Resveratrol may reduce age-related memory loss by preventing immune cell aging in the brain

Thirty-six male C57BL/6J mice, weighing 25 ± 2 g at seven weeks of age, were acquired from Jiangsu GemPharmatech Co., Ltd., Nanjing, China. Mice were housed in the specific pathogen-free (SPF) facility of the South China University of Technology Experimental Animal Center under standard conditions: room temperature 22–25 °C, relative humidity 50–60%, and a 12 h light/dark cycle. Standard chow and water were provided ad libitum. Following a 1-week acclimatization period, experimental interventions were initiated. All experimental procedures were reviewed and approved by the Animal Ethics Committee of the Experimental Animal Center, South China University of Technology (Protocol No. SPrz2024032). Mice were randomly allocated into three groups (n = 12 mice per group): control, model, and resveratrol-treated groups. Control animals received standard chow and were intraperitoneally injected with an equal volume of 0.9% sterile saline. The model and resveratrol groups were subjected to aging induction by daily intraperitoneal injection of D-galactose at 100 mg/kg for 8 consecutive weeks. Throughout the final 3 weeks, resveratrol-treated mice were given 50 mg/kg resveratrol daily by oral gavage. Mice in both the control and model groups were given an equivalent volume of 0.5% sodium carboxymethylcellulose (CMC-Na) solution by the same route and schedule.

Resveratrol was accurately weighed (50 mg) and dissolved in 10 mL of a 0.5% CMC-Na solution, which was prepared by dissolving 0.5 g of CMC-Na in 100 mL of 0.9% sterile saline. The mixture was thoroughly vortexed to produce a homogeneous suspension with a concentration of 5 mg/mL. The resveratrol suspension was freshly prepared each day and administered by oral gavage at a daily dose of 50 mg/kg body weight. To create a 10 mg/mL solution, D-galactose was dissolved in 0.9% sterile saline. Following membrane filtration for sterilization, the D-galactose solution was administered via intraperitoneal injection at a daily dose of 100 mg/kg body weight.

All mice were pre-handled for 3 days prior to testing and habituated in the behavior room for 1 h before the test. All behavior tests were analyzed by the VisuTrack software 2.0 (Shanghai Xinruan Information Technology, Shanghai, China).

The open field test was performed as follows. Briefly, the device was a 40 cm × 40 cm × 40 cm white acrylic arena that was custom-made [32]. Mice were placed at the periphery of the arena and allowed to freely explore for 5 min. Behavior was captured by a camera positioned above the arena. Total movement distance and time spent in the central area (10 cm × 10 cm) were automatically analyzed using software. After testing each mouse, the arena was cleaned with 75% ethanol.

The apparatus was a Y-shaped maze made of white acrylic [33]. Each arm was 30 cm long, 6 cm wide, and 15 cm high. Throughout the whole exam, visual cues were maintained in the testing room. In the initial test experiment, mice were allowed to visit two Y-maze arms for 5 min, while one arm was closed with a guillotine door. Mice were placed back in the start arm and given unrestricted access to all three arms for an additional 5 min following a 45 min inter-trial break. Each mouse was given a start and closed arm at random. Each mouse was tested, and then 75% ethanol was used to clear the maze. During the first 5 min of the second trial, we compared the average number of entries into the two familiar arms with the number of entries into the novel arm. VisuTrack software was used for scoring, and entries into an arm were counted when the nose point, center point, and tail base point for that arm were all within the defined area.

The rotarod test was performed as follows. Briefly, during the last three days of drug administration, mice were trained daily at fixed time points [34]. Mice were placed on the rotating rod for a 3 min habituation period at 4 rpm, followed by acceleration to the maximum speed (40 rpm) within 2 min. On the fourth day, the test was conducted with uniform acceleration from 4 rpm to 40 rpm over 2 min. Mice were allowed to remain on the device for a maximum of 5 min. Latency to fall and total walking distance were recorded. Each mouse was tested at least three times daily, with at least 1 h between trials. After testing each mouse, the rotarod apparatus was cleaned with 75% ethanol.

Brain tissues were removed after transcardial perfusion fixation with phosphate-buffered saline (PBS), followed by post-fixation with 4% paraformaldehyde (Beyotime, #P0099, Shanghai, China) and cryoprotection with 30% sucrose/PBS. Fixed frozen brain tissues were coronally sectioned using a cryostat (Leica CM1950, Nussloch, Germany) at 30 µm thickness, and three sections of dorsal hippocampal regions per brain per antibody were used for immunochemistry. The sections were subjected to antigen retrieval with sodium citrate (10 mM, pH 6.0) at 95 °C for 20 min, then blocked in 5% normal donkey serum, 5% bovine serum albumin (BSA, Millipore Sigma, #A1933, Burlington, MA, USA) and 0.1% TritonX-100/PBS for 1 h, followed by incubation with primary antibodies (pTBK1, 1:100, #5483S, Cell Signaling Technology, Danvers, MA, USA; IBA1, 1:1000, #019-19741, Wako, Tokyo, Japan; P2ry12, 1:100, #848002, Biolegend, CA, USA; TUB33, 1:500, #66375-1-lg, Proteintech, Wuhan, China; CD68, 1:100, #MCA1957, Bio-Rad, Hercules, CA, USA; β-gal, 1:200, #15518-1-AP, Proteintech, Wuhan, China), diluted with 1% BSA and 0.02% tween in PBS overnight. The samples were washed with PBST 3 times and incubated in secondary antibodies (donkey anti-rabbit IgG AlexaFluor 594, 1:2000, Thermo Fisher Scientific, #A32754, Waltham, MA, USA; donkey anti-mouse IgG AlexaFluor 488, 1:2000, Thermo Fisher Scientific, #A32766, Waltham, MA, USA) for 2 h at room temperature in darkness. After PBST washing, DAPI (Beyotime, #C1006) staining solution was added and incubated for 3 min at room temperature in the dark. Sections were mounted with anti-fade mounting medium. Subsequently, images were observed and captured using a fluorescence or confocal microscope and analyzed with the ImageJ image analysis system.

All confocal imaging was performed on a Nikon A1 using NIS-Elements C software 4.0 (Nikon, Tokyo, Japan) at the inverted microscope stand using the confocal mode with 63×/1.4 oil immersion objectives. Images of 1024 × 1024 pixels were acquired as confocal stacks with a z-interval of 20 μm using a system-optimized setup to image cells. For imaging IBA1, β-gal, pTBK1 and STING, a 552 nm laser line was used, and emission was collected at 565–650 nm; for imaging CD68, P2ry12, GM130, and TUB33, a 488 nm laser line was used, and emission was collected at 490–600 nm. All co-localization images were scanned frame by frame in the sequential scanning mode, which showed no cross-talk among multiple channels. Gain and off-set were set at values that prevented saturated and empty pixels. After image acquisition, all images were applied with lightning deconvolution. ImageJ was used to perform background subtraction, thresholding, and quantification. Final data analysis was performed using Microsoft Excel, and graph rendering was performed in GraphPad Prism 10.0.

SA-β-galactosidase staining was performed on fresh-frozen tissue sections according to established protocols with minor modifications (Cell Signaling Technology, #9860). Briefly, 30 μm cryosections were thawed and washed with PBS. After three washes with PBS, sections were incubated in SA-β-gal staining solution consisting of 1 mg/mL X-Gal, 5 mM potassium ferricyanide, 5 mM potassium ferrocyanide, and 2 mM MgCl₂ in citric acid-sodium phosphate buffer (pH 6.0). Incubation was carried out at 37 °C for 24 h in a humidified chamber protected from light. Tissue sections were then rinsed with PBS and examined under a bright-field microscope. HMC-3 cells were stained in the same SA-β-gal staining solution at 37 °C for 24 h following the instructions. Senescence-associated β-galactosidase activity was visualized as an insoluble blue precipitate within cells. For double-labeling experiments, immunohistochemical staining was subsequently performed on the same sections following SA-β-gal staining. All staining steps were performed simultaneously for comparative experimental groups to minimize inter-assay variability.

Western blot analyses were performed as follows. Cells and tissues from each group were lysed using Western IP lysis buffer (Beyotime, #P0013, Shanghai, China) to extract total proteins. Protein concentrations were measured with the BCA Pierce Protein Assay Kit (Thermo Fisher Scientific, #23225, Waltham, MA, USA) and normalized to the lowest concentration. Proteins were separated by SDS-PAGE and transferred to nitrocellulose membranes. Membranes were blocked with 5% non-fat milk for 1 h at room temperature, followed by incubation with primary antibodies (TBK1, 1:2000, #3504S; pTBK1, 1:1000, #5483S; STING, 1:1000, #13647S; pSTING, 1:1000, #62912S; GAPDH, 1:10,000, #2118S, Cell Signaling Technology) in antibody dilution buffer overnight at 4 °C. Secondary antibodies (anti-mouse or anti-rabbit HRP-conjugated) were applied for 1 h at room temperature. Protein bands were visualized using enhanced chemiluminescence substrate and imaged with a ChemiDoc XRS BioRad (BioRad, Hercules, CA, USA) imaging system with Image Lab software 6.1. Antibodies used are listed in the. Supplementary Table S1

Murine microglial cell line BV-2 and human microglial cell line HMC-3 were purchased from Shanghai Cell Bank, Chinese Academy of Sciences. BV-2 cells were cultured in DMEM basal medium, while HMC-3 cells were maintained in MEM basal medium. Both media were supplemented with 10% (w/v) fetal bovine serum (FBS) and 1% penicillin/streptomycin. Cells were incubated at 37 °C in a humidified atmosphere of 5% CO₂/95% air (v/v).

The dsDNA probes were generated by annealing complementary single-stranded DNA oligonucleotides (sense: ACATCTAGTACATGTCTAGTCAGTATCTAGTGATTATCTAGACATACATGATCTATGACATATATAGTGGATAAGTGTGG; anti-sense: CCACACTTATCCACTATATATGTCATAGATCATGTATGTCTAGATAATCACTAGATACTGACTAGACATGTACTAGATGT) [35]. The annealing reaction was assembled in a total volume of 100 µL, comprising 20 µL of 5× annealing buffer, 100 µM of each oligonucleotide (sense and anti-sense), and nuclease-free PCR-grade water. The mixture was subjected to a stepwise thermal annealing protocol as follows: 95 °C for 4 min, 85 °C for 4 min, 82 °C for 4 min, 78 °C for 4 min, 75 °C for 4 min, 72 °C for 4 min, 70 °C for 10 min, followed by a gradual cooling step at 69 °C with a −1 °C per cycle ramp for 58 cycles (1 min per cycle), and a final hold at 10 °C for 1 min. Cells were transfected with dsDNA using the jetPRIME^® (Polyplus, #101000046, Strasbourg, France) transfection reagent according to the manufacturer’s instructions. Briefly, cells were seeded in 6-well plates at a density of 300,000 cells per mL in DMEM supplemented with 10% FBS and cultured for 24 h at 37 °C under 5% CO₂. Prior to transfection, the medium was replaced with 1 mL of fresh complete medium per well. For each well, 2 µg of annealed dsDNA was diluted in jetPRIME^® buffer to a total volume of 100 µL. As a negative control, an equivalent volume of 1× annealing buffer was used in place of dsDNA. Subsequently, 2 µL of jetPRIME^® reagent was added to each mixture, vortexed briefly, and incubated at room temperature for 10 min. The transfection complex was then added dropwise to the cells. After gentle swirling to ensure even distribution, cells were incubated at 37 °C in a humidified atmosphere containing 5% CO₂ for the duration required by each experiment.

Resveratrol and H-151 stock solutions were prepared in dimethyl sulfoxide (DMSO) at 0.7 mM or 1 mM, respectively, and stored at −20 °C. Working solutions of resveratrol (0.7 µM) and H-151 (1 µM) were freshly prepared in DMEM prior to use. Control cells received equivalent volumes of DMSO in all experiments. Cells were pre-treated with resveratrol or H-151 for 1 h. Following pre-treatment, dsDNA was added and cells were co-treated for 9 h with the original compounds present.

Cells were fixed with 4% paraformaldehyde for 15 min, permeabilized with 0.1% Triton X-100 for 10 min, and washed with PBS. After blocking with 5% bovine serum albumin for 1 h, cells were incubated with primary antibodies (pTBK1, 1:100, #5483S, Cell Signaling Technology; pSTING, 1:100, #62912S, Cell Signaling Technology; STING, 1:200, #19851-1-AP, Proteintech; GM130, 1:200, #610822, BD Bioscience, Franklin Lakes, NJ, USA; P21, 1:100, #67362-1-Ig, Proteintech) overnight at 4 °C. Following PBS washes, samples were incubated with fluorescent secondary antibodies for 1 h, counterstained with DAPI, and visualized using an inverted fluorescence microscope.

Cell culture supernatants were collected and centrifuged at 500× g, 4 °C to remove cell debris and dead cells. TNF-α secreted in supernatants was measured using a mouse TNF-α ELISA kit (Invitrogen, #BMS607-3, Carlsbad, CA, USA) according to the manufacturer’s instructions. Cells were washed with ice-cold PBS and lysed in RIPA buffer (Beyotime, #P0013B) containing 1X protease and phosphatase inhibitor cocktail (Beyotime, #P1045). Lysates were clarified by centrifugation, and protein concentration was determined using the Pierce™ BCA Protein Assay Kit (Thermo Fisher Scientific, #23225, Waltham, MA, USA).

Total RNA was isolated from frozen hippocampus and cortex tissue or from cells with Trizol (Invitrogen, #15596018CN, Carlsbad, CA, USA) and was reverse transcribed using a Master Mix kit (TaKaRa, #RR092S, Kyoto, Japan), and the resulting complete DNA (cDNAs) were diluted 10-fold before use. To evaluate the level of gene expression, real-time quantitative PCR (RT-qPCR) with SYBR Green dye was applied. PCR reactions were set up in an 8-well PCR tube strip with a 10 μL final volume. The reaction mixture contained 5 μL SYBR™ Green PCR Master Mix (TaKaRa, #CN830S), 10 μM of each primer (forward and reverse) and 1 μL cDNA. All samples were run in duplicate. Amplification was carried out for 39 cycles with the standard program according to the manufacturer’s specifications in a CFX Opus96 RT-qPCR machine. Relative amounts of mRNA were calculated as the comparative C_t after normalization to the GAPDH control C_t value.

The Trizol reagent (Invitrogen, #15596018CN) was used separately to extract total RNA from the cell samples. Agilent 2200 was used to assess the RNA quality, which was stored at −80 °C. RNA with an RNA integrity number (RIN) greater than 7.0 was used to generate cDNA libraries. The VAHTS Universal V6 RNA-seq Library Prep Kit for Illumina (vazyme, Inc., Nanjing, China) was used to create the cDNA libraries for every RNA sample in accordance with the guidelines provided by the manufacturer (Shanghai NovelBio Pharmaceutical Technology Co., Ltd., Shanghai, China). In general, the following phases make up the protocol: Using oligo(dT) magnetic beads, poly-A-containing mRNA was isolated from 1 μg of total RNA. It was then broken up into 200–600 bp using divalent cations at 85 °C for 6 min. First- and second-strand cDNA synthesis was carried out using the cleaved RNA fragments. Second-strand cDNA synthesis, which permits the removal of the second strand, was carried out using dUTP mix. The cDNA segments were ligated using indexed adapters, end repaired, and A-tailed. Uracil DNA glycosylase was used to purify the ligated cDNA products and eliminate the second-strand cDNA. The cDNA libraries were produced by PCR enrichment of purified first-strand cDNA. Agilent 2200 was used for quality control, and DNBSEQ-T7 was used for a 150 bp paired-end run of sequencing.

Prior to read mapping, adaptor sequences and low-quality reads were eliminated from the raw reads to provide clean reads. The Star was then used to align the clean reads to the mouse genome (mm10, Ensembl100). Gene counts were obtained using HTseq, and gene expression was ascertained using the RPKM technique.

We applied the DESeq2 algorithm to filter the differentially expressed genes, after the significant analysis, p-values and FDR analysis were subjected to the following criteria: (i) fold change > 2 or <0.5; (ii) FDR < 0.05.

To further clarify the biological consequences of the experiment’s differentially expressed genes, gene ontology (GO) analysis was carried out. The GO annotations were obtained from Gene Ontology (http://www.geneontology.org/↗), UniProt (http://www.uniprot.org/↗), and NCBI (http://www.ncbi.nlm.nih.gov/↗). Fisher’s exact test was used to determine which GO categories were significant (p-value < 0.05).

To determine the significant route of the genes that were differentially expressed based on the KEGG database, pathway analysis was employed. To choose the significant pathway, we used Fisher’s exact test, and a p-value of less than 0.05 was used to set the significance threshold.

All data in graphs are presented as the mean ± standard error of the mean (SEM). Statistical analyses were performed using GraphPad Prism (GraphPad Software, Boston, MA, USA). Two-group comparisons were made using a two-tailed unpaired Student’s t-test. Multiple comparisons were performed by either one- or two-way ANOVA, followed by Dunnett’s or Sidak’s multiple comparisons test used as indicated in the figure legends and methods section. A statistically significant difference was assumed at p < 0.05. For differential expression analysis of transcriptomic data, significance was set as an FDR-corrected p-value (q-value).

To determine the effects of resveratrol on aging-associated cognitive decline, we established an acute aging animal model by intraperitoneally injecting 8-week-old male C57BL/6J mice with D-galactose (D-gal, 100 mg/kg/day) for 8 consecutive weeks [36]. The D-gal model is widely used to induce accelerated brain aging in rodents. D-gal accumulates intracellularly when administered at high doses, overwhelming normal metabolism and generating excessive reactive oxygen species via galactose oxidation and advanced glycation end products. Sustained metabolic and oxidative stress activates senescence pathways (e.g., p21, p16), and promotes SASP, collectively leading to cognitive decline and other hallmarks of brain aging [37]. Beginning in the 6th week of D-gal administration, resveratrol was delivered by oral gavage. Throughout the treatment period, no significant change in body weight was observed in the animals (Supplementary Figure S1). After 3 weeks of resveratrol treatment the, cognitive and motor functions of those animals were evaluated. Behavioral assessments included open field, forced alternation, and rotarod assays (Figure 1A). At 16 weeks of age, we conducted an open field test to assess spontaneous locomotion in aging mice with or without resveratrol treatment (Figure 1B). The open field test is a standard assay that examines spontaneous locomotor activity and anxiety-like behavior in rodents. It exploits the innate conflict between exploratory drive and avoidance of exposed areas; rodents normally prefer to remain near walls and avoid the central zone, which is interpreted as an index of anxiety-like behavior in a novel open environment [38,39]. In this experiment, we quantified anxiety-related exploratory behavior using the number of central square crossings. Interestingly, resveratrol-treated aged mice demonstrated a pattern of exploration more similar to that of vehicle-injected mice, with a greater tendency to enter and traverse the central squares compared with untreated aged mice (Figure 1C,D). No differences were observed in total distance traveled or average speed among the three groups within the arena, suggesting that drug treatment did not affect spontaneous locomotor activity in this test (Figure 1E,F). Subsequently, the forced alternation task was used to evaluate spatial recognition memory (Figure 1G). Interestingly, the animal travel heatmaps revealed differential exploration behaviors among groups (Figure 1H). D-gal-induced aging animals showed no preference to enter the new arm when compared to vehicle-injected mice, suggesting a recognition memory deficit (Figure 1I). On the other hand, mice given resveratrol showed a preference for the new arm over the ones they knew, indicating that functional recognition memory had been restored (Figure 1I). It suggested that resveratrol could enhance the spatial working memory of aging animals. There was no significant difference observed among groups in total arm entries or total distance traveled, indicating that D-gal or drug treatment did not affect the free exploration activity of those animals in this experiment (Figure 1J,K). In addition, to evaluate the motor function of those mice, rotarod tests were performed. The rotarod assay is a widely accepted method for assessing motor coordination, balance, and neuromuscular performance in rodents [40,41]. It measures the ability of an animal to maintain position on a rotating rod, which reflects integrated motor control and endurance. In our study, intraperitoneal injection of D-gal significantly reduced both the latency to fall and the distance traveled on the rotarod compared with controls. This decline indicates impaired motor coordination and balance in the aging model. Mice treated with resveratrol showed a trend toward increased time and distance on the rotarod compared to the D-gal-induced aging mice (Figure 1L,M). Therefore, the above behavioral results suggested that resveratrol could ameliorate the cognitive and behavioral dysfunction induced by aging.

To assess the effects of resveratrol on brain aging, senescence-associated β-galactosidase (SA-β-gal) staining was performed in mouse brain sections. Elevated SA-β-gal activity is a well-established marker of cellular senescence. In the present study, D-gal-injected mice showed a marked increase in SA-β-gal staining in the hippocampal cornu ammonis 1 (CA1) and dentate gyrus (DG) subregions, as well as in the cerebral cortex, compared with vehicle-injected controls. Interestingly, resveratrol administration significantly reduced the SA-β-gal staining in both the hippocampus and cortex of aging mice (Figure 2A,B). This result suggests that resveratrol alleviated brain senescence induced by D-gal. Because oxidative DNA damage contributes to cellular senescence, we next examined 8-hydroxy-2′-deoxyguanosine (8-OHdG), a widely used marker of oxidative DNA damage. Elevated 8-OHdG levels were observed in the D-gal group, whereas resveratrol markedly reduced oxidative DNA damage (Supplementary Figure S2A,B), suggesting that the anti-senescence effect of resveratrol may involve suppression of oxidative stress. Accumulation of oxidative DNA lesions such as 8-OHdG is known to increase during brain aging. To evaluate cell-type-specific senescence, we assessed microglia as key immune regulators in brain aging. Confocal co-localization staining of the microglial marker P2ry12 with SA-β-gal showed enhanced β-galactosidase activity in microglia of D-gal-treated mice (Figure 2C). Resveratrol treatment significantly lowered this signal (Figure 2C,D). Consistently, co-staining of IBA1 with the senescence marker p21 further confirmed enhanced microglial senescence in the D-gal group and its attenuation following resveratrol treatment (Supplementary Figure S2C,D). Notably, supplementary data showed that neuronal senescence markers did not change significantly in resveratrol-treated mice (Supplementary Figure S2E). Collectively, these findings highlight that D-gal predominantly affects microglial aging in this experimental paradigm, and resveratrol preferentially attenuates microglial senescence rather than neuronal senescence under these conditions. In addition, we further assessed microglial activation using IBA1 and CD68 co-staining. The number of IBA1-positive microglia and CD68 signal intensity were significantly increased in D-gal-treated mice, consistent with microglial overactivation (Figure 2E). Resveratrol reduced both microglial numbers and CD68 expression (Figure 2E,F), supporting its role in suppressing microglial activation. Collectively, these results indicated that intraperitoneal injection of D-gal could induce a senescent state in microglia in the mouse brain, and resveratrol treatment could suppress the microglial activation and senescence.

To further investigate the effects of resveratrol on neuroinflammation in aging mice models, we selected to measure several key SASP factors, including Cxcl-10, Ccl-2, Ifi-44, Il-1β, and Il-18. We evaluated SASP expression in both the cortex and the hippocampus of aging mice and resveratrol-treated mice. RT-qPCR analysis showed that D-gal-induced aging mice exhibited elevated transcripts of Cxcl-10, Ccl-2, Ifi-44, Il-1β, and Il-18 (Figure 3A). Interestingly, resveratrol effectively suppressed the expression of these SASP factors in both brain regions, indicating suppression of age-related inflammatory signaling (Figure 3A). Next, to determine whether resveratrol effects could be recapitulated at the cellular level, microglial cells were transfected with double-stranded DNA (dsDNA) to induce senescence. Cytosolic dsDNA acts as a trigger of cellular senescence because it activates the innate cytoplasmic DNA-sensing pathway, leading to inflammatory signaling, and promotes key features of cellular senescence, including SASP factor expression [42]. After dsDNA exposure, optical microscopy and quantitative analysis revealed a marked increase in SA-β-gal activity, confirming induction of senescence (Figure 3B,C). Importantly, compared to dsDNA alone, resveratrol therapy significantly decreased the percentage of SA-β-gal-positive cells (Figure 3B,C). Consistent with these findings, immunofluorescence analysis of the senescence marker p21 showed that dsDNA treatment markedly increased the nuclear accumulation of p21. In contrast, resveratrol treatment significantly attenuated this dsDNA-induced nuclear accumulation of p21 (Figure 3D,E). We then examined the expression of SASP factors in the microglial cell model. dsDNA treatment significantly upregulated pro-inflammatory mediators, including the protein level of TNF-α (Figure 3F), and mRNA expression of Cxcl-10, Ccl-2, Ifi-44, Il-1β, and Il-18 (Figure 3G–K). Resveratrol effectively inhibited this SASP induction, restoring lower expression levels of these cytokines and chemokines (Figure 3G–K). These findings thus further support that resveratrol reduced both senescence and associated inflammatory responses in microglial cells.

To systematically explore the cellular phenotypes associated with resveratrol inhibition of STING, we performed RNA sequencing analysis in BV-2 cells. Differentially expressed genes (DEGs) analysis between the dsDNA group and the resveratrol-treated group revealed widespread transcriptomic changes. In total, 540 genes were highly expressed in the dsDNA group, whereas 61 genes were significantly downregulated following resveratrol treatment. Genes associated with the Cytosolic DNA-Sensing Pathway and neuroinflammation were annotated (Figure 4A). Gene set enrichment analysis (GSEA) revealed significant activation of immune and DNA-sensing pathways in the dsDNA group. Enriched signatures included Allograft Rejection (NES = 1.678, FDR < 0.001), Graft-versus-Host Disease (NES = 1.680, FDR < 0.001), and Cytosolic DNA-Sensing Pathway (NES = 1.660, FDR < 0.001). Additional activation of the RIG-I-Like Receptor Signaling Pathway and Autoimmune Thyroid Disease was also observed (Supplementary Figure S3A,B). Conversely, resveratrol treatment markedly suppressed these pathways. Notably, Allograft Rejection (NES = −1.689, FDR = 0.0048), Graft-versus-Host Disease (NES = −1.713, FDR = 0.0026), and the Cytosolic DNA-Sensing Pathway (NES = −1.667, FDR = 0.0064) were downregulated following resveratrol treatment (Figure 4B; Supplementary Figure S3C). Pathways involved in antigen processing and presentation also showed reduced enrichment. Given that cGAS and STING function as core sensors of cytosolic DNA and drive downstream immune signaling, the suppression of DNA-sensing signatures by resveratrol suggests targeted modulation of this pathway. Recent studies have established that cGAS-STING activation occurs in response to cytosolic DNA accumulation, DNA damage, and neuroinflammation [6]. Furthermore, a Venn diagram identified 46 overlapping DEGs between comparisons (Figure 4C). These overlapping genes were visualized in a clustered heatmap to illustrate expression patterns (Figure 4D). To further characterize the functional relevance of these genes, we performed gene ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analyses on the 46 overlapping genes. GO analysis showed that molecular functions were enriched for cytokine activity, CARD domain binding, and type I interferon receptor binding (Figure 4E). Biological processes included response to bacterium, immune system processes, and inflammatory response. Enriched cellular components were the extracellular space, the extracellular region, and the external side of the plasma membrane (Supplementary Figure S3D–G). KEGG pathway analysis highlighted significant enrichment in the TNF signaling pathway and the Cytosolic DNA-Sensing Pathway, further linking the overlapping gene set to immune signaling and DNA recognition (Figure 4F). To validate transcriptomic findings, selected genes related to the Cytosolic DNA-Sensing Pathway and neuroinflammation were assessed by RT-qPCR in BV-2 cells. These validations confirmed the expression trends observed in RNA-seq (Figure 4G). Collectively, these data indicate that resveratrol markedly alters gene expression programs associated with the cGAS-STING signaling pathway. The results support a model in which resveratrol attenuates cellular senescence and neuroinflammatory signaling by suppressing cytosolic DNA-sensing and related immune pathways.

To further determine whether resveratrol modulated cGAS-STING signaling in the aging brain, we examined key pathway components in distinct brain regions. Western blot analysis showed that the phosphorylated STING (pSTING) levels were significantly elevated in aging model mice compared with vehicle-treated controls (Figure 5A–D). Resveratrol treatment markedly reduced pSTING expression toward levels observed in control mice. We next evaluated phosphorylated TBK1 (pTBK1), a downstream effector of activated STING. Aging mice exhibited pronounced increases in pTBK1 relative to controls. Resveratrol administration substantially lowered pTBK1 levels in both the cortex and hippocampus (Figure 5A–D). Notably, the cortex showed larger age-related increases in pSTING and pTBK1 than the hippocampus, consistent with regional vulnerability to aging processes. Immunofluorescence staining further supported these findings. Compared with aging controls, resveratrol-treated mice displayed reduced pSTING and pTBK1 immunoreactivity in affected brain regions (Figure 5E,F). Therefore, these findings show that aging-related activation of the cGAS-STING pathway occurs in the mouse brain and can be effectively inhibited by resveratrol. This indicates that inhibition of cGAS-STING signaling may represent a key mechanism by which resveratrol mitigates aging-related neuroinflammatory responses.

To extend the in vivo findings of resveratrol-mediated attenuation on cGAS-STING signaling in brain tissue, we next investigated the effects of resveratrol at the cellular level. In these experiments, dsDNA was used to induce STING pathway activation, and we subsequently assessed the effect of resveratrol on the phosphorylation levels of key signaling proteins following activation. Western blot analysis showed that resveratrol treatment effectively inhibited the activation of the STING signaling pathway induced by dsDNA in BV-2 cells (Figure 6A,B). Specifically, the phosphorylation levels of STING and its key downstream protein TBK1 were significantly increased after dsDNA induction, but they were significantly reduced by resveratrol treatment, indicating that resveratrol effectively inhibited STING and its downstream signaling pathway (Figure 6A,B). This inhibition resembled the effect of the STING inhibitor H-151, indicating that resveratrol effectively attenuates the STING-mediated signaling pathway. Upon STING pathway activation, STING oligomerizes and recruits TBK1, leading to TBK1 undergoing autophosphorylation (pTBK1), and in turn phosphorylates STING (pSTING) on key serine residues (e.g., S366), creating a docking site for interferon regulatory factor 3 [13]. These phosphorylated species show pronounced punctate perinuclear staining rather than the diffuse endoplasmic reticulum (ER) and cytosolic pattern seen in resting cells [43]. Thus, we further employed immunofluorescence staining to visualize the spatial distribution and cellular localization of the two key signaling proteins in situ. Interestingly, there was obvious punctate perinuclear staining of pTBK1 and pSTING in the dsDNA-treated cells, instead of diffuse ER localization seen in resting cells (Figure 6C,E). Conversely, both resveratrol and H-151 treatment significantly suppressed pSTING and pTBK1 signal intensity and altered their subcellular distribution induced by dsDNA (Figure 6C–F). Thus, these results corroborate our prior transcriptomic data identifying the cGAS-STING pathway as a major altered signaling cascade in this cell model.

STING acts as a scaffold protein that recruits downstream signaling components upon binding cyclic dinucleotides. Its translocation from the ER to the Golgi apparatus is a defining step in signal propagation. To investigate how resveratrol modulates STING signaling, we examined the subcellular localization of STING relative to the Golgi. Immunofluorescence assays were performed to assess the co-localization of STING with the Golgi marker GM130. In BV-2 cells, two hours of dsDNA stimulation resulted in a pronounced accumulation of STING around the Golgi, as evidenced by increased overlap with GM130 signals (Figure 7A). This finding is consistent with activation of the pathway and the trafficking of STING toward its signaling platform. In contrast, pre-treatment with resveratrol prior to dsDNA stimulation markedly reduced the amount of STING localized to the Golgi apparatus (Figure 7A). Instead, STING immunosignals were more diffusely distributed throughout the cytoplasm, indicating that resveratrol blocked the translocation of STING to the Golgi (Figure 7A,B). This effect on intracellular trafficking was distinct from that of the STING inhibitor H-151, which did not alter STING localization under the same conditions (Figure 7A,B). To further determine whether the inhibitory effect of resveratrol occurs upstream or downstream of STING activation, cells were directly stimulated with the STING agonist 2′,3′-cGAMP to bypass upstream DNA-sensing events. Even under these conditions, resveratrol treatment significantly suppressed STING translocation from the ER to the Golgi (Supplementary Figure S4A,B). Collectively, these findings suggest that resveratrol blocked a critical step in STING signal transduction by preventing its trafficking from the ER to the Golgi (Figure 7C). This observation reinforced our earlier transcriptomic identification of cGAS-STING signaling as a major altered pathway by resveratrol treatment.