the DG region of the hippocampus, after LPS challenge, the total fluorescence intensity of NeuN-positive cells per field was significantly decreased in the DG regions, as indicated in Figure 10

Discussion
Neurodegenerative diseases are characterized by chronic microglial overactivation and oxidative damage, wherein excessive levels of free radicals can overwhelm antioxidant response systems and lead to oxidative damage in the brain. Evidence indicates that superoxide anion and hydrogen peroxide are a major factor contributing to the initiation and/or progression of various neurodegenerative diseases [32]. As depicted in Figure 11, we found that BAP31 regulated LPS-induced superoxide anion and hydrogen peroxide production, and p22 phox /keap1 knockdown or apocynin treatment inhibited superoxide anion production induced by BAP31 deficiency. Additionally, BAP31 regulated microglial superoxide anion and hydro-gen peroxide production accompanied by alteration of inflammatory cytokine expression, which eventually led to oxidative damage and neuronal cell death.
The superoxide anion, an important free radical, is the source of other ROS. Hydrogen peroxide, a key reactive oxygen species, is produced at low levels during normal cellular metabolism and at higher concentrations under pathological conditions. An increased production of superoxide and hydrogen peroxide has been shown to mediate neuron death in neurodegenerative disease [33][34][35]. Therefore, our study examines the effects of superoxide anion and hydrogen peroxide in microglial cells. Under physiological circumstances, NOX likely generates only low levels of superoxide anion in the CNS [36,37], and BAP31 deficiency had a weak effect on superoxide production. Under pathophysiological conditions, superoxide anion reacts irreversibly with several cellular constituents including protein phospholipids and nuclear DNA, causing lipid peroxidation [38,39], membrane

22
Oxidative Medicine and Cellular Longevity damage, dysregulation of cellular processes, and genome mutations [40][41][42]. We found that BAP31 deficiency upregulated LPS-induced superoxide and hydrogen peroxide production. Superoxide may be protonated to form hydroperoxyl radical or dismutated to form hydrogen peroxide, both of which may contribute to the initiation of lipid peroxidation [43][44][45]. Consistent with previous results, BAP31 deficiency exacerbated superoxide anion production following LPS treatment, resulting in lipid damage in microglia. Previous studies have shown that LPS leads to superoxide anion production by upregulating the expression of NOX in microglial cells [46,47]. We also found that BAP31 deficiency upregulated LPS-induced superoxide anion production through the expression of NOX subunits, including p22 phox , p40 phox , p67 phox , and gp91 phox , indicating that BAP31 might regulate superoxide anion production via NOX. An increasing body of research suggests that gp91 phox forms a heterodimer with the smaller membrane-associated p22 phox protein, and together they make up the central component of NADPH oxidase, wherein p22 phox contributes to the maturation and stabilization of the heterodimer that it forms with gp91 phox [48]. As observed for Nox2, the Nox1, Nox3, and Nox4 proteins also form heterodimers with p22 phox , which is essential for their activity, and p22 phox and gp91 phox immunoreactivity is observed almost exclusively on microglia in the CNS [49]. Our study showed that BAP31 upregulated the protein level of p22 phox and that knockdown of p22 phox reversed superoxide anion production and inflammatory cytokine release. Consistent with our findings, p22 phox mutation inhibits inflammatory oxidative damage in endothelial cells and vessels [50]. CRISPR/Cas9mediated knockout of p22 phox leads to a loss of Nox1 and Nox4 activity. Downregulation of p22 phox ameliorates the inflammatory response during angiotensin II-induced oxidative stress by regulating MAPK and NF-κB pathways in ARPE-19 cells [51], indicating that BAP31 regulates superoxide anion production by p22 phox and BAP31 has protective effects on the homeostasis in the CNS. NOX enzyme cytosolic subunits are typically in the cytoplasm in resting cells, but most NADPH oxidases are activity-dependent enzyme complexes and their activation usually requires the translocation of cytosolic subunits to the membrane-bound subunits p22 phox and NOX isoforms. Under pathophysiological conditions, NOX enzyme cytosolic subunits translocate to membranes in response to cellular activation. Upregulated p22 phox promotes the further activation of NOX and results in a  Figure 11: A schematic illustration of BAP31 regulating superoxide anion production and neuroinflammation in microglia. BAP31 deficiency upregulates LPS-induced superoxide anion and hydrogen peroxide production through p22 phox and the keap1/Nrf2/HO-1 signaling pathway, excess superoxide anion, and hydrogen peroxide cooperate with inflammatory cytokine to induce the damage and death of neurons.
23 Oxidative Medicine and Cellular Longevity massive increase in superoxide anion. In this study, we discovered that lack of BAP31 increases the expression level of p22 phox . However, the relative changes between BAP31 deficiency and scramble cells showed no significant differences after LPS treatment, indicating that the sensitivity of p22 phox upregulation to LPS could be reduced as a result of BAP31 deficiency. Alternatively, the relative expression levels of those indicators that were not affected by BAP31 deficiency increased after the LPS treatment.
Keap1 is a cysteine-based mammalian intracellular sensor for electrophiles and oxidants [52]. The keap1-Nrf2 system plays important roles in the antioxidant response and contributes to cytoprotection from various redox disturbances [53]. Our study found that BAP31 deficiency upregulated the expression of keap1 and inhibited the translocation of Nrf2; knockdown of keap1 reversed superoxide anion production and inflammatory cytokine release caused by BAP31 deficiency by reversing the translocation of Nrf2. Similar to our results, repression of keap1 expression increases Nrf2 activity [54], and Nrf2 activation ameliorates inflammation and tissue damage in a sickle cell model [55], indicating that BAP31 may also regulate superoxide anion production via the keap1/Nrf2/HO-1 signaling pathway and that the reversal of keap1/Nrf2/HO-1 signaling might alleviate the superoxide anion production induced by BAP31 deficiency.
It is beginning to be recognized that superoxide anion polarizes microglia toward a classical proinflammatory phenotype, and excess superoxide anion accelerates the inflammatory response [56]. Here, we found that BAP31 deficiency exacerbated superoxide anion and proinflammatory cytokine production and decreased superoxide anion production by p22 phox /keap1 silencing in BAP31-deficient cells alleviated inflammatory cytokine production, indicating that BAP31 influenced inflammatory cytokine production through superoxide anion production.
A previous study has reported that microglial activation leads to noxious effects on neurons and participates in the pathophysiology of neurodegenerative diseases [57]. A key determinant of microglial neurotoxicity is the release of excitotoxins, including superoxide anion and inflammatory cytokines [58]. Consistent with the results of previous studies, our conditioned coculture in vitro studies indicated that superoxide anion and the inflammatory cytokines under conditions of BAP31 deficiency exacerbated the damage and death in neurons. In vivo, we applied the Cre-LoxP system to produce conditional BAP31 knockdown mice, and our results showed that BAP31 deficiency exacerbated neuronal oxidative damage and death by exacerbating superoxide anion and inflammatory cytokine production in vivo. Upon treatment with apocynin, the survival rate of neurons was elevated in vitro and in vivo, which is consistent with a report showing that apocynin prevents learning and memory deficits by protecting the neurons from superoxide and inflammatory cytokines [25,43,59,60]. These results indicate that apocynin has a therapeutic effect on neuroinflammation caused by BAP31 deficiency, and BAP31 might be an important regulator of superoxide anion production and neuroinflammation.
In summary, we illustrated that BAP31 deficiency may contribute to chronic unremitting inflammation in neurodegenerative diseases by promoting superoxide anion and inflammatory cytokine production, indicating that BAP31 might be a key regulator of microglial-related inflammation and neurotoxicity.

Conclusions
The present results suggest that BAP31 deficiency contributes to microglia-related superoxide anion production through p22 phox and keap1 pathways. The excess superoxide anion and inflammatory cytokines work together to induce damage and death in neurons. These studies highlight that BAP31 is an important regulator of superoxide anion production and neuroinflammation, and as such, the downstream regulators or agonists of BAP31 could be considered as a potential therapeutic targets in microglia-related superoxide anion production and neuroinflammation.