Acteoside Repressed Microglia M1 Polarization Through Inhibited NF-κB Signalling Pathway and AMPK-Mediated Mitochondria Function Recovery

Background: Alzheimer's disease (AD) is the most frequent type of dementia. While acteoside (ACT), a compound isolated from Cistanche tubulosa, possesses neuroprotective properties. However, the underlying mechanism in regulating microglia polarization remains ill-dened. Methods: Herein, AlCl 3 -induced AD model in zebrash larvae was applied to uncover the therapeutic ecacy of ACT. BV-2 cells were used to demonstrate the role of ACT on microglia polarization. RNA-Sequence, HPLC-Q-TOF-MS, western blot and molecular docking were combined to conrm its mechanism. Results: ACT signicantly ameliorated the experimental dyskinesia and nervous system disorders in zebrash. Subsequently, it suppressed M1 polarization and promoted to the M2 phenotype in LPS-induced BV-2 cells. We rst demonstrated that ACT exerted profound transcriptomic impact, which involved regulation of key signaling pathways in inammation, arginine biosynthesis, as well as pantothenate and CoA biosynthesis, correlating with mitochondria function. ACT treatment reduced microglia M1 polarization by inhibiting the NF-κB signalling pathway. And the metabolic pathways were further conrmed by HPLC-Q-TOF-MS. In addition, ACT rectied excessive ROS to restore mitochondria function through AMPK-mediated PGC-1α and UCP-2 upregulation, consistent with metabolic changes. Intriguingly, ACT may directly bind to both NF-κB and AMPKα, as evidenced by molecular docking. Conclusions: The research provided an infusive mechanism of ACT and illustrated a new perspective based on mitochondrial dysfunction to reveal the connection between metabolism and microglia polarization.

Acteoside (ACT), a phenylethanoid glycoside, is primarily derived from Cistanche tubulosa. Increasing evidence has suggested that ACT possessed numerous pharmacological activities, including neuroprotective [8] , anti-in ammatory [9] and antioxidant [10] effects. Particularly, ACT has been reported to improve learning and memory impairment as well as upregulate energy metabolism in streptozotocininduced rats [11] . It also has been suggested to inhibit neuronal apoptotic cell death and mitochondrial damage in the experimental autoimmune encephalomyelitis mice [12] . However, fewer studies have been focused on the effect of ACT on microglia M1/M2 polarization. Especially, various mechanisms, such as repair of mitochondria function and the regulation of cell metabolism, have not been performed. In addition, the mechanism of ACT contributed to microglia M1/M2 polarization has remained unexplored.
The present report was aimed to investigate the therapeutic e cacy of ACT as well as underly molecular mechanism of ACT in AD. Herein, ACT showed signi cant neuroprotection effect in AlCl 3 -induced AD zebra sh larvae. In addition, ACT effectively inhibited M1 polarization and promoted the M2 phenotype in LPS-induced BV-2 cells. RNA-Sequencing (RNA-Seq) integrated with metabolomics method to better understand the underlying mechanism of ACT in regulating microglia polarization. The crosstalk between metabolism and microglia polarization in terms of mitochondrial function were investigated. This study will provide a new respect for the further investigation of ACT as a potential therapeutic agent for treating AD.

Animals and model grouping
Wild-type zebra sh (AB strain, 4 months old) were chosen in this study (Nanjing Qi Wu Biotechnology Co., Ltd.). They were maintained under 14/10 h light/dark cycle at 28°C, following the previous method [13] .
Natural fertilized and normal developed embryos were generated and cultured to 3 days post-fertilization (dpf) in an illumination incubator. All zebra sh experiments were carried out under the supervision of the Animal Ethics Committee of China Pharmaceutical University.
Zebra sh larvae were divided into six groups and treated from 3 dpf to 7 dpf: control group, model group, model + donepezil hydrochloride (DPZ) group, model + ACT groups. The control group was maintained in the medium with 0.2% DMSO and the model group was treated with 150 μM AlCl 3 (pH 5.8). The model + DPZ group was co-treated with AlCl 3 and 8 μM DPZ. The model + ACT groups were co-treated with AlCl 3 and different concentrations of ACT (200, 100, 50 μM). ACT (HPLC purity ≥ 98%) was obtained from Baoji Herbest Bio-Tech Co., Ltd. (Baoji, China). AlCl 3 ·6H 2 O and DPZ were purchased from Shanghai Aladdin Bio-Chem Technology Co., LTD. (Shanghai, China).

Behavioral analysis
Zebra sh larvae movements were recorded with a ViewPoint behavioral analyzer (Zebralab 2018, ViewPoint Life Sciences Co., Ltd.) at 28°C. Brie y, the behavioral parameters and result processing were consistent with the method we established earlier [13] . Here, average speed (AS), speed change (ΔS), dyskinesia recovery rate (DRR), and response e ciency (RE, %) were selected to evaluate dyskinesia recovery in zebra sh.
Determination of acetylcholinesterase (AChE) and choline acetyltransferase (ChAT) activity After treated from 3 dpf to 7 dpf, zebra sh larvae were collected to measure AChE and ChAT activity. Based on the manufacturer's protocol, the activity was detected by the enzyme-linked immunosorbent assay (ELISA) kits (MLBIO biotechnology Co. Ltd., Shanghai, China). And the protein concentrations of different samples were determined by BCA method.

Cell viability assay
Cell Counting Kit-8 (CCK-8) assay (JianCheng Bioengineering Institute, Nanjing, China) was used to evaluate the viability of BV-2 cells. The cells were seeded in a 96-well plate (1×10 4 cells/well, Wuxi NEST Biotechnology Co., Ltd.). Brie y, the medium was removed at the end of the treatment, and 100 μL of serum-free medium containing CCK-8 solution was added to each well for 2 h at 37℃. The absorbance was measured at 450 nm with a microplate reader (Bio-Tek Instrument, Winooski, VT, USA). Cell viability is expressed as a percentage of the control group. The experiment was repeated three times.
Nitric oxide (NO) production assay NO was determined by measuring nitrite levels in the BV-2 culture supernatant using Griess reagent.
Brie y, at the end of the treatment, the medium (100 μL) was transferred to a new 96-well plate. The same volume of Griess reagent was added to each well and reacted for 15 min in the dark. The absorption at 540 nm was determined by a Microplate Reader.

In ammatory cytokines levels in supernatant
The concentrations of TNF-α, IL-1β and IL-10 in BV-2 cell supernatant were determined by ELISA kits according to the manufacturer's instructions (MLBIO biotechnology Co. Ltd., Shanghai, China).

Observation of cellular morphology
To determine the effect of ACT on BV-2 cells M1/M2 polarization, the cells were plated in 6-well dish and observed under the inverted microscope (Nikon ECLIPSE Ti2, Japan).

Cellular metabolism determination by HPLC-Q-TOF-MS analysis
BV-2 cells were seeded in 6-well dish separately (n=6/group). After treatment, the medium was removed, and the cells were washed three times with cold PBS. Then immediately exposed to liquid nitrogen to suppress cells metabolism. The cells were harvested with cold 80% methanol (1 mL/well) and the suspension were transferred to a 2 mL Eppendorf tube. To facilitate protein precipitation, vigorously vortexed for 1 min and centrifuged at 13,000 rpm for 15 min at 4°C. The cell suspension was transferred to a new 2 mL Eppendorf tube and dried under a stream of nitrogen and stored at -80°C until analysis. The dried residue was reconstituted in 150 μL of pre-cooled 25% acetonitrile. In order to ensure the stability and accuracy of the sequence analysis, equal volumes (10 μL) of each cell sample were combined as quality control (QC) samples. During metabolite detection, these samples were injected after every six cell samples to con rm their stability. A 1 μL aliquot was injected for HPLC-Q-TOF-MS.
HPLC-Q-TOF-MS analysis was performed on Agilent 1290 HPLC system connected with the Aglient 6530 Quadrupole Time-of-Flight (Q-TOF) mass spectrometer (Agilent Technologies, Santa Clara, CA, USA). The separation was carried out on an ACQUITY UPLC BEH C 18 column (2.1×100 mm, 1.7 μm). The mobile phase was composed of 0.1% formic acid-water (v/v; A) and acetonitrile (B).

Measurement of mitochondrial membrane potential (MMP)
MMP was detected using uorescent probe JC-1 (Beyotime, China) in accordance with the manufacturer's instructions. Brie y, cells from different groups were rinsed with PBS and incubated with JC-1 staining solution for 20 min at 37°C. After staining, cells were washed twice using staining buffer. Then, uorescent signals were detected by ow cytometry (BD Accuri C6).

Measurement of mitochondrial adenosine 5'-triphosphate (ATP)
ATP concentration in mitochondria was detected by an ATP Assay Kit (Beyotime, China) in accordance with manufacturer's instructions. Brie y, the culture medium of BV-2 cells from different groups were discarded, and cells were homogenized with lysis buffer on ice. The supernatant obtained after centrifugation (12,000 g, 5 min) was used to determine the ATP concentration. The luminescence (luciferase-catalyzed uorescein reaction) was detected by EnVision Multimode Microplate Reader (PerkinElmer).

Measurement of intracellular reactive oxygen species (ROS) level
ROS Assay Kit (Beyotime, China) was used to measure ROS level. The cells from different groups were incubated with DCFH-DA (10 μM) for 20 min at 37°C. After probe loading, cells were washed three times with DMEM. Then, uorescent signals were detected by ow cytometry (BD Accuri C6).

Transmission electron microscopy (TEM)
BV-2 cells were seeded in 6-well dish. The medium was removed and 1 mL of 2.5% glutaraldehyde was rapidly added to each well. Then the cells were transferred to a 1.5 mL Eppendorf tube and centrifuged at 1000 rpm for 3 min. The cells were xed overnight with new 2.5% glutaraldehyde at 4°C. After xation, dehydration and embedding, the cells were observed with an HT7800 transmission electron microscope (Hitachi, Tokyo, Japan).

RNA-seq and bioinformatic data analysis
Total RNA from BV-2 cells (n=3/group) were extracted using Trizol reagent (Vazyme Biotech, China) according to the reagent manufacturer's instructions. All analytical samples were sent to Majorbio (Shanghai Majorbio Bio-pharm Technology Co., Ltd.) for performing the RNA sequence assay. The data were analyzed on the free online platform of Majorbio Cloud Platform (www.majorbio.com). The parameters for the differential expression analysis were P-adjust < 0.05 and |log2FC| ≥ 1. The original sequence data have been submitted to the database of the NCBI Sequence Read Archive (SRA).
Quantitative real time polymerase chain reaction (qRT-PCR) The total RNA of BV-2 cells in each group was harvested using 500 μL RNA-easy TM Isolation Reagent (Vazyme Biotech, China), and reverse transcription reaction was conducted with FastKing-RT SuperMix (TIANGEN Biotech, China). Reactions were performed according to the manufacturer's protocol. cDNA was subjected to qRT-PCR assays with speci c primers and TransStart TOP Green qPCR SuperMix (TransGen Biotech, China). The primers are listed in Table S1 (see Additional le 1) and β-actin was used as the internal control. The 2 -ΔΔCT method was used for quantitative analysis.

Molecular docking
Molecular docking analysis were performed using Autodock software (Version 4.2). The a nity between ACT and proteins were observed by AutodockTools software. The three-dimensional (3D) protein structures of AMPKα (PDB ID: 5g5j), and NF-κB (PDB ID: 4q3j) were retrieved from the Protein Data Bank (https://www.rcsb.org).

Statistical analysis
All data are expressed as the mean ± standard deviation (SD). The differences between the different groups were analyzed by one-way analysis of variance (ANOVA), followed by Tukey's multiple comparison test. P <0.05 was considered statistically signi cant.

Results
ACT alleviated dyskinesia and improved cholinergic system function in zebra sh larvae AlCl 3 -induced AD model in zebra sh larvae was used to demonstrate the effect of ACT on AD. Firstly, the zebra sh larvae movement within light/dark cycles was observed and their swimming trails were recorded (Fig. 1a, 1b). AS and ΔS of zebra sh movement induced by AlCl 3 in the corresponding time after administration were calculated. The results showed that different doses of ACT effectively increased the AS and ΔS of zebra sh (Fig. 1c). DRR and RE uncovered a more intuitive comparison of ACT and DPZ (Fig. 1d). Accordingly, ACT alleviated dyskinesia, exhibiting similar effects as DPZ.
It's generally agreed that the cholinergic system plays an important role in learning and memory processes. Thus, the activities of AChE and ChAT were used to reveal the effect of ACT. AlCl 3 exposure in zebra sh rendered with brain cholinergic alteration (Fig. 1e). It was outstanding that ACT treatment suppressed the activity of AChE. In addition, the activity of ChAT exhibited a decrease after ACT treatment. In brief, ACT showed a profound impact on cholinergic system function in AlCl 3 -induced AD zebra sh larvae.
ACT suppressed M1 polarization and promoted M2 polarization in LPS-induced BV-2 cells The effects of ACT on microglia polarization were studied in vitro using BV-2 microglial cells. LPS signi cantly decreased BV-2 cell viability after treated 24 h. Fortunately, ACT increased the cell viability of LPS-induced BV-2 cells (Fig. 2a). In addition, the morphology of BV-2 cells was observed. After 24 h of LPS stimulation, it showed that BV-2 cells underwent a M1 polarization state. And the morphological changes were prevented by ACT co-treatment (Fig. 2b).
Moreover, the results indicated that unlike BV-2 cells stimulated by LPS, BV-2 cells co-treated with ACT displayed signi cantly suppressed TNF-α (Fig. 2c), IL-1β (Fig. 2d) and NO (Fig. 2f) expressions in cell supernatant. These are classical pro-in ammatory cytokines as the indicators of M1 microglia polarization. Similar to the results of ELISA, the results of qPCR discovered that TNF-α, nitric oxide synthase (iNOS), IL-1β and CD86 mRNA expressions were signi cantly inhibited by ACT treatment compared with the LPS group (Fig. 3a).
ACT regulated M1/M2 polarization via the inhibition of NF-κB signaling pathway in LPS-induced BV-2 cells The transcriptomic analysis was performed by RNA-seq to understand the mechanism of ACT in BV-2 cells from an overall level. PCA illustrated that the control, LPS, and ACT groups could be well distinguished (Fig. 4a). It revealed 899 differentially expressed genes (DEGs) between the control group and LPS group, whereas 49 DEGs between the LPS group and ACT group (Fig. 4b). Consistently, Gene ontology (GO) enrichment analysis (Fig. 4c) and Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analysis (Fig. 4d) uncovered that the effect of ACT was involved in the NF-κB signalling pathway. The set of genes associated to NF-κB signalling pathway was further con rmed and their homeostasis were certainly affected by LPS. As expected, ACT signi cantly affected their expressions (Fig. 4e).
NF-κB signalling pathway is a classical pathway to regulate the progression of in ammation, ultimately resulting in the release of pro-in ammatory factors. To gain mechanistic support, key protein of the NF-κB signalling pathway was evaluated by western blot analysis. LPS stimulation led to the activation of NF-κB, associated with promoting M1 polarization. Consistent with RNA-seq analysis, ACT inhibited LPSstimulated NF-κB phosphorylation (Fig. 4f). Therefore, ACT relieved the LPS-induced M1 polarization via the NF-κB signalling pathway in BV-2 cells.
ACT impaired arginine biosynthesis as well as pantothenate and CoA biosynthesis in LPS-induced BV-2 cells RNA-seq demonstrated that the pathways affected by ACT also included arginine (Arg) biosynthesis as well as pantothenate and CoA biosynthesis (Fig. 4d). And it has been suggested that LPS stimulation causing BV-2 cells metabolism disorders associated with M1 polarization [14] . Thus, untargeted cell metabolome by HPLC-Q-TOF-MS was used to identify the effect of ACT on the cell metabolism. PCA (Fig. 5a) and PLS-DA (Fig. 5b) illustrated that the control, LPS, and ACT groups could be well distinguished based on intracellular metabolites. The levels of various metabolites in LPS-induced BV-2 cells were changed after ACT treatment (Fig. 5c). Compared with the control group, there were 11 metabolites changed signi cantly in LPS group (Table S2, see Additional le 2). Whereas 14 metabolites were distinctly altered after the treatment of ACT (Table S3, see Additional le 3), involving 11 metabolic pathways (Fig. 5d). The effect of ACT mainly consisted of regulating amino acid metabolism (phenylalanine, tyrosine and tryptophan biosynthesis, D-Glutamine and D-glutamate metabolism, Arg biosynthesis, phenylalanine metabolism), nucleotide metabolism (purine metabolism, pyrimidine metabolism), energy metabolism (nitrogen metabolism), as well as metabolism of cofactors and vitamins (pantothenate and CoA biosynthesis). Interestingly, the metabolic pathways obtained by metabolome were consistent with RNA-seq, including Arg biosynthesis as well as pantothenate and CoA biosynthesis. The above showed that ACT could regulate Arg biosynthesis as well as pantothenate and CoA biosynthesis in LPS-stimulated BV-2 cells.

Act Mitigated Lps-induced Bv-2 Mitochondrial Dysfunction
Mitochondria are at the core of metabolic pathways. Evidence is evolving that mitochondria are key players in microglial M1/M2 polarization. Previous research demonstrated that LPS could cause mitochondrial dysfunction. An overview of the mitochondria status in morphology and cell distribution were judged by TEM. After LPS stimulation, BV-2 cells showed nucleus chromatin condensation, decreased cytoplasm, as well as fewer mitochondria (Fig. 6a). In addition, the mitochondrial cristae of the LPS group were disarranged or even disappeared, exhibiting partial cristolysis, reduced size, and roundshaped morphology. Interestingly, ACT treatment could alleviate LPS-induced morphological changes on mitochondria. By turns, we examined the mitochondria function of LPS-treated BV-2 cells, including MMP, and mitochondrial ATP production. The MMP (Fig. 6b, 6c) and ATP production in mitochondria (Fig. 6d) were signi cantly improved in cells treated with ACT compared to those in the LPS group. Mitochondrial dysfunction may be related to the increased level of ROS in the cell. Flow cytometry analysis revealed that the contents of ROS were overloading in the LPS group (Fig. 6e, 6f). Fortunately, ACT eliminated excessive ROS. It suggests that ACT may restore mitochondria function by clearing ROS.
ACT restored mitochondria function through the upregulation of PGC-1α and UCP-2 Peroxisome proliferative activated receptor-γ co-activator-1α (PGC-1α) plays an important role in mitochondrial biogenesis [15] . The stimulation of LPS decreased the expression of PGC-1α, exhibiting mitochondria dysfunction. Remarkably, PGC-1α gene mRNA and protein expression were signi cantly reversed by ACT treatment in LPS-treated BV-2 cells (Fig. 7a).
The mitochondrial uncoupling protein-2 (UCP-2) also was known to regulate mitochondrial functions. As a downstream protein of PGC-1α, it can control LPS-induced MMP depolarization and ROS production. Recent reports indicate that it is central to the process of microglial activation, with opposite regulation of M1 and M2 polarization [16] . Western blot results showed that UCP-2 protein level was decreased after LPS stimulation. The co-treatment of LPS and ACT could upregulate the expression of UCP-2 as compared to the LPS treatment. The mRNA expression level of UCP-2 also showed similar changes (Fig. 7b). Taken together, ACT restored mitochondria function through the upregulation of PGC-1α and UCP-2.
ACT repressed microglia M1 polarization through mitochondrial function recovery via AMPK activation AMP-activated protein kinase (AMPK), as the key cellular energy sensor, plays an important role in maintaining cell metabolism balance. At the same time, PGC-1α is a downstream protein of AMPK. Results uncovered that LPS inhibited the activation of AMPK, resulting in cell metabolism disorders and mitochondrial dysfunction. It's noteworthy that ACT could dose-dependently increase the protein expression of p-AMPK (Fig. 8a). It's suggested that ACT might increase the expression of PGC-1α and restore mitochondrial function by activating AMPK signalling pathway.
In this study, to investigate whether the activation of AMPK contributed to the regulation effect of ACT on M1/M2 polarization, compound C (CC) was employed to inhibit the effect of AMPK. In contrast to the downregulated NO level in ACT treatment group, CC partly blocked the effect of ACT on NO level (Fig. 8d). Based on these results, ACT could regulate M1/M2 polarization of BV-2 cells by the activation of AMPK.
ACT bond to and inhibited NF-κB as well as activated AMPKα Molecular docking was applied to con rm whether ACT binds to the NF-κB and AMPKα proteins. Findings demonstrated that the binding energy of ACT and NF-κB was − 8.4 kcal/mol, which of ACT and AMPKα was − 10.8 kcal/mol. Signi cant a nities veri ed that ACT directly bound to NF-κB and AMPKα (Fig. 9). Subsequently, the possible binding modes and interactions within the amino acid pocket were further explored, including Phe A146, Pro A147, Asn A240, Leu A236, Arg A232, His A183, Arg A239, Glu A179, Cys A149, and Tyr A227 of NF-κB (Fig. 9c) as well as Phe A213, Gly A481, Ala A370, Leu A482, Arg A212, Ile A369, Arg A106, Phe A215, Phe A108, Thr A309, Ser A119, and Thr A224 of AMPKα (Fig. 9f). These results indicated that ACT might directly affect NF-κB and AMPKα to attenuate BV-2 microglia M1 polarization and promoted the M2 phenotype.

Discussion
AD is a progressive neuronal and cognitional dysfunction disease, with complex dysregulated mechanisms [17] . Accumulating evidence has demonstrated a signi cant association between microgliadriven in ammation in the brain. It seems to play a critical role in the progression of AD. Microglia are macrophages in brain [18] . It could be activated to a classically M1 in ammatory phenotype, characterized by enhanced secretion of proin ammatory cytokines [4] . Excessive M1 activation could accelerate neuron damage and neurodegeneration, even exacerbate to AD [19] . Thus, it's imperative to seek new therapeutic approaches aimed at controlling microglia polarization points that could provide adaptive bene ts.
Our previous work has veri ed that ACT had signi cant effects of improving the learning and memory ability, and protecting the neurons in rat [20] . Consistently, the present study also proved that ACT could relieve AlCl 3 -induced dyskinesia and cholinergic system disorder in zebra sh. Excitedly, ACT presented remarkable anti-in ammatory activities in LPS-induced BV-2 cells. The transcriptomic pro le con rmed the signi cant changes in LPS-induced cells compared with control cells, as well as ACT-treated cells compared with LPS-induced.
ACT suppressed M1 polarization by inhibiting the NK-κB pathway. Except for the NF-κB pathway, RNA-seq also discovered that ACT treatment could affect arginine biosynthesis as well as pantothenate and CoA biosynthesis. Interestingly, the two metabolic pathways were further con rmed by HPLC-Q-TOF-MS analysis. It's widely reported that iNOS could metabolize Arg to NO and citrulline whereas Arg-1 could hydrolyze Arg to ornithine and urea, associated with neuron repair [21] . LPS stimulation led to the upregulation of iNOS (Fig. 3a) and downregulation of Arg-1 (Fig. 3b), resulting in increased NO level (Fig. 2f). The data uncovered that ACT alleviated the increased NO level through arginine biosynthesis.
Pantothenic acid (PA) is the primary substrate for pantothenate kinase [22] , as a rate-limiting metabolite in CoA biosynthesis. PA is the obligate precursor of acetyl-CoA, which of particular importance for cholinergic neurons [23] and participate in tricarboxylic acid cycle (TCA cycle) [24] . Recent study showed that elevated concentration of CoA would lead to altered mitochondrial morphology, and lower ATP content [22] . LPS-induced BV-2 cells exhibited a decrease in the number of mitochondria and a change of mitochondrial shape. After induced by LPS, the production of ROS increased in BV-2 cells. Then the overladen ROS caused membrane phospholipid to be attacked by free radical [25] . It led to the loss of MMP, in turn mitochondrial dysfunction and ATP depletion. It was outstanding that ACT treatment mitigated the decrease of MMP and ATP content. These data suggested that ACT induced mitochondrial dysfunction by regulating pantothenate and CoA biosynthesis.
It has been extensively reported that microglia polarization is closely associated with cell metabolism [14] . Particularly, as the metabolic hub, mitochondria play remarkable roles in regulating cell metabolism.
Recently, mitochondria have been positioned as a key determinant point in microglia polarization [26] . To better understand the mechanism of ACT, we judged the functional axis of mitochondria by western blot analysis. It revealed that ACT induced mitochondrial dysfunction by the activation of AMPKα/PGC-1/UCP-2 axis.
PGC-1α and UCP-2 are both related to mitochondrial biogenesis [27,28] , and they can be thought as the master regulator of ROS [29] . Reports indicate that PGC-1α-mediated mitochondrial biogenesis and reduction of ROS are dependent on induction of UCP-2 [27][28][29] . Due to overloading ROS, the expression of PGC-1α and UCP-2 was down-regulated in LPS-induced BV-2 cells. It suggested that ACT could eliminate excessive ROS through PGC-1α and UCP-2, thus restoring the mitochondrial function. According to the literature, the alteration of PGC-1α in BV-2 cells could contribute to regulating polarization. Interestingly, previous report has found that increased PGC-1α expression inhibited the NF-κB activity in LPS-induced BV-2 cells [30] . It quali ed the relationship between PGC-1α and NF-κB in our study.
The expression of PGC-1α is affected by upstream pathway proteins, such as AMPK. AMPK is the key protein for the maintenance of cellular homeostasis [31] , playing various roles in promoting M2 polarization of microglia [32] . It modulates metabolic pathways in cells [33] . We found that ACT promoted the activation of AMPK. At the same time, the application of compound C (AMPK inhibitor) blocked the effect of ACT on attenuating LPS-induced NO excess. Therefore, ACT also suppressed LPS-stimulated M1 polarization via AMPK signaling pathway.
It's the rst time to report the mechanism of ACT on regulating microglia polarization (Fig. 10). The data supported that ACT could be developed as a therapeutic agent for neurodegenerative disease associated with neuroin ammation, such as AD. Especially, we linked the microglia polarization to cell metabolism, explaining the effect of ACT through the alteration of mitochondria function. The identi cation of this metabolic axis, the targeting of this as a unique entity, could allow much better therapeutic approaches against microglia M1 polarization, particularly in AD.