DNA damage in tissue-resident macrophages leads to age-related neurodegeneration

Neurodegenerative disorders are a growing challenge for the elderly yet their etiology remains elusive. Here, we show that persistent DNA damage in tissue-resident macrophages carrying an ERCC1-XPF DNA repair defect leads to cerebellar ataxia in mice. We nd that cytoplasmic chromatin fragments accumulate in the brain microglia of progeroid and naturally aged mice stimulating a type-I Interferon (IFN-I) response and are then packaged in extracellular vesicles (EVs) leading to Purkinje cell death and neurodegeneration in Er1 CX/− animals. To reduce neuroinammation, we developed an EV-based strategy to deliver recombinant DNase I specically in inamed Er1 CX/− microglia in vivo. Our approach rapidly removes dsDNAs from the cytoplasm of microglial cells and in secreted EVs; it alleviates the IFN-I response, decreases Purkinje cell death and delays the onset of neuronal decline in Er1 CX/− animals. Thus, brain microglia causally contribute to neurodegeneration allowing for the development of promising therapeutic strategies against age-related neuroinammation. strategy deliver recombinant DNase 1 nuclease in inamed Er1 CX/− microglia cells in vivo. show that the EV-delivered nuclease cargo rapidly removes dsDNAs from the cytoplasm of microglia cells as well of microglia-derived Er1 CX/− EVs, thereby maximizing the benecial outcome of the treatment. At the cellular level, we nd that the EV-based removal of cytoplasmic DNAs restores LaminB1 disorganization in the nuclear membrane of Er1 Cx/− microglia cells and considerably lowers the secreted IFNα levels in microglia cell culture media. In mice, the EV-based therapeutic strategy lowers the percentage of activated Er1 Cx/− microglia the substantial decrease of Purkinje cell death that considerably improves motor coordination in Er1 CX/− therapeutic promising strategy to combat age-related and improve the outcome of disorders with aging


Introduction
To counteract DNA damage, cells rely on a series of partially overlapping DNA repair pathways to preserve their vital genetic information and faithfully transmit it to progeny 1 . Besides cancer, the great majority of DNA repair defects in man gives rise to a heterogeneous group of progeroid (premature aginglike) syndromes that present frequently or exclusively with mild to severe neurological symptoms 2 .
Indeed, in view of the DNA repair defect, the high oxidative load and metabolic by-products generated during normal brain metabolism are thought to trigger irreparable DNA modi cations driving neuronal cell death and the progressive degeneration of the central or peripheral nervous systems 3 .
Xeroderma Pigmentosum F-Excision Repair Cross Complementation group 1 (XPF-ERCC1) is a heterodimeric, structure-speci c endonuclease complex required for lesion excision in nucleotide excision repair (NER) 4,5 that plays an analogous role in the repair of highly cytotoxic DNA interstrand crosslinks (ICLs) 6 . Patients with mutations in XPF may present with mild symptoms of the NER-de cient syndrome XP or exhibit adult-onset progressive neurologic deterioration, including chorea, ataxia, cognitive de cits and brain atrophy 7 . Recently, a single patient was discovered with mutations in ERCC1 resulting in severe pre-and postnatal developmental defects as well as pronounced neurological alterations associated with cerebellar hypoplasia and blunted cortical gyri 8, 9 We and others have recently shown that DNA damage-driven in ammation causally contributes to agerelated, tissue degenerative changes [10][11][12][13] . However, the relative contribution of compromised genome integrity in distinct cell types, e.g. neurons or glial cells to age-related neurodegenerative features remains unknown. Using Er1 CX/mice with an engineered ERCC1-XPF defect only in tissue-resident macrophages, we provide evidence for a fundamental mechanism by which irreparable DNA damage leads to the accumulation of chromatin fragments in the cytoplasm of microglial cells. Using functional assays coupled to an advanced simultaneous 2-3 multiphoton analysis on acute brain slices, we nd that cytoplasmic chromatin fragments are sensed, stimulating a type-I IFN response. We nd that cytoplasmic chromatin fragments are packaged in microglia-derived Er1 CX/-EVs that selectively target Purkinje cells leading to their apoptotic death and the premature onset of neurodegeneration in Er1 CX/animals. To remove cytoplasmic dsDNAs from the cytoplasm of activated microglial cells and lessen the neuroin ammatory load in Er1 CX/mice, we developed an extracellular vesicle (EV)-based strategy to deliver recombinant DNase 1 nuclease in vitro and in vivo, targeting CD11b + cells of the brain. We nd that treatment with the EV-delivered nuclease cargo rapidly eliminates in ammatory dsDNA moieties from the cytoplasm of microglia cells and in microglia-secreted EVs. This approach alleviates the Type-I IFN response and Purkinje cell death, leading to the substantial delay in the premature onset of cerebellar ataxia in Er1 CX/animals. Taken together, our ndings highlight the prominent role of tissue-resident macrophages in age-related neurodegeneration opening new avenues for the development of novel intervention strategies against neuroin ammatory disorders.

Results
Loss of ERCC1 in tissue-resident macrophages triggers progressive ataxia in mice. How DNA damage leads to the premature onset of heterogeneous pathological features in NER patients and accompanied animal models remains an intriguing question arguing for cell type-speci c responses against genotoxic threats. Tissue-resident macrophages are a heterogeneous group of immune cells that reside in distinct tissue environments and are vital for tissue homeostasis and defense against foreign pathogens or environmental challenges 14 . To dissect the impact of irreparable DNA damage in tissue-resident macrophages, we intercrossed animals homozygous for the oxed Ercc1 allele (Ercc1 F/F ) with mice carrying the CX3CR1-Cre transgene in an Ercc1 heterozygous background (from now on denoted as Er1 CX/− animals). CX3CR1 is a CX3C chemokine receptor 1 for fate-mapping studies of the tissue-resident monocyte and macrophage compartment 15 . Confocal microscopy studies in CX3Cr1-Cre crossed with the Rosa YFP transgenic animals and in Er1 CX/− animals con rmed the Cx3Cr1-driven YFP expression ( Figure 1A) and the absence of ERCC1 expression in Er1 CX/− tissue-resident macrophages, respectively ( Figure 1B). Together, these ndings indicate the normative ERCC1 expression levels in Er1 CX/− tissues or cells other than the targeted cell population. Er1 Cx/− mice are born at the expected Mendelian frequency and present no developmental defects or other pathological features. At 4-months of age, however, Er1 Cx/ − mice manifest progressively signs of ataxia that become clearly evident in the 6-months old animals. In line, when the 6-months old wild-type mice (Er1 F/− ; wt.) are suspended by their tails, the animals extend and shake their hind limbs to maintain balance (Supplementary Video le 1). Instead, Er1 Cx/− mice keep their hind limbs in a clasped position ( Figure 1C; Supplementary Video le 2) and walk with a wide gait compared to age-matched littermate control animals. Rotarod assessment reveals that hind limb coordination de ciency is apparent in ~70% of the 6-months old Er1 Cx/− animals compared to Er1 F/− littermate controls ( Figure 1D). Beginning at 8-months of age, Er1 Cx/− animals develop kyphosis ( Figure   1E) and ne tremor to front legs. The premature onset of neurodegenerative features in Er1 Cx/− animals prompted us to assess the morphological and phenotypic characteristics of CNS-resident macrophages.
We nd that brain microglia in 6-months old Er1 Cx/− animals form nger-like protrusions, a hallmark of microglia activation that involves cellular locomotion and increased antigen presentation 16 ( Figure 1F).
Fluorescence-activated cell sorting (FACS) analysis of freshly isolated CD11b + cells derived from Er1 Cx/− brains revealed a substantial increase in cell size ( Figure 1G), in the expression of MHCII and CD86 proteins as well as the number of MHCII + CD86 + cells compared to CD11b + cells of Er1 F/− animals ( Figure   1H and). However, when compared to LPS-treated, fully activated pro-in ammatory microglia, Er1 Cx/− microglia appear to be in a primed, yet not fully activated sate ( Figure 1H and Supplementary Figure   S1A). Further work reveals a comparable number of microglial cells in the 3-and 6-months old Er1 Cx/− and Er1 F/− brains ( Figure 1I). The lack of in ltrating monocytes as assessed by the histological evaluation of 3-and 6-months old Er1 Cx/− brains ( Figure 1J), the normative CD45 expression levels in the 6-months old Er1 Cx/− brains ( Figure 1J; Supplementary Figure S1B) and the comparable number of Ly6C + bone marrow-derived macrophages in the 6-months old Er1 Cx/− and Er1 F/− brains ( Figure 1K) diminishes the possibility of peripheral immune cell in ltration in Er1 Cx/− brains.
Accumulation of cytoplasmic chromatin fragments triggers a type-I IFN response in Er1 Cx/− microglia.
Phosphorylated histone H2A.X (γ-H2A.X)-containing foci accumulate at sites of DNA breaks 17 . FACS analysis in single cell suspensions of Er1 F/+ brains reveal that the great majority of γ-H2A.X + cells are CD11b + NeuN-cells (Figure 2A; upper panel). Intriguingly, we nd that the percentage of γ-H2A.X + CD11b + NeuN − cells increases gradually from 3-to 12-months old Er1 F/+ brains indicating that microglia cells accumulate rapidly DNA damage compared to neuronal cell types with aging. Consistent with the DNA repair defect, we nd a signi cantly higher percentage of γ-H2A.X + CD11b + NeuN − cells in Er1 Cx/− brains compared to littermate controls across all age groups (Figure 2A; lower panel). The phosphorylated Ataxia telangiectasia-mutated protein (pATM) is a central mediator of the DNA damage response. Confocal microscopy studies con rmed the signi cant increase in the number of γ-H2A.X+ CD11b+ and pATM+ CD11b+ cells in Er1 Cx/− brains compared to Er1 F/− controls ( Figure 2B; as indicated). Interestingly, γ-H2AX and pATM accumulate in the nucleus as well as the cytoplasm of microglial cells in Er1 Cx/− brains. Unlike in Er1 F/− microglia cells, we also detect the presence of 4′,6-diamidino-2phenylindole (DAPI) dense foci marking the presence of chromatin in the cytoplasm of Er1 Cx/− freshly isolated microglial cells ( Figure 2C). DNA damage triggers the release of micronuclei containing whole or fragmented chromosomes into the cytoplasm 18 . Chromatin fragments are subject to autophagic degradation 19 or else accumulate stimulating a type-I IFN immune response 20 . The presence of chromatin fragments in the cytoplasm of cells prompts us to test whether nuclear DNA leakage in the cytoplasm of Er1 Cx/− microglia cells triggers the activation of a type-I IFN response, known to be associated with anti-viral and immune modulating functions 21 . Evaluation of the bioactive murine type-I IFN levels by means of the B16-Blue™ IFN-α/β cell line in the extracellular milieu (brain lavage) of the 6months old Er1 Cx/− and aged-matched Er1 F/− littermate control animals reveals signi cantly higher type-I IFN levels in Er1 Cx/− brains ( Figure 3A). Consistently, we nd an increase in the mRNA levels of several interferon signature genes (ISGs) i.e. ifnβ, i t2, i 207 and i 44 in Er1 Cx/− whole brain lysates ( Figure 3B). FACS analysis in the 6-months old Er1 Cx/− cerebella and cortices reveal an increase in pSTING (phosphorylated form of stimulator of interferon genes) protein levels known to act as a sensor of cyclic d-GMP and as an adaptor protein mediating the interferon response, once activated by DNA sensors like cGAS ( Figure 3C) 22 . In Er1 Cx/− microglial cells, we nd that cytoplasmic DAPI+ chromatin colocalizes with the autophagy protein LC3β known to be involved in membrane tra cking and substrate delivery to lysosomes 23 and lamin B1, a nuclear lamina protein associated with chromatin domains 24 that is markedly disorganized in Er1 Cx/− microglial cells compared to controls ( Figure 3D; as indicated and Supplementary Figure S2A). Consistently, we nd that pATM + DAPI + chromatin structures accumulate in the cytoplasm of Er1 Cx/− cells and are surrounded by P62, a sequestering protein that delivers substrates to forming autophagosomes 25 ( Figure 3E; as indicated and Supplementary Figure S2B). Together, these ndings indicate a physiological budding response of Er1 Cx/− microglial cells to remove irreversibly damaged chromatin fragments through nucleophagy. We recently showed that type-I IFN mediates the increase of lysosomal pH to delay autophagosomal degradation, thereby favoring antigen presentation and cytokine secretion 26 . In agreement, we nd that the lysosomal pH of freshly isolated CD11b + cells is signi cantly increased indicating that lysosomal acidity is compromised in Er1 Cx/− brain-derived microglia cells ( Figure 3F). The latter may well explain the pronounced accumulation of chromatin fragments in the cytoplasm of DNA repair-de cient Er1 Cx/− microglial cells. Next, we test whether cytoplasmic chromatin fragments also accumulate with natural aging. In line, we nd that DAPI-stained chromatin fragments accumulate in the cytoplasm of microglial cells derived from 24-months old naturally aged mice cells ( Figure 3G). Similar to 6-months old Er1 Cx/− animals, FACS analysis revealed an increase in pSTING protein levels in naturally aged cerebella ( Figure 3H and Supplementary Figure S2C). Er1 Cx/− microglia elicit antiviral-like response that triggers Purkinje neuronal cell death. A number of animal models carrying inborn defects in DDR (e.g. Atm −/− ) and genome maintenance (e.g. Csb m/m -Xpa −/ − ) develop cerebellar ataxia associated with Purkinje cell death 27,28 . However, unlike in DNA repairde cient animal models, Purkinje cells in Er1 Cx/− mice are DNA repair-pro cient and do not accumulate γ-H2AX foci in their nuclei ( Figure 4A). Nonetheless, TUNEL assay and staining for activated Caspase-3 in 3-and 6-months old Er1 Cx/− brain sections reveal increased cell death in the Purkinje cell layer of cerebellum ( Figure 4B; as indicated). Loss of Purkinje cells is also in line with the defective coordination of hind limbs and the ne tremor seen in Er1 Cx/− animals ( Figure 1C-D). Further work reveals that myelination is unaffected in the 6-months old Er1 Cx/− mice as assessed by confocal microscopy of uoromyelin and western blotting of MBP1 protein levels (Supplementary Figure S3A-B). FACS analysis of freshly isolated brain single cell suspensions using antibodies against Calbindin (for Purkinje cells), CD11b + (for microglia), Annexin V (for apoptosis) and Propidium Iodide (for cell viability) reveals cell death in ~30% of Purkinje cells in the 6-months old Er1 Cx/− brains ( Figure 4D). Instead, we nd that Er1 Cx/ − microglia show no difference in Annexin V or Propidium Iodide indicating that Er1 Cx/− brain microglia is tolerant to DNA damage ( Figure 4D and Supplementary Figure S4A). We previously showed that type-I IFNs and their downstream effectors are upregulated in Er1 Cx/− brains ( Figure 3A-B) at a level su cient to elicit a type I-IFN response in IFN-responsive cells. In Er1 Cx/− cerebella, we nd that type I-IFN receptor (IFNAR) protein levels are higher in the neuronal population than in microglia cells; the latter is not evident in Er1 Cx/− cortices or the hippocampi (Supplementary Figure S4B). Likewise, ifnα and ifnβ mRNA levels are preferentially higher in the 6-months old Er1 Cx/− cerebella compared to age-matched Er1 Cx/− hippocampi or cortices indicating the marked sensitivity of neurons inhabiting the cerebellum to microglia-elicited type I-IFN stimuli (Supplementary Figure S5). To test whether type-I IFN-mediated stimuli affect the survival of Purkinje cells, we performed ow cytometry studies to evaluate the IFNAR protein levels in this neuronal cell type. Our analysis in the 6-months old Er1 Cx/− cerebella revealed that the great majority of Purkinje cells (~63%) have signi cantly higher IFNAR protein levels compared to the Purkinje cell population of littermate control animals ( Figure 4E-F). Further analysis revealed that out of all IFNAR + cells in the Er1 Cx/ brain, Purkinje cells in the cerebellum are the only ones undergoing apoptosis ( Figure 4G). Taken together, our ndings indicate that ablation of ERCC1 in brain-resident macrophages is associated with loss of Purkinje cells that preferentially respond to microglia-elicited antiviral stimuli.
Er1 Cx/− microglia secrete extracellular vesicles carrying γH2AX-associated chromatin. Secreted Type I IFNs enhance innate immune responses via autocrine and paracrine mechanisms and induce expression of ISGs that trigger apoptosis in targeted cells to inhibit viral replication and spread 29 . In this work, the presence of cytoplasmic chromatin fragments in Er1 Cx/− microglia and the observed cell death of DNA repair-pro cient Purkinje cells prompted us to examine whether microglia cells secrete extracellular vesicles (EVs) carrying chromatin fragments that, in turn, target IFNAR + Purkinje cells in Er1 Cx/− mice. In support, sucrose gradient ultracentrifugation and FACS analysis for CD11b revealed the increased presence of microglia-derived (CD11b + ) EVs in the brain lavage of 6-months old Er1 Cx/− mice compared to Er1 F/− littermate control animals ( Figure 5A). Scanning and transmission electron microscopy in this fraction revealed that microglia-derived EVs maintain a size of ~100nm ( Figure 5Bi); Er1 Cx/− vesicles associate with a higher density cargo as shown by their darker center marking the presence of positively charged DNA moieties (also supported by the enriched phosphorus found by scanning electron microscopy analysis) surrounded by a double membrane (Figure 5Bii, Figure 5iii and Supplementary Figure S6). Western blot analysis con rmed the enrichment of CD11b and ALIX, known to be associated with the endosomal sorting complex required for transport 30 ( Figure 5B). FACS analysis with the DNA binding benzothiazole dye (PicoGreen) speci c for dsDNA in microglia-derived CD11b + EVs in the brain lavage of 6-months old Er1 Cx/− and Er1 F/− animals revealed a signi cantly higher percentage of Er1 Cx/− EVs carrying dsDNA moieties ( Figure 5C). Further analysis revealed that the great majority of dsDNA moieties in Er1 Cx/− EVs associate with the DNA damage marker γH2AX ( Figure 5D) and nuclear LaminB1 ( Figure 5E) further con rming the nuclear origin of accumulated γH2AX chromatin structures in Er1 Cx/− EVs. Western blot analysis of EVs derived from Er1 Cx/− and Er1 F/− brain lavages con rmed the enrichment of γH2AX and LaminB1, P62, a reporter of autophagic activity 25 and β-adaptin, one of the components of the AP-2 adaptor complex that binds ataxia telangiectasia-mutated gene (ATM) to assist with the tra cking of cytoplasmic vesicles 31 ( Figure 5B). Taken together, our ndings suggest that Er1 Cx/− microglia release EVs that are of nuclear origin carrying γH2AX-associated chromatin.
Microglia-derived Er1 Cx/− EVs target IFNα-responsive Purkinje cells triggering apoptosis. Next, we sought to examine whether Purkinje cells preferentially receive microglia-derived Er1 Cx/− EVs. To do so, we exposed Er1 F/− cultures of acute brain slices with EVs derived from 6-months old Er1 Cx/− and Er1 F/− brains that were labeled with the lipophilic green uorescent dye PKH67. To monitor the selective uptake of microglia-derived Er1 Cx/− EVs by calbindin + cells, we subjected Er1 F/− brain slices to simultaneous 2-3 multiphoton microscopy analysis allowing visualizing at least 200µm detection depths. The latter approach revealed that Er1 Cx/− EVs are preferentially taken up by Purkinje cells when compared to Er1 F/− EVs; of note, the selective targeting of Er1 Cx/− EVs to Purkinje cells is further propagated when Er1 F/− brain slices are exposed to IFNα. ( Figure 6A). Further work revealed the co-localization of PicoGreenstained dsDNA and calbindin in IFNα-treated Er1 F/− cerebella exposed to Er1 Cx/− EVs indicating that the preferential targeting of Er1 Cx/− EVs to Purkinje cells is followed by the release of the Er1 Cx/− EV dsDNA cargo in recipient cells ( Figure 6B). Staining of acute brain slices with caspase-3 revealed that the exposure of IFNα-treated Er1 F/− brains to Er1 Cx/− EVs for 6 hours is su cient to trigger Purkinje cell death ( Figure 6C). Thus, microglia-derived Er1 Cx/− EVs preferentially target and released their dsDNA cargo to IFNα-responsive Purkinje cells leading to apoptosis.
Intranasal delivery of DNase I-loaded EVs delays the DNA damage-driven neurodegenerative features in Er1 Cx/− mice. To examine the functional contribution of accumulated γH2AX-associated chromatin fragments in the in ammatory response of Er1 Cx/− microglia cells, we used NIH3T3 cells to generate EVs loaded with recombinant (pH-independent) DNase I. To maximize the selective targeting of DNase I EVs to Er1 Cx/− microglia cells, the NIH3T3-derived EVs were also decorated with a custom anti-CD11b peptide derived from a combination of a CD63 binding sequence i.e. CRHSQMTVTSRL 32 and the αMI-domain binding peptide CP05 i.e. RKLRSLWRR 33 . Using this approach, we nd that exposure of Er1 Cx/− microglia to DNase I EVs e ciently removes the great majority of cytoplasmic DAPI-stained chromatin fragments and ameliorates the DNA damage-driven LaminB1 disorganization in Er1 Cx/− microglia nuclei ( Figure 7A).
Consistently, exposure of Er1 Cx/− microglia to DNase I EVs leads to the marked decrease of secreted IFNα levels in the media of these cells ( Figure 7B). We reasoned that the NIH-and microglia-derived EVs can be fused, thereby minimizing the DNA load of Er1 Cx/− EVs to maximize the bene cial outcome of DNase Iloaded EVs. In agreement, FACS analysis of puri ed EVs from Er1 CX/− brain lavages pre-stained with PicoGreen and co-incubated for 4 hours with DNase I-loaded or control EVs revealed that the microgliaderived (CD11b + ) Er1 Cx/− EVs contain substantially less dsDNA compared to corresponding controls ( Figure 7C). Next, we sought to test for the in vivo e cacy of DNase I-loaded EVs in ameliorating the neurodegenerative features seen in the 6-months old Er1 Cx/− mice. To do so, 3-months old Er1 Cx/− animals were exposed to an intranasal delivery of DNase I-loaded EVs twice a week after treatment with a vasoconstrictor to prevent drainage of EVs from blood vessels into the tissues lining the nasal passages.
Importantly, we nd that treatment of Er1 Cx/− animals with DNase I-EVs leads to the substantial decrease in Purkinje cell apoptosis ( Figure 7D) and in the percentage of activated MHCII + CD86 + Er1 Cx/− microglia cells ( Figure 7E). Next, we monitored the motor coordination of the 3-months old Er1 Cx/− animals treated with DNase I-EVs by measuring the latency to fall from a rotarod apparatus. Our analysis revealed that hind limb coordination in the 3-months old Er1 Cx/− mice treated with DNase I-EVs improves over a period of 6 weeks when compared to age-matched Er1 Cx/− mice treated with control EVs ( Figure 7F). Further work revealed the substantial decrease of PicoGreen-stained dsDNA ( Figure 7G) in the microglia-derived (CD11b + ) EVs from Er1 Cx/− mice treated with DNase I-loaded EVs ( Figure 7H). Thus, the intranasal delivery of DNaseI-loaded EVs substantially reduces type I IFN levels and the microglia-derived dsDNAcarrying EVs, thereby delaying the premature onset of neurodegenerative features seen in Er1 Cx/− mice. Taken together, our ndings reveal that an intranasal delivery of DNase I EVs substantially decreases Purkinje cell death delaying the premature onset of neurodegenerative features seen in Er1 Cx/− mice.

Discussion
Until recently, endogenous DNA damage in post-mitotic neurons was thought to be the primary cause of age-related neurodegenerative disorders seen in DNA repair-de cient patients and accompanying animal models 34 . Using progeroid animals carrying a ERCC1-XPF DNA repair defect only in tissue-resident macrophages, we show that the gradual accumulation of persistent DNA lesions in brain microglia is su cient to trigger Purkinje cell death and cerebellar ataxia in mice. Importantly, Er1 Cx/− mice are born with Mendelian frequency, they grow normally, are fertile and show no visible pathological signs until adulthood. Beginning at 4-months, however, Er1 Cx/− mice exhibit marked signs of cerebellar ataxia associated with Purkinje cell death. The latter is unexpected as in Er1 Cx/− animals, Purkinje cells are pro cient in DNA repair with no signs of DNA damage accumulation. Instead, we nd that irreparable DNA lesions lead to the buildup of H2AX-assciated chromatin fragments in the cytoplasm of NERdefective, tissue-resident Er1 Cx/− macrophages. Importantly, cytoplasmic H2AX-associated chromatin fragments also accumulate in the microglial cells of naturally aged animals. Nuclear dsDNAs can passively diffuse into the cytosol when the nuclear envelope breaks down in mitotically dividing microglia cells 18 . In view of the DNA repair defect in Er1 Cx/− animals, the latter could also re ect a physiological response of the nucleus to remove byproducts of DNA damage repair or irreversibly damaged DNA fragments.

Materials And Methods
Animals. Animals homozygous for the oxed Ercc1 allele (Ercc1 F/F ) were intercrossed with mice carrying the CX3CR1-Cre transgene in an Ercc1 heterozygous background (Er1 Cx/− animals). Mice lacking the CX3CR1-Cre transgene in an Ercc1 homozygous background were used as wild-type controls (Er1 F/+ ). All animals were maintained in grouped cages in a temperature-controlled, pathogen-free animal facility (IMBB-FORTH) on a 12h light/dark cycle and were fed a normal diet (Lactamin, Stockholm, Sweden). Mice had access to water ad libitum. This work received ethical approval by independent Animal Ethical Committee at IMBB-FORTH. All relevant ethical guidelines for the work with animals were adhered to during this study.
Collagenase activity was halted with the addition of ice-cold medium to the resultant homogenate. After centrifugation, cells were resuspended in DMEM and further homogenized using a syringe (21G needle). Filtration through a sterile pre-moistened 40µm cell strainer was used to separate the clumped cells, meninges and tissue fragments. Following centrifugation, cells were resuspended in full medium. Primary microglial cells were isolated through CD11b + magnetic-bead selection after density gradient centrifugation on Percoll. Brie y, Percoll density gradient was prepared in polystyrene tubes by layering 5ml of 35% Percoll solution on top of 3ml 75% Percoll solution, in which the brain cells were resuspended after the nal centrifugation. 2ml of 1XPBS were loaded on top of the Percoll density gradient and microglia along with lymphocytes were separated by centrifugation at 800g for 40min at 4°C. The cell band formed between the 75 and 35% layer was harvested, cells were washed with PBS and diluted in standard growth medium. For CD11b + microglia cell isolation, the human and mouse CD11b (microglia) Microbeads (MACS, Miltenyi Biotec) were used. Microglia cells at a density of >50.000 cells per well were placed either on poly-L-lysine coated glass coverslips in a 24-well plate or directly in each well of 24-well plate, depending on the experiment. For the LPS treatment (50 ng/ml), microglia cells were plated on wells of a 24-well plate and incubated at 37 o C for 24 hrs. For confocal microscopy, microglia cells were plated (and cultured where needed) on poly-L-lysine treated coverslips and were incubated at 37 o C for 18 hrs and 3 hrs, respectively. All treatments were performed 20min after their complete attachment on culture plate/slide.
Immunochemistry on primary neuronal and microglia cells. Whole brain and brain areas (cortex, cerebellum, hippocampus) from Er1 Cx/− and Er1 F/+ animals were excised and neuronal and microglia cells were isolated as previously described. Once isolated, cells were placed on poly-L-lysine coated coverslips, xed with 4% F/A for 15min maximum, RT and washed 3X with 1X PBS, for 5min, RT. Permeabilization/Blocking was performed (B1 solution:1% BSA,0,5% Triton in 1X PBS) for 1h, RT. Primary antibodies in B1 solution were added on the coverslips, O/N, at 4 o C. The following day, coverslips were washed thrice (B2 solution: 0,5% Triton in 1X PBS) for 10min, RT and secondary antibodies were added, along with DAPI for 2h. Finally, coverslips were washed thrice with B2 solution for 10 min, RT and then they were put on microscope slides with 80% glycerol. Imaging was performed using SP8 confocal microscope (Leica).
Histology, Immunohistochemistry. Er1 Cx/− and Er1 F/+ animals were perfused and their brains were Samples were immunostained with the corresponding uorescently labeled antibodies for 1.5h, RT. A separate 10min incubation was carried out in DAPI and the slides were coverslipped with 80% Glycerol. Three 5min washes with 1X PBS were performed in between incubations. For histological analysis of Er1 F/+ and Er1 Cx/− tissues, samples were xed in 4% formaldehyde, para n embedded, sectioned and stained with Harris's Hematoxylin and Eosin Y solution. The TUNEL Staining was performed on brain cryosections using the in situ cell death detection kit, Fluorescein (11684817910, Roche Diagnostics, Mannheim, Germany), according to the manufacturer's protocol. In brief, the sections were xed in 4% formaldehyde for 1 h, rinsed with PBS (5 min, 2 times) RT and permeabilized in 0,1% Triton in 0,1% sodium citrate at 4 o C, for 8 min. The slides were again rinsed with PBS (5min, 2 times), and incubated in 50µL TUNEL reaction mixture for 1h at 37°C, dark. The reaction was terminated by rinsing the samples with PBS (5min, 2 times) and the sections were sealed and detected by a light microscope. The nuclei were stained with DAPI (1:500). A positive control was also performed using DNase I (50U/ml), MNase (10U/ml) and proteinase K (20µg/ml).
Quantitative PCR (QPCR). Quantitative PCR was performed with a CFX Connect Real-Time PCR Detection system device (BIORAD). The generation of speci c PCR products was con rmed by melting curve analysis. Each primer pair was tested with a logarithmic dilution of a cDNA mix to generate a linear standard curve (crossing point (CP) plotted versus log of template concentration), which was used to calculate the primer pair e ciency (E = 10 (−1/slope) ). Hypoxanthine guanine phosphoribosyltransferase1 (Hprt-1) mRNA was used as an external standard. For data analysis, the second derivative maximum EV isolation, labelling, loading and treatments. EVs were puri ed using the differential ultracentrifugation protocol. Brie y culture medium was centrifuged sequentially at 300 g, (10 min), 2000 g (10 min), and 10000 g (30 min) to remove dead cells and cell debris. Extracellular vesicles were isolated with an ultracentrifugation at 100.000g for 2h and were then puri ed using a 90-10% sucrose gradient. Puri ed EVs were collected after a nal ultracentrifugation at 100.000g for 2h. All ultracentrifugations were performed at 4 o C. For PKH67 staining, EVs were incubated with PKH67 (company, 500 mL 0.2 mM) for 5 min at room temperature (RT). Labelled EVs were diluted in 500 mL 1% BSA, and then pelleted at 100,000 g, washed with 1 mL PBS to remove excess dye, re-suspended in 1 mL PBS and then pelleted at 100,000 g before nal re-suspension. For the extravesicular labelling of EVs against antibodies and uorochromes, brain lavage-derived EVs were incubated with both of them at dark, ( were isolated and half of them were loaded with DNase I (Pulmozyme, Roche) and the chimeric peptide (3518, 1µgr/1µg EVs), as it was previously described, while the rest remained empty (naive EVs). Their administration was performed intranasally twice a week for 6 weeks, in 3-month-old mice and its effect on the motor coordination of Er1 Cx/− mice was monitored by rotarod latency assay.
Immunoblot analysis and antibodies. Immunoblotting brain cells were lysed in RIPA buffer (50 mM Tris-HCl at pH 8, 150 mM NaCl, 0.5% sodium deoxycholate, 1% Nonidet P-40 and 0.1% SDS) supplemented with protease and phosphatase inhibitors (Complete EDTA Free; Roche Applied Science) and equal amounts of proteins (50µgr) were subjected to SDS-PAGE on 7%, 10% and 14% gels and then transferred to PVDF membrane (Amersham Hybond). Membranes were blocked with 5% skimmed milk or 5% BSA in TBST and then incubated with primary antibodies. For western blot analysis of EVs, EV pellets were resuspended in 5X Laemli buffer, sonicated for 5 circles and loaded in the gel. Samples were normalized using antibodies for housekeeping genes (β-tubulin). The image was resolved by ECL system (Thermo

Multiphoton microscopy
A single wavelength fs laser source was used to excite simultaneously three-and two-photon uorescence (3p-F and 2p-F, respectively) and perform three -color multiphoton imaging microscopy in the acute brain slices. For this purpose, a custom-build multiphoton microscope (Supplementary Figure  S7A) was used, based on a 1030 nm fs laser (Pharos-SP, Light Conversion, Vilnius, Lithuania), which is passing through a pair of galvanometric mirrors (6215 H, Cambridge Technology, Bedford, MA, USA) before entering into an inverted microscope (Axio Observer Z1, Carl Zeiss, Jena, Germany) 37,38 . The beam is then re ected by a short pass dichroic mirror (FF700-SDi01, Semrock, Rochester, NY, USA) placed at the turret box of the microscope and is focused in to the sample plane with a 20x 0.8NA objective-lens (Plan-Apochromat 20x/0.8NA, Carl Zeiss). The emitted uorescence is collected by the same objective and is ltered by a short pass lter (FF01-680/SP, Semrock) to ensure that no laser light is reaching the detectors. Then, the beam is split by a long-pass dichroic mirror (509-FDi01, Semrock), which re ects wavelengths shorter than 509 nm and let pass wavelengths longer than 509nm. The re ected wavelengths are further ltered by a band-pass lter (FF1-458/64, Semrock), which allows passing the wavelengths in the range of 458 ± 32 nm) before reaching a detector, based on a photomultiplier tube PMT-blue (H9305-04, Hamamatsu, Hizuoka, Japan). The transmitting wavelengths longer than 509 nm, are reaching a second long-pass dichroic mirror (FF-580-FDi01, Semrock) which re ects the wavelengths shorter than 580 nm and let pass the wavelengths longer than 580 nm. The re ected wavelengths are further ltered by a band-pass lter (FF01-527/20, Semrock), which allows passing the wavelengths in the range of 527 ± 10 nm) before reaching a second detector, the PMT-green (H9305-04, Hamamatsu).
Finally, the wavelengths longer than 580 nm are passing through a band-pass lter (FF-595/31, Semrock), which allows passing the wavelengths in the range of 595 ± 15.5) before reaching the third PMT-red (H9305-04, Hamamatsu). In our experiments the PMT-blue detects the uorescence emitted from the dye DAPI, while the PMT-green and the PMT-red detect the uorescence originating from the dyes Alexa Fluor 488 (or PKH67) and Alexa Fluor 555, respectively (Supplementary Figure S7B-C). Quanti cation of colocalization analysis was performed using JACoP plugin in ImageJ, while the co-localization indicators presented in Figure 6 were calculated using the Manders' coe cient 39 . supernatants were added, along with a positive (murine IFNα) and negative control (growth medium). The plate was incubated at 37 o C for 3h. After 3h incubation, SEAP levels were detected by using a spectrophotometer (Techam) at 620-655 nm.

B16 -Blue
Acute brain slices (SNAPSHOT method). Brains from Er1 Cx/− and Er1 F/+ animals were excised and sliced (400µm) using a vibratome. Acute brain slices were transferred in a 12-well plate containing fresh ACSF 5% FBS) and each slice was incubated with the diluted primary antibodies in a small plastic bag made by using a Manual Impulse Sealer for 6 to 10 days at 4°C on a platform rotator or a 360° rotisserie wheel. After incubation with primary antibodies, slices were washed with permeabilizing/washing solution three to ve times over the course of a day. Secondary antibodies, along with DAPI were diluted to the required concentrations in staining solution and slices were incubated with the diluted secondary antibodies in a small plastic bag for 4 to 6 days at 4°C on a platform rotator or a 360° rotisserie wheel. Fluorophores were protected from exposure to light by wrapping the bags in aluminum foil. The acute brain slices were washed once again with permeabilizing/washing solution three to ve times over the course of a day and they were incubated with DAPI for 4 more hours. Acute brain slices were then rinsed three to ve times in PBS and prepared for imagining. To image the tissue slices, each slice was placed on a microscope slide prepared with the slide, cover glasses, and Krazy Glue, using a transfer pipet with the tip cut off. A small drop of PBS was added on top of the brain slice before the placement of the cover glass over the brain slice. Finally, corn oil was added to each side of the microscope slide and imaging was performed with a two-photon scanning microscope.
Rotarod assay. To assess motor impairment, Er1 Cx/− and Er1 F/+ animals were subjected to rotarod assay test. Brie y, mice need to keep their balance on a rotating rod by walking forward. One day before testing, mice were trained at a constant rotating mode of 5 rpm for 2 min. During testing, mice were initially placed in their lanes, with the rod rotating at 5-rpm constant speed to allow their positioning. Once all mice were able to walk forward, the acceleration test was performed, in which the rod accelerated from 5 rpm to 70 rpm in 60 sec. The time (latency) it took each mouse to fall off the rod rotating under continuous acceleration (from 5 to 70 rpm) was recorded, as well as the reason for trail end (e.g., falling, jumping). The temperature, humidity, ventilation, noise intensity and lighting intensity were controlled and maintained at levels appropriate for mice. All mice were kept in a uniform environment before and after testing to avoid anomalous results being obtained.
Quanti cation and Statistical analysis. A two-tailed t-test was used to extract the statistically signi cant data by means of the IBM SPSS Statistics 19 (IBM) and the R software for statistical computing (www.rproject.org). Data analysis is discussed also in the Method Details section. Experiments were repeated at least 3 times. The data exhibited normal distribution (where applicable). There was no estimation of group variation before experiments. Error bars indicate standard deviation unless stated otherwise    Accumulation of cytoplasmic chromatin fragments triggers a type I IFN response in Er1Cx/-microglia.

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download.