cGAS–STING drives ageing-related inflammation and neurodegeneration

Wild-type C57BL/6J mice and Sting1^−/− (STING-deficient) (025805) mice (aged 19–20 months) were purchased from The Jackson Laboratory. Mice were single-housed to prevent aggression-related injuries and fed standard chow diets. Aged mice were treated either with vehicle or 750 nmol H-151 in 200 µl PBS 5% Tween-80 as described in Fig. 1. In brief, aged mice were injected intraperitoneally with 750 nmol H-151 5 days a week for 8 weeks. After 8 weeks of treatment break, the mice were treated for another 8 weeks and then analysed using physical or cognitive functional tests. The mice were euthanized and tissues were collected for further analysis. Pharmacokinetic and tissue distribution experiments were performed by Pharmaron. Sting1^−/− mice and wild-type litters were bred and aged within the institute.

Cgas^R241E mice were generated using CRISPR–Cas9-mediated gene editing using a long single-strand oligo in C57BL/6J zygotes. The long single-stranded oligo contains exon 2 in reversed orientation and the R241E point mutation. This exon is flanked with lox71 and lox66 recombination sites in the opposite orientation and homology arms. The long single-strand was injected in mouse zygotes together with Cas9 protein (Integrated DNA Technologies) and two designed gRNAs (5′-TTTATAGGCACCCTATGTACAGG-3′ and 5′-CTGACCGCACGACTTACCCTGGG-3′) targeting the intronic region surrounding exon 2. Cgas^R241E mice were bred on B6;129Gt(ROSA)26Sor^{tm1(cre/ERT)Nat}/J (The Jackson Laboratory, 004847) and C57BL/6-Tmem119^{em1(cre/ERT2)Gfng}/J (The Jackson Laboratory, 031820) to generate Rosa26-creER^T2-Cgas^WT/R241E mice and Tmem119-creER^T2-Cgas^WT/R241E (mg-Cgas^R241E) mice.

To induce microglial expression of Cgas^R241E in mg-Cgas^R241E) mice, tamoxifen (75 mg per kg body weight) was intraperitoneally injected into 8–10-week-old mice once every 24 h on 5 consecutive days. Some mice also received 750 nmol H-151 daily, starting with tamoxifen (4-OHT, Sigma-Aldrich, T5648) injections. Within 1 week after the last administration of tamoxifen, some mice were analysed using cognitive tests and euthanized for tissue analysis.

To test the in vivo specificity of H-151, mice (aged 8–10 weeks) received 15 mg per kg of a STING agonist (DMXAA; InvivoGen) intraperitoneally. H-151 (750 nmol) was administered to one group 30 min before the injection of the STING agonist and, 3 h later, the mice were euthanized for tissue collection.

The cages were kept at 18–24 °C ambient temperature under 40–60% humidity. The mice were maintained under a 12 h–12 h light–dark cycle from 06:00 to 18:00. Food and water were available ad libitum. Mice were euthanized by anaesthetization by an injection of an overdose of sodium pentobarbital (150 mg per kg). Animal experiments were approved by the Service de la Consommation et des Affaires Vétérinaires of the canton of Vaud or obtained by IFM Therapeutics and were performed in accordance with the respective legal regulations.

This test was performed to assess hippocampus-dependent spatial long-term memory and learning. The Morris water maze consists of a circular tank filled with water containing a non-toxic coloured paint to contrast with the mouse fur, and maintained at room temperature (23 ± 1 °C). To escape the water, the mice had to find a platform. This escape platform was submerged and was not visible for the animals. The tank was placed in a room with constant extra-maze visual cues and ambient light throughout all the experiment. Mice were trained during a maximum of 5 days to find the escape platform using visual cues as spatial points of reference. Each day, each mouse went through a maximum of 6 trials. The maximum duration of each trial was 120 s. At the end of each trial, the mouse was placed back into its home cage under a heating lamp for recovery and resting time.

On each day of the experiment, the animals were moved to the experimental room and left undisturbed for 30 min in their home cage. On the first day, the mice were placed into a conditioning chamber, and a tone was presented followed by a foot shock. The tone–foot-shock pairing was repeated a maximum of four times during the training session. On the second day, the mice were re-exposed to the context with no tone, and the freezing response was monitored. Mice only received a foot shock on the first day of the experiment and the foot shock lasted a maximum of 2 s with a maximum intensity of 0.5 mA.

The assessment of limb strength in aged mice was performed using a commercially available automatic grip strength meter. This test is based on the natural tendency of the mouse to grasp a bar or a grid when it is suspended by the tail and gently pulled backwards. The peak resistance force after the mouse lost grip from the grid was recorded as it was being pulled away from the device.

For the exercise test, the mice were placed on a treadmill. Plexiglas walls limit the treadmill and, next to the running lane, an electric grid dispenses small shocks when mice touch it to force them to run. Mice were habituated to the treadmill for 5 min before the test. Mice started running at a very moderate speed (15 cm s⁻¹). The speed gradually increased at 3 cm s⁻¹ every 12th minute. The device automatically monitored the run distance and the number of shocks received or the time of air-puff stimulation. When a mouse received more than five shocks in two consecutive minutes, it was considered that the mouse reached exhaustion and the mouse was placed back into its home cage with full access to food and water. The shocks (0.1 mA) were of very low intensity, enough to cause an unpleasing feeling (like a sting) to keep them running and far from resulting in damage to their integrity.

We performed whole-mount immunostaining followed by confocal microscopy to examine the infiltration of F4/80⁺ macrophages in adipose tissue. The mice were perfused with PBS before the collection of inguinal white adipose tissue. The tissue depots were subdivided into 0.5–0.75 cm³ sized pieces, fixed in 1% paraformaldehyde (PFA) for 30 min at room temperature with gentle rocking, and washed in PBS three times for 10 min under gentle rocking. The tissue was blocked for 30 min in 5% BSA in PBS, then incubated with primary anti-F4/80 antibodies (Thermo Fisher Scientific, A3-1, MA1-91124) diluted 1:100 in blocking buffer overnight at 4 °C. The samples were washed with times with PBS before incubation with secondary anti-rat-Alexa 488 antibodies (Thermo Fisher Scientific, A-21208) diluted 1:350 in blocking buffer for 1 h at room temperature with gentle rocking. After three washes with PBS, the samples were incubated for 20 min in a lipid- and nucleus-staining cocktail composed of 1 µg ml⁻¹ DAPI (Sigma-Aldrich, D9542) and 0.25 µg ml⁻¹ BODIPY (Thermo Fisher Scientific, D3922) and washed three times in PBS before imaging. The images were taken on an inverted scanning confocal microscope (Zeiss LSM 700 Inverted) and assembly of 3D reconstructions was accomplished by taking z-stack images.

For mouse tissue experiments, formalin-fixed paraffin-embedded mouse tissue blocks were cut into 3 μm sections for immunohistochemistry staining and 6 μm sections for immunofluorescence staining and placed onto slides. The sections were first deparaffinized and rehydrated, then heat-induced epitope retrieval was performed and the sections were permeabilized with PBS 1% (v/v) Triton X-100. The samples were stained with primary antibodies (IBA1, Abcam, ab178846, rabbit, 1:10,000; IBA1, Abcam, ab5076, goat, 1:300; Mac3, Becton Dickinson, 553322, 1:150; GFAP, Agilent, Z033429-2(AGI), 1:100; NeuN, Merck, MAB377, 1:100; synaptophysin, Cell Signaling, 36406S, 1:500; and B2M, Abcam, ab75853, 1:100) overnight at 4 °C. For immunofluorescence analysis, the sections were then stained with fluorescently labelled secondary antibodies (anti-rabbit-AF488 (A-21206, 1:300), anti-rabbit-AF568 (A-10042, 1:250), anti-goat-AF488 (A-11055, 1:200), all from Thermo Fisher Scientific) for 90 min at room temperature. For immunohistochemistry analysis, the sections were stained with HRP-conjugated secondary antibodies (anti-rabbit-HRP, 711035152, 1:300; and anti-mouse-HRP, 715035150, 1:300; both from Jackson ImmunoResearch) followed by DAB staining and Harris counterstaining. Images from immunohistochemistry staining were acquired using either a Zeiss AxioScan 7 slide scanner or Olympus VS120 whole-slide scanner and collected using the Olympus OlyVIA software. Data were manual counts of parenchymal IBA1⁺ microglial cells, Mac3⁺ cells and NeuN⁺ cells. Cell counts were normalized to the hippocampal area. Images from immunofluorescence staining were analysed using wide-field fluorescence microscope (Zeiss Axioplan) and processed using ImageJ.

Microglia isolated from young and aged mice were imaged using the Leica SP8 confocal microscope with a white light laser. Slices or microglia were imaged with a 63× oil-immersion objective (NA = 1.45, Leica), with the standard settings across samples labelled the same way. z-Stacks were subsequently deconvolved using the Huygens Deconvolution Software (Scientific Volume Imaging), and 3D views were rendered using IMARIS (Bitplane). The following parameters were used for the generation of the surfaces in IMARIS for the visualization of pSTING and IBA1. Hippocampal slices from the aged mouse in Extended Data Fig. 6b, pSTING: manual threshold: 90, smoothing: 0.141 μm. Hippocampal slices from the aged mouse in Extended Data Fig. 6b, IBA1: manual threshold: 25.7, smoothing: 0.5 μm. Isolated microglia from young and aged mouse brains in Extended Data Fig. 6c, pSTING: manual threshold: 110, smoothing: 0.120 μm.

For quantitative analysis of the proportion of pSTING⁺ and IBA1⁺ cells, maximum intensity projection using the s.d. projection type was rendered using ImageJ. A cut-off intensity threshold of 32 was applied to the pSTING channel, and the number of pSTING⁺ cells and IBA1⁺ cells and the total number of cells in the field of view was enumerated.

For IMARIS images of microglia, morphometrical analysis of microglial cells was conducted as previously described⁵⁸. In brief, the animals were perfused with ice-cold PBS. The brains were fixed in 4% PFA overnight. The tissue was then dehydrated in 30% sucrose and embedded in Tissue Tek (Sakura). Cryostat sections (thickness, 50 µm) were blocked in 5% BSA supplemented with 0.5% (v/v) Triton X-100 in PBS. The sections were then incubated with anti-IBA1 primary antibodies (Abcam, ab178846, 1:1,000) for three nights. Donkey anti-rabbit Alexa Flour 568 antibodies (Thermo Fisher Scientific, A10042, 1:500) were used as the secondary antibodies and were applied for one night. Nuclei were visualized using DAPI. Imaging was performed using the SP8 confocal microscope (Leica) and a 63× objective (step size, 0.4 µm). The cells were visualized in 3D and reconstructed using IMARIS (v.9.6, Bitplane).

For neuronal culture staining, cell culture coverslips were fixed in 4% PFA for 10 min, washed twice with PBS, then permeabilized in 2% saponin, 0.1% (v/v) Triton X-100 in PBS for 15 min at room temperature, and blocked in PBS with 1% BSA for 1 h at room temperature. The coverslips were incubated overnight at 4 °C with primary antibodies against MAP2 (Sigma-Aldrich, M4403, 1:250) and IBA1 (Abcam ab178846, 1:250). The samples were washed with PBS and incubated with secondary antibodies (anti-rabbit-AF568 (A-10042), anti-mouse-AF488 (A-11001), both 1:250) for 90 min at room temperature. Cell nuclei were stained by incubating with 5 µg ml⁻¹ DAPI for 10 min, and the coverslips were mounted with fluorescent mounting medium (Agilent, S302380). Images were acquired using a wide-field fluorescence microscope (Zeiss Axioplan). The same settings were applied to all images within the same experiments. All analyses were performed using ImageJ. A list of all of the antibodies used is provided in Supplementary Table 2.

All material used in this study was obtained from the Cohort of Obese Patients of Lausanne with an ethically approved license by the commission of Vaud Canton (CER-VD project PB_2018-00119). The coded samples were collected under signed informed consent conforming to the guidelines of the 2000 Helsinki declaration. Human adipose tissue was resected during gastric bypass surgery from participants with obesity. Three of the participants were male, and eight of the participants were female. The mean age of the participants was 48.7 years old; s.d., 9.7; range, 30–60. No participant was known to have a malignancy. Greater omental adipose tissue was obtained from each participant. Adipose tissue was cut into small pieces, washed with PBS three times and cultured in medium containing 1 mM sodium pyruvate (BioConcept), 2 mM glutamine (Life Technologies), MEM vitamins (Life Technologies), MEM non-essential amino acids (Thermo Fisher Scientific) and antibiotics. The explants were treated with H-151 (daily, 1–2 µM), DMSO, or with 20 µM quercetin and 1 µM dasatinib (Chemie Brunschwig). After incubation for 6 days, part of the adipose tissue was fixed before SA-β-galactosidase staining (with or without tissue clearance), and the remaining—if sufficient—was collected for RNA isolation. The conditioned medium was collected (in three to five replicates per patient) for ELISA analysis of IL-6, IL-8 and MCP-1.

The adipose tissue from patients was immersed in 4% formalin (Sigma-Aldrich) for 2 days on a rotary shaker at 4 °C, then washed for 4 days on a rotary shaker in PBS at 4 °C. The tissue was washed three times for 30 min in PBS on an orbital shaker at room temperature, then sectioned to 500 µm slices with a vibratome. The sample was permeabilized in 2% PBS-T (2% Triton X-100 in PBS solution containing 0.05% sodium azide) for 2 days on an orbital shaker at room temperature, then cleared in RapiClear (RapiClear 1.52, SUNJin Lab, RC152001) overnight at room temperature. The cleared tissue was mounted in fresh RapiClear reagent and imaged using a confocal microscope (Zeiss LSM 700 Inverted).

Both cells and tissue explants were washed twice with PBS, then fixed for 5 min with 2% formaldehyde (Sigma-Aldrich) and 0.2% glutaraldehyde (Sigma-Aldrich) in PBS at room temperature, washed with PBS and then incubated overnight at 37 °C in staining solution with 40 mM citric acid NA phosphate, 5 mM K₄Fe(CN)₆, 5 mM K₃Fe(CN)₆ (Fluka analytical, 34272), 150 mM NaCl, 2 mM MgCl₂ and 1 mg ml⁻¹ X-Gal (Roche, R0404) in water. The samples were washed twice with PBS before imaging by microscopy (Zeiss Axio Vert.A1).

Human and mouse tissues were snap-frozen in liquid nitrogen and stored at −80 °C until processing. RNA was isolated using the TRIzol–chloroform (Thermo Fisher Scientific, 15596018) method and a tissue homogenizer (Thermo Fisher Scientific). For cells, RNA was isolated using the RNeasy Mini Kit (Qiagen, 74004) (cell lines) or the RNeasy Micro Kit (Qiagen, 74104) (primary microglia), according to the manufacturer’s instructions.

RNA was isolated from the control and irradiated WI-38 cells, treated daily or not from day 10 to day 20 with STING inhibitor H-151 (0.5 µM) using the RNeasy Mini kit (Qiagen, 74004). RNA was further processed for sequencing by the Gene Expression Core Facility GECF at EPFL. mRNA-seq libraries were prepared using the TruSeq mRNA stranded LT (Illumina kit). The samples were sequenced using the NextSeq 500 system sequencing with 1 × 75 cycle (single read), high output mode (expected a minimum of 400 million reads), chemistry v2. Sequencing data were processed using HTSstation online software from the Bioinformatics and Biostatistics Core Facility at EPFL. Heat maps were produced from normalized expression data using Cluster 3.0 for computation and JTreeview for visualization.

RNA isolated from mouse brains was processed as follows: RNA quality was controlled on the TapeStation 4200 (Agilent), confirming that all were of good quality (scores > 8.1). Libraries for mRNA-seq were prepared using the stranded mRNA ligation method (Illumina) starting from 800 ng RNA, according to the manufacturer’s instructions. Libraries, all bearing unique dual indexes, were subsequently loaded onto the NovaSeq 6000 flow cell (Illumina) and sequenced according to the manufacturer instructions, yielding for each sample at least 40 million pairs of 60-nucleotide-long reads. Reads were trimmed of their adapters using BCL Convert (v.3.9.3; Illumina) and quality-controlled using fastQC (v.0.11.9).

For the fibroblasts, RNA was mapped to the human genome assembly hg38 (gencode v36, Ensembl 102) using STAR aligner (v.2.7), and counts were generated with HTSeq Count. In total, 13,006 protein-coding genes (cpm > 1 in at least 2 samples) were retained for analysis. Differential gene expression analysis was performed in R (v.4.0) using voom-Limma (v.3.28).

For bulk hippocampus, RNA was mapped to the mouse genome assembly GRCm39 (release 109) using STAR aligner (v.2.7.10b), and counts were generated using featureCounts (v.2.0.1). DEG analysis was performed using the R (v.4.2.2) package DESeq2 (v.1.38.2).

Nuclei from mouse brain were extracted by homogenizing mouse brain tissues in Nuclei EZ Lysis Buffer (Millipore Sigma) using a douncer. Nuclei were labelled with DAPI (10 µg ml⁻¹, Sigma-Aldrich) and sorted for sequencing. To enrich microglial cells, in addition to DAPI staining, nuclei were stained with anti-RBFOX3/NeuN-647 (Novus Biologicals) and anti-Olig2-488 (Merck), and DAPI⁺NeuN⁻Olig2⁻ cells were sorted for analysis. The gating strategy used for microglial nucleus sorting is depicted in Supplementary Fig. 2. Sorted mouse brain cell nuclei were washed and resuspended in PBS 1% BSA supplemented with 0.2 U µl µg⁻¹ RNase inhibitor, checked for the absence of significant doublets or aggregates and loaded into a Chromium Single Cell Controller (10x Genomics) in a chip together with beads, reverse transcription master mix reagents and oil to generate single-cell-containing droplets. Single-cell gene expression libraries were then prepared using the Chromium Single Cell 3′ Library & Gel Bead Kit v3.1 (PN-1000268) according to the manufacturer’s instructions (protocol CG000315 Rev C). Quality control was performed using the TapeStation 4200 (Agilent) and the QuBit dsDNA high sensitivity assay (Thermo Fisher Scientific) according to the manufacturer’s instructions. The sequencing libraries were loaded onto an Illumina NovaSeq Flow Cell and sequenced using read lengths of 28 nucleotides for read1 and 90 nucleotides for read2, at a depth of around 80,000 reads per cell. Illumina BCL convert was used to demultiplex reads, after which 10x Genomics Cell Ranger Single Cell Software Suite (v.7.0.0) was used to perform barcode processing and 3′ gene counting using 10x Genomics custom annotation of mouse genome assembly mm10.

For the microglial and hippocampus datasets, the median sequencing depth was 107,267 and 100,912 reads per cell, with 91.5% and 94.6% of the reads confidently mapping to the genome and 9,748 and 21,585 nuclei captured, respectively. Sample processing and analysis were performed using the R package Seurat⁵⁹ (v.4.1.1.9003). For each dataset, we filtered out low-quality and potential doublet cells on the basis of violin plot distributions of genes, UMI counts and mitochondrial gene expression (as determined using the Seurat PercentageFeature function) per cell. Filters were set on a per-experiment basis as sequencing run and cell type significantly influence filtering cut-offs. For both single-nucleus microglial and whole-hippocampus datasets, only genes detected in a minimum of three cells, and cells with a minimum of 200 genes were included. For single microglia, cells with ≥5% mitochondrial gene expression, ≥4,500 genes and ≥10,000 UMIs were excluded and, in single hippocampi, cells with ≥5% mitochondrial gene expression, ≥8,000 genes and ≥45,000 UMIs were excluded. After filtering, we retained 5,360 (1,001, 514, 924, 1,468 and 1,453 per sample, 1:2 Cgas^WT/R241E, 3:5 Cgas^WT/WT) and 21,500 (2,076, 2,655 2,383, 2,769, 3,550, 3,422, 2,547 and 2,098 per sample, 1:4 Cgas^WT/R241E, 5:8 Cgas^WT/WT) nuclei per dataset, respectively. After merging, the samples were evaluated for potential batch effects. As we observed individual sample-based cell type cluster offsetting, probably a result of mean expression differences, we performed the Seurat Integration⁶⁰ preprocessing workflow with a SCTransform (v2) normalization (negative binomial model) for cell clustering⁶¹, both for the microglial and hippocampus datasets. Integration was performed on the basis of 3,000 anchor genes, and subsequent reductions in dimensionality were performed on the basis of the top 30 principal components. Unsupervised clustering was performed at a resolution of 1 and 0.8 for the microglia and hippocampi, respectively.

Cluster identity was determined on the basis of known cell type marker genes and cross-referenced to The Human Protein Atlas single-cell reference data (https://www.proteinatlas.org/↗) (Supplementary Table 6). Cell identity gene expression scores were assigned to individual cells on the basis of the overall expression of given gene lists using the AddModuleScore function (based on a function of Seurat⁶²) (Extended Data Fig. 9a,f) and visualized using the FeaturePlot_scCustom function from the R package scCustomize (v.0.7.0), allowing for subsequent analysis of specific cell types (such as microglia, oligodendrocytes) (Supplementary Table 5). In the microglial dataset, contaminating non-microglial cells were filtered out on the basis of the microglial score, and the remaining cells were further subsetted and reclustered, as described above (Extended Data Fig. 9a). Particular attention was given to the removal of MRC1⁺ macrophages, which often display a similar expression profile to microglia, and can be an unwanted source of variation. Unsupervised clustering was performed at a resolution of 0.8 for the subsetted dataset. After subletting, an average of 1,571 and 3,026 genes, 2,762 and 9,112 UMIs, and 0.069% and 0.028% mitochondrial gene content were found per nuclei in the microglial and hippocampal datasets, respectively. To identity subtypes of microglia, we used the same method as described above, using the Seurat AddModuleScore function for gene lists related to IFN, DAM, neurodegenerative and H-MG (Supplementary Table 6—gene sets were derived from refs. ^38,39, Fig. 4e and Extended Data Fig. 9b). After cluster cell type identification for both datasets, differential gene expression analysis was performed using the FindMarkers function of Seurat, using the MAST test (v.1.22.0)⁶³, which implements a two-part hurdle model, with a Bonferroni correction. Data were normalized using the NormalizeData (LogNormalize) function in Seurat before differential gene analysis. Microglial cluster identity DEGs for DAM, IFN and neurodegenerative-associated microglia were calculated with the H-MG used as a reference.

For the comparison to the aged¹³ and disease-associated microglia⁴⁰, raw datasets and metadata were acquired from the Gene Expression Omnibus (GEO: GSM4505405↗ and GSE127892↗, respectively). Data were processed using R and the Seurat package. For the aged dataset, raw counts were extracted for 24-month-old microglia only, and merged with the relevant metadata, including previously identified Louvain clusters. In the case of the disease-associated microglial dataset, no cluster annotations were readily available; as such, the analysis was recreated from the raw counts using the materials and methods of the published article. We identified all relevant clusters from the previous article, as well as an additional subcluster of ARMs that showed high expression of MHC class I genes (H2-Aa, H2-Ab1, H2-Eb1), representative of an activated microglial state, which we labelled F_MHC-ARM. We individually integrated these datasets with our final annotated microglial dataset using Seurat’s standard integration workflow with 2,000 variable features, 11 PCA dimensions for clustering, and a clustering resolution of 0.5 and 0.3 for the aged and disease datasets, respectively. After integration, we used the AddModuleScore function as described above to assign an IFN, DAM and neurodegenerative score and compared clustering of the previously annotated cell types between datasets.

For mouse organ tissue, cells and human patient tissue, RNA was reverse-transcribed using the RevertAid First Strand cDNA Synthesis reagents (Thermo Fisher Scientific), and RT–qPCR was performed in duplicates or triplicates using the Maxima SYBR Green Master Mix (Thermo Fisher Scientific) on QuantStudio 6/7 qPCR instruments. A list of the primer sequences used is provided in Supplementary Table. 1

Cell pellets were lysed in a lysis buffer containing 10 mM Tris (pH 8), 1 mM EDTA, 0.5 mM EGTA, 1% (v/v) Triton X-100, 0.1% sodium deoxycholate, 0.1% SDS, 140 mM NaCl, protease and phosphatase inhibitors (protease inhibitor cocktail I, animal-free, CALBIOCHEM). Protein concentration was measured using the BCA Pierce Protein assay kit (Thermo Fisher Scientific) and normalized to the lowest concentration. The primary antibody was incubated in 2.5% milk or BSA in PBS-T overnight at 4 °C. The secondary anti-mouse or anti-rabbit HRP-conjugated antibodies were incubated for 1 h at room temperature. Proteins were visualized with the enhanced chemiluminescence substrate ECL (Pierce, Thermo Fisher Scientific) and imaged using the ChemiDoc XRS BioRad Imager and Image Lab Software. Imaging was performed in two channels: chemiluminescence and colorimetry. Uncropped images are provided in Supplementary Fig.. A list of all of the antibodies used is provided in Supplementary Table. 1 2

Splenocytes: to isolate mouse splenocytes, the spleen was mashed through a 70 µm cell strainer using the plunger end of a syringe, cells were resuspended in medium and centrifuged at 800g for 3 min. Red-blood cells were lysed in 2 ml ACK lysing buffer (Life Technologies, A1049201) for 5 min at room temperature, inactivated by 20 ml complete medium, centrifuged at 800g for 3 min and washed once more with PBS. The pellet was kept at −80 °C until further analysis.

Tail-tip fibroblasts: a 2 cm portion of the mouse tail was cut, incubated in ethanol 70% for 5 min and air dried. The tissue was minced with scissors and incubated with digestion buffer containing 2.5 mg ml⁻¹ collagenase D (Sigma-Aldrich, 11088866001) and 1.25 mg ml⁻¹ Pronase (Millipore, 53702) in complete medium for 90 min at 37 °C, shaking at 200 rpm. Single cells were isolated by filtering through a 70 µm cell strainer, centrifuged at 580g for 7 min, resuspended in 10 ml medium and plated into a 10 cm dish.

Microglia: The whole brain from adult mice was removed and minced into small pieces. The brain pieces were further dissociated using the Adult Brain Dissociation Kit (Miltenyi Biotec, 130-092-628) according to the manufacturer’s instructions. Dissociated cells were filtered through a 100 µm cell strainer. The debris was removed by gradient centrifugation using the Debris Removal Kit (Miltenyi Biotec, 130-109-398). The cells were labelled with anti-CD11b-beads (Miltenyi Biotec, 130-093-634) and positively selected by magnetic sorting.

Human fibroblast cells (BJ, WI-38) were cultured under 5% CO₂ and 5% O₂ at 37 °C in Dulbecco’s modified Eagle medium (DMEM) (Life Technologies) containing 10% (v/v) FCS, 1% (v/v) penicillin (100 IU ml⁻¹)–streptomycin (100 μg ml⁻¹). Mouse tail-tip fibroblasts were cultured under 5% CO₂ and 20% O₂ at 37 °C. Mouse bone-marrow-derived macrophages were generated using L929-cell-conditioned medium as a source of granulocyte/macrophage colony-stimulating factor. Mouse cells (BV2) were cultured under 5% CO₂ and 20% O₂ at 37 °C in minimum essential medium (MEM) GlutaMAX Supplement (Gibco, 41090036) containing 10% (v/v) FCS, 1% (v/v) penicillin (100 IU ml⁻¹)–streptomycin (100 μg ml⁻¹). Cells were repeatedly tested for mycoplasma using specific primers. Cell lines were purchased from ATCC (BJ, CRL-4001, WI-38, CCL-75) and amsbio (BV-2, AMS.CL-0493-1).

Isolated microglia were cultured in DMEM (Life Technologies) containing 10% (v/v) FCS, 1% (v/v) penicillin (100 IU ml⁻¹)–streptomycin (100 μg ml⁻¹), GlutaMAX (Thermo Fisher Scientific, 31331028), GM-CSF (50 ng ml⁻¹, ImmunoTools, 12343123) and M-CSF (100 ng ml⁻¹, ImmunoTools, 12343112). Cells were treated with H-151 (0.5 μM daily) or VBIT-4 (10 μM every other day, Selleckchem, S3544) for 4 days before gene expression analysis. The relative survival of primary microglia treated or not with H-151 (1 μM) was assessed 24 h after treatment using the CellTiter-Blue Cell Viability Assay (Promega, G808A), according to the manufacturer’s instructions.

sgRNA (5′-ATATTCTTGTAGCTCAATCC-3′) was cloned into px458-GFP vector (Addgene, 48138). BV2 cells were transfected with the generated vector. Then, 2 days after the transfection, live GFP-positive single cells were sorted into 96-well plates. Growing cells were tested for deletion using immunoblot analysis.

Primary mouse cortical neuron cultures derived from wild-type mice were cultured on 35-mm-well plates (5 × 10⁵ cells per well) coated with Cultrex poly-lysine (Trevigen) in medium consisting of Neurobasal (Invitrogen), B27 supplement (Invitrogen), l-glutamine (Invitrogen) and penicillin–streptomycin (Invitrogen). For young/aged microglial co-culture experiments, primary microglia isolated from young (8 to 12 weeks old) and aged (24 to 27 months olds) mice were added 1:1 onto hippocampal neurons, 4 h after they were seeded onto coverslips, and additionally treated with anti-TNF (Bio X Cell, XT3.11, BE0058, 25 μg ml⁻¹) or anti-IFNAR (Sigma-Aldrich, MARI-5A3, 10 μg ml⁻¹) neutralizing antibodies. For microglia Cgas^R241E co-culture experiments, primary microglia isolated from Rosa26-creER^T2-Cgas^WT/R241E mice were added 1:2 onto cortical neurons, 4 h after they were seeded onto coverslips, and 4-OHT (Sigma-Aldrich, T176, 600 μM) and mouse anti-TNF neutralizing antibodies (Bio X Cell, clone XT3.11, BE0058, 25 μg ml⁻¹) or anti-IFNAR neutralizing antibodies (Sigma-Aldrich, MARI-5A3, 10 μg ml⁻¹) were added or not to the medium and the cells were co-cultured for 72 h. For conditioned medium experiments, macrophages were isolated from Rosa26-creER^T2-Cgas^WT/R241E mice and differentiated in the presence or absence of 4-OHT (600 μM). Their conditioned medium (from day 5 to day 8) was collected, filtered through a 22 μm mesh and added (diluted 1:2 in neurobasal medium) onto neuronal cultures for 72 h in addition or not to mouse anti-TNF (Bio X Cell, XT3.11, BE0058, 25 μg ml⁻¹) or anti-IFNβ (R&D, MAB8234, 10 μg ml⁻¹) neutralizing antibodies. For the analysis of co-culture experiments, counts were averaged per slide from 4 to 5 fields of view, obtained from 3 mice (Rosa26-creER^T2-Cgas^WT/R241E microglia) or averaged per mouse (n = 3), obtained from 3 to 4 fields of view (aged microglia).

The Click-iT EdU reaction was performed according to the instructions of the Click-iT imaging kit (molecular probes by Life Technologies). Four images per condition were acquired using a ×40 magnification objective. The number of EdU-positive cells and DAPI-positive cells was determined using ImageJ.

Human fibroblasts (WI-38 and BJ) were either irradiated with 12 Gy (RS-2000 X-Ray Irradiator), treated with 1 μM abemaciclib (Selleckchem, S7158) or with 5 μg ml⁻¹ bleomycin (Chemie Brunschwig, TRCB595750). The medium was changed every 72 h and, after 10 days, the cells were treated daily with STING inhibitor H-151 for 10 more days (0.5 µM). To obtain replicative senescent cells, WI-38 cells were serially propagated (at each passage, after reaching 80% confluence, cells were trypsinized and diluted 1:4) until proliferation ceased (approximately 90 days), and were then treated daily with STING inhibitor H-151 (0.5 µM) for 1 to 3 weeks. Mouse BV2 microglial cells were irradiated with 10 Gy (RS-2000 X-Ray Irradiator), the medium was exchanged every other day and the cells were treated daily with DMSO or H-151 (1 µM) from day 4 to day 6 after irradiation.

BV2 cells were pretreated with 2′,3′-didoxycytidine (ddC, Sigma-Aldrich, D5782) for 6 days (20 µM, renewed every other day) before irradiation (10 Gy), the medium was exchanged every other day and the cells were collected for analysis on day 6. To verify the integrity of ρ⁰ BV2 cells, they were stimulated with LPS (from Escherichia coli serotype EH100(ra), Enzo, 100 ng ml⁻¹) or dsDNA 90-mer transfection (2 μg per well of a 6-well plate, using Lipofectamine 2000 reagent, Thermo Fisher Scientific, according to the manufacturer’s instruction) and collected 3 h after.

For whole-cell mtDNA analysis, control and ddC-treated BV2 cells were collected on day 6. Total DNA was purified using the DNeasy Blood and Tissue Mini Kit (Qiagen, 69504) according to the manufacturer’s instructions. For analysis of the cytosolic fraction in BV2 cells, DNA was isolated from either control or irradiated cells (wild-type and ρ⁰, collected on day 6 after irradiation) using the Mitochondria/Cytosol Fractionation Kit (BioVision, K256) according to the manufacturer’s instructions. For the analysis in young and aged mouse primary microglia, cells were fractionated using digitonin isolation buffer (150 mM NaCl, 50 mM HEPES, 25 μg ml⁻¹ digitonin) for 10 min on ice, then centrifuged at 2,000g for 5 min to separate cytosolic and membranous extracts. DNA was isolated from all fractions using the DNeasy Blood and Tissue Mini Kit (Qiagen, 69504) according to the manufacturer’s instructions. Isolated DNA (diluted to 20 ng ml⁻¹) was used as a template for qPCR analysis of mitochondrial DNA sequences Mito and CoI expression, using B2m as a nuclear control.

Super-resolution Airyscan images were acquired on a Zeiss LSM 980 with Airyscan microscope (Carl Zeiss). Samples were prepared on the CellCarrier 96 ultral microplate (PerkinElmer). TOMM20 and dsDNA were labelled with monoclonal rabbit antibodies against TOMM20 (Abcam, ab232589, 1:500) and monoclonal mouse antibodies against dsDNA (Sigma-Aldrich, MAB1293, 1:500), respectively. The samples were then washed and incubated with Alexa Fluor 568/488 secondary antibodies. The nucleus was stained with Hoechst 33342 (Sigma-Aldrich, B2261). Data were collected using a ×63/1.4 NA objective for the majority of experiments, under the control of the Zeiss ZEN software; 405, 488 and 561 nm laser lines were used. In cases in which z-stacks were collected, the software-recommended optimal slice sizes were used. Airyscan image processing was performed using the Airyscan processing function with the standard mode in the ZEN software. To maintain clarity and uniformity throughout the paper, the images were pseudocoloured.

Acquired z-stacks were imported into IMARIS (Bitplane). For cytosolic DNA focus quantification inside and outside the mitochondrial outer mitochondrial, a surface was created using the TOMM20 channel and spots were generated from the dsDNA channel. The shortest distances between spots to surfaces were obtained and analysed to distinguish dsDNA inside or outside mitochondrial outer mitochondrial. Distances more than 1 μm were excluded. Distances with positive or negative values were defined as outside or inside the mitochondrial outer mitochondrial. Images were analysed using IMARIS and ImageJ, including the snapshot images in 3D-view mode.

Human BJ fibroblasts were irradiated (12 Gy) and, when fully senescent (day 10), were transfected with siSTING or siNC (20 ng of RNA per 6 well) using Lipofectamine 2000 reagent (Thermo Fisher Scientific) according to the manufacturer’s instruction and cultured for 72 h thereafter. Silencer select predesigned siRNAs were purchased from Life Technologies (hsSting siRNA 1, S50644; hsSting siRNA 2, S50646).

Supernatants of human explant tissue and fibroblast cell cultures were collected and centrifuged at 1,000g at 4 °C to remove cell debris and dead cells. ELISA was performed according to the instructions of the Human IL-6, IL-8 and MCP-1 ELISA sets from BD Biosciences (BD OpetEIA, 555220, 555244, 555179, respectively).

Mouse tissues and cells were lysed in Pierce RIPA Buffer (for tissues, disruption was obtained with the homogenizer (Thermo Fisher Scientific)), and a protease inhibitor cocktail (Sigma-Aldrich, 11836170001) was added to prevent protein degradation. The protein concentration of the lysate was measured using the Pierce BCA Protein Assay (Thermo Fisher Scientific, 23227) to normalize the 2′3′-cGAMP concentrations measured by enzyme-linked immunosorbent assay (ELISA) according to the manufacturer’s instructions (Cayman Chemical, CAY-501700-96).

Adult mice (aged) were perfused, through the heart, with 50 ml of a buffered mix of 1% glutaraldehyde and 2% PFA in 0.1 M phosphate buffer (pH 7.4). The animal was left for 2 h after the perfusion had finished, and the brain was then carefully dissected from the skull and placed into PBS. Next, 80-µm-thick sections were cut with a vibratome, in the sagittal plane, and sections containing the CA1 region of the hippocampus were collected. The sections were then post-fixed in potassium ferrocyanide (1.5%) and osmium (2%), then stained with thiocarbohydrazide (1%) followed by osmium tetroxide (2%). Staining was performed overnight in uranyl acetate (1%) followed by washing in distilled water at 50 °C, and another staining with lead aspartate at the same temperature. The sections were dehydrated in increasing concentrations of ethanol and then embedded in Spurr’s resin and hardened at 65 °C for 24 h between glass slides.

To collect serial electron microscopy images of microglia in the CA1 region of the hippocampus, small (approximately 0.5 mm square) blocks were trimmed from the rest of the section using a razor blade and glued to an aluminium stub using conductive glue. Trimming with a glass knife produced a small (approximately 300 × 300 µm) block that was then mounted inside a scanning electron microscope (Zeiss Merlin, Zeiss NTS) holding a block face cutting microtome (3View, Gatan). Layers of resin, 50 nm thick, were cut from the block surface, and sequential images, targeting microglia, were collected after each layer was removed. An acceleration voltage of 1.7 kV was used with a pixel size of 7 nm and a dwell time of 1 µs. A series of nearly aligned images was collected, and the final alignment was performed using ImageJ. Counts of mitochondria were made through the image series, and each one was scored according to its morphology. The number of those appearing with heavily disrupted membranes and distorted morphologies was counted against those that appeared normal.

Data are presented as mean ± s.e.m. unless otherwise indicated. The sample number (n) indicates the number of patients, mice, ROIs or cellular experiment repeats, as specified in the figure legends. We used two-sided Student’s t-tests to compare paired or independent samples, one-way analysis of variance (ANOVA) followed by Tukey’s or two-sided ANOVA followed by Sidak’s multiple-comparison tests as indicated and an adjusted P < 0.05 was set as the cut-off. Data distribution was assumed to be normal, but this was not formally tested. The experimenters were not blinded to the experimental conditions, and no randomization was performed. Details of the statistical analysis are provided in the corresponding figure legends. Excel v.16.72 was used to combine data from multiple experiments or datasets. Prism v.9.0 was used to generate graphs and calculate statistics using appropriate statistical tests depending on the data, including two-sided paired and unpaired t-tests, and one-way or two-way ANOVA. Adjusted P values were assessed using appropriate correction methods, such as Tukey, Sidak and Geisser–Greenhouse.

Further information on research design is available in thelinked to this article. Nature Portfolio Reporting Summary