Introduction
Excessive reactive oxygen species (ROS) production after ischemia/reperfusion contributes to acute brain injury after stroke. Nicotinamide adenine dinucleotide phosphate oxidase (NADPH oxidase, NOX) is the main enzyme responsible for ROS generation [
1]. NOX2, formerly gp91phox, is the most widely distributed NOX isoform in the central nervous system (CNS) and has been extensively studied in various CNS diseases, including cerebral ischemia. Previous studies have demonstrated that NOX2 is upregulated in the ischemic penumbra and core after ischemic stroke onset [
2‐
4], and NOX2-mediated ROS production promotes brain injury poststroke [
3]. On the other hand, the NOX2 inhibitor apocynin reduces ROS generation and ameliorates neuronal death and cerebral injury induced by stroke both in vitro and in vivo [
5]. Therefore, NOX2 promotes acute brain injury after stroke. Considering these detrimental effects of NOX2, other studies have suggested that NOX2 may also play beneficial roles in certain pathological conditions. Genetic deficiency or system inhibition of NOX2 increased hypoxia/ischemia-induced infarct and brain injury in newborn mice 7 days after stroke [
6]. In addition, NOX2 is beneficial in the peripheral vasculature after hind limb ischemic injury by increasing capillary density and perfusion and contributing to the proliferation of endothelial cells [
7]. However, whether NOX2 has dual roles in brain injury after stroke remains unknown.
The excessive ROS generated by NOX2 can cause inflammasome activation, which mediates the inflammatory response. NLRP3 inflammasomes, which are members of the NOD-like receptor family, are multiprotein complexes that serve as platforms for caspase activation and regulate cytokine maturation, inflammation and cell death (pyroptosis) and are also the most extensively studied inflammasomes in CNS diseases [
8,
9]. Furthermore, NLRP3 inflammasome activation has been reported in cerebral ischemia/reperfusion injury (CIRI) models, as well as in stroke patients [
10‐
12]. Overall, NOX2 appears to play an important role in the activation of ROS-dependent NLRP3 inflammasomes [
11].
Moreover, NOX2 promotes angiogenesis. NOX2 was reported to be localized in angiogenic blood vessels in the rat brain up to 7 days poststroke with reperfusion [
13]. In addition, NOX2-derived ROS are thought to act as effectors of angiogenesis and may function in brain recovery and regeneration after cerebral ischemia [
14]. Furthermore, ROS exert biphasic effects characterized by the promotion of oxidative injury in the early phase and increasing angiogenesis in the late phase in cerebral ischemia.
Autophagy, which is an evolutionarily conserved process, eliminates damaged proteins and organelles that accumulate in eukaryotes [
15]. NOX serves as a gate keeper to regulate the activation of autophagy based on cell type and cellular conditions [
16,
17], and NOX-mediated effects on autophagy are ROS-dependent [
18]. The autophagy machinery is linked with the inflammasome, and autophagy inhibits NLRP3 inflammasome activation, while autophagy inhibition or deficiency promotes NLRP3 inflammasome activity and the release of IL-1β and IL-18 [
19,
20]. In addition, autophagy induction can promote vascular growth, while autophagy inhibition blocks angiogenesis, and this process is mediated by ROS production [
21]. Furthermore, both autophagy and angiogenesis are associated with the PI3K/Akt/NF-kB pathway. However, whether NOX2 regulates autophagy, the NLRP3 inflammasome and angiogenesis in ischemic stroke is still poorly understood, and whether the PI3K/Akt/NF-kB pathway is involved in these processes is unknown.
In this study, we first tested the hypothesis that NOX2 plays dual roles in acute brain injury and delayed functional recovery after ischemic stroke. We found that NOX2-mediated ROS production as necessary and beneficial for brain functional recovery. In addition, we investigated the mechanisms of the distinctive effect of NOX2–ROS on the ischemic brain during different periods of stroke and showed that these effects were linked with dynamic autophagy activation, which regulated the NLRP3 inflammasome and angiogenesis through the PI3K/Akt pathway.
Materials and methods
Animals
Male wild-type C57BL/6J mice (n = 300; 25–30 g, 8–10 weeks) were purchased from Hubei Experimental Animal Research Center (Hubei, China; Nos. 43004700018817 and 43004700020932). All animal experimental protocols were approved by the Animal Experimentation Ethics Committee of Wuhan University (No. WDRM-20170504) and were conducted according to the Animal Care and Use Committee guidelines of Renmin Hospital of Wuhan University. Animals were housed in a room with controlled humidity (65 ± 5%) and temperature (25 ± 1 °C) under a 12/12-h light/dark cycle with free access to food and water for at least 1 week before the experiments.
Drug administration
Apocynin (178385, Sigma, St Louis, MO, USA) was dissolved in dimethyl sulfoxide and then diluted with sterile saline to the desired concentrations. Different concentrations of apocynin were intraperitoneally (i.p.) administered to the animals at a dose of 2.5 mg/kg 8 h after reperfusion as described by Qin et al. [
22]. 3-MA (M9281, Sigma-Aldrich, St Louis, MO, USA) was dissolved in physiological saline and administered (15 mg/kg, i.p.) 1 and 3 h after occlusion [
23]. Rapamycin (553210, Sigma-Aldrich, St Louis, MO, USA) was stored at room temperature and dissolved in DMSO with gentle heating to yield a clear, colorless solution and then injected (10 mg/kg, i.p.) immediately after MCAO [
24].
MCAO model
The MCAO model was established as previously described [
25]. In brief, C57BL/6J wild-type mice were anesthetized with 5% isoflurane in O
2 by a facemask, followed by ligation of the left middle cerebral artery with a 6-0 monofilament (Doccol Corp., Redlands, CA, USA). After 1 h of occlusion, the monofilament was removed to initiate reperfusion. A homeothermic heating pad was used to monitor and stabilize the body temperature at 37 ± 0.5 °C. The same procedure, but without monofilament ligation, was performed on sham-operated mice.
Infarct volume measurement
The mice were deeply anesthetized, euthanized with an overdose of isoflurane and decapitated 3, 7 and 14 days after MCAO. The brains were collected after transcranial perfusion with saline followed by 4% paraformaldehyde. After postfixation with 4% paraformaldehyde for 72 h, brain tissues were cut into 50-µm coronal sections, dipped in 0.1% cresyl violet solution for 30 min, and then rinsed in distilled water. The stained sections were fixed by serial dehydration in ethanol (70%, 90%, 100%) and xylene. The infarct volume was measured and analyzed by a blinded observer using ImageJ v1.37 (NIH, Bethesda, MA, USA), as described previously [
26,
27], and the data were normalized and are presented as a percentage of the nonischemic hemisphere to correct for edema.
Assessment of neurological deficits
Neurological deficit scores were evaluated 3, 7 and 14 days after MCAO as described previously [
28]. The score ranged from 0 (without observable neurological deficit) to 4 (no spontaneous motor activity and loss of consciousness).
Rotarod test
Accelerating rotarod (SD Instruments, San Diego, CA) instruments were used to test the motor coordination of the mice. Each mouse was placed on a 2.75 cm diameter rotating rod every other day before MCAO onset for a total of 9 training days, and the rotation speed of the rod increased from 5 to 10 rpm every 5 min. The time between the beginning of the mouse staying on the rod and falling from the rod was determined up to a maximum duration of 300 s. After the training period, MCAO surgery was conducted, and rotarod testing was performed by a blinded observer on Days 0, 3, 7, 10, and 14 poststroke. The scores were calculated by averaging the three repetitive time records of each mouse each day.
Ischemic core and penumbra segmentation
The separation of ischemic core and penumbra was performed as previously descriptions with minor revision based on the average infarct volume measured by TTC or cresyl violet staining in our pilot experiments and the heterogeneity of penumbra as a dynamic and changeable process over time [
29‐
33]. Mice were deeply anesthetized and euthanized with an overdose of isoflurane, the brain tissue was quickly collected and put on ice and the olfactory bulb, cerebellum and low brain stem were removed. Firstly, the brain tissue was cut on a coronal plane 3 mm backward from the top of the frontal lobe into three slices with thickness of 3 mm (section 1), 4 mm (section 2) and 3 mm (section 3), respectively. Then, the section 2 was taken out and the midline was identified between the ipsilateral (left) and contralateral (right) hemispheres. Next, a sagittal cut 1.5 mm far from the midline was conducted from top to bottom on the ipsilateral hemisphere, and a transverse diagonal cut was also made at around “1 o'clock” position. Thus, the tissue outside “1 o'clock” was the infarct core area, and the cortical tissue between sagittal cut and “1 o'clock” was the ischemic penumbra (Additional file
1: Fig. S1).
High-throughput RNA sequencing (RNA-Seq)
Total RNA was extracted from the penumbral region of the ischemic hemisphere on Day 7 after MCAO using TRIzol reagent (Invitrogen, Carlsbad, CA, USA). Subsequently, total RNA was quantified by using a NanoDrop ND-1000, and 1 ~ 2 μg of total RNA was used to construct the RNA-Seq libraries. mRNA was enriched using the NEBNext® Poly(A) mRNA Magnetic Isolation Module (New England Biolabs). Libraries were then constructed using a KAPA Stranded RNA-Seq Library Prep kit (Illumina) according to the manufacturer's protocol. The sequencing library was examined by an Agilent 2100 Bioanalyzer using an Agilent DNA 1000 chip kit (Agilent, part # 5067-1504). All samples were sequenced on an Illumina HiSeq 4000 with 150 bp paired-end reads. After quality control was performed, the raw sequencing data were aligned to the mouse genome (GRCm38) using Hisat2 software. Finally, differentially expressed genes were defined as those with a fold change ≥ 1.5 and p value ≤ 0.05. Cluster analysis was performed using Cluster 3.0 software. Gene Ontology (GO) biological process analysis was performed using DAVID, and KEGG was used for pathway analysis.
Measurement of ROS
To assess ROS production, the brain was carefully and quickly isolated, cut into 4.0 μm sections and placed on chilled microscope slides. The samples were incubated in physiological saline containing 10 μmol dihydroethidium (DHE; Sigma–Aldrich) for 30 min at 37 °C in the dark. The brain sections were washed twice with PBS and placed under an automatic fluorescence microscope (BX63, Olympus Optical Ltd, Tokyo, Japan).
Immunofluorescence staining
Immunofluorescence analysis was performed as previously described [
26]. Ischemic and sham-operated mice were euthanized and perfused with cold PBS, followed by fixation with 4% paraformaldehyde for 2 days. The ischemic brains were cut into 50-μm sections, and the free-floating slices were blocked with 0.1 M PBS containing 5% fetal bovine serum and 0.3% Triton X for 1 h at room temperature. After being washed, the slices were incubated at 4 °C overnight with the following primary antibodies: anti-NLRP3 (1:200; ab4207, Cell Signaling Technology, Boston, USA), anti-LC3B (1:200; ab104224, Abcam, Cambridge, England), and anti-CD31 (1:100, GB11224, Servicebio, Wuhan, China). The slices were then rinsed and incubated with an Alexa 594-conjugated antibody (1:200; ANT030, Millipore, Billerica, MA) or an Alexa 488-conjugated antibody (1:200; ANT024, Millipore, Billerica, MA) for 2 h at room temperature. After being thoroughly rinsed, the nuclei were stained with DAPI (94010, Vector Laboratories, Burlingame, CA, USA). All slices were photographed using a confocal fluorescence microscope (BX63, Olympus Optical Ltd, Tokyo, Japan). The number of immunoreactive cells in predefined areas was quantified using ImageJ software (Media Cybernetics Inc., Rockville, MD, USA). Six different fields for each mouse and six mice for each group were counted. All counts were conducted by blinded observers.
Western blot analysis
Western blotting was carried out as previously described [
25]. Cortical sections 1.0 to 2.0 mm from ipsilateral brain tissue were harvested and homogenized in cold RIPA buffer (C1053, Applygen, Beijing, China) containing a protease inhibitor cocktail (G2006, Servicebio, Wuhan, China). The homogenates were centrifuged at 4 °C at 10,000×
g for 30 min, and then the supernatants were harvested. Protein levels were determined with a BCA kit (G2026, Servicebio, Wuhan, China). Protein samples (20 μl/lane) were separated by electrophoresis on 4–15% sodium dodecyl sulfate–polyacrylamide gels and then transferred onto PVDF membranes (Millipore, Billerica, MA, USA). The membranes were then placed into 5% nonfat milk in PBS/0.1% Tween and blocked for 1 h, followed by incubation overnight with mouse anti-p62 (1:300, #16177, Cell Signaling Technology, Boston, USA), anti-NLPR3 (1:1000; ab4207, Cell Signaling Technology, Boston, USA), anti-ASC (1:1000; 67824, Cell Signaling Technology, Boston, USA), anti-CL-caspase-1 (1:500; 89332, Cell Signaling Technology, Boston, USA), anti-HMGB1 (1:1000; ab18256, Abcam, Cambridge, England), anti-IL-1β (1:500; ab8320, Abcam, Cambridge, England), anti-NOX2 (1:1000; ab18256, Abcam, Cambridge, England), anti-Beclin-1 (1:1000; ab18256, Abcam, Cambridge, England), anti-PI3K (1:1000; ab8378, Abcam, Cambridge, England), anti-p-PI3K (1:1000; ab8378, Abcam, Cambridge, England), anti-Akt (1:1000; ab8378, Abcam, Cambridge, England), anti-p-Akt (1:1000; ab8378, Abcam, Cambridge, England), anti- NF-κB p65 (1:1000; ab8378, Abcam, Cambridge, England), anti-phosphorylated NF-κB p65 (1:1000; 3033S, Cell Signaling Technology, Boston, USA) and anti-VEGF (1:1000; 2042744, Millipore, Billerica, MA) at 4 °C. After being washed with PBS/0.1% Tween, the membrane was incubated with IRDye-labeled secondary antibodies (1:10,000; c60405-05, Li-Cor Bioscience, USA) at room temperature for 1–2 h. Images were acquired with an Odyssey western blot analysis system (LI-COR, Lincoln, NE, USA). The relative band intensity was calculated using Quantity One v4.6.2 software (Bio–Rad Laboratories, Hercules, USA) and then normalized to the GAPDH loading control. These experiments were performed three times.
Statistical analysis
The results are shown as the mean ± SD and were analyzed using SPSS 22.0 software. The specific statistical methods are described in each figure legend. One-way ANOVA was performed when comparing different groups, whereas differences in two groups were evaluated using unpaired Student's t tests. P ≤ 0.05 denotes significant changes.
Discussion
While the role of NOX2-derived ROS in the brain remains controversial and has been suggested to be deleterious in some studies, recent evidence has emerged to support a neuroprotective role of this factor [
14,
38‐
40]. The present study was the first to present novel results that NOX2 plays opposing roles in the acute and chronic phases of ischemic stroke by using the NOX2 inhibitor apocynin. While NOX2 inhibition attenuated acute infarction and improved neurological scores and behavioral performance on Day 3 poststroke, it blocked the functional recovery from 7 to 14 days. We found that ROS production increased after stroke in a NOX2-dependent manner, indicating that NOX2 is a major source of ROS after MCAO. Similarly, we demonstrated that the NOX2 inhibitor apocynin increased angiogenesis and reduced the NLRP3 inflammasome in the acute phase of stroke but suppressed angiogenesis and promoted NLRP3 inflammasome activation in the later stage of stroke. Then, by using a specific autophagy inhibitor and inducer, we provided evidence that autophagy, which is also a double-edged sword, is involved in NLRP3 inflammasome activation and the promotion of angiogenesis induced by NOX2–ROS during recovery. Thus, we have demonstrated that angiogenesis and anti-inflammatory effects were enhanced by NOX2/ROS-induced autophagy through phosphorylation of the PI3K/Akt/NF-kB signaling pathway. To the best of our knowledge, this is the first study to demonstrate that NOX2 has distinctive effects during different phases after ischemic stroke. These novel findings provide insights for re-evaluating the role of ROS in brain injury and functional recovery after stroke.
Our results regarding the dual roles of NOX2 in stroke are supported by previous studies. First, the contribution of NOX2/ROS to ischemic brain damage during the acute period of stroke is well known [
3,
41‐
43]. It has been well established that the abrupt overproduction of ROS after ischemia/reperfusion exacerbates acute brain injury. NOX2 is the classic phagocytic NOX, and its primary role is the generation of free radicals, especially ROS, during ischemic stroke [
22,
44]. Our results showed that the production of NOX2 and ROS were both significantly increased 3 days after stroke and gradually decreased to the levels in the sham group on Day 14. Apocynin, a NOX2 inhibitor, decreased NOX2 expression and reduced ROS production. Therefore, it is reasonable to conclude that NOX2 inhibition attenuates acute brain injury by blocking ROS production. Second, the dual effects of ROS have been observed in other models, as excessive ROS cause detrimental effects and physiological levels of ROS result in benefits for the ischemic brain [
45,
46]. In addition, another study suggests that ROS exert biphasic effects on the ischemic brain, and the acute damaging effect occurs during the early phase, while the beneficial effects on neurovascular remodeling and functional restoration occur during the recovery stage [
23]. In our current study, we found that ROS are NOX2-dependent, and inhibiting NOX2 with apocynin improved brain damage, reduced neurological deficiencies and increased mouse survival on Day 3 after MCAO. In addition, we found that NOX2 inhibition in the ischemic brain by apocynin resulted in increased mortality and reduced functional recovery on Days 7 and 14 after stroke. Our results suggested the dual effects of NOX2–ROS during the different stages of brain ischemia, which were not only consistent with the theory regarding the biphasic roles of ROS but also provided solid evidence for this conclusion.
Our findings demonstrated that the underlying pathological mechanism by which NOX-2 mediates brain injury and functional recovery is linked to neuroinflammation and angiogenesis. The NLRP3 inflammasome, which is the best characterized inflammasome to date, contributes significantly to brain injury and neuroinflammation after stroke [
22]. The inflammasome can be activated by many factors, including environmental irritants, endogenous danger signals, pathogens, and various pathogen-associated molecular patterns (PAMPs). Substantial evidence indicates that ROS are proximal signals for NLRP3 inflammasome activation in CIRI [
47]. ROS promote tissue inflammation and activate the immune response through the NLRP3 inflammasome during ischemia/reperfusion [
48]. To confirm whether the ROS-induced NLRP3 inflammasome contributes to the dual functions of NOX2, we evaluated the dynamic changes in the NLRP3 inflammasome after MCAO and found that the number of NLRP3
+ cells increased on Day 3 but gradually decreased thereafter. However, the NOX2 inhibitor apocynin reduced NLRP3
+ cells on Day 3 but promoted inflammation during brain functional restoration, suggesting a critical role of NOX2 in the neuroinflammatory response. In contrast with neuroinflammation, angiogenesis promotes cell repair and survival after stroke. In the recovery phase of stroke, the generation of ROS-dependent growth factors promotes angiogenesis and induces the proliferation and differentiation of vascular smooth muscle cells to affect vascular remodeling [
49]. In addition, NOX2 has been shown to be advantageous in vascular proliferation after hind limb ischemic injury [
7,
50]. Moreover, NOX2 was expressed on new blood vessels in the rat brain 7 days after stroke [
13,
51]. In the current study, we found that the expression of the blood vessel marker CD31 was reduced on Day 3 but gradually increased at 7–14 days, and apocynin enhanced CD31 levels on Day 3 but inhibited its expression during brain recovery.
We further hypothesized that autophagy is involved in the effects of NOX2 on brain injury and recovery and investigated these events. As an intracellular bulk degradation system that exists ubiquitously in eukaryotes, autophagy also plays dual roles in CIRI [
15]. Previous studies have suggested that oxidative stress initiates autophagosome formation and increases autophagic flux after ischemia/reperfusion [
52‐
54]. In turn, autophagic activity could inhibit the activation of NLRP3 inflammasomes in CIRI [
55]. Autophagosomes recognize the NLPR3 inflammasome through selective autophagy receptors and remove ubiquitylated inflammasomes or eliminate activators, including defective organelles, damaged DNA and mitochondrial ROS, to negatively activate the NLPR3 inflammasome [
20,
56]. In contrast, autophagy inhibition or deficiency results in the activation of the NLPR3 inflammasome [
19]. In the current study, we showed that the expression of autophagy-related proteins was increased along with NOX2 and ROS, and the expression of these factors was downregulated by apocynin. Consistently, the specific autophagy inhibitor 3-MA and the autophagy inducer rapamycin blocked or activated autophagy and promoted or suppressed the NLPR3 inflammasome, respectively. In addition, apocynin blocked autophagy while enhancing NLPR3 inflammasome expression compared to 3-MA or rapamycin 7 and 14 days poststroke. These results strongly indicate that NOX2/ROS inhibit inflammation through autophagy activation, playing a neuroprotective role during the delayed stage of stroke.
Autophagy is also coupled with angiogenesis [
57], and it is involved in the development and fundamental roles of vascular endothelial cells [
58]. This autophagy-initiated angiogenesis is mediated by ROS [
21] and is a key repair mechanism for brain recovery after stroke. Insufficient autophagy is a detrimental factor that induces EC phenotypic senescence and dysfunction [
59]. Thus, manipulating autophagy by genetic methods or stimulators/inhibitors can modulate angiogenesis. Consistent with these previous studies, our current study confirmed that NOX2-derived ROS-dependent autophagy was involved in regulating angiogenesis and preventing brain injury, thus accelerating brain recovery after stroke.
The molecular mechanisms by which NOX2–ROS-autophagy influences vascular proliferation and NLRP3 inflammasome inhibition are complex and poorly understood. To further explore the molecular mechanisms, we used high-throughput RNA sequencing to analyze the effect of NOX2–ROS on the ischemic brain during the recovery stage. The results indicated that the PI3K–Akt signaling pathway was enriched in the apocynin treatment group 7 days after MCAO, which indicated that this pathway might be involved in NOX2–ROS-mediated long-term recovery after ischemic insult. Moreover, the regulation of angiogenesis and the NLRP3 inflammasome is well known to be linked with the PI3K/Akt pathway. PI3K/Akt activation upregulates VEGF expression and accelerates blood vessel regeneration, while pharmacological inhibition of the PI3K/Akt signaling pathway decreases VEGF secretion and inhibits angiogenesis [
60]. However, whether angiogenesis and inhibition of the NLRP3 inflammasome induced by NOX2/ROS/autophagy is mediated by the PI3K/Akt pathway remains unknown. In the current study, we confirmed the RNA sequencing results and demonstrated that the NOX2 inhibitor apocynin and the autophagy inhibitor 3-MA inhibited the protein expression of VEGF, phosphorylated PI3K and Akt, and NF-kB while promoting the expression of NLRP3 inflammasome-related proteins. In addition, VEGF and phosphorylated PI3K/Akt/NF-kB were synchronously upregulated, but NLRP3 inflammasome-related proteins were downregulated with increasing NOX2 and autophagy. Therefore, our results suggest that the PI3K/Akt/NF-kB signaling pathway is involved in angiogenesis and inhibiting inflammation induced by NOX–ROS-dependent autophagy after stroke. Our finding is consistent with the study by Du showing that autophagy triggered angiogenesis through VEGF and Akt [
61]. We are aware of other conflicting findings that autophagy blocks PI3K/Akt activation and exerts antiangiogenic effects on neuroblastoma cells [
62]. Although these discrepancies may be dependent on different cell types and cellular demands [
58], further research is needed to clarify autophagy-induced angiogenesis and the underlying mechanisms related to the PI3K/Akt/NF-kB signaling pathway after ischemic stroke.
However, there are a number of limitations in our current study, which need to be addressed by further comprehensive investigations. First, the standard between high and low ROS levels was missing, and we only measured different levels and analyzed the statistical significance of ROS in the current study. Second, the molecular mechanism by which NOX/ROS induce autophagy after stroke is not yet known and needs to be further screened by new technologies, such as transcriptome or proteome analysis, verified and clarified.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.