Animals and treatments

Male APPswe/PS1dE9 (APP/PS1) double-transgenic mice and wild-type (WT) littermates (Strain background: C57BL/6JGpt; provided by GemPharmatech) were housed under a 12 h light/dark cycle with food and water ad libitum. All animal experimental procedures were approved by the Ethics Review Committee of Nanjing Drum Tower Hospital (No. 2024AE01014).

3.5-month-old APP/PS1 and WT mice were injected intraperitoneally every other day for 4.5 months. Since EDA has shown protective effects in AD animal models [14, 15], we used EDA as a positive control. EDA and EDB (provided by Simcere Pharmaceutical) were diluted with normal saline (SA). The standard dose of EDB was set at 10 mg/kg, consisting of 10 mg/kg EDA and 2.5 mg/kg d-Bor [16-18]. We further set a low-dose group with 5 mg/kg EDB, and the positive control group with 10 mg/kg EDA. The grouping was summarized as follows: WT + SA (WT mice treated with SA), APP + SA (APP/PS1 mice treated with SA), APP + EDA 10 (APP/PS1 mice treated with 10 mg/kg EDA), APP + EDB 5 (APP/PS1 mice treated with 5 mg/kg EDB), APP + EDB 10 (APP/PS1 mice treated with 10 mg/kg EDB). A flow chart was shown in Fig. 1A.

Cell culture and treatments

Primary neuronal cell culture has been previously described [19, 20]. And primary microglia were co-cultivated with primary neurons at a 1:3 ratio to establish the neuron-microglia co-culture system [21].

Recombinant human Aβ1−42 (Sigma-Aldrich, AG968) was incubated at 37 °C for 5 d to obtain aggregated Aβ. The primary cells were pre-treated with Paquinimod (Paq; MedChemExpress, ABR-215757), EDA or EDB for 2 h, and then insulted with 2 µM Aβ for 24 h.

S100A9 (NCBI Gene ID: 20202) overexpressed lentivirus (LV-S100A9) and corresponding control lentivirus (LV-Con), knockdown lentivirus (LV-shRNA) and corresponding control lentivirus (LV-Blank) were constructed by OBiO Technology Inc. and GeneUniversal Inc. The vector used for the lentivirus was pcSLenti-EF1-mCherry-F2A-Puro-CMV-MCS-WPRE. Primary cultured microglia were transduced at a multiplicity of infection of 10. After 8 h of transfection, the virus-containing medium was replaced with complete medium and incubated for 2 d before treatment.

Behavioral tests

Mice were allowed to adapt to the experimental environment for 1 h and then subjected to behavioral tests [20, 22]. After each mouse was tested, the instruments were cleaned using 75% alcohol.

Open field

The open field test was used to evaluate the locomotor ability and anxiety of mice. The experimental box was divided into 16 identical squares. The 4 corner ones served as the corner zone and the 4 middle ones served as the center zone. Mice were allowed to move freely for 10 min. The moving speed and time spent in the corner/center zone were recorded by ANY-maze software (Stoelting).

Novel object recognition (NOR)

The NOR test was used to evaluate the short-term memory of mice. Mice were habituated to the box for three consecutive days before testing. During the familiarization session, mice were allowed to explore two identical objects freely for 10 min. At least 30 min later, one of these objects was replaced with a new-shaped object. During the test session, mice were allowed to explore two objects freely for 5 min. The discrimination index was equal to the percentage of time spent exploring the new object over the total time spent exploring objects.

Y-maze

The Y-maze test was used to evaluate the short-term memory of mice. A Y-shaped maze with three identical arms at 120° angle (35 cm × 8 cm × 6 cm) was introduced. Mice were allowed to move freely for 8 min. Spontaneous alternations were regarded as nonoverlapping entries into three arms continuously [23].

Morris water maze (MWM)

The MWM test was used to assess the learning and memory of mice. The instrument consisted of a circular pool and an underwater escape platform. During the 5-day acquisition trial, mice were trained to locate the escape platform within 60 s. If the mouse couldn’t find the platform within the stipulated time, it would be guided onto the platform and habituated for 30 s. During the probe test, mice were allowed to explore freely for 60 s with the platform removed. The mean swimming speed, number of platform crossings, latency to the platform, time spent in target quadrant and latency to the target quadrant were recorded with ANY-maze software (Stoelting).

Electrophysiology

Acute 300-µm hippocampal slices were prepared as previously described [20]. The slices were placed into the microelectrode array and perfused continuously with oxygenated artificial cerebrospinal fluid (2 mL/min, 32℃). Field excitatory postsynaptic potentials (fEPSPs) were conducted from the CA1 stratum radiatum using the MEA-2100-60-System (Multi Channel Systems). The slope of fEPSPs was measured to analyze the input-output (I/O) relationships of synapses. In long-term potentiation (LTP) experiments, the stimulation intensity was 50% of the maximum evoked response and the stimulation frequency was 100 Hz (three trains, 1-s duration, 10-s interval time). The initial slopes of fEPSPs were normalized by the average value at baseline. Data were acquired by LTP-Director software and analyzed by LTP-Analyzer software (Multi Channel Systems).

Western blotting (WB)

Western blotting was performed as previously described [20]. Total protein was extracted from hippocampal tissue using RIPA lysis buffer supplemented with protease and phosphatase inhibitors, and further separated by SDS-PAGE and transferred to PVDF membrane. After being blocked with 5% non-fat milk, the membranes were incubated overnight at 4 °C with primary antibodies. Primary antibodies used were as follows: mouse-anti β-actin (1:2000; Sigma, A5441), mouse-anti GAPDH (1:10000; Proteintech, 60004-1-Ig); rabbit-anti PSD-95 (1:1000; Abcam, ab18258), rabbit-anti Syn-1 (1:1000; Abcam, ab64581), rabbit-anti MAP-2 (1:1000; Bioworld, bs3487), rabbit-anti BACE1 (1:1000; Cell Signaling Technology, 5606), rabbit-anti ADAM10 (1:1000; Abcam, ab1997), rabbit-anti S100A9 (1:1000; Proteintech, 26992-1-AP; cell signaling technology, 73425), rabbit-anti S100A8 (1:1000; Abclonal, A20474), rabbit-anti JAK2 (1:1000; Cell Signaling Technology, 3230), rabbit-anti Phospho-JAK2 (1:1000; Cell Signaling Technology, 3776), mouse-anti STAT3 (1:1000; Cell Signaling Technology, 9139), rabbit-anti Phospho-STAT3 (1:1000; Cell Signaling Technology, 9145), rabbit-anti NF-κB p65 (1:1000; Cell Signaling Technology, 8242), rabbit-anti Phospho-NF-κB p65 (1:1000; Cell Signaling Technology, 3033), rabbit-anti p38 MAPK (1:1000; Bioworld, ap0424), rabbit-anti Phospho-p38 MAPK (1:1000; Bioworld, bs4766), rabbit-anti ERK (1:1000; Cell Signaling Technology, 4370), rabbit-anti Phospho-ERK (1:1000; Cell Signaling Technology, 4695), rabbit-anti JNK (1:1000; Cell Signaling Technology, 9252), rabbit-anti Phospho-JNK (1:1000; Bioworld, bs4322), rabbit-anti RAGE (1:1000; Abcam, ab3611), mouse-anti TLR4 (1:1000; Santa Cruz Biotechnology, sc-293072), rabbit-anti Phospho-Threonine (1:1000; Jingjie PTM BioLabs, PTM-705RM), mouse-anti HA (1:1000; Santa Cruz Biotechnology, sc-7392). Following TBST (0.1% Tween 20/Tris-buffered saline) washes, the membranes were incubated with HRP-conjugated secondary antibodies (Bioworld) for 2 h at room temperature. Bands were visualized by the Gel-Pro system (Tanon) and quantified using ImageJ Fiji software.

Golgi staining

Mouse brain tissues were treated with a FD Rapid Golgi stain kit (FD Neurotechnologies). The 100-µm brain slices were cut and underwent Golgi staining according to the manufacturer’s instructions. The stained pyramidal neurons in the hippocampal CA1 region were visualized using a bright field microscope (Olympus, IX73). Neuronal Sholl analysis and dendritic spine counting were performed with ImageJ Fiji software.

Oxidative stress assay kits

The antioxidant capacity was detected using a Total Antioxidant Capacity Assay Kit with FRAP method (Beyotime, S0116), and the activity of antioxidant enzymes was detected using a Total Superoxide Dismutase (SOD) Assay Kit with WST-8 method (Beyotime, S0101S) and a Total Glutathione Peroxidase (GSH-Px) Assay Kit with NADPH method (Beyotime, S0058) in mouse hippocampus or primary neurons according to the manufacturer’s instructions. The levels of oxidative damage were detected using a Lipid Peroxidation Malondialdehyde (MDA) Assay Kit (Beyotime, S0131S), a Reactive Oxygen Species (ROS) Assay Kit (Beyotime, S0033S) and a Protein Carbonyl Content Assay Kit (Solarbio, BC1275). In addition, a Mitochondrial Membrane Potential Assay Kit with TMRE method (Beyotime, C2001S) were used to measure alterations in mitochondrial membrane potential of primary neurons.

Quantitative real-time PCR (RT-qPCR)

Total RNA was isolated from mouse hippocampus or primary microglia using TRIzol reagent (Invitrogen) and converted to cDNA using the Evo M-MLV RT Mix Kit (Accurate Biotechnology, AG11728). Quantitative real-time PCR was then performed on a Step One Plus PCR system (Applied Biosystems) using the SYBR Green kit (Applied Biosystems). Each gene expression level was normalized relative to that of β-actin (Gene: Actb). The specific primer sequences were shown in Table S1.

Enzyme linked immunosorbent assay (ELISA)

For in vivo experiments, after anesthesia with inhaling a 5% isoflurane-oxygen mixture, mice were transcardially perfused with PBS, and then brains were removed to further isolate the hippocampus. We used RIPA buffer containing phosphatase and protease inhibitors to extract proteins from samples, and added zirconia beads (Beyotime, F6632) to the RIPA-immersed tissue and homogenized the tissue thoroughly by a grinder. For in vitro experiments, the cell culture medium was collected and centrifuged to remove cell debris, and the supernatant was used for the assay. To determine the levels of IL-1β, IL-6, TNF-α and S100A9 in mouse hippocampus or cell supernatant, the ELISA assays were performed according to the manufacturer’s instructions using a Mouse IL-1β ELISA Kit (Bioworld, CEK1788), a Mouse IL-6 ELISA Kit (Bioworld, CEK1785), a Mouse TNF-α ELISA Kit (Bioworld, CEK1783) and a Mouse S100A9 ELISA Kit (Cusabio, CSB-EL020642MO).

Immunofluorescence staining (IF)

After anesthesia with inhaling a 5% isoflurane-oxygen mixture, mice were transcardially perfused with PBS and 4% paraformaldehyde (PFA). The brain samples were collected and fixed in 4% PFA for 24 h. The 20-µm brain slices were cut after dehydration by sucrose. Primary cultured cells were fixed for 15 min with 4% PFA after removing the cell culture medium.

Immunofluorescence staining was performed as previously described [20]. Brain sections or primary cultured cells were sequentially permeabilized, blocked and incubated with primary antibodies overnight at 4 °C, followed by corresponding secondary antibodies (Invitrogen) on the next day. Primary antibodies were as follows: mouse-anti Aβ (6E10; 1:300; BioLegend, 803001), rat-anti CD68 (1:200; Bio-Rad, MCA1957), rabbit-anti IBA1 (1:500; FUJIFILM Wako Pure Chemical Corporation, 019-19741), goat-anti IBA1 (1:500; Abcam, ab5076), rabbit-anti S100A9 (1:100; Abclonal, A9842), goat-anti S100A9 (1:100; R&D system, AF2065), goat-anti AXL (1:500; R&D Systems, AF854), rat-anti TREM2 (1:500; R&D Systems, MAB1729), rabbit-anti PSD-95 (1:500; Abcam, ab18258), rabbit-anti Syn-1 (1:200; Abcam, ab64581), rabbit-anti Phospho-Threonine (1:50; Jingjie PTM BioLabs, PTM-705RM). After being counterstained with DAPI, fluorescent images were captured by a confocal microscope (Olympus, FV3000) and analyzed using ImageJ Fiji software.

Phagocytosis assays

We used Imaris software (Bitplane) to process 3D reconstitution for z-stack fluorescent images. The volume of CD68+ phagosomes, microglia, Aβ and synaptic puncta were calculated using surface rendered images. For the analysis of in vivo phagocytosis, the internalized Aβ index was equal to the ratio of the volume of Aβ and CD68 signal colocalized inside the cell to the volume of total Aβ plaques after 3D reconstitution of IF images. The engulfed PSD-95/Syn-1 index was equal to the ratio of the volume of PSD-95/Syn-1 and CD68 signal colocalized to the volume of microglia [24, 25].

FITC-Aβ1−42 (AnaSpec, AS-60479) was aggregated at 37 °C for 24 h with agitation [26]. Primary microglia were seeded in black-walled 96-well plates (Beyotime, FCP965). After 24 h of drug treatment, 0.2 µM FITC-Aβ was added for an indicated time, and the fluorescence was measured at 485-nm excitation/538-nm emission by a Spark Multimode Plate Reader (Tecan, Spark 10 M) [26].

Quantitative proteomics and bioinformatics analysis

Hippocampal proteins were extracted from the brain tissues of APP + SA and APP + EDB mice. The proteomics experiments and subsequent bioinformatics analysis in our study were supported by OE Biotech Co. Hierarchical cluster analysis was used to identify the differentially expressed proteins (DEPs) between groups. DEPs were enriched and analyzed by Gene Ontology (GO) database, focusing on molecular function and biological process. The mass spectrometry proteomics data have been deposited to the ProteomeXchange Consortium (http://proteomecentral.proteomexchange.org) via the iProX partner repository with the dataset identifier PXD054845.

Cellular thermal shift assay (CETSA)

Following the protocol provided by Martinez et al. [27], microglia were harvested and resuspended in buffers supplemented with protease inhibitors, and then freeze-thawed 3 times with liquid nitrogen. The lysate was subsequently centrifuged at 20,000 ×g for 20 min at 4 °C. The supernatant was divided into two aliquots and then treated with EDB or vehicle respectively for 30 min. These two lysates were further divided into smaller aliquots and incubated at each temperature point from 36 to 75 °C for 3 min. After cooling, these temperature-gradient lysates were centrifuged and separated for WB assay.

The mutant S100A9 plasmids or corresponding WT plasmids were transfected into BV2 cells using Lipofectamine 3000 (Invitrogen, L3000015). Specific information for plasmids was detailed in Table S2. After 6–8 h transfection, the plasmid-containing medium was replaced with complete medium, and the cells were cultured for 24 h. Then we collected the cells to perform CETSA.

Surface plasmon resonance (SPR)

The SPR experiments and subsequent affinity analysis in our study were supported by BetterWays Inc. A 3D photocrosslinked chip was printed. EDA and d-Bor were used as the stationary phases, while different concentrations of S100A9 (MedChemExpress, HY-P74583) were used as the mobile phase. For the interaction testing session, S100A9 flowed through the chip surface at 0.5 µL/s. The binding reaction temperature was 4 °C; the binding time was 600 s, and the dissociation time was 360 s. For the surface regeneration session, Glycine-HCl (pH = 2.0) solution was used as the regeneration solution, and the flow rate was 2 µL/s, and the regeneration time was 300 s. The intensity of the chip resonance was continuously recorded.

Molecular docking analysis

An automated docking was performed using the AutoDock 4.2.6 software [28] (https://autodock.scripps.edu/). S100A9 monomer (extracted from 6ZDY, Mus musculus; 5I8N, Homo sapiens), EDA ligand (PubChem CID: 4021), d-Bor ligand (PubChem CID: 6552009) were subjected to AutoDock respectively. Interactive docking with Genetic Algorithm protocol was carried out for ligands to the active sites. The docked compound was selected according to the predicted binding energy. The protein-ligand complexes were imported to the PLIP platform [29] (https://plip-tool.biotec.tu-dresden.de/) for further analysis. The final patterns were drawn with PyMOL 3.7 software (https://pymol.org/2/).

Co-Immunoprecipitation (Co-IP)

As previously described [30], the protein lysates of primary microglia were incubated with HA antibody (Cell Signaling Technology, 3724) or the same amount of rabbit IgG control (Bioworld, BD0051) overnight at 4 °C. The next day, Protein A beads (Millipore) were added for another 4 h. After being centrifuged and washed 5 times, the beads were combined with the same volume of 2 × loading buffer and boiled. We used the final supernatant to perform WB assay.

Thioflavin-T (ThT) fluorescence assay

Twenty µM ThT, which can bind to the amyloid β-sheet structures, was added to each well in black-walled 96-well plates. Recombinant mouse S100A9 (75 µM; MedChemExpress, HY-P74583) was incubated with EDB (molar ratios of 1:0, 1:1 and 1:10) at 37 °C for 6 h with continuous 420-rpm shaking. The fluorescence was measured each 10 min at 430-nm excitation/495-nm emission by a Spark Multimode Plate Reader (Tecan, Spark 10 M) [31].

Atomic force microscopy (AFM)

Recombinant mouse S100A9 (MedChemExpress, HY-P74583) was dissolved in PBS (pH 7.4). We used the 1:0 and 1:10 molar ratios of S100A9 to the corresponding EDB during incubation (37 °C, 6 h). Twenty µL of each sample was deposited on the surface of mica, washed 5 times with deionized water and left dry overnight at room temperature. According to previous studies [31, 32], AFM imaging was carried out by Shiyanjia Lab with a BioScope Catalyst AFM (Bruker) and a PicoPlus AFM (Molecular Imaging).

Statistical analysis

Statistical analyses were performed using GraphPad Prism 9. Differences between groups were assessed, using the detailed statistical tests which were indicated in the figure legends. We used Shapiro-Wilk test to determine normal distribution. If the data were normally distributed, we chose the appropriate tests (e.g. t test, one-way ANOVA, two-way ANOVA) depending on the number of groups and variables involved in the statistics, and results were represented as the means ± SD. If the data were not normally distributed, we used the nonparametric tests instead (e.g. Mann-Whitney test, Kruskal-Wallis test), and results were represented as the median ± IQR. The Bonferroni post hoc correction was used for one/two-way ANOVA, and the Dunn’s post hoc correction was used for Kruskal-Wallis test. Statistical significance was set at p < 0.05.